Activation of the phospholipid scramblase TMEM16F by nanosecond pulsed electric fields (nsPEF) facilitates its diverse cytophysiological effects

Nanosecond pulsed electric fields (nsPEF) are emerging as a novel modality for cell stimulation and tissue ablation. However, the downstream protein effectors responsible for nsPEF bioeffects remain to be established. Here we demonstrate that nsPEF activate TMEM16F (or Anoctamin 6), a protein functioning as a Ca2+-dependent phospholipid scramblase and Ca2+-activated chloride channel. Using confocal microscopy and patch clamp recordings, we investigated the relevance of TMEM16F activation for several bioeffects triggered by nsPEF, including phosphatidylserine (PS) externalization, nanopore-conducted currents, membrane blebbing, and cell death. In HEK 293 cells treated with a single 300-ns pulse of 25.5 kV/cm, Tmem16f expression knockdown and TMEM16F-specific inhibition decreased nsPEF-induced PS exposure by 49 and 42%, respectively. Moreover, the Tmem16f silencing significantly decreased Ca2+-dependent chloride channel currents activated in response to the nanoporation. Tmem16f expression also affected nsPEF-induced cell blebbing, with only 20% of the silenced cells developing blebs compared with 53% of the control cells. This inhibition of cellular blebbing correlated with a 25% decrease in cytosolic free Ca2+ transient at 30 s after nanoporation. Finally, in TMEM16F-overexpressing cells, a train of 120 pulses (300 ns, 20 Hz, 6 kV/cm) decreased cell survival to 34% compared with 51% in control cells (*, p < 0.01). Taken together, these results indicate that TMEM16F activation by nanoporation mediates and enhances the diverse cellular effects of nsPEF.

Permeabilization of biological membranes by high voltage pulsed electric fields (PEF), 2 known as electroporation, is a versatile technique in biomedicine used for intracellular delivery of drugs, plasmid DNA, and siRNA, as well as for tissue and tumor ablation (1). More recently, the pulse duration has been shortened into the nanosecond range. The effects of such short pulses have been shown to reach into the cell interior (2,3), thereby giving rise to specific biological effects and opening of new opportunities to control cell function.
The best known primary effect of nanosecond pulsed electric fields (nsPEF) is the permeabilization of membranes including the plasma membrane, endoplasmic reticulum, and mitochondria (2)(3)(4)(5)(6). nsPEF triggers the formation of pores of nanometer size (nanopores) on the cell membrane (7)(8)(9). Nanopores lifetime are in the order of seconds or minutes (7,8,10). We previously reported that nanopores are not "plain holes" in the plasma membrane such as larger pores generated by longer pulses. They have peculiar conductive properties analogous to ion channels, such as voltage sensitivity, current rectification, and ion selectivity (8,11,12). The inward rectification, which is regarded by some authors as a hallmark of nanopore formation (12), is manifested by a non-linear enhancement of the inward current at the most negative membrane potentials in nsPEFtreated cells. Nanopores display preferential permeability to cations (8,11,12), which is consistent with expectations that they are lined up with negatively charged hydrophilic phosphate groups of the membrane phospholipids (13).
Nanoporation initiates a plethora of downstream physiological changes including rapid phosphatidylserine (PS) externalization (13,14), cytoskeleton disassembly (15)(16)(17)(18), cell swelling and blebbing (19 -21), modulation of endogenous ion channels (22,23), and necrosis and apoptosis (24 -26). The diversity of these effects suggests that plasma membrane nanoporation may be the initial step in a complex chain of signaling cascades leading to deregulation of multiple physiological functions. Indeed, nanopore opening causes an immediate rise in the intracellular free Ca 2ϩ concentration. Ca 2ϩ is a versatile signal that triggers processes as diverse as cell motility, gene transcription, muscle contraction, and membrane trafficking (27). Cells may interpret nsPEF-induced Ca 2ϩ transients as authentic Ca 2ϩ signals and amplify them by the Ca 2ϩ -induced Ca 2ϩ -release process (3,4,6,28,29).
Among proteins directly modulated by Ca 2ϩ , the TMEM16 family of membrane proteins, also known as anoctamins, plays key roles in a variety of physiological functions that range from ion transport, to phospholipid scrambling and to regulating other ion channels (30 -32). In particular, TMEM16F, or Anoctamin 6 (ANO6), was found to be an essential effector for the Ca 2ϩ -dependent exposure of PS on the cell surface (33). Suzuki and colleagues (33) reported that knockdown of Tmem16f in the Ba/F 3 cells decreases the rate of PS externalization induced by the Ca 2ϩ ionophore, whereas TMEM16F overexpression strongly enhances PS exposure. Sequencing of chromosomal DNA also demonstrates that Scott's syndrome patients carry loss-of-function mutations in the gene encoding TMEM16F (33,34). Scott's syndrome is a congenital bleeding disorder caused by the loss of Ca 2ϩ -dependent PS exposure (29). This syndrome is accompanied by other cellular defects such as impaired bleb formation in platelets and absence of microvesicles shedding in both platelets and erythrocytes (35,36).
Following the discovery that TMEM16F is a Ca 2ϩ -dependent scramblase defective in patients with Scott's syndrome, Yang et al. (37) proposed that the increase in Ca 2ϩ required to induce PS externalization depends on the Ca 2ϩ -permeable non-selective cation channel activity of TMEM16F itself. The properties of ion channels associated with TMEM16F are a matter of debate. In addition to a non-selective cation channel activity, TMEM16F has been reported to function as a swelling-activated Cl Ϫ (38), outwardly rectifying Cl Ϫ (39), and Ca 2ϩ -activated Cl Ϫ channel (40 -43). These results have been obtained under different experimental conditions, which may explain some of these differences. Overall the data support the idea that TMEM16F has a non-selective pore or, as proposed by Whitlock and Hartzell (44), the existence of multiple open conformations with different ion permeability.
There is also controversy regarding the link between ion channel and scramblase activity of TMEM16F. It is not known if anions, cations, and phospholipids move through one pore, different pores formed by TMEM16F dimers, or whether additional accessory proteins are required. Whitlock and Hartzell (44) recently proposed that TMEM16F-mediated PS externalization is associated with leakage of ions through the lipid scrambling pathway between the protein and the scrambling lipid head groups.
TMEM16F activation has also been associated with programmed cell death. Several studies reported that TMEM16F is activated during apoptosis (39,40,45). Martins and colleagues (39) found that cell death induced by staurosporine in Jurkat cells was reduced in Tmem16f-silenced cells. They also showed that TMEM16F facilitates apoptotic cell shrinkage (39).
Because TMEM16F functions closely mirror nsPEF-induced bioeffects, in the present study we asked whether TMEM16F is a downstream effector of nanoporation. Using confocal microscopy and whole-cell patch clamp we show that TMEM16F expression affects PS externalization, cell blebbing, calcium influx, nanopore currents, and cell death induced by nsPEF. These results demonstrate that TMEM16F is activated in response to nsPEF, and that TMEM16F activation facilitates diverse cellular effects downstream from nanoporation.

