Sites of Protein Kinase A Activation of the Human ClC-2 Cl– Channel*

Human ClC-2 Cl– (hClC-2) channels are activated by protein kinase A (PKA) and low extracellular pHo. Both of these effects are prevented by the PKA inhibitor, myristoylated PKI. The aims of the present study were to identify the PKA phosphorylation site(s) important for PKA activation of hClC-2 at neutral and low pHo and to examine the relationship between PKA and low pHo activation. Recombinant hClC-2 with point mutations of consensus phosphorylation sites was prepared and stably expressed in HEK-293 cells. The responses to forskolin plus isobutylmethylxanthine at neutral and acidic pHo were studied by whole cell patch clamp in the presence and absence of phosphatase inhibitors. The double phosphorylation site (RRAT655(A) plus RGET691(A)) mutant hClC-2 lost PKA activation and low pHo activation. Either RRAT or RGET was sufficient for PKA activation of hClC-2 at pHo 7.4, as long as phosphatase inhibitors (cyclosporin A or endothal) were present. At pHo 6 only RGET was needed for PKA activation of hClC-2. Low pHo activation of hClC-2 Cl– channel activity was PKA-dependent, retained in RGET(A) mutant hClC-2, but lost in RRAT(A) mutant hClC-2. RRAT655(D) mutant hClC-2 was constitutively active and was further activated by PKA at pHo 7.4 and 6.0, consistent with the above findings. These results show that activation of hClC-2 is differentially regulated by PKA at two sites, RRAT655 and RGET691. Either RRAT655 or RGET691 was sufficient for activation at pHo 7.4. RGET, but not RRAT, was sufficient for activation at pHo 6.0. However, in the RGET691(D) mutant, there was PKA activation at pHo 6.0.

Human ClC-2 Cl Ϫ (hClC-2) 1 channels are activated by PKA, an activation that is prevented by treatment with a permeant PKA inhibitor, myristoylated protein kinase inhibitor, mPKI (1,2). These Cl Ϫ channels are also activated by reduced external pH (pH o ), but this activation requires PKA (1)(2)(3)(4)(5). Arachi-donic acid also activates hClC-2, but this activation is not inhibited by mPKI and therefore is independent of PKA (2). ClC-2 plays a role in Cl Ϫ transport by a variety of tissues (6), and low pH o and PKA regulation of ClC-2 may be of physiological relevance. The goal of these studies was to determine the structural basis for low pH o and PKA activation of human ClC-2 Cl Ϫ channels.
Human ClC-2 contains seven potential phosphorylation sites in the C terminus, and two in the N terminus (Table I). All but two consensus sites (RRAT655 and RGET691 in human ClC-2 and RRQS651 and KRKS749 in rabbit ClC-2) are conserved in rat (6). Human and rabbit, but not rat, ClC-2 are activated by PKA (1)(2)(3)(4)(5)(6). These sites are absent in rat ClC-2 (6), a channel that does not show PKA activation, although it has been shown to be phosphorylated without changes in function (7). The phosphorylation sites RRAT655 and RGET691 of human ClC-2 were the focus of the present studies.
ClC-2 (human and rabbit) has been shown to be activated by PKA in planar lipid bilayer studies and in hClC-2-expressing HEK-293 cells treated with forskolin plus IBMX (2,5). However, an unsuccessful attempt to activate 36 Cl transport in human IB3 cells containing hClC-2 has been reported (8), despite finding low pH o activation. Further studies of the relationship between low pH o and PKA activation of hClC-2 were warranted because low pH o activation was found to be dependent upon PKA activation (2). The present studies used sitedirected mutagenesis and functional assays in the presence and absence of phosphatase inhibitors and a PKA inhibitor, mPKI, to determine the structural basis for PKA activation and PKA-dependent low pH o activation of hClC-2.
The site(s) of PKA activation of the hClC-2 Cl Ϫ channel is not known. Moreover, it is not known whether PKA activation of the channel at pH o 7.4 and at reduced pH o occur at the same site(s). Site-directed mutagenesis of hClC-2 was used to determine which of the sites (RRAT and RGET), or whether both, was important under these two different conditions. Wild-type and mutant hClC-2 Cl Ϫ channels were stably expressed in HEK-293 cells, and Cl Ϫ channel function was assessed by measuring whole cell Cl Ϫ currents in response to activation of PKA by forskolin and IBMX at pH o 7.4 and 6.0. Arachidonic acid, which activates hClC-2 in a PKA-independent manner (2), was used to determine whether the mutant channels were expressed and present in the plasma membrane. The effects of protein phosphatase inhibitors were used to determine whether PKA activation was sufficient to overcome the action of endogenous protein phosphatase. These studies delineate the structural basis for low pH o and PKA activation of hClC-2 and suggest a possible role for PKA regulation of hClC-2 in physiological settings.

