Targeting myeloid cell coagulation signaling blocks MAP kinase/TGF-β1–driven fibrotic remodeling in ischemic heart failure

Despite major advances in acute interventions for myocardial infarction (MI), adverse cardiac remodeling and excess fibrosis after MI causing ischemic heart failure (IHF) remain a leading cause of death worldwide. Here we identify a profibrotic coagulation signaling pathway that can be targeted for improved cardiac function following MI with persistent ischemia. Quantitative phosphoproteomics of cardiac tissue revealed an upregulated mitogen-activated protein kinase (MAPK) pathway in human IHF. Intervention in this pathway with trametinib improves myocardial function and prevents fibrotic remodeling in a murine model of non-reperfused MI. MAPK activation in MI requires myeloid cell signaling of protease-activated receptor 2 linked to the cytoplasmic domain of the coagulation initiator tissue factor (TF). They act upstream of pro-oxidant NOX2 NADPH oxidase, ERK1/2 phosphorylation, and activation of profibrotic TGF-β1. Specific targeting with the TF inhibitor nematode anticoagulant protein c2 (NAPc2) starting 1 day after established experimental MI averts IHF. Increased TF cytoplasmic domain phosphorylation in circulating monocytes from patients with subacute MI identifies a potential thromboinflammatory biomarker reflective of increased risk for IHF and suitable for patient selection to receive targeted TF inhibition therapy.


Introduction
Proper wound healing depends on effective hemostasis, which allows regeneration and reconstitution of defective tissue function (1). Coronary thrombosis causing myocardial infarction (MI) (2) leads to irreversible tissue injury, particularly when timely reperfusion is not achieved. Activation of the innate and adaptive immune system mediates aspects of both cardiac dysfunction and functional repair (3). These immune responses support wound repair and collagen deposition that may lead to restoration of cardiac function or inflammation-mediated adverse left ventricular (LV) remodeling resulting in ischemic heart failure (IHF) (4). Targeting innate immune responses for secondary prevention after an MI was effective but did not result in a net clinical bene-fit owing to side effects (5), warranting tailored and more specific antiinflammatory approaches in coronary heart disease, in particular for the prevention of IHF.
Myeloid cells express tissue factor (TF), which forms a complex with factor VIIa (FVIIa) for FXa generation (6), and TF significantly contributes to multifaceted intravascular cell activation in immunothrombosis (7). Long-term anticoagulation is an established therapy to prevent major cardiovascular events (8,9). Preclinical evidence suggests beneficial effects of anticoagulants (10) or blockade of protease-activated receptors (PARs) (11) in the infarcted myocardium. Inhibition of cardiomyocyte-specific TF or thrombin attenuates myocardial injury and inflammation (12), but the underlying mechanisms linking coagulation proteases to immune cell activation and outcomes after MI remain unclear.
The temporal transition of an inflammatory phase to fibrotic remodeling after MI depends on the release of growth factors like transforming growth factor-β1 (TGF-β1) (13) activating myofibroblasts (14). Coagulation proteases mediate clotting-independent PAR activation to regulate not only hematopoiesis (15) and tumor development (16) but also cardiac fibrosis in a model of heart failure with preserved ejection fraction by modulating TGF-β1 signaling (17). Transformation of tissue resident fibro-Despite major advances in acute interventions for myocardial infarction (MI), adverse cardiac remodeling and excess fibrosis after MI causing ischemic heart failure (IHF) remain a leading cause of death worldwide. Here we identify a profibrotic coagulation signaling pathway that can be targeted for improved cardiac function following MI with persistent ischemia. Quantitative phosphoproteomics of cardiac tissue revealed an upregulated mitogen-activated protein kinase (MAPK) pathway in human IHF. Intervention in this pathway with trametinib improves myocardial function and prevents fibrotic remodeling in a murine model of non-reperfused MI. MAPK activation in MI requires myeloid cell signaling of proteaseactivated receptor 2 linked to the cytoplasmic domain of the coagulation initiator tissue factor (TF). They act upstream of pro-oxidant NOX2 NADPH oxidase, ERK1/2 phosphorylation, and activation of profibrotic TGF-β1. Specific targeting with the TF inhibitor nematode anticoagulant protein c2 (NAPc2) starting 1 day after established experimental MI averts IHF. Increased TF cytoplasmic domain phosphorylation in circulating monocytes from patients with subacute MI identifies a potential thromboinflammatory biomarker reflective of increased risk for IHF and suitable for patient selection to receive targeted TF inhibition therapy.
Targeting myeloid cell coagulation signaling blocks MAP kinase/TGF-β1-driven fibrotic remodeling in ischemic heart failure Results Proteomic identification of coagulation and innate immune activation in human IHF. To gain insight into the molecular pathophysiology of the failing human heart in chronic IHF, we performed an unbiased label-free quantitative proteomic and phosphoproteomic profiling of cardiac tissue obtained from 5 patients with severe advanced IHF compared with 5 control donor hearts. High-resolution-accuracy mass spectrometric analysis allowed for the quantification of 2,714 proteins and 10,601 phosphopeptides, of which 208 proteins and 685 phosphopeptides were significantly changed with fold changes greater than 2 or less than 0.5 in at least 60% of all measurements in 1 group ( Figure  1, Supplemental Figures 1-6, and Supplemental Tables 1 and 2; supplemental material available online with this article; https:// doi.org/10.1172/JCI156436DS1). Gene ontology analysis revealed a significant enrichment of pathways related to complement and coagulation cascades, innate immune pathways, platelet activation, and endocytosis in the IHF group relative to control (Figure 2A and Supplemental Table 3). Thrombosis and its sequelae are established hallmarks of acute MI and atherothrombotic complications in vascular diseases. The prominent changes in coagulation, complement, and innate immune responses suggested a novel and specific coagulation-inflammatory signature that persisted after acute ischemic events and might have ultimately led to chronic IHF in our samples of severe human ischemic heart disease.
