RNA-sequencing reveals that STRN, ZNF484 and WNK1 add to the value of mitochondrial MT-COI and COX10 as markers of unstable coronary artery disease

RESEARCH ARTICLE RNA-sequencing reveals that STRN, ZNF484 and WNK1 add to the value of mitochondrial MT-COI and COX10 as markers of unstable coronary artery disease Paul HolvoetID 1 *, Bernward Klocke 2 , Maarten VanhaverbekeID 3 , Roxane Menten 1 , Peter Sinnaeve 3 , Emma Raitoharju 4,5 , Terho Lehtima ¨ki 4,5 , Niku Oksala 6 , Christian Zinser 2 , Stefan Janssens 3 , Karin Sipido 1 , Leo-Pekka Lyytikainen 4,5 , Stefano CagninID 7,8 1Department of Cardiovascular Sciences, KU Leuven, Leuven, Belgium, 2Intrexon Bioinformatics Germany, Munich, Germany, 3Department of Clinical Cardiology, UZ Leuven, Leuven, Belgium, 4Department of Clinical Chemistry, Fimlab Laboratories, Tampere, Finland, 5Finnish Cardiovascular Research Centre, Faculty of Medicine and Life Sciences University of Tampere, Tampere, Finland, 6Division of Vascular Surgery, Department of Surgery, Tampere University Hospital, Tampere, Finland, 7Department of Biology, CRIBI Biotechnology Centre, Padova, Italy, 8CIR-Myo Myology Centre, University of Padova, Padova, Italy *[email protected] Abstract Markers in monocytes, precursors of macrophages, which are related to CAD, are largely unknown. Therefore, we aimed to identify genes in monocytes predictive of a new ischemic event in patients with CAD and/or discriminate between stable CAD and acute coronary syndrome. We included 66 patients with stable CAD, of which 24 developed a new ischemic event, and 19 patients with ACS. Circulating CD14+ monocytes were isolated with magnetic beads. RNA sequencing analysis in monocytes of patients with (n = 13) versus without (n = 11) ischemic event at follow-up and in patients with ACS (n = 12) was validated with qPCR (n = 85). MT-COI, STRN and COX10 predicted new ischemic events in CAD patients (power for separation at 1% error rate of 0.97, 0.90 and 0.77 respectively). Low MT-COI and high STRN were also related to shorter time between blood sampling and event. COX10 and ZNF484 together with MT-COI, STRN and WNK1 separated ACS completely from stable CAD patients. RNA expressions in monocytes of MT-COI, COX10, STRN, WNK1 and ZNF484 were independent of cholesterol lowering and antiplatelet treatment. They were independent of troponin T, a marker of myocardial injury. But, COX10 and ZNF484 in human plaques correlated to plaque markers of M1 macrophage polarization, reflecting vascular injury. Expression of MT-COI, COX10, STRN and WNK1, but not that of ZNF484, PBMCs paired with that in monocytes. The prospective study of relation of MT-COI, COX10, STRN, WNK1 and ZNF484 with unstable CAD is warranted. PLOS ONE | https://doi.org/10.1371/journal.pone.0225621 December 10, 2019 1 / 18 a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 OPEN ACCESS Citation: Holvoet P, Klocke B, Vanhaverbeke M, Menten R, Sinnaeve P, Raitoharju E, et al. (2019) RNA-sequencing reveals that STRN, ZNF484 and WNK1 add to the value of mitochondrial MT-COI and COX10 as markers of unstable coronary artery disease. PLoS ONE 14(12): e0225621. https://doi. org/10.1371/journal.pone.0225621 Editor: Carmine Pizzi, University of Bologna, ITALY Received: May 15, 2019 Accepted: November 9, 2019 Published: December 10, 2019 Copyright: ©2019 Holvoet et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Data Availability Statement: The full set of RNA sequencing data is available at https://www.ncbi. nlm.nih.gov/geo/query/acc.cgi?acc=GSE129935. All other relevant data are within the paper and its Supporting Information files. Funding: This work was funded by the Bijzonder Onderzoeksfonds of the KU Leuven (PF/10/014; Centre of Excellence), by the University of Padova (CPDA139317), and by Cariplo foundation (20161006), Italian Ministry of Health (GR-2011– 02346845), AIRC (IG 2015 Id.17773), Agenzia Introduction Several millions of patients in Western countries are hospitalized each year for chest pain. In approximately half of the cases, chest pain is of cardiac origin [1]. Among these patients approximately 50% exhibit underlying coronary artery disease (CAD) that eventually leads to an acute coronary syndrome (ACS). ACS encompasses the clinical spectrum ranging from unstable angina through acute myocardial infarction (AMI). Since we aim to search biomarkers, a non-invasive approach by performing analyses in peripheral blood was considered to be more convenient and translatable to clinical practice. Moreover, since atherosclerosis is a systemic disease in which monocytes and derived macrophages play a crucial role, we believe that measuring monocyte behaviour in peripheral blood reflects the activity inside the coronary vessel wall. This view is also supported by previous findings where gene expression in peripheral whole blood samples appeared to mirror gene expression changes in the atherosclerotic vascular wall [2]. Previously, we measured members of the cytochrome oxidase (COX) IV complex, because it has been proposed that mitochondrial dysfunction resulting in mitochondrial oxidative stress contributes to development of age-related metabolic changes and CAD [3,4]. We demonstrated that low MT-COI in monocytes of coronary artery disease patients identified a population at risk for new cardiovascular events [5]. However, we had selected