Isolation and Characterization of Tropomyosin-containing Microfilaments from Cultured Cells*

We have developed a new method for the rapid isolation of tropomyosin-containing microfilaments from cultured cells using anti-tropomyosin monoclonal antibodies. Anti-tropomyosin monoclonal antibodies in- duce the bundle formation of microfilaments, which can be easily collected by low speed centrifugation. Electron microscopic studies of the isolated microfilaments show periodic localization of tropomyosin along the microfilaments of nonmuscle cells with a 33-34 nm repeat. Furthermore, the isolated microfilaments have the ability to activate the Mg2+-ATPase activity of skeletal muscle myosin to almost the same extent as skeletal muscle F-actin (filamentous actin). This microfilament isolation method is applicable to a variety of cell types, including REF-52 cells (an established rat embryo line), L6 myoblasts, 3T3 fibroblasts, Chinese hamster ovary cells, baby hamster kidney (BHK-21) cells, mouse neuroblastoma cells, gerbil fi- broma cells, and chicken embryo fibroblasts. Sodium dodecyl sulfate-polyacrylamide gel analysis shows that, in addition to actin, microfilaments isolated from REF-52 cells contain five species of tropomyosin with apparent M, = 40,000, 36,500, 35,000, 32,400, and 32,000, a-actinin, and as yet unknown proteins with apparent M, = 83,000 and 37,000. The molar ratio of total tropomyosin (dimer) to actin in the isolated mi- crofilaments


Isolation and Characterization of Tropomyosin-containing Microfilaments from Cultured Cells*
(Received for publication, September 1,1982) Fumio MatsumuraS, Shigeko Yamashiro-Matsumura, and Jim Jung-Ching Ling From the Cold Spring Harbor Laboratory, Cold Spring Harbor, New York 11 724 We have developed a new method for the rapid isolation of tropomyosin-containing microfilaments from cultured cells using anti-tropomyosin monoclonal antibodies. Anti-tropomyosin monoclonal antibodies induce the bundle formation of microfilaments, which can be easily collected by low speed centrifugation. Electron microscopic studies of the isolated microfilaments show periodic localization of tropomyosin along the microfilaments of nonmuscle cells with a 33-34 nm repeat. Furthermore, the isolated microfilaments have the ability to activate the Mg2+-ATPase activity of skeletal muscle myosin to almost the same extent as skeletal muscle F-actin ( It is generally believed that actin-containing microfilaments play central roles in cell motility and cell shape changes of cultured cells (1)(2)(3)(4)(5)(6)(7). Over the past decade, studies with immunofluorescent microscopy on cultured cells have led to the conclusion that microfilaments are composed of several proteins in addition to actin, such as myosin (8-lo), tropomyosin (11,12), a-actinin (12)(13)(14), and filamin (15,16). These proteins have also been identified and characterized biochemically in nonmuscle cells (2)(3)(4)(5)(17)(18)(19)(20)(21)(22). However, it still remains to be shown whether changes in the protein composition and/or structural organization of microfilaments occur concomitantly with changes in biological activities of cultured * This work was supported in part by Cancer Center grant to Cold Spring Harbor Laboratory from the National Cancer Institute (PO1 CA13106-11). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore he hereby marked "aduertisernent" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

filamentous actin). This microfilament isolation method is applicable to a variety of cell types, including REF-52 cells (an established rat embryo line), L6 myoblasts, 3T3 fibroblasts, Chinese hamster ovary cells, baby hamster kidney (BHK-21) cells, mouse neuroblastoma cells, gerbil fibroma cells, and chicken embryo fibroblasts. Sodium dodecyl sulfate-polyacrylamide gel analysis shows that, in addition to actin, microfilaments isolated from
i. Supported  cells such as spreading, mitosis, movement, differentiation, oncogenic transformation, etc. To begin to answer this question, it is necessary to develop a suitable method for isolation of microfilaments in their intact form from cultured cells under various physiological conditions. Recently, Schloss and Goldman (23) have succeeded in preparing a population of microfilaments from cultured rat embryo cells by utilizing the isolation method for muscle native thin filaments (24,25). This procedure isolated microfilaments from BHK1-21 or 3T3 cells that were first induced to adopt a well spread morphology by subculturing in serum-depleted medium or by selecting only those cells tightly adherent to culture dishes. This method, however, was limited and could not be easily used to isolate microfilaments from a variety of different cell types at various biological stages, such as during mitosis and differentiation.
