Factor H-related protein 1 (FHR-1) is associated with atherosclerotic cardiovascular disease

Atherosclerotic cardiovascular disease (ACVD) is a lipid-driven inflammatory disease and one of the leading causes of death worldwide. Lipid deposits in the arterial wall lead to the formation of plaques that involve lipid oxidation, cellular necrosis, and complement activation, resulting in inflammation and thrombosis. The present study found that homozygous deletion of the CFHR1 gene, which encodes the plasma complement protein factor H-related protein 1 (FHR-1), was protective in two cohorts of patients with ACVD, suggesting that FHR-1 accelerates inflammation and exacerbates the disease. To test this hypothesis, FHR-1 was isolated from human plasma and was found to circulate on extracellular vesicles and to be deposited in atherosclerotic plaques. Surface-bound FHR-1 induced the expression of pro-inflammatory cytokines and tissue factor in both monocytes and neutrophils. Notably, plasma concentrations of FHR-1, but not of factor H, were significantly (p < 0.001) elevated in patients with ACVD, and correlated with the expression of the inflammation markers C-reactive protein, apolipoprotein serum amyloid protein A, and neopterin. FHR-1 expression also significantly correlated with plasma concentrations of low-density lipoprotein (LDL) (p < 0.0001) but not high-density lipoprotein (HDL). Taken together, these findings suggest that FHR-1 is associated with ACVD.

Inflammation is a hallmark of many diseases and represents an innate immune response to tissue injury and infection, as well as metabolic stress. Although inflammation is usually followed by clearance and healing processes, sustained immune responses that accompany permanent inflammation can cause serious inflammatory injury in the host. These findings have been observed in patients with atherosclerotic cardiovascular disease (ACVD), a condition that has a substantial negative effect on public health 1 . Innate immune reactions, including those involving complement, recognize modified self surfaces such as oxidized lipids and mark these surfaces for immune reactions. Complement proteins have been localized in atherosclerotic plaques, where they induce opsonization, recruit phagocytic cells, and generate pro-inflammatory mediators [3][4][5] . According to strong complement activation in atherosclerotic plaques, inhibition of the terminal complement pathway was found to substantially reduce plaque formation and inflammation 6,7 .

Results
Homozygous chromosomal deletion of CFHR1 protects from ACVD. Homozygous deletion of a chromosomal fragment containing the CFHR3 and CFHR1 genes (∆CFHR3/1) has been reported to be protective against diseases such as IgAN 14 and age-related macular degeneration (AMD) [15][16][17] , but it is a risk factor for atypical hemolytic-uremic syndrome (aHUS) 18,19 . We hypothesized that the protective effect of ∆CFHR3/1 could be explained by the previously described pro-inflammatory activity of FHR-1 13 and assessed whether ∆CFHR3/1 also protects against ACVD. Assessment of a cohort of patients with advanced ACVD showed that the frequency of ∆CFHR3/1 was 1.2% (3/244), compared with 5.2% (28/525) in a healthy age-matched (> 50 year) control cohort (Table 1). These healthy controls had been included in a case control study of patients with major depression disorders, with this subgroup of healthy individuals evaluated by multiplex ligation-dependent probe amplification (MLPA) assays to determine their ∆CFHR3/1 status 20 . Similarly, the frequency of ∆CFHR3/1 was found to be 0.8-1.2% in patients with AMD, compared with 4.9-5.2% in control individuals 11,[15][16][17] .
Because of the relevance of these findings, we determined the frequency of homozygous CFHR3/1 deficiency in a second, larger cohort of patients with ACVD (CVD, cohort 2), finding a similarly low frequency of homozygous CFHR1-deficient patients (16/1217, 1.3%; Table 2) when compared with an age-matched control group (Table 1). Except for one patient (0.08%), all CFHR1 deficient patients also showed homozygous deficiency of CFHR3. In total, non-deficiency resulted in an approximately 4.5-fold higher risk of developing ACVD. Taken together, these new findings indicate that ∆CFHR1 is associated with protection against ACVD.
