EBioMedicine 66 (2021) 103338 Contents lists available at ScienceDirect EBioMedicine journal homepage: www.elsevier.com/locate/ebiom Research paper Differential miRNAs in acute spontaneous coronary artery dissection: Pathophysiological insights from a potential biomarker Marta Lozano-Prietoa,#, David Adlamb,#, Marcos García-Guimaraesc,d, Ancor Sanz-Garcíae, Paula Vera-Tomea, Fernando Riveroc, Javier Cuestac, Teresa Bastantec, Anna A. Baranowska-Clarkeb, Alicia Varaa, Enrique Martin-Gayoa,f, nchez-Madrida,h,i, Miguel Vicente-Manzanaresg, Pilar Martính,i, Nilesh J Samanib, Francisco Sa b,i,z a,i,z, Fernando Alfonso , Hortensia de la Fuente * a Department of Immunology, Instituto de Investigacio n Sanitaria, Hospital Universitario de la Princesa, Universidad Auto noma de Madrid, Madrid, Spain b Department of Cardiovascular Sciences, University of Leicester and NIHR Leicester Biomedical Research Centre, Glenfield Hospital, Leicester, UK c Department of Cardiology, Instituto de Investigacion Sanitaria, Hospital Universitario de la Princesa, Madrid, Spain d Hospital del Mar, Parc de Salut Mar, Barcelona, Spain e n Sanitaria, Hospital Universitario de la Princesa, Madrid, Spain Data Analysis Unit, Instituto de Investigacio f Facultad de Medicina, Universidad Auto noma de Madrid, Madrid, Spain g ncer USAL-CSIC, 37007, Salamanca, Spain Instituto de Biología Molecular y Celular del Ca h Vascular Pathophysiology Area, Centro Nacional de Investigaciones Cardiovasculares, i CIBER de Enfermedades Cardiovasculares, Spain A R T I C L E I N F O A B S T R A C T Article History: Background: Spontaneous Coronary Artery Dissection (SCAD) is an important cause of acute coronary syn- Received 29 October 2020 dromes, particularly in young to middle-aged women. Differentiating acute SCAD from coronary athero- Revised 24 March 2021 thrombosis remains a major clinical challenge. Accepted 25 March 2021 Methods: A case-control study was used to explore the usefulness of circulating miRNAs to discriminate Available online xxx both clinical entities. The profile of miRNAs was evaluated using an unbiased human RT-PCR platform and confirmed using individual primers. miRNAs were evaluated in plasma samples from acute SCAD Keywords: and atherothrombotic acute myocardial infarction (AT-AMI) from two independent cohorts; discovery Spontaneous coronary artery dissection microRNAs cohort (SCAD n = 15, AT-AMI n = 15), and validation cohort (SCAD n = 11, AT-AMI n = 41) with 9 Acute myocardial infarction healthy control subjects. Plasma levels of IL-8, TGFB1, TGBR1, Endothelin-1 and MMP2 were analysed Biomarker by ELISA assays. Findings: From 15 differentially expressed miRNAs detected in cohort 1, we confirmed in cohort 2 the differ- ential expression of 4 miRNAs: miR-let-7f-5p, miR-146a-5p, miR-151a-3p and miR-223-5p, whose expres- sion was higher in SCAD compared to AT-AMI. The combined expression of these 4 miRNAs showed the best predictive value to distinguish between both entities (AUC: 0.879, 95% CI 0.721.0) compared to individual miRNAs. Functional profiling of target genes identified an association with blood vessel biology, TGF-beta pathway and cytoskeletal traction force. ELISA assays showed high plasma levels of IL-8, TGFB1, TGFBR1, Endothelin-1 and MMP2 in SCAD patients compared to AT-AMI. Interpretation: We present a novel signature of plasma miRNAs in patients with SCAD. miR-let-7f-5p, miR- 146a-5p, miR-151a-3p and miR-223-5p discriminate SCAD from AT-AMI patients and also shed light on the pathological mechanisms underlying this condition. Funding: Spanish Ministry of Economy and Competitiveness (MINECO): Plan Nacional de Salud SAF2017-82886-R, Centro de Investigacio  n Biome dica en Red de Enfermedades Cardiovasculares (CIBERCV). Fundacio  n BBVA a equipos de Investigacio  n Científica 2018 and from Caixa Banking Founda- tion under the project code HR17-00016 to F.S.M. Instituto de Salud Carlos III (AES 2019): PI19/00565 to F.R, PI19/00545 to P.M. CAM (S2017/BMD-3671-INFLAMUNE-CM) from Comunidad de Madrid to FSM and PM. The UK SCAD study was supported by BeatSCAD, the British Heart Foundation (BHF) PG/ * Corresponding author. E-mail address: hortensia.fuente@salud.madrid.org (H. de la Fuente). # These authors contribute equally z Joint senior and corresponding authors https://doi.org/10.1016/j.ebiom.2021.103338 2352-3964/© 2021 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/) 2 M. Lozano-Prieto et al. / EBioMedicine 66 (2021) 103338 13/96/30608 the NIHR rare disease translational collaboration and the Leicester NIHR Biomedical Research Centre. © 2021 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/) miRNA candidates were confirmed by individual miRNA PCR primer sets in the validation cohort with the AT-AMI subset 2. The SCAD 1. Introduction group from the validation cohort was not divided further as the num- ber of patients was already low. Spontaneous coronary artery dissection (SCAD) is an important cause of acute coronary syndromes (ACS), occurring predominantly in young to middle-aged women. [1] It is characterized by the devel- 2.2. Patients and sample collection opment of an intramural coronary haematoma leading to a compres- sive false lumen that develops in the absence of an atherosclerotic, Patients presenting acutely with AMI (STEMI and NSTEMI) under- traumatic or iatrogenic aetiology. The resulting compromise of the going urgent coronary angiography were included. Diagnosis of true lumen leads to progressive myocardial ischemia and acute myo- STEMI and NSTEMI was performed according to recent guidelines cardial infarction (AMI). [2] Accurate diagnosis of SCAD is critical as based on clinical, electrocardiographic findings and data from cardiac management differs in key aspects from conventional atherothrom- biomarkers. [12] Patients with systemic inflammatory disease or botic ACS (AT-AMI). In particular, percutaneous coronary interven- immunosuppressive treatment were excluded. The