Introduction

Top Abstract Introduction Experimental Procedures Results Discussion References

Mast cells are critical for the generation of allergic reactions (Galli, 1997), but increasing evidence indicates they may also be involved in the development of inflammation (Theoharides, 1996). Regulation of their secretion is, therefore, important for our understanding of a key biological process and for the development of effective antiallergic/anti-inflammatory molecules. Immunologic stimulation of mast cells results in both tyrosine and serine/threonine phosphorylation of the IgE receptor (Beaven and Metzger, 1993; Scharenberg and Kinet, 1994). The mast cell secretagogue compound 48/80 (48/80) induces phosphorylation of four other proteins with molecular masses of 42, 59, 68, and 78 kDa, of which the first three are phosphorylated within 10 s of challenge (Sieghart et al., 1978; Theoharides et al., 1981). The 59-kDa phosphoprotein was identified as vimentin, one of the intermediate filaments of cytoskeletal proteins (Izushi et al., 1992), and is known to bind to the plasma membrane (Georgatos and Marchesi, 1985). The 78-kDa protein incorporates phosphate 2 min after challenge with 48/80, when secretion has run its course, and in response to the clinically available "antiallergic" drug disodium cromoglycate (cromolyn), commonly referred to as a "mast cell stabilizer" (Sieghart et al., 1978). This finding led to the premise that this 78-kDa protein may be involved in regulation of mast cell secretion (Theoharides et al., 1980).

The 78-kDa protein was shown to be homologous to moesin and was phosphorylated on a number of sites (Ser⁵⁶, Thr⁶⁶, Ser⁷⁴) in response to cromolyn (Correia et al., 1996), an action that appeared to be mediated by a protein kinase C (PKC) isozyme (Wang et al., 1999). A single threonine residue (Thr⁵⁵⁸) was also shown to be phosphorylated in moesin during platelet activation by thrombin (Nakamura et al., 1995). We hypothesized that phosphorylation of moesin at certain sites by distinct PKC isozymes may stimulate secretion, whereas phosphorylation at other sites may promote secretion. Moesin (Lankes and Furthmayr, 1991) and the structurally related ezrin and radixin (ERM) belong to the erythrocyte band 4.1 superfamily considered important in linking the plasma membrane to cytoskeletal components (Furthmayr et al., 1992). The ERM family is increasingly being shown to also be involved in signal transduction (Tsukita and Yonemura, 1997).

Cloning and sequencing of cDNA from various species identified highly conserved domains of 300 amino acids in the amino terminus and 30 amino acids in the carboxyl terminus (Lankes et al., 1993). It was postulated that this latter sequence contains an actin binding site, possibly linking actin to the plasma membrane (Funayama et al., 1991). Changes in the conformational state of actin have been reported during exocytosis in rat mast cells (Koffer et al., 1990) and in rat basophilic leukemia (RBL) cells (Aunis and Bader, 1988; Ludowyke et al., 1994). Moreover, the importance of a physical cytoskeletal barrier in preventing exocytosis has been documented in chromaffin cells (Vitale et al., 1995) and in secretory cells in general (Aunis and Bader, 1988).

In this work, we cloned and sequenced moesin cDNA from RBL cells to generate antiserum for Western and immunocytochemical analyses. The deduced amino acid sequence of the recombinant protein showed 99% similarity to human moesin and had 12 possible serine/threonine residues located within putative PKC phosphorylation sites. Phosphorylation in response to cromolyn appeared to shift moesin from the soluble to the precipitable fraction, indicating its possible association with some other protein(s). In vivo phosphorylation appeared to increase the association of moesin with actin and with secretory granules, as shown by confocal and ultracryomicroscopy.

	Experimental Procedures

Top Abstract Introduction Experimental Procedures Results Discussion References

Materials. PKC (, , and  mixture) purified from rat brain and recombinant rabbit PKC and - expressed in Baculovirus system were purchased from Upstate Biotechnology Inc. (Lake Placid, NY). Phosphatases were also purchased from Upstate Biotechnology Inc. Horseradish peroxidase-conjugated goat anti-rabbit IgG (heavy and light chains) was purchased from Zymed Laboratories (San Francisco, CA). [-³²P]dCTP, [³²P]orthophosphate, [-³²P]ATP, and -³⁵S-dATP were purchased from Amersham Co. (Arlington Heights, IL). Cromolyn and 48/80 were purchased from Sigma Chemical Co. (St. Louis, MO).

Mast Cell Purification. Rat peritoneal mast cells were obtained from male Sprague-Dawley rats (Charles River Labs, Wilmington, MA) and were purified (>90%) by centrifuging them through 22.5% metrizamide (Accurate Scientific Co., Westbury, NY) as described previously (Theoharides et al., 1980). They were then suspended (10⁶/ml) in Locke's solution (150 mM NaC1, 5 mM KC1, 5 mM HEPES, 2 mM CaC1₂, 1 g of dextrose/l, and 1 g of BSA/l, pH 7.2). In certain experiments, mast cells were kept in short-term culture for 24 h in 50-ml culture flasks using Dulbecco's minimal essential medium (D-MEM) supplemented with 15% fetal calf serum (Life Technologies, Grand Island, NY) and 25 Ci of [³⁵S]methionine (70-85 Ci/mmol, Amersham) without methionine or cysteine.

RBL 2H3 Cell Culture. RBL cells were kindly provided by Dr. Henry Metzger (National Institutes of Health, Bethesda, MD) and were grown in stationary cultures in D-MEM supplemented with 15% fetal bovine serum (Life Technologies), as previously described (Tamir et al., 1982).

Construction and Screening of cDNA Library. The mRNA was purified from 5 × 10⁸ RBL cells using Fast Track mRNA kit (Invitrogen, San Diego, CA). Random hexanucleotide primers were used for first-strand synthesis. The cDNA library constructed was inserted at the EcoRI site of the  Zap II vector (Stratagene, La Jolla, CA). Two primers corresponding to human moesin cDNA sequences 1269 to 1292 and 1604 to 1579, respectively, were synthesized. Using synthesized primers, a portion of human moesin cDNA (1269-1604) was amplified by polymerase chain reactions (PCRs; denaturation at 94°C for 1 min, annealing at 55°C for 1 min, and extension at 72°C for 1 min and carried out for 25 cycles) from human fetal brain library, labeled with [-³²P]dCTP (Amersham) using Radprime DNA labeling system (Life Technologies), and used as probe to screen RBL cDNA libraries. In brief,  Zap II plaques were transferred onto nylon filters (Amersham) and screened using radiolabeled probe (Amersham). The filters were washed with a final stringency of 0.1× standard saline citrate, 0.1% SDS at 65°C. Recombinant  Zap II from positive plaques was converted to pBluescript SK phagemids using helper phage R408 according to the manufacturer's protocol.

