The regulatory domain of cystic fibrosis transmembrane conductance regulator (CFTR) regulates channel activity when several serines are phosphorylated by cAMP-dependent protein kinase. To further define the functional role of individual phosphoserines, we studied CFTR containing previously studied and new serine to alanine mutations. We expressed these constructs in Fischer rat thyroid epithelia and measured transepithelial Cl− current. Mutation of four in vivo phosphorylation sites, Ser660, Ser737, Ser795, and Ser813 (S-Quad-A), substantially decreased cAMP-stimulated current, suggesting that these four sites account for most of the phosphorylation-dependent response. Mutation of either Ser660 or Ser813 alone significantly decreased current, indicating that these residues play a key role in phosphorylation-dependent stimulation. However, neither Ser660 nor Ser813 alone increased current to wild-type levels; both residues were required. Changing Ser737 to alanine increased current above wild-type levels, suggesting that phosphorylation of Ser737 may inhibit current in wild-type CFTR. These data help define the functional role of regulatory domain phosphoserines and suggest interactions between individual phosphoserines.
- Cl− channel
- cystic fibrosis transmembrane conductance regulator
- regulatory domain
- cystic fibrosis
activity of the Cl− channel of the cystic fibrosis transmembrane conductance regulator (CFTR) is controlled by three cytosolic domains, two nucleotide-binding domains (NBDs), and a regulatory (R) domain (20). The gating kinetics of CFTR are controlled by ATP binding and hydrolysis by the NBDs (8, 24). However, the physiological regulation of channel function depends primarily on phosphorylation of the R domain. CFTR activity is regulated by phosphorylation of the R domain by cAMP-dependent protein kinase (PKA) and by dephosphorylation of the R domain by several phosphatases (8, 24). Protein kinase C, cGMP-dependent protein kinase, and tyrosine phosphorylation also stimulate CFTR channel activity.
The R domain contains eight serines (residues 660, 686, 700, 712, 737, 768, 795, and 813) and one threonine (residue 788) at consensus phosphorylation sites as well as several serines at less well-conserved PKA phosphorylation sites (20). Several studies have provided insight into which of these serines are phosphorylated. When CFTR is studied in vitro, PKA phosphorylates serine-660, -700, -712, -737, -753, -768, -795, and -813 (5, 15, 17). When isolated R domain proteins and peptides are studied, PKA phosphorylates these residues plus serine-686 (16, 17, 22, 26). When CFTR is phosphorylated in cells by the addition of cAMP agonists, Ser660, Ser700, Ser737, Ser795, and Ser813 are phosphorylated (5,6, 17). This evidence for phosphorylation of multiple different serines by PKA suggests that phosphorylation-dependent regulation of CFTR may be very complex.
To understand their stimulatory role, two quantitative studies have mutated individual phosphoserines and evaluated channel activity. One study (29) was in excised cell-free patches of membrane, and the other (28) was in intact cells (Xenopus oocytes). The results of these two studies showed important functional differences. Additional studies (4, 5, 14, 19, 22) have either been semiquantitative and/or have evaluated CFTR in which multiple serines have been mutated simultaneously. Therefore, the goal of this study was to quantitatively examine the function of CFTR-bearing individual phosphoserine mutations in an intact cell. This allowed us to evaluate the similarities and differences between data obtained in membrane patches and in Xenopus oocytes. We expressed the variants in Fischer rat thyroid (FRT) epithelia (23). These cells form high-resistance epithelia with little if any transepithelial electrogenic transport and no cAMP-stimulated current. Functional and immunocytochemical studies indicate that recombinant CFTR is expressed in both apical and basolateral membranes of this epithelia; thus cAMP agonists open CFTR Cl− channels in both membranes, allowing transepithelial Cl− flow. Importantly, this model allows us to quantify CFTR activity by measuring transepithelial current (23).
MATERIALS AND METHODS
We expressed CFTR in FRT epithelia (23, 30). These cells form polarized epithelia with high transepithelial resistance, and they do not express endogenous cAMP-stimulated Cl− channels. FRT cells were cultured in Coon's modified Ham's F-12 medium (GIBCO BRL, Life Technologies, Grand Island, NY) supplemented with 5% FCS, 100 U/ml of penicillin, and 100 μg/ml of streptomycin at 37°C in a humidified atmosphere of 5% CO2. Cells were seeded directly onto permeable filter supports (0.4-μm pore size, 27-mm diameter Millicell HA filters; Millipore, Bedford, MA) at a density of 7.5 × 105 cells/cm2. They were transfected 5 days after seeding when transepithelial electrical conductance was <1.7 mS/cm2.
