Microheterogeneity of actin gels formed under controlled linear shear

Flow and High Affinity Binding Affect the Elastic Modulus of the Nucleus, Cell Body and the Stress Fibers of Endothelial Cells

Cell mechanical properties are important in the adhesion of endothelial cells to synthetic vascular grafts exposed to shear flow. We hypothesized that the local apparent elastic modulus of the nucleus and the cell body would increase to a greater extent for cells adherent via the dual ligand (integrin-fibronectin/avidin-biotin) and exposed to flow, than for cells treated with either ligand alone. High affinity avidin-biotin bonds and in vitro flow exposure were used to improve adhesion to grafts thereby altering the mechanical properties of endothelial cells. Introduction of the dual ligand chemistry at the cell-substrate interface increased the apparent elastic modulus of the cells as compared to cells adherent with the fibronectin-integrin bonds only. Cells cultured on the dual ligand surface exhibited higher elastic moduli of the nucleus and cell body relative to cells cultured on fibronectin alone. Exposure of cells to flow increased the apparent elastic modulus of the cell body, nucleus, and stress fibers of cells adherent to the fibronectin surface. A similar effect was seen for cells adherent to the dual ligand surface, although there was little effect on the elastic modulus of the nucleus. While the dual ligand surface produces an increase in adhesion strength, focal contact area and elastic modulus, the change in elastic modulus after exposure to flow is due only to an increase in stress fibers and not an increase in contact area.

Mechanical shear can accelerate the gelation of actin filament networks

Rearrangements of the filamentous actin cytoskeleton at the leading edge of motile cells occur under large mechanical stresses. Contrary to conventional wisdom, we show that mechanical deformations applied during gelation can accelerate the rate of gelation and produce F-actin networks that are stiffer and mechanically more resilient than those polymerized under low or high shear deformations. Above a threshold shear strain amplitude, F-actin networks collapse and become soft and liquidlike. This effect of shear-induced strengthening of polymerizing networks depends on the state of hydrolysis of the actin-bound adenosine triphosphate.

Mechanics and multiple-particle tracking microheterogeneity of alpha-actinin-cross-linked actin filament networks

Cell morphology is controlled by the actin cytoskeleton organization and mechanical properties, which are regulated by the available contents in actin and actin regulatory proteins. Using rheometry and the recently developed multiple-particle tracking method, we compare the mechanical properties and microheterogeneity of actin filament networks containing the F-actin cross-linking protein alpha-actinin. The elasticity of F-actin/alpha-actinin networks increases with actin concentration more rapidly for a fixed molar ratio of actin to alpha-actinin than in the absence of alpha-actinin, for networks of fixed alpha-actinin concentration and of fixed actin concentration, but more slowly than theoretically predicted for a homogeneous cross-linked semiflexible polymer network. These rheological measurements are complemented by multiple-particle tracking of fluorescent microspheres imbedded in the networks. The distribution of the mean squared displacements of these microspheres becomes progressively more asymmetric and wider for increasing concentration in alpha-actinin and, to a lesser extent, for increasing actin concentration, which suggests that F-actin networks become progressively heterogeneous for increasing protein content. This may explain the slower-than-predicted rise in elasticity of F-actin/alpha-actinin networks. Together these in vitro results suggest that actin and alpha-actinin provides the cell with an unsuspected range of regulatory pathways to modulate its cytoskeleton's micromechanics and local organization in vivo.

Cryoatomic force microscopy of filamentous actin

Cryoatomic force microscopy (cryo-AFM) was used to image phalloidin-stabilized actin filaments adsorbed to mica. The single filaments are clearly shown to be right-handed helical structures with a periodicity of approximately 38 nm. Even at a moderate concentration ( approximately 10 microg/ml), narrow, branched rafts of actin filaments and larger aggregates have been observed. The resolution achieved is sufficient to resolve actin monomers within the filaments. A closer examination of the images shows that the branched rafts are composed of up to three individual filaments with a highly regular lateral registration with a fixed axial shift of approximately 13 nm. The implications of these higher-order structures are discussed in terms of x-ray fiber diffraction and rheology of actin gels. The cryo-AFM images also indicate that the recently proposed model of left-handed F-actin is likely to be an artifact of preparation and/or low-resolution AFM imaging.