cytochrome oxidase as a target a priori, and did not perform an unbiased analysis. Therefore, in this study we performed unbiased RNA sequence analysis followed by several modelling approaches to indentify the best prognostic markers predicting a future event in stable CAD patients and the best markers of ACS at time of blood sampling. Thus, our first aim was to search for markers improving the prediction of a new ischemic event in stable CAD patients during a 5-year follow-up. Our second aim was to compare gene in stable CAD and ACS patients. We performed an exploratory RNA sequencing (RNA-Seq) analysis of RNA isolated from monocytes, precursors of macrophages, followed by selective quantitative validation of robustly differentially expressed genes with qPCR. We confirmed that MT-COI predicted a new event but that striatin (STRN) added to the power. In addition, we found that COX10 and zinc finger 484 were markers of ACS. Then, we determined if those markers were related to cardiac troponin T in ACS patients, i.e. reflecting myocardial injury [6,7]. To our surprise, the identified markers did not correlate with cardiac troponin, in this study we measured their expression in atherosclerotic plaques to determine their correlation with markers of vascular injury, in particular M1 macrophage markers. The current work identifies 2 novel markers in addition to members of the COX complex IV, which may improve discrimination between stable and unstable CAD patients. Materials and methods Patients and follow-up All study subjects gave written informed consent. We did not include minors. The clinical study performed at University Hospital Leuven conforms to the principles outlined in the Declaration of Helsinki, and was approved by the Medical Ethics Committee of the University Hospital Leuven. We included 142 consecutive patients undergoing quantitative coronary angiography (QCA). In 66 patients, CAD was diagnosed as an epicardial coronary stenosis of at least 50% in at least 1 of the coronary arteries. This cohort was studied before, but in here we defined CAD as coronary stenosis of 50 instead of 30% [5]. These patients were all stable at time of blood sampling. Twenty-four (36.4%) of those 66 CAD patients experienced at least 1 new ischemic event during a 5-year follow-up: cardiovascular death (n = 2), ischemic stroke Markers of unstable CAD PLOS ONE | https://doi.org/10.1371/journal.pone.0225621 December 10, 2019 2 / 18 Nazionale di Valutazione del Sistema Universitario e della Ricerca (ANVUR) (FFABR2017), and by grants from the Competitive Research Funding of the Tampere University Hospital (Grant 9M048 and 9N035 for T.L.), the Emil Aaltonen Foundation (T. L.), the Pirkanmaa Regional Fund of the Finnish Cultural Foundation, the Research Foundation of Orion Corporation, the Jenny and Antti Wihuri Foundation, and the Academy of Finland Grant no. 104821), the Finnish Foundation for Cardiovascular Research, the Yrjo¨Jahnsson Foundation and European Union 7th Framework Program, grant number 201668, AtheroRemo and EU Horizon 2020 (grant 755320 for TAXINOMISIS). P.S. and M.V. are researchers of the Research Foundation Flanders. The funding organizations had no role in design, in the collection, analysis, and interpretation of data; in the writing of the manuscript; and in the decision to submit the manuscript for publication. Bernward Klocke is employed by Intrexon Bioinformatics Germany which was hired by KU Leuven to deliver service in modeling. Intrexon Bioinformatics Germany provided support in the form of salary for author B. K., but did not have any additional role in the study design, data collection, decision to publish, or preparation of the manuscript. The specific role of this author is articulated in the ‘author contributions’ section. Competing interests: We have the following interests: Bernward Klocke is employed by Intrexon Bioinformatics Germany which was hired by KU Leuven to deliver service in modeling. There are no patents, products in development or marketed products to declare. This does not alter our adherence to all the PLOS ONE policies on sharing data and materials. (n = 2), recurrent AMI (n = 1), and recurrent ischemia requiring recurrent ischemia requiring reintervention (n = 19, of which one patient had in-stent restenosis) [5]. Recurrent ischemia (n = 19) was defined as patients with signs of ischemia (ECG changes, angina pectoris, positive stress test), and significant new or progressive coronary lesions on angiography (n = 19, of which one patient had in-stent restenosis. As we hypothesized that monocyte behaviour is related to plaque stability, we included all major adverse cardiovascular events including stroke. Because ischemic stroke and acute myocardial infarction share similar biological processes, and they reflect a frequently used combined endpoint in clinical trials in both stable and unstable coronary artery disease stroke was also included as major cardiovascular event. The two cardiac deaths included one patient with refractory cardiogenic shock and electrical instability because of critical 3-vessel disease. The second patient developed ventricular fibrillation and recurrent MI because of severe in-stent restenosis and died shortly thereafter. We also included 19 ACS patients, 5 with ST-segment elevation myocardial infarction (STEMI) and 14 with non-ST-elevation