Mathematical modeling of Myosin induced bistability of Lamellipodial fragments
 881 Downloads
Abstract
For various cell types and for lamellipodial fragments on flat surfaces, externally induced and spontaneous transitions between symmetric nonmoving states and polarized migration have been observed. This behavior is indicative of bistability of the cytoskeleton dynamics. In this work, the Filament Based Lamellipodium Model (FBLM), a twodimensional, anisotropic, twophase continuum model for the dynamics of the actin filament network in lamellipodia, is extended by a new description of actin–myosin interaction. For appropriately chosen parameter values, the resulting model has bistable dynamics with stable states showing the qualitative features observed in experiments. This is demonstrated by numerical simulations and by an analysis of a strongly simplified version of the FBLM with rigid filaments and planar lamellipodia at the cell front and rear.
Keywords
Actin Mathematical model CytoskeletonMathematics Subject Classification
35Q92 34D20 92C17 00A721 Introduction
In a variety of physiological processes such as wound healing, immune response, or embryonic development, crawling cells play a vital role (Ananthakrishnan and Ehrlicher 2007). Cell motility is the result of an interplay between protrusion at the ’front’ edge of the cell (w.r.t. the direction of movement), retraction at the rear, as well as translocation of the cell body (Small and Resch 2005). It only occurs when the cell is polarized with a front and a differently shaped rear (Kozlov and Mogilner 2007).
Both protrusion and retraction involve the socalled lamellipodium, a thin, sheetlike structure along the perimeter of a cell, consisting of a meshwork of actin filaments. Factin is a polar dimer that forms inextensible filaments with a fastgrowing plus (barbed) end and a slowgrowing minus (pointed) end (Holmes et al. 1990).
The barbed ends abut on the membrane at the leading edge (Mogilner 2009) and have a high probability of polymerization (i.e. elongation of the filament by insertion of new actin monomers), whereas at the pointed ends mostly depolymerization (removal of one monomer) or disassembly of larger parts through severing of the filament occurs. Once a balance between polymerization and depolymerization is reached, each incorporated monomer is being pushed back by newly added monomers. Using the filament itself as a frame of reference, this can be described as movement of monomers from the barbed end towards the pointed end, a process called treadmilling (see Manhart et al. 2015a and the references therein for an overview of the involved processes and proteins). New filaments are nucleated predominantely by branching off existing filaments. The resulting meshwork is an (almost) twodimensional array of (almost) diagonally arranged actin filaments with decreasing density towards the cell body (Small et al. 1995; Vinzenz 2012).
The lamellipodium is stabilized by the cell membrane (surrounding the entire cell Mitchison and Cramer 1996; Vallotton et al. 2005), adhesions to the substrate (Li et al. 2003; Pierini et al. 2000), crosslinking proteins (Nakamura et al. 2007; Schwaiger et al. 2004) and myosin II filaments (Svitkina et al. 1997), the latter two binding to pairs of filaments. Some of the long filaments from the lamellipodium extend into the region behind, where (through the contractile effect of myosin II) forces are generated which pull the lamellipodium backwards (Small and Resch 2005).
Fish epidermal keratocytes are fastmoving cells with a relatively simple shape (circular, when stationary and crescentmoonshaped, when moving Lee et al. 1993), which makes them ideal subjects for analysis. Furthermore, they exhibit a lamellipodium with a smooth edge and a fairly uniform distribution of filaments (Lacayo et al. 2007; Small and Resch 2005; Theriot and Mitchison 1991). During the transition from the stationary to the moving state, the lamellipodium in the rear of the cell collapses and the rear bundle is formed, where myosin II generates a contractile force (Svitkina et al. 1997; Tojkander et al. 2012; Verkhovsky et al. 1997).
Various approaches to continuum mechanical modeling of the lamellipodium exist (see e.g. Kruse et al. 2006; Rubinstein et al. 2005). Bistability results similar to this work have been obtained in Ziebert et al. (2012), where a phasefield approach is used to describe the interplay between the cell shape, the mean orientation of the filaments within the network and the actin–myosin interaction. A strongly simplified model is developed in Kozlov and Mogilner (2007), where bistability has been obtained analytically as the consequence of the properties of a free energy functional containing contributions from the lamellipodium and a possible rear bundle.
