CNS resident macrophages enhance dysfunctional angiogenesis and circulating monocytes infiltration in brain arteriovenous malformation

Myeloid immune cells present abundantly in both ruptured and unruptured brain arteriovenous malformations (bAVMs). The role of central nervous system (CNS) resident and circulating monocyte-derived macrophages in bAVM pathogenesis has not been fully understood. RNA sequencing using cultured cells and bAVM samples revealed that downregulation of activin-like kinase 1 (ALK1) or endoglin (two bAVM causative genes) increased pro-angiogenic, endothelial inflammation and innate immune signaling, which provided endogenous underpinnings of the active inflammation in bAVM. To further understand the role of CNS resident macrophages in bAVM development and hemorrhage, we administrated a colony-stimulating factor 1 receptor (CSF1R) inhibitor to bAVM mice with endothelial Alk1 deletion. Transient depletion of CNS resident macrophages at early stage of bAVM development remarkably mitigated the subsequent phenotype severity of bAVM. This therapeutic effect exhibited a prolonged inhibition of angiogenesis, dysplastic vasculature formation, and infiltration of CNS resident and circulating monocyte-derived macrophages during bAVM development. Transient depletion of CNS resident macrophages also reduced the dysplasia vessels and improved the integrity of endothelial tight junctions in established bAVMs. Administration of CSF1R inhibitor also prevented severe hemorrhage of bAVMs. Thus, endothelial AVM causative gene mutation can activate CNS resident macrophages promoting bAVM progression. CNS resident macrophages could be specific targets to mitigate the development and severity of bAVMs.


Introduction
Arteriovenous malformations (AVMs) are active angiogenic lesions consisting of tangles of abnormal vessels shunting blood directly from arteries to veins without a true capillary bed [1]. Innate immune response in bAVM is characterized by macrophage in ltration. An abnormally high number of macrophages are present in and around vascular walls in human brain AVM (bAVM) specimens, with or without hemorrhage, suggesting that macrophage accumulation is not simply a response to hemorrhage [2][3][4][5][6]. Polymorphisms in in ammatory cytokines and elevated expressions of in ammation-related genes in AVM patients further suggest their active roles in bAVM pathogenesis [7][8][9]. We have found increased macrophage burden in bAVMs in mouse models generated through conditional deletion of endoglin (Eng) or activin-like kinase 1 (Alk1, also known as ACVLR1), in combination with focal angiogenic stimulation [10][11][12][13]. ENG and ALK1 are causative genes of hereditary hemorrhagic telangiectasia (HHT), which is a familial disorder with high prevalence of bAVMs [14,1]. The activated macrophages were also found abundantly in bAVM mouse model with endothelial KRAS gain of function, which is a somatic mutation identi ed in sporadic bAVMs [15]. Macrophage accumulation in bAVM represents unresolved in ammation, which then can enhance abnormal vascular remodeling and the severity of the bAVM phenotype. It is unclear whether these known gene mutations directly contribute to neuroin ammation in bAVMs. Understanding the mechanism of macrophage accumulation may afford an opportunity to improve bAVM patient care [6].
In addition, persistent proin ammatory differentiation of macrophage has been found to be critical in bAVM progression. We identi ed both the central nervous system (CNS) resident and circulating myeloid cells as potential precursors of proin ammatory macrophages in bAVM nidus [16]. Recent studies of human bAVM tissues unraveled the crosstalk between immune cells and vascular components [17]. More importantly, a heterogenous spectrum of myeloid cells with typical molecular signature and spatial distribution were clustered, which are distinct in ruptured bAVM and associated with instability of AVM vasculature. The resident myeloid cells were the major immune cells in ltrating along the perivascular space and deeper brain tissue, while some speci c monocytes were over-represented in ruptured AVMs. It is undetermined which cell clusters act as initiators and which ones respond subsequently. Moreover, potential interplay between CNS endogenous and circulating myeloid cells is yet to be determined. Therefore, further exploration is needed to elucidate the distinct role of myeloid cell subsets and catalog more speci c therapeutic targets in bAVM.
