Report

Caspase-mediated cleavage of phospholipid flippase for apoptotic phosphatidylserine exposure

See allHide authors and affiliations

Science  06 Jun 2014:
Vol. 344, Issue 6188, pp. 1164-1168
DOI: 10.1126/science.1252809

How cells haul down their “eat me” flags

Dead and dying cells expose a membrane lipid called phosphatidylserine (PS) on their cell surface as a sort of “eat me” signal. Segawa et al. identified the membrane enzyme responsible for flipping any PS that inadvertently makes it way from the inner to the outer leaflet of the plasma membrane lipid bilayer. Without the enzyme, macrophages gobbled up healthy cells.

Science, this issue p. 1164

Abstract

Phospholipids are asymmetrically distributed in the plasma membrane. This asymmetrical distribution is disrupted during apoptosis, exposing phosphatidylserine (PtdSer) on the cell surface. Using a haploid genetic screen in human cells, we found that ATP11C (adenosine triphosphatase type 11C) and CDC50A (cell division cycle protein 50A) are required for aminophospholipid translocation from the outer to the inner plasma membrane leaflet; that is, they display flippase activity. ATP11C contained caspase recognition sites, and mutations at these sites generated caspase-resistant ATP11C without affecting its flippase activity. Cells expressing caspase-resistant ATP11C did not expose PtdSer during apoptosis and were not engulfed by macrophages, which suggests that inactivation of the flippase activity is required for apoptotic PtdSer exposure. CDC50A-deficient cells displayed PtdSer on their surface and were engulfed by macrophages, indicating that PtdSer is sufficient as an “eat me” signal.

In eukaryotic cells, phospholipids are asymmetrically distributed between the outer and inner leaflets of the plasma membrane (1). Phosphatidylcholine (PtdCho) and sphingomyelin are located primarily in the outer leaflet, whereas phosphatidylserine (PtdSer) and phosphatidylethanolamine (PtdEtn) are restricted to the cytoplasmic leaflet. Disruption of asymmetrical phospholipid distribution—in particular, PtdSer exposure on the cell surface—is important in various biological processes. For example, platelet activation leads to PtdSer exposure, which in turn activates clotting factors (1), and apoptotic cells expose PtdSer as a signal to be engulfed by phagocytes (2). Scramblases nonspecifically transport phospholipids bidirectionally (1). TMEM16F is essential for PtdSer exposure in activated platelets (3), whereas Xkr8 (XK-related protein 8) supports phospholipid scrambling after being cleaved by caspase during apoptosis (4). Flippases transport aminophospholipids from the extracellular to the cytoplasmic side (1). Some members (mammalian ATP8A1 and its orthologs in yeast and Caenorhabditis elegans, and ATP11C) of the type 4 subfamily of P-type adenosine triphosphatases (P4-ATPases) have been proposed to act as flippases (58). Earlier studies reported that the flippase is inactivated during apoptosis (9, 10), but its identity is unclear, as is the mechanism of its inactivation.

We used a haploid genetic screen in human KBM7 cells (11) to identify the plasma membrane PtdSer flippase. KBM7 cells incorporated 1-oleoyl-2-{6-[(7-nitro-2-1,3-benzoxadiazol-4-yl)amino]hexanoyl}-sn-glycero-3-PtdSer (NBD-PS), NBD-PtdEtn (NBD-PE), and NBD-PtdCho (NBD-PC) with different efficiencies (Fig. 1A). KBM7 cells were mutagenized with a gene trap vector. About 1.0% of the cells defective in NBD-PS internalization were expanded and subjected to a second sorting step to obtain “low flipping” (LF) cells (Fig. 1B). Gene trap insertion sites were recovered by polymerase chain reaction from LF cells and identified by deep sequencing (12). A proximity index (12) for genomic regions containing multiple insertions in close proximity identified two genes: CDC50A (cell division cycle protein 50A) and ATP11C (ATPase type 11C) (Fig. 1B). Information from a genome database (http://genome.ucsc.edu/cgi-bin/hgTracks?org=human ) and 5′-RACE (rapid amplification of cDNA end) analysis (fig. S1) indicated that most of the insertions were assigned to intron 1 in the transcriptional direction (Fig. 1C).

