Phagocyte-Myocyte Interactions and Consequences during Hypoxic Wound Healing

Abstract

Myocardial infarction (MI), secondary to atherosclerotic plaque rupture and occlusive thrombi, triggers acute margination of inflammatory neutrophils and monocyte phagocyte subsets to the damaged heart, the latter of which may give rise briefly to differentiated macrophage-like or dendritic-like cells. Within the injured myocardium, a primary function of these phagocytic cells is to remove damaged extracellular matrix, necrotic and apoptotic cardiac cells, as well as immune cells that turn over. Recognition of dying cellular targets by phagocytes triggers intracellular signaling, particularly in macrophages, wherein cytokines and lipid mediators are generated to promote inflammation resolution, fibrotic scarring, angiogenesis, and compensatory organ remodeling. These actions cooperate in an effort to preserve myocardial contractility and prevent heart failure. Immune cell function is modulated by local tissue factors that include secreted protease activity, oxidative stress during clinical reperfusion, and hypoxia. Importantly, experimental evidence suggests that monocyte function and phagocytosis efficiency is compromised in the setting of MI risk factors, including hyperlipidemia and ageing, however underlying mechanisms remain unclear. Herein we review seminal phagocyte and cardiac molecular factors that lead to, and culminate in, the recognition and removal of dying injured myocardium, the effects of hypoxia, and their relationship to cardiac infarct size and heart healing.

Physiological relevance of the cardiac innate immune response post myocardial infarction (MI) and the importance of phagocytosis to heart repair

Heart failure after MI is a leading cause of morbidity and mortality in the industrialized world (Lloyd-Jones, Adams et al. 2010). MI often occurs secondary to atherosclerotic plaque destabilization, the precursor to atherothrombosis (Tabas 2005). Infarction triggers inflammatory cell recruitment, which is a critical component of healing after tissue injury (Frangogiannis, Smith et al. 2002). A diverse population of bone marrow and spleen-derived immune cells are recruited to the heart after ischemia (Swirski, Nahrendorf et al. 2009) and function to promote clearance of damaged cardiac myocytes and repair of damaged myocardium (Thorp 2012). In contrast to inflammation during atherosclerosis (Libby 2012), immune cell mobilization after a heart attack is relatively acute and resolving. Although the initial inflammatory response may last just a few weeks, the activation state of recruited myocardial immune cells, molded by MI risk factors, may in turn modify cardiac infarct size and subsequently, the extent of cardiac remodeling and heart function (Panizzi, Swirski et al. 2010).

Myocardial phagocytosis

A central function of recruited leukocytes to sites of sterile injury is the degradation and phagocytosis of degraded extracellular matrix and dying and necrotic cells. This in turn promotes fibrogenic, and potentially angiogenic, responses that contribute to filling the void of lost and non-regenerative cardiac myocytes. Recent data collectively and directly link efferocytosis by inflammatory immune cells (Vandivier, Henson et al. 2006), i.e., the phagocytosis of apoptotic cells, to wound healing in the myocardium and in turn implicate phagocytosis receptors on monocytes and macrophages as a significant link between acute inflammation resolution and organ function (Wan, Yeap et al. 2013). Importantly and in the elderly, sub-optimal dying-cell clearance may lead to maladaptive cardiac remodeling and tissue repair, thereby accelerating the transition to heart failure (Chen and Frangogiannis 2010). Below, we focus on basic cellular and molecular mechanisms of inflammation and its resolution in the myocardium post infarction, with a focus on phagocyte-mediated interactions with dying tissue. These concepts form a testable working model (Figure 1. Working Model of Phagocyte and Cardiac Myocyte Interactions) that predict relationships between phagocyte-mediated dying-cell recognition, efferocytosis, infarct size, tissue-reparative signaling, and myocardial remodeling in the hypoxic heart.

Figure 1. Working Model of How the Cell Biology of Inflammation post Myocardial Infarction regulates Heart Failure.

Advanced atherosclerosis promotes atherothrombotic myocardial infarction (MI), the latter of which is characterized by recruitment of neutrophils and monocyte subsets that can differentiate into macrophages or dendritic-like cells. Phagocytes promote clearance of dying cardiac and immune cells. However, inherent inefficiency or MI associated risk factors promote inefficient dying-cell clearance, leading to secondary necrosis and further loss of non-regenerative cardiomyocytes. These acute events can affect later cardiac remodeling and inflammation that may lead to heart failure.

