2025 Volume 48 Issue 5 Pages 595-605
Transmembrane protein 16F (TMEM16F), identified as the causative gene for Scott syndrome, which causes blood coagulation disorders, is known to function as not only a scramblase that bi-directionally transports phospholipids in the lipid bilayer but also a Ca2+-activated ion channel with low intracellular Ca2+ sensitivity. However, how the dual functions of TMEM16F are controlled remains poorly understood. In this study, we investigated the properties of amino acid residues in human TMEM16F involved in the linkage between phospholipid and ion transports and the regulation of their transports using flow cytometry and whole-cell patch-clamp recordings. We demonstrated that ion and phospholipid transports induced by elevation of intracellular Ca2+ concentration were tightly coupled in human embryonic kidney HEK293T cells overexpressing wild-type TMEM16F or its mutants. Mutations of amino acid residues in the hydrophilic subunit cavity of TMEM16F indicated that both substrates were transported through its subunit cavity. Importantly, the tail current analysis suggests that conformational changes of TMEM16F by the channel gating are required for its phospholipid transport. These results suggest that ion channel activities of human TMEM16F modulate its scramblase activities.
The plasma membrane consists of a lipid bilayer composed of various types of phospholipids. The inner layer is rich in phosphatidylserine (PS), phosphatidylethanolamine, and phosphatidylinositol, while the outer layer is rich in phosphatidylcholine and sphingomyelin. This asymmetry of phospholipids in the plasma membrane is important for various biological activities. This asymmetry is maintained by an ATP-dependent flip-flop mechanism mediated by flippases and floppases. However, the phospholipid asymmetry is disrupted during blood coagulation reaction and apoptotic cell death. One of the hallmarks of this asymmetric disruption is the extracellular exposure of PS, mediated by scramblases that bi-directionally transport phospholipids in an intracellular Ca2+ concentration ([Ca2+]i)-dependent manner.1) Mouse transmembrane protein 16F (TMEM16F) has been identified as a phospholipid scramblase involved in blood coagulation.2)
The mammalian TMEM16 family with 10 transmembrane domains consists of 10 isoforms that function as ion channels or phospholipid scramblases activated by elevation of [Ca2+]i. Among these TMEM16 members, TMEM16A and 16B have been reported to function as Ca2+-activated Cl− channels.3–5) On the contrary, some isoforms, including TMEM16F, have been shown to function as phospholipid scramblases.2,6,7) Scott syndrome is an inherited disorder that presents with blood coagulation defects.8–10) In patients with Scott syndrome, a splice mutation in TMEM16F generates a truncated form with no C-terminus. The impaired phospholipid transport by its deficiency inhibits platelet PS exposure, resulting in blood coagulation failure.2,11) Thus, the TMEM16 family is a unique family with a mixture of membrane transporters that act as ion channels or phospholipid scramblases. Interestingly, several groups, including us, have reported that the overexpression of TMEM16F results in Ca2+-activated ion channel function with low sensitivity to [Ca2+]i.12–16) These results suggest that the TMEM16F could be involved in the transport of ions and phospholipids. Besides, structural analyses have revealed two characteristic cavities in the TMEM16 family that form dimers: a hydrophobic dimer cavity resulting from the complex formation and a hydrophilic subunit cavity possessed by each subunit.17–19) Previous structural studies suggest that highly hydrophilic subunit cavities of the TMEM16F are involved in the transports of ions or phospholipids,17,20) but little is known about how their subunit cavities regulate phospholipid and ion transports. Therefore, we investigated how mutations of amino acid residues in the hydrophilic subunit cavity affect the phospholipid and ion transport functions of human TMEM16F.
Human embryonic kidney HEK293T cells (kindly provided by Prof. Makoto Tominaga) were grown in Dulbecco’s Modified Eagle’s Medium (D-MEM with low glucose: FUJIFILM Wako Pure Chemical Corporation, Osaka, Japan) supplemented with 10% fetal bovine serum (FBS), 100 units/mL penicillin, and 100 μg/mL streptomycin at 37°C in 5% CO2.
Plasmid Construction and Site-Directed MutagenesisHuman full-length TMEM16F cDNA (NM_001025356) was initially cloned into the bicistronic pCINeo/IRES-green fluoresence protein (GFP) vector with NheI and EcoRI, as previously described.14) Then, the TMEM16F cDNA was enzymatically subcloned into the pcDNA6/V5-His vector (Thermo Fisher Scientific, Waltham, MA, U.S.A.) with NheI and EcoRI. To express the V5-tagged TMEM16F, the stop codon of TMEM16F was mutated in the TMEM16F-containing pcDNA6/V5-His vector using QuickChange Lightning Site-Directed Mutagenesis Kits (Agilent Technologies, Santa Clara, CA, U.S.A.). Single-point mutants of the TMEM16F were constructed using QuickChange Lightning Site-Directed Mutagenesis Kits or KOD -Plus- Mutagenesis Kit (TOYOBO, Osaka, Japan). The full-length sequence of each gene was confirmed by using the ABI PRISM 3130 Genetic Analyzer (Thermo Fisher Scientific).
