Protein kinase C epsilon-dependent regulation of cystic fibrosis transmembrane regulator involves binding to a receptor for activated C kinase (RACK1) and RACK1 binding to Na+/H+ exchange regulatory factor.

Protein kinase C (PKC) regulation of cystic fibrosis transmembrane regulator (CFTR) chloride function has been demonstrated in several cell lines, including Calu-3 cells that express native, wild-type CFTR. We demonstrated previously that PKC epsilon was required for cAMP-dependent CFTR function. The goal of this study was to determine whether PKC epsilon interacts directly with CFTR. Using overlay assay, immunoprecipitation, pulldown and binding assays, we show that PKC epsilon does not bind to CFTR, but does bind to a receptor for activated C kinase (RACK1), a 37-kDa scaffold protein, and that RACK1 binds to Na(+)/H(+) exchange regulatory factor (NHERF1), a binding partner of CFTR. In vitro binding assays demonstrate dose-dependent binding of PKC epsilon to RACK1 which is inhibited by an 8-amino acid peptide based on the sequence of the sixth Trp-Asp repeat in RACK1 or by an 8-amino acid sequence in the V1 region of PKC epsilon, epsilon V1-2. A 4-amino acid sequence INAL (70-73) expressed in CFTR shares 50% homology to the RACK1 inhibitory peptide, but it does not bind PKC epsilon. NHERF1 and RACK1 bind in a dose-dependent manner. Immunofluorescence and confocal microscopy of RACK1 and CFTR revealed colocalization of the proteins to the apical and lateral regions of Calu-3 cells. The results indicate the RACK1 binds PKC epsilon and NHERF1, thus serving as a scaffold protein to anchor the enzyme in proximity to CFTR.

Cystic fibrosis is an autosomal recessive genetic disorder caused by mutations in a gene encoding a protein called CFTR, 1 cystic fibrosis transmembrane regulator (1). CFTR is a highly regulated chloride channel expressed in apical membranes of epithelia of the intestine, pancreas, sweat gland secretory coil, and conducting airways (2). In the airways, CFTR plays a major role in maintaining optimal humidity and electrolyte balance and in achieving efficient mucociliary clearance through the secretion of fluid and electrolytes. Functioning in concert with CFTR is a basolateral Na-K-2Cl cotransport protein, which mediates uptake of chloride for secretion. Although CFTR is regulated primarily by cAMP-dependent protein kinase A (PKA), it is stimulated to a modest extent by protein kinase C (PKC). In addition, our laboratory and others reported that inhibition of PKC activity using a general PKC inhibitor chelerythrine prevented forskolin-or epinephrinestimulated CFTR function (3)(4)(5). This laboratory established that activity of PKC⑀ is necessary for cAMP-dependent CFTR function. However, the intracellular signaling mechanism that accounts for PKC⑀ regulation of CFTR is not known. The goal of the present study is to elucidate this signaling mechanism.
Only a short term incubation with chelerythrine is necessary to modulate CFTR function, suggesting that PKC⑀ is proximal to and closely associated with CFTR. Alternatively, PKC⑀ could regulate CFTR indirectly through an interaction with CFTRassociated proteins. Mounting evidence suggests that active and inactive PKC isotypes are localized near their target substrates by binding to anchoring or scaffold proteins (6,7). Because constitutive activity of PKC⑀ is necessary for modulation of CFTR function, we focused, in these studies, on proteins called RACKs, or receptors for activated C kinase. RACK1 was first cloned from a rat brain cDNA library and shown to be a homolog of the ␤ subunit of heterotrimeric G proteins (8). RACK1 shares with G␤ highly conserved repeating units, called WD repeats, which usually end in Trp-Asp (WD). RACK1 and G␤ belong to an ancient regulatory protein family of WD repeat proteins (9). Although few WD repeat proteins are enzymes, several interact with other proteins through the WD repeat region to form complexes. The WD repeat regions of G␤ and RACK1 form a rigid seven-blade ␤ propeller structure, arranged in a ring, which is thought to function as an adaptor or scaffold motif (10,11). RACK1 is thought to function in intracellular signaling by recruiting and anchoring multiple proteins, including enzymes, in the same signaling cascade thus tethering the proteins in the proper subcellular localization for their function. PKC was the first enzyme identified as a RACK1-binding partner (for review, see Refs. 6 and 7). It is now proposed that RACK1 binds PKC in an isotype-specific manner, which may also be cell type-specific (8,12). A PKC⑀specific RACK has been identified by expression cloning and shown to bind activated PKC⑀ at a site in the amino-terminal variable region (V1) of PKC⑀ (13,14). PKC␤II also interacts with RACK1; however, binding involves two domains on PKC␤II, a C2, or Ca 2ϩ binding, domain (15) and a 5-amino acid motif in the V5 domain at the carboxyl terminus (16).