Circulating monocytes can in ltrate the CNS and differentiate into macrophages in brain vascular disorders [18]. The recruitment of monocytes into tissues, including CNS, depends on monocyte chemoattractant protein (CCL2) and its receptor (CCR2). We found a delayed but persistent accumulation of CCR2 + cells in the bAVM lesions of mouse model [16]. CNS resident macrophages are established in the CNS since embryo and maintain their population independently of circulating monocytes. There are two major clusters of CNS resident macrophages described based on anatomical location, morphology and molecular signatures: microglia (parenchymal speci c macrophages) and non-parenchymal macrophages (further classi ed as perivascular, subdural meninges and choroid plexus macrophages) [19][20][21]. The ionized calcium-binding adaptor molecule 1 (Iba1), CX 3 CR1, and colony stimulating factor 1 receptor (CSF1R) markers are expressed across all of the CNS resident populations [19]. The stimulation from CSF1R is critical for the development and maintenance of CNS resident macrophages [22].
In this study, we demonstrated the endothelial mutation of AVM causative genes can upregulate neuroin ammatory signaling pathways, and CNS resident macrophages actively involved in bAVM pathogenesis.

Methods
Animals 8-to 10-week-old Alk1 f/f ;Ccr2 RFP/+ Cx3cr1 GFP/+ mice in C57BL background with two alleles of Alk1 gene (exons 4 to 6) anked by loxP sites [25], red uorescent protein (RFP) gene knocked into one allele of Ccr2 gene, and green uorescent protein gene (GFP) knocked into one allele of Cx3cr1 gene [26] and PdgfbCreER; Eng f/f mice with two alleles of Eng gene (exons 4 to 5) anked by loxP sites [27] were used. Equal numbers of male and female mice were included.
Experimental procedures for using laboratory animals were approved by the Institution of Animal Care and Use Committee of the University of California, San Francisco.

Induction of bAVM through stereotactic injection of viral vector
Brain AVMs were induced in Alk1 f/f ;Ccr2 RFP/+ Cx3cr1 GFP/+ mice through stereotactic intracerebral injection of viral vectors as described in our previous paper [10]. Mice were anesthetized through inhalation of 4% iso urane and placed in a stereotactic frame with a holder (David Kopf Instruments, Tujunga, CA). A burr hole was drilled in the pericranium 2 mm lateral to the sagittal suture and 1 mm posterior to the coronal suture. A total of 2 µL viral vector suspension containing 2×10 9 genome copies (gc) of AAV-VEGF (an adeno-associated viral vector expressing human vascular endothelial growth factor) and 2×10 7 plaqueforming units (PFU) of Ad-Cre (an adenoviral vector carrying CMV promoter driving Cre recombinase expression) were injected into the basal ganglia 3 mm beneath the brain surface. Ad-GFP and AAV-LacZ were used as control for Ad-Cre and AAV-VEGF, respectively. Mice were randomly assigned to each treatment groups using ipping a coin method.
Control mice were treated with 3 doses of corn oil and intra-brain injection of AAV-VEGF. Brain AVM tissues were collected 8 weeks after model induction. Total RNAs were isolated form bAVM lesions and brain angiogenic regions (controls) for sequence. To check the knockdown e ciency, the RNAs were reverse-transcribed into cDNA using SuperScript III First-Strand Synthesis System (Invitrogen, Carlsbad, CA). Real-time PCR was performed using TaqMan Fast Advanced Master Mix (Applied Biosystems, Foster City, CA). Gene-speci c primers and probes purchased from Applied Biosystems were used: ALK1 (Hs00953798_m1), and GAPDH (Hs02758991_g1). The relative gene expression was calculated using the comparative threshold cycle (CT) and normalized to GAPDH (ΔCT).
Total RNAs isolated from HUVECs with ALK1 downregulated more than 80%, and from bAVMs and brain angiogenic region of control mice were sent to Novogene Co (Davis, CA) for sequence using the company's standard protocol (Supplemental material 1). The outcome data were also analyzed by Novogene Co.
Administration of inhibitor of CSF1R (PLX5622) PLX5622 (180 mg/kg/day of body weight, Plexxikon Biotech Company, South San Francisco, CA) was incorporated in chow and oral administered for 7 days starting at 1 week or 8 weeks after model induction. The placebo chow was administrated in the same pattern to the control group.
