Circulating microRNA profiles based on direct S‐Poly(T)Plus assay for detection of coronary heart disease

Abstract Coronary heart disease (CHD) is one of the leading causes of heart‐associated deaths worldwide. Conventional diagnostic techniques are ineffective and insufficient to diagnose CHD with higher accuracy. To use the circulating microRNAs (miRNAs) as non‐invasive, specific and sensitive biomarkers for diagnosing of CHD, 203 patients with CHD and 144 age‐matched controls (126 high‐risk controls and 18 healthy volunteers) were enrolled in this study. The direct S‐Poly(T)Plus method was used to identify novel miRNAs expression profile of CHD patients and to evaluate their clinical diagnostic value. This method is an RNA extraction‐free and robust quantification method, which simplifies procedures, reduces variations, in particular increases the accuracy. Twelve differentially expressed miRNAs between CHD patients and high‐risk controls were selected, and their performances were evaluated in validation set‐1 with 96 plasma samples. Finally, six (miR‐15b‐5p, miR‐29c‐3p, miR‐199a‐3p, miR‐320e, miR‐361‐5p and miR‐378b) of these 12 miRNAs were verified in validation set‐2 with a sensitivity of 92.8% and a specificity of 89.5%, and the AUC was 0.971 (95% confidence interval, 0.948‐0.993, P < .001) in a large cohort for CHD patients diagnosis. Plasma fractionation indicated that only a small amount of miRNAs were assembled into EVs. Direct S‐Poly(T)Plus method could be used for disease diagnosis and 12 unique miRNAs could be used for diagnosis of CHD.


| INTRODUC TI ON
Coronary heart disease (CHD) is a life-threatening disease and remains a leading cause of heart-associated deaths for adults worldwide, which develops over time due to genetic and environment factors with complex pathology. 1 Early diagnosis, effective prevention and therapy for CHD pose a major challenge to the entire medical community. 2 Without the help of well-established invasive coronary angiogram (CAG) techniques, CHD is hard to diagnose.
CAG is difficult to perform if multiple vessels are affected or the artery is narrowed at multiple locations. On the other hand, CAG may not be effective against very hard atherosclerotic plaques. 3 In recent years, application of non-invasive molecular biomarkers is emerging as a powerful approach to diagnosis and prediction of CHD, and cir- With the hypothesis that muscle-or heart-specific miRNAs are released into circulation from the injured heart, 13 circulating miR-NAs demonstrate significant dynamic change in human serum and plasma. To date, non-invasive and blood-based studies have examined miRNA expression profiles to identify novel miRNA biomarkers for CHD diagnosis. Although these findings suggest that some circulating miRNAs might be potential diagnosis markers, most of the results are based on a limited number of patients and few specific miRNAs. 14,15 And almost no related studies could be used as an auxiliary technique for CHD prevention, prediction, diagnosis and the effectiveness of therapies, due to their time-consuming operation and imprecise quantification.
In order to promote clinical application of circulating miRNAs as biomarkers, an accurate, convenient and inexpensive profiling approach is needed. By combining S-Poly(T)Plus method and extraction-free miRNAs isolation technique, we precisely quantify miRNAs expression in 1 hour, making this method effective in monitoring CHD progression. More importantly, longitudinal measurements of miRNAs in CHD patients may provide further insight into individual temporal patterns and the patient's ensuing risk for disease progression and adverse outcome. 16 Here, using plasmas obtained from CHD patients and high-risk individuals, we unearthed a group of miRNAs that can serve as non-invasive biomarkers for the diagnosis of CHD, and we evaluated their performances. This quick quantification method has a great potential to be used in clinical investigation.

| Samples collection and processing
A total of 5 mL of venous blood was obtained into ethylenediaminetetraacetic acid (EDTA)-containing tubes (BD, USA) from donors after overnight fasting. Samples were centrifuged at 1600 g for 10 minutes at 4°C to remove blood cells, followed by centrifugation at 16 000 g for 10 minutes at 4°C to completely remove cell debris. 18 To guarantee the quality of samples, the haemolytic plasma which appeared pale red or pink was excluded from consideration. Plasma was collected and stored in aliquots into RNase/DNase-free tubes at −80°C until analysis.

