Human Type IV P-type ATPases That Work as Plasma Membrane Phospholipid Flippases and Their Regulation by Caspase and Calcium*

In plasma membranes, flippases translocate aminophospholipids such as phosphatidylserine and phosphatidylethanolamine from the extracellular to the cytoplasmic leaflet. Mammalian ATP11C, a type IV P-type ATPase, acts as a flippase at the plasma membrane. Here, by expressing 12 human type IV P-type ATPases in ATP11C-deficient cells, we determined that ATP8A2 and ATP11A can also act as plasma membrane flippases. As with ATP11C, ATP8A2 and ATP11A localized to the plasma membrane in a CDC50A-dependent manner. ATP11A was cleaved by caspases during apoptosis, and a caspase-resistant ATP11A blocked apoptotic PtdSer exposure. In contrast, ATP8A2 was not cleaved by caspase, and cells expressing ATP8A2 did not expose PtdSer during apoptosis. Similarly, high Ca2+ concentrations inhibited the ATP11A and ATP11C PtdSer flippase activity, but ATP8A2 flippase activity was relatively resistant to Ca2+. ATP11A and ATP11C were ubiquitously expressed in human and mouse adult tissues. In contrast, ATP8A2 was expressed in specific tissues, such as the brain and testis. Thus, ATP8A2 may play a specific role in translocating PtdSer in these tissues.

The plasma membrane of eukaryotic cells is composed of an inner and an outer leaflet. Phospholipids, a major component of plasma membranes, are asymmetrically distributed between the leaflets (1); aminophospholipids such as phosphatidylserine (PtdSer) 2 and phosphatidylethanolamine (PtdEtn) are confined to the inner leaflet, whereas the outer leaflet is rich in phosphatidylcholine (PtdCho) and sphingomyelin (SM) (1,2). This asymmetrical distribution of phospholipids is disrupted in various biological processes. Cells undergoing apoptosis expose PtdSer on the cell surface as an "eat me" signal that is recognized by phagocytes, which respond by engulfing the apoptotic cell (3,4). Activated platelets expose PtdSer as a scaffold for blood-clotting factors (5). Three types of membrane proteins regulate the distribution of phospholipids in the plasma membrane: aminophospholipid translocases (flippases) specifically transport PtdSer and PtdEtn from the outer to the inner leaflet in an ATP-dependent manner (1,2); floppases transport phospholipids (PtdCho in particular) from the inner to the outer leaflet in an ATP-dependent manner, and phospholipid scramblases nonspecifically scramble phospholipids between the inner and outer leaflets in an energy-independent manner (1).
Scramblase systems can be Ca 2ϩ -activated or caspase-activated (6,7). In the transmembrane protein (TMEM) 16 family, which has 10 transmembrane regions (8), five of the 10 human family members (TMEM16C, D, F, G, and J) elicit Ca 2ϩ -activated phospholipid scrambling (7,9). Caspase-activated scrambling is executed by members of the Xkr (Xk-related protein) family. The human Xkr family consists of nine membrane proteins having six transmembrane segments; three of these proteins (Xkr4, Xkr8, and Xkr9) scramble phospholipids in apoptotic cells (6,10). These Xkr members carry a caspase recognition site in the C-terminal tail, and this site must be cleaved by caspase to support phospholipid scrambling (6,10).
Several Type IV P-type ATPases (P4-ATPases) have been shown to be phospholipid flippases (11)(12)(13)(14), but it remains unclear which human P4-ATPases function as flippases at the plasma membrane. Recently, we identified ATP11C, a member of the P4-ATPase family, as a flippase at the plasma membrane (15). Like other P4-ATPases (16 -18), ATP11C requires chaperoning by CDC50A to localize to the plasma membrane. CDC50A-deficient cells almost completely lose the ability to translocate PtdSer from the outer to the inner leaflet and constitutively expose PtdSer. Although ATP11C-deficient cells lose most of their flippase activity, they retain sufficient flippase ability to maintain an asymmetrical phospholipid distribution and do not expose PtdSer (15). These results suggested that other P4-ATPases might have compensated for the lack of ATP11C in these cells. In fact, ATP8A2, ATP8B1, and ATP11A were proposed to have PtdSer flippase activity at plasma membranes (11,19,20). However, how they are regulated in various biological processes and whether other members of the P4-ATPase family have the plasma membrane flippase activity is not clear.
