Cell Migration: A Physically Integrated Molecular Process

Cell migration plays a central role in a wide variety of biological phenomena. In embryogenesis, cellular migrations are a recurring theme in important morphogenic processes ranging from gastrulation to development of the nervous system. Migration remains prominent in the adult organism, in normal physiology as well as pathology. In the inflammatory response, for example, leukocytes immmigrate into areas of insult, where they mediate phagocytic and immune functions. Migration of fibroblasts and vascular endothelial cells is essential for wound healing. In metastasis, tumor cells migrate from the initial tumor mass into the circulatory system, which they subsequently leave and migrate into a new site. Finally, cell migration is crucial to technological applications such as tissue engineering, playing an essential role in colonization of biomaterials scaffolding. As with many other cellular processes, the molecular components involved in cell migration are being identified at a rapid rate, and determination of how they participate in migration is following only somewhat more slowly. But also, like most other cell functions, the manner in which these components work together as a dynamic, integrated system to give rise to migration is only beginning to be studied. Understanding cell migration as an integrated process requires an appreciation of chemical and physical properties of multicomponent structures and assemblies, including their thermodynamic, kinetic, and mechanical characteristics, because migration is a process that is physically coordinated both spatially and temporally. Only when it is understood as an integrated system will its alteration via genetic, pharmacologic, or materials-based interventions acquire a truly rational basis. In this article, we offer a perspective on cell migration emphasizing the physicochemical nature of underlying molecular mechanisms. Owing to imposed space and citation constraints, we focus on a limited set of issues, stressing conceptual insights. Readers interested in further discussions and literature citations are referred to some excellent reviews of relevant topics published in the past couple of years (34Ginsberg M.H Schwartz M.A Schaller M.D Integrins emerging paradigms of signal transduction.Annu. Rev. Cell Dev. Biol. 1995; 11: 549-600Crossref PubMed Scopus (1467) Google Scholar, 36Hall A Small GTP-binding proteins and the regulation of the actin cytoskeleton.Annu. Rev. Cell Biol. 1994; 10: 31-54Crossref PubMed Scopus (768) Google Scholar, 41Huttenlocher A Sandborg R.R Horwitz A.F Adhesion in cell migration.Curr. Opin. Cell Biol. 1995; 7: 607-706Crossref Scopus (449) Google Scholar, 45Janmey P.A Phosphoinositides and calcium as regulators of cellular actin assembly and disassembly.Annu. Rev. Physiol. 1994; 56: 169-191Crossref PubMed Scopus (471) Google Scholar, 67Oliver T Lee J Jacobson K Forces exerted by locomoting cells.Semin. Cell Biol. 1994; 5: 139-147Crossref PubMed Scopus (76) Google Scholar, 76Schafer D.A Cooper J.A Control of actin assembly at filament ends.Annu. Rev. Cell Dev. Biol. 1995; 11: 497-518Crossref PubMed Scopus (172) Google Scholar, 83Sheetz M.P Cell migration by graded attachment to substrates and contraction.Semin. Cell Biol. 1994; 5: 149-155Crossref PubMed Scopus (63) Google Scholar, 85Stossel T.P On the crawling of animal cells.Science. 1993; 260: 1086-1094Crossref PubMed Scopus (905) Google Scholar, 87Sun H.-Q Kwiatkowska K Yin H.L Actin monomer binding proteins.Curr. Opin. Cell Biol. 1995; 7: 102-110Crossref PubMed Scopus (171) Google Scholar, 88Theriot J.A Regulation of the actin cytoskeleton in living cells.Semin. Cell Biol. 1994; 5: 193-199Crossref PubMed Scopus (41) Google Scholar). Here we examine, while focusing on their coordination, distinctive aspects of locomotion: morphological polarization, membrane extension, formation of cell–substratum attachments, contractile force and traction, and release of attachments. Almost all of the information we attempt to integrate comes from in vitro studies, mainly concerning movement across two-dimensional substrata. We nonetheless believe that much of the mechanistic understanding is relevant and useful for in vivo situations even in