Quantitative assessment of vessel morphology and macrophages Brain samples were collected 8 weeks or 9 weeks after model induction. After being anesthetized with iso urane inhalation, Cy5-uorescein-conjugated lycopersicon esculentum lectin (Vector Laboratories, Burlingame, CA) was injected via jugular vein to stain endothelial cells. Mice were then perfused with heparinized PBS through left cardiac ventricle to clear blood from vasculature followed by 4% paraformaldehyde. Brain samples were collected and incubated in 4% paraformaldehyde containing 20% sucrose until they sunk to the bottom of the solution. Brain samples were then snap-frozen in dry ice and sectioned.
Coronal sections (20-µm-thick) were cut using a cryostat (Leica, CM1900, Germany). Two coronal sections from each mouse were chosen, 0.5 mm anterior and 0.5 mm posterior to the needle track. Sections were coverslipped with Vectashield mounting medium with 4'-6-diamidino-2-phenylinidole (DAPI, Vector Laboratories) to label cell nuclei. Three images were taken under 20 x objective eld of each section (left, right, and below the injection site) using a uorescent microscope (Keyence BZ-9000, Itasca, IL).
Vascular density (number of vessels per mm 2 ), RFP + and GFP + cells (per mm 2 ) and claudin 5 + , CD31 + vessels were quanti ed in using NIH Image 1.63 software. Dysplasia vessels were counted manually.
Dysplasia index (the number of vessels with lumen diameter larger than 15 µm/mm 2 ) were used to express the quantify of dysplasia vessels.

Prussian blue staining
Iron Stain Kit (Sigma-Aldrich, St. Louis, MO) was used to detect iron deposition. Slides were incubated in freshly prepared working iron stain solution for 15 minutes, washed in distilled water, and counterstained with pararosaniline solution for 3 minutes. Two sections per brain within the injection site were chosen for staining. Data presented as percentage of Prussian blue positive area in the hemisphere.

Statistics
For quanti cation of vessels density, dysplasia index, RFP + and GFP + cell numbers, vascular claudin 5 coverage and Prussian blue staining areas, section numbers were scrambled. The quanti cation was done by two researchers blinded to the treatment groups independently. Inter-observer discrepancy was controlled within one standard deviation, and the means were used for further analysis. Data are presented as mean ± standard deviation (SD). All data were analyzed through t test for two samplecomparation, or one-way ANOVA for multiple sample-comparation followed by Tukey's multiple comparisons using GraphPad Prism 9 software. A P value < 0.05 was considered to be signi cant. Sample sizes were indicated in gures.

Results
Downregulation of ALK1 or Eng in endothelial cells upregulated pro-in ammatory and innate immune signaling.
To understand the roles of Alk1 and Eng genes in endogenous angiogenesis and in ammation, we knocked down ALK1 in HUVECs and knocked out Eng in mouse endothelial cells. The transcriptional pro les of ALK1 de cient HUVECs, brain angiogenic regions of Eng de cient mice (bAVMs) and control mice were analyzed by RNAseq. We found that down regulation of ALK1 in HUVECs upregulated the expression of 507 genes and downregulated the expression of 563 gene compared to scrambled siRNA treated HUVECs. The gene transcriptional pro les are distinctive between ALK1 down regulated and control HUVECs (Fig. 1a). Gene Ontology (GO) enrichment analyses showed that knockdown of ALK1 in HUVEC increased the transcription of genes regulating angiogenesis, innate immune response and chemokine-mediated signaling pathway, including neutrophil and monocyte chemotaxis, interleukin-1 beta production and secretion, and tumor necrosis factor production (Figs. 1b & 1c).