Protein-protein interaction analysis with Cytoscape STRING-DB app identified phosphorylation of mitogen-activated protein kinase 1 (MAPK1, or extracellular signal-regulated kinase [ERK]) at Thr185 and/or Tyr187 as a central node in the integrated network kinase analysis ( Figure 2B). In addition, Integrative Inferred Kinase Activity (InKA) analysis revealed that MAPK1 is a (hyper)active kinase (average InKA score of 49) in IHF but not in controls ( Figure 2C). Thus, our analysis uncovered a central signaling pathway that may be related to the coagulation and innate immune system changes in human IHF. We therefore performed preclinical and clinical studies to establish the relevance of the MAPK pathway in MI and its progression to IHF as well as to elucidate the upstream coagulation-related signaling pathway involved and potential specific therapeutic targeting.
Blockade of MAPK1 in a preclinical model of MI ameliorates the development of IHF. We used the well-established preclinical mouse model of MI induced by permanent ligation of the left anterior descending coronary artery (LAD) to study the development of IHF. We first localized phosphorylated ERK1/2 by immunofluorescence staining in the infarcted zone 7 days after MI ( Figure 3A).
blasts to activated myofibroblasts depends on TGF-β1 signaling (18), but cellular sources and mechanisms that regulate TGF-β1 production and activation in the infarcted myocardium remain incompletely understood.
In the present study, we identified persistent activation of coagulation as a prominent local feature in human IHF. We conducted genetic and pharmacological investigations to dissect crucial coagulation protease-mediated signaling events that promote excess cardiac fibrosis and adverse remodeling. These experiments identified an unexpected role for myeloid cell TF-PAR2 signaling as a crucial driver for TGF-β1 activation and a potential target for therapeutic intervention in MI to avert the development of IHF. Figure 1. The proteomic signature of human IHF. Proteins isolated from cardiac tissue from human IHF (n = 5) and non-IHF donor hearts (controls, n = 5) were analyzed by label-free quantitative proteomics. Samples were measured in quadruplicate (proteins) and triplicate (phosphopeptides) liquid chromatography-tandem mass spectrometry (LC-MS/MS) runs. Heatmaps show the quantity profile for 208 proteins and 685 phosphopeptides significantly differing (P < 0.05, fold change > 2 or < 0.5, and identified in >60% of all LC-MS/MS replicates in either control or IHF group), log 10 -transformed and row-wise normalized using z score and sorted by descending fold change.
or cardiomyocytes, were a prominent location for ERK1/2 activation throughout the course of acute MI in border and infarcted zones, but not remote and unaffected regions of the myocardium ( We then evaluated the functional contributions of the central MAPK hub identified by the (phospho)proteome analysis. We blocked the mitogen-activated extracellular signal-regulated kinase (MEK), an ERK1/2 activator, with the MEK1/2 inhibitor trametinib ( Figure 3C). Long-term, high-dose trametinib regimens in cancer therapy have cardiotoxic side effects (19). Interestingly, short-term treatment with trametinib at a reduced dose of 1 mg/ kg/d initiated 1 day and continued for 6 days after permanent LAD ligation attenuated the deterioration of cardiac function 7 days ERK1/2 phosphorylation was detectable as early as 1 day after MI (Supplemental Figure 7) in the infarcted zone and increased from day 1 to day 7 in the infarcted and border zones ( Figure 3B) but not in nonischemic remote regions of the infarcted hearts (Supplemental Figures 8 and 9). Colocalization of ERK1/2 phosphorylation with CD45 + immune cells, α-smooth muscle actin-positive (α-SMA + ) myofibroblasts, CD31 + endothelial cells, and cTNT + cardiomyocytes showed that ERK1/2 activation primarily occurred in infiltrating immune cells in the infarcted zone at day 7 ( Figure 3B). Note that α-SMA did not colocalize with CD31 (Pearson's coefficient value -0.2267), implying that α-SMA + cells are primarily myofibroblasts and not smooth muscle cells in the ischemic myocardium. Immune cells, but not myofibroblasts, endothelial cells,  Figure 1. (B) Differentially regulated (P < 0.05) nonphosphorylated and phosphorylated proteins were analyzed for their previously known protein-protein interactions between each other by Cytoscape STRING-DB search and sorted by the number of interactions in a circular layout. Among the present kinases, MAPK1 shows the most interactions, followed by mTOR, TTN, CDK13, PI4KA, and TAOK2. (C) Volcano plot depicting significantly differing Integrative Inferred Kinase Activity (InKA) scores between IHF and control groups. cytes (3) and subsequent expansion of Ly6C lo monocytes and macrophages (21) orchestrate the inflammatory reaction within the infarcted myocardium. Trametinib significantly reduced mRNA levels of monocytic C-C chemokine receptor 2 (Ccr2), which is after MI (Figure 3D), while it did not significantly impact infarct size at days 1 and 3 after MI (Supplemental Figure 10A).
