Gating modes of calcium‐activated chloride channels TMEM16A and TMEM16B

Key points

• Calcium‐activated chloride channels TMEM16A and TMEM16B support important physiological processes such as fast block of polyspermy, fluid secretion, control of blood pressure and sensory transduction.

• Given the physiological importance of TMEM16 channels, it is important to study how incoming stimuli activate these channels. Here we study how channels open and close and how the process of gating is regulated.

• We show that TMEM16A and TMEM16B display fast and slow gating. These gating modes are regulated by voltage and external chloride.

• Dual gating explains the complex time course of the anion current.

• Residues within the first intracellular loop of the channel influence the slow gating mode.

• Dual gating is an intrinsic property observed in endogenous calcium‐activated chloride channels and could be relevant to physiological processes that require sustained chloride ion movement.

Abstract

TMEM16A and TMEM16B are molecular components of the physiologically relevant calcium‐activated chloride channels (CaCCs) present in many tissues. Their gating is dictated by membrane voltage (V _m), intracellular calcium concentrations ([Ca²⁺]_i) and external permeant anions. As a consequence, the chloride current (I _Cl) kinetics is complex. For example, TMEM16A I _Cl activates slowly with a non‐mono‐exponential time course while TMEM16B I _Cl activates rapidly following a mono‐exponential behaviour. To understand the underlying mechanism responsible for the complex activation kinetics, we recorded I _Cl from HEK‐293 cells transiently transfected with either TMEM16A or TMEM16B as well as from mouse parotid acinar cells. Two distinct V _m‐dependent gating modes were uncovered: a fast‐mode on the millisecond time scale followed by a slow mode on the second time scale. Using long (20 s) depolarizing pulses both gating modes were activated, and a slowly rising I _Cl was recorded in whole‐cell and inside‐out patches. The amplitude of I _Cl at the end of the long pulse nearly doubled and was blocked by 100 μm tannic acid. The slow gating mode was strongly reduced by decreasing the [Cl⁻]_o from 140 to 30 mm and by altering the sequence of the first intracellular loop. Mutating ₄₈₀RSQ₄₈₂ to AVK in the first intracellular loop of TMEM16B nearly abolished slow gating, but, mutating ₄₄₈AVK₄₅₁ to RSQ in TMEM16A has little effect. Deleting ₄₄₈EAVK₄₅₁ residues in TMEM16A reduced slow gating. We conclude that TMEM16 CaCCs have intrinsic V _m‐ and Cl⁻‐sensitive dual gating that elicits complex I _Cl kinetics.

Key points

• Calcium‐activated chloride channels TMEM16A and TMEM16B support important physiological processes such as fast block of polyspermy, fluid secretion, control of blood pressure and sensory transduction.

• Given the physiological importance of TMEM16 channels, it is important to study how incoming stimuli activate these channels. Here we study how channels open and close and how the process of gating is regulated.

• We show that TMEM16A and TMEM16B display fast and slow gating. These gating modes are regulated by voltage and external chloride.

• Dual gating explains the complex time course of the anion current.

• Residues within the first intracellular loop of the channel influence the slow gating mode.

• Dual gating is an intrinsic property observed in endogenous calcium‐activated chloride channels and could be relevant to physiological processes that require sustained chloride ion movement.

Abbreviations

BME

basal medium Eagle

CaCC

calcium‐activated chloride channel

DMEM

Dulbecco's modified Eagle medium

E_r

reversal potential

G

conductance

HEK‐293

human embryonic kidney 293 cells

I_Cl

chloride current

I_tail

tail current

MEM

minimum essential medium

p

permeability

TA

tannic acid

TMEM16

transmembrane protein 16

V_m

membrane voltage

W

fractional contribution

WT

wild type

Introduction

Since their discovery, almost 40 years ago (Bader et al. 1982; Miledi, 1982), calcium‐activated chloride channels (CaCCs) have been actively studied because they fulfil many important physiological functions such as smooth muscle contraction, control of blood pressure, regulation of cardiac and neuronal excitability, fluid secretion in exocrine glands, sensory transduction and cell proliferation (Large & Wang, 1996; Hartzell et al. 2005; Duan, 2009; Romanenko et al. 2010; Duvvuri et al. 2012; Qu et al. 2014; Matchkov et al. 2015). The molecular identity of CaCCs was revealed in 2008 by three independent groups (Caputo et al. 2008; Schroeder et al. 2008; Yang et al. 2008). These groups used different approaches that uncovered two members of the TMEM16 family of transmembrane proteins, TMEM16A (ANO1) and TMEM16B (ANO2), as key structural components of CaCCs.

The CaCC gating mechanism has been studied by changing the [Ca²⁺]_i within the physiological range and using voltage protocols consisting of steps lasting less than a few seconds. Such manoeuvres helped establish the role of membrane voltage (V _m) and Ca²⁺ in gating (Arreola et al. 1996; Nilius et al. 1997; Kuruma & Hartzell, 1999 a; Pérez‐Cornejo et al. 2004). It is also well established that the kinetics of endogenous Ca²⁺‐activated Cl⁻ currents in Xenopus oocytes are complex and cannot be described by a mono‐exponential function (Arreola et al. 1996; Pérez‐Cornejo et al. 2004). Some reports suggested the presence of two Ca²⁺‐activated Cl⁻ conductances to explain the complex kinetics (Boton et al. 1989). In other cases, up to three Cl⁻ conductances were reported after increasing intracellular Ca²⁺ but the complex behaviour was explained by assuming the presence of only one type of channel activated by Ca²⁺ in a V _m‐dependent manner (Kuruma & Hartzell, 1999 a,1999 b). In addition to Ca²⁺ and V _m, permeant anions also influence I _Cl kinetics. For example, in rat parotid acinar cells, anions with permeability ratios (p_x/p _Cl) > 1 act by accelerating activation kinetics in a V _m‐independent manner and slowing deactivation kinetics. Furthermore, the V _m needed to reach half maximum activation is decreased in a linear fashion as p_x/p _Cl increases (Pérez‐Cornejo et al. 2004). These data indicate that permeation is coupled to gating.

Gating of TMEM16A and TMEM16B in heterologous expression systems also shows a complex mechanism (Xiao et al. 2011; Betto et al. 2014). Both channels share the same anionic selectivity and show strong outward rectification when [Ca²⁺]_i is in the sub‐micromolar range (Yang et al. 2008; Pifferi et al. 2009; Stephan et al. 2009; Xiao et al. 2011; Cenedese et al. 2012; Adomaviciene et al. 2013; Betto et al. 2014). As with I _Cl generated by endogenous CaCC, I _Cl generated by activation of TMEM16A cannot be described by a single exponential function (Scudieri et al. 2013; Xiao & Cui, 2014). In contrast, TMEM16B I _Cl follows a mono‐exponential time course and appears to reach steady‐state at depolarized potentials (Adomaviciene et al. 2013). In both TMEM16A and TMEM16B, gating is modified when external Cl⁻ is substituted by other anions. Anions that permeate better than Cl⁻ increase the apparent open probability of TMEM16A (Xiao et al. 2011) and slow activation and deactivation of TMEM16B while the opposite was observed with less permeant anions (Betto et al. 2014). Strikingly, the apparent affinity of TMEM16A and TMEM16B for intracellular Ca²⁺ increases when highly permeant anions are present in the bath solution (Betto et al. 2014; Reyes et al. 2014), another indication that permeation and gating are coupled.

