Annals of Biomedical Engineering, Vol. 33, No. 2, February 2005 (©2005) pp. 131–141 DOI: 10.1007/s10439-0058972-9 Computation of Adherent Cell Elasticity for Critical Cell-Bead Geometry in Magnetic Twisting Experiments JACQUES OHAYON and PHILIPPE TRACQUI Laboratoire TIMC-IMAG, équipe DynaCell, CNRS, UMR 5525, Institut de l’Ingénierie et de l’Information de Santé (In3 S), 38706 La Tronche cedex France (Received 26 March 2004; accepted 4 August 2004) Abstract—Quantification of the cell elastic modulus is a central issue of micromanipulation techniques used to analyze the mechanical properties of living adherent cells. In magnetic twisting cytometry (MTC), magnetic beads of radius R, linked to the cell cytoskeleton through transmembrane receptors, are twisted. The relationships between imposed external torque and measured resulting bead rotation or translation only provide values of the apparent cell stiffness. Thus, specific correcting coefficients have to be considered in order to derive the cell elastic modulus. This issue has been highlighted in previous studies, but general relationships for handling such corrections are still lacking while they could help to understand and reduce the large dispersion of the reported values of cell elastic modulus. This work establishes generalized abacuses of the correcting coefficients from which the Young’s modulus of a cell probed by MTC can be derived. Based on a 3D finite element analysis of an hyperelastic (neo-Hookean) cell, we show that the dimensionless ratio hu /2R, where hu is the cell height below the bead, is an essential parameter for quantification of the cell elasticity. This result could partly explain the still intriguing question of the large variation of measured elastic moduli with probe size. ei j h hu h∞ p u (Aα , Bα ) (Aβ , Bβ ) (Aαi , Bαi ) (Aβi , Bβi ) C (Cβ , Dβ ) Keywords—Cytomechanical model, Finite element method, Magnetic twisting cytometry, Correcting coefficients, Cell stiffness. (Cβi , Dβi ) Ecell G Gθ NOMENCLATURE ax , a y a1 d (e1 , e2 , e3 ) (er , e , eφ ) eeff side lengths of the representative cellular volume rheological constant of the neo-Hookean strain energy function lateral bead translation along the ox axis Cartesian unit base vectors associated to coordinates (x, y, z) spherical unit base vectors associated to coordinates (r, , φ) local effective strain field Gδ I I1 R Re T V W Address correspondence to Pr. J. Ohayon or Dr. Ph. Tracqui, Equipe Dynacell, Laboratoire TIMC-IMAG, CNRS, UMR 5525, Institut de l’Ingénierie et de l’Information de Santé (In3 S), 38706 La Tronche cedex, France. Electronic mail: [email protected]; [email protected] α components of the deviatoric strain field cell height distance between the bead bottom and the substrate infinite cell height (approximated in our computation by a 20-µm-high cell) Lagrangian multiplier resulting from the incompressibility constraint displacement vector functions of γ used for the fit of the correcting coefficient α functions of γ used for the fit of the correcting coefficient β coefficients of the polynomial functions Aα (γ ) and Bα (γ ) coefficients of the polynomial functions Aβ (γ ) and Bβ (γ ) right Cauchy-Green strain tensor functions of γ used for the fit of the correcting coefficient β coefficients of the polynomial functions Cβ (γ ) and Dβ (γ ) cell Young modulus used for linear material shear modulus for linear material apparent cell stiffness related to the bead rotation apparent cell stiffness related to the bead translation identity matrix first invariant of the right Cauchy-Green strain tensor C bead radius radius of a spherical external volume immersing the rigid bead external magnetic torque applied bead volume strain energy function defining the elastic properties of the cell medium correcting coefficient related to the apparent cell stiffness Gθ 131 C 2005 Biomedical Engineering Society 0090-6964/05/0200-0131/1 132 β γ δ ε θ θs λ σ ξ J. OHAYON and P. TRACQUI correcting coefficient related to the apparent cell stiffness Gδ half-angle of bead immersion or embedding angle normalized lateral bead translation along the ox axis (δ = d/R) linear lagrangian strain tensor of components εi j total bead rotation angle slipping bead rotation angle extension ratio Cauchy stress tensor of components σi j normalized ratio h u /2R INTRODUCTION In living tissues, cells are exposed to a variety of mechanical stresses and strains which determine cell shape. In return, cell deformability is a central feature of the cellular mechanochemical signaling pathways which control a variety of fundamental cell functions, including spreading, migration, proliferation and gene expression.6,11,15,20 In this context, the characterization of cell rheological properties, and especially the quantification of cell elastic modulus, are crucial for understanding both physiological and pathological cell behavior, as well as for evaluating the action of pharmacological drugs which potentially modify cell architecture and thus cell rheological properties.14 Among the different micromanipulation methods used to analyze cell mechanical response,19 magnetic twisting cytometry (MTC) has been widely used, since