ABSTRACT
The expression and intracellular distribution of four contractile proteins (actin, myosin, caldesmon and tropomyosin) in normal fibroblasts and their transformed counterparts by Rous or avian sarcoma virus were compared. By analyzing the isoformal expression of actin, caldesmon and tropomyosin using two-dimensional gel electrophoresis, only tropomyosin showed significant alteration in its isoformal expression accompanied by transformation. Morphological study revealed that in normal cells, myosin, caldesmon and tropomyosin were distributed periodically along stress fibers, but were excluded from focal adhesions (adhesion plaques), at which stress fibers terminate. By contrast, the contractile proteins were concentrated within the protrusions of the ventral cell surface of transformed cells, which are cell-adhesive structures with high motility (podosomes). Regional analysis indicated that the contractile proteins do not show diffuse distribution within podosomes. Myosin, some caldesmon and tropomyosin in association with F-actin were localized in the region surrounding the core domains of podosomes. A major part of the caldesmon was, however, located in the core domain with short F-actin bundles. In order to compare the stability and the molecular organization of stress fibers with that of the short F-actin bundles within podosomes, the dorsal plasma membranes of the cells were removed by lysis and squirting. Then, the ruptured cells were treated with various buffers containing high salt, ATP or Ca2 /calmodulin. Myosin, caldesmon and tropomyosin were strongly associated with stress fibers of the ruptured normal fibroblasts even in a buffer containing high salt or Ca2 /calmodulin. On the other hand, myosin and tropomyosin within podosomes were easily extracted by lysis and squirting. And, the remaining caldesmon in podosomes was separated from the short F-actin bundles with high salt or Ca2 /calmodulin buffer. The present findings suggest that the high motility of podosomes from transformed cells is based on the actomyosin system, and that the stable adherence of focal adhesions of normal cells is due to a lack of this system. The accumulation of contractile proteins and their dynamic association within podosomes might be the cause of the short half-life of the structures. In relation to its localization in the core domain of podosomes without myosin and tropomyosin, the function of caldesmon has been discussed.
INTRODUCTION
When transformed by Rous sarcoma virus (RSV), fibroblasts alter their well-spread shapes into rounded ones and lose their prominent stress fibers. Upon transformation, stable adhesive structures of normal cells (focal adhesions, adhesion plaques, focal contacts or immotile contacts) disappear and distinctive adhesive structures are newly formed. These structures are called rosettes, rosette contacts (David-Pfeuty and Singer, 1980), podosomes (Tarone et al., 1985) or motile contacts (Sobue, 1990). Podosomes exhibit a longer distance from the tip of contact to the substratum when compared with focal adhesions (Tarone et al., 1985; Chen, 1989). Cellular phenotypic changes induced by tumor promoters such as phorbol esters transiently resemble those produced by RSV in terms of cytoskeletal rearrangement and podosome formation. Reorganization of some membrane skeletal proteins has been investigated during transformation by RSV or by tumor promoters. For example, vinculin, a-actinin (David-Pfeuty and Singer, 1980) and talin (Burridge and Connel, 1983; Marchisio et al., 1987) concentrated in focal adhesions are translocated into podosomes. Nonerythroid spectrin (calspectin or fodrin), which is another major component of the membrane skeleton and undercoats the plasma membrane of normal cells, is specifically accumulated into podosomes of transformed cells by phorbol ester (Sobue et al., 1988) or by RSV (Sobue et al., 1989). Not only transformed cells but also macrophages and osteoclasts have podosomes, in which a-actinin, talin and vinculin are accumulated as well (Marchisio et al., 1987; Zambonin-Zallone et al., 1989).
Podosomes can move vertically and transversely on a time scale of minutes while they degrade the extracellular matrix with proteases (Chen, 1989). Such characteristics of podosomes have been well documented in relation to metastasis and invasion by malignant cells. In comparison with this proteolytic activity of the structures, less attention has been paid to the molecular basis of their high motility, which is another important feature for metastasis and invasion. It would be reasonable to consider that active movement of podosomes is caused by the actomyosin system, which converts the chemical energy of ATP into mechanical force.
Recent biochemical studies have suggested that the actomyosin system in smooth and nonmuscle cells is dually regulated by myosin-linked and actin-linked mechanisms (reviewed by Sobue and Sellers, 1991). The myosin-linked mechanism is based on phosphorylation of myosin by Ca2+/calmodulin-dependent myosin light-chain kinase and dephosphorylation by myosin phosphatase. Caldesmon and tropomyosin are crucial components in the actin-linked mechanism. The former protein is an inhibitory factor for the actin-myosin interaction, in which the caldesmoninduced inhibition can be released by Ca2+/calmodulin (Sobue et al., 1982, 1985a). Recently, the two isoforms of caldesmon have been identified in a wide variety of tissues and cells. They can be easily distinguished by their mobilities in sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE): h-caldesmon (Mr of 120,000 to 150,000) and l-caldesmon (Mr of 70,000 to 80,000); the Mr values of two isoforms deduced from their primary structures are much smaller than those calculated by SDS-PAGE (Hayashi et al., 1989b, 1991; Bryan et al., 1989). h-Caldesmon is predominantly expressed in smooth muscle cells, whereas l-caldesmon is widely distributed in non-muscle cells (Sobue et al., 1985b; Ueki et al., 1987). Immunocytochemically, l-caldesmon in normal fibroblasts is found to be distributed with periodicity along stress fibers (bundles of actin filaments) (Owada et al., 1984; Bretscher and Lynch, 1985). Tropomyosin is a potent modulator of the actomyosin system. The intracellular distibution of tropomyosin is the same as that of l-caldesmon (Bretscher and Lynch, 1985). Despite such evidence, no one has investigated the regional distribution of contractile proteins within cell-adhesive structures such as focal adhesions and podosomes.
