Microfilaments in cellular and developmental processes.

In our opinion, all of the phenomena that are inhibited by cytochalasin can be thought of as resulting from contractile activity of cellular organelles. Smooth muscle contraction, clot retraction, beat of heart cells, and shortening of the tadpole tail are all cases in which no argument of substance for alternative causes can be offered. The morphogenetic processes in epithelia, contractile ring function during cytokinesis, migration of cells on a substratum, and streaming in plant cells can be explained most simply on the basis of contractility being the causal event in each process. The many similarities between the latter cases and the former ones in which contraction is certain argue for that conclusion. For instance, platelets probably contract, possess a microfilament network, and behave like undulating membrane organelles. Migrating cells possess undulating membranes and contain a similar network. It is very likely, therefore, that their network is also contractile. In all of the cases that have been examined so far, microfilaments of some type are observed in the cells; furthermore, those filaments are at points where contractility could cause the respective phenomenon. The correlations from the cytochalasin experiments greatly strengthen the case; microfilaments are present in control and "recovered" cells and respective biological phenomena take place in such cells; microfilaments are absent or altered in treated cells and the phenomena do not occur. The evidence seems overwhelming that microfilaments are the contractile machinery of nonmuscle cells. The argument is further strengthened if we reconsider the list of processes insensitive to cytochalasin (Table 2). Microtubules and their sidearms, plasma membrane, or synthetic machinery of cells are presumed to be responsible for such processes, and colchicine, membrane-active drugs, or inhibitors of protein synthesis are effective at inhibiting the respective phenomena. These chemical agents would not necessarily be expected to affect contractile apparatuses over short periods of time, they either do not or only secondarily interfere with the processes sensitive to cytochalasin (Table 1). It is particularly noteworthy in this context that microtubules are classed as being insensitive to cytochalasin and so are not considered as members of the "contractile microfilament" family. The overall conclusion is that a broad spectrum of cellular and developmental processes are caused by contractile apparatuses that have at least the common feature of being sensitive to cytochalasin. Schroeder's important insight (3) has, then, led to the use of cytochalasin as a diagnostic tool for such contracile activity: the prediction is that sensitivity to the drug implies presence of some type of contractile microfilament system. Only further work will define the limits of confidence to be placed upon such diagnoses. The basis of contraction in microfilament systems is still hypothetical. Contraction of glycerol-extracted cells in response to adenosine triphosphate (53), extraction of actin-like or actomyosin-like proteins from cells other than muscle cells (54), and identification of activity resembling that of the actomyosin-adenosine triphosphatase system in a variety of nonmuscle tissues (40, 54) are consistent with the idea that portions of the complex, striated muscle contractile system may be present in more primitive contractile machinery. In the case of the egg cortex, calcium-activated contractions can be inhibited by cytochalasin. If, as seems likely, microfilaments are the agents activated by calcium, then it will be clear that they have the same calcium requirement as muscle. Biochemical analyses of primitive contractile systems are difficult to interpret. Ishikawa's important observation (31), that heavy meromyosin complexes with fine filaments oriented parallel to the surface of chondrocytes and perpendicular to the surface of intestinal epithelial cells, implies that both types of filaments are "actin-like" in this one respect. Yet, it is very likely that these actin-like filaments correspond respectively to the cytochalasin-insensitive sheath of glial and heart fibroblasts and the core filaments of oviduct microvilli. No evidence from our studies links contractility directly to these meromyosin-binding filaments. Apart from this problem, activity resembling that of the myosin-adenosine triphosphatase has been associated with the microtubule systems of sperm tails and cilia (55), but those organelles are insensitive to cytochalasin in structure and function. Clearly, a means must be found to distinguish between enzymatic activities associated with microfilament networks, microfilament bundles, microtubules, and the sheath filaments of migratory cells. Until such distinctions are possible, little of substance can be said about the molecular bases of primitive contractile systems. Three variables are important for the control of cellular processes dependent upon microfilaments: (i) which cells of a population shall manufacture and assemble the filaments; (ii) where filaments shall be assembled in cells; and (iii) when contractility shall occur. With respect to distribution among cells, the networks involved in cell locomotion are presumed to be present in all cells that have the potential to move in cell culture. In this respect, the networks can be regarded as a common cellular organelle in the sense that cytoplasmic microtubules are so regarded. In some developing systems, all cells of an epithelium possess microfilament bundles (7, 13), whereas, in others, only discrete subpopulations possess the bundles (5, 6). In these cases the filaments can be regarded as being differentiation products associated only with certain cell types. These considerations may be related to the fact that microfilament networks are associated with behavior of individual cells (such as migration, wound healing, and cytokinesis), whereas the bundles are present in cells that participate in coordinated changes in shape of cell populations. With respect to placement in cells, two alternatives are apparent, namely, localized or ubiquitous association with the plasma membrane. Microfilament bundles of epithelial cells are only found extending across the luminal and basal ends of cells. In this respect they contrast with desmosomal tonofilaments and with microtubules, each of which can curve in a variety of directions through the cell. The strict localization of microfilament bundles probably rests upon their association with special junctional complex insertion regions that are only located near the ends of cells. In the case of mitotically active cells, the orientation of the spindle apparatus may determine the site at which the contractile ring of microfilaments will form (4, 56); this raises the question of what sorts of cytoplasmic factors can influence the process of association between filament systems and plasma membranes. In contrast to such cases of localized distribution, contractile networks responsible for cell locomotion are probably found beneath all of the plasma membrane, just as the network of thrombosthenin may extend to all portions of the periphery of a blood platelet. This ubiquitous distribution probably accounts for the ability of a fibroblast or glial cell to establish an undulating membrane at any point on its edge, or of an axon to form lateral microspikes along its length. The third crucial aspect of control of these contractile apparatuses involves the choice of when contraction shall occur (and as a corollary the degree or strength of contraction that will occur). In the simplest situation, contraction would follow automatically upon assembly of the microfilament bundles or networks. In cleavage furrows of marine embryos (4), for instance, microfilaments are seen beneath the central cleavage furrow and at its ends, but not beyond, under the portion of plasma membrane that will subsequently become part of the furrow. This implies that the furrow forms very soon after the contractile filaments are assembled in the egg cortex. In other cases, microfilaments are apparently assembled but not in a state of (maximal?) contraction. Thus, networks are seen along the sides of migratory cells, although such regions are not then active as undulating membrane organelles. Similarly, microfilament bundles occur in all epithelial cells of the salivary gland (13), or pancreatic anlage (7), although only the ones at discrete points are thought to generate morphogenetic tissue movements. Likewise, bundles begin to appear as early as 12 hours after estrogen administration to oviduct, although visible tubular gland formation does not start until 24 to 30 hours. Finally, streaming in plant cells can wax and wane, depending upon external factors such as auxin (57). All of these cases imply a control mechanism other than mere assembly of the microfilament systems and even raise the possibility that within one cell some filaments may be contracting while others are not. In discussing this problem, it must be emphasized that different degrees of contraction or relaxation cannot as yet be recognized with the electron microscope. In fact, every one of the cases cited above could be explained by contraction following immediately upon some subtle sort of "assembly." Inclusive in the latter term are relations between individual filaments, relations of the filaments and their insertion points on plasma membrane, and quantitative alterations in filament systems. Furthermore, the critical role of calcium and high-energy compounds in muscle contraction suggest that equivalent factors may be part of primitive, cytochalasinsensitive systems. The finding that calcium-induced contraction in the cortex of eggs is sensitive to cytochalasin strengthens that supposition and emphasizes the importance of compartmentalization of cofactors as a means of controlling microfilaments in cells.