MI. Patients with a history of cancer, renal insufficiency, inflammatory disease, treatment with corticosteroids or immunosuppressive drugs, and significant valvular disease were excluded. Isolation of monocytes For isolation of monocytes, blood was collected on BD Vacutainer 1 CPT™Mononuclear Cell Preparation Tube—Sodium Citrate (Becton Dickinson). CD14 + monocytes were isolated from the Histopaque-1077 leukocyte fraction using CD14 microbeads and LS column in a MidiMagnetic Cell Isolation Separator (MACS, Miltenyi), as previously published [5]. Only monocytes wich were isolated within 2 hours of blood sampling were analyzed further. The purity of the isolated CD14+ monocytes was >95%, as previously published. The numbers of CD14+ monocytes isolated from 1 ml (approximately 2.5x10 5 ) of blood were similar for stable CAD patients without and with a new ischemic event and ACS patients. RNA sequencing RNA from monocytes was isolated at KU Leuven as described previously [5]. RNA sequencing (RNA-Seq) was performed by Biogazelle (Gent, Belgium) on a subset of patients which were randomly selected for RNAseq, as a discovery cohort. In the next step, validation with qPCR was performed in all patients. Libraries for mRNA sequencing were prepared using the TruSeq stranded mRNA sample prep kit (Illumina). In one run we analyzed 36 samples from three groups of patients. Group 1: stable CAD patients without new ischemic event (n = 11); group 2: stable CAD patients with new event (n = 13); group 3: ACS patients (n = 12) (see methods in S1 Appendix). We performed Mann Whitney test in search of transcripts distinguishing between stable CAD patients with and without a new ischemic event; thus ACS patients were not included. We also wanted to identify transcripts which separated ACS from the two groups of stable CAD patients. Therefore, we performed multinomial logistic regression analysis combined with random forest as a complementary algorithm for modelling using the R-package “randomForest” by Andy Liaw and Matthew Wiener [8] to compare expressions in the three groups of patients (S1 Appendix). QPCR analysis Expression of selected genes in monocytes, and PBMCs, was validated with qPCR analysis on the whole cohort of 85 patients as described in S1 Appendix. QPCR analysis of genes in whole blood and PBMCs was performed as in monocytes. The refinement of models from qPCR data of all 85 patients was performed as described in S1 Appendix. Markers of unstable CAD PLOS ONE | https://doi.org/10.1371/journal.pone.0225621 December 10, 2019 3 / 18 Coronary atherosclerotic plaques Then, we validated the expression in coronary atherosclerotic plaques. We performed blinded qPCR analysis of extracts from coronary atherosclerotic plaques of 7 patients (patients 1 and 2 and 4–8) and 5 (out of 6) control samples, collected at University of Padova, Italy, described in reference [9]. This study conforms to the principles outlined in the Declaration of Helsinki. The vascular samples were classified according American Heart Association recommendation [10]. Five were classified as type 7 and two as type 6. Controls were derived from the fragment of coronary samples not affected by plaque formation and with similar expression of several reference genes. We also measured cytokines, chemokines and growth factors in the plasma of these patients and controls. Peripheral vascular plaques Gene expressions patterns in peripheral arterial samples were analyzed previously with Illumina HumanHT-12 v3 Expression BeadChip (Illumina, San Diego, CA) analyzing 47,000 transcripts of all known genes, gene candidates, and splice variants in the Tampere Vascular Study [11–13]. This study conforms to the principles outlined in the Declaration of Helsinki. In the present study, extracts from 29 carotid, 15 abdominal and 24 femoral atherosclerotic plaques and 24 non-atherosclerotic left internal thoracic artery control samples were analyzed in detail. The samples were taken from patients subjected to open vascular surgical procedures in the Division of Vascular Surgery and Heart Centre, Tampere University Hospital. The study has been approved by the Ethics Committee of Tampere Hospital District. Three plaques were classified as type 3, 4 as type 4, 20 as type 5 and 30 as type 6. The type of 11 plaques could not be determined. Even though the results from plaques were not replicated with qPCR, the accuracy of the RNA micro-array array has been verified by TaqMan qPCR for many other genes in samples from the Tampere Vascular Study [14]. Other measurements Total and HDL-cholesterol and triglyceride levels were determined with enzymatic methods (Boehringer Mannheim). LDL-cholesterol levels were calculated with the Friedewald formula. Plasma glucose was measured with the glucose oxidase method (on Vitros 750XRC, Johnson & Johnson). Hs-CRP (Beckman Coulter) was measured on an Image 800 Immunochemistry System; Hs-Troponin T (TNT) on a Modular E system (Roche Diagnostics). All laboratory assessments were performed without knowledge of clinical data. Diabetes mellitus was defined as fasting serum glucose levels >125 mg/dl or therapy with oral hypoglycaemic agents or insulin. Plasma was used to analyze cytokine, chemokines and growth factors using the Bioplex instrumentation (Bio-Rad). We used the human cytokine 27-plex panel according to the manufacturer’s