Anoctamin 9/TMEM16J is a Cation Channel Activated by cAMP/PKA Signal

Anoctamins are membrane proteins that consist of 10 homologs. ANO1 and ANO2 are anion channels activated by intracellular calcium that meditate numerous physiological functions. ANO6 is a scramblase that redistributes phospholipids across the cell membrane. However, the others are not well characterized. We found ANO9/TMEM16J is a cation channel activated by a cAMP-dependent PKA. Intracellular cAMP activated robust currents in whole-cells expressing ANO9 and inhibited by PKA blockers. A cholera toxin and purified PKA also activated ANO9. The cAMP-induced ANO9 currents were permeable to cations. The mutation of a possible phosphorylation site at Ser245 elicited a block of the cAMP-dependent activation. High levels of Ano9 transcripts were found in intestines. Human intestinal SW480 cells showed cAMP-dependent currents. We conclude that ANO9 is a cation channel activated by the cAMP/PKA pathway and could play a role in intestine function.


INTRODUCTION
The Anoctamin/TMEM16 family consists of transmembrane proteins in 10 isoforms, ranging from ANO1/TMEM16A to ANO10/TMEM16K. Anoctamins are expressed in numerous major tissues and are thought to mediate various physiological functions. The best known anoctamin gene is Ano1, which is a Cl − channel activated by Ca 2+ (Caputo et al, 2008;Schroeder et al, 2008;Yang et al, 2008). ANO1 is known to mediate transepithelial fluid movements such as salivation in the salivary glands, mucin secretion in the airway and Cl − and fluid secretion in the intestine (Huang et al, 2012a;Jang & Oh, 2014;Namkung et al, 2011;Ousingsawat et al, 2009;Romanenko et al, 2010). ANO1 is also highly expressed in small-diameter dorsal-root ganglion neurons, implicating its role in nociception as a heat sensor (Cho et al, 2012). In addition, ANO1 plays an important role in controlling smooth muscle contraction (Bulley et al, 2012) and pacemaking activity in the intestine (Cobine et al, 2017;). More importantly, ANO1 has been implicated in tumorigenesis (Britschgi et al, 2013;Duvvuri et al, 2012;Jia et al, 2015) and benign prostate hyperplasia (Cha et al, 2015).
Unlike ANO1 and ANO2, ANO6/TMEM16F has dual functions. ANO6 is a small conductance calcium activated cation channel (SCAM) that is permeable to divalent ions . Strikingly, ANO6 is known to be a scramblase that disrupts polarized phospholipids in the plasma membrane. The polarized phospholipids attract immunological signals necessary for activation of T lymphocytes (Hu et al, 2016;Xu et al, 2008;Zhang et al, 2011). Mutations in Ano6 cause a rare bleeding disease, Scott syndrome (Castoldi et al, 2011;Suzuki et al, 2010). Thus, ANO6 is both a channel and an enzyme. Some scramblase activity has also been observed in ANO4, ANO8 and ANO9 .
ANO5 is involved in skeletomuscular function; its mutations cause gnathodiaphyseal dysplasia, an autosomal dominant inherited bone disorder (Tran et al, 2014). However, its biophysical properties and mechanism of activation are unknown.
Although the functional roles of some genes in the Anoctamin family are well studied, the role of ANO9/TMEM16J remains poorly understood (Picollo et al, 2015).
ANO9 is found in the human nasal and colonic epithelium and expressed in the respiratory, digestive, skeletal and integumentary systems during development (Rock & Harfe, 2008). ANO9 is also implicated in the metastasis of colorectal cancer (Li et al, 2015). Despite its expression pattern and a possible role in colorectal tumorigenesis, the function and activation mechanism of ANO9 is not known. The present study aimed to determine if ANO9 is a channel and, if so, how it is activated. Surprisingly, we found that ANO9 is a cation channel activated by the cAMP/PKA pathway.

