Neutrophil extracellular traps and fibrocytes in ST-segment elevation myocardial infarction

Sample collection from patients and healthy controls

We enrolled patients with STEMI (n = 50) undergoing primary percutaneous intervention (pPCI) for a coronary TIMI flow of 0. Patients were included if they met the following criteria: (1) chest pain at the time of coronary angiography; (2) new ST-segment elevations of ≥ 2 mm on at least one chest lead or elevations of ≥ 1 mm on at least one limb lead within 20 min of angiography; and (3) suitable coronary anatomy for thrombus aspiration. All other inclusion and exclusion criteria, and general procedural considerations for use of a thrombectomy device, were applied as previously described [38]. All patients received 250 mg of acetylsalicylic acid and were heparinized at an activated coagulation time of > 300 s (4000–10000 IE). Patients with immunosuppression, acute or chronic infections, or autoimmune diseases were excluded. In brief, in all 50 patients, we collected 10–20 ml of blood in close proximity to the culprit site via a commercially available thrombectomy catheter. Blood drawn from the femoral sheath served as comparator. To account for dilutions, all analyses were normalized to femoral hematocrit. Particulate thrombus material was recovered in 21 of these patients, separated using a 40 µm cell strainer and processed for subsequent histology and immunofluorescent staining. In the remaining patients, no thrombus could be harvested due to its disintegration during thrombus aspiration. In 21 patients, venous whole blood was drawn 72 h after pPCI. Furthermore, venous whole blood from healthy donors (n = 21, mean age 51 ± 8, no prior diseases or medication) was drawn. All blood samples were immediately centrifuged at 1000g for 10 min, and plasma was stored in aliquots at − 80 °C until further analysis. Histological sections of left ventricular myocardium were harvested from patients who suffered from STEMI, or from patients free from cardiac disease. Schematic representation of sampling and experimental procedures is summarized in Supplemental Fig. S1.

Measurement of soluble NET surrogate markers

Double-stranded DNA (dsDNA) was measured using the Quant-iT PicoGreen Kit as previously described [38]. In brief, plasma samples or standard were incubated with PicoGreen reagent, a fluorescent dye staining free dsDNA, for 5 min in 96-well microplates. Fluorescence was measured using a Promega GloMax Discover microplate reader (excitation 480 nm, emission 520 nm) and is directly proportional to dsDNA concentration in samples. Values were normalized to the standard curve.

Citrullinated histone H3 (citH3) was measured as previously described [64], with minor modifications. Streptavidin-coated 96-well plates were incubated with anti-histone biotin provided in the Cell Death Detection ELISA PLUS Kit for 2 h. Plates were washed and incubated with 50 µl undiluted plasma or standard per well for 1.5 h. After washing, anti-citrullinated histone H3 was incubated for 1 h. Plates were washed and incubated with horse-radish peroxidase conjugate antibody for 1 h. After washing, enzymatic reaction was started by adding BM Blue POD substrate. Plates were stopped after 20 min using 2 M H₂SO₄. Optical density was measured on a Promega GloMax Discover microplate reader (450 nm, reference 620 nm) and normalized to the standard curve.

Assessment of enzymatic infarct size

Enzymatic infarct size was calculated as the area under the curve of creatine-phosphokinase isoform MB (CK-MB AUC) using the trapezoidal formula [10] and expressed as arbitrary units.