TransfectionThe expression vectors of human TMEM16F were transiently transfected into HEK293T cells using Lipofectamine 2000 (Thermo Fisher Scientific) or Polyethylenimine MAX (Polysciences, Warrington, PA, U.S.A.). The experiments were performed 24 h after the transfection.
Phospholipid Scrambling AssayPhospholipid scrambling activities of human TMEM16F were measured using a flow cytometry analysis with phycoerythrin (PE)-conjugated annexin V (Annexin V-PE: Medical & Biological Laboratories, Tokyo, Japan). Briefly, HEK293T cells transfected with empty or TMEM16F-containing pCINeo/IRES-GFP vector were exposed to a test solution with 5 μM ionomycin, a Ca2+ ionophore, for 2 min and then reacted with Annexin-PE as described in the instruction manual. To compare the Ca2+-induced phospholipid scrambling activities in the cells exogenously expressing TMEM16F, 10000 cells showing GFP fluorescence were analyzed. As shown in Fig. 1a, the cells stained with Annexin V-PE are divided into two groups depending on the intensity of fluorescence. The cells with high fluorescence intensity were defined as Annexin V-positive cells, and their percentage was compared under each condition, as previously reported.21)
(a) Representative histograms of Ca2+-induced phospholipid scrambling in mock-transfected and TMEM16F-expressing cells. (b) Quantification of annexin V-positive cells under each condition in a (n = 3). (c) Representative membrane currents in mock-transfected and TMEM16F-expressing cells. The inset indicates an applied step pulse. [Ca2+]I was 30 μM. Arrowheads: Zero-current levels. (d) Current (I)-voltage (V) relationships of membrane currents in mock-transfected (n = 6) and TMEM16F-expressing cells (n = 5). Sharps indicate significant differences to mock (p = 0.02, 0.0007, 0.0009 at +60, +80, +100 mV, respectively). (e and f) Representative TMEM16F currents (e) and the corresponding I–V relationships (f) with or without NPPB in TMEM16F-expressing cells (n = 13). Sharps indicate significant differences to control (p = 0.00008, 0.000008, 0.000003, 0.000002, 0.00002 at +20, +40, +60, +80, +100 mV, respectively). (g) Effect of NPPB on Ca2+-induced phospholipid scrambling in TMEM16F-expressing cells (n = 4). (h and i) Representative TMEM16F currents (h) and the corresponding I–V relationships (i) with or without niflumic acid in TMEM16F-expressing cells (n = 7). Sharps indicate significant differences to control (p = 0.003, 0.0008, 0.00007, 0.000001, 0.00003 at +20, +40, +60, +80, +100 mV, respectively). (j) Effect of niflumic acid on Ca2+-induced phospholipid scrambling in TMEM16F-expressing cells (n = 4). (k) Extracellular Cl− dependence of Ca2+-induced phospholipid scrambling in TMEM16F-expressing cells (n = 3). (l) Temperature dependence of Ca2+-induced phospholipid scrambling in TMEM16F-expressing cells (n = 3).
The test solution contained 140 mM NaCl, 5 mM KCl, 10 mM 2-[4-(2-hydroxyethyl)-1-piperazinyl]ethanesulfonic acid (HEPES), 2 mM ethylenediamine-N,N,N’,N’-tetraacetic acid disodium salt (EDTA-2Na), 2 mM MgSO4, and 1.379 mM CaSO4 (pH 7.4). The free Ca2+ concentration of the test solution is calculated to be 30 μM using CaBuf software (kindly provided by Dr. G. Droogmans). In some experiments, the test solutions with different Cl− concentrations were made by substituting with aspartate.
Electrophysiological ExperimentsIn patch-clamp experiments, transfected HEK293T cells were distinguished by GFP fluorescence using an inverted fluorescence microscope (Eclipse TE2000-U: Nikon, Tokyo, Japan). Whole-cell recordings were performed using a patch-clamp amplifier (Axopatch 200B: Molecular Devices, Sunnyvale, CA, U.S.A.). Currents were low-pass filtered at 1 kHz and digitized at 10 kHz with an A/D converter (DigiData1550A: Molecular Devices). pClamp10.6 (Molecular Devices) was used for data acquisition and data analysis. The pipette resistance was 2–4 MΩ. Before current recordings, series resistance was compensated by 70% to minimize voltage errors.
