Angiopoietin‐2 associates with poor prognosis in Moyamoya angiopathy

Abstract

Objective

Moyamoya angiopathy (MA) is a rare cerebrovascular disorder characterized by recurrent ischemic/hemorrhagic strokes due to progressive occlusion of the intracranial carotid arteries. The lack of reliable disease severity biomarkers led us to investigate molecular features of a Caucasian cohort of MA patients.

Methods

The participants consisted of 30 MA patients and 40 controls. We measured cerebrospinal fluid (CSF) levels of angiogenic/inflammatory factors (ELISA). We then applied quantitative real‐time PCR on cerebral artery specimens for expression analyses of angiogenic factors. By an immunoassay based on microfluidic technology, we examined the potential correlations between plasma protein expression and MA clinical progression. A RNA interference approach toward Ring Finger Protein 213 (RNF213) and a tube formation assay were applied in cellular model.

Results

We detected a statistically significant (p < 0.000001) up‐regulation of Angiopoietin‐2 (Ang‐2) in CSF and stenotic middle cerebral arteries (RQ >2) of MA patients compared to controls. A high Ang‐2 plasma concentration (p = 0.018) was associated with unfavorable outcome in a subset of MA patients. ROC curve analyses indicated Ang‐2 as diagnostic CSF biomarker (>3741 pg/mL) and prognostic plasma biomarker (>1162 pg/mL), to distinguish stable‐from‐progressive MA. Consistently, MA cellular model showed a significant up‐regulation (RQ >2) of Ang‐2 in RNF213 silenced condition.

Interpretation

Our results pointed out Ang‐2 as a reliable biomarker mirroring arterial steno‐occlusion and vascular instability of MA in CSF and blood, providing a candidate factor for patient stratification. This pilot study may pave the way to the validation of a biomarker to identify progressive MA patients deserving a specific treatment path.

Introduction

Moyamoya angiopathy (MA) is a rare, chronic, and disabling cerebrovascular disease, with variable prevalence and clinical presentation across ethnicities. MA often leads to recurrent ischemic or hemorrhagic strokes in children and young adults, due to progressive intracranial internal carotid artery narrowing. ¹ , ² , ³ To re‐establish a proper brain perfusion, a continuous but dysfunctional angiogenesis signaling occurs in MA, leading to a compensatory but defective network of basal collateral vessels described as a “puff of smoke.” ⁴ , ⁵ Of note, the imbalanced expression of pro‐ and anti‐angiogenic molecules, cytokines, and growth factors has been hypothesized to be an essential MA feature. ⁶ However, key triggering factors associated with stenotic alterations remain poorly understood and the lack of reliable experimental models have hampered the development of tailored pharmacological therapies. The ability to assess any potential treatment efficacy is furthermore undermined by the amount of co‐influencing factors, such as age, co‐morbidity conditions, and cerebral hemodynamic impairment. Surgical treatment, mainly based on direct and indirect revascularization, represents the preferred procedure for MA patients, for improving cerebral hemodynamics and decreasing the pathological collateral network development. ⁷ Although the genetic components predisposing Caucasian patients to MA are little characterized, it is conceivable that acquired infectious and inflammatory conditions may precipitate MA in genetically predisposed individuals. ⁸ , ⁹ The main susceptibility gene variant identified in East Asian MA patients encodes for the ubiquitin‐ligase Ring Finger Protein 213 (RNF213). ¹⁰

Previous investigations failed to propose reliable factors to monitor disease progression or improve patient's stratification, nor they were truly useful in elucidating the molecular mechanisms underlying basis of the disease pathogenesis. ¹¹ So far, it has not been possible to identify a single measurable biomarker whose trend was preserved between the site of the lesions typical of the disease (cerebral arteries) and the peripheral circulation (blood). These premises prompted us to investigate valuable MA specimens (i.e., cerebral arteries, cerebrospinal fluid, CSF) as a source of biomarkers to be validated later also in peripheral blood, which can be easily obtained by a minimally invasive procedure.

Our pilot study aimed to depict a preliminary framework of MA pathogenesis, by identifying altered proteins in a representative sample of a typical Caucasian MA population. We believe that biological information deriving from patients might sharpen MA prognosis as well as etiologic work‐up, with respect to purely clinical and neuroimaging information. We consider that a better understanding of specific altered elements and their correlation with MA clinical outcomes could help to guide therapeutic approaches, with a possible impact also in the prevention of the disease course.

