Anoctamins are a family of Ca²⁺-activated Cl⁻ channels

Ca²⁺-activated Cl⁻ currents (CaCC) are abundant and are present in nearly every cell type, where they fulfill very different functions (Hartzell et al., 2005). It is now well accepted that TMEM16A (anoctamin 1, Ano1) forms the plasma membrane localized Ca²⁺-activated Cl⁻ channel (Yang et al., 2008; Schroeder et al., 2008; Caputo et al., 2008). Voltage- and calcium-dependent gating of Ano1 has been examined in detail, and is linked to the first intracellular loop (Xiao et al., 2011). Ano1 is essential for Cl⁻ secretion in a number of epithelial tissues (Ousingsawat et al., 2009; Rock et al., 2009), for smooth muscle contraction (Thomas-Gatewood et al., 2011; Manoury et al., 2010), the function of nociceptive neurons and smooth muscle pacemaker cells (Liu et al., 2010; Hwang et al., 2009; Huang et al., 2009). Ano2, the closest relative of Ano1, has been shown to form a CaCC in olfactory receptors (Sagheddu et al., 2010; Billig et al., 2011).

The family of anoctamins consists of 10 different proteins, but only Ano1, Ano2 and Ano6 have been examined in more detail (Ferrera et al., 2010; Billig et al., 2011; Kunzelmann et al., 2009). For Ano6 a dual role as Fas ligand-activated Cl⁻ channel and phospholipid scramblase has been described (Martins et al., 2011; Suzuki et al., 2010). However, whether Ano3–10 also produce Ca²⁺-activated Cl⁻ currents similar to Ano1 and Ano2, is currently unclear. A recent report claimed that Ano3–7 are intracellular proteins (Duran et al., 2012), although we demonstrated earlier that most anoctamins overexpressed in HEK293 cells can be detected in the cell membrane (Schreiber et al., 2010).

Anoctamins do not show any obvious homology to other ion channels. Ano1 contains eight predicted transmembrane helices, intracellular NH₂- and COOH-ends and a pore, formed by the fifth and sixth transmembrane helices, containing a p-loop dipping back into the membrane (Yang et al., 2008). The other anoctamins (2–10) show a high degree of structural similarity, which is particularly obvious for the putative pore region (Kunzelmann et al., 2009). Apart from some indirect evidence showing enhanced ATP-induced halide permeability in Ano6- and Ano7-expressing cells (Schreiber et al., 2010), it is not clear whether other anoctamins also produce Ca²⁺-activated Cl⁻ currents. We demonstrated earlier that Ano6 produces an outwardly rectifying chloride channel, which is activated during stimulation of Fas receptors upon induction of apoptosis (Martins et al., 2011). Notably, at high intracellular Ca²⁺ concentrations as induced by Ca²⁺ ionophores, Ano6 was reported to operate as a membrane phospholipid scramblase (Suzuki et al., 2010). In contrast to Ano1 and Ano2, Ano6 Cl⁻ currents were not activated by physiological Ca²⁺ concentrations (Martins et al., 2011). As demonstrated in the present report, high intracellular Ca²⁺ levels are required to activate Ano6. Although plasma membrane expression of some anoctamins is very poor, the present experiments also demonstrate for the first time that essentially all anoctamins are able to produce Cl⁻ currents when activated by an increase in [Ca²⁺]_i. We speculate that anoctamins may operate as plasma membrane or intracellular Cl⁻ channels.

