Protein kinase A phosphorylation potentiates cystic fibrosis transmembrane conductance regulator gating by relieving autoinhibition on the stimulatory C terminus of the regulatory domain

Cystic fibrosis transmembrane conductance regulator (CFTR) is a chloride channel activated by protein kinase A (PKA) phosphorylation on the regulatory (R) domain. Phosphorylation at several R domain residues stimulates ATP-dependent channel openings and closings, termed channel gating. To explore the protein segment responsible for channel potentiation and PKA-dependent activation, deletion mutations were constructed by removing one to three protein segments of the R domain including residues 708–759 (ΔR708–759), R760–783, and R784–835, each of which contains one or two PKA phosphorylation sites. Deletion of R708–759 or R760–783 had little effect on CFTR gating, whereas all mutations lacking R784–835 reduced CFTR activity by decreasing the mean burst duration and increasing the interburst interval (IBI). The data suggest that R784–835 plays a major role in stimulating CFTR gating. For ATP-associated regulation, ΔR784–835 had minor impact on gating potentiation by 2′dATP, CaATP, and pyrophosphate. Interestingly, introducing a phosphorylated peptide matching R809–835 shortened the IBI of ΔR708–835-CFTR. Consistently, ΔR815–835, but not ΔR784–814, enhanced IBI, whereas both reduced mean burst duration. These data suggest that the entirety of R784–835 is required for stabilizing the open state of CFTR; however, R815–835, through interactions with the channel, is dominant for enhancing the opening rate. Of note, PKA markedly decreased the IBI of ΔR708–783-CFTR. Conversely, the IBI of ΔR708–814–CFTR was short and PKA-independent. These data reveal that for stimulating CFTR gating, PKA phosphorylation may relieve R784–814–mediated autoinhibition that prevents IBI shortening by R815–835. This mechanism may elucidate how the R domain potentiates channel gating and may unveil CFTR stimulation by other protein kinases.


