Rnf-213 Knockout Induces Pericyte Reduction and Blood-Brain Barrier Impairment in Mouse

Moyamoya disease (MMD) is a rare cerebrovascular disorder characterized by progressive occlusion of the internal carotid artery and the formation of an abnormal compensatory capillary network at the base of the brain. Genomics studies identified Ring finger protein 213 (RNF213) as a common genetic factor that increases the susceptibility to MMD in East Asian people. However, the function of RNF213 and its roles in pathogenesis of MMD is unclear. Here, we showed that genetic knockout of Rnf213 in mice causes significant pericyte reduction and blood-brain barrier impairment in the cortex. These phenotypes are accompanied with microglia activation and elevated level of proinflammatory cytokines. Additionally, Rnf213-deficient mice showed reduced expression of tight junction proteins, including Occludin, Claudin-5, and ZO-1. Together, these data suggested that RNF213 might contribute to the pathogenesis of MMD through disruption of pericyte homeostasis and blood-brain barrier integrity by dysregulation of inflammatory responses and tight junction formation.


Introduction
Moyamoya disease (MMD) is a rare cerebrovascular disease that significantly increases the risk of ischemic and hemorrhagic stroke in both children and adults [1][2][3].MMD is characterized by chronic progressive stenosis in the terminal portions of bilateral internal carotid arteries in the circle of Willis, with compensatory development of collateral vessels network which possess a "puff of smoke" appearance and therefore termed "moyamoya vessels" in cerebral angiography [4].The pathological characters of moyamoya vessels include intimal thickening and reduplication of the elastic lamina, partial dilatation with discontinuity of the elastic lamina, microaneurysm formation, dilative change of the vessels with medial fibrosis, and rupture of the vascular wall [5].Recent genome-wide association analysis and exome sequencing identified ring finger protein 213 (RNF213) as one of the strongest susceptibility genes for MMD in East Asian people [6,7].RNF213 is localized on chromosome 17q25.3and encodes a 591 KDa protein that possesses a AAA-type ATPase domain, an alpha-2-macroglobulin Wei Li and Xingyang Niu have contributed equally to this work and share first authorship.
domain, and a ring finger domain from its amino to carboxyl terminus [6].Previous biochemistry studies revealed that the ring finger domain of RNF213 might have E3 ubiquitin ligase activity, suggesting a potential role of RNF213 in proteasome-mediated degradation pathways [8,9].Our recent study revealed that knockout of rnf213 in zebrafish causes mulberry-like cluster of disordered-flow vascular channels [10], suggesting a potential role of rnf213 in cerebral cavernous formation.However, previously published studies showed that Rnf213(− / −) mice did not reveal stenosis or occlusion or gross malformations in the circle of Willis development or significant difference in thickness of vascular wall compared with wild-type mice [11][12][13].These results suggest that the cerebrovascular pathology in Rnf213-associated MMD might be secondary to other defects, which might be manifested by other factors, such as aging, genetic backgrounds, and environmental exposures.
Blood-brain barrier (BBB) is a highly selective barrier that sanctions the entry of macromolecules, cells, and pathogens from blood into the central nervous system (CNS) [14,15].BBB is mainly formed by vascular endothelial cells, which are sealed by continuous complexes of tight junctions [16,17].The function and homeostasis of BBB is also highly dependent on extracellular matrix, astrocytes, and pericytes [18].BBB disruption can lead to substantial leakage of peripheral molecules into the CNS, which might cause deleterious inflammatory and neurotoxic responses.Although the etiology of MMD remains unclear, previous studies suggested that BBB impairment might play critical roles in the pathogenesis of MMD [19,20].Pathology studies showed that MMD patients have significantly higher level of BBB impairment compared to atherosclerotic cerebrovascular disease patients [21].Importantly, recent studies indicated that RNF213 might be a key regulator of BBB integrity [22], suggesting that RNF213 might contribute to the pathogenesis of MMD through disruption of BBB integrity.
Pericytes are mural cells of brain micro-vessels uniquely positioned within the neurovascular unit (NVU) between endothelial cells, astrocytes, and neurons [23].Pericytes ensheath the capillary wall and make direct contact with vascular endothelial cells.The inter-cellular signal transduction between pericytes and vascular endothelial cells plays critical roles in maintaining key neurovascular functions of the brain, including formation and maintenance of the BBB, angiogenesis, and regulation of capillary blood flow [24].The proliferation, migration, and recruitment of pericytes to the vascular wall are controlled by endothelial-secreted platelet-derived growth factor B (PDGF-BB), which binds to the platelet-derived growth factor receptor beta (PDGFR-β) on pericytes to initiate downstream signaling pathways [24,25].Previous studies reported that the expression level of platelet-derived growth factor receptor (PDGFR) was significantly reduced in MMD patient samples compared to controls [26].Furthermore, pericyte loss and dysfunction is found in neurological disorders associated with neurovascular dysfunction and BBB breakdown, such as brain arteriovenous malformations [27], cerebral cavernous malformations (CCM) [28,29], stroke [30], cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy [31], and microaneurysm [32].These studies suggest that pericyte dysfunction might contribute to neurovascular diseases through disruption of BBB.
Herein, we have found that Rnf213 deficiency causes significant pericyte reduction in mouse.We propose that Rnf213 might be a key regulator of pericyte maturation and the BBB permeability.