In this investigation, we have performed comparative studies on the localization of contractile proteins in the above two cell-adhesive structures. As a result, we have identified myosin, caldesmon and tropomyosin in association with actin filaments within podosomes, but not in focal adhesions. We have obtained further evidence that the contractile system within podosomes is less stably constructed than that in stress fibers.
MATERIALS AND METHODS
Cell culture
BY1 cell is a clonal cell line of Rous sarcoma virus (RSV)-transformed 3Y1 cells derived from Fisher rat embryos (Kimura et al., 1975). 77N1 cells are transformants of NRK cells by avian sarcoma virus (ASV) (Hirai et al., 1983). All of these cell lines were generous gifts from Dr. R. Hirai (Tokyo Metropolitan Institute of Medical Science). Each cell line was seeded on glass coverslips and cultured in 5% CO2 and 100% humidity in Dulbecco’s modified Eagle’s medium supplemented with 10% fetal calf serum (FCS) for 3Y1-BY1 cells and 5% FCS for NRK-77N1 cells at 37°C.
Antibodies
Antibodies against caldesmon and tropomyosin were raised in New Zealand White rabbits. As immunogens, were purified caldesmon and tropomyosin from chicken gizzard smooth muscle (Bretscher, 1984; Tanaka et al., 1990). The antibodies against tropomyosin were purified from antisera using affinity columns coupled with the protein. Due to the ability of the anti-tropomyosin polyclonal antibody to crossreact with all of the tropomyosin isoforms in nonmuscle cells (present result; and Matsumura et al., 1983b), it was used for examining the isoformal diversity of tropomyosin during transformation. The caldesmon antibody was purified from antiserum by an affinity column coupled with the COOH-terminal 35,000 Mr fragment of h-caldesmon (Hayashi et al., 1989a), which is the conserved domain of h- and l-caldesmons and shares in common with the actin-, tropomyosin- and calmodulin-binding activity (Hayashi et al., 1991). Polyclonal antibody against platelet myosin and monoclonal antibody against chicken gizzard tropomyosin (clone no. TM311) were purchased from Sigma. The former antibody immunoreacted with nonmuscle myosin heavy chain and the latter with the high Mr types of tropomyosin.
Immunofluorescence microscopy
The intracellular distribution of contractile proteins, except for actin, was examined by indirect immunofluorescence microscopy. Filamentous actin (F-actin) was stained with rhodamine-labelled phalloidin (Sigma). The cells grown on coverslips were fixed using 3.5% formaldehyde in phosphate buffered saline (PBS) for 15 minutes at room temperature and then permeabilized with 0.2% Triton X-100 in PBS. After blocking with 1% BSA in PBS, the cells were incubated with the first antibodies diluted in PBS containing 1% BSA and 0.02% Triton X-100 for 30 minutes at 37°C, and then stained with fluorescein isothiocyanate (FITC)-labelled anti-rabbit or mouse IgG (Miles-Yeda). For double labelling of both F-actin and another protein, the second antibody solution simultaneously included rhodamine-labelled phalloidin. After extensive washing with PBS, the coverslips were mounted and viewed with a Zeiss fluorescence microscope equipped with interference reflection microscopy (IRM) optics. IRM was used for measuring the distance between the ventral plasma membrane and the substratum (Izzard and Lochner, 1976).
Lysis and squirting of cells
The original procedure established by Nermut (1982) was modified as follows. All subsequent steps were carried out at room temperature. Cells grown on glass coverslips were rinsed with PIPES (piperazine-N,N -bis(2-ethanesulfonic acid)) buffer (20 mM PIPES, 100 mM KCl, 5 mM MgCl2, 3 mM EGTA, pH 6.0) for 10 seconds, followed by incubation in hypotonic buffer (5 mM PIPES, 3 mM MgCl2, 3 mM EGTA, pH 6.0) for 4 minutes, exchanging the buffer 4 times. After incubation, the coverslips were squirted with the PIPES buffer using a syringe. Next, the ruptured cells were briefly rinsed with PIPES buffer (pH 7.0). For some experiments, the ruptured cells were incubated in PIPES buffer (pH 7.0) containing ATP, Ca2+, high salt or calmodulin. After the above treatments, the cells were fixed with formaldehyde in PBS and immunostained with the specific antibodies or labelled with rhodamine-phalloidin and observed by microscopy.