Immune cell mobilization to the heart

Although the extent of resident phagocytes in the human heart is unclear, the healthy murine heart harbors a population of resident macrophages (Pinto, Paolicelli et al. 2012). At the time of this Review’s publication, Epelman et al., provided evidence for both embryonic and adult-derived resident cardiac macrophages (Epelman, Lavine et al. 2014), that auto-renew similar to other tissues (Hashimoto, Chow et al. 2013), independent of recruitment from circulating monocyte pools. Cardiac injury elicits additional myeloid cells, largely a consequence of CC, CXC, and CX3C chemotactic factor release, many of which are found elevated in the bloodstream and lymph post MI (Frangogiannis, Smith et al. 2002). Chemokine release promotes influx of neutrophils, monocytes, and to a lesser extent, lymphocytes, into the heart, where they become tissue macrophages. Many chemotactic factors are recognized similarly by both neutrophils and monocytes, however, selective trafficking can be partially explained by differential expression or post-translational sialylation of G-protein-linked transmembrane chemokine receptors (Frommhold, Ludwig et al. 2008). Upon cardiac injury, release of damage associated molecular patterns from dying cardiomyocytes triggers chemokine synthesis by endothelial and resident cardiac cells. This process is later reinforced by newly recruited immune cells that additionally secrete chemokines as a consequence of pattern recognition receptor signaling. Evidence suggests that CXCL8 (IL-8) is induced by both endothelial cells and fibroblasts and in the myocardium, within the first hour of clinical reperfusion. Endothelial CXCL8 activates neutrophils through recognition of neutrophil CXCR1 and CXCR2 (Kukielka, Youker et al. 1995). CXCR2 recognizes CXCL2 (macrophage inflammatory protein 2-α, or MIP2-α) as well, and CXCL2 production by macrophages is also chemotactic for neutrophils (Belperio, Keane et al. 2002). Another important chemokine, CCL2, or monocyte chemo-attractant protein (MCP)-1, promotes recruitment of inflammatory monocytes through the CCR2 receptor. CCR2 null mice exhibit diminished acute cardiac macrophage accumulation and attenuated ventricular dilation after permanent coronary occlusion (Kaikita, Hayasaki et al. 2004). CCR2 also has affinity for CCL5 (RANTES) and CCL5 antagonists reduce myocardial reperfusion injury in hyper-lipidemic mice, yet this is not the case in Ccr5−/− mice, potentially due to enhanced expression of compensatory chemokines (Braunersreuther, Pellieux et al. 2010) or effects on myeloid progenitors in the bone marrow and circulation (Ergen, Boles et al. 2012).

Besides myeloid effects, a member of the CXC subfamily, Stromal cell-Derived Factor (SDF)-1 (CXCL12), both suppresses apoptosis in cardiomyocytes and promotes chemotaxis of endothelial progenitors to assist in angiogenesis (Imtiyaz, Williams et al. 2010). Furthermore, non-hematopoietic-derived cells are also targeted. For example, the CXC chemokine Interferon-gamma-inducible Protein (IP)-10 (CXCL10) is markedly induced in myocardial infarcts and exerts anti-fibrotic actions, inhibiting fibroblast migration to promote wound contraction and attenuate adverse remodeling (Bujak, Dobaczewski et al. 2009).

In addition to prototypical chemokines, angiotesin II (ANG II) is key for splenic monocyte mobilization post MI (Swirski, Nahrendorf et al. 2009). Utilizing congenic markers to track origins of cardiac monocytes from transplanted spleens, Swirski et al. calculated that the spleen contributes ~40% of monocytes to the ischemic myocardium (Swirski, Nahrendorf et al. 2009). Deficiency of the angiotensin receptor Atgr1α, as well as angiotensin-converting enzyme (ACE) inhibitor therapy, led to failure of splenic monocyte release after MI and reduced monocytic cells within the ischemic myocardium (Leuschner, Panizzi et al. 2010). In vitro, angiotensin II receptor engagement in monocytes promoted cytoskeletal rearrangement and monocyte migration. In vivo, these related pathways were augmented after activation of the sympathetic nervous system. That is, heightened sympathetic nervous system activity after cardiac injury promoted hematopoietic stem and progenitor cell (HSPC) release from bone marrow, as well as amplified extramedullary monocytopoiesis after experimental MI (Dutta, Courties et al. 2012). Enhanced monocyte mobilization required β3-adrenoceptor signaling on monocytes as infarcted mice treated with a β3- antagonist exhibited decreased HSPCs in the blood, as well as increased stem cell retention factors, compared to non-treated MI mice (Dutta, Courties et al. 2012). Importantly, recruitment of immune cells may continue after the acute inflammatory stage in the heart as chronic heart failure patients exhibit significantly elevated levels of CXC cytokines (Dahl, Husberg et al. 2009).

Cardiac “find me” vs “keep out” signals

After margination to post capillary venules, immune cells transmigrate past endothelial cells and chemotax towards the site of infarction. Directed migration to the ischemic core of an MI requires trafficking through a gradient of reducing oxygen tension. Apoptotic cells, primarily located in the zone bordering the infarct (Whelan, Kaplinskiy et al. 2010), likely secrete local so-called find-me signals, which is the first essential step for phagocyte recruitment in tissue. Find-me molecules are soluble chemo-attractants released by dying cells to establish a chemotactic gradient to attract phagocytes (Ravichandran 2011). Many of these signaling pathways act on RhoGTPases, which regulate cytoskeleton rearrangement to promote cellular migration (Singer, Tian et al. 2005). Known find-me signals include lipids, such as lyso-phosphatidyl-choline (LPC) and sphingosine-1-phosphate (S1P). LPC, one of the better-characterized find-me signals, is externalized and excreted during apoptosis (Lauber, Bohn et al. 2003). Secreted LPC interacts with G-protein-coupled receptor G2A, stimulating macrophage chemotaxis towards apoptotic cells (Peter, Waibel et al. 2008). LPC accumulates during ischemia in myocardium via thrombin activation of Ca2⁺-independent phospholipases (Daleau 1999), consistent with its role as a find-me signal in the damaged heart. S1P, another lipid find-me signal is produced by sphingosine kinase 1 (SPHK1) for recognition by S1P receptors on distal cells. Apoptotic stress induces SPHK1 activation, which can then promote S1P secretion (Matloubian, Lo et al. 2004). In addition to lipid find-me signals, proteinaceous tissue recruitment factors include cytokines and chemokines, including fractalkine (CX3CL1), which is cleaved by caspase-3 during apoptosis. In turn, the released fractalkine extracellular domain interacts with CX3CR1 on macrophages for cell recruitment (Truman, Ford et al. 2008). Nucleotides, including ATP and UTP, originate from both apoptotic and necrotic cells, and also likely act as find-me signals in the myocardium. In apoptotic cells, the plasma membrane channel pannexin 1 (PANX1) may serve as a conduit for nucleotide release after cleavage by caspases 3 and 7 (Chekeni, Elliott et al. 2010). During ischemia, cellular stress increases glycosylation of PANX1, resulting in enhanced ATP release from myocytes to promote fibroblast transformation (Dolmatova, Spagnol et al. 2012). Also, ATP can guide neutrophil chemotaxis via purinergic P2Y2 and A3 adenosine receptors in vitro and in vivo (Ayata, Ganal et al. 2012). Knockdown of P2y2 inhibits migration (Chen, Corriden et al. 2006), all consistent with the possibility that ATP released from PANX1 may act as a find-me signal in the heart.

