Clearance of apoptotic neutrophils and resolution of inflammation

Summary

The engulfment of apoptotic cells by phagocytes, a process referred to as efferocytosis, is essential for maintenance of normal tissue homeostasis and a prerequisite for the resolution of inflammation. Neutrophils are the predominant circulating white blood cell in humans, and contain an arsenal of toxic substances that kill and degrade microbes. Neutrophils are short lived and spontaneously die by apoptosis. This review will highlight how the engulfment of apoptotic neutrophils by human phagocytes occurs, how heterogeneity of phagocyte populations influences efferocytosis signaling, and downstream consequences of efferocytosis. The efferocytosis of apoptotic neutrophils by macrophages promotes anti-inflammatory signaling, prevents neutrophil lysis, and dampens immune responses. Given the immunomodulatory properties of efferocytosis, understanding pathways that regulate and enhance efferocytosis could be harnessed to combat infection and chronic inflammatory conditions.

Introduction

The ability of macrophages to remove injured host cells was described in the late nineteenth century by Elie Metchnikoff, the father of phagocyte biology. He described the homeostatic removal of dead cells and referred to the process as physiologic inflammation. deCathelineau and Henson later coined the term efferocytosis, meaning ‘to carry to the grave’, to refer to the phagocytosis of apoptotic cells (1). Professional and non-professional phagocytes are involved in the recognition and elimination of apoptotic cells from circulation. Neutrophils are the predominant circulating white blood cell in humans and play a critical role in host defense against bacterial and fungal infections. Neutrophils have an inherently short lifespan, and only live for approximately 24 hours (2). Older neutrophils undergo constitutive, spontaneous apoptosis and are cleared by efferocytosis. It is estimated that 10⁹ neutrophils/kg of body weight are turned over every day in the absence of inflammation (3).

The transfer of autologous radiolabeled neutrophils into patients with inflammation results in the trafficking of these cells from the blood to abscesses, lung, spleen, liver, and bone marrow. Retention of the radiolabelled cells in the spleen, liver, and bone marrow after the disappearance from the blood suggests these organs are the sites of neutrophil destruction (4, 5). In vitro, the appearance of apoptotic features in neutrophils, including nuclear condensation and DNA-laddering, coincides with their removal by macrophages, a subset of phagocytes (6). In some instances, removal by macrophages precedes the appearance of gross apoptotic features (7). These data indicate that efferocytosis of neutrophils occurs in vivo and involves recognition and engulfment by phagocytes.

The removal of apoptotic neutrophils from circulation regulates granulopoiesis (8), prevents secondary lysis and spillage of noxious neutrophil substances (9, 10). Moreover, apoptotic neutrophils trigger an anti-inflammatory and pro-resolving response in macrophages, monocytes, and dendritic cells (11-13). Thus, it can be surmised that efferocytosis of apoptotic neutrophils is a fundamental, innate process required for tissue homeostasis and immunity, and its dysregulation can lead to unwanted excessive inflammation, autoimmunity, and exacerbation of infection.

Human Phagocytes

The complexity and heterogeneity of human phagocytes complicates our current understanding of efferocytosis and defining how phagocyte diversity contributes to efferocytosis is essential for untangling the role of specific ligands and receptors. Unlike processes regulated by adaptive immunity, efferocytosis can occur in a non-autologous system and between different species (14). For example, mouse phagocytes can engulf apoptotic human cells and vice versa. A great amount of work has been done identifying the important receptor-ligand interactions and signaling pathways engaged during efferocytosis (15, 16); however, not everything that is known about this process can be applied to efferocytosis of apoptotic neutrophils by human phagocytes. For example, engulfment of apoptotic thymocytes involves CD14, whereas the engulfment of apoptotic neutrophils by macrophages does not (17). In addition, the phagocytosis of apoptotic neutrophils by fibroblasts involves a mannose/fucose specific lectin, whereas engulfment by unstimulated macrophages does not (18, 19). The remarkable plasticity and host-to-host variation of human phagocytes also influences the requirement for certain receptors and confounds existing data, making these cells notoriously difficult to study in the context of efferocytosis. Because the source of apoptotic cells, the species of origin, and the heterogeneity of the phagocyte population can influence the requirement for specific factors (20), the over generalization of this process should be avoided. Therefore, this review will focus primarily on the engulfment of human neutrophils by human macrophages and, to a lesser extent, dendritic cells.

Death of the neutrophil

Neutrophils are the first immune cells recruited to the site of injury or infection. Antimicrobial killing by neutrophils relies on the generation of toxic chemicals and delivery of pre-synthesized destructive enzymes from granules to phagosomes. While these degradative enzymes contribute to a beneficial antimicrobial response inside phagosomes, they can contribute to tissue damage if released extracellularly. Teleologically, efferocytosis of apoptotic neutrophils serves to prevent neutrophil lysis and subsequent release of cytotoxic and inflammatory components.

The true first step in efferocytosis is the induction of apoptosis of the target cell, therefore understanding how and when neutrophils die is crucial. Cellular apoptosis is characterized by phosphatidylserine (PS) externalization, DNA fragmentation, and membrane blebbing, resulting in phagocytosis by neighboring cells. Neutrophils undergo constitutive apoptosis after approximately 24 hours in circulation, although the exact timing is a subject of debate (21). Neutrophil apoptosis is usually initiated by one of two signaling cascades: the intrinsic or extrinsic apoptotic pathways. The intrinsic pathway relies on the loss of mitochondrial membrane integrity caused by the imbalance of pro- and anti- apoptotic factors in the cell and subsequent release of cytochrome c from the mitochondria. Cytochrome c facilitates caspase-9-dependent assembly of the Apaf-1-containing apoptosome and eventual activation of executioner caspase-3. The extrinsic pathway responds to extracellular signaling through death receptors to drive caspase-8-dependent activation of caspase-3. Fas-L, TNF-α, and TRAIL are all examples of soluble proteins that mediate extrinsic apoptosis through engagement of their cognate receptors expressed on the surface of neutrophils (22, 23).

