Critical Role of Lipid Scramblase TMEM16F in Phosphatidylserine Exposure and Repair of Plasma Membrane after Pore Formation

SUMMARY

Plasma membrane damage and cell death during processes such as necroptosis and apoptosis result from cues originating intracellularly. However, death caused by pore-forming agents, like bacterial toxins or complement, is due to direct external injury to the plasma membrane. To prevent death, the plasma membrane has an intrinsic repair ability. Here, we found that repair triggered by pore-forming agents involved TMEM16F, a calcium-activated lipid scramblase also mutated in Scott’s syndrome. Upon pore formation and the subsequent influx of intracellular calcium, TMEM16F induced rapid “lipid scrambling” in the plasma membrane. This response was accompanied by membrane blebbing, extracellular vesicle release, preserved membrane integrity, and increased cell viability. TMEM16F-deficient mice exhibited compromised control of infection by Listeria monocytogenes associated with a greater sensitivity of neutrophils to the pore-forming Listeria toxin listeriolysin O (LLO). Thus, the lipid scramblase TMEM16F is critical for plasma membrane repair after injury by pore-forming agents.

In Brief

Pore-forming agents like bacterial toxins or complement kill cells by attacking the plasma membrane. Wu et al. show that the calcium-activated lipid scramblase TMEM16F promotes plasma membrane repair after pore formation by enhancing membrane fluidity and facilitating release of extracellular vesicles containing damaged membranes.

Graphical Abstract

INTRODUCTION

In addition to initiating intracellular responses to external cues, the plasma membrane serves as a physical barrier protecting cells from external attack (Andrews and Corrotte, 2018; Christie et al., 2018). Agents such as pore-forming proteins can kill cells directly by binding and damaging the plasma membrane (Christie et al., 2018; Hamon et al., 2012; Radoshevich and Cossart, 2018; Seveau, 2014). Pore-forming agents include a variety of bacterial toxins, such as listeriolysin O (LLO) and streptolysin O (SLO), which create pores in cellular membranes for pathogenicity. They also include components of the complement cascade, which is implicated in elimination of pathogens and can also cause damage to host mammalian cells.

The process of cell death induced by pore-forming agents is distinct from intracellularly regulated cell death mechanisms such as necroptosis and pyroptosis (Andrews and Corrotte, 2018; Brito et al., 2019; Draeger et al., 2011; Idone et al., 2008). Indeed, pore formation triggers cell death because of signals initiated in the outer leaflet of the plasma membrane upon binding of pore-forming agents to the plasma membrane. In contrast, processes such as necroptosis and pyroptosis lead to cell death via death cascades initiated intracellularly. Although these intracellular pathways also cause alterations in plasma membrane permeability, these changes are secondary to activation of intracellular effectors by the death cascades.

To maintain cellular integrity, the plasma membrane has an intrinsic repair capacity (Andrews and Corrotte, 2018; Brito et al., 2019; Draeger et al., 2011; Idone et al., 2008). This function leads to removal of damaged membranes by processes such as release of extracellular vesicles (also known as microparticles, microvesicles, or ectosomes), endocytosis, and resealing. However, the molecular pathways and cellular biology processes responsible for this process are only partly understood. In the case of pore-induced repair, there is firm evidence that it is dependent on an increase in intracellular calcium, which is triggered by the influx of extracellular calcium during pore formation (Idone et al., 2008; Jimenez et al., 2014). Pore-induced repair also requires components of the endosomal sorting complexes required for transport (ESCRT) machinery, which is needed for scission of extracellular vesicles or endocytic vesicles involved in eliminating damaged membranes.

It has also been reported that pore formation triggers cell surface externalization of the phospholipid phosphatidylserine (PS), which is normally concentrated in the inner leaflet of the plasma membrane (Carrero et al., 2008; Fӧller et al., 2007). Although the biological significance of PS exposure in this setting is unclear, one possibility is that PS exposure simply reflects the transition from plasma membrane damage to full-blown cell death, which is often accompanied by PS exposure (Arandjelovic and Ravichandran, 2015). Alternatively, PS exposure may be indicative of a repair program aimed at preventing cell death.

PS externalization at the plasma membrane is the consequence of lipid redistribution (“scrambling”) in the plasma membrane by the so-called “lipid scramblases” (Whitlock and Hartzell, 2017; Yang et al., 2018). The lipid scramblase TMEM16F, also known as anoctamin 6 (ANO6), is a member of the TMEM16 family of transmembrane scramblases (Suzuki and Nagata, 2014; Suzuki et al., 2010; Whitlock and Hartzell, 2017; Yang et al., 2012). It is expressed ubiquitously, albeit reportedly in greater amounts in immune cells. TMEM16F plays a specific role in lipid scrambling in response to elevated intracellular calcium (Suzuki et al., 2010; Whitlock and Hartzell, 2017; Yang et al., 2012). Most notably, TMEM16F-mediated PS exposure is required for platelets to aggregate and release their pro-coagulant extracellular vesicles (Fujii et al., 2015; Suzuki et al., 2010; Yang et al., 2012). Consequently, inactivating mutations of TMEM16F result in a bleeding disorder in humans known as Scott’s syndrome (Suzuki and Nagata, 2014; Suzuki et al., 2010; Whitlock and Hartzell, 2017; Yang et al., 2012). In contrast, TMEM16F does not participate in lipid scrambling in the context of apoptosis, which involves the caspase-activated scramblase Xkr8 (Suzuki et al., 2013). Likewise, it is not implicated in lipid scrambling during necroptosis, which engages an as-yet unidentified scramblase (Gong et al., 2017). Recently, TMEM16F was also implicated in the release of extracellular vesicles containing the inhibitory immune cell receptor PD-1 in T cells following treatment with the calcium ionophore ionomycin (Bricogne et al., 2019). However, the physiological relevance of this finding remains to be clarified.

Here we elucidated the plasma membrane repair mechanism triggered by pore-forming agents. We found that lipid scrambling induced by pore-forming agents was mediated by TMEM16F, which enabled membrane blebbing, release of extracellular vesicles, reduced pore formation, and improved cell survival. Mice lacking TMEM16F exhibited compromised control of Listeria monocytogenes, a bacterium utilizing the pore-forming toxin LLO for pathogenicity (Hamon et al., 2012; Radoshevich and Cossart, 2018; Seveau, 2014). This defect correlated with a diminished ability of neutrophils, key components of the first line of defense against Listeria (Hamon et al., 2012; Radoshevich and Cossart, 2018; Witter et al., 2016), to resist the toxic effect of LLO. Therefore, TMEM16F is a key component of the plasma membrane repair mechanism triggered by pore-forming agents.