TMEM16F activation contributes to nsPEF-induced phosphatidylserine externalization
Fig . 1A shows that a single 300-ns pulse (25.5 kV/cm) causes a sustained externalization of PS on the plasma membrane of HEK 293 cells. Similarly to what has been already reported (46), we found that PS appears more at the anode-facing pole of the cell and within minutes distributes uniformly on the cell membrane. Externalization of PS can occur through the activation of scramblases, which are Ca 2ϩ -dependent. Therefore, to gain insight into the mechanism involved in PS exposure after nsPEF, we studied its dependence on extracellular Ca 2ϩ . HEK 293 cells were treated with nsPEF in bath solutions containing either 2 or 0 mM Ca 2ϩ and PSexternalizationwasmeasuredovertimeusingthecalcium-independent binding with FITC-tagged Lactadherin. Fig. 1B shows that in the absence of Ca 2ϩ , one 300-ns pulse (25.5 kV/cm) did not trigger PS externalization pointing at the involvement of a Ca 2ϩdependent scramblase activity. The identification of TMEM16F as an essential component for the Ca 2ϩ -dependent exposure of PS on the cell surface (33) prompted us to investigate whether this scramblase is relevant for nsPEF-induced PS externalization. To block TMEM16F scramblase activity we used the Ca 2ϩ -activated chloride channel inhibitor-AO1 (CaCCinh-AO-1). This inhibitor was reported to block both the Ca 2ϩ -dependent Cl Ϫ channel and scramblase activities (47). HEK 293 cells were pre-treated with 25 M CaCCinh-AO-1 for 5 min in the presence of Cy5-labeled Annexin V. In control samples, cells where preincubated with the same dilution of the inhibitor vehicle DMSO. After this incubation time, a single 300-ns pulse (25.5 kV/cm) was delivered to a small group of cells and the PS exposure on the cell surface was monitored by confocal microscopy. At 8 min post-nsPEF, blocking TMEM16F decreased nsPEF-induced PS exposure by 42% as compared with control samples (Fig. 1C).
To further prove the involvement of TMEM16F in nsPEF-induced PS externalization, the expression of the endogenous Tmem16f in HEK 293 was knocked down by specific siRNA. As shown in Fig. 1D, transfecting two different Tmem16f siRNA (siRNA Tmem16f-I and -II) decreased the expression level of the messenger RNA (mRNA) to 26 and 11% of that in the siRNA control cells, respectively. The rate of nsPEF-induced PS exposure was decreased in both siRNA Tmem16f transformants (Fig. 1E). At 5.5 min after treatment, PS externalization in siRNA Tmem16f-I and -II cells was reduced by 33 and 49%, respectively. Notably, the impairment in the scramblase activity efficiency correlated with the level of Tmem16f silencing.
To summarize, we found that nsPEF-induced PS externalization is dependent on extracellular Ca 2ϩ , and that both chemical and expression block of TMEM16F reduced PS exposure. Our results suggest that Ca 2ϩ coming through nanopores activates TMEM16F.

