Phosphorylation by protein kinase C is required for acute activation of cystic fibrosis transmembrane conductance regulator by protein kinase A.

Protein kinase A (PKA) stimulates Cl secretion by activating the cystic fibrosis transmembrane conductance regulator (CFTR), a tightly regulated Cl− channel in the apical membrane of many secretory epithelia. The CFTR channel is also modulated by protein kinase C (PKC), but the regulatory mechanisms are poorly understood. Here we present evidence that PKA-mediated phosphorylation alone is not a sufficient stimulus to open the CFTR chloride channel in the presence of MgATP; constitutive PKC phosphorylation is essential for acute activation of CFTR by PKA. When patches were excised from transfected Chinese hamster ovary cells, CFTR responses to PKA became progressively smaller with time and eventually disappeared. This decline in PKA responsiveness did not occur in the presence of exogenous PKC and was reversed by the addition of PKC to channels that had become refractory to PKA. PKC enhanced PKA stimulation of open probability without increasing the number of functional channels. Short-term pretreatment of cells with the PKC inhibitor chelerythrine (1 μM) reduced the channel activity that could be elicited by forskolin in cell-attached patches. Moreover, in whole cell patches, acute stimulation of CFTR currents by chlorophenylthio-cAMP was abolished by two chemically unrelated PKC inhibitors, although an abrupt, partial activation was observed after a delay of >15 min. Modulation by PKC was most pronounced when basal PKC phosphorylation was reduced by briefly preincubating cells with chelerythrine. Constitutive PKC phosphorylation in unstimulated cells permits the maximum elevation of open probability by PKA to reach a level that is ∼60% of that attained during in vitro exposure to both kinases. Differences in basal PKC activity may contribute to the variable cAMP responsiveness of CFTR channels in different cell types.

The regulatory domain of cystic fibrosis transmembrane conductance regulator (CFTR) 1 has nine dibasic consensus se-quences for phosphorylation by protein kinase A (PKA) and at least seven potential sites for phosphorylation by protein kinase C (PKC). There is direct evidence for phosphorylation of serines 660, 700, 712, 737, 753, 768, 795, and 813 by PKA (1)(2)(3) and serines 660, 686, 700, and 790 by PKC (2). In freshly excised membrane patches, CFTR Cl Ϫ channels are activated by exposure to the PKA catalytic subunit and MgATP (4,5). Altering the four major sites of in vivo phosphorylation reduces forskolin-stimulated Cl Ϫ permeability (1,6), whereas removal of all 10 dibasic sites and the additional monobasic site Ser 753 further reduces, but does not eliminate, PKA stimulation of CFTR in both intact cells and excised patches (6,7). There is little doubt that PKA-mediated phosphorylation is the primary mechanism controlling CFTR Cl Ϫ channel activity, consistent with early evidence that cAMP stimulates transepithelial secretion, in part by increasing apical membrane Cl Ϫ conductance (8 -12).
The role of PKC in regulating the CFTR channel is less certain. Exogenous PKC causes a small stimulation of channel activity when added during the first few minutes after patches are excised (4,13). PKC enhances both the rate and magnitude of subsequent PKA stimulation of open probability (p o ) by approximately 2-fold and 50%, respectively (4). Potentiation of cAMP responses has also been observed in phorbol-stimulated Xenopus oocytes expressing CFTR (14), HT-29 cells (15), and pancreatic duct cells, in which the PKC agonist ␤-phorbol 12,13-dibutyrate increased cAMP-induced whole cell CFTR Cl Ϫ current by 31% although having no effect on basal currents (16). Potentiation has received several different interpretations. The effects of PKC on excised patches (4) and phorbols on T84 and C127 were suggested to involve PKC regulation of the CFTR channel itself (17,18). However, to explain the apparent increase in Cl Ϫ current by PKC in pancreatic duct cells (expressed as pA/pF), it was proposed that PKC reduces cell membrane area rather than increasing channel activity (16). By contrast, exocytotic addition of new channels to the plasma membrane was suggested to explain the potentiation in HT-29 cells (15).