The voltage and time dependence of the TMEM16F currents were observed with step pulses from −100 to +100 mV in 20-mV increments. To analyze the voltage-dependent deactivation of TMEM16F channels, tail currents were measured by applying the step pulse with a post-pulse of −100 mV. The tail currents recorded at −100 mV were fitted by a double exponential function to calculate the fast and slow time constants.
To measure the TMEM16F currents, the following solutions were used. The pipette solution (in mM) contained 110 CsCl, 2 MgSO4, 1 ethylene glycol bis(2-aminoethyl ether)-N,N,N,N-tetraacetic acid (EGTA), 1 Na2ATP, 10 HEPES, 15 Na-HEPES, and 50 d(–)-mannitol (pH 7.3). An appropriate amount of CaSO4 was added to the pipette solution to achieve each free Ca2+ concentration using the CaBuf software. The bathing solution (in mM) consisted of 110 CsCl, 5 MgSO4, 12 HEPES, 7 tris(hydroxymethyl)aminomethane (Tris), and 100 d(–)-mannitol (pH 7.4).
ImmunocytochemistryHEK293T cells expressing V5-tagged TMEM16F were fixed in methanol at −20°C for 7 min and permeabilized in phosphate-buffered saline with 0.1 mM CaCl2 and 1 mM MgCl2 (PBS++) containing 0.3% Triton X-100 and 0.1% bovine serum albumin (BSA) at room temperature for 15 min. The permeabilized cells were subjected to a blocking reaction in the goat serum dilution buffer (GSDB: 450 mM NaCl, 17% goat serum, and 0.3% Triton X-100 in 20 mM phosphate buffer (pH 7.4)) at room temperature for 1 h. They were treated with V5-tag mouse monoclonal antibody (1 : 100: Thermo Fisher Scientific) or anti-Flotillin 2 rabbit antibody (1 : 200: Merck, Darmstadt, Germany) at 4°C overnight and then reacted with Alexa Fluor 488-conjugated mouse IgG antibody (1 : 100: Thermo Fisher Scientific) or Alexa Fluor 546-conjugated rabbit IgG antibody (1 : 100: Thermo Fisher Scientific) at room temperature for 1 h, respectively. These antibodies were diluted with GSDB. The cells were incubated with 4′,6-diamidino-2-phenylindole (DAPI) solution (1 : 1000: Dojindo Laboratories, Kumamoto, Japan) at room temperature for 15 min. Immunofluorescent images were visualized using a laser scanning confocal microscope (LSM700: Zeiss, Jena, Germany).
Cell Surface Biotinylation AssayHEK293T cells expressing V5-tagged TMEM16F were incubated in PBS(++) added 0.5 mg/mL EZ-Link Sulfo-NHS-LC-Biotin (Thermo Fisher Scientific) at 4°C for 1 h and then washed three times with PBS(++) containing 50 mM glycine and 0.2% BSA for 5 min. Subsequently, the cells were solubilized with a lysis buffer (150 mM NaCl, 50 mM Tris–HCl (pH 7.4), 1 mM EDTA, and 1%Triton X-100) supplemented with protease inhibitors (10 μg/mL phenylmethylsulfonyl fluoride, 1 μg/mL leupeptin, and 1 μg/mL pepstatin A) on ice for 30 min. After centrifugation at 15000 × g for 20 min at 4°C, the supernatant was diluted to a protein concentration of 1 mg/mL. The protein concentration of samples was quantified with the Protein Assay BCA Kit (FUJIFILM Wako Pure Chemical Corporation). The resulting supernatant was treated with Avidin-Agarose (Merck) and then incubated at 4°C for 16 h. The pellet after centrifugation at 2000 × g for 2 min at 4°C was solubilized in a sample buffer (4% sodium dodecyl sulfate, 20% glycerol, 5% β-mercaptoethanol, 130 mM Tris–HCl (pH 6.8)) at 37°C for 30 min. The samples were then centrifuged at 8000 × g for 2 min at 24°C, and the supernatant was used as the biotinylated sample.
Western BlottingThe biotinylated samples were electrophoresed in 7.5% polyacrylamide gels and transferred to polyvinylidene fluoride membranes. The membranes were incubated with 5% skim milk (FUJIFILM Wako Pure Chemical Corporation) dissolved in Tris-buffered saline (TBS-T: 150 mM NaCl, 25 mM Tris–HCl (pH 7.4), and 0.1% Tween20) at room temperature for 1 h. After blocking, the membranes were reacted with primary antibodies at 4°C overnight and then with horseradish peroxidase (HRP)-conjugated secondary antibody at room temperature for 1 h. The antibodies were diluted with TBS-T containing 5% skim milk. Chemiluminescent signals were detected using Western Lightning ECL Pro (PerkinElmer, Inc., Waltham, MA, U.S.A.) with an ImageQuant LAS 4000 system (GE Healthcare, Little Chalfont, U.K.), and quantified using Image J software.
