TMEM16A and TMEM16B proteins are CaCCs (Ca2+-activated Cl− channels) with eight putative transmembrane segments. As shown previously, expression of TMEM16B generates CaCCs characterized by a 10-fold lower Ca2+ affinity and by faster activation and deactivation kinetics with respect to TMEM16A. To investigate the basis of the different properties, we generated chimaeric proteins in which different domains of the TMEM16A protein were replaced by the equivalent domains of TMEM16B. Replacement of the N-terminus, TMD (transmembrane domain) 1–2, the first intracellular loop and TMD3–4 did not change the channel's properties. Instead, replacement of intracellular loop 3 decreased the apparent Ca2+ affinity by nearly 8-fold with respect to wild-type TMEM16A. In contrast, the membrane currents derived from chimaeras containing TMD7–8 or the C-terminus of TMEM16B showed higher activation and deactivation rates without a change in Ca2+ sensitivity. Significantly accelerated kinetics were also found when the entire C-terminus of the TMEM16A protein (77 amino acid residues) was deleted. Our findings indicate that the third intracellular loop of TMEM16A and TMEM16B is the site involved in Ca2+-sensitivity, whereas the C-terminal part, including TMD7–8, affect the rate of transition between the open and the closed state.
TMEM16A and TMEM16B, also known as ANO (anoctamin) 1 and ANO2 respectively, are major components of CaCCs (Ca2+-activated Cl− channels), a class of ion channels involved in various physiological roles, including epithelial secretion, sensory transduction, nociception, control of neuronal excitability and regulation of smooth muscle contraction [1–7].
TMEM16A and TMEM16B belong to a protein family (ANOs) that includes eight other members (TMEM16C–TMEM16H, TMEM16J and TMEM16K or ANO3–ANO10). Although the function of TMEM16A and TMEM16B as CaCCs has been confirmed in multiple studies, the role of the other members of the family is less well established. In one study, TMEM16C–TMEM16G were found in intracellular compartments . In other studies, these proteins showed functions that are more consistent with a plasma membrane localization. In particular, TMEM16F expression was associated with phospholipid scramblase activity  and with the appearance of anion or cation channels [10,11]. The differences between TMEM16A/B and the other ANOs are consistent with the extent of amino acid sequence conservation. TMEM16A and TMEM16B are close paralogues, with a nearly 60% sequence identity . In constrast, the other ANOs are more distant in evolutionary terms.
At present, little is known about the structure–function relationship of TMEM16A and TMEM16B. When expressed in heterologous expression systems, TMEM16A and TMEM16B generate Cl−-permeable channels that are activated by both membrane depolarization and by cytosolic Ca2+. In particular, the apparent affinity of the channels for Ca2+ is increased by shifting the membrane potential to more positive values [13,14]. The molecular mechanisms underlying the Ca2+- and voltage-dependence of channel gating are poorly known. Analysis of the amino acid sequence of TMEM16A and TMEM16B does not reveal any canonical voltage-sensing or Ca2+-binding domain.
According to predictive bioinformatic tools, TMEM16A and TMEM16B share a putative topology consisting of eight TMDs (transmembrane domains) and cytosolic N- and C-termini. Most algorithms that predict TMDs agree in identifying the position of transmembrane helices 1–4 and 7–8. In contrast, the topology of the region between transmembrane segments 4 and 7 is uncertain. In one study, on the basis of the mutagenesis of TMEM16A that resulted in altered ion selectivity, the region between the fifth and the sixth TMDs was proposed to form a re-entrant loop exposed to the extracellular milieu and involved in the formation of the channel pore . In a more recent study, a revised topology for TMEM16A has been proposed . According to that study, the region corresponding to the re-entrant loop would actually completely cross the plasma membrane, forming the sixth TMD and the third intracellular loop .
Studies performed on TMEM16A protein have begun to identify the possible regions involved in voltage- and Ca2+-sensing. The first intracellular loop contains a segment of eight amino acids, EEEEEAVK, that is important in the control of both Ca2+ and voltage sensitivity . Interestingly, the sequence EAVK corresponds to a microexon that can be alternatively spliced . In TMEM16B, the first intracellular loop is also involved in the control of channel properties, such as voltage dependence . In another study, the revision of TMEM16A topology showed two highly conserved glutamic acid residues, Glu702 and Glu705, in the third intracellular loop, as possible Ca2+-binding sites . Mutagenesis of these residues to glutamine caused a dramatic decrease, by more than 100-fold, in the apparent Ca2+ affinity of TMEM16A channels. These findings indicate that Glu702 and Glu705 are the major determinants of Ca2+-sensitivity of TMEM16A channels.
