<<

Journal of Science 108, 161-171 (1995) 161 Printed in Great Britain © The Company of Biologists Limited 1995

Characterization of moesin in the sea urchin Lytechinus variegatus: redistribution to the plasma membrane following fertilization is inhibited by cytochalasin B

Eric Scott Bachman and David R. McClay* Box 91000 LSRC, Research Drive, Duke University, Durham, NC 27708, USA *Author for correspondence

SUMMARY

We have investigated the distribution and function of an regions of cell-cell junctions. We show that SUmoesin is ezrin-radixin-moesin-like (ERM) molecule in the sea present in -rich regions of the embryo. Finally, we urchin. A sea urchin homologue of moesin was cloned that show that the location of SUmoesin requires an intact actin- shares 75% amino acid similarity in the conserved N- based cytoskeleton. SUmoesin fails to localize to the plasma terminal region to other moesin molecules. A 6.3 kb membrane after fertilization in the presence of cytocha- message is transcribed late in embryogenesis and is present lasin B. Furthermore, SUmoesin loses its apical position in in adult tissues. Polyclonal antibodies were generated to the region of cell-cell junctions in the presence of cytocha- proteins expressed by a bacterial expression vector, and lasin B in later stages of embryogenesis. This effect is affinity purified. These antibodies recognize a single 75 kDa reversible, and the microtubule inhibitor colchicine has no protein that is present throughout development in approx- effect. These results show that SUmoesin becomes associ- imately equal abundance, and specifically they immuno- ated with apical plasma membrane structures early in precipitate a single protein. We show by immunolocaliza- development, and that SUmoesin is both coincident with tion that SUmoesin has two predominant patterns during actin and requires the assembly of actin filmaments to development. First, SUmoesin is rapidly redistributed after maintain its localization. fertilization from a location throughout the egg to a location in the egg cortex. Later in embryogenesis, Key words: moesin, sea urchin, Lytechinus variegatus, cytochalasin SUmoesin is localized to the apical ends of cells in the B, plasma membrane

INTRODUCTION determine in mammalian cell lines. The push to describe the functions of ERM proteins has been bolstered by a recent The organization of the cortical cytoskeleton into microfila- report on another ERM protein called merlin, which is the gene ments and associated plasma membrane structures is thought associated with human neurofibromatosis 2, may be involved to be influenced by proteins that are encoded by a family of in regulating normal cell growth and may act as a tumor sup- genes known as ezrin, radixin and moesin (now called ERM; pressor (Trofatter et al., 1993). reviewed by Bretscher, 1991; Tsukita et al., 1993). These Other biochemical evidence supports a role for ERM proteins all share extensive similarity at their N terminus to the proteins as cytoskeleton-plasma membrane linkers. Analysis of erythrocyte protein 4.1, which is thought to link the plasma ERM proteins has been difficult, however, because antibodies membrane to the cytoskeleton by binding glycophorin to the generated to individual ERM proteins have cross-reacted with actin/spectrin network in erythrocytes (Conboy et al., 1986). each other, and the tissue distribution of ERM members Because of their structural similarity to 4.1 and talin, ERM overlaps (Berryman et al., 1993). Consistent with what is proteins are presumed to play a role in linking the actin-based known about protein 4.1, the localization of ezrin to the plasma cytoskeleton to the plasma membrane (Rees et al., 1990; membrane and cytoskeleton in cell lines is mediated by the N Lankes and Furthmayr, 1991). This role is supported by the and C terminus, respectively (Algrain et al., 1993). To date, localization of ERM proteins in actin-rich surface structures ERM family members have been shown to interact with actin such as microvilli, membrane ruffles and adherens junctions and the transmembrane molcule CD44 (Tsukita et al., 1989, (Sato et al., 1992). A recent report also demonstrated that ERM 1994; Sato et al., 1991). There is little information about the proteins may be functionally redundant, because they may all distinctive function of individual ERM family members, and be necessary for microvillus assembly and cell-cell adhesion even less is known about these proteins during development. (Takeuchi et al., 1994). These data taken together suggest that Model systems have offered important insights into the the roles of individual ERM proteins may be difficult to mechanisms of cytoskeletal-plasma membrane interactions.