Apoptosis of neutrophils can be accelerated, delayed, or derailed during several inflammatory and infectious conditions (reviewed in (2)). For example, phagocytosis of Fc- and complement-opsonized latex beads accelerates the apoptotic program of neutrophils (24). Some viral and bacterial pathogens also accelerate activation of apoptosis, whereas others initiate apoptosis and then abruptly cause lysis (25, 26). Conversely, neutrophil lifespan can be extended by G-CSF, GM-CSF, IL-8, bacterial components, C-reactive protein, serum amyloid A and other pro-inflammatory mediators (27-29). The ability of soluble mediators to extend neutrophil lifespan could ensure that complete bacterial clearance occurs prior to the onset of apoptosis and phagocytosis by macrophages. In contrast, it has been speculated that pathogens that accelerate apoptosis are able gain silent access into macrophages via efferocytosis (30). Despite the ability of some pathogens to exploit apoptosis, it remains an evolutionarily conserved pathways designed to promote normal cell turn over.

Macrophage heterogeneity

The role of phagocytes extends beyond host defense. Phagocytes play an important role during tissue homeostasis as they readily migrate towards and phagocytose senescent cells in circulation. Resident tissue macrophages, infiltrating macrophages, dendritic cells and non-professional phagocytes are all capable of engulfing apoptotic neutrophils (31).

The behavior of phagocytes, or phenotype, depends on specific stimuli derived from the extracellular milieu. In vivo, the number, frequency, and combination of exogenous stimuli that polarize macrophages toward different phenotypes seems almost limitless. Although more expansive naming systems have been proposed (32), this review will use the simplistic nomenclature of M1, M2a, M2b, and M2c to define macrophage subsets where relevant (33). M1 macrophages are generated by stimulation with IFN-γ and LPS, M2a macrophages with IL-4 or IL-13, M2b macrophages by immune complexes in combination with IL-1β or LPS, and M2c macrophages with IL-10, TGF-β or glucocorticoids (33). Because treatment with different exogenous stimuli program the macrophage to express a defined set of genes, the different macrophage subsets are beneficial in specialized contexts. For example, M1 macrophages upregulate the expression of those genes that are important in antimicrobial responses, whereas M2a macrophages upregulate the expression of genes that are important for anti-parasitic responses (34). Unsurprisingly, macrophage subsets also can express different sets of receptors, thereby influencing their phagocytic ability and efficiency. The mechanisms by which macrophage subsets and dendritic cells engulf apoptotic neutrophils, and their role in resolving inflammation, is addressed in the following sections.

Untangling the phagocytic synapse

Changes on the cell membrane of apoptotic cells initiate tethering to and engulfment by phagocytes. These changes to the cell surface include direct modification of lipids and proteins and the ‘indirect’ binding, or bridging, of serum components to the modified surface antigens. The receptor-ligand interactions between apoptotic cells and phagocytes form an engulfment synapse. This section will highlight some of the nuances regarding the discrimination of apoptotic neutrophils by human macrophages

Neutrophils downregulate a number of membrane proteins as they age, including CD16 (FcγRIII), CD31 (PECAM-1), CD32, CD35 (Complement receptor 1), CD43, CD45, CD44, CD47 (IAP), CD50 (ICAM-3), CD55 (Complement regulatory protein DAF), CD62L (L-selectin), CD63, CD66b, CD87 (uPAR), CD88 (C5a receptor), and CD120a (tumor necrosis factor receptor 1), thus significantly changing the apoptotic neutrophil surface (35-40). Some of these proteins, including CD31 and ICAM-3, are also functionally modified (41, 42), whereas other proteins, including ICAM-1, are upregulated on apoptotic neutrophils (41-43). Finally, new antigens are expressed on the surface of apoptotic neutrophils. Apoptotic neutrophils release intracellular proteins, including Annexin-I and calreticulin, which associate with the membrane and act as ligands for efferocytosis (reviewed in (44)). Furthermore, inactivation of an aminophospholipid translocase and activation of a lipid scramblase in apoptotic cells causes phophatidylserine (PS) accumulation on the cell's outer leaflet (45). Many serum components also recognize and bind to surface exposed PS, thereby amplifying the number of ligands present on apoptotic neutrophils. Since healthy neutrophils have no PS, or only transient PS, in the outer-leaflet of the cell membrane, surface detection of PS by labeled Annexin-V and exclusion of vital dyes has become the gold standard for detection of apoptosis.

These newly exposed ligands expressed on or bound to apoptotic neutrophils are recognized by a diverse group of receptors expressed on the surface of professional phagocytes. When studying the first identified efferocytosis receptors, α_vβ₃ integrin and PS receptor, investigators realized that different macrophage subsets exhibit a differential requirement for these two receptors (14). Given that the species and type of phagocyte and apoptotic cell alter the preference or requirement for different receptor-ligand pairs, the discussion of receptors will be limited to those that have been studied using primary human phagocytes and neutrophils.

Integrins

Unstimulated human macrophages primarily use the α_vβ₃ integrin (vitronectin receptor, CD61) system to recognize and engulf apoptotic neutrophils (Fig. 1). Savill and colleagues were the first to demonstrate that integrin α_vβ₃ recognizes the bridging molecule thrombospondin, which in cooperation with CD36, tethers and engulfs the apoptotic neutrophil (46, 47). Cultured macrophages produce thrombospondin and blockade of thrombospondin production or α_vβ₃ integrin on unstimulated macrophages reduces efferocytosis of neutrophils (17, 47, 48). Ligand redundancy exists for the α_vβ₃ integrin receptor. Serum-component milk fat globule-epidermal growth factor 8 (MFG-E8, lactadherin) also acts as bridging molecule between PS on apoptotic cells and α_vβ₃ integrin on phagocytes (49). These data demonstrate that in some instances where macrophages do not supply bridging molecules, serum may play an important role in efferocytosis.

Figure 1. Examples of receptor-ligand interactions during efferocytosis of apoptotic neutrophils.