Animals and Surgery
Mice at 2-day postnatal (P2) or 3-week, and 8-10 weeks of age were used in the experiments.Mice were housed in a specific pathogen-free environment under the condition of optimal temperature, humidity, and light-dark shift cycle with food and water ad libitum.Rnf213-deficient (Rnf213(− / −)) mice on a C57BL/6 background were designed and purchased from the Laboratory Animal Center.Wild-type (WT) C57BL/6 were purchased from experimental animal center of Sun Yat-Sen University.Rnf213 homozygous knock-out mice are used to mate with WT mice to generate Rnf213 heterozygous knock-out (Rnf213(+ / −)) mice.Rnf213(+ / −) mice are bred to generate Rnf213(− / −) and WT littermates.Rnf213 homozygous littermate mice were used as the experiment group and the WT littermate mice were used for the control.All strains were maintained in the condition described above.All animal experiments were performed according to the guidelines of the National Institutes of Health Guide for the Care and Use of Laboratory Animals and approved by the Sun Yat-Sen University Animal Program Animal Care and Use Committee.All animal experiments were performed in a double-blinded manner.

In Vivo Two-Photon Microscope
Age-matched mice (8-10 weeks) were exposed to the thinned-skull window preparation as described in previous study [33].In brief, mice were anesthetized and fixed with a custom-fabricated metal frame by holding the head with a cyanoacrylate and dental cementon the stage of Leica DM6000 CFS (Lecia Germany) equipped with a waterimmersion objective lens (25 ×).The skull over right somatosensory cortex was carefully thinned to ~ 20-30 μm within an area 3 mm in diameter.Imaging was performed within 30 min of window construction.Data acquisition and laser scanning were performed using Leica Application Suite Advanced Fluorescence 2.5 software, at a wavelength of 860 nm.To monitor the BBB permeability using detection of leaked dyes, 100 μl Rhodamine B isothiocyanate-Dextran (1.0% in saline, 70 KDa molecular weight, Sigma-Aldrich) was injected intravenously to visualize the brain vasculature.Red fluorescence channel was used for detecting the fluorescence intensity in the extravascular compartment.Images of the XYZ stacks (512 × 512 pixels) were collected to a depth of 150 μm (2-μm step size) below the cortical surface, at 5, 10, 20, 30, 45, and 60 min after the injection.

