Skip to main content
PLOS ONE logoLink to PLOS ONE
. 2010 Aug 18;5(8):e12043. doi: 10.1371/journal.pone.0012043

The Non-Equilibrium Thermodynamics and Kinetics of Focal Adhesion Dynamics

Joseph E Olberding 1, Michael D Thouless 2, Ellen M Arruda 3, Krishna Garikipati 4,*
Editor: Markus J Buehler5
PMCID: PMC2923603  PMID: 20805876

Abstract

Background

We consider a focal adhesion to be made up of molecular complexes, each consisting of a ligand, an integrin molecule, and associated plaque proteins. Free energy changes drive the binding and unbinding of these complexes and thereby controls the focal adhesion's dynamic modes of growth, treadmilling and resorption.

Principal Findings

We have identified a competition among four thermodynamic driving forces for focal adhesion dynamics: (i) the work done during the addition of a single molecular complex of a certain size, (ii) the chemical free energy change associated with the addition of a molecular complex, (iii) the elastic free energy change associated with deformation of focal adhesions and the cell membrane, and (iv) the work done on a molecular conformational change. We have developed a theoretical treatment of focal adhesion dynamics as a nonlinear rate process governed by a classical kinetic model. We also express the rates as being driven by out-of-equilibrium thermodynamic driving forces, and modulated by kinetics. The mechanisms governed by the above four effects allow focal adhesions to exhibit a rich variety of behavior without the need to introduce special constitutive assumptions for their response. For the reaction-limited case growth, treadmilling and resorption are all predicted by a very simple chemo-mechanical model. Treadmilling requires symmetry breaking between the ends of the focal adhesion, and is achieved by driving force (i) above. In contrast, depending on its numerical value (ii) causes symmetric growth, resorption or is neutral, (iii) causes symmetric resorption, and (iv) causes symmetric growth. These findings hold for a range of conditions: temporally-constant force or stress, and for spatially-uniform and non-uniform stress distribution over the FA. The symmetric growth mode dominates for temporally-constant stress, with a reduced treadmilling regime.

Significance

In addition to explaining focal adhesion dynamics, this treatment can be coupled with models of cytoskeleton dynamics and contribute to the understanding of cell motility.

Introduction

A focal adhesion (FA) is a type of cell-substrate attachment mediated by bonds between the transmembrane protein, integrin, and an extracellular matrix (ECM) protein such as fibronectin. In this communication, we consider FAs in fibroblasts, although they are observed in other mesenchymal cell types as well. The integrins in FAs are believed to associate with over 50 cytoplasmic “plaque proteins” [1] among which are tensin, paxillin [2], [3], vinculin [2][4], talin and zyxin [3]. Many of these proteins have been detected in fluorescence studies of FAs. Intracellular attachments are formed to actin stress fibers [4] by the association of vinculin and talin to the FAs, followed by the binding of these proteins to F-actin [5][7]. It has been demonstrated by numerous experiments that force can be transmitted to FAs by the actin stress fibers. The force can be generated either by actomyosin contractility, or by an external manipulation such as with a micropipette [2], [4], [8]. The FAs transmit this force to the ECM. FA growth and resorption is strongly dependent on this force, as shown by a number of studies [2], [4], [9], [10].

In fibroblasts, a rich dynamic behavior of FAs is obtained when they are subject to force [11]. The FAs grow into elongated structures at a rate of Inline graphic [2]. They tend to be elongated in the cell-substrate interfacial plane with the long axis aligned with the force component in this plane. In the case of cell-generated tension the FA area is proportional to the force on it with a stress Inline graphic kPa [4].

FA dynamics have been visualized via fluorescence in cells that are contracting or loaded by external force [8], [12]. These studies show unbinding of proteins at the ends that are distally-located with respect to the stress fiber or point of application of external force. The associated “peeling” has been studied by mounting cells and synthetically-constituted vesicles on substrates, and modelled by applying methods of fracture mechanics extended to reaction-diffusion systems [13], [14]. However, the aspect of FA dynamics that has attracted the attention of experimental cell biologists and theoretical biophysicists alike is the force-mediated growth of the proximal end by the binding of proteins—an observation that is surprising because it runs counter to intuition gained from fracture processes such as tape peeling. The relative velocities of the proximal and distal ends combine to create different regimes of FA dynamics: growth at both ends, treadmilling that consists of growth at one end due to binding and resorption at the other end due to unbinding [2], [8], and resorption at both ends. The directions of growth, treadmilling and resorption are aligned with the force [2], [8], [12]. Treadmilling is directed toward the attached stress fiber when induced by actomyosin contractility, and is directed along the external force otherwise. Under such force-mediated growth, “focal complexes”, typically observed at sizes Inline graphicm [12], mature into larger structures recognized as FAs.

Fluorescence studies reveal the binding and unbinding of labelled proteins, but do not explain the biophysics of FA growth under force. Some experimental papers have aimed to explain this growth by hypothesizing molecular mechanisms such as the exposure of cryptic self-association sites on fibronectin by the applied tensile force that is transmitted to fibronectin after formation of integrin-fibronectin bonds [6]. This hypothesis suggests that tension causes a conformational change in the fibronectin molecule, exposing a previously cryptic site, and enabling polymerization of the ECM's fibronectin network. In turn, this allows more integrins to bind to the fibronectin network, and promotes FA growth. A tension-induced conformational change also has been hypothesized for vinculin activation during assembly of FA plaque proteins [1].

A few theoretical studies [8], [15][19] have sought to explain the biophysics of FA growth based on the above experimental observations. Of interest for the general mathematical treatment of adhesion of cellular structures driven by strain energy and chemistry is the work of Freund and Lin [20]. More recent studies have considered the strength of receptor-ligand binding in a statistical mechanics setting [21], the clustering of receptor-ligand bonds via a stability analysis [22], and the evolution of a bond over a free energy landscape under influence of a force [23]. A recent molecular dynamics study [7] has attempted to shed light on the force-induced conformational change of talin that enables its binding with vinculin.

In this work we treat FA dynamics as a rate process governed by Harmonic Transition State Theory [24]. Starting from this basis, we write the rates as being driven by out-of-equilibrium thermodynamic driving forces, and modulated by kinetics. We also make a connection to classical non-equilibrium thermodynamics with linear response theory as laid out by de Groot and Mazur [25]. We consider a number of mechanisms, both chemical and mechanical, that affect the out-of-equilibrium thermodynamic driving forces. We frame the discussion in terms of the symmetry, with respect to distal and proximal ends, that each mechanism imposes on the thermodynamic driving forces. Our central finding is that one contribution to the work done by actin stress fiber-transmitted force is antisymmetric and enables the treadmilling mode of FA dynamics, and that once the symmetry is broken, it can be further skewed by a force-enhanced kinetic effect. Other chemical and mechanical effects cause symmetric growth or resorption.

Analysis

Fig. 1 represents an FA experiencing proximal end growth via binding, and distal end resorption via unbinding under a tensile force transmitted by the actin stress fibers. The binding and proximal end growth of FAs is intriguing for its force-dependence, instances of which have been cited in the Introduction. This suggests that, while chemistry has a central role by the very fact that binding and unbinding are taking place, mechanical forces are able to influence the dynamics to a significant degree. We seek to explain this coupling of chemistry and mechanics (chemo-mechanics) in terms of non-equilibrium thermodynamics and kinetics.

Figure 1. Physical and mathematical model of a FA.

Figure 1

The FA is the grey-colored parallelogram. Binders are shown as double ellipses, either free or bound in complexes. The bound complexes have been depicted to be larger, of length Inline graphic in the Inline graphic-direction and their elastic response is represented by the springs. The elastic elements have length Inline graphic along the Inline graphic-direction. (Note that Inline graphic is an arbitrary length and has no effect on the potential; see the section titled “Driving force due to elasticity”.) The dotted arrows are actin stress fibers, which transfer force to the attached complexes. The bundle of actin stress fibers transmits total force Inline graphic. Also shown are the proximal and distal ends, Inline graphic, the centroid Inline graphic, length Inline graphic and domain boundaries Inline graphic.

Physical aspects

Consideration of forces that are constant in time

Absent any experimental interventions, the total force transmitted to an FA by the actin stress fibers can be time-dependent due to its coupling with actomyosin contractility and the dynamics of stress fibers. Balaban and co-workers [4] used patterned elastomeric substrates to target single FAs in stationary human foreskin fibroblasts. They found the force on a FA to be linearly related to its area, with a stress of Inline graphic kPa on the dynamically growing/resorbing FAs. However, in earlier work Galbraith and Sheetz [26] estimated that in chicken embryo fibroblasts migrating on a surface patterned with micromachined pads, the force generated by a single FA was Inline graphic nN. Their observation time frame was of the order of 1800 s, during which time the cells migrated over distances of Inline graphicm. This suggests two conclusions: (a) During cell locomotion, the system consisting of actin stress fibers attached to a single FA is capable of maintaining a roughly constant force over 1800 s. (b) The FA dynamics that interests us may have been taking place in these experiments but remained unobserved since the cell migration velocities of Inline graphic dominate FA growth velocities of Inline graphic [2]. These conclusions suggest additionally that it would be of interest to study FA dynamics under constant force. Using the methods of Shen and co-workers [27] it appears that it would be feasible to load individual FAs by micropillar actuation and subject them to constant-in-time force control. In such an experiment, the law of mechanical equilibrium would require that the balancing force from stress fibers on the FA remain constant in time. Motivated thus, we consider the dynamics of a single FA under constant total force, Inline graphic, as a means to probe the biophysics. We also note that at constant force and temperature the Gibbs free energy is the relevant thermodynamic potential, and it allows some simplification of the mathematical treatment in this preliminary study. Since we consider the case of constant force Inline graphic, the elasticity of actin stress fibers does not enter our formulation. Therefore we do not model the actin stress fibers explicitly, but we note that their effect is represented by the total force, Inline graphic, transmitted by the fibers.