Sequencing. Positive clones were analyzed by sequencing using Bst Polymerase (Bio-Rad, Hercules, CA). T3 and T7 primers, as well as synthesized primers according to the region already sequenced, were used to sequence the cDNA.

Expression of cDNA RBL Moesin in Escherichia coli. One of the RBL moesin cDNA sequences cloned was PCR-amplified with CAC CAT GCC GAA GAC GAT C as 5' primer and T3 as 3' primer, and the PCR product was cloned into pCR II vector (Invitrogen) and then subcloned into T7 expression vector pET28 (Novagen, Milwaukee, WI) such that the cDNA is in-frame. E. coli JM109 was transformed with the recombinant plasmid. The expression was induced by isopropyl -D-thiogalactoside (IPTG), as previously described (Studier et al., 1990). After IPTG induction, a fusion protein with poly(histidine) on the amino terminus was expressed (Studier's method). The recombinant moesin expressed was purified through binding of the poly(histidine) tail to a nickel column (Invitrogen) under denaturing conditions according to the manufacturer's protocol. The purified moesin fusion protein was dialyzed against PBS containing 1% Triton X-100 and then against PBS without the detergent.

Western Blotting and Immunoprecipitation. Polyclonal antiserum was generated by immunizing rabbits with purified recombinant moesin, which was expressed in E. coli. The specificity of the anti-moesin serum had been reported previously and was characterized by immunoprecipitation (Correia et al., 1996; Wang et al., 1999). It was further characterized as described under Results. The Western blot analysis and immunoprecipitation were performed as described previously (Correia et al., 1996). For immunoblots, samples were solubilized in 3× SDS solubilizing buffer (9% SDS, 0.192 M Tris·HCl, 20% v/v glycerol, 5% 2-mercaptoethanol, 0.04% bromophenol blue). The samples were boiled for 5 min. Proteins were transferred from gel to nitrocellulose membranes by semidry blotting. The transfer buffer used was Tris (25 mM), glycine (190 mM), and methanol (20%). Transfer time was typically carried out at 15 V for 45 min. Blocking of nonspecific binding sites on the membrane was carried out using a 3% BSA in PBS with shaking for 2 h at room temperature. Primary antibody was then applied at a dilution of 1:2000. The secondary antibody used was horseradish peroxidase-conjugated anti-rabbit IgG, and detection was carried out with diaminobenzidine. For immunoprecipitation, mast cells were loaded with ³²P_i and treated with cromolyn as described previously. Mast cells or RBL cells were lysed with 1% SDS and boiled for 5 min. To the sample, a 1:1 dilution of stock inhibitor solution was added at double strength, consisting of PBS containing 20 mM sodium pyrophosphate, 100 mM NaF, 2 mM EGTA, 2 mM EDTA, 5% Nonidet P-40, and protease inhibitors [100 µg/ml leupeptin, 1 mM phenylmethylsulfonyl fluoride (PMSF), and 100 mM benzamidine]. Proteins in the supernatant were precleared by shaking with normal rabbit serum and 50 µl of protein A linked to Sepharose for 1 h at 4°C and then recentrifuged again. To this clear supernatant, 50 µl of anti-moesin serum was added and shaken for 1 more h at 4°C followed by the addition of 50 µl of protein A linked to Sepharose, and the sample was shaken for an additional 30 min at 4°C. The solution was centrifuged at 10,000g for 1 min, and the supernatant was processed for SDS-polyacrylamide gel electrophoresis (PAGE). These experiments were performed four times.

Purification of Nonphosphorylated RBL Moesin. RBL cells in Locke's solution without BSA were incubated with 1 µM staurosporine for 10 min at 37°C. Mast cells were then lysed by addition of 10 volumes of solution A (20 mM Tris-HCl, 0.5 mM EDTA, 0.5 mM EGTA, 10 mM 2-mercaptoethanol, pH 7.5) containing 1% Triton X-100 and proteinase inhibitors (100 µg/ml leupeptin, 1 mM PMSF, 100 mM benzamidine). After centrifugation at 20,000g for 60 min, the supernatant was loaded on a heparin-agarose column (15 × 57 mm; Sigma Chemical Co.) preequilibrated with equilibration buffer (solution A also containing 1 mM PMSF and 40 mM sodium pyrophosphate). Moesin was eluted with a linear gradient of (0-500 mM) NaCl (200 ml) in equilibration buffer at a flow rate of 1 ml/min. Fractions containing moesin were identified by immunoblotting using our polyclonal serum that had previously been shown to recognize only moesin (Correia et al., 1996; Wang et al., 1999). The positive fractions were pooled and diluted. In some instances, RBL moesin was denatured in 8 M urea.

Cell Permeabilization. Purified mast cells were washed three times with HEPES buffer (20 mM HEPES, pH 7.5, 10% sucrose, 1 mM PMSF) and resuspended in the same buffer at 5 × 10⁶ cells/ml. Digitonin was added at final concentration of 25 µg/ml. The cell suspension was kept on ice for 5 min and then centrifuged. The cell permeabilization was monitored by trypan blue exclusion. This experiment was performed four times.

Solubility of Purified Moesin. Purified recombinant moesin was dissolved in PBS containing 1% Triton X-100, but precipitated in PBS without detergent, suggesting that the solubility of the protein was low. The precipitated moesin was sequentially dissolved in PBS containing 1% Triton X-100, dialyzed against PBS without detergent, and then centrifuged to remove the precipitate. Most of the precipitated moesin was dissolved by repeating this method three times.