We have developed a new method for the rapid isolation of microfilaments from a variety of cell types including REF-52, gerbil fibroma (CCL-146), BHK-21, NIH/3T3, Chinese hamster ovary, mouse neuroblastoma, L6 myoblasts and myotubes, and chick embryo fibroblasts. This method is based on the previous observations that anti-tropomyosin monoclonal antibodies cause the aggregation of thin filaments from smooth muscle into ordered bundles which are easily collected by low speed centrifugation (26). Since tropomyosin is one of the major components of microfilaments (11,12), the antitropomyosin monoclonal antibody is suitable to use for the immunoprecipitation of microfilaments. In this report, we describe the method for the isolation of microfilaments from cultured cells and their morphological and biochemical characterization. We report that the isolated microfilaments show a double helical structure similar to the morphology of Factin, and the tropomyosin is found to be arranged along microfilaments with a periodicity of 33-34 nm. SDS-polyacrylamide gel analysis of the microfilaments from REF-52 cells shows that in addition to actin, several other proteins including multiple forms of tropomyosin are present. by Dr. W. C. Topp, Cold Spring Harbor Laboratory. Cell lines were maintained in DMEM containing 10% fetal calf serum in an atmosphere of 5% COz and 95% air at 37 "C unless specified otherwise. For BHK-21 cells, 10% calf serum supplemented with 10% tryptose phosphate broth replaced fetal calf serum. All cells were passaged using 0.05% trypsin in PBS.
Isolation of Microfilaments from Cultured Cell.-Monolayer cells were washed 3 times with PBS at room temperature and then extracted for 2 min at room temperature with Triton/glycerol solution (0.1 M PIPES, pH 6.9, 5 mM MgC12, 0.2 mM EGTA, 0.05% Triton X-100, 4 M glycerol) to stabilize the cytoskeleton. While the extracted cells were still attached to culture dishes, they were rinsed 3 times with PBS containing 5 mM MgC12 and 0.2 mM EGTA. After removal of the washing solution, the Triton/glycerol residues of cells (10-20 plates of 100-mm culture dish) were collected as follows. To the first dish 0.2 ml of buffer A was added and then the dish was tilted to facilitate scraping the residue into one corner with a rubber policeman. The residues plus buffer A from this dish were transferred to the second dish and the scraping procedure was repeated. The accumulated residues plus the initial 0.2 ml of buffer A were then transferred to the subsequent dishes until all dishes were scraped of their residues. This allowed the collection of the residues from 10-20 dishes to be restricted to a final volume of 1.0-1.5 ml. After addition of ATP and phenylmethylsulfonyl fluoride to a final concentration of 5 mM, respectively, the cell residues were homogenized at 0 "C by 70 strokes in a motor-driven Potter-Elvehjem (glass/Teflon) homogenizer (400 rpm). After centrifugation of the homogenate at 12,800 X g for 15 min at 4 "C in an Eppendorf centrifuge, f i o volume of ascites fluid of monoclonal antibody (LCK16, JLF15, or JLH2) was added to the supernatant, and the mixture was incubated for 30 min at room temperature to allow antibody to aggregate microfilaments into bundles. The resultant microfilament bundles were collected by centrifugation at 12,800 X g for 5 min at 4 "C in an Eppendorf centrifuge and washed once with buffer A containing 5 mM ATP, followed by washing twice with buffer A. The final pellet was resuspended in 50-100 pl of buffer A and used as a microfilament fraction. Protein concentration was determined by the method of Lowry et al. (27).
There appears to be no way to estimate the yield of microfilaments. As an alternative way we determined the amount of actin (measured by radioactivity) in various fractions at each step of the purification and expressed the yield as a percentage of the original total actin amount. Cells (one 100-mm culture dish, 106-107 cells) were in uivo labeled for 15 h with 250 pCi of [35S]methionine (1110 Ci/mmol) in methionine-free DMEM containing 2.5% fetal calf serum. Samples from each step of the isolation were analyzed by SDS-polyacrylamide gel electrophoresis as shown in Fig. 1. After staining with 0.15% Coomassie brilliant blue in 50% methanol, 10% acetic acid for 2 h followed by destaining with 7.5% methanol, 7.5% acetic acid, the gels were soaked for 30 min with distilled water and then dried on filter paper. The actin bands were cut out and sliced into about 1 mm thickness. Each slice was incubated for 24 h in 0.2 ml of 2% SDS to elute protein and the radioactivity was measured by a Beckman liquid scintillation counter.