These facts suggest that TMEM16A and TMEM16B could have more than one gating mode perhaps regulated by permeant anions. To gain further insights into the gating mechanisms of these Cl⁻ channels we analysed the V _m‐dependent activation of TMEM16A and TMEM16B channels in transfected HEK‐293 cells and in dissociated acinar cells. Our data show that TMEM16A, TMEM16B and endogenous CaCC displayed fast and slow gating modes which depend on external Cl⁻ and V _m.

Methods

Cell culture and transfection

Human embryonic kidney (HEK) 293 cells were cultured in Dulbecco's modified Eagle medium (DMEM from GIBCO; Carlsbad, CA, USA) supplemented with 10% FBS and 0.1% penicillin–streptomycin at 37°C in a 95% O₂/5% CO₂ atmosphere. The mouse TMEM16A cDNA (ac variant) or mouse TMEM16B (retinal isoform containing exon 13) were cloned into pIRES2‐EGFP (Clontech, Mountain View, CA, USA) and pEGFP‐N1 vector, respectively. TMEM16A/RSQ was constructed by mutating ₄₄₉AVK₄₅₁ residues in the first intracellular loop of TMEM16A into RSQ. Conversely, TMEM16B/AVK was constructed by mutating ₄₈₀RSQ₄₈₂ residues in the first intracellular loop of TMEM16B into AVK. TMEM16A Δ₄₄₈EAVK₄₅₁ was made by deleting residues ₄₄₈EAVK₄₅₁ from the first intracellular loop of TMEM16A. Mutations were made using a Quick change protocol (Agilent, Santa Clara, CA, USA). HEK‐293 cells were transfected with 1 μg μl⁻¹ of the cDNA using Polyfect transfection reagent (Qiagen, Valencia, CA, USA), according to the manufacturer's instructions. Cells were used 12 h after transfection and were seeded at low density for patch‐clamp experiments. For inside‐out experiments, HEK‐293 cells stably transfected with TMEM16A were plated onto poly‐l‐lysine‐coated coverslips.

Single parotid acinar cell dissociation

Single acinar cells were isolated from parotid glands dissected from C57BL/6 mice, 2–3 weeks old, as previously reported (Casas‐Pruneda et al. 2009). Animals were anaesthetized deeply with CO₂ prior to tissue collection and then killed by exsanguination, a procedure in agreement with the Official Mexican Norm NOM‐062‐ZOO‐199 (rules for the production, care and use of laboratory animals). Glands were minced in Ca²⁺‐free minimum essential medium (MEM; Gibco) supplemented with 1% BSA (Fraction V; Sigma, St Louis, MO, USA). Tissue was then treated for 20 min (37°C) with a 0.02% trypsin solution (MEM‐Ca²⁺‐free + 1 mm EDTA + 2 mm glutamine + 1% BSA). Digestion was stopped with 2 mg ml⁻¹ of soybean trypsin inhibitor (Sigma) and the tissue was further dispersed by two sequential treatments of 60 min each with collagenase (100 U ml⁻¹ Type CLSPA; Worthington Biochemical, Lakewood, NJ, USA) in MEM‐Ca²⁺‐free + 2 mm glutamine + 1% BSA. The dispersed cells were centrifuged and washed with basal medium Eagle (BME; Gibco)/BSA‐free. The final pellet was re‐suspended in BME/BSA‐free + 2 mm glutamine, and cells were plated onto poly‐l‐lysine‐coated glass coverslips (5 mm diameter) for electrophysiological recordings.

Electrophysiological recordings and solutions

Transfected HEK‐293 and acinar cells were studied using conventional whole‐cell and inside‐out patch‐clamp configurations. An inverted microscope equipped with LED illumination was used to identify green fluorescent protein (GFP) fluorescent cells. Patch pipettes were fabricated from Corning 8161 glass capillaries (Warner Instruments, Hamden, CT, USA) using a PP830 electrode puller (Narishige, Tokyo, Japan); the electrode resistance was 3–5 MΩ for whole cells or 1–2 MΩ for inside‐out patches. The stimulation protocol consisted of V _m steps from a holding potential of −60 mV. The steps were from −100 to +140 mV for depolarizations lasting 0.5 s and from −20 to +140 mV for 20 s duration pulses. Cells were repolarized to −100 or −60 mV, respectively. The time interval between two pulses was 7 and 45 s for depolarizations of 0.5 and 20 s, respectively. Data were acquired using an Axopatch 200B amplifier and pClamp10 software (Molecular Devices, Sunnyvale, CA, USA), filtered at 5 kHz and digitized at 10 kHz. The bath was grounded using 3 m KCl agar‐bridge connected to an Ag/AgCl reference electrode. The experiments were performed at room temperature (21–23°C).

Different intracellular solutions were prepared as indicated in Table 1. In all cases, the osmolarity of the solutions was 290–300 mosmol l⁻¹, and the pH adjusted to 7.3 with TEA‐OH. Additionally, the MAXCHELATOR program was used to calculate the free Ca²⁺ concentration (maxchelator.stanford.edu).

Table 1.

Composition of intracellular solutions (mm) used for electrophysiological experiments

	TMEM16A (whole‐cell)	TMEM16A (inside‐out)	Parotid acinar cells (whole‐cell)	TMEM16B (whole‐cell)
TEA‐Cl	30	17	9.6	20.2
CaCl2	5.24	11.5	15.2	9.9
EGTA‐TEA	25.24	25	25	—
Hepes	50	50	50	50
d‐Mannitol	85	85	85	85
HEDTA	—	—	—	25

The extracellular solution contained (in mm): 139 TEA‐Cl, 20 Hepes, 0.5 CaCl₂, 110 d‐mannitol, adjusted to pH 7.3 with TEA‐OH or NaOH, osmolarity 380–400 mosmol l⁻¹. To evaluate the effect of [Cl⁻]_o TEA‐Cl was reduced from 139 to 29 mm, and d‐mannitol was used to maintain the osmolarity. The external solution was hypertonic to preclude activation of endogenous volume‐sensitive chloride currents (Hernández‐Carballo et al. 2003). Osmolarity values were adjusted by adding d‐mannitol and measured using the vapour pressure point method (VAPRO, Wescor Inc., South Logan, UT, USA). The inhibition of TMEM16A Cl⁻ currents was achieved by applying 100 μm tannic acid to cells bathed in an extracellular solution containing 139 mm NaCl instead of 139 mm TEA‐Cl to avoid precipitation of tannic acid. All chemicals were purchased from Sigma‐Aldrich.

Data analysis

Raw traces of I _Cl were recorded between −100 and +140 mV using depolarizations lasting 0.5 s and from −20 to +140 mV using 20 s depolarization from the same cell bathed first in 140 and then in 30 mm Cl⁻. For a given cell, all I _Cl traces (irrespective of V _m, [Cl⁻]_o or pulse duration) were normalized to the I _Cl value measured at the end of a 0.5 s depolarization to +100 mV in 140 mm Cl⁻. Normalized I _Cl traces from different cells obtained at the same V _m, [Cl⁻]_o and pulse duration were then averaged and plotted. The I–V curves were constructed using the normalized I _Cl traces.