Keep-out/Keep-away signals

Local find me signals are balanced by local keep-out/keep-away signals. In some instances, apoptotic cells selectively recruit monocytes as opposed to neutrophils. For example, monocytes in contrast to neutrophils are selectively recruited after injecting apoptotic cell supernatants into an air-pouch model of inflammation (Elliott, Harriman et al. 2009). Additionally, lactoferrin from apoptotic cell supernatants “kept out” neutrophils but not monocytes (Bournazou, Pound et al. 2009). Interestingly, the apo-form of lactoferrin can function as a mimetic of hypoxia by stabilizing the hypoxia inducible factor HIF-1α (Zakharova, Korneeva et al. 2012). Consistent with this, lactoferrin increased in patients during ischemia, just prior to reperfusion (Fiane, Videm et al. 2003). Growth differentiation factor-15 (GDF-15), a TGF-β-related cytokine, is also a keep-out signal. For example, GDF-15 is induced in the infarcted heart and Gdf15 deficient mice exhibit enhanced recruitment of neutrophils to the infarcted myocardium. GDF-15 activates the small Rho GTPase CDC42, inhibits activation of another small GTPase RAP1, and furthermore counteracts chemokine-triggered conformational activation and clustering of adhesive β2 integrins (Kempf, Zarbock et al. 2011). Though other inflammatory signals from the surrounding parenchymal pattern recognition receptor response also strongly influence inflammatory cell recruitment (Singh, Swaminathan et al. 2012), the ratio of find-me signals to keep-away may be important in regulating local responses of phagocytes in close proximity to dying cells.

Cardiac neutrophils

Release of bone marrow-retained circulating polymorphonuclear granulocytes (PMNs), or neutrophils, is one of the earliest and most robust coordinated steps of leukocyte mobilization to the heart. After experimental coronary artery ligation in rodents, PMNs are recruited during the first few hours following blockage of blood flow, with levels peaking as early as one day after injury and further heightened after clinically relevant reperfusion (Yan, Anzai et al. 2013). Much of our in vivo insight into PMN contributions to inflammation has been achieved after targeting PMN or endothelial adhesion molecules. PMNs bind endothelial adhesion molecules such as L- and P- selectins, integrins, and intercellular adhesion molecules (ICAMs), and monoclonal anti-L- and P-selectin antibodies reduce myocardial necrosis and enhance coronary endothelial function in association with reduced PMN margination (Ma, Weyrich et al. 1993). Conversely, impaired PMN trafficking associated with combined P-selectin and ICAM-1 deficiency exhibited no significant difference in infarct size after ischemia followed by reperfusion (Briaud, Ding et al. 2001). The selectins are involved in early PMN attachment and rolling, however firm adhesion and transmigration past the endothelial layer requires PMN integrin leukocyte function associated antigen-1 (LFA-1 or CD11a/CD18)-dependent activation of endothelial integrins. Inhibiting LFA-1 has been shown to reduce infarct size in primates (Aversano, Zhou et al. 1995), however decreased PMN recruitment in mice treated with vascular cell adhesion molecule (VCAM-1) antibodies was not associated with decreased myocardial injury (Bowden, Ding et al. 2002). However it is important to note that both LFA-1 and VCAM are expressed by other immune cells during MI (Meisel, Shapiro et al. 1998) and therefore PMN-specific effects of these molecules will require conditional blockade approaches. PMNs also express PECAM (platelet endothelial cell adhesion molecule), and to a lesser extent CD99, both adhesion molecules necessary for transmigration across the endothelium, independent of adhesion (Muller 1995). Combined antibody blocking of PEACAM and CD99 had an additive effect to reduce transmigration, suggesting these two molecules regulate distinct steps in PMN extravasation (Lou, Alcaide et al. 2007). How PECAM/CD99-dependent recruitment of PMNs to the heart and their influence to cardiac pathology is unclear.