162 E. S. Bachman and D. R. McClay

For example, protein 4.1 is encoded at the locus responsible for Lab.). Gametes were obtained by intracoelomic injection of 0.5 M the coracle mutation in Drosophila (Fehon et al., 1994). This KCl, and eggs were dejellied by passing through cheese cloth and mutation causes a failure of dorsal closure during embryogene- washed 3× with artificial sea water (ASW). Later stage embryos were sis, and mutants in the coracle gene interact with a hypermor- maintained at room temperature with stirring. phic EGFR-Ellipse allele, suggesting the possibility that their Cloning moesin in the sea urchin protein products interact. We felt that understanding the func- Poly(A)+ RNA was prepared at various stages from 0.2 ml packed tion of ERM proteins would be advanced by studying these embryos processed with the Micropure Kit (Pharmacia). One micro- molecules in another model developmental system, the sea gram of RNA from both the midgastrula (14 hours) and pluteus stages urchin. Sea urchin embryogenesis involves a dynamic sequence (24 hours) was used to generate first-strand cDNA using M-MuLV of cytoskeletal-plasma membrane interactions that leads to the reverse transcriptase (New England Biolabs). Degenerate oligonu- formation of microvilli, polarized epithelia and cell-cell junc- cleotide primers to the ERM family (B. McCartney and R. Fehon, tions. Accordingly, we have cloned moesin in the sea urchin, unpublished data) were used in a PCR reaction. Conditions were: and characterized moesin protein during development. 94¡C, 40 seconds; 45¡C, 40 seconds; 72¡C, 40 seconds, for 40 cycles. The actin-based cytoskeleton of the sea urchin changes dra- A 340 bp band was isolated and a second round of PCR was carried out matically. After fertilization, a massive reorganization of the under identical conditions using an internal primer. The resultant 300 bp band was ligated into the TA vector (Invitrogen) and sequencing cytoskeleton occurs when a large G-actin pool is rapidly poly- was performed using the dideoxy method (USB). The clone was merized, forming microvilli and a robust, filamentous inserted into the Bluescript SK− vector, which was used in all subse- cytoskeletal cortex in the zygote (Spudich and Spudich, 1979; quent cloning. Screening of cDNA libraries was performed using 32P- Spudich et al., 1982; Spudich, 1992). A partially defined series labeled probes made to the PCR product and cDNA clones (Strata- of exocytotic events are stimulated by the fertilization reaction, gene). Filters were washed for 15 minutes 3× at high stringency (0.2× and these too involve cytoskeletal proteins (Otto et al., 1980; SSC, 0.1%SDS at 60¡C). All clones were obtained from Lambda Zap Bonder et al., 1989). Proteins such as fascin, α-actinin and libraries made to poly(A)+ RNA isolated from L. variegatus embryos spectrin bundle actin filaments together, but none have been at 2 stages: mid-gastrula (constructed by Stratagene with RNA pro- described that link actin filaments or the cytoskeleton to the duced in this lab.) and prism (kindly provided by Gary Wessel). plasma membrane during sea urchin development (Mabuchi et Northern analysis al., 1985; Bryan, 1986; Fishkind et al., 1987). Later in devel- A 1.2% agarose/ gel (Sambrook et al., 1989) was loaded opment, cells become polarized and some proteins require the with poly(A)+ RNA (5 µg/lane) from the indicated stages, blotted onto actin cytoskeleton for maintenance of this polarity (Nelson, a nylon membrane and hybridized using the 300 bp PCR product as 1988). At least five isoforms of actin are associated with well- a probe. The blot was washed at high stringency (as above), stripped described events such as , cell-cell junction formation and reprobed with a ubiquitin fragment from L. pictus. and maintenance of cell polarity (McClay et al., 1983; Spiegel and Howard, 1983; Davidson, 1986; Andreuccetti et al., 1987). Generation of fusion proteins and antibodies All of these cytoskeletal events can be experimentally manip- Clones used in the generation of fusion proteins were produced using the PCR with specific primers made to the 5′ end of moesin. These ulated in the sea urchin to better understand the role of ERM ′ proteins during embryogenesis. primers contained 5 BamHI sites followed by a sequence specific for moesin. Amplification by the PCR was performed using these 5′ Changes in the plasma membrane also occur during sea primers, the M13 reverse sequencing primer and 20 ng of template urchin development, and some of these may involve ERM pro- DNA from a 1700 bp moesin cDNA clone at 60¡C. A near full-length teins, which are included in a group of proteins that are believed clone (526 amino acids) and a shorter clone (450 amino acids) were to influence the association of the cytoskeleton with the plasma ligated into pGEX-3 and pGEX-2 expression vectors, respectively, membrane (Luna and Hitt, 1992). One such influence may be and induced using 0.7 mM IPTG (Pharmacia). Soluble fusion proteins mediated by an interaction of ERM proteins with the integral were purified using glutathione/agarose beads (Sigma) and injected membrane protein CD44 (Tsukita et al., 1994). Reorganization into a rabbit and three mice (Harlow and Lane, 1988). Serum was and recycling of membranes occurs after fertilization during prepared by ammonium sulfate precipitation followed by affinity µ vesicle exocytosis and microvillus formation. The restructuring purification (Pierce) using an affinity column containing 300 g of of the early embryo as it divides and cells become polarized is the larger (526 amino acid) fusion protein. accompanied by an 100× increase in the surface area to volume Immunofluorescence and western analysis ratio by the blastula stage (Wolpert and Mercer, 1963). Against Early embryos (prehatching) were fixed for 30-45 minutes in 3.7% this backdrop and the developmental potential of sea urchin paraformaldehyde/ASW at RT, while later (posthatching) stages were embryos, we have cloned and characterized a moesin homo- fixed for 10 minutes at −20¡C in methanol. Embryos were washed 3× logue in the sea urchin Lytechinus variegatus. We present evi- in ASW and once in block (10% goat serum/0.1% Triton X- dence for a subcellular distribution of moesin in a structure that 100/ASW), to block nonspecific binding and permeabilize. Multiple is recruited to the plasma membrane after fertilization, and washes in ASW and one wash in block was performed in between the maintained at the plasma membrane/cortex in a manner that primary and secondary antibody incubations. Primary antibodies to requires an intact actin cytoskeleton. SUmoesin were prepared as above, and secondary reagents were all commercially available. Affinity-purified rabbit polyclonal antibody was used at 1:1000 overnight, and secondary Indocarbocyanine (Cy3)- conjugated goat anti-rabbit IgG (Jackson Immuno Research) was incu- MATERIALS AND METHODS bated at 1:500 for 2 hours at RT. Embryos