Cellular morphology is determined by motility, force sensing, and force generation that must be finely controlled in a dynamic fashion. Contractile and extensile functions are integrated with the overall cytoskeleton, including linkages from the cytoplasmic cytoskeleton to the extracellular matrix and other cells by force sensing. During development, as cells differentiate, variations in protein expression levels result in morphological changes. There are two major explanations for motile behavior: either cellular motility depends in a continuous fashion on cell composition or it exhibits phases wherein only a few protein modules are activated locally for a given time. Indeed, in support of the latter model, the quantification of cell spreading and other motile activities shows multiple distinct modes of behavior, which we term “phases” because there exist abrupt transitions between them. Cells in suspension have a basal level of motility that enables them to probe their immediate environment. After contacting a matrix-coated surface, they rapidly transition to an activated spreading phase. After the development of a significant contact area, the cells contract repeatedly to determine the rigidity of the substrate and then develop force on matrix contacts. When cells are fully spread, extension activity is significantly decreased and focal complexes start to assemble near the cell periphery. For each of these phases, there are significant differences in protein activities, which correspond to differences in function. Thus overall morphological change of a tissue is driven by chemical signals and force-dependent activation of one or more motile phases in limited cell regions for defined periods.
- extracellular matrix
- cellular processes
- dynamic phase transitions
cellular processes have evolved over many millions of years to be robust in the face of a variety of environmental challenges. Cells can rapidly adapt to perform important processes in different matrix environments, temperatures, osmolarities, nutrient environments, and levels. Engineering principles for a machine with robust function include modular design and compartmentalization (8, 15). Modular design involves breaking the overall function of a machine into subfunctions. Each subfunction is performed by a functional module that is controlled by or signals to other modules. For example, the simple act of a cell in suspension adhering to a substrate involves the overt steps of extending a series of actin-based filopodia to the surface (filopodial sensing of the surface) resulting in signaling to the cytoskeleton to spread onto the surface (3). In this example, one subfunction would be applying force to the membrane for extension and that functional module would be the actin cytoskeleton polymerization machinery. In addition, there is a periodic sensing of force at the periphery by receptor-like protein tyrosine phosphatase (RPTP)-α that is important for extensive cell spreading (16). To understand the complete process, we must build a list of the required activities and the functional modules needed to complete them.
There are important differences in strategy between studies of protein expression profiles and protein-protein interactions versus the studies of actual behavior of the organisms (9). The profile of the proteins expressed in a cell and the known protein-protein interactions can be said to have emergent properties (i.e., cell motility) resulting from the complex set of interactions (18). On the other hand, observations of the behavior of cells can be dissected into functional pathways involving key proteins or protein groups that contribute specific functions to the overall behavior. The first approach is useful when little or no a priori knowledge of the system is available, because the belief is that the overall behavior is a function of all components (7). Ultimate understanding is dependent on following outputs for all of the physiological inputs of cells with a given genotype. In contrast is the postulate that the cell has a limited number of functional modules and different behaviors result from controlled regional and temporal activation of those modules (8). For (actin based) cell motility, there is considerable knowledge of the motile machinery even at the protein level. The location and function of a large number of proteins is known in detail. Their structural role and/or position in the various cytoskeletal, membrane, or signaling pathways has been elucidated (1, 11, 12). Nevertheless, even in this case, one typically has to rely on a coarse-grained description to understand the overall motile behavior. Both approaches should converge on a unified description of cellular behavior, but the grouping of functions in modules has a benefit in providing a simple means of classifying behaviors for interim analyses.
In a series of recent papers, we identified and quantitatively characterized a number of phases of motile behavior (2, 3, 6). We will review these results and proceed to a definition of protein modules based on a functional analysis of these phases of motility. As an organizing principle, we propose three hierarchical classes of interacting functional modules. These are structural and motor proteins, local regulatory proteins, and proteins involved in general signaling pathways. Quantification of cell motility behavior via well-defined phases with phase-specific protein activity establishes a direct link between cellular phenotype and genotype.
