Biological profile of monocyte-derived macrophages in coronary heart disease patients: implications for plaque morphology

The prevalence of a macrophage phenotype in atherosclerotic plaque may drive its progression and/or instability. Macrophages from coronary plaques are not available, and monocyte-derived macrophages (MDMs) are usually considered as a surrogate. We compared the MDM profile obtained from coronary artery disease (CAD) patients and healthy subjects, and we evaluated the association between CAD MDM profile and in vivo coronary plaque characteristics assessed by optical coherence tomography (OCT). At morphological analysis, MDMs of CAD patients had a higher prevalence of round than spindle cells, whereas in healthy subjects the prevalence of the two morphotypes was similar. Compared to healthy subjects, MDMs of CAD patients had reduced efferocytosis, lower transglutaminase-2, CD206 and CD163 receptor levels, and higher tissue factor (TF) levels. At OCT, patients with a higher prevalence of round MDMs showed more frequently a lipid-rich plaque, a thin-cap fibroatheroma, a greater intra-plaque macrophage accumulation, and a ruptured plaque. The MDM efferocytosis correlated with minimal lumen area, and TF levels in MDMs correlated with the presence of ruptured plaque. MDMs obtained from CAD patients are characterized by a morpho-phenotypic heterogeneity with a prevalence of round cells, showing pro-inflammatory and pro-thrombotic properties. The MDM profile allows identifying CAD patients at high risk.

Morphological characterization of MDMs. MDMs of CAD patients had two predominant and different morphotypes, round and spindle. Differently from that observed in healthy subjects, who are characterized by a similar prevalence of these two morphotypes, in CAD patients, the prevalence of round MDMs was significantly higher than that of spindle or undefined MDMs (Fig. 1a,b). Furthermore, the frequency of round MDMs was particularly higher than that of spindle MDMs in AMI patients ( Fig. 1 inset).
Both MDM morphotypes obtained from healthy subjects 10 and CAD patients expressed the CD68 macrophage marker (data not shown).

Efferocytic capacity of MDMs and levels of CD14 receptor and TG-2 protein. The efferocytic
capacity of the MDMs of CAD patients, either SA or AMI (9.0 ± 1.96% and 9.90 ± 2.65% of MDMs that had engulfed apoptotic Jurkat T-cells, respectively; P = 0.21), was lower than that of healthy subjects' MDMs ( Fig. 2a).
Efferocytosis comprises the recognition, tethering, and uptake of apoptotic cells, and several factors are involved in these processes. CD14 is a member of a complex recognition system utilized by macrophages for the clearance of apoptotic lymphocytes 14 . However, we found that the count of CD14-positive MDMs (88.11 ± 5.46% and 90.26 ± 5.08% in healthy subjects and in CAD patients, respectively; P = 0.09) and the Mean Fluorescence Intensity for CD14 receptor (36.80 ± 23.40 and 38.80 ± 19.10 in healthy subjects and in CAD patients, respectively; P = 0.73) was similar between healthy subjects and CAD patients.
Considering the relevant role of transglutaminase (TG)-2 in the formation of an efficient phagocyte portal in macrophages 15 , we verified whether changes in the expression of this protein may be responsible for the impaired efferocytosis of CAD MDMs. MDMs obtained from CAD patients had markedly lower levels of TG-2 than those obtained from healthy subjects (Fig. 2b). Of note, in healthy subjects, TG-2 levels were higher in round MDMs (Fig. 2c,d), the morphotype with a greater capacity to bind/uptake apoptotic cells 10 . In contrast, in CAD patients, TG-2 levels were similar in round and spindle MDMs, suggesting a potential contribution of this protein in the reduction of the apoptotic cell clearance detected in these patients (Fig. 2c,d).
Levels of CD206 and CD163 receptors. The mannose receptor CD206 and the haptoglobin/hemoglobin scavenger receptor CD163 have been associated with the non-inflammatory, anti-atherogenic macrophage phenotype 16,17 . A significant decrease of CD206 levels in both MDM morphotypes has been observed in NSTEMI and STEMI patients as compared to healthy subjects (Fig. 3a,b). In addition, CD163 levels were significantly lower only in the round MDM morphotype of NSTEMI and STEMI patients when compared to healthy subjects (Fig. 3c,d).
