Re: An Apoptotic ‘Eat Me’ Signal: Phosphatidylserine Exposure - cell 리뷰논문

[문형철]

beyond reason

An Apoptotic ‘Eat Me’ Signal- Phosphatidylserine Exposure.pdf

2.32MB

Figure 1. The Disruption of Asymmetrical Phospholipid Distribution.

In normal cells, flippase (ATP11C) continuously flips phosphatidylserine (PtdSer) in the plasma membrane, and scramblases (TMEM16F and Xkr8) are inactive.

In activated cells, increases in intracellular Ca2+ transiently activate TMEM16F to scramble phospholipids, probably inactivating flippase activity of ATP11C. When cells return to a normal state, a decrease in intracellular Ca2+ inactivates TMEM16F, and ATP11C restores the asymmetrical PtdSer distribution. Thus, the Ca2+-induced PtdSer exposure is reversible.

By contrast, when cells undergo apoptosis, caspases cleave ATP11C and Xkr8 to inactivate flippase and activate scramblase, respectively, leading to irreversible PtdSer exposure.

PtdSer and PtdEtn are colored in red, and PtdCho and sphingomyelin are in blue. Caspase recognition sites for ATP11C and XKR8 are boxed by a red line. Active caspases are shown by scissors.

Figure 2. The Two-Step Engulfment of Apoptotic Cells. In the tethering step, the phosphatidylserine (PtdSer) receptor Tim4 tightly binds PtdSer on the apoptotic cell and recruits it to the macrophage surface. In the tickling or uptake step, soluble proteins such as protein S/Gas6 or MFG-E8 bind PtdSer on apoptotic cells and activate their receptors (MerTK or integrin, respectively) on phagocytes, leading to Rac1 activation and actin polymerization. Both tethering and tickling are essential steps in the efficient engulfment of apoptotic cells in mouse-resident peritoneal macrophages. Whether a similar two-step engulfment mechanism that occurs in the engulfment of apoptotic cells in other macrophages remains to be studied.

Abbreviations: Gas6, growth arrest-specific 6; MFG-E8, milk fat globule EGF factor 8.

Figure 3. Constitutive Exposure of Phosphatidylserine (PtdSer) in Living Cells and their Engulfment. In normal resting cells, a flippase(s) such as ATP11C translocates PtdSer and phosphatidylethanolamine (PtdEtn) from outer to inner leaflets, while Ca2+- or caspase-dependent scramblases are inactive. This keeps, PtdSer, the ‘eat me’ signal to the inner leaflet of the plasma membrane, and macrophages cannot engulf them. When cells lose flippases, for example, by the lack of CDC50A (cell division cycle protein 50A), cells cannot maintain the asymmetrical distribution of phospholipids, thus exposing PtdSer to the cell surface of living cells with no involvement of scramblases. The PtdSer exposed on the cell surface under these conditions is stable, does not flip-flop at the plasma membrane, and is efficiently and tightly recognized by PtdSer receptor or PtdSer-binding proteins. Thus, living, PtdSer-exposing cells that lack flippase are engulfed by macrophages in a PtdSer-dependent manner, indicating that PtdSer is sufficient and necessary for cells to be eaten. By contrast, when aspartic acid at amino acid position 430 of the scramblase TMEM16F is mutated to glycine (D430G), the mutant seems to respond to an endogenous low level of Ca2+, and scrambles phospholipids at the plasma membrane, thus exposing PtdSer in living cells. In this case, a flippase(s) is also active, and PtdSer rapidly moves between the inner and outer leaflets of the plasma membrane. This will reduce the affinity of the exposed PtdSer to the PtdSer receptor or its binding molecule, and prevent the engulfment of PtdSer-exposed cells.

Review

An Apoptotic ‘Eat Me’ Signal: Phosphatidylserine Exposure

Katsumori Segawa1 and Shigekazu Nagata1,*

Apoptosis and the clearance of apoptotic cells are essential processes in animal development and homeostasis. For apoptotic cells to be cleared, they must display an ‘eat me’ signal, most likely phosphatidylserine (PtdSer) exposure, which prompts phagocytes to engulf the cells. PtdSer, which is recognized by several different systems, is normally confined to the cytoplasmic leaflet of the plasma membrane by a ‘flippase’; apoptosis activates a ‘scramblase’ that quickly exposes PtdSer on the cell surface.

The molecules that flip and scramble phospholipids at the plasma membrane have recently been identified. Here we discuss recent findings regarding the molecular mechanisms of apoptotic PtdSer exposure and the clearance of apoptotic cells.

Phosphatidylserine Exposure during Apoptosis

Every day, billions of senescent or damaged cells in the body undergo apoptosis (see Glossary) (Box 1) and are cleared by macrophages. Macrophages engulf apoptotic but not healthy cells [1], suggesting that apoptotic cells expose an ‘eat me’ signal to phagocytes. Many ‘eat me’ signals have been proposed, including phosphatidylserine (PtdSer), carbohydrates (amino sugars or mannose), intercellular adhesion molecule-3 (ICAM3), and calreticulin [2].