Unstimulated macrophages (MΦ) primarily use receptors α_vβ₃ integrin and CD36 to efferocytose apoptotic neutrophils (PMN, A, left). Ligands on apoptotic PMN for α_vβ₃ integrin include thrombospondin (THBD) and MFG-E8, whereas ligands for CD36 likely include oxidized-LDL-like- or oxidized PS-moieties (A, left). Accessory interactions at the synapse between apoptotic PMN and unstimulated MΦ also include complement receptors CR3 and CR4 recognition of iC3b-opsonized apoptotic PMN (A, middle), and LRP-1 recognition of plasma membrane-associated CRT (A, right). Alternatively, β-glucan-stimulated MΦ recognize PS on apoptotic neutrophils via PSRs (B), and glucocorticoid-treated MΦ enhance efferocytosis through MerTK recognition of PS-bridging molecules Gas6 and Protein S (C).

In contrast to the role of α_vβ₃ in macrophages, α_vβ₃ does not play a large role in DC efferocytosis (50); however α_vβ₅ integrin might serve as an alternative receptor for DC efferocytosis. Integrin α_vβ₅ also recognizes MFG-E8 and is required for DC engulfment of apoptotic red blood cells, HeLa cells, and apoptotic, mycobacteria-laden neutrophils (50). In addition to α_vβ₅ integrin, CD36, DC-SIGN, and the TAM family of tyrosine kinases also play important roles in DC efferocytosis (50-52).

Finally, the interaction of apoptotic neutrophils and macrophages may involve β₂ integrins, and ligands for β₂ integrins are present on apoptotic cells. For example, α_Lβ₂ (lymphocyte function-associated antigen 1, LFA-1) can recognize neutrophil ICAM-3 (intracellular adhesion molecule-3) and α_Mβ₂ (complement receptor 3, CR3) can recognize deposited complement component iC3b (43, 53). In addition, β₂ is involved in lipoxin-mediated efferocytosis pathways (54); however other data suggests that the role of β₂ integrins in the engulfment of apoptotic cells may be context dependent (43, 55).

PS receptors

Ingestion of apoptotic cells correlates with loss of phospholipid asymmetry and appearance of PS on the outer leaflet of the plasma membrane (17, 56, 57). PS is the best characterized “eat me” signal on the surface of apoptotic cells, and its stereospecific recognition has been shown to function in the interactions leading to engulfment of apoptotic cells. However, unlike many other cell types (14, 56), human macrophages do not inherently favor recognition of apoptotic neutrophils via the PS-PS receptor (PSR) system as demonstrated by studies showing blockade of PS with PS-liposomes does not reduce macrophage efferocytosis of neutrophils (17). Macrophages that are stimulated with β-glucan, on the other hand, down regulate α_vβ₃ integrin, causing macrophages to switch from the α_vβ₃ integrin recognition to PSR/PS-dependent recognition and uptake is blocked by PS-liposomes (Fig. 1) (17). Additional studies are required to determine which of the PS receptors (BAI-1, TIM4, Stabilin-1 and -2, or RAGE (58-60) modulate tethering and/or internalization of apoptotic neutrophils by human macrophages.

CD36

CD36 is constitutively used as an efferocytosis receptor (Fig. 1). Blockade of CD36 reduces efferocytosis by human macrophages relying on the PS-receptor or α_vβ₃ integrin for efferocytosis of apoptotic neutrophils. It is unlikely that CD36 is an additional receptor for PS as inhibition of phagocytosis by anti-CD36 antibodies and PS-liposomes is partially additive (17). Rather, it is suggested that the CD36 ligand is an oxidized-LDL-like or oxidized PS moiety (17, 59, 61). The cooperative nature of CD36 and PS-engagement in human macrophages was examined further by exploiting biotin-avidin binding to label erythrocytes with protein, antibody, or lipid (62). In this study, CD36 promoted tethering, but not engulfment, of erythrocytes, and addition of PS-liposomes promoted internalization (62). These data suggest that some receptors serve a tethering purpose and are constitutively engaged, whereas other receptors engage the cytoskeleton machinery, and their activity may depend on the polarization state of the macrophage.

Complement receptors

These first studies on efferocytosis argued that additional serum components, such as complement factors, were not essential for efferocytosis (18). Conflict in the literature arises because macrophages secrete many of the proteins it uses as bridging molecules. In addition, the role of specific serum factors is dependent on the context. For example, complement-related collectins mediate efferocytosis by alveolar macrophages, which is consistent with the high levels of the collectins surfactants A and D (SP-A and SP-D) in the lung (63). Defects in efferocytosis are associated with several pathological conditions that manifest in the lung (64); therefore, complement directed therapeutics might be extremely promising in this context.

In addition, the clearance of apoptotic cells exhibiting features of secondary necrosis (frequently referred to as late apoptotic cells) seems to preferentially involve complement opsonization. In the latter example CR3 and CR4 mediate efferocytosis of complement-opsonized apoptotic neutrophils (Fig. 1) (53). Furthermore, dying neutrophils downregulate complement regulatory proteins (38), which may facilitate deposition of opsonic complement fragments on the cell membrane (65). Accordingly, serum enhances efferocytosis of neutrophil preparations containing higher percentages of late apoptotic neutrophils (20).

MerTK

Mer tyrosine kinase (MerTK), AXL, and TYRO3 are a family of efferocytosis receptors that recognize the serum proteins and PS-bridging molecules, Gas6 and Protein S (Fig. 1) (66). Mice deficient in MerTK signaling display a defect in the engulfment of apoptotic cells and develop autoantibodies, and mice with a triple deficiency in MerTK, AXL, and TYRO3 develop autoimmune disease (67, 68). The importance of MerTK in efferocytosis of apoptotic neutrophils is evident when macrophages are programmed to an M2c or M2c-like phenotype (69). Treatment with IL-10 and M-CSF, or glucocorticoid steroids, generate M2c macrophages, which preferentially engulf apoptotic cells compared to M1, M2a, and M2b macrophages (70). Following polarization, M2c macrophages upregulate MerTK surface expression, and synthesize Protein S and Gas6 and secrete them into serum (69-71). Accordingly, blockade of MerTK in M2c macrophages restores macrophage efferocytosis to baseline levels (69).