This work is based on the FBLM (Manhart et al. 2015a; Ölz and Schmeiser 2010a, b), a twodimensional, anisotropic, twophase model derived from a microscopic (i.e. individual filament based) description, accounting for most of the phenomena mentioned above. It describes the actin network in terms of two transversal families of locally parallel filaments, stabilized by transient crosslinks and substrate adhesions. In Sect. 2 the FBLM is presented and extended by a model for actin–myosin interaction between the two families. We assume that myosin filaments can connect only when the families are antiparallel enough and they are described as transient, similar to crosslinks. They tend to slide the two families relative to each other, and they are assumed to have a turning effect, making the two families more antiparallel. The derivation of the additional myosin terms is presented in detail. Readers not consulting (Manhart et al. 2015a; Ölz and Schmeiser 2010a, b) may take this as representative also for the derivation of the adhesion and crosslink models.
The properties of the actin–myosin model are expected to produce the desired bistable behavior. This is demonstrated by numerical simulations in Sect. 8, which indicate the existence of two stable states, a rotationally symmetric nonmoving state and a polarized state, where the cell moves. The moving state is characterized by a more antiparallel network in the rear of the cell, where actin–myosin interaction is active. Complete collapse of the network and consequential generation of a rear bundle are avoided, since the FBLM is (so far) unable to deal with such topological changes.
The occurrence of bistability is also proven analytically for a strongly simplified model. In Sect. 3 the complexity of the model is reduced in a first step by assuming rigid filaments. Then a planar, translationally invariant lamellipodium is considered in Sect. 4, which reduces the model to a system of three ordinary differential equations. Here we also neglect the effects of branching and capping, assumed to be in equilibrium, as well as filament severing within the modelled part of the lamellipodium, implying a constant actin density there. Bistability is obtained for this model in Sect. 5. Finally, in Sect. 6 a cell (fragment) is replaced by a pair of connected backtoback planar lamellipodia, and the existence of stable stationary (symmetric) as well as moving (polarized) states is proven. The same bistable behavior is observed in the simulations of the full model in Sect. 8.
Figure 1 depicts the main components of the simplified version of the FBLM (D and E) together with one keratocyte and three fragments (A–C). The crescentmoon shaped cells and cell fragments are moving, whereas the circularly shaped fragment remains stationary. One can also observe that in moving fragments myosin can predominantely be found at the cell rear. In Fig. 1e, the idealized model obtained in Sects. 3, 4, 5, 6 is illustrated. It can be interpreted as description of lamellipodial sections at the front and at the rear of the cell. The main model ingredients are depicted in Fig. 1d: (e) diagonally arranged filaments (red) together with the membrane (dark green) and arrows indicating inward pulling forces due to stress fibers in the interior of the cell (dashed green line), (d) the cell membrane (green, with arrows indicating the force acting on the barbed ends due to membrane tension), (a) crosslinks [blue, producing friction between the filament families and a turning force trying to establish an equilibrium angle—arrows show the forces acting on the cross links due to resistance to stretching and bending above or below a certain angle (visualized by the black dashed line)], (b) myosin filaments (pink, trying to slide the filament families and to make them antiparallel—straight arrows indicate that myosin moves towards the plus ends along both filaments while curved arrows illustrate the tendency to establish an angle of \(180^\circ \) between the filaments, analogously to the crosslinks), (c) adhesions (yellow, connecting a filament through the membrane with the substrate thus producing friction relative to the substrate).
2 Adding actin–myosin interaction to the Filament Based Lamellipodium Model (FBLM)
The terms in the first line of (1) correspond to the filaments’ resistance against bending with stiffness parameter \(\mu ^B\), to friction relative to the substrate as a consequence of adhesion dynamics with adhesion coefficient \(\mu ^A\), and to the constraint (2) with the Lagrange multiplier \(\lambda _\text {inext}\).

Myosin filaments connect pairs of crossing actin filaments.

Due to the motor activity of the myosin heads they “walk” towards the barbed ends of actin filaments with a fixed speed \(v^M\).