Based on these data, we subjected PAR2 fl/fl LysM Cre mice to permanent LAD ligation for 7 days. Myeloid cell PAR2-deficient mice showed reduced TGF-β1 activation with unaltered latent TGF-β1 levels, reduced phosphorylation of the TGF-β1 target small mothers against decapentaplegic homolog 2 (SMAD2) in the myocardium ( Figure 5C), and protection from cardiac dysfunction ( Figure  5D). Immune cell recruitment into the infarcted myocardium was not diminished in myeloid cell PAR2-deficient mice (Supplemental Figure 12D). Four weeks after LAD ligation, PAR2 fl/fl LysM Cre mice compared with PAR2 fl/fl littermate controls had significantly improved survival ( Figure 5E) and improved cardiac function as well as decreased collagen deposition based on sirius red staining of myocardial cross sections ( Figure 5F and Supplemental Figure  12E). Thus, myeloid cell PAR2 supports local TGF-β1 activation and drives cardiac remodeling following MI.
Costaining for TF and CD45 revealed a significant increase of CD45/TF double-positive cells in the infarcted myocardium (Figure 6A). Confocal microscopy imaging of infarcted myocardium of TF fl/fl LysM Cre mice confirmed TF localization in myeloid cells (Supplemental Figure 13A). TGF-β1-mediated SMAD2 activation in the infarcted myocardium is known to start at day 7 and persist until 28 days after MI (22,28). There was no difference in infarct size and cardiomyocyte apoptosis 1 day after MI in TF fl/fl LysM Cre mice relative to TF fl/fl littermate controls (Supplemental Figure 13, B and C). In line with myeloid cell PAR2 deletion, TF deficiency led to significantly reduced cardiac ERK1/2 activation ( Figure  6B), TGF-β1 activation, and SMAD2 phosphorylation ( Figure  6C) as well as mRNA expression of TGF-β1-induced COLO1A1, COLO3A1, and ACTA2 (Supplemental Figure 14A). This resulted in improved cardiac function ( Figure 6D) without influencing myeloid cell infiltration into the infarcted myocardium 7 days after MI (Supplemental Figure 14B).
Long-term follow up of TF fl/fl LysM Cre mice revealed reduced collagen deposition, improved survival ( Figure 6, E and F), and better cardiac function 4 weeks after MI in comparison with TF fl/fl controls (Supplemental Figure 14C). Immunofluorescence staining revealed an increase of both CD45 + p-SMAD2 + and α-SMA + p-SMAD2 + cells in the infarcted myocardium 7 days after MI (Figure 7A). Myeloid cell TF deletion did not reduce CD45 + p-SMAD2 + cells, but α-SMA + p-SMAD2 + cells as well as total α-SMA protein expression in the myocardium were significantly reduced in TF fl/fl LysM Cre mice compared with TF fl/fl controls (Figure 7, A and B). Thus, TF-PAR2 signaling in infiltrating myeloid cells is responsible for hyperactivation of the MAPK pathway, TGF-β1 activation, and activated myofibroblast accumulation leading to fibrotic cardiac remodeling after MI and the development of IHF.
TF cytoplasmic domain signaling is linked to NOX2/ERK-dependent TGF-β1 activation in permanent MI. We next asked whether TF makes direct signaling contributions to profibrotic TGF-β1 activation. Ligation of TF by FVIIa activates rac and p38 dependent on the TF cytoplasmic domain (29). In the context of PAR signaling, the TF cytoplasmic tail binds the regulatory subunit of PI3 kinase and rac adaptor p85 (27) and recruits the NADPH oxidase for endosomal translocation and reactive oxygen species (ROS), primarily superoxide anion (O 2

•−
), production (26). We found that essential for myeloid cell recruitment. Importantly, trametinib did not reduce mRNA levels of IL-6, TNF-α, and CCL (Ccl2), implying that the local release of inflammatory mediators after MI (4) was independent of the ERK1/2 pathway ( Figure 4A). Cardiomyocyte apoptosis was marginally reduced and CD45 + immune cell infiltration was unchanged 1 and 3 days after MI (Supplemental Figure 10, B and C). In contrast, recruitment of CD45 + CD11b + Ly6C hi and Ly6C lo myeloid cells into the infarcted heart was significantly reduced by trametinib at day 7 ( Figure 4B and Supplemental Figure 11), in line with the reduced Ccr2 mRNA levels. Trametinib therapy also had no significant effects on cardiac function in sham-operated mice from day 1 to day 7 and in ligated mice from day 1 to day 3 (Supplemental Figure 10D) and did not significantly interfere with p-JNK/SAK or p38 MAPK activation (Supplemental Figure 12A), demonstrating specificity.
