INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

REGULATION of the cystic fibrosis transmembrane conductance regulator (CFTR) Cl channel by protein kinases has been intensively studied during the past several years (4, 7, 12, 13, 16, 17, 35). Phosphorylation by protein kinase A (PKA) increases open probability (P_o) (35), bursting rate (20, 23), apparent ATP affinity (20, 23), and ATP hydrolysis rate (20). Phosphorylation by protein kinase C (PKC) is also required for CFTR to respond to PKA (17). These actions of kinases are antagonized by protein phosphatases, which probably vary among different cell types. CFTR channel activity declines rapidly when patches are excised from stimulated Chinese hamster ovary (CHO) or airway epithelial cells into bath solution lacking PKA [~10 s at 37°C (35); 100 s at 22°C (2)]. This spontaneous decline in channel activity, or rundown, is slower or absent when channels are studied in patches from guinea pig myocytes (15), transfected fibroblasts (4), or Hi-5 insect cells (39). Characterization and molecular identification of the phosphatases regulating CFTR has become a priority because they are potential therapeutic targets in the treatment of cystic fibrosis (2).

Serine and threonine phosphatases are functionally classified into two types, protein phosphatase 1 (PP1) and protein phosphatase 2 (PP2). The latter is subclassified into protein phosphatase 2A (PP2A), protein phosphatase 2B (PP2B), and protein phosphatase 2C (PP2C) (see Refs. 24, 34). PP1 preferentially dephosphorylates the -subunit of phosphorylase kinase and is sensitive to inhibitors 1 and 2, whereas PP2 preferentially acts on the -subunit of phosphorylase kinase and is insensitive to these inhibitors. PP2A activity does not require particular ions or cofactors, in contrast to PP2B, which requires Ca²⁺ and calmodulin, and PP2C, which requires relatively high levels of Mg²⁺ (EC₅₀ ~1.5 mM; Ref. 8). PP1 and PP2A are both sensitive to okadaic acid and calyculin A but can be distinguished by using appropriate concentrations of these inhibitors (24, 34). PP2B can be identified by its sensitivity to inhibitors such as deltamethyrin, cyclosporin, or FK-506 (10, 34). No specific inhibitors of PP2C are available. Rundown and CFTR dephosphorylation are both inhibited by phenylimidazothiazoles (2, 3), but at much higher concentrations than are needed to inhibit alkaline phosphatase.

CFTR channel rundown in excised patches is relatively insensitive to okadaic acid (2, 35), suggesting regulation by a robust membrane-associated protein phosphatase other than PP1 and PP2A. These results do not exclude regulation of CFTR by PP2A on the cell, however, since any cytosolic PP2A would be lost from patches after excision. Indeed, Reddy and Quinton (29) showed that okadaic acid (10⁸ M) inhibits deactivation of CFTR currents in permeabilized sweat ducts. In guinea pig cardiac myocytes, ~40% of the deactivation after forskolin washout was blockable by okadaic acid or microcystin (15). Exogenous PP1 and PP2B are not effective in regulating CFTR currents in excised patches from fibroblasts (4), although there is evidence that CFTR activity can be stimulated in these cells by the PP2B inhibitors cyclosporin A or deltamethyrin, suggesting that PP2B can regulate CFTR (11). By default, PP2C has been proposed as a CFTR phosphatase because it is insensitive to okadaic acid and microcystin (15, 35). PP2C dephosphorylates CFTR and deactivates macroscopic CFTR Cl currents (38). Deactivation by PP2C and other phosphatases has not been studied at the single-channel level.

The goal of this study was to characterize the effects of PP1, PP2A, PP2B, and PP2C on single CFTR channels exposed to comparable levels of phosphatase activity. For comparison, the kinetics of CFTR channels were also examined during deactivation by endogenous phosphatases. Finally, we investigated the possible role of PP2A in intact T84 epithelial monolayers by examining the ability of calyculin A to inhibit deactivation of short-circuit current (I_sc) after washout of forskolin. The results suggest that CFTR deactivation is mediated primarily by a PP2C-like phosphatase in CHO, baby hamster kidney (BHK), and T84 epithelial cells, although PP2A and PP2B both cause partial deactivation in vitro. These results have been reported in preliminary form (21, 22).