Studying CFTR function after phorbol treatment is problematic when both its expression (19) and degradation (20) are strongly influenced by PKC activity. Another problem has been the difficulty of distinguishing acute PKC-induced changes in p o from changes in the number of functional channels (N). Finally, the responsiveness of CFTR to PKA stimulation varies widely among different cell types even when channels are studied in excised patches. The reasons for this variation are not known.
Here we present evidence that PKC-mediated phosphorylation not only potentiates CFTR responses but is, in fact, essential for acute PKA activation of CFTR channels. This requirement for PKC only became apparent when the PKA response was studied after prolonged rundown in excised patches or after short-term pretreatment with PKC inhibitors, presumably because constitutive PKC phosphorylation was slowly removed by phosphatase activity under these conditions (4,21,22).

EXPERIMENTAL PROCEDURES
Cell Culture-Chinese hamster ovary (CHO) cells and baby hamster kidney (BHK) cells that had been stably transfected with plasmids directing the expression of wild-type CFTR were provided by Drs. X.-B. Chang and J. R. Riordan (Mayo Clinic, Scottsdale, AZ; Ref. 6) and F. Siebert (University of Toronto, Toronto, Ontario, Canada; Ref. 7). CHO and BHK cells were cultured as described previously (4,42), subcultured on glass coverslips in 35-mm Petri dishes, and used in patch clamp experiments 2-5 days after plating.
Single Channel Studies-Single channels were recorded using the cell-attached and inside-out patch clamp configurations (23; also see Ref. 4). The bath and pipette solutions contained 150 mM NaCl, 10 mM TES, and 2 mM MgCl 2 (pH 7.2). Fresh MgATP (1 mM final concentration) and forskolin (5 M) were added to the bath before each experiment. Experiments were carried out at room temperature (ϳ22°C). Single channel currents were amplified (Axopatch 200; Axon Instruments Inc., Foster City, CA), recorded on video tape (PCM 418; Medical Systems Corp, Greenvale, NY), and analyzed as described previously (4). The pipette voltage was held at Ϫ30 mV during cell-attached recordings and then switched to ϩ30 mV after excision so that the potential across the membrane patch would be similar in both configurations (approximately Ϫ30 mV). Channel activity was allowed to run down for 10 min after excision to allow dephosphorylation by membrane-associated phosphatase activity prior to the addition of protein kinases. To estimate the N in the patch, AMP-PNP (1 mM) was added to the bath at the end of each experiment to lock CFTR channels in the open state so they could be counted (24). MgATP (1 mM) was added to the pipette solution from a 100 mM stock in TES buffer. Stock cpt-cAMP and 8-bromo-cAMP solutions (50 mM) were prepared in distilled water using their sodium salts and used at a final concentration of 100 M. The PKC inhibitor Gö6976 (Calbiochem) was prepared in ME 2 SO and diluted in bath solution immediately before use (final concentration, 500 nM). It was also included in the pipette solution, and experiments were begun 4 min after obtaining the whole cell configuration to allow its diffusion into the cell (25). When chelerythrine was used, it was prepared as described above and added at a final concentration of 1 M for 30 min prior to recording whole cell currents.