TGF-β1 has been implicated in adverse LV remodeling and development of IHF (4,22). In the infarcted zone, trametinib not only prevented phosphorylation of ERK, but also reduced activation of TGF-β1, the major driver of adverse cardiac fibrotic remodeling ( Figure 4C). Likewise, mRNA levels of COLO1A1 and COLO3A1, coding for collagen type I and III α1 chains, as well as Posn and ACTA2, coding for the fibrosis markers periostin and α-SMA, as potential targets of TGF-β1 signaling were significantly reduced in the infarcted myocardium (Supplemental Figure 12B). Next, we investigated the role of the MAPK pathway for TGF-β1 activation directly in immune cells. Protein expression analysis on isolated peripheral blood mononuclear cells (PBMCs) from the LAD-ligated mice revealed an increased ERK1/2 activation along with the increased TGF-β1 activation ( Figure 4D). Thus, infiltrating CD45 + cells were the major source of ERK1/2 phosphorylation in the infarcted zone, and the MEK1/2 inhibitor trametinib prevented TGF-β1 activation in circulating PBMCs, suggesting that TGF-β1 activation was specifically regulated by immune cells after MI.
TF cytoplasmic domain phosphorylation in experimental MI and in the clinical settings of IHF. Immunohistochemical staining of ischemic myocardium showed a relative abundance of TGF-β1 + Ly6C + inflammatory cells compared with TGF-β1 + CD31 + cells in control mice, but not in TF ΔCT mice ( Figure 9A). To further analyze the role of the TF cytoplasmic tail in myeloid cell TGF-β1 activation, we generated bone marrow (BM) chimeras of TF ΔCT and WT mice. After 9-10 weeks of confirmed engraftment (Supplemental Figure  16, A and B), permanent MI was induced in chimeric mice for analysis 7 days later. Whereas CD45 + CD11b + myeloid cell recruitment to the infarcted myocardium was indistinguishable between transplant groups (Supplemental Figure 16C), only chimeras with BM from TF ΔCT mice (TF ΔCT →WT) had reduced cardiac NOX2 expression, ERK1/2 phosphorylation, and TGF-β1 activation ( Figure 9B) paralleled by attenuated SMAD2 phosphorylation after MI (Supplemental Figure 17A). Six weeks after permanent LAD ligation, only TF ΔCT →WT mice showed improved cardiac function, thereby phenocopying the TF ΔCT animals (Supplemental Figure 17B). Thus, this preclinical evidence links myeloid cell TF-PAR2 signaling to TF cytoplasmic tail-dependent NOX2 activation, ERK phosphorylation, and TGF-β1 activation in MI and adverse cardiac remodeling leading to IHF.
Phosphorylation of the TF cytoplasmic domain was detectable specifically in CD45 + cells 7 days after experimental MI in WT but not in TF ΔCT mice ( Figure 10A), suggesting that activation of the TF-dependent profibrotic pathway can be identified by measurement of this posttranslational modification of TF. We employed previously validated phosphorylation-specific antibodies against the TF cytoplasmic domain (32) to translate this finding to human IHF. The numbers of CD45 + cells stained for phosphorylated TF were markedly increased in LV tissue samples obtained from IHF patients as compared with donor heart tissue ( Figure 10, B and C, and Supplemental Table 4). This increased TF phosphorylation was accompanied by upregulation of IL-6, CCL2, and CCR2 in the heart tissue (Supplemental Figure 17C), indicating myeloid cell recruitment, as well as increased TGF-β1 activation and downstream phosphorylation of SMAD2 ( Figure 10D), indicating subsequent cardiac fibrotic remodeling.
TF was upregulated on circulating PBMCs in mice after MI ( Figure 11, A and B). We next asked whether TF phosphorylation in liquid biopsies could be used for identifying patients at risk for IHF. Patients with subacute, prolonged MI that is not timely reperfused have a 2-fold higher risk of death and of developing heart failure compared with MI patients presenting early for reperfusion therapy (33). In a sample taken from our observational MICAT study, we focused on patients with subacute MI compared with stable coronary artery disease (CAD) admitted for percutaneous coronary intervention (PCI) (Supplemental Table 5). TF cytoplasmic domain TF fl/fl LysM Cre mice with permanent LAD ligation had significantly reduced cardiac NOX2 expression compared with controls and that monocytes isolated from PAR2 -/mice had reduced NOX2 expression when exposed to hypoxia plus cytokines (Supplemental Figure 15, A and B). Likewise, O 2 •− formation in the ischemic myocardium of PAR fl/fl LysM Cre mice was reduced in comparison with PAR fl/fl littermate controls (Supplemental Figure 15C).
In line with results in TF fl/fl LysM Cre and PAR fl/fl LysM Cre mice, cytoplasmic tail-deficient mice (TF ΔCT mice) had markedly reduced CD45 + NOX2 + immune cell infiltration ( Figure 8A) and O 2 •− formation ( Figure 8B) in the infarcted myocardium. In addition, circulating mononuclear cells isolated from TF ΔCT mice after MI had significantly decreased expression of NOX2 and regulatory subunit p67 phox (Supplemental Figure 15D) in comparison with WT mice 7 days after permanent LAD ligation. TGF-β1 and ROS can act as a feed-forward mechanism for fibrosis (30), and NOX2 significantly contributes to oxidative stress and cardiac remodeling after MI (31). Importantly, NOX2 -/monocytes exposed to cytokine mix and hypoxia had significantly decreased ERK1/2 phosphorylation and TGF-β1 activation, indicating a central role for NADPH oxidase-derived ROS ( Figure 8C). In line with the reduced O 2 •− production, infarcted myocardium of TF ΔCT mice relative to WT showed markedly reduced ERK1/2 phosphorylation, decreased TGF-β1 activation, and reduced TGF-β1 signaling, based on phosphorylation of SMAD2 and profibrotic α-SMA induction ( Figure 8D and Supplemental Figure 15E). Importantly, the phosphorylation of the alternative TF signaling target p38 MAPK (29) (Supplemental Figure 15F) was not altered in TF ΔCT mice, underscoring the specificity of the ERK pathway in cardiac remodeling after MI.