Measurement of functional extracellular vesicle-associated tissue factor

We measured functional extracellular vesicle-associated tissue factor (EV-TF) activity as previously described [22, 28]. We opted for measuring EV-TF instead of sole TF due to the higher reliability of the assay [24]. In brief, EVs were isolated from citrated platelet poor plasma by centrifugation at 18,000g, 20 min at 4 °C, washed twice with HBSA (137 mmol/l sodium chloride, 5.38 mmol/l potassium chloride, 5.55 mmol/l glucose, 10 mmol/l HEPES, 0.1% bovine serum albumin, pH 7.5) buffer and resuspended in 200 µl HBSA. EV suspension was then incubated with an antibody for human TF (hTF1, 1 µl) or control antibody (mouse IgG, 1 μl) for 15 min at 21 °C. 50 μl aliquots were then added to 96-well plates in duplicates. 50 μl of HBSA containing 10 nmol/l factor VIIa, 300 nmol/l factor X (FX) and 10 mmol/l calcium chloride was added to each sample and incubated for 2 h at 37 °C. FXa generation was stopped by the addition of 25 μl of disodium ethylenediaminetetraacetate (EDTA) buffer and 25 μl of the chromogenic substrate Pefachrome FXa 8595 (4 mmol/l) was added and incubated at 37 °C for 15 min. Absorbance at 405 nm was measured using a Multiscan Spectrum microplate reader (Thermo Scientific). TF-dependent FXa generation (pg/ml), which represents EV-associated TF activity, was determined by subtracting the amount of FXa generated in the presence of hTF1 from the amount of FXa generated in the presence of the control antibody.

Isolation of NETs

NETs were isolated as previously described [46]. 30 ml of heparinized whole blood of healthy donors was layered on top of 15 ml Lymphocyte Separation Medium and centrifuged at 800g for 30 min at 21 °C with brakes off. Plasma and peripheral blood mononuclear cell (PBMC) layers were discarded. Then, 20 ml 6% dextran and 20 ml sterile phosphate-buffered saline (PBS) were gently mixed with the remaining cell suspension containing erythrocytes and neutrophils. After incubation for 30 min at 21 °C, the supernatant containing neutrophils was collected and washed with sterile PBS. Remaining erythrocytes were lysed using red cell lysis buffer (154 mmol/l ammonium chloride, 10 mmol/l potassium hydrogen carbonate, 0.1 mmol/l EDTA). Neutrophil purity was assessed using a XN-350 Hematology Analyzer (Sysmex) and was usually above 95%. 5 × 10⁶/ml neutrophils resuspended in RPMI + 3% fetal calf serum (FCS) were seeded into 6-well cell culture plates and were stimulated with 500 nmol/l phorbol myristate acetate for 4 h at 37 °C. Supernatant was discarded, and ice-cold PBS was used to detach generated NETs from the bottom of the wells. After centrifugation at 450 g for 10 min at 4 °C, cell-free, NET-rich supernatant was collected and stored at − 80 °C. Concentration of NETs was measured using PicoGreen reagent as described above.

Isolation of peripheral blood mononuclear cells

15 ml of heparinized blood from healthy donors and STEMI patients was layered onto 15 ml of lymphocyte separation medium in 50 ml tubes and centrifuged for 15 min, 800g at 21 °C with brakes off. The PBMC layer was collected and washed with PBS + 10% FCS. Cells were then incubated with red cell lysis buffer to remove remaining erythrocytes and washed with PBS.

In vitro stimulation of fibrocytes with NETs

To assess the impact of NET exposure on fibrocyte differentiation, 1 × 10⁶/ml isolated PBMCs were seeded into 8-well chamber slides and stimulated with 500 ng/ml isolated NETs, NETs with 40 IE/ml DNase 1, NETs with 20 µg/µl anti-TLR-4 blocking antibody, NETs with DNase and anti-TLR-4 blocking antibody, or vehicle control. Medium was changed after 24 h, and cells were again stimulated as described above. After another 48 h of stimulation, cells were washed with medium, fixed in 4% paraformaldehyde for 1 h and stained with hematoxylin for 10 min. Slides were then rinsed with distilled water and mounted using glass coverslips. Fibrocytes were identified as spindle-shaped cells in twenty random fields (20×) of each experimental condition. Images were acquired using an Olympus DP72 microscope by an observer blinded to experimental conditions. Cells were then counted using ImageJ software (Version 1.51j8).