The human P4-ATPase family has 14 members (16,17). In this study, we expressed 12 human P4-ATPases in an ATP11Cdeficient cell line and found that only ATP8A2, ATP11A, and ATP11C had a strong ability to flip PtdSer at the plasma membrane. These proteins localized to the plasma membrane in the presence of CDC50A. Similar to ATP11C, ATP11A carried two caspase recognition sites in its cytoplasmic region that were cleaved during apoptosis. In contrast, ATP8A2 was not cleaved or inactivated during apoptosis, and cells expressing ATP8A2 did not expose PtdSer during apoptosis. Although high Ca 2ϩ concentrations inhibited ATP11A and ATP11C ATPase activity, ATP8A2 was relatively resistant to Ca 2ϩ -mediated inhibition. ATP11A and ATP11C were ubiquitously expressed in various human and mouse tissues, whereas ATP8A2 was specifically expressed in the brain and testis. These results suggest that in most mammalian cells, the major flippases at the plasma membrane are ATP11A and ATP11C. ATP8A2 may have a specific role in flipping phospholipids in the brain and testis.

Experimental Procedures
Mice, Cell Lines, Plasmids, Recombinant Proteins, Antibodies, and Reagents-C57BL/6J mice were purchased from Japan Clea. All mouse studies were approved by the Animal Care and Use Committee of the Research Institute for Microbial Diseases and the Immunology Frontier Research Center of Osaka University. W3 cells were transformed from mouse T-cell lymphoma (WR19L) to express mouse Fas (21). W3-Ildm cells, derived from WR19L cells, expressed Fas and a caspase-resistant form of the inhibitor of caspase-activated DNase (22). W3 and W3-derived cells were cultured in RPMI 1640 containing 10% FCS. ATP11C ED22 is an ATP11C-null W3 cell line; CDC50A ED29 is a CDC50A-null W3-Ildm cell line (15). HEK293T cells were cultured in DMEM containing 10% FCS. The pMXs-puro retrovirus vector and pGag-pol-IRES-bsr packaging plasmid (23) were provided by Dr. T. Kitamura (Institute of Medical Science, University of Tokyo, Tokyo, Japan). The pCMV-VSV-G plasmid was provided by Dr. H. Miyoshi (Riken Bioresource Center, Riken). We purchased pAdVAntage from Thermo Fisher.
Transformation of Mouse Cell Lines-W3 cells and their derivatives were transformed using a retroviral vector carrying VSV-G envelope genes. The cDNAs were inserted into the pMXs-puro vector and introduced into HEK293T cells using FuGENE 6 (Promega) together with pGag-pol-IRES-bsr, pCMV-VSV-G, and pAdVAntage. The resultant retrovirus was concentrated by centrifugation at 6,000 ϫ g for 16 h at 4°C and used to infect cells; the infected cells were cultured in the presence of 1 g/ml puromycin. If necessary, GFP-positive cells were sorted using a FACSAria II (BD Biosciences). To examine the subcellular localization of human P4-ATPases, CDC50A ED29 cells or CDC50A ED29 transformants expressing hCDC50A were infected by a retrovirus carrying GFP-tagged P4-ATPase, and the GFP-positive transformants were sorted using a FACSAria II. The sorted cells (2 ϫ 10 5 ) were suspended in phenol red-free DMEM containing 10% FCS and 12.5 M DRAQ5, plated on a glass-bottomed dish (Iwaki, AGC Techno Glass), incubated for 7 min at 37°C, and observed by a confocal microscope (FV1000-D; Olympus).
FasL-induced Apoptosis, and Detection of Phosphatidylserine-Cells were treated with 33 units/ml FasL at 37°C to induce apoptosis. To detect exposed PtdSer, the cells were incubated at 25°C for 5 min or 4°C for 15 min with Cy5annexin V diluted 1:2,000 in 10 mM HEPES-KOH buffer (pH 7.4) containing 140 mM NaCl and 2.5 mM CaCl 2 . The cells were then incubated with 200 nM Sytox Blue and analyzed using a FACSAria II or FACSCanto II.
To treat with caspase, the affinity-purified ATP11A or its mutants (100 ng) were incubated at 20°C for 1 h with 200 ng of caspase 3 in 10 l of 40 mM MES/Tris buffer (pH 7.0) containing 5% glycerol, 7 mM DTT, 150 mM NaCl, 0.05% LMNG. After caspase 3-treatment, aliquots were used for the ATPase assay or separated with SDS-PAGE on a 7.5% gel, visualized by SYPRO Ruby protein gel stain (Thermo Fisher) and analyzed with an ImageQuant LAS4000 Biomolecular Imager (GE Healthcare).
