Exposure of phosphatidylserine on the cell surface

Abstract

Phosphatidylserine (PtdSer) is a phospholipid that is abundant in eukaryotic plasma membranes. An ATP-dependent enzyme called flippase normally keeps PtdSer inside the cell, but PtdSer is exposed by the action of scramblase on the cell’s surface in biological processes such as apoptosis and platelet activation. Once exposed to the cell surface, PtdSer acts as an ‘eat me’ signal on dead cells, and creates a scaffold for blood-clotting factors on activated platelets. The molecular identities of the flippase and scramblase that work at plasma membranes have long eluded researchers. Indeed, their identity as well as the mechanism of the PtdSer exposure to the cell surface has only recently been revealed. Here, we describe how PtdSer is exposed in apoptotic cells and in activated platelets, and discuss PtdSer exposure in other biological processes.

Facts

• ATP11A and ATP11C, members of the P4-ATPase family, act as flippases at the plasma membrane and are cleaved by caspase during apoptosis.

• Of the TMEM16-family proteins, which carry 10 transmembrane segments, TMEM16F and four other family members function as Ca²⁺-dependent scramblases at the plasma membrane.

• TMEM16F exposes phosphatidylserine (PtdSer) and release microparticles on activated platelets for blood clotting. It is also involved in releasing hydroxyapatite in osteoblasts for bone mineralization.

• XK-related protein 8 (Xkr8) and two other Xkr-family proteins are cleaved during apoptosis and promote apoptotic PtdSer exposure.

Open Questions

• What are the physiological roles of the plasma membrane flippases that are present only in the brain and testis?

• What are the physiological roles of TMEM16-family members expressed only in the brain and intestine? Do their Ca²⁺-dependent scramblase activity at the plasma membrane plays a specific role there?

• What are the physiological roles of Xkr-family members expressed only in the brain and intestine? Do their caspase-dependent scramblase activities play a specific role?

• Is PtdSer exposure in activated lymphocytes, pyrenocytes, aged reticulocytes, capacitated sperm, tumor-associated endothelial cells and enveloped viruses regulated by the P4-type ATPase, TMEM16 and Xkr families?

• How the flippases and scramblases translocate phospholipids between inner and outer leaflets of plasma membranes?

PtdSer is Distributed Asymmetrically in the Plasma Membrane

In eukaryotic cells, phospholipids in the plasma membrane are distributed asymmetrically.^{1, 2} The amine-containing phospholipids PtdSer and phosphatidylethanolamine (PtdEtn) are confined to the cytoplasmic leaflet of the plasma membrane, while phosphatidylcholine (PtdCho) and sphingomyelin (SM) are more concentrated in the exoplasmic leaflet. Phospholipid distribution in the plasma membrane is regulated by three types of phospholipid translocases.³ Flippase (or aminophospholipid translocase) specifically translocates PtdSer and PtdEtn from the outer to the inner leaflet of the lipid bilayer, in an ATP-dependent manner. Floppase is ATP-dependent, and is thought to translocate phospholipids, especially PtdCho, from the inner to outer leaflet. Scramblase non-specifically translocates or scrambles phospholipids between the lipid bilayers in both directions, without consuming ATP.

Plasma Membrane Flippases and their Down-regulation by Ca²⁺ and Caspase

Red blood cells can incorporate aminophospholipids in an ATP-dependent manner,⁴ and this ability was termed ‘flippase’ activity. Subsequently, flippase was abundantly detected in the chromaffin granules of bovine adrenal glands⁵ and was partially purified. Since the purified granule flippase had an ATPase activity that was biochemically similar to ATPase II, a major ATPase in chromaffin granules,⁶ ATPase II was proposed to be the flippase. The molecular cloning of bovine chromaffin flippase/ATPase (ATPase II) cDNA indicated that ATPase II is a type IV P-type ATPase (P4-ATPase),⁷ and ATPase II was thus designated ATP8A1.^{8, 9} The P4-ATPases, which exist only in eukaryotic cells, comprise a large family, with 5, 6, 15 and 14 members in yeast, Caenorhabditis elegans, mice and humans, respectively.^{10, 11, 12} P4-ATPases have 10 transmembrane segments and 2 large cytoplasmic loops that contain nucleotide-binding-site and ATPase domains. P4-ATPases are chaperoned to their proper subcellular location by CDC50, which carries two transmembrane segments with cytoplasmic N- and C-termini.¹³ CDC50A seems to be necessary for the flippase and lipid-transport activity of these ATPases.¹⁴

