Modeling migration in cell colonies in two and three dimensional substrates with varying stiffnesses
 M. Dudaie^{1},
 D. Weihs^{2},
 F. J. Vermolen^{3}Email author and
 A. Gefen^{1}
https://doi.org/10.1186/s4048201500059
© Dudaie et al. 2015
Received: 30 April 2015
Accepted: 14 December 2015
Published: 23 December 2015
Abstract
Mechanotaxis is the directed migration of a cell due to forces it senses from the substrate, which are caused mainly by the presence of other cells or by external traction forces. The resulting cell movement plays important biological roles for example in wound healing, the functions of the immune system, organogenesis and metastatic diseases. We present a model to simulate collective cell migration based on the forces that cells exert on elastic substrata. It characterizes the influence of cell and substrate stiffness on the collective migration of cells. The simulations initially represent a twodimensional (monolayer) problem, and are then extended to represent migration in a threedimensional extracellular matrix. The model is generic and can be utilized to study a variety of biological processes where migration occurs including tissue repair, cancer and infiltration of white blood cells to an infection site.
Keywords
Mechanotaxis Cells mechnotrasduction Extracellular matrix Simulation MetastasisBackground
The ability to move is an essential feature of living cells, without which many biological processes could not occur, including organ development and growth, wound healing and the normal immune response to infection. In the context of disease, understanding the migration behavior of cells is particularly important in the cascade leading to the formation of cancer metastases. Some of the key mechanisms that stimulate cell migration are chemotaxis (movement along a gradient of a soluble chemoattractant), haptotaxis (directional movement along a gradient of ECMbound chemoattractants) and mechanotaxis [cell mobility triggered by mechanical cues such as substrate stiffness gradients (durotaxis), or migration towards a mechanically strained area (tensotaxis)].
Tensotaxis occurs when a cell responds to signals resulting from mechanical strains induced in its substrate or changes in the substrate stiffness [1]. To understand cell migration as an outcome of mechanotaxis, and in particular regarding the behavior of multiple cells in cultures, a theory is needed to assess the influence of the cellular forces applied on the (extracellular) environment, the effects of cell proliferation and death, the interactions between the cells, as well as the elastic properties of the substrate. A mathematical framework can then serve to design experiments and extract additional information from experimental work [2, 3]. A model makes it possible to isolate controlling parameters and factors in the experiments, and a simulation enables the testing of the individual contributions of the parameters and sensitivity to changes in their values upon migration.
The aim of this study is to describe a novel cell migration model based on physical and analytical theories characterizing the connections and influences between a cell and its surroundings (neighboring cells and the extracellular environment). One of the goals is to produce a 3D model that lends itself to simulating the migration process of diverse cells and situations, thus providing better insights into cell communication, specifically as regards mechanotaxis.
Mechanotaxis
Cell movement occurs when the cell applies forces to the substrate (in two dimensions, 2D) or to the extracellular matrix or medium (in three dimensions, 3D). These localized forces can be sensed by neighboring cells [4–7] which, in turn, migrate in the general direction of these signals.
This type of communication can also cause a group of cells to move and migrate in a specific direction, for example to repair a wound as a result of woundsite contraction. The magnitude and transfer distance of the mechanical signal that takes place as cells deform their surroundings depend on the elastic modulus of the substrate. Several studies have shown that cells will migrate towards a stiffer substrate when situated on a soft one, and will tend to move upwards on a stiffness gradient. Winer et al. showed that cell stiffness can change as a function of the substrate stiffness [11]. The intensity of the mechanical signal defines the cell’s velocity and the overall rate of motion of the colony.
Effect of stiffness of the extracellular matrix
The extracellular matrix (ECM) is a noncellular structural component that exists between and around the cells, in all tissues and organs, and contains many fibrous proteins and polysaccharides [12]. The composition of ECM is unique to each tissue and produces a different range of elastic moduli, which define the density and spatial organization of the protein molecules. These longchained molecules provide the ECM with its elasticity and stiffness. The structure and mechanics of the ECM play a major role in cell migration, as many cells receive biochemical and biomechanical cues through it. It thus acts as a medium for celltocell communication.
