Mechanisms of CFTR regulation by syntaxin 1A and PKA

It is not surprising that many layers of regulation exist for ubiquitous proteins such as ion channels as they play central roles in inter- and intracellular signaling in fluid and ion transport and in the development of memory. Ion channels and transporters are regulated by various mechanisms,including membrane potential, ligands, phosphorylation state, other ion channels, intracellular calcium concentration and second messengers(Hille, 1992). Syntaxin 1A is a target membrane t-SNARE that is classically thought to mediate vesicle trafficking along with other SNARE complex proteins including SNAP-25 and synaptobrevin (Wu et al.,1999). The first functional association observed for syntaxin 1A and ion channels/transporters was found in the brain where membrane-anchored syntaxin 1A modulates N-type calcium channel activity(Bennett et al., 1992). Regulation of both ion channels and transporters by syntaxin 1A has since been observed in a number of systems including CFTR, the amiloride-sensitive epithelial sodium channel, the GABA transporter and potassium channels(Bezprozvanny et al., 1995;Bezprozvanny et al., 2000;Deken et al., 2000;Fili et al., 2001;Naren et al., 2000;Naren et al., 1997;Naren et al., 1998;Peters et al., 1999;Qi et al., 1999;Rettig et al., 1996;Saxena et al., 1999;Sheng et al., 1996;Sheng et al., 1994;Yang et al., 1999). Despite the observations that syntaxin 1A functionally interacts with multiple ion channels, the mechanism of channel regulation by this t-SNARE is not known and may include indirect regulation by modulation of membrane trafficking or direct protein-protein interactions.

The problem of channel regulation has increased complexity in the case of CFTR, where cAMP not only activates the channel through protein-kinase-A-dependent phosphorylation but in some investigations appears to mobilize heterologously expressed CFTR into the plasma membranes of Xenopus oocytes (Takahashi et al., 1996). The question of whether an enhancement of membrane trafficking accompanies CFTR activation, independent of syntaxin modulation,remains controversial. For example, CFTR activation in Calu-3 cells (a human lung adenocarcinoma-derived line) as well as Madin Darby canine kidney (MDCK)cells is not accompanied by changes in membrane insertion and retrieval(Chen et al., 2001;Loffing et al., 1998;Moyer et al., 1998). However,a number of other investigations carried out in a variety of cell types demonstrated that cAMP facilitates translocation of CFTR from an intracellular compartment to the plasma membrane(Bradbury et al., 1992;Howard et al., 2000;Peters et al., 1999;Schwiebert et al., 1994;Takahashi et al., 1996).

Experiments carried out in this study sought to resolve two related issues:(a) is PKA activation or SNARE protein modulation of CFTR linked to changes in membrane turnover and (b) does syntaxin 1A modulate CFTR via direct effects on the gating of channels residing in the plasma membrane versus alterations in membrane traffic. Using a combination of membrane capacitance measurements coupled with FM1-43 fluorescence we demonstrate that cAMP-dependent changes in membrane turnover that accompany CFTR activation are cell specific and can be revealed in some cell types only when dynamin-dependent endocytosis is inhibited. In addition, cAMP-dependent changes in membrane turnover are independent of the disruption of syntaxin 1A-CFTR interactions.

We have previously shown that syntaxin 1A and CFTR physically interact and that this interaction results in a negative modulation of cAMP-mediated CFTR function (Naren et al., 1997;Naren et al., 1998;Naren et al., 1999;Naren et al., 2000). Furthermore, three different reagents rescue CFTR from inhibition by plasma-membrane-anchored syntaxin 1A, namely, botulinum toxin C1, the syntaxin-binding protein Munc 18 and the cytosolic domain of syntaxin 1A(Naren et al., 1997). The interaction between CFTR and membrane-anchored syntaxin 1A is dynamic(reversible) and is dependent upon an intact actin cytoskeleton. Data in this investigation demonstrate that while changes in membrane turnover do not appear to be dependent upon the presence or absence of a direct interaction between membrane-anchored syntaxin 1A and CFTR, the interaction does result in a decrease in single channel open state probability. Disruption of the t-SNARE-channel interaction by the cytosolic domain of syntaxin 1A results in a significant increase in CFTR open-state probability in excised patches. These results indicate that binding of full length syntaxin 1A to CFTR inhibits PKA-dependent activation through direct protein-protein interaction. We speculate that second messenger pathways that regulate the physical interaction of these two proteins via alterations in the actin cytoskeleton or CFTR-syntaxin-Munc-18 interactions will fine tune channel activity through subtle changes in membrane-resident channel gating and not alterations in membrane turnover.