We have also analyzed the changes of gene transcription in bAVMs of mice with Eng deleted in endothelial cells. We found 1243 genes were upregulated, and 1830 genes were downregulated in bAVM, compared to controls (brain angiogenic region of corn oil treated mice). GO enrichment and Kyoto encyclopedia of genes and genomes (KEGG) analyses showed that Eng endothelial de ciency increased the transcription of genes upregulating angiogenesis, leukocyte transendothelial migration, macrophage differentiation and macrophage chemotaxis (Fig. 1d-1f). GO enrichment analyses also show that Eng endothelial de ciency increased the transcription of gene sets related to myeloid leukocyte differentiation and migration, myeloid leukocyte activation, phagocytosis, activation and regulation of innate immune response, neutrophil chemotaxis, myeloid cell development, positive regulation of chemotaxis, glial cell proliferation and migration, phagocytic vesicle. Differential analysis shown that the transcription of Cx3cr1 (adjusted P < 0.001) and colony stimulating factor 1 (Csf1) gene (adjusted P = 0.003) were increased signi cantly in the bAVM of Eng endothelial de cient mice.
These data indicate that down regulation of ALK1 or Eng expression in endothelial cells increases endothelial cell in ammation, leukocyte extravasation from vessels, and activates tissue resident macrophages.
Transient depletion of CNS resident macrophages reduced the burden of Cx3cr1 + and Ccr2 + macrophages in bAVM To investigate the speci c function of CNS resident macrophage in bAVM in ammation, Cx3cr1 + CNS resident macrophages were transiently depleted through oral administration of PLX5622. We rst tested the e ciency of PLX5662 on depletion of CNS resident macrophages by administration of it to wild-type (WT) mice for 7 days. We found 93% of Iba1 + cells were depleted in the brains of treated mice (Fig. 2).
We next administrated PLX5662 in mice with bAVMs generated by focal deletion of Alk1 gene plus angiogenic stimulation [10] to test the effect of transient depletion of CNS resident macrophage on bAVM pathogenesis. PLX 5622 treatment for 7 days starting either at 1 week after model induction when the bAVM development begun (P = 0.002) or 8 weeks after model induction when bAVMs have established (P < 0.001) reduced Cx3cr1 + resident macrophages in bAVM lesions (Fig. 3).
Administration of PLX 5622 starting at 1 week after model induction when bAVM development begun also reduced circulating monocyte-derived macrophages (Ccr2 + ) in bAVMs (P < 0.001 compared to controls) (Fig. 3). However, PLX 5622 treatment did not reduce the number of circulating monocytederived macrophages in established bAVMs. Therefore, transient depletion of resident macrophages at early stage of bAVM development can reduce circulating monocytes in ltration into bAVM lesion, suggesting a pivotal role of resident macrophage in bAVM in ammation.

Transient depletion of CNS resident macrophages at the beginning of bAVM development reduced angiogenic activity in bAVMs
To explore the role of CNS resident macrophages on bAVM angiogenesis, we assessed the blood vessel densities within bAVMs. The vessel densities of bAVMs in mice treated with PLX5622 at the beginning of bAVM development (434.1 ± 128.9 vessels/mm 2 ) were lower than that in vehicle-treated mice (609.4 ± 40.5 vessels/mm 2 , P = 0.016) (Fig. 4). However, the vessel densities were similar in the bAVMs of mice treated with PLX5622 and vehicle 8 weeks after model induction (Fig. 4). Therefore, transient depletion of CNS resident macrophages at the early stage of bAVM development reduced the angiogenic activity, and the effect persistent after the treatment has stopped. Transient depletion of CNS resident macrophages alleviated bAVM severity. We next evaluate the effect of PLX5622 treatment on bAVM severity by examining the abnormal vessels with dilated and irregular lumen. PLX5622 treatment starting at 1 week after the model induction inhibited bAVM development. Even though the PLX5622 treatment had stopped for 6 weeks, there were still fewer dysplasia vessels in treated group (13.9 ± 4.2 vessels/mm 2 ) than control group (23.1 ± 2.4 vessels/mm 2 , P = 0.002). Administration of PLX5622 starting at 8 weeks after the model induction when bAVMs have already established also reduced the number of abnormal vessels (treated: 12.0 ± 4.6 vessels/mm 2 vs. control: 21.6 ± 3.9 vessels/mm 2 , P = 0.002) (Fig. 4). Notably, the number of Cx3cr1 + CNS resident macrophages was positively correlated with the number of abnormal vessels in the bAVMs (DI, r 2 = 0.63, P < 0.01, Fig. 5). Taken together, these data show that transient depletion of CNS resident macrophage can not only exhibit a long-term effect of ameliorating bAVM development but also reduce the severity of established bAVMs.