RESULTS

Lipid Scrambling Induced by Pore Formation Is Part of a Calcium-Dependent Repair Mechanism

To elucidate the repair mechanism triggered by pore formation, we focused on immune cells because these cells can be easily obtained in large numbers and are particularly prone to injury by pore-forming toxins (Brito et al., 2019; Hamon et al., 2012; Radoshevich and Cossart, 2018; Witter et al., 2016). As reported for non-immune cell types (Carrero et al., 2008; Fӧller et al., 2007), primary mouse thymocytes treated with a low dose (1 nM) of the Listeria-derived pathogenic toxin LLO displayed rapid (within less than 2 min) binding to annexin V or the C2 domain of lactadherin (LactC2), which both detect surface PS (van Engeland et al., 1996; Hou et al., 2011), implying PS externalization (Figure 1A; Figures S1A and S1B). There was also rapid incorporation of propidium iodide (PI) and 4′,6-diamidino-2-phenylindole (DAPI) in addition to a reduction in carboxyfluorescein succinimidyl ester (CFSE) labeling, in keeping with loss of membrane integrity (Figure 1A).

Figure 1. Bacterial Toxins and Other Pore-Forming Agents Can Induce PS Exposure in the Absence of Detectable Loss of Plasma Membrane Integrity.

(A) Thymocytes from wild-type mice were treated with LLO (1 nM) for the indicated times at room temperature (RT). Representative histograms of PI, DAPI, CFSE, and annexin V staining are shown at the top, whereas statistics for multiple experiments (PI, n = 12; DAPI, n = 7; CFSE, n = 10; annexin V, n = 8) are shown at the bottom. MFI, mean fluorescence intensity.

(B) Thymocytes were treated (+) or not (−) with LLO (1 nM) for 5 min, followed by staining with annexin V and PI. Representative dot plots are shown on the left, whereas statistics for multiple experiments (n = 5) are shown on the right. Percentages of cells in each quadrant are indicated in the dot plots on the left. Statistics for annexin V⁺PI⁻ and PI⁺ cells are shown on the right.

(C) Same as (B), but thymocytes were treated or not with streptolysin O (SLO, 500 U/mL) for 10 min, saponin (0.001%) for 1 min, or complement (100%) for 30 min. Representative dot plots are shown on the left, whereas statistics for multiple experiments (SLO, n = 7; saponin, n = 9; complement, n = 8) are shown on the right.

*p < 0.05, **p < 0.01, ***p < 0.001 (two-tailed Student’s t tests). Data are means ± SEM

Interestingly, when individual cells were analyzed, ~50% of LLO-treated cells were positive for annexin V but remained negative for PI, indicating that lipid scrambling was taking place even in the absence of compromised membrane integrity (Figure 1B). Analogous effects were seen with other pore-forming agents; namely, the bacterial toxin SLO (which is implicated in the pathogenicity of β-hemolytic Streptococcus strains), the detergent saponin (which creates pores by extracting membrane cholesterol and phospholipids), and the complement cascade (a component of the mammalian immune system capable of killing pathogens and, sometimes, host cells through pore formation) (Brito et al., 2019; Christie et al., 2018; Hamon et al., 2012; Jimenez et al., 2014; Seveau, 2014; Figure 1C; Figure S1B). Similar effects were noted in various lymphoid cell lines, both T and B cells (Figures S1C–S1E).

In addition to PS exposure, pore-forming agents induced an increase in intracellular calcium in thymocytes, as described for other cell types (Bouillot et al., 2018; Figure 2A). Importantly, prevention of calcium entry by the chelator ethylene glycol tetraacetic acid (EGTA) caused a marked augmentation of PI staining in LLO-treated cells compared with cells treated with LLO alone (Figure 2B). No effect was seen with other inhibitors; namely, inhibitors of actin polymerization (latrunculin A), myosin activation (blebbistatin), the mitogen-activated protein kinase pathway (U0126), phosphatidylinositol 3-kinase (wortmannin), and Src-family protein tyrosine kinases (SU6656). Conversely, addition of ionomycin, a calcium ionophore that mimics the effect of elevated intracellular calcium, diminished the permeabilizing effect of LLO on PI, DAPI, and CFSE compared with LLO alone (Figures 2C and 2D). It also caused greater externalization of PS. Although ionomycin alone induced PS exposure, as already documented (Suzuki et al., 2010), it caused no increase in PI staining, implying that it caused lipid scrambling without detectable pore formation (Figure 2E).

Figure 2. Pore-Forming Agents Trigger a Calcium-Dependent Protective Mechanism.

(A) Calcium influx in thymocytes treated or not with the indicated reagents. The arrow represents time of addition. a.u., arbitrary units; Iono, ionomycin. Representative of n = 2.

(B) Same as in Figure 1A, except that thymocytes were pretreated or not with DMSO, the actin polymerization inhibitor latrunculin A (Lat A, 1 μM), the myosin inhibitor blebbistatin (Bleb, 50 μM), the mitogen-activated protein kinase pathway inhibitor U0126 (1 μM), the phosphatidylinositol 3-kinase inhibitor wortmannin (Wort, 1 μM), the Src kinase inhibitor SU6656 (100 nM), or the calcium chelator EGTA (2 mM) prior to addition of LLO. Representative histograms of PI staining are shown on the left, whereas statistics for multiple experiments (n = 6) are shown on the right.

(C and D) CFSE-loaded thymocytes were treated or not with LLO (1 nM) alone or LLO with Iono (5 μM) for 5 min. Staining with annexin V and DAPI was analyzed. Representative dot plots are shown in (C), whereas statistics for multiple experiments (DAPI, n = 7; CFSE, n = 5; annexin V, n = 15) are shown in (D). (E) Annexin V and PI staining of thymocytes treated or not with Iono. Representative dot plots are shown at the top, whereas statistics for multiple experiments (n = 8) are depicted at the bottom.

NS, not significant; *p < 0.05; **p < 0.01; ***p < 0.001 (two-tailed Student’s t tests). Data are means ± SEM.

Hence, lipid scrambling induced by pore-forming agents such as bacterial toxins and complement was part of a calcium-dependent mechanism aimed at limiting the deleterious effect of pore formation.

Essential Role of TMEM16F in Plasma Membrane Repair Induced by Pore-Forming Agents

Because TMEM16F is a calcium-activated lipid scramblase (Suzuki and Nagata, 2014; Suzuki et al., 2010; Whitlock and Hartzell, 2017; Yang et al., 2012), we tested whether it might be implicated in the repair mechanism triggered by pore-forming agents. To this end, the effect of TMEM16F deficiency was tested, using cells from TMEM16F-deficient (knockout [KO]) mice. Compared with control mice, TMEM16F KO mice lacked the TMEM16F protein, which normally migrates as multiple polypeptide species in protein gels, as already reported (Suzuki et al., 2010; Figure 3A). Only a non-specific immunoreactive product remained. Loss of TMEM16F had no effect on T cell development or distribution of lymphoid subsets, as described previously (Hu et al., 2016; Figures S2A and S2B). However, as documented elsewhere (Fujii et al., 2015; Ousingsawat et al., 2015; Yang et al., 2012), it caused abnormal clot formation in vitro (reduced “clot rate”) and reduced embryonic or perinatal viability (Figures S2C and S2D). TMEM16F deficiency in mice has also been reported to cause a bone mineralization defect (Ousingsawat et al., 2015), although we did not test this phenotype in our mouse strain.