Previous studies have demonstrated that Ano1 can be activated by stimulation of G-protein-coupled receptors, such as purinergic P2Y₂ receptors, which increase [Ca²⁺]_i (Yang et al., 2008). Because Ano2 has also been demonstrated to operate as a Ca²⁺-activated Cl⁻ channel (Sagheddu et al., 2010; Billig et al., 2011), we asked whether other anoctamins are also able to produce Ca²⁺-activated Cl⁻ currents. To that end we coexpressed different anoctamins together with P2Y₂ receptors in HEK293 cells, which were stimulated by 10 µM ATP, in the presence of a cytosol-like pipette filling solution and extracellular Ringer solution (see the Materials and Methods). We observed that overexpressed Ano1 and Ano6 produce significant whole-cell Cl⁻ currents, even in the absence of ATP stimulation (Kunzelmann et al., 2011b). In cells expressing Ano1, 4, 7, 8, 9 and 10, whole-cell currents of variable magnitudes were activated by ATP (Fig. 1). Current activation by ATP was fast (below 1 s) for Ano1, but was delayed for the other anoctamins. It is currently unclear whether this is due to a reduced sensitivity towards intracellular Ca²⁺, since intracellular Ca²⁺ levels are increased equally fast in all cells by either ATP or ionomycin, when assessed by Fura-2 (Fig. 2). Increase of [Ca²⁺]_i may not only activate the Cl⁻ channel, but may also trigger inhibition of anoctamins via CAMKII, which may actually counteract further activation. CAMKII inhibition was suggested in a recent study (Tian et al., 2011) and is also demonstrated in the present paper for both, Ano1 and Ano6. For all anoctamins except of Ano4, membrane voltages were depolarized through stimulation of purinergic receptors and increase of [Ca²⁺]_i, which is expected upon increase of a Cl⁻ conductance along with cation influx through store operated Ca²⁺ channels (Fig. 1B). Surprisingly, expression of Ano6 hyperpolarized the membrane voltage of non-stimulated cells, which is probably due to its partial permeability for K⁺ ions as described further below. We wondered whether largely enhanced baseline Cl⁻ currents in Ano1-expressing cells are due to activation of by the baseline Ca²⁺ concentration of 100 nM (Fig. 2A). In fact, 30 min pre-incubation with the Ca²⁺ chelator BAPTA-AM (25 µM) abolished this baseline Cl⁻ conductance (Fig. 2C). We hypothesize that an inhibitory factor or additional protein is missing that keeps overexpressed channels closed under control conditions. As described below, this unknown protein could be another anoctamin, like Ano9, which may be able to heterooligomerize with Ano1 (Fig. 6G). Moreover, since expression of anoctamins also induced a cation permeability (cf. below), we exclude a possible contribution of endogenous Ca²⁺-activated SK4 K⁺ channels: Activation of whole-cell currents in Ano1-expressing HEK293 cells by ionomycin was identical in the absence or presence of 10 µM of the SK4 inhibitor clotrimazole (Fig. 2D). No message (RT-PCR) was detected for either SK4 or the large conductance Ca²⁺-activated (BK) K⁺ channel in HEK293 cells (data not shown).

