RNA profiling of sEV (small extracellular vesicles)/exosomes reveals biomarkers and vascular endothelial dysplasia with moyamoya disease

Abstract

The association of exosomal RNA profiling and pathogenesis of moyamoya disease (MMD) and intracranial Atherosclerotic disease (ICAD) is unknown. In this study, we investigated the RNA profiles of sEV (small extracellular vesicles)/exosomes in patients with MMD and ICAD. Whole blood samples were collected from 30 individuals, including 10 patients with MMD, 10 patients with ICAD, and 10 healthy individuals. Whole transcriptome analysis was performed using the GeneChip WT Pico Reagent kit. Transcriptional correlation was verified using quantitative reverse transcription-polymerase chain reaction (qRT-PCR). The association between functional dysregulation and candidate RNAs was studied in vitro. In total, 1,486 downregulated and 2,405 upregulated RNAs differed significantly between patients with MMD and healthy controls. Differential expression of six circRNAs was detected using qPCR. Among these significantly differentially expressed RNAs, IPO11 and PRMT1 circRNAs were upregulated, whereas CACNA1F circRNA was downregulated. This is the first study showing that the differential expression of exosomal RNAs associated with MMD pathogenesis, such as overexpression of IPO11 and PRMT1 circRNAs, may be related to angiogenesis in MMD. The downregulation of CACNA1F circRNA may be related to vascular occlusion. These results propose the utility of exosomal RNAs as biological markers in MMD.

Introduction

Moyamoya disease (MMD) is a rare chronic steno-occlusive cerebrovascular disease characterized by abnormal collateral vessels (a “smoky” abnormal vascular network) emerging at the base of the skull. ¹ Clinically, MMD is characterized by intracranial hemorrhagic and ischemic events, and revascularization surgery remains the main treatment strategy. However, the mechanisms underlying the formation of abnormal collateral vessels remain unknown, and this lack of pathogenic information hinders the development of drugs for the treatment of MMD. Intracranial atherosclerotic disease (ICAD) has one of the highest rates of recurrent ischemic stroke worldwide, despite intensive medical therapy.^2,3 Randomized controlled clinical trials have indicated that angioplasty with stenting and bypass surgery fails to improve outcomes in patients with ICAD.^4,5 Therefore, to identify new targets for effective therapy, the molecular mechanisms underlying ICAD progression need to be examined in detail.

Exosomes are small membrane particles (diameter 30–150 nm) secreted by cells that contain biologically active substances, including proteins, nucleic acids, and lipids. ⁶ Exosomes are released into circulation by various cell types, such as immune cells, platelets, and endothelial cells, to aid in communication between cells. ^7–9

Under pathogenic stimulation, affected cells secrete exosomes that transport dysregulated biological substances to target cells leading to dysregulated cell function. ^10,11 Li et al. showed that exosomes play a significant role in the elucidation of signal transduction pathways in hepatoma cells, angiogenesis, and early diagnosis of hepatocellular carcinoma. ¹² Wang et al. found that proteins in serum-derived exosomes (SDEs) from patients with MMD were differentially expressed compared with that in SDEs from healthy controls. CFL1 (Cofilin-1) andACTR2/3 (actin-related protein 2/3; also known as ARP2/3) were shown to be downregulated in SDEs of ischemic and hemorrhagic patients. Treatment with SDEs from patients with hemorrhagic MMD induced mitochondrial dysfunction in mouse cerebrovascular endothelial cells. ¹³ Nucleic acids in exosomes are protected from RNase degradation and are enriched in the bloodstream. ¹⁴ Jiang et al. revealed distinct expression profiles of circulating exosomal microRNAs (e-miRNAs) in refractory ICAD, suggesting an antiangiogenic mechanism underlying intensive medical management failure. ¹⁵ Wang et al. identified a novel and highly sensitive e-miRNA signature for MMD detection and explored potential targets. ¹⁶ However, differences in exosomal RNAs between patients with MMD, ICAD, and healthy controls, as well as how they affect the pathogenesis of MMD and ICAD, remain unknown.