cAMP activates ANO9
To determine if ANO9 is a channel, we expressed mouse ANO9 tagged on the C terminus with enhanced green fluorescence protein (eGFP) for visual identification in HEK 293T cells. The eGFP was largely localized in the plasma membrane (Fig. 1A).
Upon whole-cell formation in the HEK 293T cells transfected with Ano9-eGFP (ANO9/HEK cells), robust inward currents (65.9 ± 6.99 pA/pF, n = 36) were observed with Ehold = −60 mV when the pipette solution contained 100 M cAMP. The bath and pipette solutions contained 140 mM NaCl and 140 mM KCl, respectively. These currents were not observed in whole cells without cAMP in the pipette. In addition, the cAMP-evoked currents were completely blocked by the treatment of the cells with PKA blockers (20 M H-7 or 20 M H-89; Fig. 1B & 1C). cGMP in the pipettes (100 M) failed to evoke currents in ANO9/HEK cells (Fig. 1D). The cAMP-evoked currents in the ANO9/HEK cells began to develop 5 s to a few minutes after the formation of whole-cells. This time lag suggests a possible indirect signaling pathway for the activation by cAMP.
Cholera toxin is an oligomeric protein complex secreted by the bacterium Vibrio cholerae that leads to the direct activation of adenylyl cyclase, resulting in the overproduction of cAMP (Gabriel et al, 1994). If ANO9 is activated by cAMP and its downstream signal, then the application of cholera toxin should activate ANO9. Indeed, the application of cholera toxin to the bath solution evoked robust currents with variable time lags in ANO9/HEK cells. However, the cholera toxin-induced currents were not observed in the mock-transfected HEK 293T cells (Fig. 1E & 1F).

PKA activates ANO9
We then tested whether cAMP alone or with its downstream signal, PKA, activated ANO9. To do this, membrane patches were isolated from ANO9/HEK cells with an inside-out patch configuration. Recordings were made in a pipette solution containing 140 mM NaCl and 2 mM CaCl2 and a bath solution containing 140 mM KCl at Ehold = −60 mV. When solutions containing 1 mM cAMP, 2 mM ATP or cAMP+ATP were applied to the inside-out patches, no appreciable currents were observed ( Fig. 2A).
However, when purified PKA (10 g/reaction, 0.7 units/g) along with ATP and cAMP was applied to isolated membrane patches of ANO9/HEK cells, large macroscopic currents were observed with a prolonged time lag (Fig. 2A). These results now suggest that ANO9 is activated by PKA activity.

ANO9 is a cation channel
The cAMP-evoked currents in ANO9/HEK cells were cationic because they were to measure the reversal potential. Voltage ramps from −100 mV to +100 mV in 100 ms durations were applied. When the bath KCl concentration, initially 140 mM, was changed to 210 mM and 70 mM, the reversal potentials were changed to +6.29 ± 0.82 mV and −11.46 ± 1.09 mV (n = 8), respectively (Fig. 3C). The reversal potentials were plotted as a function of extracellular KCl concentration on a semi-logarithmic scale. A line was fitted to +37.2 mV/decade, which is relatively close to +58 mV/decade of the Nernst equation when the major carrier charge is a cation. The relative permeability ratio of K + and Cl -(PCl/PK) calculated by Goldman-Hodgkin-Katz equation (Hille, 2001) is 0.237 ± 0.034, suggesting a weak permeability to Cl -. Thus, these results clearly suggest that ANO9 exhibits a large preference for cations.
Because both ANO1 and ANO2 are known to be voltage-activated (Cenedese et al, 2012;Xiao et al, 2011), we investigated if ANO9 was also activated by voltage alone.
We increased the voltage in 20 mV steps from −120 mV to 120 mV and applied it to the whole cells of mock-and ANO9-transfected HEK cells. The I-V curves of ANO9transfected cells were outwardly rectifying (Fig. 3G). However, there was no difference between the amplitudes and shapes of the I-V curves of mock-and ANO9-transfected cells, suggesting that voltage alone does not activate ANO9.