Cystic fibrosis transmembrane conductance regulator (CFTR) is a chloride channel activated by protein kinase A (PKA) phosphorylation on the regulatory (R) domain. Phosphorylation at several R domain residues stimulates ATP-dependent channel openings and
closings, termed channel gating. To explore the protein segment responsible for channel potentiation and PKA-dependent activation, deletion mutations were constructed by removing one to three protein segments of the R domain including residues 708 -759 (⌬R 708 -759 ), R 760 -783 , and R 784 -835 , each of which contains one or two PKA phosphorylation sites. Deletion of R 708 -759 or R 760 -783 had little effect on CFTR gating, whereas all mutations lacking R 784 -835 reduced CFTR activity by decreasing the mean burst duration and increasing the interburst interval (IBI). The data suggest that R 784 -835 plays a major role in stimulating CFTR gating. For ATP-associated regulation, ⌬R 784 -835 had minor impact on gating potentiation by 2dATP, CaATP, and pyrophosphate. Interestingly, introducing a phosphorylated peptide matching R 809 -835 shortened the IBI of ⌬R 708 -835 -CFTR. Consistently, ⌬R 815-835 , but not ⌬R 784 -814 , enhanced IBI, whereas both reduced mean burst duration. These data suggest that the entirety of R 784 -835 is required for stabilizing the open state of CFTR; however, R 815-835 , through interactions with the channel, is dominant for enhancing the opening rate. Of note, PKA markedly decreased the IBI of ⌬R 708 -783 -CFTR. Conversely, the IBI of ⌬R 708 -814 -CFTR was short and PKA-independent. These data reveal that for stimulating CFTR gating, PKA phosphorylation may relieve R 784 -814 -mediated autoinhibition that prevents IBI shortening by R 815-835 . This mechanism may elucidate how the R domain potentiates channel gating and may unveil CFTR stimulation by other protein kinases.
Cystic fibrosis is a genetic disease caused by dysfunction of the cystic fibrosis transmembrane conductance regulator (CFTR) 2 (1). CFTR is an epithelial Cl Ϫ channel composed of two membrane-spanning domains (MSDs), two nucleotidebinding domains (NBDs), and a regulatory (R) domain (1,2). It is well-known that cAMP-dependent PKA phosphorylation on the R domain activates the CFTR Cl Ϫ channel (3). Then cycles of ATP binding and hydrolysis at one or two ATPase sites in the interface of two NBDs control CFTR openings and closings, termed channel gating (2). Gating motions are initiated by ATP-induced NBD dimerization, subsequently leading to structural rearrangements within MSDs (4 -7). Finally, dephosphorylation of CFTR by phosphatases ceases channel activity (3). 10 PKA phosphorylation sites are found in the R domain (1), but how they control CFTR channel gating remains unclear.
The R domain is mostly unstructured (8) and may include residues from 634 to 835 (R 634 -835 ) (3). However, the ⌬RS660A mutation that deletes a large part of the R domain (⌬R 708 -835 ) generates substantial basal activity so that this mutation greatly attenuates cAMP-stimulated CFTR current (9) and PKA-dependent CFTR activation (10). These findings suggest that R 708 -835 may contain inhibitory protein segments that directly prohibit CFTR activation. Interestingly, the ⌬RS660A mutation also greatly reduces the open probability (P o ) of CFTR (11)(12)(13), whereas the phosphorylated peptide that matches R  in the R domain evidently enhances the Cl Ϫ current and P o of ⌬RS660A-CFTR because of an increase in the channel opening rate (11). These data suggest that the phosphorylated R domain is able to stimulate the channel activity of CFTR. In addition, previous work found that R 817-838 containing no PKA phosphorylation site is also required for stimulating CFTR gating (14).
PKA phosphorylation sites Ser 768 , Ser 795 , and Ser 813 in the R domain seemingly are important for regulating CFTR gating (2). Evidence shows that the S768A mutation enhances the stimulation of CFTR activity by isobutylmethylxanthine (15) and PKA (16). S768A also increases the P o of CFTR (16 -18). By contrast, both S795A and S813A mutations greatly decrease P o (11,17) by reducing the channel opening rate (11). However, how phosphorylation at these serine residues regulates CFTR gating remains unclear. A major obstacle for addressing this question is huge conformational changes of CFTR after PKA phosphorylation (6,19). By sequentially adding phosphate groups to serine or threonine residues (10,16,20), PKA phosphorylation enhances NBD dimerization (21) and alters interaction patterns of the R domain with NBDs (6,22,23), intracellular loop 3 (18,24), front and back halves of Split⌬R-CFTR (25), and different intracellular molecules such as the SLC26A3 sulfate transporter and antisigma factor antagonist domain (22). These data suggest that multiple CFTR domains may mediate phosphorylationinduced gating stimulation.
Because phosphorylation alters CFTR conformation (6,19) and the interaction sites of the R domain (6,7,22,23), an R domain segment may regulate CFTR activity (e.g. inhibition or stimulation) differently before and after PKA phosphorylation. A caveat is that previous studies using site-directed mutations (10, 11, 16 -18, 24) for prohibiting PKA phosphorylation at one or several serine residues may also disturb phosphorylationinduced peptide movements and interactions. In other words, these site-directed mutations may eliminate or alter the postphosphorylation function of target residues, but they may also retain or modify their prephosphorylation actions. To avoid this confounding factor, deletion mutations were used in this study for removing all pre-and postphosphorylation function of residues.
To investigate the segmental function of the R domain in regulation of CFTR gating, this study first searched which protein segment plays a major role in regulating CFTR gating. Then the study further explored whether the R domainmediated stimulation is caused by ATP-associated gating regulation and how PKA phosphorylation elicits the gating stimulation.