were viewed using confocal microscopy. A mouse anti-actin antibody made to chicken actin (gift Materials from NIH hybridoma bank) was used in double-labeling experiments. Sea urchins (Lytechinus variegatus) were obtained from Susan Decker DAPI (Molecular Probes) and rhodamine/phalloidin (Sigma) were (University of Miami) and Gail Cannon (Duke University Marine used for labeling nuclei and filamentous actin, respectively. DAPI (1 Cloning of moesin in the sea urchin 163 mg/ml) was added with secondary antibody at 1:1000 for 2 hours. Rho- damine/phalloidin was added 1:1 (v/v) from a 4 mg/ml stock in methanol to live embryos for 20 minutes on ice, washed 3× in ASW, and immunostaining was carried out as above with anti-moesin. Protein was prepared for western analysis by homogenizing embryos in 0.1% Triton X-100/150 mM NaCl/1 mM PMSF. Approx- imately 25 µg/lane was loaded as measured by the Bradford assay and confirmed by Ponceau staining following SDS-PAGE (Laemmli, 1970). Western blotting was performed using the moesin primary anti- bodies at 1:1000 dilution and secondary alkaline phosphatase-conju- Fig. 1. Cloning and fusion protein strategy for sea urchin moesin. gated goat anti-rabbit antibody at 1:5000 dilution (Cappel). Leupeptin The full-length ORF and flanking untranslated sequence is indicated ′ ′ (1 µg/ml) was added to some homogenization cocktails to prevent as the 5 -3 sequence with the start and stop position indicated. The proteolysis. black bar with arrows above shows the region that was amplified by PCR and the primers used. Regions used in generating fusion Immunoprecipitation proteins are indicated at the top. cDNA clones that were sequenced Homogenates of sea urchin embryos were prepared by homogenizing are designated c1.6, c8.0 and c7.0. approximately 100 mg wet eggs or embryos in 1 ml of 1.0% Triton X-100 buffer containing 150 mM NaCl, 50 mM Tris, pH 8.0, 2 mM fragment was isolated and ligated into the TA cloning vector PMSF, 1 mM leupeptin and processed with 20 strokes of Dounce homogenization followed by 10 minutes of centrifugation at 10,000 and 10 inserts were sequenced using the dideoxy method. All rpm (7,000 g/p) in a Beckman microfuge. The supernatant (1 ml) was 10 inserts were identical in nucleotide sequence, and were precleared for 2 hours at 4¡C with 40 µl Protein A beads followed by highly similar to ezrin/radixin/moesin by sequence comparison centrifugation to remove the beads. The cleared supernatant was (Altschul et al., 1990). On the basis of deduced amino acid incubated overnight with 1 µg of affinity-purified rabbit polyclonal sequence comparisions (see below), we refer to this as sea antibody, followed by addition of 40 µl Protein A-Sepharose beads urchin moesin, or SUmoesin. The 10 clones were also identical for 2 hours followed by 5 spins at 14,000 rpm (6,000 RCF) with 1 ml to 10 clones generated from a pluteus cDNA pool. This washes in 150 mM NaCl, 0.1% Triton X-100. The washed and boiled SUmoesin clone corresponds to amino acids 200-310 of human beads were resuspended in SDS sample buffer/BME and analyzed by radixin, and was used for further screening and expression SDS-PAGE/western analysis as above. studies. We conclude that the PCR generated a single gene In vivo labeling product at both stages of development. One ml cultures of embryos (10%, v/v) at 5 stages were incubated with 100 µCi of [35S]Met-Cys trans label (ICN) for 1 hour, briefly Isolation of cDNA clones centrifuged and washed 3× in 1.5 ml sea water, homogenized and cen- A near full-length clone containing the start codon and 138 bp trifuged as above. The supernatant (1 ml) was counted by scintilla- of upstream sequence was obtained from a mid-gastrula cDNA tion, standardized to 7000 cpm/ml, precleared 2× with 40 µl Protein library, and overlapping clones covering the remaining 3′ end A beads before adding 1 µg of affinity-purified antibody as described. were obtained from a prism library (Fig. 1). These clones were Washing and analysis by SDS-PAGE was performed as above, but sequenced on both strands using the dideoxy chain termination proteins were detected by autoradiography after impregnating the gel method (USB). The predicted open reading frame of 573 in PPO (Wessel, 1986). amino acids (Mr 68,000; pI 5.31) contains an N-terminal region Cytochalasin B/colchicine experiments that is highly similar to ezrin/radixin/moesin from many Two ml cultures containing approximately 5%, v/v, of dejellied eggs species, while the divergent C terminus does not contain a were fertilized by dilute sperm in the presence of 10 µM PABA polyproline stretch and is most similar to moesin (Fig. 2). The (p-aminobenzoic acid). After 15 seconds, cytochalasin B (5 mg/ml in overall identity of SUmoesin to human radixin and human DMSO) or colchicine (20 mg/ml in water) were added to final con- moesin is 54%. We are therefore using the name SUmoesin centrations of 10 and 100 µM, respectively. Control cultures with the understanding that the sequence differences are not contained 0.1% DMSO. Fertilization was observed to be 100% in clearly distinctive (Lankes and Furthmayr, 1991). There are 4 these experiments (n=4). Fertilization envelopes were stripped by potential consensus tyrosine phosphorylation start sites at µ passing fertilized eggs through 73 m Nitex. After 10-15 minutes, fer- amino acids 161, 192 (261, 262), and 329 according to tilized eggs were fixed in 3% paraformaldehyde/ASW for 30 minutes computer analysis (Geneworks), and these tyrosines have and treated as above. In washout experiments, cultures were washed ′ 3× in ASW over 5 minutes and fixed at the indicated time points. superscripts (P). The 5 start contains the consensus Kozak Later stage embryos were grown in 2 ml samples of 5%, v/v, in sequence and the stop codon is followed by multiple stop Costar culture dishes with gentle shaking. Cytochalasin, colchicine codons. In addition, the predicted secondary structure from the and DMSO control were present for 1 hour at the same concentration carboxy end of the molecule, spanning amino acids 300-500, as above. Recovery from treatment was facilitated by gently spinning is alpha-helical (MacVector). The sequence is identical to other cultures down after 1 hour of treatment, washing replicate cultures 3× ERM family members at the highlighted amino acids. in ASW and letting cultures recover for 30 minutes in ASW. Sea urchin moesin is transcribed late in RESULTS embryogenesis Northern analysis using 5 µg of poly(A)+ RNA per lane was Cloning of an ezrin/radixin/moesin-like homologue performed and the blot was washed under high stringency in the sea urchin using the 300 bp PCR product as a probe (Fig. 3). A 6.3 kb A 300 bp PCR product was amplified from cDNA derived from message is present in the ovary, presumably in oocytes, and is mid-gastrula and pluteus stage sea urchin larvae (Fig. 1). This abundant in adult gut tissue. The mature egg has little or no 164 E. S. Bachman and D. R. McClay