Of these, PtdSer is the most studied and the most likely ‘eat me’ signal candidate. PtdSer normally localizes to the inner leaflet of the plasma membrane, but is exposed to the cell surface during apoptosis [3]. Red blood cells that are heavily loaded with PtdSer are recognized and engulfed by macrophages [4]. However, apoptotic cells are not engulfed by macrophages either in vitro or in vivo when PtdSer is masked by PtdSer-binding proteins or competed with PtdSer-bearing liposomes [5,6]. Although PtdSer exposure and PtdSer-dependent apoptotic cell engulfment are evolutionarily conserved from lower organisms (Caenorhabditis elegans and Drosophila) to mammals [7,8], the molecular mechanism that externalizes PtdSer to the plasma membrane during apoptosis has only recently been identified.

Flippases in the Plasma Membrane

In healthy cells, phospholipids are maintained asymmetrically in the plasma membrane [3]. The amine-containing lipids, PtdSer and phosphatidylethanolamine (PtdEtn), are localized to the cytoplasmic leaflet of the plasma membrane, while phosphatidylcholine (PtdCho) and sphingomyelin are concentrated in the exoplasmic leaflet. ATP-dependent aminophospholipid translocase(s) or flippase(s) are proposed to establish this asymmetrical phospholipid localization by transporting PtdSer and PtdEtn from the outer to the inner leaflet of the lipid bilayer (Box 2). Candidate flippases include type IV P-type ATPases (P4-ATPases); this large family has 15 murine and 14 human members of membrane proteins carrying ten transmembrane segments [9]. CDC50 functions as a chaperone for the proper subcellular localization of P4-ATPases, and is necessary for their lipid transport activity [10]. Flippase activity was originally found in bovine chromaffin granules, and ATPase II, a P4-ATPase (now called ATP8A1), was surmised to function as a flippase [11]. It was also reported that Drs2p, a yeast ortholog of ATP8A1, was responsible for flipping PtdSer in plasma membranes [12]. However, Drs2p mainly localizes to intracellular vesicles of the trans-Golgi network [13]; therefore, ATP8A1 and Drs2p likely flip PtdSer in intracellular vesicles rather than in plasma membranes. By contrast, Dnf1p and Dnf2p predominantly localize to the plasma membrane in yeast, and promote translocation of PtdSer, PtdEtn, and PtdCho [14]. The exposure of PtdEtn on the plasma membrane is increased in the mutants lacking both dnf1 and dnf2, supporting their essential roles as flippases [14]. By genetic screening with the human near-haploid cell line KBM7, the P4-ATPase family member ATP11C and its chaperone CDC50A were found to be required for flipping PtdSer and PtdEtn at the plasma membrane [15]. ATP11C localizes to the plasma membranes in a CDC50A-dependent manner, and ATP11C deficiency severely reduces the PtdSer and PtdEtn flippase activity in WR19L (mouse T-lymphoma cell line) and KBM7 (human myeloma cell line) cells. However, ATP11C-null cells maintain an asymmetrical PtdSer distribution even though flippase activity is reduced by approximately 80%, suggesting that factors other than ATP11C help generate PtdSer asymmetry. Consistent with this possibility, asymmetrical PtdSer distribution is normal in the plasma membrane of ATP11C-deficient mouse pro-B cells and thymocytes, where PtdSer flippase activity is significantly reduced [16]. By contrast, CDC50A-deficient KBM7 and WR19L cells lose their PtdSer flippase activity at plasma membranes, and constitutively expose PtdSer to the cell surface [15]. CDC50A functions as a chaperone for multiple P4-ATPases, from yeast to mammals [17,18], and KBM7 and WR19L cells express several P4-ATPases. It is likely that ATP11C directly flips PtdSer at the plasma membrane and is responsible for most of the PtdSer flippase activity in WR19L and KBM7 cells, while other molecules have weak, but enough, flippase activity to maintain asymmetrical PtdSer distribution. These findings suggest that CDC50A-regulated molecules with weak flippase activity exist. Furthermore, these molecule(s) may not localize to the plasma membrane, since a flippase located at trans-Golgi network might still regulate asymmetrical PtdSer distribution at the plasma membrane [19]. CDC50A-overexpressing ATP11C-deficient human and mouse cell lines in which the flippase activity is strongly reduced are available [15]. The question of whether other P4-ATPases have flippase activity can be addressed by introducing the 14 human P4-ATPases individually into these cell lines, and by assaying their ability to incorporate fluorescent labeled PtdSer. ATP11C belongs to the P4-ATPase subfamily of the P-type ATPases [9]. Among P-type ATPases, the tertiary structures of P2-type ATPases, including sarcoplasmic reticulum Ca2+ ATPase (SERCA), Na+ ,K+ -ATPase, and H+ , K+ -ATPase, have been studied the most [20–22]. Na+ ,K+ -ATPase and H+ ,K+ -ATPase are heterodimers consisting of a P-type ATPase catalytic /-subunit and a b-subunit that functions as a chaperone to stabilize the newly synthesized /-subunit and help it exit the endoplasmic reticulum (ER) [23]. Similarly, CDC50A, a chaperone for ATP11C, may also form a complex with ATP11C as a subunit of the flippase. Like SERCA and Na+ ,K+ -ATPase, ATP11C carries ten transmembrane regions, and their cytoplasmic region contains a nucleotide-binding (N), phosphorylation (P), and actuator (A) domain. These proteins use a sophisticated series of conformational changes to translocate Ca2+ or Na+ into the lumen [20]. The amino acid sequence of ATP11C is well conserved with those of SERCA and Na+ ,K+ -ATPase, suggesting that ATP11C may have a similar structure and transport mechanism. However, phospholipids are much larger (approximately 40 times) than Ca2+ and Na+ , and this issue has been referred to as the ‘giant substrate problem’ [24]. Recent mutagenesis analyses with yeast P4-ATPases of Dnf1 and Drs2 revealed several crucial residues for the substrate specificity [25,26]. A detailed structural analysis of P4-ATPase such as ATP11C using X-ray crystallography will be needed to clarify the mechanism of transporting phospholipids from the outer to inner leaflets in the plasma membrane.