Methods

MA patient and control subject cohorts

This was an observational study conducted on MA patients, diagnosed following the literature criteria. ¹² The full methodology has been previously reported. ¹³ From the original population of 160 patients, consecutively enrolled between November 2014 and September 2023 at the Neurology IX Unit of the Fondazione IRCCS Istituto Neurologico “C. Besta” (Milan, Italy), a subgroup of 30 adult Caucasian patients was selected, based on CSF and vessel specimen availability. Demographic/clinical characteristics including cerebrovascular risk factors were collected. CSF, middle cerebral artery (MCA), and superficial temporal artery (STA) samples were obtained following STA‐MCA bypass, according to MA guidelines. ¹³ , ¹⁴ All patients underwent conventional catheter digital subtraction angiography and morphological imaging by MRI. MA was classified into bilateral/unilateral types as observed on conventional angiography. Computerized tomography scan and MRI confirmed diagnosis of ischemic/hemorrhagic stroke. The MA severity was assessed by the National Institutes of Health Stroke Scale (NIHSS), by the Suzuki grading and the modified Rankin scale (mRS). ¹³ , ¹⁵ A population of 40 adult Caucasian subjects suffering from hemodynamic insufficiency by atherosclerotic cerebrovascular diseases or isolated intracranial aneurysms was collected as control group (CTRL). Control STA/MCA specimens and/or CSF were sampled in patients undergoing STA‐MCA bypass for hemodynamic cerebrovascular insufficiency or a craniotomy to clip intracranial unruptured aneurysms in the anterior circulation (mostly MCA or anterior communicating artery). A multidisciplinary team discussed surgical indications in all the patients.

Human sample collection

CSF was collected from MA and CTRL subjects from cisternal spaces, mainly sylvian fissure, after arachnoid incision during the neurosurgical intervention, in compliance with ethical regulations and MA guidelines. ¹³ , ¹⁴ After collection, the CSF samples were briefly centrifuged (2000 g for 15 mins) to remove any cellular debris, aliquoted into polypropylene tubes and stored at −80°C. STA/MCA samples were obtained from MA patients and CTRL subjects undergoing revascularization bypass, due to MA or unrelated atherosclerotic cerebrovascular diseases. Specimens were shock‐frozen in liquid nitrogen and stored at −80°C until RNA extraction. Twenty‐four milliliters of peripheral blood was collected by venipuncture from MA patients in tubes containing EDTA (Vacuette®, Preanalitica s.r.l., Caravaggio, Italy) and then centrifuged for 10 min at 300 g. Plasma was stored in aliquots at −80°C until use.

Cellular model

HUVEC cells (ATCC® CRL‐1730TM) were cultured in EGM‐2 medium (EGM™‐2 Bullet Kit, CC‐3162, Lonza Walkersville MD USA) with 10% of FBS (One Shot™, Gibco® Life Technologies, ThermoFisher, Waltham, MA, USA) at 37°C, 5% CO₂, after seeding in plates pretreated with 1 μg/cm² of collagen Bornstein and Traub type IV from human placenta (Sigma‐Aldrich C7521, St. Louis, MO, USA).

ELISA

CSF concentrations of selected proteins were assessed using highly sensitive ELISA kits. Specifically, we analyzed Angiopoietin‐2 (Ang‐2; Boster Biological Technology, Pleasanton, CA, USA), Metalloproteinase‐9 (MMP‐9; ThermoFisher), Chemokine (C‐C motif) ligand 5 (CCL5/RANTES; ThermoFisher), Vascular Endothelial Growth Factor‐A (VEGF‐A; Boster Biological Technology), Interleukin 8 or Chemokine (C‐X‐C motif) ligand 8 (IL‐8/CXCL8; ThermoFisher), and Interleukin 6 (IL‐6; ThermoFisher). The assays were conducted in a 96‐well plate according to the manufacturer's instructions. ¹⁶ The absorbance was measured on a plate reader at 450 nm.

Gene‐expression studies: Quantitative real‐time PCR analyses

Tissue specimens (MCA and STA) were processed with RLT lysis buffer with 1% of β‐mercaptoethanol and subjected to mechanical lysis by a tissue homogenizer (Mikro‐Dismembrators, B‐BRAUN, Hessen, Germany). Samples were treated with Proteinase‐K (Euroclone) and incubated at 55°C for 10 min, then centrifuged at 12,000 rpm for 3 min. Total RNA was extracted from MCA/STA using RNeasy Fibrous Tissue Mini Kit (Qiagen, Marshall Street Redwood City, CA, USA) and from cell cultures using RNeasy Plus Mini Kit (Qiagen) and quantified by NanoDrop (NanoPhotometer® N60/N50, Implen, Westlake Village, CA, USA). RNA was reverse‐transcribed with iScript Advanced cDNA Synthesis Kit (BIORAD, Hercules, CA, USA), according to the manufacturer's protocol by Mastercycler Ep Gradient Thermal Cycler (Eppendorf, Hamburg, Germany). Quantitative Real‐Time PCR (qRT‐PCR) analyses were performed by CFX‐96 Real Time PCR Detection System (BIORAD). cDNA transcripts were amplified using TaqMan assays (ThermoFisher) for RNF213 (Hs00326306_m1), Ang‐2 (Hs00169867_m1), Beta‐2‐Microglobulin (β2M Hs00187842_m1), and Glyceraldehyde‐3‐Phosphate Dehydrogenase (GAPDH Hs99999905_m1). The relative mRNA expression was calculated by the 2^−ΔΔCt comparative method using β2M and GAPDH as the housekeeping genes. Control STA/MCA specimens (or negative control scramble siRNA) were chosen as the calibrators.

