Dr S. P. Hart, MRC Centre for Inflammation Research, University of Edinburgh Medical School, Teviot Place, Edinburgh EH8 9AG, UK. E-mail: firstname.lastname@example.org
Efficient phagocytic clearance of apoptotic cells is crucial in many biological processes. A bewildering array of phagocyte receptors have been implicated in apoptotic cell clearance, but there is little convincing evidence that they act directly as apoptotic cell receptors. Alternatively, apoptotic cells may become opsonized, whereby naturally occurring soluble factors (opsonins) bind to the cell surface and initiate phagocytosis. Evidence is accumulating that antibodies and complement proteins opsonize apoptotic cells, leading to phagocytosis mediated by well-defined ‘old-fashioned’ receptors for immunoglobulin-Fc and complement. In this review we summarize the evidence that opsonization is necessary for high capacity clearance of apoptotic cells, which would render putative direct apoptotic cell receptors redundant.
Apoptosis is the process of programmed cell death in vertebrates, during which the cell activates intrinsic suicide mechanisms that rapidly (within hours) lead to the characteristic features of cell shrinkage, chromatin condensation, membrane budding and eventually formation of one or more apoptotic bodies . Apoptosis and the subsequent phagocytic clearance of senescent cells are believed to play vital roles in many fundamental biological processes, including normal tissue turnover , remodelling of embryological tissues , development of the immune system  and the resolution of inflammation . Phagocytosis of senescent cells was described first in the late 19th century by the Russian biologist Elie Metchnikoff, who used a light microscope to observe injured tadpole fins. It has been assumed that macrophages and other phagocytes must be able to recognize changes on the surface of apoptotic cells that distinguish them from healthy live cells, but intensive research has failed to identify a single dominant phagocyte receptor responsible for apoptotic cell clearance. Inhibition studies using ligands and monoclonal antibodies have implicated many different and often unrelated phagocyte receptors for apoptotic cells, including the integrins αvα3  and αvα5 , CD36 , a phosphatidylserine (PS) receptor , various lectins [10,11], a scavenger receptor , CD14 , an ATP binding cassette transporter , LOX-1  and CD68 . It is notable that complete inhibition of phagocytosis of apoptotic cells has never been achieved in experimental systems, even when inhibitory antibodies or ligands have been used in combination. In contrast to studies of other intercellular adhesion processes, no demonstration of the capacity of purified receptors to mediate specific adhesion of apoptotic cells has been reported. Furthermore, unlike the worm Caenorhabditis elegans, there is a lack of convincing evidence that genetic deletions of putative mammalian apoptotic cell receptors have significant effects on apoptotic cell clearance in vivo. Apoptotic cell recognition may require co-ordinated engagement of multiple receptors such as the interaction between antigen presenting cells and T lymphocytes , or molecular redundancy may account for the bewildering array of putative apoptotic cell receptors. Publication bias also needs to be considered, because a new report of successful inhibition of apoptotic cell phagocytosis is more likely to be published than evidence of similar quality demonstrating that a particular receptor is not involved.
It is believed that phagocytosis of apoptotic cells is anti-inflammatory. For example, macrophage production of the anti-inflammatory cytokine transforming growth factor (TGF)-β is stimulated, and release of granule enzymes and proinflammatory cytokines is inhibited in response to ingestion of apoptotic cells [18–20]. One puzzle is how phagocytosis mediated by diverse receptors could lead to a common anti-inflammatory macrophage response. A possible explanation is that many of the proposed apoptotic cell receptors are really indirect modulators of binding or ingestion . In addition, apoptotic cells may become opsonized, whereby naturally occurring soluble opsonins bind to the cell surface and initiate phagocytosis. Opsonins may be constitutively present in serum, generated from inactive precursors, or released by cells. There is accumulating evidence that well defined serum opsonins such as antibodies and complement proteins may bind to apoptotic cells and mediate phagocytosis by classical phagocytic receptors (Fig. 1). In this way, opsonization of dying cells provides a mechanism for phagocytic removal that does not require an ‘apoptotic cell’ receptor. Furthermore, phagocytosis mediated by receptors for antibody or complement is very efficient, and has the capacity to dramatically accelerate clearance in response to an increased tissue load of apoptotic cells, which is absolutely required in order to avoid ‘secondary necrosis’ and further tissue damage [22,23].
