Similarities and differences between brain and skin GNAQ p.R183Q driven capillary malformations

Abstract

Capillary malformations (CM) are congenital vascular irregularities of capillary and venous blood vessels that appear in the skin, leptomeninges of the brain, and the choroid of the eye in the disorder known as Sturge Weber Syndrome (SWS). More common are non-syndromic CM found only in the skin, without brain or ocular involvement. A somatic activating mutation in GNAQ (p.R183Q) is found in ~90% of syndromic and non-syndromic CM specimens and is present in CD31^pos endothelial cells isolated from brain and skin CM specimens. Endothelial expression of the GNAQ p.R183Q variant is sufficient to form CM-like vessels in mice. Given the distinct features and functions of blood vessels in the brain versus the skin, we examined the features of CM vessels in both tissues to gain insights into the pathogenesis of CM. Herein, we present morphologic characteristics of CM observed in specimens from brain and skin. The GNAQ p.R183Q variant allelic frequency in each specimen was determined by droplet digital PCR. Sections were stained for endothelial cells, tight junctions, mural cells, and macrophages to assess the endothelium as well as perivascular constituents. CM blood vessels in brain and skin were enlarged, exhibited fibrin leakage and reduced zona occludin-1 and claudin-5, and were surrounded by MRC1^pos/LYVE1^pos macrophages. In contrast, the CMs from brain and skin differ in endothelial sprouting activity and localization of mural cells. These characteristics might be helpful in the development of targeted and/or tissue specific therapies to prevent or reverse non-syndromic and syndromic CM.

Introduction

Sturge Weber Syndrome (SWS) is a non-inherited, rare neurocutaneous disorder with atypical morphology and function of capillary and venous blood vessels. SWS is strongly associated with a cutaneous capillary malformation (CM), also known as port-wine birthmark, leptomeningeal vascular malformation in the brain, visible with contrast enhanced magnetic resonance imaging (MRI), and increased intraocular pressure and choroidal ‘hemangioma’[1, 2]. Some of the neurological manifestations include seizures, stroke-like episodes, hemiparesis, cerebral atrophy, calcification, developmental delays, and cognitive limitations. Patients with both upper and lower eyelid port wine birthmark are at high risk of developing glaucoma, which can lead to optical nerve damage and loss of vision[2]. Currently, there are no molecularly targeted drug therapies to treat SWS. Some patients, considered non-syndromic, have cutaneous CMs but no brain or ocular involvement.

As seen by histopathology, CM is composed of dilated, enlarged vessels with irregular shapes and thickened basement membranes[3]. Furthermore, CM vessels can be surrounded by multiple layers of disorganized mural cells or lack mural cell coverage altogether[3, 4]. CM endothelial cells (ECs) in non-syndromic cutaneous lesions appear to be immature as they express stem cell markers (CD133, CD166) as well as venous (EphrinB1) and arterial (EphrinB2) markers[3]. CM ECs in the leptomeninges show strong expression of HIF1α and HIF2α, which suggests sustained angiogenic signaling[5]. We recently showed strong expression of intracellular adhesion molecule-1 (ICAM-1) in ECs in leptomeningeal CM in SWS brain specimens and increased MRC1^pos/LYVE1^pos/CD68^pos macrophages; such cells were not found in non-SWS brain[4].

Syndromic and non-syndromic CM are caused by a somatic activating mutation in GNAQ[6], the gene that encodes G protein subunit alpha-q (Gα_q). The GNAQ p.R183Q mutation, located in the switch 1 domain of Gα_q, causes constitutive activation of Gα_q. The mutation is found in ~90% of port wine birthmark and ~88% of brain lesions from SWS patients[6]. Recent testing of an additional patient cohort identified mutations in the GNAQ homolog GNA11 and in GNB2 which encodes the G protein subunit beta 2, a protein that forms a heterotrimeric complex with Gα_q[7, 8]. The GNAQ mutation is highly enriched in EC isolated from brain and skin CM specimens, yet there are indications that other unknown cell types may carry the p.R183Q mutation[9-11].