FHR-1 binds to atherosclerotic plaques and stimulates blood monocytes and neutrophils. FHR-1 was found to localize to necrotic cells (Fig. 1A,B) in coronary artery plaques, but was not detected on healthy tissue adjacent to these plaques or on tissue derived from ∆CFHR3/1 patients (Fig. 1C). FHR1 co-localized with macrophages, as shown by co-staining with antibody to CD68 (Fig. 1D). Consistent with this observation, recombinant and serum FHR-1 13 , but not FHR2 derived from normal human serum (NHS), bound to oxidation-specific epitopes such as malondialdehyde-modified LDL (MDA-LDL; Fig. 1E,F) as we previously reported 13,20 . The detection of FHR2 was confirmed by ELISA (Fig. 1G). The pro-inflammatory functions of FHR-1 were evaluated using an ex vivo whole blood system, with FHR-1 triggering IL-1β release in whole blood (Fig. 1H). Both isolated primary blood monocytes 13 and monocyte-derived macrophages secreted www.nature.com/scientificreports/ pro-inflammatory cytokines and chemokines (IL-1β, IL-6, and IL-8) upon incubation with immobilized FHR-1 (Fig. 1I). In addition, immobilized but not soluble FHR-1 activated blood-derived neutrophils, with RT-qPCR showing that the levels of expression of IL-1β, IL-8, tumor necrosis factor-alpha (TNFα and the chemokine CCL3 were significantly upregulated (Fig. 1J). Because of the central role of monocytes in ACVD and our observation that FHR-1 promotes inflammation in monocytes and neutrophils, we assessed whether FHR-1 bound to necrotic surfaces, as is found in the necrotic core of atherosclerotic plaques, induces inflammation. Oxidation of LDL particles was reported to be an initial event in atherosclerotic plaque and foam cell formation, and to be involved in various pro-inflammatory mechanisms, including NLRP3 inflammasome activation 21 . FHR-1 binds to oxLDL and seems to act as a guardian of oxidized surfaces to induce immune responses. Components deposited within atherosclerotic lesions, such as cholesterol crystals 22,23 , oxLDL 21,24,25 , and complement, have been reported to induce inflammation 26 .
FHR-1 circulates in human and murine blood plasma on extracellular vesicles. Because FHR-1 was previously described as being a constituent part of lipoprotein particles 27 , we evaluated whether FHR-1 is transported on extracellular vesicles (EVs). EVs were therefore isolated from normal human serum as well as from FHR-1-deficient human serum (NHS ΔFHR-1 ) by polymer precipitation. EVs were separated from the supernatant, and both were immunoblotted with an anti-FHR-1 antibody. EVs from NHS but not from NHS ΔFHR-1 were positive for FHR-1 signals in the form of two typical glycosylated bands of 37 and 41 kDa, similar to (G) Detection of FHR-2 but not FHR1 by monoclonal antibody to FHR-2 mAb. gel gelatin. (H) Immobilized FHR-1 induces IL-1β secretion in whole human blood (diluted 1:10, 1:5, or 1:1), lipopolysaccharide (LPS) = control activator (I) Immobilized FHR-1 induces macrophage secretion of IL-1β IL-6 and IL-8 (BSA = bovine serum albumin) (J) upregulates expression of IL-1β, IL-8, TNFα, and CCL3 genes in neutrophils (zymosan = control activator; sFHR-1 = soluble FHR-1). Data in (E)-represent the mean ± SEM of three independent experiments with different donor cells. *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001 by unpaired twotailed t-tests.  Fig. 2A). The presence of FHR-1 on EVs was confirmed by staining EVs with fluorescently labeled anti-FHR-1 (Fig. 2B). EVs ranged in size from 50 to 200 nm, as measured by dynamic light scattering microscopy (DLSM) (Fig. 2C). Thus, FHR-1 is likely to be transported on exosomes, which range between 30 and 150 nm in diameter and originate from multi-vesicular bodies. EVs carrying FHR-1 (EV FHR-1 ) were isolated from this vesicle population by immune separation, with one of the 570 EVs isolated from NHS found to transport FHR-1 (Fig. 2D). Similar results were obtained with EVs isolated from mouse serum. The EVs ranged in size from 50 to 300 nm, whereas most of the FHRE-transporting vesicles ranged in size from 50 to 250 nm ( Fig. 2E), with one of the 280 EVs transporting FHRE (Fig. 2F).
