TMEM16A, A Membrane Protein Associated with Calcium-Dependent Chloride Channel Activity

See allHide authors and affiliations

Science  24 Oct 2008:
Vol. 322, Issue 5901, pp. 590-594
DOI: 10.1126/science.1163518


Calcium-dependent chloride channels are required for normal electrolyte and fluid secretion, olfactory perception, and neuronal and smooth muscle excitability. The molecular identity of these membrane proteins is still unclear. Treatment of bronchial epithelial cells with interleukin-4 (IL-4) causes increased calcium-dependent chloride channel activity, presumably by regulating expression of the corresponding genes. We performed a global gene expression analysis to identify membrane proteins that are regulated by IL-4. Transfection of epithelial cells with specific small interfering RNA against each of these proteins shows that TMEM16A, a member of a family of putative plasma membrane proteins with unknown function, is associated with calcium-dependent chloride current, as measured with halide-sensitive fluorescent proteins, short-circuit current, and patch-clamp techniques. Our results indicate that TMEM16A is an intrinsic constituent of the calcium-dependent chloride channel. Identification of a previously unknown family of membrane proteins associated with chloride channel function will improve our understanding of chloride transport physiopathology and allow for the development of pharmacological tools useful for basic research and drug development.

Electrogenic chloride transport across cellular membranes is mediated by ion channels, which have been classified on the basis of their mechanism of activation. Accordingly, there are Cl channels regulated by cyclic adenosine monophosphate (cAMP), Ca2+, cell-volume changes, and membrane potential (1). Ca2+-activated Cl channels (CaCCs) are involved in important physiological processes such as electrolyte/fluid secretion, smooth muscle excitability, and olfactory perception, but their molecular identity is still unclear and controversial (2, 3). The proteins that have been proposed as main constituents of CaCCs include CLC-3 (4), bestrophins (5, 6), and members of the chloride channel, calcium-activated (CLCA) family (7, 8). CLCA proteins are unlikely candidates because they are secreted into the extracellular medium (9). CLC-3 and bestrophin gene expression cause the appearance of Cl currents that lack the typical voltage dependence of CaCCs (2, 3, 10, 11). Therefore, it is likely that the molecular identity of CaCCs remains only partially defined.

Long-term stimulation of airway epithelial cells with interleukin-4 (IL-4) causes a marked increase in Ca2+-activated Cl secretion (12) (fig. S1). Because this effect may be caused by increased mRNA levels of the corresponding channel gene, we used this response to identify the proteins constituting the CaCC. Therefore, we performed a microarray-based gene expression analysis on resting and IL-4–treated bronchial epithelial cells and found a large set of proteins whose corresponding mRNA is markedly up-regulated by the cytokine (13). These proteins included chemokines, cell adhesion molecules, transcription factors, other regulatory factors, and a group of putative membrane proteins with unknown functions (TMTC3, TSPAN8, KIAA1126, SIDT1, and TMEM16A) that show different levels of stimulation by IL-4 (fig. S1). Up-regulation by IL-4 was confirmed by real-time reverse transcription polymerase chain reaction (RT-PCR). For example, TMEM16A mRNA was increased approximately sevenfold after IL-4 treatment.