Figure 3. TMEM16F Protects Cells from the Toxic Effect of Pore-Forming Agents.

(A) Representative anti-TMEM16F immunoblot of thymocytes of control and TMEM16F KO mice (n = 3). Asterisk indicates non-specific band.

(B) Thymocytes from control and TMEM16F KO mice were treated or not with LLO, and staining with DAPI in addition to annexin V was monitored. Representative dot plots are shown on the left, whereas statistics for multiple experiments (n = 10) are depicted on the right.

(C) CFSE-loaded thymocytes from control and TMEM16F KO mice were treated or not with LLO (1 nM). Representative histograms are shown on the left, whereas statistics of percentages of CFSE^lo cells in multiple experiments are depicted on the right (n = 9).

(D) Same as (C), but PI staining was monitored by spinning-disk microscopy (Video S1). Representative still photographs before and 5 min after addition of LLO are shown on the left. Statistics for multiple independent fields (n = 15) are shown on the right. ΔPI intensity represents the raw fluorescence intensity of individual images (determined with ImageJ software) 5 min after treatment minus before treatment.

(E–G) Same as (B), but thymocytes were treated by SLO for 10 min (E), complement for 30 min (F), or saponin for 1 min (G). Representative dot plots are shown on the left, whereas statistics for multiple experiments (SLO, n = 10; complement, n = 9; saponin, n = 9) are depicted on the right.

(H) Same as (B), except that cells were treated in calcium-free medium or standard medium. Representative histograms of PI staining are shown on the left, whereas statistics for multiple experiments (n = 6) are depicted on the right. Filled histograms, LLO-treated cells; open histograms, untreated cells.

(I) Calcium influx in control and TMEM16F KO thymocytes treated with LLO. Iono was used as positive control. Representative of n = 3.

*p < 0.05, **p < 0.01, ***p < 0.001 (two-tailed Student’s t tests). Data are means ± SEM.

See also Video S1.

Compared with control thymocytes, TMEM16F KO thymocytes had a severely reduced proportion of cells displaying PS exposure and intact membrane integrity (annexin V⁺DAPI⁻ cells) in response to LLO (Figure 3B). Furthermore, they possessed a significantly expanded (~25%) population of cells having no PS exposure and loss of membrane integrity (annexin V⁻DAPI⁺ cells) (Figure 3B). LLO-treated TMEM16F KO cells also exhibited increased leakage of CFSE compared with control cells (Figure 3C). The increase in PI entry in TMEM16F KO cells compared with control cells was confirmed by live imaging using spinning-disk confocal microscopy (Figure 3D; Video S1). Analogous effects of TMEM16F deficiency were observed in cells treated with SLO, complement, or saponin (Figures 3E–3G).

Unfortunately, the available reagents did not enable us to test directly whether calcium and TMEM16F were required for binding of LLO to the plasma membrane. To circumvent this issue, we first examined whether removal of calcium from medium interfered with the ability of LLO to increase membrane permeability. As was the case for calcium chelation with EGTA (Figure 2B), elimination of extracellular calcium resulted in increased PI entry in response to LLO compared with untreated cells (Figure 3H). This effect was seen in control and in TMEM16F KO cells. We also ascertained whether TMEM16F was required for the ability of the toxin to induce calcium flux in the presence of extracellular calcium. TMEM16F deficiency had no effect on the ability of LLO to increase intracellular calcium (Figure 3I). Similar results were obtained with ionomycin. Thus, calcium and TMEM16F were not needed for LLO to alter membrane permeability. These findings implied that calcium and TMEM16F were not required for the capacity of LLO to bind the plasma membrane.

In summary, TMEM16F deficiency caused a greater loss of membrane integrity in response to pore-forming agents. This effect was accompanied by a compromised ability to induce lipid scrambling and did not seem to be due to loss of binding of pore-forming agents to the plasma membrane.

TMEM16F Decreases Pore Size and Prevents Cell Death

To assess whether the negative effect of TMEM16F deficiency on repair was due to an increase in the size of membrane pores, we tested the effect of LLO on permeability to dextran molecules of different sizes. In the absence of LLO, there was little or no entry of either dextran-3K (molecular weight [MW] = ~3,000 g/mol) or dextran-10K (MW = ~10,000 g/mol) in control and in TMEM16F KO cells (Figures 4A and 4B). In the presence of LLO, entry of dextran-3K was greater than that of dextran-10K in either cell type, as expected from the smaller MW of dextran-3K. Although, compared with LLO-treated control cells, LLO-exposed TMEM16F KO cells did not display a greater permeability to dextran-3K, they had a more pronounced increase in permeability to dextran-10K.

Figure 4. TMEM16F Preserves Membrane Integrity and Prevents Cell Death in Response to Pore-Forming Toxins.

(A and B) Cells were treated with LLO as for Figure 3B, but Alexa 488-labeled dextran of different MWs (3K or 10K) was added during LLO treatment. Representative histograms of Alexa 488 staining are shown in (A), whereas statistics for multiple experiments (n = 6) are depicted in (B). Cells showing increased dextran incorporation are demarcated by bars and quantified as percentages at the top of the histograms.

(C and D) LDH release was measured after treatment with LLO (C) or SLO (D). Data from multiple independent experiments are shown (C, n = 7; D, n = 5). Representative of n = 3.

*p < 0.05, **p < 0.01, ***p < 0.001 (two-tailed Student’s t tests). All data are means ± SEM.

We also monitored release of lactate dehydrogenase (LDH), a large cytoplasmic molecule (MW, R ≥35,000 g/mol) liberated from cells when they are dying (Cummings and Schnellmann, 2004). Unlike control cells, TMEM16F KO cells treated with LLO released appreciable quantities of LDH (Figure 4C). This observation implied that loss of TMEM16F not only caused larger pores but also compromised cell viability. Similar effects were seen with SLO (Figure 4D).

Therefore, lack of TMEM16F resulted in larger pores and more cell death in response to pore-forming agents.

TMEM16F Promotes Membrane Blebbing and Extracellular Vesicle Shedding in Response to Pore-Forming Agents

To identify how TMEM16F was enabling repair of the plasma membrane, changes in cell morphology were studied, using spinning-disk microscopy. In these studies, the plasma membrane was detected by labeling with fluorescent antibodies against the surface markers CD4, CD8, or Thy-1. When added to normal cells, LLO induced rapid and pronounced morphological changes in the plasma membrane, characterized primarily by formation of membrane “blebs” (Figure 5A; Video S2). Some intracellular vesicles were also seen. These effects were nearly abrogated in TMEM16F KO cells. Ionomycin, which decreased the toxic effects of LLO but did not trigger appreciable pore formation (Figures 2C–2E), also induced membrane blebbing in addition to causing cell projections resembling filopodia (Figure 5B; Video S3). Both projections and membrane blebbing were abolished in TMEM16F KO cells compared with control cells. The changes induced by LLO or ionomycin were significantly reduced by the calcium chelator 1,2-bis(2-aminophenoxy) ethane-N,N,N′,N′-tetraacetic acid tetrakis(acetoxymethyl ester) (BAPTA-AM), indicating that they were dependent on increased intracellular calcium (Figures 5C and 5D; Videos S4 and S5). The projections, which were induced by ionomycin but not by LLO, were strongly inhibited by latrunculin A (Figure S3A; Video S6).