“Bad” cardiac PMNs

In a prospective observational study, patients with chest pain in the highest tertile of blood PMN-count were at increased risk for non-fatal acute MI and death (Meissner, Irfan et al. 2011). Similarly, STEMI-classified (ST segment elevation MI) patients were less likely to survive during a 30 day follow-up if they had a higher baseline PMN count (O’Donoghue, Morrow et al. 2008). Patients with neutrophilia 4 days post-MI were more likely to develop congestive heart failure (Kyne, Hausdorff et al. 2000). These findings highlight PMNs as potential therapeutic targets for recovery following MI (Akpek, Kaya et al. 2012), although it is difficult to delineate whether heightened PMNs could also simply reflect more severe cardiac damage. Consistent with this, anti-PMN antibodies reduce ischemic injury in canine models of reperfusion following ischemia (Romson, Hook et al. 1983). Some of the untoward effects of PMNs are associated with their potent secretion of inflammatory mediators. Gelatinase degranulation in PMNs follows transendothelial migration, in turn releasing matrix metallo-proteinases MMP-8 and -9. Azurophilic granules release myeloperoxidase (MPO) as well as serine proteases (Soehnlein, Weber et al. 2009). MPO (Rudolph, Goldmann et al. 2011), lipocalin-2 (Yndestad, Landro et al. 2009) and MMPs, are associated with worsened cardiac outcome following MI. MPO generates cytotoxic fomaldehyde, acrolein, and chlorinating species in the infarct zone, which is adversely associated with LV remodeling and function in mice (Vasilyev, Williams et al. 2005). MMP-8 is responsible for the degradation of fibrillar collagen, promoting PMN migration (Lin, Jackson et al. 2008). MMP-9 is up-regulated within 24 hours post injury and secreted mostly by recruited PMNs and macrophages (Lindsey, Gannon et al. 2002; Tao, Cavasin et al. 2004). Patients suffering from ischemic, chronic heart failure were more likely to have higher serum levels of PMN-associated lipocalin than control patients (Yndestad, Landro et al. 2009). PMNs can also directly promote cardiac myocyte damage and potentially infarct size through the release of reactive oxygen species (ROS) through the NADPH oxidative burst (Ciz, Denev et al. 2012). ROS may be generated after PMNs interact with platelets through the binding of Triggering Receptor Expressed on Myeloid Cells (TREM-1). Finally, ROS production can incite further leukocyte extravasation through elevated P-selectin expression (Griendling and FitzGerald 2003) and complement activation (Shingu, Nonaka et al. 1992).

Beneficial PMN functions in heart

In humans, PMN depletion strategies have caused reduced infarct size and scar formation (Cavanagh, Gough et al. 1998), consistent with specific PMN functions contributing beneficially to cardiac wound healing. For example, PMN-derived and activated MMPs degrade pre-existing extracellular matrix, allowing extravasated leukocytes to migrate to the infarct tissue for tissue repair (Cleutjens, Blankesteijn et al. 1999). Also, PMNs are phagocytic and to a lesser extent efferocytic and contribute to removal of necrotic cardiac infarction debris. To a lesser extent, PMNs may also promote clearance of apoptotic cells, however, PMN efferocytic function is less studied than in macrophages. Efferocytosis by PMNs requires lipoprotein-receptor related protein (LRP) through recognition of calreticulin on target cells (Park, Liu et al. 2008). Apoptotic cell recognition tempers PMN oxidative burst, which is heightened after reperfusion, and reduces production of TNF-α and CXCL10 (Esmann, Idel et al. 2010). As PMNs themselves become apoptotic, they produce lactoferrin to “keep out” further PMN infiltration (Curran, Demick et al. 2006). Lactoferrin also decreases PMN activation by impairing degranulation and reducing β₂ integrin expression, the latter of which suppresses cellular motility (Bournazou, Pound et al. 2009). In fact, during homeostatic conditions, cyclic flux of PMN release from bone marrow followed by elimination of aged PMNs serves as a feedback mechanism to modulate the hematopoietic niche (Casanova-Acebes, Pitaval et al. 2013). Finally, PMNs chaperone recruitment of monocytes from the blood stream. PMN-derived cathelicidin accumulates on endothelial proteoglycans to bind monocyte formyl-peptide receptor 2 (Swirski and Robbins 2013). In turn, this increases binding of integrins MAC1 and VLA-4 to cell adhesion molecules (Wantha, Alard et al. 2013).

Blood Ly6c^HI monocytes and Ly6c^LO monocytes and cardiac tissue Ly6c^LO macrophages