ATP8A1 and its yeast ortholog Drs2p localize mainly to intracellular vesicles such as granules and trans-Golgi networks,^{5, 15, 16} and a Drs2p deficiency in yeast has no effect on the PtdSer-flippase activity at plasma membranes.¹⁷ Thus, Drs2p, and probably mammalian ATP8A1, appear to be flippases that function in intracellular vesicles. Dnf1p and Dnf2p, two yeast P4-ATPases present in the plasma membrane, have been proposed to be plasma-membrane flippases,¹⁸ but their function to flip PtdSer is controversial.¹⁹ TAT-1, one of five P4-type ATPases found in C. elegans, was reported to function as a plasma-membrane flippase;²⁰ however, it is also intracellularly localized²¹ suggesting that it may act as a flippase at intracellular membranes, too. Thus, the molecular identity of the plasma-membrane PtdSer flippase remained uncertain.

To identify plasma-membrane PtdSer flippases in mammalian cells, we performed a forward genetic screen using KBM7,²² a human myeloid-cell line with a near-haploid karyotype.²³ KBM7 cells were randomly mutagenized with gene-trap retroviruses,²⁴ and cells that could not efficiently incorporate fluorescently labeled PtdSer were collected by repeated cell sorting. We sequenced virus-insertion sites in the sorted population using a next-generation sequencer, and identified ATP11C and CDC50A as flippase candidates (Figure 1). ATP11C, a P4-ATPase, localized to the plasma membrane in a CDC50A-dependent manner, and an ATP11C deficiency severely reduced the PtdSer-flippase activity (by ~80%) at the plasma membranes. On the other hand, CDC50A-deficient cells completely lost the ability to flip PtdSer at the plasma membrane, and constitutively exposed PtdSer on the cell surface.²² These results identified ATP11C as the major plasma-membrane PtdSer flippase, but also indicated that other CDC50A-dependent P4-ATPases may contribute to the flipping of PtdSer at the plasma membrane. In fact, by establishing stable transformants expressing each member of human P4-ATPase family in ATP11C-null cells, not only ATP11C but also ATP11A and ATP8A2 were found to be localized at plasma membranes, and to flip PtdSer and PtdEtn²⁵ (Table 1). ATP11A and ATP11C were ubiquitously expressed in various cells, while ATP8A2 is expressed only in the brain and testis.

Figure 1

Structure of flippases and its cleavage by caspase. The structure of ATP11A/ATP11C and CDC50A is schematically shown. ATP11A and ATP11C carry 10 transmembrane segments. The ATPase domain in the cytoplasm is divided into A, actuator; N, nucleotide-binding; P, phosphorylation domains. CDC50A carrying two transmembrane regions functions as a chaperone for proper localization of ATP11A and ATP11C at plasma membranes. CDC50A forms a complex with ATP11A or ATP11C in the plasma membrane, and may be necessary for the flippase activity. ATP11A and ATP11C contain 2 and 3 caspase-recognition sites (Red) that flank a α-helix (Blue), and are cleaved during apoptosis for the PtdSer exposure. TAT-1 and TAT-3, P4-ATPases in C. elegans also carry putative caspase-recognition sites in the corresponding positions. DTKT, the conserved phosphorylation site, is underlined

Table 1 Human P4-ATPase family

Increasing the intracellular Ca²⁺ in human erythrocytes inhibits their ability to incorporate aminophospholipids,²⁶ suggesting that Ca²⁺ can regulate the flippase activity at plasma membranes. In fact, ATP11A- or ATP11C-mediated PtdSer flipping at the plasma membrane was inhibited by increasing the intracellular Ca²⁺ concentration.²⁵ Their PtdSer-dependent ATPase activity was also inhibited by a high concentration of Ca²⁺, suggesting that Ca²⁺ directly binds to ATP11A and 11C to inhibit their activity.

