Polymorphism of reconstituted human epidermal keratin filaments: determination of their mass-per-length and width by scanning transmission electron microscopy (STEM)

We have determined the mass-per-length (MPL) and the width of unstained freeze-dried reconstituted human epidermal keratin filaments by scanning transmission electron microscopy (STEM). Filaments were reassembled from keratins extracted from four different sources: cultured human epidermal cells (CHEC), human callus (CAL), and the living layers (LL) and stratum corneum (SC) of normal human epidermis. MPL histograms of all four keratin filament types could be fitted by a superposition of two or three Gaussians, with their respective major peaks located between 17 and 20 kDa/nm. We interpreted the multiple MPL peaks to represent different polymorphic forms of the reconstituted filaments. The number of subunits per filament cross section calculated from MPL peak positions, average subunit molecular weight, and an axial repeat of the subunits within the filament of 46.5 nm revealed an average difference between polymorphic variants of 7.5 +/- 0.9 subunits. These data suggest that reconstituted human epidermal keratin filaments are made of two to four 8-stranded "protofibrils" (i.e., made of two laterally aggregated 4-stranded protofilaments), in agreement with earlier observations. The average widths of unstained freeze-dried keratin filaments were larger than those of negatively stained filaments: 12.6 nm (9.6 nm) for CHEC, 12.3 nm (9.7 nm) for CAL, 11.6 nm (8.3 nm) for LL, and 11.3 nm (7.9 nm) for SC keratin filaments, with the values in brackets corresponding to negatively stained samples. Assuming the MPL to be proportional to the square of the filament width, there is a good correlation between the MPL and width measurements both for filaments within a given type as well as among those reconstituted from different types of keratin extracts.     

X-ray diffraction study on ordered, disordered and reconstituted intercellular lipid lamellar structure in stratum corneum

From small angle X-ray diffraction for the stratum corneum of hairless mouse, it was obtained that in the normal stratum corneum, the 1st, 2nd and 3rd order diffraction peaks for the intercellular lipid lamellar structure appear at 13.8, 6.87 and 4.59 nm, respectively and also a broad hump for the 4th order reflection appears as observed by the previous researchers. In the damaged stratum corneum prepared by the treatment of sodium dodecyl sulfate, these small-angle diffraction peaks disappear and only the broad maxima remain around the 1st, 2nd and 3rd order diffraction peaks. These facts indicate that in the normal stratum the lamellar structure is ordered and in the damaged stratum corneum the lamellar structure is disordered. Furthermore, in the reconstituted lamellar structure obtained by immersing into the dilute suspension of the mixture of ceramide 3, cholesterol and stearic acid, the 1st, 2nd and 3rd order diffraction peaks reappear at 13.3, 6.67 and 4.44 nm, respectively. This fact indicates that the reorganization of the ordered lamellar structure takes place by adding the mixture to the damaged stratum corneum.     

Supramolecular forms of actin from amoebae of Dictyostelium discoideum.

Actin purified from amoebae of Dictyostelium discoideum polymerizes into filaments at 24 degrees upon addition of KCl, as judged by a change in optical density at 232 nm and by electron microscopy. The rate and extent of formation of this supramolecular assembly and the optimal KCl concentrations (0.1 M) for assembly are similar to those of striated muscle actin. The apparent equilibrium constant for the monomer-polymer transition is 1.3 muM for both Dictyostelium and muscle actin. Although assembly of highly purified Dictyostelium actin monomers into individual actin filaments resembles that of muscle actin, Dictyostelium actin but not muscle actin was observed to assemble into two-dimensional nets in 10 mM CaCl2. The Dictyostelium actin also forms filament bundles which are 0.1 mum in diameter and which assemble in the presence of 5 mM MgCl2. These bundles formed from partially purified Dictyostelium actin preparations but not from highly purified preparations, suggesting that their formation may depend on the presence of another component. These actin bundles reconstituted in vitro resemble the actin-containing bundles found in situ by microscopy in many non-muscle cells.     

Assembly of smooth muscle myosin into side-polar filaments

The in vitro assembly of myosin purified from calf aorta muscle has been studied by electron microscopy. Two types of filament are formed: short bipolar filament similar to those formed from skeletal muscle myosin, and longer "side-polar" filaments having cross bridges with a single polarity along the entire length of one side and the opposite polarity along the other side. Unlike the case with skeletal myosin filaments, antiparallel interactions between myosin molecules occur along the whole length of side-polar filaments. The side-polar structure may be related to the in vivo form of myosin in vertebrate smooth muscle.     