diagnosis of SCAD tion (PCI) in SCAD is associated with poorer clinical and angiographic by angiography was made according to well-established criteria. results, and a higher rate of procedural complications; whilst sponta- [1,3] All angiograms were reviewed by at least 2 experienced angiog- neous healing of the dissected coronary segment occurs at follow-up raphers to determine the presence of SCAD versus an atherosclerotic in most SCAD patients treated conservatively. Therefore, contrary to culprit coronary lesion. The use of intracoronary imaging was recom- guideline-based management for AT-AMI, a conservative approach to mended when deemed safe and feasible in cases with suspected revascularisation where possible is the consensus optimal manage- SCAD but ambiguous angiographic findings. Patients with unclear ment for most SCAD patients. [15] culprit lesions or poor-quality angiograms were excluded. Left ven- At present, there are no known diagnostic biomarkers with speci- tricular function was assessed by transthoracic echocardiography ficity for acute SCAD and invasive angiography remains the first line performed systematically before discharge. All patients otherwise diagnostic tool despite a well-recognized increased risk of iatrogenic received routine clinical care according to current best practice. dissection. Although typical angiographic appearances are well [1,3,13,14] described, ambiguous cases are common and whilst intracoronary Plasma samples from discovery cohort (SCAD n = 15 patients, AT- imaging can be helpful, complications may arise even during invasive AMI n = 15) were from UK-SCAD. Samples from validation cohort imaging. [6] Therefore, the identification of non-invasive biomarkers (SCAD n = 11, AT-AMI n = 41) were from ESP-SCAD, as were plasma allowing an early and precise diagnosis is key to improving diagnostic samples from 9 control subjects. Samples were collected before hepa- accuracy, enhancing an appropriate management of the condition rin administration in EDTA tubes during the 06 days following the and ultimately constituting a novel form of non-invasive diagnosis. acute event. Blood was processed no later than 24 h from collection, MicroRNAs (miRNAs) are non-coding, short RNA molecules being kept at 4 °C until processing. (1925 nucleotides) that mediate post-transcriptional gene expres- sion regulation by inhibiting mRNA translation or promoting its deg- 2.3. Sample size radation. [7] miRNAs are an important form of intercellular communication [8] with relevant roles in physiological and patholog- Because the low prevalence of SCAD, the sample size in this study ical processes. Circulating miRNAs can be detected in bodily fluids, was not calculated using standard procedures. We included all sam- which makes them attractive targets as non-invasive diagnostic or ples available at the participating research centres. prognostic biomarkers for different diseases. [911] In this study we sought to determine the plasma miRNA profile of 2.4. Ethics patients with acute SCAD compared with samples of AT-AMI patients and to assess their usefulness to discriminate both clinical entities. The UK SCAD study (UK-SCAD), (ISRCTN42661582) and the Sam- Additionally, this profile aids gaining insight into the pathological ple and Data Collection for Cardiovascular Research Study were molecular mechanisms underlaying SCAD. approved by the UK National Research Ethics Service (14/EM/0056 and 09/H0406/114) and the UK Health Research Authority. The Hos- 2. Methods pital Universitario Princesa SCAD (ESP-SCAD) sample study was approved by the institutional ethics committee for clinical research 2.1. Study design (PI-815). These approvals enabled prospective recruitment of patients presenting with SCAD and AT-AMI and collection of clinical, To explore the usefulness of circulating microRNAs to discrimi- angiographic, imaging and biomarkers data. All patients gave written nate SCAD and AT-AMI patients, we analysed 2 independent cohorts informed consent and study procedures were conducted in accor- of patients (cohort 1, discovery cohort; and cohort 2, validation dance with the Declaration of Helsinki. cohort). The discovery cohort included n = 15 SCAD and n = 15 AT- AMI patients; the validation cohort included n = 11 SCAD and n = 41 2.5. RNA isolation, RT-PCR and analysis AT-AMI patients. AT-AMI patients from validation cohort were sepa- rated in 2 subsets, AT-AMI subset 1 (19 males/7 females) and AT-AMI Blood samples were centrifuged 20 min at 4 °C and 2000xg, ali- subset 2 (15 females). Differentially expressed miRNAs identified in quoted and stored at -80 °C for RNA isolation. All samples were tested the discovery cohort were analysed in the validation cohort with the for haemolysis at 414 nm absorbance in a NanoDrop One spectropho- AT-AMI subset 1. An unbiased human PCR panel that allows the eval- tometer (Thermo Scientific). RNA was extracted using miRNeasy uation of 179 miRNAs was used in this phase. Finally, validated Serum/Plasma Advanced Kit (Qiagen), following the manufacturer’s M. Lozano-Prieto et al. / EBioMedicine 66 (2021) 103338 3 above 36, Cq values were converted to relative quantities (RQ) and Research in context the normalization factor (NF) was calculated as the geometric mean Spontaneous Coronary Artery Dissection (SCAD), is an impor- of miR-17-5p, miR-20-5p, miR-30e-5p, miR-106a-5p and miR-let-7i- tant cause of acute myocardial infarction(AMI) particularly in 5. The Normalized Relative Quantities (NRQ) were obtained by divid- young to middle-age women but has a poorly understood path- ing the RQs by the sample specific NF. ophysiology. Accurate diagnosis of SCAD is critical as manage- ment differs in key ways from conventional atherothrombotic AMI. At present, there is no adequate non-invasive diagnostic 2.7. ELISA assays test for SCAD. Furthermore, invasive angiography, the first line diagnostic tool for SCAD, is known to be associated with an Plasma levels of C-X-C Motif Chemokine Ligand 8 (CXCL8 also increased risk of iatrogenic complications in this population. known as IL-8), Transforming Growth Factor beta1 (TGFB1), Trans- New non-invasive biomarkers are therefore needed to improve forming Growth Factor Beta Receptor 1 (TGFBR1), Endothelin-1 approaches to the clinical diagnosis of this condition. Micro- (EDN1) and Matrix Metalloproteinase-2 (MMP2) were analysed using RNAs are short RNA molecules with relevant roles in multiple a custom ELISA multiplex array (RayBiotech Life, GA, USA) according physiological processes. The presence of miRNAs in bodily flu- to manufacturer’s instructions. ids, makes them attractive targets as non-invasive diagnostic or prognostic disease biomarkers. 