To confirm that FHR-1 transporting EVs (EV FHR-1 ) captured with anti-FHR-1 coated beads indeed represent EVs, EV FHR-1 were stained with the vesicle marker anti-tetraspanin (CD9) and with SYTOX orange, a nucleic acid stain that permeates EVs and stains EVs transporting small RNAs. Fluorescent live-cell imaging showed CD9 and nucleic acid staining, confirming that FHR-1 is located on nucleic acid-transporting EVs (Fig. 2G), in contrast to EVs isolated from NHS ∆FHR1 . (Fig. 2H). Capture of CD9 vesicles from EVs transporting FHR-1 with anti-CD9 coated beads revealed that these EVs transported both CD9 and FHR-1 (EV FHR-1&CD9 ), with these EVs ranging in size from 30 to 200 nm as measured by DLSM (Fig. 2I). These findings confirm that FHR-1 is transported on extracellular levels.

FHR-1 concentrations are elevated in ACVD patients and correlate with disease markers.
To determine whether FHR-1 contributes to inflammation in ACVD, serum FHR-1 concentrations were measured in 244 patients (cohort 1; Table 1) with advanced ACVD awaiting bypass surgery, and in healthy subjects. FHR-1 concentrations were significantly higher in ACVD patients than in healthy individuals (39.2 ± 1.9 μg mL −1 vs. (26.5 ± 2.3 μg mL −1 , p < 0.0001; Fig. 3A) in contrast to complement factor H (Fig. 3B). FHR-1 concentrations were similar in men and women with ACVD ( Fig. 3C), with FHR-1 concentrations above normal (26.5 μg mL −1 ) correlating with patient body mass index (BMI) (Fig. 3D). Higher blood concentrations of LDL-cholesterol and higher LDL-to HDL-cholesterol ratios have been reported to increase the risk for ACVD 2 . The present study found that FHR-1 concentrations significantly correlated with LDL concentrations and LDL/HDL ratios, but not with HDL concentrations (Fig. 3E-G), suggesting a close association between FHR-1 and atherosclerosis. FHR-1 concentrations correlated with LDL concentrations in both men and women with ACVD (Fig. 3H,I).