specificity [9]. Cytokines were analyzed with Assayfit Pro (AssayCloud, Netherlands) using a five parameters logistic regression for the interpolation of calibration points. Statistical analysis Two groups of continuous variables were compared with Mann Whitney test; three groups with Kruskal–Wallis test followed by Dunn’s multiple comparisons test. Categorical data were compared with Fisher’s exact test. RNA expressions were compared with paired Friedman test followed by Dunn’s multiple comparisons test. Non-parametric correlations between expression in monocytes and atherosclerotic plaques or between expression in monocytes and blood parameters were determined (GraphPad Prism 6). ROC, Kaplan-Meier and COX proportional hazards regression analysis (MedCalc statistical software) was performed to determine the Markers of unstable CAD PLOS ONE | https://doi.org/10.1371/journal.pone.0225621 December 10, 2019 4 / 18 additive diagnostic value of selected genes in separating stable CAD patients with versus without new ischemic event to age, gender, (ex)-smoking, BMI, blood pressure, type-2 diabetes, HDLand LDL-cholesterol, triglycerides and hs-C-reactive protein. P-values of less than 0.05 were considered as statistically significant. We calculated power based on sample size, group means and standard deviations and sample ratio (at 1% error rate). Results Patients Baseline characteristics of patients with stable CAD with (n = 42) and without a new ischemic event (n = 24) and ACS (n = 19) are shown in Table 1. There was no baseline difference in stable CAD patients with versus without a new ischemic event. ACS patients had higher hs-CRP and LDL-cholesterol levels than the two groups of stable CAD patients. ACS patients were somewhat older than stable CAD patients who developed a new ischemic event and tended to be treated less often with a statin. Use of blood pressure lowering and antiplatelet drugs was not different. All ACS patients had received P2Y12 inhibitors and aspirin at time of hospitalization, before blood sampling. Table 1. Demographic and clinical characteristics of stable CAD patients according to new coronary event and ACS patients. Stable CAD Without event (n = 42) Stable CAD With event (n = 24) ACS (n = 19) P-value Follow-up (days) 1943±369 2058±247 (NS) -- -- Diseased vessels (0, 1, 2, 3; n) 0, 13, 16, 13 0, 8, 12, 4 2, 7, 7, 3 0.380 Age (years) 59±6.4 55±7.2 62±11 $ 0.016 Sex (n and % male) 36 (86) 22 (92) 15 (79) 0.492 SBP (mmHg) 142±19 142±20 140±21 0.890 DBP (mmHg) 80±12 80±14 83±13 0.651 Smoker; ex-smoker (n and %) 12;21 (29; 50) 7;11 (29; 46) 6;7 (31; 37) 0.897 BMI (kg/m 2 ) 27±2.9 26±4.0 28±4.6 0.222 Type 2 diabetes (n and %) 6 (14) 4 (17) 0 (0) 0.188 TG (mg/dL) 149±92 111±38 147±85 0.149 LDL-C (mg/dL) 87±32 92±29 122±52��/ $ 0.0031 HDL-C (mg/dL) 44±12 46±12 46±12 0.895 Hs-CRP (mg/L) 3.09±2.66 2.28±2.28 8.56��/ $ $ 0.009 Hs-TNT (ng/μl) admission -- -- 0.27±0.43 -- Hs-TNT (ng/μl) peak -- -- 1.20±1.38 -- ACE-inhibitor use (n and %) 12 (29) 3 (13) 4 (21) 0.164 Angiotensin-II-receptor antagonist use (n and %) 6 (15) 4 (17) 1 (5.2) 0.317 Beta-blocker use (n and %) 21 (50) 11 (46) 7 (37) 0.634 Ca-antagonist use (n and %) 8 (19) 4 (17) 1 (5.2) 0.374 Statin use (n and %) 23 (55) 16 (67) 6 (32) 0.069 Antiplatelet drug (n and %) 29 (69) 19 (79) 9 (47) 0.082 Data shown are means ±SD. Abbreviations: BMI, body mass index; C, cholesterol; DBP, diastolic blood pressure; hs-CRP, high sensitivity C-reactive protein; hs-TNT: high sensitivity troponin T; SBP, systolic blood pressure; TG, triglycerides. Three groups of continuous variables were compared with Kruskal-Wallis followed by Dunn’s multiple comparison test; two groups were compared with Mann-Whitney test. Categorical data were compared with Fisher’s exact test. �� P<0.001 compared to stable CAD patients without event during follow-up $ P<0.05 and $ $ P<0.01 compared to stable CAD patients with event during follow-up. P-value for statin use comparing total group of stable CAD patients with ACS patients was 0.03. https://doi.org/10.1371/journal.pone.0225621.t001 Markers of unstable CAD PLOS ONE | https://doi.org/10.1371/journal.pone.0225621 December 10, 2019 5 / 18 RNA sequencing and validation of gene expression by qPCR analysis The full set of RNA sequencing data is available at https://www.ncbi.nlm.nih.gov/geo/query/ acc.cgi?acc=GSE129935. Analysis of RNA sequencing data identified 20 genes with differential expression according to the patient group (S1 Appendix). We validated their expression with qPCR (Table 2). MT-COI, COX10 and STRN were different between stable CAD patients with and without new ischemic event. RNA expression of COX10 was similar in ACS patients and stable CAD patients with new ischemic event. STRN increased from stable CAD patients without new ischemic event to stable CAD patients with ischemic event to ACS patients. RNA expressions of SMIM9, TET2, RNF121, KDM5A, PRRC2C, TAF1, WNK1, ZNF484, AAMP and STK6 were different in ACS patients than in the two groups of stable CAD patients; their RNA expressions were similar in stable CAD patients with and without new ischemic event (Table 2). MT-COI and STRN are associated with new ischemic event in stable CAD patients The gene expression heat map shows separation of stable CAD patients with and without new ischemic event according to the gene expression of MT-COI, STRN