Gene‐expression studies: RT² Profiler Array

The Human Angiogenic Growth Factors & Angiogenesis Inhibitors RT² Profiler PCR Array (330231/PAXX‐024Y, Qiagen) was used. Total RNA from MCA specimens was purified and reverse‐transcribed with RT² First Strand Kit (Qiagen). cDNA was mixed with RT² SYBR Green Mastermix and placed into 96‐well array. The amplification program was: 95°C, 10 min for 1 cycle, 95°C, 15 s and 60°C, 1 min for 40 cycles. Analyses were performed by CFX‐96 Real Time PCR Detection System (BIORAD). The relative mRNA expression was calculated by the 2^−ΔΔCt comparative method using ACTB, β2M, GAPDH, HPRT1, and RPLP0 as the housekeeping genes. Control MCA specimens were chosen as the calibrators.

ELLA system

An automated enzyme‐linked immunoassay based on a microfluidic technology (ELLA, ProteinSimple, Bio‐techne, Minneapolis, MN, USA) was used for evaluating plasma concentrations of Angiopoietin‐1 (Ang‐1), Ang‐2, Platelet Derived Growth Factor Subunit B (PDGF‐BB), and VEGF‐A, by using 16 × 4 Multianalyte Cartridge (Bio‐techne). The fluorescent signals inside ELLA were used for protein quantification based on master calibration curves, according to the manufacturers' protocol. ¹⁷ , ¹⁸ The assay time was 80 min. Before the immunoassay, plasma samples were centrifuged at 3000 rpm 4°C for 5 min and diluted 1:5. Analyses were automatically performed in triplicate by ELLA.

Correlation between plasma biomarkers and MA clinical progression

We selected two age/sex‐matched groups of MA patients that were markedly different in their disease progression, as estimated by neurological/neuroradiological follow‐up assessment. Specifically, the two subsets of MA patients were defined as “stable” or “progressive” based on mRS score, number of new cerebrovascular events and clinical disease progression, between T0 (study enrolment at the baseline) and T1 (30 ± 6 months after T0). To define a patient as “progressive,” all of the abovementioned criteria have to be worsened at the T1 respect to T0. To define a patient as “stable,” none of the objective characteristics has to be changed at the T1 re‐evaluation. Plasma samples were collected at T0 and corresponding patients were classified as “stable” or “progressive” later, during the T1 clinical re‐assessment.

RNA interference approaches in HUVEC

For RNA interference (RNAi) studies, 120.000 cells/mL HUVEC were seeded in EGM™‐2 with 10% of FBS in 6‐well plates and cultured for 24 h at 37°C, 5% CO₂. Two Silencer Select siRNA (s33658 and s33568, Ambion™, ThermoFisher) targeting two distinct regions of RNF213 mRNA and a negative control scramble siRNA (Ambion™ 4390843, ThermoFisher) were used. Solution A [Opti‐MEM™ (Gibco, ThermoFisher) + 20 μM siRNA/negative control scramble] and Solution B [Opti‐MEM™ + Lipofectamine RNAiMAX reagent (ThermoFisher)] were mixed to obtain solution C and incubated at RT for 15 min. The growth medium was replaced with Opti‐MEM™, and then, aliquots of the different solution C (RNF213‐siRNA or scramble siRNA, respectively) were added to HUVEC cells and then incubated at 37°C, 5% CO₂. After 6 h and Opti‐MEM™ removal, HUVEC cells were incubated in EGM™‐2 at 37°C, 5% CO₂ for 48 h. The effectiveness of the RNF213 RNAi was verified through qRT‐PCR.

Tube formation assay

A Matrigel tube formation assay was performed to assess HUVEC angiogenic ability in different experimental conditions (e.g., negative control or RNF213 siRNA transfected cells). Cells were seeded in EGM™‐2 (250 cells/μl) into 96‐well plates at 37°C, 5% CO₂. The 96‐well plates were previously coated with Matrigel (Corning® Matrigel® Growth Factor Reduced, Corning, AZ, USA) and incubated (1 h, 37°C, 5% CO₂). Images were taken after 16 h, and tube formation was quantified using Wimasis Image analysis software (https://www.wimasis.com/en/WimTube; Onimagin Technologies SCA, Cordoba, Spain). Analyses were performed in triplicate for at least three independent experiments.