Interestingly antagonizing the nuclear receptor peroxisome proliferator-activated receptor (PPAR)-γ upregulates MerTK and Gas6 but fails to enhance uptake of apoptotic neutrophils (71), suggesting that MerTK expression alone is insufficient. Of interest, conventional M2c macrophages also exhibit rounding and higher levels of Rac activity, which may further augment the MerTK efferocytosis pathway (72). Similar to glucocorticoid steroids and PPAR-γ antagonists, complement component C1q also regulates macrophage expression of MerTK and Gas6 (73), but to determine if C1q plays a role in efferocytosis of apoptotic neutrophils by human macrophages requires further investigation.

LRP1

Lipoprotein receptor-related protein (LRP or CD91) recognizes a diverse set of ligands, including calreticulin (CRT, Fig. 1) (74). CRT normally functions as a chaperone protein in the endoplasmic reticulum (75). During apoptosis, CRT can be detected on the membrane of cells, including neutrophils (40). It is not fully understand how CRT is trafficked from the endoplasmic reticulum to the cell membrane during apoptosis; however, once at the membrane, CRT has the capacity to induce membrane ruffling and macropinocytosis in the murine macrophage cell line J774 macrophages (40). Compared to human monocytes, expression of LRP on resting human macrophages is low, and inhibition of LRP on human macrophages causes a small, but significant, reduction in efferocytosis of human neutrophils (76). It is possible that LRP may play a larger role in efferocytosis by other cell types or different macrophage subtypes; although, activation of macrophages to M1, M2a, M2b, or M2c subtypes fails to upregulate gene expression of LRP in human macrophages (77).

Receptor redundancy and specificity

Blockade of a single receptor has never been shown to completely abolish efferocytosis suggesting that either multiple, redundant efferocytosis pathways exist or individual receptors cooperate with other receptors in a synergistic manner. Efferocytosis receptors, including SR-A, LOX1, CD68, CD300, ATP binding cassette (ABC) A1 and A7, and ligands, including lysophosphatidylserine (lyso-PS), CCN1, C1q, MBL, and Annexin I, may all participate in neutrophil efferocytosis by human macrophages under certain conditions (74, 78-83). For example, previous data suggest a minor, or non-existent role, for lectin-mediated efferocytosis of apoptotic neutrophils (17, 18); however, more recent studies show that galectin-3-mediated enhanced macrophage efferocytosis of neutrophils occurs via a putative lectin-receptor (84).

The data discussed in this section highlight how different macrophage subtypes, and possibly macrophages cultured in different conditions, might generate conflicting results regarding the requirement for particular ligands and receptors; therefore, it is imperative that when studying efferocytosis we carefully characterize both the phagocyte and apoptotic neutrophil in a given experimental system. Overall, we can surmise that the efferocytosis pathways initiated in vivo involves the integration of signals from the microenvironment, tissue, individual cell, and presence or absence of accompanying inflammation.

Preferred methods for measuring efferocytosis in vitro

To continue to evaluate efferocytosis in primary human cells, in vitro assays are essential. Characterization of macrophage subtypes and death modalities in neutrophils are two aspects of improving these experimental systems, but it will also be imperative to distinguish between macrophages with tethered and internalized apoptotic cells.

Multi-color flow cytometry has proved advantageous for measuring efferocytosis. In this method, macrophages are co-cultured with pre-labeled apoptotic neutrophils and following a period of phagocytosis, neutrophils are subsequently counterstained with a separate label or marker. This method allows investigators to distinguish between macrophages with tethered neutrophils, which emit fluorescence for both labels, and macrophages with completely ingested neutrophils, which only emit fluorescence for the first label. This experimental system is good because it is sensitive, high-powered and unbiased; however, some minor caveats exist. Using conventional flow cytometry, it is not possible to determine if a single macrophage contains both engulfed and tethered apoptotic neutrophils. For assessing efferocytosis and infection, it is not impossible to discern if neutrophils laden with microbes or individual neutrophils and microbes are in the same compartment in the macrophage. These considerations often confound data regarding silent entry of pathogens into macrophages.

Microscopy of efferocytosis also has significant merit for the reasons addressed above. In particular, three-dimensional reconstruction from a stack of serial cross sections confocal images of cells can be useful to confirm complete internalization of apoptotic neutrophils and microbe-laden neutrophils by macrophages. Together, these approaches can help identify molecular processes involved in efferocytosis in disease and infection.

Intracellular signaling and digestion

Phagocytosis requires actin cytoskeleton rearrangements. Our understanding regarding cytoskeletal rearrangement during efferocytosis has been primarily examined using cell lines and model organisms amenable to genetic manipulation. The key findings from these studies are reviewed extensively elsewhere (1, 44, 45), however, a few key features are described here. C. elegans nematodes are commonly used to study efferocytosis, because 131 cells of the 1090 total cells in the worm are reproducibly destined to die. Characterization of this model system by Sydney Brenner, H. Robert Horvitz and John Sulston was recognized with a Nobel Prize in physiology and medicine in 2002. Using this model system, several groups identified a collection of signaling molecules that are involved in two pathways for the removal of apoptotic cells: the CED-12, -2, and -5 pathway and the CED-1, -7, and -6 pathway. The mammalian homologs of CED-12, -2, and -5 are ELMO, CrkII, and Dock180, respectively. CED-1 has sequence homology to scavenger receptors and the intracellular portion of LRP, and CED-7 and CED-6 are homologous to ABC-A1 and GULP, respectively. These signaling modules converge to activate CED-10 in C. elegans and the Rho family GTPase, Rac1, in mammals (1, 44, 45).

Phagocytosis is the ingestion of particles greater than 0.5 μm and it can occur through distinct mechanisms. During Fc-mediated phagocytosis by macrophages, pseudopodia containing Fc-receptors extend and bind to sequential ligands on the target forming a tight-fitting phagosome. This process has been termed the zipper-mechanism of phagocytosis and requires Rac1 (85, 86). In C. elegans and mouse macrophage cell line Mf4/4, efferocytosis morphologically resembles the zipper mechanism of phagocytosis with the pseudopods reaching around the target to form a tight fitting phagosome (45, 87).