Phosphorylation and Dephosphorylation In Vitro. Purified recombinant moesin was used in a reaction mixture containing 20 mM Tris-HCl (pH 7.5), 0.1 mM CaCl₂, 0.5 mM MgCl₂, 0.03% Triton X-100, 60 µg/ml diolein, 0.31 µg/ml phosphatidylserine, 25 ng of PKC, and 0.2 mCi/ml [-³²P]ATP (3000 Ci/mmol; Amersham). The mixture was incubated at 30°C for 10 min. For dephosphorylation, 0.1 U of phosphatase 2B (calcineurin; Upstate Biotechnology, Lake Placid, NY) was added to moesin in the same buffer mentioned above and incubated for an additional 10 min. The reaction was terminated by adding SDS-PAGE sample buffer and boiled for 5 min. The samples were analyzed by SDS-PAGE, and the gel was dried and autoradiographed. This experiment was performed four times.

Phosphorylation In Vivo. Purified mast cells were loaded with ³²P_i and treated with cromolyn as previously described (Theoharides et al., 1980). The cells were permeabilized with 25 µg/ml digitonin for 30 s at 37°C, after which supernatant and precipitable fractions were separated by centrifugation. The samples were divided and run on two separate SDS-polyacrylamide gels: one for Western blotting and another for autoradiography. This experiment was performed four times.

Moesin Interaction with Actin. Approximately 5 µg of purified recombinant moesin was phosphorylated in vitro and was mixed with 15 µg of - or -actin that had been polymerized and covalently linked to Sepharose 4B. The purity of actin was greater than 99% as described before (Herman and Pollard, 1979; Hoock et al., 1991) and was kindly provided by Dr. Ira Herman (Department of Physiology, Tufts University, Boston, MA) in binding buffer (40 mM HEPES, pH 7.5, 150 mM KCl, 1 mM MgCl₂, 0.1 mM ATP) as described previously (Shuster and Herman, 1995). It was incubated overnight with shaking at 24°C and then centrifuged at 15,000g for 1 min to separate the pellet from the supernatant. The pellets were washed three times with binding buffer, and SDS sample buffer was finally added to both the supernatant and pellets. The samples were boiled for 5 min and subjected to SDS-PAGE, and Western blot analysis was performed with anti-moesin antibody when necessary. In some experiments, moesin purified from RBL cells was denatured with 8 M urea and dialyzed and then mixed with actin as described above. In the experiments using mast cell extracts, mast cells were treated with cromolyn for 30 s as previously described (Theoharides et al., 1980), homogenized, and centrifuged at 3000g for 5 min at 4°C, and the supernatant was used in the actin binding assays as above. Controls included Sepharose alone with the radiolabeled moesin because Sepharose beads may bind phosphate, but these controls were negative.

Moesin Immunohistochemistry. Purified rat peritoneal mast cells were immediately fixed in 4% paraformaldehyde in suspension. Frozen sections were cut at 7 µm and treated with a 1:200 dilution of rabbit anti-rat moesin polyclonal antibody at room temperature for 1 h. The sections were then incubated with a 1:200 dilution of goat anti-rabbit IgG-biotin (Vector Labs, Burlingame, CA) for 30 min, followed by further exposure to streptavidin-rhodamine (Pierce, Rockford, IL) for 30 min. The sections were then mounted in aqueous mounting medium. This experiment was performed more than 10 times.

Moesin and Actin Double Immunohistochemistry. After streptavidin-rhodamine incubation for moesin labeling, mast cells were treated with mouse anti--actin monoclonal antibody (Sigma Chemical Co.) at a 1:1000 dilution for 1 h at room temperature. The cells were then incubated with horse anti-mouse IgG-biotin (Vector Labs) at a 1:200 dilution for 30 min at room temperature. Streptavidin-fluorescein (Pierce, Rockford, IL) was added to the slides for incubation for 30 min at room temperature. The cells were mounted in aqueous mounting medium and observed under a light microscope (Nikon; Don Santo Corp., Natick, MA). Normal rabbit serum, instead of rabbit anti-moesin antibody, and PBS, instead of -actin antibody, were used as negative controls, respectively. The images shown in Figs. 8 (A-D) and 9 were obtained with a confocal laser scanning imaging system (Odyssey.XL; Noran Instruments, Middleton, WI) equipped with a krypton-argon laser at 42% intensity using the wavelengths of excitation 455 to 488 nm and emission 529 nm and attached to a Silicon Graphics computer. The images in Fig. 8, E to G, were obtained with the same confocal microscope at 42% intensity (Fig. 8E) and 80% laser intensity (Fig. 8, F and G). Image analysis was performed using Image Pro Plus imaging software (Media Cibernetics, Silver Springs, MD). Each gel was analyzed separately after it was digitized from a 5 × 7-inch image on photographic paper. For immunocytochemical analysis, photographic images for each sample were printed at 600× magnification on 8 × 10-inch photographic paper and were digitized. First, a low threshold was established that extinguished the last image pixel. The illuminated pixel values, representing areas, were then read from the program. This experiment was performed four times.

Electron Microscopy. Various attempts were made at preserving the antigenicity of moesin for electron microscopy using 1) 4% glutaraldehyde with 4% paraformaldehyde, 2) 4% acrolein, 3) 1% osmium, or 4) a combination of the above without success in detecting moesin. We then used ultracryomicroscopy. Because of the importance and difficulty of this technique, the procedure is described in detail here. Cells were fixed for 10 s by microwave (900-watt model 3440; Pelco, Redding, CA) in 4% paraformaldehyde (Beil et al., 1994) and remained in 4% paraformaldehyde for 10 min. They were then diluted in 50 ml of PBS. The cells were washed in PBS three times and then treated with 0.2% gelatin, and 7-µl aliquots were placed on a glass Petri dish on ice. Solidified samples were infused with 1.8 M sucrose in PBS for 2 h on ice with two changes. Thereafter, each pellet was removed from the Petri dish, mounted on cryopins using 2.4 M sucrose, and immediately immersed in liquid nitrogen. The pins were then placed in precooled plastic Eppendorf tubes and stored at 80°C until sectioning. The pins were inserted in a precooled LKB-FC4 ultracryostat, and thin (7-µm) sections were cut with glass knives. The sections were transferred from the knife to waiting Formuar-carbon-stabilized coated grids. The grids were placed in a 2% gelatin solution in PBS to await immunocytochemistry. These experiments were performed more than 20 times.