Apoptosis is mediated by caspases, which cleave >400 cellular substrates to induce cell death.²⁷ Apoptosis is almost universally accompanied by PtdSer exposure, which also requires caspase activation.²⁸ Flippase is inactivated during apoptosis,^{29, 30} suggesting that flippase might be a caspase target. In fact, ATP11A and ATP11C carry two or three evolutionarily well-conserved caspase 3 recognition sites in its large cytoplasmic domains²² (Figure 1). Point mutations in these caspase-recognition sites generate a caspase-resistant flippase. Cells expressing this caspase-resistant flippase fail to expose PtdSer on the cell surface during apoptosis and are not engulfed by macrophages, indicating that the caspase-mediated cleavage of the flippase is essential to expose PtdSer during apoptosis. In this regard, it is noteworthy that the cell transformants expressing ATP8A2 that does not carry a caspase-recognition site fail to expose PtdSer during apoptosis, suggesting that ATP8A2 may have a specific function in the brain and testis.

Like many other processes in the cell-death pathway, apoptotic PtdSer exposure is phylogenetically well-conserved.³¹ Dying cells in C. elegans³² and Drosophila³³ expose PtdSer downstream of caspase. Among the five P4-ATPases found in C. elegans, TAT-1 and TAT-3 have putative caspase-recognition sites in positions corresponding to those of human ATP11A and 11C (Figure 1).

Calcium-Dependent Scramblase

When platelets are activated, PtdSer is exposed on the cell surface and forms a scaffold for coagulation factors.³⁴ This process is regulated by intracellular calcium, suggesting the involvement of one or more calcium-activated phospholipid scramblases.³⁵ Although PLSCR1 (phospholipid scramblase 1) was once reported to be a scramblase,^{36, 37} its molecular properties and the phenotypes of PLSCR-deficient mice and Drosophila ruled PLSCR1 out as a phospholipid scramblase.^{38, 39}

To identify phospholipid scramblases, we characterized calcium-dependent PtdSer exposure using the mouse cell line Ba/F3.⁴⁰ Cells stimulated with the Ca²⁺ ionophore A23187 in the presence of 0.5 mM Ca²⁺ exposed PtdSer to the cell surface within 15 min and underwent necrotic cell death. However, cells treated with A23187 without extracellular Ca²⁺ transiently exposed PtdSer; the PtdSer was internalized again within 12 h. We used this property of transient, Ca²⁺-dependent PtdSer exposure to establish a Ba/F3 subline that strongly exposes PtdSer. Ba/F3 cells were stimulated with A23187 in the absence of Ca²⁺, and a population (1–5%) that strongly exposed PtdSer was collected by flow cytometry and expanded. This process was repeated 19 times, with a decreasing ionophore concentration each time (from 1 μM to 63 nM). This procedure produced Ba/F3-PS19, a Ba/F3 subline that strongly exposes PtdSer at an A23187 concentration (63 nM) too low to induce PtdSer exposure in the parental Ba/F3 cells.