Box 1. Caspase Activation during Apoptosis

Apoptosis is executed by two distinct intrinsic and extrinsic pathways [97]. Caspases, which are cysteine proteases, are activated by both intrinsic and extrinsic pathways to execute apoptosis. In the intrinsic pathway, which is activated during animal development or by genotoxic stimuli (antitumor drugs or g-rays), the upregulation of BH-3 proteins of the Bcl-2 family triggers Bax/Bak oligomerization. The oligomerized Bax/Bak causes mitochondria to release cytochrome c, which forms a complex with Apaf-1 to activate caspase 9. This leads to the sequential activation of the downstream effector caspases, caspase 3 and 7. The extrinsic pathway is initiated by death ligands such as Fas ligand (FasL), tumor necrosis factor (TNF), and TNF-related apoptosis-inducing ligand (TRAIL). The engagement of FasL with its receptor (Fas) assembles the death-signaling complex (DISC), which is composed of Fas, an adaptor protein (FADD), and procaspase 8. Procaspase 8, which is processed to its active form in the DISC, transduces death signaling via two different pathways. Type I cells (e.g., thymocytes) produce large amounts of active caspase 8, which directly activates caspase 3. By contrast, caspase 8 activation is weak in type II cells such as hepatocytes, in which active caspase 8 cleaves a BH3-only protein Bid to activate Bax/Bak and trigger the activation of downstream effector caspases. More than 500 substrates are cleaved by effector caspases [98], which are responsible for DNA fragmentation [99], membrane blebbing [100], nucleotide release [101], and PtdSer exposure.

Box 2. Flippase and Scramblase

Flippase and scramblase are attractive names. Unfortunately, five proteins of different categories are called flippase, and a molecule that seems to have nothing to do with scrambling of phospholipids is called PLSCR (phospholipid scramblase). The first category of flippases comprises a family of proteins that translocate specific phospholipids from the inner to outer leaflets in an ATP-dependent manner [3,102]. They are also called ‘aminophospholipid translocases’, and some P4-type ATPases work as flippases of this category. The second category consists of proteins at the ER that translocate dolichol-conjugated sugars from the cytoplasm to ER lumen for N-glycosylation of proteins. ATPases belonging to the ABC transporter family are flippases of the third category, and they work as drug transporters. In the fourth category, bacterial membrane proteins translocate lipopolysaccharides from the cytoplasmic to the periplasmic membrane surface, while another category contains yeast protein that catalyzes sequence (Flippase recognition target or FRT)- specific DNA recombination. In this review, flippases of the first category, ATP-dependent aminophospholipid translocases at plasma membranes, are discussed. In 1996, Basse et al. [33], by assaying scramblase activity in a reconstituted liposome system, isolated a 32-kDa protein with scramblase activity from human erythrocytes; this protein was designated a phospholipid scramblase (PLSCR1). The cDNA for human PLSCR1 was molecularly cloned [103] and used to identify four human, four mouse, and two Drosophila PLSCR homologs [104,105]. PLSCR1 is a cytoplasmic protein, and PLSCR1-deficient cells have normal phospholipid-scrambling activity [106]. Although this was attributed to redundant roles of the four PLSCR homologs in the mouse genome, a deletion of both PLSCR genes in Drosophila did not affect scramblase activity [104], arguing against this possibility. Recently, PLSCR1 was found to be involved in signal transduction, particularly in the interferoninduced antiviral pathway [93]. Thus, PLSCR is a misnomer.