Interestingly, TMEM16A and TMEM16B channels have significantly different properties. In particular, TMEM16B channels have a nearly 10-fold reduced Ca2+-sensitivity with respect to TMEM16A channels. TMEM16A channels are activated by cytosolic Ca2+ concentrations in the 300–600 nM range, whereas TMEM16B channels require micromolar concentrations. Furthermore, Cl− currents elicited by TMEM16B expression are characterized by activation and deactivation kinetics much faster than those of TMEM16A . Importantly, the glutamic acid residues Glu702 and Glu705 are also present, at equivalent positions, in TMEM16B protein. Therefore the difference between TMEM16A and TMEM16B suggests the existence of other regions involved in Ca2+ sensitivity.
The aim of the present study was to compare TMEM16A and TMEM16B chloride channels at the amino acid sequence and functional levels and then, by the generation of TMEM16A–TMEM16B chimaeras, to identify the regions responsible for the functional differences between the two types of channels.
Our results indicate that many regions of TMEM16A protein, including the N-terminus and the first intracellular loop, can be replaced by equivalent regions of TMEM16B protein without dramatic alterations in channel properties. Instead, we identified intracellular loop 3 as a domain important for calcium sensitivity and the C-terminal part of the protein, including TMD7–8, as relevant for channel kinetics.
Cell culture and transfection
HEK (human embryonic kidney)-293 and HEK-293 MSR cells (Life Technologies) were cultured in DMEM (Dulbecco's modified Eagle's medium)/Ham's F12 medium supplemented with 10% fetal bovine serum, 2 mM L-glutamine, 100 units/ml penicillin and 100 μg/ml streptomycin.
For the functional assay using HS-YFP [halide-sensitive YFP (yellow fluorescent protein)] , HEK-293 MSR cells were seeded in 96-well microplates (30000 cells/well) in 150 μl of antibiotic-free culture medium. After 24 h, cells were co-transfected with plasmids carrying the coding sequence for TMEM16A (wild-type, mutant and chimaeric) and for the YFP–H148Q/I152L mutant . For each well, 0.2 μg of total plasmid DNA and 0.5 μl of Lipofectamine™ 2000 (Invitrogen) were first pre-mixed in 50 μl of Opti-MEM medium (Invitrogen) to generate transfection complexes (60 min at room temperature, 20–24°C), and then added to the cells. After 24 h, the complexes were removed by replacing with fresh culture medium. The YFP-based functional assay was performed after 24 h.
For immunofluorescence, HEK-293 MSR cells were seeded in a μ-Chamber 12 well (Ibidi) at a density of 25000 cells/well and transfected with the protocol described above.
For immunoblotting, HEK-293 MSR cells were seeded in 60-mm Petri dishes. Sub-confluent monolayers were transfected using 4 μg of plasmid DNA and 10 μl of Lipofectamine™ 2000 for each Petri dish.
For patch-clamp experiments, 500000 HEK-293 cells in 1.5 ml of antibiotic-free culture medium were mixed with 0.5 ml of Opti-MEM medium containing 2 μg of plasmid DNA and 5 μl of Lipofectamine™ 2000 (previously incubated for 60 min at room temperature to allow formation of lipid–DNA complexes). The cell suspension was then plated as 6–7 drops in 35 mm Petri dishes. After 6 h, 2 ml of culture medium without antibiotics were added to each Petri dish. After 24 h, the complexes were removed by replacement with fresh culture medium. Patch-clamp experiments were carried out 2 days after transfection.
Transiently transfected HEK-293 MSR cells were rinsed twice with 200 μl of PBS and incubated for 30 min with 60 μl of PBS (137 mM NaCl, 2.7 mM KCl, 8.1 mM Na2HPO4, 1.5 mM KH2PO4, 1 mM CaCl2 and 0.5 mM MgCl2, pH 7.4). After incubation, 96-well microplates were transferred to a microplate reader (FluoStar Galaxy; BMG Labtech) equipped with high-quality excitation (ET500/20X: 500±10 nm) and emission (ET535/30M: 535±15 nm) filters for YFP (Chroma Technology). For each well, cell fluorescence was measured for 2 s before and 12 s after injection of 165 μl of modified PBS containing 137 mM KI instead of NaCl (final I− concentration in the well, 100 mM). This solution also contained ionomycin (1 μM) as a Ca2+-elevating agent. After background subtraction, cell fluorescence recordings were normalized for the initial value measured before addition of I−. The signal decay caused by HS-YFP fluorescence quenching was fitted with a double exponential function to derive the quenching rate that corresponds to maximal influx of I− into the cells.