Assembly of vimentin in vitro and its implications concerning the structure of intermediate filaments

After dialysis against 10 mM-Tris-acetate (pH 8.5), vimentin that has been purified in the presence of urea is present in the form of tetrameric 2 to 3 nm X 48 nm rods known as protofilaments. These building blocks in turn polymerize into intermediate filaments (10 to 12 nm diameter) when they are dialyzed against a solution of physiological ionic strength and pH. By varying the ionic conditions under which polymerization takes place, we have identified two classes of assembly intermediates whose structures provide clues as to how an intermediate filament may be constructed. The structure of the first class, seen when assembly takes place at 10 to 20 mM-salt at pH 8.5, strongly suggests that one of the initial steps of filament assembly is the association of protofilaments into pairs with a half-unit axial stagger. Increasing the ionic strength of the assembly buffer leads to the emergence of short, full-width intermediate filaments at approximately 50 mM-salt at pH 8.5. In the presence of additional protofilaments, these short filaments elongate to many micrometers when the ionic strength and pH are further adjusted to physiological levels. The electron microscope images of the assembly intermediates suggest that vimentin-containing intermediate filaments are made up of eight protofilaments, assembled such that there is an approximately 22 nm axial stagger between neighboring protofilaments. We propose that this half-unit staggering of protofilaments is a fundamental feature of intermediate filament structure and assembly, and that it could account for the 20 to 22 nm axial repeat seen in all intermediate filaments examined so far.     

Isolation of intermediate filament assemblies from human hair follicles

We used developing human hair follicle cells for the isolation of hard alpha-keratin structural components. Intracellular dispersions examined by electron microscopy contained both individual alpha-keratin filaments and the tactoid-like filament assemblies observed in situ organized along subfibrillar arms of macrofibrils. The assemblies of average width 47 nm were composed of closely packed alpha-keratin filaments and originated from the initial filament arrays observed in sections of developing mammalian hair follicles. We have distinguished two types of assemblies: the para-like or hexagonally packed and the ortho-like spiral or whorl type. Axial banding extended across the width of filament assemblies, which suggested that hard alpha-keratin filaments pack in lateral register and form a lattice that contains interfilamentous bridges. We observed axial banding patterns with periods ranging from 20 to 22 nm, consistent with the 22-nm periodic structure deduced from x-ray diffraction studies and present in models proposed for hard alpha-keratin and other intermediate filaments. Preliminary biochemical studies of filaments and filament assemblies indicate that they consist of the closely related group of proteins (low-sulfur proteins) ubiquitous among extracts of hard mammalian keratins. Isolated hard alpha-keratin filament assemblies provide a new and valuable structural entity for investigating the assembly mechanisms involved in the formation of the filament-matrix framework found in hard mammalian keratin appendages.     

Structure and elasticity of desmin protofibrils explored with scanning force microscopy

Desmin filaments form the intermediate filament system of muscle cells where they play important role in maintaining mechanical integrity and elasticity. Although the importance of desmin elasticity and assembly-disassembly dynamics in cellular mechanics is being increasingly recognized, the molecular basis of neither desmin's elasticity nor its disassembly pathway is well understood. In the present work, we explored the topographical structure of purified and reconstituted desmin filaments by using scanning force microscopy. With the addition of divalent cation chelators ethyleneglycoltetraacetic acid or ethylenediaminetetraacetic acid, the filaments disassembled on a time scale of hours to days into stable, thin fibrillar components with variable (up to micrometer) length, smooth surface and uniform thickness, which are identified as protofibrils. Desmin protofibrils appear as elastic structures with a persistence length of 51.5 nm, and their Young's modulus (10.6 MPa) far exceeds that of the mature filament (3.7 MPa). Protofibrillar bundling within the desmin filament results in high longitudinal tensile strength at a large bending flexibility. The stability of protofibrils suggests that desmin may disassemble along a pathway quite distinct from its assembly.     