To test NET-induced fibrocyte activation, PBMCs (1 × 10⁶/ml) were resuspended in RPMI containing 10% FCS, l-glutamine, 1% MEM non-essential amino acids, 50 µg/ml penicillin, 50 µg/ml streptomycin, 50 µg/ml gentamycin and 2.5 µg/ml fungizone, seeded into 24-well cell culture plates and incubated for 72 h at 37 °C, which leads to spontaneous differentiation of a fraction of monocytes into fibrocytes [50]. Cells were incubated with Protein Transport Inhibitor (BD GolgiPlug^®) for 5 min and subsequently stimulated with 500 ng/ml of isolated NETs, NETs + 40 IE/ml DNase 1 (Dornase alfa, Pulmozyme^®) or vehicle control for 6 h. Medium was aspirated, and cells were detached using trypsin–EDTA. After three washing steps using PBS, fibrocytes were characterized using flow cytometry as described below.

Flow cytometry

Fibrocytes from in vitro stimulation experiments were incubated with primary fluorescent antibodies against collagen-I, CD34, CD45 and primary unconjugated goat IgG BMPRII for 15 min. After washing with PBS, a secondary fluorescent anti-goat IgG antibody was added for 15 min. In addition, intracellular IL-6 was measured. After incubation with extracellular primary antibodies, cells were incubated with FIX & PERM fixation medium for 15 min. After washing, cells were incubated with FIX & PERM permeabilization medium and intracellular IL-6 antibody for 15 min.

Whole blood samples collected in EDTA-containing tubes were incubated with primary fluorescent antibodies against collagen-I, CD34, CD45, CD11b and primary unconjugated goat IgG BMPRII for 15 min. Red blood cells were lysed using BD FACS lysis solution. After washing with PBS, a secondary fluorescent anti-goat IgG antibody was added for 15 min. Cells were then washed and resuspended in PBS for analysis.

Cells were analyzed with a BD FACSCanto II and FACSDiva Software (BD Biosciences). Cells were separated from debris by adjusting forward and side scatter. Leukocytes were identified based on CD45 expression. Frequency of circulating fibrocytes, defined as CD45⁺CD34⁺collagen-I⁺ cells [50], was computed as fibrocytes/10⁶ CD45⁺ cells. Surface marker expression is given as mean fluorescence intensity (MFI). An example of gating from a representative whole blood patient sample is shown in Supplemental Fig. S2. MFI values and statistics are provided in Supplemental Tables 2, 3, and 4.

Immunofluorescence

Upon collection, tissue specimens were immediately immersed in 7.5% formalin for 24 h, embedded in paraffin and then cut into 3 µm sections for immunofluorescence staining. For staining of fibrocytes, samples were stained using primary antibodies against CD34, CD45 and collagen-I at 4 °C overnight. As secondary antibodies, DyLight 550, DyLight 755 and DyLight 650 were used at room temperature for 1 h. For staining of NETs, culprit site thrombi were incubated using a primary antibody against DNA-Histone H1 at 4 °C overnight. As secondary antibody, DyLight 755 was used at room temperature for 1 h. Nuclei were stained using DAPI. Images were acquired with an Axio Observer Z1 fluorescence microscope. Analysis of images was performed using TissueQuest software (TissueGnostics, version 4.01.0128).

Trichrome staining

Modified trichrome stains were performed as previously described [17]. Image acquisition and analysis were performed using an Olympus DP72 microscope.

Echocardiographic assessment of left ventricular function

In a subset of 33 patients, transthoracic echocardiography was performed 3 [2, 4] days after STEMI. Echocardiography at long-term follow-up (24 ± 8 months) was performed in 24 patients. Left ventricular ejection fraction (LVEF) was measured using biplane modified Simpson’s method. Myocardial contractility was assessed and scored for each segment as described [26]: (1) normal or hyperkinesia; (2) hypokinesia; (3) akinesia; (4) dyskinesia. Wall Motion Score Index (WMSI) was computed as the sum of the score of each segment divided by the total number of segments.