Whole-cell membrane currents were recorded in transiently transfected HEK-293 cells. The extracellular solution had the following composition: 150 mM NaCl, 1 mM CaCl2, 1 mM MgCl2, 10 mM glucose, 10 mM mannitol and 10 mM sodium Hepes, pH 7.4. The pipette (intracellular) solution contained: 130 mM CsCl, 10 mM EGTA, 1 mM MgCl2, 10 mM Hepes and 1 mM ATP, pH 7.4, plus CaCl2 to obtain the desired free Ca2+ concentration: 1 mM for 0.007 μM, 6 mM for 0.088 μM, 7 mM for 0.137 μM, 8 mM for 0.235 μM, 9 mM for 0.529 μM, 9.5 mM for 1.1 μM, 9.76 mM for 2.3 μM, 9.85 mM for 3.6 μM and 9.9 mM for 5.2 μM (calculated with the Ca/Mg/ATP/EGTA Calculator v2.2b available at http://www.stanford.edu/~cpatton/maxc.html). During experiments, the membrane capacitance and series resistance were analogically compensated using the circuitry provided by the EPC7 patch-clamp amplifier. The usual stimulation protocol to generate current–voltage relationships consisted of 600 ms-long voltage steps from −100 to +100 mV in 20 mV increments starting from a holding potential of −60 mV. The waiting time between steps was 4 s. The same protocol was used to calculate the membrane conductance, the activation time constants and the deactivation time constant at different Ca2+ concentrations. Membrane conductance was calculated from the tail current measured at −60 mV following the pulse at +100 mV. The same tail current was fitted with a single exponential function to obtain the deactivation time constant (τdeact). An additional protocol to evaluate the process of deactivation at different membrane potentials consisted of a single pre-pulse at +80 mV followed by steps at different voltages in the range between −80 and +80 mV. Membrane currents were filtered at 1 kHz and digitized at 5 kHz with an ITC-16 (Instrutech) AD/DA converter. Data were analysed using the Igor software (Wavemetrics) supplemented by custom software provided by Dr Oscar Moran (Institute of Biophysics, CNR, Genova, Italy). Data are reported as representative traces or means±S.E.M.
Generation of chimaeric constructs and TMEM16A mutants
The wild-type TMEM16A(abc) coding sequence was cloned into the pcDNA 3.1 plasmid as described previously . The wild-type TMEM16B(ac) coding sequence was obtained from GeneCopoeia.
To generate chimaeras (Table 1), we used a strategy based on gene synthesis provided by GenScript. For this purpose, we introduced a BspEI restriction enzyme site by mutating a single nucleotide at position 1097 of the TMEM16A(abc) coding sequence. This mutation did not change the amino acid sequence, nor the extent of protein expression and activity as shown by functional measurements. By using the BspEI site, an endogenous EcoRI site localized at position 1836, and KpnI and XhoI flanking the cloning site, we could split the TMEM16A coding sequence in three parts of grossly similar length. Accordingly, each chimaera was generated by designing the synthesis of a DNA fragment flanked by a pair of restriction sites and containing the desired combination of TMEM16A and TMEM16B sequence. This fragment was cut and pasted into the wild-type TMEM16A sequence using the proper restriction enzymes. Further details on chimaera generation may be provided on request.
|Chimaera||Regions replaced (aa)|
|TMD1–2||347–380 and 429–450|
|TMD3–4||511–543 and 556–583|
|TMD7–8||779–816 and 879–906|
|Chimaera||Regions replaced (aa)|
|TMD1–2||347–380 and 429–450|
|TMD3–4||511–543 and 556–583|
|TMD7–8||779–816 and 879–906|
The TMEM16A mutants were generated using the QuikChange® XL site-directed mutagenesis kit (Stratagene). All chimaeras and mutants were checked by DNA sequencing.
Transfected HEK-293 MSR cells were rinsed with 200 μl of PBS and incubated with 5 μg/ml wheat germ agglutinin–Alexa Fluor® 488 conjugate (100 μl/well for 10 min at 37°C) to label cell membranes. After two washes with 200 μl of PBS, cells were fixed by adding 100 μl of Bouin's solution (HT10132, Sigma–Aldrich) for 10 min at room temperature. Cells were washed three times, blocked with 1% BSA in PBS for 2 h and then incubated overnight at 4°C with 100 μl of rabbit polyclonal anti-TMEM16A antibody (Ab53212, Abcam) diluted 1:200 in PBS containing 1% BSA and 0.3% Triton X-100. Following incubation with primary antibody, cells were rinsed twice in PBS and incubated with 100 μl of secondary goat anti-rabbit Alexa Fluor® 555 antibody (Invitrogen) diluted 1:200 in PBS with 1% BSA for 1 h in the dark. After further washes, cells were covered with mounting medium and coverslips, and then analysed using a laser-scanning spectral confocal microscope TCS SP2-AOBS (Leica Microsystems). Image analysis was performed using Leica and ImageJ software.