To better understand the effect of exosomal RNAs in the pathogenesis of MMD and ICAD, we first examined the RNA expression profiles of SDEs isolated from patients with MMD and ICAD, and healthy controls. Moreover, we investigated the effect of hsa_circRNA_0000583, hsa_circRNA_0090577, and hsa_circRNA_0051937 on human brain microvascular endothelial cells (HBMEC), which show different exosome expression between MMD and healthy controls.

Methods

Patients and blood samples

Whole blood samples used in this study were obtained from patients admitted to Peking University International Hospital between June 2019 and November 2021. This experiment was approved by the Ethics Committee of Beijing Tiantan Hospital (KY 2020-045-02). This study complied with the ethical standards of the Helsinki Declaration of 1975, and written informed consent was obtained from each participant. All 20 patients and 10 healthy controls (group C, 10 participants) enrolled in this study were evaluated for MMD (group A, 10 patients, 2 hemorrhagic MMD and 8 ischemic MMD) or ICAD (group B, 10 patients) using digital subtraction angiography. We performed detailed consultations and physical examinations on patients with MMD to ensure that they did not have any underlying diseases, such as hypertension, diabetes, hyperlipidemia, hyperthyroidism, and surgical history, that could affect the results of this study. Ten age-and sex-matched healthy participants were included in this study. Additionally, to identify the result of differential expression of RNAs, 5 additional participants of each group were enrolled in the study. These participants were also evaluated using digital subtraction angiography and did not have any underlying diseases that could affect the results.

Isolation of SDEs and RNA extraction

SDEs were isolated from serum samples obtained from patients with MMD and ICAD and healthy controls using ultracentrifugation. After adding VEX Exosome Isolation Reagent (from plasma) to the blood samples and centrifugation at 10200 rpm (10000 ×g) for 5 min, the isolated SDEs were collected from the pellet at the bottom of the tube.

The GeneChip WT Pico Reagent Kit (Affymetrix, California, USA) was used to extract RNA from whole blood samples. Following formalin fixation and paraffin embedding of the collected materials, RNA samples were processed for amplification using the GeneChip WT Pico Reagent kit (Affymetrix). The cRNA was converted to biotinylated sense-strand DNA hybridization targets for unbiased transcriptome coverage. The ratio of the absorbance at 260 nm and 280 nm for all samples ranged from 1.8 to 2.0, with concentrations ranging from 100–150 ng/µL. All samples were dissolved in DEPC water (RNase-free) before preparation and stored at −80°C in a Forma 700 ultra-low-temperature refrigerator (Thermo Ltd, USA). Four participants in group A and three in group C were treated with RNA amplification due to slightly lower amounts of RNA extracted. The Kit we used is GeneChip® WT Pico Kit (902623, WT Pico Kit).

GeneChip hybridization and data quality control

After total RNA isolation, synthesis, purification, and quantification of the first-strand cDNA and second-cycle ss-cDNA, fragmented and terminally labeled ss-cDNA samples were used for WT array hybridization. The GeneChip 3000 7 G system (Applied Biosystems, Massachusetts, USA) was used for chip scanning after washing and staining with the GeneChip Fluidics Station 450. The GeneChip 3000 7 G system generated the signal values of each probe by capturing the fluorescence and using GeneChip operating software (GCOS). The GeneChip WT Pico Reagent Kit was used to perform the necessary quality control for all samples to ensure that the total RNA samples were free of genomic DNA.

In this study, the quality of raw data was strictly controlled to ensure the accuracy and integrity of total RNA. Four types of spike-in probes (BioB, BioC, BioD, and CreX) were used on the WT chip to verify the quality of hybridization, and the P values for detecting the signal of these probes were also selected (P < 0.01) (Figure 1(a)).Based on the base reaction characteristics of the probes, which can be positive or negative, on the WT chip, the P values for detecting the signal of each probe were selected (P < 0.01) (Figure 1(b)), the histogram is shown in Figure 1(c). The group information corresponding to the sample analysis name is in the supplementary material Table S1.The fluorescence signal was reflected by streptavidin-phycoerythrin (SAPE) distributed on different probes, which could be screened to reflect RNA ID and expression according to the signal sites and strength.

Figure 1.