Our motivation here was to assess differences between the CM in brain and skin by comparing histopathological characteristics. Our goal was to understand how the single amino acid change in Gα_q disrupts capillary-venous morphology in these two distinct tissues. Such insights could help in the development of targeted molecular and tissue-directed therapies to alleviate the morbidity and social impacts of CM and SWS.

Materials and Methods

Reagents and antibodies:

Primary antibodies:

Alpha smooth muscle actin Cy3 conjugated (mouse, 1:500, Sigma Aldrich, C6198); Calponin (rabbit, 1:100, abcam, ab46794); Claudin-5 (mouse, 1:50, Invitrogen, 35-2500); Desmin (rabbit, 1:100, abcam, ab15200); Endocan/Esm1 (goat, 1:50, R&D Systems, AF1810-SP); LYVE1 (goat, 1:150, R&D systems, AF2089SP); MRC1 (mouse, 1:100, Millipore Sigma, AMAB90746); NG2 (rabbit, 1:100, abcam, ab183929); Zona occludin-1 (mouse, 1:100, Thermo Fisher, 33-9100).

Secondary antibodies:

Goat anti-Rabbit IgG (H+L), Alexa Fluor^™ 488 (1:200, Invitrogen, A11008); Donkey anti-Mouse IgG (H+L), Alexa Fluor^™ 488 (1:200, Invitrogen, A11008); Donkey anti-Goat IgG (H+L), Alexa Fluor^™ 594 (1:200, Invitrogen, A32758); Donkey anti-Mouse IgG (H+L), Alexa Fluor^™ 488 (1:200, Invitrogen, A32766).

Other reagents:

Bovine serum albumin (Millipore Sigma, A8806); 4',6-diamidino-2-phenylindole - DAPI (Invitrogen, R37606); ddPCR SuperMix (BioRad, 1863010); FormaPure kit (Beckman Coulter, C16675); Normal goat serum (abcam, ab7481), Ulex europaeus agglutinin-I, UEAI Rhodamine (1:100, Vector labs, RL-1062); UEAI-Daylight 649 (1:100, Vector labs, DL-1068); Universal HIER antigen retrieval (abcam, ab208572);

Droplet digital polymerase chain reaction:

Human brain and skin specimens (Table 1) were obtained under a human subject protocol (IRB-P00003505) approved by the Committee on Clinical Investigation at the Boston Children’s Hospital. Droplet digital PCR (ddPCR) was used to measure the variant allelic frequency in tissue sections as previously described [12]. Briefly, DNA was isolated from formalin-fixed paraffin embedded (FFPE) tissue sections by scraping off the entire section from a slide using a FormaPure kit. ddPCR with probes to detect the GNAQ R183Q mutation (Table S1) was run with 15ng DNA per sample along with ddPCR SuperMix. The droplet fluorescence was read in a QX200 droplet reader and analyzed in QuantaSoft software (BioRad). In the analysis step, mutant droplets were normalized to wild type droplets for each specimen.

Table 1.

Patient demographic information for brain specimens.

	Age	Sex	Diagnosis	Mutation	Variant Allelic Frequency (%)	Location of the specimen
Patients with CM brain
1	2 years	Female	Sturge Weber Syndrome	GNAQ R183Q	2.96	Lateral temporal lobe
2	4 years	Female	Sturge Weber Syndrome	GNAQ R183Q	4.59	Left lateral temporal lobe
3	14 years	Male	Sturge Weber Syndrome	GNAQ R183Q	0.15	Right temporal lobe
4	4 years	Male	Sturge Weber Syndrome	GNAQ R183Q	2.5	Right posterior lateral temporal tip
Patients without CM brain
1	3 months	Female	n.a.	n.a.	Not detected	Right medial temporal lobe
2	3 years	Male	n.a.	n.a.	Not detected	Left frontal lobe
Patient with CM skin
1	12 years	Male	Port-wine stain	GNAQ R183Q	7.4	Face
2	15 years	Male	Port-wine stain	GNAQ R183Q	5.3	Face
3	28 years	Male	Port-wine stain/Sturge Weber Syndrome	GNAQ R183Q	11	Face
4	23 years	Female	Port-wine stain	GNAQ R183Q	4.1	Leg
Patient without CM skin
1	17 years	Male	n.a	n.a.	Not detected	Breast
2	7 years	Female	n.a.	n.a.	Not detected	Nevus

n.a. – not available

The 4 CM brain specimens and 2 brain specimens without CM are the same as those used in our previous study[4].