Stimulation by 10 µM ATP raised intracellular Ca²⁺ levels to about 1 µM, which is mainly due to release of Ca²⁺ from intracellular ER stores (Fig. 2A) (Barro-Soria et al., 2010). This Ca²⁺ increase may not be sufficient to activate every member of the anoctamin family. We therefore stimulated the cells with the Ca²⁺ ionophore ionomycin (1 µM), which raised [Ca²⁺]_i in a more sustained manner to levels well above 1 µM (Fig. 2B). In fact whole-cell currents could be activated in cells expressing Ano5 or Ano6, when stimulated by ionomycin, suggesting that these channels require higher and maybe more sustained Ca²⁺ levels to be activated (Fig. 3A). Close to the plasma membrane ionomycin raises [Ca²⁺]_i probably to much higher levels than suggested by Fura-2 measurements. We examined the Ca²⁺ dependence of Ano6 by varying cytosolic (patch pipette) Ca²⁺ concentrations and detected substantial activation of Cl⁻ currents only at concentrations as high as 10 µM or larger (Fig. 3D). Moreover, Ano6 shows little inactivation during continuous stimulation with 1 µM ionomycin (Fig. 3B). Anion selectivity was similar for Ano1, 4, 6 and 10 and was I⁻>Br⁻>Cl⁻>HCO₃⁻, suggesting a low field strength anion selectivity. HCO₃⁻ was conducted surprisingly well by Ano6 (Fig. 3C). It is therefore entirely possible that Ano6 serves as a channel for bicarbonate secretion in airways and other epithelial tissues. Notably, at very high [Ca²⁺]_i of 0.1 and 1 mM, whole-cell Cl⁻ currents were also activated in mock-transfected cells, which express Ano6 endogenously (Fig. 3D,E). Replacement of extracellular Cl⁻ with gluconate strongly inhibited whole-cell currents and depolarized membrane voltages, indicating activation of whole-cell Cl⁻ currents at high [Ca²⁺]_i in both mock-transfected and Ano6-overexpressing cells (Fig. 3D–F). We speculate that endogenous anoctamins may be activated only at very high [Ca²⁺]_i (Almaça et al., 2009). Indeed when endogenous Ano6 (together with Ano1, 8 and 9) were knocked down simultaneously by siRNA, we found reduced baseline currents. Currents activated by 1 mM [Ca²⁺] in the patch pipette filling solution were largely attenuated (Fig. 3G). Because of the lack of suitable antibodies we assessed knockdown of anoctamins by real time RT-PCR. Expression levels (relative to β-actin) were very different for the endogenous anoctamins, with anoctamin 6 being the most abundant anoctamin (n = 3 for all): Ano1 1.4×10⁻⁶ (downregulation by 52%); Ano6 3.1×10⁻⁴ (downregulation by 82%); Ano8 1.3×10⁻⁴ (downregulation by 42%); Ano9 2.0×10⁻⁸ (downregulation by 76%). Thus Ano6 and maybe Ano8 have probably the largest impact on endogenous Ca²⁺-activated Cl⁻ conductance. It remains currently unclear why endogenous anoctamins require such high [Ca²⁺]_i to be activated.

In more than 20 cell lines and in oocytes we examined endogenous expression of anoctamins. Ano6 and Ano8 were always expressed most abundantly (Kunzelmann et al., 2009). Thus, there is no ‘empty’ system without background expression of endogenous anoctamins. However, given the low levels for endogenous anoctamins compared to overexpressed proteins (about 1000 times higher according to real time RT-PCR analysis), possible heterooligomerization of endogenous and overexpressed proteins does not appear to be a major problem in the present study.

Cl⁻ currents activated by ionomycin or high intracellular (pipette) Ca²⁺ concentrations demonstrated rather linear current/voltage relationships (Fig. 4). We compared currents activated by ionomycin in the presence of extra-and intracellular cations (cytosol-like pipette filling solution and bath Ringer solution), with currents detected in the absence of extra- and intracellular cations (patch pipette 132 mM CsCl, bath 132 mM NMDGCl; see the Materials and Methods). In general, replacing intracellular K⁺ by Cs⁺ and extracellular Na⁺ by impermeable NMDG⁺ markedly reduced whole-cell currents produced by some anoctamins (Fig. 4). We further examined possible cation permeability of several anoctamins (1, 6 and 10) by replacing intracellular K⁺ and extracellular Na⁺ by NMDG⁺ separately (Fig. 5). Indeed we found that removal of intracellular K⁺ depolarized membrane voltage and reduced outward currents, while removal of bath Na⁺ hyperpolarized membrane voltage and reduced inward (and outward) currents (Fig. 5). Particularly in Ano6-expressing cells a permeability for K⁺ ions was detected. Indirect evidence for cation conductance of anoctamins has already been provided earlier (Kunzelmann et al., 2011b; Kunzelmann et al., 2012). Moreover, anoctamins are also permeable for Ca²⁺, as reported for Ano6 (Yang et al., 2011). Taken together these results suggest that anoctamins are either poorly selective for anions or lead to parallel activation of a cation conductance.