(a) The quality control of hybridization: bioB < bioC < bioD < cre (P < 0.01), which indicated the hybridization of probes. (b) Positive and negative quality control: neg < pos (P < 0.01), indicating the accuracy of the total RNA samples. (c) Sample quality control bar chart. (d–h) Heatmap of MMD patients (Group A) and healthy controls (Group C). (i) In the scatter plot, red represents up and blue represents down.

RNAomics analysis

The original data of all RNA samples in the system were imported into R 3.6.0 (https://cran.r-proje ct.org/), tidyr, tibble, and dplyr for batch processing of the original data. The ChAMP 2.21.1 package was used for inter-group comparisons of data. The Limma package was used to partition and pre-process the difference data.

Selection of candidate RNAs and evaluation of gene expression and pathways

Differentially expressed RNAs and their gene pathways were screened to identify the potential pathogenesis of MMD and ICAD. Candidate RNAs were screened based on the adjusted P values of logFC with FDR correction. (adj. P value < 0.05, FC >1.5) The most significant differential expressed RNAs were selected as focus RNAs according to the adj. P value and FC. With selecting hsa_circRNA_0000583, hsa_circRNA_0090577, hsa_circRNA_0051937, hsa_circRNA_0076043, hsa_circRNA_0074463, and hsa_circRNA_0001491 as candidate RNAs, the expression of these target RNAs in the SDEs of MMD, ICAD, and healthy controls were detected using PCR. Based on the target gene names of the differentially expressed RNAs, PubMed and other online databases were used to search for relevant research involving these genes and hotspot pathways.

Bioinformatics analysis

The original data sets of the MMD, ICAD, and healthy control groups were sampled using the Limma package. Logarithmic differences in RNA expression were calculated after further statistical normalization. The differences in the expression levels of RNAs between patients with MMD, patients with ICAD, and healthy controls were represented by the logFC of all probes and the adjusted P value obtained using the unpaired t-test. The original genes corresponding to the identified RNAs were annotated and their interactions were predicted using the GPL-related search interface in the Gene Expression Omnibus (GEO) database. The Gene Ontology (GO) database (http://david.abcc.ncifcrf.gov/) was used to predict gene function. The Kyoto Encyclopedia of Genes and Genomes (KEGG, http://www.genome.ad.jp/kegg/) was used for gene pathway analysis.

Lentivirus infection

To study the function and interaction among focal RNAs, we knocked down the target RNAs in human brain microvascular endothelial cells (HBMECs) using lentivirus infection. The target sequence was designed for the target genes according to the principle of RNAi sequence design. Four types of plasmids were designed for gene knockdown: sh-hsa_circRNA_0000583, sh-has_circRNA_0090577, sh-hsa_circRNA_0051937 plasmid, and shRNA-NC control (Table 3). The lentiviral vector were constructed (Figure S1), the lentiviruses were cultured, packaged, and detected using RT-PCR. Equal number of HBMECs were transfected with equal volume of the four designed lentiviruses and cultured for 48 h. Consequently, PCR was used to check the transfection efficiency of each group.

Table 3.

Sequence of sh-hsa_circRNA_0000583 plasmid, sh-has_circRNA_00905 77 plasmid, sh-hsa_circRNA_0051937plasmid, and shRNA-NC control plasmid.

Plasmid	Sequence
sh-hsa_circRNA_0000583	5′-CCGGTTGTAGGTCAGGCCCCTGTGATCGAGTCACAGGGGCCTGACCTACAATTT TG-3′
sh-has_circRNA_0090577sh-hsa_circRNA_0051937shRNA-NC control	5′-CCGGTTGTAGGTCAGGCCCCTGTGATCGAGTCACAGGGGCCTGACCTACAATTTTG-3′5′-CCGG GGCATCCACGAGAATTTTGTATCGAGTACAAAATTCTCGTGGATGCCTTTTG-35′-CCGGCAACAAGATGAAGAGCACCAACTCGAGTTGGTGCTCTTCATCTTGTTGTTTTTG-3′

shRNA：short hairpin RNA.

Cell proliferation assay (CCK8)

HBMECs were harvested by centrifugation during the logarithmic growth phase and seeded on 96-well plates. The cells were cultured at 37°C in the presence of 5% CO₂ for 12 h. After three PBS washes at 24, 48, and 72 h, 10 µL of 10% CCK8 solution was added to each well and incubated for 2 h. Absorbance was determined at 450 nm using a Berthold LB941 microplate multifunctional enzyme plate analyzer (Berthold Technologies, Germany) after incubation.