FUNCTIONAL PHASES OF CELL SPREADING
There are several functional requirements for a cell to adhere, spread, and move on a given substrate. First, the cell has to test whether the extracellular matrix is chemically and mechanically suitable for adhesion (16). In case of favorable interactions with the substrate, it establishes a large contact area in a second step (Fig. 1B; phase 1). Third, the cell then probes the elasticity of the extracellular matrix by periodically applying force to the substrate (Fig. 1,C and B; phase 2). On sufficiently stiff substrates, it continues spreading, approaching its maximal area of contact. Finally, the cell moves forward in a particular direction after polarization is triggered by either internal signals or external chemical gradients (corresponding to phase 3 in Fig. 1B). This scenario is rather general and applies to fibroblasts, endothelial cells, and presumably in a similar form to even neuronal growth cones. The stages of cell motility as seen in our two-dimensional (2-D) spreading assay can be considered phases in a quite rigorous sense, because there are well-defined dynamic phase transitions between them (Fig. 2; Ref. 2). The different functional roles that the motile cell phases play are reflected in different protein localization and activity patterns associated with these functions (6).
Motility phases involve a characteristic subset of functions that are organized in a specific spatial and temporal order. They involve distinct sets of protein modules. Distinct does not mean that two functionally different modules are necessarily composed of totally disjunct sets of individual proteins. In general, there still may be common protein molecules. One way to describe motility phases is to say that they can be compared with different gears in a car. A different set of automobile components are used in reverse and first gears, including backup lights, nevertheless there are common single components in the gear box. The transition between the first two phases of cell spreading is rapid and similar to shifting gears. The details of these processes are important for a complete understanding.
In early spreading, the cell goes from a rough sphere to a thick disk on a 2-D surface with about the same cross-sectional area as the sphere. Actin filaments throughout the cell cortex are depolymerized, and the g-actin monomers are assembled into new filaments at the 2-D surface by complexes that receive a signal from the liganded integrins. Necessary steps include 1) local sensing of the matrix coating of the surface, which activates a) general breakdown of cortical actin filaments and cortical structure and b) local assembly of actin filaments at matrix-coated surfaces; 2) continued actin filament assembly that depends on new binding of surface integrins to new regions of the surface; and 3) slow rearward transport of the newly assembled actin filaments (see Fig. 1 from Ref. 6). In the first step, there is a threshold to the activation that is a function of both fibronectin density and time. Higher fibronectin concentrations decrease the lag time before spreading instead of increasing the rate of spreading (3). Principal components involved in setting the rate of early spreading include the Src family tyrosine kinases and tyrosine phosphatases.
On a practical level, the functional modules involved in this phase of cell spreading are worth considering. The spreading phase is activated by a cell-wide process that induces disassembly of filaments generally and spreading locally. The Src family kinases appear to be part of the functional protein module that activates this phase (4). Cytoskeleton remodeling may be catalyzed by actin filament severing through gelsolin or other proteins like motors. Filament disassembly by cofilin is necessary to generate g-actin. Cell edge extension requires the assembly of actin filaments from g-actin. During this phase, actin assembly in isotropic early spreading has a vertical component that keeps the growing edge on the surface, where there may be further activation. This phase will automatically stop when either the cell reaches a critical area or receives another dominant signal.
CONTRACTILE PHASE OF SPREADING
In the contractile phase of cell spreading, there is a dramatic increase in the rate of rearward actin movement. This phase is required for continued cell spreading, because inhibition of myosin or myosin light chain kinase (MLCK) blocks reward actin movement and contraction. In addition, further spreading requires rigid substrates. All of these factors indicate that force generation on the substrate is necessary for spreading. In cells with stress fibers, the generation of force on the substrate can be easily understood. However, at these early times of spreading, there are no stress fibers. Thus it is difficult to understand how the cytoskeleton is organized to support force generation from one side of the cell to the other. One possibility is that there is a central framework of filaments around the internal domain of cytoplasm or endoplasm that is linked to myosin. Myosin pulling on the cortical actin would then create a tension in the framework of filaments around the endoplasm, and force could then be transmitted from one side of the cell to the other. The tension that the cell creates is clearly due to myosin, and the major question is how the myosin is organized to enable it to generate force to move actin inward and ultimately to generate force on the surrounding matrix.
