The proteoglycan mimecan is associated with carotid plaque vulnerability and increased risk of future cardiovascular death

Background and aims: A vulnerable plaque is an atherosclerotic plaque that is rupture-prone with a higher risk to cause cardiovascular symptoms such as myocardial infarction or stroke. Mimecan or osteoglycin is a small leucine-rich proteoglycan, important for collagen fibrillogenesis, that has been implicated in atherosclerotic disease, yet the role of mimecan in human atherosclerotic disease remains unknown. Methods: 196 human atherosclerotic carotid plaques were immunostained for mimecan. Smooth muscle cells, macrophages and intraplaque haemorrhage were also measured with immunohistochemistry. Neutral lipids were stained with Oil Red O and calcium deposits were quantified. Plaque homogenate levels of MCP-1, IL-6 and MIP- 1 β were measured using a Proximity Extension Assay and MMP-9 levels were measured using Mesoscale. Gly-cosaminoglycans, collagen and elastin were assessed by colorimetric assays and TGF- β 1, β 2 and β 3 were measured using a multiplex assay. Mimecan gene expression in THP-1 derived macrophages was quantified by qPCR and protein expression in vitro was visualized with immunofluorescence. Cardiovascular events were registered using medical charts and national registers during follow-up. Results: Mimecan correlated positively with plaque area of lipids, macrophages, intraplaque haemorrhage and inversely with smooth muscle cell staining. Mimecan also correlated positively with plaque levels of MMP-9 and MCP-1. Mimecan was upregulated in THP-1 derived macrophages upon stimulation with MCP-1. Patients with high levels of mimecan (above median) had higher risk for cardiovascular death. Conclusions: This study indicates that mimecan is associated with a vulnerable plaque phenotype, possibly regulated by plaque inflammation. In line, plaque levels of mimecan independently predict future cardiovascular death.


Introduction
The rupture or erosion of an atherosclerotic plaque with subsequent thrombosis leads to myocardial infarction or stroke, which are the most common causes of death globally [1]. Despite recent advances in preventive treatments, there is still a considerable unmet need for new treatment targets to stabilize atherosclerotic plaques, avoiding ruptures or erosions. A plaque that is prone to rupture, often called a vulnerable plaque, is characterized by a large lipid core, covered by a thin fibrous cap, with degraded extracellular matrix (ECM) proteins, poor in smooth muscle cells and rich in inflammatory infiltrates. In contrast, a stable plaque has a smaller lipid core covered by a thick and less inflamed fibrous cap, rich in smooth muscle cells and collagen fibers [2,3]. The balance between the degradation and formation of ECM components is of great importance for plaque stability. Yet, these processes are poorly understood in the context of human atherosclerosis.
Mimecan, also known as osteoglycin, is an ECM component, namely a small leucine-rich proteoglycan (SLRP). Mimecan affects several biological processes including the regulation of collagen fibrillogenesis and angiogenesis [4][5][6]. Mimecan is expressed in atherosclerotic tissue in rabbits, human coronary arteries and is downregulated in intimal vascular smooth muscle cells (VSCMs) [7,8]. Mimecan is upregulated in rat carotid artery after balloon injury suggesting that it is involved in vascular remodelling [7]. However, in humans the role of mimecan in atherosclerotic plaques is still unknown.
In the present study, we investigated the associations between mimecan and 1) human plaque components in a large number of atherosclerotic lesions obtained from patients undergoing carotid endarterectomy; 2) inflammation in an in vitro model of THP-1 derived macrophages; 3) the risk for future cardiovascular events and death.