and COX10 determined by RNA sequencing (Fig 1). Table 2. RNA expressions, measured with qPCR, in monocytes from stable CAD patients according to new ischemic event and ACS patients. Gene Stable CAD Without event (n = 42) Stable CAD With event (n = 24) ACS (n = 19) P-value MT-COI 1.08±0.31 0.89±0.25�� 0.97±0.11 0.026 COX10 1.15±0.39 0.99±0.23�0.94±0.13�� 0.028 STRN 0.93±0.22 1.05±0.21�1.28±0.21��� /$ $ <0.0001 SMIM19 0.86±0.37 0.86±0.38 1.11±0.17� /$ 0.027 TET2 0.84±0.39 0.86±0.30 1.12±0.19�� /$ 0.0083 RNF121 0.86±0.32 0.86±0.37 1.12±0.16�� /$ 0.0082 KDM5A 0.77±0.41 0.77±0.31 1.12±0.18��� /$ $ 0.0007 PRRC2C 0.77±0.39 0.87±0.35 1.34±0.37��� /$ $ $ <0.0001 TAF1 0.75±0.35 0.77±0.37 1.26±0.28��� /$ $ $ <0.0001 WNK1 0.81±0.31 0.87±0.28 1.35±0.23��� /$ $ $ <0.0001 ZNF484 0.84±0.47 0.82±0.46 1.68±0.39��� /$ $ $ <0.0001 AAMP 1.01±0.15 1.03±0.16 0.87±0.13��� /$ $ <0.001 STK36 0.99±0.20 1.05±0.19 1.18±0.15��� /$ $ <0.001 Data shown are means ±SD of ratios compared to healthy controls. Three groups of continuous variables were compared with Kruskal-Wallis followed by Dunn’s multiple comparison tests. �P<0.05 �� P<0.01 ��� P<0.001 compared to stable CAD patients without event during follow-up $ P<0.05 $ $ P<0.01 $ $ $ P<0.001 compared to stable CAD patients with event during follow-up. Abbreviations: AAMP: angio associated migratory cell protein; COX10: cytochrome oxidase 10; KDM5A: lysine demethylase 5A; MT-COI: mitochondrial cytochrome c oxidase, subunit I; PRRC2C: proline rich coiled-coil 2C; RNF121: ring finger protein 121; STK36: serine threonine kinase 36; SMIM19: small integrated membrane protein 19; STRN: striatin; TAF1: TATA-box binding protein associated factor 1; TET2: tet methylcytosine dioxygense 2; WNK1: WNK lysine deficient protein kinase 1; ZNF484: zinc finger protein 484. https://doi.org/10.1371/journal.pone.0225621.t002 Markers of unstable CAD PLOS ONE | https://doi.org/10.1371/journal.pone.0225621 December 10, 2019 6 / 18 ROC analysis confirmed that MT-COI, STRN and COX10 RNA are related to new ischemic event in CAD patients (Fig 2). Power for separation of patients with and without new ischemic event was 0.97 for MT-COI (at 1% error rate), 0.90 for STRN and 0.77 for COX10. The accuracy for separating stable CAD patients with and without a new ischemic event increased from 67 to 80% by adding MT-COI and STRN to established cardiovascular risk factors age, gender, (ex)-smoking, Fig 1. Gene expression heat map. Stable CAD patients with and without new ischemic event are separated according to gene expression of MT-COI, STRN and COX10 determined by RNA sequencing. https://doi.org/10.1371/journal.pone.0225621.g001 Markers of unstable CAD PLOS ONE | https://doi.org/10.1371/journal.pone.0225621 December 10, 2019 7 / 18 BMI, blood pressure, type-2 diabetes, HDLand LDL-cholesterol, triglycerides and hs-C-reactive protein. AUC increased from 0.66 (0.53–0.77) to 0.84 (0.73–0.92). Kaplan-Meier curves showed that MT-COI and STRN were related with the time between blood sampling and occurrence of new ischemic event (Fig 3). COX proportional hazards regression analysis confirmed that MT-COI and STRN, but not COX10, were independently related to time to new event adjusting for age, gender, (ex)-smoking, BMI, blood pressure, type-2 diabetes, HDLand LDL-cholesterol, triglycerides and hs-Creactive protein. HR was 5.94 (95% CI: 2.47–14) for MT-COI and 4.47 (95% CI: 1.75–11) for STRN. That COX10 was not independently related may be due to the high correlation between MT-COI and COX10 (Rs = 0.45; P<0.0001). MT-COI, WNK1, STRN, COX10 and ZNF484 separate ACS from stable CAD patients The gene expression heat map shows complete separation of ACS from stable CAD patients with and without new ischemic event according to the gene expression of MT-COI, WNK1, STRN, COX10 and ZNF484 determined by RNA sequencing (Fig 4). Finally, these four genes significantly improved the clustering of patients into three groups compared to the whole collection of genes with significant differences in expression in the 3 groups in RNA-Seq analysis (Fig 5). We then performed modelling in search of minimal gene groups which discriminated between ACS patients and the whole group of stable CAD patients, independent of a new Fig 2. Receiver operating characteristics curve (ROC) analysis. RNA expressions of MT-COI, STRN and COX10 are related to occurrence of new ischemic event in stable CAD patients. Expressions were measured with qPCR. The area under curve was 0.69 (95% CI: 0.57–0.80) for MT-COI (circles), 0.66 (95% CI: 0.53–0.77) for STRN (triangles), and 0.63 (95% CI: 0.51–0.75) for COX10 (diamonds). The ROC curves were not statistically different. https://doi.org/10.1371/journal.pone.0225621.g002 Markers of unstable CAD PLOS ONE | https://doi.org/10.1371/journal.pone.0225621 December 10, 2019 8 / 18 ischemic event. We identified three models containing 3 genes which distinguished ACS from the whole group of CAD patients with an accuracy of 90%; COX10 and ZNF484 were present in all 3 models. The minimal model, with an accuracy of 90% consisted of COX10 and ZNF484. The sensitivity of COX10 and ZNF484 was 89% (95% CI: 52–99%). The specificity was 91% (95% CI: 52–99%). AUC was 0.96 (95% CI: 0.89–0.99) (Table 3). We then determined if RNA expressions correlated with TNT peak levels in ACS patients. MT-COI (Rs = -0.045; p = 0.86), COX10 (Rs = -0.12; p = 0.61), STRN (Rs = -0.21; p = 0.39), WNK1 (Rs = -0.041; p = 0.87) and ZNF484 (Rs = -0.044; p = 0.86) did not correlate with TNT. Comparison of gene expression in monocytes with that in PBMCs and whole blood Originally, we measured gene expression in monocytes because they are the precursors of macrophages. However, dependency on monocytes may hamper further validation, because their isolation is difficult to implement in clinical practice. Therefore, we also measured gene expression in whole blood and PBMCs of the same patients (n = 20). The expressions of MT-COI (0.71±0.10 vs. 0.88±0.11), COX10 (1.06±0.16 vs. 0.99±0.14), STRN (1.04±0.19 vs. 1.02±0.13) and WNK1 (0.95±0.23 vs. 1.14±0.15), but not ZNF484 (1.65±0.44 vs. 0.76±0.14), in PBMCs matched with these in monocytes. The expressions of COX10 (1.11±0.23), STRN (1.04±0.22) and WNK1 (0.8800780.26) in whole blood corresponded to these in monocytes; that of MTCOI did not. RNA expression in monocytes is independent of cholesterol lowering and antiplatelet treatment Finally, we divided CAD patients according to medical therapies. The expression of three RNA tertiles of MT-COI (χ 2 = 3.78; p = 0.15), COX10 (χ 2 = 1.92; p = 0.38), STRN (χ 2 = 1.37; p = 0.50), WNK1 (χ 2 = 2.75; p = 0.25) and ZNF484 (χ 2 = 0.014; p = 0.99) did not depend on statin use. The number of patients in the three tertiles of MT-COI (χ 2 = 2.27; p = 0.32), Fig 3. Kaplan-Meier analysis. Kaplan-Meier curves show that low RNA expression of MT-COI (left) and high RNA expression of STRN (right) were related with a shorter time between blood sampling and occurrence of new ischemic event. Adjusted HR, determined by COX proportional hazards regression analysis, of a new ischemic event in patients in lowest compared to patients in highest tertile of MT-COI was 5.14 (95% CI: 1.85–14). Adjusted HR of a new ischemic event in patients in highest compared to patients in lowest tertile of STRN was 2.88 (95% CI: 1.06–7.87). Adjustment was made for age, gender, (ex)-smoking, BMI, blood pressure, type-2 diabetes, HDLand LDL-cholesterol, triglycerides and hs-C-reactive protein. https://doi.org/10.1371/journal.pone.0225621.g003 Markers of unstable CAD PLOS ONE | https://doi.org/10.1371/journal.pone.0225621 December 10, 2019 9 / 18 References 1. Lenfant C. Chest pain of cardiac and noncardiac origin. Metabolism. 2010;59 Suppl 1:S41–6. Epub 2010/09/21. https://doi.org/10.1016/j.metabol.2010.07.014 PMID: 20837193. 2. Sinnaeve PR, Donahue MP, Grass P, Seo D, Vonderscher J, Chibout SD, et al. Gene expression patterns in peripheral blood correlate with the extent of coronary artery disease. PLoS One. 2009; 4(9): e7037. Epub 2009/09/15. https://doi.org/10.1371/journal.pone.0007037 PMID: 19750006; PubMed Central PMCID: PMC2736586. 3. Wallace DC. A mitochondrial paradigm of metabolic and degenerative diseases, aging, and cancer: a dawn for evolutionary medicine. Annu Rev Genet. 2005; 39:359–407. Epub 2005/11/16. https://doi.org/ 10.1146/annurev.genet.39.110304.095751 PMID: 16285865; PubMed Central PMCID: PMC2821041. 4. Wallace DC. The mitochondrial genome in human adaptive radiation and disease: on the road to therapeutics and performance enhancement. Gene. 2005; 354:169–80. Epub 2005/07/19. https://doi.org/10. 1016/j.gene.2005.05.001 PMID: 16024186. 5. Holvoet P, Vanhaverbeke M, Bloch K, Baatsen P, Sinnaeve P, Janssens S. Low MT-CO1 in Monocytes and Microvesicles Is Associated With Outcome in Patients With Coronary Artery Disease. J Am Heart Assoc. 2016; 5(12). Epub 2016/12/07. https://doi.org/10.1161/JAHA.116.004207 PMID: 27919931; PubMed Central PMCID: PMC5210432. 6. Chapman AR, Lee KK, McAllister DA, Cullen L, Greenslade JH, Parsonage W, et al. Association of High-Sensitivity Cardiac Troponin I Concentration With Cardiac Outcomes in Patients With Suspected Acute Coronary Syndrome. JAMA. 2017; 318(19):1913–24. Epub 2017/11/12. https://doi.org/10.1001/ jama.2017.17488 PMID: 29127948; PubMed Central PMCID: PMC5710293. 7. Park KC, Gaze DC, Collinson PO, Marber MS. Cardiac troponins: from myocardial infarction to chronic disease. Cardiovasc Res. 2017; 113(14):1708–18. Epub 2017/10/11. https://doi.org/10.1093/cvr/ cvx183 PMID: 29016754; PubMed Central PMCID: PMC5852618. 8. Svetnik V, Liaw A, Tong C, Culberson JC, Sheridan RP, Feuston BP. Random forest: a classification and regression tool for compound classification and QSAR modeling. J Chem Inf Comput Sci. 2003; 43 (6):1947–58. Epub 2003/11/25. https://doi.org/10.1021/ci034160g PMID: 14632445. 9. Cagnin S, Biscuola M, Patuzzo C, Trabetti E, Pasquali A, Laveder P, et al. Reconstruction and functional analysis of altered molecular pathways in human atherosclerotic arteries. BMC Genomics. 2009; 10:13. Epub 2009/01/13. https://doi.org/10.1186/1471-2164-10-13 PMID: 19134193; PubMed Central PMCID: PMC2654039. 10. Stary HC, Chandler AB, Dinsmore RE, Fuster V, Glagov S, Insull W Jr., et al. A definition of advanced types of atherosclerotic lesions and a histological classification of atherosclerosis. A report from the Committee on Vascular Lesions of the Council on Arteriosclerosis, American Heart Association. Circulation. 1995; 92(5):1355–74. Epub 1995/09/01. https://doi.org/10.1161/01.cir.92.5.1355 PMID: 7648691. 