NBD Lipid Internalization Assay-NBD lipid internalization was assayed essentially as described previously (15,32). In brief, 7 ϫ 10 5 cells were incubated with 1 M NBD-PS, NBD-PE, NBD-PC, or NBD-SM at 20°C in 200 l of Hanks' balanced salt solution (HBSS) containing 1 mM MgCl 2 and 2 mM CaCl 2 . The cells were collected by centrifugation, resuspended in HBSS containing 5 mg/ml fatty acid-free BSA to extract nonincorporated lipids, and analyzed by a FACSAria II or FACSCanto II. In some cases, 2 ϫ 10 5 cells in 200 l of HBSS containing 1 mM CaCl 2 and1 M NBD-PS were stimulated at 20°C for 4 min with A23187 Ca 2ϩ ionophore and analyzed as above. To determine the apoptotic scramblase activity, 5 ϫ 10 5 cells were treated with FasL, incubated at 25°C for 3 min with 1 M NBD-PC, suspended in HBSS containing 5 mg/ml BSA, and analyzed by a FACSCanto II.
Real Time RT-PCR-Total RNA prepared from various mouse tissues using an Isogen RNA isolation kit (Nippon Gene) and the RNeasy mini kit (Qiagen) was reverse-transcribed with a high capacity RNA to cDNA kit (Thermo Fisher). Human cDNA panels for various tissues (human MTC panel I and II) were purchased from Takara Bio and used as templates for real time RT-PCR. The cDNA was amplified using LightCycler 480 SYBR Green I Master (Roche Diagnostics). Primers for real time RT-PCR were as previously described (15).

Results
Plasma Membrane Flippase Activity of P4-ATPases-The phospholipid flippase activity is severely reduced in ATP11Cnull W3 cells (ATP11C ED22 ) (15). To examine the flippase activity of other human P4-ATPases, each P4-ATPase was Flag-tagged and introduced, along with Flag-tagged human CDC50A, into ATP11C ED22 cells using retrovirus-based expression vectors. Stable transformants were analyzed by Western blotting with an anti-Flag mAb. Because the native cDNA sequences of ATP8A2, ATP8B1, ATP8B3, ATP8B4, ATP9B, ATP10A, ATP10B, ATP10D, ATP11B, and ATP11C produced only low levels of these proteins in ATP11C ED22 cells, their nucleotide sequences were edited to optimize translational efficiency by considering mRNA stability, codon usage, and secondary structures. As a result, the transformants expressed comparable protein levels for all of the family members except ATP8B3 and ATP10A (Fig. 1A). Because ATP8B3 is exclusively expressed in the testis (33,34), it might require the testis-specific chaperone CDC50C for stable expression. However, ATP10A is expressed in various mouse tissues, so it is not clear why its expression was low in ATP11C ED22 cells.
We examined the flippase activity of 12 ATPases (ATP8A1, 8A2, 8B1, 8B2, 8B4, 9A, 9B, 10B, 10D, 11A, 11B, and 11C) by incubating the ATP11C ED22 transformants with a fluorescence-conjugated phospholipid (NBD-PS, -PE, -PC, or -SM). As shown in Fig. 1B, the ATP11C-deficient ATP11C ED22 cells slowly incorporated NBD-PS; its incorporation was strongly enhanced in cells transformed with ATP8A2, ATP11A, or ATP11C. In these cells, the incorporation reached its maximum within 6 min. In Table 1, the rate of flippase activity (measured by the incorporated NBD-PS at 2 min) was compared among transformants expressing the various ATPases. Considering the protein expression levels (Fig. 1A), ATP8A2, 11A, and 11C appeared to have similar PtdSer flipping activities. ATP8A1 weakly enhanced the NBD-PS incorporation, but little or no enhancement was observed with the other P4-ATPases. ATP11C ED22 cells significantly incorporated NBD-PE, and the rate of incorporation was enhanced 2.4 -2.7-fold by the cell transformation with ATP11A or ATP11C but not with other P4-ATPases ( Fig. 1B and Table 1). On the other hand, none of the P4-ATPases examined, includ-ing ATP8A2, ATP11A, and ATP11C, enhanced the incorporation of NBD-PC or NBD-SM (Table 1), confirming that they were aminophospholipid-specific.