Statistical analyses

For ELISA/ELLA and demographic/clinical statistical analyses, data were expressed as mean ± SD, and statistical significance was calculated through nonparametric Mann–Whitney U by using GraphPad Prism 8 (GraphPad Software, Inc., San Diego, CA). The statistical significance of the mRNA expression data, calculated by the 2^−ΔΔCt comparative method, was determined considering the relative fold‐change threshold. We considered as upregulated transcripts with 2^−ΔΔCt >2 and as downregulated those with 2^−ΔΔCt <0.5. The sample size for each experiment and the number of biological and technical replicates are reported in figure legends. Values of at least three independent experiments were shown. For each investigated protein, numerical homogeneous groups were compared between MA and CTRL subjects. p values are schematized as follows: *<0.05; **<0.01; ***<0.001; ****<0.0001. Data are shown as mean ± SD. Prognostic biomarker accuracy was calculated using receiver‐operating characteristic (ROC) analyses. Simple linear regression model was used to estimate the relationship between protein levels. The authors were blinded to the experimental protocol while performing the experiments and the statistical calculations.

Results

Demographic and clinical characteristics of the participants

Thirty MA Caucasian patients, in whom it was possible to collect arterial specimens and/or intraoperative CSF samples, were included in the present study. ¹³ , ¹⁴ We summarized MA patient information by reporting the mean age (± SD), sex (female %), CVD type (HS, IS, TIA, Other), National Institutes of Health Stroke Scale score (NIHSS; 0–5; 6–10), MA presentation (Unilateral or Bilateral), Suzuki grading (I‐III; IV‐VI), and modified Rankin Scale score (mRS; 0–1; 2–3; Table 1). The patients had a mean age of 42 years (range 22–58 years), and most of them were female (76.7%). The disease presented with a hemorrhagic stroke in 13.3% of patients, an ischemic event in 36.7% and a transient ischemic attack in 36.7% of the cases. In the remaining 13.3% of the cases, the disease was diagnosed for headache, trauma, or incidentally. Forty control subjects (52.5% female) with a mean age of 46.9 years (range 21–78 years) and suffering from hemodynamic insufficiency by unrelated atherosclerotic cerebrovascular diseases or isolated intracranial aneurysms were included in the CTRL group.

Table 1.

The demographic, clinical, and neuroradiological features of the subjects in the MA and control group.

	MA	CTRL
N	30	40
Mean Age (y ± SD)	42.0 ± 10.2	46.9 ± 16.1
Sex (% F)	76.7%	52.5%
CVD type (%)
HS	13.3%	0%
IS	36.7%	12%
TIA	36.7%	42%
Other	13.3%	48%
NIHSS score
0–5	90%	na
6–10	10%	na
MA presentation (%)
U	16.7%	na
B	83.3%	na
Suzuki grading
I–III	40%	na
IV–VI	60%	na
mRS score
0–1	86.7%	89%
2–3	13.3	11%

B, bilateral; CTRL, controls; CVD, cerebrovascular disease; F, female; HS, hemorrhagic stroke; IS, ischemic stroke; MA, Moyamoya angiopathy; mRS, modified Rankin scale; N, number of subjects; na, not available; NIHSS, National Institutes of Health Stroke Scale; TIA, transient ischemic attack; U, unilateral; y, years.

CSF angiogenic and inflammatory proteins

Based on our previous results highlighting the expression of different inflammatory/angiogenic mediators in plasma from MA patients, ¹⁹ we first investigated these pathways by performing ELISA analysis on CSF samples, to corroborate a possible pathophysiological mechanism. The CSF expression levels of angiogenic/vasculogenic (Ang‐2; MMP‐9; VEGF‐A) and inflammatory factors (CCL5/RANTES; IL‐8; IL‐6) are shown (Fig. 1). A very highly significant increase of Ang‐2 concentration (p < 0.000001) was observed in CSF of MA patient in comparison with CTRL subjects (Fig. 1A). MA patients displayed also a higher CSF level of VEGF‐A (p = 0.0435) and MMP9 (p = 0.0056) compared to CTRL. Contrary to previous results in plasma, inflammatory factors did not show any significant modulation in CSF (Fig. 1B). The ROC curve identified Ang‐2 CSF level >3742 pg/mL as a cutoff value to discriminate between MA patients and CTRL subjects with unrelated cerebrovascular diseases (AUC = 0.915; Sens. = 72.4%; Spec. = 91.3%), as shown in Figure 1C.