To detect the TMEM16A protein and the chimaeras in Western blot experiments (and immunofluorescence experiments), we chose the polyclonal Ab53212 antibody, developed against three epitopes in the TMEM16A protein. We could not use the monoclonal antibody SP31, which is characterized by better sensitivity and specificity , because it was unable to detect TMD5–LOOP–TMD6 and LOOP3 chimaeras. Transfected HEK-293 MSR cells were lysed in lysis buffer (20 mM Hepes, pH 7, 150 mM NaCl, 1 mM EGTA and 1% Igepal) containing Complete™ protease inhibitor cocktail (Roche). Lysate concentration was quantified using Bradford reagent (Sigma–Aldrich). Total lysates (20 μg) were separated on to a NuPAGE Novex Bis-Tris 4–12% gel (Life Technologies) and transferred on to a nitrocellulose membrane (Bio-Rad Laboratories) for Western blotting. TMEM16A protein was immunodetected by a rabbit polyclonal antibody (Ab53212) at 1:1000 dilution, followed by an HRP (horseradish peroxidase)-conjugated anti-rabbit antibody (Millipore) at 1:10000 dilution. Membranes were also stripped with the Restore Western Blot Stripping Buffer (Thermo Fisher Scientific) and incubated with anti-actin goat polyclonal IgG (I19, Santa Cruz Biotechnology) followed by an HRP-conjugated anti-goat secondary antibody (Santa Cruz Biotechnology).
All antibodies were dissolved in 5% (w/v) non-fat dried skimmed milk powder in TBS-T (Tris-buffered saline with Tween 20). Protein bands were visualized using the ECL (enhanced chemiluminescence) advanced Western blotting detection kit (GE Healthcare).
Direct recording of chemiluminescence was performed using the Molecular Imager ChemiDoc XRS System (Bio-Rad Laboratories).
Protein sequence alignment between TMEM16A (NM_018043.4) and TMEM16B (NM_020373) was performed using ClustalW2.
TMD prediction was performed using various software such as Phobius, TMHMM, Spoctopus, Scampi_m, Thumbup, Mensat3 and Mensat svm. For each TMD, its location was chosen as the largest region comprising the amino acid residues identified as being part of a TMD in all software used.
Data are presented as representative traces/images or as means±S.E.M. Statistical analysis was performed with the InStat software (GraphPad). Significant differences among data were determined with the Student's t test.
Expression of TMEM16A and TMEM16B in HEK-293 cells generated Ca2+-activated Cl− currents with quite different properties (Figure 1). As shown by representative traces and by current–voltage relationships (Figures 1A–1D), TMEM16A-dependent currents were activated when the pipette (intracellular) solution contained a free Ca2+ concentration in the range 0.088–1.1 μM. In contrast, TMEM16B currents required higher Ca2+ concentrations (2.3–5.2 μM). By plotting the whole-cell conductance against the intracellular free Ca2+ concentration at a given membrane potential we found a nearly 10-fold difference in apparent Ca2+ affinity. For example, by fitting the data obtained at +100 mV with the Hill equation we found half-effective concentrations of 0.25 and 3 μM for TMEM16A and TMEM16B respectively (Figure 1E). The Hill coefficients were 2.3 for TMEM16A and 4 for TMEM16B. These values indicate the existence of multiple Ca2+-binding sites for each channel, as also found previously for native CaCCs . It is important to note that the high Ca2+ concentrations required to activate TMEM16B imply conditions that are not optimal for the EGTA Ca2+-chelating agent in the pipette solution. This means that the half-effective concentrations determined for TMEM16B are less precise than those obtained for TMEM16A. The suboptimal buffering capacity of EGTA at high Ca2+ concentrations may also account for the higher Hill coefficient of TMEM16B compared with TMEM16A.
Properties of TMEM16A and TMEM16B as channels
Activation and deactivation kinetics were also different for TMEM16A and TMEM16B currents. In TMEM16A-transfected cells, voltage pulses to positive membrane potentials elicited an instantaneous outward current followed by a time-dependent outward relaxation (Figure 1A) that could be fit with a double-exponential function. The activation kinetics were not significantly affected by intracellular Ca2+ concentration or membrane potential. For example, the time constant corresponding to the fastest component, τact, was 300–400 ms, independent of the Ca2+ concentration (Figure 1F) or the applied membrane potential (Supplementary Figure S1A at http://www.biochemj.org/bj/452/bj4520443add.htm). We also looked at the slow component. Despite an apparent trend, the largely variable behaviour among different experiments did not allow us to make conclusions about the Ca2+ dependence of the slow activation process (Supplementary Figure S1B). When the membrane potential was returned to the negative holding membrane potential (−60 mV), a large fraction of the current decayed in a single exponential function. The time constant of current deactivation, τdeact, was not affected by the membrane potential (Supplementary Figure S1C), but was instead modified by the intracellular Ca2+ concentration (Figure 1G). By changing the Ca2+ concentration from 0.137 μM to 1.1 μM, τdeact increased from 40.8±4.1 to 119±4.1 ms (n=3–9).
In TMEM16B-transfected cells, Ca2+-activated Cl− currents had much faster kinetics. Furthermore, TMEM16B activation lacked the slow component that was observed for TMEM16A and therefore it could be clearly fitted to a single exponential function. In the presence of 2.3 μM intracellular Ca2+, τact was 23.5±0.8 ms, whereas τdeact was 7.3±0.4 ms (n=3). Similarly to TMEM16A, activation kinetics did not vary as a function of Ca2+ and voltage (Figure 1F and Supplementary Figure S1A). Deactivation kinetics showed a modest sensitivity to membrane potential, with a slight increase of τdeact upon membrane depolarization (Supplementary Figure S1C).