11. Oksala N, Levula M, Airla N, Pelto-Huikko M, Ortiz RM, Jarvinen O, et al. ADAM-9, ADAM-15, and ADAM-17 are upregulated in macrophages in advanced human atherosclerotic plaques in aorta and carotid and femoral arteries—Tampere vascular study. Ann Med. 2009; 41(4):279–90. Epub 2009/03/ 03. https://doi.org/10.1080/07853890802649738 PMID: 19253070. 12. Niinisalo P, Oksala N, Levula M, Pelto-Huikko M, Jarvinen O, Salenius JP, et al. Activation of indoleamine 2,3-dioxygenase-induced tryptophan degradation in advanced atherosclerotic plaques: Tampere vascular study. Ann Med. 2010; 42(1):55–63. Epub 2009/11/28. https://doi.org/10.3109/ 07853890903321559 PMID: 19941414. 13. Oksala N, Parssinen J, Seppala I, Raitoharju E, Kholova I, Hernesniemi J, et al. Association of neuroimmune guidance cue netrin-1 and its chemorepulsive receptor UNC5B with atherosclerotic plaque expression signatures and stability in human(s): Tampere Vascular Study (TVS). Circ Cardiovasc Genet. 2013; 6(6):579–87. Epub 2013/10/15. https://doi.org/10.1161/CIRCGENETICS.113.000141 PMID: 24122613. 14. Raitoharju E, Seppala I, Lyytikainen LP, Levula M, Oksala N, Klopp N, et al. A comparison of the accuracy of Illumina HumanHT-12 v3 Expression BeadChip and TaqMan qRT-PCR gene expression results in patient samples from the Tampere Vascular Study. Atherosclerosis. 2013; 226(1):149–52. Epub 2012/11/28. https://doi.org/10.1016/j.atherosclerosis.2012.10.078 PMID: 23177970. 15. Poyton RO, McEwen JE. Crosstalk between nuclear and mitochondrial genomes. Annu Rev Biochem. 1996; 65:563–607. Epub 1996/01/01. https://doi.org/10.1146/annurev.bi.65.070196.003023 PMID: 8811190. 16. Villani G, Attardi G. In vivo control of respiration by cytochrome c oxidase in wild-type and mitochondrial DNA mutation-carrying human cells. Proc Natl Acad Sci U S A. 1997; 94(4):1166–71. Epub 1997/02/18. https://doi.org/10.1073/pnas.94.4.1166 PMID: 9037024; PubMed Central PMCID: PMC19762. Markers of unstable CAD PLOS ONE | https://doi.org/10.1371/journal.pone.0225621 December 10, 2019 16 / 18 17. Holvoet P, Lee DH, Steffes M, Gross M, Jacobs DR Jr., Association between circulating oxidized lowdensity lipoprotein and incidence of the metabolic syndrome. JAMA. 2008; 299(19):2287–93. Epub 2008/05/22. https://doi.org/10.1001/jama.299.19.2287 PMID: 18492970; PubMed Central PMCID: PMC2562739. 18. Park K, Gross M, Lee DH, Holvoet P, Himes JH, Shikany JM, et al. Oxidative stress and insulin resistance: the coronary artery risk development in young adults study. Diabetes Care. 2009; 32(7):1302–7. Epub 2009/04/25. https://doi.org/10.2337/dc09-0259 PMID: 19389821; PubMed Central PMCID: PMC2699736. 19. Acin-Perez R, Bayona-Bafaluy MP, Bueno M, Machicado C, Fernandez-Silva P, Perez-Martos A, et al. An intragenic suppressor in the cytochrome c oxidase I gene of mouse mitochondrial DNA. Hum Mol Genet. 2003; 12(3):329–39. Epub 2003/01/30. https://doi.org/10.1093/hmg/ddg021 PMID: 12554686. 20. Fukuda R, Zhang H, Kim JW, Shimoda L, Dang CV, Semenza GL. HIF-1 regulates cytochrome oxidase subunits to optimize efficiency of respiration in hypoxic cells. Cell. 2007; 129(1):111–22. Epub 2007/04/ 10. https://doi.org/10.1016/j.cell.2007.01.047 PMID: 17418790. 21. Holvoet P, Vanhaverbeke M, Geeraert B, De Keyzer D, Hulsmans M, Janssens S. Low Cytochrome Oxidase 1 Links Mitochondrial Dysfunction to Atherosclerosis in Mice and Pigs. PLoS One. 2017; 12(1): e0170307. Epub 2017/01/26. https://doi.org/10.1371/journal.pone.0170307 PMID: 28122051; PubMed Central PMCID: PMC5266248. 22. Diaz F, Thomas CK, Garcia S, Hernandez D, Moraes CT. Mice lacking COX10 in skeletal muscle recapitulate the phenotype of progressive mitochondrial myopathies associated with cytochrome c oxidase deficiency. Hum Mol Genet. 2005; 14(18):2737–48. Epub 2005/08/17. https://doi.org/10.1093/hmg/ ddi307 PMID: 16103131; PubMed Central PMCID: PMC2778476. 23. Tarasenko TN, Pacheco SE, Koenig MK, Gomez-Rodriguez J, Kapnick SM, Diaz F, et al. Cytochrome c Oxidase Activity Is a Metabolic Checkpoint that Regulates Cell Fate Decisions During T Cell Activation and Differentiation. Cell Metab. 2017; 25(6):1254–68 e7. Epub 2017/06/08. https://doi.org/10.1016/j. cmet.2017.05.007 PMID: 28591633; PubMed Central PMCID: PMC5562283. 24. Tan H, Yang K, Li Y, Shaw TI, Wang Y, Blanco DB, et al. Integrative Proteomics and Phosphoproteomics Profiling Reveals Dynamic Signaling Networks and Bioenergetics Pathways Underlying T Cell Activation. Immunity. 2017; 46(3):488–503. Epub 2017/03/14. https://doi.org/10.1016/j.immuni.2017. 02.010 PMID: 28285833; PubMed Central PMCID: PMC5466820. 25. Geng S, Chen K, Yuan R, Peng L, Maitra U, Diao N, et al. The persistence of low-grade inflammatory monocytes contributes to aggravated atherosclerosis. Nat Commun. 2016; 7:13436. Epub 2016/11/09. https://doi.org/10.1038/ncomms13436 PMID: 27824038; PubMed Central PMCID: PMC5105176. 26. Hulsmans M, Geeraert B, Arnould T, Tsatsanis C, Holvoet P. PPAR agonist-induced reduction of Mcp1 in atherosclerotic plaques of obese, insulin-resistant mice depends on adiponectin-induced Irak3 expression. PLoS One. 2013; 8(4):e62253. Epub 2013/04/27. https://doi.org/10.1371/journal.pone. 0062253 PMID: 23620818; PubMed Central PMCID: PMC3631170. 27. Schieffer B, Luchtefeld M. Emerging role of chemokine receptor 7 in atherosclerosis. Trends Cardiovasc Med. 2011; 21(8):211–6. Epub 2012/08/21. https://doi.org/10.1016/j.tcm.2012.05.012 PMID: 22902067. 