Gel electrophoresis and immunoblotting
Cells grown on dishes were harvested, homogenized in Laemmli’s sample solution containing 2% SDS (Laemmli, 1970), and then boiled for 5 minutes. Equal amounts of proteins were electrophoresed and immunoblotted (Towbin et al., 1979). Protein concentrations were determined with the BCA protein assay system (Pierce). Anti-rabbit or anti-mouse IgG coupled with alkaline phosphatase (Promega) was used as the second antibody.
Two-dimensional gel electrophoresis
The change in expression of isoforms of the contractile proteins accompaning transformation was identified by two-dimensional (2D) gel electrophoresis (O’Farrell, 1975). The cell lysates were made in 9 M urea containing 2-mercaptoethanol (5%), Triton X-100 (4%), NaF (20 μg/ml), (para-amidinophenyl)methanesulfonyl fluoride hydrochloride (p-APMSF; 10 mM) and leupeptin (10 μg/ml). The protein concentrations in the lysates were adjusted to 1.0 μg/ml. For caldesmon and tropomyosin, the heat-resistant fraction of cell lysates were further fractionated by 65% saturated ammonium sulfate. The caldesmon- and tropomyosin-enriched fraction thus obtained was used for actin-binding assays (Sobue et al., 1983) The proteins cosedimented with F-actin were analyzed by 2D gel electrophoresis. The 2D gels were stained with Coomassie brilliant blue, or were used for immunoblotting.
Other materials
Calmodulin was prepared from bovine brain (Kakiuchi et al., 1981). Actin was purified from rabbit skeletal muscle (Pardee and Spudich, 1982). ATP was purchased from Oriental Yeast (Japan). Morphological photographs were recorded using Kodak Tri-X or T-MAX 100 film and Kodak D-76 as developer.
RESULTS
Expressional change of the contractile proteins accompanying transformation
The specificity of antibodies used in this study was monitered by immunoblotting (Fig. 1). Anti-myosin antibody recognized one band with a Mr of 200,000, corresponding to myosin heavy chain, in the cell lysates. The specific antibody against the COOH-terminal 35,000 Mr fragment of h-caldesmon immunoreacted with a 80,000 Mr protein and with a trace amount of 140,000 Mr protein; the former is l-caldesmon and the latter is h-caldesmon. Monoclonal antibody against tropomyosin was able to detect three bands with Mr of 40,000, 37,000 and 36,000 corresponding to the high Mr types of tropomyosin (Matsumura, 1983b). Fig. 2 shows the expressional changes of actin, caldesmon and tropomyosin isoforms accompanying transformation by RSV. The 2D gels stained with Coomassie brilliant blue did not show significant difference between actin isoforms in 3Y1 and BY1 cells (Fig. 2A,B). Eight spots of caldesmon were visualized by immunoblotting in both normal and transformed cells. Among them, the two spots on the basic side of the BY1 cell lysates seem to be larger than those of 3Y1 cell lysates (Fig. 2C,D). This difference was not detected in a pair of NRK and 77N1 cells (data not shown). In agreement with previous reports (Matsumura et al., 1983b; Lin et al., 1984), the low Mr types of tropomyosin detected by polyclonal antibody were increased in the transformed cells (Fig. 2E,F). Additionally, the monoclonal antibody used in the present morphological study could only recognize the three higher Mr types of tropomyosin among the five isoforms as shown in Fig. 1.
Immunoblotting showing the crossreactivity of antibodies against myosin, caldesmon (CaD) and tropomyosin (TM) used in the present study. Lanes 3 and B indicate the cell lysates of 3Y1 and BY1, respectively. 10 μg of protein was applied to each lane. Antitropomyosin antibody reacts with some high Mr types of tropomyosin isoforms, which are TM1, TM2 and TM3 by the convention of Matsumura et al. (1983a). The percentages for acrylamide for SDS-PAGE were 6% for myosin, 9% for caldesmon and 12.5 % for tropomyosin.
Immunoblotting showing the crossreactivity of antibodies against myosin, caldesmon (CaD) and tropomyosin (TM) used in the present study. Lanes 3 and B indicate the cell lysates of 3Y1 and BY1, respectively. 10 μg of protein was applied to each lane. Antitropomyosin antibody reacts with some high Mr types of tropomyosin isoforms, which are TM1, TM2 and TM3 by the convention of Matsumura et al. (1983a). The percentages for acrylamide for SDS-PAGE were 6% for myosin, 9% for caldesmon and 12.5 % for tropomyosin.