In this study, we used a Calu-3 cell line to study the interaction of PKC⑀ with CFTR and other cellular proteins. The Calu-3 cell line is a serous cell line, which we have shown expresses CFTR and secretes chloride in response to elevations in cAMP (4,27). CFTR activity is also modulated by an interaction with the Na ϩ /H ϩ exchange regulatory factor (NHERF1), a 50-kDa protein that is also known as EBP50 (ezrin-radixinmoesin-binding phosphoprotein-50). NHERF1 is a regulatory factor involved in cAMP-dependent inhibition of NHE3 (28). The NHERF1 PDZ1 and PDZ2 domains bind to CFTR carboxylterminal sequence QDTRL and play a role in the regulation of channel gating in a Calu-3 cell line (29 -31). NHERF1 is thought to link CFTR to the actin cytoskeleton through its interaction with ezrin and that ezrin serves as a PKA-anchoring protein (32,33). Indeed, it has been proposed that the NHERF1-CFTR stoichiometry may modulate the magnitude of CFTR channel activity (29). In our previous studies, we demonstrated that down-regulation of PKC⑀ prevented cAMP-dependent CFTR function in Calu-3 cells but did not alter agonist-stimulated Na-K-2Cl cotransporter activity (27). In the current project, we examined the interaction of endogenous proteins involved in the signaling between PKC⑀ and CFTR. Protein-protein interactions were identified and characterized using coimmunoprecipitation, overlay assay, pulldown assay and direct binding assays. We report a new previously unidentified interaction between RACK1 and NHERF1. We also demonstrate that PKC⑀ binds to RACK1 and that the interaction involves specific sites on the endogenous proteins.

EXPERIMENTAL PROCEDURES
Cell Isolation and Culture-Calu-3 cells were grown in a submerged cell culture on 100-mm 2 tissue culture plastic or on 0.4-m pore size Transwell-Clear polyester filter inserts (Corning Costar, Cambridge, MA). For immunofluorescence, cells were seeded at a density of 3 ϫ 10 5 cells/filter with a growth area of 1.0 cm 2 . Culture medium consisted of Earle's balanced salt solution supplemented with 2.4 mg of L-glutamine and 10% fetal bovine serum. Culture medium was changed at 48 -72-h intervals until confluence was reached. Confluence was assessed by microscopic examination and by measurement of electrical resistance across the cell monolayers grown on filter inserts. Electrical resistance was quantitated using chopstick electrodes and EVOM (Epithelial Voltohmmeter, World Precision Instruments, New Haven, CT). Values were corrected for background resistance of filter alone bathed in medium.
Immunoprecipitation, Pulldown Analysis, and Immunoblot Analysis-Cells were grown to confluence, serum deprived overnight, and washed with ice-cold phosphate-buffered saline (PBS). For immunoprecipitation of PKC⑀, RACK1, and NHERF1, cells were lysed in 1 ml of lysis buffer consisting of 100 mM NaCl, 50 mM NaF, 50 mM Tris-HCl, pH 7.5, 1% Nonidet P-40, 0.25% sodium deoxycholate, 1 mM EDTA, 1 mM EGTA, 1 mM sodium vanadate, 0.1 mM leupeptin, 1 mg/ml aprotinin, and 10 M AEBSF. For CFTR, the lysis buffer contained the following detergents: 1% Triton X-100, 1% sodium deoxycholate, and 0.1% SDS. Lysates were clarified by pretreatment with agarose beads and then incubated with antibody directed against the protein of interest. Immune complexes were recovered using protein A-agarose (CFTR, NHERF1), protein G-agarose (PKC⑀), or protein L-agarose (RACK1) beads. Immune complexes attached to beads were washed and resuspended in Laemmli buffer and either heated for 5 min in a boiling water bath (NHERF1, PKC⑀, RACK1) or incubated for 30 min at room temperature (CFTR). Samples were subjected to SDS-PAGE on 6, 8, or 4 -15% gradient slab gels. Protein bands were transferred to polyvinylidene difluoride (PVDF) membrane paper for immunoblot analysis. Protein bands immunoreactive to specific antibodies were detected using enhanced chemiluminescence.
Pulldown analysis was performed using rhRACK1 coupled to Talon beads or GST-NHERF1. Total cell lysates (TCL) were prepared and incubated at room temperature for 30 min with 1 or 2 g of recombinant protein. Talon beads with attached proteins were centrifuged and washed extensively with PBS. Proteins bound to rhRACK1 were identified by immunoblot analysis for the protein of interest. To recover proteins bound to GST-NHERF1, the tagged protein was recovered by anti-GST antibody affinity chromatography. Complexed proteins were analyzed by immunoblot analysis.
Binding of proteins was studied using two methods, slot-blot binding assay or pulldown assay. To quantitate PKC⑀ binding to RACK1, 1 g recombinant human RACK1 (rhRACK1) was immobilized on PVDF membrane paper in a slot-blot apparatus (Invitrogen), blocked, incubated with rhPKC⑀ in the absence or presence of activators, and washed extensively. rhPKC⑀ binding was detected by immunoblot analysis. Binding of NHERF1 and RACK1 was determined by immobilizing 2 g of GST-NHERF1 onto PVDF membrane paper. Varying concentrations of rhRACK1 were added, and after an incubation of 45 min at room temperature, unbound material was removed by extensive washing. Bound rhRACK1 was detected by immunoblot analysis. In the second pulldown method, a solution binding assay was performed. rhPKC⑀ was incubated with PtdSer and dioctanylglycerol for 15 min at 30°C to FIG. 1. Expression of rhRACK1 in Sf9 insect cells. rhRACK1 was expressed as described under "Experimental Procedures" and subjected to immunoblot analysis using a monoclonal antibody to RACK1. Endogenous RACK1 was detected in TCL prepared from Calu-3 cells as a 37 kDa protein band. A 50 kDa protein band in starting homogenate and purified protein was immunoreactive with the antibody to RACK1. WB, Western blot. preactivate PKC⑀. rhRACK1 coupled to Talon Metal Affinity Resin was added and the incubation continued for 30 min at room temperature. Protein complexes were recovered by centrifugation, washed five times to remove unbound material, and analyzed by immunoblot analysis for PKC⑀.