the pioneering work of Wang et al.26,27 to probe mechanical properties of adherent cells on various substrate and in different experimental situations.5,7,17,18 In the MTC method, ferromagnetic beads (3.5–5.5-µm diameters) are coated with arginine–glycine–aspartic acid (RGD) peptide in order to permit their binding to integrin transmembrane mechanoreceptors.23 After beads premagnetization using a short uniform magnetic pulse, a magnetic torque is then created by a uniform magnetic field.17,27 The resulting measured bead rotation or/and lateral bead translation reflect the mechanical resistance of the cell to the imposed deformation. Recently, this experimental approach also permits to distinguish between cortical and deep cytoskeleton responses.18 The so obtained experimental data allow the computation of the apparent cell stiffness, either from the ratio between the applied magnetic torque and the resulting angular rotation or from the ratio between the applied torque and the lateral bead displacement. Such quantifications appear quite appealing since they only involve one additional parameter, the bead radius.8,21 Nevertheless, Mijailovich et al.21 pointed out recently that such apparent cell stiffness may vary with cell height and bead embedding angle. Thus, in order to derive the intrinsic value of the cell elastic modulus, they computed a limited set of correcting coefficient for a given bead radius of 2.5 µm and considering three specific cell heights [20 µm (semi-infinite medium), 5 and 1 µm]. In this work, we extend and generalize the computation of the cell elastic modulus by establishing analytical expressions of the correcting coefficients in a whole range of experimental values taken by the bead embedding angle γ , cell height h, and bead radius. Additionally the mechanical response of deeply embedded bead (i.e. with an angle γ in the range 90◦ –180◦ ) is also analysed. As in the study of Caille et al.,1 we assumed the cell behaves as an hyperelastic (neo-Hookean) material which takes into account the contribution of the strain tensor nonlinear terms, while preserving a linear stress-strain relationship. We used a three-dimensional finite element method to extensively compute and analyze the cell mechanical response to applied torques for a large set (800 cases) of different cell geometries. Our goals are (i) to point out the most relevant cellular geometrical parameters in the computation of the correcting coefficients, and (ii) to establish a set of generalized abacuses from which the Young’s modulus of a cell probed by MTC can be computed, as well as critical values of the cell height below which correcting coefficients must necessary be considered. METHODS Finite Element Formulation of the Cell Model Cell Geometry The cell is represented by parallelepipedic volume element of height h with rectangular basal area of size ax × a y , where ax = 2a y = 36 µm. Considering for example a mean cell height h = 4 µm, we obtained a cellular volume of 2592 µm3 , similar to the measured volume of 2500 ± 300 µm3 obtained by Schneider et al.24 in dog kidney epithelial cells. A single bead is embedded in the middle of the upper surface of the cellular volume. Using the symmetrical plane passing across the center of the bead, we restricted the analysis to half the cellular volume (Fig. 1). The main parameters defining the cell-bead model geometry are the cell length a y , the cell height h, the bead radius R and the embedding angle γ (see Fig. 1). Experimental values of the bead embedding angles, measured with alveolar epithelial cell,17 range from 15◦ to 90◦ : we thus considered this latter value as a first step in our analysis, before considering larger bead embedding angles up to 180◦ . Finite Element Model The cellular medium is assumed to be an homogeneous quasi-incompressible and hyperelastic neo-Hookean continuum,12 characterized by a strain-energy function already used to model the mechanical response of tissues and living cells1 and given by W = a1 (I1 − 3) (1) Generalized Elastic Modulus Estimation of Adherent Cell where a1 is the cellular material constant (in Pa), while I1 is the first invariant of the right Cauchy–Green strain tensor C [I1 = Trace(C)].11 For this incompressible medium, the cell shear modulus G is related to the Young’s modulus Ecell by the equation E cell = 3 G. Furthermore, an additional explicit relationship between Ecell and the material constant a1 can be obtained for uniaxial tension or compression of the medium: in this case, E cell = 2a1 (2 + λ−3 ), where λ is the extension ratio in the direction of the uniaxial stress. Notice that for small extension ratio (λ ∼ 1), this expression reduces to E cell ∼ 6a1 . Computations of the bead rotations induced by increasing value of the applied magnetic torque were performed using a finite element approach (Ansys 8.0 software, Ansys, Inc., Cannonsburg, PA). The bead surface was modeled as a rigid shell since its stiffness is far larger than the cell stiffness. Boundary Conditions The boundary conditions were written on half of the cell volume. The conditions imposed at the cell boundaries are as follows (Fig. 1): (i) zero displacement, modeling full cell attachment to the substrate, was imposed to the cell membrane in contact with the rigid substrate, (ii) zero normal displacement condition was imposed on the cell section belonging to the plane of symmetry, (iii) free boundary conditions were assumed for all the other cell surfaces, (iv) no-slip condition at the bead-cell interface. 