Voltage- and calcium-dependent gating of TMEM16A/Ano1 chloride channels are physically coupled by the first intracellular loop

Abstract

Ca²⁺-activated Cl⁻ channels (CaCCs) are exceptionally well adapted to subserve diverse physiological roles, from epithelial fluid transport to sensory transduction, because their gating is cooperatively controlled by the interplay between ionotropic and metabotropic signals. A molecular understanding of the dual regulation of CaCCs by voltage and Ca²⁺ has recently become possible with the discovery that Ano1 (TMEM16a) is an essential subunit of CaCCs. Ano1 can be gated by Ca²⁺ or by voltage in the absence of Ca²⁺, but Ca²⁺- and voltage-dependent gating are very closely coupled. Here we identify a region in the first intracellular loop that is crucial for both Ca²⁺ and voltage sensing. Deleting ₄₄₈EAVK in the first intracellular loop dramatically decreases apparent Ca²⁺ affinity. In contrast, mutating the adjacent amino acids ₄₄₄EEEE abolishes intrinsic voltage dependence without altering the apparent Ca²⁺affinity. Voltage-dependent gating of Ano1 measured in the presence of intracellular Ca²⁺ was facilitated by anions with high permeability or by an increase in [Cl⁻]_e. Our data show that the transition between closed and open states is governed by Ca²⁺ in a voltage-dependent manner and suggest that anions allosterically modulate Ca²⁺-binding affinity. This mechanism provides a unified explanation of CaCC channel gating by voltage and ligand that has long been enigmatic.

Ca²⁺-activated Cl⁻ channels (CaCCs) play manifold roles in cell physiology (1, 2), including epithelial secretion (3, 4), sensory transduction and adaptation (5–8), regulation of smooth muscle contraction (9), control of neuronal and cardiac excitability (10), and nociception (11). This myriad of functions has attracted attention for more than 25 years (12, 13), but a lack of consensus regarding their molecular composition has stymied a mechanistic understanding of their gating. Recently, two members of the TMEM16/anoctamin family (Ano1 and Ano2) were identified as CaCC channels (14–16) and shown to be essential for salivary exocrine secretion (14, 17, 18), gut slow-wave activity (18, 19), tracheal secretion (18, 20, 21), and olfactory transduction (5–7).

Ano1 and Ano2 are well suited for their diverse roles because they are dually gated by voltage (V_m) and intracellular Ca²⁺ concentration ([Ca²⁺]_i), so that their activity is tuned by the interplay between metabotropic and ionotropic inputs (14, 16, 22–24). The molecular mechanisms underlying V_m and Ca²⁺ gating are unknown, however. Unlike typical V_m-gated channels or ligand-gated channels, CaCCs exhibit both V_m dependence and ligand gating that are strongly coupled and apparently reciprocally related. Ano1 does not contain obvious V_m-sensing or Ca²⁺-sensing domains (14). Because Ca²⁺ is often stabilized in proteins by oxygen atoms (25), it seemed reasonable to hypothesize that acidic amino acids contribute to the Ca²⁺-binding site. Ca²⁺ sensors in two other Ca²⁺-activated channels, BK and Best1, are associated with sequences rich in acidic amino acids (26, 27). The Ano1 sequence has a similar domain in the first intracellular loop (amino acids 430–480), composed of residues ₄₄₄EEEEEAVK₄₅₁ (Fig. S1). Here we show that this region plays a key role in transducing both V_m and Ca²⁺ signals.

Results

Voltage and Ca²⁺ Synergistically Gate Ano1.

At low [Ca²⁺]_i (<1 μM), WT Ano1 (splice variant a,c) was activated synergistically by Ca²⁺ and depolarization. In the absence of Ca²⁺, no current was evident at V_m between −100 mV and +100 mV, but as [Ca²⁺]_i was increased, an outward current was activated by depolarization and deactivated by hyperpolarization (Fig. 1 A–D and F). As [Ca²⁺]_i was increased, outward rectification (Fig. 1F) and the fraction of total current exhibiting time-dependence (Fig. 1E) were reduced. V_m-dependent activation of Ano1 was evaluated by plotting normalized conductance versus V_m (G/G_max vs. V_m curves; Fig. 1G). The data were well fit by the Boltzmann equation,

Fig. 1.