three dimensions. It is likely that cells interact with their surroundings by means of the same types of receptors in vivo as in vitro, and that physical interactions of cells with their environment play important roles in regulating function in both cases. We further attempt to suggest generalizations across a wide spectrum of migratory cell types, including amoebae, leukocytes, fibroblasts, and neurons, looking for broad similarities among physical mechanisms. Observations of various cells demonstrating rapid, slow, or negligible locomotion on particular substrata may be explained as much by quantitative differences in physicochemical properties affecting how intracellular forces are generated and transmitted to the environment, using related processes, as by fundamentally distinct underlying mechanisms. Qualitative differences in migratory behavior may readily derive from quantitative differences in parameters that govern the integration of molecular components, altering relative balances of rates and forces. This is not surprising, since cell migration can be shifted between “on” and “off” by quantitative changes in the concentrations of molecular components (41Huttenlocher A Sandborg R.R Horwitz A.F Adhesion in cell migration.Curr. Opin. Cell Biol. 1995; 7: 607-706Crossref Scopus (449) Google Scholar), such as adhesion receptors, cytoskeletal-linking proteins, and extracellular matrix ligands. But cell migration can also be modified by quantitatively changing physicochemical properties such as receptor–ligand binding avidity (29Duband J.-L Dufour S Yamada S.S Yamada K.M Thiery J.-P Neural crest cell locomotion induced by antibodies to β1 integrins a tool for studying the roles of substratum molecular avidity and density in migration.J. Cell Sci. 1991; 98: 517-532PubMed Google Scholar) and strength of receptor–cytoskeleton interactions (48Kassner P.D Alon R Springer T.A Hemler M.E Specialized functional properties of the integrin α4 cytoplasmic domain.Mol. Biol. Cell. 1995; 6: 661-674Crossref PubMed Scopus (96) Google Scholar). Thus, a productive view of cell migration, as well as other complex cell behavioral functions, will be that of a physically integrated molecular system in which changes in behavior are affected by quantitative alterations in the parameters characterizing kinetic and mechanical features of the molecular interactions. To migrate, cells must acquire a spatial asymmetry enabling them to turn intracellularly generated forces into net cell body translocation. One manifestation of this asymmetry is a polarized morphology, i.e., a clear distinction between cell front and rear. Concentration gradients of stimuli are not required to elicit this response. Polarization in macroscopically homogeneous stimulus environments may arise from perceived spatial or temporal stimulus gradients caused by microscopic nonuniformities or by kinetic fluctuations in receptor–ligand binding. An early event in polarization, at least for neutrophils, following stimulation of rounded cells by chemoattractant ligands is a change in filamentous, F-actin distribution from azimuthal symmetry around the cell rim to concentration at a particular region (14Coates T.D Watts R.G Hartman R Howard T.H Relationship of F-actin distribution to development of polar shape in human polymorphonuclear leukocytes.J. Cell Biol. 1992; 117: 765-774Crossref PubMed Scopus (99) Google Scholar). Additional molecular rearrangements can ensue, leading to cellular spatial asymmetries involved in migration, such as forward redistribution of chemosensory signaling receptors (86Sullivan S.J Daukas G Zigmond S.H Asymmetric distribution of the chemotactic peptide receptor on polymorphonuclear leukocytes.J. Cell Biol. 1984; 99: 1461-1467Crossref PubMed Scopus (64) Google Scholar), integrin adhesion receptors (58Lawson M.A Maxfield F.R Ca2+- and calcineurin-dependent recycling of an integrin to the front of migrating neutrophils.Nature. 