Three-dimensional electron microscopical visualization of the cytoskeleton of animal cells: immunoferritin identification of actin- and tubulin-containing structures.

Cytoskeletons prepared by Triton X-100 treatment of tissue culture cells appear in stereo electron microscopy as a highly organized and interconnected three-dimensional matrix of different fibrous elements. Microfilament bundles and also tonofilament-like bundles are readily discerned when present in the cell type. In addition thinner fibers, some of which branch (smallest diameter 30--40 A), as well as fibers of larger diameter, some of which correspond to microtubules, can be seen. Since such cytoskeletons are an open, membrane-free system, individual fibrous organizations can be identified by specific antibodies. An indirect immunoferritin procedure using antibodies to tubulin or actin visualizes microtubules or actin-containing structures. Stereo electron microscopy of cytoskeletons decorated with actin antibody reveals, in addition to the F-actin-containing microfilament bundles, an extended fine actin lattice. This actin net is displayed throughout the cytoplasm not only between the microfilament bundles but also in those regions of the cytoskeleton that in the intact cell correspond to the submembraneous regions. Thus all actin-containing fibrous cytoplasmic structures may be interconnected in the living cell.     

Microtubules and microfilaments during cell spreading and colony formation in PK 15 epithelial cells.

We have studied the distribution of microtubules and microfilaments during the cell spreading and subsequent colony formation in PK 15 pig kidney epithelial cells using indirect immunofluorescence. During the cell spreading on a solid substratum, microtubules grew out from the region around the nucleus, and a collar of microfilament bundles formed around the cell periphery. Although virtually all well-spread cells showed a complex microtubular network, distinctly different patterns of stress fibers were observed. In small colonies, the most commonly observed pattern was a ring of microfilament bundles that appeared to be in register between adjacent cells and encircled the entire colony in a fashion similar to that seen in single cells. In large colonies (more than 50 cells), approximately 60% of the cells displayed clearly stained microfilament bundles, either at the cell periphery or throughout their cytoplasm, whereas in the remaining 40%, no microfilament bundles were observed and only the outline of the cells was delineated by interaction with anti-actin. Such "negative" cells were seen in groups alongside "positive" cells (i.e., cells possessing extensive stress fiber networks) within the same colony. Independent of their stress fiber phenotype, all cells maintained a flattened shape and an extensive network of microtubules. We suggest that dense microfilament bundles are not a uniform feature of well-spread PI 15 cells in culture and that a loss of microfilament bundle occurs in some cells.     

Vinculin, an intracellular protein localized at specialized sites where microfilament bundles terminate at cell membranes.

As intracellular protein of 130,000 molecular weight was recently isolated in this laboratory from chicken gizzard smooth muscle. By immunofluorescence observations of cultured chicken fibroblasts, it was shown to be concentrated on the ventral surfaces of the cells where they formed focal adhesions to the substratum [Geiger, B. (1979) Cell 18, 193-205]. Focal adhesions are sites where, inside the fibroblast, microfilament bundles are known to terminate at the cell membrane. The suggestion was made that this new protein (herein named "vinculin") might be involved in the linkage of the termini of microfilament bundles to membranes in various cell types. To explore this possibility, in the present study we examined several chicken tissues, including intestinal epithelium, gizzard smooth muscle, and cardiac striated muscle, by immunoelectron microscopic labeling for vinculin on ultrathin frozen sections of the specimens. In each case, the immunolabeling for vinculin was concentrated close to membrane sites where microfilament bundles terminate: at the zonula adherens in the junctional complex of the brush border of epithelial cells; at the membrane-associated sense plaques of smooth cells; and at the fascia adherens of the intercalated disk membranes of cardiac muscle cells. These results suggest therefore that vinculin may participate in the anchoring of microfilament bundles to specific membrane sites in various cells.     

Phosphotyrosine-containing proteins are concentrated in focal adhesions and intercellular junctions in normal cells.

