MicroRNA expression in circulating exosomes as potential cerebral atherosclerosis biomarkers: comparative profiling study

Full Text

Introduction

Cerebral atherosclerosis (CA) is a chronic inflammatory disease of the arteries supplying blood to the brain, accompanied by the deposition of oxidized lipoproteins in the vascular wall. This highly prevalent condition accounts for over one-third of ischemic cerebrovascular events; however, existing treatment and prevention approaches (specifically, antiplatelet, lipid-lowering, and anti-inflammatory therapies) fail to achieve a reliable reduction in the proportional contribution of CA to cerebrovascular disease morbidity and mortality.

Atherogenesis (irrespective of location) involves several relatively sequential abnormal processes: endothelial dysfunction, subintimal lipid infiltration, inflammatory infiltration with foam cell formation, and atherosclerotic plaque development [1]. Rupture of the latter (commonly termed an unstable plaque) leads to in situ thrombosis and stroke via atherothrombosis or artery-to-artery embolism mechanisms; in cases of prolonged CA progression, a hemodynamic stroke subtype may occur due to significant stenosis or occlusion of a major artery. Regardless, the development and progression of CA represent a complex, multicomponent process involving numerous cellular, biochemical, and epigenetic interactions. The latter notably includes alterations in microRNAs (non-coding oligonucleotides that enable post-transcriptional modification of gene expression) — significant regulatory mechanisms for most biological processes [2].

Russian Center of Neurology and Neurosciences has conducted pioneering research to study the microRNA expression profile in the leukocytes and circulating microRNAs. Thus, pro- and antiatherogenic microRNAs have been described [3, 4], along with the expression pattern of several microRNAs associated with various pathogenetic atherogenesis stages. In light of recent discoveries, particular interest in this area lies in studying microRNAs circulating within exosomes — small (40–160 nm in diameter) extracellular vesicles of endosomal origin that are secreted by nearly all cell types and surrounded by a phospholipid bilayer, containing proteins, lipids, and nucleic acids that reflect the origin of the donor cell [5]. Previously, these structures were viewed solely in the context of removing intracellular “waste”. It is now believed that they play a key role in intercellular communication and signaling, participating in the regulation of physiological and pathological processes, and are actively studied as diagnostic biomarkers and potential therapeutic agents for targeted molecule delivery [6]. Exosomes protect microRNAs from degradation, participate in their selective sorting and transport, and reflect the functional state of source cells, including endothelium, vascular smooth muscle cells (SMCs), macrophages, and platelets. Unlike the total circulating microRNA fraction, the exosomal pool carries additional biological information related not only to expression levels but also to active secretion mechanisms and intercellular signaling [7].

There are data on the role of exosomal microRNAs in coronary atherosclerosis and other forms of cardiovascular pathology, while evidence of their involvement in the CA pathogenesis remains scarce [8, 9]. Some studies were conducted in small cohorts, often without accounting for the vascular lesion sites, and data obtained from plasma or leukocyte fractions are not always reproducible when analyzing exosomes. Moreover, there are almost no studies dedicated to systemic profiling of exosomal microRNAs in patients with atherosclerotic lesions of extra- and intracranial arteries.

In this regard, comprehensive analysis of exosomal microRNAs in patients with CA represents a relevant scientific and clinical task aimed at deepening the understanding of molecular disease mechanisms and identifying novel diagnostic and prognostic markers.

Materials and Methods

This prospective study enrolled patients admitted to the Russian Center of Neurology and Neurosciences in 2023–2025. The main inclusion criterion for the study group was angiographically confirmed atherosclerosis of extra-/intracranial arteries (duplex scanning and/or magnetic resonance/computed tomography angiography) (n = 43; median age 68 years). The comparison group comprised patients with other cerebrovascular disease of non-atherosclerotic origin (n = 17; median age 46 years). The study was approved by the local ethics committee of the Russian Center of Neurology and Neurosciences (Protocol No. 11-4/19 dated November 20, 2019) and conducted in accordance with the principles of the Declaration of Helsinki. All participants provided written informed consent.