Quantification of Pericyte Coverage and Numbers
The quantification analysis of pericyte coverage and numbers was restricted to CD13-positive pericyte that were associated with brain capillaries defined as vessels with ≤ 6 μm in diameter, as previously described [34,35].
For pericyte coverage, 10-µm maximum projection z-stacks (area 640 × 480 μm) were reconstructed, and the areas occupied by CD13-positive (pericyte) and lectin-positive (endothelium) fluorescent signals on vessels ≤ 6 μm were subjected separately to threshold processing and analyzed using Image J. First, black and white 8-bit images for CD13 and lectin signals were thresholded separately using Otsu's thresholding plugin that minimize the intraclass variance of the thresholded black and white pixels.After thresholding, the integrated signal density for each thresholded image was calculated.In order to express the integrated signal density as the area of the image (in pixels) occupied by the fluorescent signal, the integrated signal density was divided by 255 (the maximum pixel intensity for an 8-bit image).The integrated pixel-based area ratios of CD13 and lectin fluorescent signals were used to determine pericyte coverage as a percentage (%) of CD13-positive surface area covering lectin-positive endothelial capillary surface area per field, as previously reported [34].In each animal, 4-6 randomly selected fields in the somatosensory cortex S1 region (S1Cx) were analyzed in 4 non-adjacent sections (~ 100 μm apart), and averaged per mouse.
For pericyte numbers, 10-µm maximum projection z-stacks were reconstructed, and the number of CD13-positive pericyte bodies that co-localized with DAPI-positive nuclei on the abluminal side of lectin-positive endothelium on vessels ≤ 6 μm counted using Image J Cell Counter plugin, as previously described.In each animal, 4-6 randomly selected fields (640 × 480 μm) in the cortex were analyzed in 4 non-adjacent sections (~ 100 μm apart), and averaged per mouse.The number of pericytes was expressed per mm 2 of tissue.

Evans Blue Dye (EBD) Extravasation
2% EBD (3 mL/kg, Sigma) was injected via the tail vein in adult mice (8-10 weeks old) and juvenile mice (3 weeks old).Two hours post-injection, mice were anesthetized and perfused transcardially with saline solution, followed by perfusion with 4% PFA.Whole brain samples were collected for analysis.EBD extravasation was performed as previously described [36].Briefly, the brain specimens were weighed (wet weight of each sample was 50 mg), homogenized in 1 ml of 50% trichloroacetic acid, and centrifuged at 15,000 × g for 20 min.0.5 ml of the resultant supernatant was added to 1.5 ml of anhydrous ethanol.The absorbance was measured at 620 nm for a colorimetric assay using a fluorescence spectrophotometer (Ex620 nm, Em680 nm) to determine the EBD concentration.The EBD content (per mg of wet weight) within the brain tissue was used to determine the BBB permeability rate of EBD.For immunofluorescence, brains were dissected and post-fixed in 4% paraformaldehyde for 2-6 h.Embedded brain samples were coronally cut into 15-μm-thick slices using a freezing microtome (Thermo NX50).Brain slices were blocked with 5% BSA in 0.3% Triton X-100 for 60 min at room temperature and then using DAPI to visualize nuclear.Slices were shielded with Fluoroshield™ and covered with glass.Images were captured with a fluorescence microscope (Nikon, Tokyo, Japan).

RNA Extraction and Gene Expression Analyses
Total RNA was extracted using the Trizol reagent and whole RNA was reverse transcribed and converted to complementary DNA (cDNA) through Prime Script RT Master Mix (Takara, RR036A).Relative gene expression was evaluated by RT-qPCR (NovoStart®SYBR qPCR SuperMix plus) on real-time PCR detection system (BioRad CFX96 Touch).Relative gene expression for each target gene was calculated as the 2 (-ddct) values after normalization to the housekeeping gene; all the gene name and primer sequences were detailed as follows: F-NF-kB:GCG TAC ACA TTC TGG GGA GT, R-NF-kB: CCG AAG CAG GAG CTA TCA AC.F-IL-6:CCG GAG AGG AGA CTT CAC AG, R-IL-6:TCC ACG ATT TCC CAG AGA AC.F-IL-1β:GAA GTC AAG AGA AAA GTG G, R-IL-1β:ACA GTC CAG CCC ATA CTT T. F-iNOS: TCC TAC ACC ACA CCA AAC , R-iNOS: CTC CAA TCT CTG CCT ATC C.