Results of two-dimensional gel analysis for actin (A,B), caldesmon (C,D) and tropomyosin (E,F) are shown (left, acidic; right, basic). The proteins of 3Y1 and BY1 cells are shown in A,C,E and B,D,F, respectively. Isoelectric focussing of caldesmon was performed on the nonequilibrated form. For actin, cell lysates were made in 9 M urea containing 2-mercaptoethanol, Triton X-100, protease inhibitors and NaF. For caldesmon and tropomyosin, the heat-stable fraction was precipitated using 65% saturated ammonium sulfate following co-precipitation with F-actin. The precipitated actin-binding proteins were resolved in 9 M urea solution. The percentages of acrylamide for the SDS-polyacrylamide gel of the second dimension were 11% for actin, 9% for caldesmon and 12.5% for tropomyosin. Coomassie blue staining was used for actin and tropomyosin. C and D show the results of immunoblotting using antibody against the 35,000 Mr fragment of h-caldesmon for detection of caldesmon isoforms. Arrowheads in A and B indicate β - and γ-actin. Arrows in E and F indicate the isoforms of tropomyosin recognized by polyclonal anti-tropomyosin antibody. Arrowheads in E and F indicate actin of rabbit skeletal muscle.
Results of two-dimensional gel analysis for actin (A,B), caldesmon (C,D) and tropomyosin (E,F) are shown (left, acidic; right, basic). The proteins of 3Y1 and BY1 cells are shown in A,C,E and B,D,F, respectively. Isoelectric focussing of caldesmon was performed on the nonequilibrated form. For actin, cell lysates were made in 9 M urea containing 2-mercaptoethanol, Triton X-100, protease inhibitors and NaF. For caldesmon and tropomyosin, the heat-stable fraction was precipitated using 65% saturated ammonium sulfate following co-precipitation with F-actin. The precipitated actin-binding proteins were resolved in 9 M urea solution. The percentages of acrylamide for the SDS-polyacrylamide gel of the second dimension were 11% for actin, 9% for caldesmon and 12.5% for tropomyosin. Coomassie blue staining was used for actin and tropomyosin. C and D show the results of immunoblotting using antibody against the 35,000 Mr fragment of h-caldesmon for detection of caldesmon isoforms. Arrowheads in A and B indicate β - and γ-actin. Arrows in E and F indicate the isoforms of tropomyosin recognized by polyclonal anti-tropomyosin antibody. Arrowheads in E and F indicate actin of rabbit skeletal muscle.
Differential distribution of the contractile proteins in normal and transformed cells
The intracellular distribution of the four contractile proteins in normal cells and in their transformed counterparts was compared using fluorescence microscopy. The cells examined in this study had the well-known cell-substratum adhesive structures, focal adhesions (3Y1 cells) and podosomes (BY1 cells). Both of the adhesive structures contained vinculin, talin and a-actinin as previously reported (data not shown; Marchisio et al., 1987). In 3Y1 cells, myosin, caldesmon and tropomyosin immunofluorescence was periodically distributed along stress fibers, which were stained by rhodamine-phalloidin (Fig. 3A,B, D,E and G,H). Focal adhesions in the 3Y1 cells were easily identified by IRM (Fig. 3C,F and I). Stress fibers terminated within the focal adhesions, in which myosin, caldesmon and tropomyosin were not detected (indicated by arrowheads in Fig. 3). The RSV-transformed counterparts (BY1 cells) show a roundish shape with peculiar microspikes or processes (Fig. 4). IRM revealed that BY1 cells have two distinctive types of cell adhesive structures; one type is a podosome, and the other is a closely associated contact with the substratum at the cell periphery (Nakamura et al., unpublished data). As shown in Fig. 4, fluorescence for myosin, caldesmon and tropomyosin was observed to accumulate in podosomes, but not in the contact regions at the cell periphery. Careful focussing indicated that the intense fluorescence of the short F-actin bundles in the core domain of podosomes was surrounded by the faint fluorescence of F-actin (Fig. 4B,E and H). Myosin and tropomyosin fluorescence was localized in such surrounding regions, but not found in the core domain. Consequently, myosin and tropomyosin fluorescence was seen as a ring-like distribution (Fig. 4J,L). The location of caldesmon was slightly different from that of myosin and tropomyosin. The caldesmon fluorescence was mainly distributed within the core domain and, at least in part, also in the surrounding region (Fig. 4D-E and K). Changing the level of focus, the distinct regional distribution of the three protein was maintained along the vertical axis of podosomes except in the tip region, where the three protein were absent. The differential distribution of the four contractile proteins within the podosomes demonstrated here was also confirmed by confocal laser scan microscopy (data not shown).
Localization of myosin (A), caldesmon (D) and tropomyosin (G) in 3Y1 cells was examined using double labelling of cells with antibodies to the proteins and rhodamine-labelled phalloidin for F-actin (B,E and H). In order to recognize the cell-substratum adhesive area, IRM was used (C,F and I). Arrowheads indicate the two corresponding focal adhesions among FITC fluorescence, rhodamine fluorescence and IRM pictures. Bar, 20 μm.
Localization of myosin (A), caldesmon (D) and tropomyosin (G) in 3Y1 cells was examined using double labelling of cells with antibodies to the proteins and rhodamine-labelled phalloidin for F-actin (B,E and H). In order to recognize the cell-substratum adhesive area, IRM was used (C,F and I). Arrowheads indicate the two corresponding focal adhesions among FITC fluorescence, rhodamine fluorescence and IRM pictures. Bar, 20 μm.
