Exploring extracellular vesicle surface markers and microRNA regulation following remote ischemic conditioning in patients with stroke; a randomized-controlled pilot study

Abstract

Remote ischemic conditioning (RIC) has demonstrated neuroprotective effects, yet translating these effects into clinical efficacy has proven challenging. This study aims to uncover the molecular mechanisms underlying RIC’s protective effects, which could potentially identify new therapeutic targets. We conducted the ENOS pilot study, a patient-assessor blinded, Sham-controlled clinical trial, to investigate the molecular changes in patients diagnosed with acute ischemic stroke (AIS). Patients were assigned to undergo either RIC (n = 9), involving five cycles of transient ischemia and reperfusion of the arm, or a Sham treatment (n = 9), using a similar device with less pressure to avoid creating ischemia, within 48 h of symptom onset. We collected plasma samples at three time points: at inclusion (pre-RIC), two hours post-initial RIC, and after seven days, with RIC administered twice daily. Analysis focused on brain biomarkers, specifically extracellular vesicle (EV) surface markers and circulating microRNAs (miRNAs). We identified significant differences in the regulation of CD62 on EVs and five specific miRNAs (miR-374a-5p, miR-20a-5p + miR-20b-5p, miR-19b-3p, miR-24-3p, miR-30d-5p) between the RIC and Sham groups. Notably, miR-374a-5p, known for its neuroprotective properties, was significantly increased in patients treated with RIC (p = 0.015). Furthermore, the changes of several of these miRNAs were correlated with improvements in red blood cell (RBC) deformability and aggregation during shear stress. In addition, we observed significant increases in CD62 on EVs at both the two-hour and 7-day follow-ups in the RIC group. These findings suggest that RIC may induce specific changes in EV surface markers and circulating miRNAs which could serve as future biomarkers for the RIC response. The trial was registered on clinicaltrails.gov (NCT04266639) on February 12, 2020 before trial initation on July 29, 2020.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12868-025-00993-1.

Background

Acute ischemic stroke (AIS) is a major cause of death and disability [1]. The current reperfusion treatments of AIS are highly time-sensitive [2] and are first administered after brain imaging has ruled out an intracranial hemorrhage. Therefore, adjunctive neuroprotective agents that prolong survival of neurons in the ischemic penumbra are of great interest to improve stroke outcome [3].

In pre-clinical trials Remote ischemic conditioning (RIC) has been shown to reduce injury caused by thromboembolic events, acute myocardial infarction, and other ischemic events [4, 5]. RIC is performed by repeated transient ischemia-reperfusion periods in extremities. RIC has been proven safe and feasible to apply in clinical studies but conflicting results on its clinical efficacy have been found in recent large scale clinical trials [6, 7]. In the Chinese RICAMIS trial RIC increased the likelihood of excellent neurological outcome after 90 days in patients with AIS receiving RIC compared to no RIC treatment. Correspondingly, the newly published SERIC-EVT trial showed improved functional outcome at 90 days in patients with AIS who underwent endovascular thrombectomy (EVT) with RIC intervention for 7 days [8] whilst the RESIST trial showed no direct effect on 3-month outcomes in a Danish population of stroke patients. Differences in population, standardized stroke treatment in the Chinese and Danish health care systems, RIC timing and dosage, and study design may explain some of the observed differences. The underlying mechanism of RIC remains unknown, but studies show that RIC affects multiple endogenous mechanisms that seem to involve both humoral and neurogenic pathways and enhances collateral blood flow in the ischemic area [9]. In vivo models suggest that RIC alleviates synaptic impairment during vascular cognitive impairment due to chronic cerebral hypoperfusion via the miR-218a-5p/SHANK2 pathway [10] and oxidative stress and inflammatory responses via the Nrf2/HO-1 pathway in a middle cerebral artery occlusion model which might improve neurobehavioral function [11]. In addition, during stroke, RBCs undergo severe morphological changes, likely contributing to poor deformability, changes in shear stress and impaired microcirculatory flow [12]. Furthermore, identifying a RIC biomarker could help identify RIC responders as RIC is thought to activate protective mechanisms in the body [9].

Extracellular vesicles (EVs) and their microRNA (miRNA) cargo have previously been suggested as potential carriers and regulators of the neuro- and cardioprotective effect of RIC (reviewed in [13, 14]). Further, EVs are involved in cell-to-cell communication and can cross the blood brain barrier, while their surface markers enable targeted transfer of their cargo [15]. In AIS, endothelial-cell derived EVs are associated with stroke severity, lesion volume and outcome of AIS [16]. MiRNAs can regulate gene expression by degrading or suppressing the mRNA targets [17]. These changes can be induced by pathological conditions to salvage or contribute to the recovery of cells following ischemic events. In this study, the nanoString hybridization technology was chosen among other methods to measure miRNA such as qPCR due to its relatively nonbiased quantitation of miRNA compared to RNA sequencing, while it includes 827 unique miRNAs to support the exploratory nature of this study [18]. By exploring miRNA and EVs as the possible effectors of RIC, this could create an insight into the underlying mechanisms of RIC and possibly be translated into future targets of stroke treatments. In addition, the changes correlated to RBC deformability and shear stress are of interest as these parameters have been shown to be involved in changes in the microvasculature highlighting possible molecular changes of AIS progression. Therefore, this pilot study aimed to identify whether RIC treatment in patients with AIS is associated with changes in circulating EV surface markers and miRNAs.