Morphological analysis of glutaraldehyde-fixed vimentin intermediate filaments and assembly-intermediates by atomic force microscopy

Atomic force microscopy (AFM) was used to study the morphology of vimentin intermediate filaments (IFs) and their assembly intermediates. At each time after initiation of IF assembly in vitro of recombinant mouse vimentin, the sample was fixed with 0.1% glutaraldehyde and then applied to AFM analysis. When mature vimentin IFs were imaged in air on mica, they appeared to have a width of approximately 28 nm, a height of approximately 4 nm and a length of several micrometers. Taking into account the probe tip's distortion effect, the exact width was evaluated to be approximately 25 nm, suggesting that the filaments flatten on the substrate rather than be cylindrical with a diameter of approximately 10 nm. Vimentin IFs in air clearly demonstrated approximately 21-nm repeating patterns along the filament axis. The three-dimensional profiles of vimentin IFs indicated that the characteristic patterns were presented by repeating segments with a convex surface. The repeating patterns close to 21 nm were also observed by AFM analysis in a physiological solution condition, suggesting that the segments along the filaments are an intrinsic substructure of vimentin IFs. In the course of IF assembly, assembly intermediates were analyzed in air. Many short filaments with a full-width and an apparent length of approximately 78 nm (evaluated length approximately 69 nm) were observed immediately after initiation of the assembly reaction. Interestingly, the short full-width filaments appeared to be composed of the four segments. Further incubation enabled the short full-width filaments to anneal longitudinally into longer filaments with a distinct elongation step of approximately 40 nm, which corresponds to the length of the two segments. To explain these observations, we propose a vimentin IF formation model in which vimentin dimers are supercoiling around the filament axis.     

In vitro assembly properties of mutant and chimeric intermediate filament proteins: insight into the function of sequences in the rod and end domains of IF

The factors and mechanisms regulating assembly of intermediate filament (IF) proteins to produce filaments with their characteristic 10 nm diameter are not fully understood. All IF proteins contain a central rod domain flanked by variable head and tail domains. To elucidate the role that different domains of IF proteins play in filament assembly, we used negative staining and electron microscopy (EM) to study the in vitro assembly properties of purified bacterially expressed IF proteins, in which specific domains of the proteins were either mutated or swapped between a cytoplasmic (mouse neurofilament-light (NF-L) subunit) and nuclear intermediate filament protein (human lamin A). Our results indicate that filament formation is profoundly influenced by the composition of the assembly buffer. Wild type (wt) mouse NF-L formed 10 nm filaments in assembly buffer containing 175 mM NaCl, whereas a mutant deleted of 18 NH2-terminal amino acids failed to assemble under similar conditions. Instead, the mutant assembled efficiently in buffers containing CaCl2 > or = 6 mM forming filaments that were 10 times longer than those formed by wt NF-L, although their diameter was significantly smaller (6-7 nm). These results suggest that the 18 NH2-terminal sequence of NF-L might serve two functions, to inhibit filament elongation and to promote lateral association of NF-L subunits. We also demonstrate that lengthening of the NF-L rod domain, by inserting a 42 aa sequence unique to nuclear IF proteins, does not compromise filament assembly in any noticeable way. Our results suggests that the known inability of nuclear lamin proteins to assemble into 10 nm filaments in vitro cannot derive solely from their longer rod domain. Finally, we demonstrate that the head domain of lamin A can substitute for that of NF-L in filament assembly, whereas substitution of both the head and tail domains of lamins for those of NF-L compromises assembly. Therefore, the effect of lamin A "tail" domain alone, or the synergistic effect of lamin "head" and the "tail" domains together, interferes with assembly into 10-nm filaments.     

The assembly of keratins from higher plant cells

Summary In vitro assembly and morphological characteristics of purified 58 kDa, 52 kDa, 50 kDa, and 45 kDa polypeptides in the leaves and the cotyledons of the cabbage (Brassica pekinensis Rupt.) were investigated by electron microscopy and scanning tunneling microscopy. The three or four purified intermediate filament (IF) polypeptides can spontaneously assemble into intermediate filaments in vitro with a 23–24 nm axial repeat, which indicates that keratin IFs in higher plant cells have the same molecular arrangement as in animal cells. STM images suggest that the plant keratin filaments display a pronounced structural polymorphism, which can be composed of 3 nm, 4.5 nm, or 6 nm wide keratin protofilaments.Abbreviation IF intermediate filament - STM scanning tunneling microscopy - SDS sodium dodecyl sulfate - BCIP 5-bromo-4-chloro-3-indolyl phosphate-toluidine - NBC p-nitroblue tetrazolium chloride - PMSF phenylmethyl sulfonylfluoride - HOPG high oriented pyrolytic graphite     

Comparison of 10 nm filaments from three bovine tissues

Enriched fractions of 10 nm filaments were isolated from three bovine tissues and were compared using morphological biochemical, and immunological techniques. We studied keratin filaments from hoof epidermis, 10 nm filaments from corneal epithelium, and 10 nm filaments from brain white matter. The parameters of comparison and results were as follows.