Molecular cloning of human ABPL, an actin-binding protein homologue

Based on two partial cDNA sequences, a full-length cDNA sequence for an actin-binding like protein previously named ABPL has been isolated and characterized. ABPL is homologous to the human actin-binding proteins ABP-280 and ABP-278. The predicted sequence for ABPL is 2,705 amino acids in length with a calculated molecular mass of 289 kDa. It contains an amino terminal actin-binding domain followed by 24 tandem repeats of approximately 96 amino acids. Two hinge regions, Hinge I and Hinge II, were located prior to repeats 16 and 24, respectively. An isoform of ABPL lacking Hinge I, with a calculated molecular mass of 286 kDa, was also identified by the reverse transcriptase PCR (RT-PCR) method. A comparison with genomic sequences indicated the isoform resulted from alternative RNA splicing. ABPL has a unique insertion sequence of 82 amino acids in repeat 20 that was not present in the other two homologues and has a tissue distribution that was also different from the other two homologues.

A Novel Human Actin-Binding Protein Homologue That Binds to Platelet Glycoprotein Ib

Glycoprotein (GP)Ib-IX-V is one of the major transmembrane complexes present on the platelet surface. Its extracellular domain binds von Willebrand factor (vWF) and thrombin, while its intracellular domain associates tightly with the cytoskeleton through the actin-binding protein (ABP)-280, also known as filamin. In the present study, a full-length cDNA coding for a human ABP homologue has been cloned and sequenced. This protein was identified by the yeast two-hybrid screening procedure via its interaction with the intracellular domain of GPIb. Initially, a 1.3-kb partial cDNA was isolated from a megakaryocyte-like cell line (K562) cDNA library followed by a full-length cDNA of 9.4 kb that was identified in a human placenta library. The full-length cDNA encoded a protein of 2,578 amino acids with a calculated molecular weight of 276 kD (ABP-276). The amino terminal 248 amino acids contained an apparent actin binding domain followed by 24 tandem repeats each containing about 96 amino acids. The amino acid sequence of the protein shared a high degree of homology with human endothelial ABP-280 (70% identity) and chicken filamin (83% identity). However, the 32 amino acid Hinge I region in ABP-280 that contains a calpain cleavage site conferring flexibility on the molecule, was absent in the homologue. An isoform containing a 24 amino acid insertion with a unique sequence at the missing Hinge I region was also identified (ABP-278). This isoform resulted from alternative RNA splicing. ABP-276 and/or ABP-278 were present in all tissues examined, but the relative amount varied in that some tissue contained both forms, while other tissue contained predominately one or the other.

© 1998 by The American Society of Hematology.

A Novel Human Actin-Binding Protein Homologue That Binds to Platelet Glycoprotein Ib

Glycoprotein (GP)Ib-IX-V is one of the major transmembrane complexes present on the platelet surface. Its extracellular domain binds von Willebrand factor (vWF) and thrombin, while its intracellular domain associates tightly with the cytoskeleton through the actin-binding protein (ABP)-280, also known as filamin. In the present study, a full-length cDNA coding for a human ABP homologue has been cloned and sequenced. This protein was identified by the yeast two-hybrid screening procedure via its interaction with the intracellular domain of GPIb. Initially, a 1.3-kb partial cDNA was isolated from a megakaryocyte-like cell line (K562) cDNA library followed by a full-length cDNA of 9.4 kb that was identified in a human placenta library. The full-length cDNA encoded a protein of 2,578 amino acids with a calculated molecular weight of 276 kD (ABP-276). The amino terminal 248 amino acids contained an apparent actin binding domain followed by 24 tandem repeats each containing about 96 amino acids. The amino acid sequence of the protein shared a high degree of homology with human endothelial ABP-280 (70% identity) and chicken filamin (83% identity). However, the 32 amino acid Hinge I region in ABP-280 that contains a calpain cleavage site conferring flexibility on the molecule, was absent in the homologue. An isoform containing a 24 amino acid insertion with a unique sequence at the missing Hinge I region was also identified (ABP-278). This isoform resulted from alternative RNA splicing. ABP-276 and/or ABP-278 were present in all tissues examined, but the relative amount varied in that some tissue contained both forms, while other tissue contained predominately one or the other.

© 1998 by The American Society of Hematology.

