RACK1 interacts with filamin-A to regulate plasma membrane levels of the cystic fibrosis transmembrane conductance regulator

Abstract

Mutations in cystic fibrosis transmembrane regulator (CFTR), a chloride channel in the apical membranes of secretory epithelial cells, underlie the fatal genetic disorder cystic fibrosis. Certain CFTR mutations, including the common mutation ΔF508-CFTR, result in greatly decreased levels of active CFTR at the apical membrane. Direct interactions between CFTR and the cytoskeletal adaptors filamin-A (FlnA) and Na⁺/H⁺ exchanger regulatory factor 1 (NHERF1) stabilize the expression and localization of CFTR at the plasma membrane. The scaffold protein receptor for activated C kinase 1 (RACK1) also stabilizes CFTR surface expression; however, RACK1 does not interact directly with CFTR and its mechanism of action is unknown. In the present study, we report that RACK1 interacts directly with FlnA in vitro and in a Calu-3 airway epithelial cell line. We mapped the interaction between RACK1 and FlnA to the WD4 and WD6 repeats of RACK1 and to a segment of the large rod domain of FlnA, consisting of immunoglobulin-like repeats 8–15. Disruption of the RACK1-FlnA interaction causes a reduction in CFTR surface levels. Our results suggest that a novel RACK1-FlnA interaction is an important regulator of CFTR surface localization.

the cystic fibrosis transmembrane regulator (CFTR) is a cAMP-regulated chloride channel expressed primarily in the apical plasma membrane of epithelial cells in the airway, intestine, pancreas, kidney, sweat gland, and male reproductive tract. CFTR coordinates with a basolateral Na⁺-K⁺-2Cl⁻ cotransporter to regulate efficient transepithelial ion and water transport. CFTR plays critical roles in the maintenance of fluid homeostasis in the airway lumen, in airway mucociliary and fluid clearance, and in tracheal secretion (47, 63). Genetic defects that cause CFTR hypofunctioning lead to cystic fibrosis (CF), the most common genetic disease in Caucasians (16, 24). CF-causing mutations of CFTR result in disruption of ion and fluid homeostasis and altered regulation of other ion transporters, leading to pancreatic insufficiency, high concentrations of NaCl in sweat, meconium ileus, male infertility, and airway disease (13). In contrast, certain bacterial infections in the intestine promote hyperactivity of CFTR, which causes secretory diarrhea, a leading cause of mortality in early childhood (62).

CFTR in epithelial apical membranes is activated by elevation of cAMP levels and protein kinase A (PKA)-mediated phosphorylation (6, 37). Protein kinase C (PKC) isoforms also regulate CFTR activity (4, 52). However, many additional partners influence the amount, residence time, stability and activity of CFTR at the apical membrane. Adaptor proteins organize multiprotein signaling complexes that convey receptor-mediated or environmental signals to CFTR (19). In particular, the adaptors Na⁺/H⁺ exchanger regulatory factor 1 (NHERF1)and filamin-A (FlnA) stabilize apical membrane localization of CFTR. These adaptors also tether CFTR to the cortical actin cytoskeleton and to microtubules (18, 25, 58).

The WD40 repeat/β-propeller scaffold protein receptor for activated C kinase 1 (RACK1) is another adaptor that regulates CFTR. We previously found that PKC-ε potentiates maximal cAMP-dependent activation of CFTR (30). We also determined that RACK1 interacts directly with NHERF1 (31, 33). The RACK1:NHERF1 complex localizes to a C-terminal PDZ-binding motif of CFTR that interacts with the PDZ domains of NHERF1 (53). Suppression of RACK1 levels using siRNA drastically reduces the amount of apically localized CFTR in Calu-3 cells, a serous epithelial cell line (1). Surprisingly, however, disruption of the RACK1:NHERF1 interaction has relatively little effect on surface CFTR levels (31). Significantly, RACK1 does not interact directly with CFTR (1, 31) suggesting that RACK1 regulates CFTR surface expression through proteins other than NHERF1.

An N-terminal motif of CFTR binds to the cytoskeletal protein FlnA, which functions both as a structural component of the actin cytoskeleton and as an adaptor among actin filaments, signaling proteins, and diverse membrane proteins (44, 64). FlnA is a ∼250 kDa, extended multidomain protein, containing N-terminal globular actin-binding domains followed by 24 immunoglobulin-like repeats (Ig repeats) termed Ig1 to Ig24. Repeats Ig1-Ig23 are organized into two extended rod-like domains, consisting, respectively, of Ig1-Ig15 and Ig16-Ig23 (46). Individual Ig repeats, in particular repeats 17, 19, 21, and 23, interact with unstructured, extended C- or N-terminal motifs in the cytosolic domains of numerous membrane proteins, transporters, channels, and receptors, including CFTR (14, 22). Disruption of the CFTR:FlnA interaction specifically affects surface residence times and reduces apical surface expression of CFTR (43, 55, 58), potentially due to increased endocytosis and inefficient CFTR recycling.