Immunolabeling for Ultracryomicroscopy. Gelatin was melted at 37°C, and grids were removed with a loop. The grids were transferred to a small Petri dish with 50 mM glycine in 1× PBS with 0.05% sodium azide, pH 7.2. The grids were rinsed twice in fresh 50 mM glycine for 15 min, each at 37°C, and then once in PBS with 1% BSA and 0.05% sodium azide, pH 7.2, for 10 min at room temperature. Grids were transferred with a loop to 2% normal goat serum (NGS) in PBS without Tween for 15 min at room temperature. This step was repeated in PBS with 1% BSA, 2% NGS, and 0.05% sodium azide, pH 7.2, with one rinse in the same medium.

Preparation of Incubation Chamber. A large plastic-covered tray was set up with wet paper towel strips on both sides. A 5- to 6-inch piece of Parafilm was positioned in the center of the tray, and 25-µl drops for each of the primary antibodies were pipetted (one drop/grid); positions were recorded. Rabbit anti-moesin serum was used at 1:20 in PBS with 2% BSA and 0.05% sodium azide, pH 7.2. Using nonmagnetic anticapillary forceps, the grid was removed from NGS. Excess NGS was wicked off, ensuring the grids did not dry. Each grid was positioned on a 25-µl drop of the primary antibody, with the sectioned side down in the incubation chamber, and grids were allowed to incubate for 30 min, with caution taken to not allow the grids to sink. Grids were rinsed by floating them on large drops of 50 mM glycine six times over 30 min in 1× PBS and 0.05% sodium azide, pH 7.2, at room temperature. The secondary gold-conjugated antibody was microfuged before use and was rinsed once with PBS, 1% BSA, and 0.05% sodium azide, pH 7.2. Aliquots (25-µl drops) of the secondary antibody were distributed at one drop/grid on Parafilm in the incubation chamber. Using nonmagnetic anticapillary forceps, grids were removed from the glycine, excess glycine was wicked off (with care taken to ensure the grids did not dry), and the grids were transferred to the 25-µl drops of secondary antibody. The grids were allowed to incubate for 60 min at room temperature in the incubation chamber. The goat anti-rabbit IgG antibody labeled with 10 nm gold particles was used at 1:30 in cold PBS with 1% BSA and 0.05% sodium azide, pH 7.2. The grids were rinsed by floating them six times on large drops of 50 mM glycine in PBS with 0.05% sodium azide, pH 7.2, during 30 min. They were then washed with double distilled water twice for 2 min each. Grids were stained with freshly made Millipore-filtered uranyl acetate oxalate, pH 7.5, for 5 min and were rinsed twice with double distilled water for 2 to 3 min each. Grids were transferred with forceps to each drop of Methocel (Sigma Chemical Co.) in the Petri dish on ice and were left for 10 min on the last drop. Methocel had been spun in Eppendorf at the highest setting for 5 min before use. The grids were removed from the Methocel with a loop and sufficient Methocel was drawn off to produce a gold film when dry. While grids were drying, they were stored in the desiccating cabinet. When dry, grids were removed from loops by cutting the film around each grid.

Statistics. Comparison of the data obtained with image analysis was performed with nonparametric analysis using the Mann-Whitney U test. Significance is denoted by P < .05.

	Results

Top Abstract Introduction Experimental Procedures Results Discussion References

Cloning of Rat Moesin cDNA. From 5 × 10⁵ plaques screened, 11 positive clones were obtained. The insertion sizes of the clones were between 1.5 and 3.5 kb (Fig. 1). When the ends of those clones were sequenced, seven of them showed high homology with the human moesin cDNA. Those seven clones overlapped and spanned the entire human moesin cDNA. Moreover, one of the clones, pRM9, had a poly(A) tail, which is an indication that it was the 3' end nontranslated region of the corresponding mRNA.

View larger version (17K): [in this window] [in a new window]   Fig. 1.   Homology of amino acid sequences of moesin from different species. The underlined amino acids indicate the putative PKC phosphorylation sites (consensus: K/R X X T/S). The consensus sequence of moesin was generated by the PRETTY program brand. Only the amino acid residues different among the species are shown vertically in addition to the consensus sequence.

Sequencing of Rat Moesin cDNA After determining the nucleotide sequence and deducing amino acid sequence of the rat moesin cDNA, the sequence was deposited in the National Institutes of Health GenBank with access number AF004811. The protein contains 577 amino acids, which is the same as human moesin. Rat moesin showed 100% homology to mouse and 99% to human moesin (Lankes and Furthmayr, 1991). The calculated molecular mass of the protein was 67.3 kDa, and the pI was 6.37. PKC can recognize specific motifs and phosphorylate serine and threonine residues within them. These motifs are (R/K_1-3, X_2-0)-S/T-(X_2-0, R/K_1-3), S/T-(X_2-0, R/K_1-3), and R/K_1-3, X_2-0)-S/T (Allen and Katz, 1991). When the deduced amino acid sequence of moesin was searched, 12 possible phosphorylation sites were found (Fig. 1), of which at least three (Ser⁵⁶, Thr⁶⁶, Ser⁷⁴) had previously been reported to be phosphorylated by cromolyn in vivo (Correia et al., 1996). These 12 sequences are apparently conserved, as published sequences from different species showed little variation in those regions (Fig. 1).

Expression of Rat Moesin cDNA in E. coli. The moesin cDNA was cloned in expression vector pET-28. Expression of moesin was induced by IPTG in E. coli. After induction, a fusion poly(histidine) tail was inserted at the amino terminus, and moesin was purified to homogeneity with a nickel column through binding to the histidine tail. The isolated protein had a molecular mass of 78 kDa.

Specificity of Anti-Moesin Serum. Western blot analysis was performed on total RBL or rat peritoneal mast cell lysate first using an "anti-moesin" monoclonal antibody from Transduction Laboratories (Lexington, KY); this antibody identified both ezrin and moesin (Fig. 2, lanes 1 and 2; note that the RBL cells contained much more ezrin than moesin, whereas the reverse was true for mast cells). Our own anti-moesin polyclonal antiserum recognized only moesin in both RBL and mast cells (Fig. 2, lanes 3 and 4). Radiolabeled control or cromolyn-treated mast cell lysates was immunoprecipitated using our anti-moesin polyclonal serum and then analyzed by SDS-PAGE. Autoradiography identified a single phosphorylated band corresponding to a protein with 78-kDa molecular mass (Fig. 2, lane 5), whereas immunoprecipitation recognized only one band at the corresponding range (Fig. 2, lane 6). Immunoprecipitation of cromolyn-treated nonradiolabeled mast cells with either control nonimmune serum or the anti-moesin serum showed that only the immune serum (Fig. 2., lane 8) recognized moesin; consequently, the precipitated band was not due to moesin aggregation, which would otherwise also have shown with the nonimmune serum (Fig. 2, lane 7).