To isolate scramblases, we constructed a cDNA library in a retrovirus-based mammalian expression vector using 2.5-6.0-kb cDNA prepared from the Ba/F3-PS19 cells. The original Ba/F3 cells were infected with the retroviral cDNA library, stimulated with A23187 and sorted by PtdSer exposure. After undergoing this process three times, a significant population of Ba/F3 cells was found to constitutively expose PtdSer. These cells carried transmembrane protein (TMEM) 16 F with a point mutation (Asp to Gly at amino-acid position 409) that appears to have been spontaneously introduced during the sorting procedure of the Ba/F3-P19 cells. Ba/F3 and other mouse cell lines transformed with the mutant TMEM16F constitutively exposed both PtdSer and PtdEtn, and internalized PtdCho and SM. Treating the transformants with an intracellular Ca²⁺ chelator, BAPTA-AM, blocked the PtdSer exposure. These results indicated that TMEM16F supports Ca²⁺-dependent phospholipid scrambling.⁴⁰ A topological analysis of TMEM16F based on its amino-acid sequence suggested that it carries eight transmembrane regions. However, a recent structural analysis revealed that a TMEM16F ortholog from the fungus Nectria haematococca carries 10 transmembrane regions⁴¹ (Figure 2). Structural and chemical cross-linking analyses indicated that TMEM16F exists as a homodimer.^{41, 42}

Figure 2

Structure of TMEM16F. (a) A schematic representation of TMEM16F. The scrambling domain (SCRD) between transmembrane regions IV and V, and the Ca²⁺-binding site comprised of Aspartate and Glutamate in transmembrane regions VI–VIII are shown by red asterisks. (b) A tertiary structure of TMEM16 from N. haematococca. From the study by Brunner et al.⁴¹

TMEM16F is one of the TMEM16-family proteins, also known as anoctamins; this family has 10 members (TMEM16A-H, J and K; or Ano1-10).⁴³ In 2008, three groups independently reported that TMEM16A, which is expressed ubiquitously, functions as a Ca²⁺-activated Cl⁻ channel.^{44, 45, 46} Since TMEM16B, which is specifically expressed in the retina, also acts as a Ca²⁺-dependent Cl⁻ channel, all of the TMEM16-family members were thought to be Cl⁻ channels.^{43, 47, 48} Accordingly, several groups reported that some TMEM16-family members, including TMEM16F, act as ion channels.^{49, 50, 51, 52} The expression of the wild-type TMEM16F makes the cells sensitive to Ca²⁺-ionophore-induced scrambling of phospholipids.⁴⁰ Mouse fetal-thymocyte and embryonic-fibroblast (MEF) cells lacking TMEM16F lose phospholipid-scrambling ability,⁵³ clearly indicating that TMEM16F is indispensable for Ca²⁺-dependent phospholipid scrambling. On the other hand, while TMEM16A and 16B are robust Cl⁻ channels under physiological conditions, we were not able to detect Cl⁻-channel activity for TMEM16F.⁵³ Two groups recently confirmed that the ability of TMEM16F to support phospholipid scrambling, and concluded that TMEM16F’s ion-channel activity is a consequence of phospholipid translocation⁵⁴ or a secondary activity triggered under extreme, not necessarily physiological, conditions.⁵⁵

Meanwhile, we expressed the 10 TMEM16 members in a TMEM16F-deficient cell line and showed that not only TMEM16F, but also TMEM16C, 16D, 16G and 16J can scramble phospholipids at the plasma membrane⁵³ (Table 2). TMEM16E, 16H and 16 K localize to intracellular membranes, where they might also function as phospholipid scramblases. The TMEM16 family is represented in all eukaryotes.^{56, 57} Saccharomyces cerevisiae and Aspergillus fumigatus carry a single TMEM16 ortholog, whose biochemical and physiological functions are unknown. Nematodes carry two genes encoding TMEM16 homologs, ANOH1 and ANOH2; ANOH1 was recently found to be involved in Ca²⁺-dependent PtdSer exposure,⁵⁸ as we will discuss later.

Table 2 Human TMEM16 Family

A mutation analysis with TMEM16A indicated that six Glutamate and Aspartate residues located in transmembrane segments of VI–VIII are involved in direct binding of Ca²⁺.⁵⁹ These amino-acid residues are evolutionally well-conserved, and the structure of TMEM16 from N. haematococca indicates that these Ca²⁺-binding sites are positioned close to the cavity between two subunits.⁴¹ Regarding the scrambling of phospholipids, Yu et al.⁵⁴ compared the amino-acid sequences of TMEM16F paralogues from 24 animals using a program for Type II-divergence,⁶⁰ and showed that a domain of 35 amino acids located between the transmembrane IV and V is sufficient to confer the lipid scrambling activity to TMEM16A. This domain was named as a scrambling domain (SCRD). It would be interesting to study whether any other members of the TMEM16 family have the SCRD or not.