Blood samples for laboratory testing were obtained via cubital venipuncture in the morning after fasting using vacuum systems and serum separator tubes with clot activator and gel. Sample collection, transportation, storage, and other preanalytical procedures were performed in accordance with the National Standard of the Russian Federation. The obtained serum was aliquoted and stored at –80°C. Exosomal microRNAs were isolated from blood serum samples in 3 stages using the commercial ExoQuick Exosome Isolation and RNA Purification Kit (for Serum & Plasma, SBI). The SeraMir Exosome RNA Amplification Kit (SBI) was used for exosomal microRNA quantification. Expression profiles of 384 exosomal microRNAs were analyzed in all patient samples. For each sample, a 384-well chemically pure blank (DNA/RNA-free) plate was used, into which 384 ready-made primers from the kit were dispensed. Real-time PCR was subsequently performed on a Veriti thermal cycler (Veriti Thermo FS) with a 384-well plate using in-house software.

Statistical analysis was performed using the R programming language v. 4.4.1 in the RStudio software environment (v. 2025.05.1) with the following packages: “tidyverse”, “gtsummary”, “corrplot”, “pheatmap”, “igraph”, “ggplot2”, “ggrepel”, “scales”, and “circlize”. Descriptive statistics were represented by medians, as well as upper and lower quartiles for continuous variables and frequencies for discrete variables. For comparisons between two independent groups, the Wilcoxon–Mann–Whitney test (for continuous variables) or Pearson’s χ² test (for discrete variables) was used. Expression levels of exosomal microRNAs were normalized using the global mean of all endogenous microRNAs detected in the sample, followed by calculation of ΔCt for each microRNA. Since the analysis was based on ΔCt, a positive contrast value corresponded to a higher ΔCt in the atherosclerosis group, i.e., lower expression levels. Statistical significance of differential expression was assessed using linear models (limma) with empirical Bayes variance moderation, which provides stable estimates with limited biological replicates. The primary differential microRNA expression analysis was performed without covariate adjustment due to the pilot nature of the study and limited sample size. To evaluate potential impact of demographic and clinical confounders, a sensitivity analysis was additionally performed. For the 9 microRNAs identified in the primary analysis, linear regression models were used, including the comparison group, age, sex, and presence of diabetes as covariates (p_cov).

The heatmap was constructed based on normalized –ΔCt values for the 30 most significant microRNAs selected by minimal adjusted p-value (False Discovery Rate, FDR). The color scale ranged from blue (low expression) to red (high expression). Hierarchical clustering was performed using Euclidean distance and Ward’s method. Row annotation reflected sample group affiliation: control (light gray) and atherosclerosis (dark gray). The heatmap was used solely for visualizing data structure and unidirectional changes, not as a statistical test. Genes identified in the optimal microRNA–target gene interaction network were functionally annotated using Gene Ontology, KEGG, and Reactome databases. Based on their predominant involvement in endothelial dysfunction, lipid metabolism, inflammatory signaling, oxidative stress, vascular SMC proliferation, and vascular remodeling, the genes were assigned to corresponding stages of atherogenesis. The significance level was 0.1, and all tests were two-sided.

Results

The general clinical characteristics of the study patients are presented in Table 1. The main group comprised patients with CA, among whom 15 (35%) had a history of ipsilateral ischemic stroke (occurring >6 months prior) with corresponding vascular territory symptoms. The remaining patients met the criteria for asymptomatic status in terms of CA, though most exhibited nonspecific symptoms (mild/subjective cognitive impairments, complaints of tinnitus/dizziness, etc.). Comparison group patients were statistically significantly younger than patients with CA (46 vs. 68 years; p < 0.001). A trend toward sex distribution differences was observed (women 53%, men 28%; p = 0.067), which given the limited sample size precludes interpreting the groups as fully comparable. Additionally, this group included 2 (12%) patients with prior non-atherothrombotic cerebrovascular events. Overall, the expected comorbidity burden was higher in the main group; notably, type 2 diabetes occurred exclusively in this group, reflecting CA progression features.