Immunofluorescence Staining:

5μm sectioning of human brain and skin specimens (Table 1) was performed by the Boston Children’s Hospital Department of Pathology. Sections were deparaffinized in xylene and series of ethanol incubation followed by 1X universal HIER antigen retrieval for 20 mins at 95°C. Sections were incubated in blocking buffer (10% normal goat serum, 0.3% Triton X-100, 1% bovine serum albumin in phosphatase buffer saline) at room temperature for 1 hour followed by primary antibody incubation at room temperature for 1 hour or at 4°C overnight. The primary antibody was diluted in a buffer containing 1% normal goat serum, 0.3% TritonX-100, and 1% bovine serum albumin. Sections were thoroughly washed (1X phosphatase buffer saline) before and after incubating in secondary antibody for 1 hour at room temperature. Lastly, slides were counterstained for 4′,6-diamidino-2-phenylindole (DAPI) for 5-10 mins before mounting. Images were taken on Zeiss Laser Scanning 880 confocal microscope and analyzed in Fiji software.

Image Quantification: Quantification of sprouting vessels:

Vessels with sprouting UEAI^pos cells were quantified manually. Any UEAI^pos vessel with two or more filopodia was considered a “sprouting vessel”. 5-10 images were taken from each non-serial section for each patient case at 10X or 20X magnification. All the images were analyzed, and the data was normalized to total number of vessels/mm².

Quantification of mural cell positive vessels and distance:

For mural cell quantification, all the vessels were manually categorized for three unique features: mural cell layer, no mural cell layer and distant mural cell layer. The quantification of distance between the calponin-positive or desmin positive vessels from the UEAI^pos endothelium was also manually measured. 5-10 images at 20X magnification were analyzed for each patient case.

Quantification of macrophage colocalization:

MRC1 and MRC1/LYVE1 colocalization were quantified as previously described[4]. In brief, for automatic colocalization, the Pearson Correlation Coefficient (PCC) and the Manders Overlap Coefficients served as the primary indicators for fluorescent colocalization. The PCC serves to represent how closely two markers follow a simple linear relationship. The Manders Overlap Coefficients--denoted by tM1 and tM2--reveal the proportion of cell area that shows colocalization of markers in comparison to the total measured fluorescence of one marker[13]. Next, the Costes Method of colocalization analysis was applied using the Coloc2 plugin in Fiji [14]. For each case, colocalization analyses were applied to each of the four to five images captured from each non-sequential sample. The brain specimens analyzed in this study are same as previously published[4].

Movat pentachrome Staining:

Tissue sections from the CM brain and skin specimens along with the non-CM brain and skin controls (Table 1) were sent for Movat Pentachrome staining to (iHisto, Salem, MA). MoticEasyScan Infinity 60 microscope was used for bright field scanning of each tissue section slide.

Statistical analysis:

Data are provided as mean ± standard error mean. Statistical packages in GraphPad Prism 9 were utilized for data normality and variance between the groups. Statistical differences were determined either by a two tailed t-test or One-way ANOVA (Brown-Forsythe and Welch ANOVA test) followed by Dunnett T3 post hoc multiple comparison test with significance considered as p<0.05.

Results

Sprouting endothelial cells in CM.