133 RESULTS In MTC experiments, the bead rotation induced by the applied magnetic torque deforms the adherent cell. Our finite element computations simulate the cell resistance to both bead rotation and bead lateral translation (the vertical translation toward the substrate remains negligible compared to the lateral translation). Thus, cell stiffness can be computed from the imposed torque and the resulting bead motion for a given cell-bead geometry, considering either bead rotation or bead lateral translation as described in the following paragraph. Background for the Calculation of Apparent Cell Stiffness from Bead Rotation or Bead Lateral Translation In the case of a bead embedded in an infinite linear elastic and incompressible medium, one can obtain an analytical relationship between the applied mechanical torque T and the resulting bead rotation θ, as a function of the medium shear modulus G and bead radius R (see Appendix): T = 6V θ G (2) where V is the bead volume. The apparent stiffness modulus of the cell is computed from either bead rotation θ or associated lateral bead translation d using the following relationships21 : G θ = T /6V θ (3) G δ = T /6V δ (4) where δ is the normalized bead translation (δ = d/R). In order to take into account the differences between the infinite medium case and the real cell-bead geometry, FIGURE 1. Finite element mesh of the half cell volume of size ax × ay/2 × h. The arrow indicates the orientation of the bead rotation and torque applied at the bead center. Free boundary conditions are considered, except when explicitly indicated in the figure. 134 J. OHAYON and P. TRACQUI several authors9,21 introduced two correcting coefficients α and β which incorporate the effect of the geometrical parameters γ , h, and R on the apparent cell stiffness estimations. The correcting coefficients α and β are defined by G θ = αG (5) G δ = βG (6) In the case of a bead embedded in an infinite medium (see Appendix), the correcting coefficient α is equal to 1, but because no bead translation occurs (δ = 0) the coefficient β tends to infinity. In real experiments, variations of the bead radius can significantly modify the mechanical response of adherent cells for low values of the cell height. It is thus of importance to study the sensitivity of the correcting coefficients values to modifications of the geometrical parameters and to explicit the expression of α and β as functions of γ , h, and R. Influence of the Bead Radius and Cell Height on the Cell Mechanical Response Spatial Distribution of Effective Strains with Variations of Bead Radius and Cell Height Figure 2 highlights the effects of increasing bead radius and decreasing cell height on the simulated spatial distri- butions of effective strains within the cell. In all cases, the same 15◦ bead rotation θ has been imposed, which corresponds to a torque amplitude (in pN µm) ranging from 1.65 Ecell [Fig. 2(A)] to 12.58 Ecell [Fig. 2(D)]. The bead embedding angle has been kept to a constant value γ taken as 65◦ . Figure 2 shows that decreasing cell height from 5 to 2 µm has no major effect on the effective strain values for bead radius of 1.5 µm [Fig. 2(A) and 2(C)]. This is no longer the case when considering bead radius of 2.5 µm: the maximum effective strain remains close to the previous values (56% compared to 48% for 1.5-µm bead radius) for 5-µm-high cell [Fig. 2(B)], but increases sharply to 80% for 2-µm-high cell [Fig. 2(D)]. This influence of the degree of bead embedding at low cell heights is more precisely analyzed in the next paragraph. Nonlinear Corrections of the Apparent Cell Stiffness with Variations of Bead Radius and Bead Embedding As in Mijailovich et al.,21 we first consider the percentage of bead diameter as the geometrical parameter characterizing bead embedding in the cell, while bead radius is taken as a second geometrical parameter. We then simulated the cell response for 800 different realistic cellbead geometries, each corresponding to a given value of R (1 µm ≤ R ≤ 5 µm), γ (15◦ ≤ γ ≤ 180◦ ), and h (1 µm ≤ FIGURE 2. 