Scramblases in the Plasma Membrane

PtdSer is rapidly exposed to the cell surface during apoptosis [1,27], platelet activation [28], cell fusion [29], mineral deposition in osteoblasts [30], and other processes. However, once established, PtdSer asymmetry is disrupted only slowly by passive diffusion (t1/2 of several days) [31]. These observations led to the postulation that phospholipid scramblases exist that nonspecifically and bidirectionally transport or scramble phospholipids in the plasma membrane [32]. One molecule called PLSCR (phospholipid scramblase) was claimed to be a scramblase [33], but it seems to be an artifact obtained by a cell-free assay system (Box 2). Recently, two membrane proteins responsible for phospholipid scrambling have been identified: transmembrane protein 16F (TMEM16F) and Xk-related protein 8 (Xkr8) [34,35] (Figure 1). TMEM16F, which contains ten putative transmembrane domains, localizes to the plasma membrane. Using in vitro reconstitution assays, purified TMEM16F orthologs from fungus were found to have calcium-dependent scramblase activity [36,37], suggesting that TMEM16F alone functions as a scramblase. TMEM16F belongs to the TMEM16 family, which has ten murine family members of which TMEM16C, -16D, -16F, -16G, and -16J support calcium-induced phospholipid scrambling [39]. TMEM16F is ubiquitously expressed in various tissues, while TMEM16C is expressed in the brain, TMEM16D in the brain, eye, and ovary, TMEM16G in the intestine, and TMEM16J in the intestine and skin [39]. A mutation in the TMEM16F gene causes a mild bleeding disorder in humans called Scott syndrome [34], indicating that TMEM16F is indispensable for the PtdSer exposure in activated platelets required for blood clotting.

By contrast, when platelets aretreated with ABT-737, an apoptosis inducer,they undergo apoptosis and expose PtdSer. It was reported that apoptotic platelets require caspase but not calcium to expose PtdSer [38], suggesting the existence of two different scramblases, Ca2+-dependent and caspase-dependent. In fact, TMEM16F is dispensable for lipid scrambling during apoptosis, while Xkr8 is required for apoptotic PtdSer exposure [39]. The Xkr8 protein, which contains six putative transmembrane domains, localizes to the plasma membrane [35]. Of the nine murine Xkr family members, Xkr4, Xkr8, and Xkr9 are cleaved by caspase and scramble phospholipids during apoptosis [40]. Xkr8 is expressed ubiquitously, but Xkr4 and Xkr9 are expressed in the brain and intestine, respectively. Thus, apoptotic Xkr8-null mouse embryonic fibroblasts or immortalized fetal thymocytes do not expose PtdSer. The C. elegans Xkr8 ortholog CED8 promotes PtdSer exposure during programmed cell death [35,41], indicating that Xkr8/CED8-mediated phospholipid scrambling is an evolutionarily conserved mechanism for apoptotic PtdSer exposure. Whether Xkr8 acts alone as a scramblase or if other molecules are required remains to be studied. 

Flippase and Scramblase Regulation in Apoptosis

Early studies reported that apoptotic PtdSer externalization is caspase-dependent [42] and accompanied by scramblase activation and flippase inactivation at the plasma membrane [32,43]. Both Xkr8 scramblase and ATP11C flippase have caspase recognition sites [15,39]. ATP11C has three caspase-3 recognition sites at its middle or nucleotide-binding domain, flanking a conserved /-helix. Caspase cleavage at these sites apparently inactivates flippase activity of ATP11C. Asp-to-Ala point mutations in the caspase recognition sites prevent ATP11C cleavage without affecting its flippase activity. Cells expressing caspase-resistant ATP11C could not expose PtdSer during apoptosis [15], indicating that the flippase must be inactivated for apoptotic PtdSer exposure. Xkr8 is cleaved by caspases at a C-terminal cleavage site [39]. Apoptotic cells expressing a caspase-resistant Xkr8 could not expose PtdSer, suggesting that caspase cleavage activates the scramblase activity of Xkr8.

These findings reveal how PtdSer is kept at the inner leaflet of the plasma membrane in living cells, and how this asymmetrical distribution is disrupted to expose PtdSer during apoptosis. In healthy cells, ATP11C flippase actively or ATP dependently translocates PtdSer and PtdEtn from the outer to the inner leaflet of the plasma membrane, while Xkr8 scramblase remains inactive, confining PtdSer to the inner leaflet (Figure 1). During apoptosis, caspases simultaneously inactivate ATP11C flippase and activate Xkr8 scramblase, leading to rapid PtdSer exposure on the cell surface. This model is true at least in several human and mouse cell lines studied [15,39]. However, ATP11C and Xkr8 is a member of a large P4-type ATPase and Xkr families, respectively. It is possible that other members of the P4-type ATPase or Xkr family specifically or redundantly support the flippase and scramblase activity in other cell types.

