Polarization of pseudopodial activities: cytoskeletal mechanisms

Polarization of pseudopodial activities can be defined as a more or less stable division of the cell surface into pseudopodium-extending active zones and non-active zones. Polarization is the basis of numerous phenomena involved in morphogenesis and cell motility; the spectrum of these phenomena ranges from the crawling of amoebae to axon elongation. Direction of polarization is often determined by external signals such as chemotactic gradients in the liquid medium, geometrical and chemical heterogeneities of the substratum surface, cell-cell contacts etc. These external signals that apparently act via cell membrane receptors may induce pseudopodial reactions; that is, extension, attachment and contraction of pseudopodia. Local cytoskeletal reorganization, and, more specifically, local alteration of the actin-myosin cortex are essential components of individual pseudopodial reactions. The most typical of these is the formation of the network of actin microfilaments. Global reorganization of cytoskeleton usually follows series of local pseudopodial reactions. Their function is to ‘memorize’ the results of previous pseudopodial reactions and to modulate the sensitivity of various cell surface sites to future pseudopodium-inducing signals. This reorganization may facilitate the extension of pseudopodia from the sites where previous reactions were successful; that is, where the previously extended pseudopodia were able to attach themselves to the substratum. At the same time, global reorganization may completely stop pseudopodial activity in the ‘unsuccessful’ zones of the cell surface. In other words, this reorganization stabilizes certain distributions of active and non-active zones of the surface originally induced by external signals (see reviews by Vasiliev and Gelfand, 1977; Vasiliev, 1982).

Most facts concerning the nature of this global reorganization of the cytoskeleton that stabilizes polarization have been obtained in experiments with two cell types: amoeboid cells of the myxomycete Dictyostelium discoid-eum and cultured mammalian fibroblasts. Here I will compare the results obtained with these two very different cells.

Pseudopodia in amoebae can be induced by specific ligands, e.g. by chemotactic substances (see review by Spudich, 1989). A characteristic feature of the network of microfilaments within pseudopodia is the absence of conventional myosin, so-called myosin II. Therefore, local pseudopodial activity can lead to development of global asymmetry of actin-myosin distribution: most of the conventional myosin of amoebae undergoing chemotaxis is located in the posterior region, where there is very little pseudopodial activity.

Spudich and collaborators (see review by Spudich, 1989) have shown that global asymmetry of myosin II distribution is essential for polarization of amoeboid cells. In these experiments the cloned gene of the heavy chain of conventional myosin was used to disrupt the cellular myosin gene by homologous recombination. In this way mutant amoebae depleted of conventional myosin were obtained. These myosin-less cells had reduced cell polarity and tended to extend pseudopodia over most of their surface. Owing to the loss of polarity, chemotactic movement of these mutant cells was about five times less efficient than that of wild-type cells. On the basis of these experiments, Spudich (1989) developed a concept of the polarization mechanism based on asymmetrical relaxation-contraction of the cortex. According to this concept, an unstimulated cell has a contractile cortical shell of actin-myosin network; this network prevents the extension of pseudopodia. An external signal, e.g. a chemoattractant, acting on the membrane of one cell pole induces local disassembly of this network, removing the barrier for pseudopodial extension. The actin-myosin network in the posterior part of the cell remains contracted and the pseudopodial activity remains inhibited.

The concept of Spudich’s is a very important development of an older and more general concept of ‘cortical flow’ (see reviews by Dunn, 1980; Bray and White, 1988; Giuliano and Taylor, 1990). This concept postulates that actin-myosin material in polarized cells continuously circulates between the pseudopodium-extending active edge and more central parts of the cortex: the pseudopodial network of microfilaments acquires contractility and moves centripetally, being incorporated on its way into the contractile cortical structures of the central and posterior parts of the cell. Simultaneously, gradual dismantlement of previously formed contractile structures takes place in the posterior and central zones of the cortex. This dismantlement provides transportable material (nonpolymerized actin and other proteins? fragmented actin microfilaments?) that is transferred back to the active edge and used there for the formation of networks. The results of Spudich and collaborators discussed above suggest that cyclic actin-myosin flow is essential and sufficient for polarization of Dictyostelium.