Tissue factor levels and thrombin generation. Tissue factor (TF) represent the main cellular activator of the blood coagulation cascade, allowing the thrombin generation. The MDMs of CAD patients had a significantly greater TF expression than those of healthy subjects (Fig. 4a). Moreover, a significantly greater TF fluorescence intensity was detected in both MDM morphotypes of CAD patients when compared to their counterparts in healthy subjects (Fig. 4b). This enhanced antigenic expression of TF displayed a functional consequence in thrombin generation. Indeed, the lag time and the time to peak in thrombin generation were lower in MDMs of CAD patients as compared to healthy subjects. In contrast, the peak of thrombin generation and the endogenous thrombin potential (ETP) did not change (Fig. 4c). Furthermore, an increasing gradient in TF levels going from MDMs (spindle and round) of healthy subjects to MDMs of SA, NSTEMI, and STEMI patients was observed. Of note, the highest TF levels were detected in both morphotypes of STEMI patients (Fig. 4d), suggesting a potential greater MDM contribution to thrombus formation in patients experiencing an acute coronary event.
www.nature.com/scientificreports www.nature.com/scientificreports/ Matrix metalloproteinase-9 activity. MDMs showed a progressive matrix metalloproteinase-9 (MMP-9) activity increase in SA or AMI patients, reaching a maximum in STEMI patients, as compared with healthy subjects (Fig. 5), suggesting a greater in vivo potential for MDMs to induce plaque vulnerability or rupture of the fibrous cap.
MDM profile, angiographic analyses, and OCT plaque features. Representative OCT images of coronary atherosclerotic plaques are showed in Fig. 6.
The prevalence of in vitro round MDMs was significantly higher in patients presenting a thin-cap fibroatheroma (TCFA), a lipid plaque, a ruptured plaque, and a thrombus (Fig. 7a-d) www.nature.com/scientificreports www.nature.com/scientificreports/ inversely correlated with plaque cap thickness (R = −0.65; P < 0.0001), and it positively correlated with the maximum lipid quadrant (R = 0.36; P = 0.04). The quantification of CAD severity can be captured using coronary angiography, and several scoring systems were developed to assess it. At angiographic analysis of CAD severity, round MDMs tended to correlate with the Bogaty extent (R = 0.35; P = 0.08) and severity (R = 0.37; P = 0.07) scores, and they significantly correlated with the Sullivan score (R = 0.57, P = 0.002). Moreover, the intra-plaque macrophages identification was associated with a higher round MDMs prevalence (Fig. 7e), and the latter positively correlated with normalized standard deviation (NSD) (R = 0.66; P < 0.001; Fig. 7f).
TF levels in round and spindle in vitro obtained MDMs positively correlated with the in vivo detection of a ruptured plaque (R = 0.36, P = 0.05 and R = 0.47, P = 0.01, respectively). Additionally, TF expression in both MDM morphotypes positively correlated with the presence of thrombus (R = 0.36, P = 0.05 and R = 0.47, P = 0.01, respectively). Finally, TF levels in round and spindle MDMs positively correlated with the intra-plaque macrophages presence (R = 0.57, P = 0.001 and R = 0.64, P < 0.001, respectively), and they tended to positively correlate with NSD (P = 0.08 and P = 0.11).
All these observations may suggest that the profile of MDMs may be related to high-risk features of the coronary plaque morphology and activity.

Discussion
In this report, we show for the first time that the spontaneous in vitro differentiation of monocytes into MDMs of CAD patients results in a higher prevalence of round MDMs, having a lower efferocytic capacity and an enhanced thrombin generation capacity, as compared to healthy subjects. Additionally, in CAD patients, the propensity of monocytes to differentiate into round MDMs is associated with the detection of vulnerable and/or ruptured plaque, as assessed by OCT intracoronary imaging.