Engulfment of Apoptotic Cells

Macrophages and immature dendritic cells as well as epithelial and mesenchymal cells are capable of engulfing apoptotic cells in a process known as phagocytosis (as well as efferocytosis to distinguish it from the complement-mediated phagocytosis) [44]. Macrophages are present throughout animal tissues and their relatives in the liver and brains are called Kupffer cells and microglia, respectively. These cells engage to engulf apoptotic cells as professional phagocytes. However, macrophages in varying tissues have different gene expression‎ profiles [45], supporting the idea that macrophages are heterogeneous [46]. Accordingly, macrophages seem to use a different set of molecules to engulf apoptotic cells that is dependent on tissue context.

Nematodes do not carry macrophages, and neighboring nonprofessional cells engulf apoptotic cells [47]. Similarly, clearance of apoptotic cells, although delayed, occurs in macrophagedeficient mice, in which mesenchymal neighbor cells engulf apoptotic cells with less efficiency [48]. Professional and nonprofessional phagocytes seem to recognize PtdSer on apoptotic cells as an ‘eat me’ signal. Whereas, either a cell surface PtdSer receptor, or a secreted protein that strongly binds PtdSer on apoptotic cells and a membrane protein on the macrophage can be found only in professional phagocytes. Here, we focus on engulfment by professional phagocytes.

Secreted PtdSer-Binding Molecules

At least three secreted proteins bind PtdSer. Milk fat globule EGF factor 8 (MFG-E8), originally identified in mammary glands, is a 46-kDa glycoprotein expressed in thioglycollate-elicited peritoneal macrophages, tingible body macrophages in the germinal centers of the spleen, and granulocyte macrophage colony-stimulating factor (GM-CSF)-induced bone marrowderived immature dendritic cells [49,50]. MFG-E8 binds PtdSer on apoptotic cells via its C-terminal Factor VIII homologous domain, and binds integrin-/vb3 on phagocytes via its N-terminal RGD motif. Thus, MFG-E8 assists apoptotic cell engulfment by forming a bridge between apoptotic cells and phagocytes [50]. Accordingly, the ability of thioglycollate-elicited peritoneal macrophages and tingible body macrophages to engulf apoptotic cells is impaired in MFG-E8-deficient mice [51].

Two related serum proteins, growth arrest-specific 6 (Gas6) and protein S (PROS1), specifically bind PtdSer [52] and TAM (Tyro3, Axl, and MerTK) tyrosine kinase receptors [53] to form bridges between apoptotic cells and macrophages. Gas6 and protein S, which share approximately 40% amino acid sequence identity, bind PtdSer via an N-terminal g-carboxylated Gla domain. Gas6 and protein S are important for blood clotting; a deficiency in either protein causes platelet dysfunction [54,55], suggesting that their roles are not redundant. MerTK, Axl, and Tyro3 are related TAM tyrosine kinase receptors, and their extracellular region consists of two IgG-like and two fibronectin (FN)-like domains. Gas6 and protein S bind to the TAM receptor via a sex hormone-binding globulin domain (SHBG). Whether Gas6 and protein S differ in their affinity for the three TAM receptors has not been studied in detail.

MerTK is expressed in a specific set of macrophages, including resident and thioglycollate-elicited peritoneal macrophages [56,57]. MerTK-deficient resident peritoneal macrophages cannot engulf apoptotic cells [58,59]. Furthermore, MerTK-deficient mice develop a lupus-like autoimmune disease, and knocking out additional TAM receptor genes exaggerates this phenotype [60], suggesting that macrophages and immature dendritic cells might use the three TAM receptors differently. 

PtdSer Receptors

Several proteins have been proposed to serve as the membrane receptor that recognizes PtdSer. A hamster monoclonal antibody (Kat5) that inhibits apoptotic cell engulfment by mouseresident peritoneal macrophages was identified, and used to screen a peritoneal macrophage cDNA library in a mammalian expression‎ vector. This procedure led to the identification of Tim4 (T cell immunoglobulin and mucin domain-containing molecule 4) [61], a type I membrane protein that is highly expressed in resident peritoneal macrophages. Tim4 strongly (
2 nM Kd) and specifically binds PtdSer. Established Tim4-deficient mouse lines revealed that resident, but not thioglycollate-elicited, peritoneal Tim4-null macrophages lack the ability to engulf apoptotic cells [59,62,63]. The mouse Tim family has three members: Tim1, Tim3, and Tim4. Tim1 is expressed in injured kidney cells and plasmacytoid dendritic cells, and binds PtdSer with an affinity similar to Tim4 [61,64]. Tim3, which is expressed in thioglycollate-elicited peritoneal macrophages and CD8+ dendritic cells, may also bind PtdSer to promote apoptotic cell engulfment [65,66], but this is controversial [61].