Filamins also interact with numerous soluble, cytosolic signaling proteins, including FilGAP (42), migfilin (22), SH2B1β (46), TRAF2 (26), and several nonreceptor kinases (15, 40, 61). Because downregulation of RACK1 and disruption of the FlnA:CFTR interaction have similar effects on CFTR surface expression, we investigated whether RACK1 and FlnA interact with each other. In the current study, we report that FlnA and RACK1 indeed bind to each other in vitro and in the Calu-3 epithelial cell line. We map the interaction to the WD4 and WD6 repeats of RACK1 and to a segment of FlnA consisting of Ig repeats 8–15. Significantly, we find that blocking the RACK1:FlnA interaction using specific RACK1-derived peptides suppresses CFTR apical membrane expression. In sum, our results indicate that a novel interaction between the RACK1 scaffold protein and the cytoskeletal adaptor FlnA regulates CFTR apical membrane expression in polarized epithelial cells.

EXPERIMENTAL PROCEDURES

Materials.

Anti-RACK1 monoclonal antibody was obtained from BD Biosciences, anti-RACK1 (N-terminal) antibody from Sigma, anti-CFTR antibody from R&D Systems, mouse anti-FlnA antibody from Millipore, and anti-His₆ monoclonal antibody from Calbiochem. Polyclonal anti-GST antibodies, horseradish peroxidase-coupled secondary antibodies (anti-rabbit IgG, anti-goat IgG, anti-mouse IgG, and anti-mouse IgM), and protein l-agarose beads were purchased from Santa Cruz Biotechnology and protein G beads from Invitrogen. The BioPORTER protein delivery system was obtained from Gene Therapy Systems. All other chemicals were reagent grade.

Cell culture and isolation.

Calu-3 cells were grown in cell culture on 100-mm² tissue culture plastic or on 0.4-μm pore size Transwell-clear polyester filter inserts with a growth area of 4.4 cm² for biotinylation experiments (Corning Costar). For immunofluorescence, cells were seeded at a density of 0.2 × 10⁶ cells/filter with a growth area of 1.0 cm². MEM culture medium was supplemented with 2.4 mg of l-glutamine and 10% fetal bovine serum. Cell cultures were grown in submerged culture at 37°C under 5% CO₂ humidified air. Culture medium was changed at 48-h intervals until desired confluence was reached, as assessed by microscopic examination.

Immunofluorescence microscopy.

Calu-3 cell monolayers at 50% confluence were dual labeled for proteins of interest as previously described (33). Briefly, serum-deprived cells were washed with PBS, fixed in fresh 4% paraformaldehyde for 15 min at room temperature, washed three times with PBS, and permeabilized with 0.2% Triton X-100 in 10% normal goat serum in PBS. Fixed, permeabilized cells were stained for 1 h at room temperature with primary antibodies and subsequently washed three times with PBS. Fluorophore-tagged secondary antibody was applied at a 1:200 or 1:500 dilution for 1 h at room temperature. After four final washes with PBS, the polyester filters were cut from their support inserts and mounted in Vectashield mounting medium with DAPI (Vector Laboratories) on glass microscope slides.

Images were collected with a Leica TCS-SP2 laser scanning spectral confocal microscope (Leica Lasertechnik) using an HCX Plan Apo 63×, 1.32 NA objective at a zoom of 1.5. Optical sections of 512 × 512 pixels were digitally recorded in 8× line-averaging mode and z-sectioning was done in 1-μm increments. Images were processed in ImageJ (10, 51) and analyzed for colocalization using the intensity correlation analysis and JACoP plugins (2, 29). Thresholds were selected by the method of Costes et al. (11). Colocalization was estimated according to the Pearson, overlap, and Manders coefficients (34, 35) and the intensity correlation quotient (29). For each sample, colocalization was quantified for 6 or more fields, each consisting of 10–20 adjoining cells or single isolated cells that exhibited clear immunofluorescence for both species of interest. Image brightness and contrast were adjusted in Photoshop (Adobe Systems) for improved reproduction.

Pulldown analysis and immunoprecipitation from cell lysates.

Calu-3 cells were grown to confluence, serum deprived overnight, and washed with ice-cold PBS. Cells were lysed in 1 ml lysis buffer consisting of 100 mM NaCl, 50 mM NaF, 50 mM Tris·HCl, pH 7.5, 1% NP-40, 0.25% sodium deoxycholate, 1 mM EDTA, 1 mM EGTA, 1 mM Na vanadate, and protease inhibitor cocktail set III (Calbiochem). FlnA or RACK1 was immunoprecipitated from Calu-3 total cell lysate, as previously described (1, 33). Total cell lysates were incubated at 4°C for 1 h with antibodies directed against FlnA or RACK1. Immune complexes were recovered using protein G-agarose beads for FlnA or protein L-agarose beads for RACK1 and heated to boiling for 5 min in Laemmli buffer. Samples were subjected to SDS-PAGE for immunoblot analysis, and exposed bands were quantitated by densitometry.

Recombinant protein expression and purification.