To further analyze these candidate channels by gene silencing, we used CFPAC-1, a pancreatic cell line with abundant CaCC activity (14), and CFBE41o–, a cell line derived from human bronchial epithelium (15). We transfected small pools of small interfering RNA (siRNA) against each of the putative membrane proteins up-regulated by IL-4. siRNA against TMC5, an unknown membrane protein with possible channel function but not affected by IL-4, served as a control. siRNA-transfected CFPAC-1 and CFBE41o–cells were assessed for CaCC activity with an assay based on a halide-sensitive yellow fluorescent protein (YFP) (16, 17). Cells with stable YFP expression were stimulated with uridine 5′-triphosphate (UTP) (100 μM), which elicits a purinergic receptor–mediated increase of intracellular Ca2+. The Ca2+ increase triggered a rapid fluorescence decrease due to a large I influx through CaCCs (Fig. 1A). Cells transfected with siRNA against TMEM16A showed a 60 to 70% reduction in Ca2+-dependent I influx as compared with cells treated with control siRNA or siRNA against other targets (Fig. 1B and fig. S2). To confirm these results, we transfected CFPAC-1 and CFBE41o–cells with a single siRNA against TMEM16A obtained from a commercial source different from where we obtained the siRNA for the first screening (Fig. 1C and fig. S2). Anti-TMEM16A RNA duplexes caused in both cell types a substantial reduction in CaCC activity. The same degree of inhibition by anti-TMEM16A siRNA was obtained when Ca2+-elevation was triggered with ionomycin (1 μM) instead of UTP, which indicated that TMEM16A silencing affects a step downstream of purinergic receptor activation and cytosolic Ca2+ increase. CaCC activity was also unaffected when cells were transfected with siRNA against two TMEM16A homologs, TMEM16F and TMEM16K.

Fig. 1.

Down-regulation of Ca2+-dependent anion transport by TMEM16A silencing. (A to C) CaCC assay based on the halide-sensitive YFP in pancreatic CFPAC-1 cells. (A) Representative traces showing cell fluorescence quenching upon the addition of extracellular I plus UTP (arrow). (B) Summary of Ca2+-activated I influx in untreated cells, in cells transfected with non-targeting siRNA (NT-siRNA), or transfected with siRNA pools against the indicated genes (mean ± SEM, n = 8 experiments per condition). (C) Ca2+-activated I influx as in (B) but with a single siRNA against TMEM16A versus three nontargeting siRNA with different guanosine and cytosine content (mean ± SEM, n = 5 experiments per condition). Cells were stimulated with UTP (100 μM) or ionomycin (1 μM). ** P < 0.01 versus nontargeting siRNA. (D and E) Short-circuit current (Isc) recordings on polarized monolayers of primary human bronchial epithelial cells. Ca2+-dependent Cl secretion was triggered with apical UTP (100 μM). Cells were transfected with nontargeting or anti-TMEM16A siRNA and then left in control conditions (D) or incubated with IL-4 (E). (Top) Representative traces. (Bottom) Summary of results (mean ± SEM, n = 4 experiments per condition). ** P < 0.01 versus nontargeting siRNA. (F) Whole-cell membrane currents (Im) from CFPAC-1 cells. Pipette (intracellular) solution contained 600 nM free Ca2+. (Top) Representative currents elicited at voltages in the –100- to +100-mV range. (Bottom) Membrane currents measured at the end of voltage pulses (Vm) (mean ± SEM, n = 10 to 20 experiments) are plotted against the applied membrane potential. Conditions were nontransfected, transfected with nontargeting siRNA, transfected with anti-TMEM16A, and nontransfected but with nominal 0 Ca2+ in the intracellular solution. The currents measured in TMEM16A-silenced cells were significantly smaller than those of control-transfected cells at +20 mV (P < 0.05) and at +40 to +100 mV (P < 0.01).

We also measured the activity of CaCCs and the effect of TMEM16A silencing with the short-circuit current technique (17). Primary cultures of human bronchial epithelial cells were transfected with nontargeting or anti-TMEM16A single siRNA at the time of plating on porous membranes. After 8 to 10 days, when the cells were differentiated and polarized, they were treated for 24 hours with or without IL-4 (10 ng/ml). Addition of UTP (100 μM) to the apical membrane elicited a transient increase of the current due to Ca2+-dependent Cl secretion (Fig. 1D), and this response was up-regulated in cells treated with IL-4 (Fig. 1E). In agreement with data obtained with the YFP assay, the UTP-dependent Cl current was significantly reduced in cells previously transfected with anti-TMEM16A siRNA, with or without IL-4 treatment (Fig. 1, D and E). This effect was not due to inhibition of the purinergic-dependent Ca2+ signal, because the Ca2+ increase triggered by UTP was not diminished after transfection with TMEM16A-specific siRNA as compared with control-transfected cells (fig. S3). This result indicates that TMEM16A is not involved in intracellular Ca2+ signaling or homeostasis but is probably more directly involved in Cl transport. Silencing of TMEM16A also caused CaCC activity inhibition when the basolateral membrane was permeabilized (fig. S4).