Our treatment remains applicable if the stress is constant in time. This case has been worked out in the appendix. The greater complexity entailed by a force that varies in time with the dynamics of actomyosin contractility and stress fiber remodelling will be the subject of a subsequent study. The physical mechanisms that are identified in this work may remain valid in that situation, but their specific actions will probably differ.

A reaction-limited process

The formation of a bound complex requires diffusion of integrins in the cell membrane to a ligand binding site, followed by integrin-ligand binding, and subsequent diffusion of plaque proteins to the integrin site for formation of the integrin-ligand-plaque protein complex. Once formed, these complexes appear to be immobile in the interfacial plane of the cell membrane and substrate. The concentration of complexes will be denoted by Inline graphic. Diffusion and reaction-limited regimes have been considered for cell adhesion by Bell [28], although not explicitly for FA dynamics. Here we restrict ourselves to the reaction-limited regime by considering the diffusion of integrin and plaque proteins to happen relatively fast. The FA is able to exchange mass via binding and unbinding with molecules from the reservoirs of proteins in the cell membrane and cytosol, and the substrate reservoir of free ligands. These reservoirs are considered to remain at equilibrium. For concentration-dependent potentials of cytosolic proteins and free ligands, the implication is that these species can be represented by a single free binder species at fixed concentration, Inline graphic. The FA itself is not at equilibrium and is able to dynamically grow and undergo resorption (see File S1 and Movies S1, S2, S3, S4, S5, S6, S7, S8). Our computations confirm this setting of the reaction-limited regime to be a relevant limiting case, and a good model for the more detailed reaction-diffusion problem. We have provided computations of the full reaction-diffusion formulation as supporting information (File S1; Movies S9 and S10).

The FA geometry is predominantly one-dimensional

Images and supplementary information from several studies [1], [2], [4], [8], [11] have not indicated any noticeable variation in the FA width even as it grows and shrinks in length due to binding/unbinding of ligands, integrins and plaque proteins. More recent work has shown some width-wise variation, but it is significantly less than changes in length [10]. Motivated thus, we suppose that the FA maintains its width, say Inline graphic, even as its length changes. (However, also see the section titled “Summary of results in relation to mechanisms” where the tendency for weaker widthwise growth is explained in terms of this model.) The geometry and dynamics being predominantly one-dimensional, Inline graphic and Inline graphic will both be given units of numbers per unit length. Also, as discussed above, Inline graphic is taken to be an equilibrium value.

Bounds on the size of a complex

Arnold and co-workers [29] studied osteoblast attachment on substrates that were deposited with protein-functionalized gold nanodots of Inline graphic nm diameter for inter-nanodot separations of Inline graphic and Inline graphic nm. Their studies showed that FAs fail to develop if the separation between nanodots exceeds Inline graphic. We interpret this result to imply that there is a maximum allowed spacing between bound complexes in the lengthwise direction above which the FA does not form. This maximum value will be denoted by Inline graphic in our model. As the minimum spacing, Inline graphic, we use the integrin molecule packing separation of Inline graphic nm [30].

The elasticity model

We exploit the fact that FA formation involves integrin-ligand, integrin-plaque protein and plaque protein-actin bonds, and that the dominant direction of bonding is perpendicular to the cell membrane-substrate interface in the absence of force. While some bonds are understood to be formed between plaque proteins, it is not clear that there is sufficient lateral bonding to confer some shear stiffness on the FA. We will neglect the shear stiffness. Our mechanical model of the FA is therefore not an elastic layer, but a row of elastic elements arrayed in the lengthwise direction, each of which can only resist force along the ligand-integrin-plaque protein axis (Fig. 1). Bell [28] made the same observation on the deformation mechanism in adhesion bonds. We consider elastic elements, each of size Inline graphic. Consider an elastic element located at some point along the FA, and suppose that this element contains Inline graphic bound complexes. The average size of a complex within this elastic element is Inline graphic, and therefore the concentration averaged over the element is Inline graphic. The bounds introduced above for complex size lead to bounds on the concentration: Inline graphic, and Inline graphic. Here, Inline graphic is the concentration below which any assemblage of molecules is not considered to constitute an FA. The elastic element has a concentration-dependent Young's modulus Inline graphic. Following the elementary model of a linear elastic rod of cross-sectional area Inline graphic and height Inline graphic, the stiffness of each elastic element along the ligand-integrin-plaque protein axis is Inline graphic. (Also see the treatment of bond elasticity by Qian and co-workers [31].) Our model considers the substrate to be mechanically rigid motivated by observations that it is in this limit that FAs are found to develop their maximum size [32][35].

An additional elastic contribution can arise from the change in cell membrane curvature as bonds form or dissociate at the proximal and distals ends of the FA. We let the bending modulus be denoted by Inline graphic and the cell membrane's curvature be denoted by Inline graphic. As depicted in Fig. 1 the cell membrane is straight over the FA domain, i.e., on Inline graphic. However, a curvature of Inline graphic is possible for Inline graphic and Inline graphic. When FA growth takes place by binding of a new elastic element at Inline graphic or Inline graphic, the cell membrane must be unbent over a length Inline graphic to change the curvature from Inline graphic to zero. Unbinding allows the cell membrane to regain the curvature Inline graphic. The free energy changes associated with stretching of elastic elements and bending/unbending of the cell membrane contribute to the total free energy change that drives FA dynamics, as will be shown below.

Complexity in focal adhesion dynamics

When FAs are subject to force by the contractile stress fibers, a positive feedback is created between the force and size of stress fibers, and FA size. The formation of an initial focal complex, to which they are anchored, allows the stress fibers to generate a force that in turn drives the further growth of stress fibers. Increased force can then be transmitted to the focal complex, causing it to grow into an FA by binding of ligand-integrin-plaque protein complexes. The process continues with growth of the FA and stress fibers. This feedback has been reviewed by Bershadsky and co-workers [36], and modelled by Besser and Safran [37] using reaction-diffusion equations and a force-controlled stress fiber growth model. In the present paper, however, we wish to focus on explaining FA growth in a constant force experiment independent of any feedback between the FA and stress fibers. Consequently we will not address this feedback. The emphasis here is on the specific thermodynamic mechanisms that enable the various regimes of FA dynamics.

FA dynamics will be modelled over a region Inline graphic, which is a line segment on the interfacial plane between the cell membrane and substrate as shown in Fig. 1. Since the FA's geometry evolves, the positions of its distal and proximal ends are time dependent, and are denoted by Inline graphic and Inline graphic, respectively. Its centroid is denoted by Inline graphic and its length is Inline graphic. Its height remains fixed at Inline graphic, and it has width Inline graphic (not shown) in the interfacial plane along a direction perpendicular to the x-axis.

Proximal and distal edge velocities lead to FA treadmilling, growth and resorption

FA growth at either end requires that one complex, which we estimate to have the average size, Inline graphic, be added to attain a concentration Inline graphic at the newly-formed end. At a binding rate Inline graphic, the time required to add a complex at the corresponding end is Inline graphic. Over this time interval the corresponding edge of the FA advances by a distance Inline graphic. This gives the velocities of distal and proximal ends, respectively:

graphic file with name pone.0012043.e069.jpg (1)

respectively. For a force with positive Inline graphic-component as in Figs. 1, 2, 3, positive velocities, Inline graphic, Inline graphic, correspond to motion in the positive Inline graphic-direction, and therefore to distal unbinding and proximal binding, respectively.

Figure 2. Schematic of the bound and unbound states represented by corresponding reaction coordinates, and the energy barriers to transitions between them.

Figure 2

Figure 3. Schematic.

Figure 3

(a) FA geometry and loading. (b) Addition of a complex at the distal end makes the center of the FA move opposite to the direction of the horizontal force component, increasing its potential. (c) Addition of a complex at the proximal end makes the center of the FA move in the direction of the horizontal force component, decreasing its potential.

The Treadmilling Mode results when Inline graphic and Inline graphic. Proximal binding and distal unbinding causes the corresponding edges to move in the positive Inline graphic-direction maintaining the FA size while the centroid translates in the force's direction. The Growth Mode is operative in all cases with Inline graphic. The Resorption Mode is operative in all cases with Inline graphic. In the section titled “State diagrams of focal adhesion dynamics” we describe the development of these modes and their combinations. It is important to note that FA motion never happens with the same set of molecules making up the FA. Therefore, it does not move like a rigid body, but appears to do so as a combination of proximal and distal binding/unbinding.