Whole cell currents were recorded using pipettes pulled from thickwalled borosilicate glass (1.5-3 M⍀). After obtaining seals, pipette capacitance was canceled using the internal circuitry of the patch clamp (Axopatch 1C; Axon). The whole cell recording configuration was obtained by applying excess suction, and cell capacitance was measured and then canceled. In most experiments, V m was held at 0 mV for 1 s between alternating pulses to Ϯ 60 mV (1-s duration). Current-voltage relationships were generated by stepping from Ϫ100 to ϩ100 mV in 10-mV increments, returning to the holding potential of 0 mV between pulses. Currents are expressed per unit cell capacitance (pA/pF) to allow comparison of data from different cells. Outward (positive) current corresponds to Cl Ϫ ions entering the cell. The Nernst prediction for a perfectly Cl Ϫ -selective membrane was Ϫ39 mV under the conditions used. The leak current at ϩ60 mV (ϩ63.4 Ϯ 14.7 pA/pF) was small compared with the mean cAMP-stimulated CFTR current (8.6 Ϯ 1.4%), and therefore was not subtracted. Mean pipette capacitance was 7.87 Ϯ 0.08 pF (n ϭ 46), mean whole cell capacitance was 13.1 Ϯ 0.6 pF (n ϭ 51), and the average access resistance of the pipettes was 15.6 Ϯ 1.0 M⍀ (n ϭ 47). Data

Effect of PKC on CFTR Responsiveness to PKA in Excised
Membrane Patches-Exposing transfected CHO cells to forskolin activated CFTR Cl Ϫ channels in 15 of 15 cell-attached patches (Fig. 1). The p o under cell-attached conditions was 0.32 Ϯ 0.038 (n ϭ 4). As in previous studies, p o declined spontaneously when patches were excised into a bath solution containing 1 mM MgATP. When exogenous PKA (200 units/ml) was added to the bath, the magnitude of the CFTR channel response depended on the time that had elapsed since excision. Robust PKA stimulation was observed ϳ2 min after patches were excised, whereas PKA usually failed to activate CFTR Cl Ϫ channels when patches were excised and continuously exposed to 1 mM MgATP for 10 min. When patches were excised for 10 min and allowed to become refractory to PKA, responsiveness could be fully restored by adding PKC and the lipid cofactor DiC 8 to the cytoplasmic side in the continued presence of PKA (Fig. 1B). This stimulation after the addition of PKC and DiC 8 was elicited by the PKA already present in the bath rather than by PKC itself, because addition of PKC and DiC 8 after the same delay had no effect in the absence of PKA (Fig. 1C). Moreover, CFTR responsiveness to PKA did not decline if PKC was present during the 10-min period following excision (Fig. 1D). Taken together, these results suggest that PKA phosphorylation alone is not a sufficient stimulus to open the CFTR channel; constitutive phosphorylation by PKC is also required. Fig. 2 summarizes the dependence of the PKA response on time after excision and the effect of exogenous PKC on channel responsiveness to PKA. All channels were locked open at the end of each experiment using AMP-PNP to determine N. This number was then used to calculate p o , as described under "Experimental Procedures." CFTR channels were consistently reactivated if PKA was added within 2 min after excision; however, this stimulation of p o was progressively reduced with time after excision ( Fig. 2A). The mean p o values during the PKA responses are shown in Fig. 2B. The interval between excision and PKA addition was inversely related to the p o during PKA stimulation, which eventually declined to 0. By contrast, responsiveness to PKA did not decline when patches were exposed to PKC, DiC 8 , and MgATP throughout the 10-min period after excision. Single-channel p o in the presence of PKA and PKC was 0.46 Ϯ 0.07 (n ϭ 3), similar to that observed for the same patches during forskolin stimulation prior to excision (p o ϭ 0.42 Ϯ 0.02; p Ͼ 0.05; Fig. 2C). This p o is higher than that calculated for patches excised from unstimulated cells into the bath solution containing PKA and ATP (0.22 Ϯ 0.04; Ref. 24), when the number of channels in each patch was corrected using AMP-PNP. Thus, in addition to having a permissive effect on p o (Fig. 1B), PKC also potentiated the PKA response under these conditions by increasing the maximum p o elicited by PKA approximately 2-fold, as reported previously (4).

Mechanism of PKC Dependence-To further investigate
whether PKC regulates the number of channels opened by PKA or the p o of individual channels, patches containing a small number of CFTR channels were excised from CHO cells and exposed to PKA after a delay of 5 min. Adding PKC and DiC 8 to the same patches 5 min after PKA addition did not increase the maximum number of channels open simultaneously, which was identical to the value of N determined using AMP-PNP at the end of the experiment (Fig. 3A). By contrast, p o of single channels calculated from the mean number of channels open and N was approximately doubled (p Ͻ 0.005; Fig. 3B). When BHK cells, which express more than a 10-fold higher density of CFTR channels, were pretreated with the PKC inhibitor chelerythrine (1 M) for 5 min, adding PKA to the excised patches caused only a small increase in channel activity, although the number of channels locked open by AMP-PNP was high, indicating the channels were active when exposed to PKA alone but simply had low p o . For example, the patch shown in Fig. 3D  locked open by AMP-PNP (Fig. 3D), further evidence that channels were already functional after PKA alone. Thus, PKC acts by enhancing PKA stimulation of p o rather than by increasing N. PKC shortened the mean closed time (p Ͻ 0.01) but had no effect on mean burst duration (open time; p Ͼ 0.05) during PKA stimulation (Fig. 3C). The lower p o in Fig. 3D (Ͻ0.05) compared with Fig. 3B (ϳ0.2) is explained by pretreatment of the former with chelerythrine to reduce constitutive PKC phosphorylation. These results suggest there is some constitutive PKC activity in unstimulated cells, which permits submaximal responses to PKA stimulation. This possibility is examined below using cell-attached and whole cell patches.