Deletion of TF's cytoplasmic tail had no effect on infarct size and cardiomyocyte apoptosis 1 day after MI, based on Masson's trichrome and TUNEL staining (Supplemental Figure 15, G and H). Because early TGF-β1 signaling is required for late cardiac Figure 5. A profibrotic MEK1/2-TGF-β1 pathway is linked to PAR2-mediated ROS signaling in monocytes. (A and B) Protein expression analysis of monocytes isolated from WT mice and pretreated in vitro with trametinib (10 μM) for 1 hour (A), or isolated from PAR2 -/versus WT mice (B). Cells were stimulated with an inflammatory cytokine cocktail containing IL-6, TNF-α, and CCL2 at a concentration of 20 ng/mL with and without hypoxia for 4 hours. Western blotting of p-ERK1/2 (normalized to total ERK1/2) and activated TGF-β1 (normalized to GAPDH). Ordinary 1-way ANOVA, Šidák's multiple-comparison test; n = 5 replicates (2-3 mice were pooled for each sample). (C-E) PAR2 fl/fl and PAR2 fl/fl LysM Cre littermates were subjected to permanent LAD ligation and investigated after 7 days; n = 5-10 animals per group. (C) Western blot analysis of activated TGF-β1 (normalized to GAPDH) and p-SMAD2 (normalized to total SMAD2) in the infarcted myocardium. Representative blots and quantification of biological replicates. signaling function on myeloid cells results in beneficial cardiac remodeling by limiting excess fibrosis, improves cardiac function, and attenuates the development of IHF after MI.

Discussion
With the presented experiments, we expand our view of the pathophysiology of coagulation and inflammation in MI and IHF and uncover a crucial function of coagulation-related signaling in adverse cardiac remodeling. Treatment of arterial thrombosis and vascular occlusion is central to the therapy of acute coronary syndromes. Here we provide novel insights into a non-canonical mechanism of coagulation in the infarcted myocardium beyond thrombosis, as well as TF-PAR2 signaling leading to TGF-β1 activation, and unravel the precise role of myeloid cells in this context ( Figure 13).
Based on unbiased (phospho)proteomics in human IHF, MAPK1/ERK activation plays a crucial role in chronic ischemic heart disease. Inhibition of ERK1/2 activation in preclinical models of cardiac ischemia promotes angiogenesis (37) and reduces cardiomyocyte apoptosis (38,39). We now show that activation of ERK1/2 during MI specifically occurs in myeloid cells infiltrating the ischemic heart and that intervention with the MEK inhibitor trametinib significantly diminishes the CCR2-dependent (40) Ly6C hi monocyte recruitment into the infarcted myocardium. In addition, we establish a link between hyperactivation of the MAPK pathway and TGF-β1-mediated cardiac remodeling driven by inflammatory pro-oxidant myeloid cells infiltrating the infarcted heart.
Activation of TGF-β1 is required for its biological functions (41) and particularly for activating myofibroblasts in adverse cardiac remodeling after MI (42). We show with isolated cells that cytokine-primed monocytes exposed to hypoxia activate latent TGF-β1 dependent on NOX2 and MEK1/2 signaling. Furthermore, experiments with isolated monocytes from NOX2 -/and PAR2 -/mice in combination with in vivo analysis of myeloid cell-specific PAR2 and TF deletion confirmed that myeloid cell-derived TF-PAR2 signaling is the driver for TGF-β1 activation in MI. Disruption of the TF-PAR2 signaling complex on myeloid cells significantly reduces TGF-β1-mediated SMAD2 activation localized to myofibroblasts, which play a crucial role in post-MI remodeling (43,44), but not to immune cells (28). These results demonstrate that the non-canonical signaling properties of coagulation factors in innate immune cells modulate profibrotic mesenchymal cells for tissue homeostasis in post-MI cardiac remodeling (45).
Platelet-derived TGF-β1 activation depends on reduced protein disulfide isomerase (46), which also plays a critical role in TF decryption (47)(48)(49) and signaling (24,50). Notably, TF decryption is favored by complement activation (23,24), which was a prominent feature of the cardiac tissue of IHF patients identified by our proteomic screen. Independent of clotting activation, the TF-FVIIa complex furthermore interacts with β 1 integrin (51) and regulates ROS production by endosomal NOX2 trafficking, which depends on the TF cytoplasmic tail (26). ROS can regulate fibroblast proliferation and collagen synthesis in MI (52), but major cellular sources of ROS and underlying mechanisms of how they mediate cardiac remodeling in IHF have been elusive. Our presented data show that myeloid cell TF-PAR2 linked through the TF cytoplasmic tail is required for increased phagocyte-type NADPH phosphorylation in circulating monocytes and plasma levels of active TGF-β1 were significantly increased in subacute MI compared with stable CAD patients (Figure 11, C-E). These data indicated that myeloid cell TF phosphorylation could serve as a marker to identify patients at increased risk of developing IHF and adverse remodeling following MI with ongoing cardiac ischemia.