Molar ratios of actin to multiple forms of tropomyosin in the isolated microfilaments were determined from densitometer traces of one-dimensional SDS gels as described (47,56). Gels were stained quantitatively with fast green (56) and scanned with a Hoefer densitometer (GS 300). The molar ratios were calculated using values corrected for differential dye uptake between skeletal muscle actin and smooth muscle tropomyosin (47).
Preparation of Muscle Proteins-Myosin was prepared from rabbit skeletal muscle by the method described by Perry (28). Skeletal muscle actin was extracted for 10 min at 0 "C from acetone powder of rabbit skeletal muscle prepared by the method of Ebashi and Ehashi (29) and polymerized overnight a t 4 "C by addition of 3 M KC1 to a final concentration of 30 mM. F-actin was collected by centrifugation (100,000 X g, 2 h) and treated with 0.6 M KC1 according to the method of Spudich and Watt (30).
Smooth muscle tropomyosin was prepared from frozen chicken gizzards by the method of Ebashi et al. (31) except that tropomyosin was fractionated by the addition of solid ammonium sulfate to a concentration of between 30 and 36 g/100 ml rather than between 25 and 45 g. The purities of actin, myosin, or tropomyosin (more than 95% pure) were determined by SDS-polyacrylamide slab gel electro-phoresis with a low concentration of bisacrylamide (12.5% acrylamide, 0.104% bisacrylamide) (32, 33).
Preparation of Monoclonal and Conventional Antibodies against Tropomyosin-Hybridoma clones JLF15, JLH2, and LCK16 were isolated and characterized as described (34,35). The high titer antibodies obtained from the ascites fluids of hybridoma-bearing mice were used for all experiments. Hybridoma clones JLF15 and LCK16 secrete IgM antibodies while clone JLH2 secretes IgA antibodies. The antibody JLF15 recognizes only tropomyosin, whereas LCK16 and JLH2 are found to react with both tropomyosin and vimentin (34,35).
For preparation of conventional antibody against tropomyosin, purified smooth muscle tropomyosin (about 0.5 mg in PBS) emulsified in complete Freund's adjuvant was injected into rabbits. After 1 month, the rabbits were boosted with similar amounts of antigen in incomplete Freund's adjuvant. The booster injection was repeated once after 2 weeks. Antiserum was collected 1 week after the last injection and tested by Ouchterlony double diffusion tests against several purified proteins, including a-actinin, vinculin, filamin, tropomyosin, and troponin. Only tropomyosin formed a precipitating line with this antibody. The specificity of the antibody was further tested by Western blots (36) and by immunofluorescence staining. Immunoautoradiograms of Western blots showed that the antibody reacted only with tropomyosin when total cell lysates from chicken embryo fibroblasts were used. Immunofluorescence staining of rabbit myofibrils, chick embryo fibroblasts, and gerbil fibroma cells gave the typical staining pattern of tropomyosin, i.e. I-band stain of myofibrils and a periodic stain of microfilament bundles of cultured cells (11).
Immunoprecipitation-Immunoprecipitation with the conventional antibody against tropomyosin was carried out using formalinfixed Staphylococcus aureus following Kessler's procedure (37) at 0-4 "C. The cell extract was prepared from REF-52 cells labeled in vivo with 250 pCi of [35S]methionine (1110 Ci/mmol) for 15 h in methionine-free media. After washing 3 times with PBS, the cells (60-mm dish) were lysed in 100 pl of SDS gel sample buffer (2% SDS, 15% glycerol, 100 mM DTT, 80 mM Tris-C1, 0.001% bromphenol blue, pH 6.8). The cell lysates, after homogenizing by passage through a No. 26 needle at least 6 times, were boiled at 100 "C for 3-5 min and diluted (1:30) with lysing buffer (50 mM Tris, pH 8.0,0.5% Triton X-100, I mM EDTA, 1 mM phenylmethyl sulfonyl fluoride, 100 mM NaCl). The lysates were incubated for 1 h on ice with 5 p1 of rabbit antiserum against smooth muscle tropomyosin. After a subsequent incubation with 100 pl of S. aureus for 30 min, the antigen-antibody-S. aureus complex was pelleted and washed 3 times with lysing buffer containing 0.05% SDS and once with PBS to remove the detergent. The complexes were solubilized in 50 p1 of SDS gel sample buffer and boiled at 100 "C for 2 min. After centrifugation to remove S. aureus, the supernatant was analyzed by 12.5% SDS-polyacrylamide gel electrophoresis (32, 33) or two-dimensional gel electrophoresis (38,39). Radioactive proteins in the gel were detected by fluorography (40) on Kodak XR-1 film.