View larger version (30K): [in this window] [in a new window]   Fig. 2.   Characterization of the anti-moesin polyclonal antibody (n = 4). Immunoblots were performed on RBL and mast cells as described under Experimental Procedures. Lanes 1 and 2, monoclonal anti-moesin antibody (Transduction Labs) recognized both ezrin and moesin; note that RBL cells have much more ezrin than moesin, whereas the opposite was true for mast cells. Lanes 3 and 4, anti-rat moesin serum recognized a single band on RBL and mast cells; no cross-reactivity with ezrin was observed. The arrowheads indicate the position of ezrin and moesin. Lanes 5 and 6, immunoprecipitation (IP) of the 78-kDa phosphoprotein. Extracts from cromolyn-treated 32Pi-labeled mast cells were precipitated with anti-rat moesin serum. Exposure of the film was for 72 h at 70°C. The arrowheads indicates the position of phosphomoesin. Lanes 7 and 8, immunoprecipitation of moesin from unlabeled, cromolyn-treated (0.1 m for 30 s at 37°C) mast cells using nonimmune (control) serum (lane 7) or anti-moesin (anti-m) serum (lane 8); note only our antiserum generated a band, indicating that this band is not due to moesin aggregation, which would have also appeared with nonimmune serum.

Western Blot Analysis of Moesin in Permeabilized Rat Mast Cells. Digitonin generates lesions on the plasma membrane that are big enough to permit soluble proteins to leak out of the cell, whereas structural elements such as cytoskeletal components and membrane proteins remain in place. When mast cells were permeabilized with digitonin (25 µg/ml), approximately 50% of moesin was found in the supernatant or soluble fraction (Fig. 3). Increasing concentrations of NaCl (0, 150, 500 mM) decreased the amount of moesin in the precipitable fraction pellet with an apparent increase in the soluble supernatant fraction (Fig. 3, compare lanes 3 and 6), suggesting that moesin was loosely associated with some membranous/cytoskeletal elements and could be disrupted by high salt concentrations. The influence of Mg²⁺ and Ca²⁺ on the binding of moesin to the precipitable fraction was also investigated, but there was no obvious effect (data not shown). In mast cells loaded with ³²P_i and treated with cromolyn to induce the phosphorylation of moesin, there was no apparent difference in the amount of moesin identified by Western blot analysis between the control and cromolyn-treated samples or between the corresponding pellet and supernatant fractions (Fig. 4, "Western"; compare lanes with and without cromolyn). Treatment of mast cells with 0.1 mM cromolyn for 30 s at 37°C induced phosphorylation of the 78-kDa protein in whole-cell extract (compare lanes 1 and 2) and in the pellet (compare lanes 3 and 4). In contrast, phosphorylated moesin was not apparent (lanes 5 and 6) in the supernatant fraction (Fig. 4, "phosphorylation"; compare pellet and supernatant lanes with and without cromolyn), suggesting that it may bind firmly to the membrane/cytoskeleton. Even though the background of the autoradiograph of the supernatant fraction (lanes 5 and 6) is lighter than that of the pellet, the amount of protein loaded on the two gels was the same (Fig. 4).

View larger version (55K): [in this window] [in a new window]   Fig. 3.   Salt sensitivity of moesin binding to membranes (n = 4). Mast cells were permeabilized with digitonin (25 µg) in HEPES buffer (20 mM HEPES, pH 7.4, 10% sucrose, 1 mM PMSF) plus the indicated NaCl concentration, centrifuged at 1000g for 30 min to separate the pellet and supernatant fractions, subjected to SDS-PAGE, transferred onto nitrocellulose membrane, and probed with rabbit anti-rat moesin serum.

View larger version (96K): [in this window] [in a new window]   Fig. 4.   Fractionation of phosphorylated moesin (n = 4). Mast cells were loaded with 32Pi, treated with cromolyn as before, and separated into supernatant and precipitable fractions. The samples were run on SDS-PAGE. One gel (top) was used for Western blot analysis to show total moesin, and another (bottom) was used for autoradiography to show the phosphorylated form of moesin. In each case, one lane (lanes 1, 3, and 5) contained cell extracts not treated with cromolyn (), whereas the other lanes (2, 4, and 6) contained cell extracts after treatment with 0.1 mM cromolyn for 30 s at 37°C. Note that the apparent phosphorylation of moesin increased in the pellet after treatment of mast cells with cromolyn, whereas total protein detected by Western blot analysis was unaffected.

Interaction of Moesin with Actin In Vitro. To investigate the possibility that moesin may be binding to actin in vivo, whole mast cell homogenates were used in the Sepharose-actin binding assays after overnight incubation in [³⁵S]methionine. Under control conditions, no binding of moesin was observed (results not shown). When mast cells were treated with cromolyn to induce phosphorylation of moesin in vivo (Fig. 5, lane 1), some increased binding to Sepharose-linked -actin, compared with Sepharose-linked -actin, was seen (compare Fig. 5, lane 1, -actin, and lane 2, -actin). We then investigated whether recombinant moesin could bind to actin and whether in vitro phosphorylation influenced any binding. Purified recombinant moesin bound very weakly to Sepharose - or -actin (Fig. 6, lanes 1 and 2). Phosphorylation in vitro with a purified PKC mixture (-, -, and -isozymes) did not affect the weak binding to Sepharose-linked -actin (Fig. 6, compare lanes 1 and 3) but appeared to increase the binding to Sepharose-linked -actin (Fig. 6, compare lanes 2 and 4). Image analysis confirmed this result. To exclude the possibility that the 78-kDa band may be PKC, which autophosphorylates itself and has a molecular mass similar to that of moesin, immunoprecipitation was performed after the reaction and confirmed that the phosphorylated protein was moesin (data not shown). Our results showing apparent association of moesin with actin could not be due to moesin self-association and aggregation because 1) immunoprecipitation with nonimmune serum did not generate a band indicating that moesin did not self-aggregate, 2) there was apparently greater association with -actin, and 3) there was greater association with -actin after in vitro phosphorylation. If the band observed was due to self-association of moesin, there would have been similar results for all experimental conditions.