PtdSer Exposure in Activated Platelets

Normal, growing cells have a high intracellular ATP concentration, while the Ca²⁺ concentration is kept low by Ca²⁺-ATPase, a Ca²⁺ pump that removes Ca²⁺ from the cells.⁶¹ Under these conditions, ATP11A and ATP11C actively and specifically translocates PtdSer and PtdEtn from the outer to the inner leaflet of the lipid bilayer, creating their asymmetrical distribution in the plasma membrane. When platelets are activated with thrombin and collagen, the intracellular Ca²⁺ concentration increases locally and transiently,⁶² likely reaching several hundred micromolar near or beneath the plasma membrane.⁶³ This local Ca²⁺ increase would transiently activate TMEM16F’s scramblase function and inactivate ATP11A’s and 11C’s flippase (Figure 3), thereby swiftly exposing PtdSer on limited regions of the cell’s surface. Intracellular Ca²⁺ levels decrease when the cell returns to a resting state, causing TMEM16F to stop scrambling, and ATP11A’s and 11C’s flippase activity to resume, thus re-establishing the asymmetrical distribution of PtdSer in the plasma membrane.

Figure 3

The molecular mechanism for PtdSer exposure in cells with high Ca²⁺-concentration. The flippase comprised of P4-ATPase (ATP11A or ATP11C) and CDC50A, and a Ca²⁺-dependent scramblase (TMEM16F) are schematically shown. In activated platelets, the intracellular Ca²⁺ concentration increases and activates TMEM16F to scramble phospholipids, while it inactivates P4-ATPases and reduces their flipping activity. When the Ca²⁺ concentration returns to normal level, TMEM16F stops scrambling phospholipids, while P4-ATPases resume flipping PtdSer and PtdEtn. Thus, PtdSer is only transiently exposed to the cell surface in this process, and likely depends on the intracellular concentration of ATP and Ca²⁺. The constant flipping of PtdSer prevents the PtdSer-exposing cells to be engulfed by macrophages⁷⁸

Caspase-Dependent Scramblases

In contrast to the Ca²⁺-induced PtdSer exposure, TMEM16F-deficient cells have normal apoptotic PtdSer exposure,⁵³ confirming previous suggestions that two independent scrambling systems exist in human lymphocytes or mouse platelets.^{64, 65} To identify the scramblases responsible for apoptotic PtdSer exposure, we again performed expression cloning using the Ba/F3-PS19 cDNA library. Assuming that scramblases, like channels, should be large membrane proteins, we first screened a library of long cDNAs (>2.5 kb) and obtained the TMEM16F cDNA, encoding 911 amino acids. However, this library did not appear to contain the cDNA for a scramblase for apoptotic PtdSer exposure, so we switched to a library of smaller cDNAs (1–2.5 kb). The library was introduced into Ba/F3 cells, and their transformants were screened for cells that strongly exposed PtdSer when treated with A23187. By repeated sorting, we found a cell population that constitutively exposes PtdSer. These cells carried the cDNA for Xkr8, which consists of 401 amino acids and has 6–10 transmembrane regions (depending on the topography-prediction program) with cytoplasmic N- and C-termini⁶⁶ (Figure 4). Except for Ba/F3 cells, transformation with Xkr8 did not cause constitutive PtdSer exposure in the cell lines we tested (mouse WR19L T-cell leukemia, and human Raji lymphoma and PLB985 leukemia), but strongly enhanced the apoptotic PtdSer exposure. MEF and mouse fetal-thymocyte cell lines lacking Xkr8 lost the ability to scramble phospholipids (to expose PtdSer and PtdEtn and internalize PtdCho and SM) on apoptosis, while the Ca²⁺-dependent phospholipid scrambling was not affected. It was also found that human Raji and PLB985 cells that are defective apoptotic PtdSer exposure^{67, 68} do not express Xkr8 due to hypermethylation in the Xkr8 promoter region. Xkr8 carries a well-conserved caspase 3 recognition site in its C-terminal tail region, and its cleavage by caspases 3/7 is essential for its scramblase activity⁶⁶ (Figure 4).