It has been reported that CM blood vessels are predominantly enlarged and irregular [3, 4] and express angiopoietin-2, a potent angiogenic factor that can promote vascular remodeling[15]. Using fluorescently tagged Ulex europeus agglutinin-I (UEAI) to visualize human ECs in CM histological sections, we observed what appear to be actively sprouting ECs in numerous blood vessels (Figure 1A-D). We quantified the number of UEAI^pos blood vessels with two or more sprouting EC in brain and skin CM specimens as well as control brain and skin specimens for comparison (Figure 1E). Interestingly, there was a significant (p<0.0001) increase in sprouting UEAI^pos blood vessels in CM skin (38.2 ± 3.4%) compared to non-CM skin (9.8 ± 3.6%). In contrast, there was a trend but no significant increase (p=0.6628) in sprouting ECs in CM brain (9.9 ± 2.5%) compared to non-CM brain (4.5 ± 2.9%). There was a significant (p<0.0001) increase in blood vessels with sprouting ECs in CM skin compared to CM brain (Figure 1E). This suggests a stronger angiogenic milieu may exist in the CM skin versus CM brain. Alternatively, features of the blood brain barrier may restrain sprouting behavior or the increased sprouting in the skin CM could be due to the increased (p=0.0473) GNAQ p. R183Q allelic frequency in the skin CM specimens versus brain CM specimens (Table 1) available for our study (Supplemental Figure 1). To further confirm that CM vessels are sprouting, we stained for Esm1, a secreted protein that is expressed by endothelial tip cells [16]. We found very few cells stained by the anti-Esm1 in CM brain as compared to CM skin, where we found majority of the ECs express Esm1 (Figure 1F).

Figure 1. Ulex europaeus agglutinin I (UEAI) staining for endothelium in capillary malformation of brain and skin.

UEAI (yellow) and counter staining for nuclei by DAPI in (A) CM brain (n=4) (B) CM skin (n=4) (C) non-CM brain (n=2) (D) non-CM skin (n=2). High magnification images for CM brain (green box) and skin (red box) are presented as inserts. Scalebar = 100μm. See Table 1 for information on tissue specimens. (E) Quantification of % UEAI^pos sprouting vessels/total vessels. The p-values were calculated by Brown-Forsythe and Welch ANOVA test followed by Dunnett T3 multi comparison test. 4-5 non-sequential sections were analyzed per tissue specimen. (F) Tip cell marker, endothelial cell specific molecule-1 (Esm1, magenta), UEAI (yellow) and counter staining for nuclei by DAPI in CM brain (left) and CM skin (right). Single channels are shown in the bottom row below each merged image. White arrows point to Esm1^pos ECs in CM brain. Scalebar = 50μm

Extravascular fibrin in CM brain and skin specimens.

We next looked at microenvironmental features. We performed Movat Pentachrome histological staining on the CM brain and skin specimens, and their respective controls to assess the extracellular matrix environment. Extracellular matrix is crucial for the maintenance of the vessel architecture as well as providing structure and function to the blood vessel. Previous Martius Scarlet Blue histological staining revealed extravascular fibrin in CM brain[4], which was confirmed here by Movat Pentachrome staining of CM brain specimens (n=4) (Figure 2A) and extended to reveal extravascular fibrin in CM skin specimens (n=4) (Figure 2C). In both brain and skin, extravascular fibrin was not detected in the controls (Figure 2B, D). These results uncover an apparent fibrin leakage in both brain and skin CM, suggesting comprised vascular integrity in both brain and skin CM from patients.

Figure 2. Movat pentachrome staining for elastin fibers and fibrin leakage in capillary malformation of brain and skin.

(A) CM brain (n=4) (B) CM skin (n=4) (C) non-CM brain (n=2) (D) non-CM skin (n=2). Elastic Fibers (black); Collagen (yellow); Fibrin (bright red); Nuclei (blue/black). Two high magnification images for each condition are presented as inserts. Scalebar = 200μm

Zona occludin1 and Claudin-5 in CM brain and skin specimens.

Endothelial tight junctions are needed to create the critically important EC-EC barrier between the circulation and the abluminal environment. The fibrin leakage detected (Figure 2) and described previously[4] suggests a potential weakening of EC-EC junctions. To assess this, we immunostained brain and skin CM sections for the tight junction protein zona occulin-1 (ZO1) and fluorescently tagged UEAI at the same time. Strikingly, we found majority of the blood vessels (UEAI^pos) in the brain CM (Figure 3A) and skin CM (Figure 3C) were negative for anti-human ZO1 immunostaining. ZO1^pos/UEA1^pos blood vessels were detected in control brain (Figure 3B) and skin (Figure 3D). We analyzed a second junctional protein Claudin5, a transmembrane protein of tight junctions[17]. We found enlarged vessels to be devoid of Claudin-5 in both CM brain (Figure 3E) as well as CM skin (Figure 3F) compared to their respective controls (Figure 3G, H). Overall, the lack of ZO1 and Claudin-5 in CM blood vessels could be responsible for a diminished endothelial barrier and fibrin leakage.