Proteins, immunoprecipitated with 1 mg each of CFTR (Genzyme C-terminal monoclonal), non-immune serum (normal Mouse, Santa Cruz Biotech) or syntaxin 1A IgG immobilized on Protein A/G beads(Naren et al., 2000), were probed using an affinity-purified polyclonal antibody raised against nucleotide binding domain 1 (residues 430-580) of CFTR. Details have been published previously (Naren et al.,1997).

Whole-cell recordings were obtained from 16HBE14o- human airway epithelial and from HT29-CL19A human colonic epithelial cells using methods described previously (Hamill et al.,1981) and solutions published previously(Naren et al., 1997). The cAMP cocktail used to elicit CFTR activation in the whole-cell experiments contained 1 mM IBMX, 10 μM Forskolin and 400 μM cpt-cAMP. Peptides were added to the pipette solution at their specified concentrations. Single-channel recordings were obtained from 16HBE14o- cells using the inside-out configuration. The extracellular (pipette) solution contained 140 mM NMDG-Cl, 2 mM MgCl₂, 10 mM HEPES and 200 μM DIDS (to pharmacologically block non-CFTR anion channels) titrated to a pH of 7.4 with NMDG. The intracellular (bath) solution contained 140 mM NMDG-Cl, 10 mM EGTA,2 mM MgCl₂, 8 mM Tris and 5 mM Mg-ATP adjusted to a pH of 7.4. Once the inside-out configuration was achieved, 75 U/ml of PKA catalytic subunit(Promega, Madison, WI) was added to activate CFTR. Subsequently, either 350 nM GST-Syn1AΔC or 350 nM GST-Syn1AΔH3 were added to the bath. In this manner, the same inside-out patch served as its own internal control. Single-channel currents were filtered at 1 kHz and sampled at 2 kHz. All experiments were conducted at room temperature (22-24°C) using an EPC-9 patch clamp amplifier (HEKA Electronik GmbH, Lambrecht, Germany) and using the Pulse V 8.31 acquisition program (HEKA Electronik GmbH, Lambrecht, Germany). Data analysis was performed using Tac V4.1.1 (Bruxton Corp., Seattle, WA,USA). Statistical significance of results was determined using the Student's t-test.

The EPC-9 includes a built-in data acquisition interface (ITC-16,Instrutech, Port Washington, NY, USA). The software package controlled the stimulus and data acquisition for the software lock-in amplifier in the `sine+ dc' mode as described (Gillis,2000). The temporal resolution of the capacitance data was 40 mseconds per point using a 1 kHz, 20 mV sine wave. The holding potential in the capacitance experiments was -10 mV. The fluorescence intensity of FM1-43 excited at 470 nm was simultaneously measured with a photomultiplier system as described (Katnik and Nelson,1993) using a Leitz inverted DM-IRB microscope. The photo-multiplier output was collected at 510 nm and averaged online. These data were stored together with the C_m and G_mmeasurements. Pipette and bath solutions were as described above. Peptides were added to the pipette solutions as in whole-cell experiments. All experiments were conducted at room temperature (22-24°C). Statistical significance of results was determined using the Student's t-test.