Figure 1.

CSF concentrations of angiogenic (A) and inflammatory (B) protein factors released in CSF of MA patients and CTRL subjects with unrelated cerebrovascular diseases. Angiopoietin‐2 (Ang‐2), matrix metalloproteinase‐9 (MMP‐9) chemokine (C‐C motif) ligand 5 (CCL5/RANTES), Vascular Endothelial Growth Factor‐A (VEGF‐A), interleukin 8 (IL‐8/CXCL8), and interleukin 6 (IL‐6) were evaluated by ELISA and expressed as pg/mL or ng/mL, as indicated. The boxes represent data obtained in the range 25th–75th percentile; the line across the boxes indicates the median value; the lines above and below the boxes indicate extreme values (10th or 90th percentile). The statistical significance (*p < 0.05; **p < 0.01; ****p < 0.0001) was calculated through nonparametric Mann–Whitney U test. Values of at least three independent experiments are shown. For each of the investigated proteins, numerically homogeneous groups of MA patients and CTRL subjects were compared. (C) The receiver‐operating characteristic (ROC) curve for circulating Ang‐2 on discriminating MA patients from CTRL subjects with unrelated cerebrovascular disease; AUC, area under the ROC curve.

Angiopoietin‐2 mRNA expression analysis in MCA/STA

To assess whether the pronounced CSF increase of Ang‐2 and other angiogenic factors specifically reproduced the molecular pattern of MA stenotic arteries, we firstly investigated the GE profile of angiogenic growth factors/angiogenesis inhibitors in MCA from MA patients (n = 10) and CTRL subjects (n = 5), who underwent STA‐MCA bypass due to hemodynamic insufficiency caused by unrelated atherosclerotic cerebrovascular diseases. We found that MA MCA samples show a peculiar GE profile, where angiogenic growth factors/angiogenesis inhibitors are mostly deregulated. Among the 39 out of 84 analyzed gene transcripts expressed at a detectable level in MCA tissues, we found a marked up‐regulation of Ang‐2 expression (fold‐change induction: 2^−ΔΔCt = 3.20) in MA stenotic arteries as compared to controls (data not shown).

By using more specific qRT‐PCR TaqMan assays, we performed a relative quantification of Ang‐2 mRNA expression in stenotic MCA and normal STA used for revascularization by‐pass, both in MA patients and CTRL subjects. The Ang‐2 expression level was confirmed to be much higher in MCA samples from MA patients (n = 13; RQ >2), as compared to MCA obtained from CTRL subjects (n = 6). Interestingly, Ang‐2 mRNA expression did not differ in MA STA specimens compared to CTRL STA specimens, thus highlighting a peculiar feature of the stenotic MCA from MA patients (Fig. 2).

Figure 2.

Quantitative Real‐Time PCR analysis of Angiopoietin‐2 mRNA expression in MCA/STA from MA patients and CTRL subjects with unrelated cerebrovascular diseases. The relative mRNA expression level was calculated by the 2^−ΔΔCt comparative method using β2M and GAPDH as housekeeping genes. Control STA/MCA specimens were chosen as the calibrator samples.

Angiopoietin‐2 protein expression analysis in plasma

We currently improved our previous analysis on plasma samples of MA patients, ¹⁹ by using an innovative, automated, and more sensitive antibody‐based immunoassay (ELLA). ¹⁷ , ¹⁸ This approach allowed obtaining highly reproducible validated data and consistent biomarker detection in plasma samples, which have been easily collected by a minimally invasive procedure. We compared the plasmatic levels of Ang‐2 and other angiogenic factors (VEGF‐A, PDGF‐BB, and Ang‐1) between two age/sex‐matched groups of MA patients that were markedly different in their disease progression. Specifically, the two subsets of MA patients were defined as “stable” or “progressive” based on mRS score, number of new cerebrovascular events, and clinical disease progression, between T0 (study enrolment at the baseline) and T1 (30 ± 6 months after T0; Table 2).

Table 2.

The demographic and clinical features of the MA patients in the stable vs progressive disease group. ^a

	Stable MA	Progressive MA	p value
N	16	7
Mean age (y ± SD)	40 ± 11	48 ± 10	ns (0.1796)
Sex (% F)	87.5%	100%	ns (0.5257)
Baseline mRS
0–1	100%	86%	ns (0.4117)
2–3	0%	14%
Follow‐up mRS
0–1	100%	43%	*** (0.0002)
2–3	0%	57%
New CVE
Yes	0%	86%	**** (<0.0001)
No	100%	14%

F, female; MA, Moyamoya angiopathy; N, number of subjects; ns, not significant; SD, standard deviation; y, years.

