Thyone sperm were demembranated with Triton X-IO0 and, after washing, extracted with 30 mM Tris at pH 8.0 and 1 mM MgCI~. After the insoluble contaminants were removed by centrifugation, the sperm extract was warmed to 22°C. Actin filaments rapidly assembled and aggregated into bundles when KCI was added to the extract. When we added preformed actin filaments, i.e., the acrosomal filament bundles of Limulus sperm, to the extract, the actin monomers rapidly assembled on these filaments. What was unexpected was that assembly took place on only one end of the bundle--the end corresponding to the preferred end for monomer addition. We showed that the absence of growth on the nonpreferred end was not due to the presence of a capper because exogenously added actin readily assembled on both ends. We also analyzed the sperm extract by SDS gel electrophoresis. Two major proteins were present in a 1:1 molar ratio: actin and a 12,500-dalton protein whose apparent isoelectric point was 8.4. The 12,500-dalton protein was purified by DEAE chromatography. We concluded that it is profilin because of its size, isoelectric point, molar ratio to actin, inability to bind to DEAE, and its effect on actin assembly. When profilin was added to actin in the presence of Limulus bundles, addition of monomers on the nonpreferred end of the bundle was inhibited, even though actin by itself assembled on both ends. Using the Limulus bundles as nuclei, we determined the critical concentration for assembly off each end of the filament and estimated the Kd for the profilin-actin complex (,-,10 ~tM). We present a model to explain how profilin may regulate the extension of the Thyone acrosomal process in vivo: The profilin-actin complex can add to only the preferred end of the filament bundle. Once the actin monomer is bound to the filament, the profilin is released, and is available to bind to additional actin monomers. This mechanism accounts for the rapid rate of filament elongation in the acrosomal process in vivo.

We have reinvestigated the effects of Ca ++ and ATP on brush borders isolated from intestinal epithelial cells. At 37°C, Ca ++ (1/tM) and ATP cause a dramatic contraction of brush border terminal webs, not a retraction of microvilli as previously reported (M. S. Mooseker, 1976, J. Cell Biol. 71:417-433). Terminal web contraction, which occurs over the course of 1-5 min at 37°C, actively constricts brush borders at the level of their zonula adherens. Contraction requires ATP, is stimulated by Ca ++ (1 /xM), and occurs in both membrane-intact and demembranated brush borders. Ca÷÷-dependent-solation of microvillus cores requires a concentration of Ca ÷+ slightly greater (10/~M) than that required for contraction.Under conditions in which brush borders contract, many proteins in the isolated brush borders become phosphorylated. However, the phosphorylation of only one of the brush border proteins, the 20,000 dalton (20-kdalton) light chain of brush border myosin (BBMLC2o), is stimulated by Ca ÷+. At 37°C, BBMLC2o phosphorylation correlates directly with brush border contraction. Furthermore, both BBMLC2o phosphorylation and brush border contraction are inhibited by trifluoperazine, an anti-psychotic phenothiazine that inhibits calmodulin activity. These results indicate that Ca ++ regulates brush border contractility in vitro by stimulating cytoskeleton-associated, Ca +÷-and calmodulin-dependent brush border myosin light chain kinase.Ca ++ has been implicated as a regulator of both motility and cytoplasmic structure in nonmuscle ceils. For example, changes in free Ca ++ concentration may regulate motility in vertebrate nonmuscle and smooth muscle cells by controlling phosphorylation of the regulatory light chain of myosin by a Ca ++-calmodulin-dependent myosin light chain kinase. Phosphorylation of the regulatory light chain of smooth and nonmuscle myosin increases its actin-activated Mg++-ATPase activity, presumably increasing force production and causing motility (for review, see reference 1).Ca ++ may also regulate cytoplasmic structure by disrupting the bundles or isotropic networks of cross-linked actin filaments that have been postulated to contribute to the gel-like properties of cytoplasm (see reference 12, for review). Raising the concentration of free Ca ++ in cytoplasmic extracts disrupts supramolecular actin filament structures. This indicates that Ca ++ may regulate the rapid changes in cytoplasmic viscosity that can occur in vivo. Recently, two major classes of Ca ++-sensitive actin-binding proteins which could modulate the reversible breakdown of actin filament structures have been identified. These include proteins which cross-link actin filaments in the absence, but not in the presence, of Ca ++ (9,50,58,60,79) and proteins which reversibly cut or shorten filaments in the presence of Ca ++ (6,13,26,30,46,53). The mechanism by which changes in structure are coordinated with motility in vivo is still a matter of speculation (32).One of the most extensively ordered arrays of contractile prote...