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Cell culture. BHK cells stably expressing wild-type CFTR were plated at low density on glass coverslips 3-5 days before use in patch-clamp experiments. The T84 line was obtained from American Type Culture Collection (Rockville, MD) and studied between passages 77 and 115. T84 cells were plated at a density of 400,000/cm² on porous supports (Millipore, Toronto, ON, Canada), which had been coated with a gel of type I collagen. The growth medium was a 50:50 mixture of DMEM and Ham's F-12 medium and was supplemented with 15 mM HEPES, fetal bovine serum (5%), penicillin (100 IU/ml), and streptomycin (100 µg/ml). Monolayers were studied 8-12 days after plating, when transepithelial resistance had reached ~1,500  · cm².

Phosphatases. Recombinant human PP1 catalytic subunit was purchased from Calbiochem (La Jolla, CA). PP2AI (smooth muscle phosphatase I) and PP2C (smooth muscle phosphatase II) were prepared from turkey gizzard smooth muscle as described previously (26, 27). PP2A was further purified by sequential chromatography on DEAE-Sephacel, -aminooctyl-Sepharose, and an affinity column of thiophosphorylated 20,000 M_r myosin light chains coupled to Sepharose 4B. The PP2C fraction from a Sephacryl S-300 column was chromatographed on DEAE-Sephacel and on the affinity column mentioned above. PP2C bound to the column in the presence of Mg²⁺ and was selectively eluted using EDTA. Bovine brain PP2B was purchased from Boehringer Mannheim (Laval, QC, Canada). Purity of the protein phosphatase preparations was assessed by SDS-PAGE (19) in a 12.5% Microslab gel and then stained with Coomassie blue. For comparison with the four protein phosphatases, some patch-clamp experiments were also carried out using bovine intestinal alkaline phosphatase type VII-S (Sigma, St. Louis, MO). The activity of this enzyme was 2,000-3,000 U/mg enzyme [where 1 unit hydrolyzes 1.0 µmol of p-nitrophenyl phosphate (PNP)/min at 37°C]. Alkaline phosphatase was used at a final concentration of 80 U/ml.

Patch-clamp studies. BHK cells were placed in a recording chamber (200 µl vol), containing (in mM) 150 NaCl, 2 MgCl₂, and 10 TES (pH 7.4). In most experiments, this solution also contained 0.5 mM MgATP and 100 nM PKA catalytic subunit (prepared in laboratory of Dr. M. P. Walsh, University of Calgary). Patch-clamp experiments were carried out at room temperature (22°C). Pipettes were pulled in two stages (PP-83, Narishige Instrumentation Laboratory, Tokyo, Japan) and had resistances of 4-6 M when filled with 150 mM NaCl solution. The bath was grounded through an agar bridge having the same ionic composition as the pipette solution. Single-channel currents were recorded from both cell-attached and excised patches; the pipette potential was held at +30 mV. Single-channel currents were amplified (Axopatch 1B, Axon Instrument, Foster City, CA), recorded on videocassette tape by a pulse-coded modulation-type recording adapter (DR384, Neurodata Instrument, New York), and low-pass filtered during playback using an eight-pole Bessel filter (900 LPF, Frequency Devices, Haverhill, MA). Final records were sampled at 0.5 kHz and analyzed using a laboratory microcomputer system and DRSCAN, a pCLAMP-compatible program developed in this laboratory for analyzing long records.