^a The two subsets of MA patients were defined as “stable” or “progressive” based on modified Rankin scale (mRS) score, number of new cerebrovascular events (CVE), and clinical disease progression, between T0 (study enrolment at the baseline) and T1 (30 ± 6 months after T0). The statistical significance (***p < 0.001; ****p < 0.0001) was calculated through nonparametric Mann–Whitney U test.

For all the included MA subjects, protein expression levels have been determined in preoperative plasma samples collected at T0. Specifically, we found that the Ang‐2 expression levels at T0 were higher in MA patients showing a T1‐progressive disease (n = 7), as compared to patients showing a T1‐stable disease (n = 16). These findings have not been evidenced for the other investigated circulating proteins (i.e., VEGF‐A, PDGF‐BB, and Ang‐1; Fig. 3A).

Figure 3.

(A) Analysis of plasma Ang‐2, VEGF‐A, PDGF‐BB, and Ang‐1 protein expression level in MA patients with different disease progression. (B) The receiver‐operating characteristic (ROC) curve for plasma circulating Ang‐2 on discriminating stable MA patients from progressive ones; AUC, area under the ROC curve.

Logistic regression and ROC curve analyses performed to assess the predictive power of Ang‐2 suggested that it might function as a useful plasma biomarker indicating the progression of MA from moderate‐to‐severe (Fig. 3B). Specifically, the ROC curve showed a T0 Ang‐2 plasma concentration >1162 pg/mL as the discriminating value for detecting patients with worse MA progression (AUC = 0.813; Sens. = 71.43%; Spec. = 81.25%). We applied simple linear regression model to determine the reciprocal correlation among the plasma levels of selected proteins. Interestingly, the investigated angiogenic factors (i.e., VEGF‐A, PDGF‐BB, and Ang‐1) displayed a similar trend (Fig. 4A–C; p < 0.0001), but none of them correlates with Ang‐2 tendency, as appreciable from the not significant p values (Fig. 4D–F).

Figure 4.

Simple linear regression model to determine the correlation between plasma level of Ang‐1/VEGF‐A (A), Ang‐1/ PDGF‐BB (B), VEGF‐A/PDGF‐BB (C), Ang‐1/Ang‐2 (D), VEGF‐A/Ang‐2 (E), PDGF‐BB/Ang‐2 (F) in MA patients.

RNF213 gene silencing in endothelial cells

Since there is no successful MA animal model, we established an experimental cell model mimicking the pathophysiological condition, to better clarify at a cellular level the potential molecular mechanisms involving Ang‐2 in MA. Human endothelial cell (EC) cultures underwent gene silencing by a RNA interference (RNAi) approach toward RNF213, the main susceptibility gene for MA. To assess the impact of RNAi on angiogenic function of EC, we carried out a tube formation assay after transient transfection of specific siRNA. The total tube length (p = 0.021), the total number of branching points (p = 0.028), loops (p = 0.012), and tubes (p = 0.007) were found decreased in EC silenced for RNF213, as compared to negative control transfected cells (Fig. 5A). The effects of RNF213 downregulation on the formation of vessel structures by EC are shown in Figure 5B, in comparison with negative control transfected cells. In agreement with the evidence found in biological samples, the analysis of mRNA expression through qRT‐PCR showed a significant up‐regulation (RQ >2) of Ang‐2 expression in RNF213 silenced condition, as compared to negative control one (Fig. 5C).

Figure 5.

(A) Matrigel tube formation assay values obtained after transient transfection of specific siRNA toward RNF213 and negative control (siRNA scramble) in HUVEC cells. (B) Representative image of tube formation assay analysis in RNF213 silenced condition, as compared to negative control (siRNA scramble). (C) qRT‐PCR analysis of RNF213 and Angiopoietin‐2 gene‐expression levels in HUVEC cells following RNAi approach toward RNF213. The relative mRNA expression level was calculated by the 2^−ΔΔCt comparative method using β2M and GAPDH as housekeeping genes. HUVEC cells transfected with negative control siRNA were chosen as the calibrator samples and their relative mRNA expression levels were arbitrarily set to 1.

Discussion

There is an urgent need for translational research on circulating biomarkers for MA patients' stratification, to drive personalized therapies and provide prognostic advice to patients and relatives. It is critical to foster current studies on identifying MA‐specific fingerprints versus nonspecific markers of common pathophysiological processes, in order to clarify the molecular mechanisms underlying the cerebrovascular instability of MA patients. Although previous investigations attempted to identify valuable factors for disease progression monitoring and patient's stratification improvement, the crucial molecular mechanisms underlying MA pathogenesis remain elusive.

