Human project was approved by the committee of Institutional Review Board (IRB) of Shanghai Jiao Tong University, Shanghai, China. Written informed consent was obtained from patients according to the Helsinki Declaration, and the Helsinki Declaration was followed during the human studies. Peripheral blood was collected from stroke patients who were recruited consecutively from September 2014 to June 2015 in Ruijin Hospital. Control group were enrolled in January 2015 from subjects who came for medical examination and denied previous history of stroke attack. Subjects in each group were matched with age and gender. Demography feature, related previous history including hypertension, diabetes mellitus (DM), hyperlipidemia, cardiopathy, associated laboratory test, and imaging information including blood glucose, blood lipid, electrocardiogram, cardiac ultrasonography, carotid artery ultrasonography, MR imaging, and MR angiography were also collected for analysis. The risk factors were defined as the following: (1) hypertension: blood pressure ≥140/90 mmHg; (2) DM: fasting glucose ≥7.0 mmol/L or 2-h postprandial blood glucose ≥11.1 mmol/L; (3) hyperlipidemia: triglycerides >1.7 mmol/L or total cholesterol >5.7 mmol/L or low-density lipoprotein >4.3 mmol/L. The exclusive criteria included recurrent stroke or stroke onset longer than 72 h, renal or liver failure, acute infectious disease, tumor, hematologic disease, and patients who are unable to cooperate with physical examination.

The severity of stroke was evaluated with NIHSS scores. The patients were classified into three groups: large artery atherosclerotic stroke (LA), cardioembolism (CE), and small artery stroke (SA) according to the Trial of Org 10172 in Acute Stroke Treatment (TOAST) classification. Patients were also classified into the following four groups: total anterior circulation infarct, partial anterior circulation infarct (PACI), posterior circulation infarct (POCI), and lacunar infarct (LACI) according to Oxfordshire Community Stroke Project (OCSP) definition. The infarct volume was calculated by ABC/2 method (A and B represent the largest diameter of the infarct and its largest perpendicular diameter, while C represents the thickness of the slices where the infarct lesion was visible). 3-month mRS was used to assess the neurologic outcome of stroke patients.

Whole blood was drawn into promoting coagulating tubes and centrifuged at 1,600 g for 10 min. Serum was aspirated and stored at −80°C. ExoQuick exosome precipitation solution (System Biosciences, CA, USA) was used to precipitate serum exosomes. Briefly, centrifuged the serum sample at 3,000 g for 15 min to remove cell debris; Then, added 125 μL ExoQuick exosome precipitation solution to 500 μL serum and mixed well; after refrigerating at 4°C for 30 min, centrifuged the mixture at 1,500 g for 30 min and removed the supernatant; Centrifuged another 5 min to spin down residual fluid and resuspended exosomes pellet by PBS or nuclease-free water. All centrifugations were done at 4°C.

The size and morphology feature of exosomes were examined by transmission electron microscopy (TEM) (TEM, Phillip CM120, 60 kV). Protein were extracted from exosomes resuspension solution with RIPA lysis buffer and separated on 8% polyacrylamide gel before transferring to a nitrocellulose filter membrane. The blotting membrane was blocked with skim milk and incubated with CD9 antibody (1:1,000 dilution, Abcam, Cambridge, UK)/CD63 antibody (1:200 dilution, Santa Cruz Biotechnology Inc., Santa Cruz, CA, USA)/TSG101 (1:500 dilution, Abcam, Cambridge, UK) overnight. After TBST washing, secondary HRP-conjugated antibody was added for 1 h (1:5,000 dilution, HuaAn Biotechnology Inc., Hangzhou, China). The proteins were finally detected using chemiluminescence (Thermo Scientific, Rockford, IL, USA).

Total RNA was isolated from exosomes using miRNeasy Mini Kit (Qiagen, Hilden, Germany) mainly according to the manufacturer’s instructions. The qiazol volume was 700 µl, and the final elute volume was 20 µl. Single-strand cDNA was synthesized using universal cDNA synthesis kit (Exiqon, Vedbaek, Denmark) following the condition: 42°C for 1 h and then 95°C for 5 min.

The expression of miRNA was tested by real-time PCR using SYBR Green master mix (Exiqon, Vedbaek, Denmark) on an Applied Biosystems 7900 real-time PCR machine (Life Technologies). The reaction conditions were as follows: 95°C for 10 min followed by 40 cycles of 95°C for 10 s and 60°C for 1 min. The relative expression of miR-223 was normalized to the endogenous control miR-16 expression using the comparative cycle threshold (CT) method. When comparing different groups, the data were logarithm-transformed.

Statistics were calculated using SPSS 19 (IBM SPSS Inc., NY, USA). The Kolmogorov–Smirnov test was used to determine the normality of continuous variables distribution. If normal, Student’s t-test was used to compare differences between groups, if abnormal, Welch correction was used. Proportions were compared using the chi-square test or Fisher’s test.