(1)

(2)

Transepithelial experiments. T84 cell monolayers were studied in water-jacketed Ussing-type chambers (Vangard International, Neptune, NJ) that had been fitted with Teflon adaptors for holding cell culture inserts. The control solution contained (mM) 115 NaCl, 2.5 K₂HPO₄, 1.5 CaCl₂, 1 MgSO₄, 10 glucose, and 25 NaHCO₃ and was gassed with 95% O₂-5% CO₂ at 37°C. Forskolin (10 µM) was used to stimulate I_sc, which, in this preparation, is carried entirely by net Cl secretion. All chemicals were from Sigma except calyculin A, which was from Calbiochem (San Diego, CA). Monolayers were short-circuited using a conventional voltage clamp (DVC-1000, WPI Instruments, Sarasota, FL), which applied brief voltage pulses every 2 min to monitor transepithelial resistance. I_sc values were transferred to a computer spreadsheet program for calculations and graphics.

Statistics. Values are presented as means ± SE. Significance was assessed at the 95% confidence level using the Student's t-test. Histograms of _o were fitted with Gaussian curves by least squares using Origin software (version 4.1, Microcal Software, Northampton, MA).

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Characterization of protein phosphatase preparations. Figure 1 shows an SDS-polyacrylamide gel of the four protein phosphatase preparations used in patch-clamp experiments, stained with Coomassie blue. The PP1 and PP2B preparations did not contain detectable contamination. The PP2A had 8% contaminant with M_r of ~90,000, whereas the PP2C profile shows two impurities (M_r ~30,000), constituting <3% of the protein. Western blot analysis with specific antibodies revealed that the PP2A preparation did not contain detectable PP2C or vice versa (data not shown). This was confirmed by functional assays; okadaic acid (0.01 nM) inhibited 50% of the PP2A activity but had no effect on the PP2C preparation. Representative activity curves of PP2C using phosphorylated myosin light chains as substrate under different conditions are shown in Fig. 2. Orthophosphate release was linear during the first 75 s under all conditions and during the first 2 min under the high-salt conditions used in patch-clamp experiments. Phosphatase activities calculated from the initial linear release rates are summarized in Table 2. All four enzymes were inhibited by the patch-clamp solution containing 150 mM NaCl. This was most pronounced for PP1 (72%) and PP2A (66%). PP2B activity was negligible in the absence of Ca²⁺ and calmodulin. The activity of PP1 was relatively labile, declining two- to threefold when diluted with BSA and stored on ice for several hours.

View larger version (80K): [in this window] [in a new window]   Fig. 1.   SDS-PAGE of protein phosphatases (PP) used in this study. PP2B (60 and 16 kDa; lane 1), catalytic subunit of PP1 (38 kDa; lane 2), PP2A (60, 55, and 36 kDa; lane 4), and PP2C (43 kDa; lane 5) were applied to a 12.5% Laemmli gel. Molecular masses are from Bio-Rad molecular mass standards, in lane 3, in order of descending size: myosin, -galactosidase, phosphorylase, serum albumin, ovalbumin, carbonic anhydrase, trypsin inhibitor, and RNase.

View larger version (13K): [in this window] [in a new window]   Fig. 2.   Activity of PP2C assayed under 4 different conditions. Time course of 32PO4 release from 1 µM radiolabeled myosin light chains at 30°C. Phosphatase activity was measured in solution 1 (high-salt; ) containing 150 mM NaCl, which was used for patch-clamp recording, same solution lacking NaCl (solution 2, low-salt; ), conventional Tris-buffered solution containing dithiothreitol (DTT, solution 3; ), and same Tris-buffered solution lacking DTT (). See Table 1 for composition of solutions.

Two channel behaviors suggest that phosphatase activity is unevenly distributed. CFTR channel activity was usually observed on BHK cells after incubation with 20 µM forskolin for 5-10 min. Channel activity usually persisted for >5 min after patches were excised into bath solution containing 0.5-1 mM MgATP (Fig. 3, A-C); however, activity did decline rapidly in about one-fourth of the excised patches (Fig. 3, D-F). These results contrast with our previous observations with CHO and airway epithelial cells, in which CFTR channel activity always declined rapidly (2, 35). We tentatively attribute the variable rundown to an uneven distribution of phosphatase activity in the plasma membrane of BHK cells (perhaps exacerbated by a very high level of CFTR expression), since other factors such as time after plating and recording conditions were kept constant. Channel activity declined to 5-10% of the starting value within 100 s in those patches that exhibited rundown and could be restimulated by adding PKA as in previous studies of CHO, T84, and airway epithelial cells (2, 35, 36).