Effect of Inhibiting Constitutive PKC Activity in Intact Cells and Whole Cell Patches-To investigate whether endogenous PKC activity is required for cAMP activation of CFTR in intact cells, we tested the effect of preincubating cells with chelerythrine, a specific PKC inhibitor (26,27). Pretreatment with chelerythrine abolished forskolin activation of CFTR channels in 13 of 16 cell-attached patches (Fig. 4). In parallel control experiments forskolin activated CFTR Cl Ϫ channels in 15 of 15 cell-attached patches. Similarly, under control conditions, cpt-cAMP (100 M) activated whole cell CFTR chloride currents in 6 of 6 cells with a time to half-maximal activation of 2.1 Ϯ 0.6 min under control conditions but in other experiments had no effect within 12 min when cells were pretreated with chelerythrine (Fig. 5). Moreover, in further studies, a selective inhibitor of the Ca 2ϩ -dependent ␣ and ␤ 1 isoforms of PKC (Gö6976; Refs. 28 -30) also prevented acute activation of CFTR whole cell Cl Ϫ currents by cpt-cAMP (Fig. 5), although partial activation of CFTR was observed after prolonged treatment with cpt-cAMP in the presence of either chelerythrine or Gö6976. The time to half-maximal activation of this delayed response was 23.9 Ϯ 6.2 min in chelerythrine-treated cells and 21.5 Ϯ 5.1 min in Gö6976-treated cells, a lag that was approximately 10-fold longer than for control cells, which had been treated with cpt-cAMP in the absence of PKC inhibitors (p ϭ 0.0022 and 0.0022, respectively; Mann-Whitney U test; Fig. 5, D and E). The delayed response was abrupt, suggesting it may be caused by sudden activation of an alternative pathway that circumvents inhibition of PKC activity. Similar results were obtained and interbursts (T c ) during exposure of the patches to PKA or to both PKA and PKC. Note that only the interburst duration is affected by PKC potentiation. D, inhibiting constitutive PKC phosphorylation reduces the p o response to PKA without reducing the N. This recording, which is from a patch containing Ͼ40 channels, begins immediately after the patch was excised from a cell that had been pretreated with chelerythrine for 5 min. Addition of PKA stimulated CFTR channel activity (see expanded trace) but caused only a slight increase in macroscopic current. Nevertheless, subsequent addition of AMP-PNP gradually increased the current by at least 20-fold, consistent with the locking open of a large number of channels that initially had low open probability (p o Ͻ0.05) during stimulation by PKA alone. Adding PKC and DiC 8 did not cause additional channels to lock. Conditions were the same as in Fig. 1. when chelerythrine-treated cells were exposed to 8-bromo-cAMP rather than cpt-cAMP (data not shown). By contrast, a delayed response did not develop during very long recordings in the absence of cAMP analogs (range, 31-50 min; n ϭ 3). These results suggest that the inhibition of PKC can eventually be overcome by activation of an alternative pathway, which restores the responsiveness of CFTR channels to PKA. DISCUSSION The CFTR chloride channel is regulated by PKA-and PKCmediated phosphorylation; however, the relationship between these kinases has not been studied in detail. Previous work suggested that PKC increases the stimulation of p o by PKA approximately 2-fold (4); however, no precautions were taken in those early experiments to minimize PKC phosphorylation prior to testing kinase effects. A major finding of the present study is that PKC regulation of CFTR is mainly constitutive, at least in the two mammalian expression systems studied. Using somewhat different protocols we found that PKC not only potentiates responses but is required for acute stimulation of CFTR by PKA. This dependence on PKC was observed in excised and whole cell patches and in intact cells, but only when protocols were designed to allow dephosphorylation of PKC sites, which may explain why a strict dependence on PKC has not been observed previously. Although PKC is permissive and modulates the PKA responses over a wide range in excised patches (p o ϭ 0 -0.7), basal PKC activity in resting CHO cells appears sufficient to maintain PKA responsiveness near its maximal level.