GST-tagged RACK1 was expressed in BL21(DE3)-competent cells from pGEX-6p-1 vector as previously described (1). When necessary, the GST tag was cleaved from RACK1 using PreScission protease (GE Healthcare). Truncated RACK1 constructs containing WD repeats 1–4 or WD repeats 3–7 were also cloned into pGEX-6p-1 to make GST fusion proteins. RACK(WD1–4) encodes amino acids 1–189 and consists of the N terminus, WD repeats 1–4, and loops 1–4. RACK(WD3–7) protein encodes amino acids 102–318 and consists of WD repeats 3–7, loops 3–6, and C terminus. GST-RACK(WD1–4) or GST-RACK(WD3–7) were expressed in Rosetta2(DE3) cells (EMD Biosciences), purified on glutathione-Sepharose 4B beads (GE Healthcare), and analyzed by immunoblot with antibodies directed against GST or RACK1. The GST-tags were removed using PreScission protease as needed. Individual RACK1 WD peptides were designed as previously described (32) and synthesized by the Molecular Biotechnology Core Facility, Cleveland Clinic Lerner Research Institute (Cleveland, OH).

Full-length FlnA was immunopurified from 10 ml Calu-3 cell lysates, stringently washed with 0.5 M NaCl followed by 1% Tween, and resuspended in 550 μl PBS. Actin and RACK1 were not detected in immunoprecipitates of FlnA after the sequence of stringent washes (data not shown). His₆-tagged constructs containing Fln-Ig repeat bundles were a kind gift of Dr. Fumihiko Nakamura (Harvard Medical School). Ig bundle constructs Ig(1–8), Ig(8–15), Ig(16–23), or Ig(16–24) were expressed in Rosetta2(DE3) cells and purified using fast protein liquid chromatography using an Akta FPLc and HisTRAP HP columns (GE Healthcare).

In vitro binding assays.

For solid phase slot blot binding experiments, 50 μl of immunoprecipitated FlnA were immobilized per slot. Tagless RACK1 was overlaid onto the FlnA and incubated at room temperature for 25 min. Unbound material was removed by washing, and bound protein was detected by immunoblot analysis for RACK1. Membrane paper was reprobed for FlnA. Exposed bands were quantitated by densitometry, and data were normalized as the ratio of (RACK1/FlnA) densitometric units for each slot. In all slot-blot binding experiments testing binding to RACK1 or to RACK1 deletion constructs, 50 μl of the purified, tagless RACK1 construct at 0.1 mg/ml were immobilized per slot.

For experiments testing inhibition of binding by WD repeat constructs, 100 μg WD repeats were preincubated with 100-μl aliquots of immunopurified FlnA for 30 min at 30°C. Tagless RACK1 was immobilized on PVDF paper and overlaid with a 100-μl aliquot of the FlnA-WD peptide mixture. The incubation continued at room temperature for 30 min. Unbound material was removed by washing, and bound FlnA was detected by immunoblot analysis. Membrane paper was reprobed with antibody to RACK1. Exposed bands were quantitated by densitometry, and data were normalized as the ratio of (FlnA:RACK1) densitometric units for each slot.

To investigate binding of the FlnA Ig-repeat bundles, His₆-Ig repeat bundle alone or a solution containing an Ig bundle plus 100 μg WD4 or WD6 repeat was overlaid on immobilized RACK1. After incubation and washing, membrane paper was probed for His₆ and reprobed for RACK1. Exposed bands were quantitated by densitometry, and data were normalized as the ratio of (His₆/RACK1) densitometric units for each slot.

BioPORTER delivery of proteins.

RACK1-WD repeats were delivered into Calu-3 cells using the BioPORTER protein delivery system as described previously (31). BioPORTER reagent was dissolved in methanol, aliquoted in 10-μl portions, and dried under an N₂ stream. Dried reagent was reconstituted in 50 μl HPSS per filter insert containing an aliquot of protein in PBS and incubated for 5 min. The total volume of the peptide-BioPORTER complex per filter insert was adjusted to 500 μl for 24-mm filter inserts or 300 μl for 12-mm filter inserts with Hank's balanced salt solution containing 10 mM HEPES, pH 7.5 (HPSS). Apical surfaces of cells were incubated with the BioPORTER-protein complex or BioPORTER reagent alone for 2.5 h at 35°C. The apical solution was replaced with HPSS, and the incubation was continued for 2 h at 35°C.

Cell surface biotinylation.

Polarized Calu-3 cell monolayers were grown on 24-mm diameter Transwell permeable supports. Cell-surface proteins were biotinylated using EZ-Link sulphosuccinimidyl-2-(biotinamido)ethyl-1,3-dithiopropionate (sulfo-NHS-SS; Pierce). Cells were rapidly cooled to 4°C, washed in PBS (pH 8.2), supplemented with 0.1 mM CaCl₂ and 1 mM MgCl₂, and then incubated with 1 mg sulfo-NHS-SS-biotin/ml for 30 min at 4°C. The biotin solution was discarded, and a second incubation with fresh biotin solution was repeated. Nonreacted sulfo-NHS-SS-biotin was quenched by washing cells with PBS, pH 8.2, containing 100 mM glycine, 0.1 mM CaCl₂, and 1 mM MgCl₂. Cells were harvested in CFTR lysis buffer, and biotinylated proteins were isolated using streptavidin-agarose beads (GE Healthcare), eluted into SDS sample buffer supplemented with 50 mM dithiothreitol, and separated by SDS-PAGE. Biotinylated CFTR was detected by immunoblot analysis using a monoclonal antibody directed to the C terminus of CFTR. Exposed bands were quantitated by densitometry.