Ng [13], Dufort [7], and Lu [14] among others found that cells, especially cancerous ones, show a preference for a stiffer ECM. Generally, cells will tend to migrate towards a stiffer ECM zone, a feature that is associated with wounds and tumors that are much stiffer than normal tissue. Studies have reported ECM elasticity changes as a function of the type of cells inhabiting the substrate. The elasticity of the cell can be changed depending on the substrate type [8, 15–18] which is exploited in experiments on the subject.
The cell life cycle
Another important factor in the migration process takes place as part of the cell life cycle. During proliferation or death (necrosis, anokis or apoptosis), the cells interact differently with the substrate. There are indications of mechanical interactions of cells with the ECM during proliferation or after cell death such as regulation of processes [19] or pattern forming [20]. The mechanotaxic behavior of cancer cells is unusual in that there is a higher proliferation to celldeath rate and a different elastic modulus. However, substrate stiffness and dimensionality can have opposite effects on cell spread and viability. These relations and effects can be explored with migration models and simulations.
Cell migration models
The present study considers cells that are migrating through a space as discrete objects, rather than considering cell densities treated by continuumscale models using partial differential equations. The approach where cells are treated as discrete objects falls within the class of cellbased models. Cellularbased modeling approaches can be classified into two important subclasses: cellularautomata models, where the cell shape and position is represented by ‘occupying’ or ‘not occupying’ certain control areas (or volumes) that are dictated by a discrete lattice. In this subclass, one can find the cellularPotts models by Glazier and Graner [21], Merks and Koolwijk [32] and van Oers et al. [33]. The work of Borau et al. [31] represents a 3D voxel FE model for a single cell in a probabilistic setting, where also the interaction between the migration and deformation of the cell and the deformation of the cell nucleus were also taken into account. In the second subclass, which will be considered in this study, cells are allowed to move continuously over a specified domain where the stiffness varies over the domain. This continuous cellular approach was developed in earlier studies by Byrne and Drasdo [34], Groh and Louis [35] and in Neilson et al. [36]. A review on particle methods applied to tumor growth and wound healing is given in Vermolen [37]. Various cell migration models with cell–cell contacts have been developed and focus mainly on group migration and single cell movement rather than interactions between (proximal or distant) neighboring cells. Today’s more powerful computers make it possible to run more simulations that control migration thorough various interactions, although most are based on 2D models [21–23]. Here by contrast, we model mechanical signals transmitted through the substrate to which the cell adheres that affect its rate of migration, orientation and directionality.
Several analytical models for cell mechanotaxis have been developed. Geris et al. [24] described a finite element model based on celltocell springlike interactions, with a drag force caused by the medium viscosity. ReinhartKing et al. [25] produced a video analysis of cell migration, and suggested a mean displacement coefficient as a function of time, speed and direction of persistence. In a model developed by Vermolen and Gefen [4], the cells influence one another by mechanosensing traction forces arising from cell movement. During motion, a cell exerts a pulling force on the substrate, making a small deformation that a neighboring cell can sense and respond to by moving accordingly. Our model extends Vermolen and Gefen’s 2D cell migration model to a 3D spatial simulation. In the present approach, the deformation of the cell is not taken into account and all cells are assumed to be spherical in the 3D simulations. It is important to note that cell morphology and stiffness changes are required during migration, and especially in 3D migration through narrow passages, as will occur e.g. during cancer cell invasion, see Dvir et al. [30]. The likelihood of cells crossing the passage will be affected by the cell nucleus deformability. This aspect of motion will require more detailed definition of the cell structure and dynamics and is outside the scope of the current manuscript. In Borau et al. [31], a modeling study has been developed for a single cell migrating and deforming in 3D, and their modeling incorporated the interaction with the deformation of the cell nucleus indeed. One can foresee that their approach can be integrated with ours to extend the modeling to simulate en mass migration (of cell populations) in 3D. The next section outlines the physical model. “Results and discussion” presents the simulation results for illustrative cases of cell clustering and cancer metastasis.
Methods
The physical model presented here is composed of an analytical model and an algorithm developed to perform the cell mechanotaxis simulations in 2D and 3D under conditions where the number and types of cells, the complex substrate and other factors vary. The analytical model is based on the Potts model which assumes that each cell interacts with its neighboring cells and the substrate. To mimic physical ground truth, the model parameters are based on experimental results, and on stochastic movement.
