Filament organization revealed in platinum replicas of freeze-dried cytoskeletons

This report presents the appearance of rapidly frozen, freeze-dried cytoskeletons that have been rotary replicated with platinum and viewed in the transmission electron microscope. The resolution of this method is sufficient to visualize individual filaments in the cytoskeleton and to discriminate among actin, microtubules, and intermediate filaments solely by their surface substructure. This identification has been confirmed by specific decoration with antibodies and selective extraction of individual filament types, and correlated with light microscope immunocytochemistry and gel electrophoresis patterns. The freeze-drying preserves a remarkable degree of three-dimensionality in the organization of these cytoskeletons. They look strikingly similar to the meshwork of strands or "microtrabeculae" seen in the cytoplasm of whole cells by high voltage electron microscopy, in that the filaments form a lattice of the same configutation and with the same proportions of open area as the microtrabeculae seen in whole cells. The major differences between these two views of the structural elements of the cytoplasmic matrix can be attributed to the effects of aldehyde fixation and dehydration. Freeze-dried cytoskeletons thus provide an opportunity to study--at high resolution and in the absence of problems caused by chemical fixation--the detailed organization of filaments in different regions of the cytoplasm and at different stages of cell development. In this report the pattern of actin and intermediate filament organization in various regions of fully spread mouse fibroblasts is described.     

Immunofluorescence evidence for cytoskeletal rearrangement accompanying pigment redistribution in goldfish xanthophores

Immunofluorescence and phase-contrast microscopic studies of goldfish xanthophores with aggregated or dispersed pigment show two unusual features. First, immunofluorescence studies with anti-actin show punctate structures instead of filaments. These punctate structures are unique for the xanthophores and are absent from both goldfish dermal non-pigment cells and a dedifferentiated cell line (GEM-81) derived from a goldfish xanthophore tumor. Comparison of immunofluorescence and phase-contrast microscopic images with electron microscopic images of thin sections and of Triton-insoluble cytoskeletons show that these punctate structures represent pterinosomes with radiating F-actin. The high local concentration of actin around the pterinosomes results in strong localized fluorescence such that, when the images have proper brightness for these structures, individual actin filaments elsewhere in the cell are too weak in their fluorescence to be visible in the micrographs. Second, whereas immunofluorescence images with anti-tubulin show typical patterns in xanthophores with either aggregated or dispersed pigment, namely, filaments radiating out from the microtubule organizing center, immunofluorescence images with anti-actin or with anti-intermediate filament proteins show different patterns in xanthophores with aggregated versus dispersed pigment. In cells with dispersed pigment, the punctate structures seen with anti-actin are relatively evenly distributed in the cytoplasm, and intermediate filaments appear usually as a dense perinuclear band and long filaments elsewhere in the cytoplasm. In cells with aggregated pigment, both intermediate filaments and pterinosomes with associated actin are largely excluded from the space occupied by the pigment aggregate, and the band of intermediate filaments surrounds not only the nucleus but also the pigment aggregate. The patterns of distribution of the different cytoskeleton components, together with previous results from this laboratory, indicate that formation of the pigment aggregate depends at least in part on the interaction between pigment organelles and microtubules. The possibility that intermediate filaments may play a role in the formation/stabilization of the pigment aggregate is discussed.     

Distribution of intermediate filaments in amphibian oxyntic cells

Summary Intermediate filaments of toad oxyntic cells were isolated and analysed by SDS-PAGE. The major proteins of the residue were identified as actin and a 51,000 dalton polypeptide. Immunological crossreactivity between toad oxyntic cell intermediate filament components and anti-prekeratin, was shown by double immunodiffusion tests and indirect immunofluorescence. The immunofluorescent decoration of oxyntic cells and the electron microscope images are coincident in locating the intermediate filaments mainly at the cortical and perinuclear basal zones. Furthermore, the cortical zone appears especially rich in prekeratin-like material at its adluminal third. This results in a cup-like structure that encloses the cell portion occupied by the tubulovesicular system, which does not contain intermediate filaments. The translocation of membranes occurring during the secretory cycle of the oxyntic cell, has been attributed to a system of contractile proteins. The disposition of the prekeratin-like material suggests a role for intermediate filaments in the generation of movement, produced by actin and myosin interaction, by providing a fixed plane for the anchoring of actin microfilaments.     

Aldolase-localization in cultured cells: cell-type and substrate-specific regulation of cytoskeletal associations.