View larger version (25K): [in this window] [in a new window]   Fig. 5.   Interaction of mast cell extract containing moesin with actin cross-linked to Sepharose 4B (n = 6). Purified mast cells were labeled with 25 Ci (70-85 Ci/mmol; Amersham) [35S]methionine in D-MEM without methionine or cysteine overnight at 37°C before incubation with or without 104 M cromolyn for 30 s at 37°C. They were then solubilized with 0.5% Triton X-100 and centrifuged at 10,000 rpm for 10 min. The supernatant was incubated with Sepharose-linked - or -actin overnight at 4°C with shaking. The mixture was then centrifuged at 10,000 rpm for 30 s. The pellets were washed three times. Both supernatant and pellets were treated with SDS sample buffer and subjected to SDS-PAGE, Western blot analysis, and autoradiography. Lane 1, -actin. Lane 2, -actin. Arrowhead indicates the position of moesin; other bands may represent either nonspecific binding of radiolabeled proteins to Sepharose or proteins that may have interacted with moesin during treatment with cromolyn and were brought down during Western blot analysis.

View larger version (66K): [in this window] [in a new window]   Fig. 6.   Interaction of purified recombinant moesin, with or without in vitro phosphorylation, to actin cross-linked to Sepharose 4B (n = 6). Moesin expressed in E. coli was purified, phosphorylated with 0.2 µg/ml PKC isozyme mixture (Upstate Biotechnology) and 50 µM ATP, and incubated with actin-Sepharose 4B at room temperature for 5 h in "binding" buffer. Actin-Sepharose 4B was centrifuged and washed with "binding" buffer three times. SDS-sample buffer was added to this pellet, which was then boiled and subjected to SDS-PAGE. The gel was stained with Coomassie brilliant blue R-250. MW, protein molecular weight markers (Life Technologies). Lane 1, moesin with Sepharose-4B-linked -actin. Lane 2, moesin with Sepharose-4B-linked -actin. Lane 3, phosphorylated moesin with Sepharose-4B-linked -actin. Lane 4, phosphorylated moesin with Sepharose-4B-linked -actin. Note that the apparent binding was higher when phosphorylated moesin was used with -actin. The arrowhead indicates the position of moesin.

View larger version (47K): [in this window] [in a new window]   Fig. 7.   Interaction of denatured purified RBL moesin with -actin cross-linked to Sepharose 4B (n = 6). Moesin was purified from RBL cells using a heparin-agarose column. It was denatured with 8 M urea and incubated with Sepharose-linked -actin overnight at 4°C with shaking. The mixture was centrifuged at 10,000 rpm for 30 s. The pellets were washed three times. Both supernatant and pellets were treated with SDS sample buffer and subjected to SDS-PAGE and Western blot analysis. Note that some moesin bound to actin (lane 1), whereas some remained unbound (lane 2). Lane 1, pellet (-actin bound). Lane 2, supernatant (nonbound). The arrowhead indicates the position of moesin.

Immunocytochemical Localization of Moesin and -Actin in Rat Mast Cells. Mast cells were examined using confocal microscopy with the polyclonal rabbit anti-rat moesin serum. At a laser intensity of 42%, moesin was seen mostly close to the plasma membrane (Fig. 8A). This distribution was not homogeneous but often had a rather punctate appearance (Fig. 8A). Treatment with 10⁴ M cromolyn for 30 s did not appear to alter the distribution of surface-associated moesin (Fig. 8C). Stimulation of mast cell secretion with 0.1 µg/ml 48/80 resulted in significantly less immunodetectable moesin (Fig. 8B). This difference was confirmed with image analysis and was statistically significant (Table 1). This decrease in immunodetectable moesin was not due to secretion of moesin, as Western blot analysis of the supernatant after mast cell secretion in response to 48/80 did not detect any moesin (results not shown). Secretion must have, therefore, somehow made moesin inaccessible to or unrecognizable by the antiserum. In cells that had been treated with cromolyn as before and then stimulated by 48/80, the distribution of moesin was like that of the control; so was the overall amount of moesin as shown by image analysis (Table 1). However, an interesting finding was that approximately 20% of the mast cells treated with cromolyn and stimulated by 48/80 had a diffuse distribution of moesin throughout the cell (Fig. 8D). When the laser intensity was increased to 80%, moesin could also be detected inside the mast cells (Fig. 8, compare E with F); detection without the primary antibody gave no positive signal (Fig. 8G). Double immunocytochemistry with the polyclonal rabbit anti-moesin serum (Fig. 9A, brilliant red) and anti--actin antibody (Fig. 9B, fluorescent green) indicated that moesin and -actin could be colocalized at many sites along the cell surface (Fig. 9C; note yellow regions). Negative controls for moesin and F-actin failed to show any immunoreactivity for either antigen.

View larger version (22K): [in this window] [in a new window]   Fig. 8.   Photomicrographs from confocal microscopy using immunocytochemistry to show distribution of moesin in purified rat peritoneal mast cells (n = 10). A, control, unstimulated mast cells. B, cells treated with 0.1 µg/ml C48/80 for 5 min at 37°C. C, cells treated with 104 M cromolyn for 30 s at 37°C. D, cells treated with 104 M cromolyn for 30 s followed by 0.1 µg/ml C48/80 for 5 min at 37°C, note diffuse cytoplasmic distribution of moesin in approximately 20% of cells. Magnification, 1000×; laser intensity at 42%. E, control, unstimulated mast cells showing surface distribution of moesin immunoreactivity at 42% laser intensity; compare with F, where laser intensity was increased to 80%, now showing immunodetectable moesin inside the cell and G the same as F but without the primary antibody. Magnification, 1000×.

View larger version (15K): [in this window] [in a new window]   Fig. 9.   Photomicrograph from confocal microscopy using double immunocytochemistry for moesin and F-actin (n = 4). Moesin (A; brilliant red) and F-actin (B; fluorescent green) on purified rat peritoneal mast cells. C, note the almost identical, punctate distribution of moesin and actin on the cytoplasmic side of the cell surface. Magnification, 1000×.