TMEM16A and TMEM16B are close paralogues and show more than 60% identity at the amino acid level. The most conserved regions correspond to TMDs (70–90% identity), whereas the N- and C-termini are more divergent (48% and 35% identity respectively). Considering the different characteristics of TMEM16A and TMEM16B currents, we hypothesized that generation of chimaeric channels could represent a useful strategy to investigate the molecular mechanisms responsible for TMEM16A channel gating by Ca2+ and voltage. Therefore we generated various chimaeras by replacing selected regions in the TMEM16A protein sequence with equivalent regions of TMEM16B: the N-terminus (chimaera named as N-TERM), the C-terminus (C-TERM), the first intracellular loop (LOOP1) and the TMDs in pairs (TMD1–2, TMD3–4 and TMD7–8; see the cartoon in Figure 2A and Table 1). Regarding the chimaera involving the fifth and the sixth TMDs, we initially considered the topology proposed by Yang et al. . Together with the two TMDs we also transferred the putative re-entrant loop, which was proposed to form the channel pore (chimaera named TMD5–LOOP–TMD6). However, to take into account the revised topology proposed by Yu et al. , we generated an additional chimaera, partially overlapping the one described above, and corresponding to the third intracellular loop (chimaera named LOOP3). Finally, we also generated a chimaeric protein consisting of the TMEM16A sequence from the N-terminus to the fifth TMD and the TMEM16B sequence from the fifth TMD to the C-terminus (chimaera named 16A–16B).
Expression and function of TMEM16A–TMEM16B chimaeras
A first evaluation of the function and expression of TMEM16A–TMEM16B chimaeras was performed with the HS-YFP assay, Western blots and immunofluorescence. The HS-YFP assay allowed rapid functional evaluation of the different constructs. With this technique, CaCC activity is detected as an increase in the quenching rate of the fluorescent protein caused by Ca2+-dependent I− influx . With the exception of the 16A–16B chimaera, expression of all of the constructs caused the appearance of a Ca2+-dependent anion transport that was markedly higher than that of mock-transfected cells (Figure 2B). However, there were clear quantitative differences. Anion transport in TMEM16A-transfected cells was nearly 5-fold higher than that elicited by TMEM16B expression, in agreement with their different Ca2+-sensitivity. Anion transport for N-TERM, TMD1–2 and C-TERM was close to that detected for TMEM16A. The other constructs showed lower anion transport rates, with TMD5–LOOP–TMD6 and LOOP3 having values close to those of TMEM16B (Figure 2B).
To evaluate protein expression and subcellular localization, we carried out Western blot and immunofluorescence experiments using a commercially available polyclonal anti-TMEM16A antibody. In Western blot experiments, lysates from cells transfected with TMEM16A showed, as expected, a band migrating with an apparent size of 120 kDa (Figure 2C). A 120 kDa band of comparable intensity was seen also for N-TERM, TMD1–2, TMD3–4, TMD7–8 and C-TERM (see densitometry from multiple experiments in Figure 2D). The expression of the mature form of the protein was less intense for chimaeras LOOP1, TMD5–LOOP–TMD6 and LOOP3, and totally absent for the 16A–16B construct. Interestingly, an additional band was observed in some lanes at a higher molecular mass, 250 kDa. The pattern of expression of this band was more variable, but, in general, more abundant in relative terms for N-TERM, TMD3–4, TMD5–LOOP–TMD6 and LOOP3 (Figures 2C and 2D). The high molecular mass band probably corresponds to a dimer, as shown previously by others for TMEM16A protein [24,25].
Immunofluorescence experiments revealed for both wild-type and most chimaeric TMEM16A proteins a marked staining of the cell periphery consistent with a localization in the plasma membrane (Figure 2E). A similar pattern was obtained by labelling cells with wheat germ agglutinin, which binds to membrane glycoproteins and glycolipids. In contrast, the 16A–16B chimaera showed an exclusive cytosolic staining, compatible with an accumulation of an unfolded protein. Some intracellular staining was also observed in cells expressing TMD5–LOOP–TMD6 and LOOP3 chimaeras.
