www.oncotarget.com Oncotarget, 2020, Vol. 11, (No. 10), pp: 905-912 Research Paper DNA methylation of MMPs and TIMPs in atherothrombosis process in carotid plaques and blood tissues Cristina Gallego-Fabrega1,2, Natalia Cullell1,2,3, Carolina Soriano-Tárraga4, Caty Carrera2,5, Nuria P. Torres-Aguila2, Elena Muiño2, Jara Cárcel-Márquez2, Manuel Castro de Moura6, Alba Fernández-Sanlés7, Manel Esteller6,8,9, Roberto Elosua7, Jordi Jiménez-Conde4, Jaume Roquer4, Joan Montaner5, Jerzy Krupinski1,10,* and Israel Fernandez-Cadenas2,* 1 Neurology, Hospital Universitari Mútua de Terrassa/Fundacio Docència i Recerca MutuaTerrassa, Terrassa, Spain 2 Stroke Pharmacogenomics and Genetics, Sant Pau Research Institute, Barcelona, Spain 3 Facultat de Medicina, Universitat de Barcelona, Barcelona, Spain 4 Department of Neurology, Hospital del Mar, Neurovascular Research Group, Institut Hospital del Mar d’Investigacions Mèdiques, Universitat Autònoma de Barcelona/DCEXS-Universitat Pompeu Fabra, Barcelona, Spain 5 Neurovascular Research Laboratory, Vall d'Hebron Institute of Research, Universitat Autònoma de Barcelona, Barcelona, Spain 6 Josep Carreras Leukaemia Research Institute (IJC), Badalona, Barcelona, Spain 7 Cardiovascular Epidemiology and Genetics Research Group, Hospital del Mar Medical Research Institute, Universitat Pompeu Fabra, Barcelona, Spain 8 Department of Physiological Sciences II, School of Medicine, University of Barcelona, Barcelona, Spain 9 Institucio Catalana de Recerca i Estudis Avançats, Barcelona, Spain 10 Centre for Biomedicine, Manchester Metropolitan University, Manchester, UK * These authors contributed equally to this work Correspondence to: Cristina Gallego-Fabrega, email: cristina.gallego.fabrega@gmail.com Jerzy Krupinski, email: jkrupinski@mutuaterrassa.es Israel Fernandez-Cadenas, email: israelcadenas@yahoo.es Keywords: atherosclerotic plaque; epigenetics; DNA methylation; matrix metalloproteinases Received: October 09, 2019 Accepted: January 13, 2020 Published: March 10, 2020 Copyright: Gallego-Fabrega et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License 3.0 (CC BY 3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. ABSTRACT Background and Purpose: Polymorphisms and serum levels of Matrix Metalloproteinases (MMP) and Tissue Inhibitor of Metalloproteinases (TIMP) have been studied with regard to atheromatous plaques and ischemic stroke, while no studies of DNA methylation (DNAm) patterns of MMP or TIMP have been performed to that end. Here, we evaluate DNAm levels of the MMP and TIMP gene families in human carotid plaques and blood samples of atherothrombotic stroke patients. Methods: We profiled the DNAm status of stable and ulcerated atherosclerotic plaques obtained as pair sets from three patients who underwent carotid endarterectomy surgery. We selected 415 CpG sites, mapping into MMPs and TIMPs genes for further study. Secondly, the statistically associated CpG sites were analyzed in blood samples from two separate atherothrombotic stroke cohorts (total sample size = 307), ischemic stroke-cohort 1 (ISC-1): 37 atherothrombotic patients and 6 controls, ischemic stroke-cohort 2 (ISC-2): 80 atherothrombotic patients and 184 controls. DNAm levels from plaque tissue and blood samples were evaluated using a high-density microarray Infinium, HumanMethylation450 BeadChip and Infinium MethylationEPIC BeadChip. Results: Three CpG sites were statistically significantly associated with unstable plaque portions; cg02969624, q-value = 0.035 (TIMP2), and cg04316754, q-value = www.oncotarget.com 905 Oncotarget 0.037 (MMP24) were hypermethylated, while cg24211657 q-value = 0.035 (TIMP2) was hypomethylated. Association of cg04316754 (MMP24) methylation levels with atherothrombotic risk was also observed in blood tissue: ISC-1 