Study design
One hundred and ninety-six human carotid plaques were obtained from 194 patients who underwent carotid endarterectomy between 2005 and 2011 at the Vascular Department at Skåne University Hospital (Malmö, Sweden). Two patients underwent surgery for both right and left carotid plaques and, in those cases, only the first chronological clinical data was included for the follow up analysis. All patients gave written and oral informed consent. The study was approved by the Regional Ethical Committee and conformed with the principles of the Declaration of Helsinki. The clinical characteristics of the patients are described in Table 1. The inclusion of patients was done consecutively. Missing data was not imputated and no outliers were removed from the statistical analysis. Endpoints for the follow up were predefined and the specific ICD-codes can be found in the "Follow up" section below. Appropriate statistical adjustments are described in the "Statistical method" section. The clinical data in this study was collected by research nurses at the time of inclusion and in vitro data was compiled by the technical and research staff. The research subjects are patients being operated for a carotid stenosis, with the criteria for operation specified below. The data was processed randomly and blinded to the technical staff.
All patients were examined with a preoperative ultrasound of the carotid arteries, to evaluate the degree of stenosis, and were clinically assessed by a neurologist before endarterectomy to exclude other causes of thromboembolic events, i.e. atrial fibrillation or flutter. Patients were considered symptomatic if they suffered from amaurosis fugax, transient ischemic attack (TIA) or ischemic stroke. The indication for carotid endarterectomy was 1) ipsilateral symptoms and stenosis >70% or 2) asymptomatic patient with >80% stenosis.

Sample preparation
Carotid plaques were snap-frozen in liquid nitrogen in the operation theatre directly after surgical removal and stored at − 80 • C until sectioning. From the most stenotic part of the plaque, a 1 mm thick section was taken for histology and embedded in optimal cutting medium (OCT, Sakura Finetek Europe BV, Japan). The rest of the plaque was homogenized in a standardised way as described previously [9].

Histology and immunohistochemistry
The most stenotic part of the plaque (1 mm) was cryosectioned in 8 μm sections for histology as previously described [9]. Sections were then fixed with Histochoice (Amresco, Solon, OH, USA) and stained for smooth muscle cells (smooth muscle α-actin), macrophages (CD68), neutral lipids (Oil Red O) and intraplaque haemorrhage (Glycophorin A) as previously described [9,10]. Calcified areas in plaques were quantified as described previously [11]. Collagen fibers (yellow) were assessed using Russell-Movat Pentachrome staining. Mimecan was analysed in 196 human carotid plaques using immunohistochemistry staining with a rabbit polyclonal antibody (PA5-48255, Invitrogen, Waltham, MA) at 2 μg/mL as a primary antibody and then a MACH3 rabbit probe and horseradish peroxidase-polymer (RP531H, Biocare Medical, Pacheco, CA) was used as a secondary antibody. A rabbit IgG polyclonal isotype control antibody was used as a control for each plaque (ab27478, Abcam, Cambridge, UK).

Plaque levels of matrix metalloproteinase 9
Matrix metalloproteinase-9 was analysed with Mesoscale human MMP ultra-sensitive kit (Mesoscale, Gaithersburg, MD, USA) in plaque homogenate supernatants as previously described [2]. All analyses were performed according to the manufacturer's instructions and the results were normalized to plaque wet weight.

Biochemical assessments of plaque glycosaminoglycans, collagen and elastin levels
The ECM components glycosaminoglycans, collagen and elastin were analysed in plaque homogenate with colorimetric assays and normalized to plaque wet weight as previously described [9].  MA, USA) according to the manufacturer's instructions and measured using Luminex 100 IS 2.3 (Austin, TX, USA). Levels of TGF-β1, -β2 and -β3 were normalized to plaque wet weight.