Fig. 1 Screening for phospholipid flippase.

(A) KBM7 cells were incubated with 1.5 μM NBD-phospholipids, treated with bovine serum albumin (BSA), and analyzed by fluorescence-activated cell sorting (FACS). Mean fluorescence intensity (MFI) is plotted. (B) Mutagenized KBM7 cells that incorporated low levels (~1%) of NBD-PS were collected and sorted to generate LF (low flipping) cells. (C) Identification of gene trap insertions. Proximity index is plotted for human chromosome with number (N) of insertions. Arrowheads denote insertion positions for sense (red) and antisense orientation (green). (D) Incorporation of NBD-phospholipids into ATP11CGT, ATP11CGT-hATP11C, CDC50AGT, or CDC50AGT-hCDC50A cells, relative to incorporation into KBM7 cells (n = 3). Error bars denote SD. (E) Effect of vanadate on NBD-PS incorporation into KBM7, ATP11CGT, and CDC50AGT cells, relative to incorporation into KBM7 cells in the absence of vanadate (n = 3). Error bars denote SD. (F) Annexin V staining profile of KBM7, ATP11CGT, CDC50AGT, and CDC50AGT-hCDC50A cells; percentages of Annexin V–positive cells are shown.

ATP11C is a member of the P4-ATPase family, and CDC50A is its β subunit (fig. S2A) (13, 14). They were expressed in various tissues (fig. S2B). Cloned cell lines (ATP11CGT and CDC50AGT) that lost the expression of ATP11C or CDC50A (fig. S2C) were isolated. The ability to incorporate NBD-PS and NBD-PE, but not NBD-PC, was reduced in ATP11CGT cells; transformation with human (h)ATP11C rescued this ability (Fig. 1D). The incorporation of phospholipids in ATP11CGT was inhibited by orthovanadate (Fig. 1E), which suggests that the remaining flippase activity was due to other P4-ATPases in KBM7 cells (fig. S2D). Among three members, only CDC50A was expressed in KBM7 cells (fig. S2D), and the internalization of NBD-PS was completely defective in CDC50AGT cells but could be rescued with hCDC50A (Fig. 1D). CDC50AGT cells, but not ATP11CGT cells, exposed PtdSer on the cell surface (Fig. 1F); this result suggests that the residual PtdSer flippase in ATP11CGT cells was sufficient to maintain asymmetrical PtdSer distribution. The ability to internalize NBD-PC was also reduced in CDC50AGT cells (Fig. 1D), indicating that some P4-ATPases in KBM7 cells may promote PtdCho flipping.

To study the effect of the PtdSer flippase on apoptotic PtdSer, we chose mouse W3 cells that undergo apoptosis upon treatment with Fas ligand (FasL) (15). W3 cells expressed several P4-ATPases (fig. S3A) and incorporated NBD-PS, NBD-PE, and NBD-PC with different efficiencies (fig. S3B). ATP11C in W3 cells was mutated using the CRISPR/Cas (clustered regulatory interspaced short palindromic repeats/CRISPR-associated) system (16) (fig. S3C). Two cloned cell lines (ATP11CED22 and ATP11CED23) carrying biallelic ATP11C truncations lost the ability to internalize NBD-PS and NBD-PE (Fig. 2A and fig. S3D). Apoptotic PtdSer exposure is accompanied by the loss of PtdSer flippase activity (9, 10). Treating W3 cells with FasL reduced ATP11C from 120 to 50 kD (Fig. 2B); this reduction was prevented by a caspase inhibitor, QVD-OPh [quinolyl-valyl-O-methylaspartyl-(2,6-difluorophenoxy)-methyl ketone]. A search using Cascleave (http://sunflower.kuicr.kyoto-u.ac.jp/~sjn/Cascleave) revealed three caspase recognition sequences (sites 1 to 3) in the nucleotide-binding (N) domain of hATP11C (fig. S4A). Mutants in sites 1 to 3 were generated (fig. S4B), tagged by green fluorescent protein (GFP) at the C terminus, and expressed in ATP11CED22 cells. The transformants were treated with FasL and analyzed by immunoblotting with antibodies to GFP (Fig. 2C). The FasL treatment caused a shift of the wild-type and doubly mutated hATP11C-GFP from 140 to 80 kD. But little cleavage was observed with the triply mutated hATP11C-GFP (CasR). Incubation of the membrane fraction carrying hATP11C-GFP with caspases revealed that hATP11C was cleaved by caspases 3, 6, and 7 (fig. S5). ATP11CED22 cells expressing CasR (ATP11CED22-CasR) incorporated NBD-PS as efficiently as those expressing wild-type ATP11C (ATP11CED22-hATP11C) (Fig. 2D) and could not expose PtdSer upon FasL treatment (Fig. 2E), although caspase-3 was activated (Fig. 2E). Transformation of W3-Ildm cells [a derivative of W3 cells (17)] and human Jurkat cells with CasR, but not with wild-type hATP11C, blocked the FasL-induced PtdSer exposure (fig. S6). CasR had no effect on the FasL-induced scramblase, as measured by NBD-PC incorporation (Fig. 2F). FasL-induced cell death, cell shrinkage, and DNA fragmentation were also normal in CasR-expressing cells (fig. S7). The FasL-treated W3-Ildm cells and the hATP11C transformants were efficiently engulfed by thioglycollate-elicited peritoneal macrophages (thio-pMacs), whereas the CasR transformants were not (Fig. 2G); this suggests that the flippase must be inactivated by caspase for engulfment of apoptotic cells.