Segmental function of the R domain in ATP-dependent CFTR gating
Similar to previous work (26), the R domain was divided into three protein segments R 1 (R 708 -759 , residues from positions 708 to 759), R 2 (R 760 -783 ), and R 3 (R 784 -835 ) (Fig. 1A). All CFTR mutants in this study included a S660A mutation for data comparison (Fig. 1A).
To test segmental function of the R domain in channel gating, the single-channel activity of CFTR mutants that deleted one to three protein segments of R 1 , R 2 , and R 3 was carefully examined (Fig. 1B). All deletion mutations were without effect on the single-channel current amplitude (i) of the CFTR Cl Ϫ channel (Fig. 1, B and C). The deletion of R 1 alone (⌬R 1 ) showed little or no effect on the open probability (P o ) (Fig. 1D) and mean burst duration (MBD) of CFTR ( Fig. 1E) but appeared to decrease the interburst interval (IBI) (Fig. 1F, p ϭ 0.07, one-way ANOVA). In addition, ⌬R 2 did not alter P o , MBD, and IBI ( Fig.   1, D-F). By contrast, ⌬R 12 that removes R 1 and R 2 together mildly reduced P o (Fig. 1D) because of a large decrease in MBD ( Fig. 1E) with no change in IBI (Fig. 1F). However, the P o of those CFTR mutants with R 3 deleted, including ⌬R 3 -, ⌬R 23 -, and ⌬R 123 -CFTR, were greatly decreased (Fig. 1D, gray columns) by large decreases in MBD (Fig. 1E) and marked increases in IBI (Fig. 1F). Notably, ⌬R 123 -CFTR here is the same as ⌬RS660A-CFTR tested in early studies (9 -13), and the data ( Fig. 1, C-F) are consistent as previously reported (11,13). These data suggest that R 3 is required for stimulating CFTR gating, whereas R 1 and R 2 might contribute to MBD regulation. It is also possible that a large deletion of R 1 and R 2 together in ⌬R 12 -CFTR may disturb the function of R 3 in the MBD prolongation.
ATP-dependent CFTR gating may consist of three primary gating motions: ATP-mediated NBD dimerization, NBD/MSD coupling, and transmembrane gate movements (2,27). Therefore, the R domain by interacting with the intracellular side of CFTR may modulate channel gating via two major pathways: 1) it may alter properties of ATPase sites in the interface of two NBDs (11,21,28), and 2) it may guide NBD/MSD coupling for regulating channel gating. To test the first mechanism, three gating stimulators that may enhance ATP-associated gating regulation were used (Fig. 2), including 2ЈdATP (1 mM) as an ATP analogue (29), CaATP (1 mM) for preventing ATP hydrolysis (30), and PP i (2 mM ϩ 1 mM ATP) that may lock the posthydrolytic state of CFTR gating (31). If R 3 regulates CFTR gating through modifying ATP function in NBDs, one would expect to find that potentiation effects of stimulators on channel gating may be greatly altered in those mutants with R 3 deleted.
However, the data demonstrate that three gating stimulators, 2ЈdATP, CaATP, and PP i , largely increased the macroscopic currents ( Fig. 2, A and B), P o (Fig. 2D), and i ϫ P o values (Fig. 2E) of all tested CFTR constructs at similar levels of ϳ2-3-fold enhancements (Fig. 2, B and E). PP i also slightly decreased the i of WT and all mutant CFTRs (Fig. 2C), similar to the previous finding (32). These data suggest that ⌬R 3 may only have minor impacts on ATP-associated gating regulation.
To further explore this finding, the effects of three stimulators on the single-channel gating kinetics of WT and ⌬R 123 -CFTR were examined (Fig. 3). Only membrane patches that contained a single active CFTR channel were used in this study. For WT CFTR, channel activity and P o were greatly enhanced by three stimulators (Fig. 3, A and B) because of striking increases in MBD by all stimulators (Fig. 3C) and marked decreases in IBI by 2ЈdATP and CaATP (Fig. 3D). For ⌬R 123 -CFTR, channel activity and P o were also largely increased by three stimulators (Fig. 3, A and E) because of the MBD prolongation by CaATP or PP i (Fig. 3F) and the IBI shortening by 2ЈATP and possibly two other stimulators: CaATP (p ϭ 0.10) and PP i (p ϭ 0.06) (Fig. 3G).
Between WT and ⌬R 123 -CFTR, three major alterations in the potentiation of gating kinetics by three stimulators were observed. First, MBD was prolonged by PP i ϳ7.5-fold in WT CFTR (Fig. 3C), but only 2.1-fold in ⌬R 123 -CFTR (Fig. 3F) (p Ͻ 0.05, one-way ANOVA). Second, IBI with CaATP was reduced to ϳ26% of that with MgATP in WT CFTR (Fig. 3D), but to Stimulation of CFTR gating by the R domain ϳ70% in ⌬R 123 -CFTR (Fig. 3G) (p Ͻ 0.05, one-way ANOVA). Third, it is of interest that MBD was greatly prolonged by 2ЈdATP ϳ2.7-fold in WT CFTR (Fig. 3C), but without a change in ⌬R 123 -CFTR (Fig. 3F). These data suggest that ⌬R 123 mildly modulated the stimulation of CaATP and PP i on CFTR gating kinetics. The data also suggest that the R domain plays an important role in the MBD prolongation by 2ЈdATP.
To explore the protein segment responsible for 2ЈdATP-mediated MBD prolongation, the single-channel kinetics of other deletion mutants were examined in the presence of 2ЈdATP (Fig. 4). The data demonstrate that 2ЈdATP stimulated the channel activity (Fig. 4A) and P o (Fig. 4B) of all deletion mutants. It also increased the MBD of WT, ⌬R 1 -CFTR, and ⌬R 3 -CFTR to ϳ2.4-, 2.2-, and 1.8-fold, respectively (Fig. 4C). However, 2ЈdATP did not alter the MBD of those deletion mutants with R 2 removed, including ⌬R 2 -, ⌬R 23 -, and ⌬R 123 -CFTR (Fig. 4C). These data suggest that R 2 is required for the MBD prolongation by 2ЈdATP.