Expression and subcellular distribution of filamin isotypes in endothelial cells and pericytes

Two principal forms of the actin binding protein, filamin, are expressed in mammalian cells: nonmuscle and muscle isotypes (FLN-1 and FLN-2). A protein that copurifies with an alpha-naphthyl acetate hydrolyzing esterase from human omentum microvessel endothelial cells (EC) is isolated by nondenaturing electrophoresis, sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis and electroblotting. The purified protein is subjected to in situ trypsin cleavage, reversed-phase high performance liquid chromatography (HPLC) and automated Edman degradation. Six peptide fragments from the protein are identified to have 60-66% identity with nonmuscle filamin (ABP-280). Two of these peptides are 100% identical to a previously sequenced human muscle filamin fragment. Polyclonal antibody is produced using a 16-residue synthetic peptide corresponding to a structural beta-sheet region of muscle filamin. Compared with a variety of vascular cells evaluated, retinal pericytes express an abundance of both muscle and non-muscle filamin isotypes. Pericytes contain at least 10 times more muscle filamin than human umbilical vein EC and at least three times the amount expressed in human omentum microvessel and bovine pulmonary artery EC. Differential detergent fractionation indicates that both filamin isotypes are primarily localized in the cytosol and membrane/organelle fractions of pericytes. Another actin crosslinking protein, alpha-actinin, is primarily found in the cytosol and cytoskeletal fractions. The dynamic regulation of actin microfilament organization in pericytes may be controlled in part by the two filamin isotypes, which in turn may contribute to pericyte contractility.

The structure, function, and assembly of actin filament bundles

The cellular organization, function, and molecular composition of selected biological systems with prominent actin filament bundles are reviewed. An overall picture of the great variety of functions served by actin bundles emerges from this overview. A unifying theme is that the actin cross-linking proteins are conserved throughout the eukaryotic kingdom and yet assembled in a variety of combinations to produce actin bundles of differing functions. Mechanisms of actin bundle formation in vitro are considered illustrating the variety of physical and chemical driving forces in this exceedingly complex process. Our limited knowledge regarding the formation of actin filament bundles in vivo is contrasted with the elegant biophysical studies performed in vitro but nonetheless reveals that interactions with membranes, nucleation sites, and other organizational components must contribute to formation of actin bundles in vivo.

Tracer diffusion through F-actin: effect of filament length and cross-linking

We have determined diffusion coefficients for small (50- to 70-nm diameter) fluorescein-thiocarbamoyl-labeled Ficoll tracers through F-actin as a function of filament length and cross-linking. fx45 was used to regulate filament length and avidin/biotinylated actin or ABP-280 was used to prepare cross-linked actin gels. We found that tracer diffusion was generally independent of filament length in agreement with theoretical predictions for diffusion through solutions of rods. However, in some experiments diffusion was slower through short (< or = 1.0 micron) filaments, although this result was not consistently reproducible. Measured diffusion coefficients through unregulated F-actin and filaments of lengths > 1.0 micron were more rapid than predicted by theory for tracer diffusion through rigid, random networks, which was consistent with some degree of actin bundling. Avidin-induced cross-linking of biotinylated F-actin did not affect diffusion through unregulated F-actin, but in cases where diffusion was slower through short filaments this cross-linking method resulted in enhanced tracer diffusion rates indistinguishable from unregulated F-actin. This finding, in conjunction with increased turbidity of 1.0-micron filaments upon avidin cross-linking, indicated that this cross-linking method induces F-actin bundling. By contrast, ABP-280 cross-linking retarded diffusion through unregulated F-actin and decreased turbidity. Tracer diffusion under these conditions was well approximated by the diffusion theory. Both cross-linking procedures resulted in gel formation as determined by falling ball viscometry. These results demonstrate that network microscopic geometry is dependent on the cross-linking method, although both methods markedly increase F-actin macroscopic viscosity.