Possible PKA phosphorylation site
Because ANO9 was activated by PKA, we identified its phosphorylation site. Five putative PKA phosphorylation consensus sites (R-R/K-X-S/T, R-X-X-S/T, and R-X-S/T) 8 were predicted by a computer program, GPS 2.0 (Xue et al, 2008). These putative sites were the S85, S120, S245, S321 and T412 residues in ANO9. Using mutagenesis, each Ser or Thr residue of the predicted phosphorylation sites was replaced with alanine. Each mutant of ANO9 tagged with eGFP was expressed in HEK293T cells and tested for activation by cAMP. Each ANO9 mutant except T412A was well expressed in the plasma membrane of HEK293T cells ( Supplementary Fig. 1). Robust cation currents were observed in HEK293T cells transfected with S85A, S120A and Oddly, the Ser245 residue is located in the extracellular loop of the predicted topology of ANO9 based on the modeling extracted from the crystal structure of nhTMEM16 (Brunner et al, 2014) (Fig. 4E). In this model, the N-and C-termini are present in the cytosolic side of the protein (Predicted model, Fig. 4E). If this is true, the Ser245 is hardly phosphorylated by PKA. However, the actual topology of ANO9 may differ from that predicted by the structure of nhTMEM16. If the Ser245 residue is present in the intracellular side, then the N-and C-termini may be present in the 9 extracellular side of the protein as shown in a revised model in Fig. 4E. To determine if the Ser245 residue is located on the intracellular side, we constructed fusion proteins in which the HA epitope tag was attached to the N-terminus or inserted next to the Ser245 of ANO9. These HA-tagged mutants of ANO9 were transfected to HEK cells.
The HA antibody was applied to the extracellular side of the transfected cells and was detected by a fluorescent-tagged secondary antibody to anti-rabbit IgG. As shown in Fig. 4F, the HA tag was detected by the HA antibody applied extracellularly to nonpermeabilized as well as permeabilized HEK cells that were transfected with HAtagged ANO9 in the N-terminus, suggesting that the actual N-terminus of ANO9 is present in the extracellular side. In addition, the HA-tag was detected only in permeabilized cells expressing the ANO9 mutant where the HA tag was inserted next to the Ser245 residue whereas it was not detected by the antibody in nonpermeabilized cells (Fig. 4F). These results suggest that the Ser245 residue is located inside the cell.

Calcium augments the cAMP-induced ANO9 currents
Because ANO1 and ANO2 are known to be activated by intracellular Ca 2+ , we also tested for the activation of ANO9 by Ca 2+ . When 1 M Ca 2+ was applied to the pipette, no appreciable whole-cell currents were activated. When 10 M Ca 2+ was applied, small currents were observed ( Fig. 5A & 5B). When 20 M cAMP was applied together with 10 M Ca 2+ , much larger currents than those activated by 10 M Ca 2+ alone were observed. These current responses to the co-application of Ca 2+ and cAMP appeared to be dose dependent because larger currents were observed when 100 and 200 M cAMP were applied along with 10M Ca 2+ (Fig. 5A & 5B). Although the intracellular Ca 2+ in a physiological concentration rarely activated ANO9, Ca 2+ can augment the ANO9 response to cAMP.

Intracellular sodium inhibits ANO9
Unexpectedly, we also found that a high intracellular concentration of Na + inhibits the activity of ANO9. Intracellular cAMP vigorously activated ANO9 when 0 mM Na + was added to the pipette (intracellular) solution. However, when 50 mM NaCl was added to the pipette solution, the cAMP-evoked whole-cell currents were markedly reduced ( Fig. 6A & 6B). In contrast, when 50 mM CsCl was added to the pipette solution, the cAMP-evoked currents were comparable to those of the pipette solution containing 150 mM KCl (Fig. 6A & 6B). These results suggest that high intracellular concentration of Na + inhibited the cAMP-dependent ANO9 activity.

Ano9 is rich in intestines
Tissue distribution of Ano9 was determined. Ano9 mRNAs were mainly observed in the digestive system. The small intestine, colon and stomach expressed large number of copies of Ano9 mRNA. A small amount of Ano9 mRNA was also detected in the tongue, kidney, eye, lung, and bladder (Fig. 7A).
Consistent with high expression in the colon, ANO9 is known to be expressed in colorectal cancer cells (SW480) (Li et al, 2015). Indeed, the ANO9 specific immunofluorescence was present in the plasma membrane of SW480 cells when probed with ANO9 antibody (Fig. 7B). We therefore determined if cAMP activated the cation currents in these cells. Shown in Fig. 6C, the intracellular application of 100 M cAMP resulted in large inward currents in 15 out of 20 SW480 cells (Fig. 7C). The cAMP-evoked currents were blocked by co-challenge with 20 M H-89 in the pipette solution. In addition, a knock-down of Ano9 in the SW480 cells after transfection with siRNA of Ano9 markedly reduced the cAMP-evoked currents (Fig. 7C & 7D). These results demonstrate that the native ANO9 in colorectal cells is activated by intracellular cAMP.