Alternatively, two pathways of phagocytosis have been described which result in the formation of spacious phagosome: complement-mediated and macropinocytosis. In complement-mediated phagocytosis, complement coated particles interact with their cognate receptor, thereby triggering the sinking of the target into the macrophage (85). Unlike Fc-mediated phagocytosis, this pathway activates Rho family GTPase RhoA. In macropinocytosis, extensive membrane ruffling entraps particles as the membrane folds back onto itself, and like Fc-mediated phagocytosis, macropinocytosis requires Rac1 (1). PS-dependent efferocytosis closely resembles macropinocytosis (64) and PS-liposomes trigger extensive membrane ruffling and bystander engulfment (74). Moreover, PS-dependent efferocytosis by fibroblasts or J774 macrophages is sensitive to the macropinocytosis inhibitor, amiloride (40, 62). Together, these data indicate that multiple pathways are involved in efferocytosis, but much more work needs to be done to discern shared and distinct features of integrin- and complement-mediated macrophage efferocytosis of apoptotic neutrophils and conventional phagocytosis mediated by the same receptors.

Engulfed apoptotic cells are degraded through the process of phagosome maturation (44, 88). Studies comparing the rate of phagosome maturation in human macrophages showed that phagosomes containing apoptotic neutrophils acidify more rapidly than phagosomes containing antibody-opsonized apoptotic neutrophils. The rapid acidification of phagosomes containing apoptotic cells is dependent on Rho GTPase activation of ezrin-radixin-moesin (ERM) proteins, suggesting that (1) phagosome cargo dictates processing efficiency and (2) efferocytosis relies on a balance of Rho and Rac signaling (89).

Enhancing efferocytosis and implications in disease

Impaired efferocytosis has been observed in many human inflammatory and autoimmune diseases, including cystic fibrosis, chronic obstructive pulmonary disease, asthma, idiopathic pulmonary fibrosis, rheumatoid arthritis, systemic lupus erythematous, glomerulonephritis, and atherosclerosis (64). The failure to clear apoptotic cells may exacerbate inflammation (as discussed below), and therapies targeted at enhancing efferocytosis may improve outcomes in these conditions. In proven therapies and novel experimental conditions, enhanced expression of efferocytosis receptor-ligand pairs and activation of signaling moieties contributes to increased efferocytosis.

Glucocorticoid treatment leading to M2c programming is the quintessential example of a therapeutic agent that drives enhanced efferocytosis. Enhanced efferocytosis mediated by glucocorticoids is attributed to (1) enhanced expression of MerTK and ligands Gas 6 and protein S (69), (2) enhanced expression of Annexin I and its receptor ALXR (90), and (3) enhanced activation of Rac1 (72). The upregulation efferocytosis ligands and receptors following exposure to glucocorticoid steroids could also be attributed to peroxisome proliferator-activated receptor (PPAR)-γ activity during the differentiation from monocyte to macrophage (91). Inhibiting activation of PPAR-γ with antagonist GW9962 during differentiation reduces CD36 and AXL expression and reduces basal and glucocorticoid steroid-enhanced efferocytosis of neutrophils by human macrophages (71, 91). Interestingly, the same concentrations of GW9962 enhance MerTK and Gas6 expression by human macrophages, demonstrating that efferocytosis receptors MerTK, AXL, and CD36 are differentially regulated (71, 91). Treating human macrophages with low concentrations (0.1-10 μM) PPAR-γ activating ligand rosiglitazone during monocyte to macrophage maturation, increases CD36 expression, but is insufficient to alter MerTK expression or stimulate efferocytosis of apoptotic neutrophils by human macrophages (71, 91). Although there were no effects with rosiglitazone, it is worth noting here that PPAR-γ activating ligand pioglitazone normalizes the defective efferocytosis in monocytes from patients with chronic granulomatous disease (CGD) (92), and may be a promising therapeutic in this context.

Several therapeutics and experimental manipulations enhance efferocytosis by altering the RhoA-Rac1 balance in cells in favor of Rac activation. These treatments include lovastatin, N-acetylcysteine and CD44-ligation (93-95). Endogenous lipid mediator, Lipoxin-A4, also enhances Rac1 activity in human macrophages and promotes efferocytosis in a CD36, α_vβ₃, and β₂-dependent manner (54, 96). Accordingly, microparticles derived from activated neutrophils that contain pro-resolving lipids and cause macrophages to produce pro-resolving lipids, including lipoxins, enhance efferocytosis of apoptotic neutrophils by GM-CSF-treated macrophages (97, 98). Lyso-PS signaling through G2A also stimulates Rac1 activity and potentiates efferocytosis by murine macrophages and potentially human macrophages as well (83).

Macrolide antibiotics have a variety of immunomodulatory properties (99). Included in their repertoire is the ability of macrolides belonging to the 14-member and 15-member macrolide family, including erythromycin and azithromycin, to enhance efferocytosis of neutrophils by human alveolar macrophages. Enhanced efferocytosis driven by erythromycin is dependent on PS-PRS signaling, but not integrin signaling (100). In addition, azithromycin upregulated expression of mannose receptor (MR) in human alveolar macrophages (101), and MR might play a role in neutrophil efferocytosis. General immunosuppression by steroids and statins could be detrimental in the context of infection, whereas the dual bactericidal and pro-resolving action of macrolides would be more advantageous. Likewise, data supports that macrolide antibiotics are clearly beneficial in the treatment of community-acquired pneumonia (102).

Regulating efferocytosis

Neutrophils can expose PS on their surface in the absence of apoptosis (103, 104). Since neutrophils and other cells (105) can expose PS on their surfaces, yet resist engulfment, then other mechanisms must exist to control this process. The regulatory membrane proteins that prevent efferocytosis are referred to as ‘don't eat me’ signals and protect viable, healthy cells from phagocyte engulfment.

‘Don't eat me’ signals are difficult to study because the output is the absence of efferocytosis, and to date, only a small number of these signals have been fully described. The regulation of ‘don't eat me’ signaling is complex, and evidence suggests that at least three features contribute to turning on and off ‘don't eat me’ signaling, including activity, expression, and location of the ‘don't eat me’ molecules. CD31 and CD47 are the best characterized don't eat me signals that prevent the ingestion of viable leukocytes. In viable neutrophils, CD31 interacts homotypically with CD31 on phagocytes to mediate detachment. In apoptotic cells, CD31 is inactivated and these cells remained attached to the phagocytes and are internalized (42). Little is known about the deactivation of CD31 during apoptosis, but it is worth exploring.