V5-tag mouse monoclonal antibody (1 : 5000), Na+, K+-ATPase α1 isoform (α1NaK) antibody (1 : 5000: Santa Cruz Biotechnology, Dallas, TX, U.S.A.), and ERp57 antibody (1 : 2500: Enzo Life Science, Farmingdale, NY, U.S.A.) were used as primary antibodies. HRP-conjugated mouse IgG antibody (1 : 5000: Merck) was utilized as a secondary antibody. The membranes were stripped and re-probed for α1NaK as a loading control. The total amount of two bands, 110 and 130 kDa, was quantified in TMEM16F.
Statistical AnalysisData are shown in box plots or as means ± standard error of the mean of n observations. Statistical analysis was performed by paired or unpaired Student’s t-test for comparisons between two groups and one-way ANOVA with Tukey’s post hoc test for comparisons between more than three groups. A p value under 0.05 was considered significant.
In mock- or human TMEM16F-transfected HEK293T cells, we measured phospholipid scramblase activities using flow cytometry with Annexin V-PE. The percentage of annexin V-positive cells was increased to roughly 8% in mock cells treated with 5 μM ionomycin, a Ca2+ ionophore, while TMEM16F-overexpressing cells exhibited approximately two times higher phosphatidylserine (PS) exposures in the presence of ionomycin than mock cells (Figs. 1a, 1b). As previously reported,14) on the other hand, we observed strong outwardly rectifying ion currents showing time-dependent activation in response to elevation of [Ca2+]i to 30 μM in TMEM16F-overexpressing cells (Figs. 1c, 1d). Mock cells showed small Ca2+-activated ion currents with a weak outward rectification and no time-dependent activation, but the electrophysiological properties of currents are quite different from those of TMEM16F currents (Figs. 1c, 1d). These results suggest that the expression of TMEM16F induced the activities of Ca2+-induced phospholipid scramblase and Ca2+-activated ion channel.
To investigate whether the ion channel activities are associated with the phospholipid scramblase activities, we examined the effects of ion channel blockers on the TMEM16F currents and the PS exposures in the TMEM16F-overexpressing cells. The activities of not only ion channel but also phospholipid scramblase were significantly decreased in the presence of 100 μM 5-nitro-2-(3-phenylpropylamino)benzoic acid (NPPB, Figs. 1e–1g), which is reported to block mouse TMEM16F currents,22) or 300 μM niflumic acid (Figs. 1h–1j), which we have previously reported to reduce human TMEM16F currents.14) Interestingly, the decrease in extracellular Cl− concentration increased the number of annexin V-positive cells in a concentration-dependent manner (Fig. 1k). These results indicate that the Ca2+-induced phospholipid scrambling activities might be correlated with the Cl− flux in TMEM16F-overexpressing cells.
It has been reported that once mouse TMEM16F was activated, phospholipids flip-flopped even at 4°C, suggesting that mouse TMEM16F does not require its conformational change during phospholipid transport.23) However, we observed no Ca2+-induced PS exposures at 4°C in human TMEM16F-overexpressing cells (Fig. 1l), suggesting that a conformational change of human TMEM16F is required for its phospholipid transport.
The Subunit Cavity of Human TMEM16F Is Crucial for the Ion and Phospholipid TransportsThe mutation of an arginine residue into glutamate (R621E) in the extracellular loop between transmembranes 5 and 6 of mouse TMEM16A, which has only Ca2+-activated Cl− channel function, was reported to alter its ion selectivity.5,24) Therefore, we used the alanine-scanning mutagenesis to elucidate how the corresponding arginine residue of human TMEM16F affects its functions. As shown in Fig. 2a, the immunocytochemistry with the V5-tag antibody showed that wild-type TMEM16F and its R591A mutant were co-stained with flotillin 2, a plasma membrane marker. Besides, the cell surface biotinylation assay demonstrated that the R591A mutant was equally expressed in the plasma membrane to the wild-type TMEM16F (Figs. 2b, 2c). In Western blotting with the V5-tag antibody, we detected two bands in the TMEM16F samples but not in the mock sample. This suggests that these bands are derived from the exogenous TMEM16F proteins and that human TMEM16F, which has six N-glycosylation sites, may undergo post-translational modifications. As shown in Supplementary Fig. 1, the sizes of TMEM16F in the biotinylated sample were decreased by treatment of PNGase F, suggesting that glycosylated TMEM16F proteins function in the plasma membrane of the cells. The R591A mutation significantly decreased the Ca2+-activated ion currents with strong outward rectification (Figs. 2d, 2e) and the Ca2+-induced PS exposures (Fig. 2f) to mock cell levels compared with the wild-type TMEM16F.