In additional short-circuit current experiments on CFPAC-1 cells, UTP triggered a fast transient current increase due to the activation of apical CaCCs (14, 18). This current was absent in cells transfected with anti-TMEM16A siRNA (fig. S5). The currents in TMEM16A-silenced cells were similar to those of cells treated with niflumic acid, a classical blocker of CaCCs channels (fig. S5).

Whole-cell patch-clamp experiments were carried out on CFPAC-1 cells to further confirm the silencing of CaCCs by anti-TMEM16A siRNA. Using a micropipette (intracellular) solution containing 600 nM free Ca2+, we recorded typical CaCC currents (14) with time-dependent activation at positive membrane potentials (Fig. 1F). The steady-state current-voltage relationship showed a strong outward rectification. Returning to the negative holding potential at the end of positive test pulses generated inward tail currents that slowly inactivated. CFPAC-1 cells tranfected with anti-TMEM16A siRNA showed a marked reduction of CaCC currents as compared with cells transfected with control siRNA (Fig. 1F). Outward currents at positive voltages, and corresponding tail currents, were inhibited. In fact, membrane currents after TMEM16A silencing, were similar to those of nonsilenced cells recorded with a Ca2+-free micropipette solution (Fig. 1F).

In silico gene expression data indicate that TMEM16A is preferentially expressed in exocrine glands and organs rich in glands. In mice (see for tissue distribution), TMEM16A is highly expressed in the mammary glands, prostate, large intestine, lung, trachea, uterus, and vomeronasal organ. We compared expression between CFPAC-1 and 9HTEo–cells. Real-time RT-PCR in CFPAC-1 cells, which have a large CaCC activity (14), showed that expression of TMEM16A mRNA was ∼200-fold higher than in 9HTEo–cells, a tracheal epithelial cell line in which UTP and other Ca2+-elevating agents induce a small activation of a different channel, the swelling-activated Cl channel (19).

TMEM16A is alternatively spliced, generating multiple protein isoforms with various combinations of alternative protein segments a, b, c, and d (fig. S6). All isoforms were predicted to maintain a basic structure consisting of eight transmembrane helices with N and C termini lying on the cytosolic side. To determine which isoform is expressed in bronchial epithelial cells, we performed RT-PCR experiments with primers flanking the entire coding sequence. Amplification products were cloned in plasmids, and 22 clones were fully sequenced. We found that bronchial epithelial cells expressed an isoform, TMEM16A(abc), that has 982 amino acids (13).

To further validate the involvement of TMEM16A in CaCC function, we transiently cotransfected COS-7 and HEK-293 cells with a plasmid carrying the halide-sensitive YFP and plasmids coding for various TMEM16A isoforms (fig. S6). In both cell types, transfection of TMEM16A cDNAs caused a substantial increase of the halide transport triggered by ionomycin as compared with cells transfected with the YFP alone. The effect was particularly dramatic for isoforms (abcd), (abc), and (ac): Up to 60 to 80% of cell fluorescence was quenched by I influx in the first 10 s. Accordingly, Ca2+-dependent I influx was 25 to 30 times larger than in mock-transfected cells (fig. S6). Conversely, the Ca2+-dependent halide transport generated by TMEM16A(0) was considerably smaller than that generated by transfection with other isoforms.