1995; 377: 75-79Crossref PubMed Scopus (478) Google Scholar), and integrin–cytoskeleton linkages (79Schmidt C.E Horwitz A.F Lauffenburger D.A Sheetz M.P Integrin–cytoskeletal interactions in migrating fibroblasts are dynamic, asymmetric, and regulated.J. Cell Biol. 1993; 123: 977-991Crossref PubMed Scopus (295) Google Scholar). An important consequence of polarization is that extension of active membrane processes, including both lamellipodia and filopodia, takes place primarily around the cell front, so that directional turning is generally accomplished gradually, with cell locomotion taking on a persistent random walk character. The overall rate of cell migration in the absence of stimulus gradients is thus dependent on two independent quantities: linear cell locomotion speed and directional persistence time (57Lauffenburger D.A Linderman J.J Receptors. Oxford University Press, New York1993Google Scholar). Molecular interventions can thus be usefully examined specifically in terms of changes in speed, persistence, or both. For instance, some alterations in integrin–cytoskeleton linkage influence migration speed but not persistence (80Schmidt C.E Chen T Lauffenburger D.A Modulation of cell migration speed via genetic manipulation of integrin/cytoskeleton linkage.J. Cell. Eng. 1995; 1: 3-12Google Scholar). Variations among cell types in speed and persistence may also reflect differences in their spatial and temporal coordination of the various mechanisms involved in locomotion, such as force generation and adhesion. Over a spectrum of cell types, speed and persistence under optimal conditions appear to be inversely related, with slower-moving cells exhibiting greater persistence (57Lauffenburger D.A Linderman J.J Receptors. Oxford University Press, New York1993Google Scholar), with the rapidly moving but highly persistent fish epidermal keratocyte serving as a notable exception. This relationship may reflect coordination between directional signaling and physical movement processes. Lamellipodia are broad, flat, sheet-like structures, whereas filopodia are thin, cylindrical, needle-like projections. Cytoplasmic organelles are excluded from these structures, which abundantly contain actin and actin-associated proteins. Both can extend reversibly into three dimensions around the cell, even when the cell is crawling on a two-dimensional substratum. Actual speeds of cell translocation are not strongly correlated with the velocity of membrane protrusive flow (15Condeelis J Life at the leading edge.Annu. Rev. Cell Biol. 1993; 9: 411-444Crossref PubMed Scopus (399) Google Scholar), but a possible relationship between cell migration speed and the frequency of membrane extensions has not yet been rigorously examined. Extension of both lamellipodia and filopodia in response to migratory stimuli is almost universally found coupled with local actin polymerization. Intervening details are complex and poorly understood (15Condeelis J Life at the leading edge.Annu. Rev. Cell Biol. 1993; 9: 411-444Crossref PubMed Scopus (399) Google Scholar, 85Stossel T.P On the crawling of animal cells.Science. 1993; 260: 1086-1094Crossref PubMed Scopus (905) Google Scholar). An increase in the number of sites for actin polymerization is a first step, followed by net addition of monomeric, G-actin monomers to these F-actin growth sites predominantly near the membrane, in spite of fast turnover due to depolymerization. New open barbed-end sites for actin polymerization may arise by a combination of mechanisms, including uncapping of already-existing filaments, their severing, or both, as well as de novo formation of new actin trimeric nucleation sites (87Sun H.-Q Kwiatkowska K Yin H.L Actin monomer binding proteins.Curr. Opin. Cell Biol. 1995; 7: 102-110Crossref PubMed Scopus (171) Google Scholar, 88Theriot J.A Regulation of the actin cytoskeleton in living cells.Semin. Cell Biol. 1994; 5: 193-199Crossref PubMed Scopus (41) Google Scholar). In neutrophils, there is an increase in the total number of cortical actin filaments following chemosensory stimulation, without significantly altering the distribution of filament lengths (9Cano M.L Lauffenburger D.A Zigmond S.H Kinetic analysis of F-actin depolymerization in polymorphonuclear leukocyte lysates indicates that chemoattractant stimulation increases actin filament number without altering the filament length distribution.J. Cell Biol. 1991; 115: 677-687Crossref PubMed Scopus (107) Google Scholar), implying either that most increased polymerization occurs from new nucleation sites or that severing and uncapping occur concomitantly in coordinated tandem. The gelsolin family is an attractive candidate for regulation of actin nucleation sites, because it regulates both severing and uncapping of actin filaments (85Stossel T.P On the crawling of animal cells.Science. 