Tmem16f knockdown affects both nsPEF-induced membrane blebbing and rise in cytosolic free Ca 2؉
In addition to the PS externalization, nsPEF initiates rapid membrane blebbing (20,21). Morphologically these blebs resemble pseudopods and exhibit a fast directional growth toward the anode (20,21). Fig. 2A shows that nsPEF induced pseudopod-like blebs in cells transfected with the control siRNA, whereas knockdown of Tmem16f gene expression significantly impaired this response. After nanoporation, only 20% of silenced cells developed blebs compared with 53% of the control ones (Fig. 2B, left graph). Moreover, among the actively blebbing cells, Tmem16f expression affected the development of the blebs with the average bleb length for silenced cells decreased to 9.1 Ϯ 0.9 m compared with 12.9 Ϯ 0.8 m in control cells (p Ͻ 0.01, Fig. 2B, right graph). Notably, membrane blebbing in response to P2X purinergic receptor 7 (P2X 7 R) activation was found to be dependent on Tmem16f expression (48) and cell bleb formation is compromised in platelets from Scott's syndrome patients (35,36).
A downstream effect of nanoporation by nsPEF is the transient mobilization of calcium ions that involves both Ca 2ϩ uptake from the outside and release from the endoplasmic reticulum (3,4,49,50). A rise in cytosolic free calcium [Ca 2ϩ ] i has been shown to have a pivotal role in bleb formation (51)(52)(53). We therefore asked whether the defects in membrane blebbing seen in Tmem16f-silenced cells correlated with an impairment in nsPEF-induced rise of [Ca 2ϩ ] i . In the presence of 2 mM external Ca 2ϩ , HEK 293 cells responded to one 300-ns pulse, 25.5 kV/cm with an immediate rise in [Ca 2ϩ ] i (Fig. 2C), which lasted for minutes after the treatment. In silenced cells, the shape of the transients was similar to the control cell ones but reduced by 25%.
These results show that TMEM16F serves as a Ca 2ϩ -permeable channel in nsPEF-treated cells. Both calcium influx and cell blebbing, two characteristic immediate responses to nsPEF, are reduced in Tmem16f-silenced cells.

Interplay between nanopore-and TMEM16F-conducted currents
Next, we explored if nsPEF increases Cl Ϫ permeability of the cell membrane, and whether it is mediated by Ca 2ϩ -mediated TMEM16F activation. In Fig. 3A, the whole-cell configuration was established 2 min prior the delivery of one 300-ns pulse at 4.2 kV/cm. The current-voltage (I-V) dependence was measured by applying the same voltage step protocol (Ϫ100 to ϩ80 mV) at 30 s prior to nsPEF and again at 60 s after it. nsPEF increased the membrane conductance at both positive and negative potentials (Fig. 3A, left I-V curve). Substitution of the bath NaCl with sodium gluconate reduced the Cl Ϫ concentration from 152.4 to 12.4 mM, thereby attenuating currents at the most positive membrane potentials with little or no impact on inward currents at negative potentials (Fig. 3A, right I-V curve). This result shows that nsPEF induces Cl Ϫ currents. In principle these currents can flow through both nanopores and endogenous ion channels (including Ca 2ϩ -dependent TMEM16F channel), although our previous studies showed that nanopores are cation-selective, with about 10-fold higher conductivity to K ϩ than to Cl Ϫ (12). To separate between Ca 2ϩ -activated Cl Ϫ current (presumably through TMEMF16) and other pathways of Cl Ϫ influx, we varied the Ca 2ϩ chelator load into the pipette. , and on TMEM16F activity (C-E). HEK 293 were treated with a single 300-ns 25.5 kV/cm pulse, and PS externalization was monitored by time lapse confocal microscopy using either Cy5-labeled Annexin V (A, C, and E) or lactadherin-FITC (B). The nsPEF treatments were done at 87 s into the experiment (black arrows), after acquiring 3 pre-exposure images as a baseline. Panel A shows for a group of cells treated with nsPEF, DIC and Annexin V fluorescence images taken at the indicated time points. Panel B shows the lack of PS externalization in the absence of extracellular Ca 2ϩ . Cells were treated with nsPEF in either 2 or 0 mM CaCl 2 bath solutions. DIC and lactadherin-FITC fluorescence images were taken at the indicated time points. In panel C, the scramblase activity of TMEM16F was blocked by incubation of cells with 25 M CaCCinh-AO-1 for 5 min. In control samples, the inhibitor vehicle DMSO was diluted the same way. Sham-exposed cells were not exposed to nsPEF but subjected to all the same manipulations. In panel D cells were transfected with two siRNA targeting Tmem16f transcripts (siRNA Tmem16f-I and -II) or scrambled siRNA and specific silencing was verified by real-time quantitative PCR. Tmem16f mRNA levels were normalized to b-act mRNA and are shown as relative expression. Panel E shows that silencing Tmem16f with two different siRNA sequences, Tmem16f-I (left) and -II (right), reduced the Annexin V emission curve in response to nsPEF. Mean Ϯ S.E.