Pharmacological targeting of TF-FVIIa improves cardiac function by preventing TGF-β1 activation. We next targeted TF signaling pharmacologically after MI. Nematode anticoagulant protein C2 (NAPc2) blocks TF signaling and coagulation by forming an inhibited TF-FVIIa-FX(a) complex (34). NAPc2 has been tested in PCI without increasing bleeding risk in combination with antiplatelet and standard heparin therapy (35) and was recently evaluated in a clinical trial in COVID-19 (36) without major safety concerns. Recapitulating the findings in mouse monocytes, isolated human monocytes exposed to inflammatory cytokines and hypoxia had increased NOX2 expression and TGF-β1 activation, which were suppressed by NAPc2 along with ERK1/2 phosphorylation ( Figure 12A). In our preclinical model 7 days after MI, circulating myeloid cells had increased levels of NOX2 and active TGF-β1 that were attenuated by NAPc2 treatment starting 1 day after acute MI (Figure 12, B and C). In addition, NAPc2 improved cardiac function ( Figure 12D), attenuated cardiac infiltration of CD11b + myeloid cells (Supplemental Figure 18A), and diminished ERK1/2 phosphorylation, NOX2 expression, TGF-β1 activation, and the downstream target α-SMA in the infarcted heart ( Figure 12E and Supplemental Figure 18B). Next, we investigated beneficial effects of NAPc2 in the adverse cardiac remodeling and the transition from non-reperfused MI to IHF. Compared with vehicle-treated mice, mice with short-term NAPc2 treatment from day 1 to day 7 post-MI showed significantly reduced cardiac fibrosis ( Figure 12F) and improved cardiac function at day 28 after MI ( Figure 12G). This protection from cardiac damage resulted in improved survival ( Figure 12H). Collectively, our results show that targeting of TF (35) but, excitingly, disrupts signaling in TF-dependent coagulation, resulting in diminished fibrosis after experimental MI. The identified phosphorylation of the cytoplasmic tail of TF may serve as a biomarker applicable to liquid biopsies of patients with MI. It may facilitate clinical development of strategies to specifically target coagulation and TF-PAR2 signaling for myeloid cell reprogramming. This concept of interrupting coagulation-inflammatory signaling after MI has the potential to prevent TGF-β1 activation, to attenuate excess cardiac fibrotic damage, and to avert the development of IHF.

Methods
Human heart samples. Ischemic heart samples were obtained from the left ventricular (LV) wall of explanted hearts after cardiac transplantation or from heart tissue obtained during implantation of an LV assist device. Donor hearts were used as controls and acquired from ano-oxidase-derived ROS production and TGF-β1 activation to promote cardiac remodeling and propagate IHF.
In the clinical setting, patients with coronary no-reflow and/ or delayed presentation after onset of symptoms -so called subacute MI -show signs of thromboinflammation and are characterized by worse clinical outcome (33,53,54). However, biomarkers other than Q waves or T wave inversion in electrocardiography (55) are currently not established to predict poor outcomes of patients with MI that is not timely reperfused. Our data indicate that TF phosphorylation of circulating monocytes may serve as a marker for patients at increased risk of developing IHF and adverse remodeling following MI. In addition, our proof-of-principle experiments with pharmacological inhibitors delineate potential highly specific avenues to target this pathway. The putative clinical benefit is exemplified by NAPc2. This drug not only has dual antithrombotic action in the absence of excess bleeding LysM Cre mice at day 7 after MI. Ordinary 1-way ANOVA, Šidák's multiple-comparison test; n = 4-7 animals per group. Data are shown as mean ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.
(PLGS) (version 3.02 build 5, Waters Corp.). The human reference proteomes (entries: 20,365; UniProtKB/Swiss-Prot) were used for peptide identification. The false discovery rate (FDR) for peptide and protein identification was assessed searching a reversed decoy database and set to a 1% threshold for database search in PLGS. Label-free quantification analysis was performed using ISOQuant as described previously (59). For each protein, absolute sample amounts were estimated using TOP3 quantification (60).
Phosphopeptide analysis. Protein was digested after reduction overnight with trypsin (TPCK-treated, Sigma-Aldrich), phosphopeptides were enriched on preloaded TiO 2 spin tips, and bound phosphopeptides were eluted, desalted using Pierce graphite spin columns (Thermo Fisher Scientific), lyophilized, and reconstituted for LC-MS analysis on an Ultimate 3000 nanoUPLC (Thermo Fisher Scientific) and Nanospray Flex electrospray ionization source (Thermo Fisher Scientific). All samples were measured in triplicates. Mass-to-charge analysis of the eluting peptides was performed using an Orbitrap Exploris 480 (Thermo Fisher Scientific) in data-dependent acquisition mode. The MS proteomics data were deposited to ProteomeXchange via the PRIDE partner repository with the data set identifier PXD024727. Raw data processing for discovery phosphoproteomics analysis was performed in Proteome Discoverer v2.4 (Thermo Fisher Scientific) using Sequest HT Search Engine in the processing workflow. UniProtKB/ Swiss-Prot entries of the human reference proteomes (entries: 20,365) were used as database for peptide and protein identification with maximum allowed missed cleavages of 2, and maximum precursor and fragment ion mass tolerance of 10 ppm and 0.02 Da, respectively. The phosphosite localization probabilities were determined using IMP-ptmRS (https://ms.imp.ac.at/index.php?action=ptmrs) with Phos-phoRS mode enabled. Features were detected with Minora Feature Detector. In the consensus workflow, label-free quantification (LFQ) was performed by Feature Mapping and subsequent Precursor Ions Quantifier using the top 3 intense unique and razor peptides. Raw data processing for Integrative Inferred Kinase Activity (InKA) analysis was performed in MaxQuant v1.6.14.0 (61) using the same search engine settings as in Proteome Discoverer including LFQ with Match Between Runs enabled. The results were then submitted for InKA analysis (61).