| miRNAs profiling
To identify potential biomarkers, we profiled miRNAs from pooled plasma and individual plasma. The whole study flow chat is shown in Figure 1. Firstly, we prepared three pools of 18 CHD patients and three pools of 18 high-risk controls. Quantitative global profiling of plasma miRNAs was performed using the direct S-Poly(T)Plus approach 19 to screen from each pool (Files S1 and S2), and comparing the level of each miRNA in CHD and high-risk groups. Secondly, TA B L E 1 Demographical and clinical features of coronary heart disease (CHD) patients and high-risk controls in the discovery set and training set  Ultimately, selected miRNAs were evaluated in plasma from CHD patients and healthy volunteers.

| Extraction-free miRNAs isolation and quick quantification
Plasma for miRNAs detection was treated based on our optimized direct extraction method. Briefly, 20 μL thawed plasma was mixed thoroughly with 20 μL 2 × lysis buffer and 1 μL protease K, followed by incubation for 20 minutes at 50°C, 5 minutes at 95°C to denature protease K completely. The jelly products were centrifuged at 14 000 g for 5 minutes at 4°C to remove precipitants. The supernatant was preceded immediately for RT reaction.
Quantification was performed through S-Poly(T)Plus method as described before. The level of miRNAs was calculated using 2 −ΔCt and normalized to global mean Ct value. Exogenous spike-in cel-miR-54 was measured to evaluate the stability and to normalize candidate miNRAs. All sequences of miRNAs in this study were downloaded from miRBase 22. 20 TaqMan probe and miRNA-specific primer sequences (File S1) were designed in the laboratory and synthesized by IDT (Integrated DNA Technologies) and GENEWIZ.
Candidate miRNAs were further validated by Sanger Sequencing.

| Extracellular vesicles isolation, verification and miRNAs expression profiling
Extracellular vesicles (EVs) were isolated from plasma with differential centrifugation/ultracentrifugation ( Figure 5A). About 1 mL plasma was diluted to 20 mL with ice-cold PBS. The diluted plasma was centrifuged at 300 g for 10 minutes at 4°C to remove TA B L E 3 Demographical and clinical features of coronary heart disease (CHD) patients and healthy volunteers in evaluation set  to isolate miRNAs. 19 microRNAs were analysed using RT-qPCR as described above. Non-parametric Mann-Whitney tests were used to compare miRNA levels between the CHD groups and high-risk groups in discovery set. Student's t test was used to compare the differences in other variables between the two groups. P < .05 was considered statistically significant.

| Baseline clinical characteristics of the study population
We recruited 347 participants including 203 CHD patients, 126 high-risk controls and 18 healthy volunteers. All the CHD patients were selected on the basis of clinical parameters (eg chest pain and palpitation, history and laboratory value) combined with angiographic documentation (Figure S1). High-risk controls were recruited from a large pool of individuals seeking a routine chest examination without obvious cardiovascular obstruction.

| Identification of candidate miRNAs in discovery set
To identify novel miRNAs biomarkers for CHD diagnosis, we collected plasma from CHD patients and high-risk controls. Firstly, we performed S-Poly(T)Plus analysis to screen candidate miRNAs that showed obvious alteration in three paired plasma samples between CHD patients and high-risk controls ( Figure 1A,B). These 18 high-risk controls which were selected in discovery set were reused in discovery, training and validation steps. As is shown in Table 1, there were no significant differences in the distribution of smoking, alcohol consumption, age and sex between these two groups. We compared the miRNA quick quantification method with conventional TRIzol isolation method, and the results indicated that quick quantification method could accurately and sensitively quantify circulating  Figure S2D; File S2).
miRNA quantitative analyses showed that the levels of these miRNAs were significantly increased in CHD patients ( Figure 4A). To further explore the potential use of altered miRNAs as novel biomarkers for CHD, we built ROC (Receiver Operating Characteristic)curves and calculated the AUC (Area Under Curve)for these biomarkers, which ranged from 0.580 to 0.767, respectively ( Figure 4B). To estimate the classification performance of the 12-miRNAs-based biomarker, we calculated the diagnostic sensitivity and specificity of this panel for CHD detection, which were 97.1% and 87.5%, respectively. Furthermore, the ROC curve for this panel revealed a pronounced diagnostic accuracy, evidenced by the AUC of 0.971 (P < .001), which was much better than that of 12 individual miRNAs ( Figure 4B). These data suggested that these 12 circulating miRNAs might be a group of appropriate biomarkers for discriminating CHD patients from high-risk controls.