EXPERIMENTAL PROCEDURES
Transfected HEK-293 Cells-HEK-293 cells stably transfected with wild-type or mutant human ClC-2 cDNA in the mammalian expression vector pcDNA3.1 using LipofectAMINE were prepared as described previously (2).
Site-directed Mutagenesis-Site-directed mutagenesis of single phosphorylation sites was carried out using a Transformer II site-directed mutagenesis kit (Clontech). Four mutants, as follows, were prepared and confirmed by sequencing: RRAT655(acc)3 A(gcc); RGET691 (acc)3 A(gcc); a double mutant containing both of these changes; and RRAT655(acc)3 D(gac). The entire cDNA was also verified by sequencing. Sequential PCR was used to introduce the point mutation RGET691(acc)3 D(gac) into a 1363-bp cDNA fragment of wild-type human ClC-2, which was then cloned into PCR2.1Topo. After confirming the sequence, a SanDI/HindIII fragment, which included the mutation, was cloned into the full-length human ClC-2 cDNA and sequenced over the ligation sites.
Measurement of Whole Cell Cl Ϫ Currents-Whole cell currents were measured as described previously (5). Forskolin and arachidonic acid in Me 2 SO were diluted into the bath solution resulting in a final concentration of Me 2 SO of 0.1% or less. Controls containing equivalent amounts of Me 2 SO alone were always performed. Statistical significance of the difference between two means was determined using the Student's t test using N ϭ number of cells.
Materials-HEK-293 cells were obtained from American Type Culture Collection (ATCC). HEPES, Me 2 SO, Tris, EGTA, and inorganic and organic salts were obtained from Sigma. Forskolin, arachidonic acid, cyclosporin A, nodularin, endothal, and mPKI were from Calbiochem. IBMX was from Aldrich. LipofectAMINE, pcDNA3.1, and PCR2.1Topo were from Invitrogen. Fig. 1A shows typical Cl Ϫ current recordings from HEK-293 cells expressing wild-type hClC-2 Cl Ϫ channels at pH o 7.4 with no addition (control) and following addition of 5 M forskolin plus 20 M IBMX (forskolin/IBMX). The cells were subsequently treated with 1 M arachidonic acid. Both treatments increased Cl Ϫ currents. Similar treatments of nontransfected HEK-293 cells had no effects (Table IV). Cl Ϫ currents in HEK-293 cells expressing the double site (RGET(A) plus RRAT(A)) or the single site (RRAT(A)) mutant hClC-2 Cl Ϫ channel were not activated by forskolin/IBMX but were activated by arachidonic acid. This suggests that RRAT is involved in the PKA activation of hClC-2 at pH o 7.4. HEK-293 cells expressing the single site RGET(A) mutant hClC-2 were activated by forskolin/ IBMX, and the level of activation was greater than in HEK-293 cells expressing wild-type hClC-2, indicating that the remaining site, RRAT, is involved in PKA activation of the channel at pH o 7.4. The single site RRAT(D) mutant hClC-2 showed high basal Cl Ϫ channel activity that was further increased with forskolin/IBMX. The RGET(D) site was also constitutively active, thus confirming the importance of these sites to PKA activation of hClC-2 at pH o 7.4. I/V curves normalized to capacitance under each condition are shown in Fig. 1B. The summarized data expressed as slope conductance (nS/pF) are shown in Fig. 1C and Table II. The extent of arachidonic acid activation was similar in wild-type and mutant channels, demonstrating that the mutant hClC-2 channels were expressed to levels similar to that of the wild-type channel in HEK-293 cells.

Wild-type and Mutant hClC-2 Cl Ϫ Channel Activity at pH o 7.4 -
Wild-type and Mutant hClC-2 Cl Ϫ Channel Activity at pH o 6.0 - Fig. 2A shows typical Cl Ϫ current recordings of wild-type and mutant hClC-2 channels stably expressed in HEK-293 cells, first recorded at pH o 7.4 and then recorded at pH o 6.0, followed by forskolin/IBMX and then arachidonic acid. The I/V curves normalized to capacitance are shown in Fig. 2B, and the summarized data expressed as slope conductance (nS/pF) are shown in Fig. 2C and Table III. Nontransfected HEK-293 cells had very low currents, and these treatments had no effect on the current (Table IV) (Table III). Fig. 3 and Table V show that low pH o activation of wild-type hClC-2 and RGET(A) mutant hClC-2 was abolished by mPKI. Thus, acid activation in the basal state is dependent upon PKA phosphorylation of RRAT.