Fig. 2 Cleavage of ATP11C during apoptosis.

(A) Incorporation of NBD-PS into ATP11CED22, ATP11CED23, ATP11CED22-hATP11C, and ATP11CED23-hATP11C cells, relative to incorporation into W3 cells (n = 3). Error bars denote SD. (B) W3 and ATP11CED22 cells were incubated for 1 hour with FasL in the presence or absence of 20 μM QVD-OPh. Membrane fractions were analyzed by immunoblotting with antibodies to mATP11C or Fas. Middle panel, longer exposure. (C) ATP11CED22, ATP11CED22-hATP11C-GFP, or ATP11CED22-mutant hATP11C-GFP cells were incubated for 45 min with FasL, and analyzed by immunoblotting with antibody to GFP or tubulin. (D) W3, ATP11CED22, ATP11CED22-hATP11C, and ATP11CED22-CasR cells were incubated with NBD-PS, treated with BSA, and analyzed by FACS. MFI is plotted (n = 3). Error bars denote SD. (E) Annexin V staining profile of W3, ATP11CED22, ATP11CED22-hATP11C, and ATP11CED22-CasR cells that were untreated (red) or FasL-treated for 2 hours (blue). Right panel: Immunoblotting of the cell lysates with antibody to cleaved caspase 3 or tubulin. (F) W3, ATP11CED22, ATP11CED22-hATP11C, and ATP11CED22-CasR cells were treated for 1 or 2 hours with FasL, incubated with 0.5 μM NBD-PC for 4 min, and analyzed by FACS. MFI is plotted (n = 3). Error bars denote SD. (G) W3-Ildm, W3Ildm-hATP11C, and W3Ildm-CasR cells were treated with FasL, labeled with pHrodo (a pH indicator), incubated for 1 hour with thio-pMacs, and analyzed by FACS.

CDC50A was then mutated using the CRISPR/Cas system in W3-Ildm cells that expressed only CDC50A in the CDC50 family (fig. S8A). Cloned cell lines (CDC50AED29 and CDC50AED62) carrying a biallelic truncation (fig. S8B) could not support the localization of hATP11C-GFP at the plasma membrane (Fig. 3A), could not internalize NBD-PS (Fig. 3B), and exposed PtdSer (Fig. 3C and fig. S9A). Transformation of CDC50AED cells with hCDC50A supported the localization of ATP11C at the plasma membrane, rescued the flippase activity, and blocked PtdSer exposure, confirming the chaperone-like function of CDC50 for P4-ATPase (5, 13, 14). W3 cells expressing a constitutively active form of TMEM16F (D430G-L) (3, 18) internalized NBD-PS (Fig. 3B) and bound MFG-E8, a PtdSer-binding protein (19), in HEPES-buffered saline containing 2.5 mM CaCl2 but not in Iscove’s modified Dulbecco’s medium with 10% fetal calf serum (IMDM–10% FCS) (Fig. 3D). In contrast, CDC50AED cells bound MFG-E8 in IMDM–10% FCS. Observation by microscope confirmed the binding of MFG-E8 to CDC50AED cells. The doubling time of CDC50AED29 cells (13.7 ± 0.14 hours) was slightly longer than that of W3-Ildm cells (12.2 ± 0.39 hours) (fig. S9B), but CDC50AED29 cells responded normally to FasL for apoptosis (fig. S9C).