WT Ano1 gating by voltage and Ca²⁺. (A–E) Representative I_Cl,Ano1 in transfected HEK293 cells at the indicated free [Ca²⁺]. Voltage protocol is shown above A. (F) Steady-state current–voltage relationships (n = 5–9). (G) G/G_max vs. V_m curves for Ano1 in excised patches activated by 0.6 μM (▲), 1 μM (▪), and 2 μM (●) Ca²⁺. (H) Instantaneous tail current density at −100 mV after prepulses to V_m plotted vs. free [Ca²⁺] and fitted to the Hill equation. (I) EC₅₀ (▪) and n_H (●) values from fits in H plotted vs. V_m.

where G/G_max is normalized conductance; z is the equivalent gating charge associated with voltage-dependent channel opening; V_0.5 is the membrane potential (V_m) where G/G_max is half-maximal and is related to the conformational energy associated with voltage-independent channel opening; and F/RT = 0.039 mV⁻¹. At 1 μM Ca²⁺, V_0.5 was 64 ± 0.9 mV (Fig. 1G, black squares); doubling [Ca²⁺] to 2 μM shifted the G/G_max vs. V_m curve to the left by −145 mV (Fig. 1G, red circles) with no significant effect on z. Because Ca²⁺ shifts the G/G_max vs. V_m curves so dramatically, a complete G/G_max vs. V_m curve could be recorded for only a narrow range of [Ca²⁺]_i. For these [Ca²⁺], z was not obviously Ca²⁺-dependent (z = 0.40–0.46). This indicates that Ca²⁺ does not change the V_m sensitivity (z) of the Ano1 channel, but rather shifts V_0.5, the energy associated with V_m-independent gating.

Although these data may imply that Ano1 is a simple ligand-gated channel, Ano1 is actually more complicated, because Ca²⁺ gating is strongly influenced by V_m. The EC₅₀ of Ano1 for Ca²⁺ decreases ∼15-fold from 5.9 ± 2.5 μM at −100 mV to 0.4 ± 0.1 μM at +100 mV while the Hill coefficient, n_H, indicative of the cooperativity of Ca²⁺ binding, increases from 1.0 ± 0.1 to 2.0 ± 0.4 (Fig. 1 H and I). These data support a mechanism of Ano1 gating with V_m and Ca²⁺ converging on the switch between closed and open conformations.

Glutamic Acids in the First Intracellular Loop Contribute to Channel Gating.

The first intracellular loop of Ano1 contains five consecutive glutamic acids (₄₄₄EEEEE₄₄₈), resembling the “Ca²⁺ bowl” of BK channels (26) and the acidic cluster in hBest1 (27) that are linked to Ca²⁺ sensing (Fig. S1). The last glutamic acid of this cluster, ₄₄₈E, is the first residue of a naturally occurring alternatively spliced segment, ₄₄₈EAVK₄₅₁ (28). To investigate this region in detail, we mutated the first four glutamic acids to alanines (₄₄₄EEEE/AAAA₄₄₇) and deleted ₄₄₈EAVK₄₅₁ (ΔEAVK) (current traces; Fig. S2). The ΔEAVK deletion and, to a lesser extent, the ₄₄₄EEEE/AAAA₄₄₇ mutation, shifted G/G_max vs. V_m curves at each [Ca²⁺]_i to the right without altering the slope (Fig. 2). For ΔEAVK, the position of the G/G_max vs. V_m curve on the V_m axis was relatively insensitive to Ca²⁺: 2 μM Ca²⁺ had no effect, and 25 μM Ca²⁺ shifted the curve only to a position comparable to the WT curve at 1 μM Ca²⁺. The slope of the relationship between V_0.5 and [Ca²⁺]_i was reduced ∼50-fold for ΔEAVK and ∼3-fold for ₄₄₄EEEE/AAAA₄₄₇ compared with WT (Fig. 2C). Because V_0.5 represents the conformational energy associated with V_m-independent channel opening, these mutations decrease the ability of Ca²⁺ to reduce the activation energy required for the channel to open. This could be caused by either lower Ca²⁺-binding affinity or decreased stability of the Ca²⁺-bound open state.