The role of aldolase as a true F- and G-actin binding protein, including modulating actin polymerization, initiating bundling, and giving rise to supramolecular structures that emanate from actin fibrils, has been established using indirect immunofluorescence, permeabilization of XTH-2 cells and keratocytes, and microinjection of fluorescence-labeled aldolase. In addition, binding to intermediate filaments, vimentin, and cytokeratins has been demonstrated. In permeabilized cells in the presence of fructose-1,6-bisphosphate (20-2000 microM) aldolase shifts from association with actin fibres to intermediate filaments. Plenty of free binding sites on microtubules have been revealed by addition of fluorochromed aldolase derived from rabbit skeletal muscle. However, endogenous aldolase was never found associated with microtubules. Differences in actin polymerization in the presence of aldolase as revealed by pyrene-labeled actin fluorimetry and viscosimetry were explained by electron microscopy showing the formation of rod-like structures (10 nm wide, 20-60 nm in length) by association of aldolase with G-actin, which prevents further polymerization. Upon the addition of fructose-1,6-bisphosphate, G-actin-aldolase mixture polymerizes to a higher viscosity and forms stiffer filaments than pure actin of the same concentration.     

Rapid purification of intact tonofilaments from newborn rats : Comparison with glial filaments and neurofilaments

A rapid, new procedure for the isolation of intact tonofilaments from newborn rat skins is described. The filament preparations show two major protein subunits on SDS-PAGE with molecular weights of 58000 and 66000 D. An antiserum prepared against the 58000 D protein reacted specifically with the tonofilament preparation, but not with the protein subunits of neurofilaments, glial filaments, tubulin or actin. This specificity is confirmed by indirect immunofluorescence: anti-P58 reacts with the epidermis, whereas antisera against the neurofilament or glial filament proteins and anti-tubulin do not. These data suggest that epidermal filaments represent a class of intermediate filaments distinct from either glial filaments or neurofilaments.     

Confocal analysis of cytoskeletal organisation within isolated chondrocyte sub-populations cultured in agarose

This study reports the cytoskeletal organisation within chondrocytes, isolated from the superficial and deep zones of articular cartilage and seeded into agarose constructs. At day 0, marked organisation of actin microfilaments was not observed in cells from both zones. Partial or clearly organised microtubules and vimentin intermediate filaments cytoskeletal components were present, however, in a proportion of cells. Staining for microtubules and vimentin intermediate filaments was less marked after 1 day in culture however than on initial seeding. For all three cytoskeletal components there was a dramatic increase in organisation between days 3 and 14 and, in general, organisation was greater within deep zone cells. Clear organisation for actin microfilaments was characterised by a cortical network and punctate staining around the periphery of the cell, while microtubules and vimentin intermediate filaments formed an extensive fibrous network. Cytoskeletal organisation within chondrocytes in agarose appears, therefore, to be broadly similar to that described in situ. Variations in the organisation of actin microfilaments between chondrocytes cultured in agarose and in monolayer are consistent with a role in phenotypic modulation. Vimentin intermediate filaments and microtubules form a link between the plasma membrane and the nucleus and may play a role in the mechanotransduction process.     

The apoptotic process of human bladder carcinoma T24 cells induced by retinoid

Breakdown of the cytoskeletal network and redistribution of cytoplasmic organelles are early events of programmed cell death. Previous studies showed that retinoic acid induces programmed cell death in many tumor cell lines and that cytokeratins, particularly cytokeratin 18, are affected in the early events of apoptosis. In this study, patterns of cytoplasmic intermediate filaments (cytokeratin 18), actin filaments, and microtubules, as well as Bax and Bcl-2 proteins in human bladder carcinoma T24 cells were examined before and after retinoic acid treatment by immunocytochemistry and conventional electron microscopy. Our results demonstrate that the redistribution of Bax and Bcl-2 proteins in the subcellular compartment of T24 cells is correlated with reorganization of the cytoplasmic intermediate filament network and that cytokeratins are cleaved by caspases, as revealed by the M30 antibody which recognizes a specific caspase cleavage site within cytokeratin 18. The cytoskeletal architectures of microtubules are not significantly affected in the early apoptotic process, from our observations. We suggest that the breakdown in the intermediate filament network associated with the aggregation of mitochondria and lysosome may be a crucial event in the apoptotic process and that aggregation of cytoplasmic Bax may accelerate apoptotic death.     