Statistical Analysis
The Image J software (National Institutes of Health, Bethesda, MD, USA) was used to analyze the immunofluorescence results.All the data were showed in the form of mean ± SD.P < 0.05 is the predetermined standard of statistical significance.Normality test was performed for all the dataset we collected.If the data distribution follows normal distribution, parametric analysis (Student's t-test) was performed.If the date distribution does not follow normal distribution, non-parametric analysis (Mann-Whitney test) was performed.The statistical analysis method used for each dataset is described in the figure legend.All the graphs were created by GraphPad Prism 8.

Blood-Brain Barrier Impaired in Rnf213(− / −) Mice
To investigate the effect of the RNF213 gene on the cerebrovascular phenotype, we generated Rnf213 gene-deficient mice using CRISPR-Cas9 genome editing (supplementary Fig. 1).Rnf213 knockout is validated using Sanger sequencing and immunoblotting of Rnf213 protein.To investigate whether the BBB is impaired in Rnf213(− / −) mice, we injected EBD via the tail vein to assess the permeability of whole cerebral vessels in adult (8-10 weeks old) and juvenile (3 weeks old) Rnf213(− / −) mice and agedmatched WT.In adult and juvenile mice injected with EBD, the uniform bluish color of the brain was observed in the Rnf213(− / −), but not in age-matched WT mice (Fig. 1a and  d).Measurement of EBD extravasation showed significant EBD leakage into the brain parenchyma in adult and juvenile Rnf213(− / −) mice but not in WT mice (Fig. 1b and e).Microscopy analysis of coronal sections further revealed that EB fluorescence (red) was detected in the brain parenchyma of adult and juvenile Rnf213(− / −) mice (Fig. 1c and f).In contrast, EB retention was seen only in the endothelium in age-matched adult WT and juvenile animals (Fig. 1c and f).
Next, we performed in vivo time-lapse multiphoton imaging of cortical vessels with intravenous injection medium size macromolecular dextran (MW = 70,000 Da, Rhodamine B isothiocyanate-Dextran) in 8-10 weeks old Rnf213(− / −) and age-matched WT mice.In vivo time-lapse multiphoton imaging revealed that the intravascular dye began to leak into the extravascular space at the time of 10 min post-injection in 8-10-week-old Rnf213(− / −) mice, with the intravascular dye being accumulated in the extravascular space over time (Fig. 1g).Quantification of the tertramethylrhodamine fluorescence intensity in the extravascular compartment showed that the average fluorescence intensity in the extravascular compartment was significantly increased over the 60-min time of observation post injection (Fig. 1h).In contrast, no intravascular dye leakage was observed in age-matched WT mice at any time point (Fig. 1g).Together, these data suggested that Rnf213 deficiency caused the BBB impairment.Recent studies suggested that RNF213 might regulate cerebral endothelial cell functions in vitro, and potentially control the BBB integrity [22].To investigate the role of RNF213 in BBB-associated cell types in vivo, we generated Rnf213(− / −) mice and immuno-stained for surface markers of endothelial cell (Lectin +) and pericytes (CD13 +).Confocal images were collected from coronal sections of cortex to compare pericyte number and pericyte coverage in WT and Rnf213(− / −) mice.We found a substantial reduction of pericyte number in the cortex of Rnf213(− / −) mice compared to the corresponding agematched WT mice.In the cortex of P2 animals, ~ 14% pericyte reduction was observed in Rnf213(− / −) mice compared to the age-matched WT (Fig. 2a, b).The difference of pericyte number between Rnf213(− / −) mice and age-matched WT showed progressive increases with age, with ~ 23% and ~ 30% pericyte reduction observed in the cortex of 3-week (Fig. 2d, e), and 8-10-week-old (Fig. 3a, c) animals, respectively.In addition, pericyte number in WT mice showed a steady trend of increase until maturity in adulthood (~ 8 weeks), while the increase of pericytes in Rnf213(− / −) mice is significantly slower compared to WT mice (Supplementary Fig. 2a).Consistent with previous studies [11,37], there was no difference in growth and development between Rnf213(− / −) mice and WT mice (Fig. 3b).
Consistent with the loss of pericyte number, pericyte coverage determined as a percentage (%) of CD13-positive pericyte surface area covering lectin-positive endothelial surface area (see "Materials and Methods" for details), showed ~ 13%, ~ 22%, and ~ 27% reduction in the cortex of P2 (Fig. 2a, c), 3-week (Fig. 2d, f), and 8-10-week-old (Fig. 3a, d) Rnf213(− / −) mice compared to the age-matched WT, respectively.Consistent with the quantification of pericyte number, a progressive decrease of pericyte coverage with age in the cortex of Rnf213(− / −) mice was observed.In contrast, the pericyte coverage showed no significant change at different age-stage in the age-matched WT group (Supplementary Fig. 2b).