3D color maps showing the cell deformed shapes and the spatial distributions of the effective strains eeff in the neigh borhood of the bead (e eff = 2ei j ei j /3, with ei j the components of the deviatoric strain tensor). Four different cell-bead geometries have been considered for given values γ = 65◦ and ay = 18 µm. For each geometry, a bead rotation of 15◦ (0.262 rad) has been imposed, giving rise to a lateral bead translation d. Parameters and variables values are respectively (with Ecell in Pascals): (A) h = 5 µm, R = 1.5 µm, d = 0.222 µm, T = 1.651 Ecell pN µm; (B) h = 5 µm, R = 2.5 µm, d = 0.371 µm, T = 7.779 Ecell pN µm; (C) h = 2 µm, R = 1.5 µm, d = 0.208 µm, T = 1.828 Ecell pN µm; (D) h = 2 µm, R = 2.5 µm, d = 0.316 µm, T = 12.577 Ecell pN µm. The bead has been removed in order to visualize more clearly the spatial strain distributions in the neighborhood of the cell-bead contact area. Generalized Elastic Modulus Estimation of Adherent Cell FIGURE 3. Influence of the bead radius R and of the percentage of embedding angle on the correcting coefficient α values for two extreme cell heights h∞ = 20 µm (A) and h = 1 µm (B). Solid points correspond to iso-values of bead radius when the percentage of bead diameter is varied and to iso-values of bead embedding angle when bead radius is varied. h ≤ 20 µm). The associated correcting coefficient α is then computed from Eqs. (3) and (5) (Fig. 3). When the bead is embedded in a semi-infinite medium (approximated here by a 20-µm-high cell), the coefficient α does not change with the bead radius and becomes significantly a nonlinear function of the bead embedding angle for values larger than 90◦ [Fig. 3(A)]. This graph is drastically modified when the cell height decreases up to 1 µm: the correcting coefficient α now exhibits a sharp nonlinear variation with the bead embedding [Fig. 3(B)]. Similar results have been obtained for the correcting coefficient β (results not shown). Taken as a whole, these results indicate that a complete set of abacus for the functions α(γ , h, R) and β(γ , h, R) must be provided in order to estimate the intrinsic cell elastic modulus of adherent cell probed by MTC. Analytical Expressions of the Correcting Coefficients α and β In our simulations, the ratio ξ = h u /2R, i.e. cell height below the bead over bead diameter (Fig. 1), appears as a key 135 FIGURE 4. Best fits obtained for the two correcting functions α(hu /2R ) [(A), r 2 = 0.990] and β(hu /2R ) [(B), r 2 = 0.984] for the fixed value γ = 50◦ of the bead embedding angle. The dimensionless geometrical invariant hu /2R is the ratio of the cell height below the bead over the bead diameter. dimensionless geometrical parameter for the quantification of the cell response. This property is supported by the series of fits performed, at given embedding angles γ , against the computed values of α and β when the ratio ξ is increased. Based on experimental measurements22 of hu , ξ is taken in the range 0.1 ≤ ξ ≤ 10. A very satisfactory fit (exemplified in Fig. 4 when γi = 50◦ ) has been obtained when assuming for each embedding angle γ i hyperbolic-like relationships of the form Bα (γi ) α(γi , ξ ) = Aα (γi ) + (7) ξ β(γi , ξ ) = Aβ (γi ) + Bβ (γi ) + Cβ (γi )ξ 2 Dβ (γi ) + ξ (8) In a second step, we then try to find mathematical expressions of the functions Aα (γ ), Bα (γ ), Aβ (γ ), Bβ (γ ), Cβ (γ ), and Dβ (γ ), from which the correcting coefficients α(γ , ξ ) and β(γ , ξ ) can be continuously computed. We found that the best fit (Fig. 5) to the simulated cell response for the 800 geometries we considered is obtained when considering the following expressions: Aα (γ ) = Aα1 γ + Aα2 γ 2 + Aα3 γ 3 Bα (γ ) = Bα0 + Bα1 γ + Bα2 γ 2 + Bα3 γ 3 (9) (10) 136 J. OHAYON and P. TRACQUI FIGURE 5. Best mathematical fits of the six functions Aα (γ) [(A), r 2 = 0.999], Bα (γ) [(B), r 2 = 0.998], Aβ (γ) [(C), r 2 = 0.999 if 15◦ ≤γ≤90◦ and r 2 = 0.998 if 90◦ ≤γ≤180◦ ], Bβ (γ) [(D), r 2 = 0.998 if 15◦ ≤γ≤90◦ and r 2 = 0.999 if 90◦ ≤γ≤180◦ ], Cβ (γ) [(E), r 2 = 0.998] and Dβ (γ) [(F), r 2 = 0.999] given in Eqs. (9)–(14) and used for the estimation of the two correcting coefficient α and β [see Eqs. (7) and (8)]. Aβ (γ ) = Aβ0 + Aβ1 γ + Aβ2 γ 2 + Aβ3 γ 3 (11) Bβ (γ ) = Bβ0 + Bβ1 γ + Bβ2 γ 2 + Bβ3 γ 3 (12) Cβ (γ ) = Cβ0 + Cβ1 γ + Cβ2 γ (13) 2 Dβ (γ ) = Dβ0 e Dβ1 γ (14) An excellent goodness of fit has been obtained with numerical values of coefficients given in Table 1. Let us remark that in the limiting case where the bead-cell interface reduces to one contact point (i.e. when γ → 0◦ and h u → ∞), the correcting coefficients α and β tend to zero. (ii) by Laurent et al.17 when considering the approximated analytical solution α = sin3 (γ )/2 obtained in the case of a bead embedded in a semi-infinite medium. Figure 6(A) and 6(B) show that, for all considered embedding angles, our results are in very good agreement with those of Mijailovich et al.21 for the three cell heights considered in their analysis, namely h ∞ = 20 µm, h = 5 µm, and h = 1 µm. On the other hand, our results only agree with those of Laurent et al.17 for bead embedding angles smaller than 20◦ [see enlargement in Fig. 6(A)]: above this value, the approximation based on the sin3 (γ ) function gives relative error on α larger than 50%. Comparison with Previous Results For the range of experimentally observed embedding angles22 (i.e. 15◦ ≤ γ ≤ 90◦ ), we compared the values of the correcting coefficients α and β to those derived—(i) by Mijailovich et al.21 using finite element simulations and— Elaboration of a Set of Abacuses for Analyzing Magnetic Twisting Cytometry Data We took benefit of the analytical expressions of the correcting coefficients α(γ , ξ ) and β(γ , ξ ) established above to Generalized Elastic Modulus Estimation of Adherent Cell 137 TABLE 1. Numerical values of the coefficients defining the mathematical expressions of the interpolation functions Aα (γ), Bα (γ), Aβ (γ), Bβ (γ), Cβ (γ), and Dβ (γ) given