Nanosecond pulsed electric fields activate TMEM16F
With 2 mM extracellular Ca 2ϩ , the cell cytoplasm was dialyzed using pipette solutions containing either 0.5 or 5 mM EGTA, or 5 mM BAPTA. At 30 s after one 300-ns pulse at 4.2 kV/cm, the I-V curves from cells dialyzed with the three solutions were very similar (Fig. 3B), suggesting that nanoporation was not affected by the availability of cytosolic free Ca 2ϩ . This situation, however, changed over time: at 150 s after nsPEF, the current at the most positive membrane potentials in 0.5 mM EGTA cells was similar to the one at 30 s, whereas in 5 mM EGTA cells was significantly reduced. This difference became even more pronounced when using BAPTA which, compared with EGTA, has much faster Ca 2ϩ -binding kinetics (54). In the subsequent experiments, we proved that this Cl Ϫ current was indeed dependent on Tmem16f expression (Fig. 4). Using a pipette solution containing 0.5 mM EGTA, at 30 s after treatment with a single 300-ns pulse we recorded in both control and Tmem16fsilenced cells similar I-V curves. However, at 270 s after nsPEF, whereas the inward currents were still very similar, the outwardly ionic current of silenced cells was significantly reduced compared with control cells. These results show that the TMEMF16-mediated Cl Ϫ conductance is activated by nsPEF in a Ca 2ϩ -dependent manner with a few minutes delay after nsPEF exposure.

TMEM16F overexpression increases nsPEF cytotoxicity
The cytotoxicity of nsPEF have been demonstrated in multiple cell types in vitro (24,25,(55)(56)(57)(58). Because TMEM16F acti-vation by nsPEF exacerbates cellular responses that have been linked to cell death such as PS externalization and cell blebbing (59), we anticipated that TMEM16F overexpression would increase nsPEF cytotoxity. First we studied the sensitivity to nsPEF of HEK 293 exposed in suspension in electroporation cuvettes. Fig. 5A shows the effect of the increasing pulse number on cell survival, whereas other treatment parameters were fixed (20 Hz, 300-ns pulse duration, 6 kV/cm). A train of 120 pulses, which caused a 45% decrease of viability, was chosen to study the effect of TMEM16F overexpression on cell survival after nsPEF. Cells were exposed to nsPEF 48 h after transfection with either TMEM16F or a control construct. TMEM16F overexpression dropped cell survival to 34% compared with 51% in control samples (Fig. 5C). Considering that, on average, only about 50% of cells overexpressed the construct (Fig. 5B), the survival in the overexpressing cell subpopulation was estimated to be at ϳ17%.

Discussion
In the present study we show that TMEM16F, a protein that functions as a Ca 2ϩ -dependent phospholipid scramblase and Ca 2ϩ -activated Cl Ϫ channel, is a downstream effector of nsPEF.
In light of our results, a key question is how nsPEF activates TMEM16F. Nanopores can remain open for minutes before being resealed (7,8,10), thus leading to rapid influx of Ca 2ϩ , ionic imbalance, and cell swelling (24,60). Notably, both intracellular Ca 2ϩ elevation and cell swelling have been linked to