Fig. 3 PtdSer exposure in CDC50A−/− cells.

(A) Observation of hATP11C-GFP expressed in CDC50AED29 or CDC50AED29-hCDC50A cells by confocal fluorescence microscopy. DIC, differential interference contrast. Scale bar, 5 μm. (B) Incorporation of NBD-PS into CDC50AED29, CDC50AED62, CDC50AED29-hCDC50A, CDC50AED62-hCDC50A, and W3-D430G-L cells, relative to incorporation into W3-Ildm cells (n = 3). Error bars denote SD. (C) W3-Ildm, CDC50AED29, and CDC50AED62 cells and their hCDC50A transformants (red) were stained with Cy5–Annexin V or fluorescein isothiocyanate (FITC)–MFG-E8 in Annexin V buffer. (D) W3-Ildm, CDC50AED29, and W3-D430G-L cells were incubated at room temperature for 5 min with FITC-MFG-E8 in HEPES-buffered saline or in IMDM–10% FCS. MFI is indicated (n = 3). Lower panels: W3-Ildm, CDC50AED29, and CDC50AED62 cells were stained with FITC-MFG-E8 in IMDM–10% FCS and observed by confocal fluorescence microscopy. Scale bar, 10 μm.

When living CDC50AED cells were cultured with thio-pMacs in medium containing 1.0% methylcellulose, more than 20% of the macrophages engulfed CDC50AED cells (Fig. 4A). The thio-pMacs did not engulf W3-Ildm cells, W3-D430G-L cells, or CDC50AED-hCDC50A cells. The engulfment of CDC50AED cells was inhibited by D89E [a mouse MFG-E8 (milk fat globule EGF factor 8) mutant carrying a point mutation in the RGD domain], which masks PtdSer (19). Similar to the engulfment of apoptotic cells (20), living CDC50AED cells were not engulfed by MerTK−/− thio-pMacs (Fig. 4B). Among 132 engulfment events observed (Fig. 4C and movies S1 and S2), about 80% were with living cells, while 20% involved apoptotic cells. As reported for the entosis of epithelial cells (21), the engulfment of living CDC50AED cells was reversible until a certain point. About 3% of the engulfed cells were released from the macrophages before they were transferred into lysosomes (Fig. 4D and movie S3). The release of engulfed cells was not observed with caspase-positive cells. Examination with electron microscopy showed that the engulfed living cells had a swollen morphology (Fig. 4E) different from the apoptotic cells with a condensed morphology. Engulfment of living PtdSer-exposing cells by thio-pMacs was also observed with CDC50A-null KBM7 cells (fig. S10). Subcutaneous transplantation of W3-Ildm cells into nude mice induced tumors in 8 of 11 recipients, and the tumor size was approximately 4.3 g after 4 weeks (Fig. 4F). CDC50AED cells, but not their hCDC50A transformants, lost the ability to induce tumors, which suggests that CDC50AED cells were cleared in vivo.

Fig. 4 Engulfment of viable cells.