Figure 4

Structure of Xkr8, and its caspase-cleavage site. (a) The structure of mouse Xkr8, a caspase-dependent phospholipid scramblase is schematically shown. The caspase-recognition site at C-terminal region is indicated. In addition to Xkr8, Xkr4 and Xkr9 also carry the caspase-recognition site at the C-terminal tail region, and work as a caspase-dependent scramblase. (b) The amino-acid sequences of mouse Xkr8 (mXkr8), CED8 of C. elegans and CG32579 of Drosophila are aligned to obtain the maximal homology. The caspase-recognition sites at the C-terminal region of mXkr8⁶⁶ and at the N-terminal region of CED8⁷¹ are highlighted in green. Transmembrane regions are underlined. reproduced from Suzuki et al.⁶⁶ with permission

Xkr8 belongs to the Xkr family, which has nine and eight members in humans and mice, respectively.⁶⁹ To examine whether any other Xkr-family proteins function as apoptotic scramblases, we expressed all of the Xkr-family members in human PLB985 or Xkr8^−/− mouse fetal-thymocyte cells. Our results showed that not only Xkr8, but also Xkr4 and Xkr9 support apoptotic PtdSer exposure⁷⁰ (Table 3). Like Xkr8, Xkr4 and Xkr9 carry a caspase-recognition site in their C-terminal region, and this site is cleaved during apoptosis to activate the scramblase and expose PtdSer. Xkr8 is ubiquitously expressed in various tissues, and is expressed strongly in the testes. Xkr4 is ubiquitously expressed at low levels, but is strongly expressed in the brain and eyes. Xkr9 is strongly expressed in the intestines. Whether Xkr4 and Xkr9 are redundant to Xkr8 or play specific roles in the brain/eyes and intestines remains to be studied. PtdSer is exposed in apoptotic cells not only in mammals, but also in Drosophila and C. elegans.³¹ Accordingly, flies and nematodes carry an Xkr8 orthlog (CG32579 in D. melanogaster, and CED8 in C. elegans)⁶⁶ (Figure 4). CED8 has a caspase (CED3)-recognition site in its N terminus⁷¹ and is indispensable for CED3-dependent PtdSer exposure.⁶⁶ On the other hand, a Drosophila ortholog (CG32579) has no apparent caspase-recognition site, and it is not clear how this molecule is activated or whether it is involved in apoptotic PtdSer exposure.

Table 3 Human Xkr family

PtdSer Exposure on Apoptotic and Other Dying Cells

Although apoptosis has been considered to be responsible for all programmed cell death,⁷² recent studies indicate that cell death can also occur by programmed necrosis⁷³ by necroptosis or pyroptosis, which are induced by tumor necrosis factor (TNF) and bacterial infection, respectively. Cells that undergo apoptosis, necroptosis or pyroptosis expose PtdSer on their surface and are engulfed by phagocytes in a PtdSer-dependent manner.^{74, 75, 76, 77}