Figure 3. Loss of tight junction protein, zona occludin1 (ZO1) and Claudin-5 in capillary malformation of brain and skin.

ZO1 (magenta), UEAI (yellow) and counter staining for nuclei by DAPI in (A) CM brain (n=4) (B) CM skin (n=4) (C) non-CM brain (n=2) (D) non-CM skin (n=2). Claudin-5 (magenta), UEAI (yellow) and counter staining for nuclei by DAPI in (E) CM brain (n=4) (F) CM skin (n=4) (G) non-CM brain (n=2) (H) non-CM skin (n=2). Single channel images are on the right side of each merged image. Scalebar = 50 μm.

Disorganized mural cell coverage in CM skin compared to brain specimens.

Mural cells are important in vascular development, vascular patterning, and vascular contractility and tone. Furthermore, mural cells induce endothelial quiescence and thereby dampen endothelial sprouting. We have recently shown that some of the enlarged vessels in the CM brain lack mural cells expressing the markers calponin, desmin, and alpha smooth muscle actin (αSMA)[4]. Herein, we compared these findings to CMs in the skin (Figure 4 A, B). As previously shown, enlarged vessels tend to lack mural cells, identified by reduced immunostaining for calponin, desmin and αSMA. A notable difference in mural cell coverage of CM vessels in skin specimens was that the mural cells were separated from the UEAI^pos endothelium. This was particularly striking for the calponin^pos and desmin^pos cells in CM skin. In contrast, αSMA^pos and NG2^pos cells were detected on the abluminal side of the UEAI^pos endothelium suggesting heterogenous mural cell phenotypes in CM skin. By quantification, we found that 33% of the CM blood vessels in skin have mural cell coverage (magenta bar), ~38% have no mural cell coverage (black bar) and ~38% have a distant mural cell layer, labeled by anti-calponin or anti-desmin antibody (purple bar) (Figure 4E). The distance between the calponin^pos and desmin^pos cells and the UEAI^pos endothelium ranged from 5-15μm (Figure 4F) In contrast, control brain and skin specimens had a single layer of mural cell coverage surrounding the UEA1^pos endothelium (Figure 3B, D). Overall, CMs in the brain have fewer calponin^pos vessels compared to desmin and αSMA and NG2. Strikingly, CMs in the skin have distant mural cells labeled by anti-calponin and anti-desmin, which might be positive for αSMA.

Figure 4. Mural cell environment in capillary malformation of brain and skin.

Antibody staining for calponin, desmin, NG2 and alpha-smooth muscle actin (αSMA) (magenta) and UEAI (yellow) and nuclei staining with DAPI (blue) in (A) CM brain (n=4), (B) non-CM brain (n=2), (C) CM skin (n=4) (D) non-CM skin (n=2). Scalebar= 100μm. See Table 1 for information on tissue specimens. (E) Quantification of % number of vessels with four distinct features: mural cell layer (magenta), no mural cell coverage (black) and distant mural cell layer (purple) in CM brain, non-CM brain, CM skin, non-CM skin. (F) Quantification of vessels with distance between the mural cell layer (Calponin (left) and Desmin(right)) and UEAI^pos endothelium (μm).

Increased MRC1^pos and LYVE1^pos macrophages in brain and skin CM specimens.