The mononuclear phagocyte system (MPS) of monocytes, macrophages, and dendritic, collectively scavenge damaged matrix, microparticles, dead cells, and regulate inflammation. In the heart, monocyte residence time has been approximated to 20 hours. Sustained levels post MI are supplied by extramedullary splenic hematopoiesis, as well as from bone marrow sources (Leuschner, Rauch et al. 2012). There are two monocyte/macrophage subsets in the heart identified by expression of surface markers and characterized by inflammatory phenotype: Ly6C^HI monocytic (analogous to CD14+ CD16− in humans) and Ly6C^LO monocytic/macrophage cells (CD14+ CD16+ in humans) (Shantsila and Lip 2009; Shantsila, Wrigley et al. 2011; Tapp, Shantsila et al. 2011; Nahrendorf and Swirski 2013). In human MI patients, individuals with prolonged prevalence of proinflammatory CD14⁺/CD16⁻ cells have decreased myocardial salvage, a measure of the amount of healthy tissue in the infarct (Tsujioka, Imanishi et al. 2009). Ly6C^HI cells exhibit a pro-inflammatory phenotype and are found early in the infarct after occlusion, peaking in mice ~3 days post permanent occlusion of the murine left anterior descending artery. These cells also promote removal of necrotic debris (Nahrendorf, Swirski et al. 2007). Ly6c^HI monocytes express high levels of CCR2, respond to MCP-1, and produce TNFα and proteolytic enzymes. Ly6c^HI monocytes typically are not associated with high efficiency of apoptotic cell clearance, however, recent studies suggest that cross-talk with macrophages may be partially responsible. For example, 12/15 lipooxygenase (LO), expressed by alternatively activated macrophages (described below), generate phospholipid oxidation motifs on the macrophage that sequester soluble molecules that bridge apoptotic cell receptors and apoptotic targets. This in turn reduces monocyte efferocytosis efficiency (Uderhardt, Herrmann et al. 2012). Ly6c^LO, CX3CR1^HI, CCR2^LO monocyte/macrophages emerge in the myocardium soon after Ly6c^HI subsets and are critical for myocardial repair, where they secrete pro-fibrotic and angiogenic cytokines. Ly6c^LO pro-reparative functions are tied to the Ly6c^HI response, as clodronate-mediated depletion during the predominantly Ly6c^HI monocyte phase delays healing. Ly6C^LO cells “patrol” endothelial capillaries during homeostasis, and therefore are positioned to respond immediately after MI (Auffray, Fogg et al. 2007). Alternatively, Ly6c^LO cells need not extravasate into tissue to promote wound healing, as Carlin et al. (Carlin, Stamatiades et al. 2013) reported that Ly6c^LO cells can be interestingly retained by endothelial cells in the kidney vasculature to recruit PMNs. In this example, recruited PMNs promoted lysis of compromised endothelium, which was subsequently cleared up by monocytes. In the heart, and in terms of absolute numbers, the vast majority of Ly6C^LO cells accumulate in the cardiac wound after Ly6c^HI monocytes peak, eventually outnumbering Ly6C^HI cells in the later stages of the cardiac inflammatory response (Nahrendorf, Swirski et al. 2007). Interestingly, the apoptotic cell receptor MERTK is expressed predominantly by Ly6C^LO cells post MI, suggesting distinct clearance mechanisms utilized by monocyte subsets (Wan, Yeap et al. 2013).

Cardiac macrophages and dendritic-like cells

Monocytes differentiate into macrophage-like cells and proliferate in the presence of Macrophage-Colony Stimulating Factor (M-CSF), which is elevated in canine infarcts (Frangogiannis, Mendoza et al. 2003). Monocytosis is associated with higher numbers of mature macrophages in the infarct on day 5 (Panizzi, Swirski et al. 2010). Within the myocardium, the early macrophage phenotype is similar to pro-inflammatory/activated M1-like macrophages (F4/80+, CD86+), which is followed by a phenotypically similar anti-inflammatory M2-like macrophage profile (F4/80+, CD206+). Furthermore, increasing the M2/M1 ratio after mesenchymal stem cell therapy was associated with improved regional function at the mid-anterior infarct zone, an effect that was abrogated upon clodronate depletion of phagocytic cells (Ben-Mordechai, Holbova et al. 2013). Dendritic-like cells peak at day 7 in experimental models of MI. Following dendritic cell ablation, mice exhibited enhanced inflammation and extracellular matrix degradation in the infarcted myocardium, leading to wall thinning, impaired neo-angiogenesis, and increased infiltration of Ly6C^HI monocytes. This suggested that at least immature CD11c+ dendritic-like cells may play a protective role in post-MI repair (Anzai, Anzai et al. 2012). Other cell lineages classically associated with the adaptive immune arm have been discovered in injured myocardium, including IL-10 and TGF-β producing regulatory T cells, and B-cells (Zouggari, Ait-Oufella et al. 2013), both of which may interact with phagocytes to control inflammation and potentially other aspects of chronic heart healing.

Macrophage-mediated efferocytosis and inflammation resolution in the heart

Whereas efficient efferocytosis activates pro-resolving/anti-inflammatory pathways in the phagocyte (Serhan and Savill 2005; Birge and Ucker 2008), defective efferocytosis leads to secondary post-apoptotic necrosis and expansion of tissue necrosis (Vandivier, Henson et al. 2006). Previous studies have linked defective apoptotic cell clearance to diseases of chronic non-resolving inflammation such as atherosclerosis and lupus (Tabas and Glass 2013). In contrast, the extent to which efferocytosis efficiency during acute resolving inflammation may affect long-lasting organ function is much less clear. In particular, inefficient removal of dead cardiac tissue in aged hearts has been linked to progression of heart failure (Bujak, Kweon et al. 2008). Clearance of apoptotic PMNs by macrophages initiates the resolution phase of inflammation, inducing the production of IL-10, TGF-β, lipoxins, and resolvins. IL-10 appears late in the infarcted myocardium and contributes to the stabilization of the matrix by promoting macrophage production of tissue inhibitor of metallo-proteinases (Frangogiannis 2013). IL-10 knockout animals display an increased inflammatory response (Krishnamurthy, Rajasingh et al. 2009), including heightened TNF-α and MCP-1 mRNA and increased mortality rates during ischemia-reperfusion (Yang, Zingarelli et al. 2000). Lipoxins and resolvins, derived from poly-unsaturated fatty acids, are protective for cardiomyocyte reperfusion injury (Keyes, Ye et al. 2010) and promote efferocytosis of PMNs by macrophages (Schwab, Chiang et al. 2007) while reducing vascular permeability (Takano, Clish et al. 1998) and PMN infiltration (Serhan, Maddox et al. 1995). Enhanced clearance of PMNs may feed-back through an IL-23 pathway to affect granulopoiesis and PMN production (Stark, Huo et al. 2005). In a mouse model of acute kidney ischemia reperfusion injury, PMN-associated production of IL-17/IL-23 was shown to be required for further PMN infiltration and IFN-γ production (Li, Huang et al. 2010). Finally, If PMNs are cleared by macrophages, the question arises: What cells are responsible for clearing macrophages that turnover in the heart? Dying macrophages may emigrate to spleen or lymph or alternatively be removed by new resident cardiac macrophages or other resident cardiac cells.