Data analysis.

Data are representative of at least three or more experiments, unless otherwise stated, and treatment effects were evaluated using a two-sided Student's t-test for unpaired samples or an unpaired t-test with Welch correction. Where appropriate, data were analyzed using a one-way ANOVA with Tukey-Kramer multiple comparisons test or Student-Newman-Keuls multiple comparisons test as a posttest. Both tests were analyzed using GraphPad InStat 3.0 software.

RESULTS

FlnA interacts with RACK1 in Calu-3 epithelial cells.

Previously, we found that downregulation of RACK1 using double-stranded silencing RNA led to reduced apical expression of CFTR (1). These data suggested that RACK1 modulates the surface stability of CFTR, despite the fact that RACK1 does not interact directly with CFTR (33). A distinct interaction between the CFTR N terminus and the cytoskeletal adaptor protein FlnA promotes and stabilizes apical membrane expression of CFTR (43, 55, 58). Since downregulation of RACK1 and disruption of the CFTR:FlnA interaction both reduce surface levels of CFTR, we hypothesized that RACK1 regulates CFTR surface levels through a FlnA:RACK1 interaction.

FlnA and RACK1 were previously shown to localize partly to the subapical membrane of airway epithelial cells in close proximity to apical CFTR (1, 55, 58). We confirmed these observations by confocal immunofluorescence microscopy of native CFTR, RACK1, and FlnA in Calu-3 cells (Fig. 1A). RACK1 and FlnA partly colocalize with each other at the plasma membranes of these cells (Fig. 1B; Table 1). To determine whether this correlates with an interaction between RACK1 and FlnA, we immunoprecipitated each protein from Calu-3 total cell lysates and probed for the putative binding partners. RACK1 is present in immunoprecipitates of FlnA and vice versa (Fig. 2).

Fig. 1.

Colocalization of endogenous filamin-A (FlnA) and receptor for activated C kinase 1 (RACK1) in Calu-3 cells. Cells were labeled with the appropriate primary antibodies to FlnA and RACK1 and probed with Alexa-Fluor 488-labeled secondary antibody (blue/cyan) or Alexa-Fluor 568-labeled secondary antibody (magenta). Colocalization of RACK1, FlnA, and/or cystic fibrosis transmembrane regulator (CFTR) is shown by itself (yellow) and on merged panels (white). Scale bars in the leftmost frames correspond to 10 μm. A: membrane localization of FlnA and RACK1, respectively, with CFTR. Top row: FlnA staining using AF568 secondary antibody and CFTR staining using AF488 secondary antibody. Middle and bottom rows: RACK1 staining using AF568 secondary antibody and CFTR staining using AF488 secondary antibody. B: colocalization of FlnA with RACK1 at Calu-3 cell membranes. Top row: FlnA staining using AF568 secondary antibody and RACK1 staining using AF488 secondary antibody. Bottom row: RACK1 staining using AF568 secondary antibody and FlnA staining using AF488 secondary antibody.

Table 1.

Colocalization of filamin, CFTR, and RACK1 in Calu-3 cells

Protein1	Protein2	Rpearson	Roverlap	M1	M2	ICQ	n (cells)
Filamin	CFTR	0.61 ± 0.04	0.71 ± 0.06	0.28 ± 0.06	0.57 ± 0.07	0.35 ± 0.05	12 (130)
Filamin	RACK1	0.70 ± 0.06	0.79 ± 0.04	0.39 ± 0.06	0.57 ± 0.04	0.37 ± 0.02	6 (47)
CFTR	RACK1	0.50 ± 0.03	0.60 ± 0.03	0.24 ± 0.04	0.35 ± 0.04	0.29 ± 0.04	8 (80)

All values shown are means ± SE. CFTR, cystic fibrosis transmembrane regulator; RACK1, receptor for activated C kinase 1; R_pearson, Pearson's correlation coefficient (−1; negative correlation, e.g., exclusion; 0: no correlation; 1: positive correlation); R_overlap, overlap correlation coefficient, similar to Pearson's coefficient but without subtraction of the mean intensity values in each channel (0: negative correlation; 1: positive correlation); M1 and M2: Manders coefficients, quantifying the fractional intensity overlap of the first protein's immunofluorescence signal with that of the second protein (M1) and vice versa (M2); ICQ, intensity correlation quotient, as defined by Li et al. (Ref. 6; −0.5: exclusion; 0: random with uncorrelated intensities; 0.5: colocalization with completely correlated intensities between the 2 channels); n (cells): n, total number of imaged fields; cells, total number of cells in these fields

Fig. 2.

Coimmunoprecipitation of RACK1 and FlnA from Calu-3 cell lysates. Calu-3 total cell lysates were treated with anti-FlnA or anti-RACK1 antibodies and immunoprecipitated (IP) with protein G-agarose or protein L-agarose, respectively. Immunoprecipitates and total cell lysates (TCL) were probed for RACK1 and FlnA. Duplicate immunoprecipitations are illustrated. RACK1 coimmunoprecipitates with FlnA and vice versa. IB, immunoblot.