EdU cell proliferation assay

Cells from each group were seeded in 96-well plates and cultured for 24 h at 37°C in the presence of 5% CO₂. EdU was dissolved and diluted to a ratio of 10:1 with deionized water. Cells were then washed twice with PBS. Thereafter, 500 µL of Click-iT® reaction mixture was added to each well and incubated at room temperature of 20° for 30 min in the dark. The reaction mixture was removed and the cells were washed once with 100 µL of PBS containing 3% BSA. The cells were stained with DAPI and immediately observed using a Zeiss LSM laser confocal microscope (Zeiss, Germany). A random field of view was selected to capture photos.

Apoptosis detection using flow cytometry

The Annexin V/PI double staining kit (Invitrogen, California, USA) was used for flow cytometry analysis. The experimental grouping was the same as mentioned above, and the cells were washed twice with PBS. After EDTA-free trypsin treatment, the cell suspension was centrifuged at 800 rpm for 5 min, and the supernatant was removed. Staining buffer (0.5 mL) was then added to resuspend the cells. Then, 5 µL FITC staining solution and 5 µL PI staining solution were added, and the cells were cultured in the dark at room temperature for 15 min. After incubation, apoptosis was analyzed in vitro using a DXFlex flow cytometer (Beckman Coulter, California, USA).

Tubule formation assay

Preparation of organ-like medium

Matrigel glue stored at −20°C was transferred to a refrigerator at 4°C for thawing 24 h before the experiment. The pipetting gun head was kept in a refrigerator at −20°C for precooling. The spearhead was removed and placed on ice for 30 min before the experiment. Then, 50 µL of overnight melted Matrigel glue was added to each well of the 96-well plate with slight shaking to avoid bubble formation and incubated at 37°C for 30 − 60 min to solidify the Matrigel glue.

Cell implantation

HBMECs in each group were routinely digested, centrifuged, and resuspended in basic medium to obtain a cell density of 200,000 cells/mL. Next, 100 µL of the single-cell suspension was slowly added along the wall into the well without touching the glue surface. Three replicate wells were used for each sample, and the 96-well plate was incubated at 37°C for 6 h.

Collection and analysis of tube formation

The lumen was completely formed after 6 h of incubation. The 96-well plates were observed using a microscope and images were captured. The radiographs were crosschecked by two skilled operators. Each experiment was repeated three times. ImageJ software (https://imagej. nih. gov/ij/) was used to analyze the length of each group of cells forming branching lumens.

Results

Exosomal RNA profile in patients with MMD and ICAD

The clinical data of the patients are presented in Table 1. To clarify the exosomal RNA distribution in patients with MMD or ICAD and its differences compared to healthy controls, exosomal RNA analysis was performed on blood samples from patients with MMD (n = 10) and ICAD (n = 10). The original probe data was quality-controlled and confirmed. In total, 135,750 probe sets were used for RNA analysis, and whole-transcriptome analysis was performed for all probes. Exosomal RNA distribution followed an equilibrium distribution pattern in all samples, and exosomal RNA levels differed in most patient groups (>80% and <20%). Whole transcriptome analysis revealed significant differences between the patient group and healthy controls.

Table 1.

Characteristics and clinical information of participants.

Characteristics	MMD group (n = 10)	ICAD group (n = 10)	Control group (n = 10)
Mean age ± SD (years)	30.4 ± 19.5	54.7 ± 16.1	33.3 ± 5.7
Male	2 (20%)	6 (60%)	5 (50%)
Female	8 (80%)	4 (40%)	5 (50%)
Type
Unilateral	0 (0%)	7 (70%)
Bilateral	10 (100%)	3 (30%)
Hemorrhagic	2 (20%)
Ischemic	8 (80%)
Mean duration of symptoms ± SD (years)	1.8 ± 1.8	1.7 ± 3.0
Suzuki stage (mainly the high-level side)
1	0 (0%)
2	1 (10%)
3	1 (10%)
4	4 (40%)
5	2 (20%)
6	2 (20%)

SD：standard deviation; MMD: moyamoya disease; ICAD: intracranial Atherosclerotic disease.