(a) Representative immunocytochemistry images showing localization of TMEM16F in cells expressing wild-type TMEM16F and R591A mutant. Scale bars, 10 μm. (b) Surface biotinylation assay in mock-transfected cells and cells expressing the wild-type and R591A mutant. Western blots of TMEM16F (110 and 130 kDa), Na+, K+-ATPase α1 isoform (α1NaK) (95 kDa), and ERp57 (57 kDa) in the total cell lysate (Input) and the cell surface biotinylation sample (Cell surface). α1NaK was used as a loading control. (c) Quantification of surface expression of the wild-type and R591A mutant (n = 3). The total amount of two bands, 110 and 130 kDa, was quantified in TMEM16F. (d and e) Representative wild-type and R591A mutant currents (d) and the corresponding I–V relationships (e) (WT: n = 9, R591A: n = 9). [Ca2+]i was 30 μM. The dotted line indicates the mean currents in mock cells shown in Fig. 1d. Sharps indicate significant differences to WT (p = 0.04, 0.03, 0.02, 0.02 at +40, +60, +80, +100 mV, respectively). (f) Ca2+-induced phospholipid scrambling in mock cells and the cells expressing the wild-type and R591A mutant (n = 3).
Two sites where phospholipid headgroups are thought to interact, the intracellular site (Sc) and the extracellular site (Se) of fungal Nectria haematococca TMEM16 (nhTMEM16) and mouse TMEM16F, were reported to be critical for phospholipid scrambling.23,25) Interestingly, since the two amino acid residues at the Sc site of scrambling domain are conserved in many of the TMEM16 family members that exhibit scramblase function, but are different in TMEM16A that exhibits ion channel function,16,23) we replaced the amino acid residues in human TMEM16F with the corresponding residues human in TMEM16A (E528G and K529C). On the contrary, the two amino acid residues at the Se site are conserved among the TMEM16 family.23) Therefore, these amino acid residues in the Se site of human TMEM16F were replaced with alanine (R477A and E603A). Here, we examined how these mutations in the Sc and Se sites of human TMEM16F affect scramblase and channel function. In the mutants of Se sites (R477A and E603A) that were equally expressed in the plasma membrane compared with the wild-type TMEM16F (Figs. 3a, 3e), the Ca2+-induced PS exposure was decreased to mock levels (Figs. 3b, 3f). Interestingly, these Se site mutations similarly decreased the TMEM16F-dependent Ca2+-activated ion currents (Figs. 3c, 3d, 3g, 3h). Besides, although the expression of the Sc site mutants (E528G and K529C) in the plasma membrane was comparable to that of the wild-type TMEM16F (Fig. 3i), the phospholipid scrambling activities were significantly reduced to mock levels in these Sc site mutants compared with the wild-type TMEM16F (Fig. 3j). Intriguingly, these Sc site mutants exhibited the significantly reduced outwardly rectifying ion currents, albeit to a lesser degree (Figs. 3k, 3l). These results suggest that these mutants substituted for corresponding amino acid residues of human TMEM16A exhibit partial ion channel activities but not scramblase activities.
(a) Surface biotinylation assay in cells expressing the wild-type and R477A mutant (110 and 130 kDa). α1NaK (95 kDa): a loading control. The total amount of two bands, 110 and 130 kDa, was quantified in TMEM16F (n = 4). (b) Ca2+-induced phospholipid scrambling in mock cells and the cells expressing the wild-type and R477A mutant (n = 3). (c and d) Representative wild-type and R477A mutant currents (c) and the corresponding I–V relationships (d) (WT: n = 6, R477A: n = 6). [Ca2+]i was 30 μM. The dotted line indicates the mean currents in mock cells shown in Fig. 1d. Sharps indicate significant differences to WT (p = 0.04, 0.004, 0.00004 at +60, +80, +100 mV, respectively). (e) Surface biotinylation assay in cells expressing the wild-type and E603A mutant (110 and 130 kDa). α1NaK (95 kDa): a loading control. The total amount of two bands, 110 and 130 kDa, was quantified in TMEM16F (n = 3). (f) Ca2+-induced phospholipid scrambling in mock cells and the cells expressing the wild-type and E603A mutant (n = 4). (g and h) Representative wild-type and E603A mutant currents (g) and the corresponding I–V relationships (h) (WT: n = 8, E603A: n = 3). [Ca2+]i was 30 μM. The dotted line indicates the mean currents in mock cells shown in Fig. 1d. Sharps indicate significant differences to WT (p = 0.04, 0.02, 0.02 at +60, +80, +100 mV, respectively). (i) Surface biotinylation assay in cells