 

Table 1. The general clinical characteristics of the study patients

Parameter		All patients (N = 60)	CA (N = 43)	Comparison group (N = 17)	p
Age, years	Ме [Q1; Q3]	65 [52; 72]	68 [60; 72]	46 [45; 53]	< 0,001
Women	n/N (%)	21/60 (35)	12/43 (28)	9/17 (53)	0,067
Stroke	n/N (%)	17/60 (28)	15/43 (35)	2/17 (12)	0,110
Type 2 diabetes	n/N (%)	16/60 (27)	16/43 (37)	0/17 (0)	0,003

 

In accordance with the primary study objective, all patients underwent exosomal microRNA profiling. After quality control, including removal of spike-in controls, no-template control wells, placenta-specific microRNAs (e.g., miR-324-3p, miR-369-5p), and low-expressed microRNAs (detected in fewer than 5 samples per group), 337 endogenous microRNAs were retained for differential expression analysis.

Data normalization was performed using the global mean of all endogenous microRNAs detected in the sample, followed by ΔCt calculation for each microRNA. Differential expression analysis was conducted using the limma method with empirical Bayes correction, with statistical significance defined as FDR < 0.1 and |log₂FC| > 0.5 (corresponding to >1.4-fold change). Statistically significant differential expression of 9 exosomal microRNAs was identified (Fig. 1), exclusively showing decreased expression in CA patients (Table 2; Fig. 2).

 

Fig. 1. Volcano plot of exosomal microRNA expression.

Red indicates microRNAs with statistically significant differences between groups. The X-axis shows log₂ fold change (positive values indicate lower expression in the comparison group); the Y-axis shows –log₁₀ adjusted p-value (values >1.0 correspond to p < 0.1).

 

Table 2. Exosomal expression of statistically significant differentially expressed microRNAs

MicroRNA	Expression in CA	logFC	p	FDR	Biological fold change	pcov
hsa-miR-18a	â	9.497686112	< 0.0001	0.003069536	0.001383285	0.0008
hsa-miR-539	â	7.243347486	< 0.0001	0.007019844	0.006599866	0.002
hsa-miR-20a	â	7.260677171	0.000173963	0.019541841	0.006521063	0.09
hsa-miR-100	â	7.326407473	0.000252608	0.021282248	0.006230625	0.001
hsa-miR-10b	â	5.954562429	0.000336715	0.022694558	0.01612494	0.10
hsa-let-7f	â	5.582010663	0.000603783	0.03391246	0.020876003	0.045
hsa-miR-148b	â	22.19720587	0.000938846	0.045198729	2.08E-07	0.583
hsa-miR-187	â	7.981040114	0.001518526	0.063967923	0.003957925	0.025
hsa-miR-374a	â	17.46588418	0.002207399	0.082654839	5.52E-06	0.730

 

Fig. 2. Box plot of statistically significant differentially expressed exosomal microRNAs.

The Y-axis shows normalized ΔCt values; dots represent individual samples; the thick line indicates the median; vertical lines show 1.5 × IQR.

 

To visualize expression patterns of the most significant microRNAs, a heatmap of 30 microRNAs with the lowest FDR values was generated (Fig. 3). The heatmap reveals that all identified differentially expressed microRNAs show reduced levels in the atherosclerosis patient group compared to the control group. Hierarchical clustering confirmed a pathogenic relationship among these changes: samples separated into two main clusters corresponding to clinical groups, while microRNAs with similar expression profiles grouped together, suggesting biological coherence of the alterations. However, the observed sample clustering reflects the exploratory nature of the analysis and should be interpreted with caution.

 

Fig. 3. Heatmap of the top 30 differentially expressed exosomal microRNAs in patients with CA compared to the control group.

Color reflects the normalized expression level (–ΔCt): red indicates high expression, blue indicates low expression. Samples are grouped by study group affiliation (light gray — control, dark gray — atherosclerosis). Hierarchical clustering was performed across rows (samples) and columns (microRNAs).

 

Principal component analysis was used as an exploratory visualization of the overall data structure (Fig. 4). The first two components (PC1 and PC2) explained approximately 25% of the total variability, reflecting significant interindividual heterogeneity in exosomal microRNA profiles (principal component analysis was not considered as the primary statistical test for intergroup differences).

 

Fig. 4. Principal component analysis of exosomal microRNAs.

Each point represents one sample; color denotes group affiliation: light blue — control, red — CA. The mean silhouette width was 0.55, indicating a moderately pronounced cluster structure that requires cautious interpretation due to group heterogeneity and imbalance.

 

Further analysis included assessment of significant microRNA–target gene interactions for microRNAs demonstrating the highest significance level (Fig. 5).

 

Fig. 5. Chord diagram of microRNA–target genes based on the miRTarBase database (MIENTURNET online resource).