We have used a high-affinity polyclonal antibody directed against phosphotyrosine (P-Tyr) to localize P-Tyr-containing proteins in normal and transformed cells in culture by immunofluorescence microscopy experiments. The distribution of the proteins with modified tyrosine was compared with that of F-actin in these cells. Cells infected with Abelson murine leukemia virus were found to contain elevated levels of P-Tyr, as expected. Various permanent lines of fibroblastic and epithelial cells exhibited lower, but easily detectable, levels of P-Tyr. The P-Tyr in fibroblasts was concentrated at the focal contacts at the termini of actin-containing microfilament bundles and, in the epithelial cells examined, at the intercellular junctions. Early passages of primary cultures of chicken embryo fibroblasts and chicken embryo heart cells also showed detectable levels of P-Tyr in focal contacts and cell-cell junctions. However, P-Tyr was not detectable in later passages of chicken embryo fibroblasts. The concentration of P-Tyr-containing proteins in intercellular junctions in normal cells suggests that these are sites of significant biochemical regulatory activities which may be important in the control of normal cell adhesivity, motility, and shape.     

Changes in microfilament organization and surface topogrophy upon transformation of chick embryo fibroblasts with Rous sarcoma virus.

A series of morphological changes occurred when chick embryo fibroblasts infected with the NY68 mutant of Rous sarcoma virus were shifted from nonpermissive temperature (41degrees) to permissive temperature (37 degrees). We observed three distinct stages in cell morphology and surface topography that were correlated with a reduction in the organization and assembly of actin-containing microfilament bundles. Our observations suggest that control of microfilament organization and surface topography are responsive to the presence of a functioning transforming gene (src) product of Rous sarcoma virus.     

Assay for early cytoplasmic effects of the src gene product of Rous sarcoma virus

When microinjected into normal fibroblasts, cytoplasmic extracts of cells transformed by Rous sarcoma virus caused dissolution of microfilament bundles. This activity was not found in extracts of normal cells. The maximum effect was seen within 30 min of injection, and the activity could still be measured after a 10-fold dilution of the cytoplasmic extracts (14 mg/ml original protein concentration). The activity was trypsin sensitive and was destroyed by boiling, but was not RNase sensitive. Protein synthesis was not required for the disruption of actin-containing stress fibers by the injected activity. Microinjected cytoplasts prepared from normal 3T3 cells also showed dissolution of microfilament bundles, indicating that the cell nucleus was not required for expression of activity. Extracts made from fibroblasts transformed by Rous sarcoma virus having a temperature-sensitive mutation in the src gene were also temperature sensitive in the microinjection assay. Thus, the activity of extracts from cells infected with src mutant virus, but not from cells infected with wild-type virus, was destroyed either by in vitro incubation of the extract at the nonpermissive temperature before injection or by incubation of recipient cells at the nonpermissive temperature after injection.We conclude that the microinjection assay can detect a cytoplasmic activity coded for by the src gene of Rous sarcoma virus and that an early direct or indirect target of the src gene product is the cytoskeleton and cell motility system. This result is discussed in relation to the hypothesis that submembranous arrays of microfilaments, microtubules, and their associated proteins interact with cell surface receptors to form a surface modulating assembly that functions as a key regulator of cell growth.     

Ultrastructure of chicken cardiac muscle as studied by double immunolabeling in electron microscopy.

The ultrastructural localization of alpha-actinin and vinculin in chicken cardiac muscle was studied by double indirect immunoelectron microscopy, using ferritin and iron-dextran (Imposil) as the electron-dense markers conjugated to the secondary antibodies, on ultrathin frozen sections of fixed tissue. Fixation and immunolabeling procedures were developed that permitted maximal retention of the two proteins at their natural sites as well as their adequate labeling. alpha-Actinin was found both on the Z-bands, as expected, and near the fascia adherens of the intercalated discs, whereas vinculin was confined to the latter sites. At the fascia adherens, the double labeling results clearly showed that vinculin was situated closer to the membrane than was alpha-actinin. These results, coupled with earlier observations, suggest that vinculin may participate in the linkage of actin-containing microfilament bundles to membranes in a variety of cell types.     

Characterization of antibodies to smooth muscle myosin kinase and their use in localizing myosin kinase in nonmuscle cells.