1993; 260: 1086-1094Crossref PubMed Scopus (905) Google Scholar). At micromolar and greater concentrations of calcium, and in the presence of low levels of the chemoattractant-induced phosphoinositides, the severing activity of gelsolin becomes significant, shortening filaments and increasing their number but leaving them capped. At less than micromolar concentrations of calcium, gelsolin dissociates from actin filaments, opening barbed ends for new polymerization. No discernable relationship has been found between calcium levels and membrane protrusion activity, however, so it is unclear whether these severing and uncapping activities are appropriately coordinated for promoting membrane extension (15Condeelis J Life at the leading edge.Annu. Rev. Cell Biol. 1993; 9: 411-444Crossref PubMed Scopus (399) Google Scholar). A modest correlation of gelsolin expression level with cell migration rate has been found in some studies (23Cunningham C Stossel T.P Kwiatkowski D Enhanced motility in NIH 3T3 fibroblasts that overexpress gelsolin.Science. 1991; 251: 1233-1236Crossref PubMed Scopus (259) Google Scholar, 98Witke W Sharpe A.H Hartwig J Azuma T Stossel T.P Kwiatkowski D Hemostatic, inflammatory, and fibroblast responses are blunted in mice lacking gelsolin.Cell. 1995; 81: 41-51Abstract Full Text PDF PubMed Scopus (376) Google Scholar) though not others (1Andre E Brink M Gerisch G Isenberg G Noegel A Schleicher M Segall J.E Wallraff E A Dictyostelium mutant deficient in severin, an F-actin fragmenting protein, shows normal motility and chemotaxis.J. Cell Biol. 1989; 108: 985-995Crossref PubMed Scopus (77) Google Scholar, 18Cooper J.A Bryan J Schwab B Frieden C Loftus D.J Elson E.L Microinjection of gelsolin into living cells.J. Cell Biol. 1987; 104: 491-501Crossref PubMed Scopus (99) Google Scholar). It is possible that gelsolin activity may be important for aspects of cell locomotion other than lamellipod/filopod extension. Uncapping of actin filament barbed ends would permit growth of existing filaments even in the absence of severing. Members of the calcium-independent capping protein family (76Schafer D.A Cooper J.A Control of actin assembly at filament ends.Annu. Rev. Cell Dev. Biol. 1995; 11: 497-518Crossref PubMed Scopus (172) Google Scholar), such as capping protein β2, appear to be the barbed-end regulator of predominant importance in neutrophils (28DiNubile M.J Cassimeris L Joyce M Zigmond S.H Actin filament barbed-end capping activity in neutrophil lysates the role of capping protein β-2.Mol. Biol. Cell. 1995; 6: 1659-1671Crossref PubMed Scopus (63) Google Scholar) and perhaps other cell types as well. The time constant for F-actin recapping by capping protein is roughly a few seconds, consistent with the window needed to account for new actin polymerization kinetics. For new polymerization arising from uncapped F-actin barbed ends, however, there would be an increase in the filament length distribution and not in the number of filaments, in contrast with the findings of 9Cano M.L Lauffenburger D.A Zigmond S.H Kinetic analysis of F-actin depolymerization in polymorphonuclear leukocyte lysates indicates that chemoattractant stimulation increases actin filament number without altering the filament length distribution.J. Cell Biol. 1991; 115: 677-687Crossref PubMed Scopus (107) Google Scholar. Although some nucleation activity by capping protein has been found in vitro, no such effect is noticeable in vivo, at least with Dictyostelium (42Hug C Jay P.Y Reddy I McNally J.G Bridgman P.C Elson E.L Cooper J.A Capping protein levels influence actin assembly and cell motility in Dictyostelium.Cell. 1995; 81: 591-600Abstract Full Text PDF PubMed Scopus (146) Google Scholar). As with gelsolin, a positive correlation of cell migration rate with capping protein expression level may result from an effect on actin cytoskeleton related to cell body translocation rather than membrane extension (42Hug C Jay P.Y Reddy I McNally J.G Bridgman P.C Elson E.L Cooper J.A Capping