To construct normalized conductance–voltage (G _Norm–V _m) curves we measured the I _Cl magnitude at the end of test pulse and calculated the conductance (G) as I _Cl/(V _m–E _r). E _r, the reversal potential, was determined by interpolation from the tail current vs. V _m plot. Tail currents (I _tail) were generated by repolarizing between +20 and −60 mV in 20 mV decrements after activating the channel with a voltage step to +100 mV that lasted 20 s. The calculated G values were then normalized as follows:

$$G_{\mathrm{Norm},140\mathrm{Cl}}\left(V_{\mathrm{m}}\right)=\frac{G_{140\mathrm{Cl}}\left(V_{\mathrm{m}}\right)}{G_{20\mathrm{s},+100\mathrm{mV},140\mathrm{mM}\mathrm{Cl}}}$$ (1)

$$G_{\mathrm{Norm},30\mathrm{Cl}}\left(V_{\mathrm{m}}\right)=\frac{G_{30\mathrm{Cl}}\left(V_{\mathrm{m}}\right)}{G_{20\mathrm{s},+100\mathrm{mV},140\mathrm{mM}\mathrm{Cl}}}$$ (2)

Finally, we averaged normalized G values from different cells and plotted those as a function of V _m.

To determine activation or deactivation time constants, I _Cl or I _tail were fitted with either mono‐ or bi‐exponential functions (eqn (3)). The Simplex method and the sum of square errors minimization method were used (Clampfit; Molecular Devices):

$$I_{\mathrm{Cl}}=A_f·\left(1-e^{-\frac{\mathrm{t}}{{\tau }_{\mathrm{f}}}}\right)+A_{\mathrm{s}}·\left(1-e^{-\frac{\mathrm{t}}{{\tau }_{\mathrm{s}}}}\right)+C$$ (3)

where τ_f and τ_s are fast and slow time constants. The contribution of each component to the total current was calculated as W _f = A _f/(A _f + A _s + C), W _s = A _s/(A _f + A _s + C) and W _C = C/(A _s + A _f + C). In the figures, fits are shown as red lines, and the dashed black line indicates I _Cl = 0. For clarity, only W _f and W _s are shown in the figures. W _c vs. V _m for TMEM16A, TMEM16B, acinar cells and TMEM16A/RSQ at 140 mm [Cl⁻]_o and for TMEM16A and TMEM16B at 30 mm [Cl⁻]_o shows a linear relashionship (∼0.72 at +20 mV and ∼0.09 at +140 mV). Decreasing [Cl⁻]_o from 140 to 30 mm modified W _c vs. V _m curves in a parallel fashion; W _C increased ∼0.2 and ∼0.3 for TMEM16A and TMEM16B, respectively. All data were analysed using Origin8 (Origin Lab, Northampton, MA, USA). Experimental data are presented as mean ± SEM of n independent experiments. Statistically significant differences between means were determined using a Student's t‐test. An asterisk in the plots indicates statistically significant differences at P ≤ 0.001.

Results

Dual gating in TMEM16A and TMEM16B chloride channels

To analyse the chloride current (I _Cl) kinetics, we recorded I _Cl from HEK‐293 cells expressing wild‐type (WT) TMEM16A or WT TMEM16B that were dialysed with 0.2 or 2.5 μm internal Ca²⁺ concentration, respectively. A V _m step protocol with 0.5 or 20 s pulses that changed the membrane V _m between −20 and +140 mV in increments of 20 mV was used. Figure 1 A and B shows normalized I _Cl values from a cell expressing TMEM16A. With pulses lasting 0.5 s, both I _Cl activation and deactivation followed mono‐exponential kinetics (red lines and inset). At +140 mV, the activating and deactivating time constants were 256.2 ± 32.9 and 44.0 ± 2.2 ms (n = 6), respectively. However, as the duration of the V _m pulse was lengthened from 0.5 to 20 s, the I _Cl kinetics changed rather dramatically, as the current did not reach steady state but instead kept slowly increasing (Fig. 1 B). Average I _Cl at +140 mV with 0.5 and 20 s depolarizations was 1.13 ± 0.13 and 3.4 ± 0.36 nA, which corresponded to 0.12 ± 0.008 and 0.34 ± 0.04 nA pF⁻¹, respectively (n = 6). Figure 1 C shows normalized I _Cl vs. V _m relationships. The graph shows that long depolarizations significantly increased I _Cl (V _m > +100 mV); for example, the current increased nearly 3‐fold at +140 mV. Moreover, activation and deactivation currents were no longer described by a mono‐exponential fit. For example, the inset in Fig. 1 B shows a bi‐exponential fit to I _tail recorded at −60 mV after a depolarization to +140 mV. The resulting fast (τ_f) and slow (τ_s) time constants were 150.6 ± 9.2 ms and 2.6 ± 0.3 s (n = 6), respectively. A detailed analysis of TMEM16A I _Cl kinetics observed with the 20 s depolarizations is shown in Figure 1 D–F. The analysis shows that τ_f and τ_s are V _m‐dependent, and their values increased at depolarized V _m. At +20 mV, τ_f and τ_s were 181.3 ± 28.6 ms and 0.7 ± 0.56 s, respectively, but changed to 448.1 ± 26.3 ms and 19.1 ± 3.79 s at +140 mV. Also, the fractional contribution of the fast component (W _f) displayed a bell shape between +60 and +100 mV and reached a maximum value at +80 mV (Fig. 1 F, magenta circles). The slow component (W _s) was weakly V _m‐dependent and changed from 0.09 ± 0.02 at +20 mV to 0.80 ± 0.04 at +140 mV (Fig. 1 F, cyan circles).

Figure 1. Fast and slow gating in TMEM16A channel .

Whole‐cell chloride currents recorded from HEK‐293 cells stably transfected with TMEM16A that were dialysed with an intracellular solution containing 0.2 μm Ca²⁺. A, normalized I _Cl traces elicited by the V _m protocol depicted at the top. Pulse duration was 0.5 s. Red lines are mono‐exponential fits to I _Cl and I _tail recorded at +140 and −100 mV; from these fits the τ_on = 256.2 ± 32.9 and τ_off = 44.0 ± 2.1 ms were calculated. The y scale bar is the normalized I _Cl at +100 mV; it is also valid for the traces in B. B, normalized I _Cl recordings obtained with the V _m protocol shown above the traces. Pulse duration was 20 s. Inset is a representative bi‐exponential fit to I _tail with τ_f = 150.6 ± 9.2 ms and τ_s = 2.6 ± 0.3 s. C, normalized I _Cl–V _m relationships from short (0.5 s, black circles) or long (20 s, grey circles) depolarization pulses. Current values were normalized as described in the data analysis section. I _Cl values shown in black and grey circles correspond to those obtained with 0.5 or with 20 s pulses, respectively. *P ≤ 0.001 by Student's t‐test; n = 6; mean ± SEM. D–F, fast (τ_f) and slow (τ_s) time constants of activation and relative weight of fast (W _f) and slow (W _s) components as a function of V _m. Time constants were obtained by fitting the raw data with a bi‐exponential function.