View larger version (22K): [in this window] [in a new window]   View larger version (13K): [in this window] [in a new window]   View larger version (21K): [in this window] [in a new window]   View larger version (20K): [in this window] [in a new window]   Fig. 3.   Two types of channel activity in patches excised from baby hamster kidney (BHK) cells. A-C: examples of patches in which cystic fibrosis transmembrane conductance regulator (CFTR) channels remained active after excision (typical of ~3/4 of BHK patches). A: patch recording obtained before and after excision at arrow. Upward deflections indicate channel openings in this and all subsequent traces (holding potential +30 mV). After stimulation of cells by forskolin (10 µM), patches were excised into protein kinase A (PKA)-free bath solution containing MgATP (0.5 mM). B and C: mean number of channels open (NPo) and mean burst duration (o), respectively, calculated during 5-s segments from 3 patches. D-F: examples of patches in which channel activity rapidly declined after excision (typical of ~1/4 of BHK patches). D: patch recording showing spontaneous rundown after excision at arrow. E and F: NPo rapidly declined with no change in o; data pooled from 3 patches.

Burst duration during spontaneous rundown in excised patches. To assess alterations in single-channel kinetics during rundown, we calculated that _o values during each 5-s interval after P_o had declined by at least 90% and compared them with values obtained from patches having stable channel activity (Fig. 3, C and F). The distributions of open burst durations in both groups of patches could be fitted with Gaussian functions having similar means (n = 7 patches with rundown, n = 5 patches without rundown; P = 0.24; see Fig. 4). Thus, when membrane-associated phosphatase activity was present in a particular patch, it reduced CFTR channel activity by >90% without significantly altering _o.

View larger version (25K): [in this window] [in a new window]   Fig. 4.   Histograms of o in patches with and without rundown. o was calculated during each 5-s interval as described in MATERIALS AND METHODS. A: distribution of o in a patch with stable activity, i.e., without rundown. B: distribution of o in a patch displaying rundown of CFTR channel activity. Both distributions were fitted by 2 Gaussian curves. Overall mean and SD areas derived from fits of each Gaussian.

Effect of calyculin A on CFTR channels from BHK cells. Channel activity declined slowly in 20 of 84 patches excised in the presence of low (50-100 nM) PKA activity and 0.5 mM MgATP. The contribution of membrane-associated PP1/PP2A was assessed in some of the patches that displayed rundown by addition of the potent PP2A inhibitor calyculin A while rundown was in progress. Calyculin A (100-1,000 nM) did not block further rundown in 10 of 12 patches, nor did it enable the low PKA activity present in the bath solution to restimulate channels in any of the patches (n = 10). Figure 5 shows a representative experiment in which CFTR channel activity was high in the cell-attached configuration during forskolin stimulation (indicated by letter a) and declined spontaneously when excised into bath solution containing 100 nM PKA (compare with b and c). Rundown continued after the addition of 0.1 and 1.0 µM calyculin A (d and e, respectively) but was reversed by raising the PKA concentration from 0.1 to 0.3 µM (f). These results confirm that a fraction of membrane patches from BHK cells contain robust phosphatase activity resembling that in CHO cells (35). This predominant phosphatase is insensitive to calyculin A and independent of Ca²⁺ and calmodulin and therefore unlikely to be PP1, PP2A, or PP2B. However, because calyculin A did slow rundown in 3 of 10 patches, excised membrane patches from BHK cells can apparently contain some PP1 or PP2A activity that mediates a small fraction of the deactivation.

View larger version (19K): [in this window] [in a new window]   Fig. 5.   Lack of effect of calyculin A on channel rundown in an excised patch. CFTR channel activity was stimulated by preincubating cells with forskolin (20 µM). After recording channel activity in cell-attached configuration (see segment marked a), patch was excised at 1st arrow into bath solution containing low PKA activity (100 nM). Rundown was still observed under these conditions (compare b and c) and continued after addition of 100 nM calyculin A (d) and 1,000 nM calyculin A (e). Decline in channel activity was caused by dephosphorylation, since it was partially restored by raising concentration of PKA to 350 nM (f).