Stimulation of CFTR gating by the R domain
The data suggest that all tested R domain deletions had minor effects on the IBI shortening by 2ЈdATP. Without a significant contribution to CFTR gating potentiation by three stimulators (Figs. 2-4), R 3 might stimulate CFTR gating by modulating coupling of NBDs and MSDs.
To support this hypothesis, R 3 may provide essential peptide interactions for potentiating CFTR gating, as shown in previous work (11,14). Therefore, five synthesized peptides were tested in the following experiment, each matching a short segment of the R domain (Fig. 5A).
However, peptide P 3-2 significantly enhanced the I (Fig. 5C) and single-channel activity (Fig. 5D) of ⌬R 123 -CFTR. Although P 3-2 did not alter the i and MBD of ⌬R 123 -CFTR (Fig. 5, E and  G), it evidently enhanced P o (Fig. 5F) by reducing IBI (Fig. 5H). The data suggest that R 809 -835 , covered by P 3-2 (Fig. 5A), may generate crucial interactions for the IBI shortening or, in other words, for accelerating the channel opening rate of CFTR, consistent with previous findings (11,14). To search specific small segments contributing to the IBI shortening by P 3-2 (Fig. 5H) and gating regulation by R 3 (Figs. 1-4), CFTR mutants that remove the whole, one-half, or one-seventh of R 3 were examined (Fig. 6, A and B).

R 815-835 in R 3 dominantly stimulates the IBI shortening
In the presence of ATP (1 mM) and PKA, all deletion mutations did not alter i (Fig. 6B and data not shown). By contrast, the P o (Fig. 6C) and MBD (Fig. 6D) of these mutants were all lower than that of WT CFTR. The data suggest that the entirety of R 3 regulates the MBD of CFTR.
For the IBI regulation by R 3 (Fig. 1F), ⌬R 3 and ⌬R 815-835 , but not ⌬R 784 -814 , apparently increased the IBI of CFTR (Fig. 6E). These data and the data of P 3-2 (Fig. 5H) suggest that in R 3 , R 815-835 plays a major role in stimulating the IBI shortening. By using mutations that remove both R 1 and R 2 (⌬R 12 ), further investigation demonstrates that the IBI of CFTR was not changed by ⌬R 12 (Fig. 1F) but was slightly increased by ⌬R 708 -814 that deletes R 784 -814 from ⌬R 12 (Fig. 6E) and markedly prolonged by ⌬R 123 that deletes both R 784 -814 and R 815-835 (Fig. 6E). Moreover, the IBI of ⌬R 123 -CFTR was much longer than that of

Stimulation of CFTR gating by the R domain
⌬R 708 -814 -CFTR (Fig. 6E, p Ͻ 0.05, one-way ANOVA). These data suggest that R 815-835 -mediated IBI shortening is dominant for stimulating CFTR gating but may need the assistance of R 784 -814 for achieving the full effect of R 3 .
Two PKA phosphorylation sites Ser 795 and Ser 813 are present in R 784 -814 but none in R 815-835 (Fig. 6A). In addition, sitedirected mutations S795A and S813A that abolish PKA phosphorylation at these two serine residues both greatly increase IBI (11). These data suggest that PKA phosphorylation of R 784 -814 may control R 815-835 -mediated IBI shortening.

Phosphorylation of R 784 -814 relieves strong autoinhibition of IBI shortening
After a patch of the cell membrane was excised from a HeLa cell, CFTR Cl Ϫ currents across the patch membrane were recorded at 1 mM ATP for 3 min initially (Fig. 7B). After PKA was added directly into the chamber, the currents were continually recorded for 6 min or more (Fig. 7B). The number of active channels (N) was increased by PKA in three of seven membrane patches for ⌬R 12 -CFTR (Fig. 7C) and

Stimulation of CFTR gating by the R domain
one of seven membrane patches for ⌬R 708 -814 -CFTR (Fig.  7C). Because N may be underestimated at 1 mM ATP because of short recording time, these data suggest that the majority of both mutants in membrane patches were already partially active before PKA was added, especially for ⌬R 708 -814 -CFTR.
For each experiment, N in Fig. 7C was used to calculate the P o of two mutants in the absence and presence of PKA. The data demonstrate that PKA greatly stimulated the channel activity (Fig. 7B) and P o (Fig. 7D) of ⌬R 12 -CFTR by slightly increasing MBD (Fig. 7E) but markedly decreasing IBI (Fig. 7F). Note that the effects of PKA on the P o and IBI of ⌬R 12 -CFTR (Fig.  7, D and F) might be less marked because of possible underestimation of N at 1 mM ATP for three patches (Fig. 7C). The data suggest that ⌬R 12 -CFTR retains the PKA-dependent activation mechanism.