Subsequently, we recorded I _Cl from HEK‐293 cells expressing TMEM16B to examine whether this channel displayed fast and slow gating. Cells were dialysed with 2.5 μm Ca²⁺ _, the channels were activated using 0.5 or 20 s depolarizations, and the activation and deactivation kinetics were analysed (n = 9 for ≤+120 mV; n = 4 at +140 mV). Figure 2 A shows I _Cl recorded with 0.5 s test pulses. As described by others (Pifferi et al. 2009; Stephan et al. 2009; Cenedese et al. 2012) I _Cl activation of TMEM16B was faster than TMEM16A. A mono‐exponential function (solid red lines) described both activation and deactivation kinetics of I _Cl with time constants of 22.3 ± 1.9 and 6.4 ± 0.6 ms, respectively. In contrast, 20 s depolarizations uncovered an additional, slow component that resulted in a slow increase in the current. Figure 2 B shows normalized current traces depicting the fast activation followed by the slow component observed in TMEM16B. I _Cl at +140 mV with 0.5 and 20 s depolarizations was 0.65 ± 0.12 (n = 9) and 1.71 ± 0.59 nA (n = 4), which corresponded to 0.05 ± 0.008 and 0.14 ± 0.045 nA pF⁻¹, respectively. Figure 2 C shows normalized I _Cl vs. V _m relationships. The more depolarized the V _m pulses, the larger the I _Cl amplitude; for this channel the current increased nearly 2‐fold at V _m > +120 mV. I _Cl elicited by long depolarizations was fit using a bi‐exponential function to determine the time constants and the relative contribution of each component to the total current. τ_f, τ_s, W _f and W _s were all V _m‐dependent parameters (Fig. 2 D–F) that followed the same trend as in TMEM16A. However, W _f showed a pronounced bell shape that was smaller in TMEM16B than in TMEM16A at V _m < +100 mV.

Figure 2. Fast and slow gating in TMEM16B channel .

Whole‐cell chloride currents recorded from HEK‐293 cells transitorily expressing TMEM16B and dialysed with 2.5 μm Ca²⁺. A, average of normalized I _Cl of TMEM16B induced by the V _m protocol shown in Fig. 1 A. Red lines: mono‐exponential fits to activating and deactivating (inset) currents with τ_on = 22.3 ± 1.9 and τ_off = 6.4 ± 0.6 ms, respectively. The y scale bar is the normalized I _Cl at +100 mV; this is also valid for traces in B. B, average of normalized current recordings obtained with the 20 s depolarizations protocol shown in Fig. 1 B. Inset: bi‐exponential fit to I _tail after a +140 mV depolarization with values of τ_f = 9.8 ± 1.6 ms and τ_s = 2.4 ± 0.3 s. C, normalized I _Cl–V _m relationships from short (0.5 s, black circles) or long (20 s, grey circles) depolarization pulses. *P ≤ 0.001 by Student's t‐test; n = 9 for ≤+120 mV, n = 4 for +140 mV; mean ± SEM. D–F, fast (τ_f) and slow (τ_s) time constants and relative weight of fast (W _f) and slow (W _s) components as function of V _m. Time constant values were obtained as indicated in Fig. 1 D.

To rule out contributions from any Cl⁻ currents intrinsic to HEK‐293 cells that could be activated by long depolarizations, we studied I _Cl in HEK‐293 cells transfected with an empty vector. Figure 3 A shows a typical recording obtained when using 20 s long pulses from −20 to +140 mV, in 20 mV increments. Current amplitudes were practically negligible (n = 6). Furthermore, to prove that I _Cl generated by long depolarizations resulted from activation of TMEM16A, we used 100 μm tannic acid, a concentration that fully inhibited the I _Cl. Figure 3 B shows that 97.2 ± 0.4% of the I _Cl activated with a 20 s pulse to +120 mV was inhibited by tannic acid (black vs. blue traces, n = 7). To further strengthen our assertion that dual gating is an intrinsic property of the channel, we recorded I _Cl from inside‐out patches using long depolarizations from −20 to +120 mV (Fig. 3 C). Dual gating was evident under this experimental condition too, thus ruling out the need for intracellular components or the delivery of additional channels to the plasma membrane (n = 6; except for +120 mV, n = 5). Our data show that TMEM16A and TMEM16B channels have fast and slow gating modes that are V _m‐dependent.

Figure 3. Dual gating is an intrinsic property of TMEM16 channels .

A, representative whole‐cell voltage clamp recordings from mock transfected HEK‐293 cells using the protocol depicted at the top (n = 6). B, inhibition of TMEM16A current (black trace) by acute exposure to tannic acid (TA; 100 μm, blue trace) in the same cell. I _Cl was evoked by a 20 s step depolarization of +120 mV as depicted at the top (n = 7). An external solution containing NaCl was used, in this case to preclude tannic acid precipitation. C, representative inside‐out patch clamp recording of TMEM16A I _Cl using the V _m protocol depicted above the traces. The patch was exposed to 54.7 nm of free intracellular Ca²⁺ and 40/140 mm of intra‐ and extracellular [Cl⁻], respectively (n = 6 macro‐patches; except for +120 mV, n = 5).

As a final test of dual gating, we studied the behaviour of endogenous CaCCs. For this, we recorded currents in acinar cells obtained from freshly dissociated mouse parotid gland (Schroeder et al. 2008; Yang et al. 2008; Romanenko et al. 2010). Acinar cells were dialysed with 0.15 μm internal Ca²⁺ and exposed to 140 mm Cl⁻. I _Cl was generated using a V _m step protocol that changed the membrane V _m during 0.5 or 20 s from −100 to +140 mV in 20 mV increments (n = 5–8; except for +140 mV, n = 4). Figure 4 A shows characteristic recordings obtained using 0.5 s pulses. Activation and deactivation (inset) current kinetics obtained at +140 mV were fitted with a mono‐exponential function (red lines) with time constants of 448 ± 48.0 and 125.5 ± 17.8 ms, respectively. As with TMEM16A and TMEM16B, the 20 s depolarizing pulses revealed the presence of a second gating component (Fig. 4 B). I _Cl at +140 mV with 0.5 and 20 s depolarizations was 0.70 ± 0.12 (n = 8) and 0.93 ± 0.22 nA (n = 4), which corresponded to 0.05 ± 0.008 and 0.07 ± 0.02 nA pF⁻¹, respectively. A bi‐exponential fit to I _tail at −60 mV after a 20 s depolarization to +140 mV gave τ_f and τ_s values of 304.7 ± 34.5 ms and 2.8 ± 0.2 s, respectively (Fig. 4 B, inset). Figure 4 C shows normalized I _Cl vs. V _m relationships; long depolarizations significantly increased the I _Cl amplitude (nearly 1.5‐fold). Also, τ_f and τ_s were V _m‐dependent but different from those determined for TMEM16A (Fig. 4 D and E). At +20 mV, τ_f was 215.6 ± 34.9 ms, while at +140 mV τ_f had a value of 772.8 ± 74.9 ms; likewise, τ_s was 1.2 ± 0.02 s at +20 mV and 10.6 ± 0.5 s at +140 mV. W _f was V _m‐dependent with an approximate value of 0.16 for V _m ≤ +40 mV and 0.50 for V _m ≥ +60 mV (Fig. 4 F, magenta circles). Contrary to data observed in HEK‐293 cells after TMEM16A overexpression, the W _s from endogenous CaCCs in acinar cells remained almost constant around 0.22 at +20 mV to 0.32 at +140 mV (Fig. 4 F, cyan circles). Thus, fast and slow gating modes control activation in endogenous CaCCs.

Figure 4. Endogenous calcium‐activated chloride channels from mouse parotid acinar cells display dual gating .