Effect of phosphatases on NP_o. Patches that had stable CFTR activity after 5-min exposure to low PKA (50-100 nM) were used to test the effects of exogenous phosphatases (solutions 1-3, Table 1). PP1 (final activity 1.6 pmol/min in 1 ml of bath buffer) did not affect P_o (P > 0.38), consistent with a previous study (4). By contrast, addition of PP2A (0.78 pmol/min) under the same conditions caused a slow decline in P_o from 0.46 to 0.12 within 6.2 ± 0.6 min (P < 0.005; Fig. 6A). PP2B (0.4-0.8 pmol/min) caused some decrease in P_o, but this just reached statistical significance (P = 0.04) after a 10-min exposure in solution 2 containing Ca²⁺ and calmodulin. PP2C was the most potent phosphatase tested, causing P_o to decline rapidly from 0.48 to 0.11 within 3.4 ± 0.8 min (final activity, 0.16-0.32 pmol phosphate transferred/min; P < 0.0004; Fig. 6B). Alkaline phosphatase at much higher levels (80 µmol phosphate transferred/min when assayed using PNP as substrate) caused slow deactivation (P < 0.002; Fig. 6C) as reported previously (2, 35). The effects of exogenous phosphatases on P_o are summarized in Fig. 7.

View larger version (30K): [in this window] [in a new window]   Fig. 6.   Patch recordings of PP2A, PP2C, and alkaline phosphatase effects on CFTR activity in excised patches. After stable channel activity had been recorded in excised patches for several minutes, exogenous phosphatases were added at arrows. A: PP2A caused 95% deactivation of CFTR channel in ~5 min; B: PP2C reduced channel activity by a similar amount within ~1 min; C: deactivation by alkaline phosphatase began after a delay of several minutes and required ~10 min for activity to decline to low levels.

View larger version (23K): [in this window] [in a new window]   Fig. 7.   Summary of PP1, PP2A, PP2B, PP2C, and alkaline phosphatase effects on open probability (Po) in excised patches. Patches were excised from BHK cells into bath solution containing 100 nM PKA at time 0. Po was computed for each 1-min segment. For clarity, values plotted represent Po determined during preceding minute (i.e., up to and including time point). Po of CFTR channels in most control patches increased gradually and then stabilized (A; n = 5). After recording of channel activity for 5 min, PP1 (C; n = 5), PP2A (D; n = 4), PP2B (E; n = 5), PP2C (F; n = 50), or alkaline phosphatase (G; n = 4) was added to bath. PP1 did not cause induced deactivation under these conditions. PP2A, PP2C, and alkaline phosphatase deactivated CFTR channels with different time courses. PP2B caused some deactivation that reached statistical significance after 10 min. For comparison with spontaneous rundown occurring in ~25% patches, excisions of patches with rundown are aligned in B with time at which exogenous phosphatases were added in other panels to patches having stable Po. Fast decline after PP2A (time constant  = 5.03 min) or PP2C ( = 0.69 min) was fitted by single exponential functions (dotted lines in D and F, respectively). Spontaneous decline in Po in patches with rundown was very similar to that induced by addition of exogenous PP2C ( = 0.69 min).

Effect of PP2A and PP2C on burst duration. To study phosphatase effects on channel gating, we used patches from forskolin-stimulated BHK cells having stable channel activity after excision. Phosphatases were added after recording channel activity for several minutes (Fig. 8). Representative control data under these conditions are shown in Fig. 3A, and the mean control values are given in Fig. 4A. Forskolin stimulation before excision was the only stimulus used to activate CFTR in these experiments, since exposure of excised patches to both phosphatase and PKA could complicate interpretation of gating effects. Both PP2A (Fig. 8A) and PP2C (Fig. 8B) caused deactivation of CFTR channels under these conditions, although the effect of PP2A was slower, in qualitative agreement with the results obtained when low PKA activity was also present (compare Fig. 7).