Cytokine plaque levels and in vitro stimulation of THP-1 cells
MCP-1, IL-6 and MIP-1β were measured in plaque homogenate supernatants using Proximity Extension Assay (PEA) technique using the Proseek Multiplex CVD96x96 reagents kit (Olink Bioscience, Uppsala, Sweden) as previously described [12]. Data are presented as arbitrary units. Human blood monocytes cells (THP-1, 88081201 -SIGMA) were cultured in RPMI-1640 medium (11875093, Thermo Fisher Scientific, Waltham, USA) supplemented with 10% FBS (10270106, Thermo Fisher Scientific, Waltham, USA) and 50 U/mL penicillin-streptomycin. Prior to the experiment, cells were treated with Phorbol 12-myristate 13-acetate (PMA, cat #78139 Sigma, Saint Louis, USA) for 24 h, to differentiate the cells into macrophages, and then washed 3 times with PBS and medium changed every 48 h for 6 days. Cells were seeded onto 6-well plates and cell culture slides and stimulated with 5 and 15 ng/ml of MCP-1 (RP-8648, Thermo Fisher Scientific, Waltham, USA) and 1 and 2 ng/ml of TGF-β2 (T2815, Sigma, Saint Louis, USA) for 24 h. Immunofluorescence microscopy was used to detect mimecan staining with a rabbit polyclonal antibody (PA5-48255, Invitrogen, Waltham, MA) as a primary antibody at 1 μg/ml and a secondary polyclonal goat anti-rabbit antibody with Alexa 488 fluorescent dye at 0.5 μg/ml. An isotype rabbit IgG antibody was used as a control antibody (Abcam, 37415) at 1 μg/ml.
For quantification of mimecan in immunofluorescence staining, cells were counted and areas (and cells) exhibiting positive mimecan immunoreactivity were analysed by applying colour threshold measurements (from which positive immunoreactivity corresponding to the isotype control were subtracted) using Adobe Photoshop CS6, Fiji software [13,14] and QuPath v0.1.2 [15]. Cell supernatants were used for mimecan detection, and RNA was extracted from the cells for gene expression analysis.
Total RNA was isolated with the RNeasy Mini Kit (74106, QIAGEN) following manufacture's instruction, and 1 μg was retrotranscribed using the High Capacity RNA-to-cDNA Kit (4387406, Applied Biosystems). Quantitative real time PCR (qPCR) was performed using TaqMan Fast Advanced MasterMix (1710122, Applied Biosystems) and the following primers: 18s (Hs99999901_s1), GAPDH (Hs02786624_g1) and Osteoglycin (Hs00247901_m1). Each reaction was performed in triplicate (n = 3) and results were normalized by geometric average of two internal controls (18s and GAPDH).
Events were verified by patient medical charts and by standardized telephone interviews. Events occurring within 72 h after carotid endarterectomy were considered procedure-related and were not included in the analysis. For patients suffering from multiple events, only the first one was taken into account in the survival analysis.

Statistical methods
Continuous variables were non-normally distributed and are thus presented as median (IQR), while categorical variables are expressed as percentages. For continuous variables, Mann-Whitney U test and Spearman's rank correlation were used, whereas for categorical variables Chi-square test was performed. For the in vitro experiments in THP-1 cells, one-way ANOVA with Holm Sidak's adjustment for multiple comparisons was used for statistical comparison. Kaplan-Meier survival analyses were performed for mimecan divided in above or below median values. Significant differences between the groups were assessed by the Log-rank test. Cox proportional hazard regression analysis (hazard ratios (HR) with 95% CI) was used and multivariate models were performed controlling for covariates that were significantly different between the groups above/below median (age, sex, diabetes, BMI, HDL, triglycerides, estimated glomerular filtration rate). A p-value <0.05 was considered statistically significant. Statistical analysis was performed using SPSS 24.0 (IBM Corp., Amonk, NY, USA).

Data statement
Due to the sensitive nature of the personal data, sharing of data is only available upon reasonable request.