expressing the wild-type, E528G, and K529C mutants (110 and 130 kDa). α1NaK (95 kDa): a loading control. The total amount of two bands, 110 and 130 kDa, was quantified in TMEM16F (n = 4). (j) Ca2+-induced phospholipid scrambling in mock cells and the cells expressing the wild-type, E528G, and K529C mutants (n = 4). (k and l) Representative currents of the wild-type, E528G, and K529C mutants (k) and the corresponding I–V relationships (l) (WT: n = 28, E528G: n = 19, K529C: n = 16). [Ca2+]i was 30 μM. The dotted line indicates the mean currents in mock cells shown in Fig. 1d. Sharps indicate significant differences to WT (E528G: p = 0.002, 0.000008, 0.00002, 0.0008, 0.002 at +20, +40, +60, +80, +100 mV, respectively; K529C: p = 0.01, 0.0003, 0.0001, 0.0002, 0.001 at +20, +40, +60, +80, +100 mV, respectively). (m) Surface biotinylation assay in cells expressing the wild-type and Y562W mutant (110 and 130 kDa). α1NaK (95 kDa): a loading control. The total amount of two bands, 110 and 130 kDa, was quantified in TMEM16F (n = 3). (n) Ca2+-induced phospholipid scrambling in mock cells and the cells expressing the wild-type and Y562W mutant (n = 4). (o and p) Representative wild-type and Y562W mutant currents (o) and the corresponding I–V relationships (p) (WT: n = 8, E603A: n = 8). [Ca2+]i was 30 μM. The dotted line indicates the mean currents in mock cells shown in Fig. 1d. Sharps indicate significant differences to WT (p = 0.03, 0.004, 0.0006, 0.0001, 0.00004 at +20, +40, +60, +80, +100 mV, respectively).
Furthermore, it has been reported that a tyrosine residue located near the center of the subunit cavity functions as an inner activation gate of mouse TMEM16F.26) Therefore, we constructed a human TMEM16F mutant (Y562W) that substituted the corresponding tyrosine residue into bulky tryptophan, and investigated its function. The Y562W mutation similarly reduced both phospholipid scrambling and ion channel functions triggered by elevated [Ca2+]i to mock levels, although the mutant showed the plasma membrane expression comparable to the wild-type TMEM16F (Figs. 3m–3p).
These results suggest that ions and phospholipids are transported through the subunit cavity of human TMEM16F.
The First Intracellular Loop of Human TMEM16F Regulates Its Gating KineticsSince an aspartate residue (D409) in the first intracellular loop of mouse TMEM16F is reported to regulate the intracellular Ca2+ sensitivity,2) we next analyzed two functions using several mutants (D408G, D408N, and D408E) at the corresponding aspartic acid residue of human TMEM16F. Although these mutants had equivalent expressions in the plasma membrane (Fig. 4a), all mutants exhibited enhanced Ca2+-induced PS exposure compared with the wild-type TMEM16F (Fig. 4b). However, the increased phospholipid scrambling activities became significantly lower in the D408E mutants (Fig. 4b). On the contrary, these mutants enhanced the Ca2+-activated ion currents in a similar way to the phospholipid scrambling activities (Figs. 4c, 4d). Figure 4e indicates the Ca2+ sensitivity of channel functions in the wild-type TMEM16F and its D408 mutants. The half-activation concentrations (EC50) for [Ca2+]i were 9.4 ± 0.1, 3.7 ± 0.9, 3.8 ± 0.4, and 3.8 ± 0.5 μM in the wild-type, D408G, D408N, and D408E mutants, respectively.
(a) Surface biotinylation assay in cells expressing the wild-type and D408 mutants (110 and 130 kDa). α1NaK (95 kDa): a loading control. The total amount of two bands, 110 and 130 kDa, was quantified in TMEM16F (n = 5). (b) Ca2+-induced phospholipid scrambling in mock cells and the cells expressing the wild-type and D408 mutants (n = 6). (c and d) Representative wild-type and D408 mutant currents (c) and the corresponding I–V relationships (d) (WT: n = 15, D408G: n = 5, D408N: n = 6, D408E: n = 10). [Ca2+]i was 30 μM. The dotted line indicates the mean currents in mock cells shown in Fig. 1d. Sharps indicate significant differences to WT (D408G: p = 0.00002, 0.00002, 0.00001, 0.00001, 0.04 at +20, +40, +60, +80, +100 mV, respectively; D408N: p = 0.005, 0.0001, 0.000007, 0.0000009, 0.006 at +20, +40, +60, +80, +100 mV, respectively; D408E: p = 0.01, 0.04 at +80, +100 mV, respectively). Daggers indicate significant differences to D408G (p = 0.0004, 0.002, 0.007, 0.02, 0.000003 at +20, +40, +60, +80, +100 mV, respectively). (e) Intracellular Ca2+ dependence of wild-type and D408 mutant currents (WT: n = 4–21, D408G: n = 5–8, D408N: n = 4–12, D408E: n = 4–10).