2.8. Statistical analysis Added value of this study When continuous data did not fulfil normality criteria, a log trans- We present the first evidence that the circulating miRNA signa- formation was applied; otherwise, a non-parametric test was used to ture in acute SCAD is distinct from atherosclerotic AMI. In two detect differences between samples. Univariate analyses (pairwise) cohorts from different national centres of excellence, we iden- and multiple test corrected by false discovery rate (FDR) (adjusted p tify 4 candidate miRNAs bearing promise as biomarkers for value <0.05) were used to identify those microRNAs differentially acute SCAD. We also report that molecules regulated by these expressed in AT-AMI and SCAD patients from both cohorts. FDR was miRNAs are also increased in plasma samples of SCAD patients. calculated by the Benjamini-Hochberg method. microRNAs differen- These findings provide key pathophysiological insigths into tially expressed in cohort 1 and cohort 2 were further analysed for SCAD and related arteriopathies. differential expression depending on age, sex, diabetes mellitus, smoking condition and the haemolysis levels of plasma using a multi- Implication of all the available evidence ple linear regression analysis. When miRNA levels were not affected The identification of non-invasive diagnostic biomarkers allow- by any of these factors, univariate (pairwise) analysis followed by a ing early, and precise diagnosis could prevent SCAD patients correction for multiple comparison was applied. Selected microRNAs from being inappropriately treated for an atherosclerotic myo- were also confirmed by individual RT-PCR in cohort 2. The results of cardial infarction, ensuring a tailored clinical approach specific the ROC show the confidence interval (CI) of 95% and the p-value of to the pathology. the hypothesis testing (H0: AUC = 0.5). Differences between AT-AMI and control subjects were analysed using a univariate test followed by a correction for multiple comparisons. Differences in cytokine plasma levels between SCAD and AT-AMI were analysed using two- tailed level of significance Mann-Whitney U test. Data were analysed instructions and eluted using 20 ml of RNAse-free water. RNA sam- using STATA (STATA 14.0 for MAC, StataCorp, College Stataion, Texas, ples were stored at 80 °C. Reverse transcription was performed USA) and R software (version 3.5.1 http://www.R-project.org (last with 2 ml of RNA using miRCURY LNA RT Kit (Qiagen) according to accessed 9-20-2020); the R Foundation for Statistical Computing, manufacturer’s instructions in a final reaction volume of 20 ml. cDNA Vienna, Austria) and graphs were performed with GraphPad Prism 8 samples were stored at 20 °C. RT-PCR assays were performed using (Prism 8 for macOS, GraphPad Software, LLC) and R software. For the ready-to-use miRCURY LNA miRNA Focus PCR Panels and miRCURY functional profiling of the microRNA gene targes, data were analysed LNA SYBR Green PCR Kit (Qiagen) following the manufacturer’s by Fisher exact test and Bonferroni correction, the complete Homo instructions in a CFX384 PCR detection system (Bio-Rad). sapiens genome was used as reference list. Only those biological pro- UniSp3 RNA spike-in Cq values were used as inter-plate calibra- cesses with a corrected p value <0.05 were considered statistically tors following the manufacturer’s instructions. Then, data were ana- significant. lysed using the global mean normalization method. [15] Briefly, after exclusion of values above 36, Cq values were converted to relative quantities (RQ) and sample specific normalization factor (NF) was cal- 2.9. miRNA target identification and their association with biological culated as the geometric mean of the RQs of all expressed targets per processes sample. Normalized Relative Quantities (NRQ) were obtained by dividing the RQs by the sample specific NF. miRTarBase database was used to identify validated miRNA tar- gets for those miRNAs differentially expressed between SCAD and 2.6. Confirmatory RT-PCR using individual primers AT-AMI (miR-let-7f-5p, miR-146a-5p, miR-151a-3p and miR-223- 5p). Then, the association of functional target genes with biological miRCURY LNA miRNA PCR Assay primer kits (Qiagen) for miR-let- processes was determined using the PANTHER Classification System 7f-5p, miR-146a-5p, miR-151a-3p and miR-223-5p were used to per- with terms from Gene Ontology. form the individual PCR assays following the manufacturer’s instruc- tions in a CFX384 PCR detection system (Bio-Rad). To select the most stable miRNAs suitable for the normalization, NormFinder algorithm 2.10. Role of the funding source [16] was applied to miRNA expression data obtained from the miRNA PCR panels. The most stable miRNAs were miR-17-5p, miR-20-5p, The funding source for this study had no role in this study (design, miR-30e-5p, miR-106a-5p and miR-let-7i-5. After exclusion of values data collection, data analysis, interpretation or writing). 