The brush border of intestinal epithelial cells consists of a tightly packed array of microvilli, each of which contains a core of actin filaments. It has been postulated that microvillar movements are mediated by myosin interactions in the terminal web with the basal ends of these actin cores (Mooseker, M. S. 1976. J. Cell Biol. 71:417-433). We report here that two predictions of this model are correct: (a) The brush border contains myosin, and (b) myosin is located in the terminal web. Myosin is isolated in 70% purity by solubilization of Triton-treated brush borders in 0.6 M KI, and separation of the components by gel filtration. Most of the remaining contaminants can be removed by precipitation of the myosin at low ionic strength. The yield is -1 mg of myosin/30 mg of solubilized brush border protein. The molecule consists of three subunits with molecular weights of 200,000, 19,000, and 17,000 daltons in a 1:1:1 M ratio. At low ionic strength, the myosin forms small, bipolar filaments with dimensions of 300 • 11 nm, that are similar to filaments seen previously in the terminal web of isolated brush borders. Like that of other vertebrate, nonmuscle myosins, the ATPase activity of isolated brush border myosin in 0.6 M KC1 is highest with EDTA (i /xmol Pi/ mg-min; 37~ intermediate with Ca ++ (0.4 /xmol Pi/mg-min), and low with Mg ++ (0.01 /~mol Pi/mg-min). Actin does not stimulate the Mg-ATPase activity of the isolated enzyme. Antibodies against the rod fragment of human platelet myosin cross-react by immunodiffusion with brush border myosin. Staining of isolated mouse or chicken brush borders with rhodamine-antimyosin demonstrates that myosin is localized exclusively in the terminal web.KEY WORDS antibody brush border intestine myosin . terminal webThe arrangement of actin and its companion proteins in the cortex of cells must be highly ordered, given the anisotropic nature of most movements involving the cell surface. Cytokinesis is a classic example (26,35). From "guilt by association" evidence, we surmise that a key feature of this arrangement is the interaction of actin with the plasma membrane (4, 6, 20, 22-25, 32, 33, 37, 40). It seems reasonable to assume that the ani-444 J. CELL BIOLOGY 9 The Rockefeller University Press

Calmodulin is present in brush borders isolated from intestinal epithelial cells and is one of the major components of the microvillar filament bundle . Calmodulin was purified from either demembranated brush borders or microvilli by a simple boiling procedure . The boiled supernate derived from the microvillus cores contained one major polypeptide of 20,000 daltons . The supernate from the brushborder preparation contained the 20,000-dalton subunit and a second protein of 30,000 daltons . The 20,000-dalton subunit has been identified as calmodulin by several criteria: (a) heat resistance, (b) comigration with brain calmodulin on alkaline urea gels and SDS gels, both cases in which the 20,000-dalton protein, like calmodulin, exhibits a shift in electrophoretic mobility in the presence of Ca", and (c) 4-5-fold activation of 3',5'-cyclic nucleotide phosphodiesterase in the presence but not the absence of Ca" . With a cosedimentation assay it was determined that brush-border calmodulin does not bind directly to actin . In the presence of Ca" (>5 X 10-' M) there was a partial release of calmodulin from the microvillus core, along with a substantial conversion of microvillus actin into a nonpelletable form. The dissociation of calmodulin was reversed by removal of Ca" . If microvillus cores were pretreated with phalloidin, the Ca"-induced solubilization of actin was prevented, but the partial dissociation of calmodulin still occurred. The molar ratio of calmodulin:actin is 1 :10 in the demembranated brush border and 1 :2-3 in the microvillus core . No calmodulin was detected in the