Methods

Study design and study population

The plasma samples for this study were obtained as part of the ENOS trial (approved by the Central Denmark Region Committees on Health Research Ethics ID no. 1-10-72-184-19), which is a single-center, patient-assessor blinded, randomized Sham-controlled trial [19]. The inclusion criteria were: Time from symptom onset to randomization within 48 h, age between 18 and 80 years, independent daily living (mRS 0–2), and documented ischemic stroke on MRI. Patients were excluded if they presented with prior stroke, dementia, or other known neurological conditions as well as diabetes, pregnancy, or known upper extremity peripheral arterial stenosis that could affect the results of the measurements. Hypertension was reported if the patient was diagnosed prior to or during hospitalization. Patients were randomly assigned to RIC/Sham (1:1) using a simple, secure randomization procedure in Research Electronic Data Capture System (REDCap) (Vanderbildt University, Nashville), without stratification [20]. All three patients receiving thrombectomy were randomized to RIC treatment and were therefore excluded from this study. In addition, two RIC patients withdrew their consent. Matching groups of patients treated with RIC (n = 9) or Sham (n = 9) were selected based on age, sex, and reperfusion therapy. Patients diagnosed with AIS were enrolled in the study within 24 h of admission to Aarhus University Hospital, Denmark and randomized to either RIC or Sham treatment. Both treatments were performed using an automatic RIC device that was placed on the upper arm of the patients (non-paretic side). RIC was performed with 5 cycles of 5 min of occlusion + 5 min of deflation inflation. The RIC device occluded with a dynamic pressure according to the patient’s blood pressure adding a minimum of 200 mmHg to ensure arterial occlusion. In contrast, the cuff pressure of the Sham device was fixed at 20mmHg disregarding the level of systolic blood pressure. Data on medical history, clinical characteristics, and treatment were obtained from medical records and the Danish Stroke Registry [19].Patients were assessed at baseline and at follow-up using clinical scores; National institute of Health Stroke Scale (NIHSS), Physical Activity Score for the Elderly (PASE) and Montreal Cognitive Assessment (MoCA) for early detection of mild cognitive impairment [21]. Data on medical history, clinical characteristics, and treatment were obtained from medical records and the Danish Stroke Registry.

Blood sampling and plasma isolation

Blood samples were drawn at three different timepoints for all patients with AIS: before and two hours after the first RIC or Sham treatment, and finally at follow-up after seven days of twice daily RIC/Sham treatment. All blood samples (2 × 3.0 mL EDTA and 2 × 3.5 mL citrate, Vacutainer (BD, Franklin Lakes, NJ, United States)) were spun for 25 min at 3000g at room temperature (RT), at low brake and plasma was collected and stored at -80 °C.

NanoString nCounter MiRNA profiling

Plasma miRNAs were quantified using the Human v3 miRNA nCounter Assay measuring 827 human miRNAs (nanoString, Seattle, Washington, USA) following the manufacturer’s guidelines. According to nanoStrings recommendation, RNA was purified from 800 µl plasma using the Plasma/Serum Circulating and Exosomal RNA Purification Mini Kit (Slurry Format), (Norgen Biotek, Toronto, Canada) followed by sodium/ethanol precipitation of the 100 µL eluted RNA using linearized acrylamide as carrier and resuspended in 8 µL DNase free H₂O. Spike-in solutions (5 µL of 200pM) of osa-miR414, cel-miR-248, and ath-miR159a were added during the purification process as recommended by nanoString (Accession: MIMAT0001330, MIMAT0000304, MIMAT0000177, Integrated DNA Technologies, Iowa, USA). The nanoString panels were prepared according to the manufactures protocol using 3 µL precipitated RNA and the supplied assay controls in the annealing, ligation, and purification steps. Finally, 30 µL reaction mix was loaded on the nanoString panels.

The raw miRNA data was imported into the nSolver 4.0 software (nanoString Technologies). Systems Quality Control (QC) including imaging QC, binding density Q and positive control linearity QC was performed on all samples using default settings. A positive Control Limit of Detection (LOD) QC was performed using two standard deviations (SDs) above the mean of the negative controls. No issues were detected during QC. Raw data were exported, and a background threshold was subsequently set for the raw data (average of negative controls + 2 SDs) and used as LOD. A miRNA was considered above LOD when the average raw expression across all samples were greater than LOD. To account for technical variation, a positive control normalization and top100 normalization was performed in nSOLVER, using the geometric mean of all positive spike-in controls except for the control named F, as recommended by the nanoString Gene Expression Data Analysis Guidelines - MAN-C0011-04, due to F being below the limit of detection for the system, and therefore outside the linear range of the other controls. Positive-control-normalized count data above LOD was then used as input for differential expression analysis.