A rate process driven by non-equilibrium thermodynamics, and modulated by kinetics

The framework of Harmonic Transition State Theory, as proposed by Vineyard [24] underlies our theoretical treatment. Consider the binding and unbinding of complexes, as depicted in Fig. 2, applied to an elastic element located at any point along the FA. The net binding rate expressed in terms of number of complexes is

graphic file with name pone.0012043.e079.jpg (2)

where Inline graphic is the Gibbs free energy barrier between the bound state and the transition surface, Inline graphic is the corresponding barrier between the unbound state and the transition surface, Inline graphic is the Boltzmann constant, Inline graphic is the temperature, and Inline graphic is an effective frequency associated with atomic vibration. We recognize that Inline graphic is the free energy difference between the binder-ligand pair in the bound and unbound states, and that this is also the chemical potential, Inline graphic, of the complexes. Accordingly we write,

graphic file with name pone.0012043.e087.jpg (3)

Suppose first that net binding occurs; i.e., Inline graphic in Eq. (2). Using Eq. (3) it is easy to see that Eq. (2) can be written as

graphic file with name pone.0012043.e089.jpg (4)

with Inline graphic signifying that binding is thermodynamically favored. Next, consider unbinding; i.e., Inline graphic. We also can write (2) as

graphic file with name pone.0012043.e092.jpg (5)

signifying that unbinding is thermodynamically favored. Of course, all three forms (2), (4) and (5) give Inline graphic for Inline graphic; i.e., if the bound and unbound states are in thermodynamic equilibrium.

As proposed by Bell [28], the energy barrier to bond breakage, Inline graphic can be lowered by a suitably oriented force, a result which can be expressed as

graphic file with name pone.0012043.e096.jpg (6)

where Inline graphic is the force-independent component of Inline graphic. Here, we have estimated the force transmitted to the bond on a single complex to be Inline graphic, since the total force Inline graphic is spread over a length Inline graphic. The parameter Inline graphic is the projection onto the force of the vector connecting the endpoints of the path along which the bond is deformed up to rupture. Following Bell's analysis we use Inline graphic. The unbinding rate therefore can be boosted by an exponential dependence on the force as shown immediately below. A recent paper [38] has reported a force-activated increase of the binding rate between integrins and ligands due to the hypothesized presence of catch bonds. Such an effect would be represented by an exponential dependence of the binding rate on Inline graphic in Eq. (8)Inline graphic.

graphic file with name pone.0012043.e106.jpg (7)

For the values listed in Table 1 this results in force activated unbinding kinetics at Inline graphic pN.

Table 1. Parameters used in the numerical examples.

Parameter (Symbol) Value Units Remarks
Force-independent part of potential (Inline graphic) Inline graphic J Models no growth for vanishing force.
Temperature (Inline graphic) Inline graphic K
Actin stress fiber angle (Inline graphic) Inline graphic degrees Effect of varying Inline graphic can be subsumed by varying Inline graphic.
Young's modulus (Inline graphic) Inline graphic kPa Estimate for soft, gel-like biological materials.
Bending modulus (Inline graphic) Inline graphic N.mInline graphic Estim. from Inline graphic and cell memb. thickness Inline graphic nm.
Cell memb. curvature (Inline graphic) Inline graphic mInline graphic Estim. from cell height Inline graphicm.
Initial FA length (Inline graphic) Inline graphic Inline graphicm Typical focal complex length.
FA width (Inline graphic) Inline graphic Inline graphicm Estim. from images in [4] and [2].
Upper bound on complex size (Inline graphic) Inline graphic nm Motivated by [29].
Lower bound on complex size (Inline graphic) Inline graphic nm Integrin packing density; see [30].
Conformational displacement (Inline graphic) Inline graphic nm Estimated from integrin domain size.
Kinetic coeff. (Inline graphic) Inline graphic sInline graphic For sliding velocities Inline graphicnmInline graphicsInline graphic; see [2].
Ref. displacement for unbinding (Inline graphic) 1.48 nm From ref. [28].
Max. conc. (Inline graphic) Inline graphic mInline graphic Inline graphic.

Eqs. (4) and (7) hold if there is no restriction on the local supplies of bound complexes and free binders, respectively. When the supplies are limited, the rates in Eqs. (4) and (7) must also be proportional to the concentrations of bound complexes and free binders, respectively. The rates therefore vary along the FA, and it proves convenient to express them as rates of change of concentration rather than of numbers. Since Inline graphic, we note that the concentration of bound complexes is low for the parameters listed in Table 1, and a discrete treatment in terms of numbers of bound complexes rather than concentrations may be considered preferable on these grounds. However, we have found the ease of representing non-uniform fields via concentrations in Eq. (8) to be advantageous in this study. The proximal and distal edge velocities in (1) and the computed results also reflect this dependence on local concentrations. Finally we write kinetic coefficients for binding and unbinding, Inline graphic and Inline graphic, respectively. The rate equations are summarized below. (Rigorously, proportionality factors must be introduced in extending Eqs. (4) and (7), which express rates of change of numbers, to (8), which is a concentration rate. However, these factors can be absorbed into Inline graphic and Inline graphic.)

graphic file with name pone.0012043.e155.jpg (8)

We note that a number of reactions are actually involved even in the simplified system considered here. These reactions include the binding/unbinding of ligands with integrins, of integrins with plaque proteins and of plaque proteins with actin. The single reaction rate of Eq. (8) can be viewed as representing the rate determining step of this cascade.

We now detail the specific thermodynamic driving forces that make up the chemical potential, Inline graphic.

Thermodynamic driving force from the work of complex addition at edges

Consider first the addition of a complex at the distal or proximal end to extend the FA domain. At time Inline graphic the distal and proximal ends of the FA are Inline graphic. At time Inline graphic the FA is deemed to extend over the regions Inline graphic and Inline graphic if the local concentration in these regions increases smoothly from Inline graphic to Inline graphic at time Inline graphic. This concentration can be achieved by adding a single complex over the average length Inline graphic. If the complex is added at the distal end the center of the FA shifts distally by Inline graphic. Conversely, addition of one complex over length Inline graphic at the proximal end shifts the FA's center proximally by Inline graphic. Since the newly-added complex bears some portion of the total force, Inline graphic, the result is a translation of the center of action of the force. Considering a force distributed uniformly over the FA for simplicity (but not necessity; computations including all the quantitative differences induced by non-uniform force distribution are also presented in this paper), the point of action of the total force translates with the FA centroid (see Fig. 3). A thermodynamic driving force arises due to the corresponding work done:

graphic file with name pone.0012043.e170.jpg (9)

The term, Inline graphic, represents one contribution to the chemical potential, Inline graphic. Its effect is antisymmetric with respect to complex addition at the distal and proximal ends. It is responsible for enabling the treadmilling regime of FA dynamics by favoring complex addition at the proximal end, but extracting an energetic cost for complex addition at the distal end.

Chemical driving forces

Let Inline graphic be the binding enthalpy. Using the concentration of free binders, Inline graphic, the mixing entropy is Inline graphic. The chemical driving force that contributes to the potential is

graphic file with name pone.0012043.e176.jpg (10)

Since bound complexes are immobile, they do not contribute to the mixing entropy or the resultant chemical potential. Also, consistent with the representation of the proteins in a cascade of reactions by a single free binder, we use a single mixing entropy term. Clearly, if Inline graphic is equal between distal and proximal ends, Inline graphic is also equal. In this case Inline graphic is symmetric between distal and proximal ends, and therefore does not influence FA treadmilling. This term causes symmetric binding if Inline graphic and conversely, unbinding for Inline graphic.

Driving force due to elasticity

For uniform force distribution that has been assumed for simplicity of presentation (but not necessity; see the section titled “Sensitivity studies”, where a crack-like force distribution has been considered), the force acting on an elastic element of length Inline graphic is Inline graphic. Since the elastic elements are modelled to have no shear stiffness as observed also by Bell [28], the ligand-integrin-plaque protein axis aligns with the force direction. The elastic free energy due to stretching of the element is

graphic file with name pone.0012043.e184.jpg
graphic file with name pone.0012043.e185.jpg (11)

As shown above, when the force is held constant the free energy is inversely proportional to the elastic modulus. The rigid substrate does not contribute an elastic free energy under force control since it has an effectively unbounded elastic modulus. In the above equation, the first term on the right hand-side of the first line is the elastic strain energy due to stretching of the elastic element, and the second is the change in potential of the force. This equation uses the fact that mechanical equilibrium is established at each instant for the current value of concentration Inline graphic. Using the number of complexes in the elastic element, Inline graphic, the elastic contribution to the chemical potential due to bond stretching is

graphic file with name pone.0012043.e188.jpg (12)

A further contribution comes from the elastic bending energy [39] as discussed in the section titled “The elasticity model”. Binding at the proximal or distal ends requires that the cell membrane be straightened out from its preferred curvature, Inline graphic, to zero curvature. The elastic free energy to straighten the cell membrane out over the length Inline graphic of an elastic element is

graphic file with name pone.0012043.e191.jpg (13)

Using Inline graphic gives the contribution to the chemical potential due to bending of the cell membrane

graphic file with name pone.0012043.e193.jpg (14)

The total elastic contribution to the chemical potential is

graphic file with name pone.0012043.e194.jpg (15)

The length Inline graphic of the elastic element makes its appearance during the derivation of Eqs. (12) and (14). For this reason, the elastic stiffness used in Eq. (11) is proportional to Inline graphic. Note that stiffness is a structural parameter and includes the effect of size in contrast with the elastic modulus Inline graphic, which is a material parameter and is independent of size. We point out that Inline graphic and Inline graphic are independent of Inline graphic, and therefore so are the resulting dynamics. The length, Inline graphic, has no further role in the formulation. For equal concentrations, Inline graphic, at the distal and proximal ends, these elastic contributions to the potential are also symmetric. Addition of a complex at either end results in the same value of Inline graphic.