Connections with FDR < 0.1 and >3 microRNAs regulating a single gene are highlighted. Line thickness corresponds to interaction strength; no more than 2 microRNAs per gene are shown; color coding indicates the proposed role of the microRNA/gene in atherogenesis.

 

Analysis of the microRNA–target gene network revealed key hub genes regulated by multiple microRNAs and associated with various stages/mechanisms of atherogenesis. The most represented genes included those involved in epigenetic regulation of endothelial function (DNMT1, MSL3, SMCHD1), lipid metabolism and anti-inflammatory signaling (RORA, NCOA6), oxidative stress and macrophage activation (SESN3), vascular SMCs proliferation (CCND1, CTDSPL2), as well as cellular migration and vascular remodeling (PAFAH1B1, BTBD7).

Discussion

For the first time in a Russian population, a pilot study profiling exosomal microRNAs in patients with CA was conducted, revealing a differentiated expression pattern characteristic of cerebral vascular diseases. Among the core pool of statistically significant microRNAs (hsa-miR-18a, hsa-miR-539, hsa-miR-20a, hsa-miR-100, hsa-miR-10b, hsa-let-7f, hsa-miR-148b, hsa-miR-187, and hsa-miR-374a), the most notable finding was coordinated depletion of exosomal microRNAs, manifested by a unidirectional decrease in their levels in circulating exosomes of patients with CA.

The reduced exosomal expression of hsa-miR-18a and hsa-miR-20a, members of the miR-17-92 cluster, is of particular interest. miR-18a is known to actively promote proliferation, migration, and differentiation of vascular SMCs [10], and prior studies have shown its elevated plasma levels in atherosclerosis patients [11]. Other researchers have demonstrated the cardioprotective effect of reduced miR-18a expression [12], mediated through inactivation of the Akt/mTOR axis. miR-20a appears to be an anti-inflammatory microRNA that inhibits TLR4–NF-κB/NLRP3 cascades in endothelial cells, whose expression is suppressed by oxidized low-density lipoproteins [13]. This study provides the first-ever demonstration of significantly reduced expression of these microRNAs in exosomes of patients with CA.

hsa-miR-100 is highly expressed in the endothelium and exhibits anti-inflammatory properties by enhancing autophagy in the vascular wall [14]. We previously demonstrated reduced plasma levels of this microRNA in patients with intracranial atherosclerosis [15]; a similar change in its exosomal expression confirms its potentially significant role as an atheroprotective molecule in the cerebral atherosclerosis. Interpretation of reduced miR-10b expression in our study proved more complex: several studies identify hyperexpression of this microRNA as a proatherogenic factor, particularly in patients with unstable angina [16] and intracranial atherosclerosis [17]. This discrepancy might be attributed to the lack of a clear correlation between leukocyte and exosomal expression for certain microRNAs. On the other hand, at least in unstable angina, miR-10b expression was higher in patients with actively progressing coronary atherosclerosis, whereas in our study blood samples were collected over a significant time interval from the acute cerebrovascular episode (if present in the medical history).

hsa-let-7f (lethal-7) was one of the first discovered microRNAs, notable for its high conservation across animal species. In the context of atherosclerosis, data demonstrate this microRNA’s high tropism for plaque structures and its role in shaping a potential atheroprotective response [18]. At the same time, reduced exosomal expression of let-7f was detected in patients with diabetes — one of the most significant risk factors for CA [19].

An important atheroprotective factor confirmed in our study was reduced expression of hsa-miR-148b. According to X. Zhang et al., this microRNA is significantly less expressed in atherosclerotic plaques, and its upregulation inhibits proliferation and migration of vascular SMCs [20]. The observed decrease in exosomal miR-539 levels in CA aligns with experimental data indicating its inhibitory role in SMCs proliferation and migration [21]. Deficiency of miR-539-5p may contribute to abnormal vascular wall remodeling and intimal thickening. Exosomal depletion of this microRNA may reflect a shift toward a proatherogenic phenotype in vascular SMCs.

miR-187-3p is known as a key regulator of interleukin-10-mediated anti-inflammatory responses in macrophages. For example, its insufficient levels increase production of proinflammatory factors like tumor necrosis factor-α and interleukin-6 [22]. Reduced exosomal expression of miR-187-3p in CA underscores the role of chronic inflammation in this pathology. The decreased expression of miR-374 in CA in our study contradicts data on its proatherogenic role in a cohort of patients with asymptomatic carotid atherosclerosis [23]. Such discordance between the general circulating fraction and the exosomal fraction appears typical for microRNAs: the exosomal pool reflects not only expression levels but also selective microRNA transport.