The analytical model
Consider a mechanotaxis situation between two cells; i.e., a mechanical signal between the cells caused by the movement of one of the cells, which exerts a traction force on the substrate and causes the second cell to move towards it.
To simplify the model and calculations, the cells are treated as circles or spheres with radius R. In addition, since the deformations caused by cell movement are small compared to the substrate thickness, the deformations are assumed to behave like a spring.
We ignore interactions with the ECM after death or while proliferating, due to the very short simulation time interval and length of these processes.
The dimension parameter, \(\alpha_{i}\), is defined by the cell viability and interaction with the ECM as: \(\alpha_{i} = \left( {\frac{{F_{i} }}{{\hat{F}}}} \right)^{2} \beta_{i} \frac{{R_{i}^{3} }}{f}\); f is the friction force and equals \(\mu F_{i}\). Thus \(\alpha_{i} = \beta_{i} \frac{{R_{i}^{3} }}{{\mu \hat{F}}}\), or 0 if the cell is not viable (cells do not move after death or negligibly while proliferating). We also define the mobility of the cell, \(\beta\); i.e., its ability to move as a function of its own elastic modulus and the substrate’s elastic modulus.
Cell migration is semirandom and not only determined by the environment. Randomness is thus introduced by defining two components of cell velocity: movement resulting from the cell surroundings; i.e. mechanotaxis, and a velocity vector randomized under the assumption that a cell would continue to migrate approximately at the same velocity with minor changes, which can be modeled by a normal distribution. We define a probability P _{mp} that in a certain timeframe (TF) the cell velocity will derive from either mechanotaxis or from this random walk.
Simulation algorithm
List of parameters used in the simulation
Parameter  Value  Units 

E _{ s }  5  kPa 
E _{ c }  0.5  kPa 
F (cell force)  1  nN 
d (MDS length)  30  μm 
ε (MDS)  1.06 × 10^{−34}  Pa 
R (cell radius)  2  μm 
p (likelihood of death)  0.05 %  – 
q (likelihood of split)  0.5 %  – 
P_{mp} (random var.)  50 %  – 
β _{ i }/μ  10/0.2  s^{−1} 
Δt (Time step)  2  s 
Probabilities of proliferating or death were calculated to trigger these processes. Cancerous cells were simulated by setting the likelihood of proliferation higher than the likelihood of death to generate a tumor growth model.
This simulation, by controlling each cell E _{ c }, the size and traction force and the ECM elastic modulus value, makes it possible to examine numerous conditions and variables to investigate mechanotaxisrelated phenomena and effects.
Results and discussion

The two cell problem (2D).

Different cells with different ECM (2D + 3D).

Tumor growth.
Other situations can also be modeled using this simulation such as wound healing, apoptosis, etc.
The two cell problem
Let us review the simple problem of two cells on the same ECM. Placing two cells at a distance from each other, but within a detection range d, will result in the cells moving towards each other.
Figure 6b depicts the same process, with the addition of stochastic movement, for different P_{mp} (E_{s} = 10 kPa). It shows the time difference between cells that move mostly by themselves (P_{mp} = 0.9) and cells moving under mechanotaxis effects (P_{mp} = 0 or 0.1).
For both simulations, the starting distance was chosen to be 9.6 μm. A larger distance or a change in the ECM elastic modulus will make the signal weaker than the MDS and the cell will wander (solid and dashed lines in Fig. 6a). An analytical solution to this problem is presented in [4].
These results illustrate the basic movement in the model, and the conflicts between mechanotaxis versus other types of movement (stochastic in our example) and between different elastic moduli of the substrate. Changing the cells type will change the cell velocities, according to Eq. (10).
Two cells, two substrates
This behavior can be accounted for by looking at Eqs. 10 and 12; cells on the lower E _{ s } side (left side on Figs. 7, 8) have greater velocities than cells on the higher E _{ s } side; hence, there is a greater likelihood for these cells to migrate to the higher E _{ s } side. Furthermore, cells on the lower E _{ s } will have a higher maximal distance for MDS (d) than cells on higher E _{ s }, since these cells receive more signals from the other zone and will be drawn over.