Transient depletion of CNS resident macrophages enhanced blood brain barrier (BBB) integrity and attenuated hemorrhage in bAVMs Our previous study revealed impairment of vascular integrity in bAVMs [11]. To explore the impact of CNS resident macrophages on BBB integrity of established bAVM, the expression of tight junction protein was quanti ed. We found PLX5622 treatment restored the expression of claudin 5 in established bAVM (Fig. 6), suggesting that transient depletion of CNS resident macrophage can rescue abnormal vessels and BBB integrity independently from circulating monocyte in ltration.
We further tested whether transient depletion of CNS resident macrophage reduce or prevent bAVM hemorrhage. Hemorrhage in bAVMs were detected by analyzing iron deposition using Prussian staining.
PLX5622 treatment starting at 1 or 8 weeks after model induction prevented severe hemorrhage (Prussian blue staining area > 1% of total hemisphere). No mouse in treated groups had hemorrhage area lager that 1% of total hemisphere, the average hemorrhage areas were 0.28% for mice received the treatment starting at 1 week and 0.29% for mice received the treatment starting at 8 weeks after model induction. In contrast, the hemorrhage was more severe in controls. Two out of six mice in 1-week vehicle group and two out of seven mice in 8-week vehicle group have hemorrhage area over 1% of total hemisphere (Fig. 7).

Discussion
Our ndings demonstrated that the genetic dysfunction of ENG and ALK1 in endothelial cells promote transendothelial leukocyte extravasation and activate innate immune system. Notably, CNS resident macrophages play an important role in promoting the development of bAVM by enhancing abnormal angiogenesis and in ltration of both resident and circulating in ammatory cells. Through pharmacological depletion of CNS resident macrophages transiently at different stages of bAVM development, we were able to investigate the temporal role of resident macrophages in bAVM development. In the initial stage of bAVM development, CNS resident macrophages enhance abnormal angiogenesis and the recruitment of blood-borne monocytes into the angiogenic foci. During the time window preceding the in ltration of Ccr2 + circulating cells, a transient suppression of CNS resident macrophages reduced the recruitment of Ccr2 + circulating cells and mitigated the development of bAVM. The effect extended 6 weeks after ceasing PLX5622 treatment. In the established bAVMs with abundant dysplastic vessels and in ammatory cells, CNS resident macrophage continued to play a critical role in inducing vascular impairment and hemorrhage. Short-term treatment with the PLX5622 in mice with the 'mature' bAVMs, that are more approximate to the clinical scenario, has restored dysplasia vessels towards normal phenotype and improved vascular integrity, which ultimately lead to reduction of severe hemorrhage.
Although several germline and somatic mutations have been reported in human bAVMs, the mechanism by which these mutations endogenously amplify the dysfunctional angiogenesis and maintain in ammation during bAVM development, is not fully understood. Recent studies suggested that clonal expansion of mutant endothelial cells contributes to this process [29]. However, the burden of pathogenic mutations is not consistently associated with the phenotype severity of human bAVMs, which are typically heterozygotes or somatic mosaicism [30]. Therefore, the synergistic cofactors upon genetic mutation have yet to be identi ed. The present study highlights the temporal modulating role of CNS resident macrophages in abnormal angiogenesis and in ammation during bAVM development.