Permissive attachment of inactive CD31 on apoptotic cells and active CD31 on macrophages further augments efferocytosis. Macrophage contact with apoptotic cells causes macrophage membrane depolarization, a process that facilitates firm binding of apoptotic cells, and CD31 homophilic ligation prevents macrophage repolarization through CD31-dependent inhibition of the voltage-gated potassium channel ether-à-go-go-related gene (ERG) (106). In summary, active CD31 promotes detachment of viable cells from phagocytes, whereas inactive CD31 prompts attachment of apoptotic cells to phagocytes, homotypic CD31 ligation in macrophages, and efferocytosis.

Apoptotic cells downregulate CD47 and sequester this molecule away from the phagocytic synapse during efferocytosis. In viable cells, including neutrophils and red blood cells, CD47 interacts with immunoreceptor trypsine-based inhibitory motif (ITIM)-containing protein SIRPα. Previous studies postulate that this interaction causes activation of inhibitory tyrosine phosphatases SHP-1 and SHP-2 to shut down actin-cytoskeleton rearrangement and phagocytosis (40, 107). Consistent with this earlier observation, blockade of either CD47 on the neutrophil or SIRPα on the phagocyte triggers extensive membrane ruffling and ingestion of viable cells (40).

CD300a (108), plasminogen activator inhibitor-1 (PAI-1) (109), and the urokinase receptor (uPAR) (110) have been identified as ‘don't eat me’ receptors using either cell lines or animal models. In addition, soluble factors including vitronectin (111), TNF-α (112, 113), HMGB1 (114-116), prototypic tissue pentraxin PTX3 (117), extracellular histones (118), and PGE₂-mediated elevation of intracellular cAMP (119) inhibit efferocytosis pathways, and understanding these inhibitory signaling pathways may be of therapeutic value.

Production of anti-inflammatory mediators

Because an enormous number of neutrophils are eliminated every day, any accompanying inflammation would be devastating for an individual. Efferocytosis of neutrophils is considered an anti-inflammatory and pro-resolving event. As such, harnessing the anti-inflammatory and pro-resolving nature of efferocytosis could be targeted therapeutically to combat chronic inflammatory conditions, including graft-verses-host disease, Crohn's disease, scleroderma, and atopic dermatitis (120). The anti-inflammatory nature of neutrophil efferocytosis is discussed here.

The first studies investigating whether apoptotic neutrophils trigger cytokine production by macrophages show that efferocytosis is a silent process, resulting in no production of pro-inflammatory cytokines (121, 122). Subsequent studies showed that efferocytosis is not merely a silent process, but actually immunosupressive. Engulfment of unopsonized, UV-irradiated, apoptotic neutrophils by macrophages actively inhibits LPS- and zymosan- dependent production of IL-1β, GM-CSF, TNF-α, and IL-10. In parallel, 18-hour coculture of apoptotic neutrophils with macrophages stimulates the production of anti-inflammatory mediators, TGF-β and PGE₂, but not IL-10 (12). We observed that macrophages produced increased amounts of TGF-β and IL-10 when co-cultured with apoptotic neutrophils for 18 hours in serum-containing media (unpublished data); however, macrophages given a shorter, one hour exposure to apoptotic neutrophils (in which more than 50% of the macrophages ingest apoptotic neutrophils) do not produce elevated IL-10 after an additional 12 hours in culture in serum-free media (26). These data suggest that the production of anti-inflammatory cytokines relies on either the continual presence of apoptotic neutrophils or serum components.

Immunosuppression caused by apoptotic neutrophils is receptor mediated. Using monocytes, which differ from macrophages in that they are unable to engulf apoptotic cells at least in vitro (123), allowed investigators to show that contact between dying cells and monocytes is sufficient to induce immunosuppression (11, 124). Monocytes cultured with apoptotic neutrophils suppress proinflammatory cytokine production following LPS stimulation and increased production of TGF-β and IL-10 (12, 124). Interestingly, Fc-mediated phagocytosis of antibody-opsonized neutrophils by macrophages or monocytes fails to inhibit LPS-mediated pro-inflammatory cytokine production (12, 124), whereas CD36 ligation on monocytes is sufficient to induce the anti-inflammatory effect on LPS-activated monocytes (11). It is also plausible that complement fragment iC3b on apoptotic cells promotes anti-inflammatory cytokine production, because iC3b recognition by monocytes promotes IL-10 production and dampens IL-12 production (125, 126).

The study of resolving inflammation has helped to identify a novel class of lipid mediators called specialized pro-resolving mediators (SPM). SPM, including lipoxins, resolvins, protectins and maresins, have potent anti-inflammatory properties (127). The Serhan group has elegantly demonstrated that efferocytosis of apoptotic neutrophils by macrophages leads to increased production of SPM including, Resolvin D1 (RvD1), RvD2, and lipoxin B₄. Additionally, PGE₂ production accompanies SPM production, and together these soluble mediators play a critical role in resolving inflammation (97). Blockade of soluble mediators, including TGFβ, PAF, or PGE₂ restored LPS-mediated cytokine production demonstrating that these anti-inflammatory mediators have a paracrine effect of blocking proinflammatory cytokine production by macrophages (12).

Prevention of secondary necrosis and necrosis

Neutrophils are full of noxious substances, including proteolytic enzymes and danger-associated molecular patterns that propagate inflammation and cause host tissue damage. Unfortunately, when efferocytosis is delayed necrotic corpses can accumulate. Many groups have compared how apoptotic neutrophils and necrotic neutrophils stimulate macrophages. Lysed neutrophils cause macrophages to produce the pro-inflammatory cytokine, TNF-α, as well as IL-10, whereas apoptotic neutrophils or necrotic T cells do not (128). The proinflammatory effects of lysed neutrophils are dependent on neutrophil elastase (128). In mice, coculture with necrotic neutrophils causes up regulation of costimulatory molecule CD40 on murine bone marrow-derived macrophages (BMDM) and enhance T cell responses, whereas colculture with apoptotic neutrophils has no effect (129), suggesting that necrotic debris has adjuvant like properties, whereas apoptotic cells do not. Finally, imaging studies conducted in laser-injured mice revealed that the cell death of one neutrophil amplifies chemotaxis of surrounding cells (130).