1. 1. Corneal epithelial filaments and keratin filaments repolymerized after a buffered 8 M urea extract of the tissue was dialyzed against a low ionic strength (0.005 M) buffer. However, a greater yield of repolymerized corneal epithelial filaments was obtained if the urea-soluble fraction was dialyzed against the same buffer containing 0.17 M NaCl. Brain filaments harvested by cell fractionation did not repolymerize when similarly treated.

2. 2. Electrophoretic patterns of proteins of filament-enriched fractions from the three sources were different in sodium dodecyl sulphate (SDS) polyacrylamide gels, except for one co-migrating band.

3. 3. Peptide mapping by limited proteolysis of the eluted co-migrating proteins showed few similarities.

4. 4. Amino acid analysis of the co-migrating proteins revealed numerous differences.

5. 5. Antibodies to the co-migrating corneal epithelial filament and brain filament proteins reacted only with their own antigen and whole filament type, and antibody to total keratin filament protein cross-reacted only with keratin filaments.

     

Immunohistochemical analysis of stratum corneum components in oral squamous epithelia

The generation of a stratum corneum in squamous epithelia involves marked changes in morphology and in the expression of cell products. We have examined the expression of some of the components involved in this process in oral squamous epithelia with different terminal differentiation patterns by use of immunofluorescent techniques. Involucrin and transglutaminase are involved in formation of cornified envelopes consistently seen in the stratum corneum. Both components were present in keratinized oral epithelia (palatal epithelium and hyperkeratinized buccal epithelium). The nonkeratinized normal buccal epithelium stained positive as well. Filaggrin, a protein derived from a precursor present in keratohyalin granules, is proposed to aggregate keratin filaments in the cornified layer. Although the staining differed markedly in quantity, this component was likewise detected in both keratinized and nonkeratinized epithelia. The staining patterns for different keratin polypeptides, however, showed qualitative differences between the different epithelia. Thus, it seems that the keratin composition shows differentiation-specific characteristics, whereas the presence of other important components needed to generate a stratum corneum is not as closely related to the terminal differentiation pattern of oral epithelia.     

Probing of the structural stability of vimentin and desmin-type intermediate filaments with Ca2+-activated proteinase, thrombin and lysine-specific endoproteinase Lys-C

Intermediate filaments (IFs) reconstituted from purified, delipidated vimentin and desmin as well as respective protofilaments were subjected to degradation by Ca2+-activated neutral thiol proteinase, thrombin and lysine-specific endoproteinase Lys-C, respectively. The breakdown products were analyzed by SDS-polyacrylamide gel electrophoresis and negative stain electron microscopy. While Ca2+-activated proteinase and thrombin caused rapid and complete degradation of IFs with kinetics not significantly different from those of the degradation of protofilaments, lysine-specific endoproteinase did not exert any electron microscopically detectable effect on filament structure. Although both types of subunit proteins were truncated at their non-alpha-helical, C-terminal polypeptides by this proteinase, they were still able to assemble into 10 nm filaments. Closer electron microscopic inspection of IFs treated with Ca2+-activated proteinase revealed numerous ruptures along the filaments already at very early stages of digestion. SDS-polyacrylamide gel electrophoresis of the processed filaments in conjunction with previous biochemical characterizations of the breakdown of protofilaments by Ca2+-activated proteinase showed that these inhomogeneities primarily arose from degradation of the arginine-rich, non-alpha-helical N-termini of the filament proteins. These findings demonstrate that, although the N-terminus of vimentin and desmin is essential for filament stability, it is still highly susceptible to proteolytic attack in particular and very likely to posttranslational modification in general. Such structural modifications of the N-termini of IF proteins might exert great influences on the intracellular distribution and molecular organization of IFs in various physiological and pathological conditions.     