Differential expression analysis was performed using the DESeq2 package [22] in R [23]. The miRNA analysis was conducted in different designs: (1) Within-group comparison to test for differences from baseline (0 h) to 2 h and 7-day in the RIC or Sham group (paired design) and (2) Between-group comparison analyzing difference-of-differences between Sham and RIC (interaction model) for baseline to 2 h and from baseline to 7-day. A miRNA was denoted as differentially expressed when p < 0.05 in both comparisons described above. This intentionally conservative consistency rule ensures that a feature only is highlighted if it is significant both for the within-RIC change over time and for the RIC-vs-Sham interaction at the same time window (0→2 h or 0→7 d). This intersection test reduces false positives at the per-feature level compared with a single test. Under independence, the chance a null feature meets both p < 0.05 tests is 0.0025.

Characterization of surface markers on EVs with the EV array

To study the EV surface- and surface associated biomarkers on, plasma EVs were captured and visualized using the EV Array protocol outlined by Jørgensen et al., 2013 with a 54 EV marker antibody panel (Table 1), which were diluted to 200 µg/mL in 50 mM trehalose in phosphate-buffered saline (PBS) and printed in triplicates on epoxysilane-coated slides (75.6 × 25 mm, Schott MINIFAB, Germany) using the microarray printer sciFLEXARRAYER S12 (Scienion AG, Berlin, Germany) with a PDC60. 50 mM trehalose in PBS was applied as a negative control and biotinylated goat anti-mouse IgG (Novus Biologicals, Centennial, CO, USA) with final concentrations of 5, 10, and 20 µg/mL were used as positive control. Following the print procedure, the slides were kept at RT until further analysis. Before testing plasma EVs, the slides underwent a 1 h incubation at 120 RPM orbital shaker with a slowly added the manufacturer recommended blocking solution (50 mM ethanolamine, 0.1% sodium dodecyl sulfate, and 100 mM Tris [pH 9]) and then incubated at 200 RPM for 15 min with wash buffer containing low concentration of Tween20 to protect bound EVs (0.05% Tween20^® in PBS). Subsequently, 40 µl of plasma was diluted 2.5X in wash buffer and added to the slides in a multi-well hybridization cassette (Arrayit Corporation, Sunnyvale, CA, USA), which was incubated for 2 h RT at 450 RPM, followed by overnight incubation at 4 °C without stirring.

Table 1.

An investigative panel comprising 54 EV markers was utilized for the EV Array. Each marker (200 µg/mL) was printed onto a slide to create a microarray

Antibody	Supplier	Catalogue no.	Clone
Annexin V	R & D Systems	AF399	-
CD41	Biolegend	303,702	HIP8
CD9	Ancell	156 − 020	SN4/C3-3A2
CD62 E/P	R & D Systems	BBA1	BBIG-E(13D5)
EpCam	Santa Cruz Bio	Sc-59,782	0.N.277
CD63	Biorad	MCA2142	Mem-259
CD8α	R & D Systems	MAB1509	37,006
EGFR	Ab Online	ABIN191750	-
CD62 E (E-selectin)	Thermo Scientific	MA1-22165	1.2B6
CD81	Ancell	302 − 020	1.3.3.22
Hsp90	Abcam	ab13495	IGF1
CD142 (anti-hTF)	R & D Systems	MAB2339	323,514
CD13 (hAPN)	R & D Systems	MAB3815	498,001
CD3	BD Biosciences	555,337	Hit3a
Flotillin-1	Abcam	ab41927	-
CD42a (GP9)	LS Bio	LS-C45240	FMC-25
TSG101	Abnova	H00007251-mol	5B7
CD106 (Vcam1)	R & D Systems	MAB809	HAE-2Z
ICAM-1 (CD54)	eBioscience	BMS1011	R6.5
Claudin-1	Abcam	ab63070	-
CD14	BD Biosciences	555,396	M5E2
CD56	BD Biosciences	559,043	3G8
VEGFR2	Biolegend	359,902	7D4-6
tPA	R & D Systems	AF7449	-
CD31 (PECAM-1)	R & D Systems	AF806	-
VE-Cadherin	R & D Systems	AF938	-
CD105 (Endoglin)	LS Bio	LS-C149182	-
Glypican-1	R & D Systems	AF4519	-
CD235a (glycophorin A)	R & D Systems	MAB1228	R10
CD42b	R & D Systems	MAB4067	486,805
APP	R & D Systems	PPS044	-
Occludin	Abcam	ab167161	EPR8208
CD138/ syndecan-1	R&D Systems	AF2780	-
Osteopontin	R&D Systems	MAB14332	223,112
Adenosine A2a R	R&D Systems	MAB9497-100	599,717
MMP-9	R&D Systems	MAB936	36,020
TIMP-1	R&D Systems	MAB970	63,515
Hsp70	R&D Systems	MAB1663	242,707
Vimentin (VIM)	Avivasysbio	OAEE00559	VI-RE/1
TNF RII	R&D Systems	MAB726	22,210
TNF RI	R&D Systems	MAB225	16,803
Claudin 3	R&D Systems	MAB4620	385,021
PDGF R beta	R&D Systems	AF385	-
BDNF	R&D Systems	MAB648	37,141
GFAP	R&D Systems	MAB2594	273,807
MCP-1	R&D Systems	MAB679	23,007
S100B	R&D Systems	MAB18201	721,703
VEGFR1	R&D Systems	MAB321	49,560
NG2	R&D Systems	MAB2585	LHM-2
HIF1a	R&D Systems	MAB1935	241,812
ADAMTS13	R&D systems	MAB4245	442,323
TIMP-4	R&D systems	MAB974	153,934
vWF-A2	R&D systems	MAB2764	210,909
NMDAR2A	Novus	NBP2-22405	S327-95