Currents were recorded from acinar cells dialysed with 0.15 μM free Ca²⁺, and bathed in a solution containing 140 mM Cl⁻. A, membrane potential was varied between −100 and +140 mV in 20 mV increments by delivering step depolarizations of 0.5 s from a holding Vm of −60 mV and then repolarized to −100 mV. B, alternatively, Vm was varied between −20 and +140 mV by delivering step depolarizations of 20 s from a holding potential of −60 mV and then repolarizing to −60 mV. Insets: red lines show mono‐ (τ_on = 448.0 ± 48 ms and τ_off = 125.5 ± 17.8 ms) or bi‐exponential (τ_f = 304.7 ± 35.4 ms and τ_s = 2.8 ± 0.2 s) fits to I _tail recorded after a +140 mV depolarization. The y scale bar in A is the normalized I _Cl at +100 mV; this is also valid for the traces in B. C, normalized I _Cl–V _m relationships from short (0.5 s, black circles) or long (20 s, grey circles) depolarization pulses. *P ≤ 0.001 by Student's t‐test; n = 4–7; mean ± SEM. D and E, V _m dependence of fast and slow time constants, respectively. F, V _m dependence of the relative contribution of fast (W _f) and slow (W _s) components to the total current.

Dual gating depends on extracellular chloride

We have previously reported that the apparent open probability of TMEM16A is altered by decreasing [Cl⁻]_o (Xiao et al. 2011), which could alter current kinetics. Thus, we decided to evaluate the role of external Cl⁻ on the V _m dependence of fast and slow gating modes in TMEM16A and TMEM16B channels. Using 20 s depolarizations we recorded I _Cl from cells bathed in solutions containing 140 and 30 mm Cl⁻. Figure 5 A shows I _Cl recordings from a cell expressing TMEM16A, dialysed with 0.2 μm Ca²⁺ and bathed in 30 mm Cl⁻ (n = 6). At 30 mm [Cl⁻]_o, I _Cl was significantly decreased (about 30‐fold at +140 mV). Figure 5 B shows the conductance determined with 140 (grey circles) and 30 mm (black circles) [Cl⁻]_o. This graph illustrates the sharp decrease in conductance, which indicates a reduction in the apparent open channel probability caused by low external Cl⁻. Figure 5 C–E shows the corresponding τ_f, τ_s, W _f and W _s parameters obtained using 30 mm Cl⁻. The V _m dependence of τ_f and τ_s was similar to that determined using normal [Cl⁻]_o (Fig. 1 D and E). However, the magnitude of these constants was different at depolarized V _m. At +20 mV, τ_f was 126.9 ± 11.9 ms and τ_s was 0.65 ± 0.20 s, which are very close to the values obtained using 140 mm Cl⁻ (τ_f = 181.3 ± 28.6 ms and τ_s = 0.7 ± 0.56 s). In contrast, at +140 mV, τ_f and τ_s values were 390.9 ± 59.1 ms and 8.5 ± 2.9 s, respectively, with 30 mm [Cl⁻]_o and 448.1 ± 26.3 ms and 19.1 ± 3.79 s with 140 mm Cl⁻. Low [Cl⁻]_o also brought a significant variation in the fractional contributions of the W _f and W _s components. Figure 5 E shows that the V _m dependence of W _f was different from that observed with 140 mm Cl⁻ (see Fig. 1 F). With 30 mm Cl⁻, W _f increased from nearly 0 to a plateau value of about 0.54. At depolarized potentials, W _f values were larger in 30 than in 140 mm Cl⁻. In turn, W _s remained below 0.1 at V _m < +100 mV and at +140 mV reached a value of 0.3, which is significantly smaller than the value of 0.8 observed in 140 mm Cl⁻.

Figure 5. Extracellular chloride dependence of TMEM16 channels dual gating .

Top, TMEM16A data; bottom, TMEM16B data. Both data sets were recorded from HEK‐293 cells bathed in a solution containing 30 mm Cl⁻ and stimulated with the protocol shown in Fig. 1 B. A and F, representative I _Cl from cells bathed in a solution containing 30 mm Cl⁻ (n = 6 or 4 for TMEM16A and TMEM16B, respectively). Insets: bi‐exponential fits to I _tail generated by a 20 s depolarization to +140 mV. Values for TMEM16A τ_f = 71.6 ± 11.9 ms and τ_s = 2.6 ± 0.7 s; and for TMEM16B τ_f = 5.9 ± 0.9 ms and τ_s = 2.6 ± 0.4 s. B and G, normalized conductance against V _m at 140 and 30 mm [Cl⁻]_o. *P ≤ 0.001 by Student's t‐test; mean ± SEM. C and H, fast time constants as a function of V _m. D and I, slow time constants as a function of V _m. Time constants were obtained by fitting the I _Cl data shown in A and F with a bi‐exponential function. E and J, relative weight of fast (W _f) and slow (W _s) components of I _Cl as a function of V _m.

For TMEM16B, dual gating also displayed a notable extracellular Cl⁻ dependence. Figure 5 F shows I _Cl recordings from a cell exposed to 30 mm Cl⁻ and dialysed with 2.5 μm Ca²⁺ (n = 4). Compared to I _Cl recorded with 140 mm Cl⁻ (Fig. 2), the I _Cl with low Cl⁻ displayed very fast activation; there was little or no slow component. By estimating the conductance from I _Cl values and comparing it to that obtained with 140 mm Cl⁻ _o (Fig. 5 G), we found that apparent open channel probability sharply decreased just like in TMEM16A. The V _m dependence and magnitude of τ_f (Fig. 5 H) was quite similar in 140 and 30 mm Cl⁻. In contrast, τ_s in 30 mm Cl⁻ had a lower value at all V _m when compared to the time constants obtained with 140 mm Cl⁻ _o. For example, τ_s was nearly 10‐fold smaller at V _m ≥ +100 mV (Fig. 5 I). Figure 5 J shows W _f and W _s as a function of V _m. Note that at 30 mm Cl⁻ most of I _Cl was due to the contribution of W _f while W _s was close to 0 at all V _m. Thus, dual gating in TMEM16A and TMEM16B is strongly affected by external Cl⁻. With low external Cl⁻, the slow gating mode is nearly abolished, and the channels are activated in the fast mode by Ca²⁺ and V _m.