Theoretical estimates of mechanical properties of the endothelial cell cytoskeleton

Current modeling of endothelial cell mechanics does not account for the network of F-actin that permeates the cytoplasm. This network, the distributed cytoplasmic structural actin (DCSA), extends from apical to basal membranes, with frequent attachments. Stress fibers are intercalated within the network, with similar frequent attachments. The microscopic structure of the DCSA resembles a foam, so that the mechanical properties can be estimated with analogy to these well-studied systems. The moduli of shear and elastic deformations are estimated to be on the order of 10(5) dynes/cm2. This prediction agrees with experimental measurements of the properties of cytoplasm and endothelial cells reported elsewhere. Stress fibers can potentially increase the modulus by a factor of 2-10, depending on whether they act in series or parallel to the network in transmitting surface forces. The deformations produced by physiological flow fields are of insufficient magnitude to disrupt cell-to-cell or DCSA cross-linkages. The questions raised by this paradox, and the ramifications of implicating the previously unreported DCSA as the primary force transmission element are discussed.

Role of the Dictyostelium 30 kDa protein in actin bundle formation

We have studied the formation of bundles in mixtures of actin with the Dictyostelium 30 kDa actin-bundling protein as a function of 30 kDa protein concentration, actin concentration, and filament length. The presence of the 30 kDa protein promotes formation of filament bundles at actin concentrations and filament lengths that are not spontaneously aligned into liquid crystalline domains in the absence of the 30 kDa protein. Bundle formation in the presence of the 30 kDa protein was observed over a broad range of actin filament lengths and concentrations. Bundling was filament length dependent, and short filaments were more efficiently bundled. Bundles formed at actin concentrations as low as 2 microM. The volume fraction of the bundled portion and concentrations of actin and the 30 kDa protein in the bundled portion were measured using a sedimentation assay. Bundles have concentrations of actin and 30 kDa protein that are 10-20 and 5-20 times, respectively, greater than that of the bulk solution. Computer modeling reveals that bundling of actin by a bundling protein increases both the mean length and the polydispersity of the length distribution, factors which lower the actin concentration required for spontaneous alignment within the bundle. We propose that entropy-driven spontaneous ordering may contribute to bundle formation in two ways. Bundling of actin creates longer aggregates with a more polydisperse length distribution in which actin aligns spontaneously within the bundle at very low concentrations. In addition, bundling creates locally high concentrations of actin within these aggregates that will spontaneously align, providing an additional driving force for bundle ordering.

Capping protein levels influence actin assembly and cell motility in dictyostelium

Actin assembly is important for cell motility, but the mechanism of assembly and how it relates to motility in vivo is largely unknown. In vitro, actin assembly can be controlled by proteins, such as capping protein, that bind filament ends. To investigate the function of actin assembly in vivo, we altered the levels of capping protein in Dictyostelium cells and found changes in resting and chemoattractant-induced actin assembly that were consistent with the in vitro properties of capping protein in capping but not nucleation. Significantly, overexpressers moved faster and underexpressers moved slower than control cells. Mutants also exhibited changes in cytoskeleton architecture. These results provide insights into in vivo actin assembly and the role of the actin cytoskeleton in motility.

Formation of liquid crystals from actin filaments

Actin is cross-linked by actin-binding proteins in the cytoplasm to form either isotropic or highly oriented anisotropic structures. The inherent orientation among actin filaments could influence whether an isotropic or highly oriented anisotropic structure is formed. A highly oriented state can arise spontaneously through the formation of liquid crystals as predicted by polymer theory. In this study, the ability of filamentous actin to form liquid crystalline domains was detected using the anisotropic component of scattered light and by observation of birefringence. As liquid crystalline domains formed, the intensity of the anisotropic component of scattered light increased, and birefringent macroscopic oriented domains were directly observed. The formation of liquid crystalline domains was dependent on the concentration of actin filaments and on the average filament length controlled by varying the ratio of gelsolin to actin monomers. The concentration of actin filaments required to form liquid crystalline domains increased moderately as the average length was decreased. At a fixed actin concentration, orientation among the filaments attained a maximum value at a ratio of actin to gelsolin in the range from 1500 to 2000 and decreased as the ratio was increased or decreased from this range. The results are not well explained by theoretical treatments for liquid crystal formation by monodisperse, charged worm-like chains. Differences from the theoretical predictions for formation of liquid crystals are most likely due to the polydisperse filament length of actin. This phenomenon may have important effects on the structural and rheological properties of the cytoplasm in living cells.