Increase in intracellular Ca²⁺ activates anoctamins only transiently. On the other hand, overexpression of Ano1 in HEK293 and other cell types leads to channels that are partially active even under baseline Ca²⁺ concentrations of 100 nM (Figs 1, 2) (Kunzelmann et al., 2011b). We observed that during transient activation/deactivation of Ano1 whole-cell Cl⁻ currents declined even below the initial high current levels, This was observed during continuous stimulation by ATP (10 µM) or ionomycin (1 µM), or when 1 µM Ca²⁺ was provided through the patch pipette filling solution (1 µM; Fig. 6A). This suggests that increase in [Ca²⁺]_i initially activates the channel but then triggers inactivation by a Ca²⁺-dependent process. Spontaneous current inactivation was not due to a loss of ATP or calmodulin, since we also observed inactivation when both components (3 mM ATP, 2 µM calmodulin) were present in the patch pipette. Because Ca²⁺/calmodulin-dependent kinase II (CAMKII) inhibits overexpressed Ano1 currents (Tian et al., 2011), we hypothesized that an increase of Ca²⁺ may activate CAMKII, which in turn inhibits Ano1. In fact, in the presence of the CAMKII blocker KN62 (5 µM), inactivation of the channel was significantly attenuated (Fig. 6B). Similar to Ano1, also Ano6 contains four putative CAMKII phosphorylation sites in the N-terminus (supplementary material Fig. S2). Thus inhibition of CAMKII by 10 µM KN62 also increased activation of Ano6 by 1 µM ionomycin [ΔG_Iono = 29.6±3.6 (w/o KN62) versus 43.3±4.5 nS (w/KN62); n = 5].

Moreover, when patch pipettes were loaded with pre-activated CAMKII (5000 U/ml), activation of whole-cell currents in Ano1-overexpressing cells was significantly reduced (Fig. 6C). Thus CAMKII inhibits overexpressed Ano1, which is reminiscent to inhibition of CaCC by CAMKII in smooth muscle cells (Greenwood et al., 2001). Along this line, a splice variant of Ano1 (ac-Ano1) had been reported to have a higher Ca²⁺ sensitivity than the splice form abc-Ano1 (Ferrera et al., 2009), which we used in our experiments. As reported by Ferrera et al., we also found that the channel is activated at [Ca²⁺]_i below 100 nM (Fig. 6D). However, we also found that ac-Ano1 currents inactivated at lower Ca²⁺concentrations than the splice form abc-Ano1 (Fig. 6D). We tried to further localize the molecular site within Ano1 that is in charge for its only transient activation. Since activation of Ano6 produces more sustained currents (Martins et al., 2011), we generated a chimeric protein in which the N-terminus of Ano1 was replaced with that of Ano6 (Ano1N6). Notably, whole-cell currents generated by the chimeric Ano1N6 channel were no longer transient, but generated a rather sustained current when activated by 1 µM ionomycin (Fig. 6E,F). This suggests that the N-terminus controls both activation and inactivation of anoctamins. Notably, potential binding sites for calmodulin have been found in the N-terminus of Ano1, along with a number of putative phosphorylation sites, including those for CAMKII (Kunzelmann et al., 2009).

The data presented above indicate that Ano6, 8 and 9, when overexpressed in HEK293 cells, produce Cl⁻ currents upon increase in intracellular Ca²⁺ by ionomycin. However, although all three paralogs are expressed endogenously in HEK293 cells, no currents are activated in mock-transfected cells (Fig. 3A) (Almaça et al., 2009). However, when increasing [Ca²⁺]_i to very high levels (≥10 µM), endogenous anoctamins were activated (Fig. 3B,E). This may be due to a lack of sufficient membrane expression of endogenous anoctamins. Endogenous anoctamins are expressed at much lower levels than overexpressed proteins and may be inhibited by endogenous accessory proteins, such as CAMKII. Alternatively, specific Ca²⁺ release and/or influx pathways need to be triggered to activate these endogenous anoctamins. The overall linear structures appear similar in all anoctamins (supplementary material Fig. S1) and an analysis of putative functional domains revealed a number of similar motifs and putative phosphorylation sites in all ten anoctamins (supplementary material Fig. S2). Notably, Ano1 exists as a stable dimer (Fallah et al., 2011; Sheridan et al., 2011), and may even form heterooligomeric proteins with other anoctamins (Schreiber et al., 2010). In the present study we therefore coexpressed different anoctamins and found that only coexpression of Ano9 inhibited Ano1 currents activated by ATP, suggesting a specific interaction of different anoctamin paralogs (Fig. 6G).