To determine whether RACK1 and FlnA interact directly with each other, we tested the binding of the respective purified proteins in vitro using a solid phase slot blot assay. Endogenous FlnA was immunopurified from Calu-3 lysate and treated with high-stringency washes to remove coimmunoprecipitated proteins. Purified FlnA was immobilized on PVDF paper in a slot blot binding apparatus. Figure 3 illustrates typical results showing binding of recombinant, untagged RACK1 to the immobilized FlnA. A test for linear trend between column means yielded a P = 0.0075, indicating a significant concentration-dependent interaction.

Fig. 3.

Direct binding of RACK1 to FlnA. FlnA was immunoprecipitated from Calu-3 cells, immobilized on PVDF paper in a slot-blot filter apparatus, and overlaid with recombinant (untagged) RACK1. The membrane was probed with anti-RACK1 and anti-FlnA antibodies. A: representative binding experiment showing that RACK1 binds to immobilized FlnA in a concentration-dependent manner. B: means ± SE values from 5 experiments showing concentration-dependence of RACK1 binding to FlnA. Binding was quantitated as the intensity ratio of RACK1 to FlnA as measured by densitometry. Results of ANOVA (P = 0.0086) and Student-Newman-Keuls Multiple Comparison posttest: *P < 0.05 for 1 vs. 4 μl RACK1; **P < 0.01 for 1 vs. 8 μl RACK1.

The WD4 and WD6 repeats of RACK1 mediate its interaction with FlnA.

RACK1 adopts a β-propeller structure (60) consisting of seven consecutive WD repeats. Individual WD repeats have previously been shown to act as binding sites for RACK1 binding partners (36). To better define how RACK1 binds to FlnA, we synthesized peptides corresponding to each of the seven WD repeats of RACK1 (Fig. 4A) and examined their ability to competitively inhibit binding of RACK1 to FlnA (Fig. 4, B and C). Peptides WD4 and WD6 significantly inhibited binding of the FlnA:RACK1 interaction in a concentration-dependent manner (Fig. 4D).

Fig. 4.

RACK1 interacts with FlnA through its WD4 and WD6 repeats. Peptides encoding individual WD repeats of RACK1 were tested for their ability to compete with full-length RACK1 for binding to FlnA. A: structure of human RACK1 (50). Colored areas show the location of each WD peptide, corresponding to the equivalently colored residues in the sequence. B: immunopurified FlnA was incubated with WD peptides and overlaid onto immobilized RACK1 using a slot-blot filter apparatus, the membrane was probed with anti-FlnA antibody and FlnA binding was quantitated by densitometry. Representative binding experiment; the first 2 lanes contain no immobilized RACK1, showing negligible nonspecific binding of FlnA. Preincubation of FlnA with peptides from WD4 and WD6 repeats reduces FlnA binding to RACK1. C: quantitation of FlnA binding to RACK1 in the presence of WD repeat peptides (means ± SE from 9 experiments). WD4- and WD6-peptides compete with full-length RACK1, causing significant (*P < 0.05; **P < 0.01) reductions in bound FlnA. D: concentration-dependent reduction of FlnA binding to RACK1 after preincubation with WD4- or WD6-peptides, showing that the peptides specifically compete for FlnA.

To verify that FlnA interacts with the WD4 and WD6 repeats, we made RACK1 deletion constructs containing WD repeats 1–4 and WD repeats 3–7, respectively. As shown in Fig. 5, A and B, both truncation constructs compete with full-length RACK1 for binding to FlnA. However, the effect of WD3–7 is not additive compared with WD1–4, suggesting some redundancy between the WD4 and WD6 binding sites. We next performed slot blot assays to directly measure the binding of the deletion constructs to FlnA. WD(1–4) binding was blocked by WD4 repeat peptide, but not WD1 repeat (Fig. 5C). WD(3–7) binding was blocked by WD4- and WD6-peptides but not a WD7-peptide (Fig. 5D). Taken together, these results show that RACK1-WD4 and -WD6 repeats specifically mediate the interaction between RACK1 and FlnA.

Fig. 5.