Actin binding to lipid-inserted alpha-actinin

The interaction of alpha-actinin with lipid films and actin filaments was investigated. First alpha-actinin was incorporated in lipid films at the air/water interface. Injection of alpha-actinin into the subphase of a lipid monolayer led to a significant increase of the surface pressure only for lipid films consisting of a mixture of a negatively charged lipid with a high proportion of diacylglycerol. These alpha-actinin-containing films were transferred onto silanized quartz slides. Photobleaching experiments in the evanescent field allowed quantification of the lateral number density of the lipid-bound alpha-actinin. In combination with the area increase from the monolayer experiments, the photobleaching measurements suggest that alpha-actinin is incorporated into the lipid film in such a way that actin binding sites are accessible from the bulk phase. Binding experiments confirmed that the alpha-actinin selectively binds actin filaments in this configuration. We also showed that, in contrast to actin filaments which are adsorbed directly onto planar surfaces, the alpha-actinin-bound actin filaments are recognized and cleaved by the actin-severing protein gelsolin. Thus we have constructed an in vitro system which opens new ways for investigations of membrane-associated actin-binding proteins and of the physical behavior of actin filaments in the close neighborhood to membranes.

A 27,000-D core of the Dictyostelium 34,000-D protein retains Ca(2+)-regulated actin cross-linking but lacks bundling activity

Actin cross-linking proteins are important for formation of isotropic F-actin networks and anisotropic bundles of filaments in the cytoplasm of eucaryotic cells. A 34,000-D protein from the cellular slime mold Dictyostelium discoideum mediates formation of actin bundles in vitro, and is specifically incorporated into filopodia. The actin cross-linking activity of this protein is inhibited by the presence of micromolar calcium. A 27,000-D fragment obtained by digestion with alpha-chymotrypsin lacks the amino-terminal six amino acids and the carboxyl-terminal 7,000 D of the intact polypeptide. The 27,000-D fragment retains F-actin binding activity assessed by cosedimentation assays and by 125I-[F-actin] blot overlay technique, F-actin cross-linking activity as assessed by viscometry, and calcium binding activity. Ultrastructural analyses indicate that the 27,000-D fragment is deficient in the bundling activity characteristic of the intact 34,000-D protein. Actin filaments are aggregated into microdomains but not bundle in the presence of the 27,000-D fragment. A polarized light scattering assay was used to demonstrate that the 34,000-D protein increases the orientational correlation among F-actin filaments. The 27,000-D fragment does not increase the orientation of the actin filaments as assessed by this technique. A terminal segment(s) of the 34,000-D protein, lacking in the 27,000-D fragment, contributes significantly to the ability to cross-link actin filaments into bundles.

Morphological changes in liposomes caused by polymerization of encapsulated actin and spontaneous formation of actin bundles

Spherical giant liposomes that had encapsulated skeletal-muscle G-actin were made by swelling a dried lipid mixture of dimyristoyl phosphatidylcholine/cardiolipin, 1:1 (wt/wt), in a solution of G-actin/CaCl2 at 0 degree C. Polymerization of the encapsulated G-actin into actin filaments was achieved by raising the temperature to 30 degrees C. We observed the subsequent shape changes of the liposomes by dark-field and differential interference-contrast light microscopy. After approximately 40 min, which was required for completion of actin polymerization, two shapes of liposome were evident: dumbbell and disk. Elongation of the dumbbell-shaped liposomes was concomitant with actin polymerization. Polarization microscopy showed that actin filaments formed thick bundles in the liposomes and that these filaments lay contiguous to the periphery of the liposome. Localization of actin filaments in the liposomes was confirmed by observation of rhodamine phalloidin-conjugated actin filaments by fluorescence microscopy. Both dumbbell- and disk-shaped liposomes were rigid and kept their shapes as far as actin filaments were stabilized. In contrast, liposomes containing bovine serum albumin were fragile, and their shapes continually fluctuated from Brownian motion, indicating that the actin bundles served as mechanical support for the liposome shapes.