Localization of myosin (A,J), caldesmon (D,K), tropomyosin (G,L) and F-actin (B,E and H) and in BY1 cells. C,F, and I are the IRM pictures corresponding to the micrographs on the left. Arrowheads indicate the two corresponding podosomes. Some podosomes cannot be recognized in the IRM pictures, implying that they do not make contact with the substrata. Myosin, caldesmon and tropomyosin are sparsely distributed in the cytoplasm; however, they are specifically accumulated within podosomes. Note the ring-like immunofluorescence of myosin (A,J) and tropomyosin (G,L) around the core domain of podosomes. By contrast, caldesmon tends to be located in the core with the short F-actin bundles (D,E,K). Arrows indicate the region that closely attaches to the substratum at the cell periphery, where the contractile proteins do not accumulate. Bar, 20 mm. A-I and J,K,L are at the same magnification.
Localization of myosin (A,J), caldesmon (D,K), tropomyosin (G,L) and F-actin (B,E and H) and in BY1 cells. C,F, and I are the IRM pictures corresponding to the micrographs on the left. Arrowheads indicate the two corresponding podosomes. Some podosomes cannot be recognized in the IRM pictures, implying that they do not make contact with the substrata. Myosin, caldesmon and tropomyosin are sparsely distributed in the cytoplasm; however, they are specifically accumulated within podosomes. Note the ring-like immunofluorescence of myosin (A,J) and tropomyosin (G,L) around the core domain of podosomes. By contrast, caldesmon tends to be located in the core with the short F-actin bundles (D,E,K). Arrows indicate the region that closely attaches to the substratum at the cell periphery, where the contractile proteins do not accumulate. Bar, 20 mm. A-I and J,K,L are at the same magnification.
Stability and molecular organization of contractile proteins in normal and transformed cells
By introducing the lysis/squirting technique, we compared the stability and the molecular organization of contractile proteins in stress fibers of normal cells and in podosomes of transformed cells. In addition, by removing the cytoplasm by the lysis/squirting technique, the fluorescence of the proteins on the stress fibers or in the adhesive structures could be more clearly observed. In 3Y1 cells, the fluorescence of myosin, caldesmon and tropomyosin along stress fibers remained intact even after lysis/squirting (Fig. 5A-D). Notably, myosin and tropomyosin fluorescence within the podosomes of BY1 cells was completely absent after the same treatment. Some caldesmon and most of the sparsely distributed actin filaments in the region surrounding the core domain were also extracted. By contrast, the short F-actin bundles and their associated caldesmon were unaffected (Fig. 5E-J).
Using the lysis/squirting technique, the ruptured cells were prepared as shown here. (A,B,C, and D) and (E,F,G,H,I, and J) show ruptured 3Y1 and BY1 cells, respectively. (A and B) show myosin immunoreactivity and rhodamine-phalloidin fluorescence in the same 3Y1 cell. (C and D) show caldesmon and tropomyosin immunofluorescence, respectively. Any protein in the 3Y1 cells maintains its original distribution along stress fibers even after the treatment by lysis/squirting. In BY1 cells, myosin (E) and tropomyosin (I) immunofluorescence around podosomes was lost after the same treatment, whereas the corresponding short F-actin bundles in podosomes (F, J) are maintained. (G and H) show caldesmon and F-actin fluorescence, respectively, in podosomes. Only caldesmon can attach to the short F-actin bundles in podosomes even after lysis/squirting treatment.
Using the lysis/squirting technique, the ruptured cells were prepared as shown here. (A,B,C, and D) and (E,F,G,H,I, and J) show ruptured 3Y1 and BY1 cells, respectively. (A and B) show myosin immunoreactivity and rhodamine-phalloidin fluorescence in the same 3Y1 cell. (C and D) show caldesmon and tropomyosin immunofluorescence, respectively. Any protein in the 3Y1 cells maintains its original distribution along stress fibers even after the treatment by lysis/squirting. In BY1 cells, myosin (E) and tropomyosin (I) immunofluorescence around podosomes was lost after the same treatment, whereas the corresponding short F-actin bundles in podosomes (F, J) are maintained. (G and H) show caldesmon and F-actin fluorescence, respectively, in podosomes. Only caldesmon can attach to the short F-actin bundles in podosomes even after lysis/squirting treatment.
Using the ruptured cells prepared by the lysis/squirting technique, we further investigated the stability of contractile proteins. When ruptured 3Y1 cells were incubated in a high salt buffer (20 mM PIPES, 300 mM KCl, 5 mM MgCl2, 3 mM EGTA, pH 7.0) for 5 minutes at room temperature, the overall distribution of the four contractile proteins along stress fibers did not change (Fig. 6A-D). When ruptured BY1 cells were incubated in the same high salt buffer, almost all the caldesmon immunoreactivity in the podosomes was abolished in association with a reduction in F-actin fluorescnce (Fig. 6E,F). When ruptured 3Y1 cells were incubated in a buffer containing 100 mM KCl and 2 mM ATP, no alteration was observed in the localization of the four contractile proteins (the periodic distribution of myosin along stress fibers in ruptured 3Y1 cells is shown in Fig. 6G,H). When 2 mM ATP was included in the high salt buffer, myosin, caldesmon and tropomyosin fluorescence on stress fibers was completely absent and the stress fibers became thinner, wavy and loosely bundled (Fig. 6I-L). a-Actinin in 3Y1 cells, which has the bundling activity of F-actin and is periodically distributed along stress fibers (Bretscher and Lynch, 1985; Sobue et al.,1989), was simultaneously extracted with myosin, caldesmon and tropomyosin in a buffer containing both ATP and high salt (data not shown).