Excessive and dysfunctional angiogenesis is a key process in bAVM development. Increased levels of VEGF and its receptors were found in human bAVM tissue. The single-cell pro le of human bAVM tissue identi ed a differentiation trajectory of endothelial cells to angiogenic clusters, which enriched with proangiogenic and pro-permeability gene transcription. These endothelial cells were predicted to be the major contributors of dysregulated angiogenic communication network in bAVM, including angiopoietin, VEGF, TGF-b signaling [17]. Here, we provided direct evidence, both in vitro and in vivo, that the ALK1 and ENG gene deletion in endothelial cells could endogenously activate pro-angiogenic signaling. The KRAS G12V -induced bAVM also revealed increased VEGF-A mRNA expression and phosphorilated-VEGFR2 which con ned to the KRAS mutant endothelial cells [15]. Our previous data indicated a higher proliferation of genetic aberrant endothelial cells than the wild-type endothelial cells in the early stage of bAVM development [29]. These endothelial cells with somatic mutations underwent clonal expansion in response to the angiogenic signaling. And the burden of the mutant endothelial cells determined the ratio of dysplastic vessels with dilated lumen in the mosaicism. Notably, the vessel density was not correlated with the number of mutant endothelial cells in our mouse model. It was congruent with the insigni cant association between presence or allelic burden of causative gene mutation and human bAVM lesion size [30]. These previous ndings suggested that the aberrant endothelium was critical to induce the vascular dysplasia, while the extent of angiogenesis or lesion size was not determined by the burden of endothelium mutation alone.
In the present study, transient depletion of CNS resident macrophages in the early stage of bAVM development reduced the extent of angiogenesis via measuring the average vessel density, suggesting a more crucial crosstalk between mutant endothelial cells and immune cells during the early development of bAVM in determining the ultimate size of angiogenic lesion. Our previous data shown that the angiogenesis extent would affect not only the lesion size but also the hemorrhage of bAVM [31]. Although the angiogenic extent could not be rescued in a fully formed bAVM through transient depletion of CNS resident macrophages, other bAVM phenotypes were attenuated, such as hemorrhage, impairment of vascular integrity and morphology. Therefore, the pharmacological depletion of resident macrophage might be a promising therapy for bAVM.
Persistent in ammation has been observed in bAVMs, which accompanies dysfunctional angiogenesis. Myeloid immune cells expressing Iba1 were identi ed as major in ammatory components in human bAVM. The increased Iba1 + cells were composed of two clusters: the P2RY12 − perivascular macrophage and the P2RY12 + microglia predominantly in adjacent brain of bAVM. Previous data has demonstrated a good correlation between the expression of Iba1 and Cx3cr1 in cells localized within perivascular space and brain parenchyma [32,23], indicating that Cx3cr1 may also be an optimal marker for CNS resident macrophages, including both microglia and perivascular macrophage. The fate mapping analysis of these brain Cx3cr1 + cells con rmed their common prenatal origin and their continuous residence in CNS since embryo [32,19]. These cells subsequently differentiate into a range of CNS resident macrophages in speci c niches: microglia, perivascular, meningeal, and choroid plexus macrophages [19,32]. However, there is also evidence that in ammation in bAVM involves more than just resident immune cells. Our previous study revealed that both the circulating monocyte-derived macrophage (Ccr2 + ) and resident microglia (Cx3cr1 + ) are abundantly in ltrated and persistently activated in bAVM [16]. The increased level of monocyte-derived macrophages was even more prominent than that of resident microglia. These ndings raised question regarding the extent to which the unresolved in ammation could be ascribed to, the CNS resident macrophages or the circulating monocytes. CD34 + peripheral blood cells of patient with ALK1 or ENG gene mutation are more likely to differentiate into activated myeloid immune cells [16]. In the present study, we revealed that the CNS resident macrophages play a major role in recruiting both resident and monocyte-derived macrophages into bAVMs and promoting bAVM progression. The CSF1R inhibitors have been shown to be effective in eliminating Iba1 + and Cx3cr1 + cells present in both the parenchyma and perivascular space of CNS, which represent microglia and perivascular resident macrophages, respectively but not impede the in ux of circulating monocytes into the brain [23]. In this study, we shown that depletion of CNS resident macrophages reduced the recruitment of Ccr2 + circulating monocyte at the early stage of bAVM development. Depletion of CNS resident macrophages transiently after bAVM has formed did not prevent the in ux of Ccr2 + cell into the bAVM, but reduced hemorrhage and vascular dysplasia, suggesting that CNS resident macrophages play a central role in inducing peripheral monocytes into bAVMs and enhancing bAVM hemorrhage. Thus, transiently depletion of CNS resident macrophage could be development into a therapy to reduce bAVM progression and hemorrhage.