RACK1-WD(1–4) and -WD(3–7) interact with FlnA. A: representative immunoprecipitation experiment, showing that coimmunoprecipitation of native FlnA with RACK1 from Calu-3 cell lysates is reduced by preincubation with WD(1–4) or WD(3–7). B: means ± SE from 5 coimmunoprecipitation experiments. Preincubation with 100 μl of WD(1–4) or WD(3–7) leads to a significant (*P < 0.05) reduction in coimmunoprecipitation of FlnA with RACK1 from Calu-3 lysates. WD(3–7) competes more effectively for binding to FlnA than WD(1–4), possibly because the WD(3–7) contains FlnA-binding epitopes on both WD4 and WD6. C: representative binding of FlnA to WD(1–4) immobilized on PVDF paper. Preincubation of WD4-peptide (but not WD1-peptide) reduces FlnA binding to WD(1–4), suggesting that the interaction is mediated by the WD4 repeat. D: representative binding of FlnA to WD(3–7) immobilized on PVDF membrane. Preincubation of FlnA with WD4- or WD6- (but not WD7-) peptides competes with WD(3–7), suggesting the interaction is mediated by the WD4 and WD6 repeats. E: means ± SE of binding of FlnA to WD(1–4), with and without preincubation with WD1 and WD4 peptides. Results shown are from 5 different experiments. WD4-peptide competes with WD(1–4), causing a very significant (*P < 0.01) or highly significant (**P < 0.001) decrease in binding of FlnA relative to control experiments. Ctrl, control. F: means ± SE of binding of FlnA to WD(3–7), with and without preincubation with WD4-, WD6-, and WD7-peptides for 4 different experiments. WD4- and WD6-peptides compete with WD(3–7), causing a significant (*P < 0.05) decrease in binding of FlnA relative to the control.

FlnA binds to RACK1 through Ig repeats 8–15.

We used constructs of multiple FlnA Ig repeats (“Ig bundles”) to ascertain which portion(s) of FlnA interact with RACK1. We first tested whether recombinant His-tagged bundles Ig(1–8), Ig(8–15), Ig(16–23), or Ig(16–24) (Fig. 6A) could pull down RACK1 in a solution binding assay (Fig. 6B). We observed that only Ig(8–15) interacted with RACK1. These results were further replicated in a slot-blot binding assay to immobilized RACK1 (Fig. 6C). The interaction of Ig(8–15) with RACK1 was competitively inhibited by recombinant RACK1-WD4- or WD6-peptide (Fig. 6D), verifying the specificity of the interaction.

Fig. 6.

FlnA (8–15) bundle interacts with RACK1. A: schematic representation of the overall Filamin domain structure. ABD, actin-binding domains. B: representative binding experiment showing that RACK1 binds to recombinant His₆-tagged Ig(8–15) bundle. RACK1 and 4 different Ig bundles were incubated together then pulled down using Talon beads. Pulldowns were probed for RACK1 and His₆: lane A, Ig(1–8); lane B, Ig(8–15); lane C, Ig(16–23); lane D, Ig(16–24). C: representative experiment showing inhibition of RACK1 binding to Ig bundles by WD peptides. RACK1 was immobilized on membrane paper and overlaid with a solution containing an Ig bundle and WD4- or WD6-peptide. Membrane paper was probed for His₆ and RACK1. WD4- and WD6-peptides block binding of RACK1 with Ig(8–15) bundle. D: means ± SE for 3 different experiments on inhibition of RACK1 binding to Ig(8–15) bundle. WD4- and WD6-peptides compete with RACK1, causing very significant (**P < 0.01) decrease in binding of RACK1 to Ig(8–15).

Disruption of the RACK1:FlnA interaction reduces surface levels of CFTR.

Previous studies from our laboratory established that apical CFTR surface expression is greatly decreased after siRNA-mediated downregulation of RACK1 (1) as well as when the FlnA:CFTR interaction is inhibited (55). We investigated whether the RACK1:FlnA interaction is important for the stabilization of CFTR surface expression by RACK1. Based on our results showing that RACK1-WD4- and WD6-peptides can compete with native RACK1 for FlnA, we used a BioPORTER reagent to deliver WD4- or WD6-peptides into Calu-3 cells. We measured levels of CFTR in the apical membrane by cell surface biotinylation followed by streptavidin pulldown. Quantitation of proteins bound to streptavidin beads showed that intracellular delivery of WD4- or WD6- peptides decreased CFTR surface expression approximately twofold (Fig. 7, A and B) without affecting total cellular amounts of CFTR, RACK1 or FlnA. As a control, WD1 and WD2 were also tested and found to not significantly affect CFTR surface expression (Fig. 7, C and D). These results show that the RACK1:FlnA interaction is indeed important for proper maintenance of surface CFTR levels.

Fig. 7.

WD4- and WD6-peptides reduce CFTR surface expression. WD peptides were delivered into Calu-3 cells using BioPORTER reagent. The apical membrane was biotinylated, and biotinylated proteins were recovered by streptavidin pulldown. Proteins were probed for CFTR, RACK1, and FlnA. In A–C: −BioP, no BioPORTER or peptide added to cells; +BioP, BioPORTER but no peptide added to cells. A: WD4 and WD6 peptides. Representative results from biotinylation experiment, showing recovery of CFTR, RACK1 and FlnA with streptavidin beads. As a control, blots were reprobed for actin, as shown at bottom, which indicate, as expected, a weak signal on beads and much stronger signal in the supernatant. B: percent surface CFTR is calculated from densitometry values as bead value/(bead + total supernatant values) × 100. means ± SE for 3 different experiments. **P < 0.01, suppression of CFTR surface expression by WD4 and WD6 repeats is very significant. C: WD1 and WD2 peptides. Representative experiment showing recovery of CFTR relative to -BioP control. D: summary of data of the effects of WD1 and WD2 peptides on recovery of biotinylated CFTR. Statistical test by ANOVA yielded a P = 0.63 indicating that WD1 and WD2 do not affect CFTR surface expression.