Fig. 2.

G/G_max vs. V_m curves for mutant Ano1 in excised patches at the indicated [Ca²⁺]_i. (A and B) Solid lines without symbols: WT from Fig. 1G at 1 μM and 2 μM Ca²⁺. (A) ΔEAVK with 25 μM (▽), 2 μM (□), and 1 μM (●) Ca²⁺. (B) ₄₄₄EEEE/AAAA₄₄₇ with 25 μM (◇), 2 μM (□), and 1 μM (●) Ca²⁺. (C) V_0.5 vs. [Ca²⁺] for WT (▪), ΔEAVK (), and ₄₄₄EEEE/AAAA₄₄₇ (△).

Ca²⁺ Dissociation Is V_m-Dependent and Slow.

To test whether these mutations alter Ca²⁺ affinity, we performed fast perfusion experiments in which excised patches were switched between zero and high [Ca²⁺] within several ms (Fig. S3). Fig. 3 displays examples of the time course of Ano1 activation at the indicated V_m when Ca²⁺ was increased rapidly from 0 μM to 20 μM (Fig. 3A) and deactivation when Ca²⁺ was withdrawn (Fig. 3B). The time course of activation was sigmoid and displayed a lag period consistent with multistep channel opening. For simplicity, current activation after the initial lag period was fit to a monoexponential equation to calculate the time constant of activation, τ_on (Fig. 3A). τ_on was decreased when [Ca²⁺] was increased, as expected if Ca²⁺ binding is a rate-limiting step in channel opening. τ_on was not strongly V_m-dependent (Fig. 3E). On removal of Ca²⁺, the currents decayed monoexponentially (Fig. 3B), and τ_off was strongly V_m-dependent, with an e-fold slowing per 70.3-mV depolarization for WT (Fig. 3F). ₄₄₄EEEE/AAAA₄₄₇ and ΔEAVK mutations had no significant effect on τ_on (Fig. 3E). In contrast, τ_off was greatly accelerated by ΔEAVK and slightly accelerated by the ₄₄₄EEEE/AAAA₄₄₇ mutation at positive V_m (Fig. 3F). At +120 mV, τ_off was 501.5 ± 10.2 ms for WT, 37.4 ± 1.1 ms for ΔEAVK, and 345.0 ± 18.8 ms for ₄₄₄EEEE/AAAA₄₄₇. Because τ_off for WT was surprisingly slow, for comparison, we performed an experiment using patches containing the BK Ca²⁺-activated K⁺ channel (Fig. S3). τ_off for the BK channel was 28 ms, ∼20-times faster than for WT Ano1.

Fig. 3.

Activation and deactivation kinetics of Ano1 with rapid Ca²⁺ and Ba²⁺ perfusion. (A–D) Representative traces of I_Cl,Ano1 in response to application (A and C) and washout (B and D) of 20 μM Ca²⁺ at the indicated holding potentials. (A and B) WT Ano1. (C and D) ΔEAVK. (E–G) V_m dependence of τ_on, τ_off, and EC₅₀ for WT Ano1 (○), ₄₄₄EEEE/AAAA₄₄₇ (▲), and ΔEAVK (●). (H and I) Representative traces of increase in current on application (H, arrowhead) and washout (I) of 1 mM Ba²⁺ (1 mM Ba²⁺, 100 μM EGTA) for WT Ano1. (J) τ_off for Ba²⁺.

τ_off is equal to the reciprocal of the rate constant(s) of the rate-limiting step(s) leading to channel closure, which could be Ca²⁺ dissociation or channel closure itself. As a test of the idea that τ_off reflects ligand dissociation, we measured τ_off for Ba²⁺-activated Ano1. Because the free energy of hydration of Ba²⁺ is smaller than that of Ca²⁺, Ba²⁺ binding to sites composed of coordinating oxygen ligands is significantly less stable than that of Ca²⁺ (25). For example, Ba²⁺ is almost inactive in activating CaM-dependent phosphodiesterase or displacing Ca²⁺ from CaM (29). The apparent EC₅₀ for Ano1 activation is >10-fold larger for Ba²⁺ than for Ca²⁺, reflected as a 20-fold faster turnoff of the Ano1 current on Ba²⁺ washout (Fig. 3 I and J). These observations support the conclusion that τ_off reflects ligand dissociation and, therefore, ΔEAVK dramatically increases the rate of Ca²⁺ dissociation from the channel at all voltages.