Nanosecond pulsed electric fields activate TMEM16F
TMEM16F activation. Ionomycin, P2Y 2 receptors activation, and direct elevation of cytosolic Ca 2ϩ using specific patch pipette solutions as well as hypotonic bath solutions all activate TMEM16F-dependent Cl Ϫ currents (38,40,41,48,61). Our results show that Ca 2ϩ coming through nanopores is sufficient to activate TMEM16F. We found that depletion of the extracellular Ca 2ϩ completely abolished PS exposure in nanoporated cells and that the cytosolic chelation of Ca 2ϩ coming through nanopores reduced TMEM16F-mediated Cl Ϫ conductance. The relevance of the osmotic cell swelling for TMEM16F channel activation in nanoporated cells has yet to be explored.
Permeabilization of the cell membrane by nsPEF causes rapid PS externalization (13,14). Results from both experiments and molecular dynamics simulations support the idea that nsPEF-induced PS translocation is a result of an electrophoretic migration of the negatively charged PS head group alongside the lipid-water interface of nanopores (62-64). Our results point, for the first time, to the involvement of activated scramblase in nsPEFinduced PS externalization. Both chemical inactivation of TMEM16F and gene silencing significantly reduced nsPEF-induced PS exposure. We also show that PS externalization is abolished when cells are treated with nsPEF in the absence of extracellular Ca 2ϩ . These results suggest that PS exposure in nanoporated cells happens exclusively in a scramblase-dependent manner. However, Ca 2ϩ affects the plasma membrane electroporation itself. We previously reported that pore formation and evolution are dependent on external Ca 2ϩ (65). In particular, we found that Ca 2ϩ first stabilizes the pores but facilitates their delayed expansion (65). Therefore one should consider that, in addition to activating TMEM16F, in nanoporated cells Ca 2ϩ may affect PS exposure by modulating the structure and stability of the pores. Indeed, direct blockage of TMEM16F did not abolish PS externalization suggesting that both scramblase activation and lateral drift throughout nanopores may drive nsPEF-induced phospholipid perturbation. The balance between these two mechanisms may depend on the pulse parameters and on whether the influx of Ca 2ϩ through the nanopores is sufficient to stimulate scramblase activity.
We also found that TMEM16F gene silencing markedly reduced cell blebbing occurring in response to nsPEF. Blebs are a result of actomyosin contraction, which causes transient detachment of the cell membrane from the actin cortex. Their expansion is rapid and dependent on actin polymerization (66). Cortical contractility is associated with cyclic intracellular calcium transients (67,68). Although it has been reported that Scott's syndrome patients have cell blebbing and microvesicles shedding defects, the mechanisms by which TMEM16F controls these processes are unknown. One can speculate that the TMEM16F non-selective cation channel activity is responsible for the rise in cytosolic free Ca 2ϩ crucial for bleb formation. Indeed, we found a 25% calcium transient reduction in Tmem16f-silenced cells. However, whether this difference is enough to explain the blebbing impairment seen in silenced cells, as well as the direct role of the TMEM16F cation channel activity needs to be further investigated.
To assess the activation of the TMEM16F channel, we focused on the Cl Ϫ currents. Using both overexpression and silencing approaches, several groups showed that TMEM16F can conduct Cl Ϫ currents in response to the elevation of the intracellular Ca 2ϩ concentration (40 -43, 69).
nsPEF caused a long-lasting increase of cell membrane conductance, both at positive and negative membrane potentials. Removal of the extracellular Cl Ϫ , chelating of the cytosolic Ca 2ϩ , and changing Tmem16f expression affected the current at the positive but not negative membrane potentials. Considering our earlier data that nanopore currents are maximized at negative membrane potentials (8,11,12), and that nanopores are strongly selective for cations (8,11,12), these new findings provide strong evidence that nsPEF activates TMEMF16 function as a Cl Ϫ channel. Moreover, this activation was observed minutes after the  TMEM16F overexpression on HEK 293 cell death (B and C). A, cells were treated with increasing numbers of 300-ns pulses (20 Hz at 6 kV/cm). Exposures were performed in 1-mm gap electroporation cuvettes and cell survival was assessed in 24 h. Results are expressed in % to sham-exposed parallel control at 24 h. Mean Ϯ S.E. for 9 independent experiments. B, expression level of TMEM16F-3xFLAG analyzed by Western blot and FACS analyses at 48 h after transfection. Extracts of cells were immunoblotted with an anti-FLAG and, as a control, with ␤-actin antibody. TMEM16F is seen as a broad band around 120 kDa. To measure the % of TMEM16F-positive cells, control and overexpressing cells were permeabilized, stained with an anti-FLAG antibody, and analyzed by FACS. Cells in the gated area were considered positive for TMEM16F expression. C, overexpression of TMEM16F increases the cytotoxicity of exposure to 120 pulses (300-ns width, 6 kV/cm, 20 Hz). The survival was measured at 24 h after nsPEF exposure and expressed in % to sham-exposed parallel control. Mean Ϯ S.E. for 6 independent experiments. *, p Ͻ 0.001.