Although the clinical picture supports the notion that defects in the clearance of apoptotic neutrophils contributes to tissue damage and unwanted inflammation, there is also evidence that in certain contexts late apoptotic and necrotic neutrophils stimulate resolution. In the set of experiments analyzing the proinflammatory effects of necrotic neutrophils, it was noted that late apoptotic neutrophils did not stimulate macrophage pro-inflammatory cytokine production, and it is thought that caspases might inactivate inflammatory substances (131). Also, under serum-free conditions human necrotic neutrophils inhibit LPS-mediated cytokine production. Alpha-defensins are anti-microbial, cationic peptides expressed by human, but not mouse, neutrophils. Alpha-defensins are inactivated in serum; however, in the absence of serum, alpha-defensins released from dying neutrophils or liberated from necrotic neutrophils by freeze-thaw have potent anti-inflammatory activity (132). Understanding how necrotic neutrophils stimulate or fail to stimulate inflammation in different contexts is an important and clinically relevant area of study.

Insights into neutrophil efferocytosis in small rodent models

Much of the work investigating the process of eliminating neutrophils in vivo relies on studies involving animal models. Although little to no intravital imaging of efferocytosis currently exists in animal models, neutrophil efferocytosis by phagocytes has been demonstrated in vivo using a variety of other analytical techniques, including radiolabeling, and confirms this process occurs in the lung, spleen, liver, and bone marrow (4, 7, 133, 134). Consistent with the trafficking and destruction of apoptotic neutrophils in the liver, blocking Kuppfer cell phagocytosis in rats causes systemic neutrophilia and a particular increase in neutrophils in the lung and spleen (7). Clearance in the bone marrow is especially interesting, since intact and viable cells appear to migrate back to bone marrow for destruction (5). The CXCR4/SDP-1α signaling axis mediates trafficking into the bone marrow (135). Aging neutrophils upregulate CXCR4 (136), and while still Annexin-V-negative, older neutrophils can migrate towards the CXCR4 ligand, SDF-1a/CXCL12. When injected into mice, the CXCR4-high neutrophils home to a hematopoietic compartment of the bone marrow, where SDF-1a is constitutively expressed (135).

In vivo evidence of efferocytosis has also emerged from studies investigating the production of leukocytes from the bone marrow, or granulopoiesis. Several studies have suggested that efferocytosis in tissues is a prerequisite for turning off neutrophil granulopoesis. Interestingly, mice deficient in factors required for neutrophil extravasation (i.e. β₂ integrin/CD18 and P-selectin) into tissue exhibit neutrophilia; therefore, Ley and colleagues speculate that turning off neutrophil granulopoiesis requires extravasation into tissues (8). The engulfment of apoptotic neutrophils dampens IL-23 production, and IL-23-dependent IL-17 production by T cells. Because IL-17 is diminished, IL-17-dependent, GCSF-driven neutrophil granulopoiesis is therefore reduced. In support of these data, transfer of normal neutrophils into CD18^-/- mice restores IL-23 levels, reduces IL-17, and normalizes neutrophil numbers in the blood (8, 137).

Specialized subsets of efferocytic macrophages are detected during resolving inflammation in mice. During the resolution of murine peritonitis a CD11b-low population of macrophages emerges. Gene expression in the CD11b-low population diverges from the M1 and M2 framework because classic macrophage phenotype markers, arginase-1 and iNOS, are not found in the CD11b-low population. Regardless, CD11b-low macrophages engulf higher numbers of apoptotic neutrophils, and produce less pro-inflammatory cytokines but more TGF-β in response to TLR-ligands. Moreover, CD11b-low cells are less phagocytic than their CD11b-high counterparts and are more likely to migrate to the lymph nodes (138). The observed decrease in phagocyte appetite could be triggered by PGE₂ production and elevation of intracellular cAMP (119) or alterations in the Rho-Rac1 balance. Looking at the effects of efferocytosis on alveolar macrophage antibacterial function, Medeiros et al. demonstrated that signaling leading to production of PGE₂ and PGE₂ signaling through the EP2 receptor (but not TGF-β signaling) activates cAMP. Likewise, activation of cAMP causes a reduction in the ability of rat alveolar macrophages to engulf Fc-osponized targets and reduces the ability of alveolar macrophages to kill Streptococcus pneumoniae (139). These data highlight the plasticity of macrophages in vivo and how extracellular stimuli influence their effector functions.

Finally, although neutrophil reverse transmigration has been reported to occur in zebrafish, mice, and humans, only an extremely small percentage of human neutrophils (0.25% in healthy patients) have this capacity (140). Therefore, disappearance of neutrophils from sites of inflammation is most likely mediated by efferocytosis in the inflamed tissues.

Regulation of dendritic cell function

DC reside in tissues as an immature cell type and subsequently mature into professional antigen presenting cells capable of stimulating a T cell response. Phagosome maturation in DC is slow and less efficient than macrophages, favoring antigen presentation to T cells (141). Expression of costimulatory molecules on DC is also a prerequisite for maturation and T cell stimulation, whereas lack of costimulatory molecules contributes to T cell tolerance (142). DC are one of the cell types capable of recognizing and engulfing apoptotic neutrophils, thus understanding how DCs promote tolerance to self-antigen rather than immunity is a subject of significant interest.

Typical of DC, the phagosomes of DC containing either apoptotic cells or antibody-opsonized apoptotic cells mature at a slower rate than in macrophages (89), indicating that phagosome maturation in DC is different than macrophages in which apoptotic cargo acidify more rapidly than antibody-opsonized cells.