Assuming that the rate-limiting steps are Ca²⁺ binding and unbinding, the apparent EC₅₀ for Ca²⁺ can be calculated as

Deletion of EAVK produces a large increase in EC₅₀ at all voltages (Fig. 3G), suggesting an important role in Ca²⁺ sensing. In contrast, ₄₄₄EEEE/AAAA₄₄₇ is similar to WT. The V_m-dependence of EC₅₀ is well described by the equation

where EC_50(0mV) is the EC₅₀ value at 0 mV, z = 2, and δ is the electric field fraction (Fig. 3G). The EC_50(0mV) values were 1.39 μM for WT and 71 μM for ΔEAVK. Values of δ were similar, 0.16 for WT and 0.12 for ΔEAVK. The V_m dependence of the EC₅₀ values agree with other published data for both Ano1 and endogenous CaCCs (14, 22, 23, 28).

Both the Ca²⁺-activated and the Ba²⁺-activated currents exhibit a sigmoid onset (Fig. 3 A and H), indicating that current activation is a highly cooperative multistep process. The sigmoid onset is more pronounced with Ba²⁺, and the activation, especially the lag period after switching to Ba²⁺ before the current begins to increase, is distinctly V_m-dependent. This V_m dependence is not as obvious with Ca²⁺, but does appear to be present (Fig. 3A). The V_m-dependent sigmoid onset of the current supports a multistep model of Ano1 activation with V_m altering the affinity of the ligand-binding site.

The V_m dependence of τ_off supports the idea that Ca²⁺ binding is V_m-dependent. If this were the case, then we would expect current deactivation produced by hyperpolarization to reflect Ca²⁺ dissociation as a result of a V_m-dependent decrease in Ca²⁺ affinity. Consequently, we predict that the time constant of current deactivation (τ_deact) would become faster with hyperpolarization but slower with increased [Ca²⁺]. This is indeed what we observe (Fig. 4A). Consistent with the rapid perfusion, ΔEAVK displays a faster τ_deact at each [Ca²⁺] tested (Fig. 4B). A smaller effect is observed with ₄₄₄EEEE/AAAA₄₄₇ (Fig. 4C).

Fig. 4.

Deactivation of I_Cl,Ano1. (A–C) V_m and [Ca²⁺] dependence of tail current deactivation. Currents were activated using a +100-mV pulse, and tail currents were measured at the indicated potentials on the x-axis. (A) WT Ano1. (B) ΔEAVK. (C) ₄₄₄EEEE/AAAA₄₄₇. n = 4–6.

Ano1 Exhibits Intrinsic Voltage Dependence.

Although there are no obvious classical V_m-sensor domains in the predicted transmembrane segments of Ano1 (Fig. S1), Ano1 can be opened by V_m in the absence of Ca²⁺ or other divalent cations by strong depolarization (Fig. 5). Under several conditions with zero [Ca²⁺]_i, depolarizations >100 mV evoke outward currents (Fig. 5 A, D, and E). The magnitude of the Ca²⁺-independent current was ∼100 pA/pF at +200 mV which is <10% that activated by maximum [Ca²⁺]_i (Fig. 1H). The Ca²⁺-independent currents are mediated by Ano1 because they are not present in HEK cells transfected with GFP alone.

Fig. 5.