In the apoptotic caspase cascade, caspases 3 and 7 in the downstream of the cascade cleave the flippases of ATP11A and ATP11C and inactivate them. At the same time, caspases 3/7 cleave the Xkr8 scramblase, and activate it (Figure 5). Cleavage by caspases, which is irreversible, thus leading to irreversible PtdSer exposure. Living cells that expose PtdSer by the constitutive active forms of TMEM16 F are not engulfed by macrophages,⁷⁸ but cells exposing PtdSer due to the lack of flippase activity are recognized and engulfed by macrophages.²² PtdSer-expressing cells by the constitutive-active scramblase, carry active flippase. They seem to have a low affinity to the PtdSer receptor on macrophages,⁷⁸ probably because the PtdSer is constantly being flipped. In the absence of flippase, the PtdSer remains exposed and does not return to the inner leaflet, increasing the cell’s affinity for the PtdSer receptor. The difference between complete, irreversible flippase inactivation and partial, transient inactivation may explain why PtdSer-exposing apoptotic cells are engulfed by macrophages, but activated platelets are not.⁷⁹ Once established, it takes days to disrupt the asymmetrical distribution of phospholipids because spontaneous translocation across the bilayer is extremely low.^{80, 81} In this regard, a scramblase(s) is essential for quick PtdSer exposure during apoptosis.

Figure 5

The PtdSer exposure in apoptotic cells. A caspase-dependent phospholipid scramblase of Xkr8 and flippase (ATP11A/ATP11C associated with CDC50A) are schematically shown. When cells undergo apoptosis, caspase 3 or caspase 7 in the downstream of the caspase cascade cleaves Xkr8 to activate its scramblase activity, while the same caspases cleave and inactivate ATP11A and ATP11C. This is the irreversible process, and the PtdSer exposed on the cell surface is recognized by macrophages for engulfment

Recently, a scramblase is reported to function in PtdSer exposure in necrotic cell death in nematodes. Dominant mutations in mec-4 (the core subunit of a mechanically gated sodium channel) in touch neurons in C. elegans cause hyper channel activity, which results in excess Ca²⁺ influx and necrotic cell death.^{82, 83} Li et al.⁵⁸ recently showed that necrotic cells in mec-4 nematodes expose PtdSer by the action of ANOH1 (a TMEM16 homolog) and CED7 (an ABCA1-transporter homolog). TNF-induced necroptosis and bacteria-induced pyroptosis are mediated by calcium influx.^{84, 85} It will be interesting to investigate whether TMEM16F or other TMEM16 proteins play a role in exposing PtdSer in necroptosis or pyroptosis.

Defects in Flippases and Scramblases

Mouse WR19L cells that lack CDC50A or that express constitutively active TMEM16F expose PtdSer but grow normally,^{22, 78} suggesting that cell growth may not require an asymmetrical distribution of phospholipids at the plasma membrane. On the other hand, and ATP11C-null mice suffer from B-cell lymphopenia, cholestasis, anemia, dystocia and hepatocellular carcinoma.^{86, 87, 88, 89} In ATP11C-null B cells and erythrocytes, the flippase activity is reduced and the population of erythrocytes that expose PtdSer is slightly increased.^{88, 89} How this small effect on flippase activity in ATP11C^−/− cells leads to the strong phenotypes seen in ATP11C-null mice is unclear. Lymphocytes or erythrocytes might express only the ATP11C flippase at a specific developmental stage; an ATP11C-null mutation may cause the PtdSer exposure in cells at this stage, leading macrophages to recognize and engulf these cells or causing the destabilization and abnormal assembly of plasma-membrane proteins.^{90, 91}

Scott syndrome, a mild autosomal bleeding disorder, is caused by the failure of Ca²⁺-dependent PtdSer exposure in activated platelets.⁹² Patients in two families carrying Scott syndrome had a homozygous null mutation or compound heterozygous mutations in the TMEM16F gene.^{40, 93} Canine Scott syndrome, a naturally occurring bleeding disorder, is found in German shepherd dogs.⁹⁴ As with human patients, the affected dogs lack procoagulant activity in activated platelets and carry loss-of-function mutations in the TMEM16F gene.⁹⁵