Expression of Recombinant Proteins-Sf9 cells (Invitrogen) were maintained at 27°C in Grace's insect medium supplemented with 10% fetal bovine serum and 10 g/ml gentamicin. Viral stocks were kindly provided by Dr. Susan Brady-Kalnay and used to infect Sf9 cells and to express recombinant human RACK1, as described by Mourton et al. (34). The RACK1 construct consisted of a 1,800-bp fragment containing the full-length human RACK1 cDNA plus a polyhistidine tag, a PKA site, a thrombin cleavage site and an HA tag. Three days after infection, cells were lysed in ice-cold buffer consisting of 1% Triton X-100, 100 mM NaCl, 20 mM Tris, pH 7.6, 1 mM benzamidine, 1 mM sodium vanadate, and 20 mg/ml protease inhibitor mixture (Calbiochem) for 30 min on ice. The lysates were centrifuged at 3,000 ϫ g for 10 min at 4°C to remove Triton-insoluble material. The supernatant was transferred to tubes containing washed Talon Metal Affinity Resin, incubated with rocking at room temperature for 30 min, and pelleted at 3,000 ϫ g for 3 min at 4°C. Talon beads were washed three times with ice-cold PBS and either used directly in experiments or treated with elution buffer (20 mM Tris, 200 mM imidazole, pH 8.0, 100 mM NaCl) to cleave HA-RACK1 from the resin. The resin was pelleted and discarded. The supernatant was added to a Millipore concentrator and centrifuged at 3,000 ϫ g for 10 -30 min at 4°C until the volume was ϳ0.5-1.0 ml. Protein content was determined using a protein assay kit (Pierce) and bovine serum albumin as the standard. Isolation of fusion protein was monitored by immunoblot analysis using a monoclonal antibody to RACK1 (Fig. 1) or a polyclonal antibody to the HA tag (data not shown). As shown in Fig.   1, the final purified protein was immunoreactive with antibody to RACK1 and migrated as a single 50 kDa protein band. The molecular mass is consistent with the addition of the various tags.
GST-NHERF1 in a pGEX4T-1 vector was expressed after addition of 100 M isopropyl-␤-D-thiogalactopyranoside and a 6-h incubation at 37°C. GST-NHERF1 fusion protein was purified using B-PER extraction as directed by the manufacturer (Pierce) followed by affinity chromatography with glutathione-Sepharose B beads. The fusion protein was evaluated by immunoblot analysis for the GST tag using a polyclonal antibody to GST.
Immunofluorescence and Confocal Microscopy-Cell monolayers were washed three times with PBS and fixed in fresh 4% paraformaldehyde for 15 min at room temperature. Cells were washed three time with PBS and permeabilized with 0.2% Triton X-100 in 10% normal goat serum in PBS. The fixed, permeabilized cells were stained for 1 h at room temperature with a monoclonal antibody directed to the carboxyl terminus of CFTR (1:100 dilution) or a monoclonal IgM antibody directed to RACK1 (1:200 dilution). Cell monolayers were washed three times with PBS. Secondary Oregon Green-conjugated anti-rabbit antibody for CFTR or Texas Red anti-mouse antibody for RACK1 was applied at 1:200 dilution for 1 h at room temperature. After three final washes with PBS, the polyester filters were carefully cut from their support inserts and mounted in Slow Fade mounting medium on glass microscope slides. Fluorescence was analyzed using a LSM410 confocal scanning microscope (Carl Zeiss, Germany) equipped with an external argon-krypton laser. Optical sections of 512 ϫ 512 pixels were digitally recorded in 16ϫ line-averaging mode, and z-sectioning was done in 1-m increments. Images were processed for reproduction using Photoshop software (Adobe Systems, Mountainview, CA).
Reverse Transcription-PCR-Reverse transcription-PCR was per-

FIG. 2. PKC⑀ binding to Calu-3 cell proteins by overlay assay.
Calu-3 cells were grown to confluence and serum deprived overnight. TCL were prepared and proteins separated on 4 -15% gradient slab gels and transferred to PVDF membrane paper. For the overlay assay, 200 ng of rhPKC⑀ was inactive (Ϫ) or preactivated in the presence of 30 g/ml PtdSer and 2 g/ml dioctanylglycerol (؉). Preactivated rhPKC⑀ was incubated with membrane paper for 60 min at room temperature. Membrane strips were washed extensively with blocking buffer and immunoblotted with polyclonal antibody to PKC⑀. A, overlay assay on 25 g of TCL reveals a protein that binds PKC⑀ and migrates to the same molecular mass as RACK1 in Fig. 1. Endogenous PKC⑀ is detected as a 75 kDa protein band. B, PKC⑀ binding is not detected in an overlay assay of immunoprecipitated (IP) CFTR. A positive control demonstrates the presence of CFTR in immunoprecipitates. C, PKC⑀ overlay assay of immunoprecipitated RACK1 shows binding at a predominant band at 37 kDa (compare first and third lanes). Endogenous PKC⑀ is in immunoprecipitates of RACK1 as a 75 kDa protein band. An inhibitory peptide, VI-RACK1, based on an 8-amino acid segment of the sixth WD repeat of RACK1, prevents binding of rhPKC⑀ to immunoprecipitated RACK1 (second lane). Immunoblot analysis of RACK1 in immune complexes that were not incubated with rhPKC⑀ and PKC activators shows the presence of RACK1 (far right lane). Results are representative of four separate experiments.