Structure of the peripheral domains of neurofilaments revealed by low angle rotary shadowing

The structure of the peripheral domains of neurofilaments (NFs) was revealed by rotary shadowing electron microscopy. NFs were isolated from bovine spinal cords by Sepharose CL-4B gel filtration and examined by low angle rotary shadowing. The peripheral domains appeared as thin, flexible, filamentous structures projecting from the intermediate filament core, with a constant density along their entire length. The average length of the projections was approximately 85 nm and the width about 4 nm. These projections appeared from regularly distributed sites, at 22 nm spacing, which seemed to correspond to the typical repeat of the alpha-helix-rich rod domain of the core filament. The density of the projections was found to be 4.1 (+/- 0.6) per 22 nm. We performed reconstitution experiments using purified NF polypeptides to confirm that the projection was indeed the NF peripheral domain. Individual components of the NF triplet, i.e. NF-L, NF-M and NF-H, were purified by DE-52 and Mono-Q anion exchange chromatographies in the presence of 6 M-urea and were assembled in various combinations into filaments. Reassembled filaments were somewhat more slender than the isolated NFs and exhibited a distinct 22 nm axial periodicity. While prominent projections were not observed in the filaments assembled from NF-L alone, reconstructed filaments containing NF-L plus either NF-M or NF-H revealed many projections. The average length of the projections in the filaments reconstructed from NF-L and NF-H was about 63 nm. The projections of reconstructed filaments from NF-L and NF-M were about 55 nm in length. The difference in the lengths of the projections might reflect the difference in the length of the carboxy-terminal tail domain between NF-M and NF-H. The results are interpreted to show that the carboxy-terminal tail domains of NFs project in a regular pattern from the core filament, which is consistent with a half-staggered organization of the tetrameric subunits.     

Keratin filaments of cultured human epidermal cells. Formation of intermolecular disulfide bonds during terminal differentiation.

Human epidermal cells grown in culture synthesize abundant keratins. These keratins are similar to those of stratum corneum of human epidermal callus in their insolubility in dilute aqueous buffers, their molecular weight range of 40,000 to 60,000, their immunolgical reactivity, and their ability to assemble into 80 A tonofilaments in vitro; but there are differences in the molecular weights of some of the proteins, the number of components, and their charge heterogeneity, related at least in part to phosphorylation. About 30% of all the proteins of living cultured keratinocytes consists of keratins, compared with over 85% of stratum corneum. All the keratins of human stratum corneum were found to be cross-linked by intermolecular disulfide bonds while most keratins of the living cells were not. As the cells mature in Methocel-stabilized suspension culture, their keratins become increasingly disulfide cross-linked. When uncross-linked tonofilaments of living keratinocytes are dissolved in 8 M urea and the filaments reconstituted in vitro their keratins become disulfide cross-linked under aerobic conditions and consequently insoluble in solutions of 8 M urea or sodium dodecyl sulfate. The results indicate that the uncross-linked state of the keratins in living cells is due to the reducing intracellular environment and not to a precursor state related to the primary structure of the proteins. The disulfide cross-links stabilizing the keratin filaments must be distinguished from the epsilon-(gamma-glutamyl)lysine cross-links stabilizing the cornified cell envelope.     

Reconstitution of the muscle thin filament from recombinant troponin components and the native thin filaments

We have developed a technique by which muscle thin filaments are reconstituted from the recombinant troponin components and the native thin filaments. By this technique, the reconstituted troponin complex is exchanged into the native thin filaments in the presence of 20% glycerol and 0.3 M KCl at pH 6.2. More than 90% of endogenous troponin complex was replaced with the recombinant troponin complex. Structural integrity and Ca²⁺ sensitivity of the reconstituted thin filament prepared by this technique was confirmed by X-ray fiber diffraction measurements and the thin filament-activated myosin subfragment 1 ATPase measurements, respectively.     

Keratinization of the outer surface of the avian scutate scale: Interrelationship of alpha and beta keratin filaments in a cornifying tissue

Summary The outer surface of adult Gallus domesticus scutate scale was studied as a model for epidermal cornification involving accumulation of both alpha and beta keratins. Electron-microscopic analysis demonstrated that the basal cells of the adult epidermis contained abundant lipid droplets and that filament bundles and desmosomes were distributed throughout the cell layers. Indirect immunofluorescence microscopy and double-labeling immunogold-electron microscopy confirmed that the stratum germinativum contained alpha keratin but not beta keratin. Beta keratins were first detected in the stratum intermedium and were always found intermingled with filament bundles of alpha keratin. As the differentiating cells moved into the outer regions of the stratum intermedium and the stratum corneum, the large mixed keratin filament bundles labeled increasingly more with beta keratin antiserum and relatively less so with alpha keratin antiserum. Sodium dodecyl sulfate-polyacrylamide gel analysis of vertical layers of the outer surface of the scutate scale confirmed that cells having reached the outermost layers of stratum corneum had preferentially lost alpha keratin. The mixed bundles of alpha and beta keratin filaments were closely associated with desmosomes in the lower stratum intermedium and with electron-dense aggregates in the cytoplasm of cells in the outer stratum intermedium. Using anti-desmosomal serum it was shown that these cytoplasmic plaques were desmosomes.     