For a given force distribution, a more sophisticated model for deformation of the FA, such as an elastic layer, would require the solution of a boundary value problem of linear elasticity with the additional complication of a concentration-dependent elastic modulus to determine stress distributions. While the quantitative results would change, the underlying physics would not be qualitatively different in our model. Alternately, in some studies [31], [40] arrays of FAs have been considered and fracture mechanics has been invoked to model the intervening gap regions between the cell membrane and substrate as cracks. With this approach a non-uniform stress distribution is obtained with stress concentrations at the distal and proximal edges. In the section titled “Sensitivity studies” we consider crack-like stress fields to demonstrate the model's robustness against variations in the stress distribution.

Driving force due to work done by force via conformational changes; protein mechanosensitivity

Conformational changes that are favorable to focal adhesion growth under tension have been hypothesized in vinculin [1], talin [7] and fibronectin [6] molecules as models of protein mechanosensitivity. Guided by these hypotheses we identify an additional possible driving force for complex formation under tensile force: Let Inline graphic be the change in internal energy of the conformation-changing molecule. This is the contribution to the chemical potential due to the conformational change in the absence of force. Additionally, if an external tensile force is transmitted to this molecule, there is a lowering of the potential energy of the system consisting of the molecule and the force mechanism if the displacement vectors of the conformational change have components aligned with the force vector. Thereby, the free energy of the system is also lowered, and these changed conformations are thermodynamically favored under an external tension. The total contribution to the chemical potential from the conformational change then is written as

graphic file with name pone.0012043.e205.jpg (16)

where Inline graphic is the component of the conformational length change along Inline graphic. This term, like Inline graphic, arises from work done by the force. While geometric details of conformational changes have been computed, e.g. for talin [7], they can be reduced to the form in Eq. (16). Like Inline graphic and Inline graphic the term, Inline graphic, is symmetric with respect to distal and proximal ends if the force distribution is equal at the two ends. For this reason it also cannot be the universal mechanism that is responsible for FA treadmilling.

The combination of Eqs. (9–16) fully specifies the chemical potential:

graphic file with name pone.0012043.e212.jpg (17)

We emphasize that of the various contributions to the chemical potential only Inline graphic is antisymmetric between distal and proximal ends. The others, Inline graphic and Inline graphic are symmetric between the ends. Therefore, while important to the total potential, Inline graphic, they do not cause symmetry breaking, which is necessary for the treadmilling mode.

The rate at which the concentration of bound complexes changes due to binding and unbinding can be computed, as described above, at any point on the FA. However, for the purpose of studying the growth and resorption of the FA structure, it is sufficient to focus on the distal and proximal ends only. This will be the aim in the remainder of this communication with special attention paid to symmetry breaking between the ends.

Results

Focal adhesion dynamics

Eq. (8) was solved numerically using the nonlinear ordinary differential equation integration routines in MATLAB. The parameters that appear in Table 1 have been fixed for the main study that follows (Fig. 4), but the model's sensitivity to some of them has been shown in Figs. 5, 6, 7, 8.

Figure 4. State diagram of the final position of the focal adhesion's centroid, Inline graphic, and length, Inline graphic at Inline graphic s as a function of force, Inline graphic.

Figure 4

Also shown are the normalized chemical potentials, Inline graphic and Inline graphic at the distal and proximal ends, respectively. The schematic diagrams indicate the dynamics corresponding to each regime. Note the various modes attained as Inline graphic is varied.

Figure 5. Effect of a crack-like force distribution, Eq. (19).

Figure 5

State diagram of the final position of the focal adhesion's centroid, Inline graphic, and length, Inline graphic at Inline graphic s as a function of total force, Inline graphic. The other parameters are as in Fig. 4, with which this state diagram should be compared. Also shown are the normalized chemical potentials, Inline graphic and Inline graphic at the distal and proximal ends, respectively.

Figure 6. Sensitivity of the model to changing the complex size, from the upper bound Inline graphic nm to the lower bound Inline graphic nm.

Figure 6

The other parameters are as in Fig. 4, with which this state diagram should be compared.

Figure 7. Sensitivity of the model to variation in the force-indpendent part of the potential, Inline graphic.

Figure 7

Here this combination has been varied between (a) Inline graphic and (b) Inline graphic. In comparison, Inline graphic in Fig. 4. The other parameters are as in Fig. 4, with which this state diagram should be compared.

Figure 8. Sensitivity of the model to variation in the size of the conformational change, Inline graphic.

Figure 8

Here this parameter has been changed to Inline graphic nm from Inline graphic nm assumed in Fig. 4. The other parameters are as in Fig. 4, with which this state diagram should be compared.

The following remarks can be made regarding our choice of parameters:

  1. Eq. (10) identifies enthalpic and entropic contributions to the chemical part of the potential, Inline graphic, and Eq. 16 identifies a force-independent (and therefore, chemical) part of the conformational change induced potential, Inline graphic. There are studies suggesting that the binding enthalpy of integrins to ligands lies in the range of Inline graphic [41] to Inline graphic [42]; however, this does not account for the further step by which plaque proteins bind to integrin. It is therefore difficult to precisely evaluate the individual contributions to Inline graphic and Inline graphic. However, experiments suggest that FAs do not grow in the absence of force [2], [4], [10], a result that is attained in our model by setting Inline graphic, and that we have adopted for the studies that follow.

  2. We have identified bounds on the complex size: Inline graphic. In generating numerical results with our model, we have first considered Inline graphic, followed by Inline graphic as part of a sensitivity study (Fig. 6).

  3. Our choice of a numerical value for the conformational change, Inline graphic is guided by the sizes of integrin and plaque protein molecules in the FA. Our first set of calculations is with Inline graphic nm, motivated by the approximate size of the integrin domain. The resulting dynamics are not qualitatively different unless this value is increased to Inline graphic nm (Fig. 8)— a sensitivity study whose results we put into perspective by comparison with experiments.

State diagrams of focal adhesion dynamics

The observed dynamics at time Inline graphic s are summarized in the state diagram of Fig. 4 by showing the FA length and displacement of its centroid on the left vertical axis versus the force, Inline graphic on the horizontal axis. This time instant has been chosen to correspond with the typical duration of a force-driven experiment [2], [4],[10]. Note that changes in length represent growth/resorption, while displacement of the centroid represents the effect of treadmilling. Also shown on the right vertical axis is the thermodynamic driving force represented by the normalized chemical potential, Inline graphic at the distal and proximal ends, respectively. The prevailing dynamic mode changes as the force is increased, and is indicated by schematic diagrams on which the direction and relative magnitudes of proximal and distal edge velocities are indicated by arrows. These modes are also identified in the text.

In the absence of force, i.e., when chemistry alone operates, Inline graphic as argued above. The FA does not change in length (Inline graphicm) and its centroid remains stationary (Inline graphicm). This is the Static Mode. As the force Inline graphic, inclined to the right in Figs. 1 and 3, increases from zero, the different mechanical contributions to Inline graphic begin to exert their influences. The distal- and proximal-end potentials become respectively, positive and negative to favor unbinding and binding from a combination of the following mechanisms: (i) the antisymmetric work of complex addition, (ii) the symmetric elastic free energy, and (iii) the symmetric conformational change. However, the combination of mechanisms makes Inline graphic more strongly negative than Inline graphic is made positive. As a result the proximal end binds more rapidly than the distal end unbinds. The resulting mode is Treadmilling with Growth. However, as the force increases and reaches Inline graphic pN the distal edge unbinds as rapidly as the leading edge binds. The length, Inline graphic remains fixed at Inline graphicm while the centroid moves rightward following the force as demonstrated by Inline graphic. This is the pure Treadmilling Mode, which dominates until Inline graphic pN, with an unchanged amount of treadmilling. Over the range Inline graphicInline graphic pN the Treadmilling Mode persists for arbitrarily large times since, according to the model, there are no changes in any of the contributions to Inline graphic (see Movie S3).