Phagocyte directed cardiac repair

Efferocytosis induces TGF-β (Fadok, Bratton et al. 1998), which also plays an important role in tissue remodeling and post-MI inflammation resolution. TGF-β activates fibroblasts and induces collagen and fibronectin production (Bassols and Massague 1988). This cytokine also reduces adhesion molecule expression and promotes the differentiation of regulatory T cells (Bujak and Frangogiannis 2007). After activation by IFNγ, LPS, lactate, or hypoxia, macrophages may produce pro-angiogenic factors, including nitric oxide and VEGF (Vascular Endothelial Growth Factor), or. During hypoxia, VEGF is up-regulated through the action of hypoxia transcription factors and through increased mRNA stability (Xiong, Elson et al. 1998). A subset of pro-angiogenic macrophages have been described, termed myeloid angiogenic cells (MACS), which are similar in function to alternatively active/M2 phagocytes, but also express endothelial cell markers including TIE2 (Tunica Interna Endothelial Cell Kinase, or TEK tyrosine kinase) and VEGFR. These cells promote angiogenesis via paracrine signaling, producing MCP-1, MMP9 and IL-8, which act on endothelial cells to activate VEGFR (Medina, O’Neill et al. 2011). Interestingly, when these cells were administered intravenously in a rat model of MI, they were found to localize to ischemic areas, causing reduced scarring and improved ventricular function (Kawamoto, Gwon et al. 2001). Whether these cells are capable of differentiating into endothelial cells or maintain their myeloid phenotype once within tissue, however, remains unclear (Chambers, O’Neill et al. 2013). Of further interest, when CD14+ monocytes were delivered to ischemic sites of oxygen-induced retinopathy, the result was an enhanced pro-angiogenic M2 macrophage phenotype that improved vascularization and reduced retinopathy-associated inflammation (Marchetti, Yanes et al. 2011). Furthermore, in a hind-limb ischemia model, deletion of one allele of hypoxia transcription factor-suppressor PHD2 (prolyl hydroxylase domain protein 2), in turn skewed macrophage polarization towards a pro-arteriogenic phenotype, thereby preventing tissue necrosis and preserving limb perfusion (Takeda, Costa et al. 2011). Finally, Hochreiter-Hufford showed that through recognition of externalized phosphatidylserine on neighboring apoptotic cells, BAI1 (brain-specific angiogenesis inhibitor), a member of the adhesion type-G protein coupled receptor family, can signal through the EMLO-DOCK180-Rac1 pathway and enhance myoblast fusion during muscle development, regeneration and repair (Hochreiter-Hufford, Lee et al. 2013; Novak, Weinheimer-Haus et al. 2014).