The rates of creation and breakage of the myosin connections depend on the forces acting on them (see below).

The connections exert twisting forces on the connected actin filaments towards the equilibrium angle \(\pi \), i.e. the antiparallel state. These forces are caused by the stiffness of myosin filaments.

The connections can be stretched against an elastic restoring force.
Step 2: Scaling The key scaling assumption of the original FBLM is that the average lifetime of a crosslink or adhesion is small compared to the average time a monomer spends inside a filament, a ratio denoted by \(\varepsilon \). In the limit \(\varepsilon \rightarrow 0\) the nonlocality in time is removed from the problem. Furthermore this reduces the effect of adhesions to friction with the substrate and that of the crosslinkers to friction between the filament families (compare Eq. (1)).
3 Rigid actin filaments in the limit of large bending stiffness
We want to derive a simplified model with rigid actin filaments. This is motivated on the one hand by the observation that filaments within the lamellipodium are typically rather straight (Vinzenz 2012). On the other hand stiff filaments can be interpreted as a description of only the outermost part of the lamellipodial region, where filaments are (locally) straight. The resulting model is mathematically much simpler and can be derived by assuming a relatively large bending stiffness \(\mu ^B\). The limit \(\mu ^B\rightarrow \infty \) will be carried out formally in this section.
4 A geometric simplification: the planar lamellipodium
Since in keratocytes the leading edge is rather smooth, we approximate a piece of lamellipodium by an infinite strip, parallel to the xaxis, and invariant to translations and to reflection. For the given data this means that the maximal filament length L and the polymerization speed v are constants. As a further simplification, we assume no filament ends inside the modeled part of the lamellipodium with the consequence \(\eta = 1\) (and \(s_0=L/2\)).
5 Forces at the filament endssteady protrusion
Lemma 1
Theorem 2
6 Coupling of two opposing lamellipodia–bistability
Theorem 3
Proof
7 Parameter dependencies
The simplifications of the two preceding sections lead to several (testable) statements. This section can be seen as a discussion of the results so far.
Parameter values
Var.  Meaning  Value  Comment 

L  Filament length  \(8\,\mathrm{\upmu m}\)  Order of magnitude as in Verkhovsky et al. (1999) 
\(A_0\)  Equilibrium inner area  \(300\,\mathrm{\upmu m}^2\)  Order of magnitude as in Verkhovsky et al. (1999) 
N  Total filament number  9000  Order of magnitude as in Koestler et al. (2008) 
\(\mu ^B\)  Bending elasticity  \(0.7\,\mathrm{pN\,\upmu m}^2\)  10 times higher than in Gittes et al. (1993) 
\(\mu ^A\)  Macroscopic friction caused by adhesions  \(0.14\, \text {pN min} \,\mathrm{\upmu m}^{2}\)  Measurements in Li et al. (2003), Oberhauser et al. (2002), estimation and calculations in Ölz et al. (2008), Ölz and Schmeiser (2010a, b) 
v  Polymerization speed  \(3\,\mathrm{\upmu m}\,\text {min}^{1}\)  In biological range 
\(\varphi _0\)  Equilibrium crosslink angle  \(70^\circ \)  Equal to the branching angle 
\(\mu ^S\)  Crosslink stretching constant  \(4.2\times 10^{3}\, \mathrm{pN\, min}\, \,\mathrm{\upmu m}^{1} \)  
\(\mu ^T\)  Crosslink twisting constant  \(4.2\times 10^{3}\, \mathrm{pN\,\,\upmu m}\)  
\(v^M\)  Myosin velocity  \(1\,\mathrm{\upmu m}\,\text {min}^{1}\)  Order of magnitudes as in Svitkina et al. (1997) 
\(\overline{\varphi }\)  Myosin cutoff  \(100^{\circ }\)  
\(\overline{\mu ^{SM}}\)  Myosin stretching constant  \(4.2\times 10^{3}\, \text {pN min}\, \,\mathrm{\upmu m}^{1} \)  
\(\overline{\mu ^{TM}}\)  Myosin twisting constant  \(1.4 \times 10^{2}\,\mathrm{pN\,\upmu m}\)  Motivated by Lemma 1, Simulation 1 
\(1.8 \times 10^{2}\,\mathrm{pN}\,\mathrm{\upmu m}\)  Motivated by Lemma 1, Simulation 2  
\(\mu ^\text {stress}\)  Stress fiber force  \(5\times 10^{2} \,\mathrm{pN \upmu m}^{1} \) 
8 Simulations with the full model
In this section we demonstrate that with the additional term describing myosin within the lamellipodium, the model is able to produce cells/cell fragments that, depending on the initial conditions, will either remain stationary or start moving. In contrast to the simulations presented in Manhart et al. (2015a, b), here the movement is achieved without a continuing external signal and without varying the polymerization speed. In the simulation, we work with the full model (6)–(8) and not with the simplifications introduced in Sects. 3 and 4. However, the qualitative results of Sect. 6 will be reproduced.