Stimulation of CFTR gating by the R domain
Importantly, the P o of ⌬R 12 -CFTR was lower than that of ⌬R 708 -814 -CFTR at 1 mM ATP but became higher after adding PKA (see # in Fig. 7D). Two alterations in gating kinetics were observed. First, the MBD of ⌬R 12 -CFTR was slightly longer than that of ⌬R 708 -814 -CFTR after adding PKA (Fig. 7E). Second, the dominant change is that after PKA was added, the IBI of ⌬R 12 -CFTR was altered from strikingly longer to mildly shorter than that of ⌬R 708 -814 -CFTR (Fig. 7F). These data suggest that unphosphorylated R 784 -814 may strongly inhibit the IBI shortening by R 815-835 . Moreover, the data also suggest that phosphorylated R 784 -814 may not only lose its inhibition on R 815-835 but may also simultaneously stimulate CFTR gating by slightly increasing MBD and decreasing IBI.

Discussion
Among three protein segments of the R domain, R 1 and R 2 play a minor role in regulation of CFTR gating. It is of great interest that R 2 is required for the MBD prolongation by 2ЈdATP. By contrast, R 3 predominantly stimulates CFTR gating by greatly increasing MBD and markedly decreasing IBI. Moreover, the data suggest that R 3 contributes little to gating potentiation by three stimulators: CaATP, PP i , and 2ЈdATP, but generates peptide interactions to decrease the IBI of CFTR. Notably, in R 3 , unphosphorylated R 784 -814 may inhibit the IBI shortening mediated by R 815-835 . PKA phosphorylation of R 784 -814 may relieve this autoinhibition

Stimulation of CFTR gating by the R domain
and further mildly stimulate CFTR gating. These findings reveal an activation mechanism by which PKA phosphorylation at Ser 795 and Ser 813 in R 784 -814 elicits a great stimulation of CFTR gating by permitting R 815-835 -mediated IBI shortening.

Segmental function of the R domain in CFTR gating
By site-directed mutations (10, 11, 16 -18, 33, 34) or deletions (11) of the R domain, previous work and this study demonstrate that PKA phosphorylation of the R domain modulates CFTR gating. By contrast, the previous study (35) that deletes the R domain by co-expression of the N-and C-terminal half of CFTR, including FLAG-cut-⌬R (N-terminal R FLAG3-633 ϩ C-terminal R 837-1480 ) and cut-⌬R-CFTR (R 1-633 ϩ R 837-1480 ) in Xenopus oocytes, demonstrates that the P o of these split CFTRs are only moderately reduced. When expressed in Chinese hamster ovary cells, cut-⌬R-CFTR exhibits phosphorylation-independent P o with channel activity and gating kinetics similar to that of WT CFTR (36). Therefore, studies on split CFTR (35,36) may suggest that the R domain is not important for regulating CFTR gating.
The discrepancies among these studies may result from confounding variations across different CFTR constructs. For this

Stimulation of CFTR gating by the R domain
study, one may argue that the peptide chain deletion in the R domain could generate structural constraints that significantly disrupt CFTR activity. However, a deletion of 52 residues in ⌬R 1 -CFTR or 24 residues in ⌬R 2 -CFTR showed little effect on CFTR gating (Fig. 1). ⌬R 12 -CFTR with a deletion of 76 residues only reduced MBD but without effect on IBI (Fig. 1, E and F). Conversely, ⌬R 3 , ⌬R 815-835 , and ⌬R 829 -835 with deletions of 52, 21, and 7 residues, respectively, all significantly altered MBD and IBI (Fig. 6, D and E). Moreover, among all synthesized peptides (Fig. 5C), only peptide P 3-2 matching R 809 -835 was able to stimulate the channel activity of ⌬R 123 -CFTR. These data suggest that disruption of CFTR gating by ⌬R 3 is mainly caused by removing function of specific residues, rather than the peptide deletion or deletion-induced structural constrains. Thus, it is speculated that two half-split polypeptides of cut-⌬R-CFTR might allow specific structural arrangements, possibly around the interaction region of R 3 , overcoming the loss of R 3 -mediated gating stimulation.
This study demonstrates that R 2 plays a minor role in MgATP-mediated CFTR gating (Fig. 1). However, PKA phosphorylation at Ser 768 in this segment is thought to be important for inhibiting CFTR activity. By abolishing phosphorylation at Ser 768 , the site-directed mutation S768A enhances the stimulation of CFTR activity by isobutylmethylxanthine (15) and PKA (16). S768A also increases the P o of CFTR (16 -18) possibly by increasing MBD (16 -18). The reason for this discrepancy is unclear. It is noted that Ser 768 and its neighboring residues may interact with NBD1 (23) or intracellular loop 3 (18), whereas phosphorylation also alters intra-and interdomain interactions of the R domain, including the region around Ser 768 (6,22,23). Because the R 2 function is seemingly associated with the MBD regulation (Figs. 1E and 4C), one possible mechanism is that without phosphorylation at Ser 768 , R 2 in S768A-CFTR may interact with the site differently from that of WT CFTR, permitting stimulation on channel gating. Future work is required to test this possibility.
Finally, the mutation S737A that abolishes the major PKA phosphorylation in R 1 appears to reduce the closed time of CFTR (17), consistent with the data of ⌬R 1 -CFTR (Fig. 1F). Overall, this study indicates that R 1 is not important for regulating CFTR gating because ⌬R 1 appeared to have only minor impact on MBD and IBI (Fig. 1, E and F).