Electron Microscopy-One drop (about 30 p1) of sample solution was applied onto a carbon-coated Formvar grid and negatively stained with 2.5% aqueous uranyl acetate. Samples were observed by a Phillips 201E electron microscope at an accelerating voltage of 80 kV.
For measurement of the periodicity of tropomyosin localization along microfilaments, collagen fibers from a rat tail were mixed with the antibody-induced bundles. Micrographs of both bundles and collagen fibers in the same field were taken and the periodicity of tropomyosin localization was measured by using the periodicity (64 nm) of collagen fibers as a standard.
Other Procedures-ATPase assay was performed a t 26 "C in 20 mM imidazole buffer (pH 7.0) containing 50 mM KCl, 5 mM MgC12, 50 p~ CaC12 with 75 pg/ml of myosin and various concentrations of F-actin (0-120 pg/ml) or microfilament fraction. ATP concentration was kept constant (0.1 mM) by using 0.2 mg/ml of pyruvate kinase (Sigma) and 2 mM phosphoenolpyruvate (Sigma) as the ATP-regenerating system. The activities were calculated from the initial rate of pyruvate liberation as determined by the method of Reynard et al. (41). Indirect immunofluorescence was performed as described by Blose (42) and Feramisco and Blose (43). in lane 4 (14 times more concentrate) was loaded on the gel. After electrophoresis, the gel was stained with Coomassie blue and destained with 7.5% acetic acid, 7.5% methanol.

Isolation of
disperse microfilaments in the supernatant, and immunoprecipitation of microfilaments by anti-tropomyosin monoclonal antibodies in a native condition.
Monolayer cells of REF-52 were subjected to this method to isolate microfilaments. The protein composition of the various fractions arising during the isolation procedure was analyzed by SDS-polyacrylamide gel electrophoresis (Fig. 1).
The Triton/glycerol soluble fraction (lane I in Fig. 1) stains very lightly with Coomassie blue due to the volume of the fraction (about 40 times more dilute compared to the other fractions). In an experiment with ["S]methionine-labeled cells, we can detect many soluble proteins including actin in this fraction, although the substantial majority of microfilament-associated proteins remains insoluble. Because there is no way to measure the amount of microfilaments, we alternatively determined the amount of actin in various fractions to estimate the yield (see "Materials and Methods"). As shown in Table I, 12-1696 of total actin was extracted by Triton/glycerol treatment. This may represent the nonfilamentous actin fraction in the cells.
T h e Triton/glycerol-insoluble residues (lane 2 in Fig. 1) were homogenized in 5 mM M$'-ATP to disperse the microfilaments. This homogenate contained actin, vimentin (the major component of intermediate filaments), and nuclear proteins as well as many other minor proteins. After centrifugation of the homogenate at 12,800 X g for 15 min (Eppendorf), about 50-60% of the total actin remained in the supernatant (lane 4 in Fig. 1). This centrifugation removed the intermediate filaments and nuclei as judged by the disappearance of the vimentin band and many low molecular weight nuclear proteins from the superantant fraction (lane 4 in Fig.  1). Microfilament bundles (Fig. 3E) produced by incubation of the supernatant with anti-tropomyosin monoclonal antibodies were pelleted by low speed centrifugation at 12,800 X g for 5 min and washed by resuspension. As Table I shows, approximately 30% of total actin was finally recovered in the microfilament fraction. SDS-polyacrylamide gel analysis showed that the microfilament fraction (lane 5 in Fig. 1) contained mainly actin, multiple forms of tropomyosin (which   will be identified later), and the heavy and light chains of the anti-tropomyosin IgM antibody. Several other minor proteins such as a-actinin (105,000 Da) and an 83,000-Da protein can be also detected in this fraction. The characterization of these microfilament-associated proteins will be discussed later.