ATPase Activity of the Three Flippases-The ATPase activity of P4-ATPases is known to be activated by phospholipids (12,30). To examine the ATPase activity of human ATP8A2, 11A, and 11C, the proteins were Flag-tagged, produced in 293T cells together with human CDC50A, and affinity-purified. The purified ATPases (ATP8A2, 11A, and 11C) were of the expected size on SDS-PAGE and were associated with proteins of ϳ78 and 72 kDa ( Fig. 2A), which were probably differently glycosylated CDC50As fused to GFP. The ATPase activity of ATP8A2, 11A, and 11C was strongly activated by POPS (Fig. 2B). A Michael-Menten kinetics analysis indicated that the K m and V max values of the three ATPases for POPS: 2.1 M and 13 nmol Pi/g protein/min, 1.9 M and 6.5 nmol Pi/g protein/min, and 1.3 M and 8.9 nmol Pi/g protein/min for ATP8A2, 11A, and 11C, respectively. POPE also activated these ATPases, but the K m for POPE was 3-15 times higher than that for POPS and differed significantly among the three ATPases (29.2 M (V max : 9.6 nmol Pi/g protein/min), 6.1 M (V max : 4.9 nmol Pi/g protein/min), and 11.0 M (V max : 6.7 nmol Pi/g protein/min) for ATP8A2, 11A, and 11C, respectively). POPC and sphingomyelin had little if any effect on the ATPase activity of ATP8A2, 11A, or 11C (Fig. 2C). As reported for the membrane ATPases (35), the POPS-stimulated ATPase activity of ATP8A2, 11A, and 11C was inhibited by orthovanadate (Fig. 2D). The significant POPE-induced ATPase activity of ATP8A2 was inconsistent with its apparently low PtdEtn-flippase activity ( Table 1), but the latter finding might have been due to the high background incorporation of NBD-PE into ATP11C ED22 cells (Fig. 1).
Subcellular Localization of P4-ATPases-Little or no evidence of flippase activity was found in the remaining eight human P4-ATPases (ATP8B1, 8B2, 8B4, 9A, 9B, 10B, 10D, and 11B). It is possible that these ATPases had low flippase activity because they did not localize to the plasma membrane. To examine this possibility, we tagged the 12 human P4-ATPases with GFP and introduced them into CDC50A-null CDC50A ED29 cells or CDC50A-transformed CDC50A ED29 cells to establish stable transformants expressing each ATPase. As expected from their flippase activity, ATP8A2, 11A, and 11C localized to the plasma membrane in the presence of CDC50A    (Fig. 3). In the absence of CDC50A, these proteins were found intracellularly, indicating that CDC50A was required for their localization at the plasma membrane. ATP8B1, 8B2, 8B4, and 10D also localized to the plasma membrane in a CDC50A-dependent manner. ATP8A1 and 11B were found mainly in intracellular vesicles and partly at the plasma membrane in the presence of CDC50A. As previously reported (36), ATP9A and ATP9B localized predominantly to intracellular vesicles, and the presence of CDC50A did not change their localization. On the other hand, the localization of ATP10B was dramatically changed by the presence of CDC50A; ATP10B-GFP was diffusely present in intracellular vesicles in the absence of CDC50A but appeared as a large patch in the presence of CDC50A (Fig. 3A). The patch-like ATP10B-GFP structure colocalized with LysoTracker (Fig. 3B), indicating that ATP10B localized to the late endosome or lysosome, as previously reported (37). Caspase Cleavage of ATP11A-Human ATP11C carries three caspase recognition sites in its nucleotide-binding domain and is cleaved by caspases for apoptotic PtdSer exposure (15). To examine whether other PtdSer-flippases are similarly cleaved during apoptosis, ATP11C ED22 transformants expressing GFP-tagged ATP8A2, 11A, or 11C were treated with FasL for 1 h and analyzed by Western blotting with an anti-GFP mAb. In FasL-treated cells, ATP11A was cleaved from 140 to 75 kDa, but ATP8A2 remained intact (Fig. 4A). A caspase inhibitor (Q-VD-OPh) blocked the cleavage of ATP11A in the FasLtreated cells, indicating that the cleavage was caspase-dependent (Fig. 4B).