Two mouse lines lacking TMEM16F in platelets have been established, by Yang et al.⁵¹ and by us.⁹⁶ Yang et al. established conventional TMEM16F-knockout mice in a 129/B6-mixed background; the phenotype of these mice differs significantly from that of human or canine Scott syndrome. Although the PtdSer exposure in activated platelets depends significantly on TMEM16F, these mice produce tissue-factor-induced thrombin without TMEM16F; however, the tail-bleeding time is greatly increased in these mice. Yang et al. concluded that this phenotype was due to a defect in TMEM16F’s cation-channel activity. On the other hand, we were unable to detect ion-channel activity in TMEM16F, and our platelet-specific deletion of the TMEM16F gene in a C57/B6 background produced mice with phenotypes matching those in human and canine Scott syndrome. That is, the thrombin/collagen-induced PtdSer exposure in platelets is fully impaired in these mice, as is the tissue-factor-induced thrombin activation. However, similar to human and canine Scott syndrome, the tail-bleeding time is not affected in our platelet-specific TMEM16F-null mouse. The reason for the differences in these two TMEM16F-null mouse lines^{51, 96} is not clear.

A defect in Xkr8 causes the loss of PtdSer exposure in apoptotic cells, and the dead cells are not engulfed.⁶⁶ Similarly, mutations in CED8, an Xkr8 ortholog in nematodes, causes the failure of PtdSer exposure in dying cells, and the inefficient or delayed engulfment of dead cells.⁷¹ Masking PtdSer or deleting molecules that recognize PtdSer, such as MFG-E8, Tim4 and TAM receptors, blocks the engulfment of apoptotic cells, and induces the development of systemic lupus erythematosus-type autoimmune diseases.^{97, 98, 99, 100, 101} Thus, it is likely that Xkr8-deficient mice develop similar autoimmune diseases, but this remains to be confirmed.

Perspectives

We have here described one flippase and two scramblase families (Ca²⁺-dependent and caspase-dependent), and have discussed how PtdSer is exposed on activated platelets and apoptotic cells. PtdSer and PtdEtn are exposed not only on activated platelets, but also on galectin-treated neutrophils,¹⁰² IgE-stimulated mast cells,¹⁰³ activated CD4 T cells,¹⁰⁴ capacitated sperm,^{105, 106} aged red blood cells,¹⁰⁷ and cells activated through the P2X₇ ATP receptor.¹⁰⁴ PtdSer exposure on galectin-treated neutrophils,¹⁰⁸ activated mast cells,¹⁰⁹ activated lymphocytes and cells activated via the P2X₇ receptor¹¹⁰ is mediated by (or at least accompanied by) increased intracellular calcium. It is tempting to speculate that PtdSer exposure in these processes involves TMEM16F or other TMEM16-family members (TMEM16C, 16D, 16F, 16G or 16J) with Ca²⁺-dependent scramblase activity.

Pyrenocytes (nucleus with a thin rim of cytoplasm surrounded by a plasma membrane) extruded from erythroblasts,¹¹¹ milk-fat globules from mammary glands,¹¹² exsosomes,¹¹³ enveloped viruses¹¹⁴ and microparticles released from activated platelets⁹⁶ expose PtdSer. Pyrenocytes do not carry mitochondria, and quickly lose their ATP once separated from erythroblasts, causing a rapid Ca²⁺influx.¹¹¹ A similar process may occur for milk-fat globules and enveloped viruses. Whether this loss of ATP and increased intracellular Ca²⁺ is responsible for activating or inactivating specific flippases or scramblases to expose PtdSer in these particles remains to be studied.

Tumor-associated endothelial cells expose PtdSer.^{115, 116} A monoclonal antibody against PtdSer was proposed as a treatment for cancer, to kill tumor-associated endothelial cells via antibody- or complement-mediated cytotoxicity.¹¹⁷ This treatment appears promising.^{118, 119, 120} Recent reports suggest that the PtdSer in the tumor-associated endothelial cells and the PtdSer exposed on tumor cells after treatment with anti-cancer drugs create an immune-suppressive tumor environment; the antibody against PtdSer reverses this effect, creating anti-tumor immunity.¹²¹ It will be interesting to study whether the flippase and scramblases discussed in this review are involved in the PtdSer exposure on tumor-associated endothelial cells.