Structural interaction of cytoskeletal components

Three-dimensional cytoskeletal organization of detergent-treated epithelial African green monkey kidney cells (BSC-1) and chick embryo fibroblasts was studied in whole-mount preparations visualized in a high voltage electron microscope. Stereo images are generated at both low and high magnification to reveal both overall cytoskeletal morphology and details of the structural continuity of different filament types. By the use of an improved extraction procedure in combination with heavy meromyosin subfragment 1 decoration of actin filaments, several new features of filament organization are revealed that suggest that the cytoskeleton is a highly interconnected structural unit. In addition to actin filaments, intermediate filaments, and microtubules, a new class of filaments of 2- to 3-nm diameter and 30- to 300-nm length that do not bind heavy merymyosin is demonstrated. They form end-to-side contacts with other cytoskeletal filaments, thereby acting as linkers between various fibers, both like (e.g., actin- actin) and unlike (e.g., actin-intermediate filament, intermediate filament-microtubule). Their nature is unknown. In addition to 2- to 3-nm filaments, actin filaments are demonstrated to form end-to-side contacts with other filaments. Y-shaped actin filament “branches” are observed both in the cell periphery close to ruffles and in more central cell areas also populated by abundant intermediate filaments and microtubules. Arrowhead complexes formed by subfragment 1 decoration of actin filaments point towards the contact site. Actin filaments also form end-to-side contacts with microtubules and intermediate filaments. Careful inspection of numerous actin-microtubule contacts shows that microtubules frequently change their course at sites of contact. A variety of experimentally induced modifications of the frequency of actin-microtubule contacts can be shown to influence the course of microtubules. We conclude that bends in microtubules are imposed by structural interactions with other cytoskeletal elements. A structural and biochemical comparison of whole cells and cytoskeletons demonstrates that the former show a more inticate three-dimensional network and a more complex biochemical composition than the latter. An analysis of the time course of detergent extraction strongly suggests that the cytoskeleton forms a structural backbone with which a large number of proteins of the cytoplasmic ground substance associate in an ordered fashion to form the characteristic image of the “microtrabecular network” (J.J. Wolosewick and K.R. Porter. 1979. J. Cell Biol. 82: 114-139).     

Tubulin, actin, and tau protein interactions and the study of their macromolecular assemblies

The intracellular polymerization of cytoskeletal proteins into their supramolecular assemblies raises many questions regarding the regulatory patterns that control this process. Binding experiments using the ELISA solid phase system, together with protein assembly assays and electron microscopical studies provided clues on the protein-protein associations in the polymerization of tubulin and actin networks. In vitro reconstitution experiments of these cytoskeletal filaments using purified tau, tubulin, and actin proteins were carried out. Tau protein association with tubulin immobilized in a solid phase support system was inhibited by actin monomer, and a higher inhibition was attained in the presence of preassembled actin filaments. Conversely, tubulin and assembled microtubules strongly inhibited tau interaction with actin in the solid phase system. Actin filaments decreased the extent of in vitro tau-induced tubulin assembly. Studies on the morphological aspects of microtubules and actin filaments coexisting in vitro, revealed the association between both cytoskeletal filaments, and in some cases, the presence of fine filamentous structures bridging these polymers. Immunogold studies showed the association of tau along polymerized microtubules and actin filaments, even though a preferential localization of labeled tau with microtubules was revealed. The studies provide further evidence for the involvement of tau protein in modulating the interactions of microtubules and actin polymers in the organization of the cytsokeletal network.     

Models of the Collective Behavior of Proteins in Cells: Tubulin, Actin and Motor Proteins

One of the most important issues of molecular biophysics is the complex and multifunctional behavior of the cell's cytoskeleton. Interiors of living cells are structurally organized by the cytoskeleton networks of filamentous protein polymers: microtubules, actin and intermediate filaments with motor proteins providing force and directionality needed for transport processes. Microtubules (MT's) take active part in material transport within the cell, constitute the most rigid elements of the cell and hence found many uses in cell motility (e.g. flagella andcilia). At present there is, however, no quantitatively predictable explanation of how these important phenomena are orchestrated at a molecular level. Moreover, microtubules have been demonstrated to self-organize leading to pattern formation. We discuss here several models which attempt to shed light on the assembly of microtubules and their interactions with motor proteins. Subsequently, an overview of actin filaments and their properties isgiven with particular emphasis on actin assembly processes. The lengths of actin filaments have been reported that were formed by spontaneous polymerization of highly purified actin monomers after labeling with rhodamine-phalloidin. The length distributions are exponential with a mean of about 7 μm. This length is independent of the initial concentration of actin monomer, an observation inconsistent with a simple nucleation-elongation mechanism. However, with the addition of physically reasonable rates of filament annealing and fragmenting, a nucleation-elongation mechanism can reproduce the observed average length of filaments in two types of experiments: (1) filaments formed from a wide range of highly purified actin monomer concentrations, and (2) filaments formed from 24 mM actin over a range of CapZ concentrations. In the final part of the paper we briefly review the stochastic models used to describe the motion of motor proteins on protein filaments. The vast majority of these models are based on ratchet potentials with the presence of thermal noise and forcing due to ATP binding and a subsequent hydrolysis. Many outstanding questions remain to be quantitatively addressed on a molecular level in order to explain the structure-to-function relationship for the key elements of the cytoskeleton discussed in this review. This revised version was published online in July 2006 with corrections to the Cover Date.     