The supernatant (lane 6 in Fig. 1) after the precipitation of microfilaments with anti-tropomyosin monoclonal antibodies contained significant amounts of actin, approximately 20% of the total actin. No significant amount of tropomyosin could be found in this antibody-supernatant fraction, as determined by two-dimensional gel electrophoresis and by further immunoprecipitation using conventional rabbit anti-tropomyosin antiserum. However, half of the actin in the antibodysupernatant could be pelleted by high speed centrifugation at 100,000 x g for 2 h. This suggests that this supernatant may contain microfilaments devoid of tropomyosin and/or membrane-associated actin.
The formation of microfilament bundles appears to be specifically caused by the anti-tropomyosin monoclonal antibodies. In order to show the specificity, the microfilaments in supernatant were incubated either with ascites fluid of other monoclonal antibodies (57), such as FM28 (against skeletal muscle myosin) and JLTl2 (against troponin-T), or with heat-denatured anti-tropomyosin monoclonal antibodies. In all cases, no microfilament bundles were detected by electron microscopy or by SDS-polyacrylamide gel analysis.
As described in a previous paper (26), the anti-actin monoclonal antibody JLA2O (34) morphologically distorted thin filaments or microfilaments, which lost their double helical structure and showed a fuzzy coats of antibody binding. This distortion may also cause a dissociation of minor proteins from the microfilaments. Therefore, even though actin is the major component of microfilaments of all eukaryotic cells, the monoclonal antibody against actin is not suitable for microfilament isolation.
We have shown that both IgM (LCK16 or JLF15) and IgA (JLH2) monoclonal antibodies against tropomyosin are equally efficient to use for the isolation of thin filaments (26) or microfilaments. Furthermore, we are also able to isolate troponin-T-containing microfilaments from chicken embryo myotubes by a using similar method with an IgG monoclonal antibody (JLT12) against troponin-T (data not shown). Thus, the microfilament isolation method described here appears to be independent of classes of antibody. Comparison with Other Methods for Microfilament Isolation-In the method described here, we used the monoclonal antibodies for immunoprecipitation of microfilaments because of their high specificity and high titer. For comparison, we also used rabbit polyclonal antibody instead of the monoclonal antibodies. When the antisera were used for the isolation of microfilaments from REF-4A cells, only tropomyosin but no detectable amount of actin was found in the immunoprecipitates (lane 2 in Fig. 2), indicating that no microfilaments were precipitated. This may be due to the presence of actin-depolymerizing factors or actin filament length regulators (gelsolin, brevin, etc.) in the crude serum (see, for review, Ref. 5 5 ) . When the IgG fraction of polyclonal antibody purified by ammonium sulfate precipitation (45% saturation) was used for the immunoprecipitation, the immunoprecipitate contained actin and multiple forms of tropomyosin. However, the immunoprecipitate with polyclonal antibodies showed no F-actin-like filaments but amorphous clumps when examined by electron microscopy. On the other hand, the immunoprecipitate with monoclonal antibodies showed ordered bundles of microfilaments ( Fig. 3) with typical F-actin-like structure. The binding of polyclonal antibodies to tropomyosin molecules a t too many sites probably caused this distortion of microfilaments and might cause a dissociation of minor proteins from microfilaments.