| Evaluation of miRNAs as sensitive and potential predictors for CHD in validation set-2
After getting confirmation of twelve circulating miRNAs as novel biomarkers for CHD, we were sufficiently interested in investigating sensitivity and specificity of candidate miRNAs for CHD prediction.
Moreover, we investigated the six miRNAs and their different combination panels in CHD cases and controls from validation set-2. The individual miR-320e, miR-378b and miR-15b-5p could reliably discriminate CHD from controls with each AUC of 0.811 (95% con- 0.581 (95% CI 0.418-0.814) ( Figure S4). Next, we combined the statistically significant miRNAs together as new biomarker which showed a better performance compared with individual miRNA ( Figure 5B). The performance of the six miRNA combined panel for CHD detection in validation set-2 was 92.9% and 89.5%, which indicated that this panel was really a comprehensive and specific indicator. We further evaluated the performance of these candidates in plasma, most of whose miRNAs alone could perfectly distinguish healthy volunteers from CHD cases, except miR-26a-5p with its AUC of 0.717 (95% CI 0.680-0.990) ( Figure S5). At the same time, a formula was estimated to predict the probability of having CHD based on the relative expression level of these candidates compared to spike-in cel-54 by performing the binary logistic regression analysis in SPSS. The relationship between the risk of having CHD and the relative expression of predictors in details is p = hsa-miR-15b-5p + hsa-miR-320e × 552 + hsa-miR-378b × 182.
Taken together, these novel findings suggest that these six circulating miRNAs, especially miR-15b-5p, miR-320e and miR-378b could be used as sensitive and independent predictors for CHD.

F I G U R E 4
Candidate miRNAs were validated using CHD patients and high-risk control individuals from validation set-1. A, Plasma from 48 CHD patients and 48 high-risk controls were detected with candidate miRNAs; (B) ROC analysis of individual miRNAs and combined miRNAs as biomarker for CHD diagnosis. Data are shown as mean ± SEM. CHD, coronary heart disease;CK, high risk control; ns, not significant, **P < .01, ***P < .001 and **** P < 0.001. P values are shown above each miRNA F I G U R E 5 Candidate miRNAs were validated using CHD patients and high-risk control individuals from validation set-2. A, Plasma from 60 CHD patients and 95 high-risk controls were detected with 10 candidate miRNAs; (B) diagnostic value of the combined miRNAs in CHD patients from second cohort. CHD, coronary heart disease; ns, not significant, * P < .05, ** P < .01, *** P < 0.01 and **** P < 0.001

| The distribution of circulating miRNAs in plasma
To investigate the distribution of these circulating miRNAs in plasma, we isolated EVs from plasma obtained from CHD patients and control individuals ( Figure 6A). Exosome vesicle was confirmed by specific protein marker TSG101, CD63, CD9 and CD81, and meanwhile big/middle vesicles were detected by calnexin ( Figure 6B). We analysed several miRNAs contents from different fractions of plasma, and our findings demonstrated that more than 80% miRNAs existed in the supernatant. About 15% miRNAs were assembled into big/ middle vesicles and less than 5% miRNAs were packaged into exosome vesicles ( Figure 6C,D). These observations indicate that free argonaut-miRNA complex may be the main form of these circulating miRNAs existing in plasma, and only a small number of miRNAs are assembled into EVs.