We further investigated the gating of TMEM16F channels by analyzing tail currents recorded at −100 mV after a prepotential of +100 mV. As shown in Fig. 5a, all tail currents in the wild-type TMEM16F and its D408 mutants were well-fitted by a double exponential function. Compared with the wild-type TMEM16F, intriguingly, both fast and slow time constants were increased in D408G mutants, but only a slow time constant was partly and significantly increased in D408E mutants (Figs. 5b, 5c). The results suggest that an aspartate residue in the first intracellular loop in human TMEM16F regulates the channel gating.
(a) Representative TMEM16F tail currents in cells expressing the wild-type and D408 mutants. The inset indicates an applied step pulse. [Ca2+]i was 30 μM. The tail currents from +100 mV were enlarged at each bottom. Red dotted lines indicate the fittings with a double exponential equation. (b) Fast and (c) slow time constants of tail current deactivation in cells expressing the wild-type (n = 7) and D408 mutants (D408G (n = 8) and D408E (n = 9)).
Here, we showed that NPPB and niflumic acid, which are the general Cl− channel blockers, inhibited not only ion channel but also phospholipid scrambling activities induced by elevation of [Ca2+]i (Figs. 1e–1j). The inhibition mechanism of these TMEM16F functions by NPPB and niflumic acid is unrevealed in this study. However, a recent report demonstrated that NPPB and niflumic acid bind to the pore region of TMEM16A.27) Since the residue to which these blockers bind is conserved between TMEM16A and TMEM16F, these inhibitors are likely able to directly block the pore region of TMEM16F. Besides, we found that 30 μM tannic acid, a blocker of mouse TMEM16F channel,28) also suppressed both the TMEM16F currents (67% inhibition at +100 mV, n = 6) and the phospholipid transport (88% inhibition, n = 3). However, since tannic acid is reported to quench the fluorescent probe using flow cytometry,28) further investigation could be required to clarify the effects of tannic acid on the phospholipid transport. These pharmacological analyses suggest that human TMEM16F channel function is associated with its phospholipid scrambling activity, consistent with previous reports in fungal nhTMEM16 and mouse TMEM16F.16,29) Surprisingly, reduced extracellular Cl− concentrations promoted Ca2+-induced PS exposure (Fig. 1k), strengthening the correlation between ion and phospholipid transports of the TMEM16F. By contrast to the previous report with mouse TMEM16F,23) human TMEM16F-expressing cells exhibited no Ca2+-induced PS exposure at 4°C (Fig. 1i). These results suggest that ion fluxes due to conformational changes of the TMEM16F are associated with its phospholipid transport.
Currently, some structural and functional analyses suggest that the hydrophilic subunit cavity of TMEM16s mediates phospholipid transport.18,20,25,29–31) There are both positive and negative amino acid residues at the intracellular and extracellular sides of this subunit cavity, called Sc and Se sites, respectively. A model has been proposed in which phospholipid transport occurs via the interaction of their heads with these Sc and Se sites.25,29) In the present study, the phospholipid scrambling activities were reduced in the cells expressing Sc and Se mutants of human TMEM16F compared with the wild-type TMEM16F, although their plasma membrane expression was comparable (Figs. 3a, 3b, 3e, 3f, 3i, 3j). Thus, it is suggested that both positive and negative charges on the Sc ad Se sites are required for phospholipid transport. Intriguingly, we also found that these Sc and Se mutants exhibit decreased TMEM16F currents (Figs. 3c, 3d, 3g, 3h, 3k, 3l). Although the current inhibition was weaker in the Sc mutants (E528G and K529C) than the Se mutants (R477A and E603A), this may be due to the replacement of amino acid residues in the Sc site of human TMEM16F with those of human TMEM16A, which functions as a Ca2+-activated Cl− channel. These results suggest that both Sc and Se sites of human TMEM16F contribute to ion transport as well as phospholipid transport. Since the subunit cavity where Sc and Se sites are located is reported to undergo structural changes in nhTMEM16,29,31,32) mutations of these Sc and Se sites in human TMEM16F could affect their structural changes. It would be interesting to clarify how the Sc site contributes to ion transport, especially since the Sc site is located in a scrambling domain, which is essential for phospholipid transport.16,23) Besides, in the structural analysis using nhTMEM16, an alternating pore/cavity model has been proposed as one of the transport mechanisms for phospholipids and ions; that is, nhTMEM16 exhibits stepwise activation through three states, closed, intermediate, and open, in the presence of Ca2+.32,33) The fact that only ion channel activity was observed in E528G and K529C mutants may support this model. Further studies are awaited to understand how these sites of human TMEM16F modulate its ion channel function.