RIGIDITY SENSING AND RADIAL TRANSPORT
Rigidity sensing is a major aspect of the contractile phase, and fibroblasts need a rigid substrate to spread fully (10). Neurons on the other hand prefer a softer substrate and seem to grow better when the material is less rigid (5). Two major models of rigidity sensing can be postulated (Fig. 3). Because rigid surfaces require higher forces for the same degree of deformation than soft, cells could measure the amount of movement for a given force. Alternatively, cells could measure the rate of rise in force, because the rate of actin movement inward in the contractile phase is relatively constant. Experiments are underway to test between these possibilities. Another important aspect of rigidity sensing is that fibroblasts orient and move toward a rigid substrate (17). For orientation, there must be a position-dependent component to the rigidity sensing. How that is conveyed to the cell body is suggested by recent findings of a radial transport of a contractile signal from the periphery to a more centrally located myosin (6).
GLOBAL PHASES AND PROTEIN MODULES
Remarkably, one finds increasing evidence for universal global phases in various cell lines and species with different genotypic and phenotypic background. Indeed, the evolutionary pressure to survive preserves function but not necessarily the associated protein modules. Thus one may have the same functional phase with different instances of realization by proteins in different tissues or species. To illustrate this with a nonbiological example, let us consider a vehicle with a motor, a gear box, and a steering wheel. The functions associated to these modules are velocity, forward/backward movement, and direction. These are common to all vehicles. However, the instances of realization in a tractor, limousine, or racing car, respectively, are quite different.
In general, the classification of the motile machinery into different protein modules is quite ambitious when striving for completeness. However, an enumeration of key elements already allows us to establish a minimal quantitative model for motility. We propose to define basic functional modules, which interact to fulfill the overall function of a given phase. Furthermore, these basic modules can be hierarchically ordered into structural elements that are the phase building blocks and regulating factors that directly affect coupling strength and activity of structural proteins. In turn, the regulatory proteins are controlled by a signaling network coordinating spatially distant and/or logically separate functional events in the cell. We would assign a particular protein to the signaling network if it does not directly interact with structural proteins. Finally, the signaling network itself consists of logical modules integrating various upstream regulators and downstream effectors. A hierarchical structure of modules does not exclude parallel signal processing or even partial overlap of functional modules. A regulatory protein of a local structural module can well be part of a signaling cascade or be considered a phase-building element in a larger subsystem that includes the local module. We note that we are not simply proposing a new terminology, e.g., exchanging “mechanisms” and “pathways” with “phases” and “regulating modules.” Instead, we have embarked on a general program establishing the existence of well-defined transitions between motility phases, including a characterization of the regulating protein modules driving these phase transitions.
A nonexhaustive list of basic functional modules of cell motility consists of the actomyosin gel, the microtubule scaffold, the plasma membrane, as well as associated proteins. The phase state of the structural network, formed by the actin cytoskeleton (with its various linking proteins) and the myosin motors, is regulated, for example, by polymerizing (Arp2/3) and depolymerizing (ADF/cofilin) proteins, as well as proteins controlling motor activity (MLCK). The membrane serves as a complex for the relay of extracellular signals and transmits forces from the extracellular matrix to the cytoskeleton via integrins. Finally, various G proteins, like Rac and Rho, form part of the signaling network affecting the regulatory modules identified as driving directly specific phase transitions.
PHASE TRANSITIONS IN SPREADING
We established clear signatures of at least two different dynamic phase transitions in the spreading behavior of mouse embryonic fibroblast cells (Fig. 2). These are 1) the initiation of fast continuous spreading after a period of basal activity, and, subsequently, 2) the start of periodic membrane retractions. How are these transitions controlled by the cell?