CSF1R inhibitors have served as effective tools for investigating the interplay of peripheral and CNS resident myeloid population. These oral administrated small-molecule inhibitors achieve robust but reversible brain-wide resident macrophage elimination without reactive in ammatory response or cytokine storm [33,34]. Remarkably, pexidartinib (PLX3397) has been approved by FDA for the treatment of tenosynovial giant cell tumor in 2019 and is already in clinical trials for CNS tumors [35]. PLX5622 used in this study, has the same potency as its predecessor in CSF1R inhibition, but is 10-fold more selective in binding with the receptors and with better brain penetrance [34]. This highly selective brain penetrant CSF1R inhibitor can consistently deplete over 90% of microglia after a 7 day-treatment [24]. A sustained treatment of PLX5622 for 24 weeks is well-tolerated in rodents [34]. Upon cessation of PLX5622 treatment, the repopulation of microglia occurred within 24 hours and returned to the normal level in 36 hours [33]. The present study suggested a prolonged impact of PLX5622 in preventing bAVM progression, which lasted after microglia have repopulated. Remarkably, the e cacy of PLX5622 in elimination of resident macrophages in bAVM was not as effective as that in normal brain area, suggesting a consistent in ltration of macrophages in bAVMs despite the 7-day PLX5622 treatment.
Macrophage recruitment is predisposed by ALK1 or Eng gene dysfunction as indicated in our RNAseq data. It is important to determine whether an extended treatment period could further improve therapeutic effects of PLX5622 in bAVM, and the long-term outcome after treatment ceased in further study.
Notably, our present ndings were not based on the assumption that CSF1R inhibitor has minimal effects on peripheral immune cells. There are disputes that the population of peripheral circulating and bonemarrow myeloid cells could also be affected by PLX5622 treatment [36]. The peripheral circulating and bone-marrow derived myeloid cells rebound at 7 days after PLX5622 cessation [36]. Monocyte-derived macrophage in CNS lesions can present competitively increase after resident population elimination with CSF1R inhibitor [18]. In the present study, in ltration of Ccr2 + cells within bAVM lesion were not affected by CSF1R inhibitor in the established bAVMs, which is different from those observed in the peripheral blood and bone marrow of healthy wild-type mice. Moreover, we did not nd a rebound or competitive increase of Ccr2 + cells after PLX5622 treatment. The mice treated transiently with PLX5622 at the early stage of bAVM development reduced Ccr2 + cells in bAVMs analyzed 6 weeks after PLX5622 treatment stopped. The improvement of vascular phenotypes was comparable between mice with PLX5622 treatment at the early stage of bAVM development and after bAVMs had established. Therefore, the bene cial effect of CSF1R inhibition in bAVM severity appears to be independent of Ccr2 + cell in ltration. We also showed that the number of CNS resident macrophages was positively correlated with the number of dysplasia vessels in bAVMs. Taken together, our data shew that CNS resident macrophages may be more detrimental for bAVM progression than circulating monocytes.
In summary (Fig. 8), the present study revealed that genetic dysfunction of ALK1 or Eng predisposed a pro-in ammatory and pro-angiogenic microenvironment for abnormal vascular development. The CNS resident macrophages, including microglia and other non-parenchymal macrophages, play fundamental roles in orchestrating dysfunctional angiogenesis and persistent recruitment of both circulating monocytes and more resident macrophages into bAVMs promoting bAVM progression. A modulation of CNS resident macrophage would redice the vascular aberrancy, restore vascular integrity, and prevent severe hemorrhage of bAVMs.   AVM mice treated with vehicle; AVMP: AVM mice treated with PLX 5622; AngP: mice with brain angiogenesis and PLX5622 treatment; CP: mice received intra-brain injection of control vectors PLX5622 treatment. N=5-7.

Figure 4
PLX5622treatment reduced dysplastic vessels in mouse bAVMs. a. Representative images of brain sections collected from lectin-perfused mice. Vessels are labeled by violet uorescence. White arrows indicate dysplastic vessels. Scale bar= 50μm. b. Quanti cations of vessel density. c. Quanti cation of dysplasia index (number of vessels with lumen diameter >15 mm/mm2. C: vehicle treated. P: PLX5622 treated. 1W and 8W: treatment started at 1 and 8 weeks after model induction. N=6 or 7.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.