Quality control of the GeneChip results of exosomal RNAs and whole transcriptome analysis

The data detected using the GeneChip WT Pico Reagent kit was summarized and sorted, and quality controls were performed to ensure the stability and reliability of the data. Exosomal RNAs were sorted into five types for subsequent analysis: mRNA, lncRNA, circRNA, small RNA, and others. Based on the differential expression thresholds (1.50-fold for downregulation and upregulation of RNA expression), we performed pairwise comparisons among the three groups using differential expression analysis. For patients with MMD, there were 1,486 downregulated RNAs and 2,405 upregulated RNAs compared with that in healthy controls. There were 834 downregulated RNAs and 1,001 upregulated RNAs in patients with ICAD compared with that in healthy controls. There were 2,184 downregulated RNAs and 1,261 upregulated RNAs in ICAD patients compared with that in MMD patients. The hierarchical clustering heat maps of group A (MMD) and Group C (HC) RNA are shown in Figure 1(d) to (h), and the volcano is shown in Figure 1(i). Detailed biogenic analysis results of groups A (MMD) and B (ICAD), group B (ICAD), and C (HC) are shown in supplementary material Fig S5-6.

RNA profiles suggested dysfunction of immunity and extracellular exosomes in MMD patients

To investigate the distinct RNA profiles in MMD patients, we performed a bioinformatics analysis of the RNA profiles of MMD patients and healthy controls. The differences among the three groups of LncRNA, circRNA, smallRNA and others are disclosed in the Supplementary materials, as shown in Tables S2-S4. We annotated 1,486 downregulated and 2,405 upregulated RNAs to biological processes, cellular components, and molecular function GO terms. The results showed that both down- and upregulated genes were annotated significantly as being part of an immune response. Moreover, 27 miRNAs were associated with neutrophil degranulation, indicating immune dysfunction in patients with MMD; 12 downregulated RNAs were annotated as immune responses, and some were also associated with the regulation of complement activation (Figure S2).

For cellular component terms, it was worth noting that 88 upregulated RNAs were significantly related to extracellular exosomes, which indicated that extracellular exosomes might play an important role in the pathogenesis of MMD. In terms of molecular function, 178 downregulated RNAs were related to protein binding and 11 upregulated RNAs were distinctly related to RNA binding, which suggested dysregulation of gene expression in MMD patients (Figure S3).

We also performed pathway enrichment analysis on the RNA profiles, and the results indicated immune dysfunction in MMD patients. Eight downregulated RNAs were associated with the NOD-like receptor signaling pathway and six downregulated RNAs were related to the toll-like receptor signaling pathway, both of which play an important role in inflammation and immunity. Moreover, 11 upregulated RNAs were significantly related to the cluster of differentiation (CD) molecules, which are cell surface markers useful for the identification and characterization of leukocytes (Figure S4).

qPCR identified the differential expression of focus RNAs in SDEs

From the differentially expressed RNAs, we selected six significant differential RNAs as the target RNAs. The relative expression of these RNAs is shown in Table 2. To identify the differential expression of target RNAs, we performed qPCR using SDEs from patients with MMD (n = 5) and patients with ICAD (n = 5) (Figure 3). The results showed that these RNAs were differently expressed in SDEs from MMD and ICAD patients, indicating that they could play an important role in the pathogenesis of MMD and ICAD.

Table 2.

The focus RNAs with significantly differential expression.

RNA	P-value	logFC (abs)	Regulation	Gene
hsa_circ_0000583	5.919855e-4 (B vs A)8.656197e-4 (A vs C)	1.959 (B vs A) 1.879(A vs C)	Down(B vs A) Up(A vs C)	None
hsa_circ_0090577	9. 092384e-4(B vs A)	1.659(B vs A)	Down(B vs A)	CACNA1F
hsa_circ_0001491	4.832206e-3(B vs A)	1.765(B vs A)	Up(B vs A)	IPO11
	1.820000e-4(A vs C)	2.235(A vs C)	Down(A vs C)
hsa_circ_0076043	2.231238e-3(A vs C)	3.249(A vs C)	Down(A vs C)	RPS10-NUDT3
hsa_circ_0051937	2.112324e-4(A vs C)	21.556(A vs C)	Up(A vs C)	PRMT1
hsa_circ_0074463	3.469654e-3(B vs C)	2.071(B vs C)	Up(B vs C)	PPP2R2B

Group A = MMD, Group B = ICAD, Group C = Heathy Controls.