Allosteric regulation of R 3 on MgATP-dependent channel gating
Previous studies demonstrate that the phosphorylated R domain enhances the NBD dimerization (21) and ATPase activity of CFTR (5,28), whereas ATP binding to CFTR is phosphorylation-independent (37). Enhanced NBD dimerization and ATPase activity by PKA phosphorylation might be due to relieving the R domain inhibition that blocks gating motions (5).
The constitutive channel activity of ⌬R 12 -CFTR at 1 mM ATP alone (Fig. 7, B-F) is consistent with the previous finding that R 2 predominantly blocks CFTR activation (26). The remaining large PKA-dependent channel activity of ⌬R 2 -CFTR (26) is likely achieved by R 3 -mediated gating stimulation (Fig. 7). It is of interest that ⌬R 3 greatly disrupted CFTR gating (Figs. 1 and   4) but showed little or no effect on the gating stimulation of 2ЈdATP (Fig. 4, C and D). The data suggest that R 3 may potentiate channel gating by the mechanism different from that of 2ЈdATP.
Because the hydrolysis rate of 2ЈdATP (V max ϭ 8.5 nmol/ mg⅐min) is slower than that of ATP (V max ϭ 47 nmol/mg⅐min) (29), it is suggested that 2ЈdATP increases MBD by slowing the hydrolysis rate. However, without R 2 , 2ЈdATP-mediated MBD was similar to that by ATP (Fig. 4C), and ⌬R 2 did not significantly alter ATP-mediated MBD (Fig. 1E). These data suggest that in R 2 -deleted mutants, the ATP hydrolysis rate might not be strictly coupling to the channel closing rate.
How the ATPase activity of NBDs regulates CFTR gating remains unclear (2,38). The major argument focuses on whether there is a strict correlation between ATP hydrolysis and channel closing (2,38). An unmatched correlation was observed in this study because ATP-and 2ЈdATP-mediated MBDs of R 2 -deleted CFTR mutants were similar (Fig. 4C). Our recent study (13) suggests that the Michaelis-Menten relationship of the ATP concentration and P o of CFTR, conventionally used for simulating ATP dependence of channel gating, may just denote the transitions of gating states. By contrast, channel opening is likely triggered by ATP (2,38), whereas the ratelimiting step of channel opening likely occurs after ATP binding to NBDs, e.g. the NBD dimerization (2,38).
Here, an interesting finding is that 2ЈdATP increased the channel opening rate ϳ2.5-fold in WT and other mutant CFTRs (Fig. 4D), despite the IBI of ⌬R 3 -, ⌬R 23 -, and ⌬R 123 -CFTR being greatly increased to ϳ5-fold longer than that of WT CFTR (Fig. 4D). Biochemical evidence suggests that 2ЈdATP may have higher binding affinity (K m ϭ 0.2 mM) to CFTR than that of ATP (K m ϭ 1.1 mM) (29). However, 2ЈdATP simply removes the 2Ј-hydroxl (-OH) group in the ribose of ATP. Most electrostatic interactions between ATP and NBDs come from phosphate groups of ATP (6). Therefore, it seems unlikely that 2ЈdATP can generate the binding force 2.5-fold higher than that by ATP in the NBD dimer.
A plausible explanation for above findings is that 2Ј-OH group in the ribose of ATP may generate steric hindrance for the NBD dimerization. Thus, for shortening IBI, 2ЈdATP might allow NBDs to dimerize faster than that by ATP. Consequently, the NBD dimer induced by 2ЈdATP might be formed differently from that by ATP, permitting R 2 to stabilize the dimer and prolong MBD.
In contrast to 2ЈdATP, R 3 may stimulate CFTR activity by modulating coupling of NBDs and MSDs (2,27), based on the data in this study. This mechanism is consistent with recent findings that in phosphorylated CFTR, the C-terminal half of R 3 (R 825-835 ) may interact with the region between NBD and MSD, away from ATPase sites in CFTR (7). Following this idea, R 3 may serve as a platform providing essential peptide interactions that reduce the energy barrier required for channel openings and also lower the energy level of the bursting state for prolonging MBD (39).
Moreover, this mechanism suggests that ⌬R 3 may disturb NBD/MSD coupling, meant slowing and destabilizing the gating motion that contains the tight and synchronous coordination between NBD dimerization and NBD/MSD coupling (27).