The actin filament severing protein actophorin promotes the formation of rigid bundles of actin filaments crosslinked with alpha-actinin

The actin filament severing protein, Acanthamoeba actophorin, decreases the viscosity of actin filaments, but increases the stiffness and viscosity of mixtures of actin filaments and the crosslinking protein alpha-actinin. The explanation of this paradox is that in the presence of both the severing protein and crosslinker the actin filaments aggregate into an interlocking meshwork of bundles large enough to be visualized by light microscopy. The size of these bundles depends on the size of the containing vessel. The actin filaments in these bundles are tightly packed in some areas while in others they are more disperse. The bundles form a continuous reticulum that fills the container, since the filaments from a particular bundle may interdigitate with filaments from other bundles at points where they intersect. The same phenomena are seen when rabbit muscle aldolase rather than alpha-actinin is used as the crosslinker. We propose that actophorin promotes bundling by shortening the actin filaments enough to allow them to rotate into positions favorable for lateral interactions with each other via alpha-actinin. The network of bundles is more rigid and less thixotropic than the corresponding network of single actin filaments linked by alpha-actinin. One explanation may be that alpha-actinin (or aldolase) normally in rapid equilibria with actin filaments may become trapped between the filaments increasing the effective concentration of the crosslinker.

Probe diffusion in solutions of filamentous actin formed in the presence of gelsolin

Dynamic light scattering was used to characterize the diffusion of monodisperse polystyrene latex spheres (PLS) of different sizes (55-, 105-, and 265-nm radii) in column-purified 0.65 mg/mL actin solutions polymerized with 100 mM KCl in the absence and presence of various amounts of the actin-binding protein gelsolin. The gelsolin and its interaction with actin was initially studied to ensure that the gelsolin could be used to produce filament populations with well-defined mean lengths. Measurements with PLS diffusion probes present showed, in the absence of gelsolin, that the effective local microviscosity in the actin solutions was 5-20 times that of water and that a large fraction of the PLS are trapped within the pores of the actin filament network, as found previously [J. Newman, K. L. Schick, & K. S. Zaner, (1989) Biopolymers 28, 1969-1980]. As the molar ratio of gelsolin to actin was increased, the diffusion coefficients of the PLS approached those in pure water while the degree of PLS trapping went to zero. Measurements of the dependence of the PLS diffusion coefficients on the ratio of actin concentration to the semidilute overlap concentration showed, for the smaller PLS, a transition occurring near the mean global overlap concentration. These results reflect the dissolution of the actin network as the gelsolin concentration is increased and illustrate the role of gelsolin/actin interactions in the control of macromolecular transport within the periphery of cells.

Modulation of contraction by gelation/solation in a reconstituted motile model

The actin-based cytoskeleton is a dynamic component of living cells with major structural and contractile properties involved in fundamental cellular processes. The action of actin-binding proteins can decrease or increase the gel structure. Changes in the actin-based cytoskeleton have long been thought to modulate the myosin II-based contractions involved in these cellular processes, but there has been some debate concerning whether maximal gelation increases or decreases contractile activity. To address this question, we have examined how contractile activity is modulated by the extent of actin gelation. The model system consists of physiologically relevant concentrations and molar ratios of actin filaments (whose lengths are controlled by gelsolin), the actin-cross-linking protein filamin, and smooth muscle myosin II. This system has been studied at the macroscopic and light microscopic levels to relate the gel structure to the rate of contraction. We present results which show that while a minimal amount of structure is necessary to transmit the contractile force, increasing the gel structure inhibits the rate of contraction, despite an increase in the actin-activated Mg(2+)-ATPase activity of myosin. Decreasing the total myosin concentration also inhibits the rate of contraction. Application of cytochalasin D to one side of the contractile network increases the rate of contraction and also induces movement comparable to flare streaming observed in isolated amoeba cytoplasm. These results are interpreted relative to current models of the relationship between the state of gelation and contraction and to the potential effects of such a relationship in the living cell.

Viscoelastic properties of vimentin compared with other filamentous biopolymer networks

The cytoplasm of vertebrate cells contains three distinct filamentous biopolymers, the microtubules, microfilaments, and intermediate filaments. The basic structural elements of these three filaments are linear polymers of the proteins tubulin, actin, and vimentin or another related intermediate filament protein, respectively. The viscoelastic properties of cytoplasmic filaments are likely to be relevant to their biologic function, because their extreme length and rodlike structure dominate the rheologic behavior of cytoplasm, and changes in their structure may cause gel-sol transitions observed when cells are activated or begin to move. This paper describes parallel measurements of the viscoelasticity of tubulin, actin, and vimentin polymers. The rheologic differences among the three types of cytoplasmic polymers suggest possible specialized roles for the different classes of filaments in vivo. Actin forms networks of highest rigidity that fluidize at high strains, consistent with a role in cell motility in which stable protrusions can deform rapidly in response to controlled filament rupture. Vimentin networks, which have not previously been studied by rheologic methods, exhibit some unusual viscoelastic properties not shared by actin or tubulin. They are less rigid (have lower shear moduli) at low strain but harden at high strains and resist breakage, suggesting they maintain cell integrity. The differences between F-actin and vimentin are optimal for the formation of a composite material with a range of properties that cannot be achieved by either polymer alone. Microtubules are unlikely to contribute significantly to interphase cell rheology alone, but may help stabilize the other networks.