A ROLE FOR SERUM IN PHAGOCYTOSIS OF APOPTOTIC CELLS
During inflammation the removal of apoptotic neutrophils is thought to be an important step in preventing the release of toxic granules and chemotactic factors into the extracellular fluid, thereby halting further injury and allowing resolution to occur . In early serum-free phagocytosis assays, macrophages specifically recognized and ingested neutrophils that had undergone spontaneous apoptosis during culture in 10% serum [24,25]. In vitro serum-free assays have been the cornerstone of most subsequent studies of the phagocytosis of apoptotic neutrophils and other apoptotic cells, and it has been naturally assumed that serum is not required for apoptotic cell clearance. On the other hand, several studies have reported a requirement for serum for phagocytosis of apoptotic cells in vitro[26–28]. Although we have found that serum present during an in vitro phagocytosis assay has no significant effect on macrophage phagocytosis of apoptotic neutrophils (our unpublished observations), it is possible that apoptotic cells might become opsonized during culture in the presence of serum, prior to their use in phagocytosis assays. Even when apoptotic cells are cultured in the absence of serum, macrophage-derived opsonins such as complement protein iC3b , C1q and MFG-E8  may be important . These observations provide evidence that opsonization of apoptotic cells is likely to play an important role in their clearance by phagocytes. Potential serum-derived opsonins include complement proteins, antibodies, collectins, pentraxins and anticoagulant proteins.
Several lines of evidence suggest that proteins of the complement cascade are required for efficient removal of apoptotic cells. Deficiency of C1q is the strongest known genetic risk factor for systemic lupus erythematosis (SLE), a human inflammatory disease chraracterized by numerous autoantibodies and circulating immune complexes , and C1q-deficient mice exhibit a lupus-like disease with autoantibodies, immune deposits and glomerulonephritis. Apoptotic cell clearance is impaired in C1q-deficient mice, which display multiple uningested apoptotic cell bodies in the kidneys . Reduced apoptotic cell phagocytosis has been confirmed in a model of peritoneal inflammation . These observations, along with those of Korb and colleagues, that exposure of self-antigens in surface blebs on apoptotic keratinocytes leads to direct binding of C1q , have led to the hypothesis that complement is required for proper processing and clearance of self-antigens . Mevorach and colleagues were able to induce production of autoantibodies in animals by injecting apoptotic thymocytes . Deficiencies of several other complement proteins also increase the risk of developing SLE, and complement components other than C1q may bind to apoptotic cells leading to ligation of macrophage complement receptors such as CR1 (CD35), CR3 (CD11b/CD18) and CR4 (CD11c/CD18). For example, opsonization of apoptotic Jurkat cells by iC3b enhanced macrophage phagocytosis, which could be inhibited by antibody blockade of CR3 and CR4 . Mevorach reported that addition of serum to phagocytosis assays increased the uptake of apoptotic cells more than threefold, but heat-inactivated serum or serum deficient in C3, factor B or C1q had a reduced effect [27,28]. Again antibodies against CR3 and CR4 partially inhibited uptake of serum-exposed apoptotic cells. Complement receptor-mediated phagocytosis appears morphologically distinct from Fc receptor-mediated phagocytosis  and may not induce a proinflammatory phagocyte response [38,39], consistent with a role for complement receptors in the clearance of apoptotic cells. It has been suggested that exposure of PS on apoptotic cells is responsible for opsonization with iC3b, as preincubation with annexin V partially inhibited complement binding . However, evidence from other studies suggests binding of proteins such as IgM or C-reactive protein (CRP) to apoptotic cells may be required for subsequent complement deposition [40,41].