Fig. 2. The predicted amino acid sequence of sea urchin moesin is aligned with human moesin and mouse radixin, where shaded regions indicate identity. The nucleotide sequence has been submitted to Genbank (accession no. U 14180). There is no polyproline stretch of amino acids in sea urchin moesin (underlined). Residues representing consensus tyrosine phosphorylation sites that were derived from computer analysis are denoted by P (Geneworks). The predicted amino acid structure is highly similar to that of other ERM members in the amino half of the molecule, and less so in the C terminus. moesin RNA; expression during embryogenesis begins at the in the sea urchin. The presence of moesin protein in eggs and late blastula stage and peaks at the pluteus larva stage. A during early embryogenesis indicates that moesin protein is ubiquitin control shows that abundant RNA has been loaded in provided maternally, because moesin RNA is absent from each lane. This developmental profile therefore is classified as those early stages. an ‘embryonic late gene’ (Davidson, 1986). Affinity-purified antibody recognizes a single 75 kDa Moesin antibody recognizes a single protein protein throughout embryogenesis Immunoprecipitation experiments were performed: (i) to Homogenates from all stages of development were prepared, confirm the western analysis; (ii) to investigate possible bio- and 25 µg of each was run on 7.5% SDS-PAGE, blotted and visualized by western analysis using affinity-purified rabbit polyclonal antibody (Fig. 4A). A 75 kDa protein is recognized at all stages of development from eggs through the 24 hour pluteus larva stage, and abundance is similar at each stage. Antibodies prepared independently to both fusion proteins (85 kb and 95 kDa) in mice and rabbit gave the same staining pattern by western analysis and by whole-mount embryo immunoflu- orescence. Repeated analysis using the three antibody sources consistently showed a single protein even when blots were overdeveloped or resolved during long runs on 7.5% SDS- PAGE. This antibody is therefore specific for SUmoesin, and does not cross-react with other potential ERM family members