Ultrastructural Immunocytochemistry. It was quite difficult to retain antigenicity of moesin for electron microscopy. We tried a number of fixatives and combinations (listed under Experimental Procedures) with little success until we used ultracryomicroscopy. Ultrastructural observations of cells fixed with microwave in 4% paraformaldehyde for 10 s indicated that moesin was mostly associated with membranous sites at the cell surface and filopodia; there also appeared some gold particles around secretory granules (Fig. 10). When the cells were fixed in 4% paraformaldehyde for 10 min, moesin-associated gold particles were clearly seen around the perigranular membranes (Fig. 10); in certain cases, the double perigranular membrane was visible around secretory granules, of which the content had been lost due to cryosectioning of the sample (Fig. 11A). When cells were treated with cromolyn, substantial perigranular distribution of moesin was noted with gold particles seen clustered around the secretory granules, of which the round shape was preserved even though they were devoid of any electron dense content (asterisk) due to the weak fixation (Fig. 11, B-D).

View larger version (167K): [in this window] [in a new window]   Fig. 10.   Electron micrographs of immunodetectable moesin in control cryosectioned rat peritoneal mast cells (n = 20). A, part of one mast cell treated with the immune serum showing numerous gold particles. B, one mast cell processed with nonimmune serum showing no gold deposition. Magnification (A and B), 34,500×. Gold particles are noted around secretory granules (C), as well as at the cell surface (D), especially associated with filopodia and the cell surface. Magnification (C and D), 63,000×.

View larger version (126K): [in this window] [in a new window]   Fig. 11.   Electron micrographs of immunodetectable moesin in cryosectioned rat peritoneal mast cells (n = 20). A, control with sparse gold particles located along the perigranular membranes of two secretory granules (asterisk) shown (bar = 0.4 µm). B to D, intense gold particle deposition around secretory granules, after incubation with 104 M cromolyn for 30 s at 37°C. The core of the secretory granules (asterisk) has been depleted by the cryotreatment (bar, 1 µm). Magnification, 85,500× (A) and 63,000× (B-D).

	Discussion

Top Abstract Introduction Experimental Procedures Results Discussion References

The RBL moesin cDNA cloned is nearly identical to human (amino acid identity 99%) (Lankes and Furthmayr, 1991), mouse (Sato et al., 1992), and porcine (Lankes et al., 1993) moesin. The full-length protein was expressed in E. coli and purified, and polyclonal antiserum was raised in rabbits. This antiserum had previously been shown to be monospecific for moesin and could distinguish moesin from ezrin (Correia et al., 1996; Wang et al., 1999), as also shown herein. ERM are members of the erythrocyte 4.1 band superfamily (Lankes and Furthmayr, 1991) and are thought to link the plasma membrane to the cytoskeleton (Furthmayr et al., 1992) through the amino- and carboxyl-terminal domains, respectively (Algrain et al., 1993). Nevertheless, the cellular distribution of ezrin and moesin is quite distinct, with ezrin being nearly ubiquitous (Berryman et al., 1993), whereas moesin is mostly found in endothelial cells (Berryman et al., 1993) and in lymphocytes (Berryman et al., 1993; Pestonjamasp et al., 1995). Moreover, moesin is the only member found in platelets (Nakamura et al., 1995). These findings suggest that moesin may have a distinct function in selected cell types.

In mast cells permeabilized with digitonin and treated with increasing salt concentrations, most of moesin identified by Western blot analysis leaked out, suggesting that it was loosely bound; similar results had previously been obtained with ezrin (Hanzel et al., 1991). Moesin phosphorylated in vivo, however, was consistently found with the precipitable fraction, indicating its association with some, possibly structural, protein(s). Control rat peritoneal mast cell extracts containing moesin failed to bind Sepharose-linked actin under the conditions used, but when mast cell moesin had been phosphorylated in vivo in response to cromolyn, there was some binding to -actin. We then examined whether purified moesin could associate with Sepharose-linked actin. A major problem was keeping moesin in solution to perform these assays. Binding was demonstrated only with recombinant moesin solubilized in Triton X-100 or if purified RBL moesin was first denatured by urea; phosphorylation in vitro slightly increased the binding of recombinant moesin to -actin, but it is not known which sites are phosphorylated in vitro and whether the same sites are phosphorylated in vivo. Unfolding ezrin with SDS exposed a masked domain in the 479 to 585 region (which includes the F-actin binding site) that permits homotypic and heterotypic associations among ERM family members (Gary and Bretscher, 1995). Recombinant or denatured RBL moesin appeared to have more exposed binding sites than their native three-dimensional form, and these sites were available for binding without requiring phosphorylation to expose them. The choice of detergent, however, appears to be a determining factor. For instance, it was recently reported that there was no interaction between purified platelet moesin, whether phosphorylated or not, in Triton X-100 or other nonionic or amphoteric detergents (Nakamura et al., 1999).

Our results showing apparent association of moesin with actin could not be due to moesin self-association and aggregation because 1) immunoprecipitation with nonimmune serum did not generate a band, indicating that moesin did not self-aggregate; 2) there was apparently greater association with -actin; and 3) there was greater association with -actin after in vitro phosphorylation; there should have been qualitatively similar results if the band observed was due to self-association of moesin.

Ezrin had been shown to bind -actin (but not -actin) indirectly, following which elevated calcium levels activated calpain-I, which cleaved ezrin into a 55-kDa fragment, thus dissociating ezrin from actin (Shuster and Herman, 1995). Ezrin was also shown to bind to moesin (Gary and Bretscher, 1993; Andréoli et al., 1994) and radixin, as well as to self-associate (Andréoli et al., 1994) through interaction of the amino-terminal residues 1 to 296 with the carboxyl-terminal residues 479 to 585, which also include the F-actin binding site (Gary and Bretscher, 1995). This interaction inhibited the ability of ezrin to support microspike extrusion, a property that depended on residues 566 to 586 (Martin et al., 1995), which are required for actin binding in vitro (Turunen et al., 1994). The first 300 amino acids in the amino terminus and 30 amino acids in the carboxyl terminus are highly conserved phylogenetically (Lankes et al., 1993), and the latter residues contain an actin binding site (Funayama et al., 1991). Using a blot overlay technique, moesin-actin interactions had previously been shown in bovine neutrophil membrane fragments, but moesin was denatured under these conditions (Keresztes et al., 1998). A moesin-enriched plasma membrane fraction from adherent granulocytes was shown to bind to F-actin by blot overlay (Keresztes et al., 1998), but here again moesin was denatured by SDS. More recently, blot overlay assays with F-actin showed that only the phosphorylated form of moesin interacted with F-actin (Nakamura et al., 1999).