Figure 3.

The relative RNA levels of six focus RNAs between healthy controls, ICAD group, and MMD group. The six focus RNAs differentially expressed in SDEs from MMD patients and ICAD patients, which indicated theses RNAs could play an important role in the pathogenesis process of MMD and ICAD.

Construction of cell lines with low expression of candidate RNAs

We selected hsa_circRNA_0000583, hsa_circRNA_0090577, and hsa_circRNA_0051 937 as candidate RNAs to identify their functions in MMD and ICAD. As shown in Figure 2ABC, the expression of candidate RNAs decreased significantly in the sh-hsa_circRNA group compared to that in the shRNA-NC group, indicating that cell lines with candidate RNA knockdown had been successfully constructed.

Effects of candidate genes on cell viability and proliferation rate

The viability of HBMEC at 24, 48, and 72 h was detected using CCK 8 assay kit. As shown in Figure 2(d), compared with the shRNA-NC group (100.00% ± 3.41%, 169.32% ± 5. 42%, 222.12% ± 8.05%), the cell viability of the sh-hsa_circRNA_0000583 (91.0 3% ± 10.04%, 127.15% ± 7.55%, 147.78% ± 9.97%), sh-hsa_circRNA_0090577 (87.1 4% ± 5.89%, 98.57% ± 1.65%, 118.63% ± 7.19%), and sh-hsa_circRNA_0051937 (89.4 5% ± 5.89%, 111.76% ± 8.69%, 131.63% ± 6.61%) groups decreased significantly. Moreover, as shown in Figure 4(e) and (f), the apoptosis increased markedly in the sh-hsa_circRNA_0000583 (16.65% ± 1.28%), sh-hsa_circRNA_0090577 (18.59% ± 0.97%), and sh-hsa_circRNA_0051937 (17.14% ± 0.81%) groups compared with that in the shRNA-NC group (7.47% ± 0.48%). EdU cell proliferation assay indicated that the cell proliferation was inhibited significantly in the sh-hsa_circRNA_0000583 (35.51% ± 3.31%), sh-hsa_circRNA_0090577 (29.95% ± 3.18%), and sh-hsa_circRNA_0051937 (31.46% ± 1. 95%) groups compared with that in the shRNA-NC group (51.74% ± 2.92%) (Figure 4(a) and (b)).

Figure 2.

(a, b, c) The relative RNA expression levels of three candidate RNAs between experimental group and shRNA-NC after plasmid transfection. The relative RNA levels of sh-hsa_circRNA_0000583, sh-has_circRNA_0090577, and sh-hsa_circRNA_0051937 compared respectively with shRNA-NC. (d) Cell viability of three RNA knockdown groups and shRNA-NC group detected after 24, 48, and 72 h incubation. Compared with shRNA-NC group, the cell viability of RNA knockdown groups decreased significantly, which identified these RNAs knockdown inhibited the cell viability.

Figure 4.

(a) EdU assay of HBMECs. The proliferating nuclei were stained with EdU (red) and Hoechst (blue) for 2 h. Three random pictures per group were used to count the numbers of EdU-positive cells and Hoechst-positive cells. (b) EdU assay of HBMECs. Data are shown as the mean ± SD from four independent experiments (n = 4). *P < 0.05, **P < 0.01. (c) Capillary formation was measured using the tube formation assay. HBMECs from each group were seeded on Matrigel-coated wells and incubated for 72 h to allow the formation of capillary-like structures (n = 4). (d) The quantitative data of tube formation assay are shown as the percentage of tube formation. (e) Apoptosis in HBMECs was assessed using flow cytometry after staining with AnnexinV and PI (a). (f) Data are shown as the mean ± SD from four independent experiments. **P < 0.01.