This study has some limitations that should be considered when interpreting its results. First, this pilot study has a hypothesis-generating nature and was conducted on a relatively limited patient sample, which reduces the statistical power of the analysis and limits the generalizability of the findings to broader populations. No independent technical validation was performed for statistically significant microRNAs within this study, requiring confirmation of the identified signals in a separate cohort and on an alternative analytical platform. Additionally, the study groups differed in age, sex, and comorbidity profile, particularly in the prevalence of diabetes, which could independently influence the microRNA profile regardless of CA.

No healthy control group and the use of a clinical comparison group limit the ability to fully distinguish the effects of CA from age-associated changes and those related to cerebrovascular pathology. Therefore, the results should be interpreted as a basis for subsequent validation in larger, age- and comorbidity-matched cohorts, with independent technical validation.

An important methodological limitation is the cross-sectional study design, which precludes assessment of the changes in exosomal microRNA expression relative to the stage of atherogenesis, disease activity, or administered therapy. Additionally, this study did not perform functional validation of the identified microRNA-target interactions in vitro or in vivo, and the network analysis of microRNA-target genes relied on bioinformatic predictions and data from previously published studies.

Exosomal fraction of microRNAs reflects not only cellular expression levels but also selective microRNA sorting and secretion, complicating direct comparison of these data with results from studies using plasma, leukocyte, or tissue fractions. Finally, no morphological characterization of atherosclerotic plaques (stability, composition, signs of inflammation) limits the ability to correlate molecular changes with the vascular wall lesion phenotype.

Conclusion

This study is the first to demonstrate in a Russian patient cohort that CA is associated with a significant and consistent reduction (depletion) in exosomal expression of several key microRNAs involved in regulating inflammation, endothelial function, and vascular remodeling. The findings support the concept that impaired exosome-mediated intercellular communication serves as a key molecular mechanism for the development and progression of atherosclerotic lesions in the cerebral arteries.

The identified profile of exosomal microRNAs and their associated target genes provides a foundation for further research aimed at developing molecular biomarkers for CA and refining pathogenetic targets for personalized prevention and therapy of cerebrovascular diseases.

About the authors

Anton A. Raskurazhev

Russian Center of Neurology and Neurosciences

Author for correspondence.
Email: annaly-nevrologii@neurology.ru
ORCID iD: 0000-0003-0522-767X

Cand. Sci. (Med.), Head, Laboratory of neuropharmacological fMRI, Institute of Clinical and Preventive Neurology

Russian Federation, Moscow

Аlla А. Shabalina

Russian Center of Neurology and Neurosciences

Email: annaly-nevrologii@neurology.ru
ORCID iD: 0000-0001-7393-0979

Dr. Sci. (Med), leading researcher, Head, Laboratory diagnostics department, Institute of Clinical and Preventive Neurology

Russian Federation, Moscow

Polina I. Kuznetsova

Russian Center of Neurology and Neurosciences

Email: annaly-nevrologii@neurology.ru
ORCID iD: 0000-0002-4626-6520

Cand. Sci. (Med.), senior researcher, 1^st Neurological department, Institute of Clinical and Preventive Neurology

Russian Federation, Moscow

Vladislav A. Annushkin

Russian Center of Neurology and Neurosciences

Email: annaly-nevrologii@neurology.ru
ORCID iD: 0000-0002-9120-2550

Cand. Sci. (Med.), neurologist, 1^st Neurological department, Institute of Clinical and Preventive Neurology

Russian Federation, Moscow

Marine M. Tanashyan

Russian Center of Neurology and Neurosciences

Email: annaly-nevrologii@neurology.ru
ORCID iD: 0000-0002-5883-8119

Dr. Sci (Med), Professor, Full Member of RAS, Deputy Director for Science, Head, 1^st Neurological Department, Institute of Clinical and Preventive Neurology

Russian Federation, Moscow

References

Supplementary files