We also compared the present method with a second microfilament isolation protocol similar to that described by Schloss and Goldman (23). This method is based on the fact that a high concentration of Mg2' (10 mM) forms paracrystals of muscle actin filaments (44), which can be easily collected by low speed centrifugation. As Fig. 2 (lane 3 ) shows, actin, myosin, tropomyosin, and several other proteins were enriched by this procedure relative to the supernatant. However, the protein pattern is much more complex than that of microfilaments isolated by the monoclonal antibodies (lane I). Moreover, the yield of microfilaments is less than half of that    of microfilaments isolated by the monoclonal antibodies. Periodic Localization of Tropomyosin along Microfilaments-In a previous paper (26), we showed that anti-tropomyosin monoclonal antibodies cross-linked between pairs of the same antigenic determinant on tropomyosin molecules within smooth muscle thin filaments, and thus caused the thin filaments to align into ordered bundles. The formation of ordered bundles greatly enhanced visualization of the localization of tropomyosin molecules along the thin filaments. Similarly, the monoclonal antibodies allow us to localize nonmuscle tropomyosin along the microfilaments at the electron microscopic level. As can be seen in Fig. 3, microfilament bundles from BHK-21 (Fig. 3B), L6 myoblasts (Fig. 3C), REF-52 cells (Fig. 30), and gerbil fibroma cells (Fig. 3E) show obvious cross-striations along the whole length of the bundles which are morphologically very similar to those from smooth muscle (Fig. 3A). The individual microfilaments in these bundles preserve the double helical structure that is typically observed in the F-actin filaments of both smooth and skeletal muscle. Since measurements based on electron microscopy can be in error due to variations in magnification, we have calculated the periodicity of the cross-striations in microfilament bundles by using the known periodicity (64 nm) of collagen fibers from rat tail as an internal standard. As Table  I1 shows, the periodicities of nonmuscle tropomyosin localized along microfilaments ranged from 33 to 34 nm, which were considerably shorter than those (37-38 nm) of muscle tropo-  spots (a, b, and c ) were chosen for the alignment of these two gels. Multiple forms of tropomyosin were indicated by the 1-5. Note that all these five polypeptides were not labeled with ['Hlproline, which is one of the characteristics for tropomyosin.
Activation of Skeletal Muscle Myosin ATPase Activity by the Isolated Microfilaments-It is well known that F-actin can activate the skeletal muscle myosin Mg2"ATPase activity (48). Therefore, we asked whether the microfilaments isolated from REF-52 cells also activate the myosin ATPase activity (Table 111). When activities were measured in the absence of DTT, the activation of myosin ATPase by REF-52 microfilaments (30 Fg/ml) was 1.8-fold. The same extent of activation required 2.3 pg/ml of skeletal muscle F-actin; that is, the isolated microfilaments only contained 2.3 pg/ml of actin which were able to activate the myosin ATPase activity. This is equivalent to 15% of the total actin in the isolated microfilaments assuming that 50% of total protein in the isolated microfilaments is actin (based on the electrophoretic pattern of microfilaments shown in lane 5 of Fig. 1). This inefficient activation of the myosin ATPase activity may be due to the fact that most of the actin in the microfilament bundles is sterically prevented from reacting with myosin by the monoclonal antibodies. Therefore, we have introduced 10 m M DTT into the microfilament bundles to dissociate the bound antibody and then measured the activation of ATPase activity. In this case, up to 80% of the actin in the microfilament fraction was as active as F-actin (Table  HI). In another experiment with microfilaments isolated from L6 myoblasts, 70% of the actin was found to be active. Therefore, we have concluded that microfilaments isolated by this method can activate the myosin M$+-ATPase activity to nearly the same extent as skeletal muscle F-actin.
Identification  Fig. 2). In addition, the microfilament fraction contained several proteins with apparent M, = 250,000, 105,000,83,000, 40,000, 37,000, 36,500, 35,000, 32,400, and 32,000 (see Fig. 1). Of these polypeptides, proteins of 40,000, 36,500, 35,000, 32,400, and 32,000 Da were identified as tropomyosin by the criteria of two-dimensional gel analysis, immunoprecipitation with conventional rabbit antiserum in the presence of SDS, lack of proline and tryptophan, heat stability, and actin binding as shown below.
These five proteins were found to have an isoelectric point around pH 4.6 (Fig. 4, A and B ) by two-dimensional gel analysis, which is the known PI for tropomyosin (17,18,21).
Immunoprecipitation of total cell lysates of REF-4A cells with rabbit anti-tropomyosin antiserum in the presence of 0.05% SDS showed five spots by two-dimensional gel analysis (Fig.  4C), whose electrophoretical mobilities were identical to those of the five proteins of microfilaments isolated from both normal (REF-52) and SV40-transformed (REF-4A) cells (Fig.  4). It should be noted that both 32,400-and 32,000-Da proteins were not completely recovered in the immunoprecipitate with rabbit antisera (compare Fig. 4, B and C). This may be due to the lower affinity of the antisera to both proteins.