Plasma samples from stroke patients were examined on the array. The optimal antibody concentration was found following the procedure of Jørgensen et al., 2015 [25]

After an additional washing procedure, the slides were incubated at RT for 2 h at 450 RPM with biotinylated detection antibodies diluted 1:1500 in wash buffer (anti-human-CD9 (clone SN4/C3-3A2), -CD63 (clone AHN16.1), and -CD81 (clone 1.3.3.22), Ancell Corporation, Stillwater, MN, United States). For detection, Cy5-labeled streptavidin (1:3000 in incubation buffer, Life Technologies, Carlsbad, CA, USA) was added after a wash, and the slides were incubated for 1 h at 450 RPM at RT. Following two washes (first with wash buffer and then with MilliQ water), the slides were dried in a microarray high-speed centrifuge (Arrayit Corporation, Sunnyvale, CA, USA). Slides were scanned as described by Jørgensen et al., 2013 [24] with the InnoScan 710AL microarray scanner (Innopsys Inc., France) at 635 nm.

EV array analysis

Data was obtained with Mapix software Ver 9.1.0 (Innopsys Inc., France) where triplicates of each specific antibody were used to calculate the mean total intensity. Data underwent quality control, where any samples with a positive-to-negative control ratio below 0.97 were excluded. To calculate the total intensity of an antibody, the relative intensity (RI) in a patient sample was adjusted by subtracting the RI of the negative control well, divided by the RI from the negative control (blank spot) in the well of a patient sample. Only EV markers with an RI exceeding that of the negative control were Log₂-transformed and retained for further analysis. The Log₂-transformed intensity values were used as input for differential abundance analysis. The analysis was performed using the Bioconductor package LIMMA [25]. Substantial inter-person variability was observed, and therefore, we normalized the intensity values for each marker to the intensity value of the canonical EV marker CD9. Statistical analysis was conducted using similar comparisons and methods as for the miRNA analysis.

Ektacytometry measuring red blood cell properties

The rheological properties of RBCs were analyzed by ektacytometry (RheoScan-AnD300, RheoMeditech, Seoul, S. Korea) using laser optics and microfluidic technology [26]. The RheoScan-AnD300 uses disposable test kits (RSD-K01, RSD-K02) consisting of a sample chamber, a micro-channel, a waste sample chamber and a rubber cap, and a disposable test chip (RSA-C01) consisting of a flat cylindrical test chamber, sample inlet, air outlet and stirrer. Test time was less than two minutes for each analysis. Further functional details of the microfluidic ektacytometer are described elsewhere [19].

Correlation analysis

Differentially expressed miRNAs and EV markers were correlated to red blood cell deformability and aggregation during shear stress. miRNA counts were subjected to variance stabilizing transformation (VST, DESeq2) and the change over time was calculated on these transformed counts (2 h – baseline; 7-days – baseline). For the EV markers, change-over-time was calculated on the normalized Log₂-tranformed intensities. Changes-over-time for the biochemical measurements were performed on the non-transformed values. A Pearson correlation analysis was then performed using the PerformanceAnalytics package in R (“R & R” from the Statistics Department of the University of Auckland, Auckland City, New Zealand) [23].

Results

In the ENOS study, 12 RIC patients and 14 Sham remained after follow-up, moreover all patients receiving thrombectomy (n = 3) were assigned to the same treatment group and excluded from this study. In the RIC group, the median (IQR) age was 67 years (61, 73) compared to 60 years (59, 68) in the Sham group, and 11% were female in RIC compared to 22% in the Sham group. The change in NIHSS assessments from baseline to 7-day follow-up were 0 (-1, 0) in RIC group, compared to − 1 (-1, 0) in patient randomized to Sham treatment. There were no significant differences in clinical scores (NIHSS, mRS), demographics or comorbidities between the two groups (Table 2). In addition, the table includes comorbidities, baseline medication use, stroke etiology and baseline biochemistry data.

Table 2.