CFTR mutants were constructed in the vaccinia virus expression plasmid pTM-CFTR4 by the method of Kunkel (12) and were cloned into the PMT-2 vector (25). Mutants containing multiple alterations were constructed by simultaneously including up to three mutagenic oligonucleotides in the reaction. Mutants containing more than three changes were generated with the use of analogous procedures with lower-order variants. Mutants were verified by restriction enzyme analysis, DNA sequencing around the sites of mutation, and in vitro transcription and translation assays. In the S-Oct-A mutant, eight serines were mutated to alanine at residues 660, 686, 700, 712, 737, 768, 795, and 813. In the S-Quad-A mutant, four serines were mutated to alanine at residues 660, 737, 795, and 813. In each of the serine to alanine mutants, S660A, S737A, S795A, and S813A, alanine replaced serine at the designated residue. In S-737/795/813-A, S-660/795/813-A, S-660/737/813-A, and S-660/737/795-A, three serines were mutated to alanine. In S-737/795-A, serines at positions 737 and 795 were changed to alanines.
We used cationic lipid-mediated transfection to transiently express wild-type and mutant CFTR in FRT cells. The cationic lipid DMRIE [N-(2-hydroxyethyl)-N,N-dimethyl-2,3-bis(tetradecyloxy)-1-propanaminum bromide] was a gift from Dr. Phil Felgner (Vical, San Diego, CA). The neutral lipid DOPE (1,2-dioleoyl-sn-glycero-3-phosphoethanolamine) was purchased from Avanti Polar Lipids (Alabaster, AL). DMRIE was mixed with DOPE at an equal molar ratio, and the individual CFTR mutants were then added to the lipid solution. Epithelia were transfected for 6 h and studied 4 days later.
Measurements of transepithelial Cl− current.
Transepithelial Cl− current was measured with methods similar to those previously described (1, 23). We mounted FRT epithelia in modified Ussing chambers (Jim's Instruments, Iowa City, IA). The apical surface of the epithelium was bathed in a solution containing (in mM) 135 sodium acetate, 1.2 CaCl2, 1.2 MgCl2, 2.4 K2HPO4, 0.6 KH2PO4, 10 HEPES, and 10 dextrose (titrated to pH 7.4 with NaOH). The composition of the submucosal solution was similar to the apical solution except that 135 mM NaCl replaced 135 mM sodium acetate. Thus there was a large Cl− concentration gradient (139.8 vs. 4.8 mM) across the epithelium. The solutions were maintained at 37°C and bubbled with 100% O2. Transepithelial voltage (referenced to the apical solution) was clamped at 0 mV with dual-voltage clamp 558C-5 (Dept. of Bioengineering, University of Iowa, Iowa City, IA). Current was recorded continuously. Under these conditions, current reflects the flow of Cl−in response to its concentration gradient. Flow of Cl−from the submucosal to the apical solution is shown as an upward current deflection. Because of the transepithelial Cl−concentration gradient, baseline current was not zero; it ranged from 3.9 to 10.3 μA/cm2. In Figures 1-6, we show the change in current induced by cAMP agonists. Solutions were changed by slowly flushing the apical and basolateral sides of the epithelium (chamber volume = 5 ml) with 30 ml of the solutions. The transepithelial conductance of transfected FRT epithelia at the time of study ranged from 0.75 to 1.77 mS/cm2, with an average of 1.28 mS/cm2. There was no difference between the conductance of control epithelia and epithelia transfected with the various mutants. Stimulation with 8-(4-chlorophenylthio)-cAMP (CPT-cAMP; 250 μM) reversibly increased transepithelial conductance from 1.17 ± 0.11 to 1.35 ± 0.12 mS/cm2; this was followed by a fall to 1.06 ± 0.13 mS/cm2 after CPT-cAMP was washed out (P < 0.0002; n= 7 experiments).
− current normalized to wild-type CFTR. Results are expressed as means ± SE withn = 4–24 experiments. To compare mean values, we used Student's unpaired t-test and ANOVA. Differences were considered significant at P < 0.05.
Forskolin, 3-isobutyl-1-methylxanthine (IBMX), and CPT-cAMP were obtained from Sigma. All other chemicals were of reagent grade.
cAMP agonists stimulate CFTR expressed in FRT epithelia.