Simulation results Figure 3 shows the evolution of the cell in two different numerical experiments over a timespan of 15 min. In both cases identical initial conditions were used, in which the left side of the cell is deformed. This could be caused either by internal fluctuations or by a pushing force from the left, two situations known to lead to symmetry breaking in fragments (Verkhovsky et al. 1999). Both simulations use the same set of parameters, with the exception of the value of the myosin twisting constant, \(\mu ^{TM}\), which is smaller in the left column than in the right column. In Fig. 3a–d it can be observed that even though the filaments on the left initially lie antiparallel enough for myosin to act on them, myosin does not establish itself there permanently. Lemma 1 gives a possible explanation for this: If myosin is too weak compared to the crosslinks, there is no myosin/crosslinkequilibrium, hence the cell reverts back to a purely crosslink dominated state. This situation corresponds to the steady state (21) in Theorem 3.
The case of a larger myosin twisting constant is depicted in Fig. 3e–h. In situation case a bundleprecursor is formed on the left, whereas the right stays myosin free. This allows the cell to change to a moving steady state (see the Supplementary Material for a movie) with a collapsed lamellipodium at the back and a noncollapsed lamellipodium at the front, a situation described by the steady states (23) and (24) in Theorem 3. One can also observe the contraction of the rear bundle leading to a more halfmoon shaped cell. Clearly if the initial deformations are so small that no myosin can attach, the cell always reverts back to the stationary state. This refers to a situation where the fluctuations or pushing force are too small to cause symmetry breaking.
9 Discussion and outlook
In this work, we extended the FBLM introduced in Manhart et al. (2015a), Ölz and Schmeiser (2010a) by a description of actin–myosin interaction. The limit of large bending stiffness led to a simplified model for rigid filaments. The additional simplification of a planar lamellipodium reduces the model to a small ODE system for the center of mass of a reference filament and the angle with the membrane. A caricature of a cell fragment has been described by considering two versions of the model coupled by membrane and stress fiber forces. Bistability has been shown in the asymptotic regime of relatively small coupling forces and adhesion strength. Since these results are almost explicit, various parameter dependencies could be obtained. Furthermore, we carried out numerical simulations based on the full FBLM supporting the main analytical result of bistable behaviour.
Our model is able to qualitatively reproduce the observed bistability of cells and cell fragments (Verkhovsky et al. 1999). Since myosin has been shown to be effective only if the angle between the filaments is large enough (Reymann 2012), our specific modeling of the myosin effect to be angledependent seems to be a reasonable choice.
In the numerical simulations we have shown that for an initially slightly asymmetric cell, one of two stable steady states is attained, depending on the parameters for myosin and crosslinks. If myosin is too weak to exert considerable twisting forces, the effects from crosslinks on the local angles between the filaments dominate and the cell reverts back to a symmetric nonmoving shape. On the other hand, if myosin forces are strong enough, a bundle precursor forms at the ’rear’ (the location of initial asymmetry), and the cell starts moving. What is apparent in these simulations is that myosin tends to locate at the back of the lamellipodium (away from the membrane). This is also in good agreement with experimental findings. It is remarkable that both for the analytical setting as well as in the numerical simulations, movement is an selforganized behavior, once an initial asymmetry has been established. It is known that keratocytes exhibit little if any chemotaxis, i.e. directed movement to outward cues.