Structure of the myosin filaments of relaxed and rigor vertebrate striated muscle studied by rapid freezing electron microscopy.

Rapid freezing followed by freeze-substitution has been used to study the ultrastructure of the myosin filaments of live and demembranated frog sartorius muscle in the states of relaxation and rigor. Electron microscopy of longitudinal sections of relaxed specimens showed greatly improved preservation of thick filament ultrastructure compared with conventional fixation. This was revealed by the appearance of a clear helical arrangement of myosin crossbridges along the filament surface and by a series of layer line reflections in computed Fourier transforms of sections, corresponding to the layer lines indexing on a 43 nm repeat in X-ray diffraction patterns of whole, living muscles. Filtered images of single myosin filaments were similar to those of negatively stained, isolated vertebrate filaments and consistent with a three-start helix. M-line and other non-myosin proteins were also very well preserved. Rigor specimens showed, in the region of overlapping myosin and actin filaments, periodicities corresponding to the 36, 24, 14.4 and 5.9 nm repeats detected in X-ray patterns of whole muscle in rigor; in the H-zone they showed a disordered array of crossbridges. Transverse sections, whose Fourier transforms extend to the (3, 0) reflection, supported the view, based on X-ray diffraction and conventional electron microscopy, that in the overlap zone of relaxed muscle most of the crossbridges are detached from the thin filaments while in rigor they are attached. We conclude that the rapid freezing technique preserves the molecular structure of the myofilaments closer to the in vivo state (as monitored by X-ray diffraction) than does normal fixation.     

Cytoskeleton-associated plectin: in situ localization, in vitro reconstitution, and binding to immobilized intermediate filament proteins

The association and interaction of plectin (Mr 300,000) with intermediate filaments and filament subunit proteins were studied. Immunoelectron microscopy of whole mount cytoskeletons from various cultured cell lines (rat glioma C6, mouse BALB/c 3T3, and Chinese hamster ovary) and quick-frozen, deep-etched replicas of Triton X-100-extracted rat embryo fibroblast cells revealed that plectin was primarily located at junction sites and branching points of intermediate filaments. These results were corroborated by in vitro recombination studies using vimentin and plectin purified from C6 cells. Filaments assembled from mixtures of both proteins were extensively crosslinked by oligomeric plectin structures, as demonstrated by electron microscopy of negatively stained and rotary-shadowed specimens as well as by immunoelectron microscopy; the binding of plectin structures on the surface of filaments and cross-link formation occurred without apparent periodicity. Plectin's cross-linking of reconstituted filaments was also shown by ultracentrifugation experiments. As revealed by the rotary-shadowing technique, filament-bound plectin structures were oligomeric and predominantly consisted of a central globular core region of 30-50 nm with extending filaments or filamentous loops. Solid-phase binding to proteolytically degraded vimentin fragments suggested that plectin interacts with the helical rod domain of vimentin, a highly conserved structural element of all intermediate filament proteins. Accordingly, plectin was found to bind to the glial fibrillar acidic protein, the three neurofilament polypeptides, and skin keratins. These results suggest that plectin is a cross-linker of vimentin filaments and possibly also of other intermediate filament types.     

A nucleation--elongation mechanism for the self-assembly of side polar sheets of smooth muscle myosin.

Self-assembled filaments of smooth muscle myosin were observed by low dose electron microscopy to be flat side-polar sheets, in which the component molecules appeared straight and close-packed. Fraying experiments released small oligomers, in which molecules were staggered in parallel by about +/- 14 nm relative to two immediate neighbours, and were bound also to an antiparallel partner via a approximately 14 nm overlap at the very tip of the tail. We suggest a filament model which preserves these packing relationships. Adding stoichiometric amounts of MgATP to the filaments caused them to disassemble completely by progressive loss of material from their ends, at a limiting rate equivalent to about 2 monomers per second per end in physiological saline. The rate of the competing association reaction varied linearly with the monomer concentration, as determined in pressure-jump experiments. This suggests that myosin monomers, rather than dimers or higher oligomers, are the building blocks of these filaments. Shearing and annealing of assembled filaments appeared negligible on a time scale of a few hours. In consequence, filament number and filament length were dependent on the rate at which monomers were supplied to the assembly reaction, and on the number of filaments already present at the start of the assembly reaction.