We expressed wild-type CFTR in FRT epithelia and examined cAMP-dependent stimulation. Figure 1shows that increasing concentrations of CPT-cAMP stimulated increasing Cl− currents. The effect was reversible with the removal of CPT-cAMP. Subsequent addition of forskolin (10 μM) and IBMX (100 μM) stimulated a current greater than that obtained with CPT-cAMP. When either CPT-cAMP or higher concentrations of forskolin and IBMX were added, there was no further increase in current in the wild type or the various mutants (data not shown). We also found that the addition of forskolin alone could generate a maximal current (seecAMP-dependent stimulation of wild-type and mutant CFTR). Therefore, we examined the effects of a submaximal (250 μM CPT-cAMP) or a maximal (10 μM forskolin and 100 μM IBMX) cAMP stimulus.
cAMP-dependent stimulation of wild-type and mutant CFTR.
To evaluate the functional role of specific phosphoserines, we studied CFTR containing various serine to alanine mutations. Representative current tracings from wild-type and variants of CFTR stimulated with CPT-cAMP are shown in Fig. 2 A; Fig. 2 B shows data from several experiments. Current measured in epithelia expressing the S-Quad-A variant was low and was not significantly different from that in control epithelia transfected with vector containing no insert. This result suggests that the residues mutated in this construct account for most of the response to a submaximal stimulus. The S660A and S813A variants generated small but significant cAMP-stimulated Cl− currents. The S795A channels generated more current, but it was still less than that produced by wild-type CFTR. Interestingly, current from the S737A variant tended to be greater than that of wild-type CFTR, although the difference was not significant.
Figure 3 shows current from the same mutants after maximal stimulation with forskolin and IBMX. With the maximal stimulus, the S-Quad-A mutant generated a small but significant current, whereas current from the S-Oct-A mutant was not different from that of control. This result is consistent with earlier reports that additional serine residues (serines-686, -700, -712, -768, and -753) may be phosphorylated and contribute to channel activity under some conditions (17, 22). The S660A and S813A variants generated currents, but they were no greater than those obtained with S-Quad-A. These results suggest that Ser660 and Ser813 may be particularly important for phosphorylation-dependent stimulation of channel activity. Current from the S795A variant was not different from that in wild-type CFTR. Interestingly, current generated by the S737A variant was more than twice that generated by wild-type CFTR. This result suggests that phosphorylation of Ser737 may have inhibited current and that its removal may have allowed maximal current stimulation.
Effect of serine to alanine mutations on the sensitivity and time course of current activation by forskolin.
The data presented above suggest that in addition to altering the maximal amount of cAMP-stimulated current, mutation of serines in the R domain may have altered the sensitivity to cAMP agonists. To test this possibility, we stimulated epithelia with increasing concentrations of forskolin. Figure 4 A shows an example of the effect of increasing concentrations of forskolin applied to an epithelium expressing wild-type CFTR. Figure 4 B shows the results from several experiments. The S737A variant generated more current than the wild type, and the S-Quad-A mutant generated less. Ten micromolar forskolin produced a maximal current response in each of the variants, and higher concentrations tended to decrease current in the S-Quad-A mutant. In Fig. 4 C, current was normalized to the maximal amount obtained with each of the variants. The data show that the S-Quad-A variant was less sensitive to forskolin than the wild type and that the S737A variant was more sensitive than wild-type CFTR.
Differences in the sensitivity to forskolin suggested that activation rates of the variant channels might be different from wild-type CFTR. Figure 5 shows the time course of current change after the addition of forskolin (10 μM). The results show that current increased more rapidly with S737A than with wild-type and S-Quad-A CFTR.
Stimulation of CFTR variants containing multiple serine to alanine mutations.
Our data suggest that Ser660 and Ser813 may play particularly important stimulatory roles. Therefore, we predicted that constructs containing only one of these residues would generate more current than S-Quad-A. To test this, we examined the effect of variants in which one or the other of these residues was present but the three other serines were mutated (S-737/795/813-A and S-660/737/795-A; Fig. 6). In contrast to our prediction, neither variant generated more current than S-Quad-A. The variant that retained Ser795 intact, S-660/737/813-A, also failed to increase current over that obtained with S-Quad-A. These data suggest that Ser660 and Ser813 may be required for stimulation to wild-type current levels but that neither serine alone is sufficient to increase current.