Mechanical properties of cytoskeletal polymers

The mechanical properties of cytoplasm are dominated by microfilaments, microtubules, and intermediate filaments, collectively termed the cytoskeleton. This review discusses how the physical properties of these biopolymer systems are related to their molecular structures and interactions, and how remodelling of these biopolymers in vivo affects cell shape and motility.

Human endothelial actin-binding protein (ABP-280, nonmuscle filamin): a molecular leaf spring

Actin-binding protein (ABP-280, nonmuscle filamin) is a ubiquitous dimeric actin cross-linking phosphoprotein of peripheral cytoplasm, where it promotes orthogonal branching of actin filaments and links actin filaments to membrane glycoproteins. The complete nucleotide sequence of human endothelial cell ABP cDNA predicts a polypeptide subunit chain of 2,647 amino acids, corresponding to 280 kD, also the mass derived from physical measurements of the native protein. The actin-binding domain is near the amino-terminus of the subunit where the amino acid sequence is similar to other actin filament binding proteins, including alpha-actinin, beta-spectrin, dystrophin, and Dictyostelium abp-120. The remaining 90% of the sequence comprises 24 repeats, each approximately 96 residues long, predicted to have stretches of beta-sheet secondary structure interspersed with turns. The first 15 repeats may have substantial intrachain hydrophobic interactions and overlap in a staggered fashion to yield a backbone with mechanical resilience. Sequence insertions immediately before repeats 16 and 24 predict two hinges in the molecule near points where rotary-shadowed molecules appear to swivel in electron micrographs. Both putative hinge regions are susceptible to cleavage by proteases and the second also contains the site that binds the platelet glycoprotein Ib/IX complex. Phosphorylation consensus sequences are also located in the hinges or near them. Degeneracy within every even-numbered repeat between 16 and 24 and the insertion before repeat 24 may convert interactions within chains to interactions between chains to account for dimer formation within a domain of 7 kD at the carboxy-terminus. The structure of ABP dimers resembles a leaf spring. Interchain interactions hold the leaves firmly together at one end, whereas intrachain hydrophobic bonds reinforce the arms of the spring where the leaves diverge, making it sufficiently stiff to promote high-angle branching of actin filaments. The large size of the leaves, their interruption by two hinges and flexible actin-binding site, facilitate cross-linking of widely dispersed actin filaments.

Liquid crystal domains and thixotropy of filamentous actin suspensions

The thixotropic properties of filamentous actin suspensions were examined by a step-function shearing protocol. Samples of purified filamentous actin were sheared at 0.2 sec-1 in a cone and plate rheometer. We noted a sharp stress overshoot upon the initiation of shear, indicative of a gel state, and a nearly instantaneous drop to zero stress upon cessation of shear. Stress-overshoot recovery was almost complete after 5 min of "rest" before samples were again sheared at 0.2 sec-1. Overshoot recovery increased linearly with the square root of rest time, suggesting that gel-state recovery is diffusion limited. Actin suspensions subjected to oscillatory shearing at frequencies from 0.003 to 30 radians/sec confirmed the existence of a 5-min time scale in the gel, similar to that for stress-overshoot recovery. Flow of filamentous actin was visualized by polarized light observations. Actin from 6 mg/ml to 20 mg/ml showed the "polycrystalline" texture of birefringence typical for liquid crystal structure. At shear rates less than 1 sec-1, flow occurred by the relative movement of irregular, roughly ellipsoidal actin domains 40-140 microns long; the appearance was similar to moving ice floes. At shear rates greater than 1 sec-1, domains decreased in size, possibly by frictional interactions among domains. Eventually domains flow in a "river" of actin aligned by the flow. Our observations confirm our previous domain-friction model for actin rheology. The similarities between the unusual flow properties of actin and cytoplasm argue that cytoplasm also may flow as domains.