p-values = 0.03, ISC- 2 p-value = 1.9 × 10-04. Conclusions: The results suggest different DNAm status of MMP24 between stable and unstable atherothrombotic carotid plaques, and between atherothrombotic stroke and controls in blood samples. INTRODUCTION RESULTS Arteriosclerosis is the underlying pathology in most Seventy-two differentially methylated CpGs were cases of cardiovascular disease (CVD), including ischemic observed between normal and ulcerated plaque portions stroke (IS), contributing to major mortality in Western (p-value < 0.05) (Supplementary Table 3). Sixty of them countries. Atherosclerosis is a process that involves a were hypermethylated in the ulcerated portion, and complex interaction between different factors and cell twelve were hypomethylated. Three CpGs passed FDR types, including cells of the vessel wall and immune adjusted p-value (or q-value), two were hypermethylated: system. During their formation, atherosclerotic lesions cg02969624, q-value = 0.036 mapping in TIMP2, and undergo different stages, starting from inflammatory cg04316754, q-value = 0.037 mapping in MMP24; endothelial activation/dysfunction and resulting in plaque whereas cg24211657 q-value = 0.036 mapping in TIMP2 vulnerability and rupture [1]. was hypomethylated (Figure 1B) (Supplementary Table 4). Matrix Metalloproteinases (MMPs) are a family Supervised clustering of the 72 significant sites of zinc-binding proteolytic enzymes that are known for distinguished all atherosclerotic lesions from their donor- their ability to cleave one or more extracellular matrix matched healthy counterparts (Figure 2). Good clustering constituents, as well as other proteins [2]. MMPs and their was also observed when using all 415 CpGs, highlighting specific inhibitors, tissue inhibitors of metalloproteinases the specificity of the MMPs and TIMPs in plaque (TIMPs), are involved in many processes, including progression discrimination (Supplementary Figure 2). wound healing, angiogenesis, inflammation, and blood- All CpGs were also interrogated in two independent brain barrier disruption [2, 3]. Uncontrolled expression of blood sample ischemic stroke cohorts of atherothrombotic MMPs can result in tissue destruction and inflammation. stroke subtype (TOAST classification), 415 CpGs in ISC-1 MMPs and TIMPs have raised considerable interest within and 355 CpGs in ISC-2 (Supplementary Table 3). One of the the atherosclerosis and IS research community, as they significantly associated CpGs identified in the initial plaque represent an attractive target for the use of current drugs study was also associated with the risk of atherothrombotic and the development of novel ones, aimed at blocking stroke. Thus, cg04316754, MMP24, was up-methylated in MMP activity [4]. control samples when compared to cases in both cohorts. In Some studies have observed associations between ISC-1, we found a significant association of cg04316754 protein plasma levels of MMPs, TIMPs, atheromatous (Moldel_2 p-value = 0.025) that was also replicated in ISC- plaque instability [5, 6], stroke progression [7]. 2 (Model_1 p-value = 1.9 × 10-04) (Supplementary Table 5) However, few studies have investigated the relationship (Figure 1A). Thus, cg02969624 and cg24211657 (TIMP2) between DNA methylation (DNAm) and atherosclerosis were up-methylated in control samples when compared to pathogenicity [8–10] and none have focused on the MMP cases, but their association was not statistically significant and TIMP gene families. Zaina et al., observed widespread in either of the two ischemic cohorts. hypermethylation in the atherosclerotic portion of 15 An overall hypermethylation of MMPs and TIMPs aorta samples compared to healthy counterparts [9]. genes was observed in plaque tissue samples, with a In a different study, Zaina