Therefore, we generated a variant that contained both Ser660 and Ser813, S-737/795-A. Figure 6 shows that S-737/795-A produced as much current as wild-type CFTR. Although there was a tendency for current to be greater with this mutant than with the wild type, the difference was not significant (P = 0.2). These results suggest that there are interacting effects of Ser660 and Ser813 in generating cAMP agonist-stimulated current.
The increase in current and the more rapid activation observed when Ser737 was mutated to alanine suggested that phosphorylation of Ser737 may inhibit current. To test this notion, we produced constructs in which Ser660, Ser795, and Ser813 were all mutated to alanine and Ser737 was left intact (S-660/795/813-A). If phosphorylation of Ser737 inhibited current independent of the other three serines, then cAMP agonists might inhibit current in S-660/795/813-A. However, Fig. 6 shows that the variant in which only Ser737 remained intact produced cAMP-stimulated current that was not different from S-Quad-A or the other variants with only one of the four serines remaining. These results suggest that the inhibitory function of Ser737 depends on the presence of the other serines.
Earlier studies have shown that the R domain contains several serines that become phosphorylated when cellular levels of cAMP increase. Several studies (5, 6, 15-17, 22, 26) have identified serines that are phosphorylated by PKA in R domain proteins, in CFTR phosphorylated in vitro, and in CFTR phosphorylated in vivo. Several studies (4, 5, 14, 19, 22) have also examined the functional consequences of mutating R domain serines, but only two (28,29) have provided quantitative data about the effect of mutating individual serines. There were important differences in the results of those two studies. Our current study in cells helps explain the differences. In addition, by examining several new CFTR variants, the data provide new insights into the control of CFTR activity.
Our study has a number of limitations. First, mutations of the serines could change the level of protein expression. However, the data show that mutations alter the time course of current activation (Fig. 5), the sensitivity of these channels to increasing concentrations of forskolin (Fig. 4), and the amount of transepithelial current (Figs. 2and 3). The correspondence of these different measurements indicates that changes in transepithelial current are the result of functional effects rather than of differences in the amount of protein expressed. This conclusion is also consistent with earlier work showing that mutation of multiple phosphorylation sites does not alter CFTR expression (4) or its distribution to the cell surface (21). Second, it is possible that the mutations alter the proportion of CFTR that is targeted to the apical or basolateral membrane, thereby affecting current. However, we are not aware of any data that suggest this might occur. Third, we have not done phosphopeptide mapping to show which serines are phosphorylated in the various mutants. However, earlier studies (5, 6, 17) have identified serines that are phosphorylated in vivo. Thus their mutation will eliminate phosphorylation at that site. Nevertheless, it remains possible that mutation of a single serine could affect phosphorylation of other serines. In this case, phosphopeptide mapping as has been done for CFTR would not yield the quantitative results that would be required to address this issue.
Ser660 and Ser813 are important for cAMP-mediated stimulation of current.
Serine-660 and -813 appeared to play a key role in phosphorylation-dependent stimulation of activity. Mutation of either residue alone significantly decreased current; with maximal stimulation by cAMP agonists, neither the S660A nor the S813A mutant gave more current than the S-Quad-A variant. With submaximal cAMP agonists, S660A- and S813A-generated current was also reduced, but it was slightly greater than that obtained with S-Quad-A. The difference between maximal and submaximal stimulation of S-Quad-A may be because of the phosphorylation of other sites in S-Quad-A such as Ser700 and sites not normally phosphorylated in vivo such as Ser686, Ser712, Ser768, and Ser753 (17, 22). Moreover, neither Ser660 nor Ser813 alone was sufficient to increase current compared with S-Quad-A. We interpret these data to mean that Ser660 and Ser813 are both required for maximal channel activation in the response to phosphorylation. However, neither was sufficient on its own to generate wild-type levels of current. The effects of these two residues were not simply additive; when expressed together, they increased current up to wild-type levels.
Our conclusion that Ser660 and Ser813 play a critical role is consistent with single-channel studies (29) showing a significant decrease in open-state probability when these residues are mutated to alanine. Data from studies in Xenopus oocytes (28) also suggest that Ser813 and Ser660 are important stimulatory sites.
Ser795 may have a minor role in cAMP-mediated stimulation of current.