Perivascular macrophages are distinguished by their anatomical location, molecular signature, as well as phagocytotic capacity [18]. There is growing evidence that the perivascular macrophages not only provide structural and functional maintenance of tissue homeostasis, but they also play a significant role in pathophysiological settings. We showed previously there are significantly increased numbers MRC1^pos/LYVE1^pos/CD68^pos macrophages in the CM brain compared to non-CM brain controls[4]. We extended the analyses to CM in skin, analyzing CM brain in parallel (Figure 5A, B, C, D). We first looked to see if there are increased MRC1^pos/LYVE1^pos macrophages in CM skin These cells were similarly elevated in skin CMs (p=0.0024) as in brain CMs (p=0.0310) compared to respective controls. (Figure 5E). When MRC1^pos cells were quantified, we found a significant (p<0.0001) increase in the number of MRC1^pos cells in CM brain compared to non-CM brain (Figure 5F). In contrast, there was no difference (p=0.2592) in MRC1^pos cells in skin CMs compared to control skin. Hence, MRC1^pos/LYVE1^pos macrophages are increased in both brain and skin CMs, but when a broader macrophage population was assessed, MRC1^pos cells, there was no significant increase in skin CMs compared to skin controls. Thus, while the perivascular macrophage component of CMs appears to be similar between brain and skin CMs, there are nuanced differences depending on the macrophage subset examined.

Figure 5. Mannose receptor C type 1 (MRC1^pos) and lymphatic vessel endothelial hyaluronan receptor 1 (LYVE1^pos) cells in capillary malformation of brain and skin.

MRC1(magenta), LYVE1(grey), UEAI (yellow), and counterstaining for DAPI (blue) in (A) CM brain (n=4) (B) CM skin (n=4) (C) non-CM brain (n=2) (D) non-CM skin (n=2). See Table 1 for information on tissue specimens. (E) Quantification of colocalized MRC1^pos and LYVE1^pos cells in the CM brain and skin comparing to non-CM brain and skin. (F) Quantification of MRC1^pos cells in the CM brain and skin comparing to, non-CM brain and skin. The p-values were calculated by Brown-Forsythe and Welch ANOVA test followed by Dunnett T3 multi comparison test. 4-5 non-sequential sections were analyzed per tissue specimen. Scalebar = 200μm.

Discussion

In this study, we show CMs in brain and skin have similarities and differences in cellular and morphologic characteristics. Both consist of enlarged vessels with apparent fibrin leakage, perhaps due to lack of ZO1, and are surrounded by MRC1^pos/LYVE1^pos macrophages. Notable differences were observed in skin CM. The endothelium of the skin CM vessels contained many ECs that appeared to be sprouting outward from the vessel lumen, which was much less apparent in brain CM. Furthermore, mural cells in skin CM appeared to be heterogenous and differentially localized compared to brain CM mural cells (Figure 6). Skin CM contained a ring of calponin^pos/desmin^pos cells located 5-15μm away from the UEAI^pos endothelium. These histopathological characteristics – both the similarities and differences – are important to consider and may hold important clues for targeted molecular or tissue specific therapies for non-syndromic and syndromic CM.

Figure 6. Summary comparing and contrasting brain and skin capillary malformation features by immunohistopathology.

Brain and skin share a common embryological origin arising from the ectodermal layer during embryogenesis[19]. It remains unclear in the literature how the common ectodermal origin contributes to the specialized functions in the brain and skin in terms of bio-mechano-physiological characteristics. During murine embryonic vasculogenesis, mesodermal progenitor hemangioblasts migrate to the yolk sac giving rise to ECs and primitive blood cells. These same precursors within the embryo locally proliferate and undergo differentiation into a primary vascular plexus giving rise to the rest of the vasculature[20, 21]. The heterogeneity in ECs is linked to genetic factors, locality, soluble mediators, cell-to-cell contact, cell matrix interactions, pH, and mechanical forces (shear stress, strain and compression forces). In a healthy state, ECs are quiescent with turnover time over hundreds of days. However, during active angiogenesis, they can proliferate with a turnover time of less than 5 days[22]. In CM skin, we found UEAI^pos blood vessels to be sprouting in addition to undetectable ZO1 immunostaining, which we speculate are cells carrying the p.R183Q mutations[9, 10]. This remains an unanswered question due to lack of tools to distinguish mutant from non-mutant ECs in CM specimens. Future studies using CM specific animal models may address the role of Gα_q in ECs contribution to angiogenesis in a specific manner.