View larger version (17K): [in this window] [in a new window]   View larger version (17K): [in this window] [in a new window]   Fig. 8.   Effects of exogenous PP2A and PP2C on CFTR channel activity. NPo was calculated before and after addition of PP2A (A) or PP2C (B) to patches that had stable channel activity for at least 1 min after excision from forskolin-stimulated cells. (s) was calculated during 2 phases: rapidly declining initial phase (i) and deactivated phase, during which NPo was reduced by >90% (ii). Note decrease in o between phases i and ii after addition of PP2A (A). By contrast, NPo was similar during phases i and ii after PP2C addition (B).

View larger version (23K): [in this window] [in a new window]   View larger version (27K): [in this window] [in a new window]   Fig. 9.   Distributions of o after exposure to PP2A or PP2C. Histograms of o calculated during each 5-s interval during 2 phases shown in Fig. 8. Data were pooled from at least 3 patches under each condition. Distributions from phase i immediately after PP2A or PP2C addition were both fitted by two Gaussian curves. A: after prolonged exposure to PP2A (phase ii), slow o component disappeared and bursts in fast component became shorter. By contrast, prolonged exposure to PP2C did not eliminate slow component or shorten o (phase ii of B).

View larger version (34K): [in this window] [in a new window]   Fig. 10.   o after channel rundown and phosphatase exposure. Summary of o calculated in control patches (no rundown), after activity had declined spontaneously to <10% of starting value, and immediately after addition of PP2A or PP2C (phase i) or after prolonged exposure to PP2A and PP2C when activity had declined to <10% of starting value (phase ii).

Effect of Mg²⁺ on CFTR rundown in excised patches from CHO cells. Unlike other protein phosphatases, PP2C activity strongly depends on free Mg²⁺ concentration ([Mg²⁺]). We examined the effect of lowering total [Mg²⁺] from 2 to 0.5 mM on the rundown of CFTR channels in patches excised from CHO cells (Fig. 11). This represents a decrease in free [Mg²⁺] from 1.45 to 0.36 mM, which is expected to reduce PP2C activity by 60-70% according to previous biochemical studies (8). CHO cells were used for these experiments because all patches excised from CHO cells displayed rapid rundown under control conditions. Channel activity was increased approximately fourfold during the first 5 min of exposure to 0 mM Mg²⁺, consistent with the Mg²⁺-dependent phosphatase activity.

View larger version (21K): [in this window] [in a new window]   Fig. 11.   Effect of reducing free Mg2+ concentration ([Mg2+]) on rundown of CFTR channel rundown in patches excised from Chinese hamster ovary cells. NPo calculated during 10-s intervals after excision, normalized to value measured in cell-attached configuration. A: normalized NPo under control conditions with 2 mM Mg2+ in bath solution. B: same experiments carried out with 0.5 mM free Mg2+. Note that channel activity was prolonged by reducing [Mg2+].

Effect of calyculin A on CFTR rundown in T84 cell monolayers. If soluble phosphatases are lost from excised patches, studying rundown might underestimate the importance of cytosolic PP2A in intact cells. We therefore examined the effect of calyculin A on I_sc across intact (unpermeabilized and undialyzed) T84 monolayers. Previous studies have shown that I_sc in this preparation provides a measure of net Cl secretion under various conditions (9) and that Cl secretion is mediated by CFTR channels (6, 37). Calyculin A (20 nM-100 nM) had no effect on I_sc when it was added to both sides (Fig. 12). Subsequent addition of 10 µM forskolin to the serosal side increased I_sc from 3 to 43 µA/cm² within ~8 min. The rate and magnitude of the forskolin stimulation were not affected by calyculin A (Fig. 12). I_sc declined exponentially back to the baseline level when forskolin was washed out. The rate of this decline was also unaffected by 20 nM calyculin A (Fig. 12A); I_sc after forskolin washout was well fitted by a single exponential having the same time constant in the absence ( = 3.6 ± 0.2 min) or presence ( = 3.5 ± 0.1 min) of calyculin A. Complete deactivation of I_sc was observed after forskolin removal despite the presence of 20 nM calyculin A. Similar results were obtained using T84 monolayers at three different passages.