Diverse functions of the anoctamin family
The anoctamins have 10 homologs with diverse cellular and physiological functions. ANO1 was first discovered as a calcium-activated chloride channel (Caputo et al, 2008;Yang et al, 2008) and is known to be involved in many physiological functions such as fluid secretion, smooth muscle contraction, nociception, tumorigenesis and cell proliferation (Cho et al, 2012;Duvvuri et al, 2012;Huang et al, 2012a;Jang & Oh, 2014;Katoh & Katoh, 2003;Namkung et al, 2011;Ousingsawat et al, 2009;Romanenko et al, 2010). ANO2 is also known to be a calcium-activated chloride channel with olfaction and learning and memory functions (Billig et al, 2011;Huang et al, 2012b;Neureither et al, 2017;Pietra & Dibattista, 2016;Stephan et al, 2009). ANO3 is known to increase modulation of the activity of the Na + -activated K + channel in dorsal-root ganglion neurons, and it also modulates nociception (Huang et al, 2013). ANO5, expressed in muscles and bones, has been implicated in skeletomuscular functions because its mutations cause a skeletomuscular disease, gnathodiaphyseal dysplasia (Tran et al, 2014). Despite their physiological implications, however, the activation mechanisms and other cellular functions of ANO3 and ANO5 are not well characterized. ANO6 was initially characterized as a scramblase that disrupted polarized phospholipids in the plasma membrane (Castoldi et al, 2011;Suzuki et al, 2010). Later, ANO6 was found to be a cation channel activated by Ca 2+ . Despite the diverse functions of anoctamins, the activation mechanisms of those other than ANO1, ANO2 and ANO6 are not well understood.
Unlike ANO1 and ANO2, ANO9 was permeated mainly by cations. The ANO9 currents were not blocked by the ANO1 blockers such as MONNA, NPPB, or tannic acid ( Fig. 1B & 1C). More importantly, ANO9 was activated by the cAMP/PKA pathway but not by Ca 2+ or voltage. However, the activation by PKA was augmented by intracellular Ca 2+ . Although ANO9 is known to be expressed in the intestines, its physiological function is largely unknown.

Role in colorectal cancer cells
ANO9 has been implicated in intestinal functions; notably, its transcription levels were much higher in intestines compared to other tissues ( Fig. 7A; (Schreiber et al, 2015;Schreiber et al, 2010)). It is expressed in the colorectal cancer cell line, SW480 cells. We observed cAMP-activated and H-89-reversible currents in the SW480 cells ( Fig. 7), and ANO9 has been linked to colorectal cancer (Li et al, 2015). The activity of ANO9 was negatively associated with tumorigenesis in the colon. The expression of ANO9 was higher in non-tumor tissue than in tumorous tissues. ANO9 expression was lower in the recurrent colorectal cancer cells than in non-recurrent colorectal cancer cells. ANO9 overexpression is known to reduce the invasion of cancer cells (Li et al, 13 2015). Lower levels of ANO9 expression have been associated with poorer prognoses in patients with higher expression levels (Li et al, 2015). Thus, ANO9 appears to play a tumor suppressor role in the intestines, but its precise pathophysiological role needs to be determined.