There is circumstantial evidence that in disease, and perhaps in health, apoptotic cells become opsonized by antibodies. The antiphospholipid syndrome (APLS) is characterized by arterial and venous thromboses and the presence of ‘antiphospholipid antibodies’. Most of the these antibodies are in fact directed against phospholipid-associated proteins such as β 2-glycoprotein I (β 2-GPI), which binds to PS or other phospholipids exposed on apoptotic cells and leads to generation of autoantibodies [42,43]. It has been confirmed that β 2-GPI antibodies in the serum of human patients with APLS bind to apoptotic cells in vitro[44,45]. Phagocytosis of apoptotic thymocytes was increased in the presence of β 2-GPI antibodies . Similarly, phagocytosis of apoptotic cells by immature murine dendritic cells was enhanced substantially by opsonization with β 2-GPI antibodies , which has implications for subsequent antigen presentation and the perpetuation of autoimmune disease. Apoptotic cells that become opsonized with antibody, particularly IgG, are likely to be recognized by ligation of macrophage Fc receptors, resulting in stimulation of a proinflammatory response in autoimmune diseases such as APLS. It seems counterintuitive that a similar mechanism may operate in health, but there are clues that IgG may mediate apoptotic cell clearance in some experimental systems. Kurosaka reported that phagocytosis of apoptotic cells by human THP-1 monocytic leukaemia cells led to production of proinflammatory cytokines, notably IL-8, MIP-2 and tumour necrosis factor (TNF)-α[48,49]. Blockade of phagocyte FcγRI by monomeric IgG inhibited proinflammatory cytokine production and increased production of anti-inflammatory interleukin (IL)-10 and TGF-β. These findings raise the possibility that the apoptotic cells were opsonized with IgG under control conditions. Similarly, human macrophages derived from monocytes in the presence of serum may retain IgG bound via FcγRI, thus priming them for anti-inflammatory cytokine responses in response to phagocytosis of apoptotic cells . Although phagocytosis of IgG-opsonized particles may induce a proinflammatory response, this may be a small price to pay for substantially increased apoptotic cell clearance, because Fc receptor-mediated phagocytosis is much more efficient than uptake of ‘naked’ apoptotic cells. We have reported recently that IgG-containing immune complexes opsonize apoptotic neutrophils , which is associated with substantially enhanced phagocytosis by macrophages and only a small increase in release of proinflammatory cytokines (Hart SP et al. submitted). Fc receptor ligation following contact with these opsonized apoptotic cells may occur in the context of other apoptotic cell–macrophage ligand–receptor interactions, which alter the interpretation of the Fcγ R-mediated signals and down-regulates any ensuing inflammatory response. It is not yet clear how the balance of these pro- and anti-inflammatory processes influences the final outcome of inflammation.
IgM has also been implicated in the opsonization of apoptotic human T lymphocytes . Experiments with protease-derived fragments suggest that the Fab′ but not the Fc portion of IgM binds to phospholipase A2-derived lysophosphatidylcholine on the surface of apoptotic cells. IgM-deficient serum resulted in reduced binding of C3, suggesting that this may be one mechanism for complement deposition.
COLLECTINS, PENTRAXINS AND ANTICOAGULANT PROTEINS
A report of serum-mediated, complement-independent uptake of apoptotic cells by murine mesangial cells  raises the possibility that other potential opsonins such as the collectins, pentraxins and anticoagulant proteins may be involved in the opsonization of apoptotic cells. The collectins mannan binding lectin (MBL), surfactant protein A (SP-A) and SP-D are calcium-dependent lectins which form tertiary structures very similar to C1q, and recognize high mannose and other molecular patterns on pathogens . An increased frequency of MBL mutations has been reported in populations of patients with SLE and rheumatoid arthritis . Collectin binding may initiate phagocytosis both directly, via putative collectin receptors, and indirectly by triggering complement deposition. It has been proposed that collectins bind to apoptotic cells and trigger ingestion by ligating a complex of calreticulin and CD91 on macrophages [56,57]. Schagat and colleagues reported that SP-A (and to a lesser extent SP-D) enhanced phagocytosis of apoptotic neutrophils by rat alveolar macrophages, although baseline phagocytosis was only about 2%. In this study, MBL and C1q had no effect on apoptotic cell phagocytosis .