Studies on tissue-resident macrophages are very challenging due to the difficulties in macrophage extraction and tissue availability. Moreover, coronary atherosclerotic plaque is a compartment not easily accessible, and www.nature.com/scientificreports www.nature.com/scientificreports/ macrophages obtained from a spontaneous differentiation of circulating monocytes may be considered a good surrogate. By investigating MDM morpho-phenotypes in the different CAD clinical presentation, we found that, differently from healthy subjects, who are characterized by a similar frequency of spindle and round cells 10 , CAD patients have mostly round MDMs. This was unexpected, as round MDMs in healthy subjects proved to exhibit anti-inflammatory properties. However, the differences between CAD patients and healthy subjects in terms of MDMs are not confined to morphology, but also reflected in different antigenic and functional profiles. In particular, despite the prevalence of round cells, MDMs of CAD patients were characterized by an anti-inflammatory properties reduction as compared to MDMs of healthy subjects, as supported by their lower levels of CD206 and www.nature.com/scientificreports www.nature.com/scientificreports/ CD163, both markers of the anti-inflammatory macrophage phenotype 16,17 . Indeed, it has been shown that macrophages expressing high levels of CD206 and CD163 had a reduced pro-inflammatory cytokine production 18 ; and, accordingly, a high expression of CD206 was peculiar of phagocytic macrophages 19 . Therefore, as plaque development may originate also from inadequate anti-inflammatory responses, the macrophage polarization imbalance may play a critical role in plaque formation 20,21 . Notably, macrophage phenotype is affected by the surrounding microenvironment and by cell-cell interactions. Indeed, the interaction between CD40L (Ligand), expressed on T lymphocytes surface, and CD40, highly expressed in pro-inflammatory macrophages 22 , induces the expression of adhesion molecules, and the release of matrix metalloproteinases and pro-inflammatory cytokines, promoting monocyte recruitment and plaque progression 23 . Thus, the imbalance in the inflammatory response in atherosclerotic plaques might be supported by a loss of anti-inflammatory properties of the round MDMs detected in CAD patients. To further investigate this observation, we evaluated the efferocytic capacity of MDMs and correlated it with angiographic and OCT CAD burden. Interestingly, we found a significant reduction of the efferocytic ability in the MDMs of CAD patients as compared to healthy subjects, and, in our cohort of www.nature.com/scientificreports www.nature.com/scientificreports/ CAD patients, coronary angiographic analyses demonstrated that, the lower was the MDM efferocytic capacity, the more severe and extensive was the coronary atherosclerotic burden. A similar behavior was observed when considering MLA and stenosis grade of the investigated lesion by means of OCT. It is well known that, at coronary plaque level, apoptotic cell death and efferocytic rates are strongly related to atherosclerotic lesion stage 24 . The apoptotic cell clearance enhances the resolution of the ongoing inflammatory process, by modulating several and critical inflammatory pathways, including IL-10 and TGF-β release, and promoting phagocyte survival [25][26][27] . Thus, even a small reduction in efferocytic ability may have relevant clinical implications. The reduced efferocytosis detected in CAD MDMs did not depend on the binding of apoptotic cell mediated by the CD14 receptor, but it was related to the expression of TG-2, a protein-cross-linking enzyme involved in an efficient phagocytic portal formation 15 . These observations further confirm the presence of significant functional differences of CAD MDMs, as compared to healthy subjects, that may be involved in atherosclerosis initiation and progression and potentially in its complications. This may pave the way to pharmacological modulation studies aimed at limiting Data derive from independent cultures obtained from 10 healthy subjects, 10 SA patients, 10 NSTEMI patients, and 10 STEMI patients. *P < 0.05, **P < 0.005. Full blots are shown in Supplementary Fig. 2. www.nature.com/scientificreports www.nature.com/scientificreports/ progression and improving stabilization of coronary atherosclerosis, by a direct targeting of specific MDM subpopulations. Of note, recently published studies found that antibodies blocking CD47, a key anti-phagocytic molecule, could restore phagocytosis and prevent atherosclerosis 28 .