in Eqs. (9)–(14). Aα1 Aα2 Aα3 Bα0 Bα1 Bα2 Bα3 15◦ ≤ γ ≤ 180◦ −6.497 × 10−2 2.966 × 10−1 −6.240 × 10−2 −2.811 × 10−3 1.133 × 10−2 7.418 × 10−3 −2.466 × 10−3 Aβ0 Aβ1 Aβ2 Aβ3 Bβ0 Bβ1 Bβ2 Bβ3 Cβ0 Cβ1 Cβ2 Dβ0 Dβ1 15◦ ≤ γ ≤ 90◦ 0 −7.205 × 10−3 −9.690 × 10−3 2.778 × 10−1 −4.673 × 10−3 2.403 × 10−2 −2.125 × 10−2 3.362 × 10−2 0 0 0 0 0 α β 90◦ ≤ γ ≤ 180◦ 29.688 −43.252 20.229 −2.734 −10.830 16.823 −8.694 1.540 0.593 −0.756 0.242 0.002 × 10−1 2.892 construct the set of abacuses which appear as the most relevant for quantifying the Young’s modulus of a cell probed by MTC. Moreover, we considered as the main dependent variable the ratio 1/ξ instead of ξ in order to exhaust the sensitivity of the correcting coefficients variations in the region of interest, i.e. when hu tends to small values, as exemplified in Fig. 4. From the mathematical expressions of α and β given in Eqs. (7) and (8), we derived the critical value of the ratio ξ = h u /2R for which the slopes of the curves α(γ i , ξ ) and β(γ i , ξ ) at given embedding angle γ i become larger than 10%. Figure 7 shows that this threshold increases with increasing embedding angles: for example, the influence of hu on the correcting coefficient α must be considered as soon as hu is lower than ∼R (i.e. ξ ∼ 1/2) when the embedding angle equals 90◦ [Fig. 7(A)]. Estimation of the cell elastic modulus from the bead translation is even more sensitive: the variation of the correcting coefficient β with hu must be taken into account for still larger values of hu , approximately equal to 2R (i.e. ξ ∼ 1) when γ = 90◦ (Fig. 7B). For values of hu larger than these critical bead sizes, the correcting coefficient α is almost constant for fixed γ : its value is then given by the function Aα (γ ) (Eq. (9)). In practical terms, the use of the abacuses can be illustrated in two cases of interest. The first one is a straightforward evaluation of the correcting coefficients α or β for a given cell-bead geometry (Fig. 8). The cell Young’s modulus is then computed from Eqs. (3) and (5) or/and Eqs. (4) and (6), knowing the amplitude of the applied magnetic torque as well the bead rotation or/and bead translation. FIGURE 6. Variation of the correcting coefficients α [Fig. 6(A)] and β [Fig. 6(B)] with increasing values of the embedding angle γ and for three different cell heights chosen for a seek of comparison with previous works: h = 1 µm (square symbols), h = 5 µm (triangular symbols) and h∞ = 20 µm (circular symbols). Our results (solid symbols) are compared to the analytical expression proposed by Laurent et al.17 (α = sin3 (γ)/2) in the semi-infinite medium case (enlargement in graph A) and to the numerical results of Mijailovich et al.21 (open symbols) obtained for finite cell height and with the same bead radius R = 2.25 µm. However, the measurement of the bead embedding angle remains a quite involved task10 which could prevent the use of the above procedure. One can thus originally underline that such a measurement of the embedding angle, as well as the measurement of the value hu , can be avoided if both bead rotation and bead lateral translation are measured. From the abacus of Fig. 9(A), one can indeed determine, from the knowledge of the cell height h and the bead radius R, the intersection point between the iso-(h/2R) values curve [for example h/2R = 0.8, Fig. 9(A)] with the horizontal line corresponding to the measured θ/δ ratio [for example θ/δ = 2, Fig. 9(A)]. Only one iso-γ curve goes through this point [solid point in Fig. 9(A)], which gives the corresponding embedding angle γ (γ = 65◦ in this example). In addition, the abscissa of this intersection point gives the 2R/hu ratio which will be used for the derivation of the correcting coefficients α and β from the abacus of Fig. 8 [solid points, Fig. 8(A) and 8(B) respectively]. 138 J. OHAYON and P. TRACQUI FIGURE 7. Critical values of the ratio ξ = hu /2R above which the slopes of the correcting functions α(ξ) [Fig. 7(A)] and β(γ i , ξ) [Fig. 7(B)] are equal to or less than 10% for a given embedding angle γ i . Below these thresholds, the correcting coefficients α and β vary significantly with decreasing values of ξ (see Fig. 4). DISCUSSION Estimation of the cell stiffness is a key issue for understanding how changes in cytoskeleton tension modulate intracellular mechanochemical signaling, i.e. mechanotransduction cell properties. Moreover, quantification of cell stiffness is also a prerequisite for analyzing the global mechanical response of cells cultured on extracellular matrices with different stiffness6 and on which they undergo a limited or complete spreading.5 On the other hand, cell elasticity values reported from experimental data in the literature exhibit large dispersions which may be related not only to differences between the micromanipulation assays or cell types,19 but also to the method used for estimating the cell elastic modulus from rough experimental data. In this work, we revisited such estimation methods when adherent cells are probed by MTC, with the aim to correct and extend previous linear models taking into account the geometrical variations of the cell-bead interface.17,21 Indeed, even if the need for corrected