et al., detected small DNAm 65.7% of CpG up-methylated in the unstable portion changes between 19 symptomatic and asymptomatic of the plaque compared to the stable counterparts plaque pairs, and a drift toward hypomethylation, (Supplementary Figure 3A). The same behavior was associated with increasing post-cerebrovascular event detected when interrogating both blood-sample cohorts, time [8], whereas others identified a set of CpGs that drift with a 70.85% of CpGs up-methylated in case samples toward hypermethylation with lesion progression [10], in compared to atherothrombotic stroke patients in ISC- 15 sample-pairs. 1 and 63.48% of the CpG up-methylated in ISC-2 Here, we present a characterization of DNAm status (Supplementary Figure 3A). This trend was observed of the MMP and TIMP gene families in donor-matched across all gene mapping compartments, mostly in the stable and ulcerated carotid artery atherosclerotic plaques gene body and gene-TSS1500, and across all CpG islands and whole blood in atherothrombotic stroke patients and regions, as well as in sites mapping into enhancer regions controls. (Supplementary Figure 3B–3D). www.oncotarget.com 906 Oncotarget DISCUSSION the genes. Our results are in line with those of others, who observed a broad trend of DNA hypermethylation in The role of MMPs in atherosclerosis has been plaque progression [8–10]. Zaina et al., observed pattern extensively evaluated and their activity is essential of genome-wide hypermethylation in atherosclerotic for many processes involved in atherosclerotic plaque plaques, suggesting that an atherosclerosis-specific DNA formation, such as infiltration of inflammatory cells, methylation profile is established in the early stages of smooth muscle cell migration and proliferation and plaque evolution [9]. In a separate study, Zaina et al., angiogenesis [11]. Furthermore, matrix degradation by identified small DNAm changes between symptomatic and MMPs causes plaque instability and rupture leading to asymptomatic plaques and an association with increasing unstable angina, myocardial infarction and stroke [12]. post-cerebrovascular event time [8]. They observed a Müeller et al., identified increased expression levels of general hypermethylation in the plaque in early post- MMP1, MMP9, MMP12, and MMP14 in vulnerable cerebrovascular event time compared to asymptomatic plaques compared to stable ones, and lower expression plaque. In another study that analyzed donor-matched levels in MMP2 and TIMP3 vulnerable plaques [5]. In atherosclerotic and normal aortic samples, the authors contrast, Sapienza et al., observed an imbalance between observed a correlation between histological grade and MMPs and TIMPs plasma levels in unstable carotid differential methylation between plaque pairs for 1,985 plaques [6]. A recent multiancestry genome-wide- CpGs, most of which drifted toward hypermethylation association meta-analysis [7] identified a locus in MMP12 with lesion progression [10]. associated with atherothrombotic stroke, where the lead MMP24 is a protease, member of the membrane- SNP was also associated with methylation and protein type MMP (MT-MMP), a subfamily in the matrix quantitative trait loci (meQTL and pQTL) [7]. metalloproteinases-family, characterized by having a In our study, unstable plaque samples had 65.7% of transmembrane domain and being expressed at the cell CpG up-methylated compared to their stable counterparts, surface. Substrates of this protease include the proteins indicating a trend to hypermethylation. These findings cadherin-2 and MMP2. Here, we present the first reported were most evident at the body and TSS1500 regions of association between MMP24 and atherosclerosis, although Figure 1: Differentially methylated levels of cg04316754 (MMP24). (A) Boxplot: differentially methylated status of cg04316754. X-axis indicates sample type, in each cohort. Y-axis indicates DNA methylation levels (β-values = 0–1).0 = 0% methylation status, 1 = 100% methylation status. (B) Statistically significant differentially methylated CpG sites in plaque tissue, annotation and statistical results in the three cohorts. www.oncotarget.com 907 Oncotarget links with other MMPs in the MT-MMP subtype have an important role for MMP14 in plaque stability [13]. been described before. For instance, MMP14, a member While MMP24 was not assessed, a study reported, higher of the MT-MMP subfamily, has been implicated in acute gene expression levels of MMP14 in vulnerable plaques myocardial infarction. Moreover, expression of MMP14 than stable plaques, (no other member of MT-MMP can influence collagen content of mouse plaques, implying subfamily was interrogated) [5]. Figure 2: Cluster analysis of donor-matched aortic plaque samples. Supervised clustering of 72 CpGs nominal association. Columns represent samples, horizontal lines represent CpGs. Notice the perfect segregation between stable and unstable plaque counterpart. www.oncotarget.com 908 Oncotarget Our results indicate a distinct methylation state of ulcerated portions in plaque tissue for MMP24 and TIMP2. the CpG site cg0431675 in MMP24 between stable and Differences in MMP24 were also observed in blood unstable carotid plaque, which we have also detected samples between atherothrombotic stroke patients and associated with atherosclerotic stroke in two independent healthy controls. The generalized hypermethylation found cohorts. A trend to hypermethylation is observed in in ulcerated portions and samples from healthy controls is unstable plaque-portion and non-stroke control samples, in line with other methylation studies. Functional analysis which falls into line with what has been postulated by of the implications of methylation levels changes in others, i.e., that ruptured plaques tend to revert to a stable MMP24 and TIMP2 have in their expression levels are structure. Zaina et al., in their study, observed an initial needed to link this finding with the biopathology of plaque hypermethylation in the transition from asymptomatic destabilization. Studies with larger sample size, to confirm plaque to early post-cerebrovascular event time, followed our results. by a reversion to initial methylation levels with increasing post-cerebrovascular event time, suggesting a process Limitations of plaque remodeling to a more stable phenotype [8]. Peeters W et al., proposed that ruptured plaques remodel The study presents a small sample size, which might to a relatively stable structure after stroke events [14]. have limited the ability to find other potential biomarkers. They identified higher macrophage infiltration in carotid Although other studies have also performed epigenetic plaques, obtained from patients operated early after experiments in a limited number of atherothrombotic stroke, compared with asymptomatic patients, as well as plaques [10]. Using samples from the same patient significantly elevated levels of a set of proinflammatory decreases the variability, thus increasing the statistical cytokines and MMP8, MMP9, and MMP2 activity [14]. power. Moreover, our blood results corroborate that TIMP2, in addition to its role as metalloproteinase MMP24 may play a role in atherosclerosis due to inhibitor, can directly suppress the proliferation of epigenetic modifications. Additionally, having whole endothelial cells. TIMP2 is a rather large gene, with blood DNA samples derived from the same atheromatous 62 CpG sites of the 450k array mapping into the gene. plaque subjects would have given more insight into useful Two CpG sites, in different regions of TIMP2, were methylation