It had previously been shown that the erythrocyte protein 4.1 binds to glycophorin in the plasma membrane (Anderson and Marchesi, 1985). Previous reports had shown moesin to be localized almost exclusively in filopodia (Furthmayr et al., 1992; Sato et al., 1992). Our present results indicate that moesin is localized close to the plasma membrane and secretory granule membranes. Even though moesin was primarily shown close to the cell surface at 42% laser intensity with confocal microscopy, increasing intensity to 80% identified moesin immunoreactivity also inside the cell. Double immunocytochemistry using confocal microscopy colocalized moesin and F-actin at many sites, mostly at the cell surface. This finding could not be an artifact of cross-reactivity as omission of the primary antibody did not yield such results; moreover, if the streptavidin reporter was simply binding to both secondary antibodies, it still would not have access to the antibody against anti-moesin, as the latter was treated with excess streptavidin at the first pass and all sites must have reacted the first time around. Similar findings had also been previously reported for ezrin (Hanzel et al., 1991), as well as for radixin (Sato et al., 1992).

Ultrastructural observations of moesin had been attempted in different cell lines (Hanzel et al., 1991) using periodate-lysine-phosphate fixation, followed by Triton X-100 permeabilization, but the results were poor (Hanzel et al., 1991). We also tried periodate-lysine-phosphate fixation without success in preserving either moesin or cellular architecture. In gastric parietal cells, ezrin was identified by electron microscopy after fixation with 2% paraformaldehyde and 0.03% glutaraldehyde for 1 h (Hanzel et al., 1991). Even then, the number of gold particles reflecting the localization of ezrin were very few and scantily distributed, despite the intense localization by light microscopy. In fact, it was reported that both the anti-ezrin (Hanzel et al., 1991) and anti-moesin (Masumoto et al., 1998) antibodies were very sensitive to fixation. We experienced similar problems when the tissue was fixed with aldehydes. In two other ultrastructural studies, a common antibody for ERM was used on mouse fibroblasts and human epidermoid cells (Sato et al., 1992), whereas specific antibodies for ERM were used on epithelial cells (Berryman et al., 1993), both with equivocal results. We have now identified moesin using techniques designed to better preserve the antigenicity of proteins for electron microscopy using cryo-ultrastructural analysis and have found moesin to be localized close to the plasma and perigranular membranes. Moreover, treatment with the mast cell stabilizer drug cromolyn resulted in preferential "clustering" of moesin around secretory granules. A surprising finding was that approximately 20% of rat peritoneal mast cells treated both with cromolyn and 48/80 demonstrated diffuse localization of moesin. A similar picture was recently reported for moesin in lymphocytes from inflammatory sites (Suzuki et al., 1998), indicating that the redistribution of moesin is possible and may have some pathophysiological significance.

We had previously reported that moesin residues Ser⁵⁶, Thr⁶⁶, Ser⁷⁴, and Ser³⁷⁴ are located at specific PKC recognition motifs, further supporting the possible involvement of PKC (Correia et al., 1996). We have also reported that the PKC isozyme most likely responsible for the phosphorylation of moesin is the calcium-independent and phorbol ester-insensitive, atypical isozyme  (Wang et al., 1999). Phosphorylation of moesin was recently shown to occur exclusively on Thr⁵⁵⁸ during platelet activation by thrombin (Nakamura et al., 1995), an action apparently due to the atypical PKC calcium- and phorbol ester-independent isozyme  (Pietromonaco et al., 1998). This Thr⁵⁵⁸ is found within the 20 C-terminal residues (557-577) of moesin (Pestonjamasp et al., 1995) and within the consensus motif KYKXL, which is required for binding to actin (Turunen et al., 1994). In fact, replacement of Thr⁵⁵⁸ with Asp was shown to promote binding to F-actin (Huang et al., 1999). The possibility that different moesin residues may be phosphorylated by different isozymes and for different functions is supported by findings with phosphorylation of ezrin. For instance, ezrin is phosphorylated on tyrosine residues by epidermal growth factor (Hanzel et al., 1991; Krieg and Hunter, 1992) and platelet-derived growth factor (Fazioli et al., 1993; Franck et al., 1993) but is phosphorylated on serine residues during parietal cell secretion (Hanzel et al., 1991). Differential phosphorylation of multiple sites had also been reported for erythrocyte protein 4.1 in response to phorbol esters or cAMP (Horne et al., 1985).

It is unclear how phosphorylation of moesin may regulate mast cell secretion, but a plausible mechanism could involve some interaction with actin or other cytoskeletal proteins. In vivo conformational changes of moesin after phosphorylation/dephosphorylation may lead to positional rearrangements with respect to the plasma membrane/cytoskeleton that could possibly regulate mast cell secretion. Protein phosphorylation appears to play a key role in a variety of processes that regulate cell function (Hunter, 1995). In neurons, phosphoproteins control secretion by regulating the fraction of the synaptic vesicles available for release, through synapsin I (McCloskey and Cahalan, 1990; Tarelli et al., 1992; Greengard et al., 1993; Levitan, 1994). In particular, synapsin I, which is highly similar to erythrocyte protein 4.1 (Baines and Bennett, 1985), promotes neuronal secretion in its phosphorylated form (Greengard et al., 1993) through dissociation from the cytoskeleton and release of secretory granules to undergo exocytosis. In mast cells, phosphorylation of moesin on such residues as Ser⁵⁶, Thr⁶⁶, Ser⁷⁴, or Ser³⁷⁴ may unmask inaccessible domains that promote cytoskeletal binding and immobilization of secretion granules, thus inhibiting mast cell secretion. In contrast, phosphorylation of other residues, such as Thr⁵⁵⁸, may induce binding to actin that results in "release" of secretory granules to undergo exocytosis. Changes in the state of actin have been reported during exocytosis in rat mast cells (Koffer et al., 1990) and in RBL cells (Aunis and Bader, 1988; Ludowyke et al., 1994). Moreover, the importance of a physical cytoskeletal barrier in preventing exocytosis has been documented in chromaffin cells (Vitale et al., 1995) and in secretory cells in general (Aunis and Bader, 1988).