formed with specific primers based on a sequence of RACK1 (Gen-Bank™ Accession NM_006098). Two overlapping sets of sense and antisense oligonucleotides were used: 5Ј-GTGGCTTTCTCCTCTGA-CAA-3Ј and 5Ј-TTAGCGGGTACCAATAGTCA-3Ј encoding a 623-bp cDNA and 5Ј-CAGGAGAGGTTGTGGTGCTA-3Ј and 5Ј-GCCTTGCT-GCTGGTACTGAT-3Ј encoding a 922-bp cDNA. Total RNA (1 g) was reverse transcribed at 50°C for 30 min, and PCR was performed using a One-Step reverse transcription-PCR kit according to the manufacturer's instructions. For cDNA synthesis, 1 g of RNA was mixed with 2 l of random hexamer primer (400 M, each dNTP), 0.6 M primer, enzyme mixture including Omniscript reverse polymerase, and diethyl pyrocarbonate-treated water to a final volume of 50 l in a 0.5-ml Eppendorf tube and incubated at 50°C for 30 min. The reaction mixture was heated to 95°C for 15 min to activate HotStar Taq polymerase and inactivate reverse transcriptase. Single-stranded cDNA was amplified with a PTC-100 thermal cycler from MJ Research (Watertown, MA). PCRs were denatured at 95°C for 2 min followed by 29 cycles of 95°C for 1 min, 62°C for 1 min, 72°C for 2 min, and a final extension at 72°C for 10 min. PCR products were analyzed by electrophoresis on 1% agarose gels and stained with Syber Green I.
Sense and antisense strands of cloned PCR products were sequenced using the dideoxynucleotide chain termination methods. Automated sequencing reactions were performed with synthetic oligonucleotide primers and fluorescent dideoxynucleotide terminators on a DNA sequencer (model 377, Applied Biosystems, Inc., Foster City, CA). Sequence data were analyzed using a BLAST (NIH) sequence similarity data base.
Materials-Peptides encoding sequences from PKC⑀, RACK1, and CFTR were synthesized by the Molecular Biotechnology Core Facility of the Cleveland Clinic Institute. Purity and structural integrity of the peptides were confirmed by high performance liquid chromatography, amino acid analysis, and mass spectrometry. Polyclonal antibodies to NHERF1 and to E3KARP and a GST-NHERF1 construct were supplied by C. H. C Yun (Emory University). Polyclonal anti-PKC⑀ and anti-GST antibodies, anti-GST-agarose beads, horseradish peroxidase-coupled secondary antibodies (anti-mouse IgM, anti-rabbit IgG, and anti-mouse IgG 2A ) and protein L-agarose were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Baculovirus-expressed recombinant PKC⑀ was obtained from Pan Vera (Madison, WI). Anti-human CFTR (carboxyl terminus-specific) monoclonal antibody was from R&D Systems (Minneapolis, MN), and anti-mouse RACK1 monoclonal antibody was from Transduction Laboratories (Lexington, KY). Sepharose and agarose beads, PCR primers, and tissue culture supplies were purchased from Invitrogen. Chelerythrine chloride was purchased from Research Biochemicals International, (Natick, MA). Precast 4 -15% SDS-polyacrylamide gels were obtained from Bio-Rad, and a One-Step reverse transcription-PCR kit was from Qiagen. An enhanced chemiluminescence reagent was purchased from Amersham Biosciences, and Transwell-Clear filter inserts were from Fisher Scientific. All other chemicals were reagent grade.

RESULTS
Overlay Assay for PKC⑀-To determine whether PKC⑀ interacts with airway epithelial cell proteins, we performed an overlay assay for PKC⑀ on Calu-3 cell proteins. Cell proteins were extracted with detergent buffer, separated by gel electrophoresis, and transferred to PVDF membrane paper. Endogenous PKC⑀ was detected as a 75 kDa protein band (Fig. 2A). The overlay assay using rhPKC⑀ revealed binding of exogenous PKC⑀ to at least six protein bands; one predominant band is a 37-kDa protein that corresponds in molecular mass to RACK1 (6, 7). We next performed overlay assays on immunoprecipitates of CFTR (Fig. 2B) and RACK1 (Fig. 2C) and found no specific binding of PKC⑀ to immunoprecipitated CFTR. Activated and inactive PKC⑀ bound to several protein bands in immunoprecipitates of RACK1; binding of activated PKC⑀ is favored (Fig. 2C, compare first and third lanes). Binding was prominent at 37 kDa, indicating PKC⑀ binding to endogenous RACK1. Immunoblot analysis of RACK1 immunoprecipitates for RACK1 demonstrated the presence of a RACK1 immunoreactive protein (Fig. 2C, right lane; see also Fig. 4B, center panel,  right lane). We also detected a protein band at 75 kDa that was immunoreactive with antibody to PKC⑀, suggesting coimmunoprecipitation of PKC⑀. Sequence analysis of cDNA amplified from reverse transcribed total mRNA of Calu-3 cells with primers for human RACK1 demonstrated that the mRNA encoded RACK1, a member of a heterotrimeric G␤ superfamily of proteins.