biomarkers of plaque vulnerability. statistically associated with plaque progression in our study, although this association was not observed between MATERIALS AND METHODS atherothrombotic stroke and control samples, either in statistical significance or directionally. Therefore, we Human atherosclerotic plaque samples cannot conclude much from those results, except that the most known activity of TIMP2 is MMP2 regulation [11]. Three atherosclerotic plaque samples were obtained It has been reported that human carotid plaque extracts at Hospital Universitari Mutua de Terrassa (Barcelona, promote platelet aggregation due to their MMP2 content, Spain), according to a protocol approved by the local which can be inhibited by TIMP2, and that the ratio of ethics committee. Patients with >70% stenosis underwent MMP2/TIMP2 of plaques potentiating platelet aggregation carotid endarterectomy surgery, following European is significantly higher than that of plaques not potentiating Society for Vascular Surgery Guidelines [16]. Plaque it. Moreover, an elevated MMP2 activity in plaques and sizes ranged from 1.5 to 3.5 cm and the samples were a high aggregation-potentiating effect of plaques have stored at –80°C until the analysis. Stable and ulcerated been associated with a higher rate of subsequent ischemic portions of plaque were macroscopically identified and cerebrovascular events [15]. sectioned prior to DNA extraction by a trained technician Our results indicate that, there is a distinct (Supplementary Figure 1). Sample selection was based on methylation pattern of MMPs and TIMPs genes between the size and definition of stable and ulcerated portions. stable and unstable carotid plaque, that might play a Sample information for donor-matched plaque pairs is role in plaque instability. Three CpGs reached statistical shown in Supplementary Table 1. significance, a bigger sample could have boosted the power of the study to identify other CpGs. While there Human whole blood samples is no previous literature linking our primary association, MMP24, with atherosclerosis, a vast catalog of studies Methylation levels of MMPs and TIMPs in plaque on RNA expression, plasma levels studies, genetics tissue were crossed with methylation levels in blood and methylation [5–8, 13, 15] in the MMP gene family tissue in two new independent cohorts. Ischemic stroke- corroborate the biopathological plausibility of its role in cohort 1 (ISC-1): 37 atherothrombotic stroke patients plaque progression. and 6 healthy controls, from the GRECOS study [17]; In summary, we characterized MMPs and TIMPs Ischemic stroke-cohort 2 (ISC-2): 80 atherothrombotic DNAm patterns in atheromatous plaque, which led us patients from the BASICMAR prospective register [18] to observe significant differences between stable and and 184 healthy controls from REGICOR population- www.oncotarget.com 909 Oncotarget based cohort (Supplementary Table 2). In both cases, only not to rule out TIMP1 CpG sites, mapped in the X atherothrombotic stroke samples were selected as defined chromosome. Recommended BMIQ method was used by the Trial of Org 10172 in Acute Stroke Treatment for data normalization. We also adjusted our data by cell (TOAST) [19]. proportion, with champ. refbase function. Local ethics committee approved the study (PR (AG) Finally, Wilcoxon rank-sum test and the Champ. 