Beyond this force the trailing edge unbinds more rapidly than the leading edge binds, and the FA length shrinks to Inline graphicm. The distal unbinding is boosted in this regime due to the force-induced enhancement of unbinding kinetics as expressed in Eq. (8b). This is the Treadmilling with Resorption Mode. Between Inline graphic pN and Inline graphic pN each increment in force causes an increment in treadmilling (see the increasing Inline graphic over this regime). However, force-induced enhancement of unbinding kinetics at the distal end causes Inline graphic to decrease rapidly with force, and finally a transition sets in at Inline graphic pN. The distal unbinding becomes so pronounced that Inline graphic, and the FA is fully resorbed. The amount of treadmilling also enters a steep decline with force. Finally, at Inline graphic pN a thermodynamic transition also takes place. The thermodynamic driving force comes to be dominated by the quadratic elastic energy term, and causes unbinding at the proximal end in addition to the distal end. This is seen in the very sharp profile of Inline graphic as it crosses into the positive half-plane. This is the Mechanical Resorption Mode in which thermodynamics and kinetics combine to cause very rapid unbinding at both ends; Inline graphic shrinks and the FA is fully resorbed very rapidly.

The appendix shows the corresponding state diagram if the stress on the FA is the controlled parameter instead of the force.

Inadequacy of linear response theory

Linear response theory of classical non-equilibrium thermodynamics as laid out by de Groot and Mazur [25] is easily recovered from Eq. (8) by expanding Inline graphic up to first-order in Inline graphic. The linearized version of the rate law, Eq. (8) is

graphic file with name pone.0012043.e282.jpg (18)

The form of Eq. (18) is a combination of Equation (18) in Chapter III, and Equation (18) in Chapter IV of de Groot and Mazur [25]. Their equations are in terms of the affinity, which is the product of a stoichiometric factor and Inline graphic. The stoichiometric factor reduces to unity in our case because of the generic reaction that we consider between one binder and one ligand. In the framework described by de Groot and Mazur, therefore, Inline graphic is the overall binding coefficient, and Inline graphic is the overall unbinding coefficient.

Note, however, that the first-order approximations in Eq. (18) hold only for Inline graphic. On examining Fig. 4, it is clear, therefore, that linear response theory is inadequate for describing FA dynamics since Inline graphic is required to access the relevant dynamic modes.

Sensitivity studies

Some authors [31], [40] have considered arrays of FAs and invoked fracture mechanics to model the intervening gap regions between the cell membrane and substrate as cracks. To demonstrate the model's robustness against variations in the stress distribution we address this case for the first of our sensitivity studies. On adapting the treatment of Lin and Freund [40] to our model the stress distribution along the FA is:

graphic file with name pone.0012043.e288.jpg (19)

The resulting state diagram is Fig. 5, and is to be compared with Fig. 4. The differences are minor; the transitions between the different regimes occur at lower force, Inline graphic, because the edge stress concentration causes the distal edge to unbind at lower force.

Fig. 6 demonstrates the model's sensitivity to variation of the complex size from the upper bound, Inline graphic to the lower bound Inline graphic. This weakens the antisymmetric work term due to complex addition, Inline graphic (Eq. (9)). When compared with Fig. 4 the main difference is that distal unbinding is weaker, as reflected by the delayed divergence of Inline graphic. This allows the development of a Treadmilling with Growth Mode from Inline graphic pN to Inline graphic pN. This is followed by a Symmetric Growth Mode over a very small force regime, and a pure Treadmilling Mode over an even smaller force regime. From Inline graphic pN to Inline graphic pN there is a Treadmilling with Resorption Mode. The decrease in Inline graphic, which is linear in Inline graphic and competes against the quadratic elastic energy Inline graphic, allows Inline graphic to make its influence felt at a lower value of Inline graphic. Therefore the thermodynamic transition to Mechanical Resorption occurs earlier for this particular choice of Inline graphic.

Fig. 7 demonstrates the sensitivity to variation in the force-independent part of the potential, Inline graphic arising from the chemical part of the potential, and the internal energy of the conformational change, respectively. We consider this sum of terms to vary in the range of the thermal energy to model the spontaneous growth and resorption of the FA in the absence of force. We draw the reader's attention the changes in the low-force regime when compared with Fig. 4 : For Inline graphic FA growth is not favored in the absence of force, causing resorption for forces Inline graphic pN. Conversely, Inline graphic causes symmetric growth for forces Inline graphic pN.

Fig 8 demonstrates the sensitivity of the model to the size of the conformational change, Inline graphic. On increasing Inline graphic to 28 nm the chemical potentials at both ends become negative at low force, and binding is strongly favored at both ends. A Treadmilling with Growth Mode sets in at Inline graphic pN. However, due to the larger conformational change, a Symmetric Growth Mode develops at Inline graphic pN. This mode is strong enough that the centroid, Inline graphic, remains at its initial position, while the length, Inline graphic, grows well beyond Inline graphicm. The Symmetric Growth Mode persists until Inline graphic pN. The force-enhanced unbinding kinetics causes a transition to Treadmilling with Resorption at about the same force level as in Fig. 4, and is followed by the thermodynamic transition to the Mechanical Resorption Mode at marginally higher force than in Fig. 4.

In File S1 we have provided a list of movies of FA dynamics in the reaction-limited regime (Movies S1, S2, S3, S4, S5, S6, S7, S8), as well as movies of computations resulting from the extension of our formulation to the reaction-diffusion regime (Movies S9 and S10).

Summary of results in relation to mechanisms

  1. The central result of this study is that FA treadmilling is explained by the antisymmetric work done by the external force while complexes are being added to or removed from the FA. This mechanism provides the required symmetry-breaking in the thermodynamic driving force, and has the compelling feature that the most elementary concept of Work is sufficient to explain treadmilling. There is no need to invoke special constitutive hypotheses for this purpose. We recognize, however, that it does not rule out such constitutive models (see “Relation to other models in the literature”). On the other hand the symmetry of the chemical, elastic and conformational change-driven components of the chemical potential makes them unlikely contenders for the role of the mechanism that enables treadmilling.

  2. Once the symmetry of the thermodynamic driving force has been broken, the kinetics lead to further asymmetry due to the force dependence of the unbinding rate. The interplay between the kinetic and thermodynamic effects determines the growth, treadmilling and resorption modes in Figs. 4, 5, 6, 7, 8.

  3. The elastic part of the chemical potential has one contribution, Inline graphic, which is quadratic in force and contributes to thermodynamically-driven resorption of the FA. We have found that the elastic contribution due to cell membrane bending, Inline graphic, has no noticeable effect on the dynamics, and is dominated by the bond stretching contribution, Inline graphic.

  4. Figs. 4, 5, 6, 7, 8 emphasize the dynamic behavior of FAs under the influence of force. However, we note that the model does not preclude the existence of a static FA under force. This only requires a non-uniform force distribution over the FA by which the chemical potential vanishes at both ends: Inline graphic.

  5. The dominant direction of force exerted by the actin stress fibers is along the axis of the FA. However, there are reports of individual filaments having non-axial orientations in the plane of the FA [43]. The conclusion reached with our model is that the force transmitted by these individual filaments does result in growth along nonaxial (e.g., widthwise) directions in the FA plane. However, since the dominant direction of actin stress fibers is axial, the FA grows substantially more in the axial direction and attains an elongated shape. We have modelled the idealized case of constant width, Inline graphic.

  6. We have tested several different forms of the rate law in place of Eq. (8). Some of these forms have reached saturation at finite rates as Inline graphic, others have had no exponential dependence on force Inline graphic, and yet others have been linear in Inline graphic. All these forms were found to be qualitatively equivalent in the sense that the state diagram Fig. 4 showed the same dynamic modes, and the sensitivity studies in Figs. 58 demonstrated the same changes in these modes. In this sense the results and our conclusions reported in this section are independent of the rate law. The quantitative results were, of course, different in that the transitions between different modes occurred at different force levels for these different rate laws. The kinetic coefficients Inline graphic and Inline graphic were also different for these different rate laws. The results comparing different rate laws have not been shown for the sake of brevity.

Discussion

Comparison with experiments

  1. The distinct modes of focal adhesion growth, treadmilling, resorption and combinations of these modes are obtained as the force is varied. The dynamics seen in the supporting movie of Nicolas and co-workers [8] and in Aroush and co-workers [10] is well-represented by the model.

  2. The state diagrams in Fig. 6 with Inline graphic nm, and Fig. 8 with Inline graphic nm include the Symmetric Growth Mode, in which distal and proximal ends grow outward at the same velocity. This mode, however, has not been observed in experiments [2], [4], [10], suggesting that these parametric values are not representative of the underlying physics.

  3. Fig. 6, shows that the FA remains robust at Inline graphic pN. For a FA of area Inline graphic this leads to a stress Inline graphic kPa, a prediction which corresponds well with the stress of Inline graphic kPa measured by Balaban and co-workers [4]. Since this state diagram was generated using the smaller size Inline graphic for the ligand-integrin-plaque protein, it suggests that this value is more representative of the physics than Inline graphic.

  4. As the second more specific prediction we note the following: For a cell radius of Inline graphicm our finding that FAs are resorbed at Inline graphicInline graphic pN (Figs. 4 and 5) suggests that Inline graphicInline graphic FAs are needed to attain a cell adhesion strength Inline graphic kPa measured by Garcia and co-workers [44].