An important issue remains to be addressed: The analytical model, dealing only with pieces of lamellipodium at the front and at the rear of the cell, avoids the transition zone separating the two parts of the lamellipodium with an intact network on the one hand and, on the other hand, compression to a rear bundle. It can be expected that due to lateral flow, actin filaments are drawn into the bundle in these transition regions. However, we expect that this happens with the pointed ends first entering the bundle, as opposed to the simulations presented here. A model supporting the necessary change of orientation in the transition zone is not available so far, and its derivation is the subject of ongoing work.
Notes
Acknowledgments
Open access funding provided by Austrian Science Fund (FWF). Financial support by the Austrian Science Fund (FWF) through the doctoral school Dissipation and Dispersion in Nonlinear PDEs (Project W1245, S.H./A.M.) as well as the Vienna Science and Technology Fund (WWTF) (Project LS13/029, C.S.).
Supplementary material
Supplementary material 1 (avi 2632 KB)
References
 Ananthakrishnan R, Ehrlicher A (2007) The forces behind cell movement. Int J Biol Sci 3:303–317CrossRefGoogle Scholar
 Barnhart EL, Lee KC, Keren K, Mogilner A, Theriot J (2011) An adhesiondependent switch between mechanisms that determine motile cell shape. PLoS Biol 9:e1001059CrossRefGoogle Scholar
 Gittes F, Mickey B, Nettleton J, Howard J (1993) Flexural rigidity of microtubules and actin filaments measured from thermal fluctuations in shape. J Cell Biol 120:923–934CrossRefGoogle Scholar
 Holmes KC, Popp D, Gebhard W, Kabsch W (1990) Atomic model of the actin filament. Nature 347:44–49CrossRefGoogle Scholar
 Koestler SA, Auinger S, Vinzenz M, Rottner K, Small JV (2008) Differentially oriented populations of actin filaments generated in lamellipodia collaborate in pushing and pausing at the cell front. Nat Cell Biol 10:306–313CrossRefGoogle Scholar
 Kozlov MM, Mogilner A (2007) Model of polarization and bistability of cell fragments. Biophys J 93:3811–3819CrossRefGoogle Scholar
 Kruse K, Joanny JF, Julicher F, Prost P (2006) Contractility and retrograde flow in lamellipodium motion. Phys Biol 3:130–137CrossRefGoogle Scholar
 Lacayo CI, Pincus Z, VanDuijn MM, Wilson CA, Fletcher DA, Gertler FB, Mogilner A, Theriot JA (2007) Emergence of largescale cell morphology and movement from local actin filament growth dynamics. PLOS Biol 5:e233CrossRefGoogle Scholar
 Lee J, Ishihara A, Theriot J, Jacobson K (1993) Principles of locomotion for simpleshaped cells. Nature 362:167–171CrossRefGoogle Scholar
 Li F, Redick SD, Erickson HP, Moy VT (2003) Force measurements of the \(\alpha 5\beta 1\) integrin–bronectin interaction. Biophys J 84(2):1252–1262CrossRefGoogle Scholar
 Manhart A (2011) A Mathematical model of actinmyosin interaction and its application to Keratocyte movement. Masters thesis, University of ViennaGoogle Scholar
 Manhart A, Oelz D, Sfakianakis N, Schmeiser C (2015a) An extended Filament Based Lamellipodium model produces various moving cell shapes in the presence of chemotactic signals. J. Theor. Biol. 382, 244–258Google Scholar
 Manhart A, Oelz D, Sfakianakis N, Schmeiser C (2015b) Numerical treatment of the Filament Based Lamellipodium Model (FBLM) arXiv:1505.04266 (to appear)
 Mitchison T, Cramer L (1996) Actinbased cell motility and cell locomotion. Cell 84:371–379CrossRefGoogle Scholar
 Mogilner A (2009) Mathematics of cell motility: have we got its number? J Math Biol 58:105–134MathSciNetCrossRefMATHGoogle Scholar