Mimecan was associated to features of plaque vulnerability
Mimecan was detected in human atherosclerotic plaques with immunohistochemistry (0.40% plaque area (IQR 0.17-0.96)). Mimecan was observed in areas of collagen fibers, close to the core and around calcified regions ( Fig. 1A and B, Supplementary Fig. 1). Mimecan  Table 1, Fig. 1E-J).
There were no significant correlations between mimecan and total levels of extracellular matrix proteins glycosaminoglycans, elastin or collagen (Supplementary Table 2). However, mimecan correlated positively with plaque levels of MMP-9, one of the extracellular matrix degrading enzymes known to be associated with human plaque vulnerability (r = 0.288, p≤0.0001, Fig. 2A).
Therefore, to examine if MCP-1 could affect mimecan mRNA expression in vitro, THP-1 cells were differentiated into macrophages with PMA and thereafter stimulated with 5 or 15 ng/ml of MCP-1 for 24 h. TGF-β is known to be important for maintaining plaque stability and the balance between plaque inflammation and fibrosis [18,19], therefore THP-1 cells were stimulated with 1 or 2 ng/ml of TGF-β2 for 24 h to evaluate if this cytokine affects mimecan gene expression. Stimulation with 15 ng/ml of MCP-1 caused a significant increase in mimecan gene expression (p = 0.0047; Fig. 2C). However, stimulation with 5 ng/ml of MCP-1 or with TGF-β2 did not significantly affect mimecan gene expression in macrophages. Mimecan upregulation in THP-1 differentiated macrophages was also visualized at protein level with immunofluorescence before and after stimulation with 15 ng/ml of MCP-1 ( Fig. 2D and E). The immunofluorescence was quantified and the number of cells with positive immunoreactivity for mimecan was higher in cells stimulated with 15 ng/ml of MCP-1 than in unstimulated cells ( Fig. 2G).

Mimecan levels are associated with age and diabetes
Patients with higher plaque levels of mimecan were significantly older compared to patients with lower levels of mimecan (71 years IQR (66-78) vs 70 years IQR (64-74), p = 0.017, Table 1). The patient group with mimecan levels above median also had a higher percentage of diabetes compared to the group with lower levels (40.8% vs 22.4%, p = 0.009, Table 1) and accordingly, HbA1c levels were significantly higher in patients with mimecan levels above median compared to the group with below median levels of mimecan (57.3 mmol/mol, IQR (45.8-64.6) vs 46.9 mmol/mol IQR (43.7-56.3), (Table 1).

Patients with high mimecan plaque levels have an increased risk of cardiovascular death
A total of 194 patients were followed up for cardiovascular events and cardiovascular death. Two plaques were excluded from the follow  The boxes in the boxplots represent the median and interquartile range and the whiskers indicate the minimum and maximum values. For the qPCR, each reaction was performed in triplicate (n = 3) and results were normalized by geometric average of two internal controls (18s and GAPDH). Spearman rank correlation test was used in (A and B), one-way ANOVA with Holm Sidak's adjustment for multiple comparisons was used in (C) and Mann-Whitney test in (D and E) for statistical analysis. Representative immunofluorescence images were taken at 40× magnification. Scale bars (D and E) represent 50 μm.

C. Tengryd et al.
up analysis since they were from patients who were operated bilaterally. Altogether, 54 patients (27.8%) experienced cardiovascular events and 21 patients (10.8%) suffered from a cardiovascular death. Patients with higher plaque levels of mimecan (above the median) had a higher risk of future cardiovascular events in a log rank test (p = 0.005, Fig. 3A) and cardiovascular death (p = 0.034, Fig. 3B).