Activation of Ano1 in zero intracellular Ca²⁺. (A–C) Representative traces of WT Ano1 (A), ₄₄₄EEEE/AAAA₄₄₇ (B), and ΔEAVK (C) activated by V_m in nominally zero Ca²⁺. (D) Mean current density at +200 mV with zero intracellular Ca²⁺ plus 5 mM EGTA, zero intracellular Ca²⁺ and Mg²⁺ plus 5 mM EGTA (EGTA + 0 Mg), or zero Ca²⁺ plus 5 mM EGTA plus 1 mM BAPTA in both intracellular and extracellular solutions (EGTA + BAPTA). n = 5–11. (E) Steady-state I-V curve from cells dialyzed with zero Ca²⁺: nontransfected cells (control; □), WT Ano1 (), Δ₄₄₄EEEEEAVKD₄₅₂ (△), ΔEAVK (▽), and ₄₄₄EEEE/AAAA₄₄₇ (♢). (n = 4–11).

To investigate the role of acidic residues in the Ca²⁺-independent current, ₄₄₄EEEEEAVKD₄₅₂ was deleted. These channels were not activated by depolarizations to +200 mV (Fig. 5E), but were activated by Ca²⁺ with an EC₅₀ of 45 μM. ₄₄₄EEEE/AAAA channels were not activated by V_m either (Fig. 5 B and E), despite the fact that their Ca²⁺ sensitivity was similar to that of WT. In contrast, ΔEAVK channels were activated more readily by depolarization compared with WT (Fig. 5 C and E). Taken together, Figs. 2–5 argue that ₄₄₄EEEE₄₄₇ and ₄₄₈EAVK₄₅₁ are critical in both Ca²⁺-dependent and V_m-dependent gating of Ano1. The reciprocal relationship of Ca²⁺ and V_m is illustrated by the disruption of ₄₄₈EAVK₄₅₁, which decreases apparent Ca²⁺ affinity while stabilizing the V_m-gated open state. The ₄₄₄EEEE₄₄₇ mutation, on the other hand, seems to stabilize the closed states of both ligand-gated and voltage-gated pathways.

Voltage Gating Is Affected by Permeant Anions.

The observation that neutralization of ₄₄₄EEEE₄₄₇ does not change the slope of the G/G_max vs. V_m curve (Fig. 2) argues that these residues are not themselves the voltage sensor. Because previous observations showing that the gating of CaCCs in Xenopus oocytes and salivary glands is affected by permeant anions (30, 31), we tested the idea that Ano1 voltage sensitivity is related to occupancy of the pore by anions. Replacement of external Cl⁻ with NO₃⁻ greatly reduced V_m-dependent gating of Ano 1, as evidenced by a reduction in outward rectification and an increase in the fraction of current activating instantaneously with depolarization (Fig. 6 A–D). Furthermore, G/G_max vs. V_m curves were shifted to negative potentials when Cl⁻ was replaced by NO₃⁻ or SCN⁻ (Fig. 6 E and F). These ions would be expected to exhibit a higher occupancy in the pore than Cl⁻, because they have lower hydration energies and higher relative permeabilities than Cl⁻ (31).

Fig. 6.

WT Ano1 gating is dependent on permeant anions. (A and B) Current traces from a cell dialyzed with 180 nM Ca²⁺ and bathed in symmetrical 150 mM Cl⁻ (A) or 150 mM NO₃⁻ (B). (C) Fraction of instantaneous current relative to total current at the end of a 700-ms pulse to +100 mV. (D) Normalized (at +100 mV) I-V curves with Cl⁻ (●) or NO₃⁻ (▪) as permeant ions. (E) G/G_max vs. V_m curves for WT Ano1 activated at different [Ca²⁺]_i with Cl⁻ (open symbols) or SCN⁻ (filled symbols) as permeant anions. Circles represent 360 nM Ca²⁺; squares, 600 nM Ca²⁺; triangles, 1 μM; diamonds, 2 μM. (F) V_0.5 determined from G/G_max vs. V_m curves for Cl⁻, NO₃⁻, and SCN⁻. Red bars represent 0.6 μM Ca²⁺; green bars, 1 μM Ca²⁺. n = 4–11. (G) Cl⁻ dependence of G/G_max vs. V_m curves. [Cl⁻]_e: 1 mM (green), 10 mM (red), 72 mM (blue), and 136 mM (black). (H) Anomalous mole fraction behavior of WT Ano1. The change in E_rev is plotted as a function of the mole fraction of iodide (with the balance Cl⁻). The red line represents the prediction from the Goldman–Hodgkin–Katz equation.