protein levels influence actin assembly and cell motility in Dictyostelium.Cell. 1995; 81: 591-600Abstract Full Text PDF PubMed Scopus (146) Google Scholar). Thus, the mechanism providing new actin polymerization sites for membrane extension in migrating cells remains unclear at present. Instead, regulation of local free G-actin levels may be a primary effector for membrane extension. Whatever the number of uncapped growth sites, the amount of F-actin could potentially be increased by raising the concentration of G-actin monomer, which exists in two pools: free G-actin and G-actin bound to a monomer-binding protein. Indeed, were there no additional source pool of G-actin besides that existing as free monomer, the increase in the amount of F-actin due to uncapping all extant filament barbed ends would be negligible (31Fechheimer M Zigmond S.H Focusing on unpolymerized actin.J. Cell Biol. 1993; 123: 1-5Crossref PubMed Scopus (83) Google Scholar). Three major families of cytoplasmic proteins that bind G-actin have been identified: β-thymosins, profilins, and ADFs/cofilins, each serving as a potential source of G-actin following release by migration stimuli (87Sun H.-Q Kwiatkowska K Yin H.L Actin monomer binding proteins.Curr. Opin. Cell Biol. 1995; 7: 102-110Crossref PubMed Scopus (171) Google Scholar, 88Theriot J.A Regulation of the actin cytoskeleton in living cells.Semin. Cell Biol. 1994; 5: 193-199Crossref PubMed Scopus (41) Google Scholar). However, the G-actin source effect of some of these proteins is negligible, while for others their function is more complex. Each of these families appears to have a distinct role in controlling F-actin levels: β-thymosins as a G-actin source, profilins as a filament elongation promotor, and ADFs/cofilins as a filament cutter (31Fechheimer M Zigmond S.H Focusing on unpolymerized actin.J. Cell Biol. 1993; 123: 1-5Crossref PubMed Scopus (83) Google Scholar, 87Sun H.-Q Kwiatkowska K Yin H.L Actin monomer binding proteins.Curr. Opin. Cell Biol. 1995; 7: 102-110Crossref PubMed Scopus (171) Google Scholar, 88Theriot J.A Regulation of the actin cytoskeleton in living cells.Semin. Cell Biol. 1994; 5: 193-199Crossref PubMed Scopus (41) Google Scholar). Theoretical analyses argue that local actin polymerization is in itself an adequate energy source for extension against the mechanical resistance provided by the cell membrane (15Condeelis J Life at the leading edge.Annu. Rev. Cell Biol. 1993; 9: 411-444Crossref PubMed Scopus (399) Google Scholar, 17Cooper J.A The role of actin polymerization in cell motility.Annu. Rev. Physiol. 1991; 53: 585-605Crossref PubMed Scopus (242) Google Scholar), and experiments demonstrate that deformation of lipid vesicles occurs following induction of actin polymerization within the vesicle interior (19Cortese J.D Schwab B Frieden C Elson E.L Actin polymerization induces a shape change in actin-containing vesicles.Proc. Natl. Acad. Sci. USA. 1989; 86: 5773-5777Crossref PubMed Scopus (111) Google Scholar). Moreover, a constant rate of new actin polymerization can lead to a constant rate of membrane extension, consistent with experimental observations, if viscous resistance by the membrane is not rate-limiting (32Felder S Elson E.L Mechanics of fibroblast locomotion quantitative analysis of forces and motions at the leading lamellas of fibroblasts.J. Cell Biol. 1990; 111: 2513-2526Crossref PubMed Scopus (93) Google Scholar). Experimental evidence so far seems to weigh against a necessity for cell body contraction in membrane extension (30Evans E Leung A Zhelev D Synchrony of cell spreading and contraction force as phagocytes engulf large pathogens.J. Cell Biol. 1993; 122: 1295-1300Crossref PubMed Scopus (74) Google Scholar, 103Zhelev D.V Hochmuth R.M Mechanically stimulated cytoskeleton rearrangement and cortical contraction in human neutrophils.Biophys. J. 1995; 68: 2004-2014Abstract Full Text PDF PubMed Scopus (44) Google Scholar), and perhaps for myosin motors more generally. Mutant cell lines defective in certain types of myosin molecules exhibit some defects in locomotion but not pseudopod extension (90Titus M.A Wessels D Spudich J.A Soll D The unconventional myosin encoded by the myoA gene plays a role in Dictyostelium motility.Mol. Biol. Cell. 1993; 4: 233-246Crossref PubMed Scopus (117) Google Scholar, 95Wessels D Soll D Knecht D Loomis W.F DeLozanne A Spudich J.A Cell motility and chemotaxis in amoebae lacking myosin heavy chain.Dev. Biol. 1988; 128: 164-177Crossref PubMed Scopus (251) Google Scholar96Wessels D Murray J Jung G Hammer III, J.A Soll D Myosin IB null mutants of Dictyostelium exhibit abnormalities in motility.Cell Motil. Cytoskel. 