Figure 5. TMEM16F-Dependent Membrane Blebbing and Extracellular Vesicle Release during Membrane Repair.

(A) Thymocytes from control and TMEM16F KO mice were fluorescently labeled with anti-CD4, CD8, or Thy-1 antibodies. They were then treated or not with LLO and analyzed by spinning-disk microscopy (Video S2). Representative still photographs of individual cells taken before and 5 min after addition of LLO are shown on the left. A statistical representation of the proportions of cells displaying membrane blebs is shown on the right. Statistics for cells in multiple independent fields are shown. Each symbol represents one field. Results are pooled from 4 independent experiments (n = 16 fields for control, n = 13 fields for KO). Each field typically contained 20–40 cells.

(B) Same as (A), but cells were treated with Iono (5 μM) (Video S3). Results are pooled from 5 independent experiments. One field per experiment was analyzed.

(C and D) Same as (A) and (B), respectively, but cells were pretreated with either DMSO or BAPTA-AM (Videos S4 and S5). Results are pooled from 3 independent experiments (C, n = 5 fields for DMSO and n = 8 fields for BAPTA-AM; D, n = 5 fields for DMSO and n = 8 fields for BAPTA-AM).

(E) Histograms of membrane receptor staining on control and TMEM16F KO thymocytes, treated or not with Iono. Representative plots of six independent experiments are shown.

(F) Immunoblots of various proteins from total cell lysates and extracellular vesicles (EVs) from control and TMEM16F KO thymocytes, treated with Iono. Representative blots from two independent experiments are shown.

Results were pooled from 4 (A), 5 (B), 3 (C), and 3 (D) mice studied in independent experiments. All data are means ± SEM. Scale bars, 5 μm (A–D).

See also Video S2, S3, S4, and S5.

To test whether the TMEM16F-dependent morphological changes in the plasma membrane resulted in a loss of surface molecules, flow cytometry analyses were performed. Ionomycin-treated cells were primarily, but not exclusively, used for these studies because ionomycin resulted in more prominent plasma membrane changes and did not cause detectable pore formation compared with LLO. The TMEM16F-dependent membrane alterations in response to ionomycin were paralleled by a reduction in cell surface staining for CD4, CD8, Thy-1, CD45, and CD48 in control cells but not in TMEM16F KO cells (Figure 5E), suggesting that surface membrane proteins were lost during repair. Similar but less extensive effects were seen with LLO (Figure S3B). To test whether protein loss occurred through extracellular vesicle release, differential ultracentrifugation was performed to recover extracellular vesicles. Ionomycin-treated control cells shed material that contained most cellular proteins, including membrane and cytosolic proteins (Figure 5F). No protein was present in preparations obtained from TMEM16F KO cells, indicating that no extracellular vesicles were released. Insufficient protein material was released from cells treated with LLO, precluding firm conclusions in a similar analysis performed with LLO.

To obtain a finer resolution of the TMEM16F-dependent plasma membrane changes during repair, scanning electron microscopy studies were conducted, using control or TMEM16F KO cells treated with LLO, ionomycin, or both. First we found that, in the absence of any treatment, control and TMEM16F KO thymocytes displayed relatively smooth surfaces, with only a few short projections resembling microvilli (Figures 6A and 6B). Second, addition of LLO caused the appearance of membrane blebs (Figure 6A, red arrowheads). However, the percentage of cells displaying bleb formation was much lower than that seen in the spinning-disk confocal microscopy analyses and was not different between control and TMEM16F KO cells (Figure 5A). This observation presumably indicated that blebs were lost during the fixation step required for electron microscopy studies. Nonetheless, we also observed larger outward membrane protrusions, denoted as “bulges” (Figure 6A, orange arrowheads), that were more frequent in control compared with TMEM16F KO cells (Figure 6B). Membrane blebs, in addition to projections (Figure 6A, blue arrowheads), were also seen with ionomycin alone. Such effects were nearly abrogated in TMEM16F KO cells. Third, we observed that LLO, but not ionomycin, triggered the appearance of “indentations” in the plasma membrane that resembled pores (Figures 6A, green arrowheads, and 6B). Unlike blebs, these indentations were more numerous in TMEM16F KO cells compared with control cells. Fourth, in keeping with the results of Figure 2, treatment with LLO plus ionomycin evoked more numerous blebs but much fewer pores compared with LLO alone. These changes took place in control cells but not in TMEM16F KO cells.

Figure 6. Electron Microscopy Studies of TMEM16F-Related Changes in Plasma Membrane Morphology and Cellular Integrity after Pore Formation.

(A and B) Scanning electron microscopy analyses of thymocytes from control and TMEM16F KO mice, treated or not with the indicated reagents. Representative photographs are shown in (A). Examples of bulges, blebs, pores, and projections are identified by arrowheads. Cells at the bottom are shown with greater magnification. These cells are boxed in white at the top. Statistical representations of the proportions of cells displaying bulges, blebs, EVs, pores, and projections are shown in (B). Each symbol represents data from one photograph. Results are pooled from 4 independent experiments.

(C) Transmission electron microscopy analyses of thymocytes from control and TMEM16F KO mice, treated or not with LLO. Two examples of cells for each condition and genotype are shown on the left, whereas statistical analyses of the proportions of permeabilized cells are shown on the right. Permeabilized cells were defined as cells with poorly defined plasma membranes and decreased cytoplasmic density. Results are pooled from n = 4 mice for control; n = 10 mice for KO.

Scale bars, 10 μm (A) and 1 μm (C). **p < 0.01, ***p < 0.001 (two-tailed Student’s t tests); ND, not detected. Results were pooled from 4 mice (A and C) studied in independent experiments. All data are means ± SEM.

Last, the alterations in the plasma membrane and inner cell structures were analyzed using transmission electron microscopy. In the absence of LLO, the morphology of control and TMEM16F KO cells was not noticeably different (Figure 6C). However, following exposure to LLO, TMEM16F KO cells, but not control cells, displayed attenuated plasma membrane definition and reduced electron density in the cytoplasmic region. The latter was consistent with a loss in intracellular content. The morphology of the nucleus was not appreciably altered. The other plasma membrane changes, including blebs and bulges, were not seen in this analysis, presumably because cells were fixed and sectioned prior to these studies.

Thus, microscopy studies showed that TMEM16F-dependent membrane repair in response to LLO or ionomycin primarily occurred through membrane blebbing and extracellular vesicle release. In the absence of TMEM16F, LLO triggered greater numbers of pore-like structures in addition to a reduction in cytoplasmic electron density, suggestive of loss of content.