CD300 is a type I transmembrane protein carrying a single extracellular immunoglobulin (Ig)-like domain and intracellular immunoreceptor tyrosine-based inhibitory motifs (ITIMs). Among seven human CD300 members, CD300a, CD300b, and CD300f bound PtdSer and PtdEtn, suggesting that they function as a PtdSer receptor [67,68]. One study found that expression‎ of CD300a or CD300f in fibroblasts promoted the engulfment of apoptotic cells [67,69], whereas another study showed that CD300a in fibroblasts inhibited the engulfment [68]. It is not clear how similar receptors that transduce negative signals promote or inhibit the engulfment of apoptotic cells. Other membrane proteins, including phosphatidylserine receptor (PSR) (Box 3), brain-specific angiogenesis inhibitor 1 (BAI1), stabilin 1 and 2, CD36, lectin-like oxidized low-density lipoprotein receptor (LOX-1), and receptor for advanced glycosylation end products (RAGE), have been proposed to be PtdSer receptors or molecules that promote apoptotic cell engulfment [70–73]. In particular, BAI1 and stabilin 2 were shown to signal via Rac1 when engulfing apoptotic cells [70,74]. PtdSer is exposed on the surface of enveloped viruses such as retrovirus, dengue virus, and influenza virus. MFG-E8 and Gas6 were shown to promote the binding of enveloped virus to the integrins- and TAM receptor-expressing cells, respectively [75,76]. Enveloped virus also bound to Tim1-, Tim4-, or CD300-expressing cells, supporting these molecules as PtdSer receptors. By contrast, BAI1-, stabilin 1/2-, or RAGE-expressing cells could not trap PtdSerexposing enveloped virus [75,77]. It is unclear how proteins that recognize PtdSer on apoptotic cells are not able to bind enveloped virus [78]. 

Box 3. PSR: PtdSer Receptor

In 2000, Fadok et al. [107] established a monoclonal antibody (mAb 217) that bound human and mouse macrophages and inhibited apoptotic cell engulfment. A PtdSer-binding membrane protein, designated PSR (PtdSer receptor), was identified by screening a phage peptide display library using mAb 217. PSR was thought to be a type II membrane protein that specifically bound PtdSer and enhanced the PtdSer-dependent apoptotic cell engulfment. PSR knockout mice were established but were embryonically lethal; this was attributed to a failure in apoptotic cell engulfment [108]. Soon afterwards, mouse PSR was reported to localize to the nucleus and to have a Jumonji domain, which is frequently found in transcription factors [109]. In an independent study [110], a PSR-null mouse line was established, which revealed that PSR-null fetal liver macrophages expressed a protein recognized by mAb 217 (the mAb originally used to identify PSR), and had the full ability to engulf apoptotic cells. The lethality of the PSR-null mice was attributed to developmental defects. While working on a transcription system, a later study [111] found that PSR is identical to JMJD6, which has histone arginine demethylase activity and functions as a chromatin remodeling factor. Recently, it was reported that PSR in C. elegans binds PtdSer and remodels chromatin [112]. However, PSR has a low affinity for PtdSer (a Kd of 200–500 nM) and low phospholipid-binding specificity. Taken with the negligible effect of the PSR-null mutation on apoptotic cell engulfment, the possibility that PSR functions as a PtdSer receptor is low [94].

The Two-Step Engulfment of an Apoptotic Cell

The expression‎ of Tim4, a PtdSer receptor, confers a PtdSer-dependent phagocytic ability on fibroblasts such as NIH3T3 cells or mouse embryonic fibroblasts (MEFs) [61]. However, the cytoplasmic region of Tim4 is both short (only 42 amino acids) and dispensable for apoptotic cell engulfment [79], suggesting that engulfment itself may be mediated by molecules other than Tim4. This finding is in agreement with the proposed two-step model for phagocytosis of apoptotic cells [80], in which phagocytes initially tether the apoptotic cell to their surface, followed by tickling to trigger PtdSer-mediated engulfment (Figure 2). Reconstitution of the engulfment system with Ba/F3 cells, a mouse B cell line that grows in suspension [81], revealed that Tim4-transformed Ba/F3 cells bound strongly to apoptotic cells via PtdSer/Tim4 interactions, but failed to engulf them. By contrast, Ba/F3 cells expressing integrin-/vb3 or MerTK did not bind or engulf apoptotic cells in the presence of MFG-E8 or protein S, respectively. However, Ba/F3 cells that coexpressed Tim4 with integrin-/vb3 or MerTK exhibited strong phagocytic activity in the presence of MFG-E8 or protein S [59,81], indicating that efficient apoptotic cell engulfment requires both the tethering (Tim4) and tickling (MFG-E8/integrin-/vb3) steps. Mouseresident peritoneal macrophages were found to express Tim4 and MerTK and to require both molecules to efficiently engulf apoptotic cells. Tim4-positive and MerTK-negative, but not Tim4-negative and MerTK-positive, macrophages bound apoptotic cells [59], suggesting that Tim4 and MerTK are required for tethering and tickling, respectively, in resident macrophages. Thus, the two-step model with tethering and tickling steps is suited for the engulfment of apoptotic cells in resident peritoneal macrophages. However, this does not discount the possibility that Tim4 is involved in both tethering and tickling steps in other types of macrophages [82].