Hypoxic effcts on phagocytes

A significant aspect of the post-MI or wound injury phagocyte response is an ischemic environment, with oxygen tensions is healing wounds reported as low as 0–3 kPa (Hunt, Twomey et al. 1967). In this context, it’s worth noting that most mechanistic studies of immune cell function are not modeled under reduced oxygen tension, although interestingly, over-confluent tissue culture conditions can reduce oxygen saturation in tissue culture media (Thompson, Binham et al. 2013). In PMNs, Hypoxia Inducible Factor 1α (HIF-1α) extends PMN lifespan during hypoxia (Hannah, Mecklenburgh et al. 1995) and this in part may be promoted by HIF-enhanced glycolysis (Walmsley, Chilvers et al. 2011). Prolonged PMN survival in the heart could extent the direct deleterious action of PMNs and further delay monocyte/macrophage-mediated anti-inflammatory signaling, secondary to efferocytosis of apoptotic PMNs. In monocytes, responses to hypoxia have only initially begun to be investigated: Knockdown of monocyte Hif-1α supressed CCR2 expression in Ly6c^HI monocytes, resulting in decreased monocyte entry into the infarct and increased cardiac ejection fraction (Dong, Khalil et al. 2010). Low oxygen importantly also modulates angiogenic and inflammatory gene expression in monocytes through increased expression of VEGF (Bosco, Puppo et al. 2008), as eluded to above. Myeloid expression of Hif-1α was required for adequate skeletal muscle regeneration. In mice with a myeloid-specific deletion of Hif-1α, there was delayed invasion of F4/80+ macrophage-like cells, suppression of myoblast proliferation in parallel with decreased cyclooxygenase-2 activity, impaired removal of necrotic cell debris, and reduced regeneration of endothelial cell structures following myoblast injury (Scheerer, Dehne et al. 2013). Interestingly, these effects were shown to be independent of phagocyte polarization phenotype between control and Hif-deficient mice. Much less has been studied regarding Ly6C^LO phagocytes and their adaptation to low oxygen levels. Future work in this area is needed to resolve the effects of hypoxia on monocyte subset levels, recruitment, clearance capacity and a possible inter-conversion between subsets during MI. Also, many questions remain regarding the involvement of each monocytic subset in post-ischemic neovascularization. While TIE2-expressing Ly6C^LO cells have been shown to promote tumor angiogenesis, in contrast, in mice with hind limb ischemic injury and adoptive transfer of bone marrow derived Ly6C^HI monocytes, there was significant improvement of blood flow recovery, an effect not seen after transfer of Ly6C^LO cells (Capoccia, Gregory et al. 2008). Similarly, CCL2/CCR2 signaling was required for adequate neo-vascularization following hind limb ischemia, as blood flow recovery was promoted by adoptive transfer of Ly6C^HI, but not Ly6C^LO phagocytes. A differential angiogenic capacity of each subset may be influenced by increased MMP-9 expression by Ly6C^HI cells, which promotes capillary branching and revascularization (Cochain, Rodero et al. 2010). HIFs are also required for myeloid cell aggregation, invasion, and motility (Kong, Scully et al. 2007), as well as during macrophage differentiation (Oda, Hirota et al. 2006). HIFs also mediate macrophage motility and energy homeostasis through the glycolytic maintenance of the cellular ATP pool, which was decreased in Hif-1α knockout cells (Cramer, Yamanishi et al. 2003). Additionally, macrophage migration in response to nitric oxide is dependent on HIF-1α signaling through small GTPase CDC42 and RAC-1 (Zhou, Dehne et al. 2009). Much of the investigation of HIF regulation in macrophages has focused on Tumor Associated Macrophages (TAMs), as hypoxia is a major component of the tumor microenvironment. In this context, Hif-1α knockout TAMs had impaired inflammatory TLR4 responses and decreased cytotoxicity to tumor-spheroid cells. Taken together these results indicate that HIF-1α is required for the maintenance of M1 polarization (Werno, Menrad et al. 2010). Duration of hypoxic exposure can also determine macrophage response. Death-resistant macrophages induced by long-term (48hrs) exposure to 1% O₂ exhibited decreased expression of Heat Shock Protein70, and increased TNFα (Degrossoli and Giorgio 2007). In macrophages, interestingly, HIF-1α does not always require low oxygen for activation. Case in point, normoxic (21% oxygen) supernatants from apoptotic cells have been shown to transcriptionally activate Hif-1α expression, in a NFAT (nuclear factor of activated T cells)-dependent manner. One of the key factors that promoted Hif-1α expression was S1P, or sphingosine 1 phosphate (Herr, Zhou et al. 2009), the find-me signal described above.

Phagocyte: Myocyote interactions: “Eat me” vs “don’t eat me” ligands

The molecular pathways responsible for phagocyte interactions with cardiomyocytes remain largely unknown. A key purpose of the aforementioned recruitment signals and recruited leukocytes is to promote interactions with dying or necrotic cardiac tissue. Phagocytes distinguish viable from non-viable cells through the aid of so-called self eat-me and don’t-eat-me signals, which are presented on the target cell surface and aided by binding of bridging molecules that interface between the target cell and phagocyte (Figure 2). Eat-me signals can be externalized phospholipids, proteins, alterations in cell-surface charge or glycosylation patterns, and nucleotides (Hochreiter-Hufford and Ravichandran 2013). Externalization of phosphatidyl-serine (PS) is one of the most conserved apoptotic markers, however, we are still just learning about the mechanism of PS externalization. Just this past year, Nagata and colleagues published that Xk-Related Protein 8 (XKR8) is required for PS externalization under apoptotic stimuli. Cells deficient for Xkr-8 failed to expose PS during apoptosis and were inefficiently engulfed by phagocytes. Interestingly, both cancer cells and terminally differentiated cardiomyocytes were found to express low levels of Xkr-8 (at the mRNA level), consistent with recent data showing reduced efferocytosis efficiency of cardiac myocytes by macrophages in vitro (data not published). In the context of the infarcted heart, eat-me signals in vivo may be affected by low oxygen levels. For example, acute hypoxia alters PS content in erythrocytes, potentially through modulation of phospholipid scramblases, aminophospholipid translocases and ATP-dependent floppases (Nie, Tian et al. 2011). In addition to PS, annexin 1, a calcium and phospholipid binding protein in the annexin superfamily, is an endogenous eat-me signal for macrophages (Swathi Arur et al.). Annexin 1 is cleaved by ADAM10 (A Disintegrin And Metalloproteinase) during cell necrosis, also contributing as a monocytic chemotactic signal (Blume, Soeroes et al. 2012). This pathway may be especially important after MI considering the extensive level of cellular necrosis. Another newly defined eat-me signal is the ficolin1-PTX3 heterocomplex, which can interact with late apoptotic or necrotic cells and enhance their clearance (Ma, Doni et al. 2013).

Figure 2. Phagocyte-Myocyte interactions.