Antibodies to myosin light chain kinase, purified from turkey gizzard smooth muscle, were developed in rabbits and purified by affinity chromatography on a myosin light chain kinase-Sepharose 4B column. The purified antibodies crossreact with purified smooth muscle myosin light chain kinase but not with a variety of contractile or cytoskeletal proteins. The antibodies inhibit the catalytic activity of smooth muscle myosin light chain kinase and there is an inverse relationship between the kinase activity and the amount of antibody present in an assay. Half-maximal inhibition of myosin kinase activity occurs at an antibody/myosin kinase molar ratio of 10:1. The affinity-purified antibodies to smooth muscle myosin kinase were used to study the location of myosin kinase in a variety of nonmuscle cells. Immunofluorescence studies indicate that myosin light chain kinase is localized on microfilament bundles (stress fibers) in cultured fibroblasts. The stress fiber staining pattern is abolished when the antibodies are incubated with purified smooth muscle myosin light chain kinase prior to staining cells, while the staining pattern is unaffected when the antibodies are incubated with actin, myosin, alpha-actinin, or tropomyosin prior to staining. Moreover, the stress fiber staining pattern is periodic in well-spread gerbil fibroma cells and experiments have demonstrated that myosin light chain kinase appears to have the same periodic distribution as myosin but an antiperiodic distribution relative to alpha-actinin. These data indicate that myosin light chain kinase and its substrate, myosin, are in close proximity and are consistent with the hypothesis that myosin light chain kinase regulates actin-myosin interactions in nonmuscle cells.     

Visualization of a system of filaments 7-10 nm thick in cultured cells of an epithelioid line (Pt K2) by immunofluorescence microscopy.

During our studies with antibodies against structural proteins of the cytoskeleton of eukaryotic cells we have observed that sera from many normal rabbits decorate a fiber system in cells of the established rat kangaroo cell line Pt K2. The display and organization of these fibers are different from those of microfilament bundles (decorated by antibody to actin) and microtubules (decorated by antibody to tubulin). This new fiber system can be further distinguished by its resistance to reorganization when cells are treated with Colcemid or cytochalasin B. The decoration of this fiber system is not detected if Pt K2 cells are fixed with formaldehyde. Such sera also appear to decorate swirls of perinuclear fibers in mouse Neuro 2a cells, and in mouse 3T3 cells treated with mitotic drugs. Comparison of the immunofluorescence pictures with electron microscopic data suggests that the sera are visualizing bundles of intermediate 7- to 10-nm filaments.     

Intracellular distributions of mechanochemical proteins in cultured fibroblasts

We have used methods that have allowed simultaneous fluorescent staining of intracellular actin together with either myosin, filamin, or tubulin in normal rat kidney fibroblasts in monolayer culture. In the main portions of the cell body, the actin, myosin, and filamin are all present in two structures: in one, the three proteins are present in the same fiber bundles (stress fibers); in the other, there is a diffuse distribution of the three proteins. On portions of the cell periphery however-in the basal regions of microspikes, in ruffles, and in regions of cell-cell contact-actin and filamin are present, but myosin is severely depleted or absent. Microtubules are present in the cell body in a distribution independent of the stress fibers and are mostly absent from the cell periphery. Microspikes and ruffles are highly dynamic structures on the cell surface, and regions of cell-cell contact generally result from the association of ruffles on the two contacting cells. Therefore, the presence of filamin and actin but not myosin in these specialized regions on the cell surface, together with the recent demonstration [Wang, K. & Singer, S. J. (1977) Proc. Natl. Acad. Sci. USA 74, 2021-2025)] that pure filamin interacts with individual F-actin filaments in solution to form fiber bundles and sheet-like structures, suggest that in vivo filamin-actin interactions play an important role in the control of actin filament structure, in cell motility, and in the stabilization of cell-cell contacts.     

Actin Antibody: The Specific Visualization of Actin Filaments in Non-Muscle Cells

Actin purified from mouse fibroblasts by sodium dodecyl sulfate gel electrophoresis was used as antigen to obtain an antibody in rabbits. The elicited antibody was shown to be specific for actin as judged by immunodiffusion and complement fixation against partially purified mouse fibroblast actin and highly purified chicken muscle actin. The antibody was used in indirect immunofluorescence to demonstrate by fluorescence light microscopy the distribution and pattern of actin-containing filaments in a variety of cell types. Actin filaments were shown to span the cell length or to concentrate in "focal points" in patterns characteristic for each individual cell.     

Visualization of actin fibers associated with the cell membrane in amoebae of Dictyostelium discoideum.

Amoebae of Dictyostelium discoideum were attached to a surface coated with polylysine, and the upper portion of the cells was sheared off with a stream of buffer. Scanning and transmission electron microscopy showed that the cytoplasmic surface of the exposed membrane was covered with fibers consisting of actin-containing filaments. The actin was identified by its solubility properties and its ability to interact with muscle myosin.     