Stimulation of CFTR gating by the R domain
Consequently, 2ЈdATP may still slowly accelerate the NBD dimerization at a speed 2.5-fold higher than that by ATP for shortening the IBI of those CFTR variants with R 3 removed (Fig. 4D). This mechanism might also allow R 3 to stimulate CFTR gating by allosterically regulating NBD dimerization (21) and ATP dependence of channel gating (11). A caveat for this proposed mechanism is that CFTR structural changes accompanying with 2ЈdATP-induced NBD dimerization remain unclear.

Autoinhibition and self-stimulation of CFTR gating by R 3
Phosphorylation is a key protein modification that activates various signaling molecules in a cell (40). However, the phosphorylation-induced activation process for most of these molecules is not well-understood. Phosphorylation may cause charge-charge interactions and regional conformational changes, which may further modulate proteinprotein interactions (40). For example, PKA phosphorylation on the protein kinase A-anchoring protein smAKAP may alter the helicity of the domain, subsequently disrupting interactions between smAKAP and the type I regulatory subunit of PKA (41).
Similarly, PKA phosphorylation greatly alters the conformation of the R domain (6,19). Therefore, the autoinhibition of the IBI shortening by R 784 -814 (Fig. 7F) may be relieved by phosphorylation-induced structural changes (Fig. 8). Consistently, mutations S795A and S813A that prevent phosphorylation on R 784 -814 both markedly prolong the IBI of CFTR (11). Moreover, providing R 815-835 -mediated peptide interactions by the peptide P 3-2 (R 809 -835 ) in this study (Fig. 5H) and NEG2 (R 817-838 ) in previous work (14) all stimulate the IBI shortening. Furthermore, the N-ethylmaleimide modification at the Cys 832 residue also reduces IBI (42). Recent structural evidence reveals that in phosphorylated CFTR, R 825-843 may be relocated from the region wedged by transmembrane helices 9, 10, and 12 (5,7) to the position interacting with helices 10 and 11 and residues 34 -39 of the lasso motif (7). Therefore, unphosphorylated R 784 -814 might disrupt interactions of R 815-835 with the above helices 10 and 11 and the lasso motif for blocking the IBI shortening.
It is unclear whether unphosphorylated R 784 -814 directly interacts with R 815-835 or binds to the interaction sites of R 815-835 to inhibit IBI shortening (Fig. 8). A direct interaction between R 784 -814 and R 815-835 is feasible because R 815-835 contains many negatively charged residues, and PKA phosphorylation sites in R 784 -814 comprise positively charged dibasic residues, including lysine and arginine residues (1). Then added phosphate groups at Ser 795 and Ser 813 after PKA phosphorylation may produce negative-to-negative repulsive forces that prevent the inhibition of R 815-835 by R 784 -814 .
It is intriguing that deletions of R 12 , R 3 (Fig. 1E), or a part of R 3 (Fig. 6D) all decreased MBD to a similar value, suggesting that the R domain may adopt a common mechanism to regulate MBD. It could be a short protein segment in R 3 essential for prolonging MBD, whereas any small deletions in R 3 or the large deletion like ⌬R 12 may be sufficient to disrupt the MBD regulation by this responsible segment.
This study addressed a major CFTR activation mechanism, namely how PKA phosphorylation of the R domain stimulates channel gating. The data suggest that PKA phosphorylation on R 784 -814 relieves an autoinhibition that blocks the IBI shortening by R 815-835 (Fig. 8). Phosphorylated R 3 may induce the protein segment relocation (7) in which R 815-835 may provide essential peptide interactions for potentiating CFTR gating (Fig. 8). This stimulatory mechanism advances our understandings of PKA-dependent CFTR activation. Similar mechanisms may be adopted by other protein kinases, such as PKC (43), cGMP-dependent protein kinase II (44) and tyrosine kinases (45) for regulating CFTR gating.

Mutagenesis and CFTR expression
Human CFTR mutants were constructed using the pTM1-CFTR4/S660A plasmid (26), in which the T7 promotor is recognized by the bacteriophage T7 RNA polymerase. Some CFTR mutants were made in the previous study (26), and others were created by the QuikChange II XL site-directed mutagenesis kit (Agilent Technologies, Santa Clara, CA) and verified by DNA sequencing.
WT and all mutant CFTRs were expressed in HeLa cells by the vaccinia virus/bacteriophage T7 hybrid expression system as described previously (10). In brief, HeLa cells were grown in Dulbecco's modified Eagle's medium (Gibco, Thermo Fisher Scientific) with 10% FBS at 37°C and 5% CO 2 . For transiently expressing CFTR in the cell membrane, HeLa cells were infected for 1 h with recombinant vaccinia viruses that encode the T7 RNA polymerase gene. After that, virus-infected cells were transfected with CFTR plasmids by Lipofectamine 2000 (Invitrogen) and then used for experiments within 12-48 h.