To determine if syntaxin 1A associates with CFTR in human airway epithelial cells as we have shown for canine tracheal epithelial cells(Naren et al., 2000), we performed a series of coimmunoprecipitation and whole-cell patch clamp studies on 16HBE14o- human airway epithelial cells. The mature form of CFTR (band `C')could be coimmunoprecipitated from 16HBE14o- cells with a syntaxin 1A monoclonal antibody (Fig. 1A). Although it has been shown that CFTR binds to the H3 domain or SNARE-motif of syntaxin 1A (Jahn and Sudhof,1999), it has also been shown that inhibition of I_CFTRis dependent on the presence of the C-terminal transmembrane domain (TMD) of syntaxin 1A (Naren et al.,1998). Both soluble syntaxin 1A lacking the C-terminal TMD(GST-Syn1AΔC) and the syntaxin-1A-binding protein Munc 18 disrupt the regulatory interaction between CFTR and the membrane-anchored t-SNARE resulting in a marked enhancement of I_CFTR(Naren et al., 1997;Naren et al., 1998). This effect was also observed in human airway epithelial cells(Fig. 1B-D). Whole-cell I_CFTR was elicited by bath application of a cAMP cocktail(Fig. 1B). When either 350 nM GST-Syn1AΔC or 300 nM GST-Munc 18 were included in the standard pipette solution in order to disrupt the syntaxin 1A-CFTR interaction,I_CFTR increased approximately threefold from approximately 30 to 90 pA/pF (Fig. 1D). Neither reagent affected basal currents in the absence of cAMP cocktail(Fig. 1D). To examine the specificity of this effect, we substituted a syntaxin 1A fusion protein lacking the H3 domain (GST-Syn1AΔH3) for GST-Syn1AΔC. As the H3 domain is necessary for syntaxin 1A binding to CFTR(Naren et al., 1998),GST-Syn1AΔH3 should not disrupt the syntaxin 1A-CFTR interaction and as expected, this construct did not rescue CFTR from membrane-anchored syntaxin 1A inhibition (data summarized in Fig. 1D). Characteristic of CFTR, currents under all conditions were insensitive to 500 μM 4,4′-diisothiocyanatostilbene-2,2′-disulphonic acid (DIDS) but were inhibited by 1 mM diphenylamine-2-carboxylic acid (DPC)(Fig. 1C). Macroscopic current kinetics and current-voltage (IV) relationships remained unchanged by the disruption of the syntaxin 1A-CFTR interaction by either GST-Syn1AΔC(Fig. 1B,C) or GST-Munc 18.

Rescue of CFTR from inhibition by membrane-anchored syntaxin 1A by the cytosolic domain of syntaxin 1A was dependent on the actin cytoskeleton. If the 16HBE14o- cells were treated with cytochalasin D (0.5 μg/ml), which disrupts the actin cytoskeleton, cAMP-dependent current activation was not significantly different from that observed under control conditions(Fig. 2). However, the syntaxin 1A cytosolic domain failed to rescue CFTR from membrane-anchored syntaxin 1A inhibition in the cytochalasin-treated cells. Thus, the presence of an intact cytoskeleton appears to maintain the reversibility of the dynamic interaction between CFTR and membrane-anchored syntaxin 1A. This result suggested to us that CFTR current augmentation could involve a trafficking event mediated via cytoskeletal interactions.

Reasoning that cytoskeleton-dependent vesicle insertion or inhibition of membrane endocytosis could account for the soluble syntaxin 1A rescue of CFTR inhibition by membrane anchored syntaxin 1A, we measured membrane turnover,I_CFTR activation and the regulation of these parameters by the cytoplasmic domain of syntaxin 1A in the 16HBE14o- cells. Membrane capacitance(C_m) is proportional to cell surface area, and the measurement of changes in C_m can give insight into the change in surface area resulting from exocytosis and endocytosis(Neher, 1988;Neher and Marty, 1982). Time resolved changes in membrane capacitance provide a direct and quantitative record of the time course of plasma membrane turnover in response to stimuli. Conductance (G_m) was markedly augmented by dissociation of the syntaxin 1A-CFTR interaction by either GST-Syn1AΔC(Fig. 3A and B) or GST-Munc18(Fig. 3B). As expected,addition of 350 nM GST-Syn1AΔH3 to the pipette solution did not result in an increase in G_m. Under all conditions tested, the capacitance of the 16HBE14o- cells remained unchanged. Thus, neither cAMP activation alone nor cAMP activation in combination with GST-Syn1AΔC or GST-Munc18 had any detectable effect on C_m in human airway cells.