Phospholipid Flippase Activity of Human P4-ATPases
We searched for caspase recognition sites using Cascleave (59) and identified two putative phylogenetically well conserved caspase 3 recognition sites in human ATP11A at positions similar to those in human ATP11C (Fig. 4C). To determine whether ATP11A is cleaved at these sites during apoptosis, the two caspase recognition sites (site 1, DMIDS; site 2, DSVDG) were mutated singly to AMIAS (Mut1) and ASVAG (Mut2), fused to GFP, and expressed in ATP11C ED22 cells. ATP11A with a mutation at only one caspase recognition site (Mut1 or Mut2) could be cleaved during FasL-induced apoptosis, whereas ATP11A with mutations in both caspase recognition sites (Mut1 ϩ 2) resisted cleavage (Fig. 4D). This was confirmed with the purified proteins. That is, ATP11A and its caspase-resistant mutants were Flag-tagged, expressed in 293T cells together with GFP-tagged CDC50A, and purified with anti-Flag antibody. As shown in Fig. 4E, incubating the purified proteins with caspase 3 confirmed that caspase 3 could directly cleave ATP11A at these two recognition sites.
Caspase Regulation of Flippase-We previously found that cells expressing a caspase-resistant ATP11C do not expose PtdSer during apoptosis (15). To examine whether this is also true of ATP11A, ATP11C ED22 cells were transformed with wild-type (WT) or mutant (Mut1, Mut2, or Mut1 ϩ 2) ATP11A. All of the transformants expressing a single-or double-mutant ATP11A had flippase activity comparable with that of WT ATP11A (Fig. 5A); thus, the mutations did not affect flippase activity. When these transformants were treated with FasL, cells expressing WT ATP11A exposed PtdSer as efficiently as cells lacking ATP11C. In contrast, FasL treatment could not induce PtdSer exposure in cells expressing Mut1, Mut2, or Mut1 ϩ 2 ATP11A (Fig. 5B). The expression of ATP11A mutants did not affect the FasL-induced caspase 3 activation or phospholipid scrambling, as determined by NBD-PC incorporation assays (Fig. 5C). These results confirmed that ATP11A, like ATP11C, must be inactivated by caspase for efficient PtdSer exposure. The ability of Mut1 and Mut2 to inhibit the PtdSer-exposure suggested that cleavage at one caspase recognition site did not abolish the flippase activity of ATP11A. To verify this, the ATPase activity of caspase 3-treated recombinant WT, Mut 1, Mut 2, and Mut1 ϩ 2 ATP11A (Fig. 4E) was measured. As shown in Fig. 5D, caspase 3 treatment abolished the PtdSer-dependent ATPase activity of WT AT11A, but it had little effect on the ATPase activity of Mut1, Mut 2, and Mut 1 ϩ 2 ATP11A, supporting that cleavage at one position of ATP11A is not sufficient to inactivate the flippase. Because ATP8A2 had no caspase recognition site and was not cleaved during apoptosis (Fig. 4A), it is possible that WT ATP8A2 would inhibit apoptotic PtdSer exposure. In fact, ATP11C ED22 transformants expressing ATP11C efficiently exposed PtdSer when treated with FasL, but those expressing ATP8A2 did not expose PtdSer (Fig. 5E). These results confirmed that ATP8A2 flips PtdSer, but unlike ATP11A and 11C, ATP8A2 is not inactivated during apoptosis.
Calcium Regulates PtdSer Flippase-Increasing intracellular Ca 2ϩ levels in human erythrocytes inhibits their ability to incorporate spin-labeled aminophospholipids (38), suggesting that Ca 2ϩ can regulate flippase activity. To confirm this finding at a molecular level, ATP11C ED22 transformants expressing ATP8A2, 11A, or 11C were incubated for 4 min with NBD-PS in the presence of various concentrations of A23187, a calcium ionophore that dose-dependently increases intracellular Ca 2ϩ (39). As shown in Fig. 6A, A23187 dose-dependently inhibited the ATP8A2, ATP11A, and ATP11C PtdSer-flippase activity, with a 3-fold difference in the sensitivity of the three flippases: the A23187 IC 50 was 1.54, 0.48, and 0.56 M for ATP8A2, 11A, and 11C, respectively. To confirm the differing sensitivities of ATP8A2, 11A, and 11C to Ca 2ϩ , their PtdSer-activated ATPase activity was measured in the presence of various concentrations of CaCl 2 . As shown in Fig. 6B, CaCl 2 inhibited the PtdSer-stimulated ATPase activity of ATP8A2, 11A, and 11C. Again, sensitivity to Ca 2ϩ differed between the three flippases: the IC 50 was 581, 206, and 130 M for ATP8A2, 11A, and 11C, respectively.