Coimmunoprecipitation of PKC⑀, RACK1, and CFTR-To characterize a potential interaction among PKC⑀, RACK1, and CFTR, lysates from Calu-3 cells were immunoprecipitated with antibodies to PKC⑀, RACK1, or CFTR and resolved on SDS-PAGE. Immunoblots of immunoprecipitates of CFTR and PKC⑀ were probed with antibodies to PKC⑀ and CFTR, respectively, are shown in Fig. 3A. CFTR was detected in immune complexes of PKC⑀ and, in a control for antibody selectivity, in TLC from T84 colonic cells. Likewise, PKC⑀ was detected in immunoprecipitates of CFTR. rhPKC⑀ was used as an antibody control for PKC⑀ in the immunoblot analysis. The results demonstrate coimmunoprecipitation of CFTR and PKC⑀ which might be related to PKC⑀ regulation of CFTR function. Because activity of PKC⑀ is necessary for its effect on CFTR, we next inhibited PKC⑀ using two different PKC inhibitors, chelerythrine and bisindolylmaleimide, and examined coimmunoprecipitation of PKC⑀ with CFTR. Fig. 3B illustrates the results. Inhibition of PKC⑀ with either PKC inhibitor did not prevent coimmunoprecipitation of PKC⑀ with CFTR.
Immunoblots of PKC⑀ and RACK1 immunoprecipitates were probed with antibodies to RACK1 and PKC⑀, respectively and are shown in Fig. 4, A and B. RACK1 was detected in immunoblots of PKC⑀ (Fig. 4A, left lane) and vice versa (Fig. 4B, left  panel, left lane). Because RACK1 can potentially bind PKC isotypes in addition to PKC⑀, we determined whether endogenous RACK1 associates with PKC␦, a PKC isotype that is necessary for hormonal activation of Na-K-2Cl cotransporter (35). The right panel of Fig. 4B illustrates the results. PKC␦ was found in TCL but was not detected in RACK1 immunopre-  (4). CFTR was immunoprecipitated from lysates and subjected to electrophoresis on 4 -15% SDS-gels. Western blot (WB) analysis for PKC⑀ showed that pretreatment with PKC inhibitors did not alter coimmunoprecipitation of PKC⑀ with CFTR, indicating that activity of PKC⑀ is not a prerequisite for coimmunoprecipitation with CFTR. The illustrated results are representative of four separate experiments. IgG, immunoglobulin G.
cipitates. As a measure of an association between CFTR and RACK1, we probed immunoblots of CFTR for RACK1 and vice versa. CFTR was not detected in RACK1 immunoprecipitates (Fig. 4C, second lane), and RACK1 was not detected in immunoprecipitates of CFTR (Fig. 4C, third lane). To confirm the lack of association of CFTR and RACK1, we used a pulldown technique with rhRACK1 bound to Talon beads to pull CFTR down from TCL of Calu-3 cells. CFTR was not detected in immunoblots of proteins associated with rhRACK1 (data not shown).
Binding of PKC⑀ to RACK1-RACK1 has been shown to bind PKC⑀ in the presence of PtdSer and dioctanylglycerol in a dose-dependent manner (12). The binding properties of RACK1 and PKC⑀ were examined using recombinant proteins. Binding studies were performed in a slot-blot apparatus with immobilized rhRACK1 (Fig. 5A) or by solution assay with rhRACK1 attached to Talon Metal Affinity Resin (Fig. 5B).
We detected binding of rhPKC⑀ to rhRACK1 using both assay methods. PKC⑀ bound to rhRACK1 in a dose-dependent manner. Immunoblot analysis and laser densitometry for the HA tag on rhRACK1 indicated recovery of equal amounts of HA-tagged rhRACK1 from the solution binding assay (Fig. 5C). PtdSer and dioctanylglycerol were required for the binding. An EC 50 (effective concentration) for PKC⑀ binding of 6 M was calculated from Hill plots of binding data.
Peptides based on an 8-amino acid sequence in the sixth WD40 repeat of RACK1 (DIINALCF, designated VI-RACK) have been shown to bind PKC⑀ (8). BLAST comparison of the amino acid sequence of CFTR and RACK1 revealed homology in 4 amino acids, 70 -73, of CFTR, INAL. We tested inhibition of the binding of the two recombinant proteins by VI-RACK and a 12-amino acid peptide based on CFTR amino acids 66 -77 and embedding the INAL sequence. The results are illustrated in Fig. 6. We found that rhPKC⑀ binds to VI-RACK in a dose-dependent manner and that binding was dependent on the presence of PKC activators (Fig. 6A). Peak binding was found at 5 nM VI-RACK. rhPKC⑀ did not bind to a CFTR peptide with an embedded INAL sequence (Fig. 6B). These results demonstrate that PKC⑀ binds at a specific site on RACK1.
An 8-amino acid segment of the V1 region of PKC⑀, denoted as ⑀V1-2, selectively inhibits PKC⑀ translocation function in intact myocytes (8) and prevents binding of PKC⑀ to rat ␤Јcoatomer protein (12). Therefore, we next investigated binding of RACK1 and PKC⑀ using ⑀V1-2 as an inhibitory peptide. If ⑀V1-2 encodes a specific binding site for interaction with RACK1, the inhibitory peptide should prevent binding of rh-PKC⑀ and rhRACK1. We used a solution binding assay in which rhRACK1 coupled to Talon Metal Affinity Resin was incubated with preactivated rhPKC⑀ in the absence or presence of varying concentrations of ⑀V1-2. The results are illustrated in Fig. 7. ⑀V1-2 dose dependently blocked the interaction between rhPKC⑀ and rhRACK1 with an IC 50 (Fig. 7, first lane). In the absence of inhibitory peptide, 4.08 Ϯ 0.42 (n ϭ 3) ng of rhPKC⑀ binds to 2 g of rhRACK1.