The resulting quantitative information for proteins and phosphosites was analyzed using R including the packages tidyverse, pheatmap, and imputeLCMD. Fold changes (FCs) of IHF group versus control group were calculated using median intensity values. Two-sided t test assuming equal variances was performed for all samples that could be confidently identified with valid intensity values in at least 60%. All samples that were confidently appearing or disappearing in IHF condition (i.e., at least 60% identified intensities in one of the conditions and only missing values in the other condition) were assigned an FC of 100 and 0.01 as well as a P value of 0.001. The resulting P values were corrected for multiple testing using the Benjamini-Hochberg (FDR) method. The resulting proteins and phosphosites were then filtered for FCs greater than 2 and less than 0.5 and an adjusted P value less than 0.05. After sorting for descending FC, the protein and phosphopeptide abundances were log 10 -transformed, normalized by z score, and plotted as heatmap. Protein-protein interaction networks were generated using Cytoscape v3.7.2 including the STRING app, Networ-kAnalyzer plug-ins, and ClueGO app (62)(63)(64)(65). Changes in InKA scores were identified by calculation of FC and performance of t test applying the same settings as for proteome and phosphoproteome data. The nymized donors whose hearts could not be used for transplantation. Failing heart donors gave written informed consent for tissue donation. Storage and use of human heart tissue specimen by the Erich and Hanna Klessmann-Institute received approval of the ethics committee of the Ruhr-University Bochum, located in Bad Oeynhausen, Germany. Acquired heart samples were divided into at least 3 pieces, snap-frozen in liquid nitrogen, and stored at -80°C by HDZ-NRW.
Proteomic profiling of human hearts. Transmural heart tissue samples from 5 control hearts and 5 patients with IHF were lysed in 7 M urea, 2 M thiourea, 1% Phosphatase Inhibitor Cocktail 3 (Sigma-Aldrich), 100 mM NH 4 HCO 3 by sonication at 4°C and centrifuged, and the protein concentration was determined with the Pierce protein assay (Thermo Fisher Scientific). For whole-proteome analysis, 20 μg was used for filter-aided sample preparation and 700 μg protein for in-solution digest followed by phosphopeptide enrichment. The proteins were processed following previously published protocol (56,57). Purified peptides were reconstituted in 0.1% (vol/vol) formic acid for liquid chromatography-mass spectrometry (LC-MS) analysis on a nanoAcquity LC (Waters Corp.) (58). Eluting peptides were analyzed in positive-mode electrospray ionization MS by ion-mobility separation-enhanced data-independent acquisition (DIA) ultra-definition data-independent acquisition MS using elevated collision energy (UDMS E ) on a Synapt G2-S HDMS mass spectrometer (Waters Corp.) as described previously (59). Acquired MS data were postacquisition lock mass corrected using [Glu1]-fibrinopeptide B. LC-MS DIA raw data processing was performed with ProteinLynx Global SERVER  (2-HE), the oxidation product of DHE, and quantification normalized to weight of the infarcted tissue. Ordinary 1-way ANOVA, Šidák's multiple-comparison test; n = 5-6 animals per group. (C) Protein expression analysis of monocytes isolated from NOX2 -/animals and stimulated with an inflammatory cytokine cocktail containing IL-6, TNF-α, and CCL2 at a concentration of 20 ng/mL with and without hypoxia for 4 hours. Western blotting of p-ERK1/2 (normalized to ERK1/2) and activated TGF-β1 (normalized to GAPDH). Ordinary 1-way ANOVA, Šidák's multiple-comparison test; n = 4 replicates per group (2-3 mice were pooled for each sample). (D) Western blot analysis of activated TGF-β1 (normalized to GAPDH) and p-SMAD2 (normalized to total SMAD2) in infarcted myocardium obtained from WT or TF ΔCT mice after 7 days of MI. Representative blots and quantification of biological replicates. Ordinary 1-way ANOVA, Šidák's multiple-comparison test; n = 5-7 animals per group. (E) Sirius red staining and deconvoluted images of fibrotic area on paraffin-embedded heart sections 4 weeks after permanent LAD ligation to induce IHF versus sham surgery. Representative images and quantification of fibrotic areas normalized to surface area. Ordinary 1-way ANOVA, Šidák's multiple-comparison test; n = 5-7 animals per group. (F) Longitudinal echocardiographic studies over 4 weeks for LVEF (%) and LVEDV (μL) in parasternal long axis M-mode. Two-way ANOVA, Bonferroni's multiple-comparisons test; n = 6-17 animals per group. (G) Kaplan-Meier survival analysis of permanently LAD-ligated versus sham-operated C57BL/6J and TF ΔCT mice over 4 weeks. Log-rank (Mantel-Cox) test; n = 15-20 animals per group. Data are shown as mean ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001. tuted in NaCl physiological saline. Dosing was performed once a day starting from 1 day after MI until day 7.