4 M. Lozano-Prieto et al. / EBioMedicine 66 (2021) 103338 3. Results Regarding the use of statins, no differences were detected in the dis- covery cohort, while in the validation cohort a higher percentage of 3.1. Differences in circulating miRNAs profile among SCAD, AT-AMI AT-AMI patients were taking statins at the time of sampling (Table 1). patients and healthy controls The analysis of discovery cohort showed a differential expression of 15 miRNAs with an FDR corrected p value <0.05 (Table 2). These To validate our approach, we first compared plasma miRNA miRNAs were then analysed in the validation cohort (AT-AMI subset expression levels of AT-AMI patients from both cohorts (n = 41) and 1). Our data showed a statistically significant (FDR corrected p value control subjects (n=9). This analysis showed 23 differentially <0.05) difference in 4 out of 15 miRNAs: miR-let-7f-5p, miR-146a- expressed miRNAs in AT-AMI patients compared with control sub- 5p, miR-151a-3p and miR-223-5p (Supplementary Fig. 1), confidence jects (Supplementary Table 1), including let-7d-5p, miR-103a-3p, intervals (95%) are shown in Supplementary Table 3. The presence of miR-106a-5p, miR-125b-5p, miR-191-5p, miR-22-3p, miR-26b-5p DM, HT, DL, smoking condition or the use of statins did not affect the and miR-30c-5p (Fig. 1) previously associated to atherosclerotic acute expression of these miRNAs (data not shown). In order to confirm myocardial infarction. [17-22] Confidence intervals (95%) of miRNAs these results, we performed PCR assays with individual PCR primer in Fig. 1 are shown in Supplementary Table 2 sets in samples from the validation cohort using the AT-AMI subset 2 (n = 15, female = 15, male = 0, Fig. 2). Our data demonstrate that the circulating levels of miR-let7f-5p, miR-146a-5p, miR-151a-3p and 3.2. SCAD and AT-AMI patients display specific signatures of plasma miR-223-5p were higher in SCAD patients compared with AT-AMI miRNAs patients (p < 0.05) (Fig. 3a), confidence intervals (95%) are shown in Supplementary Table 4. To evaluate whether the expression of miR- Next, we specifically analysed differences in the expression of cir- let-7f-5p, miR-146a-5p, miR-151a-3p or miR-223-5p could be useful culating miRNAs in SCAD patients compared to AT-AMI patients. to distinguish SCAD from AT-AMI patients, receiver operating charac- miRNAs expression was analysed in the two independent cohorts teristic (ROC) curves analysis was performed. Expression of miR-let- using an unbiased panel of 179 miRNAs. A total of 56 AT-AMI and 26 7f-5p showed an AUC of 0.8 (95%CI: 0.620.98, p = 0.004). The AUC SCAD patients were studied. Discovery cohort consisted of 15 SCAD values for miR-146a-5p were 0.83 (95%CI: 0.661, p = 0.0013). miR- patients (14 females and 1 male) and 15 AT-AMI patients (all 151a-3 showed an AUC of 0.82 (95% CI: 0.641(p = 0.0016) while the females), and validation cohort consisted of 11 SCAD patients (all AUC values for miR-223-5p were 0.80 (95% CI: 0.660.96, p = 0.0026) females) and 41 AT-AMI patients (22 females and 19 males) (Fig. 2). (Fig. 3b). The global evaluation of the 4 miRNAs showed a higher pre- The presence of diabetes mellitus (DM) was more prevalent in the dictive value compared to each individual miRNA, showing an AUC of AT-AMI group than in SCAD group (discovery cohort 13% vs 0%; vali- 0.88 (95% CI: 0.721, p = 0.0002) (Fig. 3c), specificity of 0.64 (95% CI: dation cohort 44% vs 27%). The percentage of smokers was also higher 0.510.77), sensitivity 0.72 (95% CI: 0.620.83), positive predictive in the group of patients with AT-AMI (discovery cohort: 47% vs 7%, value 0.68 (95% CI: 0.580.76) and negative predictive value of 0.83 validation cohort: 44% vs 27%). Regarding hypertension (HT), no sta- (95% CI: 0.790.87). tistically significant differences were observed in the discovery cohort. However, the percentage of AT-AMI patients with this comor- bidity was higher in the validation cohort compared to SCAD patients. 3.3. Association of miRNA-target genes with blood vessel biology Dyslipidaemia (DL) was also higher in AT-AMI group in both cohorts (discovery cohort: 13% vs 7%, validation cohort: 56% vs 9%). All To explore the possible functional implication of miR-let-7f-5p, patients were taking acetylsalicylic acid (ASA) at sampling time. miR-146a-5p, miR-151a-3p and miR-223-5p we determined the Fig. 1. Plasma miRNAs differentially expressed in control subjects compared to AT-AMI patients. Box and whiskers Min to Max plots showing plasma levels of microRNAs. Data correspond to the expression of miRNAs in AT-AMI n=41 and control subjects n=9. Expression of miRNAs was analysed by RT-PCR using a serum/plasma focus PCR panel. Data were normalized using the global mean, expression levels are shown as Normalized relative quantity (NRQ), as indicated under Methods. Differences among groups were analysed by univariate test followed by FDR multiple correction test. M. Lozano-Prieto et al. / EBioMedicine 66 (2021) 103338 5 Fig. 2. Study design and cohorts. Plasma samples (AT-AMI n = 56, SCAD n = 26) were collected in both UK (AT-AMI n = 15, SCAD n = 15) and Spain (AT-AMI n = 41, SCAD n = 11), constituting the discovery and the validation cohorts, respectively. The number of males and females of each sample set is shown in the figure. Firstly, differences in the expression of plasma miRNAs between AT-AMI and SCAD patients from the discovery cohort were assessed performing RT-PCR assays using unbiased PCR panels of 179 miRNAs. Differentially expressed miRNAs identified in the discovery cohort were then analysed in 26 AT-AMI and 11 SCAD patients from the validation cohort using the same method. Lastly, resulting dif- ferentially expressed miRNAs were confirmed by performing PCR assays with individual PCR primer sets in 15 AT-AMI (all female) not included in the previous phase and 11 SCAD patients from the validation cohort. Created with BioRender.com Table 1 Clinical and demographic variables. Cohort 1 (Discovery cohort) Cohort 2 (Validation cohort) SCAD (n = 15) AT-AMI (n = 15) SCAD (n = 11) AT-AMI (n = 41) CONTROL (n = 9) Sex (F/M) 15/0 15/0 11/0 22/19 9/0 Age mean § SD (years) 49.6 § 6.5 52.0 § 6.6 52.6 § 12.1 61.4 § 9.2 53 § 5.2 STEMI 7 7 4 25 0 NSTEMI 8 8 7 16 0 