Link between RACK1-PKC⑀ and CFTR Is NHERF1-In the absence of direct binding of PKC⑀ or RACK1 to CFTR, we next explored a binding partner for RACK1 which is proximal to CFTR. Two closely related proteins that have been found to interact with CFTR are the PDZ domain proteins NHERF1, a 50-kDa phosphoprotein with two PDZ domains, and NHERF2, or E3KARP, a 50-kDa protein that shares ϳ52% amino acid identity with NHERF1 (31,33,36). NHERF1 was found in immune complexes of RACK1 as a 50 kDa protein band, and RACK1 was detected in immune complexes of NHERF1 as a 37 kDa protein band, suggesting that the two proteins retain an association in detergent lysis buffer (Fig. 8). Although readily detected in Calu-3 cell lysates, NHERF2 was not consistently detected in immunoprecipitates of RACK1 (data not shown). The polyclonal antibody used for immunoblot analysis of NHERF1 also detected lower molecular mass proteins. One at 25 kDa may be a light chain from IgG. A higher molecular mass band at ϳ100 kDa was also observed, but its identity is not clear. It could be the dimer form of NHERF1 or another protein closely related to NHERF1. A protein band immunoreactive to antibody to NHERF1 at 50 kDa was also seen in pulldown assays using His 6 -tagged rhRACK1 (Fig. 8B). Positive controls were also run to confirm the presence of RACK1 in pulldown assays. As seen Fig. 8B, rhRACK1 was observed in pulldowns as 50 kDa protein and endogenous RACK1 in TCL as a 37 kDa protein band. If NHERF1 and RACK1 are binding partners in Calu-3 cells, we predicted that RACK1 would be pulled down with NHERF1. This was examined using GST-tagged NHERF1. The tagged protein and associated proteins were recovered by anti-GST affinity chromatography. Immunoblot analysis of recovered proteins revealed the presence of RACK1 and, as a positive control, of NHERF1 (Fig. 8C). However, PKC⑀ was not detected in this pulldown assay using GST-NHERF1 and hence, we conclude, is not likely to be a binding partner of NHERF1. Our findings thus far indicate an association between RACK1 and NHERF1 but provide no evidence for direct binding. We performed slot-blot binding assays using GST-NHERF1 and rhRACK1 to examine the binding of the two FIG. 5. In vitro binding of rhPKC⑀ to rhRACK1. A, 1 g of full-length rhRACK1 fusion protein was immobilized and blocked on PVDF membrane paper in a slot-blot apparatus. rhPKC⑀ was preactivated by incubation with mixed micelles (MM) of PKC activators, consisting of 30 g/ml PtdSer and 2 g/ml dioctanylglycerol, for 15 min at 30°C. Preactivated rhPKC⑀ was incubated with RACK1 for 30 min at room temperature. Unbound material was removed by washing, and bound rhPKC⑀ was detected by immunoblot analysis. B, rhRACK1 complexed to 1 g of Talon beads was mixed with increasing amounts of rhPKC⑀ in the presence of mixed micelles. Protein complexes were recovered by centrifugation, washed, and subjected to electrophoresis on 4 -15% SDS-polyacrylamide gels. Immunoblot analysis for PKC⑀ indicates dose-dependent binding of rhPKC⑀. C, as a control for pulldown of rhRACK1, duplicate samples of immune complexes were immunoblotted using antibodies to the HA tag. Equal amounts of HAtagged rhRACK1 were detected in pulldowns from the solution binding assay. LD, laser densitometry values.
FIG. 6. Binding of PKC⑀ to RACK1 inhibitory peptide. A, PKC⑀ was inactive (Ϫ) or preactivated (ϩ) by incubation with 30 g/ml PtdSer and 2 g/ml dioctanylglycerol (DOG) for 15 min at 30°C. rhPKC⑀ (25 ng/slot) was added to slots containing varying concentrations of a VI-RACK1 inhibitory peptide (DIINALCF) and incubated for 30 min at room temperature. Unbound material was removed by washing, and bound PKC⑀ was detected by immunoblot analysis. B, inactive or preactivated PKC⑀ (50 ng/slot) was added to slots containing varying concentrations of inhibitory peptide based on the amino acid sequence of CFTR (NPKLINALRRCF) and incubated as described in A. Immunoblot analysis for PKC⑀ indicates that activated PKC⑀ preferentially binds to RACK1 inhibitory peptide but not to CFTR inhibitory peptide. Results are representative of three or four experiments. LD, laser densitometry units; WB, Western blot.
FIG. 7. Inhibition of PKC⑀ binding to RACK1 by ⑀VI-2, a PKC⑀ inhibitory peptide. Varying concentrations of ⑀VI-2 inhibitory peptide (EAVSLKPT) were mixed with rhRACK1 coupled to Talon beads, incubated at room temperature for 30 min, and then added to preactivated rhPKC⑀. The mixture was incubated at 30°C for 20 min, centrifuged to recover Talon beads, washed, and subjected to gel electrophoresis on 4 -15% gradient slab gels. Immunoblot analysis for PKC⑀ indicates a dose-dependent inhibition of binding to rhRACK1 with increasing concentration of inhibitory peptide. Illustrated results are typical of three separate experiments. The asterisk (*) indicates that an aliquot of rhPKC⑀ was applied directly to the gel as a control for the primary antibody. LD, laser densitometry units.
proteins. As seen in Fig. 8D, RACK1 binds to NHERF1 in a dose-dependent manner. An EC 50 , calculated from the data of Fig. 8D, is 3.1 g of RACK1, equivalent to a nominal concentration of 50.1 mM.