The first intracellular loop of TMEM16 plays a role in dual gating

We have previously reported that Ca²⁺ and V _m dependence in TMEM16A are coupled by the segment ₄₄₈EAVK₄₅₁ located within the first intracellular loop of the channel (Xiao et al. 2011). Interestingly, fast activation and deactivation current kinetics are observed in channels lacking the segment ₄₄₈EAVK₄₅₁ (Xiao et al. 2011), resembling those of TMEM16B (Adomaviciene et al. 2013). Residues ₄₄₉AVK₄₅₁ present in TMEM16A correspond to residues ₄₈₀RSQ₄₈₂ in TMEM16B. Thus, to investigate the role of the first loop in dual gating we replaced ₄₄₉AVK₄₅₁ with RSQ in TMEM16A (TMEM16A/RSQ) and ₄₈₀RSQ₄₈₂ with AVK in TMEM16B (TMEM16B/AVK). The chimeras were expressed in HEK‐293 cells, and their I _Cl kinetics were analysed. Application of 0.5 and 20 s depolarizing pulses to cells expressing TMEM16A/RSQ, dialysed with 0.2 μm Ca²⁺ and bathed in 140 mm Cl⁻, generated I _Cl with typical kinetics and amplitudes similar to those recorded from WT TMEM16A (n = 8). For example, a 0.5 s depolarizing pulse to +140 mV generated a current whose deactivation kinetics are described by a mono‐exponential function with a 33.5 ± 2.6 ms time constant (Fig. 6 A). Prolonging the pulse duration to 20 s produced I _Cl activation and deactivation kinetics identical to that of WT TMEM16A. Thus, I _tail recorded at −60 mV followed a dual behaviour characterized by τ_f = 138.1 ± 14.3 ms and τ_s = 3.2 ± 0.4 s (Fig. 6 A inset). These values are very close to that determined for WT TMEM16A (see Fig. 1 B, inset). The V _m dependences of normalized I _Cl recorded with 0.5 and 20 s pulses (Fig. 6 C) were identical to those obtained from cells expressing WT TMEM16A (Fig. 1 C). I _Cl amplitude at +140 mV in TMEM16A/RSQ increased 2.9‐fold with 20 s pulses relative to shorter pulses, while for WT TMEM16A the increment was 3‐fold. The V _m dependences of τ_f and τ_s were similar for TMEM16A/RSQ and WT TMEM16A (Fig. 6 D, E), although the chimera τ_f at each V _m was nearly 1.3‐fold smaller than τ_f for WT TMEM16A. However, τ_s was similar to that of WT except at +140 mV, where it was 1.5‐fold larger. Finally, the V _m dependence of W _s for chimeric channels (Fig. 6 F, cyan) was nearly identical to that of WT TMEM16A (dashed blue line) and WT TMEM16B (dashed black line). W _f (magenta) was between WT TMEM16A (dashed blue line) and WT TMEM16B (dashed black line).

Figure 6. Dual gating of TMEM16A is partially altered by replacing ₄₄₉AVK₄₅₁ residues in the first intracellular loop .

A and B, average of normalized whole cell Cl⁻ currents recorded using 0.5 (A) or 20 s (B) depolarizations from cells expressing TMEM16A/RSQ mutant channel. The HEK‐293 cells were dialysed with 0.2 μm [Ca²⁺]_i and stimulated with the V _m protocols shown in Fig. 1 A and B, respectively. Insets: mono and bi‐exponential fits (red lines) to I _tail (black traces) that were generated by a depolarization to +140 mV. Corresponding values are τ_on = 210.6 ± 18.1 ms and τ_off = 33.5 ± 2.6 ms or τ_f = 138.1 ± 14.3 ms and τ_s = 3.2 ± 0.4 s for depolarization of 0.5 s and 20 s, respectively. The y scale bar in A is the normalized I _Cl at +100 mV; this is also valid for the traces in B. C, normalized I _Cl–V _m relationships from short (0.5 s, black circles) or long (20 s, grey circles) depolarization pulses. *P ≤ 0.001 by Student's t‐test; n = 8; mean ± SEM. D and E, V _m dependence of fast and slow time constants. F, relative contribution of fast (W _f) and slow (W _s) components to TMEM16A/RSQ total current. Blue and black dashed lines are W _f and W _s from WT TMEM16A and WT TMEM16B, respectively. *P ≤ 0.001, statistically different from TMEM16A WT by Student's t‐test.

Surprisingly, introducing AVK residues in the first intracellular loop of TMEM16B abolished slow gating. Figure 7 A and B shows normalized I _Cl from cells sequentially stimulated with 0.5 or 20 s depolarizing pulses, respectively (n = 7; except for +140 mV, n = 4). Currents generated with short pulses show a typical fast activation with a mono‐exponential time course (red line in Fig. 7 A). In contrast, long depolarizations produced a smaller I _Cl (grey trace) without a slow component at V _m < +120 mV. In Figure 7 C we superimposed the onset of the currents recorded from WT TMEM16B and TMEM16B/AVK at +120 mV. These traces show that the fast component was preserved while the slow component was absent in TMEM16B/AVK. Furthermore, Figure 7 D illustrates that the normalized I _Cl at the end of the depolarizing pulses was smaller with 20 s pulses than with 0.5 s pulses. I _Cl through TMEM16B/AVK at V _m < +120 mV was described with a mono‐exponential function; the resulting V _m dependence of τ_f was similar to that of WT (Fig. 7 E). In some cells, a slow component was evident at V _m > +140 mV, and only in those cases we were able to estimate τ_s (Fig. 7 E, cyan). This observation was accompanied by a significantly decreased W _s. At +140, +160 and +180 mV, W _s was 0.21 ± 0.02, 0.33 ± 0.05 and 0.46, respectively. Interestingly, W _f for TMEM16B/AVK (Fig. 7 F, magenta) was similar to that of WT TMEM16B determined with 30 mm Cl⁻ (blue).

Figure 7. Slow gating is absent in TMEM16B channels carrying AVK instead of ₄₈₀RSQ₄₈₂ residues in the first intracellular loop .

HEK‐293 cells were dialysed with 2.5 μm [Ca²⁺]_i and 40 mm [Cl⁻]_i. A, average of normalized I _Cl generated using 0.5 s depolarizations. The red line represents a fit with a mono‐exponential function to I _Cl, τ_on = 14.7 ± 1.1 ms. The y scale bar is the normalized I _Cl at +100 mV; this is also valid for the traces in B. B, average of normalized I _Cl generated by 20 s depolarizations. C, comparison of I _Cl recorded at +120 mV from HEK‐293 cells expressing WT TMEM16B (black) or TMEM16B/AVK (grey). The current onsets were matched. D, normalized I _Cl–V _m relationships from short (0.5 s, black circles) or long (20 s, grey circles) depolarization pulses. *P ≤ 0.001 by Student's t‐test; n = 4–7; mean ± SEM. E, V _m dependence of the time constants for the fast and slow gating mechanism (magenta and cyan circles, respectively). F, relative weight of the fast component (W _f) of I _Cl as a function of V _m for TMEM16B/AVK at 140 mm [Cl⁻]_o (magenta) and TMEM16B at 30 mm [Cl⁻]_o (blue).

Finally, we have previously shown that the naturally occurring variant of TMEM16A lacking residues ₄₄₈EAVK₄₅₁ activates very rapidly and the current traces are nearly flat during the duration of the depolarization (Xiao et al. 2011). Thus, we wondered if removing, instead of mutating, these residues would alter slow gating. For this we recorded I _Cl with 0.5 and 20 s pulses from HEK‐293 cells dialysed with 0.2 μm [Ca²⁺]_i expressing TMEM16A and lacking ₄₄₈EAVK₄₅₁ (TMEM16A‐Δ₄₄₈EAVK₄₅₁, n = 6). Figure 8 A shows that I _Cl from TMEM16A‐Δ₄₄₈EAVK₄₅₁ recorded during 0.5 s depolarizations activated following a time course characterized by τ_on = 96.2 ± 6.3 ms then deactivated very quickly. Interestingly, the second gating mode was not evident with 20 s depolarization at pulses <+120 mV (Fig. 8 B). This is clearly shown by comparing I _Cl recorded at +100 mV from HEK‐293 cells expressing WT TMEM16A or TMEM16A‐Δ₄₄₈EAVK₄₅₁ (Fig. 8 C). Figure 8 D shows that normalized I _Cl–V _m curves were the same with short or long pulses. Figure 8 E and F displays τ_f and τ_s vs. V _m and W _f and W _s vs. V _m, respectively. W _f remained around 0.4 at all V _m. Strong depolarizations, V _m ≥ +120, were needed to uncover the slow gating mode, although the relative contribution of this component was between 0.16 ± 0.2 and 0.31 ± 0.001 when the cell was depolarized in a range of +140 to +180 mV. Taken together, these experiments suggest that residues within the first intracellular loop are relevant to slow gating in TMEM16 channels and dual gating is an inherent property of TMEM16 channels.