These fluorescence micrographs show ruptured cells treated with a buffer containing 300 mM KCl (A,B,C,D,E and F), a buffer containing 100 mM KCl and 2 mM ATP (G,H), or a buffer containing 300 mM KCl and 2 mM ATP (I,J,K and L). Micrographs E,F show BY1 cells and others show 3Y1 cells. The cells in A,B,C,D,E and F were incubated with a buffer containing 20 mM PIPES, 300 mM KCl, 5 mM MgCl2 and 3 mM EGTA, pH 7.0. (A and B) show the same cells, indicating myosin and F-actin fluorescence, respectively. (C and D) show caldesmon and tropomyosin immunoreactivity. (E and F) show caldesmon and F-actin fluorescence of the same BY1 cells. Although all contractile elements were maintained in the 3Y1 cells after incubation with high salt buffer, almost all of the caldesmon and a major part of the F-actin were lost in the BY1 cells. (G and H) show the distribution of myosin and F-actin in 3Y1 cells incubated with buffer containing 20 mM PIPES, 100 mM KCl, 5 mM MgCl2, 3 mM EGTA and 2 mM ATP, pH 7.0. No alteration in fluorescence of the proteins, including caldesmon and tropomyosin, is observed. (I,J,K and L) show 3Y1 cells treated with buffer containing 20 mM PIPES, 300 mM KCl, 5 mM MgCl2, 3 mM EGTA and 2 mM ATP, pH 7.0. (I, K and L) indicate the loss of myosin, caldesmon and tropomyosin from stress fibers, respectively. J shows the distribution of F-actin in the same cell as shown in I. Arrowheads in K and L show unruptured cells that are diffusely immunostained.
These fluorescence micrographs show ruptured cells treated with a buffer containing 300 mM KCl (A,B,C,D,E and F), a buffer containing 100 mM KCl and 2 mM ATP (G,H), or a buffer containing 300 mM KCl and 2 mM ATP (I,J,K and L). Micrographs E,F show BY1 cells and others show 3Y1 cells. The cells in A,B,C,D,E and F were incubated with a buffer containing 20 mM PIPES, 300 mM KCl, 5 mM MgCl2 and 3 mM EGTA, pH 7.0. (A and B) show the same cells, indicating myosin and F-actin fluorescence, respectively. (C and D) show caldesmon and tropomyosin immunoreactivity. (E and F) show caldesmon and F-actin fluorescence of the same BY1 cells. Although all contractile elements were maintained in the 3Y1 cells after incubation with high salt buffer, almost all of the caldesmon and a major part of the F-actin were lost in the BY1 cells. (G and H) show the distribution of myosin and F-actin in 3Y1 cells incubated with buffer containing 20 mM PIPES, 100 mM KCl, 5 mM MgCl2, 3 mM EGTA and 2 mM ATP, pH 7.0. No alteration in fluorescence of the proteins, including caldesmon and tropomyosin, is observed. (I,J,K and L) show 3Y1 cells treated with buffer containing 20 mM PIPES, 300 mM KCl, 5 mM MgCl2, 3 mM EGTA and 2 mM ATP, pH 7.0. (I, K and L) indicate the loss of myosin, caldesmon and tropomyosin from stress fibers, respectively. J shows the distribution of F-actin in the same cell as shown in I. Arrowheads in K and L show unruptured cells that are diffusely immunostained.
We have previously reported that Ca2+/calmodulin regulates the binding of caldesmon to F-actin. If the concentration of Ca2+/calmodulin was high enough, caldesmon could be released from F-actin (Sobue et al., 1981). Fig. 7 shows the effect of Ca2+/calmodulin on the association of caldesmon with stress fibers or short F-actin bundles in ruptured 3Y1 and BY1 cells. In 3Y1 cells, Ca2+/calmodulin had no particular effect on caldesmon immunoreactivity along stress fibers (Fig. 7A,B). In BY1 cells, caldesmon immunoreactivity within podosomes disappeared after the addition of Ca2+/calmodulin to the incubation buffer, even when the short F-actin bundles were intact (Fig. 7C,D). As a control, caldesmon immunofluorescence in the core domain of motile contacts was unchanged in the absence of calmodulin, with or without Ca2+ (Fig. 7E,F).