SMOKING (NO/YES/EX) 10/1/4 8/7/0 8/3/0 16/18/7 6/3/0 HT (NO/YES) 12/3 11/4 8/3 15/26 0 DM (NO/YES) 15/0 13/2 8/3 23/18 0 DL (NO/YES) 14/1 13/2 10/1 18/23 0 BMI mean § SD 29.1 § 10.2 31.0 § 6.2 25.7 § 4.9 27.5 § 5.4 26.3 § 8.5 Use of statins 15 15 4 27 0 Use of ASA 15 15 11 40 0 Culprit vessel N/A 24-LAD 7 3 5 8 24-RCA 3 9 0 14 24-CX 5 3 2 3 Number of vessels affected N/A 24-0 0 0 0 2 24-1 14 5 8 21 24-2 0 9 2 11 24-3 1 1 1 7 Coronary revascularization N/A 24-Conservative 10 1 11 0 24-PCI 5 13 0 37 24-CABG 0 1 0 2 LVEF (%) mean § SD 55.6 § 6.6 61.3 § 4.7 54.3 § 12.0 57.1 § 9.5 N/A HT hypertension, DM diabetes mellitus, DL dyslipidaemia, BMI body mass index, ASA acetylsalicylic acid, LAD left anterior descending artery, RCA right coronary artery, CX circumflex artery, PCI percutaneous coronary intervention, CABG coronary artery bypass graft, LVEF left ventricular ejection fraction. N/A: not applicable. 6 M. Lozano-Prieto et al. / EBioMedicine 66 (2021) 103338 Table 2 associated with biological processes according with terms of Gene DE miRNAs in SCAD vs AT-AMI patients in Discovery cohort. Ontology. The functional profiling showed that the list of target genes miRNA ID p value (FDR) SCAD/AT-AMI was enriched (FDR corrected p value <0.05) in genes involved in blood vessel biology including blood vessel morphogenesis, circula- hsa_mir_2110 0.0233887517108984 UP tory system development, and the regulation of smooth muscle cell hsa_let_7d_3p 0.0233887517108984 UP hsa_let_7f_5p 0.0346269703856619 UP proliferation (Fig. 4a). Forty-eight genes were involved in blood ves- hsa_mir_92b_3p 0.0346269703856619 UP sel development among them Transforming Growth Factor beta-1 hsa_mir_1 0.0371248112086971 UP (TGFB1) and Endothelin-1 (EDN1) genes regulated by miR-146a-5p hsa_mir_320c 0.0371248112086971 UP and miR-let-7f-5p, respectively (Fig. 4b). A statistically significant hsa_mir_146a_5p 0.0371248112086971 UP hsa_mir_223_5p 0.0371248112086971 UP over-representation of target genes in TGFB1 signalling pathway was hsa_mir_200c_3p 0.0371248112086971 UP also identified (Fig. 4a). Target genes associated with blood vessel hsa_mir_151a_3p 0.0371248112086971 UP development/morphogenesis and TGF-beta signalling are shown in hsa_mir_361_5p 0.0498538460706333 UP Fig. 4b. Among the target genes associated with these biological pro- hsa_mir_133a_3p 0.0498538460706333 UP cesses, we detect a group of genes associated to cytoskeletal traction hsa_mir_421 0.0498538460706333 UP hsa_mir_501_3p 0.0498538460706333 UP forces (WASF2, RHOA, ROCK1, MYLK and MYH9, Fig. 4b). hsa_mir_93_3p 0.0498538460706333 UP Finally, we evaluated the plasma levels of most relevant molecules identified as targets: TGFB1, EDN1, C-X-C Motif Chemokine Ligand 8 (CXCL8 also known as IL-8), Transforming Growth Factor Beta Recep- target genes using miRtarBase (http://mirtarbase.cuhk.edu.cn/php/ tor 1 (TGFBR1), and Matrix Metalloproteinase-2 (MMP2). ELISA index.php, last accessed 10-4-20) a database of experimentally vali- assays revealed statistically significant higher levels of these mole- dated miRNA-target interactions, a total of 924 genes were identified cules in plasma samples from SCAD patients compared with AT-AMI as functional targets. We assessed whether these genes were patients (Fig. 4c, confidence intervals in Supplementary Table 5). Fig. 3. Plasma microRNA signature and its predictive value to distinguish SCAD from AT-AMI patients. a, Box and whiskers Min to Max plots showing plasma levels of miR-let- 7f-5p, miR-146a-5p, miR-151a-3p and miR-223-5p from SCAD (n = 11) and AT-AMI (n = 15) patients from validation cohort. Expression of miRNAs was analysed by RT-PCR using individual sets of primers. Data were normalized using the geometric mean of the five most stable miRNAs. Expression levels are shown as Normalized relative quantity (NRQ), as indicated under Methods. Differences among groups were analyzed by a Mann-Whitney test. b. Receiver operating characteristic (ROC) curves corresponding to miR-let-7f-5p, miR-146a-5p, miR-151a-3p and miR-223-5p expression levels from SCAD (n = 11) and AT-AMI (n = 15) patients from validation cohort. c. Receiver operating characteristic (ROC) curve corresponding to the concomitant evaluation of miR-let-7f-5p, miR-146a-5p, miR-151a-3p and miR-223-5p expression levels from SCAD (n = 11) and AT-AMI (n = 15) patients from validation cohort. b,c The bold line shows the ROC curve. The values inside the graph correspond to the area under the curve (AUC) and its 95% confidence interval; p value is also indicated. M. Lozano-Prieto et al. / EBioMedicine 66 (2021) 103338 7 Fig. 4. Target genes of miR-let-7f-5p, miR146a-5p, mir-151a-3p and miR-223-5p are associated with blood vessel biology. a, Bar graph illustrating the significance (-log p value) of blood vessel biology-associated biological processes predicted by PANTHER Classification System. Functional miRNA targets were selected using miRTarBase database and analysed to obtain functional classifications with terms from Gene Ontology. Data were analysed by Fisher’s test and Bonferroni correction. b, Functional miRNA targets associated with blood vessel biology, TGF-beta pathway and cytoskeleton are shown. Colours indicate each miRNA with its corresponding targets. SMC, smooth muscle cell. Triangles indicate cytokines and soluble factors; squares indicate plasma membrane receptor proteins; circles, intracellular proteins (cytoplasmic and/or nuclear); hexagons, proteins with extra- and intracellular function and/or presence. c, Box and whiskers Min to Max plots showing plasma levels of CXCL8, TGFB1, TGFBR1, Endothelin-1 and MMP2 in SCAD patients (n = 26) and AT-AMI patients (n = 34). Levels of these molecules were analysed using a custom multiplex ELISA assay. Differences between groups were evaluated using Mann-Whitney U test. 4. Discussion infarction. Furthermore, we identify four differentially expressed miRNAs validated from two cohorts studied. Accordingly, the