quantification of cell elastic modulus has been highlighted by Mijailovich et al.,21 general relationships allowing the derivation of this modulus over the complete range of experimental parameters values are still lacking, especially when the influence of the cell height comes into play. Considering the cell as hyperelastic (neo-Hookean) material undergoing large strains, our finite element computations first indicate that only two geometrical parameters, namely the embedding angle γ and the ratio of the under bead height over the bead diameter (i.e. the normalized geometrical parameter ξ = h u /2R), are relevant for determining the correcting coefficients α and β that allow the cell elastic modulus to be estimated from measurements of either bead rotation or lateral bead translation, respectively. Secondly, the consideration of a neo-Hookean cellular material validates our analysis in the domain of large deformations, i.e. for magnetic bead rotation going up to ∼15◦ , since the correcting coefficients α and β remain quasi-constant in this rotation range. This is true even for the lowest h u values we considered, i.e. when the nonlinear geometrical effects are important: for example, the computed relative variation α/α and β/β of the correcting coefficients are both in the order of 3% when the bead rotation increases from 1◦ to 15◦ for the cell-bead geometry considered in Fig. 2D, with h u = 2 µm and R = 2.5 µm (ξ ∼ 0.4). Consequently, our analysis kept only two varying parameters, the bead embedding angle γ and the dimensionless geometrical invariant ξ = h u /2R. We have shown that the corresponding α(γ , ξ ) and β(γ , ξ ) functions can be accurately described by mathematical functions. This result can be discussed in the light of a recent work using magnetocytometry16 and in which the relationship between the applied force and the bead displacement was reported to scale with sin3 γ . We indeed found that a good fit to our computed α and β values can be obtained with analytical functions α(γ , ξ ) and β(γ , ξ ) of the form (k1 + k2 /ξ ) sin3 γ , where k1 and k2 are real constants. However, this is only true for embedding angles γ lower than 40◦ for the fit of function β(γ , ξ ) and for values of γ lower than 30◦ for the fit of the function α(γ , ξ ). Above these values, the sin3 γ -type fitting function is no longer valid. The critical role of the geometrical invariant ξ = h u /2R has been more precisely highlighted by computing the threshold values ξ ∗ (γ ) of the ratio ξ above which α(γ , ξ ) and β(γ , ξ ) no longer vary significantly (Fig. 7), i.e. α(γ , ξ ) ∼ α(γ ) and β(γ , ξ ) ∼ β(γ ) for a given embedding angle γ . Below this threshold value ξ ∗ (γ ), the varying correcting coefficients α(γ , ξ ) and β(γ , ξ ) can be easily determined thanks to the relationships we established [Eqs. (7) and (8)]. We also analyzed the influence of deeply embedded beads and computed the corresponding abacuses for increasing values of γ up to 180◦ . Our computations of the Generalized Elastic Modulus Estimation of Adherent Cell 139 FIGURE 8. Generalized abacuses establishing the relationships between the correcting coefficients α [Fig. 8(A) and 8(C)] and β [Fig. 8(B) and 8(D)] when the two geometrical parameters 2R/hu and γ are varied. These abacuses cover the following ranges of parameter values: 0.1 ≤ 2R/ hu ≤ 10; 25◦ ≤ γ ≤ 180◦ . The iso-embedding angle curves are computed by increment of 5◦ . The two solid points, corresponding to a specific embedding angle of 65◦ , exemplify the application of theses abacuses (see text for details). correcting coefficients α(γ , ξ ) and β(γ , ξ ) also agree with and extend previous results of Mijailovitch et al.21 In this work, we assumed that the bead is firmly attached to the cell cytoskeleton through integrin trans-membranous receptors. Thus, a no slipping boundary condition has been imposed at the cell/bead interface. However, if a partial slipping θ s of the bead occurs during bead rotation, the estimation of the shear modulus G would be biased by improper evaluation of the correcting factor α. From Eqs. (3) and (5), we derived the expression of the relative error δG on the shear modulus G as δG = θs /(θ − θs ), where θ is the overall bead rotation. The influence of various degrees of slipping can be analyzed by this relationship. Typically, the relative error δG is lower than 10% if the ratio θ /θ s is larger than 10. Finally, despite of static conditions assumed for this finite element analysis, our results may be used to characterize the linear viscoelastic properties of the cytoskeletal medium.4,9,25 Indeed, knowing that the correcting coefficients α(γ , ξ ) and β(γ , ξ ) are only geometry-dependent and not frequency-dependent, the computed results can be generalized9 such that G̃ θ ( f ) = α(γ , ξ )G̃( f ) and G̃ δ ( f ) = β(γ , ξ )G̃( f ), where G̃ θ , G̃ δ , and G̃ are complex moduli and f is the excitation frequency. The associated spatial stress distribution could thus be compared to recent experimental data obtained by intracellular stress tomography.13 In conclusion, this work provides a refined analysis of the experimentally measured cell response to external torques