Interconnected endothelial junctions, namely, tight junctions, adherent junctions, and gap junctions are important for barrier function, tissue integrity, and cell-cell communication, respectively. In quiescent endothelium, inflammation, immune cell extravasation, or angiogenic stimuli can weaken the EC-EC junctions and impact vascular permeability. In the brain, the blood-brain barrier is maintained by tight junctions between ECs and the highly specialized organization of cellular environment. [23-25]. Emerging studies have also shown that there is significant crosstalk between tight junctions to regulate barrier function[26]. Moreover, during embryogenesis, ZO1 deficiency causes embryonic lethality by affecting yolk sac angiogenesis. This suggests that ZO1 may be functionally important for cellular remodeling and tissue organization in both embryonic and post-natal tissues[27]. As p.R183Q is a somatic activating mutation, it may cause an early deficiency and disruption of ZO1-containing tight junctions. This in turn could dysregulate early angiogenesis by failure to establish tight junctions and possibly other cell-cell connectors.

Mural cells (pericytes and smooth muscle cells) are found in most microvascular beds and their density and coverage depends on the tissue type[28]. Mural cells communicate with ECs through paracrine signaling by either diffusion or direct cell contact to allow for intercellular communication. Dysregulation in EC-mural cell communication can influence gene expression and thereby impact vascular permeability. Mural cells can arise from various progenitor cells depending on their origin, anatomical location and developmental time point. For example, in mice, neural crest derived mural cells give rise to choroid plexus[29] and thymus[30]. Recent studies have also shown that Col1a2 expressing-fibroblasts can give rise to pericytes[31, 32]. Mural cells are known to not only assist with the completion of vascular growth but to help establish correct blood flow patterns in the brain, also known as functional hyperemia[33, 34]. In the brain, mural cells also interact with astrocytes and neurons to establish and maintain the blood-brain barrier[35, 36]. In a pathological state, particularly in arteriovenous malformation, absence of pericytes surrounding the vessels has been reported[37-41]. In hereditary hemorrhagic telangiectasia type 2 patients with activin receptor-like kinase 1 loss of function, reduced mural cell coverage has been found with a potential mechanism linked to vascular endothelial growth factor[38]. In another study, Notch signaling has been linked to pericyte survival in arteriovenous malformations[42]. In CM brain and skin, we found many enlarged vessels devoid of mural cell coverage, Particularly in CM skin, we found mural cells separated from the endothelium by 5-15μm.

Macrophages are important for homeostasis and play diverse roles. They can sense pathogens and respond at the site of injury to begin repair processes. MRC1 – aka CD206 - expressing resident macrophages are known to conduct phagocytosis. Moreover, it has been reported that there are subsets of macrophages that express LYVE1[43-45]. We previously showed that in the CM brain, there is an increased number of MRC1^pos/LYVE1^pos/ CD68^pos cells [4]. Herein, we show MRC1^pos/LYVE1^pos cells are significantly higher in CM skin as well. Although it remains unknown where these perivascular macrophages originate from and the underlying recruitment mechanism in CMs, our previous study showed an increased adherence of macrophage cell lines (THP1 and U937) to GNAQ p.R183Q mutant ECs mediated by ICAM1[4]. As we found fibrin leakage in both brain and skin CMs, along with a lack of ZO1, one might speculate that the GNAQ p.R183Q endothelium preferentially allows macrophage adherence, and that this, coupled with the compromised tight junctions, may cause the vascular permeability.

There are limitations in our study. First, we relied on a small number of CM brain and skin samples as well as non-CM controls. The variant allelic frequency was higher in the skin CM specimens compared to brain CM specimens. Furthermore, the patients with skin CM were older than the patients with brain CM, which could also contribute to the differences in variant allelic frequency (Table 1). Lastly, we used non-leptomeningeal brain sections as controls. Despite the limited number of specimens, insights presented here comparing the tissue microenvironment are valuable and new. From the basic science perspective, it is important to take in account the tissue characteristics and the cellular environment to understand disease mechanisms as this might impact the potential drug target specific to the mutation driving the disease pathogenesis.

Supplementary Material

Supplemental Figure 1 and Supplemental Table 1

Supplemental Figure 1. Variant allelic frequency of brain vs. skin specimens (n=4). Two tailed t-test was performed. Normality of residuals was tested with Shapiro-Wilk (W) test. See Table 1 for information on tissue specimens.

Table S1: List of ddPCR primers and probes