One half consisted of cells with a constant E _{ c } of 0.5 kPa, and the other half was made up of cells with E _{ c } ranging from 0.1 to 1.1 kPa. The cells ‘preferred’ to drift towards the stiffer cell zone with little influence of substrate stiffness. Note that the effectiveness of E _{ s } virtually vanishes when dealing with large E _{ c } cells.
Metastasis
One of the most interesting situations that this simulation can reflect is metastasis, where a cell becomes malignant and reproduces faster until it extends to another region. In terms of mechanotaxis, cells with different amounts of stiffness migrate towards another ECM region.
Figure 12 shows the attraction of the tumor towards the upper tissue. Tweaking with the elasticity values can produce a full metastasis simulation. Note that both q (likelihood of proliferation) and p (likelihood of death) play a crucial role.
Cell death (black cells in the simulation) impacts the balance of the proliferation rate, and can potentially influence the migration direction.
Discussion
We presented a novel mechanotaxis model and simulation results. Cell and substrate elastic moduli were shown to have a considerable influence on the way that cells are able to move (movement rate and direction) as well as on culture migration behavior, which can start with only one cell. To understand the parameters controlling mechanotaxis, several simulations were run in order to disentangle these parameters. The data show that the ratio between the elastic modulus of the substrate and the cell, \(\lambda\), primarily influences the cells’ ability to migrate.
As shown in Eq. 10, the cell’s velocity is defined by \(\lambda\), and by selecting this correctly, we can manipulate cell movement on the basis of their elasticity, and culture them on a chosen substrate. \(\lambda\) also affects the migration time, by making the length of the simulation timestep an important parameter. Another effect of \(\lambda\) is the directivity of the migration. The results of the two elastic region simulations (“Two cells, two substrates”) show that stiffer cells produce a “louder” signal and softer cells will migrate towards them. A stiffer substrate will also influence cell movement, and cells will often tend to migrate towards the stiffer side of the substrate. An in vitro experiment [6] motivated the original assumption.
The model incorporated stochastic behavior to make the simulation more realistic and enable spontaneous migration to take place. Without free movement, the cells would be encapsulated in one region without migration. Enabling stochastic movement lengthens the simulation time (Fig. 6), and the ratio between mechanotaxis and selfinflicted movement (P_{mp}) needs to be calibrated for the simulation to mimic reality. Running the simulation in 2D or 3D alters the length of the simulation as a result of the number of cells and the third degree of freedom of movement.
Proliferation and cell death can affect simulation duration and migration behavior, depending on their rates. As cells multiply more often, a cluster of cells can separate from the main tissue and migrate towards a stiffer region, mimicking the metastasis process (Fig. 12). Here, we chose to manipulate the cells one cell at a time, in an attempt to simulate the connections between the cells, and the response of each cell to the signals it receives individually. Therefore the runtime of this simulation was dependent to a great extent on the number of cells (and hence on the proliferation and celldeath rate), and on the elastic modulus, through the threshold parameter \(\varepsilon\). Running this simulation on a powerful computer, coding it in another language such as C or C++ or running the program in a parallel computational environment such as a GPU (Graphical Processing Unit) would considerably shorten the runtime and make it possible to investigate more complex situations and different parameters.
Conclusion
The model introduced here focused on cell mechanotaxis, to analyze the connection and influence between the cellular substrate, the extracellular matrix and cell stiffness. Our goal was to assess collective effects and migration patterns of celltocell “mechanical communication”.
The findings indicate that cells tend to migrate to stiffer regions. The rate of migration is dependent on cell elasticity and the ability to move freely. These findings have implications for tissue growth, since the growth rate can be altered. We also showed that the model can simulate other biological processes such as tumor development and metastasis, where cell migration plays an important role. For instance, this could be applied to develop medication for faster wound healing or inhibiting migration to prevent metastasis from forming.
This framework can easily be extended to incorporate different types, sizes and shapes of cells, migration mechanisms such as chemotaxis, responses to obstacles, resource scavenging, etc., to investigate their affects on migration process.