Brush cells of rodent gallbladder and stomach epithelia express neurofilaments.

It has been suggested that brush cells (BCs), a distinct type of cell occurring in various epithelia of the respiratory and gastrointestinal tracts, may function as receptor cells. The major characteristics of BCs are a prominent brush border and an unusually highly ordered arrangement of cytoskeletal elements (F-actin, microtubules, and intermediate filaments). In this study we aimed to characterize the nature of the intermediate filaments in BCs by light and electron microscopic immunostaining. Gallbladder and stomach specimens from mice and rats, respectively, were fixed in various solutions, embedded either in paraffin or epoxy resin, and processed for immunodetection. Commercially available, well-characterized antibodies against neurofilaments, peripherin, and cytokeratin peptide 18 were used. The polyclonal antiserum cocktail to neurofilaments was applied as a supplement in a double-labeling procedure with anti-actin and anti-cytokeratin 18 antibodies. The results demonstrate that the BCs of both organs express two types of intermediate filaments, i.e., neurofilaments and cytokeratin 18 filaments, and that these have a compartmentalized distribution in the cytoplasm. BCs do not express peripherin. The immunodetection of intermediate filaments distinctive for mature neurons in BCs supports their putative receptor function. The co-expression of neurofilaments and cytokeratins is shown for the first time in healthy tissues.     

Immunocytochemical studies of intermediate filament aggregates and their relationship to microtubules in cultured skin fibroblasts from patients with giant axonal neuropathy

Giant axonal neuropathy (GAN) is a severe autosomal recessive disease affecting both the peripheral and central nervous systems. It is characterized by segmental axonal ballooning due to large neurofilamentous masses and abnormal aggregation of filaments in other cell types including glial cells. Coomassie blue staining of the detergent-resistant cytoskeleton of cultured skin fibroblasts from three patients with GAN revealed the presence of large cytoplasmic filamentous aggregates in the great majority of cells. The aggregates were birefringent when viewed under polarization microscopy and electron microscopy showed that they were composed of aggregates of 8 to 10 nm intermediate filaments. The aggregates stained with antisera specific for vimentin but did not stain with antibodies to actin, tubulin, or the high molecular weight (HMW) microtubule associated protein. Examination of the fibroblasts containing the vimentin aggregates with antibodies to tubulin and the HMW protein showed that they had a normal distribution of microtubules and that the microtubules present were normally associated with the HMW protein. The results suggest that giant axonal neuropathy is a generalized inborn error of organization of intermediate filaments and that a defect in microtubules or their association with HMW protein is not responsible for the observed aggregation of intermediate filaments in this disease. Further study of GAN may be useful in understanding the function of intermediate filaments.     

SB401, a pollen-specific protein from Solanum berthaultii, binds to and bundles microtubules and F-actin

We characterize a novel, pollen-specific, microtubule-associated protein, SB401, found in Solanum berthaultii. This protein binds to and bundles taxol-stabilized microtubules and enhances tubulin polymerization in a concentration-dependent manner, particularly at lower temperatures. Electron microscopy revealed that the protein decorates the entire length of microtubules. Cross-linking and electrophoresis studies showed that SB401 protein forms dimers, and suggest that dimerization could account for bundling. Double immunofluorescent staining of pollen tubes of S. berthaultii showed that SB401 protein co-localized with cortical microtubule bundles. SB401 protein also binds to and bundles actin filaments, and could connect actin filaments to microtubules. SB401 protein had a much higher affinity for microtubules than for actin filaments. In the presence of both cytoskeletal elements, the protein preferentially bound microtubules to form bundles. These results demonstrate that SB401 protein may have important roles in organizing the cytoskeleton in pollen tubes.     