imposed at the cell apical surface. In particular, consideration of the here reported abacuses is of importance to deal with variations in parameter hu due to the random localization of bead attachment at the cell surface. Indeed, the decrease of the cell height when going from the cell body to flattened lamellipods would lead to drastic variation in the estimated cell elastic modulus.2 Consideration of such variations of the correcting coefficients could be still more crucial when measuring the integrated mechanical response of cells anchored on gels with varying stiffness.6 In this case, the substrate stiffness modifies the cell spreading, and thus the mean cell height, with direct implications on the 140 J. OHAYON and P. TRACQUI APPENDIX Rigid Bead Immersed in a Finite (or Infinite) Isotropic Incompressible Medium Let us denoted (er , e , eφ ) and (r, , φ) the spherical unit base vectors and the associated physical coordinates respectively. An exact solution of a single spherical rigid bead of radius R embedded (γ = π ) in an elastic finite spherical concentric medium of radius Re can be found, in linear elasticity, when the bead is submitted to a small bead rotation θ in the parallel plane (r, φ) going through the bead center. In this test problem, relevant for twisting magnetocytometry experiments, we are looking for the relationship between applied torque T and bead rotation θ . The medium is assumed to be incompressible and isotropic, and is thus described by the constitutive law: σ = − p I + 2Gε (A.1) where σ and ε are the stress and strain tensors, I is the identity matrix, p is the Lagrangian multiplier resulting from the material incompressibility,12 and G is the material shear modulus, related to the Young’s modulus Ecell by the relationship E cell = 3G. Moreover, the linearized material incompressibility constraint is given by ∇ u = 0 (A.2) where u is the unknown displacement vector. If gravity and inertial forces are neglected, the condi or in terms of the tion for local equilibrium is ∇·[θ ] = 0, displacement vector: FIGURE 9. Generalized abacuses giving the relationship between the ratio of bead rotation over normalized bead translation (θ/δ) when the three geometrical parameters 2R /hu , h/2R, and γ are varied [Fig. 9(A)]. Notice that the ratio θ/δ can be derived from the correcting coefficients α and β using the relationship θ/δ = β/α. These abacuses are built for the following parameter ranges: 0.1 ≤ 2R/hu ≤ 10, 0.2 ≤ h/2R ≤ 10, and 25◦ ≤ γ ≤ 180◦ . The iso-γ curves are computed by increment of 5◦ , while the iso-h/2R curves [Fig. 9(A)] are computed from h/2R = 0.2 to 2 with an initial step of 0.1, then with the steps size specified on the abacus. The solid point corresponds to the values used to exemplify the application of these abacuses (see text for more details). estimated apparent cell stiffness: in the MTC experiments conducted by Doornaert et al.,5 the apparent cell stiffness almost doubles with cell spreading, i.e. when maximum cell height decreases from 22 to 9 µm. Therefore, although focusing on data analysis obtained by magnetic twisting cytometry, this study provides a framework in which finite element modeling of cell rheology would help to decrease the dispersion of cell elastic modulus estimated from experimental cell mechanical response. More fundamentally, it would improve our knowledge of the mechanisms of mechanotransduction and modified genes expression triggered by micromanipulations techniques, including MTC.3 ∇ p = µ∇ 2 u (A.3) The displacement field u must satisfy the following boundary conditions: (i) a small bead rotation θ with no translation is imposed to the bead and (ii) zero displacements are imposed on the external surface r = Re . These two conditions are expressed respectively as: u (r, , φ) = r sin () θ eφ at r = R (A.4) u (r, , φ) = 0 at r = R (A.5) The displacement field can be found by using the method of separation of variables and by looking for a u vector solution of the form u (r, , φ) = r sin ()w(r )eφ (A.6) where w(r ) becomes the unknown function of the problem. For a solution field u which satisfies the above Eqs. (A.2)(A.5), one gets the following function w(r ): 3 θ R3 Re w(r ) = 3 −1 (A.7) Re − R 3 r 3 So, the relation between the applied external torque T and the small bead rotation θ is given by: π 2π T = σr φ | R R 3 sin () d dφ (A.8) 0 0 Generalized Elastic Modulus Estimation of Adherent Cell or, in the developed form, by T = R3 R3 8 π E cell θ 3 e 3 3 Re − R 11 (A.9) In the particular case of a rigid bead immersed in an infinite medium, the torque-bead rotation relation is obtained from the limit lim T and is given by Re →∞ 8 π E cell θ R 3 = 6V Gθ 3 where V is the bead volume. T = (A.10) ACKNOWLEDGMENTS We gratefully acknowledge Dr Daniel Isabey and Pr François Gallet for helpful discussions. Ph. Tracqui and J. Ohayon are supported by a grant “ACI-Bioinformatique” from the French Centre National de la Recherche Scientifique (CNRS). REFERENCES 1 Caille, N., O. Thoumine, Y. Tardy, and J. J. Meister. Contribution of the nucleus to the mechanical properties of endothelial cells. J. Biomech. 35:177–187, 2002. 