Effect of Ca2+/calmodulin on caldesmon and F-actin in the ruptured cells. (A,B) and (C,D,E and F) show ruptured 3Y1 and BY1 cells, respectively. The cells were incubated with the buffers containing 20 mM PIPES, 100 mM KCl, 5 mM MgCl2 and 0.2 mM CaCl2 with or without 10 μg/ml of calmodulin, pH 7.0. No particular change was observed on caldesmon (A) and F-actin (B) fluorescence in 3Y1 cell. The BY1 cells lost their caldesmon immunoreactivity (C) after the treatment, although the F-actin bundles (D) were intact in the podosomes. Treatment with Ca2+ buffer without calmodulin did not affect the localization of caldesmon (E) and F-actin (F) in the BY1 cells.
Effect of Ca2+/calmodulin on caldesmon and F-actin in the ruptured cells. (A,B) and (C,D,E and F) show ruptured 3Y1 and BY1 cells, respectively. The cells were incubated with the buffers containing 20 mM PIPES, 100 mM KCl, 5 mM MgCl2 and 0.2 mM CaCl2 with or without 10 μg/ml of calmodulin, pH 7.0. No particular change was observed on caldesmon (A) and F-actin (B) fluorescence in 3Y1 cell. The BY1 cells lost their caldesmon immunoreactivity (C) after the treatment, although the F-actin bundles (D) were intact in the podosomes. Treatment with Ca2+ buffer without calmodulin did not affect the localization of caldesmon (E) and F-actin (F) in the BY1 cells.
DISCUSSION
By comparing podosomes of transformed cells with focal adhesions of normal cells, the former were shown to be the more dynamic adhesive structures with high motility (short half-life), leading to metastasis and invasion (Chen, 1989). The molecular organization of membrane skeletal proteins in podosomes has also been investigated. As a result, some membrane skeletal proteins such as talin (Burridge and Connel, 1983), vinculin, a-actinin (David-Pfeuty and Singer, 1980) and nonerythroid spectrin (Sobue et al., 1988; Sobue et al., 1989) have been identified as major constituents of this structure. The presence of short F-actin bundles within podosomes has long been noted (Carley et al., 1981). Regulatory proteins for actin filaments, fimbrin (Carley et al., 1985) or gelsolin (Wang et al., 1984) have been found within these adhesive structures. No one has clearly demonstrated the localization of other contractile proteins in podosomes. On the contrary, some authors have denied the localization of tropomyosin in podosomes (Leonardi et al., 1982; Carley et al., 1985). Using antibody against the whole h-caldesmon molecule, we previously reported the diffuse distribution of caldesmon in the cytoplasm of cells transformed by temperature-sensitive RSV (Owada et al., 1984). In that report, we were unable to detect the accumulation of caldesmon within podosomes, probably due to a lesser specificity of the previous antibody. Similarly, using polyclonal antibodies against chicken gizzard tropomyosin did not clarify the localization of tropomyosin in podosomes (data not shown). Using the specific antibody for the COOH-terminal 35,000 Mr fragment of h-caldesmon and monoclonal antibody against tropomyosin, it became possible to detect caldesmon and tropomyosin within podosomes. However, myosin, caldesmon and tropomyosin were not found in focal adhesions, implying a lack of motility of such cell adhesive structures. Consequently, we first identified the major components of the actomyosin system within podosomes as molecular bases for their dynamic motility. The actomyosin system should participate in the vertical movement of podosomes. Indeed, vertical movement has been observed by time-lapse study using cinemicrography (Chen, 1989). Furthermore, it has been reported that an increase in the cytosolic Ca2+ concentration of osteoclasts, which was induced by membrane depolarization using high K+ or BAY K 8644, or a high concentration of extracellular Ca2+, causes the disappearance of podosomes (Miyauchi et al., 1990). Although those authors have suspected the involvement of gelsolin in the turnover of podosomes, the increase in the cytosolic Ca2+ concentration also seems to be related to the Ca2+-dependent activation of the contractile system within podosomes.
Changes in expression of tropomyosin (Matsumura et al., 1983b) and actin (Okamoto-Inoue et al., 1990) during transformation have been reported. In this study, we compared the isoforms of the contractile proteins from normal and transformed cells. An apparent differential expression of tropomyosin was found (Fig. 2; Matsumura et al., 1983b). On the other hand, there was no particular difference between the isoforms of actin and caldesmon in normal and transformed cells. We also observed a similar translocation of the contractile proteins into podosomes of Balb/c 3T3 cells transformed by phorbol ester (data not shown). The rearrangement of contractile proteins by phorbol ester was observed within 30 minutes. The time course of this change is considered to be too short for de novo synthesis of the proteins during transformation. Furthermore, it has been reported that enucleated cells can acquire the transformed morphology (Beug et al., 1978). These observations suggest that a change in expression of contractile proteins is not essential for the rearrangement of the actomyosin system.