03/2007). All patients signed a written informed consent. DMP function were used to identify differentially methylated positions among groups in the univariate DNA preparation and bisulfite conversion model (Model_1), and glm for the multivariate models. Multivariate model, Model_2, included sex, age, Genomic DNA was extracted from <10 mg of frozen smoking, variables, and Model_3 included sex, smoking, plaque tissue using the QIAamp DNA Micro Kit (Qiagen, hypertension (HTA) and dyslipidemia (DL). This pipeline Hilden, Germany) tissue protocol. Genomic DNA, from was used for the three cohorts, atherosclerotic plaque, fresh whole blood samples, was obtained using the ISC-1, and ISC-2, (except: ISC-2 cohort did remove the Gentra Puregene Blood Kit (Qiagen, Hilden, Germany), CpG sites mapped in the X chromosome). following the manufacturer’s instructions, in the Plaque- tissue samples and the Blood-tissue ISC-1 samples. Statistical analysis and power calculation Blood-tissue ISC-2 samples were extracted using manual salt precipitation in Banco Nacional de ADN (Instituto de The significance threshold, for all statistical tests, Salud Carlos III, Madrid, Spain). was set at False Discovery Rate (FDR) p < 0.05. Statistical analyses were performed using Bioconductor packages Methylation assays (http://www.bioconductor.org) and R software (http:// www.cran.r-project.org). Genome-wide DNAm was assessed using the Infinium As described by Tsai P et al., [22] we have an 80% HumanMethylation450 BeadChip (450K), in plaque and power to detect p < 0.05, considering Cohen’s d effect size ISC-1 samples, and Infinium MethylationEPIC (EPIC) in between 1.93–3.33 in the paired-plaque sample study and ISC-2 samples, (Illumina Inc, San Diego, CA, USA). over 95% to detect p < 0.05, considering Cohen’s d effect size between 1.47–1.80. CpG site selection Annotation information provided by Illumina’s Abbreviations 450K BeadChip manifest [20] was used to identify CpG sites (CpGs) mapped into the MMP and TIMP family of 450K: Infinium HumanMethylation450 BeadChip; genes. A total of 27 genes (23 MMPs and 4 TIMPs) were CpG: cytosine-phosphate-guanine; CpGs: CpG sites; identified, comprising a total of 464 CpGs, of which 415 CVD: Cardiovascular disease; DNAm: DNA methylation; passed quality controls (Supplementary Table 6). DL: Dyslipidemia; EPIC: Infinium MethylationEPIC; FDR: False discovery rate; HTA: Hypertension; IS: Ischemic Stroke; ISC-1: Ischemic stroke cohort 1; Data preprocessing, quality control (QC), and ISC-2: Ischemic stroke cohort 2; meQTL: Methylation analysis quantitative trait loci; MMP: Matrix Matalloproteinases; Functions from the ChAMP Bioconductor pQTL: Protein quantitative trait loci; QC: Quality control; package v.2.9.19 [21] were used for data preprocessing, TIMP: Tissue Inhibitor of Metalloproteinases; TOAST: QC, normalization and univariate analysis. Following Trial of Org 10172 in Acute Stroke Treatment; TSS1500: the ChAMP package pipeline recommendations, Transcription start site 1500. QC metrics were examined to determine the success of the bisulfite conversion and subsequent array Author contributions hybridization. Fluorescence intensities were imported from GenomeStudio, then probe filtering was performed CGF: wrote the paper; conceived and designed the to remove probes that had failed to hybridize (detection analysis; performed the analysis; interpretation of data. P > 0.05) and were not represented by a minimum of 3 NC: contributed data and analysis; interpretation of data. beads on the array. CpG sites containing documented CST: contributed data and analysis; interpretation of single-nucleotide polymorphisms were also excluded. data. CC: contributed data and analysis; interpretation of Multidimensional scaling plots were used to evaluate sex data. NPTA: contributed data and analysis. EM: critical outliers based on chromosome X data. Multidimensional revision; interpretation of data. JCM: contributed data and scaling and principal components were also used to analysis; interpretation of data. MCM: contributed analysis check for unknown population structures. Probes tools. AFS: contributed data. ME: contributed analysis mapping to X chromosome were not removed, so as tools. RE: collected data; critical revision. JJC: collected www.oncotarget.com 910 Oncotarget data; critical revision. JR: collected data; critical revision. 