The absence of two amino acids, proline and tryptophan, is known to be characteristic of tropomyosin (17,50,51). Microfilaments were separately isolated from REF-4A cells labeled in vivo with either [:"SS]methionine, ["Hltryptophan, or ["HI proline, and were analyzed by two-dimensional gel electrophoresis. While fluorograms of ["sS]methionine-labeled microfilaments had five spots of 40,000, 36,500, 35,000, 32,400, and 32,000 Da (Fig. 5A), ['Hlproline-labeled microfilaments did not show any of these five spots (Fig. 5B) mers in skeletal muscle thin filaments (45)(46)(47). The molar ratios of actin to the total amount of tropomyosins (calculated as dimer) in the isolated microfilaments of REF-52 and REFsitometer traces of fast green-stained gels. From the periodicity data obtained by us (Table 11) and by others (17,18,21,23), it is likely that 1 nonmuscle tropomyosin dimer binds 6 actin monomers. These results together suggest that some of the actin in the isolated microfilaments are not fully covered with tropomyosin molecules.
The protein band with an apparent M , = 105,000 that also coisolated in our preparation was identified as a-actinin by two-dimensional gel analysis (co-migration with authentic aactinin isolated from chicken gizzard) and by immunoautoradiography on SDS-polyacrylamide gels with a rabbit antiserum against beef heart a-actinin (a generous gift from Dr. K. Burridge Fig. 1 with lane 1 in Fig. 2). In addition, the 37,000-Da protein appeared to be rather basic on two-dimensional gels.
Application of the Method-The method we describe here is applicable to isolation of microfilaments from a variety of reason for this is unknown. It might suggest that HeLa cells have few tropomyosin-containing microfilaments. phan-labeled microfilaments gave the same result (data not We have applied the method to examine changes in the shown). protein composition upon DNA virus-induced cell transfor-One well known characteristic for tropomyosin is its heat mation. One such example is shown in Fig. 4, A and B. On stability (52). The putative tropomyosins were tested by boil-transformation by SV40, the levels of both TM-1 and TM-2 ing the microfilament of REF-4A cells at 100 "C for 10 min. were decreased while the minor tropomyosins TM-3 and TM-The supernatant after heat treatment contained all five poly-5 were increased. Similar but more drastic changes in tropopeptides (Fig. 6), suggesting that these five proteins were myosins were found in adenovirus type 5-transformed cells. heat-stable. Preliminary binding experiments have shown In the microfilaments of adenovirus type 5-transformed cells that these five polypeptides are able to bind to skeletal F-(Ad5D.lA), TM-1 was missing entirely and TM-3 was found actin (data not shown). Based on these pieces of evidence, we to increase 10 times more than that of the normal cell microhave concluded that tropomyosin in REF-52 cells exists in filaments. Therefore, the increase in the amount of minor multiple forms with apparent molecular masses of 40,000 tropomyosin and the decrease of major tropomyosin appeared (TM-11, 36,500 (TM-2), 35,000 (TM-3), 32,400 (TM-4), and to commonly occur on cells transformed by SV40 or adeno-32,000 (TM-5) Da. It is unlikely that these five polypeptides virus type 5. Detailed analysis of these changes in tropomyosin are derived from proteolysis during isolation of microfila-patterns will be described elsewhere. ments since two-dimensional gel analysis of freshly prepared total cell lysates also showed these five polypeptides. Furthermore, the relative ratio of these five tropomyosin species This method of microfilament isolation provides us with a appeared to be same in both total cell lysates and the micro-sensitive way to detect new microfilament-associated pro- teins. As Fig. 1 and 6 show, microfilaments of REF-52 and REF-4A contained an 83,000-Da protein which has not yet been reported. The presence of an 83,000-Da protein in total cell lysates by two-dimensional gel analysis suggests that this protein is not a proteolytic fragment during isolation of microfilaments. Preliminary studies have shown that this protein is also present in the microfilaments isolated from L6 myoblasts but is missing in the microfilaments of L6 myotubes. This may suggest that the 83,000-Da protein may play an important role in myogenesis. Further characterization of the protein is in progress. Furthermore, we also identified five polypeptides (Mr = 40,000 (TM-l), 36,500 (TM-2), 35,000 (TM-3), 32,400 (TM-4), and 32,000 (TM-5)) as tropomyosin in the microfilaments from REF-52 cells. These multiple forms of tropomyosin appeared to be commonly present in rat cell lines since microfilaments from L6 myoblasts and normal rat kidney cells were found to contain the same five polypeptides judged by the criteria of the co-migration on two-dimensional gels, heat stability, and immunoprecipitation with conventional anti-tropomyosin antiserum. L6 myoblasts were reported to have three species of tropomyosin (53), which appeared to correspond to the tropomyosins TM-1, TM-2, and TM-4 of REF-52 cells. The microfilament isolation method described here allowed us to identify minor proteins with apparent M, = 35,000 (TM-3) and 32,000 (TM-5) as tropomyosins, because these proteins were greatly enriched in the microfilament fraction. It remains to be investigated whether these multiple forms of tropomyosin are functionally different and whether they bind to microfilaments randomly or selectively. Any of the methods for the isolation of microfilaments have the problem whether or not the isolated microfilaments are precisely the same structures that exist in living cells. To assess the nativeness of isolated microfilaments, we addressed two major questions associated with the method described here: 1) Does Triton/glycerol extraction immediately freeze the microfilaments at a state identical to that occurring in in uiuo? 2) Does an exchange reaction occur between microfilament components and free molecules during isolation?