Nanosecond pulsed electric fields activate TMEM16F
mobilization of cytosolic Ca 2ϩ , which is consistent with earlier reports that Ca 2ϩ -dependent Cl Ϫ current through TMEM16F channel takes minutes to develop (41,43,69).
TMEM16F is not the first ion channel found to respond to nsPEF. Indeed, several groups including ours reported activation of voltage-gated calcium channels in response to nsPEF as well as a long lasting inactivation of both voltage-gated calcium channels and voltage-gated sodium channels by an unknown mechanism (23, 29, 70 -73).
Activation of ion channels is an important step for regulated cell death (74). Our data show that nsPEF cytotoxicity is increased when TMEM16F is overexpressed in nanoporated cells. Depending on the pulse parameters, exposure method, and cell type used, in vitro studies showed that nsPEF can activate apoptosis, necrosis, a combination of these mechanisms, and autophagy (24,55,75). Although it is not uncommon that a death stimulus activates more than one cell death pathway, it is not clear how nsPEF does it. Notably, TMEM16F was reported to be activated during both apoptosis and necroptosis, a programmed form of necrosis (39,40,45,76). Activation of TMEM16F by a mild increase of intracellular Ca 2ϩ induces Cl Ϫ -selective currents, cell shrinkage, and apoptosis (39,40,45). At strong activation, e.g. by stimulation of purinergic P2Y 7 receptors, TMEM16F participates in pore formation, massive membrane blebbing, cell swelling, and necroptotic cell death (48). Based on these findings, one could expect that the mechanism and level of TMEM16F activation may influence the cell death mode in nsPEF-treated cells.
Taken together, we demonstrate that TMEM16F is an important regulator of the cellular effects of downstream nanoporation. Our results provide new insight into the mechanisms of action of nsPEF stimulation.

Cell culture, chemicals, and solutions
HEK 293 cells were kindly provided by Dr. John Catravas (Old Dominion University, Frank Reidy Research Center for Bioelectrics, Norfolk, VA). Cells were cultured in Eagle's minimum essential medium with L-glutamine (ATCC), supplemented with 10% (v/v) fetal bovine serum (certified OneShot format, FBS-OneShot, Life Technologies). One day prior to the experiments, cells were passaged and transferred onto glass coverslips (number 0 thickness, 10 mm diameter) pretreated with 20 ng/ml of Laminin (Sigma) to improve cell adhesion.
The composition of solutions utilized in different experiments is described below. Chemicals were purchased from Sigma and Life Technologies. The osmolality of the solutions was between 290 and 310 milliosmole/kg, as measured by a freezing point microosmometer (Advanced Instruments, Inc., Norwood, MA). All experiments were performed at room temperature (22 Ϯ 2°C).

nsPEF exposure methods
nsPEF was delivered to cells either attached to glass coverslips or in suspension in electroporation cuvettes. Both methods have been previously described (8,77). For cells on glass coverslips, nearly rectangular 300-ns pulses were generated by a Mosfet-switch-based circuit upon delivery of a TTL trigger pulse from pClamp software via a Digidata 1322A output (Molecular Devices). The same software and Digidata output were used to synchronize nsPEF exposure, image acquisition, and whole-cell recordings. nsPEF were delivered to a selected cell or small group of cells with a pair of tungsten rod electrodes (100 m diameter, 140 m gap). The electrodes were positioned precisely 30 m or, for the electrophysiology experiments, 40 m above the coverslip using a robotic manipulator (MP-225, Sutter Instruments, Novato, CA). In most experiments we delivered one 300-ns pulse at 25.5 kV/cm; for electrophysiology experiments pulse intensity was reduced to 4.25 kV/cm. The E field at the cell location was calculated as described previously (78), by 3D numerical simulations using a finite element analysis software COMSOL Multiphysics, release 5.0 (COMSOL Inc., Stockholm, Sweden).
For cells in suspension, trains of trapezoidal pulses of 300 ns duration (20 Hz, 6 kV/cm) from an AVTECH AVOZ-D2-B-ODA generator (AVTECH Electrosystems, Ottawa, Ontario, Canada) were delivered to 1-mm gap electroporation cuvettes (BioSmith, San Diego, CA) via a 50-to 10-ohm transition module (AVOZ-D2-T, AVTECH Electrosystems) modified into a cuvette holder. To produce pulse trains of predetermined duration at selected repetition rates, the generator was triggered externally from a model S8800 stimulator (Grass Instrument Co., Quincy, MA). The pulse amplitude and shape were monitored in all experiments using a 500 MHz, 5 GS/s TDS 3052B oscilloscope (Tektronix, Wilsonville, OR).