In contrast to the resident peritoneal macrophages, bone marrow-derived macrophages (BMDMs) do not express Tim4. In these macrophages, TAM receptors together with protein S or Gas6 seem to mediate both tethering and tickling processes [83]. What causes the different requirement of the tethering step between resident peritoneal macrophages and BMDMs is not clear. However, the isolation and timing of the assays may account for some of the differences. Indeed, resident peritoneal macrophages were used immediately after isolation from the peritoneal cavity for the engulfment assay, while BMDMs were cultured for 6–9 days in vitro before they were used for the same assay. Whether resident peritoneal macrophages and BMDMs have a similar ability to engulf apoptotic cells or not should be determined. 

‘Eat Me’ and ‘Don’t Eat Me’ Signal

The molecules involved in tethering and tickling recognize PtdSer as an ‘eat me’ signal. Is this recognition alone sufficient to trigger engulfment? Cells lacking the flippase due to a null mutation of CDC50A constitutively expose PtdSer without undergoing apoptosis [15]. Similarly, live cells expressing the constitutively active form of TMEM16F scramblase expose PtdSer [34,84]. When cultured with thioglycollate-elicited mouse peritoneal macrophages in methylcellulose, cells lacking CDC50A were engulfed, but cells expressing constitutively active TMEM16F were not [15]. We hypothesize that if flippase is inactive due to the lack of CDC50A, PtdSer should be symmetrically distributed between the inner and outer leaflets and not move between the two layers (Figure 3). In this case, PtdSer-binding proteins such as Tim4 and protein S would bind PtdSer tightly on the surface, resulting in cell engulfment. However, in cells expressing the constitutively active TMEM16F, the flippase remains active, so PtdSer can move rapidly between the outer and inner leaflets, reducing the affinity of the PtdSer-binding proteins for PtdSer on the surface. In any event, flippase-deficient cells that expose PtdSer are engulfed by macrophages in a PtdSer-dependent manner, indicating that PtdSer exposed on the cell surface is sufficient as an ‘eat me’ signal. Whether other proposed ‘eat me’ candidates, such as calreticulin or ICAM3, further enhance engulfment remains to be studied [2].

Several molecules have been proposed to act as ‘don’t eat me’ signals that inhibit engulfment [85–87]. Among these molecules, CD47 is thought to be a strong candidate because CD47- deficient red blood cells injected into mice are rapidly cleared from the bloodstream by splenic macrophages [87]. The receptor for CD47 is SIRPa, a type I membrane protein carrying two ITIM motifs, expressed in macrophages. The interaction of CD47-expressing cells with SIRP/- expressing macrophages is proposed to block the ability of macrophages to engulf the apoptotic cells, thus functioning as a ‘don’t eat me’ signal. Since human tumor cells treated with anti-CD47 mAb are killed through engulfment by macrophages [88], its clinical application for cancer patients is considered. By contrast, mouse thymocytes express a high level of CD47, which is not downregulated during apoptosis. Yet, apoptotic thymocytes are efficiently engulfed by macrophages [89,90]. Furthermore, overexpressing CD47 in mouse lymphoma cells does not affect their PtdSer-dependent engulfment by thioglycollate-elicited peritoneal macrophages [89]. Thus, the PtdSer ‘eat me’ signal appears to overcome the CD47 ‘don’t eat me’ signal, at least for the engulfment of apoptotic cells. Alternatively, the role of CD47 as a ‘don’t eat me’ signal may also be limited [91]. 

Concluding Remarks

The evolutionarily conserved apoptotic cell clearance by phagocytes requires collaboration between apoptotic cells and phagocytes. Apoptotic cells expose PtdSer as an ‘eat me’ signal that is recognized by phagocytes, which engulf the dead cells. How PtdSer is externalized to the cell surface and is recognized by phagocytes had long been a mystery [92–94]. The molecular identity of flippase, which maintains the asymmetrical distribution of phospholipids at plasma membranes, was also elusive in the mammalian system. Recent identification of the molecules that support the scrambling and flipping of phospholipids at the plasma membrane revealed an elegant, unexpectedly simple system that exposes PtdSer during apoptosis [95]. The molecular organization of flippase and scramblase, including subunit structures and active sites, can now be investigated, along with strategies of these molecules of translocating phospholipids between the inner and outer leaflets (see Outstanding Questions).