Fig. 3. Northern anaysis of sea urchin moesin. The 300 bp PCR product was used to probe an RNA blot made to poly(A)+ RNA from the stages indicated. The 6.3 kb message is present in the ovary and abundant in the adult gut, but is absent in eggs. Moesin message increases in amount steadily during embryogenesis, after its first appearance in late blastulae. An L. pictus ubiquitin probe was used as a loading control. M, mesenchyme. Cloning of moesin in the sea urchin 165

to be synthesized through the feeding pluteus larval stage. The kDa pattern of coprecipitated proteins is similar at all stages. The band that migrated at 75 kDa in the metabolically labeled material is identical in size to a protein that is detected by western analysis after immunoprecipitation using nonradioac- tive cell extracts (Fig. 5C). Moesin is localized to the plasma membrane after fertilization To investigate the cellular distribution of moesin, whole-mount indirect immunofluorescence imaging using confocal microscopy was performed on eggs and early fertilized embryos. In eggs, moesin is distributed in a punctate pattern Fig. 4. Moesin protein is present throughout embryogenesis. A throughout the cytoplasm (Fig. 6A). Within 10 minutes after developmental western blot is shown using 25 µg of total protein per fertilization, moesin becomes concentrated just beneath the stage. The protein was run on 7.5% SDS-PAGE, blotted onto plasma membrane in the egg cortex; over the same interval the nitrocellulose and incubated with affinity-purified rabbit anti- punctate cytoplasmic distribution becomes less brightly fluo- SUmoesin antibody followed by alkaline phosphatase-conjugated secondary antibody. A 75 kDa protein is present in approximately rescent (Fig. 6B). Embryos from 2-cell and early blastula equal amounts at the 5 stages indicated. developmental stages are shown in surface and cross-sectional views by confocal microscopy (Fig. 6C-F). SUmoesin in these embryos is seen in the cell cortex in a predominately punctate chemical interactions between moesin and other proteins; and distribution and continues to be seen in the cytoplasm at lower (iii) to learn when SUmoesin is translated (Fig. 5). Embryos 35 levels. The patchy staining on the surface of early embryos were incubated in [ S]Met-Cys at five stages of development may correspond to the location of microvilli. In the morula and SUmoesin was analyzed by immunoprecipitation using µ stage, moesin is localized at the cell cortex in the region of cell- 1 g of affinity-purified antibody, followed by SDS-PAGE and cell contacts and on the apical sides of cells (Fig. 6E). There autoradiography. A band of 75 kDa is first detected at the continues to be staining within the cytoplasm in a punctate hatching blastula stage. Faint bands at 205, 116 and 50 kDa pattern as well (Fig. 6F). Controls with preimmune serum, coprecipitated in both the beads only and antibody/beads preimmune serum with secondary antibody and secondary fractions (Fig. 5A). At least one of these (116 kDa) may cor- antibody alone were blank. respond to echinonectin, which is a sea urchin protein that has a high affinity for Sepharose beads (Alliegro et al., 1990). Moesin is associated with the region of cell-cell Labeled moesin protein is not detected in early embryonic contacts in later embryogenesis stages (Fig. 5B), a result that can be explained by the previous Embryos at hatched blastula, gastrula and feeding pluteus observation that moesin mRNA is not present in early stages are shown in Fig. 7. Surface (A,C) and cross-sectional embryonic stages (Fig 3). We conclude that zygotic SUmoesin (B,D) views of two stages are seen. Moesin is concentrated in is not translated until the blastula stage, after which it continues intercellular borders (Fig. 7A,C,E), and appears in the apical

Fig. 5. Immunoprecipitation of SUmoesin. (A) Extracts of [35S]Met-Cys-labeled embryos were homogenized, pelletted, precleared twice (beads 1 and 2 lanes) and incubated with 1-2 µg of anti-moesin antibody are shown from left to right in progression (A). A predominant band at 75 kDa is seen in the immunoprecipitate, with other bands faintly seen at 210, 116 and 55 kDa. Adding twice the amount of antibody gave a small increase in protein, including coprecipitating bands. (B) The same immunoprecipitated lanes from different stages are seen, and 75 kDa SUmoesin is not translated 1 hour after fertilization (1 hour lane), but is labeled during the mesenchyme blastula stage. (C) Western blot analysis of immunoprecipitated material from nonradioactive cell extracts, using affinity-purified antibody to SUmoesin, recognizes a single band at 75 kDa and the expected 55 kDa rabbit IgG antibody that brought down the immunoprecipitate. Nonspecific binding to beads used in preclearing was not detectable by western blot (BEADS). 166 E. S. Bachman and D. R. McClay

Fig. 6. Distribution of SUmoesin protein in early embryos as analyzed by indirect immunofluorescence and confocal microscopy. Eggs and embryos were fixed in paraformaldehyde and stained with affinity-purified rabbit anti-moesin antibody followed by Cy3-conjugated goat anti- rabbit secondary antibody. The stages indicated are egg (A), fertilized egg at 10 minutes (B), 2-cell embryos, surface view (C) and cross-section (D). After fertilization moesin is associated with surface structures and in close association with the plasma membrane. Morula and early blastula embryos are seen in surface and cross-sectional views (E, F). Bar, 20 µm. region where adherens junctions are known to be (Fig. probes. Moesin in fertilized zygotes is concentrated in the 7B,D,F), and in the apical regions of cells in the ectoderm and cortex near the plasma membrane, where actin is abundant as endoderm. Moesin is not present in mesenchyme tissues. In the judged by anti-actin antibody (Fig. 8A,B). Both moesin and pluteus larva, moesin is concentrated in the apical region of actin are abundant in the cortex of fertilized eggs, but the cell-cell contacts in the ectoderm, oral hood and mouth (7 E,e). merged image shows that moesin retains a cytoplasmic, Moesin staining is noticeably reduced in aboral and perianal punctate distribution as well (Fig. 8C). Surface staining of ectoderm cells. In endoderm, moesin is concentrated between moesin and actin in these fertilized eggs is seen as a pattern of cells at the apical surface of endoderm cells of the fore, mid- small particulate structures that may be microvilli (Fig. 8a,b). and hindgut (anus) regions (Fig. 7E). Merging the red and green channels shows the overlapping dis- tribution of moesin (green, FITC) and actin (red, Cy3) images Moesin is concentrated in actin-rich regions of the in yellow on the surface (Fig. 8c). Gastrula stage embryos were embryo fixed in 50% methanol, 4 mg/ml rhodamine/phalloidin for 30 On the basis of the location of moesin protein: (i) in the cortex minutes and analyzed using the same procedure as described of fertilized eggs, and (ii) in the region of cell-cell contacts above for late stage embryos. A surface view of moesin, fila- above, we asked whether moesin was localized in regions that mentous actin and their overlapping distribution is shown (Fig. contain high concentrations of actin at these two stages of 8D,E,F). embryogenesis. We analyzed the distribution of moesin and actin by double labeling embryos with rabbit anti-moesin Moesin is not recruited to the plasma membrane in (FITC-conjugated secondary antibody) and either mouse anti- the presence of cytochalasin B actin antibody or rhodamine/phalloidin (filamentous actin) The rapid redistribution of moesin after fertilization, from Cloning of moesin in the sea urchin 167