TMEM16F Protects from the Pathogenic Effect of Listeria monocytogenes in Mice

To examine the relevance of TMEM16F-mediated plasma membrane repair in vivo, we used a model of infection by the bacterium L. monocytogenes, an intracellular pathogen requiring the toxin LLO for pathogenicity (Czuczman et al., 2014; Hamon et al., 2012; Radoshevich and Cossart, 2018; Seveau, 2014; Witter et al., 2016). Following intravenous injection, Listeria rapidly infects neutrophils and macrophages in the spleen and liver, where immune cells control infection by killing the bacteria.

Control littermates and TMEM16F KO mice were infected intravenously with Listeria and, after 1 or 3 days, live Listeria bacteria were enumerated in the spleen and liver (Figure 7A). Compared with control mice, TMEM16F KO mice had greater numbers (up to 10-fold more) of bacterial colony-forming units (CFUs) in spleen and liver (Figure 7B). This difference was more pronounced in liver compared to spleen and on day 3 compared with day 1. In histological analyses of liver tissue, TMEM16F KO mice showed greater numbers of inflammatory cells compared with control mice (Figure 7C). Moreover, in serum, TMEM16F KO mice displayed higher levels of alanine transaminase (ALT), a marker of liver damage (Figure 7D).

Figure 7. TMEM16F Is Required for Protection against Listeria monocytogenes.

(A and B) Control and TMEM16F KO mice were infected intravenously with L. monocytogenes. After 1 or 3 days, the liver and spleen were harvested, and the number of remaining live bacteria in these tissues was determined using a bacterial colony formation assay. A schematic representation of the protocol is depicted in (A). Statistical analyses of remaining colony-forming units (CFUs) of Listeria per gram of tissue in multiple mice (n = 24 for day 1, n = 27 for day 3) are shown in (B).

(C) Hematoxylin and eosin (H&E) staining of liver tissue after 1 or 3 days of infection. Representative sections are shown on the left. Arrows represent inflammatory foci. Statistical analyses of the numbers of inflammatory foci per square centimeter for multiple mice (day 1, n = 19 for control and n = 18 for KO; day 3, n = 19 for both control and KO) are shown on the right. One slide was analyzed for each mouse. Scale bar, 40 μm.

(D) Alanine aminotransferase (ALT) levels were measured in serum 3 days after infection. n = 18 mice in each group. Levels in uninfected mice are shown for comparison.

(E and F) Kinetics of annexin V and PI staining for neutrophils from control and TMEM16F KO mice treated or not for different times with LLO (10 nM). Representative dot plots are shown in (E), whereas statistics for multiple experiments (n = 16) are depicted in (F).

(G) LDH release by neutrophils from control and TMEM16F KO mice treated with LLO (n = 6). Spontaneous LDH release was subtracted from values obtained after LLO treatment. *p < 0.05, **p < 0.01, ***p < 0.001 (two-tailed Student’s t tests). Data are means ± SEM.

To test whether TMEM16F might mediate its effect by protecting cells from the toxic effects of LLO at the plasma membrane (Hamon et al., 2012; Radoshevich and Cossart, 2018; Seveau, 2014), bone marrow neutrophils from control and TMEM16F KO mice were isolated and treated with LLO in vitro. In response to LLO, TMEM16F KO neutrophils became positive for PI faster compared with control neutrophils (Figures 7E and 7F). This finding implied that TMEM16F KO neutrophils were more susceptible to membrane damage by LLO. In addition, unlike control cells, all annexin V-positive TMEM16F KO cells were also PI-positive and, thus, were presumably already dying or dead (Figures 7E and 7F). TMEM16F KO neutrophils also showed greater release of LDH in response to LLO compared with control cells (Figure 7G). As was the case for thymocytes (Figure 5), treatment of control neutrophils with ionomycin, a potent agonist of TMEM16F-dependent membrane repair, induced extensive membrane blebbing and projections (Figure S4). These effects were lost in TMEM16F KO neutrophils.

Therefore, TMEM16F was required for efficient clearance of L. monocytogenes in vivo. This function was associated with the capacity of TMEM16F to promote membrane blebbing and preserve membrane integrity in LLO-treated neutrophils.

DISCUSSION

Here we found that the calcium-activated lipid scramblase TMEM16F plays a key role in the plasma membrane repair response to pore-forming agents. Based on our data, we propose the following model. First, pore-forming agents bind to the plasma membrane and trigger pore formation. These pores are initially sufficient to initiate a flux of extracellular calcium but too small for appreciable entry of PI or DAPI and release of CFSE. In the case of toxins, the smaller pores may be due to insertion of toxin monomers or oligomers, rather than polymers, in the membrane (Christie et al., 2018; Hamon et al., 2012; Radoshevich and Cossart, 2018; Seveau, 2014). The binding of pore-forming agents to the membrane is independent of calcium and TMEM16F.

Second, the increase in intracellular calcium activates TMEM16F, which provokes lipid scrambling in the plasma membrane. Scrambling presumably augments the fluidity or plasticity of the membrane, leading to formation of membrane blebs and release of extracellular vesicles (Gong et al., 2017; Jimenez et al., 2014). The release of extracellular vesicles enables removal of damaged membranes and, possibly, may help with eliminating the toxin from the membrane, although future studies with improved reagents are needed to address the latter possibility. Although the membrane repair activity of TMEM16F is assumed here to be mediated by its scramblase function, it should be pointed out that previous studies have documented that TMEM16F can also operate as a calcium-activated ion channel (Shimizu et al., 2013; Yang et al., 2012; Ye et al., 2019). Although not tested, it is plausible that some of the effects of TMEM16F in repair were caused or modulated by this ion channel activity.

Although our data implied that membrane blebbing and microparticle release took place prominently in response to LLO or ionomycin, it is possible that other mechanisms to eliminate damaged membranes, such as endocytosis and resealing, also occurred, in particular in response to other pore-forming agents. Nonetheless, it has been reported that SLO and saponin induced membrane blebbing (Jimenez et al., 2014; Keyel et al., 2011), inferring that membrane repair in response to these agents also involves microparticle release. In contrast, membrane repair in response to perforin, a pore-forming agent released by cytotoxic lymphocytes, is reportedly mediated by endocytosis and resealing (Keefe et al., 2005; Thiery et al., 2010). Perhaps the smaller size of pores created by perforin compared with other pore-forming agents (Metkar et al., 2002) or the need to simultaneously deliver other proteins, such as granzymes, inside the porous cell (Thiery et al., 2010), dictates the more prominent use of these alternative repair modalities.

The mechanisms of plasma membrane damage and repair caused by intracellular death pathways such as necroptosis and pyroptosis differ from the mechanisms reported here (Andrews and Corrotte, 2018; Brito et al., 2019; Draeger et al., 2011; Idone et al., 2008). For instance, necroptosis-induced plasma membrane damage has been shown to be secondary to functional activation of the intracellular pseudokinase mixed lineage kinase domain-like pseudokinase (MLKL) (Gong et al., 2017). Likewise, although the mechanism initiating plasma membrane repair during necroptosis is still unknown, this process has been determined to be independent of TMEM16F (Gong et al., 2017). It possibly involves another scramblase activated by necroptosis in a manner analogous to Xkr8, a caspase-regulated scramblase activated by apoptosis (Suzuki et al., 2013). In contrast, however, more distal components of the repair machinery may be shared between intracellular death pathways and pore-forming agents. Along these lines, necroptosis, pyroptosis, and pore-forming agents have been shown to evoke plasma membrane repair through the ESCRT machinery, which drives scission and release of extracellular vesicles (Gong et al., 2017; Jimenez et al., 2014; Rühl et al., 2018 ).