Role of Basal State Phosphorylation in Low pH o Activation-
Effect of Protein Phosphatase Inhibitors-To assess whether protein phosphatase inhibition could affect basal levels or forskolin/IBMX-activated levels of wild-type hClC-2 Cl Ϫ channel activity, the effects of cyclosporin A (a PP2B inhibitor) (9) and endothal (a PP2A Ͼ Ͼ PP1 inhibitor) (10) were examined at pH o 7.4 and 6.0. As shown in Table II, at pH o 7.4 cyclosporin A (20 M) added to the bath significantly increased (approximately doubled) the basal Cl Ϫ channel slope conductance of wild-type and both single-site mutant hClC-2 channels, but it was without effect on the double phosphorylation site mutant hClC-2. Endothal (500 nM) had no effect on basal activity of either wild-type or mutant channels. However both cyclosporin A and endothal increased forskolin/IBMX activation of the wild-type channel and rescued the forskolin/IBMX activation of the RRAT(A) mutant hClC-2 at pH o 7.4 (Fig. 4, Table II). These results show that both RRAT and RGET phosphorylation play a role in activation of hClC-2 at pH o 7.4. Suppression of phosphatase activity at pH o 7.4 was required to show activation of the RRAT(A) mutant channel, which retains the phosphatasesensitive RGET site. When 1 M nodularin, a cell-impermeant phosphatase inhibitor with a specificity of PP1 Ն PP2A ϾϾϾ PP2B, was added to the bath, there was no effect; however, when it was added to the pipette solution the wild-type channel showed increased activation by forskolin/IBMX, but the basal level was unaffected at pH o 7.4 (Table II). In contrast, at pH o 6 the only significant phosphatase inhibitor effects were those of cyclosporin A and nodularin (in the pipette), which increased forskolin/IBMX activation of the wild-type hClC-2 channel. No

FIG. 1. Effect of PKA activation on wild-type (WT) and phosphorylation site mutant hClC-2 Cl ؊ channel activity at pH o 7.4. A, representative scans of Cl Ϫ currents of HEK-293 cells expressing wild-type and RRAT(A) plus RGET(A), RRAT(A), RGET(A), and RRAT(D) mutant hClC-2 Cl
Ϫ channels before (C, control) and after treatment with 5 M forskolin, 20 M IBMX (F/I) followed by 1 M arachidonic acid (AA). B, I/V curves for Cl Ϫ currents normalized to cell capacitance (pA/pF). C, summarized data expressed as normalized slope conductance (nS/pF) for the data given in A and B. Data in B and C were plotted as means Ϯ S.E. Numbers in parentheses, number of cells and detailed statistical comparisons for each condition are shown in Table II. #, p Ͻ 0.01, and *, p Ͻ 0.001, with respect to control.  other effects of cyclosporin A, endothal, or nodularin on basal or activated channel activity in either wild-type or mutant hClC-2 were observed (Table III). DISCUSSION The cumulative data presented show that both RRAT655 and RGET691 are important to the activation of hClC-2 by PKA phosphorylation at pH o 7.4. RGET alone, but not RRAT alone, supported PKA-dependent activation at pH o 6.0. However, RRAT did support PKA-dependent activation in the RGET(D) mutant. In addition, these studies show that acid activation of hClC-2 Cl Ϫ channel requires phosphorylation of RRAT.
The functional significance of these phosphorylation sites has been investigated using alanine and aspartate substitution of consensus phosphorylation sites (11)(12)(13)(14) by site-directed mutagenesis. This is a widely used technique, and loss of activation by protein kinases in such mutants is generally interpreted as a specific modification that does not affect global conformation of the protein. Arachidonic acid, which acts through a separate mechanism to activate hClC-2 Cl Ϫ currents, was used to confirm structural integrity of the mutant channels and their presence on the surface of the HEK-293 cells. In addition, the substitution of only one of the two phosphorylation sites generally did not affect the ability to activate at the remaining site, supporting the argument for structural integrity of the mutant channels.