combi- To our knowledge, this is the first study that investigates the role nation of miR-let-7f-5p, miR-146a-5p, miR-151a-3p and miR-223-5p of plasma miRNAs as potential biomarkers for acute SCAD. According is a promising predictor of acute SCAD. Our data also reveal that the to the data presented herein, acute SCAD has a specific miRNA signa- functional profiling of target genes of the four differentially expressed ture that is different from that of athero-thrombotic acute myocardial miRNAs regulate genes known to be associated with vascular biology, 8 M. Lozano-Prieto et al. / EBioMedicine 66 (2021) 103338 TGF-beta signaling and SCAD. Our data also revealed that relevant the blood levels of these molecules between SCAD and AT-AMI targets for the vascular pathology such as CXCL-8, TGFB1, TGFBR1 patients. In the case of EDN1, an association between lower levels of and EDN1 are increased in plasma from acute SCAD patients. These EDN1 in plasma and the presence of the PHACTR1/EDN1 risk allele findings support further validation of miRNAs as diagnostic bio- rs9349379 does exist. [28] Further investigation is needed to deter- markers as well as their pathophysiological role in larger studies. mine whether the plasma levels of EDN1 in SCAD patients are associ- Although less common than AT-AMI, SCAD is not rare accounting ated to additional factors, and to establish the levels of this molecule for up to 35% of events in women under 50 years of age and 43% of in a healthy population. pregnancy-associated AMI. As understanding of the condition has Mutations in components of TGFB pathway result in a series of moved forward following the publication of the first American and diseases collectively known as TGFB vasculopathies. [29,30] Indeed, European consensus documents, key differences in management causal variants in SMAD2, TGFB1 and TGFBR2 have been recently between SCAD and AT-AMI have become increasingly evident. These reported in SCAD. [31] In this regard, our data identified an enrich- include different approaches to revascularization and potentially sec- ment of functional miRNA-targets in the pathway of TGFB response. ondary prevention medications. In this context, accurate diagnosis is [32,33] miR-146a-5p regulates TGFB1 expression, a crucial player in critical. Currently, invasive coronary angiography is the gold standard the development and homeostasis of the vascular system mainly diagnostic test, but many SCAD patients present with angiographic through TGFBR1 and TGFBR3. Strikingly, both receptors are regulated patterns that mimic or are actually indistinguishable from atheroscle- by miR-let-7f-5p, which is also a differential marker identified in this rotic coronary artery disease. Use of intracoronary imaging is study. Dysregulation of the TGFB1 signalling cascade has been associ- required in these cases necessitating mechanical instrumentation of ated with syndromic aortic aneurysm and dissection conditions, the coronary vessel with inherent risks in patients with SCAD, in including heritable connective tissue disorders such as Marfan syn- which the vessel wall is fragile and vulnerable to iatrogenic complica- drome (MFS), Loeys-Dietz syndrome and vascular Ehlers-Danlos syn- tions. [6] This has led to a search for acute SCAD-specific biomarkers drome (vEDS). [34] TGFB is also related to the onset of fibrosis in which can be used either to support angiographic diagnosis or even different pathologies, which strongly suggests an important function to develop a new non-invasive diagnostic paradigm. in the mechanical response of the tissue to shear stress. [35] This is In this study we have focused on miRNAs due to their potential as an important issue as collagen vascular disorders like MFS or vEDS diagnostic, prognostic, and monitoring biomarkers in many different are predisposing conditions related to SCAD. [36] miR-let-7f-5p also contexts including in AT-AMI. [911] As a proof of concept, we first seems regulate the expression of endothelin 1 (EDN1); [37] this is assessed the ability of our approach to reproduce miRNAs differen- highly expressed in vascular endothelial and smooth muscle cells, tially expressed in AT-AMI compared to controls subjects. The com- macrophages and podocytes among other type of cells, and it has parison with other AT-AMI studies is a challenge mainly due to been also recently associated with SCAD. [28,38,39] differences in normalization methods; nevertheless, our data show The finding of a group of interrelated target genes (WASF2, RHOA, differences in the expression of miR-22-3p, miR-22-5p, miR-26b-5p, ROCK1, MYLK and MYH9) suggests the involvement of a mechano- miR-30c-5p, miR-125b-5p, miR-106a-5p, and miR-191-5p, as previ- active mechanism that controls cell shape and cytoskeleton dynamics ously reported. [1722] in SCAD. In addition, mutations in MYLK cause familial aortic dissec- Our approach allowed us to identify 4 miRNAs whose presence in tion underscoring the role of mechano-sensing and mechanical plasma is higher in SCAD patients compared to AT-AMI (p value responses in the maintenance of aortic homeostasis. [40] Very recent <0.05). One of these miRNAs is miR-146a-5p, the increased presence data highlight the presence of causal variants in MYLK in SCAD. [41] in plasma of which has also been reported in patients with aortic dis- Whether these biological processes are a cause, or a consequence of section. [23] In particular, expression of miR-146a-5p, but also the SCAD pathophysiology is just prospective. Neither the cells from combination of the 4 miRNAs described here, was able to distinguish which these plasma miRNAs are secreted, nor their recipient cells can SCAD from AT-AMI patients. The AUC, specificity, sensitivity, positive be elucidated from our results, although both topics are very interest- predictive value and negative predictive value compare favourably to ing aspects to address in future investigations. a plasma protein and lipid-based approach recently published for In summary, our work provides a distinct miRNA signature that, fibromuscular dysplasia, a condition related-to and overlapping with with further prospective validation, could be a useful biomarker to SCAD. [24] The combination of these miRNAs is therefore a poten- distinguish acute SCAD from AT-AMI. It also provides novel insight tially suitable candidate to be studied in larger prospective cohorts as into the molecular mechanisms underlying this condition. a biomarker for acute SCAD. During the last years, circulating miRNAs have been proposed as potential biomarkers in numerous studies. 