Immunofluorescent detection of actin cytoskeleton in spontaneously transformed and B77-supertransformed cells using antibody to tropomyosin

Differences in immunofluorescence of actin cytoskeleton between normal rat fibroblasts and two transformed cell lines SAMIV and SAMB77 were detected by antiserum to tropomyosin. In both transformed cell lines reduction in number and shortening of microfilament bundles (stress fibers) were observed. In some transformed cells ruffling membranes (peripheral ruffles) were seen. Differences were found in actin cytoskeleton among individual cells of spontaneous transformants (SAMIV) and supertransformants (SAMB77).     

Microinjection of fluorescently labeled α-actinin into living fibroblasts

alpha-Actinin from chicken gizzard labeled with tetramethylrhodamine isothiocyanate has been incorporated into living fibroblast cells by microinjection. Fluorescent labeling of alpha-actinin was carried out such that the conjugated protein was functional in vitro as shown by its ability to bind to F-actin. Within 1-2 hr after injection, diffuse fluorescence was observed throughout the cytoplasm and only faint fluorescence was apparently associated with the stress fibers. During the ensuing 2-15 hr, however, most of the fluorescence was seen as periodicities along the stress fibers and as foci of the microfilament polygonal networks. This distribution of alpha-actinin in the living cells was strikingly similar to that found by indirect immunofluorescence localization of endogenous alpha-actinin in fixed samples of the same cell type. Control studies in which heat-treated (100 degrees C, 2 min) fluorescent alpha-actinin or tetramethylrhodamine isothiocyanate alone was injected into the cells indicated that the stress fiber and polygonal network labeling was specific for "native" fluorescently labeled alpha-actinin. These results suggest that the dynamic properties of proteins and structures in cultured mammalian cells can be studied with the use of microinjection and fluorescence microscopic techniques.     

Monoclonal antibodies against myofibrillar components of rat skeletal muscle decorate the intermediate filaments of cultured cells.

Monospecific antibodies were produced in vitro by fusing mouse myeloma cells with spleen cells from a BALB/c mouse immunized with rat skeletal myofibrils. After cloning 3 times on agarose, two stable clones were obtained and chosen for further characterization. The first clone, JLB1, produced an antibody that recognizing an antigen distributed in the M-line region and on either site of the Z line of myofibrils. The second clone, JLB7, produced an antibody reacting only with an antigen located at the M-line region of myofibrils. Both JLB1 and JLB7 antibodies decorate the typical intermediate filaments of a variety of cultured cells. Colcemid treatment of cells before reaction with both antibodies resulted in the coiling or capping (or both) of the fibers around the nucleus. Brief treatment of cells with cytochalasin B did not affect the integrity of the fibers stained by both antibodies whereas, under the same conditions, microfilament bundles visualized by another monoclonal antibody (JLA20) against actin were disassembled into many aggregates in the cytoplasm. Identical staining patterns of the intermediate filaments are obtained by double-label immunofluorescence microscopy of the same cell stained with these monoclonal antibodies and rabbit autoimmune serum (which has been shown to react with the components of the intermediate filaments). By using immunoprecipitation, protein bands at 210,000 and 95,000 daltons from chicken embryo fibroblasts were identified as the potential antigens recognized by JLB1 and JLB7 monoclonal antibodies, respectively. The widespread occurrence of these antigenic determinants in different cultured cells suggests the highly conservative property of these intermediate-filament components.     

Actin-severing activity copurifies with phosphofructokinase.

Microinjection of muscle 6-phosphofructokinase (PFK; EC 2.7.1.11) into tissue culture cells led to a reversible disintegration of microfilament bundles (stress fibers). The mode of disruption as well as of recovery of stress fibers was very similar to that found previously in experiments performed with the actin-severing protein brevin, an extracellular variant of gelsolin. PFK, like brevin, was also capable of disrupting stress fibers in detergent-extracted cells and in ethanol-fixed cells, in a Ca2+-dependent manner. When compared with heart muscle gelsolin, PFK comigrated with the 85- to 90-kDa band. Antibodies against PFK crossreacted with gelsolin from the same species. These results point to a tight association between polypeptides with similar biochemical and immunological parameters present in both preparations. They suggest hitherto unexpected cellular control mechanisms for both microfilament functions and glycolysis.     

Caldesmon and a 20-kDa actin-binding fragment of caldesmon inhibit tension development in skinned gizzard muscle fiber bundles.