DISCUSSION

We previously found that suppression of endogenous RACK1 levels using double-stranded silencing RNA drastically reduced CFTR levels at the apical membrane of epithelial cells by 80–90% (1). This suggested that RACK1 regulates CFTR levels at the membrane, although RACK1 does not directly interact with CFTR (33). A subpopulation of RACK1 localizes to apical CFTR by interacting with NHERF1 (31, 33), which in turn interacts with the C terminus of CFTR (53). NHERF1 connects CFTR to the actin cytoskeleton through the actin-binding protein Ezrin (20). NHERF1 also links CFTR to the β-adrenergic receptor, to protein kinase Cε, and (via Ezrin) to protein kinase A (27, 33, 41). These NHERF1-linked interactions potentially stabilize and/or activate CFTR at the apical membrane, although the exact mechanisms are not fully elucidated. Knockdown of NHERF1 also reduces surface CFTR levels while NHERF1 overexpression has the opposite effect (18, 24). RNA silencing of RACK1, RNA silencing of NHERF1, or elimination of the NHERF1:CFTR interaction by deletion of the CFTR PDZ-binding motif (38) all suppress apical CFTR levels to a similar degree. Surprisingly, however, a NHERF1-derived peptide that competitively inhibits the NHERF1:RACK1 interaction has little effect on surface CFTR levels (31). These data suggest that RACK1 regulates apical CFTR levels by a mechanism that is unrelated to its interaction with NHERF1.

We have now discovered that RACK1 interacts with the cytoskeletal adaptor protein FlnA, which binds to an N-terminal motif of CFTR and anchors CFTR directly to the cortical actin cytoskeleton. We and others have found that disruption of the CFTR:FlnA interaction suppresses CFTR surface levels to a comparable extent as silencing of NHERF1, silencing of RACK1, or disruption of the CFTR:NHERF1 interaction (43, 55, 58). We found that RACK1 and FlnA interact directly with each other in Calu-3 cells as well as in vitro. Intriguingly, specific disruption of the RACK1:FlnA interaction in Calu-3 cells, by transfection of recombinant WD4 and WD6 repeats from RACK1, caused a significant reduction in apical CFTR expression.

RACK1 contains seven WD40 repeats, with consensus sequence X^6–94-[GH-X^23–41-WD], that form a β-propeller domain with seven blades. (Fig. 4A). RACK1 functions as a scaffold protein and is best known as a binding partner for several isoforms of PKC. However, RACK1 also binds to NHERF1, the phosphodiesterase PDE4D5, Src, several β-integrins, HIV-Nef1, the ribosome, and possibly to heterotrimeric G protein β-subunits (12, 36, 54). RACK1 interactions with different binding partners may involve one or more of the WD-repeats (56). For example, the SH2 domain of Src binds to a single repeat of RACK1, WD6 (3). Similarly, our group found that RACK1 binds to NHERF1 through WD5 (32). A structure-based comparison of three of the four surface-exposed loops from each repeat (the 4th loop was not included in our WD repeat peptides) shows several short sequence differences in WD4 and WD6 compared with the other repeats. Loop 1 contains an acidic/hydrophobic epitope; loop 2 is longer in WD4 and more basic in WD6; and loop 3 is shorter and less polar in WD6. At present, we do not know whether these residues contribute to binding of FlnA. The binding sites of FlnA on RACK1 lie on two noncontiguous WD repeats, WD4 and WD6, while the intervening repeat does not contribute substantially to binding. FlnA Ig repeats are large enough to potentially contact both WD repeats simultaneously. However, the similar binding of WD(1–4) and WD(3–7) constructs to FlnA (Fig. 5) raises the possibility that the WD4 and WD6 sites may function independently of each other.

However, details of the interactions between RACK1 and any of its diverse binding partners are poorly defined. A number of RACK1 binding partners contain β-strand-rich domains, such as C2 domains or pleckstrin homology (PH) domains. PKC isoforms, for example, contain C2 domains, the putative RACK1 partners dynamin and spectrin contain PH domains (49), and NHERF1 contains PDZ domains. Interestingly, C2 domains and FlnA-Ig repeats have similar immunoglobulin-like protein folds, consisting of β-sandwiches with large, conformationally variable loops that connect individual β-strands (14, 48). While the topologies of C2 domains and FlnA-Ig repeats are not equivalent, their structural similarities raise the possibility that they may bind RACK1 in a similar manner.