Studies of capacitance are limited by the fact that they measure net changes in cell surface area owing to simultaneous exocytosis and endocytosis. Examination of each process independently could lend valuable insight into the underlying membrane trafficking events. Exocytosis can be examined independently of endocytosis using the fluorescent styryl dye FM1-43, since it labels but does not permeate cell membranes. Fluorescence intensity of cells in the presence of FM1-43 remains unchanged during the process of endocytosis. As vesicles fuse with the plasma membrane, however, the newly exposed surface is labeled by the FM1-43 and fluorescence intensity increases. Thus,exocytosis can be measured without contamination from endocytotic processes(Smith and Betz, 1996). As summarized in Fig. 3B, we detected no changes in FM1-43 fluorescence upon activation of CFTR in the 16HBE14o- cells with or without GST-Syn1AΔC, GST-Munc18 or GST-Syn1AΔH3 (fluorescent intensity only increased 1.1 to 1.3% under the various conditions where n=4-10).

Dynamin mediates different types of endocytosis in various cell types(Hinshaw, 2000). Disruption of its function with anti-pan-dynamin IgG (α-Dyn) inhibits rapid endocytosis in chromaffin cells (Artalejo et al., 1995; Elhamdani et al.,2000). Therefore, we carried out a set of experiments to determine if we could detect effects of syntaxin-1A-CFTR disruptive reagents on C_m or G_m by blocking endocytosis with α-Dyn. Disruption of the syntaxin-1A-CFTR interaction with either GST-Syn1AΔC or GST-Munc18 did not change capacitance in 16HBE14o- human airway epithelial cells even in the presence of α-Dyn, suggesting again that membrane recycling is not involved in the augmentation of I_CFTR by the cytosolic domain of syntaxin 1A (Fig. 3). Capacitance changes were also not apparent in the presence of GST-Syn1AΔH3 plus α-Dyn. Thus, using two different assays, we could not detect an increase in exocytosis during the time course of the current augmentation response, whether or not the interaction between syntaxin 1A with CFTR was disrupted.

The possibility that SNARE-dependent membrane turnover contributes to the regulation of I_CFTR in other epithelial cell types lead us to examine the t-SNARE-channel interaction in a cell line of colonic origin. Support for these experiments came from reports that cAMP induces the insertion of heterologously expressed CFTR into the plasma membranes of Xenopus oocytes in the absence of syntaxin 1A(Peters et al., 1999). As it had been previously shown that cAMP stimulates apical protein secretion in CFTR-expressing HT29-CL19A human colonic epithelial cells(Jilling and Kirk, 1996), we assayed G_m, C_m and FM1-43 exocytosis in these cells in order to determine if membrane trafficking contributes to CFTR regulation by syntaxin 1A in cells of gastrointestinal origin. Results of these experiments are given in Fig. 4. We failed to observe an increase in C_m in response to cAMP in the presence or absence of GST-Syn1AΔC or GST-Munc18. However, there was a significant increase in the FM1-43 fluorescence following the addition of the cAMP cocktail, indicating that the simultaneity of the processes of exocytosis and endocytosis might result in an undetectable change in membrane capacitance. The cAMP-induced change in C_m in the HT29-CL19A cells was not detectable unless dynamin-dependent endocytosis was inhibited. The addition ofα-Dyn to the pipette in the presence of the cytosolic domain of syntaxin 1A resulted in a large increase in C_m, conductance and fluorescence. This increase in G_m was, however, not significantly different from that observed with GST-syn 1AΔC alone, indicating that inhibition of membrane retrieval did not contribute to G_maugmentation by either GST-Syn1AΔC or GST-Munc18(Fig. 4B). Moreover, the increase in G_m induced by cAMP in the presence of α-Dyn was essentially complete before the increase in C_m was detectable,indicating that inhibition of membrane turnover was not causally linked to the conductance increase as seen in Fig. 5. There were no measurable changes in either capacitance or conductance induced by α-Dyn alone in the absence of activation with cAMP, as summarized in Fig. 6.