Additionally, the other mutations (R591A and Y562W) in the subunit cavity similarly suppressed the dual functions of human TMEM16F (Figs. 2, 3m–3p). The arginine residue at position 591 of human TMEM16F corresponds to R592 of mouse TMEM16F, which interacts with phospholipids,34) as well as R621, which determines the ion selectivity of mouse TMEM16A.5) Thus, it makes sense that a positive residue R591 of human TMEM16F plays an important role in ion and phospholipid transport. On the contrary, the tyrosine residue at position 562 of human TMEM16F corresponds to Y563 of mouse TMEM16F, which is located near the center of the subunit cavity and acts as an inner activation gate.26,35) Previous reports suggest that a narrow constriction in the subunit cavity of nhTMEM16 could inhibit phospholipid transport.29,31) In mouse TMEM16F, Y563 is suggested to protrude its side chain into a narrow subunit cavity to inhibit phospholipid entry.17,26) Since the Y562W mutation of human TMEM16F gives a bulkier side chain, it is considered that the tryptophan blocked the center of the subunit cavity and prevented the phospholipid entry even during the Ca2+-induced activation. Intriguingly, we demonstrated that the Y562 of human TMEM16 also contributes to ion transport (Figs. 3o, 3p). These results suggest that human TMEM16F transports both phospholipids and ions through the narrow center of the subunit cavity.
Previously, the D409G mutant of mouse TMEM16F was reported to enhance phospholipid scramblase activity.2,26) Consistently, all D408 mutants of human TMEM16F exhibited increased PS exposure induced by intracellular Ca2+ (Fig. 4b). Surprisingly, the negative charge of amino acid residue in TMEM16F could be responsible for the negative regulation of this scramblase activity. Importantly, the mutational effect of this aspartate residue on the TMEM16F current correlated with its effect on the Ca2+-induced phospholipid transport (Figs. 4b–4d). These results strongly suggest that human TMEM16F has a dual function: phospholipid scramblase and ion channels. On the contrary, all D408 mutants showed similarly high Ca2+ sensitivity to TMEM16F currents compared with the wild-type TMEM16F (Fig. 4e), suggesting that the negative charge of this residue is not important for increased Ca2+ sensitivity to the TMEM16F currents. Further studies are needed to clarify how the D408 residue of human TMEM16F regulates its Ca2+ sensitivity.
Ion channels function by bi-directional transitions between open and closed states. TMEM16F generates strong outwardly rectifying currents that exhibit a higher open probability at positive voltages than at negative voltages. Therefore, tail currents of human TMEM16F observed by holding at −100 mV from positive potentials indicate a transition from an open to a closed state. Here, we demonstrated that the tail currents could be fitted with double exponential functions (Fig. 5), suggesting the existence of three states in the gating of TMEM16F. This is consistent with the alternating pore/cavity model.32,33) Thus, we propose the transport model that human TMEM16F functions in closed, intermediate, and open state transitions, as shown in Fig. 6. The human TMEM16F D408 mutants exhibited changes in their gating kinetics (Fig. 5). The tail current deactivation in these mutants was slower than that in the wild-type TMEM16F, suggesting that a larger number of these mutants are long maintained in the open and intermediate states. Interestingly, the changes in fast and slow time constants calculated from the tail currents of D408 mutants correlated with the degree of phospholipid scramblase activities. Although more detailed studies are needed, this might indicate that the aspartate residue in human TMEM16F regulates three state transitions in the alternating pore/cavity model. These results suggest that the channel gating of human TMEM16F triggers its phospholipid transport.
Human TMEM16F exhibits stepwise activation through closed, intermediate, and open states in the presence of Ca2+. TMEM16F in the intermediate state transports ions. After transitioning to the open state, TMEM16F also transports phospholipids.
In conclusion, the present study demonstrates that ion and phospholipid transports induced by elevation of [Ca2+]i are tightly coupled in HEK293T cells overexpressing human TMEM16F and that both substrates are transported through the hydrophilic subunit cavity in human TMEM16F. We propose that conformational changes by the channel gating are required for phospholipid transport in human TMEM16F (Graphical Abstract).
This work was supported by Grants-in-Aid for Scientific Research (KAKENHI) from Japan Society for the Promotion of Science (to H.S. (22H02801 and 23K24063), T.S. (22K06827), and T.F. (23K06330)). This work was supported by JST SPRING, Grant No. JPMJSP2145 (to T.K.).
The authors declare no conflict of interest.
This article contains supplementary materials.