The pentraxins are acute-phase proteins that have been reported to bind to chromatin, small nuclear ribonucleoproteins, C1q and altered phospholipids . The pentraxin CRP (first identified as binding the C-polysaccharide of Streptococcus pneumoniae) is a useful blood marker in clinical practice as serum concentrations may increase more than 1000-fold in acute inflammation. CRP bound in a calcium-dependent, chromatin-independent manner to late apoptotic human lymphocytes . CRP opsonization increased subsequent binding of complement components C1q, factor B and iC3b, but reduced deposition of the membrane attack complex, suggesting that prior binding of CRP may be a prerequisite for complement binding. Phagocytosis-stimulated release of TGF-β from macrophages required the presence of normal serum or CRP, but was deficient with heat-inactivated or C1q-deficient serum, supporting the hypothesis that opsonization of apoptotic cells with complement is required for subsequent macrophage release of anti-inflammatory mediators. The pentraxin serum amyloid P (SAP) bound in a calcium-dependent manner to phosphatidylethanolamine on apoptotic lymphoma cells . Binding to membrane-permeable late apoptotic cells was notably stronger, but seemed to be independent of chromatin binding. This observation is important, because late apoptotic or ‘post-apoptotic’ cells have deteriorated to a stage where membrane integrity is lost, so potential opsonins may have access to the interior of the cell. It is not clear whether binding to molecules within these post-apoptotic cells would be accessible for recognition by phagocyte receptors. Other authors have reported calcium-independent pentraxin-3 binding to late apoptotic lymphoma cells, but to a lesser extent to primarily necrotic (detergent-permeabilized or freeze-thawed) cells, again providing evidence against binding mediated by cytoplasmic or nuclear components. Binding of pentraxin-3 was cross-inhibited by CRP and SAP, suggesting a common site for pentraxin binding on the apoptotic cell . Mold and colleagues have provided evidence that apoptotic cells opsonized with pentraxins are phagocytosed using an Fc receptor-mediated pathway, because macrophages from FcR γ-chain deficient mice (which lack Fcγ RI and Fcγ RIII) did not exhibit increased phagocytosis of apoptotic lymphocytes opsonized with SAP or CRP .
Recent reports suggest that the natural anticoagulant protein S may bind to apoptotic cells via a mechanism that may involve exposed PS in conjunction with C4b-binding protein. Binding of protein S has been reported to stimulate phagocytosis [63,64], suggesting that macrophages express receptors specific for protein S that confer recognition. The protein S homologue growth arrest-specific gene-6 (gas-6) may also opsonize apoptotic cells for phagocytosis mediated by Mer, a member of the Axl/Mer/Tyro3 receptor tyrosine kinase family [65,66].
Apoptosis is associated with many alterations in the protein and carbohydrate composition of the plasma membrane [10,67–70]. Some of these changes may be responsible for binding potential opsonins. In a similar way to the initial proliferation of putative apoptotic cell receptors, many different serum-derived and cell-secreted factors have been proposed to opsonize apoptotic cells and subsequently mediate their phagocytosis. However, common themes have started to emerge, with complement proteins being strongly implicated and increasing evidence that antibodies may opsonize apoptotic cells. Well-characterized receptors for Ig-Fc and complement mediate high capacity phagocytosis which would be required in the face of a large apoptotic cell load, so obviating the need for a unique apoptotic cell receptor. It remains to be determined whether the same is true following binding of other potential opsonins such as collectins, pentraxins or anticoagulant proteins. Experiments utilizing genetic knock-outs of putative opsonins and receptors for antibodies and complement will help to define the contributions of ‘old fashioned’ phagocytic pathways in apoptotic cell clearance.
This work was supported by a Medical Research Council Clinician Scientist Fellowship (SPH).