View larger version (19K): [in this window] [in a new window]   Fig. 12.   Effect of calyculin A on deactivation of short-circuit current (Isc) across T84 cell monolayers after forskolin washout. A: adding calyculin A (20 nM; n = 12) had no effect on baseline Isc, stimulation elicited by forskolin (10 µM), or rate of decay of Isc after washout of calyculin compared with control monolayers (n = 24). B: adding 100 µM calyculin A at same time as forskolin was withdrawn did not affect rate of deactivation (n = 13).

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

The goal of this study was to compare the ability of PP2C and other protein phosphatases to downregulate CFTR and assess their relative importance in deactivating CFTR channels in the epithelial cell line T84. By adding exogenous phosphatases to BHK patches having stable channel activity, we found that PP2A and PP2C both deactivated CFTR channels which had been activated in vivo (by forskolin) or in vitro (by PKA catalytic subunit). The protein phosphatase preparations were assayed biochemically with a common substrate and patch-clamp solution, so that comparable levels of PP activity could be used when recording channel activity.

Evidence that PP2C is the primary phosphatase regulating CFTR in BHK and T84 cells. Several results in this study suggest that endogenous PP2C regulates CFTR in the cells studied: 1) deactivation induced by exogenous PP2C occurred at the same rate as that mediated by endogenous, membrane-associated phosphatase; 2) burst duration did not change during deactivation by exogenous PP2C or during spontaneous rundown after excision, whereas exogenous PP2A caused burst shortening; 3) rundown in excised patches and deactivation after forskolin washout from cell monolayers were both insensitive to calyculin A; and 4) spontaneous rundown was Mg²⁺-dependent within the millimolar range, consistent with the known properties of PP2C. Recent studies (38, 40) indicate that PP2C is expressed in most tissues, including T84 cells. Our conclusions regarding the role PP2C in regulating epithelial CFTR are in agreement with recent results of Travis et al. (38), who found that okadaic acid did not inhibit deactivation of Cl current across permeabilized human airway and T84 cell monolayers. They also found that macroscopic (CFTR-mediated) current in excised patches was reduced by ~80% after exposure to recombinant PP2C. PP2A may play a larger role in regulating CFTR in cardiac (15) and sweat duct epithelial cells (29). Distinct effects of PP2A and PP2C on the gating of single CFTR channels have not been described previously.

Properties of protein phosphatase preparations. Analysis of the phosphatases by SDS-PAGE indicated that the PP2A and PP2C preparations contained some impurities (8 and <3%, respectively). However, Western blot analysis with specific antibodies did not reveal any cross-contamination, i.e., no PP2A in the PP2C preparation or PP2C contamination in the PP2A preparation. Moreover, okadaic acid did not affect phosphatase activity of the PP2C preparation but abolished that of the PP2A preparation, and the PP2C preparation exhibited no activity in the absence of Mg²⁺. We conclude that the PP2A and PP2C preparations used were functionally homogeneous. PP2A from a commercial supplier had ~1,000-fold lower activity than the PP2A preparation used here when it was assayed under the same conditions. Commercially prepared PP2B was more satisfactory, having low phosphatase activity in the absence of Ca²⁺ and calmodulin that was stimulated 30- to 40-fold by addition of these cofactors.

Effect of PP2A on burst duration. Exogenous PP2A caused a remarkable shortening of burst duration when added to BHK patches. Burst shortening did not occur during the spontaneous deactivation induced by endogenous phosphatase activity or after addition of exogenous PP2C, but has been reported previously for cardiac CFTR where it is a major mechanism by which P_o is downregulated (16). The different results would be reconciled if cardiac cells had more membrane-associated PP2A activity than the cells used here. Higher PP2A activity would explain both burst shortening and the more pronounced inhibition of rundown in cardiac cells by okadaic acid (15).