The findings, that FHR-1 induces inflammation in vitro and correlates with inflammation in AAV 13 , suggested that FHR-1 could induce inflammation in patients with ACVD. Concentrations of the acute phase protein C-reactive protein (CRP) were higher in ACVD cohort 1 (10.3 ± 1.0 mg dL −1 ) than in the healthy donors (below the detection limit of 5 mg dL −1 ; Fig. 4A), with FHR-1 concentrations positively correlating with CRP concentrations (p < 0.001; Fig. 4B) and CRP concentrations positively correlating with concentrations of serum amyloid A (SAA), an acute phase protein 28 (p < 0.001; Fig. 4C). Furthermore, ACVD patients with FHR-1 levels > 80 µg mL −1 showed increased levels of SAA (Fig. 4D). FHR-1 concentrations also positively correlated with concentrations of neopterin (Fig. 4E), a marker of cellular immune system activation 29 . Because IL-6 is the major inducer of CRP and SAA, its concentration was also analyzed, with a significant correlation observed between serum FHR-1 and IL-6 levels (p < 0.05; Fig. 4F). These findings suggest that FHR-1 is an active mediator (marker) of inflammation. Detection of FHR-1 in the EV fraction from NHS (EV NHS ) but not from homozygous FHR-1-deficient (EV ΔFHR-1 ) human serum by western blot analysis. FHR-1 was also absent from the supernatant fractions of both sera. Results shown are representative of three experiments. An uncropped gel is shown in Supplementary Information. (B) Tracking of EV NHS and EV ΔFHR-1 by live-cell imaging using CLSM. EVs were stained with anti-FHR-1 antibodies conjugated with Alexa Fluor 647 (red). Bars: 10 µm. (C-F) Size distribution and number of vesicles isolated from EV FHR-1 and EV FHRE determined by DLSM using NanoSight NTA 3.2 software. Graphs in (C,E) represent overlays of results from 3 to 4 donors. ((C), SEM ± standard error, *p ≤ 0.0255 by unpaired two-tailed t-test, n = 4), and ((E), SEM ± standard error, *p ≤ 0.0232 by unpaired twotailed t-test, n = 3). (G) Tracking by live-cell imaging using CLSM of EV FHR-1 and EV ΔFHR-1 stained with anti-CD9 antibody (vesicle marker) and SYTOX orange (nucleic acid marker), Bars: 10 µm. (H) Size distribution of EVs transporting CD9 and FHR-1 (EV CD9&FHR-1 ). EV FHR-1&CD9 were captured with anti-CD9-coated beads from EV FHR-1 , isolated from 1 mL NHS and analyzed by DLSM using NanoSight NTA 3.2 software. (I) Particle numbers (J) and sizes (K) in fractions obtained by size-exclusion chromatography of EV NHS measured by DLSM. (L) High FHR-1 content in fractions 13-16 determined by ELISA. FHR-1 (red) (M) and FHR-1, and vesicle marker CD9, were in close proximity (red) (N) in atherosclerotic tissues, mainly in blood vessels (stippled lines). Complexes were analyzed by proximity ligation assays using anti-FHR-1 and anti-CD9, and were detected by CLSM. EV FHR-1 carry nucleic acids (orange). Bars = 10 µm.

FHR-1 increases tissue factor (TF) expression in monocytes. Inflammatory mediators have been
shown to play pivotal roles in thrombus formation, a severe complication of atherosclerosis, as well as in inflammation 2,28-30 . Therefore, the involvement of FHR-1 in coagulation was evaluated. The number of thrombocytes positively correlated with FHR-1 concentration in ACVD patients (p < 0.005, cohort 1; Fig. 5A) as well as Quick test (Fig. 5B) suggesting the involvement of FHR-1 in thrombus formation. To confirm whether FHR-1-treated monocytes directly activate the coagulation cascade, we evaluated the activation of signaling pathways by RNA sequencing in monocytes incubated for 4 h with immobilized FHR-1 or bovine serum albumin (BSA). Several genes involved in the coagulation cascade, including the gene encoding TF, also called factor III and F3, were found to be significantly upregulated in FHR-1-treated monocytes (Fig. 5C). Flow cytometry showed that in vitro treatment of blood-derived monocytes with FHR-1 resulted in a substantial increase in cell surface TF expression. This effect was inhibited by the addition of pharmacologic inhibitors of phospholipase C (PLC) or NF-κB (Fig. 5D), the major signaling molecules downstream of FHR-1. To determine whether FHR-1-induced TF on monocytes is functionally active, monocytes were treated with FHR-1 for 20 h, washed, and then incubated with a chromogenic factor VIIa substrate, which is cleaved by TF. FHR-1-treated monocytes showed increased cleavage of Factor VIIa compared with untreated monocytes (Fig. 5E). In addition, inhibition of PLC or NF-κB significantly reduced Factor VIIa cleavage by FHR-1-treated monocytes (Fig. 3E).