Tubule formation of HBMEC by candidate genes

The lengths of the tubules and branch points in the sh-hsa_circRNA_0000583, sh-hsa_circRNA_0090577, and sh-hsa_circRNA_0051937 groups were significantly shorter than those in the shRNA-NC group (Figure 4(c)). Compared to that in the shRNA-NC group (69.33 ± 4.51), the number of tube formations decreased significantly in the sh-hsa_circRNA_0000583 (30.67 ± 4.04), sh-hsa_circRNA_0090577 (25.67 ± 4.04), and sh-hsa_circRNA_005193 7 (35.00 ± 3.60) groups. Tube formation decreased significantly in the sh-hsa_circRNA_0000583 (44.23% ± 5.83%), sh-hsa_circRNA_0090577 (37.02% ± 5.83%), and sh-hsa_circRNA_0051937 (50.48% ± 5.20%) groups than in the shRNA-NC group (100.00% ± 6.50%).

Discussion

To investigate the pathogenesis of MMD and ICAD, we studied the RNA profiling of sEV/exosomes from MMD and ICAD. We further performed qPCR and lentivirus infection to identify the function of the target RNAs. The RNA profiling revealed that candidate RNAs were expressed differently in patients with MMD and ICAD, and healthy controls. Moreover, we found that knockdown of these candidate RNAs decreased cell viability, proliferation, and migration of HBMECs.

In this study, we found that RNAs associated with the immune response were upregulated in patients with MMD. Further analysis revealed that the pathways such as CD molecule pathway and cytokine-cytokine receptor interaction were upregulated in the immune response. Elevated levels of inflammation-related molecules and cytokines in the serum of patients with MMD suggest that abnormal angiogenesis caused by inflammation contributes to disease pathogenesis. CD molecules are cell surface markers used for the identification and characterization of leukocytes. Fujimura et al. found that the serum level of soluble CD163 significantly increased in patients with MMD compared with that in healthy controls, which is an activating marker for CD163+ M2-polarized macrophages, which is complicated in a variety of autoimmune disorders. ¹⁷ Kang et al. revealed that patients with MMD exhibited significantly higher plasma concentrations of the cytokines MMP-9, MCP-1, IL-1β, VEGF, and PDGF-BB compared with those in healthy controls. The imbalance of cytokines seemed to correlate with disease pathogenesis. Increased plasma levels of the cytokines MCP-1 and VEGF in patients with MMD may play a role in the recruitment of vascular progenitor cells and in the formation of collateral vessels. ¹⁸

We identified candidate RNAs in SDEs from patients with MMD and ICAD. Table 2 lists the information for all of these RNAs that were found to be significantly expressed. We hypothesized that these RNAs may play an important role in the pathogenesis of MMD and ICAD. Importin-11 (Ipo11) is a novel transport receptors (karyopherins) in the human importin family that mediates the nucleocytoplasmic transport of proteins and RNA cargos. ¹⁹ It was found that importin-11 plays a role in synaptic development and functions in Drosophila. Characteristic defects in synaptic transmission in adult photoreceptors and at larval neuromuscular junctions (NMJs) can be found in importin‐11 mutants. Furthermore, it acts as a characteristic factor resembling the phenotype of bone morphogenic protein (BMP) pathway disruption. Neurons deficient in importin-β11 were viable and properly differentiated but showed distinct defects. ²⁰ Moreover, it was identified that knockout of Ipo11 nuclear import factor affects normal embryonic development and regulates embryo-lethal phenotypes in mice, which indicates that Ipo11 is essential for normal embryonic development in mice. ²¹ Relevant gene analyses for MMD are currently unavailable. For the first time, we observed that IPO11 circRNA is downregulated in patients with MMD compared with that in healthy controls. Furthermore, it is downregulated in patients with MMD compared with that in patients with ICAD, with no significant difference in expression between ICAD patients and healthy controls, suggesting that IPO11 might play a role in MMD pathogenesis.