Next, we characterized chimaeras using the whole-cell patch-clamp technique performed on HEK-293 transfected cells. The first set of experiments was carried out with an intracellular free Ca2+ concentration of 0.235 μM. Replacement of the N-terminus, the first intracellular loop, TMD1–2, and TMD3–4 did not grossly alter the appearance of TMEM16A-dependent currents (Figure 3A). Instead, replacement of TMD7–8 or of the C-terminus appeared to accelerate activation and deactivation kinetics. Negligible membrane currents were recorded in cells transfected with TMD5–LOOP–TMD6 and LOOP3 chimaeras. Raising the Ca2+ concentration to 0.529 μM evoked relatively large membrane currents in cells transfected with TMD5–LOOP—TMD6, but not with LOOP3 (Figure 3B). To induce large currents in cells transfected with the LOOP3 chimaera, the intracellular Ca2+ concentration had to be raised to the micromolar range. At 2.3 and 3.6 μM Ca2+, the membrane currents recorded in LOOP3-transfected cells were comparable in amplitude with those recorded for wild-type TMEM16A at 0.529–1.1 μM Ca2+ (Figure 4A). Accordingly, the apparent Ca2+ affinity was shifted to higher values by nearly 8-fold, with a half-effective Ca2+ concentration of 2 μM at +100 mV and 3.1 μM at −100 mV (Figures 4C and 4D). Similarly to TMEM16A, the activation kinetics of the LOOP3 chimaera did not change as a function of Ca2+. The values of the time constant describing the current activation were close to those of wild-type TMEM16A (Figure 4E). Instead, the kinetics of deactivation of the LOOP3 chimaera was sensitive to Ca2+, as also found for TMEM16A (Figure 4F). Where comparison was possible (i.e. at Ca2+ concentrations of 0.529 and 1.1 μM), the τdeact values of LOOP3 were significantly smaller than those of TMEM16A (Figure 4F).
Membrane currents elicited by TMEM16A–TMEM16B chimaeras
Properties of the LOOP3 chimaera
In contrast with LOOP3, the TMD5–LOOP–TMD6 chimaera showed only a relatively modest decrease in apparent Ca2+ affinity. Indeed, the half-effective Ca2+ concentration at +100 mV was 0.55 μM for the chimaera and 0.25 μM for TMEM16A (Figure 5G). We also analysed the behaviour of the other constructs. We found no decrease in Ca2+ sensitivity for N-TERM, LOOP1, TMD1–2, TMD3–4, TMD7–8 and C-TERM. Actually, for all of them, except TMD7–8, there was a small shift of the conductance relationship towards lower Ca2+ values (Figure 5).
Ca2+ sensitivity of TMEM16A–TMEM16B chimaeras
The different Ca2+ sensitivity between the LOOP3 and TMD5–LOOP–TMD6 constructs was surprising. We reasoned that, if LOOP3 shows a Ca2+ sensitivity that is close to that of TMEM16B, it should contain critical residues that are not possessed by the TMD5–LOOP–TMD6. However, as shown in Figure 6(A), there was a large degree of overlap between the two sequences. In particular, both chimaeras contained a sequence of TMEM16B that includes the two critical glutamic acid residues that mediate Ca2+ binding in TMEM16A  and a region of maximal sequence diversity between TMEM16A and TMEM16B (see asterisks in Figure 6A). We noted, however, that the LOOP3 sequence includes 23 amino acids that are not present in the TMD5–LOOP–TMD6 construct. This segment in TMEM16A has three amino acids, Ile757, Val772 and Ala776, that in TMEM16B are replaced by valine, aspartic acid and threonine respectively (Figure 6A). We introduced these changes in the TMEM16A backbone and studied the anion transport in HEK-293 cells. We found that the triple mutant I575V/V772D/A776T had a decreased activity, but not comparable with that of the LOOP3 chimaera (Figures 6B and 6C). For unknown reasons, we could not test multiple Ca2+ concentrations for this mutant due to large unstable currents. Therefore we could not generate a full relationship between conductance and Ca2+ concentration. We also introduced the mutations E702Q and E705Q in the TMEM16A protein that were found recently to strongly decrease apparent Ca2+ affinity . We found that these mutations indeed abolished channel activity at Ca2+ concentrations equal or lower than 1 μM. High Ca2+ concentrations, above 3.6 μM, were required to evoke significant membrane currents (Supplementary Figure S2 at http://www.biochemj.org/bj/452/bj4520443add.htm).
Mutations in the third intracellular loop
During the study of the behaviour of the other chimaeras, we found that TMD7–8 and C-TERM were characterized by faster kinetics compared with TMEM16A (Figure 7). This characteristic was particularly evident for TMD7–8. This chimaeric protein had faster activation and deactivation at high Ca2+ concentrations relative to wild-type TMEM16A (Figures 7C and 7D). Interestingly, the C-TERM chimaera had an additional unexpected characteristic. Expression of this protein was associated with the appearance of a Ca2+-independent component (Figures 7E and 7F). At +100 mV, and an intracellular free Ca2+ concentration of 0.007 μM, the membrane currents had a value of 659.4±162.3 pA for C-TERM and 65.1±6.8 pA for wild-type TMEM16A (n=8). The latter value was indistinguishable from that of HEK-293 cells without TMEM16A expression. These findings induced us to further investigate the C-terminus of TMEM16A.