The detailed morphology of the actin-myosin cortex in Dictyostelium and in fibroblasts is in many ways different, e.g. fibroblast cortex has certain differentiated structures that are absent in amoebae, such as stress fibers. However, at least one feature is common to the cortices of both cells: fibroblasts, like amoebae, have an asymmetric distribution of conventional myosin, with this protein being depleted from distal parts of active lamellae. Dynamic observations have shown that actin is concentrated in lamellar protrusions at the active edges of migrating fibroblasts from the moment of their extension, while myosin is initially absent and penetrates into protrusions several minutes later (Conrad et al. 1989). Accumulation of conventional myosin may be a prerequisite for the development of the contractility of extended pseudopodia and of their centripetal movement. There are no data showing directly the essential role of this cortical asymmetry in the maintenance of polarity in fibroblasts. We can only suggest, by analogy with amoebae, that this role is significant. At the same time experiments with fibroblasts have revealed the dependence of polarity on other types of cytoskeletal structures, on microtubules and, possibly, also on intermediate filaments. The data proving this dependence were obtained in experiments with drugs that specifically affect the microtubular system. Additional results that are important for analysis of polarization were obtained in experiments with an activator of protein kinase C, 12-O-tetradecanoylphorbol-13-acetate (TPA). I will briefly discuss the data provided by both these sets of experiments.

It was shown long ago (see reviews by Vasiliev and Gelfand, 1977; Vasiliev, 1982) that various drugs selectively depolymerizing the microtubular system, e.g. colcemid or colchicine, inhibit directional movement of fibroblasts; in cells treated with these drugs all parts of the cell edge become active. These cells also become much less elongated. Depolymerization of microtubules induced by the drugs is always accompanied by the perinuclear collapse of vimentin-containing intermediate filaments. Recent experiments have revealed additional important characteristics of the effects of microtubule-depolymerizing drugs.

1. The intensity of pseudopodial activity at the leading active edge of normal migrating cells is much higher than that at the active perimeter of colcemid-treated cells (A. D. Bershadsky, E. A. Weisberg and J. M. Vasiliev, unpublished). Thus, microtubules are needed not only for restriction of pseudopodial activity to one particular part of the cell edge but also for the development of maximal activity at this edge. The microtubular system is essential for efficient differentiation of pseudopodial activity between various parts of the cell edges.

2. Drug-induced depolymerization of microtubules is accompanied by a considerable increase in contraction of the actin cortex as evidenced by the rapid contraction of the elongated cell bodies after addition of colcemid (I. D. Karavanova and J. M. Vasiliev, unpublished), an increase in the numbers of stress fibers and an increase in the ability of cells that are spread on elastic substrata to wrinkle these substrata (Danowski and Harris, 1988; Danowski, 1988). These results suggest that microtubules and/or intermediate filaments somehow decrease the contractility of the cortex, thereby facilitating cell elongation and extension of pseudopodia at the active edge.

The effect of TPA on the shape of certain lines of fibroblastic cells is in many ways antagonistic to that of microtubule-depolymerizing drugs; this agent increases the segregation of the cell periphery into two types of domains: actin-rich lamellae with active cell edges and thin stalk-like cytoplasmic processes with stable edges (see review by Vasiliev and Gelfand, 1988). These processes have abundant microtubules and intermediate filaments but a poorly developed cortex. Dynamic observations show that stalk-like processes are formed from lamellae; this reorganization probably involves dismantling of actin microfilament networks within the lamellae and the cell body, and is accompanied by a general decrease in cortex contractility. Analysis of the effects of TPA in the presence of colcemid has shown that intermediate filaments are probably essential for the formation of stalk-like processes, while microtubules affect this reorganization indirectly by controlling the positions of intermediate filaments (Bershadsky et al. 1990).

These two groups of data suggest that polarization of fibroblasts is controlled by interactions between microtubules, intermediate filaments and cortex; microtubule-depolymerizing drugs and TPA alter polarization by shifting these interactions in opposite directions. The following hypothetical scheme (see also Fig. 1) describes possible mechanisms for these interactions in more detail:

1. External molecules, via signal-processing reactions, induce extension of lamellar pseudopodia at some parts of the cell edge; the flow of cortical material is directed toward this edge; pseudopodial formation remains inhibited in the contractile parts of the cortex proximal to the active edges and rich in conventional myosin. At this stage the mechanisms of polarization are essentially similar to those postulated for amoebae (see above).