28. Mueller PA, Zhu L, Tavori H, Huynh K, Giunzioni I, Stafford JM, et al. Deletion of Macrophage Low-Density Lipoprotein Receptor-Related Protein 1 (LRP1) Accelerates Atherosclerosis Regression and Increases C-C Chemokine Receptor Type 7 (CCR7) Expression in Plaque Macrophages. Circulation. 2018; 138(17):1850–63. Epub 2018/05/26. https://doi.org/10.1161/CIRCULATIONAHA.117.031702 PMID: 29794082; PubMed Central PMCID: PMC6343494. 29. Lahav-Ariel L, Caspi M, Nadar-Ponniah PT, Zelikson N, Hofmann I, Hanson KK, et al. Striatin is a novel modulator of cell adhesion. FASEB J. 2018:fj201801882R. Epub 2018/12/29. https://doi.org/10.1096/fj. 201801882R PMID: 30592649. 30. Garza AE, Pojoga LH, Moize B, Hafiz WM, Opsasnick LA, Siddiqui WT, et al. Critical Role of Striatin in Blood Pressure and Vascular Responses to Dietary Sodium Intake. Hypertension. 2015; 66(3):674–80. Epub 2015/07/15. https://doi.org/10.1161/HYPERTENSIONAHA.115.05600 PMID: 26169051; PubMed Central PMCID: PMC4537321. 31. Zheng S, Chen X, Hong S, Long L, Xu Y, Simoncini T, et al. 17beta-Estradiol inhibits vascular smooth muscle cell migration via up-regulation of striatin protein. Gynecol Endocrinol. 2015; 31(8):618–24. Epub 2015/07/30. https://doi.org/10.3109/09513590.2015.1021325 PMID: 26220767. 32. Lu Q, Pallas DC, Surks HK, Baur WE, Mendelsohn ME, Karas RH. Striatin assembles a membrane signaling complex necessary for rapid, nongenomic activation of endothelial NO synthase by estrogen receptor alpha. Proc Natl Acad Sci U S A. 2004; 101(49):17126–31. Epub 2004/12/01. https://doi.org/ 10.1073/pnas.0407492101 PMID: 15569929; PubMed Central PMCID: PMC534607. Markers of unstable CAD PLOS ONE | https://doi.org/10.1371/journal.pone.0225621 December 10, 2019 17 / 18 33. Bellefroid EJ, Poncelet DA, Lecocq PJ, Revelant O, Martial JA. The evolutionarily conserved Kruppelassociated box domain defines a subfamily of eukaryotic multifingered proteins. Proc Natl Acad Sci U S A. 1991; 88(9):3608–12. Epub 1991/05/01. https://doi.org/10.1073/pnas.88.9.3608 PMID: 2023909; PubMed Central PMCID: PMC51501. 34. Schirmer SH, Fledderus JO, van der Laan AM, van der Pouw-Kraan TC, Moerland PD, Volger OL, et al. Suppression of inflammatory signaling in monocytes from patients with coronary artery disease. J Mol Cell Cardiol. 2009; 46(2):177–85. Epub 2008/12/09. https://doi.org/10.1016/j.yjmcc.2008.10.029 PMID: 19059264. 35. Margolin JF, Friedman JR, Meyer WK, Vissing H, Thiesen HJ, Rauscher FJ 3rd. Kruppel-associated boxes are potent transcriptional repression domains. Proc Natl Acad Sci U S A. 1994; 91(10):4509–13. Epub 1994/05/10. https://doi.org/10.1073/pnas.91.10.4509 PMID: 8183939; PubMed Central PMCID: PMC43815. 36. Friedman JR, Fredericks WJ, Jensen DE, Speicher DW, Huang XP, Neilson EG, et al. KAP-1, a novel corepressor for the highly conserved KRAB repression domain. Genes Dev. 1996; 10(16):2067–78. Epub 1996/08/15. https://doi.org/10.1101/gad.10.16.2067 PMID: 8769649. 37. Wolf G, Greenberg D, Macfarlan TS. Spotting the enemy within: Targeted silencing of foreign DNA in mammalian genomes by the Kruppel-associated box zinc finger protein family. Mob DNA. 2015; 6:17. Epub 2015/10/06. https://doi.org/10.1186/s13100-015-0050-8 PMID: 26435754; PubMed Central PMCID: PMC4592553. 38. Ridker PM. Canakinumab for Residual Inflammatory Risk. Eur Heart J. 2017; 38(48):3545–8. Epub 2017/12/28. https://doi.org/10.1093/eurheartj/ehx723 PMID: 29281019. 39. Ridker PM, Everett BM, Thuren T, MacFadyen JG, Chang WH, Ballantyne C, et al. Antiinflammatory Therapy with Canakinumab for Atherosclerotic Disease. N Engl J Med. 2017; 377(12):1119–31. Epub 2017/08/29. https://doi.org/10.1056/NEJMoa1707914 PMID: 28845751. 40. Powell L, Wiederkehr RS, Damascus P, Fauvart M, Buja F, Stakenborg T, et al. Rapid and sensitive detection of viral nucleic acids using silicon microchips. Analyst. 2018; 143(11):2596–603. Epub 2018/ 05/10. https://doi.org/10.1039/c8an00552d PMID: 29741175. 41. Cornelis S, Fauvart M, Gansemans Y, Vander Plaetsen AS, Colle F, Wiederkehr RS, et al. Multiplex STR amplification sensitivity in a silicon microchip. Sci Rep. 2018; 8(1):9853. Epub 2018/07/01. https:// doi.org/10.1038/s41598-018-28229-9 PMID: 29959383; PubMed Central PMCID: PMC6026139. 42. Wang H, Zhang X, Xu X, Zhang Q, Wang H, Li D, et al. A portable microfluidic platform for rapid molecular diagnostic testing of patients with myeloproliferative neoplasms. Sci Rep. 2017; 7(1):8596. Epub 2017/08/19. https://doi.org/10.1038/s41598-017-08674-8 PMID: 28819248; PubMed Central PMCID: PMC5561050. 43. Goldstein BA, Navar AM, Carter RE. Moving beyond regression techniques in cardiovascular risk prediction: applying machine learning to address analytic challenges. Eur Heart J. 2017; 38(23):1805–14. Epub 2016/07/21. https://doi.org/10.1093/eurheartj/ehw302 PMID: 27436868; PubMed Central PMCID: PMC5837244. 44. Weng SF, Reps J, Kai J, Garibaldi JM, Qureshi N. Can machine-learning improve cardiovascular risk prediction using routine clinical data? PLoS One. 2017; 12(4):e0174944. Epub 2017/04/05. https://doi. org/10.1371/journal.pone.0174944 PMID: 28376093; PubMed Central PMCID: PMC5380334. 45. Mandl KD, Manrai AK. Potential Excessive Testing at Scale: Biomarkers, Genomics, and Machine Learning. JAMA. 2019. Epub 2019/02/09. https://doi.org/10.1001/jama.2019.0286 PMID: 30735228. Markers of unstable CAD PLOS ONE | https://doi.org/10.1371/journal.pone.0225621 December 10, 2019 18 / 18