Clinical characteristics of the population in the ENOS study, no differences in patient characteristics were found in the two groups

Factor	Sham	RIC
N	9	9
Age, median (IQR)	60 (59, 68)	67 (61, 73)
Female, n (%)	2 (22%)	1 (11%)
Prestroke mRS, median (IQR)	0 (0, 0)	0 (0, 0)
Smoking (current/prior), n (%)	4 (44%)	5 (56%)
Alcohol use, n (%)	1 (11%)	0 (0%)
Hypertension, n (%)	3 (33%)	4 (44%)
Hyperlipidimia, n (%)	2 (22%)	3 (33%)
Atrial fibrillation, n (%)	2 (22%)	2 (22%)
Former TCI, n (%)	1 (11%)	1 (11%)
Former AMI, n (%)	0 (0%)	1 (11%)
Angina pectoris, n (%)	1 (11%)	1 (11%)
NIHSS baseline, median (IQR)	1 (1, 2)	1 (1, 1)
NIHSS progression - baseline to 7 days (IQR)	-1 (-1,0)	0 (-1,0)
MoCA, median (IQR)	26 (25, 27)	25 (23, 26)
PASE. Median (IQR)	106.3 (76.4, 133)	145.1 (122, 191)
Baseline medication use (%)
Antiplatelet therapy, n, (%)	0 (0%)	1 (11%)
Marevan, n (%)	0 (0%)	0 (0%)
NOAK, n (%)	1 (11%)	0 (0%)
Antihypertension therapy, n (%)	2 (22%)	2 (22%)
Statins, n (%)	0 (0%)	2 (22%)
Opioid, n (%)	1 (1%)	0 (0%)
SSRI, n (%)	0 (0%)	0 (0%)
Stroke etiology (%)
Large artery atherosclerosis, n (%)	1 (11.1%)	0 (0.0%)
Small vessel disease, n (%)	2 (22.2%)	1 (11.1%)
Cardioembolic, n (%)	5 (55.6%)	2 (22.2%)
Multiple / unknown / rare, n (%)	1 (11.1%)	6 (66.7%)
tPa, n (%)	1 (11%)	2 (22%)
EVT, n (%)	0 (0%)	0 (0%)
Baseline biochemistry (IQR)
Total cholesterol, median (IQR)	5.6 (4.8, 6.4)	5.1 (4.3, 6)
LDL cholesterol, median (IQR)	3 (2.9, 4.1)	3.1 (2.4, 3.9)
HB1Ac, median (IQR)	41 (37.5, 42.5)	36 (34.5, 38.5)
eGFR, median (IQR)	90 (90, 90)	90 (82, 90)
CRP, median (IQR)	4 (4, 4)	4 (4, 4.3)

Prestroke mRS = modified Rankin scale, alcohol use = above 7 (women) or 15 (men) units per week, IQR = interquartile range, TCI = transient ischemic attack, AMI = acute myocardial infarct, NIHSS = National Institute of health stroke scale, MoCA = the Montreal cognitive Assessment, PASE = Physical activity scale for the Elderly, antiplatelet therapy (Acetylsalisylic acid, Clopidogrel, Persantin), NOAK = Non-vitamin K orale Antikoagulantia, SSRI = Selective serotonin reuptake Inhibitor, tPa = tissue type plasminogen activator, EVT = endovascular thrombectomy, LDL = Low-density lipoprotein, eGFR = estimated glomerular filtration rate, CRP = C-reactive protein

MiRNAs changes after RIC in stroke patients

Prior to nanoString analysis, the concentration of the purified miRNA was quantified (median 1540 ng/mL, IQR: 1072, 1915) to verify the presence of miRNAs in the volume needed for the analysis. In addition, top100 normalization of the miRNA data was introduced to diminish the influence of different input RNA. Five circulating miRNAs (miR-374a-5p, miR-20a-5p + miR-20b-5p, miR-19b-3p, miR-24-3p, miR-30d-5p) were differentially expressed within both the RIC-treated patients and compared to the Sham group (Fig. 1 and Supplementary Table 1, Additional File 1). Of the five miRNAs, two miRNAs were differentially expressed 2 h after first treatment, while the other three were differentially expressed at seven days follow-up (Fig. 1). hsa-miR-20a-5p + hsa-miR-20b-5p was upregulated in RIC treated patients at both 2 h after RIC (p = 0.021) and at 7 days follow-up (p = 0.003) (Fig. 1b). The nanoString hybridization technology used in this study do not distinguish between hsa-miR-20a-5p and hsa-miR-20b-5p. In addition, miR-374a-5p showed a large upregulation 2 h after RIC, which returned to baseline after seven days (Fig. 1e). In contrast, the remaining three miRNAs (miR-19b-3p, miR-24-3p, and miR-30d-5p) were only differentially expressed after 7 days of twice daily RIC treatment (Fig. 1a + c + d). When comparing the response in miR-374a-5p expression from baseline to 2 h after first RIC treatment (Fig. 1e), most of the individuals expressed explicit upregulation of the miRNA whilst the remaining shows sign of stagnating expression.

Fig. 1.