Fig. 7. Moesin localization in post- hatched embryos. Surface and cross- sectional confocal microscopic views of moesin are shown for blastula and gastrula stages. SUmoesin is localized to the region of cell-cell contacts (A), and is found in an apical location in the region of cell-cell junctions (B), in hatched blastula stage embryos. SUmoesin is in the region of cell-cell contacts in an apical location in gastrula stage embryos (C,D). In pluteus larvae, moesin is concentrated between cells mostly in the oral hood and endoderm regions of the pluteus larva (E). A magnified view of the oral hood and mouth is shown (e). Bar, 20 µm. throughout the cytoplasm to the plasma membrane, led us to a halo around the cell by confocal microscopy (Fig. 9A). ask what cytoskeletal proteins might be involved. In particu- Adding cytochalasin B, however, prevented localization of lar, we asked whether moesin requires microtubules or micro- moesin to the cell cortex (Fig. 9B). The effect of cytochalasin filaments to be associated with the plasma membrane after fer- B is reversible. Moesin staining near the plasma membrane in tilization. Microfilament and microtubule assembly are cytochalasin-treated cells is restored by rinsing 3× in ASW for disrupted in the presence of cytochalasin B and colchicine, 5 minutes befor fixation (Fig. 9C). respectively (Banzhaf et al., 1980). Cytochalasin B enters and exits cells within seconds. Accordingly, we added inhibitors of Moesin distribution is disrupted by blocking actin microtubule and microfilament formation 15 seconds after fer- polymerization tilizing eggs with dilute sperm. After 10 minutes, embryos We next asked whether the polarized, cortical distribution of were fixed and stained with anti-SUmoesin antibody as moesin seen at later stages of embryogenesis requires an intact described above. Control embryos in 0.1% DMSO looked like cytoskeleton. We again used colchicine and cytochalasin B to those shown in Fig. 6B. Colchicine-treated zygotes showed inhibit the de novo assembly and maintenance of microtubules staining concentrated at the plasma membrane that appears like and microfilaments, respectively. Embryos at the morula (4 168 E. S. Bachman and D. R. McClay

Fig. 8. Moesin is found in regions of the cell that are rich in actin. Fertilized eggs (10 minutes) are seen by indirect immunofluorescence using confocal microscopy to detect rabbit anti-moesin (A,a), and mouse anti-actin (B,b) antibodies by creating a merged image of the FITC (SUmoesin) and Cy3(actin) signals, respectively (C,c). Coincident signals (green and red) are seen as yellow. Bar, 20 µm. Gastrula stage embryos are viewed en face and analyzed as above except anti-moesin antibody-FITC (D), or rhodamine/phalloidin-stained embryos (E) are shown and the signals are merged (F). Overlapping signal is seen in yellow in the region of cell borders. hours) and hatched blastula (10 hours) stages were incubated localization of moesin is reversible. When embryos are as above in 0.1% DMSO, or in DMSO plus colchicine or incubated for 1 hour in cytochalasin B, washed thoroughly and cytochalasin for either: (i) 1 hour; or (ii) 1 hour followed by incubated in ASW alone for 30 minutes, staining returns near washing 3× in ASW and 30 minutes of recovery in ASW the plasma membrane in a confluent, apical pattern. At the before fixation. Embryos were fixed in methanol, stained with blastula stage (10 hours), colchicine again has no effect on antibody and analyzed by indirect confocal microscopy. In moesin distribution (Fig. 9G). Cytochalasin B, however, both colchicine- and cytochalasin-B-treated embryos, there are disrupts the continuous staining of moesin at the plasma binucleate cells that have not completed . This is membrane and the organization is again subapical and patchy expected if microtubules or microfilaments are unable to form (Fig. 9H). While moesin remains towards the apical end of and take part in karyokinesis or cytokinesis (Cande et al., cyotchalasin-treated cells, staining becomes punctate and 1977). Moesin remains at the plasma membrane in an apical patchy, and staining is no longer restricted to the regions of location in the presence of colchicine and DMSO (Fig. 9D). In cell-cell junctions. Association of SUmoesin with the plasma the presence of cytochalasin B, however, the staining pattern membrane is partly reversed when embryos are incubated for at the plasma membrane is disrupted and appears patchy, 1 hour and allowed to recover for 30 minutes (Fig. 9I). The vesicular and subapical (Fig. 9E). This disruption of cortical confluent appearance of moesin at the apical end of blas- Cloning of moesin in the sea urchin 169