Figure 8. Deletion of ₄₄₈EAVK₄₅₁ residues on TMEM16A affects slow gating .

Whole‐cell chloride currents recorded from HEK‐293 cells transiently expressing TMEM16A‐Δ₄₄₈EAVK₄₅₁. Recordings were made using 0.2 μm intracellular Ca²⁺. A and B, average of normalized I _Cl recordings obtained by pulsing the cells (n = 6) with 0.5 s (A) or 20 s (B)  V _m steps according to the protocol shown in Fig. 1 A. The red line in A is a mono‐exponential fit to activating current, τ_on = 96.2 ± 6.3 ms. Note that dual gating was practically absent in TMEM16A lacking ₄₄₈EAVK₄₅₁, except for ≥+120 mV (B). The y scale bar in A is the normalized I _Cl at +100 mV; this is also valid for the traces in B. C, time course of normalized I _Cl recorded at +100 mV from HEK‐293 cells transiently expressing WT TMEM16A (black) vs. TMEM16A‐Δ₄₄₈EAVK₄₅₁ (grey). D, normalized I _Cl–V _m relationships from short (0.5 s, black circles) or long (20 s, grey circles) depolarization pulses. I _Cl obtained from 0.5 s depolarizing pulses (black circles) or 20 s depolarization (grey circles). E, V _m dependence of fast (τ_f) and slow (τ_s) time constants from I _Cl induced by 20 s depolarization. F, relative weight of fast (W _f, magenta) and slow (W _s, cyan) components as a function of V _m.

Discussion

Recently there have been major advances in our understanding of the physiology, biophysical properties and even the structure of the TMEM16 protein family (Pifferi et al. 2009; Xiao et al. 2011; Yu et al. 2012; Brunner et al. 2014). However, the activation mechanism of the TMEM16 chloride channels remains uncertain. In this work, we demonstrate that Ca²⁺‐activated Cl⁻ channels TMEM16A and TMEM16B as well as endogenous CaCCs have fast and slow gating modes that are determined by V _m and external Cl⁻. Fast gating happens in tens to hundreds of milliseconds while slow gating occurs in seconds. Of these gating modes, the slow mode is more sensitive to external Cl⁻. Dual gating can be considered an intrinsic property of CaCCs as it is observed in excised inside‐out patches and is sensitive to alterations of the first intracellular loop of the channels. The presence of a slow gating mode in TMEM16A explains, in part, its complex kinetics (Arreola et al. 1996; Nilius et al. 1997; Romanenko et al. 2010; Scudieri et al. 2013; Brunner et al. 2014).

We have previously shown that fast perfusion of 20 μm Ca²⁺ to inside‐out patches excised from HEK‐293 cells expressing TMEM16A resulted in I _Cl with fast activation kinetics (Xiao et al. 2011). The time constants of activation of those currents were in the range 25–30 ms and were V _m‐independent. With 82 μm Ca²⁺ the activation kinetics is even faster (Yu et al. 2012), suggesting that Ca²⁺ regulates a fast gating process. Consistent with this idea, mutation of E705, a residue forming the Ca²⁺ binding site, increased the activation time constant 20‐fold (Yu et al. 2012). These results suggest that fast gating in TMEM16A, and perhaps in TMEM16B too, is due to binding of calcium to the Ca²⁺ pocket.

The EAVK sequence, also called ‘region c’, is part of the first intracellular loop in TMEM16 channels and is encoded by exon 13, which is regulated by alternative splicing (Ferrera et al. 2009). In a previous study, we found that TMEM16A lacking residues ₄₄₈EAVK₄₅₁ have altered Ca²⁺ and V _m dependence (Xiao et al. 2011). TMEM16A slow gating was marginaly altered when ₄₄₉AVK₄₅₁ residues were replaced by RSQ. However, in TMEM16A lacking ₄₄₈EAVK₄₅₁, the slow gating mode is dampened at moderate depolarized V _m. At highly depolarized voltages the slow gating contributes to the total current, although its contribution is minimal. These data suggest that the structural integrity of the EAVK region is critical to TMEM16A slow gating perhaps to maintain a specific spatial arrangement in the protein. The role that ₄₈₀RSQ₄₈₂ residues of the first intracellular loop of TMEM16B play on Ca²⁺ affinity were assessed by replacing them with equivalent residues from TMEM16A. The results suggest that residues ₄₇₉ERSQ₄₈₂ (₄₇₅ERAQ₄₇₈ in Scudieri et al. 2013) do not control either Ca²⁺ sensitivity or V _m dependence in TMEM16B channels (Pifferi et al. 2009; Stephan et al. 2009; Scudieri et al. 2013). These observations suggest that these residues do not control the biophysical properties of TMEM16B. However, here we show that replacing ₄₈₀RSQ₄₈₂ with AVK on TMEM16B abolished the slow gating mode without altering fast gating. Additional studies of the role of this sequence in gating together with structural studies are necessary to understand better the Ca²⁺‐ and V _m‐dependent gating of TMEM16B.

Fast and slow gating modes could be relevant to different processes. Several studies have reported a relationship between anion permeability and gating in CaCCs (Greenwood & Large, 1999; Pérez‐Cornejo et al. 2004; Xiao et al. 2011; Betto et al. 2014; Reyes et al. 2014). However, we still do not know how permeant anions alter gating. One possibility would be that permeation and gating are coupled by the Cl⁻ dependence of the slow gating mode. This idea is supported by our finding that decreasing external Cl⁻ from 140 to 30 mm blunted slow gating. For permeant Cl⁻ to alter gating, Cl⁻ needs to be sensed by a channel domain that participates in gating. As discussed above substitution of residues in the first intracellular loop diminished slow gating even when [Cl⁻]_o was maintained at 140 mm. These effects were stronger in TMEM16B than in TMEM16A.