In line with their other immunomodulatory properties of apoptotic cells, engulfment of either apoptotic or necrotic neutrophils fails to induce expression of important costimulatory molecules on immature DC (143). Even in the response to pro-inflammatory molecules TNF-α or LPS, DC cocultured with apoptotic neutrophils failed to mature to the same extent as DC treated with TNF-α or LPS alone (89, 143). In contrast, necrotic tumor cells have been shown to induce DC maturation, whereas necrotic neutrophils, apoptotic neutrophils and apoptotic tumor cell lines tend to be suppressive (144). It is currently unknown how necrotic neutrophils differ from necrotic tumor cells in their inability to induce DC maturation.

Despite strong evidence that apoptotic cells prevent DC maturation, there is evidence that neutrophil efferocytosis is important for DC cross-priming of CD8⁺ T cells (145). Cross-priming occurs when extracellular antigen gains access to the cytosol and is then loaded into MHC class I molecules and presented to cytotoxic CD8⁺ T cell. Neutrophils carrying tumor antigen, Escherichia coli, or Mycobacterium tuberculosis are able induce cross priming in DC as evidenced by lymphocyte proliferation (146-148).

The exception are Leishmania major-laden neutrophils, which fail to cross-prime DC as described above. This parasite enters the skin via the bite of an infected sand fly, where it immediately triggers neutrophil infiltration. Neutrophils fail to control parasite burden and directly promote infection (149). Likewise, neutropenia has been associated with a better immune response, suggesting that neutrophils may be a double-edged sword in Leishmania infection. To study how neutrophil effector functions and lifespan could impair immunity to Leishmania, Ribeiro-Gomes and colleagues infected mice with Leishmania major and isolated neutrophils 12 hours following inoculation. They demonstrated that cultured DC are capable of capturing sorted uninfected and parasite-laden neutrophils, albeit at an extremely low rate (∼1.2% phagocytosis). Furthermore, DC associated with parasite-laden neutrophils are less capable of CD8+ T cells cross-priming compared to uninfected neutrophils from the same mouse. In contrast, DC that capture either uninfected or Toxoplama gondii-laden neutrophils are able to support CD8+ T cell cross-priming to the same extent indicating that inhibition of cross-priming is unique to Leishmania (52). Overall, these data demonstrate that efferocytosis is important for cross-priming during infection, and is targeted by at least one pathogen.

Macrophage efferocytosis and infection

An exciting recent hypothesis is that neutrophils containing viable bacteria can hijack efferocytosis to gain silent entry into phagocytes. Many examples of this phenomenon exist, as certain parasites and bacteria have been shown to survive neutrophil-mediated killing, modulate apoptosis, and be engulfed by phagocytes; in some instances, efferocytosis of neutrophils containing viable organisms results in a detrimental consequences for the host. Yet in other instances, efferocytosis has been shown to be beneficial in promoting a host response or efferocytosis simply does not happen at all.

Chlamydia pneumoniae, a Gram-negative, obligate intracellular bacterium, Yersinia pestis, a Gram-negative, facultative intracellular bacterium, and Leishmania major, an obligate intracellular protozoan, are engulfed by human neutrophils, survive within the phagosome, and modulate apoptosis (149-152). Neutrophils containing these viable microorganisms are subsequently engulfed by human macrophages (149-152). Entry via the neutrophil appears to enhance intracellular survival in macrophages; however, it is unclear if this is due to (1) cloaked entry via efferocytosis (2) upregulation of bacterial/parasitic pro-survival factors while in the neutrophil or (3) unchecked proliferation in the neutrophil and therefore higher initial bacterial or parasitic burden delivered to macrophages or other mechanisms. Despite the higher number of microorganisms present, macrophages cocultured with microbe-laden neutrophils produced less proinflammatory cytokines and more anti-inflammatory cytokines, including TGF-β and IL-1RA (150-152). These data suggest that pathogens can exploit efferocytosis to gain entry into a host cell. Nevertheless, the relevance of this pathway to pathogen dissemination and overall manipulation of the inflammatory response remains somewhat controversial and merits further study.

In contrast to the pathogens described above, efferocytosis is important for the control of Mycobacteria tuberculosis. Typically, M. tuberculosis prevents macrophage phagosome-lysosome fusion and replicates in the phagosomal niche. Efferocytosis of mycobacteria-containing apoptotic macrophages results in phagosome-lysosome fusion and pathogen elimination (153). Although neutrophil products contribute to anti-mycobacterial responses (154), M. tuberculosis delays neutrophil apoptosis and impairs T cell responses (155), suggesting that the ability to avoid efferocytosis contributes to M. tuberculosis pathogenicity. Future studies are required to confirm this hypothesis; however, it is still plausible that neutrophil efferocytosis contributes to antibacterial responses through expedited phagosome maturation, acquisition of anti-microbial neutrophil products, or through DC-cross priming as discussed above.

Finally, our lab has shown that community-acquired methicillin-resistant Staphylococcus aureus (CA-MRSA), prevents macrophage efferocytosis of CA-MRSA-laden neutrophils. CA-MRSA initially induces a program of accelerated apoptosis in neutrophils following phagocytosis; however, the neutrophil abruptly lyses without caspase activation (26, 156). While CA-MRSA-neutrophils are still intact, they bind to macrophages, but resist internalization. CD47 is upregulated on CA-MRSA laden neutrophils, but blockade of CD47 alone is not sufficient to induce efferocytosis. The inability of macrophages to engulf CA-MRSA-laden neutrophils likely perpetuates inflammation in two complementary ways: (1) by preventing macrophage anti-inflammatory cytokine production and (2) by culminating in neutrophil lysis and release of host and bacterial constituents that can activate macrophages and contribute to tissue damage (26, 157, 158). Especially in cases of antibiotic resistance, there is an urgent need for novel targets and therapeutics to treat S. aureus and other infections. Elucidating how individual pathogens manipulate efferocytosis may provide important insights into targets for antimicrobial therapies.

Conclusions

Efferocytosis of human neutrophils by phagocytes is critical for normal tissue homeostasis and resolution of inflammation. Outstanding questions remain in several areas, including the role of macrophage heterogeneity in neutrophil efferocytosis, and how specific pathogens modulate neutrophil survival and efferocytosis. Given the anti-inflammatory nature of efferocytosis, understanding the molecular players that initiate recognition and engulfment and downstream signaling events may lead to the development of novel therapeutics targeted at curbing unwanted inflammation.