Effect of filamin and controlled linear shear on the microheterogeneity of F-actin/gelsolin gels

We have previously established [Cortese and Frieden, J. Cell Biol. 107:1477-1487, 1988] that actin gels formed under shear are microheterogeneous. In this study, the effect of cross-linking (by chicken gizzard filamin), severing (by plasma gelsolin), and shear on actin microheterogeneity are investigated using fluorescence photobleaching recovery and video microscopy. We find that filamin and shear form microheterogeneous F-actin:gelsolin gels by different mechanisms. Bundling of actin:gelsolin filaments by filamin can be explained by an increase in the apparent length of the filaments due to interfilament binding, resulting in a decrease of the polymer number concentration at which filaments organize into anisotropic phases. Some intrafilament binding of filamin to actin filaments may also be present, and those filaments coated with filamin immobilize more slowly than actin under the same polymerization conditions. The length of F-actin/gelsolin filaments seems to be a major factor in controlling the extent of bundling relative to network formation. In contrast, the effect of shear on the microheterogeneity of actin:gelsolin filaments is consistent with our previous proposal that shear aligns actin filaments, allowing filament-filament interactions and phase formation to occur. Short filaments are unable to organize into branched actin networks, but they can create large aggregates under low shear. Longer actin filaments will exist as networks with variable levels of branching and are less sensitive to shear. The effect of the intensity of a shear field on the spatial distribution of actin may involve a progressively more random orientation of actin molecules and bundles. A regular pattern develops across the sample at low shear rates (0.04-1.39 s-1), and becomes very irregular at higher shear rates (greater than 10 s-1). We suggest here that actin-binding proteins and shear can control the transition between isotropic networks and anisotropic phases by their effect on apparent length and local filament concentration, and also that this transition can have substantial effects on the resistance of cells to mechanical stress.

Viscoelasticity of F-actin measured with magnetic microparticles

Dispersed submicroscopic magnetic particles were used to probe viscoelasticity for cytoplasm and purified components of cytoplasm. An externally applied magnetic field exerted force on particles in cells, in filamentous actin (F-actin) solutions, or in F-actin gels formed by the addition of the actin gelation factor, actin-binding protein (ABP). The particle response to magnetic torque can be related to the viscoelastic properties of the fluids. We compared data obtained on F-actin by the magnetic particle method with data obtained on F-actin by means of a sliding plane viscoelastometer. F-actin solutions had a significant elasticity, which increased by 20-fold when gels were formed by ABP addition. Both methods gave consistent results, but the dispersed magnetic particles indicated quantitatively greater rigidity than the viscoelastometer (two and six times greater for F-actin solutions and for F-actin plus ABP gels, respectively). These differences may be due to the fact that, compared with traditional microrheometers, dispersed particle measurements are less affected by long-range heterogeneity or domain-like structure. The magnetometric method was used to examine the mechanical properties of cytoplasm within intact macrophages; the application of the same magnetometric technique to both cells and well-defined, purified protein systems is a first step toward interpreting the results obtained for living cells in molecular terms. The magnetic particle probe system is an effective nonoptical technique for determining the motile and mechanical properties of cells in vitro and in vivo.

Actin polymerization induces a shape change in actin-containing vesicles

We have encapsulated actin filaments in the presence and absence of various actin-binding proteins into lipid vesicles. These vesicles are approximately the same size as animal cells and can be characterized by the same optical microscopic and mechanical techniques used to study cells. We demonstrate that the initially spherical vesicles can be forced into asymmetric, irregular shapes by polymerization of the actin that they contain. Deformation of the vesicles requires that the actin filaments be on average at least approximately 0.5 micron long as shown by the effects of gelsolin, an actin filament-nucleating protein. Filamin, a filament-crosslinking protein, caused the surfaces of the vesicles to have a smoother appearance. Heterogeneous distribution of actin filaments within the vesicles is caused by interfilament interactions and modulated by gelsolin and filamin. The vesicles provide a model system to study control of cell shape and cytoskeletal organization, membrane-cytoskeleton interactions, and cytomechanics.