If channel gating is dependent on pore occupancy by permeant anions, V_m dependence would be expected to depend on [Cl⁻]. As expected, increasing [Cl⁻]_e shifts the G/G_max vs. V_m curves to more negative potentials (Fig. 6G). Furthermore, the reversal potential of the current carried by mixtures of Cl⁻ and I⁻ cannot be described by the Goldman–Hodgkin–Katz equation (Fig. 6H), supporting the idea that channel gating is dependent on ion permeation.

Discussion

The gating of Ano1 is controlled by a complex interplay among [Ca²⁺]_i, V_m, and permeant anions. The most important contribution to gating is provided by an increase in [Ca²⁺]_i under our conditions. At high concentrations, Ca²⁺ alone is able to open the channel even at very hyperpolarized potentials; however, at lower [Ca²⁺], voltage has a significant effect that is likely physiologically relevant (see below).

Although Ano1 resembles ligand-gated channels, its V_m sensitivity is unique. For example, Ano1 differs from the ligand-gated nAChR because although nAChR is weakly V_m-sensitive, its affinity for ACh is not obviously V_m-sensitive (32), whereas the Ca²⁺ sensitivity of Ano1 is clearly V_m-sensitive. Furthermore, unlike nAChRs, Ano1 can be completely closed by hyperpolarization in the presence of <1 μM Ca²⁺ and can be opened by strong depolarization itself in the absence of Ca²⁺. The observation that the channel can be opened in the absence of Ca²⁺ supports the conclusion that the “V_m sensor” is unlikely to be Ca²⁺ itself. Ano1 also differs significantly from the large-conductance Ca²⁺-activated K⁺ channel (BK), in that Ano1 has much higher sensitivity to Ca²⁺ and does not have an obvious protein-based V_m sensor.

Ca²⁺ and V_m Dependence Are Coupled by the First Intracellular Loop.

Residues ₄₄₄EEEEEAVK₄₅₁ are involved in the transduction of both Ca²⁺ and V_m signals. Deletion of ₄₄₈EAVK₄₅₁ severely decreases Ca²⁺ sensitivity and shifts V_0.5 values to positive voltages. Because V_0.5 represents the energy of V_m-independent channel opening, ΔEAVK apparently increases the activation energy required for Ca²⁺ to gate the channel open. This could be caused by decreased Ca²⁺-binding affinity or diminished stability of the Ca²⁺-bound open state. Colquhoun (33) discussed the difficulty in separating the effects of mutations on channel gating and on ligand binding; however, our fast perfusion experiments provide strong support for a major effect of ΔEAVK on Ca²⁺ binding. Nonetheless, there are several reasons why ₄₄₄EEEEEAVK₄₅₁ is unlikely to form part of the Ca²⁺-binding site. At very positive voltages, the EC₅₀ values of WT and ΔEAVK appear to be convergent (Fig. 3G), which would not be expected if ΔEAVK disrupted the Ca²⁺-binding site. Second, neutralizing ₄₄₄EEEE₄₄₈ has little effect on apparent Ca²⁺ sensitivity, but renders the channels insensitive to V_m activation in the absence of Ca²⁺ and shifts the G/G_max vs. V_m curve to positive potentials in the presence of Ca²⁺. This suggests that ₄₄₄EEEE₄₄₇ is involved in transducing depolarization to channel opening, but is unlikely to contribute to Ca²⁺ binding.

The V_m dependence of the EC₅₀ values can result if depolarization drives Ca²⁺ into a binding site within the electrical field or alters binding site conformation and Ca²⁺ affinity. The hypothesis that V_m alters binding affinity is supported by the observation that the V_m dependence of current activation and deactivation in rapid perfusion experiments is different for Ca²⁺ and Ba²⁺ (e.g., a 70 mV/e-fold change in τ_off for Ca²⁺, compared with 115 mV for Ba²⁺; Fig. 3), whereas electrophoretic effects should be similar, because these ions have similar ionic mobilities (119 cm^2.Ω^-1·M⁻¹ for Ca²⁺ vs. 127 cm^2.Ω^-1·M⁻¹ for Ba²⁺).