Discussion
In the present study, we showed that mimecan was associated with vulnerable plaque features including macrophages, neutral lipids, intraplaque haemorrhage and MMP-9. Mimecan correlated with proinflammatory cytokine MCP-1 and in vitro mimecan secretion is promoted in macrophages by MCP-1 stimulation. Finally, patients with higher plaque levels of mimecan had an increased risk for future cardiovascular death in a multivariate Cox regression model, even after adjusting for age, sex, diabetes, BMI, HDL, TG and eGFR (Model C).
The finding that mimecan was negatively correlated with smooth muscle cells area is interesting considering that mimecan mRNA has been shown to be downregulated in a subset of intimal VSMC in human coronary arteries [7] and that overexpression of mimecan inhibited proliferation and enhanced apoptosis in aortic smooth muscle cells [20]. This suggests that increased levels of mimecan might induce VSMC apoptosis and thereby limiting vascular repair/remodelling and contributing to a vulnerable plaque.
Mimecan also correlated with neutral lipids, which may indicate that mimecan affects lipid accumulation, as has previously been shown for fibromodulin, another proteoglycan in the SLRP family [21]. Mimecan correlated with glycophorin A, a marker of intraplaque haemorrhage. Interestingly, it is known that mimecan negatively regulates angiogenesis [5], which may in turn cause dysfunctional neo-vessel formation and, consequently, increased intraplaque haemorrhage. Intra plaque haemorrhage is also suggested to contribute to a vulnerable plaque phenotype [22], again providing a possible link between mimecan and the vulnerable plaque. Intraplaque haemorrhage and smooth muscle cell apoptosis, particularly in the cap, are all deleterious characteristics previously associated with plaque vulnerability [23]. Mimecan could Fig. 3. Patients with high plaque levels of mimecan have an increased risk for cardiovascular events and death. Kaplan-Meier curves for cardiovascular events and cardiovascular death for mimecan (A and B). Groups are divided in above (blue lines) and below median (red lines). Statistical differences between the groups were assessed by Log-rank test. A total number of 194 patients were assessed (above median, n = 97 and below median, n = 97) for both cardiovascular events and death. potentially contribute to plaque vulnerability by affecting collagen fibril size, as seen in mimecan deficient mice [4] where loss of mimecan results in larger collagen fibrils. Mimecan was also shown to affect collagen maturation in cardiac tissue in a mouse model for heart failure after myocardial infarction [24]. However, if mimecan affects the size or maturation of collagen fibrils in atherosclerotic plaques is not known. Mimecan was also associated with macrophages and correlated with levels of MCP-1, a pro-inflammatory cytokine important for atherosclerotic plaque progression and destabilization [25,26]. In line, mimecan gene expression was upregulated in THP-1 derived macrophages upon MCP-1 stimulation, which supports that a pro-inflammatory environment increases plaque mimecan levels. Moreover, mimecan is expressed in circulating and resident cardiac macrophages and it is also increased in cardiac inflammation during viral myocarditis [27]. Furthermore, mimecan correlated with MMP-9, an extracellular matrix degrading enzyme and a well-known component of vulnerable atherosclerotic plaques, [28,29]. MMP-9 is also produced by macrophages [30,31] and MMP-9 secretion in vitro increased in macrophage-like THP-1 cells and human peripheral blood monocytes after stimulation with MCP-1 [32].
In patients with higher mimecan (above median) levels, there was a higher frequency of diabetes. Interestingly, mimecan has previously been shown to be a coordinator of glucose homeostasis [33], which could potentially explain the associations identified between mimecan, HbA1c and diabetes diagnosis in this study.
The associations found between mimecan and vulnerable plaque phenotype, as well as increased risk for future cardiovascular death, are in line with a study showing that higher serum levels of mimecan were independently predictive for occurrence of major adverse cardiovascular events within one year after coronary angiography [34]. In a proteomic study of plaque extracts, mimecan was found to be higher in fibrotic than hemorrhagic plaques, [35]. An explanation for the discrepancy compared to the results of our study could potentially be the smaller sample size of only six plaques in each group, the different method used for measuring mimecan or the criteria for the separation of plaques into hemorrhagic or fibrotic plaques (based upon levels of haemoglobin and fibrous cap thickness).
Since this study is of observational nature, it is not possible to prove causality between mimecan and the increased risk for future cardiovascular death. Moncayo-Arlandi et al. [36] investigated ApoE − /− mimecan knockout mice and did not see any effect on plaque composition or lesion size, suggesting potential differences in murine and human mimecan function or a compensatory upregulation of other small leucine-rich proteoglycans during development.
In conclusion, mimecan correlated positively with histological features of plaque vulnerability, including macrophages, neutral lipids, and intraplaque haemorrhage, and correlated positively with the extracellular matrix degrading enzyme MMP-9. Mimecan also correlated positively with plaque levels of the pro-inflammatory cytokine MCP-1, and in vitro stimulation by MCP-1 increased mimecan expression in THP-1 derived macrophages. Finally, high plaque levels of mimecan was an independent predictor for future cardiovascular death in patients who had undergone endarterectomy. Taken together, this study suggests an association between mimecan and plaque vulnerability, possibly due to the local inflammatory activity. However, further mechanistic studies are needed to unravel potential treatment targets for patients with atherosclerosis.