The correlation between TMEM16A and CaCC channels was further investigated by stable transfection of TMEM16A in Fischer rat thyroid (FRT) cells. Such cells have low endogenous CaCC transport and are able to form tight epithelia, suitable for transepithelial Cl current measurements (16, 17). Null FRT cells responded to ionomycin with a small current increase (1 to 2 μA). The pool of FRT cells arising from stable TMEM16A(abc) transfection showed a ninefold increase in ionomycin-induced current (Fig. 2A). Isolation of pure clones from the pool led to the generation of cells with very large ionomycin-induced currents, up to 170-fold larger than in null FRT cells (Fig. 2A). These currents were strongly inhibited by niflumic acid (100 μM). We further studied pharmacological sensitivity by using the YFP assay. TMEM16A-dependent anion transport was highly sensitive to niflumic acid and 5-nitro-2-(3-phenylpropylamino) benzoic acid (NPPB) but not to diphenylamine carboxylate and cystic fibrosis transmembrane conductance regulator inhibitor–172 (CFTRinh-172) (fig. S7).

Fig. 2.

CaCC activity in stable-transfected FRT cells. (A) Short-circuit current recordings on FRT cells. (Left) Representative experiments showing the effect of apical ionomycin addition (arrows) in null cells, a pool of cells transfected with TMEM16A, or a pure clone (no. 77) with stable expression of TMEM16A(abc). (Right) Summary of results (mean ± SEM, n = 4 to 5 experiments). Currents on clone 77 were also measured in the presence of niflumic acid (100 μM). ** P < 0.01 versus null cells. (B) Immunofluorescence images from FRT cells stably-transfected with the FLAG-tagged TMEM16A(abc) (top) and from null FRT cells (bottom). Cells were stained with antibody to FLAG (green) and counterstained with 4′,6′-diamidino-2-phenylindole (blue). (C) Whole-cell membrane currents recorded in FRT cells stably-transfected with TMEM16A(abc) or in null cells at the indicated intracellular free-Ca2+ concentration. (D and E) Current-voltage relationships for TMEM16A(abc)-expressing cells showing dependence on intracellular free Ca2+ (D) or extracellular Cl (E). In the latter type of experiments, intracellular Ca2+ was 600 nM.

To study the subcellular localization of TMEM16A protein, we introduced by mutagenesis a FLAG epitope in the N terminus of isoform (abc) at the end of segment (a). This construct was transiently transfected and stable-transfected in human embryonic kidney (HEK) 293 and FRT cells, respectively. Immunofluorescence showed a pattern consistent with a plasma membrane localization of the TMEM16A protein (Fig. 2B). The protein with the FLAG epitope was fully functional (fig. S6).

Whole-cell patch-clamp experiments revealed that FRT cells with stable expression of TMEM16A(abc) manifested membrane currents with the typical voltage dependence of CaCC (Fig. 2C). The shape and size of the currents were dependent on the free-Ca2+ concentration in the pipette (cytosolic) solution (Fig. 2, C and D). In contrast, none of the null cells showed CaCC-like currents (Fig. 2C). The maximum current, measured at +100 mV in null cells with 600 nM free Ca2+ in the pipette solution, was 42 ± 8 pA (n = 10 experiments), a value 25 times smaller than in TMEM16A-transfected cells under identical conditions (n = 16 experiments; P < 0.01). TMEM16A-dependent currents were sensitive to extracellular Cl concentration. Lowering of extracellular Cl by replacement with gluconate from 154 to 4 mM strongly abolished outward currents (Cl entering the cell) and shifted the reversal potential by 41.2 ± 40.1 mV in the positive direction (n = 5 experiments) (Fig. 2E), the expected shift for a perfectly selective Cl channel being 90 mV. Such a difference may indicate that the underlying channel has a small but notable permeability to gluconate (20).

We transiently transfected the plasmid carrying the coding sequence for isoform (abcd) in HEK-293 cells for whole-cell patch-clamp analysis. As expected for a transfection with an estimated 20 to 30% efficiency, 18 out of 65 cells showed membrane currents with CaCC biophysical characteristics (Fig. 3A). This current was never observed in 27 mock-transfected cells. After averaging the results from all cells, without selecting for the ones having voltage-dependent currents, we found a considerable difference with respect to mock-transfected cells. For example, at +100 mV the currents were approximately fivefold larger in (abcd)-transfected cells (Fig. 3, A and B). Cl selectivity of currents evoked by TMEM16A transfection was demonstrated by extracellular Cl replacement with gluconate. Under these conditions, outward currents were strongly decreased and the reversal potential was shifted by 32.9 ± 1.5 mV in the positive direction. When we expressed the (0) isoform in HEK-293, we measured currents that lacked the time-dependent activation at positive voltages of (abc) and (abcd) isoforms. Although small, such currents were substantially larger than those of mock-transfected cells (Fig. 3, A and B).