Permeant anions control V_m-dependent gating of native CaCCs in Xenopus oocytes and parotid acinar cells (30, 31), as well as other Cl⁻ channels like ClC-2 (34). Permeant anions also affect Ano1 gating, and we surmise that permeant anion occupancy of the pore stabilizes the open state and thereby allosterically modulates Ca²⁺-dependent gating. Channel rectification and V_m-dependent gating can be explained by asymmetric access of permeant anions to a critical site in the pore.

We hypothesize that V_m-dependent Ca²⁺ binding may confer a majority of V_m dependence. Simple models with a V_m-dependent Ca²⁺ affinity can reproduce the Ano1 current (22, 23), but allosteric models, such as those used to explain the gating of BK channels (35, 36), are likely to be more successful in explaining the intricate gating of Ano1. However, more structural information is needed about the number of Ca²⁺-binding sites, the oligomerization state (37, 38), and the involvement of CaM (39). The first intracellular loop of Ano1 is very hydrophilic and unlikely to be located within the transmembrane voltage field (Fig. S1) (40). The physical relationship of ₄₄₄EEEEEAVK₄₅₁ to the Ca²⁺-binding site(s) and the pore, proposed to be in the reentrant loop between transmembrane domains 5 and 6 (14, 15), remains to be determined.

Ano1 Splice Variants and the Role of Calmodulin.

Human Ano1 has four different alternatively spliced segments, a, b, c, and d, corresponding to an alternative initiation site, exon-6b, exon-13, and exon-15 (28). Segments a and b are located in the N terminus, and segments c and d are in the first intracellular loop. Ferrera et al. (28) concluded that deleting segment c (ΔEAVK) in human Ano1 affects V_m dependence, but not Ca²⁺ sensitivity. This discrepancy with our conclusion is explained by our observation that ΔEAVK channels are strongly activated by V_m in the absence of Ca²⁺ (Fig. 5). Thus, the current amplitudes reported by Ferrera et al. (28) include both Ca²⁺-dependent and Ca²⁺-independent components. The EC₅₀ of ΔEAVK channels measured by voltage steps in the presence of steady [Ca²⁺] will be contaminated by a Ca²⁺-independent component that will make the EC₅₀ appear smaller than it actually is. In contrast, our fast perfusion experiments determine only the Ca²⁺-dependent component. This difference in methodology likely explains the discrepancy, but differences between human Ano1(a,b) and mouse Ano1(a) also could play a part.

A recent paper reported that CaM binds to a 22-aa region, CaM-BD1, that overlaps with the b segment and is essential for gating Ano1(a,b,c) (39). However, Tian et al. (39) showed that, unlike Ano1(a,b,c), Ano1(a,c), which lacks CaM-BD1, does not require CaM. Also, our finding that Ano1(a,c) is activated by Ba²⁺ supports a direct effect of Ca²⁺ on Ano1(a,c), because CaM is not significantly activated by Ba²⁺ (29). Some endogenous CaCCs are regulated directly by Ca²⁺, whereas others are regulated by CaM-dependent pathways (2, 41, 42), which might be explained by the expression of different splice variants.

Physiological Significance of Synergistic Gating of Ano1.

In cells in which resting V_m is typically between −90 and −30 mV and resting Ca²⁺ is ∼100 nM, Ano1 channel activation requires both depolarization and increased Ca²⁺. Elevation of Ca²⁺ to the low μM range at hyperpolarized potentials will be ineffective in opening Ano1, as will modest depolarization without a rise in intracellular Ca²⁺. This dual regulation by Ca²⁺ and V_m makes Ano1 particularly well suited to serve as a feedback regulator of V_m and [Ca²⁺]_i. Because E_Cl is dynamically regulated both temporally and spatially in many cells (1), Cl⁻ fluxes may be inward or outward, depending on local values of E_Cl, [Ca²⁺]_i, and V_m. Thus, Ano1 might have a negative or a positive feedback function. This might have important physiological and pathological implications in, for example, the genesis of cardiac arrhythmias, where an outward CaCC current normally participates in repolarization of the cardiac action potential in some species (2). However, under conditions of Ca²⁺ overload, this channel produces transient inward currents, which are arrhythmogenic. Similarly, because Ano1 rectification is regulated by [Ca²⁺], Ano1 in epithelial cells could mediate secretion or absorption, depending on the interplay of V_m, E_Cl, and [Ca²⁺]_i. Thus, the exquisite V_m and Ca²⁺ sensitivity of CaCC channels places these channels in a pivotal position for regulation of cellular excitability.