FHR-1 has been shown to activate the PLC pathway via the G protein-coupled receptor EMR2. EMR2 inhibition for 4 h also reduced the cleavage of Factor VIIa (Fig. 5F) and inhibited TNFα expression (Fig. 5G) in FHR-1-treated monocytes These results show that, in addition to pro-inflammatory genes, FHR-1 induces the expression in monocytes of TF, a protein involved in coagulation.

Discussion
Inflammation is a hallmark of many diseases and represents an innate immune response to tissue injury, infection, and metabolic stress. FHR-1 has been shown to act as an immune sentinel by binding specifically to necrotic cell surfaces and inducing pro-inflammatory cytokine release in monocytes. The present study showed that FHR-1 is associated with ACVD. Plasma concentrations of FHR-1 were higher in ACVD patients than in normal individuals and significantly correlated with LDL-but not HDL-cholesterol concentrations in both men and women patients with ACVD. Moreover, homozygous deficiency of the CFHR1 gene, which encodes FHR-1, was significantly less frequent in two cohorts of ACVD patients than in control individuals. Monocytes bearing FHR-1 were found to induce neutrophils to express pro-inflammatory cytokines. By binding to surfaces such as atherosclerotic plaques, this plasma protein is expected to activate monocytes/macrophages and neutrophilic granulocytes in ACVD patients, inducing inflammation and exacerbating this chronic disease (Fig. 6). Inflammation is a fundamental multistep and highly regulated process for fighting infections, repairing tissue injury, and re-establishing homeostasis. Although neutrophils and monocytes also cause cell damage and necrosis, their phagocytic actions and interaction are essential for proper regeneration. However, sustained immune responses in patients with persistent inflammation can lead to serious inflammatory injury. In ACVD, pro-inflammatory cytokines enhance the infiltration of immune cells into plaques and predispose vessels to future thromboembolic events 31 . The present study demonstrated that surface-bound, but not circulating, FHR-1 triggers neutrophilic granulocytes to upregulate the transcription and release of pro-inflammatory cytokines such as IL-1β, IL-8, and TNFα. FHR-1 bound to necrotic cells induces inflammation in human blood-derived monocytes by interacting with monocytes via the EMR2 receptor. The present study confirmed FHR-1 stimulation in a whole blood model. FHR-1 is a dimeric protein circulating in plasma with functions within and independent of the complement cascade. FHR-1 binding to C3b competes with factor H binding to C3b, reducing the inhibitory activities of factor H through a process called deregulation 11,12 . Moreover FHR-1 was shown to inhibit the activation of terminal pathway proteins in vitro 10 . FHR-1, but not FHR-2, on necrotic cell surfaces with exposed oxidized lipids, as in atherosclerotic plaques, activates NLRP3 in monocytes 13 . Evidence showing that FHR-1 is localized to atherosclerotic plaques is consistent with its binding to modified or damaged surfaces.
The present study also showed that FHR-1 is transported through the blood on EVs, likely exosomes. Exosomes, which originate from multi-vesicular bodies, range in size from 30 to 100 nm, but show a considerable overlap with microvesicles 100-1000 nm in size 32 . No FHR-1 signal was detected in supernatants collected after vesicle isolation, indicating that FHR-1 is predominantly associated with vesicles rather than circulating as a soluble protein in plasma. As expected, vesicles isolated from NHS∆ FHR-1 lacked FHR-1 signals, confirming the specificity of the isolated exosomes. EV FHR-1 carry vesicle markers such as CD9 and transport nucleic acids, likely miRNAs. CD9 and miRNA are EV-but not lipoprotein-specific. FHR-1 was reported to be a constituent of HDL blood particles composed of phospholipids, apoA-I, the LPS binding protein LBP, and FHR-2, of average    33,34 . Because of its ability to bind to artificial vesicles, nonlipidated plasma apoA-I may be associated with EVs 35 . The present study clearly demonstrates that these vesicles contain FHR-1, in addition to factor H 36,37 . Because FHR-1 in NHS is not inflammatory, FHR-1 on exosomes is expected to be in an inactive form but likely becomes activated due to a conformational change upon binding to necrotic surfaces. Indeed, serum-derived FHR-1 has been shown to induce inflammation upon binding to necrotic surfaces 13 . Exosomes are released by many different cells and, as transport carriers, participate actively in intercellular communications 32,35,38 . Upon binding to damaged surfaces, FHR-1 triggers inflammation and likely upregulates CFHR1 gene expression in liver cells. FHR-1 concentrations were significantly higher than in normal controls (p < 0.0001). Serum concentrations of FHR-1, however, have been reported to vary from 25 to 95 μg mL −1 , depending on the ELISA kit. To date, elevated FHR-1 concentrations have been reported in patients with AAV 13 and IgAN 39,40 , and correlate with disease progression. FHR-1 was also identified as a biomarker of tuberculosis 41 and acute cellular rejection 42 , diseases characterized by high levels of cell damage and necrosis.