Macrophages are intrinsically linked not only to atherosclerotic disease progression but also to disease activity. Indeed, the amount of monocyte-macrophages infiltrating the plaque and their detection at plaque rupture-sensitive sites are associated to plaque vulnerability 29,30 . Interestingly, we found that the propensity of circulating monocytes to differentiate into round MDMs in CAD patients was associated with a vulnerable plaque presence and with a ruptured plaque, as evaluated by OCT analyses. The risk of an acute coronary event is related to the composition and activity of the plaque, and inflammation is crucial for its instability [31][32][33] . Our observation could imply that round MDMs in CAD patients may reduce lesion integrity and increase the likelihood of an acute myocardial infarction. Indeed, the presence of a thinner fibrous cap, a larger lipid core, and a greater macrophage content were more frequently detected in CAD patients with a higher prevalence of in vitro round MDMs. Thus, although our data provide initial observational insights on the potential association between vulnerable coronary plaques and round MDMs in CAD patients, this morpho-type might represent a tool able to recognize high-risk patients of developing an acute myocardial infarction. Yet, future mechanistic studies and/or transcriptomic investigations, combined with the in vivo analysis of coronary atherosclerotic plaques, are needed in order to support our findings, to identify the involved signal transduction, and to explore the contribution of an intrinsic differentiation program or of environmental imprinting of monocytes, in determining the diversity of MDMs 34 .
Finally, it is known that plaque macrophage accumulation may favor both rupture of fibrous cap, due to pro-inflammatory cytokines and proteolytic enzymes secretion, and thrombus formation, due to TF release 35 . It has been shown that circulating TF may potentiate the thrombogenic stimulus and upregulate the inflammatory response 36,37 . TF is synthetized by macrophage-derived foam cells in atherosclerotic plaques, and its expression is increased in lipid-rich plaques 38,39 . In our report, the MDMs obtained from AMI patients showed a higher TF fluorescence than the MDMs of healthy subjects, with a peak in STEMI patients. In addition, we found that TF levels in spindle and in round MDMs positively correlated with the OCT detection of a ruptured plaque and of thrombi. Accordingly, it has been documented that TF levels were higher in atheroma of patients with acute coronary syndrome than with SA 40 . Then, TF expression may enhance the severity of acute coronary thrombosis, further Figure 5. MMP-9 activity. MDMs were incubated overnight in medium containing 0.2% fatty-acid-free albumin. The activity of MMP-9 secreted by MDMs was evaluated by the Biotrak TM activity assay system. Data are expressed as mean ± SD and derive from independent cultures obtained from 9 healthy subjects, 9 SA patients, and 18 AMI (9 NSTEMI and 9 STEMI) patients. *P < 0.05 vs. healthy subjects. www.nature.com/scientificreports www.nature.com/scientificreports/ highlighting the functional differences between the MDMs of CAD patients and those of healthy subjects. Indeed, round MDMs in CAD patients seem to lose their anti-inflammatory properties, and to acquire pro-thrombotic features, potentially contributing, at least in part, to acute coronary event initiation and severity. The mechanisms underlying this observation cannot be deduced from our study; yet, this study shows a peculiar phenotype of round MDMs in CAD patients that may potentially affect the inflammatory status and the pathogenesis of coronary atherosclerosis and atherothombosis.
Our findings might have some potential clinical implications. Although this is an observational study and a cause-effect relationship cannot be established, the definition of an MDM profile may help to predict CAD burden and coronary plaque composition, also contributing to the differentiation between stable and unstable plaques. Moreover, the development of therapies that blunt macrophage cytotoxicity, plaque growth, and destabilizing functions and enhance their natural reparative properties, may represent a new approach to curtail macrophage-mediated injury, limit coronary stenosis progression and enhance plaque stability 41 , the two most important aims when counteracting coronary atherosclerosis. Of interest, emerging evidences suggest the possibility of pharmacologically modulating the functions of macrophages [42][43][44] . In particular, our findings, if confirmed in larger study populations, may pave the way to novel diagnostic tools, able to early identify patients at www.nature.com/scientificreports www.nature.com/scientificreports/ high risk of severe CAD and/or of atherothrombosis. Further, our observations may be useful to develop novel therapies able to manipulate macrophage morpho-phenotype, with the ultimate goal of reducing CAD burden and first or recurrent acute cardiovascular events.