Indeed, several studies described peculiar MA molecular profile by different explorative approaches, including omics techniques. A recent research demonstrated the different expression profile of long noncoding RNA in MCA specimens from MA patients as compared to not injured cerebral artery. ²⁰ Transcriptomic analyses performed by Xu and colleagues revealed a differential GE in MA intracranial arteries, based on sex group. ²¹ Metabolic studies on MA serum amino acid profiles defined a promising objective diagnostic method for MA early diagnosis. ²² Ultimately, proteomic analyses on serum‐derived exosomes led to the discovery of proteins shared by ischemic/hemorrhagic MA patients and differently expressed when compared with controls. ²³ Proteomic data collected from a large MA cohort pointed out the differential expression of proteins associated with cytoskeletal structure, possibly relevant for MA cerebrovascular intimal hyperplasia. ²⁴ All studies clearly showed how the scientific community is attempting to define MA molecular mechanisms, while still not identifying a specific targetable therapeutic factor or a clinically helpful biomarker. Therefore, determining the correlation between a specific measurable biomarker and meaningful clinical endpoints is emerging as highly mandatory.

In this study, we characterized fluid/tissue samples of MA patients, assuming that altered molecular features could account for the phenotypical disorder traits. MA is characterized by a stenotic lesion in the intracerebral carotid artery with concomitant formation of “Moyamoya” vessels. We hypothesized that CSF, due to its proximity with injured vessels, may mirror pathological modifications, supplying reliable markers of disease features. Therefore, we selected a representative subgroup of patients based on (i) availability of CSF/artery samples following neurosurgical procedures and (ii) demographic and clinical features typical for our MA cohort. Of note, these characteristics are precisely representative of the whole MA population in Western countries, ²⁵ which is a relevant aspect considering the rarity of the disease.

We analyzed potential key angiogenic proteins and we found a marked increase of Ang‐2 in MA CSF samples as compared to controls. Ang‐2 is a secreted glycoprotein that can alter TIE2 receptor signaling, leading to pathological angiogenesis, vascular remodeling, endothelial destabilization, and inflammation. ²⁶ , ²⁷ , ²⁸ , ²⁹ Specifically, Ang‐2‐related inhibition of TIE2 receptor potentiates the action of VEGF‐A, weakens the integrity of EC junctions, promotes the detachment of pericytes, and is involved in the recruitment of inflammatory cells. ³⁰ A synergistic role of Ang‐2 and VEGF‐A in vascular destabilization has previously been explored. ³¹ , ³² It is noteworthy that intimal fibrous thickening is a frequently observed feature in the MA stenotic arteries. ³³ Therefore, the up‐regulation of MMP‐9, a key factor in extracellular matrix remodeling, may be responsible for EC migration from pre‐existing to newly formed blood vessels. Overall, our findings about Ang‐2, VEGF‐A, and MMP‐9 increased levels in MA CSF suggested the deregulated migration of EC and the resulting dysfunction in the formation of new collateral vessels (Fig. 6).

Figure 6.

Schematic diagram showing cellular and molecular interplay among the main angiogenesis factors identified in the present study. Their potential roles in MA pathophysiology and/or as prognostic biomarkers are highlighted. AKT, serine/threonine protein kinase PKB; Ang2, angiopoietin‐2; EC, endothelial cell; ECM, extracellular matrix; FAK, focal adhesion kinase; MMP‐9, Metalloproteinase‐9; mTOR, mammalian target of rapamycin; P13K, Phosphoinositide 3 kinase; TIE2, Tyrosine kinase with Immunoglobulin‐like and EGF‐like domains 2 receptor; VEGF‐A Vascular Endothelial Growth Factor‐A; VEGFR, vascular endothelial growth factor receptor.

Interestingly, the pronounced over‐expression of Ang‐2 was confirmed in MCA of MA patients when compared to STA specimens, and with respect to MCA/STA of control subjects, thus highlighting a specific association of Ang‐2 with the MA cerebral lesion. We supposed that the increased Ang‐2 expression in MCA might be a consequence of the stenotic damage. Indeed, the higher Ang‐2 mRNA found in MCA could induce the up‐regulation of Ang‐2 protein, as detected in CSF, for counteracting the stenosis and establishing a compensatory collateral vessel network.