Because there are numerous consensus phosphorylation sites on hClC-2, the site-directed mutagenesis approach would be very tedious, especially because quantitative assessment of the function of the Cl Ϫ channels required patch clamp technology. However, based on the initial observation that the human channel contains two phosphorylation sites that are not present in rat ClC-2, which has been repeatedly demonstrated to be insensitive to PKA activation (7,15), it was reasonable to begin this study by focusing on the sites that are unique to hClC-2. The loss of the effect of PKA in the double knock-out RRAT(A) plus RGET(A) mutant hClC-2 supported this rationale and greatly limited the number of sites that required detailed study. However, the other potential phosphorylation sites may play a role in regulation of hClC-2 under different conditions.
The RRAT(A) mutant was not responsive to PKA activation at pH o 7.4 in the absence of protein phosphatase inhibitors. Wild-type hClC-2 has been previously shown to be responsive to protein phosphatase inhibition (16). Protein phosphatase inhibitors increased PKA activation of the wild-type and the RRAT(A) mutant channel at pH o 7.4, demonstrating the importance of control of protein phosphatase activity in PKA activation studies. Thus, the report of lack of PKA activation of 36 Cl transport in the human cystic fibrosis (IB-3) cell line containing hClC-2 (8) could be explained in part by the lack of inclusion of phosphatase inhibitors in that study. The same cell line was shown to exhibit low pH o activation in electrophysiological studies. Based on our finding that low pH o activation is dependent upon PKA, it is likely that endogenous PKA activation of hClC-2 occurred in the basal state in those studies (8) showing low pH o activation.
The RRAT site alone is not responsive to PKA activation at pH o 6.0 in this model system. PKA itself was active, based on its ability to activate the wild-type and the RRAT(A) mutant hClC-2. Protein phosphatase inhibition was effective because activation of the RRAT(A) mutant containing the RGET site was rescued by protein phosphatase inhibition at pH o 7.4 and because cyclosporin A increased forskolin/IBMX-activated wild-type hClC-2 activity at pH o 6. Therefore, these results suggest that reduced extracellular pH changed the availability of the RRAT site to protein phosphatases or PKA, perhaps through direct effects on the channel. The RGET(D) mutation also changes the availability of RRAT at reduced pH, suggesting that under some conditions, phosphorylation of RGET may have similar effects.
In support of this suggestion, the RRAT(D) and RGET(D) phosphomimetic mutant hClC-2 channels were constitutively active under basal conditions at both pH o 7.4 and 6.0. Some of these could be further activated by PKA through phosphorylation at the RRAT site at pHo 7.4 and the RGET and RRAT sites at pH o 6.0. The constitutive activity of the RRAT(D) mutant and the inability of PKA to activate the RGET(A) mutant containing RRAT at reduced pH o demonstrated that this was  Although it is tempting to speculate that there is transmembrane coupling between extracellular pH and the C-terminal phosphorylation site(s), further studies will be required to determine how these pH o -dependent effects arise.
In the presence of cyclosporin A, wild-type, RGET(A), and RRAT(A) channels at pH o 7.4 show similar levels of activation by forskolin/IBMX. The difference noted with and without cyclosporin A shows that the wild-type channel has a different phosphatase sensitivity than either of the mutants.
The present study was not designed to consider the physiological significance of any particular protein phosphatase in the regulation of hClC-2 in the HEK-293 model system. Other approaches may be better suited to such studies. For example, a direct interaction between PP1 and rabbit ClC-2 has been demonstrated in a yeast two-hybrid system (18). Endothal has a partial selectivity for PP1, and PP1 inhibition by inclusion of nodularin in the pipette solution increased PKA activation of wild-type hClC-2 at both pH o 7.4 and 6. Additional studies of the action of phosphatases on native hClC-2 channels in cells may be warranted to help evaluate the physiological role of these channels.
This study has identified the residues responsible for PKA activation and PKA-dependent low pH o activation of the human ClC-2 channel (RRAT655 and RGET691) and provides some new direction for future evaluation of PKA activation of ClC-2-mediated currents in human tissues. Whereas PKA activation, per se was only 4-fold at pH o 7.4 and 2-fold at pH o 6.0, there was an 8.6-fold increase at pH o 7.4 comparing basal Cl Ϫ currents with currents in the presence of cyclosporin A and forskolin/IBMX and a 13-fold increase comparing basal currents at pH o 7.4 with currents at pH o 6.0 in the presence of cyclosporin A plus forskolin/IBMX (Tables II and III). Several reports of the kinase regulation of ClC-2 in other systems have now appeared (1-5, 16 -18), and future studies to investigate the role of PKA activation of ClC-2 in physiological function may benefit from our present studies.