The Expression of BBB Tight Junction Proteins Was Reduced in Rnf213(− / −) Mice
BBB permeability is highly dependent on the expression level of cerebrovascular endothelial tight junction proteins.4a-f).Consistent with the immunofluorescence staining, analysis of protein expression using western blot showed diminished expression of Occludin, Claudin-5, and ZO-1 in the cortex of Rnf213(− / −) mice (Fig. 4g-j).Similarly, the analysis of protein expression using western blot in juvenile Rnf213(− / −) and WT mice also showed decreased expression of Occludin, Claudin-5, and ZO-1 in the cortex of Rnf213(− / −) mice (Supplementary Fig. 3).These observations suggest that the expressions of tight junction proteins are reduced in Rnf213(− / −) mice.

Glial Activation and Elevated Levels of Inflammatory Cytokines in Rnf213(− / −) Mice
Previous literature suggested that RNF213 might regulate inflammatory responses in endothelial cells and smooth muscle cells [22,38].To determine whether Rnf213 deficiency causes neuroinflammation phenotypes in vivo, we immunostained microglia (Iba-1 +) and astrocyte (GFAP +) in adult (8-10 weeks old) and juvenile (3 weeks old) Rnf213(− / −) and age-matched WT mice brain.Compared to age-matched WT, adult Rnf213(− / −) mice showed significantly increased number of Iba-1 + cells (Fig. 5a), indicating potential microglia activation in adult Rnf213(− / −) mice.Of note, microglia in WT mice displayed a ramified morphology with long branches and small cellular body (Fig. 5a), consistent with the previously reported morphology or resting-stage microglia [39].In contrast, microglia in adult Rnf213(− / −) mice showed amoeboid morphology with a larger cellular body and shorter branches were observed, resembling the morphology of chronically activated microglia.Additionally, the intensity of Iba-1 staining in Iba-1 + cells are significantly increased in adult Rnf213(− / −) mice compared to age-matched WT animals (Fig. 5b), further corroborating that the microglia are indeed activated in adult Rnf213(− / −) mice.Consistent with the Iba-1 staining, GFAP immunostaining showed that the number of astrocytes and the density of GFAP staining are both significantly increased in adult Rnf213(− / −) mice compared with age-matched WT animals (Fig. 5a, c); these data suggested that Rnf213 deficiency indeed causes gliosis phenotypes in adult mice.Additionally, neuron coverage in the cortex (determined as a percentage (%) of NeuN-positive neuron surface area covering the cortex) showed no significant difference between adult Rnf213(− / −) mice and the age-matched WT (Supplementary Fig. 4); this observation suggests that there is no significant neuron loss in Rnf213(− / −) mice.Subsequently, we measured pro-inflammatory cytokines in the cortex of the adult Rnf213(− / −) mice and age-matched WT.RT-qPCR revealed significantly increased level of NF-kB, IL-6, and IL-1β in adult Rnf213(− / −) mice compared to age-matched WT animals (Fig. 5d-g).These observations together showed that Rnf213 deficiency causes neuroinflammation phenotypes in adult mice brains.Of note, no significant increase of Iba-1 + and GFAP + cells was observed in 3-week-old Rnf213(− / −) compared to age-matched WT mice (Supplementary Fig. 5), suggesting that the neuroinflammation phenotypes are developed in an age-dependent manner in Rnf213(− / −) mice.