Due to these promising evidences, we analyzed Ang‐2 levels in MA patient plasma. Intriguingly, a high Ang‐2 level appeared to be associated with unfavorable outcome in MA patients, thus predicting patients with a higher probability to develop new ischemic or hemorrhagic events. Peripheral blood is easily accessible, with little invasive and inexpensive sampling, providing for time‐efficient measurements of a valuable MA predicting factor. Notably, through the simple linear regression model, we found a positive correlation between other investigated angiogenic factors (i.e., VEGF‐A, PDGF‐BB, and Ang‐1), whereas Ang‐2 appears to be deregulated in a peculiar way, supporting its specific involvement in MA and showing the best characteristics for being included in a clinical scoring system for noninvasive detection. Very recently, Ang‐2 serum level was associated with capillary leak and predicted complications after cardiac surgery, thus showing promising potential to prevent morbidity and mortality due to hemodynamic instability. ³⁴

Besides the difficulty in obtaining cerebral artery specimens, the other relevant hindrance in the MA basic research is the lack of reliable animal and cellular preclinical models. We developed a preliminary in vitro model by silencing the expression of the main susceptibility gene of MA, RNF213. ³⁵ We evidenced an impaired angiogenic function of RNF213‐defective EC, by strengthening the reliability of this MA experimental cellular model. In this preclinical context, we detected a simultaneous and significant up‐regulation of Ang‐2, in agreement with biological data from patients, thus confirming the strong relationship between Ang‐2 up‐regulation and MA phenotypical features.

Overall, our results pointed up Ang‐2 as a reliable biomarker mirroring the MA cerebrovascular instability. In this regard, the identification of changes in molecular profiles providing valuable biomarkers and shedding light on mechanisms of diseases could have a crucial role in patients' risk stratification and precision medicine, allowing for the selection and customization of medical treatment. The establishment of a trustworthy Ang‐2 plasmatic cutoff value could actually stand for a better patient management, in terms of check‐up frequency, drug therapy dosage, and implementation of preventive measures, resulting in a real, though limited, benefit for the patient. Although restricted to a single biomarker, this approach is easy, fast, and immediately feasible.

In addition, we might pave the way toward the validation of Ang‐2 as a possible novel therapeutic target, by delineating effects of Ang‐2 inhibition on MA cerebrovascular instability. Recent findings suggested that targeting both Ang‐2 and VEGF‐A with a bi‐specific antibody can provide additional protective benefits in models of ocular neovascularization and pathological vascular permeability. ³⁶ Interestingly, Ang‐2 targeting was reported as a novel treatment option also for chronic kidney disease, where increased levels of Ang‐2 have been found to correlate with the disease severity and with patient's arterial stiffness. ³⁷ Indeed, the therapeutic inhibition of Ang‐2 could attenuate kidney inflammation and fibrosis in murine models of progressive kidney disease, in the form of transgenic mice overexpressing human Ang‐1. ³⁸ Moreover, the application of an Ang‐2‐specific Fc‐fusion peptide inhibitor blocked the ability of Ang‐2 to bind to TIE2, by promoting endothelial survival and inhibiting vascular rarefaction. ³⁸

The present study has some limitations that should be acknowledged. Firstly, our pilot study focused only on a limited subset of samples from MA patients in whom it was possible to collect CSF and MCA specimens. Secondly, the CTRL group was also restricted to subjects suffering isolated intracranial aneurysms or unrelated cerebrovascular diseases; therefore, the preliminary implications of our findings are worth considering to be extended to a more numerous cohort. Moreover, our preliminary in vitro model based on transient siRNA transfection, an approach that potentially affects the robustness of our results. To address these limitations, future studies could focus on advanced CRISPR/Cas9 technology to establish a more reliable experimental model of RNF213 knockout cells.

In conclusion, by combining the results of cerebral arterial, CSF, and peripheral blood sampling, we found that among several angiogenic factors, only Ang‐2 showed a consistent expression pattern within a cohort of Western MA patients. We emphasized the relevance of Ang‐2 as a specific predictive blood biomarker for patient's stratification as well as a novel potential therapeutic target for MA care.

Author Contributions

GG, AP, TC, and GP performed the measurements, analyzed the data, provided statistical analysis methods, and drafted the original draft of the manuscript. IC, NF, and AB performed the clinical evaluations of subjects. FA, IGV, and PF performed the neurosurgical procedures. AB and PF provided intellectual advice on the study. LG contributed to the general conception and supervised the execution of the study and edited‐revised the manuscript.

Funding Information

This work was partially supported by the Italian Ministry of Health (grant number RRC 2018–2023 to LG; grant number RF‐2019‐12369247 to AB; grant number RCR‐2021‐23671214 to AB).

Conflict of Interest

Nothing to report.

Funding Statement

This work was funded by Italian Ministry of Health grants RCR‐2021‐23671214, RF‐2019‐12369247, and RRC 2018‐2023.

Data Availability Statement

The study raw data will be made available subject to institutional agreements and ethical approval. Anonymized data will be shared by request from a qualified academic investigator for the sole purpose of replicating procedures and results presented in the article and as long as data transfer is in agreement with EU legislation (GDPR, https://gdpr.eu/) and decision by the institutional Ethical Committee, which should be regulated in a material transfer agreement.