Circulating miRNA changes after RIC treatment: The miRNA changes are depicted in plots showing the changes over time in each individual patient as a ratio relative to baseline expression. The box plots illustrate the miRNA expression at baseline (0 h), two hours post-initial RIC (2 h), and seven days after treatment initiation (7d) for both the RIC group (red) and the Sham group (grey). Significant p-values are indicated for changes within the RIC group as well as comparisons between the RIC and Sham groups. (a) hsa-miR-19b-3p, (b) hsa-miR-20a-5p + hsa-miR-20b-5p, (c) hsa-miR-24-3p, (d) hsa-miR-30d-5p, and (e) hsa-miR-374a-5p show regulation patterns in response to RIC treatment

Surface protein abundance on circulating EVs

To analyze EV surface proteins associated with RIC, we utilized a customized microarray platform known as the “EV Array”, a priori designed with 54 targeted stroke/brain markers (Table 1). This allowed for the precise capture and measurement of EV-attached surface proteins from the patient samples. We observed a significant upregulation of EV-bound CD62 in RIC-treated patients compared to the Sham group (Fig. 2a and Supplementary Table 1, Additional File 1). Notably, this increase was significant both from baseline to 2 h after the first RIC treatment, and at seven-day follow-up. These changes were consistent compared to the Sham group and within the RIC group. Additionally, we noted a similar pattern of expression change in the EV surface protein MCP-1 over time in both groups (Fig. 2b) suggesting that certain EV surface proteins, like MCP-1, might reflect general stroke-related biological processes rather than being specific to RIC-induced changes.

Fig. 2.

EV surface marker changes after RIC treatment: The expression of the EV-markers CD62 and MCP-1 is plotted showing the Log₂ fold changes compared to the baseline sample of each patient. The p-value is shown for both the significant changes within RIC group (red) and compared with the miRNA expression changes over time in Sham group (grey). (a) CD62 shows an upregulation in RIC treated patients compared to Sham. (b) MCP-1 showed no significant changes in between the different timepoints

Correlation of miRNAs and CD62 expression to neurological scores and red blood cell aggregation and shear stress

All patients were assessed using several clinical scores (NIHSS, MoCA and PASE-score) both at baseline and seven days follow-up. However, neither miRNA nor CD62 showed any correlation to the clinical stroke scores (See Supplementary Fig. 1, Additional File 2) [19].

In AIS, red blood cells (RBCs) face increased shear stress and undergo changes that affect their aggregation and deformability, impacting their passage through small cerebral vessels [19]. As part of the ENOS study, we investigated whether RIC influenced these RBC parameters by correlating changes in the differentially expressed miRNAs and EV surface markers with these RBC properties (Fig. 3 and Supplementary Table 1, Additional File 1) [19]. Our findings revealed a significant correlation between the expression changes of hsa-miR-19b-3p and hsa-miR-374-5p and the RBC aggregation index over time. Interestingly, hsa-miR-19b-3p exhibited a negative correlation with shear stress changes, indicating that its upregulation reduced the shear stress needed to disrupt RBC aggregation. Conversely, both hsa-miR-374-5p and CD62 levels positively correlated with the time required for RBCs to reach half of their aggregation potential. Additionally, hsa-miR-374-5p was negatively correlated with the extent of RBC aggregation over a 10-second period. ‘.

Fig. 3.

Correlation analysis of significant miRNA measurements (hsa-miR-19b-3p, hsa-miR-30d-5p, hsa-miR-374a-5p, hsa-miR-20a/20b-5p, hsa-miR-24-3p) and CD62/P-selectin. The histograms on the diagonal show the distribution of each variable. The lower triangular panels depict scatter plots with fitted lines, illustrating the linear relationships between pairs of variables. The upper triangular panels display the correlation coefficients, with significance levels indicated by asterisks (*p < 0.05, **p < 0.01, ***p < 0.001)

Discussion

The effect of RIC treatment is still under investigation; however, the need for early intervention during AIS and the minimally invasive and pliable nature of RIC, support continuous research to identify possible circulating biomarkers of RIC that could assist in assessing RIC effect. The major findings in this study are (1) increased expression of miR-374a-5p, miR-20a-5p + miR-20b-5p, miR-19b-3p, miR-24-3p, and miR-30d-5p in RIC treated patients, of these (2) two miRNAs were correlated to RBC parameters; aggregation and shear stress and finally, (3) increased plasma levels of EV-bound CD62 in RIC treated compared to Sham. This suggests that responses to RIC treatment can be reflected in miRNA and EV expression even in this cohort of mild stroke patients (median NIHSS 1) and may be a target in future clinical RIC trials. This study is single centered and consisted of a homogeneous group of patients with a high-performance level and low number of comorbidities. By comparing the individual changes from baseline to follow-up in miRNA and EV response between the two intervention groups the study sought to minimize factors that are known to affect the biomarker measurements such as baseline medication use, baseline biochemistry and risk factors including cardiovascular comorbidities, alcohol and tobacco use.