[Mg²⁺] and relative importance of PP2C and PP2A. In contrast to the present results, deactivation of macroscopic CFTR conductance in isolated, permeabilized sweat ducts is prevented by okadaic acid (29). This could reflect higher PP2A and/or lower PP2C expression in sweat duct epithelium, or it may be due to the experimental conditions used. Sweat ducts were treated with -toxin to permeabilize the basolateral membrane to small solutes such as ATP, cAMP, and phosphatase inhibitors. The free [Mg²⁺] of the bath solution (exposed to permeabilized membrane) was strongly buffered to ~0.1 mM by high concentrations of ATP, gluconate, and EGTA. Most PP2C would have been inactive if intracellular [Mg²⁺] approached this same low level, which may explain the absence of an okadaic acid-insensitive component. In a whole cell patch-clamp study of cardiac cells, in which >50% of the deactivation was okadaic acid-insensitive, Hwang et al. (15) used pipette solution containing 0.92 mM [Mg²⁺]. In our studies of excised patches, okadaic acid-insensitive rundown was routinely observed with solutions containing 1-3 mM [Mg²⁺]. Thus the relative contribution of PP2C reported for different preparations correlates with [Mg²⁺], although other explanations for the differences are also possible. Cytoplasmic free [Mg²⁺] has not been measured in the cells used here, but recent estimates for the bulk cytoplasm of other cells are in the 0.5-1.1 mM range (e.g., for review, see Ref. 32).

Other phosphatases: PP1, PP2B, and alkaline phosphatase. PP1 is a ubiquitous protein phosphatase that is known to regulate other ion channels. Exogenous PP1 (1.6-3.2 U/ml) failed to deactivate CFTR Cl channels when added to patches from BHK cells. This result agrees with that of Berger et al. (4), who found that PP1 (5 U/ml) did not deactivate CFTR when patches were excised from NIH/3T3 cells and is also consistent with the insensitivity of I_sc deactivation in T84 monolayers exposed to calyculin A (present study) and okadaic acid (38). Insensitivity to calyculin A in T84 cells excludes a role for other isoforms of PP1 (PP1 or PP1), since they would also have been inhibited.

PP2A and PP2C act at functionally distinct phosphorylation sites. The results in this study indicate that CFTR deactivation in several cell lines, including T84 epithelial monolayers, is mediated primarily by an okadaic acid- and calyculin A-insensitive phosphatase (35). These properties and the Mg²⁺ dependence of rundown point to PP2C; however, a specific inhibitor of PP2C is not presently available. About 10% of the PP2C in human HL-60 cells is membrane associated (25). Approximately 33% of PP2C activity in BHK, CHO, T84, and Calu-3 cells is in the particulate (membrane) fraction (40). A membrane-bound form of PP2C has also been reported in Paramecium (18). Although PP2A appears to mediate little, if any, deactivation in these cells under normal conditions (i.e., with physiological levels of Mg²⁺), neither PP2C nor PP2A was able to completely deactivate CFTR channels in the present study. Between 5 and 10% of the initial channel activity persisted in the presence of each exogenous phosphatase. These results suggest that multiple phosphatases may be required for complete deactivation, analogous to previous studies (15). Alternatively, PP2C may need to be associated with CFTR in the membrane to be fully effective. The qualitatively different effects of PP2A and PP2C on gating imply that they dephosphorylate functionally distinct sites on CFTR. Deactivation by PP2A was accompanied by a reduction in _o. By contrast, deactivation by PP2C was not associated with burst shortening, although it caused a similar decline in P_o. The inability of PP2C to alter burst duration was not due to weaker phosphatase activity, since PP2C caused more rapid deactivation than did PP2A, and raising the concentration of PP2C by fourfold still did not result in shorter bursts (Luo, unpublished observation). As discussed above, the PP2A and PP2C preparations used in patch-clamp experiments had comparable phosphatase activities when assayed for their ability to dephosphorylate myosin light chains. These results suggest that burst and interburst durations are regulated independently, presumably by distinct phosphorylation sites.