PS externalization and calcium imaging
To study both PS externalization and Ca 2ϩ influx in response to nsPEF, a coverslip with cells was placed in a glass-bottomed chamber (Warner Instruments, Hamden, CT) mounted on an Olympus IX81 inverted microscope equipped with an FV 1000 confocal laser scanning system (Olympus America, Center Valley, PA). The chamber was filled with a physiological solution containing (in mM): 140 NaCl, 5.4 KCl, 1.5 MgCl 2 , 2 CaCl 2 , 5 HEPES, 10 glucose (pH 7.4 with NaOH).
To detect PS externalization we used either a Ca 2ϩ -dependent Cy5-labeled Annexin V (BD Biosciences, San Jose, CA) or a Ca 2ϩ -independent bovine lactadherin-FITC (Hematologic Technologies, Essex, VT). Cy5-labeled Annexin V was diluted 1:100 in the physiological solution. Lactadherin-FITC was used at 0.25 g/ml in physiological solutions containing either 2 mM CaCl 2 or 0 mM CaCl 2 and 2 mM EGTA. To block TMEM16F, we used CaCCinh-AO-1 from Sigma. The inhibitor was resuspended in 50 mM DMSO and diluted 1:2000 to a final concentration of 25 M. In control samples DMSO was diluted the same way. Differential interference contrast (DIC) and fluorescent images were taken with a ϫ40, NA 0.95 dry objective as a time series (1 image/30 s) beginning 87 s before nsPEF exposure and continuing for up to 8 min after. Cy5-labeled Annexin V was excited with a red laser (635 nm) and the emission of the dye was detected between 655 and 755 nm. Lactadherin-FITC was excited with a blue laser (488 nm) and the emission of the dye was detected between 505 and 605 nm.
The cytosolic Ca 2ϩ was monitored by fluorescence imaging with Fluo-4 (Invitrogen). To load the cells with the dye, the coverslips were incubated for 15 min in the physiological solution containing 5 M Fluo-4/AM and 0.02% of Pluronic F-127 Nanosecond pulsed electric fields activate TMEM16F (Life Technologies), in the dark at room temperature. The coverslips were rinsed twice and then left for 15 min in the physiological solution before being transferred into the glass-bottomed chamber.
DIC and fluorescence images were taken as a time series (1 image/10 s) starting at 27 s before nsPEF treatment and continuing for 4.8 min. Fluo-4 was excited with a green laser (488 nm) and the emission of the dye was detected between 505 and 605 nm. Images were quantified using MetaMorph Advanced version 7.7.0.0 (Molecular Devices, Foster City, CA).
HEK 293 cells plated on coverslips were placed in the same microscope setup as was used for cell imaging. In most experiments the bath solution contained (in mM): 140 NaCl, 5.4 KCl, 1.5 MgCl 2 , 2 CaCl 2 , 5 HEPES, and 10 glucose. For experiments described in Fig. 3A, 140 mM NaCl was replaced with 140 mM sodium gluconate. The pipette solution contained (in mM): 140 cesium acetate, 4 MgCl 2 , 10 HEPES and either 0.5 (Figs. 3B and 4) or 5 cesium-EGTA (Fig. 3, A and B), or 5 cesium-BAPTA (Fig. 3B). In the experiment described in Fig. 3B, we blocked the 2 mM Ca 2ϩ coming through the nanopores by using pipette solutions with different Ca 2ϩ buffering efficiency. We used a solution with low Ca 2ϩ binding capacity, 0.5 mM EGTA (1.5 mM free Ca 2ϩ ), and two solutions with comparable high Ca 2ϩ binding capacity, 5 mM EGTA (60 mM free Ca 2ϩ ) and 5 mM BAPTA (160 mM free Ca 2ϩ ), but different binding kinetics. BAPTA, compared with EGTA, has much faster Ca 2ϩ -binding kinetics (54). The free calcium concentration was calculated with Ca/Mg, ATP/EGTA Calculator version 2.2b (web.stanford.edu/ϳcpatton/webmaxcS.htm). 3 The pH of both bath and pipette solutions was adjusted to 7.4.
Two minutes after the whole-cell configuration was established, membrane currents were recorded at specific times before and after nsPEF exposure by applying the same voltagestep protocol (200 ms steps from Ϫ100 to 70 mV in 10-mV increments); the holding potential between the sweeps was set at Ϫ70 mV. Data were collected using a Multiclamp 700B amplifier, Digidata 1322A A-D converter, and pCLAMP 10 software (Molecular Devices). All voltage values reported in the graphs were corrected for respective junction potentials.

Tmem16f expression knockdown and overexpression
To silence Tmem16f expression, we used two different stealth siRNA: HSS153251 (siRNA Tmem16f-I) and HSS176378 (siRNA Tmem16f-II) (Invitrogen). To transfect the siRNA we used Lipofectamine RNAiMAX (Invitrogen) according to the manufacturer's instructions. Control samples were transfected with stealth siRNA negative control medium GC duplex (Invitrogen).
To overexpress TMEM16F, we used a mouse TMEM16F-3xFLAG (80) construct kindly provided by Dr. Criss Hartzell (Emory University School of Medicine, Department of Cell Biology, Atlanta, GA). The construct was transfected in HEK 293 using Lipofectamine 2000 (Invitrogen) according to the manufacturer's instructions. Control samples were treated the same way and transfected with an empty control vector. Experiments with both silenced and overexpressing cells were done at 48 h post-transfection.