Statistical analysis

Distribution of data was analyzed using the Kolmogorov–Smirnov test and histograms (data not shown). Normally distributed numerical data are presented as mean ± standard deviation (SD); otherwise, median and interquartile range (IQR) are provided. To compare paired and normally distributed data, paired Student t test was used; otherwise, Wilcoxon signed-rank test was computed. For unpaired comparisons, unpaired Student’s t test was employed in case of normally distributed data; otherwise, Mann–Whitney U test was used. For comparison of three groups, one-way analysis of variance with Dunn’s multiple post hoc comparison was employed. Correlations between parameters were assessed using Spearman’s rank correlation. To correct for multiple testing, the Bonferroni–Holm method was used. Statistical analyses were performed using IBM SPSS 21.0. Figures were generated using GraphPad Prism 5. Data are depicted as scatter plots; lines in graphs indicate mean ± standard error of the mean, in case of normally distributed data; otherwise, median and IQR are given.

Patient characteristics

NETs accumulate at the culprit site and correlate with infarct size

In a first series of experiments, we measured concentrations of dsDNA and citH3 as surrogate markers of NET burden. Consistent with our previous findings [38], dsDNA was significantly increased at the culprit site compared with femoral site dsDNA concentration (n = 48, culprit site dsDNA 529 [429–740] vs. femoral dsDNA 404 [349–563] ng/ml, p < 0.0001, Fig. 1a). Furthermore, the specific NET marker citH3 was also highly increased at the culprit site (n = 48, culprit site 332 [123–810] vs. femoral 235 [113–434] ng/ml, p < 0.01, Fig. 1b). Healthy controls displayed lower dsDNA concentrations compared with femoral site concentrations of STEMI patients (controls n = 21, 291 [252–327] ng/ml, p < 0.0001, Fig. 1a), whereas citH3 levels did not differ significantly (controls n = 21, 192 [150–399] ng/ml, p = ns, Fig. 1b). No correlations were found between symptom to balloon time and NET burden (data not shown). Both dsDNA (Fig. 1c) and citH3 (Fig. 1d) in culprit site plasma were positively correlated with enzymatic infarct size. No differences in NET burden between culprit sites were observed, neither of soluble NET markers, nor of thrombus NET burden as measured by immunohistochemistry (Supplemental Fig. S4a–c). Similarly, multi-vessel disease was not associated with increased NET burden (Supplemental Fig. S4d–f).

Open image in new window

Tissue factor (TF) is important for the primary activation of the coagulation protease cascade and is thus paramount for hemostasis and thrombosis [9, 19]. Recently, TF was shown to be expressed and functionally active on NETs at the culprit site [60]. We found activity of circulating EV-TF to be increased at the culprit site compared to femoral control (n = 46, culprit site 0.12 [0.01–0.83] vs. femoral 0.00 [0.00–0.08] pg/ml, p < 0.001, Supplemental Fig. S5a). At 72 h after pPCI, EV-TF activity was not different compared with femoral site values (72 h n = 21, 0.02 [0.00–0.16] pg/ml, p = ns vs. femoral). EV-TF activity at the culprit site was positively correlated with dsDNA (n = 46, r_s = 0.316, p < 0.05, Supplemental Fig. S5b), but not with citH3 (n = 47, r_s = 0.275, p = ns, Supplemental Fig. S5c).

NETs induce differentiation of fibrocytes from monocytes in vitro and lead to fibrocyte activation

Monocytes have been shown to be precursors of fibrocytes [43]. Upon incubation of cells with isolated NETs, we observed an increased differentiation of monocytes into fibrocytes after 72 h (Fig. 2a). Representative examples for each condition are shown in Fig. 2b–f. Addition of DNase diminished the effect of NETs on fibrocyte differentiation. Addressing the question of mechanisms by which NETs induce monocyte differentiation towards fibrocytes, we employed a TLR-4 blocking antibody and observed that differentiation was blunted, similar to the effect of DNase. Addition of both DNase and TLR-4 blocking antibody exerted no additional inhibitory effect. Monocytes derived from STEMI patients displayed similar levels of fibrocyte differentiation compared to healthy controls (Supplemental Fig. S6).