The significantly higher plasma concentrations of FHR-1 in ACVD patients correlate with the inflammatory markers IL-1β, CRP, SAA, and neopterin. Thus, FHR-1 may represent a general cell death-triggered inflammation marker. FHR-1 is found in atherosclerotic plaques and induces isolated monocytes/macrophages and neutrophils to release pro-inflammatory cytokines, including IL-6, which elevates CRP and SAA expression. Stimulation of human monocytes/macrophages with interferon-y induces the production and secretion of neopterin, an indicator of increased cellular immune reactions and oxidative stress 29 . Monocytes and macrophages in particular play important roles in the pathogenesis of atherosclerosis [43][44][45] . Monocytes infiltrate plaques during their formation and differentiate into macrophages within these lesions 46,47 . In addition, impaired monocyte infiltration into plaques has been shown to attenuate atherogenesis 48,49 . Moreover, monocytosis is an independent risk factor for ACVD, with monocytes causally linked to the progression of atherosclerotic lesions 4,47,50,51 . Targeting inflammation in ACVD patients with anti-inflammatory drugs, such as canakinumab or cholchicine, had a significant impact on the disease as it reduced the incidence of secondary cardiovascular events, including thrombotic heart failure, as shown in the CANTOS and COLCOT trials [52][53][54][55] .
The suggested role of FHR-1 in inflammation is corroborated by studies in ACVD patient cohorts, which revealed that homozygous deficiency of CFHR1 was less frequent in ACVD patients than in control cohorts. Similarly, the frequency of homozygous deficiency of CFHR1 was lower in AMD patients (0.8-1.2%) than in control subjects (4.9-5.2%) 11,15,17 . These results are in agreement with results from a study showing that AMD was associated with atherosclerosis, and that a polymorphism in complement factor H (rs1061170) associated with AMD was also associated with an increased risk of myocardial infarction 56,57 . Homozygous CFHR1 deficiency is, in most cases, accompanied by homozygous CFHR3 deficiency because of the loss of a chromosomal fragment harboring both genes during homologous recombination 19 . However, in vitro assays showed that immobilized FHR3 did not induce inflammation in human monocytes 13 .
Taken together, these results show that FHR-1 on exosomes binds to atherosclerotic plaques, and that this binding increases the production by monocytes and neutrophils of inflammatory cytokines via NLRP3 activation, enhancing the infiltration of immune cells into plaques. As FHR-1 also correlates with CRP and induces IL-6 production and TF expression, FHR-1 may affect the risks of coagulation and thromboembolic events. However, although FHR-1 serum concentrations are substantially higher in ACVD patients than in controls, and although homozygous chromosomal deletion of CFHR1 is less frequent in two cohorts of ACVD patients than in normal Figure 6. Model of FHR-1-induced inflammation. Extracellular vesicles that transport FHR-1 (EV FHR1 ) bind to oxLDL in necrotic cores of atherosclerotic plaques and activate neutrophils and monocytes. These monocytes and neutrophils release inflammatory cytokines, which attract more immune cells. FHR-1 also induces monocytes to express tissue factor on their surfaces, which can trigger the coagulation pathway and thrombosis, the ultimate complication of atherosclerosis. www.nature.com/scientificreports/ individuals, we cannot exclude the possibility that these results are due, at least in part, to other factors, such as the deregulation of factor H by elevated levels of FHR-1 or a genetic constellation of unknown linkage disequilibrium. Further studies are therefore needed to confirm the above results and to analyze the mechanisms of action of FHR-1. Because inflammation exacerbates atherosclerosis, the inhibition of FHR-1 or its associated signaling pathways should be evaluated to determine whether inhibition of FHR1 would be beneficial to ACVD patients.