RT-PCR, Western blot, and FACS analyses
Total RNA was extracted from ϳ5 ϫ 10 5 cells using TRIzol TM (Invitrogen) following the manufacturer's protocol. RNA quantity and quality were assessed by absorbance at 260/280 nm using a NanoDrop 2000 (Thermo Fisher, Waltham, MA), with 260/280 ratios between 1.8 and 2.0. Residual genomic DNA was degraded from 1 g of total RNA using DNase I, Amplification Grade (Thermo Fisher). One g of each experimental sample was reverse transcribed into cDNA using a High Capacity cDNA Reverse Transcription Kit (Applied Biosystems, Foster City, CA) according to the suggested manufacturer's protocol. Gene expression analysis was conducted on samples using TaqMan Gene Expression Assays for Tmem16f (Hs03805835_m10.51), and the endogenous housekeeping gene b-act (Hs99999903_m1). Real-time quantitative PCR was conducted with a StepOnePlus TM Real-time PCR System (Applied Biosystems) using TaqMan TM Fast Advanced Master Mix (Life Technologies) and 10 ng of cDNA for each sample. All experimental reactions were performed in triplicate. Negative controls were included for all reactions, with nuclease-free water substituted for cDNA samples. Each cDNA sample was subsampled 3 times during each quantitative PCR. Relative fold-changes were calculated using the 2 Ϫ⌬⌬Ct method. Significance was determined by comparing the 2 Ϫ⌬Ct value using a one-way analysis of variance with a Dunnett's post hoc. Error bars represent the standard deviation of the relative-fold expression between samples.
The Western blot procedure was described previously (81). At 48 h after transfection, cells were lysed in a buffer containing 20 mM HEPES (pH 7.5), 200 mM NaCl, 10 mM EDTA, 1% Triton X-100, supplemented with the SIGMAFAST mixture of protease inhibitors (Sigma). Proteins in the lysate were separated by electrophoresis on a NuPAGE 4 -12% BisTris SDS-polyacrylamide gel (Life Technologies) and then transferred to Immune-Blot Low Fluorescence PVDF membrane (Bio-Rad). The membranes were blocked in the Odyssey blocking buffer for 1 h at room temperature (LI-COR Biosciences, Lincoln, NE). The primary mouse anti-FLAG M2 monoclonal antibody (Sigma) was diluted to a final concentration of 1 g/ml in the Odyssey blocker with 0.2% Tween 20. The secondary goat anti-mouse IgG (HϩL) antibody, conjugated with an infrared fluorophore IRDye-680LT (LI-COR Biosciences), was diluted 1Ϻ20,000 in the same buffer. The membranes were incubated at room temperature with primary and secondary antibodies for 2 and 1 h, respectively. The membranes were imaged using the Odyssey 9120 Infrared Imaging System (LI-COR Biosciences) in the 700-nm channel.
Quantification of the cells overexpressing TMEM16F was done by FACS analysis. At 48 h after transfection, 1 ϫ 10 6 HEK 293 cells per sample were fixed and permeabilized using the Fix and Perm cell permeabilization reagent (Molecular Probes, Eugen, OR) according to manufacturer's instructions. The primary anti-FLAG M2 (Sigma) and secondary Alexa Fluor 488 goat anti-mouse IgG (HϩL) (Life Technologies) antibodies were diluted to a final concentration of 2 and 1 g/ml, respectively, and incubated for 30 min at room temperature. Samples were acquired using a MACSQuant Analyzer 10 flow cytometer (Miltenyi Biotec, Bergisch Gladbach, Germany) and analyzed with FlowJo software (FlowJo, Ashland, OR).

Viability assay
At 48 h after transfection, cells were detached by treatment with 0.5 mg/ml of trypsin-EDTA, resuspended at 1.2 ϫ 10 6 cell/ml in fresh medium, and 100-l samples of this suspension were aliquoted to electroporation cuvettes for nsPEF exposures. Immediately following the exposures, the samples were diluted with fresh medium to 3 ϫ 10 5 cells/ml and aliquoted into a 96-well plate, in triplicates at 30 ϫ 10 3 cell/well and left at 37°C in the incubator. At 24 h after the exposure, 10 l of Presto Blue reagent (Life Technologies) was added to each well and the incubation continued for 1 h at 37°C. The plate was read with a Synergy 2 microplate reader (BioTek, Winooski, VT), with excitation/emission settings at 530/590 nm. The triplicate data were averaged, corrected for the background, and considered as a single experiment.

Statistical analysis
Unless specified otherwise, data are presented as mean Ϯ S.E. for n independent experiments. Statistical analyses were performed using a two-tailed t test where p Ͻ 0.05 was considered statistically significant. Statistical calculations, including data fits, and data plotting were accomplished using Grapher 11 (Golden Software, Golden, CO).