In addition to its role in plasma membrane repair, TMEM16F has been implicated in other physiological processes involving the plasma membrane (Fujii et al., 2015; Suzuki et al., 2010; Whitlock and Hartzell, 2017; Yang et al., 2012). In particular, TMEM16F is mutated and inactivated in humans with Scott’s syndrome, a mild bleeding disorder because of defective platelet activation (Suzuki et al., 2010). TMEM16F has been shown to facilitate platelet aggregation (Yang et al., 2012). In addition, TMEM16F fosters the release of platelet extracellular vesicles, which further promote hemostasis (Fujii et al., 2015). Likewise, TMEM16F has been shown to promote membrane blebbing during P2X7 receptor-mediated apoptosis and during membrane vesiculation induced by sulfhydryl-blocking reagents such as N-ethyl maleimide (Han et al., 2019; Ousingsawat et al., 2015). Furthermore, roles of TMEM16F in multivesicular body formation and PD-1 vesicular trafficking have been suggested in T cells (Bricogne et al., 2019; Hu et al., 2016). Last, possible roles of TMEM16F in cell-to-cell communication have been reported for osteoblasts during bone development and neutrophils during joint inflammation (Ehlen et al., 2013; Headland et al., 2015). These functions of TMEM16F may involve changes in the plasma membrane akin to those described here for membrane repair.

The TMEM16 family is highly conserved in eukaryotes, including fungi, plants, and flies (Whitlock and Hartzell, 2017). Although some of the TMEM16 family members, including TMEM16F, can function as ion channels, most operate as lipid scramblases. Thus, the involvement of TMEM16 molecules in plasma membrane repair may be highly conserved across the TMEM family and across evolution. Indirect support for this idea was provided by the observation that another TMEM16 family member, TMEM16E (or ANO5), is mutated in humans with myopathies associated with abnormal muscle regeneration (Bolduc et al., 2010; Whitlock and Hartzell, 2017). Although a role of TMEM16E in a membrane repair process has not been yet documented, mice lacking TMEM16E have been shown to have compromised recovery of muscle fibers following laser damage, which can induce membrane pores (Griffin et al., 2016).

Exposure of PS because of lipid scrambling is typically viewed as an “eat me” signal, which promotes phagocytosis and elimination of dying or dead cells by macrophages. However, there is firm evidence that PS exposure can also take place when cells are healthy; for instance, during platelet activation and degranulation. Our data support the notion that PS exposure is also part of an active plasma membrane repair mechanism. Thus, a function of TMEM16F and other TMEM16 family scramblases may be to constitute a “repair me” signal aimed at protecting cells from external attacks and injuries across cell types and species.

STAR★METHODS

LEAD CONTACT AND MATERIALS AVAILABILITY

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, André Veillette (andre.veillette@ircm.qc.ca). A Material Transfer Agreement may be needed for some of the materials.

EXPERIMENTAL MODEL AND SUBJECT DETAILS

To generate the TMEM16F KO mice, a strategy similar to that of Batti et al. (2016) was used. Briefly, three genomic fragments including exon 13 of the Ano6 gene, which encodes TMEM16F, from C57BL/6 (B6) mice were cloned on either side of the neomycin cassette in the vector pJA1617 (provided by Dr. J. Drouin, IRCM) (Figure S5). DNA was then transfected into the B6 mouse embryonic stem cell line Bruce-4, and cells were selected with G418. Correctly targeted ES cells were identified by PCR and injected into blastocysts to generate chimeras. After germline transmission, mice were bred with a mouse expressing the Flpe recombinase, to remove the neomycin resistance cassette. Then, Ano6^fl/+ were bred with a mouse expressing the Cre recombinase under the control of the cytomegalovirus (CMV) promoter, to delete exon 13 and generate germline TMEM16F KO mice. Mice were genotyped using standard procedures. All the mice used in the experiments were backcrossed more than six generations to C57BL/6J (The Jackson Laboratory). For in vitro experiments, either wild-type C57BL/6 (non-littermates) or heterozygous Ano6^+/− (littermates) were used as controls. For in vivo experiments, heterozygous Ano6^+/− littermates were used as controls. Both females and males were used for our studies. Animal experimentation was approved by the Animal Care Committee of the Institut de recherches cliniques de Montréal, the Animal Care Committee of the Hôpital Maisonneuve-Rosemont Research Institute or the Animal Ethics Committee of Huazhong University of Science and Technology, and, for Canadian institutions, performed as defined by the Canadian Council of Animal Care.

METHOD DETAILS

Platelet function assays

Platelet functions were measured using the SCP1 Sonoclot Coagulation & Platelet Function Analyzer (Sienco, Inc., Guangzhou, China). In essence, this instrument monitors viscoelastic changes in blood in vitro, as the samples evolve from a liquid into a clot. It calculates three measurements: the time required to initiate clot formation (“activated clotting time”; ACT), the rate of clot formation (“clot rate”), and a quantification of platelet activation and clot retraction (referred to as “platelet function”).

Cells

Thymocytes and bone marrow neutrophils were isolated from 6- to 12-week-old mice. Thymocytes were obtained directly from thymus, by making a total cell suspension. Neutrophils were isolated from femur and tibia bone marrow, using Percoll (62%; GE Healthcare) gradient centrifugation. Typically, purity of thymocytes was over 98%, while purity of neutrophils was ~85%–90%. YAC-1 (T cell thymoma), EL-4 (T cell lymphoma) and L1210 (pre-B cell leukemia) were obtained from ATCC. RMA and RMA-S (T cell lymphoma) were obtained from Benedict Chambers (Karolinska Institute). All cells were shown to be negative for Mycoplasma, either by ATCC or by the Veillette lab.

Antibodies

For flow cytometry and cell microscopy, antibodies against the following antigens were used: CD4 (GK1.5), CD8 (53–6.7), Thy-1 (30-H12), FoxP3 (FJK-16S), B220 (RA3–6B2), CD45 (30-F11), CD48 (HM48–1), TCRβ (H57–597), NK1.1 (PK136), CD11b (M1/70) and Ly6G (1A8) were from BioLegend, except the anti-FoxP3, which was obtained from eBioscience. Monoclonal antibody against tubulin (10D8) was from Santa Cruz. For immunoblotting of β-actin, mouse monoclonal antibody C4 (from Santa Cruz) was used. AntiCD18 antibodies were obtained from Abcam (cat. No. ab53009). Rabbit polyclonal antibodies against mouse TMEM16F were generated by the Veillette laboratory, using a glutathione-S-transferase (GST) fusion protein encompassing amino-acids 847–911 of mouse TMEM16F. Antibodies against Lck, Fyn, Csk, Vav-1, SHIP-1, SHP-1 or SHP-2 were generated in the Veillette laboratory and reported previously (Chen et al., 2017; Dong et al., 2012; Wu et al., 2016).