Tissue Distribution of P4-ATPases with PtdSer-Flippase Activity-Because our results indicated that ATP8A2, 11A, and 11C act as flippases at the plasma membrane and that ATP8A2 sensitivity to caspase and Ca 2ϩ differed greatly from that of ATP11A and 11C, we next used real time RT-PCR to analyze the distribution of these flippases in human and mouse tissues. As shown in Fig. 7, ATP11A and 11C were ubiquitously expressed in various tissues but with varying preferences. ATP11C was highly expressed in the liver, whereas ATP11A expression was relatively low; this difference was even more pronounced in the mouse. The levels of ATP11A mRNA in the mouse kidney and heart were 6 -7 times higher than those of ATP11C mRNA. In contrast, the ATP8A2 expression was rather tissue-specific; ATP8A2 mRNA was abundantly expressed in the human and mouse brain and testis, less abundantly in the mouse large intestine and heart, and hardly at all in other adult human and mouse tissues.

Discussion
Early biochemical studies of membranes from human red blood cells (40) and bovine chromaffin granules (41) showed a specific ATP-dependent inward incorporation of aminophospholipids. An ATPase purified from human erythrocyte membranes was shown to flip aminophospholipids in an ATP-dependent manner in a reconstituted liposome system (42).
Cloning of the ATPase cDNA revealed that this ATPase was a member of the large P4-ATPase family (43). Although many groups have since studied the biochemical characteristics of each member of the P4-ATPase family, there is confusion in the literature as to which flippases act at the plasma membrane. We previously showed that most of the PtdSer flippase activity is lost in ATP11C-null human and mouse leukemia cells. In the present study, we expressed 12 human P4-ATPases in ATP11C-null cells and found that ATP11C is not the only P4-ATPase that functions as a plasma membrane flippase; ATP11A and ATP8A2 also localize to the plasma membrane and have PtdSer-flippase activity. Accordingly, the ATPase activity of these flippases was strongly stimulated by PtdSer. The ability of ATP8A2 to flip PtdEtn was low compared with its ability to flip PtdSer. However, the ATPase activity of ATP8A2 was enhanced by high PtdEtn concentrations, suggesting that ATP8A2 may also flip PtdEtn. ATP11A, ATP11C, and ATP8A2 did not flip PtdCho or SM, and their ATPase activity was not stimulated by PtdCho or SM, confirming that they are aminophospholipid translocases.
The flippase activity of ATP11A at plasma membrane agrees with a previous report by Takatsu et al. (19). However, it is controversial where ATP8A2 is localized. One report indicates that ATP8A2 works as a flippase at plasma membranes (20), whereas Lee et al. (13) showed that ATP8A2 localizes to recycling endosomes and works as a flippase for membrane fission. Although we cannot rule out the possibility that ATP8A2 localization depends on cell type, ATP8A2 inhibition of apoptotic PtdSer exposure supports its role as a flippase at the plasma membrane. ATP8B1 is reported to localize to the plasma membrane and to flip PtdSer (11), cardiolipin (44), or PtdCho (19). Although we confirmed ATP8B1 plasma membrane localization, we did not detect any translocase activity for fluorescent-labeled PtdSer or PtdCho. The reason for this difference in findings is not clear. ATP8B1 is specifically expressed in liver, intestinal, and pancreatic cells (45), and it may be necessary to examine its function in these cells. We found three P4-ATPases (ATP8B2, ATP8B4, and ATP10D) that localized to the plasma membrane but did not support the flipping of PtdSer, PtdEtn, PtdCho, or SM. Whether these ATPases are involved in flipping other phospholipids or require different substrates, such as oligosaccharide diphosphate dolichols (46), remains to be studied.