Declarations
Authors’ contributions
MD performed the implementation of the modeling into Matlab, he also did the computations, and took the lead in the writing of the manuscript. AG and FJV took part of the supervision of the project and in the writing process. DW revised the manuscript critically and she added important biological context to the results that were obtained. All authors read and approved the final manuscript.
Acknowledgements
MD and AG acknowledge the financial support of the Tel Aviv University. FJV thanks the Delft University of Technology for the financial support. Finally, DW thanks Technion for the funding.
Competing interests
Two of the authors, F. J. Vermolen and D. Weihs are the editors in chief of this journal.
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.
Authors’ Affiliations
References
 Pleham RJ Jr, Wang YL. Cell locomotion and focal adhesions are regulated by substrate flexibility. Proc Natl Acad Sci. 1997;94:13661–5.View ArticleGoogle Scholar
 Yeung T, Georges PC, Flanagan LA, Marg B, Ortiz M, Funaki M, Zahir N, Ming W, Weaver VM, Janmey PA. Effects of substrate stiffness on cell morphology, cytoskeletal structure and adhesion. Cell Motil Cytoskelet. 2005;60:24–34.View ArticleGoogle Scholar
 Califano JP, ReinhartKing CA. Substrate stiffness and cell area predict cellular traction stresses in single cells and cells in contact. Cell Mol Bioeng. 2011;3(1):68–75.View ArticleGoogle Scholar
 Vermolen FJ, Gefen A. A semistochastic cellbased formalism to model the dynamics of migration of cells in colonies. Berlin: Springer; 2011.Google Scholar
 Throm Quinlan AM, Sierad LN, Capulli AK, Firstenberg LE, Billiar KL. Combining dynamic stretch and tunable stiffness to probe cell mechanobiology in vitro. Plos One. 2011;6(8):e23272.Google Scholar
 Dembo M, Wang YL. Stresses at the celltosubstrate interface during locomotion of fibroblasts. Biophys J. 1999;76:2307–16.Google Scholar
 Dufort CC, Paszek MJ, Weaver VM. Balancing Forces: architectural control of mechanotransduction. Mol Cell Biol. 2011;12:308–19.Google Scholar
 Paszek MJ, Zahir N, Johnson KR, Lakins JN, Rozenbers GI, Gefen A, ReinhartKing CA, Margulies SS, Debmo M, Boettiger D, Hammer DA, Weaver VM. Tensional homeostasis and the malignant phenotype. Cancer Cell. 2005;8:241–54.Google Scholar
 Ananthakrishnan R, Ehrlicher A. The forces behind cell movement. Int J Biol Sci. 2007;3:303–17.View ArticleGoogle Scholar
 Mogilner A, Oster G. Cell motility driven by actin polymerization. Biophys J. 1996;71:3030–45.View ArticleGoogle Scholar
 Winer JP, Chopra A, Kresh JY, Janmey PA. Mechanobiology of cell–cell and cell–matrix interactions. Ch. 2. New York: Springer; 2011. p. 11–22.Google Scholar
 Frantz C, Stewart KM, Weaver VM. The extracellular matrix at a glance. J Cell Sci. 2010;123:4196–200.View ArticleGoogle Scholar
 Ng MR, Brugge JS. A stiff blow from the stroma: collagen crosslinking drives tumor progression. Cancer Cell. 2009;16:455–7.View ArticleGoogle Scholar
 Lu P, Weaver VM, Werb Z. The extracellular matrix: a dynamic niche in cancer progression. JCB. 2012;196(4):395–406.View ArticleGoogle Scholar
 Geris L, Gersich A, Schugart RC. Mathematical modeling in wound healing, bone regeneration and tissue engineering. Acta Biotheor. 2010;58:355–67.View ArticleGoogle Scholar
 Lo CM, Wang HB, Dembo M, Wang YL. Cell movement is guided by the rigidity of the substrate. Biophys J. 2000;79:144–52.View ArticleGoogle Scholar
 Lopez JI, Mouw JK, Weaver VM. Biomechanical regulation of cell orientation and fate. Oncogene. 2008;27:6981–93.View ArticleGoogle Scholar