To the left is a micrograph of the elongated cardiomyocyte juxtaposed next to multiple circular macrophages. To the right is a schematic exhibiting the unknown spectrum of interacting ligands between macrophages and cardiomyocytes, including putative eat-me signals on the apoptotic cardiomyocyte, such as phosphatidylserine and calreticulin, as well as potential molecules that bridge cardiomyocytes and macrophages, such as Gas6 (growth arrest specific) or MFG-E8 (milk fat globule). Don’t eat me signals may include CD47 through interactions with Sirp1α. MERTK is important on the phagocyte side for efferocytosis of cardiomyocyte apoptotic bodies. Other putative cardiac recognition ligands and macrophage receptors are yet to be described.

“Don’t-eat-me” signals in the heart

Don’t-eat-me signals, such as CD31 and PAI-I (Plasminogen activator inhibitor), can also help prevent viable cells from being engulfed by phagocytes. The most widely studied don’t-eat-me signal is CD47, which is a membrane protein expressed on the surface of most cells. CD47 interacts with SIRPα on phagocytes, recruits phosphatases, and inhibits downstream activation of the phagocyte actin cytoskeleton, thereby preventing engulfment (Tsai and Discher 2008). It has been shown that CD47 is expressed in abundance on apoptotic neonatal cardiocytes (Izmirly, Saxena et al. 2011). However, nothing more has been studied in the heart. By showing thrombospondin-2 (a CD47 ligand) knockout mice have higher mortality and dilated cardiomyopathy, it was concluded that TSP-2 protects age-related dilated cardiomyopathy (Swinnen, Vanhoutte et al. 2009). However, whether this phenomenon requires CD47, or how CD47 may be directly involved in removal of apoptotic cells in the heart, is unknown. Furthermore, the TSP1-CD47 axis is induced in renal tubular epithelial cells (RTEC) under hypoxia (Rogers, Yao et al. 2012). Thus, it will be interesting to see if such an axis is also induced between cardiomyocytes and macrophages within the hypoxic environment of the ischemic heart.

Therapeutic implications and future directions

Though pharmacological advances have significantly reduced mortality, the residual risk of post MI-induced heart failure remains high. This necessitates the development of complementary approaches to preserve heart function. It stands to reason that clearance of dying cells after MI by recruited phagocytes may be inherently inefficient as evolutionary pressure has not selected for optimal monocyte interactions with cardiac myocytes during diseases of aging. It is tempting to speculate that enhancing efferocytosis in the heart might help wound healing after heart attack. Efficient clearance of dying cells, both in a timely manner and of a significant quantity, is a pre-requisite for resolution of inflammation and downstream reparative processes after tissue clearance. In addition to clearance efficiency, the downstream signaling responses of phagocytes post engulfment are critical to inflammation resolution. Imbalances between pro- and anti-inflammatory stimuli may contribute to clearance inefficiency. The extent of cell death in the acute inflammatory phase of MI is a critical determinant of the degree of adverse remodeling leading to heart failure (Whelan, Kaplinskiy et al. 2010). Therefore, strategies that enhance efficient resolution of inflammation and prevent unnecessary further cell death of terminally differentiated heart myocytes (cardiomyocytes) may be useful in slowing the progression to heart failure and potential autoimmune reactions. Additional MI risk factors, such as hyperlipidemia may further reduce efferocytosis efficiency and repair in the heart.

Though many potential molecular reasons may explain inefficient effferocytosis of cardiomycoytes, one candidate worth noting is at the level of apoptotic cell receptors, which could be rendered naturally dysfunctional in the setting of disease or genetic risk factors. For example, polymorphisms in the apoptotic cell receptor Mertk are associated with increased autoimmune inflammation in diseases such as Lupus (Cheong, Lee et al. 2007). In addition, ADAM metallopeptidase 17, which cleaves MERTK into a soluble inhibitory receptor (Sather, Kenyon et al. 2007; Thorp, Vaisar et al. 2011), is increased during MI (Akatsu, Nakamura et al. 2003). This may explain the identification of solMER in murine extracts post MI (Wan, Yeap et al. 2013) and further provide the impetus for investigation of human MI specimens or blood. MERTK activity may also be limited by availability of its ligand Gas6, which is required for binding to apoptotic cells (Munoz, Sumoy et al. 2004). Future therapeutic approaches must strike a balance as, although the innate immune response has helpful activity in the healing heart, maladaptive inflammation can also be detrimental. Therefore, selective approaches that target specific immune subsets or cellular pathways harbor the most potential, as broad immunosuppressive therapy post MI can be detrimental in mice and humans (Hammerman, Kloner et al. 1983; Getts, Terry et al. 2014). Targeting of Ly6c^HI monocytes after reperfusion have shown promise in (Majmudar, Keliher et al. 2013). Also, stimulation of so called pro-resolving pathways downstream of efferocytosis and phosphatidylserine recognition have proven beneficial in animal models (Harel-Adar T et al.).

Highlights.

• The molecular mediators of phagocyte-myocyte interactions are just beginning to be studied

• We review critical phagocytes populations in the heart during injury

• We discuss critical myocyte interactions with phagocytes

• We discuss how hypoxia and hypoxia-inducible factors may affect phagocyte-myocyte interactions

• We provide a working model of phagocyte-myocyte biology and physiological significance