7. Malik R, Chauhan G, Traylor M, Sargurupremraj M, Okada JM: collected data; critical revision. JK: conceived and Y, Mishra A, Rutten-Jacobs L, Giese AK, van der Laan designed the analysis; collected data; critical revision; SW, Gretarsdottir S, Anderson CD, Chong M, Adams HH, equal contribution: IFC: conceived and designed the et al, and AFGen Consortium, and Cohorts for Heart and analysis; interpretation of data; critical revision; equal Aging Research in Genomic Epidemiology (CHARGE) contribution. Consortium, and International Genomics of Blood Pressure (iGEN-BP) Consortium, and INVENT Consortium, and CONFLICTS OF INTEREST STARNET, and BioBank Japan Cooperative Hospital Group, and COMPASS Consortium, and EPIC-CVD The authors have no conflicts of interest to declare. Consortium, and EPIC-InterAct Consortium, and International Stroke Genetics Consortium (ISGC), and METASTROKE Consortium, and Neurology Working FUNDING Group of the CHARGE Consortium, and NINDS Stroke Genetics Network (SiGN), and UK Young Lacunar DNA This study was funded by the INVICTUS network, Study, and MEGASTROKE Consortium. Multiancestry Generacion, Maestro EPIGENESIS, FEDER-ERDF, genome-wide association study of 520,000 subjects and BasicMar Regist projects from the Carlos III Health identifies 32 loci associated with stroke and stroke subtypes. Institute, the AGAUR, the RecerCaixa 2013 and the Nat Genet. 2018; 50:524–37. https://doi.org/10.1038/ European Regional Development Fund (ERDF). I. s41588-018-0058-3. [PubMed] Fernandez is recipient of a research contract from Miguel Servet Program from the Carlos III Health Institute 8. Zaina S, Gonçalves I, Carmona FJ, Gomez A, Heyn H, (CPII17/00021). EPIGENESIS project (CarlosIII Institute, Mollet IG, Moran S, Varol N, Esteller M. DNA methylation Marató TV3 and Fundació MútuaTerrassa). dynamics in human carotid plaques after cerebrovascular events. Arterioscler Thromb Vasc Biol. 2015; 35:1835–1842. https://doi.org/10.1161/ATVBAHA.115.305630. [PubMed] REFERENCES 9. Zaina S, Heyn H, Carmona FJ, Varol N, Sayols S, Condom 1. Lu H, Daugherty A. Atherosclerosis. Arterioscler Thromb E, Ramírez-Ruz J, Gomez A, Gonçalves I, Moran S, Esteller Vasc Biol. 2015; 35:485–491. https://doi.org/10.1161/ M. DNA methylation map of human atherosclerosis. ATVBAHA.115.305380. [PubMed] Circ Cardiovasc Genet. 2014; 7:692–700. https://doi. org/10.1161/CIRCGENETICS.113.000441. [PubMed] 2. Kapoor C, Vaidya S, Wadhwan V, Kaur G, Pathak A, and Hitesh. Seesaw of matrix metalloproteinases 10. Valencia-Morales MP, Zaina S, Heyn H, Carmona FJ, Varol (MMPs). J Cancer Res Ther. 2016; 12:28–35. https://doi. N, Sayols S, Condom E, Ramírez-Ruz J, Gomez A, Moran org/10.4103/0973-1482.157337. [PubMed] S, Lund G, Rodríguez-Ríos D, López-González G, et al. The DNA methylation drift of the atherosclerotic aorta increases 3. Chang JJ, Stanfill A, Pourmotabbed T. The Role of Matrix with lesion progression. BMC Med Genomics. 2015; 8:7. Metalloproteinase Polymorphisms in Ischemic Stroke. https://doi.org/10.1186/s12920-015-0085-1. [PubMed] Int J Mol Sci. 2016; 17:1323. https://doi.org/10.3390/ ijms17081323. [PubMed] 11. Gresele P, Falcinelli E, Sebastiano M, Momi S. Matrix Metalloproteinases and Platelet Function. Prog Mol Biol 4. Lapchak PA, Chapman DF, Zivin JA, Hsu CY. Transl Sci. 2017; 147:133–165. https://doi.org/10.1016/ Metalloproteinase inhibition reduces thrombolytic bs.pmbts.2017.01.002. [PubMed] (tissue plasminogen activator)-induced hemorrhage after thromboembolic stroke. Stroke. 