 Merkel R, Kirchgeßner N, Cesa CM, Hoffmann B. Cell force microscopy on elastic layers of finite thickness. Biophys J. 2007;93:3314–23.View ArticleGoogle Scholar
 Asally M, Kittisopikul M, Rué O, Du Y, Hu Z, Çağatay T, Robinson AB, Lu H, GarciaOjalvo J, Süel GM. Localized cell death focuses mechanical forces during 3D patterning in a biofilm. PNAS. 2012;109(46):18891–6.View ArticleGoogle Scholar
 Santini MT, Rainaldi G, Indovina PL. Apoptosis, cell adhesion and the extracellular matrix in the threedimensional growth of multicellular tumor spheroids. Oncology/Hematology. 2000;36:75–87.View ArticleGoogle Scholar
 Graner F, Glazier JA. Simulation of biological cell sorting using a twodimensional extended potts model. Phys Rev Lett. 1992;69(13):2013–6.View ArticleGoogle Scholar
 Pompe T, Kaufmann M, Kasimir M, Johne S, Glorius S, Renner L, Bobeth M, Pompe W, Werner C. Friction controlled traction force in cell adhesion. Biophys J. 2011;101:1863–70.View ArticleGoogle Scholar
 Sakamoto Y, Prudhomme S, Zaman MH. Viscoelastic gelstrip model for the simulation of migrating cells. ICES Report; 2011.Google Scholar
 Geris L, Liedekerke PV, Smeets B, Tijskens E, Ramon H. A cell based modeling framework for skeletal tissue engineering applications. J Biomech. 2009;43:887–92.View ArticleGoogle Scholar
 RienhartKing C, Dembo M, Hammer DA. Cell–Cell mechanical communication through compliant substrates. Biophys J. 2008;95:6044–51.View ArticleGoogle Scholar
 Discher D, Janmey P, Wang YL. Tissue cells feel and respond to the stiffness of their substrate. Science. 2005;310:1139–43.View ArticleGoogle Scholar
 Palecek SP, Loftus JC, Ginsberg MH, Lauffenburger DA, Horwitz AF. Integrinligand binding properties govern cell migration speed through cellsubstratum adhesiveness. Nature. 1997;385(6):537–40.View ArticleGoogle Scholar
 Gefen A. Effects of virus size and cell stiffness on forces, work and pressures driving membrane investigation in a recptormediated endocytosis. J Biomech Eng. 2010;132:08450114.Google Scholar
 Johnson KL. Contact mechanics. Cambridge: Cambridge University Press; 1985.View ArticleMATHGoogle Scholar
 Dvir L, Nissim R, AlvarezElizondo MB, Weihs D. Quantitative measures to reveal coordinated cytoskeletonnucleus reorganization during in vitro invasion of cancer cells. N J Phys. 2015;17:143010.View ArticleGoogle Scholar
 Borau C, Polacheck WJ, Kamm RD, GarciaAznar JM. Probabilistic voxelFE model for single cell motility in 3D. In Silico Cell Tissue Sci. 2014;1(2).Google Scholar
 Merks RMH, Koolwijk P. Modeling morphogenesis in silico and in vitro: towards quantitative, predictive, cellbased modeling. Math Model Nat Phenom. 2009;4(4):149–71.MathSciNetView ArticleMATHGoogle Scholar
 van Oers RFM, Rens EG, LaValley DJ, ReinhartKing CA, Merks RMH. Mechanical cellmatrix feedback explains pairwise and collective endothelial cell behavior in vitro. PLoS Comput Biol. 2014;10(3):e1003774.View ArticleGoogle Scholar
 Byrne H, Drasdo D. Individualbased and continuum models of growing cell populations: a comparison. J Math Biol. 2009;58:657–87.MathSciNetView ArticleMATHGoogle Scholar
 Groh A, Louis AK. Stochastic modeling of biased cell migration and collagen matrix modification. J Math Biol. 2010;61:617–47.MathSciNetView ArticleMATHGoogle Scholar
 Neilson MP, MacKenzie JA, Webb SD, Insall RH. Modeling cell movement and chemotaxis using pseudopodbased feedback. SIAM J Sci Comput. 2011;33(3):1035–57.MathSciNetView ArticleMATHGoogle Scholar
 Vermolen FJ. Particle methods to solve modeling problems in wound healing and tumor growth. Comput Part Mech. 2015; doi:10.1007/s4057101500556.Google Scholar