  5. For the third prediction we note: If a single ligand-integrin-plaque protein complex has size 20–73 nm and is attached to a single actin filament, between Inline graphic and Inline graphic filaments transmit force to an FA of size Inline graphic. This requires that at a force that causes FA resorption, Inline graphicInline graphic pN, each actin filament can bear 1.6–50 pN force without itself breaking. These levels are far below the 200–600 pN rupture force of actin filaments measured using glass needle micromanipulation in vitro by Tsuda et al. [45], and even typical loads of 120–200 pN in the muscle sarcomere during isometric contraction by Oosawa [46]. Separately, rupture forces for standalone Inline graphic integrin-fibronectin complexes were Inline graphic pN as measured via AFM [47]. Therefore the force level required in the actin filaments by our model is within experimental bounds.

  6. Recent experiments [48] have shown that when fibroblasts are plated on compliant substrates that are dynamically stretched, the FAs in the fibroblasts orient with the direction of stretching. Our model is able to explain this observation when we carry out a two-dimensional extension of the antisymmetric work term, Inline graphic. We have Inline graphic, where Inline graphic is the projection of the force vector Inline graphic on the interfacial plane, and Inline graphic is the unit vector in the interfacial plane along the direction of FA extension/retraction due to complex binding/unbinding. Clearly, Inline graphic is most negative when Inline graphic is aligned with Inline graphic, as happens for proximal binding and distal unbinding in the direction of Inline graphic. Conversely, Inline graphic is most positive when Inline graphic is aligned opposite to Inline graphic in the interfacial plane, as happens for proximal unbinding and distal binding in the direction opposite to Inline graphic. Therefore, proximal binding and distal unbinding are favored in the direction of Inline graphic, and the FA aligns with the force projection Inline graphic.

  7. In experiments carried out by Cluzel et al. [49] the integrins from FA structures remained intact even after the transmission of force was disrupted by destruction of the actin cytoskeleton. However, proteins such as paxillin and vinculin were not recruited in the absence of force. Since the different proteins are not explicitly modelled, this detail is not represented in our results. Instead, our model's results represent the failure to grow the FA structure in the absence of force. This is in correspondance with the experimental observations of Cluzel et al.

  8. We have considered the effects of variation of a few parameters on the state diagrams in Figs. 48. These include the force distribution, the complex size Inline graphic, the force-independent part of the potential Inline graphic, and the conformational change Inline graphic. Through these studies we have identified broad regimes and trends described in these state diagrams, argued against some parametric values, made specific predictions, and made connections to additional experiments. Molecular structural studies and conformational changes predicted by molecular modelling will bring greater certainty to estimates for Inline graphic and Inline graphic. With better estimates for parameters, force-controlled loading of individual FAs, e.g. by actuating micropillars, could provide further benchmarks against which specific predictions of our model can be tested, and the model can be verified or falsified.

Relation to other models in the literature

In general, the papers that we review in this section have addressed the dynamics of FA response to force. Growth, treadmilling and resorption regimes have been reproduced by some of these models. Our contribution, coming after these papers, is focused on alternate, perhaps simpler, explanations of FA dynamics.

Shemesh and co-workers [16] and Besser and Safran [17] (the latter paper encompasses and elaborates upon earlier work by Nicolas and co-workers [8] and Nicolas and Safran [15]) have attempted to explain FA dynamics in thermodynamic or kinetic terms, although the details of each treatment are different from ours. The most significant manifestation of these differences is the conclusion reached in these two papers that as the external force on the FA increases beyond some level, the FA continues to grow, and does so symmetrically. In contrast our study concludes that the force-induced enhancement of unbinding kinetics, and the quadratic force-dependence of the elastic free energy ultimately cause resorption (Figs. 48). We note that the more recent work of Nicolas and co-workers [19] reaches the same conclusion as we do with regard to the ultimate resorption of the FA at high force levels.

Shemesh and co-workers require a very specific geometry of the FA: free of ligand-integrin bonds at the proximal end, and bonded at the distal end. Without such a geometry the gradient of FA stress that they find to drive transport disappears and their model fails. Also see the comments by Besser and Safran [17] in this regard.

Besser and Safran [17] and Nicolas and co-workers [19] consider the detailed geometry pf the integrin-ligand bonded region: extending slightly beyond the plaque protein layer distally and proximally. (Also see the appendix of this communication where this geometry is used to explain symmetry-breaking by the work done during addition/removal of a bound complex under stress control.) The authors hypothesize that the proximal compressive stress in the integrin-ligand bonded layer induces a molecular conformational change that favors FA growth via binding, while the distal tensile stress inhibits binding, and therefore downregulates growth. However, all the molecular conformation changes that have been hypothesized to favor FA growth [1], [6], and investigated [7] involve uncoiling by local tension, not compression. Compression in one direction can induce tension in any other direction according to the theory of elasticity. However, a negative Poisson ratio and a constraint against free deformation in the second direction are both required. These make up rather specific constitutive and microstructural assumptions, which have not been investigated to our knowledge. Nicolas and co-workers [19] consider the possibility that a force-induced gradient in conformation also may favor binding. Apart from the possible roles of compression and gradient in conformation, the treatment of the conformational change by Besser and Safran [17] and by Nicolas and co-workers [19] is essentially the same as ours.

We note that Shemesh and co-workers [16] applied the Gibbs-Duhem relation to obtain a result for potential due to elasticity that is equivalent to our elastic free energy results in Eqs. (11) and (13). However, they did not consider the change in stiffness with concentration, for which reason their elastic potential is the same as our elastic free energy. Also, in contrast with their model and the model of Besser and Safran [17] or of Nicolas and co-workers [19], our model does not include in-plane shear stress, because we have neglected shear stiffness—using a physical argument on orientation of integrin-ligand bonds that was also invoked by Bell [28]. While Nicolas and co-workers [19] have considered FA dynamics dependent on substrate stiffness, our study considers a mechanically rigid (infinitely stiff) substrate that does not store elastic energy under force-control, and therefore does not affect the thermodynamics. It is in this limiting case that FAs develop to their maximum size [32][35].

Deshpande and co-workers [18] have built in a finite strength via an elastic potential for the integrin-ligand bonds. In the present study, however, a finite strength emerges from competition between different thermodynamic driving forces for binding and unbinding, modulated by force-enhanced kinetics. Deshpande and co-workers' constitutive model assumes integrin sliding once the maximum stress is reached. They discuss this mechanism qualitatively to rationalize their constitutive model. In constrast we have presented an integrated quantitative treatment that accounts for FA growth, treadmilling and resorption. Deshpande and co-workers consider low- and high-affinity states of integrin molecules that correspond to the unbound and bound states in our work. They advance two distinct and independent hypotheses for the sliding mode (equivalent to our treadmilling mode) of FA dynamics: (i) Integrin molecules switch from high affinity states of deformed, energetically-unfavorable conformations into low affinity and low energy states. Other integrins maintain strain continuity of the FA by switching from low to high affinity, but lower energy, states. (ii) The propagation of a front of switching from high to low affinity states allows FA sliding—a mechanism that they liken to dislocation glide in crystals. These are highly specific mechanisms that await validation.

Aroush and co-workers [10] have studied the accumulation and depletion of proteins over individual FAs. They observe growth, treadmilling and resorption, and combinations of these modes, all of which are reproduced by our model. Direct comparisons are not possible because their experimental study did not employ force control, which is the case that we have considered. Instead, the FAs in their study were subject to force from actomyosin contractility. Cytoskeletal dynamics therefore affected the magnitude of force developed. However, their study did bear out the predominantly one-dimensional nature of FA dynamics, which is the case we considered. Aroush and co-workers also developed a model for the stress distribution along the FA-ECM interface based on shear-lag effects. They make the constitutive hypothesis that the unsymmetric stress profile that emerges as a main result of shear lag could be responsible for causing unbinding of proteins at the distal edge. They do not consider fundamental physical mechanisms that could favor binding or unbinding as we have done.

Importantly, the present communication finds no need for special constitutive assumptions on the response characteristics of the material in a FA to explain the treadmilling regime. Instead, treadmilling is explained by a more fundamental principle: It is driven by the asymmetry between the work done when complexes are added at the distal and proximal ends, respectively. If the findings reported here stand up to experimental validation they could be the basis for techniques by which to control FA formation by interfering with force generation of stress fibers in actin-targeted therapies. Furthermore, this model can be combined with models of cytoskeleton dynamics and contribute to the understanding of cell adhesion and motility.

Appendix: The stress-controlled case

Under stress control the appropriate thermodynamic potential differs from the Gibbs free energy, Inline graphic. Instead of Inline graphic, with internal energy Inline graphic, entropy Inline graphic and displacement Inline graphic, we have

graphic file with name pone.0012043.e373.jpg (20)

where Inline graphic is the stress, Inline graphic the volume and Inline graphic is the strain. In analogy with Fig. 2 we write Inline graphic as the barrier between the unbound state and the transition state, and Inline graphic as the barrier between the bound state and the transition state, respectively. Proceeding as in Eqs. (2–8) with Inline graphic taking the place of Inline graphic we have

graphic file with name pone.0012043.e381.jpg (21)

where the chemical potential is redefined as Inline graphic and Inline graphic is the force on a bound complex. The other quantities have the same definitions as in Eqs. (2–8).