Discussion
Recent progress in genetic association studies showed that RNF213 is a strong genetic factor for MMD.However, little is known about the exact biological function of RNF213 in the pathogenesis of MMD.Previously published studies showed that Rnf213(− / −) mice did not develop moyamoya disease-like cerebrovascular pathology, including stenosis or occlusion in the circle of Willis development, or significant change in thickness of vascular wall [11][12][13].These results suggest that the cerebrovascular pathology in RNF213 associated MMD might be secondary to other defects, which are not thoroughly investigated in previous literature.Of note, previous studies suggest that BBB damage might be a driving factor in MMD pathogenesis [20,21].Therefore, in this manuscript, we focused on the role of Rnf213 in BBB damage, and tested the hypothesis that Rnf213 knockout might cause BBB damage through dysregulation of key molecules involved in tight junction formation.These results provide experimental evidence supporting that Rnf213 might increase the risk of MMD through disruption of BBB integrity.
In the present study, we showed that Rnf213 deficiency causes pericyte reduction, BBB impairment, and glia activation, and increases pro-inflammatory cytokines in the cortex of Rnf213(− / −) mice.Additionally, we demonstrated that the Rnf213 deficiency caused BBB impairment that is associated with decreased expression of the tight junction proteins.Taken together, our data revealed a possible role of Rnf213 in dysregulating pericytes and BBB homeostasis, which might in turn initiate pathogenic mechanism leading to the development of MMD.Pericytes are known be playing critical roles in maintaining cerebrovascular homeostasis, which are more abundant in the CNS compared to peripheral tissues/organs [23,25,40].It was previously demonstrated that pericyte reduction can lead to endothelial hyperplasia and VEGF-A upregulation [41].Of note, the expression level of PDGF receptors, which plays critical roles in controlling the pericyte population, is downregulated in moyamoya smooth muscle cells [26], indicating that the alteration in mural cell population might contribute to the pathological changes in MMD.In line with these observations, we showed that the expression level of PDGFR-β and the pericyte population are significantly decreased in Rnf213(− / −) mice, suggesting that downregulation of PDGF signaling pathway and pericyte population might contribute to the early development of RNF213associated MMD.Of note, previous literature showed that RNF213 is an E3-ubiquitin ligase.Therefore, it is tempting to hypothesize that PDGFR-β reduction is due to the dysregulation of proteasome-mediated degradation pathways.Future studies will be needed to understand the detailed molecular mechanism underlying the reduction of PDGF receptor level in RNF213-deficient animals, and to investigate whether PDGFR-β-mediated pericyte loss is associated with MMD progression in human samples.Interestingly, RNF213 variants have also been linked to increased susceptibility to intracranial major arterial stenosis/occlusion [42], ischemic strokes [43], CCM [10], and intracranial aneurysms [44].Previous studies suggest that pericyte loss and degeneration are also correlated with CCM [28], strokes [30], and intracranial microaneurysms [32].These observations indicate that RNF213 might contribute to these cerebrovascular disorders through a common pathway whereby RNF213-mediated downregulation in PDGFR-β signaling pathway might result in pericyte loss and downstream pathological changes.
BBB is a specialized border that selectively permits the transportation of substances from the circulating blood into the CNS.The integrity of BBB is critical to prevent toxic substances and pathogens from entering the brain [14,15].BBB is composed of endothelial cells connected by tight junctions, with pericytes and end-feet of astrocytes all attached to the capillary basement membrane [45].Previous studies have that BBB impairment is a common pathological character in cerebrovascular diseases including acute and chronic ischemia [46], cerebral small vessel disease [47], and vascular cognitive impairment [48].In line with these studies, BBB impairment is also reported in MMD patients.Recent evidence suggests that MMD patients are more vulnerable to cerebral hyper-perfusion syndrome following a revascularization procedure [49,50].Although the underlying mechanism of the potential complication of revascularization procedure for MMD patients is still undetermined, it was speculated to be caused by the intrinsic fragility BBB structure in MMD patients.Interestingly, biomarker studies showed that the expression level of matrix metalloproteinase 9 (MMP-9), a key protease controlling the degradation of the endothelial basal lamina and BBB integrity, is significantly