The identity of some of the PtdSer receptors and PtdSer-binding molecules, along with their role in engulfing apoptotic cells have been confirmed in macrophages lacking these molecules [27]. Next, the signal transduction that mediates apoptotic cell engulfment should be examined. Although the mechanism that exposes PtdSer seems to be conserved from C. elegans to mammals (Xkr8/CED8), PtdSer recognition systems such as the PtdSer receptor (Tim4) and PtdSer-binding proteins (MFG-E8 and protein S) do not appear to be conserved. In mammals, professional phagocytes engulf apoptotic cells, while nonprofessional phagocytes engulf apoptotic cells in C. elegans. Thus, the signal transduction system for apoptotic cell engulfment in mammals may differ from that revealed in C. elegans. Apoptotic cells that are not properly engulfed and cleared are thought to undergo secondary necrosis, releasing intracellular materials from the ruptured plasma membrane and activating the immune system [27,96]. This happens when apoptotic cells cannot expose the ‘eat me’ signal or macrophages cannot recognize it. Mice deficient for the PtdSer-binding protein MFG-E8 develop an SLE-type autoimmune disease – at least, mice from a 129/B6 mixed background [51]. Therefore, it would be interesting to examine whether mice lacking scramblase have a similar phenotype, and whether human autoimmune patients have defects in the molecules involved in PtdSer exposure or recognition.

Acknowledgments

We thank all of the members of our laboratory for discussions and M. Fujii for secretarial assistance. Experiments in our laboratory were supported in part by grants-in-aid from the Ministry of Education, Science, Sports, and Culture in Japan.

References

1. Fadok, V.A. et al. (1992) Exposure of phosphatidylserine on the surface of apoptotic lymphocytes triggers specific recognition and removal by macrophages. J. Immunol. 148, 2207–2216 2. Gardai, S.J. et al. (2006) Recognition ligands on apoptotic cells: a perspective. J. Leuk. Biol. 79, 896–903 3. Leventis, P.A. and Grinstein, S. (2010) The distribution and function of phosphatidylserine in cellular membranes. Annu. Rev. Biophys. 39, 407–427 4. Tanaka, Y. and Schroit, A.J. (1983) Insertion of fluorescent phosphatidylserine into the plasma membrane of red blood cells. Recognition by autologous macrophages. J. Biol. Chem. 258, 11335–11343 5. Krahling, S. et al. (1999) Exposure of phosphatidylserine is a general feature in the phagocytosis of apoptotic lymphocytes by macrophages. Cell Death Differ. 6, 183–189 6. Asano, K. et al. (2004) Masking of phosphatidylserine inhibits apoptotic cell engulfment and induces autoantibody production in mice. J. Exp. Med. 200, 459–467 7. van den Eijnde, S.M. et al. (1998) Cell surface exposure of phosphatidylserine during apoptosis is phylogenetically conserved. Apoptosis 3, 9–16 8. Venegas, V. and Zhou, Z. (2007) Two alternative mechanisms that regulate the presentation of apoptotic cell engulfment signal in Caenorhabditis elegans. Mol. Biol. Cell 18, 3180–3192 9. Palmgren, M.G. and Nissen, P. (2011) P-type ATPases. Annu. Rev. Biophys. 40, 243–266 10. Lenoir, G. et al. (2009) Cdc50p plays a vital role in the ATPase reaction cycle of the putative aminophospholipid transporter Drs2p. J. Biol. Chem. 284, 17956–17967 11. Zachowski, A. et al. (1989) Control of transmembrane lipid asymmetry in chromaffin granules by an ATP-dependent protein. Nature 340, 75–76 12. Tang, X. et al. (1996) A subfamily of P-type ATPases with aminophospholipid transporting activity. Science 272, 1495–1497 13. Hua, Z. et al. (2002) An essential subfamily of Drs2p-related Ptype ATPases is required for protein trafficking between Golgi complex and endosomal/vacuolar system. Mol. Biol. Cell 13, 3162–3177 14. Pomorski, T. et al. (2003) Drs2p-related P-type ATPases Dnf1p and Dnf2p are required for phospholipid translocation across the yeast plasma membrane and serve a role in endocytosis. Mol. Biol. Cell 14, 1240–1254 15. Segawa, K. et al. (2014) Caspase-mediated cleavage of phospholipid flippase for apoptotic phosphatidylserine exposure. Science 344, 1164–1168 16. Yabas, M. et al. (2011) ATP11C is critical for the internalization of phosphatidylserine and differentiation of B lymphocytes. Nat. Immunol. 12, 441–449 1