RACK1 Colocalizes with CFTR to the Apical Plasma Membrane-The new findings on PKC⑀-RACK1 interaction in
Calu-3 cells imply that RACK1 and CFTR are colocalized in the same region of the cell. The intracellular localization of RACK1 relative to CFTR was determined using a double label immunofluorescence of confluent cell cultures grown on filter inserts. Shown in Fig. 9 are en face images of CFTR and RACK1 taken at three different intracellular locations. CFTR localized in an apical region, whereas RACK1 was localized predominantly in the plasma membrane of the apical and lateral domains. Merged images shows a color shift to orange-yellow, indicating colocalization of RACK1 with CFTR. Colocalization was seen predominantly in the apical region of Calu-3 cells, but it was also observed in the lateral plasma membrane. We were unable to investigate colocalization of PKC⑀ with RACK1 or with CFTR because of cross-reactivity between PCK⑀ primary antibody and fluorescent tagged-secondary antibodies to RACK1 and CFTR. DISCUSSION These studies were initiated to clarify the signaling mechanism that explained regulation of cAMP-dependent CFTR function by PKC⑀ in Calu-3 cells. An examination of the binding of activated PKC⑀ to cellular proteins by coimmunoprecipitation and overlay, pulldown and binding assays leads us to conclude that PKC⑀ binds directly to RACK1 and that RACK1 binds to NHERF1. This implies a new and unique role for the scaffold protein RACK1 in epithelial chloride channel function. Characterization of PKC⑀ binding by overlay assay provided evidence for the absence of direct binding to CFTR (Fig. 2) even though coimmunoprecipitation experiments suggest an interaction between CFTR and PKC⑀ (Fig. 3). Overlay assay also revealed that other Calu-3 cellular proteins bind PKC⑀; one is a 37-kDa protein that corresponds in molecular mass to RACK1. Lack of coimmunoprecipitation of RACK1 and CFTR indicates that the two proteins do not interact directly (Fig. 4) even though both proteins are colocalized in the apical region of Calu-3 cells (Fig. 9). Rather, we found that RACK1 and NHERF1, a scaffold protein that binds four carboxyl-terminal amino acids (DTRL) of CFTR through either of its PDZ (PDZ1 and PDZ2) domains (29,31,37,38), coimmunoprecipitate from Calu-3 cells and are binding partners (Fig. 8). NHERF1 and CFTR bind with high affinity, apparently to regulate other transport proteins, such as the renal outer medullary potassium channel and epithelial sodium channels and NHE3 (39 -41) and to function as a membrane retention signal (42). But, it is the potential dimerization of CFTR by bivalent NHERF1 which is postulated to regulate the full expression of CFTR channel function (29). NHERF1 may act as a scaffold protein to FIG. 8. Interaction between RACK1 and NHERF1. A, coimmunoprecipitation of endogenous NHERF1 and RACK1. RACK1 or NHERF1 was immunoprecipitated (IP) from TCL and subjected to electrophoresis on 4 -15% gradient slab gels and immunoblot analysis for NHERF1 and RACK1, respectively. NHERF1 was detected as a 50 kDa protein band and RACK1 as a 37 kDa protein band. As a positive control for each immunoprecipitated protein, immunoblots were stripped and reprobed. RACK1 and NHERF1 were detected in appropriate immunoprecipitates and TCL (data not shown). B, recovery of NHERF1 after pulldown (PD) of RACK1 using rhRACK1 coupled to Talon beads. 2 g of rhRACK1 was mixed with 2.5 mg of TCL protein and incubated at 30°C for 40 min. Talon beads were recovered by centrifugation, washed, and prepared for electrophoresis on 4 -15% gradient slab gels. NHERF1 was detected by immunoblot analysis in TCL and in pulldowns with RACK1 as a 50 kDa protein band (upper panel). Antibody to RACK1 detected a 50 kDa protein band in pulldowns (lower panel, left lane) which corresponds in molecular mass to rhRACK1. Endogenous RACK1 is evident as a 37 kDa protein band in TCL (lower panel, right lane). C, recovery of RACK1 after pulldown using GST-NHERF1. The pulldown assay was run as described above except 1 g of GST-NHERF1 was added/2. facilitate an interaction between CFTR and RACK1 or PKC⑀ to achieve optimal cAMP-dependent CFTR function.
These studies identify, for the first time, NHERF1 as a binding partner of RACK1 in airway epithelial cells, which express both proteins endogenously. RACK1 also binds PKC⑀, suggesting that the association of RACK1 and NHERF1 brings activated PKC⑀ close to its site of action. RACK1 and its homolog, the G␤ subunit of heterotrimeric proteins, share a unique rigid ␤ propeller structure, formed by WD repeat regions. The WD repeats have been implicated in protein-protein interactions, the first being binding of activated PKC (8). It is thought that binding to RACK1 is necessary for translocation of activated PKC and to protect activated PKC from degradation. Our results indicate that airway epithelial RACK1 binds multiple proteins that are implicated directly in the regulation of CFTR function, thus acting as a scaffold. The mode of interaction between RACK1 and NHERF1 is not known. RACK1 lacks a PDZ binding motif at its carboxyl terminus. However, we could speculate that RACK1 might express an internal motif that binds to the PDZ domains of NHERF1. Alternatively, NHERF1 might interact with RACK1 through non-PDZ motifs, such as in its interaction with ezrin (43). RACK1 is not phosphorylated by PKC (7), but our study provides convincing evidence for direct binding of PKC⑀ to RACK1 in Calu-3 cells. PKC⑀ bound to immunoprecipitates of endogenous RACK1 in an overlay assay (Fig. 2), and endogenous PKC⑀ and RACK1 coimmunoprecipitate (Fig. 4, A and B).