Open image in new window

In a next step, we stimulated cultivated fibrocytes with isolated NETs in vitro for 6 h. NETs induced an upregulation of collagen-I (Fig. 3a), BMPRII (Fig. 3b) and CD34 (Fig. 3c) compared to control. Furthermore, NETs stimulated the production of IL-6 by fibrocytes (Fig. 3d). Addition of DNase abrogated the increase of BMPRII (Fig. 3b) but had no significant impact on other markers.

Open image in new window

Fibrocytes accumulate at the culprit site and coronary thrombus in STEMI

Circulating fibrocytes were significantly increased at the culprit site compared to the femoral site (Fig. 4a). At the culprit site, fibrocytes expressed increased levels of collagen-I (Fig. 4b), while CD34 expression was unchanged (Fig. 4c). However, CD11b expression was increased at the culprit site (Fig. 4d). Expression of BMPRII was not different between the culprit site and the femoral site (Fig. 4e). No correlations were found between symptom to balloon time, fibrocyte frequency or expression of surface markers (data not shown). By immunofluorescent staining, we detected CD45⁺CD34⁺collagen-I⁺ cells in particulate coronary thrombi, confirming the presence of fibrocytes (Fig. 5a–e). In line with previous results [38], we found NETs within thrombi (Fig. 5f–h).

Open image in new window

Open image in new window

We analyzed fibrocyte frequency and function 72 h after pPCI. Compared to baseline, we observed a decline of the number of circulating fibrocytes at 72 h (Fig. 4a). This was accompanied by increased collagen-I (Fig. 4b) but unchanged CD34, CD11b and BMPRII expression (Fig. 4c–e). BMPRII and collagen-I expressions of fibrocytes were positively correlated with each other (culprit site n = 43, r_s = 0.441, p < 0.01; femoral n = 43, r_s = 0.513, p < 0.001; 72 h n = 19, r_s = 0.802, p < 0.0001).

Increased NET burden at the culprit site is associated with local fibrocyte activation

Since NETs were reported to promote fibroblast function [8], and based on our in vitro findings, we searched for a potential link between NET burden and fibrocyte function in patients. We observed a positive correlation between dsDNA at the culprit site and CD34 expression of culprit site fibrocytes (n = 47, r_s = 0.321, p < 0.05). Furthermore, a positive correlation was found between citH3 and CD34 expression (n = 47, r_s = 0.377, p < 0.01). A positive correlation was observed between dsDNA at the culprit site and BMPRII expression of culprit site fibrocytes (n = 41, r_s = 0.360, p < 0.05).

Fibrocytes are present in the heart after myocardial infarction

Fibrocytes were detected mostly in the transition zone between viable myocardium and scar of patients dying from acute STEMI (n = 4, 2.7 ± 0.6% of cells; Fig. 6). By contrast, few fibrocytes were detected in myocardial sections of normal hearts (n = 3, 0.9 ± 0.2%, p = 0.007 for STEMI vs. healthy; Supplemental Fig. S7).

Open image in new window

Fibrocyte adhesion and migration are mediated via CD11b, with intracellular adhesion molecule 1 (ICAM-1) being an important ligand of CD11b [11, 50]. ICAM-1 has been shown to be upregulated in myocardium after ischemia [31, 47, 48]. Both ICAM-1 and CD11b were expressed in the transition zone after STEMI (Supplemental Fig. S8a, b).

NET burden and fibrocyte activation at the culprit site are correlated with left ventricular dysfunction

In a subset of patients, we obtained transthoracic echocardiographic measurements at long-term follow-up. dsDNA at the culprit site at the time of pPCI was positively correlated with WMSI at follow-up (Fig. 7a). Fibrocyte numbers at the culprit site at baseline were not correlated with left ventricular function or WMSI at follow-up (data not shown). However, BMPRII (Fig. 7b) and CD11b (Fig. 7c) expressions on culprit site fibrocytes were positively correlated with WMSI.

Open image in new window