PRMT1, the major protein arginine methyltransferase in mammals, catalyzes monomethylation and asymmetric dimethylation of arginine side chains in proteins. ²² In the vascular system, PRMT1 dysregulation is predicted to be associated with endothelial dysfunction and vascular diseases via the induced upregulation of asymmetric dimethylarginine (ADMA), an endogenous inhibitor of endothelial nitric oxide synthase.^23–26 Furthermore, changes in ADMA levels in cerebrospinal fluid are linked to the development and resolution of vasospasm seen on arteriograms after subarachnoid hemorrhage (SAH), implying that endogenous inhibition of endothelial nitric oxide synthase by ADMA may play a role in the development of delayed cerebral vasospasm. ²⁷

This is the first study to find that PRMT1 circRNA is up-regulated in MMD patients compared with that in healthy controls. We used plasmid transfection to knockdown PRMT1 circRNA in HBMECs and observed that both tube length and the number of tubes formed had significantly decreased, indicating that upregulation of PRMT1 circRNA in the MMD could lead to endothelial dysfunction and vascular abnormalities. Relevant models are needed to study the role of PRMT1 in angiogenesis in MMD.

CACNA1F encodes Ca_v1.4 complex of the L-type calcium channel alpha-1 subunit, which was previously thought to be expressed specifically in the retina. Mutations in this gene result in incomplete congenital stationary night blindness (iCSNB2) in humans. ²⁸ However, CACNA1F mRNA and protein have been found in skeletal muscles, bone marrow, spinal cord, spleen, thymus,^28,29 and lymphocytes,^30–32 indicating that the CACNA1F gene has multiple functions that remain unclear. We knocked down CACNA1F circRNA in HBMECs and observed that both tube length and the number of tubes formed had decreased. We also found that knockdown of CACNA1F circRNA induced apoptosis in HBMECs, indicating that downregulation of CACNA1F circRNA could play a role in the pathogenesis of MMD.

This study had some limitations. First, the sample size may be small to identify differential expression of RNAs. And it might effect the result of study as we did not distinguish hemorrhagic MMD and ischemic MMD due to the limitation of sample size. Xing Peng et al. have discovered that dysregulated genes in peripheral blood of MMD with transient ischemic attack, cerebral infarction and intraventricular or intracerebral hemorrhage as initial symptoms all mainly played key roles in extracellular organization, inflammatory and immune responses, which was corresponded with our results. ³³ It seemed that different types of patients’ mainly symptom might not effect the pathogenesis of MMD significantly. Xia Wang et al. have found that the effects on proliferation of cerebrovascular endothelial cells were both significant in SDEs from hemorrhagic MMD and ischemic MMD. However, they also identified that some proteins related to dysfunction of immunity were significantly upregulated in hemorrhagic MMD and these proteins were distinct from these detected in healthy controls and ischemic MMD patients. ¹³ It might need to be explored and studied further whether the pathogenesis progression differed between hemorrhagic MMD and ischemic MMD.

Second, as there is currently no successful in vitro model of MMD, the candidate RNAs evaluated in this study were identified only from the perspective of HBMECs. What’s more, we just compared differential expressed RNAs among groups and identified that these dysregulated focus RNAs might lead to the pathogenesis of MMD, but did not compare RNAs profile of patients from different stages of MMD, which might explain the function of RNAs during the progression of MMD.

Yue Wang et al. have identified the function of RNF213 p.R4810K (G > A) in independent clinical cases at different periods and compared the different effects of AA and GA of RNF213. The result showed that MMD patients in the Suzuki intermediate stage were more common in the GA group and compare with the GG group, the number of adult patients in the early stage was significantly lower in the GA group, which indicated that GA had different predictive effects on Suzuki stages in MMD. ³⁴ It is identified that different genes expression might play an important role during progress of MMD and RNA profile might be the evidence. In future studies, we will explore and identify the effect and function of exosomal RNA at different periods of MMD.

Conclusion

Our study examined the exosomal RNA profiles of patients with MMD and ICAD, and identified several RNAs with significantly differential expression compared with that in healthy controls. The findings of our study provide a direction for investigating the biological markers of MMD. The effect of dysregulated RNAs on tubule formation in HBMECs also contributes to the understanding of how the dysregulation of these RNAs affects the pathogenesis of MMD. In future studies, we will explore the construction of a model based on these newly discovered dysregulated RNAs.

Availability of data and materials

All data generated or analyzed during this study are included in this published article. Some or all data, models, or code generated or used during the study are available from the corresponding author by request.