It remains now to specify the contributions to Inline graphic. From Eq. (20) it follows that the addition of a complex of length Inline graphic at the distal or proximal edge changes Inline graphic as follows: Inline graphic. Since Inline graphic is the applied stress, any non-uniformity in this quantity could induce symmetry-breaking between the proximal and distal edges. Even if Inline graphic is equal at the two edges the sign of this contribution depends on Inline graphic. Recall from the first paragraphs of the Introduction and the section titled “A reaction-limited process” that some plaque proteins bind with integrins and others bind to F-actin. Since a substrate must be available for newly-attached actin to generate force, it suggests that at least some assembly of plaque proteins must happen before actin binds to them. If this newly forming complex, with plaque proteins but no bond to actin yet, is at the proximal edge it is compressed by the actin-loaded FA behind it; alternately, if it forms at the distal edge it is stretched by the actin-loaded FA ahead of it [15]. Once the actin bond forms this complex gets stretched by the stress Inline graphic; however, proximal stretching, Inline graphic, is from a compressed reference state, while the distal stretching, Inline graphic, is from an already-stretched reference state. On this basis we have Inline graphic if Inline graphic is uniform. The associated change in Inline graphic is the chemical potential from the work of complex addition at edges:

graphic file with name pone.0012043.e397.jpg (22)

This is the symmetry-breaking mechanism under stress-control. The details of how it operates do differ from symmetry-breaking under force-control (Eq. (9)), and the mechanism summarized in Eq. (22) does require some shear stress to develop; see Ref. [15]. However, this is to be expected. Since the boundary conditions on the system have changed, its response also has changed and effects that were neglected for force-control must now be included. Importantly, the feature of symmetry-breakage remains.

Since the cases of force- or stress-control introduce no differences in the chemical enthalpy or the entropy, the contribution from chemical driving forces is the same as in Eq. (10):

graphic file with name pone.0012043.e398.jpg (23)

Following the development of the section titled “Driving force due to elasticity”, but accounting for stress-control instead of force-control, we have

graphic file with name pone.0012043.e399.jpg (24)

Using Inline graphic and Inline graphic gives

graphic file with name pone.0012043.e402.jpg (25)

where Inline graphic is an effective modulus for the combined normal-shear mode of deformation.

The driving force due to work done by conformational change of a protein is obtained by following the arguments to Eq. (16) and replacing the force Inline graphic with Inline graphic:

graphic file with name pone.0012043.e406.jpg (26)

Then the total chemical potential from all four mechanisms is Inline graphic using Eqs. (22–26). Note that for a uniform stress, Inline graphic, the only symmetry-breaking mechanism is the work done due to complex addition at the edges, and therefore this is the mechanism that largely determines the treadmilling regime. A state diagram of FA dynamics under stress-control appears as Fig. 9. While the Symmetric Growth Mode dominates the dynamics, there is a small regime of the Treadmilling Mode at Inline graphic kPa, which correspnds well with the stress Inline graphic kPa measured by Balaban and co-workers [4]. Thus, the broad qualitative conclusions do not differ in the stress-controlled case, although the details of the regimes of the various dynamic modes do differ in quantitative terms. This is perhaps not surprising: the system differs in its response for different boundary conditions.

Figure 9. State diagram of FA dynamics for stress-control (see Appendix).

Figure 9

The strains used were Inline graphic, Inline graphic, with other parameters are as in Fig. 4, with which this state diagram could be compared. The main effect of stress-control on FA dynamics is that the symmetric growth mode is strongly favored over most low stress values, as seen by the growth of Inline graphic. However, the Treadmilling Mode does hold over a narrow range for Inline graphic kPa, which compares well with the stress of Inline graphic kPa reported by Balaban and co-workers [4]. The Treadmilling Mode gets suppressed for less polarized strains Inline graphic relative to Inline graphic.

Supporting Information

File S1

(0.02 MB PDF)

Movie S1

P = 0 pN and small conformational change; no growth.

(0.37 MB AVI)

Movie S2

P = 1×10−1 pN and small conformational change; treadmilling with growth.

(0.47 MB AVI)

Movie S3

P = 1×102 pN and small conformational change; pure treadmilling.

(0.41 MB AVI)

Movie S4

P = 104 pN and small conformational change; treadmilling with resorption.

(0.20 MB AVI)

Movie S5

P = 0 pN and large conformational change; no growth.

(0.37 MB AVI)

Movie S6

P = 1×10−1 pN and large conformational change; treadmilling dominated by growth at both ends.

(0.47 MB AVI)

Movie S7

P = 1×102 pN and large conformational change; symmetric growth.

(0.42 MB AVI)

Movie S8

P = 104 pN and large conformational change; treadmilling with resorption.

(0.20 MB AVI)

Movie S9

Reaction-diffusion formulation with P = 102 pN and small conformational change; pure treadmilling is discernible.

(0.89 MB AVI)

Movie S10

Reaction-diffusion formulation with P = 102 pN and large conformational change; treadmilling is dominated by growth at both ends.

(0.92 MB AVI)

Footnotes

Competing Interests: The authors have declared that no competing interests exist.

Funding: JEO was supported by a Rackham pre-doctoral fellowship from the University of Michigan. KG was supported by a Presidential Early Career Award for Scientists and Engineers (PECASE) grant from the Department of Energy. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