(A) Thio-pMacs were incubated for 2 hours with pHrodo-labeled W3-Ildm, CDC50AED29, CDC50AED29-hCDC50A, or W3-D430G-L cells in the presence or absence of D89E (4 μg/ml) and analyzed by FACS. (B) Wild-type or MerTK−/− thio-pMacs were incubated for 2 hours with pHrodo-CDC50AED29 and analyzed by FACS. (C) pHrodo-CDC50AED29 was incubated with thio-pMacs in the presence of CellEvent Caspase-3/7 Green Detection Reagent (Life Technology). Images for pHrodo (red), CellEvent (green), and DIC were captured every 1.5 min. Scale bars, 10 μm. Engulfment of CellEvent-negative (living) and -positive (dead) cells was followed for at least 130 events in 30 fields of three independent experiments. Percentages of dead or living cell engulfment are at right. (D) Release of engulfed CDC50AED29 (arrowhead) from thio-pMacs. Scale bars, 10 μm. (E) Transmission electron micrograph of thio-pMacs engulfing CDC50AED29 or FasL-treated W3-Ildm cells. Scale bars, 2 μm. N, macrophage nucleus; L, engulfed living cells; A, engulfed apoptotic cells. (F) W3-Ildm, CDC50AED29, and CDC50AED29-hCDC50A (106 cells) were transplanted subcutaneously into nude mice (N = 6 to 11) and tumors were weighed 4 weeks later.

Our results show that ATP11C can function as a PtdSer flippase at the plasma membrane. ATP11C-deficient KBM7 cells and W3 cells exhibited reduced PtdSer flippase activity, whereas CDC50A-null cells almost completely lost the activity. Most, if not all, P4-ATPases require CDC50 family proteins as a functional subunit or chaperone (5, 22, 23). KBM7 cells and W3 cells expressed only CDC50A among CDC50 proteins, but also expressed ATP11C and other P4-ATPases, some of which may also function as a PtdSer flippase. PtdSer flippase has been considered to be specific for aminophospholipids (14). But, as reported previously with Jurkat cells (24), CDC50A-null KBM7 cells lost the ability to transport PtdCho, which suggests that some P4-ATPases may flip PtdCho. In apoptosis, Xkr8’s scramblase is activated (4) and ATP11C’s flippase is inactivated to expose PtdSer. The PtdSer exposure in activated platelets and lymphocytes is Ca2+-dependent (3, 25). Because a high Ca2+ concentration inhibits P4-ATPases (26), PtdSer exposure in these processes may be mediated by the Ca2+-dependent activation of scramblases coupled to the Ca2+-mediated down-regulation of flippase activity. Once cellular Ca2+ levels are reduced, ATP11C and/or other P4-ATPases would reestablish the asymmetric phospholipid distribution in the plasma membrane. On the other hand, caspase-mediated apoptotic PtdSer exposure is irreversible, leading to engulfment by macrophages.

The engulfment of living cells by neighboring cells or macrophages has been reported in various systems (21, 27, 28). Our finding that macrophages engulf PtdSer-exposing CDC50A-null cells supports the idea (28) that when PtdSer is exposed, even viable cells can be engulfed. The engulfment of living cells may be involved in diseases such as hemophagocytosis, neurodegeneration, and cancer (29). ATP11C-defective mice lose a large number of B cells during differentiation from progenitor B cells to precursor B cells in bone marrow (7, 30), and a weak defect in the PtdSer internalization has been detected in progenitor B cells (7). One possible explanation is that among the P4-ATPases promoting flippase activity, only ATP11C is expressed in the early stage of B cell development. Thus, ATP11C-defective cells expose PtdSer and are engulfed by macrophages.

Supplementary Materials

www.sciencemag.org/content/344/6188/1164/suppl/DC1

Materials and Methods

Figs. S1 to S10

References (3141)

References and Notes

  1. Acknowledgments: We thank B. H. Cochran for KBM7 cells, K. Higasa and M. Shimizu for help in next-generation sequencing, K. Okamoto-Furuta and H. Kohda for support in electron microscope analysis, and M. Fujii for secretarial assistance. This work was supported in part by grants-in-aid from the Ministry of Education, Science, Sports, and Culture, Japan. Kyoto University has filed a patent entitled “Method of screening agents for the treatment or prevention of cancer or apoptosis-related diseases” (inventors, Shigekazu Nagata, Katsumori Segawa; reference number: 61/978415). Plasmids encoding human ATP11C or CDC50A, human KBM7 cells lacking ATP11C or CDC50A, and mouse W3 cell lines lacking ATP11C or CDC50A are all available under material transfer agreements. T.R.B. is a co-founder, shareholder, and member of the scientific advisory board of Haplogen, an early-phase biotech company based on haploid genetics.
View Abstract

Stay Connected to Science

Navigate This Article