2000; 31:3034–40. https:// 12. Newby AC. Metalloproteinases promote plaque rupture doi.org/10.1161/01.STR.31.12.3034. [PubMed] and myocardial infarction: A persuasive concept waiting 5. Müller A, Krämer SD, Meletta R, Beck K, Selivanova for clinical translation. Matrix Biol. 2015; 44–46:157–66. SV, Rancic Z, Kaufmann PA, Vos B, Meding J, Stellfeld https://doi.org/10.1016/j.matbio.2015.01.015. [PubMed] T, Heinrich TK, Bauser M, Hütter J, et al. Gene 13. Di Gregoli K, Jenkins N, Salter R, White S, Newby expression levels of matrix metalloproteinases in human AC, Johnson JL. MicroRNA-24 regulates macrophage atherosclerotic plaques and evaluation of radiolabeled behavior and retards atherosclerosis. Arterioscler Thromb inhibitors as imaging agents for plaque vulnerability. Nucl Vasc Biol. 2014; 34:1990–2000. https://doi.org/10.1161/ Med Biol. 2014; 41:562–569. https://doi.org/10.1016/j. ATVBAHA.114.304088. [PubMed] nucmedbio.2014.04.085. [PubMed] 14. Peeters W, Hellings WE, de Kleijn DP, de Vries JP, Moll 6. Sapienza P, di Marzo L, Borrelli V, Sterpetti AV, Mingoli FL, Vink A, Pasterkamp G. Carotid atherosclerotic plaques A, Cresti S, Cavallaro A. Metalloproteinases and their stabilize after stroke: insights into the natural process of inhibitors are markers of plaque instability. Surgery. 2005; atherosclerotic plaque stabilization. Arterioscler Thromb 137:355–363. https://doi.org/10.1016/j.surg.2004.10.011. Vasc Biol. 2009; 29:128–33. https://doi.org/10.1161/ [PubMed] ATVBAHA.108.173658. [PubMed] www.oncotarget.com 911 Oncotarget 15. Lenti M, Falcinelli E, Pompili M, de Rango P, Conti V, 19. Chung JW, Park SH, Kim N, Kim WJ, Park JH, Ko Y, Guglielmini G, Momi S, Corazzi T, Giordano G, Gresele P. Yang MH, Jang MS, Han MK, Jung C, Kim JH, Oh CW, Matrix metalloproteinase-2 of human carotid atherosclerotic Bae HJ. Trial of ORG 10172 in Acute Stroke Treatment plaques promotes platelet activation. Correlation with (TOAST) classification and vascular territory of ischemic ischaemic events. Thromb Haemost. 2014; 111:1089–1101. stroke lesions diagnosed by diffusion-weighted imaging. J https://doi.org/10.1160/TH13-07-0588. [PubMed] Am Heart Assoc. 2014; 3:e001119. https://doi.org/10.1161/ 16. Eckstein H. European Society for Vascular Surgery JAHA.114.001119. [PubMed] Guidelines on the Management of Atherosclerotic Carotid 20. Bibikova M, Barnes B, Tsan C, Ho V, Klotzle B, Le JM, and Vertebral Artery Disease. Eur J Vasc Endovasc Surg. Delano D, Zhang L, Schroth GP, Gunderson KL, Fan JB, 2018; 55:1–2. https://doi.org/10.1016/j.ejvs.2017.06.026. Shen R. High density DNA methylation array with single [PubMed] CpG site resolution. Genomics. 2011; 98:288–95. https:// 17. Fernández-Cadenas I, Mendióroz M, Giralt D, Nafria doi.org/10.1016/j.ygeno.2011.07.007. [PubMed] C, Garcia E, Carrera C, Gallego-Fabrega C, Domingues- 21. Tian Y, Morris TJ, Webster AP, Yang Z, Beck S, Feber A, Montanari S, Delgado P, Ribó M, Castellanos M, Martínez Teschendorff AE. ChAMP: updated methylation analysis S, Freijo M, et al, and GRECOS Study Group. GRECOS pipeline for Illumina BeadChips. Bioinformatics. 2017; Project (Genotyping Recurrence Risk of Stroke): The Use 33:3982–84. https://doi.org/10.1093/bioinformatics/btx513. of Genetics to Predict the Vascular Recurrence After Stroke. [PubMed] Stroke. 2017; 48:1147–1153. https://doi.org/10.1161/ 22. Tsai PC, Bell JT. Power and sample size estimation for STROKEAHA.116.014322. [PubMed] epigenome-wide association scans to detect differential 18. Roquer J, Rodríguez-Campello A, Gomis M, Jiménez- DNA methylation. Int J Epidemiol. 2015; 44:1429–41. Conde J, Cuadrado-Godia E, Vivanco R, Giralt E, https://doi.org/10.1093/ije/dyv041. [PubMed] Sepúlveda M, Pont-Sunyer C, Cucurella G, Ois A. Acute stroke unit care and early neurological deterioration in ischemic stroke. J Neurol. 2008; 255:1012–1017. https:// doi.org/10.1007/s00415-008-0820-z. [PubMed] www.oncotarget.com 912 Oncotarget