Scientific Reports
Cohort 2. Subjects were selected from the INTERCATH cohort at the Department of General and Interventional Cardiology, University Heart & Vascular Center, Medical University Hamburg-Eppendorf. The INTER-CATH cohort is an ongoing, observational cohort study that has included patients undergoing coronary angiography beginning in January 2015, and has been described in detail previously 58 . In brief, all patients have been characterized according to their medical history including comorbidities, medications, lifestyle, socioeconomic information, and laboratory parameters. Coronary angiograms have been analyzed by experienced cardiologists, with the complexity in patients with confirmed coronary artery disease quantified using the SYNTAX score, which was calculated using the internet-based SYNTAX calculator version 2.10 (www. synta xscore. com). The present analyses included all patients with available SYNTAX scores but excluded all patients with acute coronary syndrome. Genomic DNA was extracted from buffy coats 59 . After quality control and measurement of DNA concentrations, genomic DNA was stored at − 80 °C until further analyses. Immunohistochemistry. Atherosclerotic plaques were obtained from patients with high-grade aortic sclerosis who underwent concomitant ascending aorta replacement during coronary artery bypass grafting. Human coronary plaques were segmented into 3-4 mm blocks, which were fixed overnight in formalin, decalcified in 0.5 M EDTA (pH 7.2), and embedded in paraffin, as previously described 13 . Tissue section (1 μm) were preincubated with a protease solution for 30 min at 40 °C, and incubated with a monoclonal anti-FHR-1 antibody and an anti-CD68 antibody 10 (1:40.000 each) overnight at 4 °C. After washing, the sections were incubated with an AffiniPure rabbit anti-mouse IgG (H + L; Jackson ImmunoResearch) secondary antibody and an anti-alkaline phosphatase antibody (APAAP complex; MyBioSource), followed by development using ZytochemPlus/ POLAP100 (Zytomed Systems). Slides were incubated with fuchsin naphthol As-Bi phosphate substrate solution for 30 min, and nuclei were counterstained with Mayer's hematoxylin solution (MilliporeX) for 1 min. Staining was quantified using ImageJ or ZEN software (version 2, blue edition).
Analysis of serum samples. FHR-1 and factor H concentrations in serum samples were determined using human FHR1 (RayBiotech) and factor H (Biomatik) ELISA kits according to the manufacturers' protocols. The absence of FHR-1 from serum samples was confirmed by western blot analysis and by PCR, as described 19 . CRP concentrations were measured using a Cobas8000 analyzer (Roche Diagnostics) on module C701. The CRP concentrations in all samples below the CRP limit of detection of 5 mg dL −1 were defined as 5 mg dL −1 . Vesicle biochemistry. Vesicle isolation. Vesicles were isolated from sterile serum and diluted with RPMI-1640. Vesicles were precipitated using ExoQuick-TC (Systems Bioscience) 60 , according to the manufacturer's protocol. Both the vesicle precipitates and supernatants were collected and stored at − 20 °C for no longer than 1 month. EVs were also isolated from 1 mL fresh human serum by size-exclusion chromatography using Sepharose CL-2B (Sigma-Aldrich), as described 61 . Sixteen consecutive 1 mL fractions were eluted from each column and stored at − 20 °C until further processing. Upon reuse, the fractions were thawed on ice.