Some limitations warrant mention. First, the small sample size of our study limits the general applicability of our findings. Second, even if the macrophage heterogeneity has been documented in tissues, these macrophage phenotypes may not correspond to phenotypes generated in vitro. Indeed, in atherosclerotic lesions, macrophages respond to several environmental stimuli, such as cytokines and modified lipids, and they interact with other cells, including erythrocytes, lymphocytes, and platelets [45][46][47] , potentially modifying their phenotypes. Thus, an assessment of monocyte functions and of their differentiation process via interaction with other cells, such as lymphocytes and platelets, and in relation to progenitors and stem cells, is needed. Third, our in vitro model may be used as a representative model of tissue macrophages 10,48,49 , however, the lack of cell turnover and of tissue-specific matrix proteins, are crucial in tissue macrophage behavior. Fourth, OCT imaging is a suboptimal reflection of true histology, mainly for macrophage infiltration 50,51 .
In conclusion, we demonstrated that the MDMs of CAD patients, as compared to healthy subjects, show a peculiar morpho-phenotype profile, which is characterized by a predominance of round MDMs, with reduced anti-inflammatory properties and a higher propensity for thrombogenicity than those of healthy subjects. Notably, this specific profile is associated with the detection of high-risk and rupture-prone coronary plaques, at OCT investigation. Assessing the functional features of macrophage phenotypes may allow to shed lights on their contribution to coronary atherosclerosis, thus providing novel diagnostic, therapeutic, and, more importantly, preventative regimens for CAD.

Methods
Study design. The study was carried out at Centro Cardiologico Monzino, Milan, Italy. Ninety consecutive CAD patients undergoing coronary angiography due to stable angina (SA) or AMI, as their first ischemic heart disease event, and showing obstructive atherosclerosis (>50% diameter stenosis by visual estimate) were enrolled. Twenty-five healthy subjects, with neither history of CAD, nor cardiovascular risk factors, nor inflammatory disorders, and specifically not taking any cardiovascular therapy, were recruited as control group.
In the initial 40 CAD patients, we performed the characterization of MDMs of CAD patients, and we compared these data to those obtained in healthy subjects. Subsequently, we evaluated the association between this profile and CAD clinical presentation, and we also assessed plaque composition in 50 patients undergoing OCT evaluation.
The study was approved by the institutional Ethic Committee (Comitato Etico IRCCS-Istituto Europeo di Oncologia e Centro Cardiologico Monzino), and it was performed according to the Declaration of Helsinki. All the study subjects provided written informed consent at the time of enrollment.

Culture of MDMs.
MDMs were obtained from a differentiation of monocytes isolated from venous blood of CAD patients and healthy subjects as described 10 . Briefly, mononuclear cells were obtained by density centrifugation on Ficoll-Paque (GE Healthcare, EuroClone, Milan, Italy) and plated (2 × 10 6 /mL) in 35 mm well plates (Primaria TM , Falcon, Sacco S.r.l, Como, Italy). After 90 minutes, non-adherent cells were removed, and adherent monocytes were cultured over 7 days in Medium 199 (Lonza, EuroClone, Milan, Italy) supplemented with 2 mM L-glutamine, 100 U/ml penicillin, 100 μg/ml streptomycin, and 10% autologous serum freshly obtained from blood clotted for 2 hours at 37 °C. Medium was not replaced throughout the culture period. The morphology of MDM was inspected by phase contrast microscopy (AxioVert 200 M, Zeiss, Milan, Italy) at 20× or 40× magnification. MDMs were defined spindle when a length >70 µm and a width <30 µm were detected, and round when width and length were similar and >35-40 µm. Cells of morphology and/or dimension that did not meet the above criteria were classified as undefined 10 .
Flow cytometric analysis of monocytes and MDMs. Monocyte subsets were analyzed in 100 μl of whole blood collected from all participants. Adherent MDMs were detached with trypsin.
Cells were stained for 10 min with monoclonal mouse anti-human CD14 APC-conjugated and CD16 FITC-conjugated antibodies (BD Biosciences, Milan, Italy) and analyzed using flow cytometry 52 . Data were analyzed using CellQuest analysis software (Becton-Dickinson, Oxford, UK).