Comparison of the three-dimensional organization of unextracted and Triton-extracted human neutrophilic polymorphonuclear leukocytes

The three-dimensional organization of the cytoplasm of randomly migrating neutrophils was studied by stereo high-voltage electron microscopy. Examination of whole-mount preparations reveals with unusual clarity the structure of the cytoplasmic ground substance and cytoskeletal organization; similar clarity is not observed in conventional sections. An extensive three-dimensional network of fine filaments (microtrabeculae) approximately 7 to 17 nm in diameter extends throughout the cytoplasm and between the two cell cortices; it also comprises the membrane ruffles and filopodia. The granules are dispersed within the lattice and are surrounded by microtrabeculae. The lattice appears to include dense foci from which the microtrabeculae emerge. Triton X-100 dissolves the plasma membrane, most of the granules, and many of the microtrabecular strands and leaves as a more stable structure a cytoskeletal network composed of various filaments and microtubules. Heavy meromyosin-subfragment 1 (S1) decoration discloses actin filaments as the major filamentous component present in membrane ruffles and filopodia. Actin filaments, extending from the leading edge of the cells, are of uniform polarity, with arrowheads pointing towards the cell body. Likewise, the filaments forming the core of filopodia have the barbed end distal. End-to-side associations of actin filaments as well as fine filaments (2--3 nm) which are not decorated with S1 and link actin filaments are observed. The ventral cell cortex includes numerous substrate-associated dense foci with actin filaments radiating from the dense center. Virtually all the microtubules extend from the centrosome. An average of 35 +/- 7 microtubules originate near the pair of centrioles and radiate towards the cell periphery; microtubule fragments are rare. Intermediate filaments form an open network of single filaments in the perinuclear space. Comparison of Triton-extracted and unextracted cells suggest that many of the filamentous strands seen in unextracted cells have as a core a stable actin filament.     

Distribution of microtubules and other cytoskeletal filaments during myotube elongation as revealed by fluorescence microscopy

Summary Distribution of microtubules and other cytoskeletal filaments in growing skeletal muscle cells (myotubes) was studied in vitro by fluorescence microscopy using fluorescin-labeled antibodies and phalloidin, a specific antiactin drug. In the distal elongating tips of myotubes, microtubules were the major cytoskeletal elements; actin and intermediate filaments were much less abundant. On the other hand, colcemidand nocodozole-treatments caused disruption of microtubules and also prompt retraction of growth tips to form myosacs, a type of deformed myotube. Actin filaments remained unaffected during the retraction. The difference in the distribution of the 3 cytoskeletal filaments in the region of growth tips was most remarkable in the case of those myotubes in the process of recovery from myosacs. In an early phase of recovery, the cellular processes extending from myosacs were enriched with both microtubules and intermediate filaments, but not with actin filaments. Later, when the processes became further developed, intermediate filaments were scarce at the extreme ends. Fluorescein-labeled actin introduced by a micro-injection method was minimally incorporated into filaments in the cellular processes. We conclude that microtubules make up the cytoskeletal element which is most responsible for elongation or spreading of growth tips of myotubes in vitro.     

The Interaction of Mip-90 with Microtubules and Actin Filaments in Human Fibroblasts

The novel microtubule-interacting protein Mip-90 was originally isolated from HeLa cells by using affinity columns of agarose derivatized with peptides from the C-terminal regulatory domain on β-tubulin. Biochemical and immunocytochemical data have suggested that the association of Mip-90 with the microtubule system contributes to its cellular organization. Here we report the interaction patterns of Mip-90 with microtubules and actin filaments in interphase human fibroblasts. A polyclonal monospecific antibody against Mip-90 was used for immunofluorescence microscopy analysis to compare the distribution patterns of this protein with tubulin and actin. A detailed observation of fibroblasts revealed the colocalization of Mip-90 with microtubules and actin filaments. These studies were complemented with experiments using cytoskeleton-disrupting drugs which showed that colocalization patterns of Mip-90 with microtubules and actin filaments requires the integrity of these cytoskeletal components. Interestingly, a colocalization of Mip-90 with actin at the leading edge of fibroblasts grown under subconfluency was observed, suggesting that Mip-90 could play a role in actin organization, particularly at this cellular domain. Mip-90 interaction with actin polymers was further supportedin vitroby cosedimentation and immunoprecipitation experiments. The cosedimentation analysis indicated that Mip-90 bound to actin filaments with an association constantK_a= 1 × 10⁶M⁻¹, while an stoichiometry Mip-90/actin of 1:12 mol/mol was calculated. Western blots of the immunoprecipitates revealed that Mip-90 associated to both actin and tubulin in fibroblasts extracts. These studies indicate that Mip-90, described as a microtubule-interacting protein, also bears the capacity to interact with the microfilament network, suggesting that it may play a role in modulating the interactions between these cytoskeletal filaments in nonneuronal cells.     