Phosphatidylserine exposure and membrane permeability

PS exposure was detected by staining cells with annexin V (BioLegend) or LactC2-FITC (Haematologic Technologies, USA). Increase in membrane permeability was monitored by staining cells with PI, DAPI or CFSE. In brief, cells were collected and washed in phosphate-buffered saline (PBS). After pelleting at room temperature (RT), cells were treated with the indicated stimuli, in 1X annexin V-binding buffer containing annexin V-APC, with or without either PI (1 μg/ml, Sigma) or DAPI (40 nM, BioLegend). LactC2-FITC was used with or without extracellular calcium, together with either PI or DAPI. In some experiments, cells were pre-loaded with CFSE (1 μM in PBS at 37°C for 10 min), prior to incubation with the stimuli. In other experiments, permeability was monitored by adding fluorescence-labeled dextran molecules of different molecular weights (3K: D34682 or 10K:D22910, from Thermo Fisher), instead of PI or DAPI.

The following pore-forming agents or stimuli were used, at the indicated concentrations and for the indicated times: listeriolysin O (LLO; ab68200, Abcam; or PRO-320, ProSpec) at 1 nM for the times indicated in the text for thymocytes or 10 nM for the times indicated in the text for neutrophils; streptolysin O (SLO; S5265, Sigma) at 500 U/ml for 15 min; saponin (47036, Sigma) at 0.001% for 30 s; pure complement (S3269, Sigma) for 30 min; or ionomycin (I0634, Sigma) at 5 μM for 2 or 5 min. Reactions were stopped by putting samples on ice, and cells were analyzed immediately by flow cytometry.

Inhibitors

The following pharmacological inhibitors were used: latrunculin A (Lat A; 428020, Calbiochem), blebbistatin (Bleb; ab120425, Abcam), U0126 (U120; Sigma), wortmannin (Wort; 12–338, EMD Millipore), SU6656 (572635, Calbiochem), EGTA (E3889, Sigma), and BAPTA-AM (A1076, Sigma). Cells were incubated with these inhibitors for 10 min at room temperature prior to treatment with the indicated stimuli.

Flow cytometry

Flow cytometry was usually performed using a CyAn ADP analyzer (Dako). Some experiments were performed using an LSR Fortessa or a FACSCalibur analyzer (BD). Calcium fluxes were detected by flow cytometry, as previously described (Dong et al., 2009). Briefly, Indo-1-loaded thymocytes were stimulated by different pore-forming agents, and fluorescence was monitored over time using an LSR Fortessa analyzer. Calcium flux was quantified as the ratio of emission at 400 nm (bound Ca²⁺) versus emission at 475 nm (free Ca²⁺). This ratio is expressed in arbitrary units (AU).

Cell viability

LDH release were measured using the Pierce LDH Cytotoxicity Assay kit, according to the manufacturer’s instructions (Thermo Fisher Scientific, Cat. No. 88953). Briefly, for thymocytes, 5×10⁶ cells were washed with PBS at RT, pelleted and resuspended in 500 μL 1x annexin V-binding buffer containing or not LLO (2 nM; 5 min) or SLO (500 U/ml; 10 min). For neutrophils, cells (1×10⁶ in 1 ml) were treated or not with LLO (10 nM; 30 min). After the treatments, cells were centrifuged for 1 min at RT (for thymocytes) or 5 min at 4°C (neutrophils), and 50 μL of supernatant was transferred to a 96-well plate. 50 μL of the LDH reaction mix was then added, and samples were incubated at RT for 30 min. Reactions were stopped by adding 50 μL of stop buffer. Absorbance was read at 490 nm and 680 nm in a PowerWave XS plate reader (Bio-Tek). Maximum LDH release was determined by lysing all cells in lysis buffer.

Immunoblots

Methods for immunoblots were described previously (Wu et al., 2016). Briefly, total cell lysates were obtained in 1x TNE buffer (1% NP-40, 2mM EDTA, 50mM Tris-Cl, pH 8.0) supplemented with sodium vanadate (180 mg/L), sodium fluoride (50 mM), PMSF (10 mg/L), and leupeptin/apoprotein (10 mg/L). Proteins were then separated by 8% SDS-PAGE and transferred to PVDF membrane. After 1.5 hours of incubation with primary antibodies, proteins were visualized using HRP-coupled secondary antibody by ECL detection (GE Health). For immunoblotting of TMEM16F, TMEM16F was first recovered by immunoprecipitation using the rabbit anti-TMEM16F serum and then processed for immunoblotting.

Isolation of extracellular vesicles

Extracellular vesicles were isolated from cells as follows: 20×10⁶ thymocytes were treated with ionomycin (5 μM) in 1x annexin V-binding buffer at RT for 5 min, then centrifuged at 3000 rpm for 5 min to remove intact cells and nuclei. Supernatants were then recovered and centrifuged at 14000 rpm for 10 min at 4C to pellet the extracellular vesicles. Supernatants were discarded and extracellular vesicle pellets were lysed in 1x sample buffer.

Live imaging

For live cell imaging, spinning disc microscopy was used. Cells were resuspended in RPMI1640 medium containing 10% FBS plus fluorescence-coupled primary antibodies. They were then seeded onto poly-L-Lysine (Sigma-Aldrich)-coated 35 mm-imaging dishes having a glass bottom (Ibidi). After incubation for 30 min at 37C in a cell culture incubator, unattached cells were removed by washing with PBS. Samples were then placed in the confocal microscope and treated with the indicated stimuli at room temperature. Images were collected every 5 s, before and after treatment. The pore-forming reagents were used as for the flow cytometry studies. The spinning disc confocal microscope was an Axon Observer inverted microscope (Zeiss) equipped with a Yokogawa CSU-1 module. Images were analyzed by the ImageJ software (NIH), and cells with blebs (round bubble-like structures at the surface of the cell) and projections (long flexible fiber-like structures at the surface of the cell) were quantified visually.

Electron microscopy

In brief, freshly isolated thymocytes were plated onto the poly-L-lysine-coated coverslips, and then treated by LLO (1 nM), ionomycin (5 μM) or both in 1x annexin V-binding buffer for 5 min at RT. Cells were immediately fixed at RT in 2 mL of fixative solution (2.5% glutaraldehyde in 0.1 M sodium cacodylate buffer). Scanning electron microscopy (SEM) were done using a FEI Inspect F50 FE-SEM microscope with an EDAX Octane Super 60 mm² SDD and the TEAM EDS analysis system. Transmission electron microscopy (TEM) analyses were performed using a FEI Tecnai 12 BioTwin 120kV TEM microscope with an AMT XR80C CCD camera system. SEM and TEM studies were performed by the Facility for Electron Microscopy Research (FEMR) of McGill University, Montreal, Canada. Cells having different morphological alterations were quantified visually (blebs: round bubble-like structures at the surface of the cell; projections: long fiber-like structures at the surface of the cell; protrusions: large bulge-like structures at the surface of the cell; pores, apparent holes in the plasma membrane).