detergent-solubilized brush-border membrane fraction .Movements of microvilli on the apical, brush-border surface of cells lining the intestine may facilitate the absorptive functions of this epithelium . In vivo observations of brush-border motility have been recorded (32, 38), although adequate documentation of these technically difficult observations is still lacking . Studies on the isolated brush border indicate that this organelle is a highly organized, motile apparatus comprised of actin, myosin, and various associated proteins that are organized in a functional configuration analogous to that determined for actin and myosin in striated muscle (5,(25)(26)(27) . In vitro experiments on contractile models of the brush border have helped to 916 reveal possible molecular mechanisms for microvillar movement (25,30) . The addition of MgATP and micromolar levels of free calcium ions to demembranated brush borders isolated from chicken intestine causes rapid retraction of the microvillar cores into and through the terminal web (25). Presumably, this contractile mechanism coupled to a relaxation mechanism, perhaps involving the extensive linkages of the microvillus cores to the plasma membrane (6,24,25,27,28), is responsible for cyclic movements of microvilli in vivo.We hope to determine the molecular basis for the observed calcium sensitivity of this contrac-J . CELL BIOLOGY

The terminal web of the intestinal brush border contains a spectrin-like protein, TW 260/240 (Glenney, J. R ., Jr., P. Glenney, M. Osborne, and K. Weber, 1982, Cell, 28:843-854 .) that interconnects the "rootlet" ends of microvillar filament bundles in the terminal web (Hirokawa, N ., R. E. Cheng, and M . Willard, 1983, Cell, 32 :953-965; Glenney J . R ., P. Glenney, and K. , /. Cell Biol., 96:1491-1496 . We have investigated further the structural properties of TW 260/240 and the interaction of this protein with actin . Salt extraction of TW 260/240 from isolated brush borders results in a loss of terminal web cross-linkers primarily from the apical zone directly beneath the plasma membrane. Morphological studies on purified TW 260/240 using the rotary shadowing technique confirm earlier results that this protein is spectrin-like and is in the tetrameric state in buffers of low ionic strength . However, examination of TW 260/240 tetramers by negative staining revealed a molecule much straighter and more uniform in diameter than rotary-shadowed molecules . At salt concentrations at (150 mM KCI) and above (300 mM KCI) the physiological range, we observed a partial dissociation of tetramers into dimers that occurred at both 0°and 37°C. We also observed (in the presence of 75 mM KCI) a concentration-dependent self-association of TW 260/240 into sedimentable aggregates .We have studied the interaction of TW 260/240 with actin using techniques of cosedimentation, viscometry, and both light and electron microscopy. We observed that TW 260/240 can bind and cross-link actin filaments and that this interaction is salt-and pHdependent . Under optimum conditions (25-75 mM KCI, at pH 7 .0) TW 260/240 cross-linked F-actin into long, large-diameter bundles. The filaments within these bundles were tightly packed but loosely ordered . At higher pH (7.5) such bundles were not observed, although binding and cross-linking were detectable by co-sedimentation and viscometry. At higher salt (>150 mM KCI), the binding of TW 260/240 to actin was inhibited . The presence of skeletal muscle tropomyosin had no significant effect on the salt-dependent binding of TW 260/240 to F-actin .The ubiquitous presence of spectrin-like proteins in the "cortical" cytoplasm of cells has been recently established by numerous biochemical and immunological studies on a wide variety of cell types and tissues (reviewed in references 1 and 2). By analogy with erythrocyte spectrin