2. Microtubules and, later, intermediate filaments grow into the extended lamella towards its active edge. I suggest that microtubules and intermediate filaments activate the ‘outward’ components of cortical flow in the cytoplasm around them; that, more specifically, they activate the dismantling of proximal contractile structures and transfer of cortex material to the anterior edge of lamellae. This activation leads to enhancement of pseudopodial extension at the active edge. At the same time the enhanced removal of actin-myosin material from the proximal cortex decreases the contractility of this cortex and facilitates its stretching and elongation. In this way the relative and absolute length of the stable cortical zones, that is, the degree of polarization, is increased. I will not discuss here possible molecular mechanisms for the postulated activation of the outward cortical flow by microtubules and intermediate filaments. They may act as transport structures needed for efficient transfer of cortex components from the central parts to the active edge. They may also transport some membraneous organelles needed for cortex reorganization. Microtubules and, intermediate filaments especially, may also stretch the microfilamental network mechanically. We know so little about these processes that it would be premature to make any detailed suggestions. The relative roles of microtubules and intermediate filaments in these reorganizations present another unsolved problem.

3. Signal processing reactions and, in particular, stimulation of protein kinase C may enhance further the outward cortical flow along microtubules and intermediate filaments. This hyperactivation may lead to the extension of large lamellae, followed by the resorption of proximal parts of these lamellae, and to their transformation into stalk-like processes; these processes retain only the thin stretched cortical layer. In other words, segregation of lamellar and stalk-like domains by TPA may be an example of overactivation of the same mechanism that controls microtubule-dependent polarization.

Thus, fibroblasts seem to have at least two levels of global cytoskeletal reorganization leading to polarization: a basic level of reorganization of the actin-myosin cortex and a higher level of reorganization of microtubules and intermediate filaments controlling the cortex. At the basic level, stable surface zones are formed due to contraction of the adjoining cortex. At the higher level, the length of these zones is increased due to partial dismantling and relaxation of the contracted cortex.

This concept is in good agreement with the findings showing that microtubule-dependent drugs do not always abolish all polarization of pseudopodial activities in fibroblasts: although the long stretches of stable edge zones disappear, the cell may retain relatively short stable zones interrupted by active edge zones. This short-range ‘microtubule-independent stabilization’ (for reviews see Vasiliev, 1982; Middleton et al. 1989) was observed in various experimental situations: in small cytoplasmic fragments cut from fibroblasts or from epitheliocytes; in primary cultures of chick embryo fibroblasts, where the cells are less elongated than in standard secondary cultures; in certain types of epithelial cultures. It seems probable that microtubule-independent stabilization may be based on the stretching of certain zones of the actin-myosin cortex; experimental evidence for this stretching was obtained by Kolega (1986). A possible role for intermediate filaments in this stabilization also needs investigation.

Dictyostelium amoebae apparently represent a cell type in which all polarization is determined by the basic-level mechanism alone. Some other cells, morphologically similar to Dictyostelium amoebae, e.g. polynuclear leucocytes of vertebrates, probably have the same type of polarization. However, one should stress that the absence of any role for microtubules and intermediate filaments in the polarization of amoebae and leucocytes remains to be proven. Fibroblasts represent cells in which both levels of polarization mechanism are usually active; some other tissue cells, e.g. neurons, possibly have the same type of polarization. Differences in control mechanisms may be associated with differences in morphology and behavior: amoeboid cells with one-level controls can change the direction of their movement much more quickly than cells with two-level polarization control; at the same time the former cells cannot acquire very elongated shapes. Different variants of polarization of the same cell, e.g. a fibroblast, can be achieved by mechanisms involving or not involving certain cytoskeletal components, especially, microtubules. This redundancy of interacting components seems to be characteristic of many phenomena involving the cytoskeleton (Bray and Vasiliev, 1989).