Caldesmon is known to inhibit actin-activated myosin ATPase activity in solution, to inhibit force production when added to skeletal muscle fibers, and to alter actin movement in the in vitro cell motility assay. It is less clear that caldesmon can inhibit contraction in smooth muscle cells in which caldesmon is abundant. We now show that caldesmon and its 20-kDa actin-binding fragment are able to inhibit force in chemically skinned gizzard fiber bundles, which are activated by a constitutively active myosin light-chain kinase in the presence and absence of okadaic acid. This inhibitory effect is reversed by high concentrations of Ca2+ and calmodulin. Therefore, caldesmon may act by increasing the level of myosin phosphorylation required to obtain full activation. Our results also suggest that caldesmon does not act to maintain force in smooth muscle by cross-linking myosin with actin since competition of binding of caldesmon with myosin does not cause a reduction in tension.     

Morphological and biochemical responses of cultured thyroid cells to thyrotropin

In the thyroid gland, TSH stimulates cAMP formation, exocytosis of precursor (noniodinated) thyroglobulin, endocytosis of thyroglobulin, and proteolytic processing of the thyroglobulin to form thyroid hormones. In this report we describe TSH effects on cAMP levels, microtubules, microfilaments, myosin fibrils, and the morphology of cultured thyroid follicle cells. The cells were normally cultured in the presence of 10 mU/ml TSH, and fresh TSH produced no stimulation when assayed for cAMP production in a 15-min assay. When such cells were cultured for up to 72 h in the absence of TSH and then assayed for cAMP production, the basal levels were much reduced, but fresh TSH stimulated cAMP levels half-maximally at 1 mU/ml and up to 50-fold at 20 mU/ml. Microtubules, myosin fibers, and microfilaments were demonstrated by indirect immunofluorescent staining. Fluorescent staining of fibers was observed in cells fixed before lysis and in cells lysed before fixation. In control cells grown without hormone, microtubules originated near the nucleus and extended to the cell periphery. Myosin-containing fibrils traversed the cell or radiated from foci. Microfilaments spanned the cell in a stress fiber pattern. After incubation with 20 mU/ml TSH and 4 mM isobutylmethylxanthine (IBMX) for 10-20 min, the microtubules in up to half of the cells appeared altered and more granular, and the cell periphery was scalloped. After 15-30 min with TSH and IBMX, normal myosin fibers were replaced with a fine lattice-work, peripheral staining disappeared, and the proportion of nonfibrous myosin increased. Stress fibers demonstrated with antibody to actin also disappeared, and the peripheral structures observed in normal cells became fragmented. Incubation with forskolin or TSH and IBMX for 2-3 h resulted in arborization of 30-60% of the cells that contained bundles of microtubules, myosin fibers, or microfilaments into dendrite-like processes and increased staining near the nucleus. At 5 h, more than 80% of the cells were arborized. These morphological changes were less pronounced with IBMX alone and minimal with TSH alone. The time course of cAMP levels observed basally or after TSH, forskolin, or TSH and IBMX was consistent with the relative effects of these agents on arborization. These studies are consistent with effects of cAMP on microtubules, myosin-containing fibrils, and microfilaments and may provide a basis for the morphological response to TSH.     

Microfilament dysfunction as a possible cause of intrahepatic cholestasis.

The effects of cytochalasin B on bile canalicular structure and function were examined. Three experimental models were used, cultured hepatocytes, isolated perfused liver, and in vivo infused liver. The techniques used were light and electron microscopy and, in selected instances, scanning electron microscopy, electron "stains" for microfilaments, and measurements of bile flow. Microfilament disruption and dilation of bile canaliculi were consistently found and closely paralleled a reduction in bile flow in both in vitro and in vivo infused animals. It is proposed that under normal circumstances, the microfilaments maintain the canaliculi in a contracted or partly contracted state. Hence, the microfilamentous network would provide tone to the canalicular system which would tend to reduce stagnation and facilitate the flow of bile. Removal of normal microfilament contractile function would be expected to produce canalicular ectasia and reduction of bile flow, as was observed. Microfilament dysfunction may therefore be a possible cause of intrahepatic cholestasis. Crucial to this hypothesis are the presence of actin-containing microfilaments in the pericanalicular web, and an action of cytochalasin B on their contractility. Evidence pertaining to these requirements is presented and discused.