In vivo Listeria monocytogenes infection

Infection of mice with L. monocytogenes was performed using Biosafety Level 2 practices and containment. Briefly, L. monocytogenes (DP-L4056) was grown overnight in bacterial culture medium (BHI) containing 200 μg/ml of streptomycin. 50 μL of bacterial culture was then added to 3 mL of fresh growth medium and grown for 2 h at 37°C. Bacteria in exponential growth phase [optical density (OD)_600nm = 0.09 to 0.16] were then collected, and enumerated using the following formula: 1 OD_{600 nm} = 0.7×10⁹ CFU/ml. After washing in PBS, bacteria were resuspended at the desired concentration (normally 5×10⁴ CFU/ml), and 200 μL of bacteria was injected i.v. in the tail of each mouse. Infected mice were sacrificed at day 1 and day 3 post infection. After sacrificing the animals, liver and spleen were harvested, homogenized, diluted and plated onto BHI plates containing streptomycin. Colony-forming units (CFU)/g of tissue were determined after 24 h of growth at 37°C. Serum was collected to monitor levels of alanine transaminase activity (ALT assay kit, Abcam). Liver histology was evaluated by staining paraffin-fixed sections (prepared by the IRCM Histology Core Facility) with hematoxylin and eosin (H&E). All H&E-stained slides were scanned and analyzed for inflammatory cell infiltrates. They were blindly analyzed and independently scored by two pathologists. The total number of inflammatory foci in each section was counted and discordances were reviewed by the two pathologists. An inflammatory focus was defined as a continuous aggregate of inflammatory cells (lymphocytes, histiocytes, neutrophils and plasma cells) with adjacent hepatocytic injury. The latter included hepatocyte collapse, eosinophilic or apoptotic changes, and architectural breakdown. Each inflammatory focus was graded according to the number of inflammatory cells. Grade I was defined as 10–50 cells, Grade II as 51–150 cells, and Grade III as more than 150 cells. The final count was standardized to 1 cm² according to the total surface area of each liver section, as calculated by the LesionMeter application (LesionMeter Team).

QUANTIFICATION AND STATISTICAL ANALYSIS

GraphPad Prism software was used for the statistical analyses. The statistical details for each experiment can be found in the figure legends. All experiments were repeated at least two times, usually 3 or more, as specified in the legends. n represents the number of animals used, or the number of pictures quantified, as indicated in the legends. Data are means ± s.e.m. Statistical analyses were performed using Student’s t test (two-sided). P values of less than 0.05 were considered to be statistically significant.

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
APC-Annexin V	Biolegend	Cat# 640941
CD4-FITC (GK1.5)	Biolegend	Cat# 100406
CD8-PECy7 (53–6.7)	Biolegend	Cat# 100722
Thy-1-A488	Biolegend	Cat# 105316
PE anti-Foxp3 (FJK-16S)	eBioscience	Cat# 12-5773-82
FITC anti-B220 (RA3–6B2)	Biolegend	Cat# 103206
PE anti-CD45 (30-F11)	Biolegend	Cat# 103106
PE anti-CD48 (HM48-1)	Biolegend	Cat# 103406
Pacific blue anti-TCRβ (H57–597)	Biolegend	Cat# 109226
PE-Cy7 anti-NK1.1 (PK136)	Biolegend	Cat# 108714
PE anti-CD11b (M1/70)	Biolegend	Cat# 101208
A647 anti-Ly6G (1A8)	Biolegend	Cat# 127610
Anti-tubulin (10D8)	Santa Cruz	Cat# sc-53646
Anti-CD18	Abcam	Cat# ab53009
Anti-actin (C4)	Santa Cruz	Cat# sc-47778
FITC-LactC2	Haematologic Technologies	Cat# BLAC-FITC
Polyclonal Anti-Lck	Veillette lab	N/A
Polyclonal Anti-Fyn	Veillette lab	N/A
Polyclonal Anti-Csk	Veillette lab	N/A
Polyclonal Anti-Vav-1	Veillette lab	N/A
Polyclonal Anti-SHIP-1	Veillette lab	N/A
Polyclonal Anti-SHP-1	Veillette lab	N/A
Polyclonal Anti-SHP-2	Veillette lab	N/A
Polyclonal Anti-TMEM16F	This paper	N/A
Bacterial and Virus Strains
L. monocytogenes (DP-L4056)	Daniel A. Portnoy	N/A
Chemicals, Peptides, and Recombinant Proteins
Percoll	GE Healthcare	Cat# 17089101
Propidium iodide	Sigma	Cat# P4170
DAPI	Biolegend	Cat# 422801
CellTrace CFSE Cell Proliferation Kit	Thermo Fisher	Cat# C34554
Dextran, Alexa Fluo 488; 3,000 MW, Anionic	Thermo Fisher	Cat# D34682
Dextran, Alexa Fluo 488; 10,000 MW, Anionic	Thermo Fisher	Cat# D22910
Listeriolysin O (LLO)	Abcam	Cat# ab68200
Listeriolysin O (LLO)	ProSpec	Cat# PRO-320
Streptolysin O (SLO)	Sigma	Cat# S5265
Saponin	Sigma	Cat# 47036
Pure complement	Sigma	Cat# S3269
Ionomycin	Sigma	Cat# I0634
Latrunculin A	Calbiochem	Cat# 428020
Blebbistatin	Abcam	Cat# ab120425
U0126	Sigma	Cat# U120
Wortmannin	EMD Millipore	Cat# 12–338
SU6656	Calbiochem	Cat# 572635
EGTA	Sigma	Cat# E3889
BAPTA-AM	Sigma	Cat# A1076
Indo-1	Thermo Fisher	Cat# I1226
Critical Commercial Assays
Pierce LDH Cytotoxicity Assay kit	Thermo Fisher	Cat# 88954
ALT assay kit	Abcam	Cat# ab105134
Experimental Models: Cell Lines
YAC-1	ATCC	TIB-160
EL-4	ATCC	TIB-39
L1210	ATCC	CCL-219
RMA	Karolinska Institute	Dr. Benedict Chambers
RMA-S	Karolinska Institute	Dr. Benedict Chambers
Experimental Models: Organisms/Strains
Mouse: TMEM16F KO:C57BL/6J	This paper	N/A
Mouse: B6.C-Tg(CMV-cre)1Cgn/J	The Jackson Laboratory	JAX: 006054
Software and Algorithms
Prism software V5.0	GraphPad Software	https://www.graphpad.com
ImageJ	NIH	https://imagej.nih.gov/ij/
FlowJo V9 and V10	FlowJo, LLC	https://www.flowjo.com

Highlights.

• Pore-forming agents evoke calcium-dependent lipid scrambling in the plasma membrane

• This effect promotes plasma membrane repair and is mediated by lipid scramblase TMEM16F

• TMEM16F induces repair by facilitating release of extracellular vesicles

• TMEM16F-deficient mice display greater susceptibility to Listeria monocytogenes