Electrophysiology
The patch-clamp technique with excised inside-out membrane patches was adopted for studying the channel activity of CFTR as described previously (13). The CFTR activity was activated and maintained by 1 mM ATP and 75 nM PKA in the bath (intracellular) solution at room temperature. To amplify CFTR Figure 8. A proposed activation mechanism indicates how PKA phosphorylation alters R 3 -mediated regulation of CFTR activity. The CFTR Cl Ϫ channel is switched from the inactive to active state by PKA phosphorylation (PKA ϩ ATP). Before PKA phosphorylation, R 784 -814 in R 3 (light purple) may generate intra-or interdomain inhibition (blue arrows) that blocks CFTR activity, whereas after PKA phosphorylation, the inhibition is relieved and R 3 -mediated peptide interactions may stimulate CFTR gating (red arrows). ATP was prepared fresh in the bath solution before each experiment. For making the CaATP solution, MgCl 2 in the bath solution was replaced by CaCl 2 , and EGTA was removed. The stock solution of synthesized R domain peptides (AnaSpec, Fremont, CA) contained 0.005 mM peptides, 10 mM KH 2 PO 4 , 1 mM EDTA, and 1 mg/ml BSA, pH 6.7, with NaOH. Peptides (5 l) were diluted in the bath solution (50 l) and phosphorylated by PKA (750 nM) for 10 min before being added into the recording chamber (ϳ500 l; Brook Industries, Lake Villa, IL) with a final concentration of 50 nM at room temperature. Recording of the CFTR current was started within 30 s after adding peptides and lasted for 10 -15 min.

Stimulation of CFTR gating by the R domain
CFTR currents were recorded using an Axopatch 200B patch-clamp amplifier (Molecular Devices, Union City, CA), filtered by an eight-pole Bessel filter (Frequency Devices, Inc., Ottawa, IL) at a corner frequency of 500 Hz and digitized by a Digidata 1322 interface (Molecular Devices) at 10 kHz. The software pCLAMP (Molecular Device) was used for data acquisition and analysis.
For membrane patches that contain many channels, the average of the macroscopic CFTR currents was measured. Moreover, the basal current in the absence of ATP and PKA was subtracted from the total current, as described previously (12). For membrane patches that contained five or less channels, single-channel analysis was performed. The number of active channels (N) in a membrane patch was determined by the maximum number of channels that opened simultaneously at one time during the entire experiment. To measure the singlechannel current amplitude (i), Gaussian distributions were fit to current amplitude histograms. To measure the open probability (P o ), event lists of open-and closed-times were created using a half-amplitude crossing criterion, whereas transitions with durations less than 1 ms were excluded (eight-pole Bessel filter rise time (T 10 -90 ) was ϳ0.73 ms at a corner frequency of 500 Hz).
To measure mean burst duration (MBD), interburst interval (IBI), and P o within a burst (P o-burst ), burst analysis was performed. The delimiter time that separates interburst closures from intraburst closures was determined from the point of intersection between the two exponential curves fitting fast and slow populations of channel closures in the closed-time histogram, as described previously (46). Event lists and values of the delimiter time were used to derive MBD and P o-burst with pCLAMP software. To obtain burst durations, opening bursts formed by only one active channel were measured. Finally, IBI was calculated using the following equation.  (Fig. 1, D and E), it requires 26 s for observing all five channels opening at the same time and 53 s for six channels. Therefore, after PKA was added, CFTR currents were recorded at each intervention for 6 min or more for WT CFTR and 10 -30 min for CFTR mutants in most experiments. PKA was regularly added into the bath solution again after 15-min recordings. However, N would be underestimated for mutant CFTR with large gating anomalies. Therefore, the reported P o vales of these CFTR mutants should be considered maximum P o values, and their IBI values should be considered minimal IBI values.
To reduce above uncertainty in N, the majority of our data were obtained from recordings with three or fewer channels in a membrane patch. For experiments tested with gating stimulators 2ЈdATP, CaATP, and PP i that greatly prolong the MBD of CFTR (29 -31), only membrane patches that contained a single active channel were used for kinetic analysis. For the purpose of illustration, recordings were further digitized at 1 kHz, in which current tracings of 10 s could clearly demonstrate the gating kinetics of CFTR with reduced data points, noise, and some very brief current transitions (e.g. those of Ͻ1 ms).

Reagents and chemicals
PKA catalytic subunits purified from bovine heart (Calbiochem) were used. Other chemicals were purchased from the Sigma-Aldrich or in the reagent grade.

Statistical analysis
The data are presented as means Ϯ S.D. of n observations. Student's paired t test or one-way ANOVA values were used to test statistically significant differences between sets of data. The differences were considered statistically significant when p Ͻ 0.05.