The present results confirm that forskolin stimulates CFTR channels in CHO and BHK cells, and that CFTR channel activity declines rapidly when patches are excised in the absence of PKA, consistent with dephosphorylation of PKA sites by membrane-associated phosphatases (4). Moreover, PKA reactivates CFTR chloride channels when added soon after patches are excised but not when more than 10 min have elapsed. We attribute this decline in responsiveness to PKA to slow dephosphorylation of PKC sites, because it did not occur in the presence of PKC and was reversed by the addition of PKC to the bath. Although PKC alone caused an apparent, partial (10%) activation in this and previous studies, the small PKC response may be a manifestation of the permissive effect of PKC. The early experiments suggesting activation by PKC alone were not preceded by a lengthy rundown period such as the one shown in Fig. 1C. Thus the stimulation by PKC may have been due to the reactivation of channels that had become dephosphorylated at a permissive PKC site but retained some residual PKA phosphorylation. Consistent with this explanation, some CFTR activity was observed in the present work when patches were excised directly into PKC-containing buffer lacking PKA, comparable with that reported previously when PKC was added immediately after channel rundown (p o ϭ 0.04, (4) and 0.034 Ϯ 0.0125, (31)). Regulation of Cl Ϫ secretion in native epithelia is complicated because PKC has multiple sites of action, including inhibition of basolateral potassium conductance. Nevertheless, there are several reports of phorbol ester stimulation of Cl Ϫ secretion in airway and other epithelia (32-36) that might be explained by basal PKA activity combined with the permissive effect of PKC described here.
PKA reactivated CFTR channels when added 2 min after excision but not when added 10 min after excision; therefore, the permissive PKC site(s) may be dephosphorylated more slowly than the PKA sites mediating initial rundown of channel activity. The rundown caused by dephosphorylation of PKA sites occurs within 100 s when patches are excised from these cells at room temperature (22). Biochemical studies will be needed to assess whether constitutive PKC phosphorylation is required for phosphorylation of CFTR by PKA as in hierarchical phosphorylation (37) or whether PKA phosphorylation of CFTR is normal in the absence of PKC and only the functional response of the channel to PKA phosphorylation is attenuated. Prestimulating CHO cells with phorbol ester did not noticeably increase phosphorylation of CFTR protein during forskolin stimulation (6); however, the functional responses may depend on trace phosphorylation at cryptic sites, which could escape detection. Interdependent PKC and PKA regulation may allow more precise control of CFTR activity, as would regulation by multiple phosphatases acting at PKA or PKC sites.
Permissive control of cAMP stimulation by PKC was also observed in intact cells; PKC inhibitors greatly reduced forskolin-stimulated CFTR activity in cell-attached patches and abolished acute stimulation of whole cell CFTR currents by cpt-cAMP. Thus constitutive, endogenous PKC activity is sufficient for activation of CFTR by forskolin or cAMP. The fact that exogenous PKC needs to be added to excised patches implies that the endogenous PKC activity responsible for priming CFTR is not membrane-associated, although it remains a formal possibility that PKC activity in the patches is membraneassociated but requires an additional cytoplasmic factor for activity. A membrane-associated PKC isoform has been shown previously to stimulate CFTR channels when added to patches excised from NIH-3T3 cells (13).