Binding of PKC⑀ to RACK1 was concentration-dependent with a preference for activated enzyme (Fig. 5, A and B). Binding was blocked by inhibitory peptides derived from the amino acid sequences of predicted binding sites on PKC⑀ and RACK1 (8,14). The inhibitory peptide ⑀V1-2, based on an 8-amino acid sequence in the V1 region of the amino terminus of PKC⑀, prevented in vitro binding of RACK1 and PKC⑀ (Fig. 7). The V1, or variable, region encodes a 145-amino acid segment at the amino-terminal region of PKC⑀ which bears close resemblance to a C2 domain of conventional PKC isotypes (44). The C2 domain together with C1 domain of conventional PKC isotypes (␣, ␤, ␥) is responsible for interaction with PKC activators. Dioctanylglycerol and phorbol esters and Ptd-Ser bind to C1 and Ca 2ϩ binds to C2 domains to achieve proper PKC activation (45). The C2-like domain of novel PKC isotypes (␦, ⑀, , ) regulates kinase activity through binding of PtdSer, lacks acidic side chains necessary for Ca 2ϩ binding, and also serves as a site of protein interactions (46). Other investigators and our current study demonstrate that C2 counterparts in PKC⑀ have sites of interaction with RACK1.
A second inhibitory peptide, based on an 8-amino acid peptide in the sixth WD repeat of RACK1 and designated VI-RACK, bound PKC⑀ in a dose-dependent manner and blocked the interaction between PKC⑀ and RACK1 (Fig. 6). We compared the sequence of VI-RACK with that of CFTR to determine whether an analogous binding site for PKC⑀ is expressed in CFTR. Amino acids 70 -73 of CFTR shared 50% homology with VI-RACK. To test binding of PKC⑀, a 12-amino acid peptide was designed to embed the 4-amino acid sequence INAL and still retain a CFTR signature. The CFTR-based peptide did not bind PKC⑀, thus direct binding is not likely to account for PKC⑀ regulation of CFTR function. Instead, our data indicate that PKC⑀ interacts with the sixth WD repeat of RACK1. Binding to RACK1 can, however, be more complicated, involving more than one binding site on RACK1 or its binding partner. For example, binding of PKC␤II to RACK1 involves two sites, one in the C2 domain, the other in the V5 domain (16,47).
We have yet to identify an additional site for the interaction of airway epithelial PKC⑀ with RACK1.
A direct target of activated PKC⑀ has yet to be defined. Only a short term incubation with the PKC inhibitor chelerythrine is necessary to inhibit cAMP stimulation of CFTR (3)(4)(5), suggesting acute regulation of CFTR function. In the Calu-3 cells, NHERF1, CFTR, and PKC⑀ are each phosphoproteins but may well represent only a small pool of potential PKC⑀ phosphorylation targets. Other proteins implicated in the regulation of CFTR include syntaxin 1A (48), which binds to CFTR at the amino terminus. The phosphorylation state of NHERF1 and Munc18, a binding partner of syntaxin 1A, are thought to be important determinants in the functional consequences of binding to CFTR.
How the interaction between PKC⑀ and RACK1 and between RACK1 and NHERF1 regulates CFTR function is not known. The activity state of PKC⑀ is often associated with its translocation. Selective activation of PKC⑀ or inhibition of PKC⑀ translocation has specific consequences in other cell types. Secretion of amyloid precursor protein (49), anionic amino acid transport (50), activation of cardiac L-type Ca 2ϩ channels (13), and protection of cardiac myocytes by ischemic preconditioning (51,52) are all regulated by PKC⑀. RACK1 has been implicated in some of these effects of PKC⑀; however, the intracellular signaling mechanism still must be determined. Nevertheless, one model that emerges from these studies is that NHERF1 recruits a stable regulatory complex and that this is important for regulation of CFTR function. A protein complex of activated PKC⑀ bound to RACK1, which binds NHERF1, would support FIG. 9. Localization of RACK1 and CFTR in Calu-3 cells. En face computer-generated images of apical, midsection, and basal planes through epithelial cell layer are shown. Serum-deprived Calu-3 cells were fixed in 4% paraformaldehyde, permeabilized in 0.2% Triton X-100, and incubated with primary antibodies (anti-carboxyl terminus for CFTR and anti-RACK1) for 60 min before adding secondary antibody (Oregon Green-conjugated anti-rabbit antibody for RACK1 and Texas Red-conjugated anti-mouse antibody for CFTR) for 60 min. En face images of RACK1 (left images), CFTR (center images), and merged images (right images) were visualized by confocal microscopy. As reported by others, CFTR is localized predominantly in the apical region of Calu-3 cells. RACK1 is found in the apical and upper lateral regions of cells. Merged images show colocalization of RACK1 and CFTR. Cells incubated with secondary antibody alone displayed no image (data not shown). cAMP-dependent CFTR function. One can conjecture that inactivation of PKC⑀ would free the enzyme from RACK1, possibly to bind to another scaffold or anchor protein, with a subsequent loss of cAMP-dependent CFTR function. RACK1 might, as a consequence, dissociate from NHERF1 or alter NHERF1-CFTR interaction. Thus, PKC⑀ may shift between anchor proteins depending on its activity state. The activity of PKC⑀ may be an important determinant of its binding partner and phosphorylation state of its target protein as well as the functional status of CFTR.