To address the first question, we examined the effect of Triton/glycerol extraction on microfilament isolation. We observed microfilaments in the homogenates of cells treated with or without Triton/glycerol using electron microscopy. These two sets of filaments were morphologically identical. However, while microfilaments in the homogenates of cells without Triton/glycerol extraction disappeared quickly, within hours, microfilaments in the homogenates of extracted cells were stable for at least 1 day. In addition, if Triton/ glycerol extraction was omitted, the yield of microfilaments was reduced to only one-fourth of that obtained by the original method. These results may mean that Triton/glycerol treatment stabilizes the microfilament structure by extracting some factors which cause disassembly and/or degradation of microfilaments. We also examined the effect of the concentration of Triton X-100 on the extractability of tropomyosin from cells. While 20-40% of tropomyosin was extracted with 0.5% Triton X-100 from chick embryo fibroblasts, no significant amounts of tropomyosin were extracted with 10-fold diluted Triton X-100 (0.05%) as judged by two-dimensional gel analysis.
To address the second problem, we examined whether an exchange reaction occurs between the actin or tropomyosin of microfilaments and free actin or tropomyosin externally added during isolation. Cells were first labeled in vivo with [35S]methionine. Purified smooth muscle tropomyosin or skeletal muscle actin was added to the Triton/glycerol residues of the cells to a final concentrations of 30 or 100 pg/ml, respec-tively, and microfilaments were isolated in the same way. Autoradiography of microfilament components analyzed by SDS-polyacrylamide gels showed no significant changes in protein patterns as compared to that of the control (data not shown). This suggests that no exchange reactions occur between free molecules and microfilaments.
Finally, the isolation of microfilaments by this method was affected by the biological states of microfilaments in uiuo. For example, the treatment of cells with cytochalasin B, which is known to disorganize microfilament bundles in cells (7, 54), reduced the yield of microfilaments isolated to one-third of that isolated from control cells, although there were no apparent differences in the protein composition of the microfilaments between the two experiments. The detailed results will be described elsewhere. Another such example is that the tropomyosin patterns of isolated microfilaments were changed in the cells transformed by DNA virus as described under "Results." These results suggest that microfilaments isolated by this method may represent, at least in part, those existing in living cells.
It should be noted that our method only detects the tropomyosin-containing microfilaments from cells. The supernatant after immunoprecipitation of microfilaments with antitropomyosin monoclonal antibodies still contains a significant amount of actin but no obvious tropomyosin. The observation that half of the actin was precipitable by high speed centrifugation may suggest the presence of other classes of actincontaining microfilaments devoid of tropomyosin. This notion is further supported by a preliminary result that a monoclonal antibody JLNZO against a-actinin (34) is able to precipitate a second class of microfilaments from the supernatant remaining after the removal of the tropomyosin-containing microfilaments from chicken embryo fibroblask2 Consequently, we are now using monoclonal antibodies against other contractile proteins such as a-actinin, filamin, and vinculin to isolate other classes of microfilaments by a similar approach. Such an approach will provide us with the detailed information regarding the protein composition and perhaps organization of microfilaments and will help us understand the function of microfilaments in cell motility and cell shape changes.