CFTR channels were weakly responsive to PKA when cells were briefly pretreated with chelerythrine (5 instead of Ͼ30 min) or when patches were excised from cells and allowed to run down for only 5 min so that some PKC phosphorylation would remain. To assess whether the permissive effect of PKC involved an increase in p o or the number of functional channels under these conditions, we took the advantage of the previous findings that the poorly hydrolyzable analog AMP-PNP causes CFTR to become locked in an open burst state in which p o Ϸ 1.0 (38,39). Exposure to PKC increased the p o without altering the number of functional (i.e. PKA-responsive and locked by AMP-PNP) channels in the patches. This result was most obvious when CHO patches containing only a few channels were used, since the probability that all the functional channels would be open simultaneously was highest under these conditions.
To ensure that increases in N were not missed by studying cells with low CFTR channel density, we also examined BHK cells that contained between 40 and several hundred CFTR channels per patch. When BHK cells were pretreated with chelerythrine for 5 min, subsequent exposure of excised patches to PKA caused only a small stimulation, although the number of functional channels as defined by their susceptibility to locking by AMP-PNP was high and was not increased further by PKC. If exocytotic insertion of CFTR channels is stimulated by PKC, it was not apparent in the patches we studied and did not contribute to the potentiation reported in this article. Kinetic analysis of patches from CHO cells revealed that PKC decreased the mean closed time during PKA stimulation and had no effect on the mean open time. These results are consistent with an increased rate of channel open-ing, such as the one that accompanies phosphorylation-induced increases in ATP affinity (24,31). In a CFTR mutant with alterations at all 10 dibasic consensus PKA sequences (10SA; Ref. 6), the reduction in p o was associated with a lower bursting rate (i.e. a longer closed time) rather than a shortening of the bursts (24). Strong potentiation by PKC was still observed in the 10SA mutant, which lacks serines at positions 660, 686, and 700; therefore, these dual PKA and PKC sites are not essential for PKC-dependent regulation of PKA responsiveness.
PKC isozymes have been classified into several distinct groups (40). The "Ca-dependent" group (␣, ␤ I , ␤ II , and ␥ isozymes) contains a putative Ca 2ϩ binding site, which is lacking in the "Ca-independent" group (␦, ⑀, , , and isozymes). Another "atypical" group (including the and isozymes) is structurally distinct and poorly characterized. Based on its sensitivity to Gö6976, the permissive regulation in CHO and BHK cells appears to be mediated by endogenous Ca-dependent ␣ or ␤ I forms. Nevertheless, PKA responsiveness in excised patches was restored by the addition of rat brain PKC, which consists mainly of the Ca-dependent ␤ II isozyme, and the role of PKC isozymes in CFTR regulation may vary according to cell type.
Although the initial experiments in this study suggested that PKA responsiveness is completely dependent on PKC phosphorylation, we were surprised to find that whole Cl Ϫ cell currents suddenly became responsive to cpt-cAMP in the presence of PKC inhibitors after a delay of more than 20 min. Since channels in excised patches did not regain responsiveness to PKA during long exposures (Ͼ30 min) in the absence of PKC, the surrogate pathway that overcomes PKC inhibition in whole cell patches is probably not membrane-delimited. In whole cell studies, similar responses were observed with two structurally unrelated PKC inhibitors, further evidence that inhibition of acute cAMP stimulation is a bona fide effect on PKC. Also, PKA responsiveness did not increase gradually in whole cell patches in a manner consistent with the loss of inhibitor potency. Rather, the delayed response developed abruptly after 20 min, as if the dependence on PKC was relieved by activation of an alternate signaling pathway.
In summary, these results indicate that PKA-mediated phosphorylation alone is not a sufficient stimulus to activate the CFTR chloride channel. We propose that under normal conditions, constitutive phosphorylation of CFTR by PKC has a permissive role in priming the channel for acute activation by PKA. Thus maneuvers that lower this constitutive phosphorylation reveal PKC effects that are much more profound than previously suspected. These results have implications for functional studies of CFTR phosphoforms; PKA-stimulated channel activity in membrane patches and bilayers will be dictated by residual PKC phosphorylation. Variations in constitutive phosphorylation by PKC may also help explain the variable responsiveness of CFTR channels to stimulation by PKA in different expression systems (7,41).