Mutation of Ser795 had only small effects on cAMP-stimulated current, suggesting that even though it is phosphorylated in vivo (5, 17, 26), the functional consequences may be minor. This result is different from studies in excised patches of membrane showing that S795A reduced open state probability (29). Unlike our results, Wilkinson et al. (28) found that in Xenopus oocytes Ser795 had a stimulatory role similar to that of Ser660. It is possible that these differences may in part be caused by the different expression systems or levels of stimulation.
Phosphorylation of Ser737 may inhibit current.
Ser737 is phosphorylated by cAMP agonists in vivo (5,17). When we expressed CFTR in which Ser737 was mutated, the channels were more sensitive to low concentrations of cAMP agonists, i.e., the dose-response curve was shifted to the left. At maximal cAMP concentrations, there was more Cl− current, and current increased more quickly on the addition of cAMP agonists. These results suggest that phosphorylation of Ser737 may inhibit current, and, consequently, when it is mutated, the functional response to phosphorylation of other serines is greater. Interestingly, the study by Wilkinson et al. (28) also found increased sensitivity to activation when Ser737 was eliminated. Taken together, these data suggest an inhibitory role for phosphorylated Ser737, at least in vivo.
Potential interactions between phosphorylation sites.
Our data, combined with earlier work, emphasize the complexity of the regulatory mechanisms and highlight some of the questions that remain. An example is the importance of Ser660 and Ser813 for stimulation; mutation of either residue alone markedly reduced cAMP-stimulated current. Yet neither serine was sufficient to increase the current response to maximal cAMP agonists; both Ser660 and Ser813 were required in the S-737/795-A variant for current to reach wild-type levels. These results suggest that there may be significant interaction between the two phosphorylated residues. Alternatively, there may be interaction within other domains of CFTR that depend on phosphorylation of both sites.
Another example is the inhibitory role of Ser737. Our data and those of Wilkinson et al. (28) suggest that phosphorylation of Ser737 may inhibit current. Yet this effect only appears when the other serines are present; when serine-660, -795, and -813 were mutated, cAMP-stimulated current was the same, irrespective of whether Ser737 was present. These results also suggest the possibility of interactions between the various phosphorylation sites.
Differences between results obtained in different model systems.
In Potential interactions between phosphorylation sites, we noted some differences in results obtained with different model systems. A particularly striking example is the different function of S737A in cells and in excised membrane patches. In FRT cells and inXenopus oocytes (28), Ser737 generated more current than wild-type CFTR, but, in patches, S737A showed the same open state probability as the wild type, and the sensitivity to ATP was reduced (29). The reason for this and other model-dependent differences is unknown, but we can speculate about possible explanations. First, patches of membrane and various cell types may contain different types or amounts of the kinases and phosphatases that control phosphorylation (2,7, 10, 13, 18, 27). Second, various model systems may express different complements of linker and adapter molecules (such as A-kinase anchoring proteins) that place the regulatory enzymes in proximity to CFTR (9, 11) or link CFTR to other regulatory molecules. Third, it is possible that when studied in cells, cAMP agonists may have additional effects that may not be observed in excised patches. An example is the effect of cAMP agonists on endocytosis and exocytosis (3). These or other factors may explain the reasons for the differences in CFTR behavior in different experimental systems. The same considerations may also contribute to some of the observed differences between various mutants when studied in a single model system such as FRT epithelia or Xenopus oocytes.
It is difficult to understand the reason for the complexity in phosphorylation-dependent regulation. It could be that the complexity allows for precise control of the channel activity of a cell under a variety of activation and metabolic states. It could be that the complexity allows for different mechanisms of regulation in tissues that have distinct physiological functions. Or it could be that our limited knowledge of CFTR structure and function only makes regulation by phosphorylation “appear” complex.
We thank Anthony Thompson, Lisa DeBerg, Philip Karp, Pary Weber, Tatiana Rokhlina, Theresa Mayhew, and our other colleagues for advice and assistance. We thank the University of Iowa Diabetes and Endocrinology Research Center DNA Core Facility (which is supported by National Institute of Diabetes and Digestive and Kidney Diseases Grant DK-25295) for oligonucleotide synthesis and sequencing.
This work was supported by the National Heart, Lung, and Blood Institute and the Howard Hughes Medical Institute.
M. J. Welsh is an Investigator of the Howard Hughes Medical Institute.
Address for reprint requests and other correspondence: M. J. Welsh, Howard Hughes Medical Institute, Univ. of Iowa College of Medicine, 500 EMRB, Iowa City, IA 52242 (E-mail:).
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2000 the American Physiological Society