References

  • 1.Zamir E, Geiger B. Molecular complexity and dynamics of cell-matrix adhesions. J Cell Sci. 2001;114:3583–3590. doi: 10.1242/jcs.114.20.3583. [DOI] [PubMed] [Google Scholar]
  • 2.Riveline D, Zamir E, Balaban NQ, Schwarz US, Ishizaki T, et al. Focal contacts as mechanosensors: Externally applied local mechanical force induces growth of focal contacts by an mDia1-dependent and ROCK-independent mechanism. J Cell Biol. 2001;153:1175–1185. doi: 10.1083/jcb.153.6.1175. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.Zaidel-Bar R, Ballestrem C, Kam Z, Geiger B. Early molecular events in the assembly of matrix adhesions at the leading edge of migrating cells. J Cell Sci. 2003;116:4605–4613. doi: 10.1242/jcs.00792. [DOI] [PubMed] [Google Scholar]
  • 4.Balaban NQ, Schwarz US, Riveline D, Goichberg P, Tzur G, et al. Force and focal adhesion assembly: a close relationship studied using elastic micropatterned substrates. Nat Cell Biol. 2001;3:466–473. doi: 10.1038/35074532. [DOI] [PubMed] [Google Scholar]
  • 5.Johnson RP, Craig SW. F-actin binding site masked by the intramolecular association of vinculin head and tail domains. Nature. 1995;373:261–264. doi: 10.1038/373261a0. [DOI] [PubMed] [Google Scholar]
  • 6.Geiger B, Bershadsky A, Pankov R, Yamada KM. Transmembrane extracellular matrix–cytoskeleton crosstalk. Nat Rev Mol Cell Biol. 2001;2:793–805. doi: 10.1038/35099066. [DOI] [PubMed] [Google Scholar]
  • 7.Lee SE, Kamm RD, Mofrad RK. Force-induced activation of talin and its possible role in focal adhesion mechanotransduction. J Biomech. 2007;40:2096–2106. doi: 10.1016/j.jbiomech.2007.04.006. [DOI] [PubMed] [Google Scholar]
  • 8.Nicolas A, Geiger B, Safran SA. Cell mechanosensitivity controls the anisotropy of focal adhesions. Proc Natl Acad Sci USA. 2004;101:12520–12525. doi: 10.1073/pnas.0403539101. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Lele TP, Pendse J, Kumar S, Salanga M, Karavitis J, et al. Mechanical forces alter zyxin unbinding kinetics within focal adhesions of living cells. J Cell Physiol. 2006;207:187–194. doi: 10.1002/jcp.20550. [DOI] [PubMed] [Google Scholar]
  • 10.Aroush DRB, Zaidel-Bar R, Bershadsky AD, Wagner HD. Temporal evolution of cell focal adhesions: Experimental observations and shear stress profiles. Soft Matt. 2008;4:2410–2417. [Google Scholar]
  • 11.Smilenov LB, Mikhailov A, Jr RJP, Mercantonio EE, Gundersen GG. Focal adhesion motility revealed in stationary fibroblasts. Science. 1999;286:1172–1174. doi: 10.1126/science.286.5442.1172. [DOI] [PubMed] [Google Scholar]
  • 12.Zamir E, Katz M, Posen Y, Erez N, Yamada KM, et al. Dynamics and segregation of cell-matrix adhesions in cultured fibroblasts. Nat Cell Biol. 2000;2:191–197. doi: 10.1038/35008607. [DOI] [PubMed] [Google Scholar]
  • 13.Evans EA. The detailed mechanics of membrane adhesion and separation. 1. Continuum of molecular cross-bridges. Biophys J. 1985;48:175–183. doi: 10.1016/S0006-3495(85)83770-X. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Dembo M, Torney DC, Saxman K, Hammer D. The reaction-limited kinetics of membrane-to-surface adhesion and detachment. Proc Roy Soc Lond Ser B. 1988;234:55–83. doi: 10.1098/rspb.1988.0038. [DOI] [PubMed] [Google Scholar]
  • 15.Nicolas A, Safran SA. Elastic deformations of grafted layers with surface stress. Phys Rev E. 2004;69:051902-1–051902-7. doi: 10.1103/PhysRevE.69.051902. [DOI] [PubMed] [Google Scholar]
  • 16.Shemesh T, Geiger B, Bershadsky AD, Kozlov MM. Focal adhesions as mechanosensors: A physical mechanism. Proc Natl Acad Sci USA. 2005;102:12383–12388. doi: 10.1073/pnas.0500254102. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Besser A, Safran SA. Force-induced adsorption and anisotropic growth of focal adhesions. Biophys J. 2006;90:3469–3484. doi: 10.1529/biophysj.105.074377. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.Deshpande VS, Mrksich M, McMeeking RM, Evans AG. A bio-mechanical model for coupling cell contractility with focal adhesion formation. J Mech Phys Solids. 2008;56:1484–1510. [Google Scholar]
  • 19.Nicolas A, Besser A, Safran SA. Dynamics of cellular focal adhesions on deformable substrates: Consequences for cell force microscopy. Biophys J. 2008;95:527–539. doi: 10.1529/biophysj.107.127399. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Freund LB, Lin Y. The role of binder mobility in spontaneous adhesive contact and implications for cell adhesions. J Mech Phys Solids. 2004;52:2455–2472. [Google Scholar]
  • 21.Lin Y, Inamdar M, Freund LB. The competition between brownian motion and adhesion in soft materials. J Mech Phys Solids. 2008;56:241–250. [Google Scholar]
  • 22.Wang J, Gao H. Clustering instability in adhesive contact between elastic solids via diffusive molecular bonds. J Mech Phys Solids. 2008;56:251–266. [Google Scholar]
  • 23.Freund LB. Characterizing the force generated by a molecular bond as it is forcibly separated. Proc Natl Acad Sci. 2009;106:8818–8823. doi: 10.1073/pnas.0903003106. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24.Vineyard GH. Frequency factors and isotope effects in solid state rate processes. J Phys Chem Solids. 1957;3:121–127. [Google Scholar]
  • 25.de Groot SR, Mazur P. Nonequilibrium Thermodynamics. New York: Dover; 1984. [Google Scholar]
  • 26.Galbraith CG, Sheetz MP. A micromachined device provides a new bend on fibroblast traction forces. Proc Natl Acad Sci USA. 1997;94:9114–9118. doi: 10.1073/pnas.94.17.9114. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Shen CJ, Fu JP, Chen CS. Patterning cell and tissue function. Cell Mol Bioeng. 2008;1:15–23. [Google Scholar]
  • 28.Bell GI. Models for the specific adhesion of cells to cells. Science. 1978;200:618–627. doi: 10.1126/science.347575. [DOI] [PubMed] [Google Scholar]
  • 29.Arnold M, Cavalcanti-Adam EA, Glass R, Blummel J, Eck W, et al. Activation of integrin function by nanopatterned adhesive interfaces. ChemPhysChem. 2004;5:383–388. doi: 10.1002/cphc.200301014. [DOI] [PubMed] [Google Scholar]
  • 30.Erb EM, Tangemann K, Bohrmann B, Muller BB, Engel J. Integrin alphaIIb beta3 reconstituted into lipid bilayers is nonclustered in its activated state but clusters after fibrinogen binding. Biochem. 1997;36:7395–7402. doi: 10.1021/bi9702187. [DOI] [PubMed] [Google Scholar]
  • 31.Qian J, Wang J, Gao H. Lifetime and strength of adhesive molecular bond clusters between elastic media. Langmuir. 2008;24:1262–1270. doi: 10.1021/la702401b. [DOI] [PubMed] [Google Scholar]
  • 32.Katz BZ, Zamir E, Bershadsky A, Kam Z, Yamada KM, et al. Physical state of the extracellular matrix regulates the structure and molecular composition of cell-matrix adhesions. Cell Biol. 2000;11:1047–1060. doi: 10.1091/mbc.11.3.1047. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.Cukierman E, Pankow R, Stevens DR, Yamada KM. Taking cell-matrix adhesions to the third dimension. Science. 2001;294:1708–1712. doi: 10.1126/science.1064829. [DOI] [PubMed] [Google Scholar]
  • 34.Beningo KA, Dembo M, Wang Y. Responses of fibroblasts to anchorage of dorsal extracellular matrix receptors. Proc Natl Acad Sci USA. 2004;101:18024–18029. doi: 10.1073/pnas.0405747102. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35.Nicolas A, Safran SA. Limitation of cell adhesion by the elasticity of the extracellular matrix. Biophys J. 2006;91:61–73. doi: 10.1529/biophysj.105.077115. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36.Bershadsky A, Kozlov M, Geiger B. Adhesion-mediated mechanosensitivity: a time to experiment, and a time to theorize. Curr Opin Cell Biol. 2006;18:472–481. doi: 10.1016/j.ceb.2006.08.012. [DOI] [PubMed] [Google Scholar]
  • 37.Besser A, Schwarz US. Coupling biochemistry and mechanics in cell adhesion: a model for inhomogeneous stress fiber contraction. New J Phys. 2007;9:425–452. [Google Scholar]
  • 38.Kong F, García AJ, Mould AP, Humphries MJ, Zhu C. Demonstration of catch bonds between an integrin and its ligand. J Cell Biol. 2009;185:1275–1284. doi: 10.1083/jcb.200810002. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 39.Helfrich W. Elastic properies of lipid bilayers: theory and possible experiments. Z Naturf C. 1973;28:693–703. doi: 10.1515/znc-1973-11-1209. [DOI] [PubMed] [Google Scholar]
  • 40.Lin Y, Freund LB. Optimum size of a molecular bond cluster in adhesion. Phys Rev E. 2008;78:021909-1–021909-6. doi: 10.1103/PhysRevE.78.021909. [DOI] [PubMed] [Google Scholar]
  • 41.Palecek SP, Loftus JC, Ginsberg MH, Lauffenburger DA, Horwitz AF. Integrin-ligand binding properties govern cell migration speed through cell-substratum adhesiveness. Nature. 1997;385:537–540. doi: 10.1038/385537a0. [DOI] [PubMed] [Google Scholar]
  • 42.Takagi J, Erickson HP, Springer TA. C-terminal opening mimics ‘inside-out’ activation of integrin α5β1. Nature Structural Biology. 2001;8:412–416. doi: 10.1038/87569. [DOI] [PubMed] [Google Scholar]
  • 43.Geiger B. 2008. Private communication.
  • 44.Garcia AJ, Ducheyne P, Boettiger D. Quantification of cell adhesion using a spinning disc device and application to surface-reactive materials. Biomat. 1997;18:1091–1098. doi: 10.1016/s0142-9612(97)00042-2. [DOI] [PubMed] [Google Scholar]
  • 45.Tsuda Y, Yasutake H, Ishijima A, Yanagida T. Torsional rigidity of single actin filaments and actin-actin bond breaking force under torsion measured directly by in vitro micromanipulation. Proc Natl Acad Sci. 1996;93:12937–12942. doi: 10.1073/pnas.93.23.12937. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 46.Oosawa F. Actin-actin bond strength and conformational change of F-actin. Biorheol. 1977;14:11–19. doi: 10.3233/bir-1977-14102. [DOI] [PubMed] [Google Scholar]
  • 47.Li FY, Redick SD, Erickson HP, Moy VT. Force measurements of the alpha5 beta1 integrin-fibronectin interaction. Biophys J. 2003;84:1252–1262. doi: 10.1016/S0006-3495(03)74940-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 48.Jungbauer S, Kemkemer R, Spatz J. 2008. Private communication.
  • 49.Cluzel C, Saltel F, Lussi J, Paulhe F, Imhof BA, et al. The mechanisms and dynamics of alpha(v)beta(3) integrin clustering in living cells. J Cell Biol. 2005;171:283–292. doi: 10.1083/jcb.200503017. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

File S1

(0.02 MB PDF)

Movie S1

P = 0 pN and small conformational change; no growth.

(0.37 MB AVI)

Movie S2

P = 1×10−1 pN and small conformational change; treadmilling with growth.

(0.47 MB AVI)

Movie S3

P = 1×102 pN and small conformational change; pure treadmilling.

(0.41 MB AVI)

Movie S4

P = 104 pN and small conformational change; treadmilling with resorption.

(0.20 MB AVI)

Movie S5

P = 0 pN and large conformational change; no growth.

(0.37 MB AVI)

Movie S6

P = 1×10−1 pN and large conformational change; treadmilling dominated by growth at both ends.

(0.47 MB AVI)

Movie S7

P = 1×102 pN and large conformational change; symmetric growth.

(0.42 MB AVI)

Movie S8

P = 104 pN and large conformational change; treadmilling with resorption.

(0.20 MB AVI)

Movie S9

Reaction-diffusion formulation with P = 102 pN and small conformational change; pure treadmilling is discernible.

(0.89 MB AVI)

Movie S10

Reaction-diffusion formulation with P = 102 pN and large conformational change; treadmilling is dominated by growth at both ends.

(0.92 MB AVI)


Articles from PLoS ONE are provided here courtesy of PLOS

RESOURCES