Properties of TMD7–8 and C-TERM chimaeras
In a previous study, we began to investigate the role of the TMEM16A C-terminus by truncating the cytosolic tail downstream of the last TMD . We found that the removal of a large part of the C-terminus (44 amino acids) did not alter channel activity as evaluated by the HS-YFP assay and the patch-clamp technique. In the present study, we have expanded the truncation to the entire region downstream of the eighth TMD (77 amino acids removed; Figure 8A). Truncated TMEM16A (W906X) was still able to traffic to the plasma membrane and to transport iodide (Figures 8B and 8C). Whole cell recordings, in the presence of various intracellular Ca2+ concentrations, showed currents similar to that of wild-type protein in terms of Ca2+ sensitivity (Figures 8D–8F). Indeed, the plot of conductance against Ca2+ concentration was essentially identical with that of wild-type TMEM16A (Figure 8F). Interestingly, the truncated TMEM16A protein, lacking the entire cytosolic C-terminus, did not show the Ca2+-independent component found for the C-TERM chimaera (see the current–voltage relationship in Figure 8E). However, the W906X mutant showed a clear difference with respect to wild-type TMEM16A. Indeed, the activation and deactivation kinetics of the W906X mutant were markedly accelerated. The difference between the mutant and wild-type protein was particularly evident at high Ca2+ concentrations (Figures 8G and 8H). In particular, the time constant for current deactivation became Ca2+-independent in the W906X mutant (Figure 8H).
Truncation of the TMEM16A C-terminus
TMEM16A and TMEM16B proteins appear to be major components of CaCCs [13,14,20]. Knockdown of TMEM16A and TMEM16B expression in different cell models and in animals causes inhibition of endogenous CaCCs [3,7,13,20,27–30]. Furthermore, expression of TMEM16A and TMEM16B in null systems evokes the appearance of large membrane currents modulated by Ca2+ and membrane potential [2,13,14,16,19,20,31].
Despite a relatively high level of amino acid identity (>60%), TMEM16A and TMEM16B show different properties. Therefore the analysis of the regions showing sequence divergence may reveal important domains affecting ion channel behaviour. For this purpose, we have replaced regions of TMEM16A with the equivalent regions of TMEM16B to look for a change in their properties.
Our results indicate that most domains of TMEM16B can be transplanted into the TMEM16A backbone without a dramatic change in its properties. In particular, the first intracellular loop of TMEM16A appears to be an important domain controlling Ca2+- and voltage-sensitivity [16,17]. However, a TMEM16A protein containing the first intracellular loop of TMEM16B did not show a decrease in apparent Ca2+ affinity, despite a 35% difference in amino acid sequence. In particular, the TMEM16B sequence has a ERAQ instead of the EAVK domain, which has been shown to strongly affect Ca2+ sensitivity in TMEM16A . Our findings indicate that the ERAQ sequence in TMEM16B is not a major determinant of the difference in Ca2+ sensitivity with respect to TMEM16A. Instead, we found that replacement of the region corresponding to the third intracellular loop changed the Ca2+-sensitivity of the chimaeric TMEM16A protein so that it became more similar to that of TMEM16B. Indeed, the half-effective Ca2+ concentration of the LOOP3 chimaera and TMEM16B at +100 mV was 2 and 3 μM respectively, whereas that of the wild-type TMEM16A protein was 0.25 μM. Although the values calculated for TMEM16B and LOOP3 chimaera may not be precise due to the low buffering capacity of EGTA at high Ca2+ concentrations, our results indicate that the Ca2+ sensitivity of these proteins is indeed different from that of TMEM16A.
We found that LOOP3 had a partial trafficking problem, with a certain level of the protein remaining in intracellular compartments as shown by immunofluorescence analysis. This conclusion appeared to be supported by the finding of a reduced intensity of the protein detected at 120 kDa, with a parallel increase in the intensity of a band with a slightly reduced molecular mass, which probably corresponds to an immature (partially glycosylated) form of the protein. However, the reduced targeting of the chimaeric protein to the plasma membrane does not explain by itself the shift in Ca2+-sensitivity. If the protein had only a problem with trafficking, the relationship between conductance and Ca2+ should not show a shift of the half-effective concentration, but only a decrease in the maximal value. Furthermore, it is important to note that two of the three epitopes that were used to generate the polyclonal antibody of our Western blot experiments are localized in the first and third intracellular loops. Therefore it is possible that the antibody does not detect the LOOP1 and LOOP3 chimaeras well, thus leading to an underestimation of actual protein expression and maturation.