 Nakamura F, Osborn TM, Hartemink CA, Hartwig JH, Stossel TP (2007) Structural basis of lamin A functions. J Cell Biol 179(5):1011–1025CrossRefGoogle Scholar
 Oberhauser AF, BadillaFernandez C, CarrionVazquez M, Fernandez JM (2002) The mechanical hierarchies of fibronectin observed with singlemolecule AFM. J Mol Biol 319(2):433–447CrossRefGoogle Scholar
 Ölz D, Schmeiser C (2010) How do cells move? Mathematical modelling of cytoskeleton dynamics and cell migration. In: Chauviere A, Preziosi L, Verdier C (eds) Cell mechanics: from single scalebased models to multiscale modelling. Chapman and Hall / CRC Press, LondonGoogle Scholar
 Ölz D, Schmeiser C (2010) Derivation of a model for symmetric lamellipodia with instantaneous crosslink turnover. Arch Ratio Mech Anal 198:963–980MathSciNetCrossRefMATHGoogle Scholar
 Ölz D, Schmeiser C, Small JV (2008) Modelling of the actin–cytoskeleton in symmetric lamellipodial fragments. Cell Adhes Migr 2:117–126CrossRefGoogle Scholar
 Pierini LM, Lawson MA, Eddy RJ, Hendey B, Maxeld FR (2000) Oriented endocytic recycling of \(\alpha \beta \)1 in motile neutrophils. Blood 95(8):2471–2480Google Scholar
 Reymann A et al (2012) Actin network architecture can determine myosin motor activity. Science 336:1310–1314CrossRefGoogle Scholar
 Rubinstein B, Jacobson K, Mogilner A (2005) Multiscale twodimensional modeling of a motile simple shaped cell. Multisc Model Simul 3:413–439MathSciNetCrossRefMATHGoogle Scholar
 Schwaiger I, Kardinal A, Schleicher M, Noegel A, Rief M (2004) A mechanical unfolding intermediate in an actincrosslinking protein. Nat Struct Mol Biol 11(1):81–85CrossRefGoogle Scholar
 Small JV, Herzog M, Anderson K (1995) Actin filament organization in the fish keratocyte lamellipodium. J Cell Biol 129:1275–1286CrossRefGoogle Scholar
 Small JV, Resch G (2005) The comings and goings of actin: coupling protrusion and retraction in cell motility. Curr Opinion Cell Biol 17:517–523CrossRefGoogle Scholar
 Svitkina TM, Verkhovsky AB, McQuade KM, Borisy GG (1997) Analysis of the actinmyosin II system in fish epidermal keratocytes: mechanism of cell body translocation. J Cell Biol 139(2):397–415CrossRefGoogle Scholar
 Theriot JA, Mitchison TJ (1991) Actin microfilament dynamics in locomoting cells. Nature 352:126–131CrossRefGoogle Scholar
 Tojkander S, Gateva G, Lappalainen P (2012) Actin stress fibers: assembly, dynamics and biological roles. J Cell Sci 125(8):1855–1864CrossRefGoogle Scholar
 Vallotton P, Danuser G, Bohnet S, Meister JJ, Verkhovsky AB (2005) Tracking retrograde flow in keratocytes: news from the front. Mol Biol of the Cell 16:1223–1231CrossRefGoogle Scholar
 Verkhovsky AB, Svitkina TM, Borisy GG (1997) Polarity sorting of actin filaments in cytochalasintreated fibroblast. J Cell Sci 110:1693–1704Google Scholar
 Verkhovsky AB, Svitkina TM, Borisy GG (1999) Selfpolarization and directional motility of cytoplasm. Curr Biol 9:11–20CrossRefGoogle Scholar
 Vinzenz M et al (2012) Actin branching in the initiation and maintenance of lamellipodia. J Cell Sci 125:2775–2785CrossRefGoogle Scholar
 Yam PT et al (2007) Actinmyosin network reorganization breaks symmetry at the cell rear to spontaneously initiate polarized cell motility. J Cell Biol 178:1207–1221CrossRefGoogle Scholar
 Ziebert F, Swaminathan S, Aranson IS (2012) Model for selfpolarization and motility of keratocyte fragments. J R Soc Interface 9:1084–1092CrossRefGoogle Scholar
Copyright information
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.