The five differentially expressed miRNAs in the RIC group compared to Sham all regulate different pathways both individually and collectively with other miRNAs. In this study, we found a significant increase in RIC treated AIS patients of miR-374a-5p. This miRNA has already been suggested to be involved in neuroprotection and myocardial protection through different miRNA-mRNA validated pathways including MAPK6 [27], PTEN/PI3K [28], and possibly through downregulation of proinflammatory cytokines [29]. In addition, several of the RIC affected miRNAs target the inflammatory pathway, which is involved in the pathophysiology of stroke. Both miR19b-3p and miR-20a-5p are part of the miR17-92 cluster, which is involved in inflammation regulation [30], while upregulation of miR-20b-5p suppressed TXNIP thereby inhibiting inflammasome formation [31]. At a slightly different level but targeting the same pathways, miR-30d-5p has been shown to be involved in M2 microglia polarization/inflammation [32]. Mir-24-3p targets Notch signaling, which has been seen to be of benefit in stroke as reducing Notch signaling attenuates neuroinflammation [33] and decreases apoptosis in rat stroke model [34–36].

Combining the results of the ENOS study [19] with our measurements of circulating miRNAs and EV surface markers in the same patients, might reveal molecules associated with increased RBC aggregation and reduced deformability following stroke in these patients. The ENOS study did not find a correlation between RIC and RBC changes, however correlation analysis of miRNA, EV changes and RBC properties (RBC deformability and shear stress) showed an interesting connection. Changes in RBC characteristics have also been reported in e.g. sepsis patients, whose RBCs have lower deformability compared with healthy controls [37]. As in our study, miRNA changes were suggested to be associated with lower RBC deformability after inducing sepsis in a murine sepsis model compared to Sham animals [38]. Interestingly, hsa-miR-374a-5p, hsa-miR-19b-3p, and hsa-miR-30d-5p found in our study were negatively correlated to RBC aggregation, and additionally, with aggregation at different levels of shear stress (hsa-miR-19b-3p) (Fig. 3). All of these miRNAs are involved in pathways regulating inflammation, which is known to affect RBC aggregation negatively [39]. Reducing inflammation and thereby RBC aggregation should result in less viscous blood and increase cerebral blood flow in the microvasculature, which could be beneficial in stroke patients [40]. However, no direct interaction between RIC induced miRNAs and reduced inflammation was examined in this study, therefore this remains speculative.

In this study, plasma CD62 was significantly upregulated on EVs from baseline to two hours after first RIC treatment. However, the CD62 antibody used in EV Array cannot distinguish between P-selectin and E-selectin but E-selectin was measured individually and did not significantly differ in the two treatment groups. Nonetheless it is unclear whether the upregulation is collectively or a result of individual upregulation of one of the proteins. Both E- and P-selectin are expressed on the surface of endothelial cells whilst P-selectin is also expressed on platelets [41, 42]. On activated platelets, P-selectin binds to leukocytes, which alters leukocyte recruitment and activation patterns contributing to vascular repair and microcirculatory disturbances in ischemic tissue [41] as well as increased inflammation [43]. In similarity. E-selectin mediates leukocyte rolling on the endothelium recruiting neutrophils, monocytes and activated T-cells [42]. Studies have shown that E- and P-selectin are increased in AIS due to the aggregation of platelets and the recruitment of leukocytes during inflammation in the brain [41]. It can be speculated that RIC induced upregulation of CD62 on EVs can act as sponges for ligands binding P-selectin thereby diminishing the P-selectin activated platelet induced inflammation during stroke. To elucidate the possible positive role of CD62 positive EVs, it would be important to study the binding potential of these EVs to leukocytes as well as characterizing the presence of CD62 on activated platelets before and after RIC in patients with AIS. Following these studies it would be interesting to test the RIC induced CD62 positive EVs in preclinical stroke models where miRNAs could be introduced either as miRNA mimics (increasing the miRNA) or as antagomirs (reducing the miRNA). Interestingly, even though there is no direct regulation by the RIC regulated miRNAs in this study and CD62, they all have a downstream effect on inflammation and immune cell adhesion.

The specific development of MCP-1 expression over time in our study, combined with previous studies, indicate that MCP-1 could be a candidate for monitoring stroke pathogenesis or patients at risk of developing stroke [44, 45]. MCP-1 showed a clear reduction over time in both groups (Fig. 2a) suggesting that inflammation is reduced in the patients during the 7-day follow-up as MCP-1 is involved in the regulation of migration and infiltration of monocytes and macrophages during inflammation [46]. Georgakis et al., 2019 reported that elevated levels of MCP-1 were associated with higher risk of stroke, in particular with large-artery stroke and cardioembolic stroke [44]. The similar expression of MCP-1 in the AIS patients with minor stroke included in this study suggests that the fluctuation of the expression could serve as a marker for stroke.