We previously showed that ATP11C is cleaved by caspases and inactivated during apoptosis to expose PtdSer to the cell surface (15); this process is essential for the dead cell to be engulfed. Because ATP11A carries two caspase recognition sites at similar positions to those in ATP11C, we examined whether ATP11A might behave similarly. Apoptosis is a universal process in mammals and occurs in most cells (47). Because the delayed clearance of apoptotic cells is detrimental for animals (4), inactivation of the ubiquitously expressed ATP11A and ATP11C during apoptosis would be reasonable for the prompt engulfment of dead cells. Thus, it was a surprise that ATP8A2 was not cleaved during apoptosis. Ϫ, untreated; ϩ, treated with FasL. In the lower panel, untreated (gray bars) or FasL-treated (black bars) cells were incubated at 25°C for 3 min with 1 M NBD-PC and analyzed by flow cytometry. Phospholipid scrambling activity was expressed as the incorporation of NBD-PC. Experiments were conducted in triplicate, and the average values were plotted with S.E. D, ATPase activity of caspase 3-treated ATP11A and its mutants. Purified ATP11A and indicated mutants were left untreated or treated with caspase 3 at 20°C for 1 h. Using aliquots of the protein (10 ng), their ATPase activity (nmol released phosphate/g protein/min) was measured and is plotted with S.E. (n ϭ 3). C3, recombinant caspase 3. E, apoptotic PtdSer exposure was inhibited in ATP8A2-expressing cells. W3 cells, ATP11C ED22 cells, or ATP11C ED22 transformants expressing Flag-tagged CDC50A with ATP11C or ATP8A2 were left untreated of treated with 33 units/ml of FasL for 2 h at 37°C, stained with Cy5-annexin V and Sytox blue, and analyzed by flow cytometry. MFI, mean fluorescence intensity.
ATP8A2 is expressed in photoreceptor cells in the eyes (12,20). Every day, distal outer segments of these photoreceptor cells are shed, expose PtdSer, and are phagocytosed by adjacent retinal pigment epithelial cells that express PtdSer-recognizing proteins (48,49). How PtdSer is exposed on the shed outer segments is not known. Within the outer segments, ATP8A2 is present in the disk membrane (12), which is derived from the plasma membrane, but later is disconnected from it (50). It is possible that the role of ATP8A2 in maintaining asymmetrical phospholipid distribution is lost in the outer segments, at least in rod photoreceptor cells, to facilitate PtdSer exposure. ATP8A2 is also widely expressed by neurons in the central nervous system and by spermatids in the testis (51). PtdSer is exposed when neurons undergo apoptosis (52) and probably when axons are pruned (53). PtdSer is also exposed when sperm undergoes capacitation (54). It will be interesting to examine how ATP8A2 is regulated to enable PtdSer exposure in these processes. PtdSer is exposed in various activated cells, including platelets (5) and lymphocytes (55,56). When cells are activated, the free Ca 2ϩ concentration is known to locally and transiently increase near or beneath the plasma membrane, reaching to more than 100 M (57, 58). The Ca 2ϩ in these activated cells would activate Ca 2ϩ -dependent phospholipid scramblase and inactivate flippase. This inhibition of flippase by Ca 2ϩ should not be complete, otherwise the activated cells would be engulfed by macrophages. Once the Ca 2ϩ concentrations return to normal levels, scramblases would stop working, and flippase would restart to translocate PtdSer and PtdEtn from outer to inner leaflet of the plasma membrane. Whether Ca 2ϩ directly binds to the flippase or not remains to be studied. PtdSer is exposed on the cell surface or vesicles not only in the engulfment of dead cells and in blood clotting, but also in many other biological processes. Two ubiquitously expressed scramblases, TMEM16F and Xkr8, are activated by Ca 2ϩ and caspase, respectively (6, 7). We found that two similar, ubiqui-  . Tissue distribution of ATP8A2, ATP11A, and ATP11C. Using human tissue cDNA panels (human MTC panels I and II) and mouse cDNA prepared from the indicated tissues of C57BL/6 mice aged 10 -12 weeks, the mRNA levels for ATP8A2, 11A, and 11C were quantified by real time RT-PCR and expressed relative to ␤-actin mRNA.
tously expressed P4-ATPases (ATP11A and ATP11C) are down-regulated by Ca 2ϩ and caspases (15). In addition, we found a caspase-and Ca 2ϩ -resistant flippase that is tissue-specifically expressed. There is a tissue-specific Ca 2ϩ -dependent scramblase (9) and a caspase-dependent scramblase (10). It will be interesting to investigate how these scramblases and flippases interact to expose PtdSer in various biological processes.
Author Contributions-K. S. and S. N. conceived the study. K. S. and S. K. performed the experiments. K. S. and S. N. wrote the manuscript.