elevated in MMD [51].Recent studies revealed that MMP-9 expression was also significantly elevated in Rnf213(− / −) mice compared to wild-type mice after common carotid artery ligation [52].Additionally, conditional knockout of RNF213 in brain endothelial cells showed increased BBB permeability [22].These data suggest that RNF213 might play critical roles in BBB homeostasis.In line with these studies, our data showed that the BBB was impaired in the cortex of adult Rnf213(− / −) mice, potentially through downregulation of tight junction proteins in the cortex of Rnf213(− / −) mice.This observation is consistent with recent transcriptomics and proteomics studies showing RNF213 deficiency significantly reduced tight junction protein expression [22].Importantly, pericytes are centrally positioned in the NVU between endothelial cells, astrocytes, and neurons, and maintain the integrity of BBB.Previous studies suggested that ablation of decrease in pericyte coverage dramatically damages the integrity of the BBB [40,45].Herein, we demonstrated that mild decrease (within 30%) of pericyte coverage caused by Rnf213 deficiency also caused significant BBB damage in mice.Indeed, previous literature suggested that pericyte loss might cause BBB leakage through dysregulation the transcytosis pathway [53].However, giving the dramatic BBB leakage observed in the Rnf213(− / −) mice, it would be technically challenging to investigate whether the leakage is caused by the dysregulated transcytosis pathway, or by the disruption of tight junction reported in this manuscript.Future studies, including electron microscopy to quantify transcytosis vehicles and RNA-seq to quantify the expression level of key genes controlling transcytosis pathways, might help to clarify the contribution of altered transcytosis in BBB damage induced by Rnf213 knockout.Besides, whether the pericyte loss is directly associated with the reduction of tight junction protein remained to be further investigated.
In addition to BBB damage, pro-inflammatory responses have also been implicated in the pathogenesis of MMD.Biomarker analysis using serum and cerebrospinal fluid showed significant elevation of pro-inflammatory markers in MMD patients [54][55][56].Pathology studies showed that the thickened intracranial arterial intima of MMD patients contains activated macrophages and infiltrated T cells [57].Importantly, lymphocyte accumulation within RNF213-deficient brain endothelial cell monolayers has been reported recently [22].In line with previous studies, we showed that microglia activation and expression of pro-inflammatory cytokines are significantly elevated in Rnf213(− / −) mice.Whether the inflammatory response is the cause or the consequence of the pathological changes, including pericyte loss and BBB damage, remained to be investigated.Of note, a recent study reported that in the absence of pericytes, the NVU becomes permissive to leukocyte entry, leading to aggravated neuroinflammation which can potentially cause autoimmunity [58,59].Our data suggests that Rnf213 deficiency in mice may play critical roles in the initiation of the immune response through disruption of pericyte homeostasis in MMD.It would be interesting to investigate whether pericyte loss precedes NUV disruption and immune cell infiltration at the early stages of MMD in patients.Nevertheless, the current study provided experimental evidence that warrants future studies on the potential protective effect of vascular therapies on neuroinflammation phenotypes in MMD.
In summary, our study showed the first time that Rnf213 deficiency causes significant pericyte reduction, BBB impairment, and neuroinflammation phenotypes in mice.The increased permeability of BBB might be caused by the reduced expression of the tight junction proteins in Rnf213(− / −) mice.Thus, our findings indicate that reduction of pericytes might be a key trigger in the pathogenesis of MMD.Future studies might include testing whether knockout of Rnf213 can exacerbate MMD-relevant cerebrovascular pathology using previously described MMD models, including ACTA2 mutant mice model [60], NEO1-deficient mice model [19], and chronic hypoperfusion model [61].

Fig. 1
Fig. 1 Blood-brain barrier broken in the cortex of Rnf213(− / −) mice.(a and d) Representative dissected brains of intravenously injected EB from WT (left) and Rnf213(− / −) mice (right).(n = 5).Note the uniform bluish color of the brain of the Rnf213(− / −) mice, which is absent in brains of WT mice.(b and e) Representative quantification of Evans blue dye extravasation into brain tissue (n = 5).Each dot represents one animal.Significance is assessed using Student's t-test.** P < 0.01.(c and f) Representative EBD fluoresence (red) is detected in the brain parenchyma after 2 h of circulation in Rnf213(− / −) mice.In control animals, EB retention is seen only in