Although structural information about vertebrate TMEM16 channels is not yet available, such information is needed to better understand the mechanism by which Cl⁻ alters TMEM16 gating. Therefore, we built a homology model for mouse TMEM16A based on the nhTMEM16 structure recently determined by X‐ray crystalography (Brunner et al. 2014) as described by Yu et al. (2015). nhTMEM16 has been shown to have phospholipid scramblase activity like the mammalian TMEM16F, but no significant ion channel activity. Nevertheless, the nhTMEM16 structure provides a reasonable starting point for clues into TMEM16A structure. The homology model is excellent within the transmembrane domains (Yu et al. 2015), but the extracellular loops are poorly modelled because of low sequence similarity in these regions. Figure 9 A shows a TMEM16A homodimer (subunits are coloured light brown and light grey) viewed from the centre of the lipid bilayer with the Ca²⁺ binding residues E702, E705, E734 and D738 (Yu et al. 2012; Tien et al. 2014) shown in green. Figure 9 B shows one monomer rotated 70 deg around an axis perpendicular to the bilayer and viewed from the bilayer. In the nhTMEM16 structure, Brunner et al. (2014) describe a hydrophilic cleft that they suggest is a likely structure for phospholipid scrambling. TMEM16A has a similar cleft that we believe forms part of the Cl⁻ conduction pathway. We believe that this is the Cl⁻ conduction pathway because residue K588 (magenta) that participates in ionic selectivity is located halfway along this cleft (Yang et al. 2012; Reyes et al. 2015) and residues near the top of the cleft have been identified as forming the vestibule and contributing to ionic selectivity (residues G628–M532 and I636–Q637; Yu et al. 2012) shown in red and residues R515, K603, R621 and R788 (Peters et al. 2015) shown in cyan. Interestingly, when these basic residues were mutated, the permeability ratio to SCN⁻ was altered in a Ca²⁺‐dependent manner. This result suggests that Ca²⁺ may induce a conformational change in this region to alter anion permeation (Peters et al. 2015). The EAVK loop lies at the cytosolic end of the hydrophilic cleft and therefore is located in a position well suited to control ion permeability. Alternatively residues located in the external mouth of TMEM16A pore may sense permeant anions as described for CFTR channel (Broadbent et al. 2015).

Figure 9. Homology model of mTMEM16A showing functional residues .

mTMEM16A was modelled on nhTMEM16 (Brunner et al. 2014) as described by Yu et al. (2015). Alpha‐helices are numbered and lines indicate the approximate location of the bilayer. A, the TMEM16A homodimer viewed from the centre of the lipid bilayer. One monomer is coloured pale brown and the other light grey. Atoms in the subunit on the right that are shown in space‐filling representation have been studied by mutagenesis and shown to have a role in channel function. B, one TMEM16A monomer was rotated 70 deg around the y‐axis and viewed from the bilayer. This view looks into the hydrophilic cleft described by Brunner et al. (2014). Red: channel vestibule residues G628–M632 and I636–Q637 (Yu et al. 2012); cyan: ionic selectivity residues K603, R621 and R787 (Peters et al. 2015); magenta: ionic selectivity residue K588 (Yang et al. 2012); green: calcium binding residues E702, E705, E734 and D738 (Yu et al. 2012; Tien et al. 2014); brown: channel gating and calcium sensitivity residues ₄₄₄EEEEEAVK₄₅₁ (Xiao et al. 2011; this study).

The exact physiological relevance of slow gating remains undetermined; cells do not regularly become depolarized to +100 mV, a voltage range where slow gating is more prominent. Despite this, an intrinsic characteristic of gating like this may be physiologically relevant under other conditions we have not yet encountered, such as changes in phosphorylation, oxidative stress or pathological conditions (e.g. cancer) where TMEM16A is overexpressed (Qu et al. 2014; Sui et al. 2014). The slow gating mode may provide a way to maintain TMEM16 channels in the open state for a long time. This property would be important in processes that need to sustain Cl⁻ movements such as salivation and fast block of polyspermy in Xenopus oocytes (Webb & Nuccitelli, 1985; Melvin et al. 2002; Catalán et al. 2015). Both processes require large Ca²⁺‐dependent chloride movements. In salivary glands, primary secretion involves Cl⁻ exit through the apical membrane into the ducts (Melvin et al. 2005). In this way the [Cl⁻]_i decreased from about 61 to 29 mm (Foskett, 1990). Thus, Cl⁻ would transiently increase on the extracellular side of the apical membrane where CaCCs are located. One outcome of this event is a positive feedback such that the open probability of CaCC remains high to sustain salivation. In the second example, fusion of the sperm with the Xenopus egg produces an intracellular calcium wave that opens CaCCs. Once the channels open the egg depolarizes and prevents fertilization by other sperm; the effect of CaCC activation on the number of eggs that fertilize can be reproduced by electrical depolarization (Glahn & Nuccitelli, 2003). Perhaps at voltages more depolarized than 0 mV, a sustained open probability is accomplished by switching channels to the slow gating mode. On the other hand, TMEM16B is expressed in sensory neurons and does not inactivate in cells dialysed with high [Ca²⁺]_i. In contrast, TMEM16B without ERSQ, which inactivates in the presence of high [Ca²⁺]_i, is the dominant variant in cortex and hippocampus; this variant is expressed in cerebellum, brain stem and olfactory epithelium (Vocke et al. 2013). This stringent tissue expression pattern of TMEM16B variants and the dependence of the slow gating mode on RSQ could signify an important functional adaptation for the physiological role each of these variants play in neurons. For example, it has been proposed that non‐inactivating TMEM16B limits transmitter release by suppressing a depolarizing‐induced Ca²⁺ influx (Dauner et al. 2013). This role would require a sustained Cl⁻ movement through channels that remain open for long periods of time. A condition like this could be achieved by activating the slow gating mode.

Finally, a dual gating mechanism of TMEM16 channels is a property shared with CLC Cl⁻ channels. There are remarkable similarities between TMEM16 and CLC Cl⁻ channels. For example, both are homodimers, their gating is strongly affected by permeant anions and protons, and both are V _m sensitive despite lacking V _m sensors (Pérez‐Cornejo et al. 2004; Chen, 2005; Miller, 2006; Fallah et al. 2011; Xiao et al. 2011; Betto et al. 2014). CLC Cl⁻ channels have two pores and each pore has a fast gate in it (Dutzler et al., 2002, 2003) while the slow gate seems to be located at the C‐terminus (Bykova et al. 2006; Garcia‐Olivares et al. 2008). Recent evidence points in this direction (Bill et al. 2015) with the first intracellular loop participating in slow gating as shown here. In addition, homology models based on nhTMEM16F scramblase structure suggest that TMEM16A has a homodimeric architecture (Peters et al. 2015; Yu et al. 2014; present data).

In conclusion, we have demonstrated here that calcium‐activated chloride channels TMEM16A and TMEM16B display fast and slow gating modes. The slow mode shows strong dependence on V _m, residues in the first intracellular loop and extracellular [Cl⁻] within the physiological range.

Additional information

Competing Interests

The authors declare no competing interests.

Author Contributions

This work was done at the Biophysics Laboratory of the Physics Institute Universidad Autónoma de San Luis Potosi, Mexico. All authors approved the final version of the manuscript and all persons who qualify for authorship are listed. SC‐R, JJDe J‐P, PP‐C, HC‐H, JA: Conception and design of the experiments. SC‐R, JJDe J‐P, JACV, JA: Collection, assembly, analysis and interpretation of data. SC‐R, JJDe J‐P, JACV, PP‐C, HC‐H, JA: Drafting the article or revising it critically for important intellectual content.

Funding

This study was supported in part by the following: grants 219949 from CONACyT, Mexico, to Jorge Arreola and Patricia Pérez‐Cornejo; grants GM60448 and EY11482 from National Institutes of Health to H. Criss Hartzell; Silvia Cruz‐Rangel is a recipient of Postdoctoral Fellowships 256034 and 290807 from CONACyT, Mexico; José J. De Jesús‐Pérez is the recipient of Graduate Student Fellowship 229968 from CONACyT, Mexico; and Juan A. Contreras‐Vite is the recipient of Graduate Student Fellowship 234820 from CONACyT, Mexico.