Calcium dependence
Ca 2+ is indispensable for anoctamin functions, and it activates ANO1 and ANO2 (Caputo et al, 2008;Schroeder et al, 2008;Stephan et al, 2009;Yang et al, 2008). Ca 2+ is also required for the scramblase activity of ANO6 and fungal TMEM16 (Malvezzi et al, 2013;Suzuki et al, 2010;Yang et al, 2012). However, Ca 2+ was not found to be essential for the activation of ANO9 because physiological concentrations of Ca 2+ rarely gated ANO9. In addition, cAMP activated ANO9 in Ca 2+ -free conditions, but Ca 2+ augmented the cAMP-induced ANO9 currents. Thus, the role of Ca 2+ in activating ANO9 cannot be totally ignored because Ca 2+ affected the ANO9 activity.
Recently, the X-ray crystal structure of nhTMEM16 was determined (Brunner et al, 2014). The Glu residues of ANO1 essential for its activation by Ca 2+ were located in the subunit cavity of ANO1 in the hydrophobic core of the membrane. The subunit cavity was lined by three glutamates, two aspartates and an asparagine residue in the -helices 6, 7, and 8 of nhTMEM16 and ANO1. These helices are known to be crucial for the channel activation of ANO1 and its scramblase activity (Brunner et al, 2014).
ANO9 has a similar amino-acid sequence in this region ( Supplementary Fig. 2): ANO9 has three glutamates, one aspartate and one serine residue in the -helices 7 and 8.
This structural constraint of ANO9 compared to ANO1 predicts the synergistic but not essential role of Ca 2+ in activating ANO9. Voltage has been found to be another 14 endogenous activator of ANO1 and ANO2, but, analogous to the role of Ca 2+ , ANO9 was not activated by voltage alone (Fig. 3F & 3G). We found that ANO9 added more to the diversity of the functions of the anoctamin family.

PKA activation
Many of the functions of the cystic fibrosis transmembrane conductance regulator (CFTR), such as the transepithelial Cl − transport in various epithelia, overlap with those of ANO1 mainly because they are both anion channels. Because of this functional overlap, ANO1 has been considered to rescue the defective CFTR functions in cystic fibrosis (Becq et al, 2011;Sondo et al, 2014). CFTR is activated by ATP and PKA and has multiple transmembrane-spanning domains in the cell membrane and a regulatory domain and two nucleotide-binding domains in the cytosolic side (Hwang & Sheppard, 2009). The binding of ATP to the nucleotide-binding domains is known to activate the CFTR while the phosphorylation of the regulatory domain by PKA is a prerequisite for the activation. Unlike the CFTR, ANO9 does not contain nucleotide-binding domains, and therefore ANO9 gating does not require ATP because it is not activated by ATP alone (Fig. 2B). However, whether the phosphorylation of ANO9 at Ser245 by PKA is indispensable or is a prerequisite for its activation remains unclear. ANO9 did not have a bulky regulatory domain with multiple phosphorylation sites as shown in the CFTR, which is the structural basis for the prerequisite requirement for activation. Because only the direct application of PKA along with cAMP and ATP activates ANO9, it is highly likely that ANO9 opens only when it is phosphorylated at Ser245 by PKA. One limitation of this hypothesis is the location of the Ser245 residue. When the location was predicted by the crystal structure of the homologous nhTMEM16 in the fungus 15 Nectria haematococca, the Ser245 residue was located in the first extracellular loop.
This location is difficult to reconcile as the PKA phosphorylation site of ANO9. However, the crystal structure was obtained with membrane-embedded crystals of nhTMEM16, which have only scramblase activity and not channel activity. Mouse ANO9 also has only 21% sequence homology with nhTMEM16. Therefore, the predicted location of the Ser245 residue in the homology modeling with nhTMEM16 might not represent the actual location in the mouse ANO9. To identify if Ser245 region is located in the intracellular region, we cloned an HA-tagged ANO9. If the N-terminal of ANO9 is in the extracellular region, then the Ser245 will be located in the intracellular loop. In live cell imaging, the HA epitope tagged N-terminal was detected by an HA-antibody without detergent. These results provide strong evidence that the location of the N-terminal of ANO9 is extracellular.

Cloning of mouse Ano9 and mutagenesis
Primers were designed using the mouse cDNA sequences of Ano9 (TMEM16J) from the NCBI database (NM_178381.3). The cDNA encoding Ano9 has been isolated from the lung of adult C57BL/6J mice. The full-length coding sequence of Ano9 All mutants were generated from the wild-type mouse Ano9 construct tagged with eGFP (pEGFP-N1-mANO9). The amino acid substitution mutants were constructed using a site-directed mutagenesis kit. The construction of mutants was verified with DNA sequencing. To generate HA-tagged mAno9 mutants, haemagglutinin (HA) epitopes (YPYDVPDYA) were added to the N-terminus of mAno9 using the HA-pcDNA3.1 vector.
The HA epitopes next to the Ser245 residue in mAno9 were introduced by sitedirected mutagenesis. The insertion was repeated four times with each set of primers.