4.1. Study limitations However, the inherent difficulties of their study and normalization strategies result in limitations in their use as diagnostic tools. [25] Our data is limited by the sample size, due to the relatively low These findings may also provide novel insights into the molecular incidence of acute SCAD and as such it has not been possible to inves- pathophysiology of acute SCAD. miRNAs participate in different tigate any potential co-variates with the potential to influence the molecular pathways associated with coronary diseases, such as endo- findings. thelial dysfunction, smooth muscle cell and platelet activation, inflammation or atheromatous plaque formation. [26,27] Interest- 4.2. Contributors ingly, we identified a statistically significant enrichment of target genes of the miRNAs signature in processes associated to the mor- MLP: Literature search, data collection, writing; DA: Literature phogenesis and development of the circulatory system, for example search, writing, data interpretation, patients recruitment; MGG: Con- LRP2, NOTCH1, TGFB1, TGFBR3, TGFBR1, SMAD2, CXCL8, EDN1, MMP2, ceived the presented idea, data interpretation; ASG: Statistical analy- among others. Increased plasma levels of IL-8, EDN1, TGFB1, TGFBR1 sis; PVT: samples management for the revised version; data and MMP2 in SCAD patients highlight the possible involvement of collection FR: Patients recruitment, data interpretation; JC: Patients these molecules in the pathophysiology of SCAD. The high levels of recruitment; TB: Patients recruitment; AB: samples management, miR-let-7f-5p, miR-146a-5p, miR-151a-3p and miR-223-5p in SCAD data base; AV samples management; EMG: Figures and network anal- patients could be a mechanism to counter-regulate the expression of ysis; MVM: Figures and data interpretation; PM: data interpretation; those molecules. Several functions described for these gene products NS: Patients recruitment, data interpretation; FSM: Supervised the make them interesting candidates to be explored in further detail in findings of this work, data interpretation; FA: Supervised the findings SCAD patients. To our knowledge, no prior studies have compared of this work, overall planning; HF: Data collection, analysis, writing M. Lozano-Prieto et al. / EBioMedicine 66 (2021) 103338 9 and interpretation. All authors have read and approved the final ver- acute myocardial infarction in patients presenting with ST-segment elevation of sion of the manuscript. MLP, HF and FSM have verified the underlying the European society of cardiology (ESC). Eur Heart J 2018;39(2):119–77. [15] D'Haene B, Mestdagh P, Hellemans J, Vandesompele J. miRNA expression profil- data. ing: from reference genes to global mean normalization. Methods Mol Biol 2012;822:261–72. 5. Data Sharing Statement [16] Andersen CL, Jensen JL, Orntoft TF. Normalization of real-time quantitative reverse transcription-PCR data: a model-based variance estimation approach to identify genes suited for normalization, applied to bladder and colon cancer data The data that support the findings of this study are available from sets. Cancer Res 2004;64(15):5245–50. the corresponding author upon request. [17] Zhong Z, Zhong W, Zhang Q, Zhang Q, Yu Z, Wu H. Circulating microRNA expres- sion profiling and bioinformatics analysis of patients with coronary artery disease by RNA sequencing. J Clin Lab Anal 2020;34(1):e23020. Declaration of Competing Interest [18] Li C, Chen X, Huang J, Sun Q, Wang L. Clinical impact of circulating miR-26a, miR- 191, and miR-208b in plasma of patients with acute myocardial infarction. Eur J Med Res 2015;20:58. DA has received in kind support for SCAD genetics research from [19] Fichtlscherer S, De Rosa S, Fox H, Schwietz T, Fischer A, Liebetrau C, et al. Circulat- Astra Zeneca Inc. and research grants for unrelated research from the ing microRNAs in patients with coronary artery disease. Circ Res 2010;107 same Company. He has received funding from Abbott Vascular Inc. to (5):677–84. [20] Jakob P, Kacprowski T, Briand-Schumacher S, Heg D, Klingenberg R, Stahli BE, support a clinical research Fellow and has undertaken unrelated con- et al. Profiling and validation of circulating microRNAs for cardiovascular events sultancy for GE inc. No other authors have potential conflicts of inter- in patients presenting with ST-segment elevation myocardial infarction. Eur est to declare. Heart J 2017;38(7):511–5. [21] Chang TY, Tsai WC, Huang TS, Su SH, Chang CY, Ma HY, et al. Dysregulation of endothelial colony-forming cell function by a negative feedback loop of circulat- Acknowledgments ing miR-146a and -146b in cardiovascular disease patients. PLoS One 2017;12(7): e0181562. We acknowledge the leadership of the ESC-ACCA SCAD Study [22] Huang W, Nussinov R, Zhang J. Computational tools for allosteric drug discovery: site identification and focus library design. Methods Mol Biol 2017;1529:439–46. Group. We thank study participants. We acknowledge Jenny Middle- [23] Xue L, Luo S, Ding H, Liu Y, Huang W, Fan X, et al. Upregulation of miR-146a-5p is ton, Jane Plume, Donna Alexander, Sue Sterland, Daniel Lawday, associated with increased proliferation and migration of vascular smooth muscle Emma Beeston, Shireen Kharodia, Tara Maitland and Andrea Marshall cells in aortic dissection. J Clin Lab Anal 