In contrast to RACK1, the interaction of FlnA with many of its binding partners is better understood. Several groups have elucidated a “canonical” binding mode for Ig repeats 4, 9, 12, 17, 19, 21 and 23, which all interact with characteristic sequences near the N or C termini of their partners (22). These sequences typically contain a linchpin serine residue, flanked by alternating hydrophobic residues. Upon binding, these motifs form β-strand-type backbone interactions with the third strand of the Ig-repeat and hydrophobic sidechain packing contacts with the fourth strand (14). The N-terminal cytoplasmic tail of CFTR, among other proteins, interacts with FlnA in this manner (55). However, the WD4- and WD6-derived peptides used in our experiments correspond to internal β-strands of RACK1 and their connecting loops (Fig. 4A). As these peptides bind FlnA and compete with full-length (folded) RACK1, they may contain atypical FlnA-interaction sequences or epitopes that are surface exposed in full-length RACK1. A structure-based comparison of three of the four surface-exposed loops from each repeat (the 4th loop was not included in our WD repeat peptides) shows several short sequence differences in WD4 and WD6 compared with the other repeats (Table 2). Loop 1 contains an acidic/hydrophobic epitope; loop 2 is longer in WD4 and more basic in WD6; and loop 3 is shorter and less polar in WD6. At present, we do not know whether these residues contribute to binding of FlnA. However, it is likely that FlnA interacts with RACK1 in a manner that differs from the “standard” FlnA binding modes that have been characterized to date.

Table 2.

Structure-based alignment of surface-exposed loop regions of WD repeat peptides

WD	Loop 1	Loop 2	Loop 3
1	GHNGWV	TTPQFP-DM	SRDKT
2	GHSHFV	ISS-DG-QF	SWDGT
3	GHTKDV	FSS-DN-RQ	SRDKT
4	SHSEWV	FSPNSSNPI	GWDKL
5	GHTGYL	VSP-DG-SL	GKDGQ
6	DGGDII	FSP-NR-YW	T-GPS
7	AEPPQC	WSA-DG-QT	YTDNL

Sequences that differ in WD repeats 4 and 6 compared with the other repeats are bolded and underlined.

Based on our studies, we hypothesize that RACK1 localizes to CFTR by at least two mechanisms. RACK1 can be positioned near the N terminus of CFTR by interacting with FlnA, and near the C terminus of CFTR by interacting with NHERF1 (Fig. 8). Disruption of the FlnA:RACK1 interaction reduces surface levels of CFTR, but the mechanism by which RACK1 stabilizes CFTR is unclear. In general, RACK1 acts as a scaffold to bring together PKC isoforms and specific substrates (31). RACK1 may promote access of enzymes such as PKCβII, PKCε, or PDE4D5 to CFTR, FlnA, or NHERF1. Phosphorylation of CFTR by PKC may modulate CFTR conformation and activity directly (4, 5) or synergize with PKA-mediated phosphorylation (8, 52). Because FlnA and NHERF1 bind to different locations on CFTR, FlnA:RACK1 complexes may position PKC to phosphorylate different sites on CFTR than NHERF1:RACK1 complexes.

Fig. 8.

Model of RACK1 localization to CFTR. RACK1 can be positioned near the N terminus of CFTR by interacting with FlnA and near the C terminus of CFTR by interacting with NHERF1. RACK1 potentially recruits the same effectors to each position but implements different mechanisms of CFTR regulation.

A simple model for RACK1 regulation of CFTR through FlnA:RACK1 complex would involve recruitment of PKC by FlnA-associated RACK1 in situ. In turn, the recruited PKC could regulate the CFTR:FlnA interaction as well as CFTR activity, internalization, and/or the interactions of CFTR with other proteins. Interestingly, phosphorylation is implicated in the regulation of interactions between FlnA and membrane proteins other than CFTR. For example, PKC-mediated phosphorylation of the D2 and D3 dopamine receptors regulates their association with filamin (9, 28). FlnA is itself phosphorylated by PKC in several nonepithelial cell-types, including fibroblasts (17) and the M2 melanoma cell line (39). Phosphorylation of FlnA positively regulates its interactions with β-integrins (7), while phosphorylation of the integrins themselves antagonizes FlnA binding (57). However, FlnA interacts directly with PKCα when expressed in HeLa cells (59), with PKCθ in activated T cells (21) and with PKCε in fibroblasts (23). Thus FlnA may not always require RACK1 to recruit some PKC isoforms in epithelial cells. Instead, RACK1 may need to recruit particular PKC isoforms for phosphorylation of a specific range of sites on CFTR or FlnA. A full understanding of the regulation of epithelial CFTR by the FlnA:RACK1 complex and the biochemical details of the FlnA:RACK1 interaction will require further investigation.

GRANTS

This research was supported by National Heart, Lung, and Blood Institute Grant HL-58598 and funds from a Cystic Fibrosis Foundation Research Development Program (to C. M. Liedtke). S. Misra was supported by National Institute of General Medical Sciences Grant GM-02701 and funds from a Ralph Wilson Medical Research Foundation Award. S. Misra, J. Amick, and R. C. Page were also supported by a philanthropic donation to the Cleveland Clinic from the Heritage Mark Foundation in memory of Lauren “Lo” Detrich.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author(s).

AUTHOR CONTRIBUTIONS

Author contributions: L.S., P.L., E.K., J.A., and R.C.P. performed experiments; L.S., P.L., S.M., and C.M.L. interpreted results of experiments; L.S., P.L., S.M., and C.M.L. edited and revised manuscript; L.S., P.L., J.A., R.C.P., S.M., and C.M.L. approved final version of manuscript; S.M. and C.M.L. conception and design of research; S.M. and C.M.L. analyzed data; S.M. and C.M.L. prepared figures; S.M. and C.M.L. drafted manuscript.