The present results suggest that endogenous and overexpressed anoctamins behave differently. We found earlier that calmodulin and 1-EBIO, a calmodulin-dependent activator of K⁺ channels, activate overexpressed Ano1 (Tian et al., 2011). In the present study we made use of a well-established technique to assess endogenous Ca²⁺-activated Cl⁻ currents in HT₂₉ cells stably expressing I⁻-sensitive yellow fluorescent protein (YFP). Cellular YFP fluorescence is quenched by iodide influx from the bath solution (Verkman and Galietta, 2009) (Fig. 7A–C). Stimulation of purinergic receptors by ATP, dose dependently activated I⁻ influx and YFP quenching, indicating activation of a Ca²⁺-dependent Cl⁻ conductance (Fig. 7A,B). ATP-activated Cl⁻ conductance was blocked by typical Ano1-inhibitors (Fig. 7C). Moreover, siRNA knockdown of Ano1 inhibited ATP-induced whole-cell currents in HT₂₉ cells in another study (Tian et al., 2012). We examined the effects of 1-EBIO on Ca²⁺-dependent Cl⁻ conductance in HT₂₉ cells. However, even at a concentration of 1 mM 1-EBIO was unable to directly activate a Cl⁻ conductance (data not shown). It only slightly and dose dependently augmented ATP (1 µM)-activated Cl⁻ conductance (Fig. 7D). A number of 1-EBIO-related compounds (DCEBIO, riluzole, CBIQ, NS8593, CYPPA, NS309, NS4591) were examined in a dose-dependent manner. However, for none of them we found activation of a Cl⁻ conductance and receptor-mediated activation by ATP was not augmented (data not shown). We further performed experiments in a human airway epithelial cell line expressing the most common mutant form of CFTR, F508del-CFTR (CFBE/F508del-CFTR). We found: (1) that ATP (10 µM) activated a whole-cell Cl⁻ conductance of 3.5±0.4 nS (n = 5), while direct application of 1-EBIO (100 µM) did not. (2) Activation of the whole-cell Cl⁻ conductance through ATP stimulation was slightly augmented by 0.7±0.08 nS (n = 5) in the presence of 1-EBIO. Taken together these results demonstrate (i) that endogenous Ano1 behaves differently than overexpressed Ano1, and (ii) that 1-EBIO is probably not useful to directly activate Ca²⁺-activated Cl⁻ secretion in airways of CF patients.

Finally, we detected differential regulation by protein kinase C of Ano1 overexpressed in HEK293 cells and endogenous Ano1 in HT₂₉ cells. While inhibition of protein kinase C by preincubation with bisindolylmaleimide 1 (BIM; 0.1 µM/2 hrs) did not affect activation (by 1 µM ionomycin) of overexpressed Ano1 (Fig. 7E) (Tian et al., 2011), it augmented endogenous Ano1 currents in HT₂₉ cells (Fig. 7F). Thus, properties and regulation of overexpressed anoctamins may not fully reflect those of native channels.

A recent report did not find membrane expression or Ca²⁺-activated Cl⁻ currents for Ano3–7 (Duran et al., 2012). This is in contrast to our earlier results which demonstrate plasma membrane expression for Ano1, 2, 5, 6, 7 and 9 when overexpressed in HEK293 and FRT cells (Schreiber et al., 2010). Moreover, it was shown recently that Ano6 is an essential component of outwardly rectifying Cl⁻ channels (ORCC) and, at the same time, operates as a membrane localized phospholipid scramblase (Martins et al., 2011; Suzuki et al., 2010). Moreover, also Ano5 and Ano7 were detected in the plasma membrane in earlier studies (Mizuta et al., 2007; Das et al., 2007). We re-examined plasma membrane expression of overexpressed anoctamins in HEK293 cells and, again, found clear membrane expression for Ano1, 4, 5, 6 and 7 and a weak membrane staining for Ano8, 9 and 10 (Fig. 8). We produced additional YFP fusion proteins for Ano6, 8 and 9. Life cell imaging of YFP fluorescence suggested (partial) membrane expression for Ano6, 8 and 9 (supplementary material Fig. S3). For Ano10 we managed to introduce a FLAG tag into the third extracellular domain and labeled non-permeabilized cells. The results show a spotted membrane expression of Ano10 (supplementary material Fig. S3). Thus the present data fully confirm those obtained in earlier studies (Schreiber et al., 2010).