Fig. 9. Moesin distribution is disrupted in the presence of microfilament inhibitors. Three stages of sea urchin embryogenesis were analyzed by indirect immunofluorescence and confocal microscopy after incubation in colchicine (100 µm, A,D,G) or cytochalasin (10 µm, B,E,H) inhibitors. The stages correspond to 10 minutes after fertilization (A,B,C), morula stage (D,E,F) and hatched blastula (G,H,I). Fertilized eggs were incubated for 10 minutes in cytochalasin B or colchicine and recovered for 5 minutes (C). Other embryos were incubated in ASW for 30 minutes after being washed from the cytochalasin B after 1 hour of treatment (F,I). tomeres in this confocal image is mostly restored. We conclude line stretch of amino acids that is found in ezrin and radixin. that the apical localization of moesin is disrupted in the Amino acids predicted from the nucleotide sequence reveals presence of reagents that prevent actin polymerization, but is four consensus tyrosine phosphorylation sites, two of which unaffected in the presence of microtubule inhibitors such as are shared by other ERM family members. Phosphorylation of colchicine. Moesin therefore requires an intact actin cytoskele- ezrin has been described in EGF-stimulated carcinoma cells, ton to remain associated with the plasma membrane. gastric parietal cells and tumor transplantation antigens (Gould et al., 1986, 1989; Hanzel et al., 1991; Yang and Tonks, 1991; Egerton et al., 1992; Krieg and Hunter, 1992; Fazioli et al., DISCUSSION 1993). At present, no evidence exists for direct regulation of ezrin by phosphorylation. It will be interesting to find out While ERM proteins have been examined in cell lines and whether SUmoesin is phosphorylated in any of its diverse isolated differentiated tissues, there have been no reports of developmental patterns. these proteins as they change during the development of an The original description of ERM proteins in mammals organism. In addition, the analysis in this report of what showed that they can overlap in distribution, and may be func- appears to be a single ERM protein simplifies an approach to tionally redundant. Cross-reactivity of antibodies among ERM understanding ERM proteins by looking at what is perhaps an family members originally made interpretation of ERM data ancestral protein to the family. What role does moesin have difficult. For that reason, we attempted to be thorough in estab- during reorganization of the cytoskeleton, establishment of cell lishing that we were investigating a single ERM protein in the polarity and formation of cell-cell junctions in the sea urchin sea urchin. SUmoesin is a single ERM family member, based embryo? We asked these questions as they pertain to the sea on the following evidence: (i) the antibodies used in this study urchin embryo by cloning and characterizing moesin, and then were affinity purified; (ii) the SUmoesin protein that we studying its distribution changes during development. describe is always a single species as shown by SDS-PAGE The ERM genes are so highly similar at the nucleotide and and western blot analysis, and that is true for three indepen- amino acid level that identifying a sea urchin protein as a par- dent antibodies; (iii) western blot analysis described above ticular ERM homologue is equivocal. We call this ERM shows only a single band even when low percentage gels are member SUmoesin, because of its similarity to other moesin run for extended times; (iv) in vivo labeling experiments show members within the C-terminal, divergent half of these that the rabbit polyclonal antibody described here specifically molecules. The 75 kDa size of the mature SUmoesin protein immunoprecipitates SUmoesin. If ezrin or other ERM also resembles most closely the ERM family, which is also members exist in the sea urchin, then these do not seem to consistent with this nomenclature. SUmoesin lacks a polypro- associate heterotypically with moesin: an interaction that 170 E. S. Bachman and D. R. McClay occurs in human cell lines (Gary and Bretscher, 1993). The of the actin-based cytoskeleton and SUmoesin suggest that faint band seen at 78 kDa in the in vivo translation experiment SUmoesin may be constitutively involved in interactions could represent a heterotypic association with another ERM between the plasma membrane and the cytoskeleton. This sub- member, but this protein is not recognized by the SUmoesin cellular structure remains to be identified. The function(s) of antibody. We conclude that our description of SUmoesin SUmoesin in the during embryogenesis may be regulated by during sea urchin embryogenesis follows the distribution of a SUmoesin itself or by the group of proteins that SUmoesin can single protein in the ERM family. interact with. We are investigating the possibility that The relocation of SUmoesin, from punctate structures in the SUmoesin is regulated by phosphorylation during develop- cytoplasm to discrete regions of localization at the plasma ment. membrane later in development, is reminiscent of the changes Whether there exist(s) one or several ERM family members in ezrin distribution during chicken erythrocyte differentiation in the sea urchin will be of interest with regard to the evolution (Birgbauer and Solomon, 1989). This movement may represent of the ERM family that now includes the tumor suppressor, an association of the protein with vesicles that are trafficked to merlin. At least one gene duplication event may have occured the plasma membrane. Ezrin, for example, is known to be during evolution, according to one report on human radixin recruited from tubulovesicles to the apical plasma membrane (Wilgenbus et al., 1993). If the sea urchin retains a smaller in gastric parietal cells, and this recruitment is concurrent with complement of ERM genes than there are in mammals, then a transient phosphorylation of ezrin (Hanzel et al., 1991). the analysis of ERM function(s) will be simplified. We are SUmoesin could be free in the egg, or associated with a class investigating the possibility that other ERM genes exist in the of vesicles that is recruited to the cell surface after fertilization sea urchin. If they do exist, they are unlike some of their ver- (McClay et al., 1990). The location of SUmoesin in the egg is tebrate counterparts because they do not cross-react immuno- distinct from the pool of G-actin, and seems to correspond to logically. It is also possible that SUmoesin may fulfill the role a region internal to the egg cortex described as Z3 (Bonder et of two proteins in mammals. The distribution of moesin in the al., 1989). sea urchin during development is similar to that of two Later in development, SUmoesin is located apically in cells different ERM members. Mammalian ezrin and moesin are in the region of cell-cell junctions. We asked how that apical found in microvilli, and membrane ruffles, and colocalize with polarity was maintained. Embryos that were incubated in actin to the cell cortex (Franck et al., 1993). SUmoesin is colchicine and cytochalasin B remained viable, but after one confined in later development to the apical surface and in hour formed binucleate cells because of incomplete cytokine- regions of cell-cell junctions, which is reminiscent of radixin sis. Cytochalasin B clearly disrupts the apical, plasma (Tsukita et al., 1989). These regions of the embryo are also rich membrane-associated appearance of moesin, wheras micro- in actin filaments. tubule inhibitors have no effect. These experiments were all We are currently testing the hypothesis that moesin is asso- reversible, indicating that moesin distribution recovers when ciated with membrane and/or cytoskeletal proteins. The avail- actin polymerization can proceed. The kinetics of cytochalasin ability of an embryonic system with a predictable temporal B and D are similar at 10 µM that was used in this study pattern of moesin localization should provide an excellent (Cooper, 1987). This reversibility argues that cytochalasin is system for this analysis. not toxic to the cell(s), although an effect on pre-established actin filaments must also be considered. Finally, it is possible We thank Dr Richard Fehon (Duke University) for providing that the plasma membrane may be disrupted and internalized primers used to clone moesin and for helpful comments. The mono- reversibly in response to cytochalasin B. Our observations are clonal antibody developed by Jim Jung-Ching Lin was obtained from consistent with a role for polymerized actin in the establish- the Developmental Studies Hybridoma Bank maintained by the Department of Pharmacology and Molecular Sciences, Johns Hopkins ment of polarity and the maintenance of SUmoesin at the apical University School of Medicine, Baltimore, MD 21205, and the ends of cells. These data do not reveal whether moesin interacts Department of Biological Sciences, University of Iowa, Iowa City, IA directly with actin or is influenced indirectly by the actin 52242, under contract N01-HD-2-3144 from the NICHD. I thank John cytoskeleton to assume its apical location in the cortex of cells. Matese for sequencing the PCR products. This work was supported At least three different patterns of SUmoesin distribution by NIH grant no. 5R01 HD 14483 to Dr David R. McClay. were observed during development: (i) the actin-rich cell cortex of fertilized eggs; (ii) the apical, plasma membrane- associated distribution as soon as cells become polarized; and (iii) in the apical regions of cell-cell junctions in later devel- REFERENCES opment. These diverse patterns, and their reversible disruption Algrain, M., Turunen, O., Vaheri, A., Louvard, D. and Arpin, M. (1993). by cytochalasin, suggest that SUmoesin may be involved Ezrin contains cytoskeleton and membrane binding domains accounting for dynamically in the organization of the these structures. This its proposed role as a membrane-cytoskeletal linker. J. Cell Biol. 120, 129- role for SUmoesin would be distinct from those of actin 139. bundling or actin organizing proteins, and would place Alliegro, M. C., Burdsal, C. A. and McClay, D. R. (1990). In vitro biological activities of echinonectin. Biochemistry 29, 2135-2141. SUmoesin in a position more proximate to the plasma Altschul, S. F., Gish, W., Miller, W., Myers, E. W. and Lipman., D. J. membrane. Whether another molecule, perhaps associated with (1990). Basic local alignment search tool. J. Mol. Biol. 215, 403-410. membranes, associates with SUmoesin and the cytoskeleton Andreuccetti, P., Lumaga, M. R. B., Cafiero, G., Filosa, S. and Parisi, E. after fertilization remains to be shown in the sea urchin. A (1987). Cell junctions during the early development of the sea urchin embryo (Paracentrotus lividus). Cell Diff. 20, 137-146. recent study established a link between moesin and CD44 Banzhaf, W. C., Warren, R. H. and McClay, D. R. (1980). Cortical (Tsukita et al., 1994). reorganization following fertilization of sea urchin eggs: sensitivity to These data that show a potential link between the formation cytochalasin B. Dev. Biol. 80, 506-515. Cloning of moesin in the sea urchin 171