In order to examine the hypothesis that syntaxin 1A regulates CFTR by direct protein-protein interactions rather than through a modulation of membrane trafficking, we performed single channel studies using excised inside-out patches obtained from 16HBE14o- cells the results of which are illustrated in Fig. 7. The extracellular (pipette) solution contained 200 μM DIDS to pharmacologically block non-CFTR anion channels. In these experiments, the reversal potential for Cl^- was zero, and Cl^- was the only permeant ion in both bath and pipette solutions (see Methods). Once the inside-out configuration was achieved, 75 U/ml of PKA catalytic subunit was added to the bath solution to activate CFTR. Patches were exposed to the catalytic subunit only if background Cl^- currents were not observed during a 300 second baseline recording period. A channel with the characteristics of CFTR was associated with the addition of the catalytic subunit to the bath as seen in Fig. 7A. Channel activity was observed following exposure of the excised patch to the catalytic subunit of PKA and was determined to be CFTR on the basis of the following criteria:(1) current activation was observed in the presence of 200 μM DIDS to the pipette (extracellular solution), (2) currents had a linear I-V with a single channel conductance of 6 pS as expected for CFTR and (3) currents were not observed in the absence of the catalytic subunit. Patches with an open-state probability of greater than approximately 0.10 following PKA activation were discarded. Low open-state probability patches were subsequently exposed to either 350 nM GST-Syn1AΔC or 350 nM GST-Syn1AΔH3 added to the bath. Selection of patches with initial low levels of current activation allowed for an accurate quantitation of CFTR rescue from inhibition by membrane-anchored sytaxin 1A via the soluble reagents. We observed increases in channel activity in 5 of the 9 patches exposed to GST-Syn1AΔC as seen in Fig. 7A. The apparent level of patch variability was probably due to cytoskeletal disruption during patch excision that prevented the rescue of CFTR from inhibition by the cytosolic domain of syntaxin 1A in the whole cell experiments(Fig. 2). However, even if the cytoskeleton were intact, such variability in response to the soluble syntaxin 1A peptide would not be unexpected as the syntaxin 1A-CFTR interaction is reversible and likely to be dynamic such that not all channels would be expected to be bound to syntaxin 1A at a given time.

Analysis of open times (τ_o) for a representative patch revealed no change prior to or after disruption of the syntaxin-1A-CFTR interaction (Fig. 7C). Closed time analysis revealed that disrupting syntaxin 1A from CFTR resulted in a decreased closed time (τ_c) consistent with an increase in the likelihood of opening events. Open probability (P_o) increased threefold (similar to the increase seen in whole-cell current measurements)after addition of GST-Syn1AΔC from 0.06±0.03 to 0.19±0.04(Fig. 7D). Channel activity,both prior to and after interruption of the syntaxin-1A-CFTR interaction, was stable for the duration of each 300 second segment of recording. The changes in P_o after addition of GST-Syn1AΔC were immediate. To examine the specificity of the response, we tested GST-Syn1AΔH3 for its effects on CFTR activity in excised membrane patches. P_o was unaffected by the soluble peptide lacking the CFTR binding domain as summarized in Fig. 7D.

Although many ion channels and transporters associate with and are regulated by syntaxin 1A, we demonstrate that this t-SNARE inhibits the PKA-dependent Cl^- channel, CFTR, by decreasing the likelihood of channel opening. Our data are most consistent with a model in which syntaxin 1A directly regulates membrane-resident CFTR (and probably other ion channels and transporters) rather than one requiring alterations in plasma membrane turnover. In whole-cell experiments, partial disruption of the actin cytoskeleton with cytochalasin D prevented the rescue of membrane-anchored syntaxin 1A-induced inhibition of CFTR by the cytosolic domain of syntaxin 1A. Thus, our observation that the increase in open-probability with single-channel recordings was not always observed with the soluble syntaxin was not unexpected as we believe that although syntaxin 1A interacts directly with CFTR, it does so in a dynamic manner dependent on an intact cytoskeleton that may or may not be disrupted by patch excision. It is possible that the loss of cytoskeletal structure might lock the syntaxin-1A-CFTR complex together irreversibly. We speculate that the subset of patches in which augmentation could be detected occurred when the residual cytoskeletal elements accompanied the excised membrane. Finally, current augmentation ranged from small to large effects (as reflected in the s.e.m. values) under whole-cell recording conditions, demonstrating the variability of this response even in non-excised states.