Quantitative fluorescent image analysis. Digital images were taken on an AxioObserver.Z1 microscope connected to a camera, and processed using the AxioVision 4.7 (Zeiss) software. Fluorescence intensity (densitometric sum of gray) was quantified, as an index of the amount of the protein investigated 10 . Data are expressed as log median of fluorescence intensity/μm 2 and interquartile range for each MDM morphotype, subtracting the negative control value. Multiple fields of view (at least three fields, 400× magnification) were taken for each culture. Means derive from 10 independent cultures obtained from different donors.
Matrix metalloproteinase-9 activity. MDMs were washed, and fresh medium containing 0.2% fatty acid free albumin was added. After overnight incubation, supernatant was collected and centrifuged, and the activity of MMP-9 secreted by MDMs was evaluated by the Biotrak TM activity assay system (Amersham, GE Healthcare, Milan, Italy).

Angiographic analyses.
Angiographic analyses for the evaluation of CAD severity and extent were focused on the assessment of the Bogaty score 57 and the Sullivan extent score 58 .
OCT image analysis. Culprit lesion, in AMI patients, was identified by angiography, electrocardiographic T-wave or ST-segment modifications, and/or regional wall motion abnormalities at echocardiogram. In SA patients, OCT analysis was performed at the MLA site. Frequency domain OCT (FD-OCT) images were acquired by a commercially available system (C7 System, LightLab Imaging Inc/St Jude Medical, Westford, MA) connected to an OCT catheter (C7 Dragonfly; LightLab Imaging Inc/St Jude Medical), which was advanced to the culprit lesion. All images were digitally recorded and stored, and every single frame (0.2 mm) was analyzed by two independent investigators from an OCT core lab, who were blinded to clinical and laboratory values (Institute of Cardiology, Catholic University of the Sacred Heart, Policlinico Gemelli, Rome, Italy) 59 .
At the MLA site or at culprit lesion, respectively in SA and AMI patients, the analysis was targeted on plaque characterization (calcified, fibrous, or lipid plaques), presence of plaque rupture, measurement of fibrous cap thickness, and presence of intracoronary thrombi and intra-plaque microchannels, as previously described [11][12][13] . When a plaque contained two or more lipid-containing quadrants, it was considered a lipid-rich plaque, and the lipid arc and the cap thickness were measured. TCFA was defined as a lipid-rich plaque with a fibrous cap thickness of ≤65 µm 60 .

OCT macrophage analysis. The presence of macrophage infiltration (MØI) in the lesions analyzed by
OCT was assessed as previously reported 59 . Briefly, macrophages were qualitatively identified on raw OCT data within a 300 × 125 µm 2 (lateral × axial) region of interest (ROI), according to the International Working Group for Intravascular Optical Coherence Tomography (IWG-IVOCT) Consensus standards 12 .
Macrophages were visualized as signal-rich, distinct, or confluent punctate regions exceeding the intensity of background speckle noise and generating a backward shadowing. For caps having a thickness <125 µm 2 , the depth of the ROI was matched to the cap thickness. Median filtering was performed with a 3 × 3 square kernel to remove speckle noise. In plaques with MØI, quantitative evaluation of macrophage content was obtained by measuring the NSD known to have a high degree of positive correlation with histological measurements of macrophage content, by using a dedicated software provided by S. Jude Medical 13,61 . NSD was measured for each pixel within each cap using a 125 µm 2 window centered at the pixel location: NSD (x, y) = [σ (x, y)125 µm 2 / (Smax-Smin)] × 100, where NSD (x, y) is the normalized standard deviation of the OCT signal at pixel location (x, y), Smax is the maximum OCT image value, and Smin is the minimum OCT image value. Pixels within the (125 × 125) µm 2 window that did not overlap with the segmented cap were excluded 13 . Statistical analysis. The distribution of continuous variables was assessed by visual inspection of frequency histograms and with the use of the Shapiro-Wilk test. Continuous variables were expressed as mean ± standard deviation (SD) or median with interquartile range, if they followed a normal or non-normal distribution,