Our functional data suggest that the LOOP3 chimaera has a Ca2+ sensitivity that resembles in large part that of the TMEM16B channel. Interestingly, this region includes the two glutamic acid residues that appear to be responsible for a large fraction of the Ca2+ sensitivity in the TMEM16A channel . However, the two glutamic acid residues are also present in the TMEM16B protein at equivalent positions. Therefore, additional amino acid residues in intracellular loop 3 appear to mediate Ca2+ sensitivity. The present study, based on the generation of chimaeras and mutagenesis of specific sites, has not revealed single amino acids that modulate Ca2+ sensitivity by themselves. Actually, our results suggest a more global effect of loop 3 in determining the response of the channels to changes in Ca2+ concentration. It is possible that the glutamic acid residues at position 702 and 705 in the TMEM16A protein, and the corresponding ones in TMEM16B, are the direct determinants of Ca2+ binding, possibly by electrostatic interaction between the divalent cation and the negative charges of the carboxylic groups. Differences in the sequence and structure of the intracellular loop 3 between TMEM16B and TMEM16A could then affect the strength of interaction between Ca2+ and the pair of glutamic acid residues. This interpretation appears to be supported by experimental findings. Mutagenesis of Glu702 and Glu705 to remove their negative charges shifts the Ca2+ sensitivity by nearly 100-fold . Instead, replacement of the intracellular loop 3 of TMEM16A with that of TMEM16B shifts the Ca2+ sensitivity by only 8-fold.
Interestingly, the LOOP3 chimaera had a reduced Ca2+ sensitivity, resembling in part that of the TMEM16B channels, but almost normal TMEM16A-like kinetics. Indeed, the rates of activation and deactivation of the chimaeric protein were not as high as those of the TMEM16B protein. This finding indicates that the process underlying the transitions of the channel between the closed and the open state is separated from Ca2+ binding. Different models have been proposed to explain the kinetics of CaCCs [23,32–34]. All models propose multiple sequential closed states where Ca2+ binding occurs. More than one Ca2+-binding site is postulated to exist to take into account the steepness of the relationship between conductance and Ca2+ concentration. Indeed, the fitting of this relationship with the Hill equation gives consistently values that are higher than one. Furthermore, all models postulate that the transition from the closed to the open state is not influenced directly by Ca2+ or membrane potential. The models differ in the site of voltage sensitivity. In one model, the voltage affects the Ca2+-binding to the closed channel . Accordingly, membrane depolarization promotes the transition of the channel to the last closed state that precedes opening. According to this model, voltage also affects the open-to-closed step. In another model, Ca2+ binds to multiple closed states in a voltage-independent manner and the channel can open from each one of these states . The voltage sensitivity in this case would reside only in the open-to-closed transitions . As stated by the authors [32,33], no model perfectly describes the behaviour of the native CaCC channels . The results of the present study are consistent with a model in which channel opening is not directly controlled by membrane potential or Ca2+, as indicated by the little changes shown by the rate of current activation under different conditions. We also favour a scheme in which Ca2+ binding to the closed states is influenced by membrane potential. According to this scheme, membrane depolarization shifts the equilibrium between states so that more channels are moved to the last state preceding channel opening. Also, this scheme predicts that channels cannot open in the absence of Ca2+, by the simple action of depolarization, as also found in our experiments . However, we also found that channel deactivation is not influenced by membrane potential, but is Ca2+-sensitive. The mechanism to explain this behaviour is unclear.
Interestingly, TMD7–8 and C-TERM chimaeras showed higher activation and deactivation rates, without a big change in apparent Ca2+ affinity. The values of the time constants describing the activation and deactivation, particularly at high Ca2+ concentrations, were smaller than those of the TMEM16A protein, although still large if compared with TMEM16B (compare Figures 1F and 1G with Figure 7). These findings would suggest that the C-terminal region in TMEM16A and TMEM16B influences the energy barrier separating the closed from the open state. This conclusion seems to be also supported by the finding that total removal of the C-terminus after the last TMD (W906X mutant) markedly accelerates the process of activation and deactivation.
Surprisingly, the C-TERM chimaera also displayed an additional feature. This protein was characterized by the appearance of a Ca2+-independent component of channel activity. At a very low Ca2+ concentration (7 nM), a concentration at which both wild-type TMEM16A and TMEM16B are totally inactive, C-TERM showed significant membrane currents. Since this feature is not a characteristic of TMEM16B, it probably derives from a perturbation of TMEM16A structure generated from the replacement of the C-terminus. This perturbation, which may involve an altered docking with another region of the protein, results in a partially ‘leaky’ channel.
In summary, the present study supports the conclusion that the third intracellular loop is a main domain controlling the Ca2+ sensitivity of TMEM16 channels. This region contains two glutamic acid residues, shared by TMEM16A and TMEM16B, which may be directly involved in Ca2+ binding . Other amino acids in the same region may be responsible in the tuning of Ca2+ affinity and therefore in determining the different Ca2+ sensitivity between TMEM16A and TMEM16B. Importantly, none of the different chimaeras displayed characteristics completely resembling those of TMEM16B. Therefore the properties of TMEM16A and TMEM16B as CaCCs appear to be determined by different domains. In this respect, the present study represents a framework for future studies.
Paolo Scudieri, Elvira Sondo and Emanuela Caci carried out the experiments. Roberto Ravazzolo and Luis Galietta analysed the data and designed the experiments. Paolo Scudieri and Luis Galietta wrote the paper.
This study was supported by the Telethon Foundation [grant number GGP10026] and by the Italian Cystic Fibrosis Foundation [grant number FFC #2/2012].