Berryman, M., Franck, Z. and Bretscher., A. (1993). Ezrin is concentrated in Tsukita, S. (1985). Alpha-actinin from sea urchin eggs: biochemical the apical microvilli of a wide variety of epithelial cells whereas moesin is properties, interaction with actin, and distribution in the cell during found primarily in endothelial cells. J. Cell Sci. 105, 1025-1043. fertilization and cleavage. J. Cell Biol. 100, 375-383. Birgbauer, E. and Solomon, F. (1989). A marginal band-associated protein McClay, D. R., Cannon, G. W., Wessel, G. M., Fink, R. D. and Marchase, has properties of both microtubule- and microfilament-associated proteins. J. R. D. (1983). Patterns of antigenic expression in early sea urchin Cell Biol. 109, 1609-1620. development. In Time, Space and Pattern in Embryonic Development. pp. Bonder, E. M., Fishkind, D. J., Cotran, N. M. and Begg, D. A. (1989). The 157-169. Alan R. Liss, Inc., New York. cortical actin-membrane cytoskeleton of unfertilized sea urchin eggs: McClay, D. R., Alliegro, M. C. and Black, S. D. (1990). The ontogenetic analysis of the spatial organization and relationship of filamentous actin, appearance of extracellular matrix during sea urchin development. In nonfilamentous actin, and egg spectrin. Dev. Biol. 134, 327-341. Organization and Assembly of Plant and Animal Extracellular Matrix (ed. R. Bradford, M. M. (1976). A rapid and sensitive method for the quantitation of Raff and W. Jeffreys), pp. 1-14. Academic Press, New York. microgram quantities of protein utilizing the principle of protein dye binding. Nelson, J. H. and McClay, D. R. (1988). Cell polarity in sea urchin embryos: Anal. Biochem. 72, 248-254. reorientation of cells occurs quickly in aggregates. Dev. Biol. 127, 235-247. Bretscher, A. (1991). Microfilament structure and function in the cortical Otto, J. J., Kane, R. E. and Bryan, J. (1980). Redistribution of actin and cytoskeleton. Annu. Rev. Cell Biol. 337-374. fascin in sea urchin eggs after fertilization. Cell Motil. 1, 31-40. Bryan, J. (1986). Isolation of fascin, an actin-bundling protein, and SU45, an Rees, D. J. G., Ades, S. E., Singer, S. J. and Hynes, R. O. (1990). Sequence actin-severing/capping protein from sea urchin eggs. Meth. Enzymol. 134, and domain structure of talin. Nature 347, 685-689. 12-23. Sambrook, J., Fritsch, E. F. and Maniatis, T. (1989). Molecular Cloning: A Cande, W. Z., Lazrides, E. and McIntosh, J. R. (1977). A comparison of the Laboratory Manual. Cold Spring Harbor Laboratory Press, Cold Spring distribution of actin and tubulin in the mammalian mitotic spindle as seen by Harbor, New York. indirect immunofluorescence. J. Cell Biol. 72, 552-567. Sato, N., Yonemura, S., Obinata, T., Tsukita, S. and Tsukita, S. (1991). Conboy, J., Kan, Y. W., Shohet, S. B. and Mohandas, N. (1986). Molecular Radixin, a barbed end-capping actin modulating protein, is concentrated at cloning of protein 4.1, a major structural element of the human erythrocyte the cleavage furrow during cytokinesis. J. Cell Biol. 113, 321-330. membrane skeleton. Proc. Nat. Acad. Sci. USA 83, 9512-9516. Sato, N., Funayama, N., Nagafuch, A., Yonemura, S., Tsukita, S. and Cooper, J. A. (1987) Effects of cytochalasin and phalloidin on actin. J. Cell Tsukita, S. (1992). A gene family consisting of ezrin, radixin and moesin. J. Biol. 105, 1473-1478. Cell Sci. 103, 131-143. Davidson, E. H. (1986). Gene Activity in Early Development, Academic Press, Spiegel, E. and Howard, L. (1983). Development of cell junctions in sea Orlando. urchin embryos. J. Cell Sci. 62, 27-48. Egerton, M., Burgess, W. H., Chen, D., Druker, B. J., Bretscher, A. and Spudich, A. and Spudich, J. A. (1979). Actin in triton-treated cortical Samelson, L. E. (1992). Identification of ezrin as an 81-kDa tyrosine- preparations of unfertilized and fertilized sea urchin eggs. J. Cell Biol. 82, phosphorylated protein in T cells. J. Immunol. 149, 1847-1852. 212-226. Fazioli, F., Wong, W. T., Sakaguch, K., Appella, E. and DiFiore, P. P. Spudich, A., Giffard, R. G. and Spudich, J. A. (1982). Molecular aspects of (1993). The ezrin-like family of tyrosine kinase substrates: receptor-specific cortical actin filament formation upon fertilization. Cell Diff. 11, 281-284. pattern of tyrosine phosphorylation and relationship to malignant Spudich, A. (1992). Actin organization in the sea urchin egg cortex. In transformation. Oncogene 8, 1335-1345. Cytoskeleton in Development (ed. E. L. Bearer), pp. 9-21. Academic Press, Fehon, R. G., Dawson, I. A. and Artavanis-Tsakonas, S. (1994). A San Diego. Drosophila homologue of membrane-skeleton protein 4.1 is associated with Takeuchi, K., Sato, N., Kasahara, H., Funayama, N., Nagafuchi, A., septate junctions and is encoded by the coracle gene. Development 120, 545- Yonemura, S., Tsukita, S. and Tsukita, S. (1994). Perturbation of cell 557. adhesion and microvilli formation by antisense oligonucleotides to ERM Fishkind, D. J., Bonder, E. M. and Begg, D. A. (1987). Isolation and family members. J. Cell Biol. 125, 1371-1384. characterization of sea urchin egg spectrin: calcium modulation of the Trofatter, J. A., MacCollin, M. M., Rutter, J. L., Murrell, J. R., Duyao, M. spectrin-actin interaction. Cell Motil. Cytoskel. 7, 303-314. P., Parry, D. M., Eldridge, R., Kley, N., Menon, A. G., Pulaski, K., Haase, Franck, Z., Gary, R. and Bretscher, A. (1993). Moesin, like ezrin, colocalizes V. H., Ambrose, C. M., Munroe, D., Bove, C., Haines, J. L., Martuza, R. with actin in the cortical cytoskeleton in cultured cells, but its expression is L., MacDonald, M. E., Seizinger, B. R., Priscilla Short, M., Buckler, A. J. more variable. J. Cell Sci. 105, 219-231. and Gusella, A. (1993). A novel moesin-, ezrin-, radixin-like gene is a Gary, R. and Bretscher, A. (1993). Heterotypic and homotypic associations candidate for the neurofibromatosis 2 tumor suppressor. Cell 72, 791-800. between ezrin and moesin, two putative membrane-cytoskeletal linking Tsukita, S., Hieda, Y. and Tsukita, S. (1989). A new 82-kD barbed end- proteins. Proc. Nat. Acad. Sci. USA 90, 10846-10850. capping protein (radixin) localized in the cell-to-cell adherens junction: Gould, K. L., Cooper, J. A., Bretscher, A. and Hunter, T. (1986). The purification and characterization. J. Cell Biol. 108, 2369-2382. protein-tyrosine kinase substrate, p81, is homologous to a chicken Tsukita, S., Itoh, M., Nagafuchi, A., Yonemura, S. and Tsukita, S. (1993). microvillar core protein. J. Cell Biol. 102, 660-669. Submembranous junctional plaque proteins include potential tumor Gould, K. L., Bretscher, A., Esch, F. S. and Hunter, T. (1989). cDNA suppressor molecules. J. Cell Biol. 123, 1049-1053. cloning and sequencing of the protein-tyrosine kinase substrate, ezrin, Tsukita, S., Oishi, K., Sato, N., Sagara, J., Kawai, A. and Tsukita, S. (1994). reveals homology to band 4.1. EMBO J. 8, 4133-4142. ERM family members as molecular linkers between the cell surface Hanzel, D., Reggio, H., Bretscher, A., Forte, J. G. and Mangeat, P. (1991). glycoprotein CD44 and actin-based cytoskeletons. J. Cell Biol. 126, 391- The secretion-stimulated 80K phosphoprotein of parietal cells is ezrin, and 401. has properties of a membrane cytoskeletal linker in the induced apical Wessel, G. M. (1986). On the structure, ontogeny and function of the microvilli. EMBO J. 10, 2363-2373. ultracellular matrix in the sea urchin embryo. Thesis, Duke University. Harlow, E. and Lane, D. (1988). Antibodies: A Labratory Manual. Cold Wilgenbus, K. K., Milatovich, A., Francke, U. and Furthmayr, H. (1993). Spring Harbor Laboratory Press, Cold Spring Harbor, New York. Molecular cloning, cDNA sequence, and chromosomal assignment of the Krieg, J. and Hunter, T. (1992). Identification of the two major epidermal human radixin gene and two dispersed pseudogenes. Genomics 16, 199-206. growth factor-induced tyrosine phosphorylation sites in the microvillar core Wolpert, L. and Mercer, E. H. (1963). An electron microscope study of the protein ezrin. J. Biol. Chem.. 267, 19258-19265. development of the blastula of the sea urchin embryo and its radial polarity. Laemmli, U. K. (1970). Cleavage of structural proteins during the assembly of Exp. Cell Res. 30, 280-300. the head of bacteriophage T4. Nature 227, 680-685. Yang, Q. and Tonks, N. K. (1991). Isolation of a cDNA clone encoding a Lankes, W. T. and Furthmayr, H. (1991). Moesin: a member of the protein human protein-tyrosine phosphatase with homology to the cytoskeletal- 4.1-talin-ezrin family of proteins. Proc. Nat. Acad. Sci. USA 88, 8297-8301. associated proteins band 4.1, ezrin, and talin. Proc. Nat. Acad. Sci. USA 88, Luna, E. J. and Hitt, A. L. (1992). Cytoskeleton-plasma membrane 5949-5953. interactions. Science 258, 955-963. Mabuchi, I., Hamaguchi, Y., Kobayashi, T., Hosoya, H., Tsukita, S. and (Received 10 August 1994 - Accepted 28 September 1994)