Targeted vesicle isolation. Specific FHR-1-transporting EVs were isolated from the total vesicle population using M-pluriBeads (pluriSelect), according to the manufacturer's instructions. The beads were coated with monoclonal antibody to human FHR-1 10 . FHRE-transporting vesicles were isolated using M-pluriBeads (pluriSelect) coated with a polyclonal antiserum to mouse FHRE generated by immunization with recombinant PLA assays were performed to evaluate the co-localization of the vesicle markers CD9 and FHR-1. Vesicles isolated by precipitation from human serum were seeded onto 6.7 mm diagnostic slides coated with poly-l-lysine. The vesicles were fixed with 4% formaldehyde, blocked with Duolink blocking solution (Sigma-Aldrich), and incubated with mouse antibody to human FHR-1 13  Western blot analysis. To detect vesicle proteins, samples were resolved in non-reducing buffer (Rotiload 1:4), loaded onto 10% polyacrylamide gels, subjected to electrophoresis, and blotted onto polyvinylidene fluoride (PVDF) membranes (Sigma-Aldrich). FHR-1 was detected by incubating the membranes with a monoclonal antibody to human FHR-1 10 (1 µg mL −1 ) and a goat anti-mouse immunoglobulin-conjugated HRP secondary antibody (1:1000; Dako, cat no. P044701-2). Pictures were captured with Fusion.
ELISA. FHR-1 concentrations were measured in size-exclusion chromatography fractions using Human FHR-1 ELISA kits (RayBiotech, cat no. ELH-CFHR1) according to the manufacturer's instructions. Absorption at 450 nm was measured on a TECAN Safire2 microplate reader.
Immunohistochemistry and PLA. Paraffin-embedded atherosclerotic plaques were prepared as described. Paraffin-embedded liver sections were deparaffinized by treatment with Roticlear (Carl Roth), 100% ethanol, and 90% ethanol. These tissue samples were boiled in 10 mM Na citrate buffer (pH 6.5) for antigen retrieval and permeabilized with 0.1% saponin (Sigma-Aldrich). Tissues were blocked with normal serum block (Biolegend) and human FcR Blocking Reagent (Miltenyi) diluted in antibody diluent (Carl Roth). Tissue samples were incubated with mouse monoclonal FHR-1 antibody 13 (1:200) and goat anti-mouse IgG (H + L) conjugated to Alexa Fluor 647 (Thermo Fisher Scientific, cat no. A-21235). Nuclei were stained with SYTOX orange. Images were captured with LSM 710 fitted with ZEN 2011 software.
For PLA, deparaffinized and blocked tissue samples were incubated with mouse anti-FHR-1 antibody 13 (1:200), with or without rabbit anti-CD9 antibody (1:200; Abcam, cat no. ab92726). PLA assays were performed using the Duolink In Situ Red Starter Kit Mouse/Rabbit (Sigma-Aldrich, cat no. DUO92101) according to the manufacturer's protocol. Images were captured with LSM 710 fitted with ZEN 2011 software. Statistical analysis. Significant differences between groups were analyzed by unpaired two-tailed Student's t-tests using GraphPad Prism version 5. Atherosclerosis patients were analyzed using unpaired two-tailed t-tests with Welch's correction. Correlations were analyzed using Spearman's correlation tests. CAIA studies were analyzed using ANOVA and t-tests. p-values ≤ 0.05 were considered statistically significant. Odds ratios were calculated using GraphPad Prism version 5. Study approval. Informed consent was obtained from all subjects and/or their legal guardian(s) and patient data, blood, and tissue samples were collected according to the protocls. The experimental protocols were approved by the local ethics committees of the Medical University Hamburg-Eppendorf (approval numbers: PV3162, PV4068, and PV5657), Friedrich-Schiller University (approval number: 5071-02/17), and Ludwig Maximilians University, and according to the Guidelines of the World Medical Association Declaration of Helsinki. License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.