Mouse model of non-reperfused MI. MI was induced by permanent ligation of the proximal LAD as described previously (68). Mice were anesthetized with medetomidine (500 μg/kg body weight), fentanyl (50 μg/kg body weight), and midazolam (5 mg/kg body weight). To antagonize the anesthesia, we injected atipamezole (2.5 mg/kg body weight) and flumazenil (0.5 μg/kg body weight). Sham surgery followed the same procedure except for ligation of the LAD. Mice received buprenorphine (0.075 mg/kg s.c.) twice daily for 2 days, starting on the day of surgery. Heart failure was defined in this study by a reduction of the LV ejection fraction (LVEF) below 35% and/or visual infarction of the LV. Animals that did not fulfill these criteria were excluded from the study.
Bone marrow transplantation. TF ΔCT , C57Ly5.1, and C57BL/6J mice aged 8-11 weeks were irradiated with a lethal dose of 9 Gy. Briefly, donor bone marrow (BM) was harvested in 2% PBS/FCS, filtered through a 70 μm cell strainer. Collected BM cells from the donor mice were washed in fresh 2% PBS/FCS and then resuspended at a log 2 of the FC and -log 10 of the adjusted P value were calculated and displayed in a volcano plot.
Animals and in vivo treatments. Nine-to 12-week-old male mice lacking 21 amino acids of the cytoplasmic tail of TF (TF ΔCT mice) on a C57BL/6J background and PAR2 -/mice on a C57BL/6N background were used along with strain-matched controls (26). PAR2 fl/fl mice were crossed to LysM Cre mice to generate conditional knockout of PAR2 on the myeloid cell compartment (PAR2 fl/fl LysM Cre+ mice) (15, 66); as controls, Cre-negative PAR2 fl/fl littermates were used. Generation of TF fl/fl LysM Cre+/mice has been previously described (67). All animals were bred and housed in the Translational Animal Research Center of the Johannes Gutenberg University, Mainz, Germany. For pharmacological interventions, C57BL/6J animals were used. LAD-ligated animals were further randomly divided into vehicle-or drug-treated groups. Trametenib (GSK1120212) was purchased from Selleck Chemicals and reconstituted in 200 μL vehicle (hydroxypropyl methylcellulose buffer) and orally administered to the mice (1 mg/kg/d) once a day starting from day 1 after MI throughout day 7. For NAPc2 treatment, mice were treated with i.p. injections (1 mg/kg/d) of NAPc2 reconsti- transplanted BM were subjected to permanent LAD ligation versus sham surgery and investigated 7 days later. Western blot analysis of NOX2 (normalized to GAPDH), p-ERK1/2 (normalized to total ERK1/2), and TGF-β1 (normalized to GAPDH) in infarcted myocardium obtained from chimeric mice. Ordinary 1-way ANOVA, Šidák's multiple-comparison test; n = 5-7 animals per group. Data are shown as mean ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001. and LVEF, posterior wall thickness in diastole, and interventricular septum thickness in diastole analyzed in parasternal long axis by means of M-mode linked to 2D B-mode images. Post-acquisition analysis was performed with Vevo LAB software (FUJIFILM VisualSonics Inc).
Clinical study. Twelve patients enrolled in the MICAT study (Mainz Intracoronary Database, ClinicalTrials.gov NCT02180178) were examined. Subacute MI was defined as follows: elevated circulating cardiac troponin I compatible with myocardial injury according to the Fourth Universal Definition of MI (69); symptoms of acute coronary syndrome >24 hours to <30 days prior to percutaneous coronary intervention; signs of subacute MI in electrocardiogram. Written informed consent was obtained from every participant. Medical history was documented, and body weight, height, and heart rate were obtained. Thirty milliliters of venous blood drawn from the cubital vein of the right arm was used for monocyte isolation. mix. For blocking of TF-FVIIa signaling on human monocytes, isolated monocytes from healthy individuals were treated with NAPc2 in the presence of OGD along with the cytokine cocktail mix followed by the protein expression analysis of p-ERK1/2, NOX2, and TGF-β1.
Western blotting. PBMCs, monocytes, or cardiac tissue were homogenized in lysis buffer (1% Triton X-100, 20 mM Tris pH 7.4-7.6, 150 mM NaCl, 50 mM NaF, 1 mM EDTA, 1 mM EGTA, 1 mM glycerol phosphate, 1% SDS, 100 mM PMSF, and 0.1% protease/phosphatase inhibitor cocktail) for 20 minutes on ice. Lysates were cleared by centrifuging at 11,000g for 15 minutes at 4°C. Total protein concentration was estimated using Lowry Assay (DC Protein Assay, Bio-Rad), and an equal protein amount in all samples was mixed in 6× Laemmli sample buffer, heated to 99°C for 10 minutes, separated according to their the manuscript. PW conceptualized, coordinated, and supervised the study, interpreted experiments, and wrote the manuscript.