To test the feasibility of using miR-223 as a biomarker for stroke, we built statistical model by multivariable logistic regression analysis. Forward stepwise selection procedures were adopted. Results were expressed as adjusted odds ratios (OR) with the corresponding 95% confidence intervals (95% CI). Correlations were estimated by Pearson correlation test. Differences were considered significant at p < 0.05.

Transmission electron microscopy and western blot were used for identification of circulating exosomes. On the one hand, the yielded vesicles were consistent with exosomes in size and morphology—30–100 nm nano-size and round-shaped (Figure 1A). Some researchers showed TEM picture of cup-shaped exosomes (24). The shape difference is caused by different methods of preparing sample (25). On the other hand, western blot proved the expression of certain exosomal marker proteins, i.e., CD9, CD63, and Tsg101 (Figure 1B). CD9 and CD63 are the most commonly used marker proteins for exosomes identification. Tsg101 involves in transmembrane cargo inclusion and is important for exosomes biogenesis (26).

The clinical characteristics of ischemic stroke patients (n = 50) and control subjects (n = 33) are shown in Table 1. Mean age and the distribution of race/ethnicity were not different between two groups (p = 0.67, 1, 1). The number of male in both groups was about 1.5 times that of female (p = 0.75). A greater percentage of risk factors in stroke patients was found including DM 36.0% (p < 0.05) and cardiopathy 30.0% (p < 0.05). In our study, stroke patients and control group had the similar percentage of hypertension (p = 0.98). We considered that this phenomenon may be due to the increased awareness of health care in people with hypertension. They were more willing to participate in our research. The mean NIHSS score of stroke patients was relatively low (3.32). According to the TOAST classification, 50.0% of patients were classified as LA, while cardioembolism 16.0% and SA 34.0%. A smaller percentage of cardioembolism subtype is because that these patients were mostly enrolled to the Department of Cardiology in our hospital. According to the OCSP classification, 22 stroke patients were classified as PACI, while POCI subgroup was 16 cases and LACI subgroup was 11 cases. Patients with good outcome (mRS ≤ 2) took up 60% and those with poor outcome (mRS > 2) accounted for 40%.

In our study, miR-16 was used as internal control. Exosomal miR-16 level was almost the same between the two groups (Table 2, p = 0.828). When compared to control group, the expression of exosomal miR-223 in the serum of patients with acute ischemic stroke were increased greatly (Table 2; Figure 2A, p < 0.001). To explore the changing tendency of miR-223 after stroke attack, we analyzed circulating exosomal miR-223 levels at 24, 48, and 72 h. Exosomal miR-223 expression was not significantly different between subgroups, although we found it had a tendency to increase by time (Figure 2B).

To eliminate the unmatched risk factors impact (DM and cardiopathy) on exosomal miR-223 expression between two groups, we performed a logistic regression analysis. The result indicated that circulating exosomal miR-223 expression was a possible risk factor of stroke (Table 3, p < 0.001). In addition, when the value of exosomal miR-223 increased by 0.1, the risk of stroke occurrence increased by about 0.7-fold. The receiver operating characteristic curve was drawn to evaluate the diagnostic value of exosomal miR-223 (Figure 3). The area under the curve was 0.859, when circulating exosomal miR-223 = −0.50, had the best sensitivity (84.0%) and specificity (78.8%) for stroke diagnosis.

To explore the relationship between circulating exosomal miR-223 and stroke subtypes, we compared exosomal miR-223 expression in stroke patients with different TOAST subtypes (Figure 4A). There was no difference in exosomal miR-223 level among TOAST subtypes, for example, LA (−0.28 ± 0.27, n = 26), cardioembolism (−0.16 ± 0.19, n = 7), and SA (−0.29 ± 0.31, n = 17, p = 0.53). Patients who were classified to cardioembolism seemed to have a higher level of exosomal miR-223 but needed further verification with larger sample. To examine the association between circulating exosomal miR-223 and NIHSS score, the Pearson correlation test was performed. Positive correlation was found between the exosomal miR-223 expression and NIHSS score (Figure 4B, r = 0.31, p = 0.03). As to the relationship between circulating exosomal miR-223 and infarct volume, no significant correlation was found in our study (Figure 4C, r = 0.20, p = 0.19). To evaluate the relationship between circulating exosomal miR-223 and 3-month prognosis, we compared the exosomal miR-223 level in stroke patients with good outcomes (mRS ≤ 2) and those with poor outcomes (mRS > 2). Stroke patients who had poor outcomes were shown to have a greater expression of exosomal miR-223 than patients who had good outcomes (Figure 4D, −0.1 ± 0.3 vs. −0.3 ± 0.3, p = 0.036).