Nevertheless, overexpressed Ano8, 9 and 10 show poor membrane expression. Since endogenous anoctamins are expressed at much lower levels, these anoctamins may not even accumulate at significant numbers in the plasma membrane of native cells. It is therefore entirely possible that Ano8, 9 and 10 have an intracellular rather than plasma membrane function. Currently available commercial antibodies are mostly of insufficient quality to stain endogenous anoctamins. Noteworthy, in contrast to Duran and colleagues (Duran et al., 2012), we expressed only human anoctamins and acutely activated anoctamins, by either receptor stimulation or by ionomycin. Since activation of most anoctamins was found to be transient, it is possible to overlook these currents, particularly if they are small, as those found for Ano5, 7, 8 and 9 (Figs 1, 2). In summary, our present results suggest that anoctamins form a family of Ca²⁺-activated Cl⁻ channels, which may have their function in the plasma membrane or in intracellular compartments. In addition to their property as Ca²⁺-activated Cl⁻ channels, anoctamins may either have some permeability for cations, or their expression activates independent cation channels. These results may help to understand the physiological function of anoctamins and their role in disease.

HEK293 and HT₂₉ cells were grown in DMEM and DMEM-F12, separately (GIBCO, Karlsruhe, Germany) supplemented with 10% fetal bovine serum at 37°C in a humidified atmosphere with 5% CO₂. Cells were plated on fibronectin- and collagen-coated 18 cm diameter coverslips and co-transfected with cDNA encoding either hTMEM16A, D, E, F, G, H, J, K, or empty pcDNA3.1 vector (mock) along with P2Y2 receptor and CD8. Expression of TMEM16A, F, H, J was suppressed each by two independent sets of RNAi. Duplexes of 25nucleotide of RNAi were designed and synthesized by Invitrogen (Paisley, UK) and Ambion (Darmstadt, Germany). RNAi was transfected using Lipofectamine 2000 (1 µg/µl) and cells were examined 48 h or 72 h after transfection. Transfections were carried out using Lipofectamine 2000 (Invitrogen) according to the manufacturer's protocol. Successful knockdown of anoctamins was demonstrated by real time RT-PCR (Kunzelmann et al., 2011a).

Following transfection (2–3 days) of His-tagged anoctamins or siRNA, respectively, overexpressing HEK293 or siRNA-transfected HT₂₉ cells were identified by incubating the cells 1–2 min with Dynabeads CD8 (Invitrogen). His-tagging of anoctamins did not interfere with its properties as a Cl⁻ channel. Coverslips were mounted on the stage of an inverted microscope (IM35, Zeiss) and kept at 37°C. The bath was perfused continuously with Ringer solution (mM: NaCl 145, KH₂PO₄ 0.4, K₂HPO₄ 1.6, D-glucose 5, MgCl₂ 1, calcium gluconate 1.3, pH 7.4) or NMDG solution (NMDG 145, HCl 130, KH₂PO₄ 0.4, K₂HPO₄ 1.6, D-glucose 5, MgCl₂ 1, calcium gluconate 1.3, pH 7.4) at the rate of 5 ml/min. For fast whole-cell patch clamping pipettes were filled with intracellular like solution containing (mM: KCl 30, potassium gluconate 95, NaH₂PO₄ 1.2, Na₂HPO₄ 4.8, EGTA 1, calcium gluconate 0.758, MgCl₂ 1.034, D-glucose 5, ATP 3, pH was 7.2) or CsCl solution containing (mM: CsCl 130, NaH₂PO₄ 1.2, Na₂HPO₄ 4.8, EGTA 1, calcium gluconate 0.758, MgCl₂ 1.034, D-glucose 5, ATP 3, pH was 7.2) and had an input resistance of 2–4 MΩ. In other experiments K⁺ in the pipette filling solution was replaced with NMDG⁺. For all solutions Ca²⁺ concentrations were adjusted to 10⁻⁷ (pipette) and 1.3 mM (bath). Experiments were conducted as described earlier (Ousingsawat et al., 2009). During experiments cells were current clamped and membrane voltages were measured. At intervals cells were voltage clamped to ±50 mV in steps of 10 mV for 1 s. The applied clamp voltages were corrected for liquid junction potentials. Because current voltage relationships were linear (except for Ano6), membrane conductances were usually calculated according to Ohm's law. For some experiments, patch pipettes (P6060S; New England Biolabs, Germany) were loaded with pre-activated calmodulin-dependent kinase II (5000 U/ml). In these experiments, CAMKII was given sufficient time (40 min) to enter the cell before stimulation with ionomycin, and the cell was current clamped during this time.