Taken together with the observations made by fluorescence microscopy, confocal laser scanning microscopy and the lysis/squirting study, we propose that there is stereotypic distribution of the contractile proteins in and around focal adhesions and podosomes (Fig. 8). A focal adhesion, or the terminal region of stress fibers of normal cells, does not contain contractile proteins such as myosin, caldesmon and tropomyosin, although they are periodically distributed along stress fibers. In RSV-transformed cells, the short F-actin bundles are oriented from the lower portion of the ventral cell body to the tip of a podosome. Around the core domain, thin microfilaments radiate from near the tip to the upper or the outer region of a podosome. Because myosin and tropomyosin have a limited location along such thin microfilaments, the immunofluorescence of the two proteins seems to have a ring-like appearance (Fig. 4). Caldesmon, however, is located mainly in the core domain with the short F-actin bundles. The lysis/squirting study supports such stereotypic distribution. By treating cells with the lysis/squirting technique, most of the thin microfilaments associated with myosin, caldesmon and tropomyosin were easily extracted from podosomes. Alternatively, the core domain composed of short F-actin bundles and caldesmon was maintained, probably due to the close binding of the bundles to the tip region of podosomes.
Localization of the four contractile proteins around a focal adhesion (A) and a podosome (B). Thick lines indicate the F-actin bundles that terminate at the focal adhesion and at the tip of the podosome. (A) Myosin, caldesmon and tropomyosin are localized along stress fibers and do not accumulate at the terminating area of stress fibers. (B) Myosin and tropomyosin are localized in the area around the core domain of the podosome with sparsely distributed thin microfilaments (shown as thin lines in B); however, caldesmon is mainly localized in the core domain with the short F-actin bundles.
Localization of the four contractile proteins around a focal adhesion (A) and a podosome (B). Thick lines indicate the F-actin bundles that terminate at the focal adhesion and at the tip of the podosome. (A) Myosin, caldesmon and tropomyosin are localized along stress fibers and do not accumulate at the terminating area of stress fibers. (B) Myosin and tropomyosin are localized in the area around the core domain of the podosome with sparsely distributed thin microfilaments (shown as thin lines in B); however, caldesmon is mainly localized in the core domain with the short F-actin bundles.
The lysis/squirting study has also suggested that stress fibers are more stable than short F-actin bundles within podosomes. Myosin, caldesmon and tropomyosin are dismembered with difficulty from stress fibers (Fig. 5) and only when both ATP and high salt were included in the processing buffer, were the three contractile proteins extracted simultaneously (Fig. 6). Myosin and tropomyosin within podosomes could be easily extracted after the cells were lysed and squirted. The remaining caldesmon was separated from the short F-actin bundles when the cells were incubated with a buffer containing Ca2+/calmodulin, although the same treatment was unable to bring about any change in the location of the contractile proteins in stress fibers (Fig. 7). Judging from these observations, the stability of stress fibers probably depends on the interaction between the four contractile proteins including caldesmonmyosin (Lash et al., 1986; Hemric and Chalovich, 1990) and caldesmon-tropomyosin (Fujii et al., 1988; Horiuchi and Chacko, 1988; Hayashi et al., 1989a). a-Actinin in stress fibers was simultaneously solubilized with myosin, caldesmon and tropomyosin in high salt + ATP buffer, suggesting that it might be involved in the stabilization of stress fibers. Alternatively, filamin (Mittal et al., 1987) and/or some unknown factors may also be stabilizers for such structures. In podosomes, some stabilizing factors might be missing or inactivated. Further analysis is required to solve this problem.
As shown in Fig. 3, caldesmon is distributed periodically along stress fibers (Fig. 3), and this periodicity resembles that of myosin and tropomyosin. Such localization of the contractile proteins is in agreement with the concept that caldesmon is a regulatory protein for the actomyosin system. In transformed cells, a major part of caldesmon is attached to short F-actin bundles within podosomes, whereas, myosin and the high Mr types of tropomyosin are not localized in the core domain. Such incompatibility with the distribution of the contractile proteins has been reported in the ruffling membrane of the cell periphery in which F-actin and caldesmon are colocalized, but not myosin II and tropomyosin (Bretsher and Lynch, 1985). The presence of low Mr types of tropomyosin in ruffling membrane has been described (Lin et al., 1988). The ruffling membrane and podosomes are composed of similar cytoskeletal proteins such as nonerythroid spectrin, a-actinin, talin, fimbrin, caldesmon and actin. Myosin I has been found in the leading edges of Dictyostelium amoebae (Fukui et al., 1989) as well as in the intestinal brush border (Bretscher, 1991), both of which are cell surface structures that contain F-actin bundles, as do podosomes. Taking into account these findings, the presence of myosin I and the low Mr types of tropomyosin in the core domain of podosomes would be suspected. If such speculation is correct, caldesmon and the low Mr types of tropomyosin might be involved in the regulation of the myosin I -F-actin interaction in the core domain of podosomes. Further study is required to clarify the regional distribution of myosin I and the low Mr types of tropomyosin.
ACKNOWLEDGEMENTS
We thank Miss Miho Oda for her skilled technique of cell culture. Leica K.K. kindly lent us a Leiz confocal laser scanning microscope. This study was partly supported by grants from the Scientific Research Fund of the Ministry of Education, Science and Culture of Japan, and from Nissan Fundation.