2019;33(4):e22843. [24] Olin JW, Di Narzo AF, d'Escamard V, Kadian-Dodov D, Cheng H, Georges A, et al. A from the Leicester NIHR BRC, UK. plasma proteogenomic signature for fibromuscular dysplasia. Cardiovasc Res 2020;116(1):63–77. Supplementary materials [25] Peters LJF, Biessen EAL, Hohl M, Weber C, van der Vorst EPC, Santovito D. Small things matter: relevance of microRNAs in cardiovascular disease. Front Physiol 2020;11:793. Supplementary material associated with this article can be found, [26] Thomas MR, Lip GY. Novel risk markers and risk assessments for cardiovascular in the online version, at doi:10.1016/j.ebiom.2021.103338. disease. Circ Res 2017;120(1):133–49. [27] Viereck J, Thum T. Circulating noncoding RNAs as biomarkers of cardiovascular disease and injury. Circ Res 2017;120(2):381–99. References [28] Adlam D, Olson TM, Combaret N, Kovacic JC, Iismaa SE, Al-Hussaini A, et al. Asso- ciation of the PHACTR1/EDN1 genetic locus with spontaneous coronary artery [1] Hayes SN, Kim ESH, Saw J, Adlam D, Arslanian-Engoren C, Economy KE, et al. dissection. J Am Coll Cardiol 2019;73(1):58–66. Spontaneous coronary artery dissection: current state of the science: a scientific [29] Loeys BL, Chen J, Neptune ER, Judge DP, Podowski M, Holm T, et al. A syndrome of statement from the american heart association.. Circulation. 2018;137(19):e523– altered cardiovascular, craniofacial, neurocognitive and skeletal development e57. caused by mutations in TGFBR1 or TGFBR2. Nat Genet 2005;37(3):275–81. [2] Gilhofer TS, Saw J. Spontaneous coronary artery dissection: update 2019. Curr [30] Bertoli-Avella AM, Gillis E, Morisaki H, Verhagen JMA, de Graaf BM, van de Beek G, Opin Cardiol 2019;34(6):594–602. et al. Mutations in a TGF-beta ligand, TGFB3, cause syndromic aortic aneurysms [3] Adlam D, Alfonso F, Maas A, Vrints C, Writing C. European society of cardiology, and dissections. J Am Coll Cardiol 2015;65(13):1324–36. acute cardiovascular care association, SCAD study group: a position paper on [31] Verstraeten A, Perik M, Baranowska AA, Meester JAN, Van Den Heuvel L, Bastia- spontaneous coronary artery dissection. Eur Heart J 2018;39(36):3353–68. nen J, et al. Enrichment of rare variants in Loeys-Dietz syndrome genes in sponta- [4] Hayes SN, Tweet MS, Adlam D, Kim ESH, Gulati R, Price JE, et al. Spontaneous cor- neous coronary artery dissection but not in severe fibromuscular dysplasia. onary artery dissection: JACC state-of-the-art review. J Am Coll Cardiol 2020;76 Circulation 2020;142(10):1021–4. (8):961–84. [32] Compton LA, Potash DA, Brown CB, Barnett JV. Coronary vessel development is [5] Alfonso F. Spontaneous coronary artery dissection: new insights from the tip of dependent on the type III transforming growth factor beta receptor. Circ Res the iceberg? Circulation 2012;126(6):667–70. 2007;101(8):784–91. [6] Jackson R, Al-Hussaini A, Joseph S, van Soest G, Wood A, Macaya F, et al. Sponta- [33] Sun F, Li X, Duan WQ, Tian W, Gao M, Yang J, et al. Transforming growth factor- neous coronary artery dissection: pathophysiological insights from optical coher- beta Receptor III is a potential regulator of ischemia-induced cardiomyocyte apo- ence tomography. JACC Cardiovasc Imaging 2019;12(12):2475–88. ptosis. J Am Heart Assoc 2017;6(6). [7] Bartel DP. MicroRNAs: genomics, biogenesis, mechanism, and function. Cell. [34] Isselbacher EM, Lino Cardenas CL, Lindsay ME. Hereditary influence in thoracic 2004;116(2):281–97. aortic aneurysm and dissection. Circulation 2016;133(24):2516–28. [8] Mittelbrunn M, Sanchez-Madrid F. Intercellular communication: diverse struc- [35] Biernacka A, Dobaczewski M, Frangogiannis NG. TGF-beta signaling in fibrosis. tures for exchange of genetic information. Nat Rev Mol Cell Biol 2012;13(5):328– Growth Factors 2011;29(5):196–202. 35. [36] Garcia-Guimaraes M, Bastante T, Antuna P, Jimenez C, de la Cuerda F, Cuesta J, [9] Rupaimoole R, Slack FJ. MicroRNA therapeutics: towards a new era for the man- et al. Spontaneous coronary artery dissection: mechanisms, diagnosis and man- agement of cancer and other diseases. Nat Rev Drug Discov 2017;16(3):203–22. agement. Eur Cardiol 2020;15:1–8. [10] Huang W. MicroRNAs: biomarkers, diagnostics, and therapeutics. Methods Mol [37] Whisnant AW, Bogerd HP, Flores O, Ho P, Powers JG, Sharova N, et al. In-depth Biol. 2017;1617:57–67. analysis of the interaction of HIV-1 with cellular microRNA biogenesis and effec- [11] Barwari T, Joshi A, Mayr M. MicroRNAs in cardiovascular disease. J Am Coll Cardiol tor mechanisms. mBio 2013;4(2):e000193. 2016;68(23):2577–84. [38] Barton M, Yanagisawa M. Endothelin: 30 years from discovery to therapy. Hyper- [12] Thygesen K, Alpert JS, Jaffe AS, Chaitman BR, Bax JJ, Morrow DA, et al. Fourth uni- tension 2019;74(6):1232–65. versal definition of myocardial infarction (2018). J Am Coll Cardiol 2018;72 [39] Saw J, Yang ML, Trinder M, Tcheandjieu C, Xu C, Starovoytov A, et al. Chromosome (18):2231–64. 1q21.2 and additional loci influence risk of spontaneous coronary artery dissec- [13] Collet JP, Thiele H, Barbato E, Barthelemy O, Bauersachs J, Bhatt DL, et al. 2020 ESC tion and myocardial infarction. Nat Commun 2020;11(1):4432. Guidelines for the management of acute coronary syndromes in patients present- [40] Wang L, Guo DC, Cao J, Gong L, Kamm KE, Regalado E, et al. Mutations in myosin ing without persistent ST-segment elevation. Eur Heart J 2020. doi: 10.1093/ light chain kinase cause familial aortic dissections. Am J Hum Genet 2010;87 eurheartj/ehaa575. (5):701–7. [14] Ibanez B, James S, Agewall S, Antunes MJ, Bucciarelli-Ducci C, Bueno H, et al. 2017 [41] Carss KJ, Baranowska AA, Armisen J, Webb TR, Hamby SE, Premawardhana D, ESC Guidelines for the management of acute myocardial infarction in patients et al. Spontaneous coronary artery dissection: insights on rare genetic variation presenting with ST-segment elevation: The Task Force for the management of from genome sequencing. Circ Genom Precis Med 2020;13(6):e003030.