In agreement with studies in Calu-3 cells that are derived from human airway submucosal glands and express high levels of CFTR(Shen et al., 1994), we found that I_CFTR activation in human airway epithelia did not involve membrane trafficking (Chen et al.,2001; Loffing et al.,1998). Rather, the increase in C_m occurred independently of the increase in current. This is not necessarily incongruous with studies that demonstrate that exocytosis of CFTR-containing vesicles are involved in current activation (Lehrich et al., 1998; Takahashi et al.,1996). As has been suggested, the mechanism by which cAMP stimulates CFTR activation may be variously regulated in different cell lines and, furthermore, may be dependent on the state of tissue differentiation(Guggino, 1998;Moyer et al., 1998).

Endocytosis is the cellular process that serves the dual function of maintaining cell surface area constant while retrieving vesicular components for recycling. The severing activity that releases endocytotic vesicles from the plasma membrane is controlled by the dynamin family of proteins, which show high levels of GTPase activity. Dynamin is found in cells in both a soluble and membrane-associated form and is thought to oligomerize around the neck of retracting vesicular structures(Henley et al., 1999). Rapid endocytosis in neurons requires GTP hydrolysis, and the critical G protein regulating this process appears to be dynamin(Artalejo et al., 1995;Artalejo et al., 1997). The increase in exocytosis induced by cAMP and unmasked by α-Dyn that we observed for the HT-29 cells is normally balanced by a compensatory increase in endocytosis in a manner analogous to that in kidney epithelial cells expressing aquaporin 2 water channels(Katsura et al., 1995). Consistent with this interpretation, we were unable to observe an effect ofα-Dyn on C_m unless exocytosis was stimulated by cAMP(Fig. 4B). There were no measurable changes in either capacitance or conductance in the presence ofα-Dyn in the absence of activation with cAMP(Fig. 6). These results indicated that membrane-anchored syntaxin 1A inhibits CFTR via direct interactions rather than by modulation of membrane insertion in both human colonic and airway epithelial cells. In addition, cAMP activates two distinct processes in colonic epithelial cells: one involving activation of CFTR and the other involving membrane trafficking.

Understanding the complexities of CFTR regulation is an important prerequisite in the design of novel therapeutic strategies targeting cystic fibrosis and perhaps secretory diarrhea. We have previously shown that syntaxin 1A binds specifically and stoichiometrically to the N-terminal tail of CFTR (Naren et al., 1998). The N-terminal tail of CFTR modulates PKA-dependent channel gating, possibly by interacting with the cytoplasmic regulatory (R) domain(Naren et al., 1999). We hypothesize that membrane-bound syntaxin 1A sterically prevents the N-terminus from interacting with its R domain, thus resulting in a decrease in P_o. The present findings demonstrate that CFTR ion channel function can be stimulated directly by reagents that disrupt the association of syntaxin 1A with its N-terminal tail. While syntaxin 1A may also influence the intracellular traffic of CFTR in some systems (e.g. Xenopus oocytes)(Peters et al., 1999), this does not appear to be the case in epithelial cells. We show, in this study,that CFTR is modulated directly by SNARE interactions. A possible physiological role of such SNARE—ion-channel interactions may well link the activities of certain ion channels and transporters to membrane traffic events such as exocytosis in tissues for which such coordination is functionally advantageous. Given the growing body of evidence regarding direct SNARE-ion transport interactions, we speculate that SNAREs play roles both in modulation of ion channels and in trafficking of proteins and vesicles.

In summary, our data demonstrate that cAMP augments membrane turnover in a cell-specific manner. The cAMP-induced changes in membrane turnover, when observed, are not correlated with cAMP-induced changes in CFTR activation. In fact, the conductance change appears to be complete before the capacitance change is initiated. Finally, dissociation of syntaxin 1A from CFTR does not alter the cAMP-induced enhancement of membrane trafficking that we observed for the colonic secretory epithelial cells. Thus, in chloride transporting epithelia, there is a clear separation of the modulatory from the trafficking or anchoring functions of the syntaxins, and this is likely to be a generalized phenomenon for all membrane ion channels that interact with the ubiquitous SNARE proteins.

The authors thank Dieter Gruenert (University of Vermont) and Steven White(University of Chicago) for graciously supplying the 16HBE14o- cells and Harry Fozzard for critical review of the manuscript. This work was supported by NIH RO1 DK57882 (DJN) and a grant from the Cystic Fibrosis Foundation (DJN).