Cell culture and transfection of Ano9
Cell culture and functional expression of ANO9 or mutants were performed in HEK 293T cells. HEK 293T cells were maintained at 5% CO2, and incubated at 37 •

Whole-cell and single-channel current recordings
Whole-cell and single-channel current recordings were obtained using either a voltage-clamp technique with an Axopatch 200B amplifier (Molecular Devices) as descried previously (Cho et al, 2012;Lee et al, 2015). Briefly, whole-cell currents were measured after breaking the plasma membrane under the pipette tips. The resistance of the glass pipettes was about 3 mΩ. The junctional potentials were cancelled to zero.
For inside-out patch recording, after a giga-seal was formed, the glass pipette was pulled quickly from the cell to isolate a membrane patch. The output of the amplifier was fed to an analog/digital converter (Digidata 1440, Molecular Devices) and stored in a personal computer. The pClamp 10 software was used for I-V curve and other biophysical analysis.

Immunofluorescent staining
SW480 cells were fixed on glass coverslips in 4% paraformaldehyde for 10 min at room temperature, permeabilized with 0.2% Triton X-100, and incubated with anti- for 10 min. Then, the cells were incubated with the anti-HA antibody (1:100) for 1 hr, washed three times, and then incubated with the Alexa Fluor 488 goat-anti-rabbit IgG for 30 min at room temperature. After washing, the coverslips were mounted and imaged using a confocal microscopy.

Small interfering RNA (siRNA) treatment
Two nucleotide oligomer siRNAs targeting mouse ANO9 or scrambled siRNA labelled with Cy3 were provided by Bioneer (Seoul, Korea) transfected into cells using

ANO9-HEK cells or HEK293T cells transfected with S245A
Ano9 mutant were loaded with Fluo3-AM (Invitrogen) containing 0.1% . After loading with Fluo3-AM for 40 min, db-cAMP alone or with H89 was applied to the cells.
The fluorescence intensities of cells were measured at 488 nm in every 5 s with a confocal microscope (LSM700, Zeiss).

Real-time quantitative PCR
Major mouse organs were isolated from three 7-week-old mice. The total RNAs from each organ were purified with ethanol precipitation or the Easy Spin TM Total RNA Extraction Kit (iNtROn Biotech) according to the manufacturer's protocol. The first strand cDNAs were reverse transcribed from the total RNA using the Transcriptor First Strand cDNA Synthesis Kit (Roche). To measure the gene expression level, we performed real-time quantitative PCR (qPCR) in the LightCycler 2.0 system (Roche) using an ANO9 specific universal probe (forward primer: cggactctcctcatgaatcc; reverse primer: tgttcacgacaaagctcaca). The absolute copy numbers per 250 ng of total RNA 21 were calculated using the absolute quantification with the external standard method in the Light Cycler software 4.0.

Statistical analysis
All results are expressed as means ± standard errors (S.E.). The statistical significances of the differences were determined by a one-way analysis of variance (ANOVA) followed by the Tukey's post-hoc test for multiple comparisons. Statistical significance was accepted at p values of less than 0.05.   (D) and the permeability ratios (PX/PK) (E) of ANO9 currents. Whole cells were formed with the pipette contained 140 mM KCl. Current-voltage relationship were obtained after the bath solution was changed to 70 mM CaCl2, 140 mM NaCl, 70 mM CsCl, and 140 mM LiCl. The permeability ratios, PX/PK was calculated using the Goldman-Hodgkin-Katz equation. (F-G) Whole-cell currents (F) and the I-V curves of ANO9-HEK cells elicited by voltage steps. Voltage steps from -120 to +120 mV in 20 mV increment were applied to whole cells of HEK cells transfected with vector alone (mock) or Ano9 (ANO9-HEK).