2 Charras, G. T., and M. A. Horton. Determination of cellular strains by combined atomic force microscopy and finite element modeling. Biophys. J. 83:858–879, 2002. 3 Chen, J., B. Fabry, E. L. Schiffrin, and N. Wang. Twisting integrin receptors increases endothelin-1 gene expression in endothelial cells. Am. J. Physiol. Cell Physiol. 280:C1475–C1484, 2001. 4 Djordjevic, V. D., J. Jaric, B. Fabry, J. J. Fredberg, and D. Stamenovic. Fractional derivatives embody essential features of cell rheological behavior. Ann. Biomed. Eng. 31:692–699, 2003. 5 Doornaert, B., V. Leblond, E. Planus, S. Galiacy, V. M. Laurent, G. Gras, D. Isabey, and C. Lafuma. Time course of actin cytoskeleton stifness and matrix adhesion molecules in human bronchial epithelial cell cultures. Exp. Cell Res. 287:199–208, 2003. 6 Engler, A., L. Bacakova, C. Newman, A. Hategan, M. Griffin, and D. Discher. Substrate compliance versus ligand density in cell on gel responses. Biophys. J. 86:617–628, 2004. 7 Fabry, B., G. N. Maksym, S. A. Shore, P. E. Moore, R. A. Panettieri Jr., J. P. Butler, and J. J. Fredberg. Selected contribution: time course and heterogeneity of contractile responses in cultured human airway smooth muscle cells. J. Appl. Physiol. 91:986–994, 2001a. 8 Fabry, B., G. N. Maksym, J. P. Butler, M. Glogauer, D. Navajas, and J. J. Fredberg. Scaling the microrheology of living cells. Phys. Rev. Lett. 87:148102-1–148102-4, 2001b. 9 Fabry, B., G. N. Maksym, J. P. Butler, M. Glogauer, D. Navajas, N. A. Taback, E. J. Millet, and J. J. Fredberg. Time scale and other invariants of integrative mechanical behavior in living cells. Phys. Rev. E 68:041914-1–041914-18, 2003. 10 Fodil, R., V. Laurent, E. Planus, and D. Isabey. Characterization of cytoskeleton mechanical properties and 3D-actin structure in twisted adherent epithelial cells. Biorheology 40:241–245, 2003. 141 Hamill, O. P., and B. Martinac. Molecular basis of mechanotransduction in living cells. Physiol. Rev. 81:685–740, 2001. 12 Holzapfel, G. A. Nonlinear Solid Mechanics. New York: Wiley, 2001. 13 Hu, S., J. Chen, B. Fabry, Y. Numaguchi, A. Gouldstone, D. E. Ingber, J. J. Fredberg, J. P. Butler, and N. Wang. Intracellular stress tomography reveals stress focusing and structural anisotropy in cytoskeleton of living cells, Am. J. Physiol. 285:C1082–C1090, 2003. 14 Hubmayr, R. D., S. A. Shore, J. J. Fredberg, E. Planus, R. A. Panettieri Jr, W. Moller, J. Heyder, and N. Wang. Pharmacological activation changes stiffness of cultured human airway smooth muscle cells. Am. J. Physiol. 271:C1660–C1668, 1996. 15 Janmey, P. A. The cytoskeleton and cell signaling: Component localization and mechanical coupling. Physiol. Rev. 78:763–781, 1998. 16 Karcher, H., J. Lammerding, H. Huang, R. T. Lee, R. D. Kamm, and M. R. Kaazempur-Mofrad. A three-dimensional viscoelastic model for cell deformation with experimental verification. Biophys. J. 85:3336–3349, 2003. 17 Laurent, V. M., S. Henon, E. Planus, R. Fodil, M. Balland, D. Isabey, and F. Gallet. Assessment of mechanical properties of adherent living cells by bead micromanipulation: Comparison of magnetic twisting cytometry vs optical tweezers. ASME J. Biomech. Eng. 124:408–421, 2002. 18 Laurent, V. M., R. Fodil, P. Cañadas, S. Féréol, B. Louis, E. Planus, and D. Isabey. Partitioning of cortical and deep cytoskeleton responses from transient magnetic bead twisting. Ann. J. Biomed. Eng. 31:1263–1278, 2003. 19 Maksym, G. N., B. Fabry, J. P. Butler, D. Navajas, D. J. Tschumperlin, J. D. Laporte, and J. J. Fredberg. Mechanical properties of cultured human airway smooth muscle cells from 0.05 to 0.4 Hz. J. Appl. Physiol. 89:1619–1632, 2000. 20 Maniotis, A. J., C. S. Chen, and D. E. Ingber. Demonstration of mechanical connections between integrins, cytoskeletal filaments, and nucleoplasm that stabilize nuclear structure. Proc. Natl. Acad. Sci. U.S.A. 94:849–854, 1997. 21 Mijailovich, S. M., M. Kojic, M. Zivkovic, B. Fabry, and J. J. Fredberg. A finite element model of cell deformation during magnetic bead twisting. J. Appl. Physiol. 93:1429–1436, 2002. 22 Ohayon, J., P. Tracqui, R. Fodil, S. Féréol, V. M. Laurent, E. Planus, and D. Isabey. Analysis of nonlinear responses of adherent epithelial cells probed by magnetic bead twisting: A finite element model based on an homogenization approach. J. Biomech. Eng. 126(6), 2004, in press. 23 Rahman, A., Y. Tseng, and D. Wirtz. Micromechanical coupling between cell surface receptors and RGD peptides. Biochem. Biophys. Res. Commun. 296:771–778, 2002. 24 Schneider, S. W., P. Pagel, C. Rotsch, T. Danker, H. Oberleithner, M. Radmacher, and A. Schwab. Volume dynamics in migrating epithelial cells measured with atomic force microscopy. Pflugers Arch. 439:297–303, 2000. 25 Trickey, W. R., T. P. Vail, and F. Guilak. The role of the cytoskeleton in the viscoelastic properties of human articular chondrocytes, J. Orthop. Res. 22: 131–139, 2004. 26 Wang, N., J. P. Butler, and D. E. Ingber. Mechanotransduction across the cell surface and through the cytoskeleton. Science 260:1124–1127,1993. 27 Wang, N., and D. E. Ingber. Probing transmembrane mechanical coupling and cytomechanics using magnetic twisting cytometry. Biochem. Cell Biol. 73:327–335, 1995.
© Copyright 2026 Paperzz