1991; 20: 301-315Crossref PubMed Scopus (113) Google Scholar). Possible redundancies with alternative myosin isoforms are difficult to rule out, however. Localization of myosin I in membrane protrusions (33Fukui Y Lynch T.J Brzeska H Korn E.D Myosin I is located at the leading edges of locomoting Dictyostelium amoeba.Nature. 1989; 341: 328-331Crossref PubMed Scopus (241) Google Scholar, 101Yonemura S Pollard T.D The localization of myosin I and myosin II in Acanthamoeba by fluorescence microscopy.J. Cell Sci. 1992; 102: 301-315Google Scholar) could play a role in other aspects of migration, such as directed transport of adhesion receptors to enhance formation of attachments (83Sheetz M.P Cell migration by graded attachment to substrates and contraction.Semin. Cell Biol. 1994; 5: 149-155Crossref PubMed Scopus (63) Google Scholar). The ability of actin polymerization to drive membrane extension requires that actin filaments possess or aquire appropriate mechanical properties. In filopodia, actin filaments are grouped into rope-like bundles, while in lamellipodia they are cross-linked into lattice-like meshwork. Filament-binding proteins have begun to be classified according to their structures and activities; they include the fimbrin/α-actinin/filamin, villin, scruin, and fascin families (63Matsudaira P Actin crosslinking proteins at the leading edge.Semin. Cell Biol. 1994; 5: 165-174Crossref PubMed Scopus (113) Google Scholar). Individual actin filaments can be bound by several different binding proteins simultaneously, permitting a diversity of organizational variations. Filament bundling and cross-linking both serve to increase the rigidity of the actin polymer network against the load of a membrane resisting deformation as a filopod or lamellipod attempts to extend. Thus, the activity of actin filament–binding proteins could be a key locus for regulation of membrane extension. Consistent with this view, ABP-120, a member of the filamin subfamily, is required for normal rates of lamellipod extension in Dictyostelium (21Cox D Condeelis J Wessels D Soll D Kern H Knecht D Targeted disruption of the ABP-120 gene leads to cells with altered motility.J. Cell Biol. 1992; 116: 943-955Crossref PubMed Scopus (94) Google Scholar), and similar findings have been obtained for a larger member of the same family, ABP-280, in melanoma cells (24Cunningham C Gorlin J Kwiatkowski D Hartwig J Janmey P Stossel T.P Requirement for actin binding protein for cortical stability and efficient locomotion.Science. 1992; 255: 325-327Crossref PubMed Scopus (498) Google Scholar). Importantly, subtle details of cross-linking structure may strongly affect membrane protrusion processes. For example, the number density of filaments in lamellipodia of cells lacking ABP-120 is at least as great as that in normal cells, but the spatial distribution is less regular and the interconnectedness is diminished (20Cox D Risdale J.A Condeelis J Hartwig J Genetic deletion of ABP-120 alters the three-dimensional organization of actin filaments in Dictyostelium pseudopods.J. Cell Biol. 1995; 128: 819-835Crossref PubMed Scopus (64) Google Scholar). The lower extension rate could thus be due to a difference in mechanical properties of the cytoskeletal network. Moreover, lamellipodia and filopodia each contain physically connected growing filaments, but their extension rates and geometries are strikingly disparate. Both types of processes can be observed growing simultaneously in a single cell at the same location (39Heidemann S.R Buxbaum R.E Growth cone motility.Curr. Opin. Neurobiol. 