For single cell fluorescence measurements HEK293 cells were grown on glass covers slips, mounted in a cell chamber and perfused with ringer solution at 8 ml/min at 37°C. Cell fluorescence measurements was measured continuously on an inverted microscope Axiovert S100 (Zeiss, Germany) with a Flua 20×/0.75 objective (Zeiss, Germany) and a high speed polychromator system (VisiChrome, Visitron Systems, Germany). Cells were loaded with 2 µM Fura-2, AM (Molecular Probes) in ringer solution or Opti-MEM (GIBCO) under experimental conditions with 0.2% pluronic (Molecular Probes) for 1 h at 37°C. Fura-2 was excited at 340/380 nm, and the emission was recorded between 470 and 550 nm using a CCD camera (CoolSnap HQ, Visitron Systems, Germany). Control of experiment, imaging acquisition, and data analysis were performed using the software package Meta-Fluor (Universal imaging, USA) and Origin (OriginLab Corporation, USA). For calibration of intracellular Ca²⁺ concentration cells were perfused with Ringer solution and Ca²⁺-free Ringer with 1 µM ionomycin, 10 µM monensin and 5 µM nigericin.

Quenching of the intracellular fluorescence generated by the iodide-sensitive enhanced yellow fluorescent protein (YFP) was used to measure anion conductance. YFP fluorescence was excited at 490 nm using a semi-automatic Novostar plate reader (BMG-Labtech, Offenburg, Germany). I⁻ influx was induced by replacing 20 mM extracellular Cl⁻ with I⁻. Background fluorescence was subtracted and autofluorescence was negligible. Changes in fluorescence induced by I⁻ are expressed as initial rates of fluorescence decrease (arbitrary units/sec).

Transfected HEK293 cells were grown on glass coverslips and fixed for 10 min with 4% (w/v) paraformaldehyde at room temperature. Cells were incubated for 5 min with 0.1% SDS in PBS. After washing, cells were permeabilized and blocked with 2% (w/v PBS) BSA and 0.04% (v/v PBS) Triton X-100 for 1 h and incubated overnight with primary anti-His AB (1∶500, Qiagen, Hilden, Germany) and anti-β-catenin (1∶500, Sigma, Taufkirchen, Germany) at 37°C. Binding of the primary antibody was visualized by incubation with a secondary donkey anti-mouse antibodies conjugated with Alexa Fluor 546 (1∶400, Molecular Probes, Invitrogen). Nuclei were stained with Hoe33342 (0.1 µg/ml PBS, Aplichem, Darmstadt, Germany). β-catenin was visualized using an Alexa-Fluor-488-labeled secondary antibody. Cells were mounted on glass slides with fluorescent mounting medium (DAKO Cytomation, Hamburg, Germany) and examined with an ApoTome Axiovert 200 M fluorescence microscope (Zeiss, Göttingen, Germany).

We thank Prof. Dr M. Amaral for providing the HT₂₉ cell line stably expressing YFP. CBIQ, NS8593, CYPPA, NS309 and NS4591 were generously provided by NeuroSearch (Ballerup, Denmark).