The onset of continuous spreading is characterized by an increase in the actin polymerization velocity at the leading edge of the lamellipodium, pushing the membrane forward. Increased polymerization is triggered by favorable contact with the extracellular matrix. We found that the time from contact until initiation of spreading decreases with fibronectin density (3), suggesting an integration of a chemical signal from integrin receptors binding to fibronectin. This signal then leads to activation of the polymerization module, which drives actin polymerization at the surface (11).
The transition to the periodic contractile phase is linked to the activity of MLCK, a protein controlling in turn the activity of myosin motors. Indeed, we found that periodic membrane retractions were absent when inhibiting MLCK. The conjecture is that the actin network contracts or is actively pulled back by myosin motor activity (6). However, we have no direct evidence yet for the involvement of myosin.
HIERARCHICAL PHASE MODEL AND MOTILITY PHASE DIAGRAM
The motile phases and phase transitions in cell spreading have been used as an example of a more general mechanism of cellular function through phases. A way to describe the control of motile phases and phase transitions is through integration of signals from integrins, force sensors, cell cycle, and hormones. In such a scheme, select components of the actomyosin cytoskeleton will be activated to start the phase. In the contractile phase, MLCK, Arp 2/3, and cofilin will be engaged in a temporal pattern of cell edge extension and retraction that belies a finely organized clock (6). Different dynamic phases correspond to different functional regions in the regulating parameter space. When all relevant regulatory parameters of a certain cellular subsystem are known, the phase is defined. This is independent of possible states of the signaling network corresponding to this set of regulatory parameters (Table 1). It is the sequence of transitions, i.e., the trajectories in this parameter space, which is determined by the cellular signaling network. To put it differently, the topology of the motility phase diagram is independent of the complex signaling network, i.e., the relative positions of all the motility phases do not depend on the trajectories in parameter space followed by the cell. This means the basic phase characteristics of the actomyosin cytoskeleton can be modeled separately, which is an important conceptual advantage. It is important to realize, however, that a given topological phase structure of possible cellular states does not mean that there is no trajectory hysteresis or memory of past events. Indeed, a particular future phenotypic trajectory will in general depend on its past course due to control by the signaling network.
For example, the lamellipodial clock in the contractile phase is turned on by one set of signals and must receive an additional set to continue. The machinery receives signals from a rigidity sensor “S1,” from integrin binding to ligand “S2,” and from the apparent tension of the plasma membrane (14) “S3.” These signals contribute to the continuation of the periodic contraction phase. S1 and S2 are positive, whereas a high apparent tension, S3, would inhibit further extension. A table of the conditional behavior of the contractile phase exemplifies the conditions that must be met for the phase to continue (Table 1). This is a rather trivial example due to the existence of only one activating state. However, different signaling pathways, both leading to activation of the same phase, would have a set of nondegenerate activating states.
CONCLUSION AND APPLICATION
Force signals are similar to other environmental signals such as matrix binding or hormones, and they all contribute to the control of the specific phase behavior of the cell. A limited number of motile phases can be activated, and the extent of each phase in time and space serves to create the shape of the cell. For the future, it will be important to define the molecular mechanisms as well as the higher level control processes. At the molecular level, we believe that both the sensing of small forces on integrins and large forces on the whole cytoskeleton are critical factors in force sensation that are affected by different mechanisms. In the first case, the extended shape of the integrins suggests that they can be sensitive to forces that would alter their conformation and binding partners (16). In the second case, we suggest that the unfolding of proteins can activate them to bind and become substrates for activation (13). By continuing to define the roles of specific molecules in specific steps of the different motility phases, we believe that functional modules in actin-based motility can be elucidated. As these modules emerge, the task of understanding the process of motility as a whole will be greatly simplified. At the whole cell level, the phases are under the control of many inputs including the signals from forces, but each phase itself depends on a limited number of protein components such that it can be used in a number of phenotypic backgrounds. Thus the critical functional modules can be considered as separate machines that are controlled by several signals that all contribute to activating them or keeping them active.
- Copyright © 2005 the American Physiological Society