Several limitations should be considered when interpreting the results of the study. The modest sample size increases uncertainty and the risk of Type I/II error. Given the exploratory pilot design (n = 9/arm), we did not compute post-hoc/observed power because it is redundant with p-values and may be misleading in small samples; results are interpreted as hypothesis-generating. Larger, prospectively powered cohorts will be required to confirm these miRNA and EV findings. Unfortunately, the study included three patients receiving thrombectomy treatment, which were randomly assigned to the same RIC-treatment group. This randomization could impact the results introducing confounding bias to the results, which explains the exclusion of these patients in the study. In this study, we only included patients who were able to perform the RIC/Sham treatments on their own and who could return for 7-day out-patient visit. The median stroke severity was very low (NIHSS 1) which may reduce the signal-to-noise ratio since the size of the ischemic injury, on average, was small. No significant correlation was found between the miRNA and EV markers and the clinical scores (NIHSS, PASE and MoCA), which limits the clinical interpretation of this study. However, the study seeks to investigate the molecular response of miRNA and EV after RIC using explorative methods to identify possible biomarkers for future targets or biomarkers with clinical properties. In addition, the medical history and medication use of the included patients (Table 2) did not show substantial differences between the two groups and in general only a minority of the included patients suffered from comorbidities and had a limited prior medication use. The results of this study will need to be reproduced in a larger heterogeneous study population, including patients both eligible and ineligible for reperfusion therapy in order to avoid selection bias. However, this explorative study design can pave the way for more targeted studies to verify the results. The differences in the miRNA and EV response within each patient could reflect the heterogeneity of the patients or indicate that some patients are responders or non-responders of RIC. Due to the size of our study population, a stratified analysis on stroke etiology and changes in biomarkers is not possible and larger future studies are needed to elucidate this.

The multiplexed, highly sensitive and high throughput platform of EV Array was used for this study. No EV standard which expresses all types of molecules is available in order to validate the specificity of each antibody, therefore the manufacturers documentation of specificity was trusted, however this poses a limitation in the study. To assure detection of the broadest possible EV populations, it was decided to use detection antibodies against CD9, CD63 and CD81, concurrently. All three antigens were targeted using a cocktail of antibodies to maximize the detection signals. These antigens were chosen because they are known to be present on EVs [47]. The ability for the EV Array to capture EVs in a quantitative manner have previously been proven in earlier studies, where detailed molecular analyses of the EVs were performed by nanotracking analysis [24], transmission electron microscopy [48, 49], and western blotting [50]. The scope of this study was to investigate the diagnostic and prognostic potential of EV phenotypes in stroke patients based on a blood sample. As only limited sampling material was available it was chosen to use an already established and verified technology, which is why it was chosen not to focus on the EV characteristics despite the recommendations by the MISEV guidelines [51].

Conclusion

In conclusion, we found several RIC-modulated miRNAs involved in neuroprotection and inflammation. Combined with the EV surface marker CD62, these changes could serve as potential effectors and indicators of successful RIC treatment in AIS. We were able to correlate miRNA changes to beneficial red blood cell aggregation and shear stress which represents an interesting avenue to be explored in future studies. Given the previously reported neuroprotective properties of miR-374a-5p, this miRNA would be very interesting to study in more detail in future studies of stroke treatments.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

AIS

Acute Ischemic Stroke

BBB

Blood Brain Barrier

CD

Cluster of differentiation

EV

Extracellular Vesicles

EVT

Endovascular thrombectomy

LOD

Limit of Detection

MiRNA

MicroRNA

MISEV

Minimal information for studies of extracellular vesicles

MoCA

Montreal Cognitive Assessment

NIHSS

National Institute of Health Stroke Scale

PASE

Physical Activity Score for the Elderly

QC

Quality control

RBC

Red blood cells

PBS

Phosphate-Buffered Saline

RIC

Remote Ischemic Conditioning

RT

Room temperature

TPA

Tissue-type plasminogen activator

Author contributions

Study conception and design: R.B.J., M.K., D.C.H., G.A., R.A.B., and K.R.D. Acquisition and analysis of data: R.B.J., J.J., L-A.C., R.B., E.K., B.E., and M.M.J. The first draft of the manuscript was written by R.B.J. The manuscript was reviewed by all authors, who also read and approved the final manuscript.

Funding

Financial support was obtained from the Harboe Foundation, the Novo Nordisk Foundation (NNF20OC00060998), and the Lundbeck Foundation via the Danish Neurological Society and Danish Society for Neuroscience.

Data availability

The data that support the findings of this study are not openly available due to data sensitivity and the European GDPR. The data are available from the corresponding author upon reasonable request. Data are located in encrypted data storage at Aarhus University.

Declarations

Ethics approval and consent to participate

The study was approved by the Central Denmark Region Committees on Health Research Ethics (ID no. 1-10-72-184-19) and the Danish Medicines Agency (jr. no. 2019081802, EUDAMED CIV-ID nr. 19-08-029484) and was registered in the Region’s Internal List of research projects (ID no. 1-16-02-333-19). Clinicaltrial.gov identifier: NCT04266639 (registration date: 2020-03-12). The study was monitored by the Good Clinical Practice unit Aalborg/Aarhus, Denmark. All patients signed an informed consent form before enrolling in the study. The study adheres to the CONSORT guidelines.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.