| Correlation of plasma circulating miRNA with angiographical and clinical factors
To determine whether the expression levels of these miRNA biomarkers are associated with clinical features of CHD patients, we estimated the correlation coefficient between miRNAs and angiographical/clinical factors ( Figure 7A; Files S3 and S4). Our data revealed that the expression levels of miR-26a-5p and miR-320e were significantly correlated and LCX and RCA narrowing were significant (P < .001) ( Figure S6).
These findings reinforce that circulating miR-320e and miR-26a-5p may act as novel biomarkers for CHD diagnosis. Based on next-generation sequencing and S-poly(T) results, 19,24 we selected 343 mature miRNAs in plasma. By using the S-Poly(T)

| D ISCUSS I ON
Plus method ( Figure 1B), we rapidly and accurately screened genome-wide miRNAs in plasma from CHD patients and control individuals ( Figure S2). The nominal EDTA concentration in blood samples is much lower than the concentration of MgCl2 in RT-PCR and PCR reaction, so it has a slight effect on these reactions. Furthermore, previous study shows EDTA is a better anticoagulant than heparin and citrate for plasma preparation. 25 In the present study, we ultimately selected a group of miRNAs as a first pass to introduce a specific and non-invasive diagnostic tool for CHD. We propose that the expression pattern of all these miRNAs may make it possible to differentiate between high-risk cases and CHD cases. Compared to high-risk CK group, miR-133b and miR-1-3p have lower expression in CHD patients while miR-499 and miR-208 have higher expression ( Figure 3). All four miRNAs are muscle-enriched, although miR-499 and miR-208 are usually expressed at extremely low levels except in cases of substantial (cardiac) muscle damage. 14 Figure 5B). Furthermore, when we detected these miRNAs in healthy volunteers and CHD cases, these candidates adequately distinguished different types of plasma ( Figure S5). Among the various miRNAs investigated in our study, miR-15b-5p, miR-155-5p, miR-149-5p, miR-199a and miR-378b [32][33][34] have been reported to be correlated with CHD. Most importantly, for the first time our study showed that the increase of the expression level of miR-361-5p, miR-29c-3p and miR-320e has a high correlation with CHD.
Previously, miRNAs have been reported to be transported in body fluids within exosomes, and once released into extracellular fluid, exosomes fuse with other cells and transfer their cargo to acceptor cell. 35 Interestingly, our results showed that all the candidate miRNAs mainly existed outside of EVs ( Figure 6C,D), which was consistent with the results of quantitative analysis of miRNA content of exosomes. 24,36 Correlation analysis indicated that miRNAs (miR-26a-5p and miR-320e) could be better biomarkers for CHD diagnosis compared to most conventional clinical factors, such as apolipoprotein A (ApoA), apolipoprotein B (ApoB), LPA (Figure 7; Figure   S6). Consistent with results of previous studies, immune system was involved in CHD patients, 37 as leucocyte was correlated with RCA narrowing.

| Study limitations
Because patients with myocardial damage were excluded from our cohort, we cannot detect different expression patterns of miR-499 and miR-208 in the following analyses. The weak correlations between miRNAs expression levels and luminal narrowing may be because of the quantification strategy of narrowing coronary artery, as single plaque stenosis in one coronary artery is hard to be distinguished from a diffuse stenotic disease in multiple vessels. As some of the participators were taking drug treatment which may cause differential expression of multiple miRNAs, the noise and difficulty of data analysis were increased.

| CON CLUS ION
In conclusion, our study of plasma circulating miRNAs showed a unique and reliable pattern of non-invasive biomarkers that have the potential to be used for early diagnosis of CHD. The biological characteristics of CHD were better understood through the study, which was conducive to the exploration of new therapies for future clinical applications to improve therapeutic efficacy and pertinence of treatment.

CO N FLI C T O F I NTE R E S T
The authors declare that they have no conflict of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.