Experimental Procedures

Electrophysiology.

HEK-293 cells were transiently transfected with 0.1–1 μg of mouse Ano1(a,c) tagged on the C terminus with EGFP (provided by Dr. Uhtaek Oh, Seoul National University) per 35-mm dish using Fugene-6 (Roche) and patch-clamped 24–72 h later at room temperature (27). Mutations were made by PCR-based mutagenesis (Quickchanger; Agilent) and confirmed by sequencing. Transfected cells were identified by EGFP fluorescence. For whole-cell recording, zero Ca²⁺ pipette solution contained 146 mM CsCl, 2 mM MgCl₂, 5 mM EGTA, 10 mM sucrose, and 8 mM Hepes (pH 7.3), adjusted with N-methyl-D-glucamine. High Ca²⁺ pipette solution contained 5 mM Ca²⁺-EGTA, instead of EGTA (free Ca2⁺ ∼25 μM). Different free [Ca²⁺] solutions were made by mixing zero- Ca²⁺ and high-Ca²⁺ solutions. The 126 μM Ca²⁺ was made by adding 0.2 mM CaCl₂ to high-Ca²⁺ solution. Br₂-BAPTA [5,5′-dibromo-1,2-bis(2-aminophenoxy)ethane-N,N,N′,N′-tetraacetic acid, 3.5 mM; Invitrogen] was used to make 1-μM to 4-μM free [Ca²⁺] solutions in some experiments. Standard external solution contained: 140 mM NaCl, 4 mM KCl, 2 mM CaCl₂, 1 mM MgCl₂, 10 mM glucose, and 10 mM Hepes (pH 7.4). The study of the effects of [Cl⁻]_e or [I⁻]_e on Ano1 gating used a pipette solution with 2.241 mM CaCl₂, 25.241 mM EGTA-TEA, and 50 mM Hepes (pH 7.3) (TEA-OH), and 85 mM d-mannitol (calculated [Ca²⁺] = 75 nM). High-Cl⁻ external solution contained 135 mM TEA-Cl, 0.5 mM CaCl₂, 20 mM Hepes, and 75 mM d-mannitol. Low-Cl⁻ external solution contained 0.5 mM CaCl₂, 200 mM Hepes, and 45 mM d-mannitol. External solutions with different [Cl⁻] concentrations were made by mixing high- and low-Cl⁻ solutions. Iodide mixtures were made by mixing high-Cl⁻ external solution with a solution in which Cl⁻ was replaced by I⁻.

Rapid Perfusion.

The fast application of Ca²⁺ to excised inside-out patches was performed using a double-barreled theta tubing (1.5 mm o.d.; Sutter Instruments) with a tip diameter of ∼50 μm attached to a piezobimorph on a micromanipulator (43). One barrel was filled with standard zero-[Ca²⁺]_i solution, and the other barrel was with intracellular solution containing Ca²⁺. Excised patches were switched between streams by applying ∼100 V to the piezobimorph. The time course of solution exchange across the laminar flow interface was estimated by liquid junction potential measurements to be 0.5 ms (10–90% rise time) for a 10-fold difference in ionic strength (Fig. S3). For BK channel recordings, CsCl was replaced with equimolar KCl.

Data Analysis.

Traces were analyzed with Clampfit 9 (Molecular Devices). Conductance (G) from tail currents measured 200 μs after repolarization to −100 mV from various test potentials was normalized to maximum conductance, G_max. G/G_max vs.V_m curves were fitted to a Boltzmann function with Origin 7 (OriginLab). Results are presented as mean ± SEM.