The cytoskeleton of isolated murine primitive erythrocytes

Summary Cytoskeletons of primitive erythrocytes have been isolated from the embryos of day 12 pregnant C57/Bl mice and examined by transmission electron microscopy, immunofluorescence microscopy, and SDS-polyacrylamide gel electrophoresis. Microtubules are the most prominent cytoskeletal component. They are found either singly or organized into loose bundles just under the plasma membrane, but do not form classical marginal bands in most cells. Immunofluorescence with a polyclonal tubulin antiserum confirms this distribution and further reveals numerous mitotic figures among the cells. Rhodamine-conjugated phalloidin and heavy meromyosin labeling reveal that actin is localized in the cortex of the primitive erythrocyte in the form of 6 nm filaments. Antibody directed against avian erythrocyte alpha spectrin demonstrates that spectrin is also found in the cortex. Occasional 10-nm intermediate filaments, observed in the primitve erythrocytes by electron microscopy, are believed to be of the vimentin class based on positive reaction of the cells with vimentin-specific antiserum. In addition, a band in erythrocyte cytoskeletons comigrates in SDS-polyacrylamide gels with vimentin isolated from mouse kidney. Spectrin and actin were also found to be associated with the membrane of primitive erythrocytes when membrane ghost preparations were analyzed by SDS-polyacrylamide gel electrophoresis.     

Apparent stiffness of vimentin intermediate filaments in living cells and its relation with other cytoskeletal polymers

The cytoskeleton is a complex network of interconnected biopolymers intimately involved in the generation and transmission of forces. Several mechanical properties of microtubules and actin filaments have been extensively explored in cells. In contrast, intermediate filaments (IFs) received comparatively less attention despite their central role in defining cell shape, motility and adhesion during physiological processes as well as in tumor progression. Here, we explored relevant biophysical properties of vimentin IFs in living cells combining confocal microscopy and a filament tracking routine that allows localizing filaments with ~20 nm precision. A Fourier-based analysis showed that IFs curvatures followed a thermal-like behavior characterized by an apparent persistence length (lp*) similar to that measured in aqueous solution. Additionally, we determined that certain perturbations of the cytoskeleton affect lp* and the lateral mobility of IFs as assessed in cells in which either the microtubule dynamic instability was reduced or actin filaments were partially depolymerized. Our results provide relevant clues on how vimentin IFs mechanically couple with microtubules and actin filaments in cells and support a role of this network in the response to mechanical stress.     

Distribution of Cytoskeletal Proteins in Neomycin-Induced Protrusions of Human Fibroblasts

The organization of actin, tubulin, and vimentin was studied in protruding lamellae of human fibroblasts induced by the aminoglycoside antibiotic neomycin, an inhibitor of the phosphatidylinositol cycle. Neomycin stimulates the simultaneous protrusion of lamellae in all treated cells, and the lamellae remain extended for about 15–20 min, before gradually withdrawing. The pattern and distribution of actin, tubulin, and vimentin during neomycin stimulation were analyzed by fluorescence and electron microscopy. F-actin in the newly formed lamellae is localized in a marginal band at the leading edge. Tubulin is colocalized with F-actin in the marginal band, but the newly formed lamellae are initially devoid of microtubules. Over a period of 10 to 20 min after the addition of neomycin, microtubules grow into the lamellae from the adjacent cytoplasm, while the intensity of tubulin staining of the marginal band decreases. Distribution of vimentin remains unchanged in neomycin-treated cells and vimentin filaments do not enter the new protrusions. Treatment of cells with colchicine and Taxol do not inhibit neomycin-induced protrusion but protrusions are no longer localized at the ends of cell processes and occur all around the cell periphery. We conclude that actin filaments are the major component of the cytoskeleton involved in generating protrusions. Microtubules and, possibly, intermediate filaments control the pattern of protrusions by their interaction with actin filaments.