1991; 1: 339-345Crossref PubMed Scopus (18) Google Scholar), but they exhibit different distributions of actin-binding proteins yielding the different spatial structures (63Matsudaira P Actin crosslinking proteins at the leading edge.Semin. Cell Biol. 1994; 5: 165-174Crossref PubMed Scopus (113) Google Scholar). It is not clear whether the underlying physical growth mechanisms of lamellipodia and filopodia are identical, nor precisely what they are. Favored candidate mechanisms at this point in time, not mutually exclusive, are the Brownian ratchet (71Peskin C.S Odell G.M Oster G.F Cellular motions and thermal fluctuations the Brownian ratchet.Biophys. J. 1993; 65: 316-324Abstract Full Text PDF PubMed Scopus (723) Google Scholar) and cortical expansion (15Condeelis J Life at the leading edge.Annu. Rev. Cell Biol. 1993; 9: 411-444Crossref PubMed Scopus (399) Google Scholar) models. In the Brownian ratchet mechanism, actin monomers may be added to filaments proximal to the cell membrane when thermal fluctuations of the membrane position allow the requisite room. In the cortical expansion mechanism, the actin filament gel is proposed to swell from local influx of water either due to increased osmotic potential, possibly resulting from filament severing, or more generally due to an entropic driving force when filament–water interactions are energetically favorable. Relative contributions of the Brownian ratchet mechanism versus the corticol expansion mechanism could be favored in filopodia and lamellipodia, respectively, on the basis of their comparative structures of a highly oriented tight bundle versus a looser network mesh. Along with a bias for membrane extension at the cell front, there may also be a preferential ability of attachments to form at the leading edge of lamellipodia and filopodia. Several observations point to the cell front as a preferential locus where adhesions form. Interference reflection microscopy (IRM) images of migrating heart fibroblasts show new focal adhesions forming at the cell front and persisting until they reach the cell rear (44Izzard C.S Lochner L.R Formation of cell-to-substrate contacts during fibroblast motility an interference reflexion study.J. Cell Sci. 1980; 42: 81-116Crossref PubMed Google Scholar). Video tracking of integrins using non-adhesion-perturbing antibodies directed against β1 integrins also reveals the front as a site where new adhesions tend to form (72Regen C.M Horwitz A.F Dynamics of β1 integrin-mediated adhesive contacts in motile fibroblasts.J. Cell Biol. 1992; 119: 1347-1359Crossref PubMed Scopus (179) Google Scholar). Nascent adhesions appear in temporal waves, initially as small aggregates that trace the geometry of the leading lamella. These aggregates increase in size and intensity as the cell migrates over them, persisting and remaining fixed on the substratum until they reach the rear, or an edge, of the cell. While specific molecules that initiate or nucleate the formation of adhesive complexes have not been identified, some evidence points to the existence of a preformed cytoskeletal complex that precedes the incorporation of adhesion molecules. IRM studies, for example, demonstrate that development of actin filament stress fibers precedes the formation of focal adhesions (43Izzard C.S A precursor of the focal contact in cultured fibroblasts.Cell Motil. Cytoskel. 1988; 10: 137-142Crossref PubMed Scopus (70) Google Scholar). Cell–substratum attachments at the leading edge that subsequently remain fixed to the substratum as the cell moves forward effectively serve to remove adhesion molecules from the leading lamella. This implies existence of mechanisms to replenish such components at the cell front. Such a process has been demonstrated with tracking of gold aggregates conjugated to reagents directed against cell surface proteins. These studies demonstrate that membrane proteins, including integrins, are directed rapidly toward the cell periphery, including the leading edge, where they tend to remain (79Schmidt C.E Horwitz A.F Lauffenburger D.A Sheetz M.P Integrin–cytoskeletal interactions in migrating fibroblasts are dynamic, asymmetric, and regulated.J. Cell Biol. 1993; 123: 977-991Crossref PubMed Scopus (295) Google Scholar). Increased concentrations of other cytoskeletally associated components are also enriched in the leading lamella, although the mechanism of their recruitment is not known (66Nobes C.D Hall A Rho, rac, and cdc42 GTPases regulate the assembly of multimolecular focal complexes associated with actin stress fi