Fig. 3.

Induction of Cl currents by transient TMEM16A expression. (A) Representative membrane currents measured in HEK-293 cells. Each panel is the overlap of currents elicited in a single cell at membrane potentials in the –100 to +100 mV range. Cells were transfected with null plasmids or with plasmids coding for (abcd) or (0) isoforms. (B) Current-voltage relationships from experiments as those shown in (A). Each point is the mean ± SEM of currents measured at the end of voltage pulses (n = 27 to 65 experiments). Values for (abcd) and (0) isoforms were significantly larger than those of mock-transfected cells at positive (P < 0.01) and negative (P < 0.05) membrane potentials). (Inset) Current-voltage relationship for a representative cell transfected with the (abcd) isoform before (no asterisk) and after (asterisk) replacement of extracellular Cl with gluconate. (C) Membrane currents from a cell transfected with TMEM16A(abcd) carrying the R563A mutation. The arrow shows tail currents with a slower decay as compared with the wild-type protein in (A). (D) Time constant values (τ) (mean ± SEM, n = 3 experiments per condition) determined by fitting tail currents of the wild-type protein and the R563A mutant with a single exponential function. Tail currents were measured at –60 mV after stepping the membrane to the indicated voltages. * P < 0.05; ** P < 0.01. (E) Membrane currents from a cell transfected with the Q757A mutant. (F) Current-voltage relationship from experiments with the Q757A mutant before and after replacement of extracellular Cl with gluconate (mean ± SEM, n = 6 experiments per condition).

To further validate the relationship of TMEM16A proteins with CaCC currents, we introduced mutations in highly conserved amino acids with predicted localization in transmembrane segments of the (abcd) isoform (fig. S6). We transiently transfected the resulting plasmids in HEK-293 cells and studied whole-cell membrane currents with the patch-clamp technique. Transfection of Lys349 → Ala349 (K349A) (21), K631A, and T830A generated currents that were not markedly different from those of wild-type protein, whereas K636A and R912A elicited very small membrane currents. R563A was associated with CaCC currents with an altered kinetic behavior: The decay of tail currents after returning to the holding potential from positive test pulses was substantially slower (Fig. 3C). Fitting the current decay with a single exponential function gave time constants consistently higher for R563A relative to wild-type protein (Fig. 3D). Q757A caused a marked alteration in voltage dependence. Transfected cells showed membrane currents larger than those in control cells but without time-dependent activation at positive voltages (Fig. 3E). Furthermore, the currents elicited by Q757A expression showed a reduced Cl selectivity because extracellular Cl replacement with gluconate shifted the reversal potential by only 23.3 ± 1.5 mV (Fig. 3F).

We have obtained evidence that TMEM16A is a membrane protein involved in Ca2+-dependent Cl transport. In particular, silencing by siRNA or up-regulation by cDNA transfection leads to the decrease or increase, respectively, of membrane currents whose biophysical properties reproduce those of classical CaCCs found in various cell types. Furthermore, specific mutagenesis of highly conserved amino acids changed intrinsic properties of the channel. TMEM16A belongs to a large family that includes other membrane proteins that may also have an ion channel function. Family members with strong sequence identity to TMEM16A, such as TMEM16B, are also possible CaCCs. Other family members, such as TMEM16F, TMEM16J, or TMEM16K, may represent other types of Cl channels with different biophysical properties and mechanisms of regulation.

Supporting Online Material

Materials and Methods

Figs. S1 to S7


References and Notes

Stay Connected to Science

Navigate This Article