Wang: Science, Volume 260(5111).May 21, 1993.pp 1124-1127.

Volume 260(5111) May 21, 1993 pp 1124-1127

Mechanotransduction Across the Cell Surface and Through the Cytoskeleton [REPORTS.] Wang, Ning; Butler, James P.; Ingber, Donald E.*

N. Wang and J. P. Butler, Respiratory Biology Program, Harvard School of Public Health, Boston, MA 02115.
D. E. Ingber, Departments of Surgery and Pathology, Children’s Hospital and Harvard Medical School, Enders 1007, 300 Longwood Avenue, Boston, MA 02115.
*To whom correspondence should be addressed.

Mechanical stresses were applied directly to cell surface receptors with a magnetic twisting device. The extracellular matrix receptor, integrin beta sub 1, induced focal adhesion formation and supported a force-dependent stiffening response, whereas nonadhesion receptors did not. The cytoskeletal stiffness (ratio of stress to strain) increased in direct proportion to the applied stress and required intact microtubules and intermediate filaments as well as microfilaments. Tensegrity models that incorporate mechanically interdependent struts and strings that reorient globally in response to a localized stress mimicked this response. These results suggest that integrins act as mechanoreceptors and transmit mechanical signals to the cytoskeleton. Mechanotransduction, in turn, may be mediated simultaneously at multiple locations inside the cell through force-induced rearrangements within a tensionally integrated cytoskeleton.

The process of recognizing and responding to mechanical stimuli is critical for the growth and function of living cells. Many sensory functions including touch, hearing, baroreception, proprioception, and gravity sensation involve specialized mechanotransduction mechanisms. Development of tissue pattern is also exquisitely sensitive to changes in mechanical stress (1). Nevertheless, the molecular mechanism by which individual cells recognize and respond to external forces is not well understood. Stretch-sensitive ion channels, adenylate cyclase, and protein kinase C change their activity in response to applied stress (2-4). However, these signaling pathways are likely to lie downstream from the initial mechanoreception event at the cell surface. For example, activation of these signaling molecules appears to be mediated though changes in the cytoskeleton (CSK) (2,4,! 5). Although changes in CSK organization are a ubiquitous response to mechanical perturbation (4,6,7), the mechanism by which forces are transmitted across the cell surface and transduced into a CSK response remains unknown.

Analysis of mechanotransduction in specialized force-sensing cells, in both plants and animals, suggests that the cell’s extracellular matrix (ECM) attachments are the sites at which forces are transmitted to cells (6,8). As in any architectural structure, mechanical loads are transmitted across the cell surface and into the cell by means of structural elements that are physically interconnected. Transmembrane ECM receptors, such as members of the integrin family, are excellent candidates for mechanoreceptors because they bind actin-associated proteins within focal adhesions and thereby physically link ECM with CSK microfilaments (9). The possibility that ECM receptors mediate mechanotransduction is supported by the finding that stretching flexible ECM culture substrata alters CSK organization and induces biochemical changes in adherent cells (10). However, in these stretching studies, it is not po! ssible to separate effects due to transmembrane force transfer from those due to global shape changes and generalized deformation of the plasma membrane and CSK.

To determine whether ECM receptors provide a specific molecular path for mechanical signal transfer to the CSK, we devised a method in which controlled mechanical loads could be applied directly to specific cell surface molecules without producing large-scale changes in cell shape (Figure 1). We modified a cell magnetometry system (11) by allowing cells to bind spherical ferromagnetic microbeads that were coated with specific receptor ligands that mediate attachment but not cell spreading (12,13). By magnetizing these surfacebound beads in one direction and then applying a second, weaker magnetic field oriented at 90 degrees, we were able to twist the beads in place and thereby exert a controlled shear stress (0 to 68 dyne/cm sup 2) on bound cell surface receptors. An in-line magnetometer was used to simultaneously measure changes in the orientation of the magnetized beads and hence to quantitate ! angular strain produced in response to the applied stress.

*Figure 1. The magnetic twisting device. Microbeads (5 times 10 sup 4 in each well) were allowed to bind to cell surfaces for 10 to 15 min, and unbound beads were removed before magnetic manipulation was initiated. Brief application of a strong external magnetic field (1000 G for 10 mu s) resulted in magnetization and alignment of the magnetic moments of all surface-bound beads. We then applied defined mechanical stresses (0 to 68 dyne/cm sup 2) without remagnetizing the beads, using a weaker twisting magnetic field (0 to 25 G) applied perpendicular to the original field. We measured the average bead rotation (angular strain) induced by the twisting field by using a magnetometer to measure changes in the component of the remanent magnetic field vector in the direction of the original magnetization as a function of time (11). In the absence of force transmission across the cell surface, the spherical beads would twist in place by 90 degrees into complete al! ignment with the twisting field, and the remanent field vector would immediately drop to zero. In contrast, transmission of force to the CSK would result in increased resistance to deformation and decreased bead rotation.*

Adherent endothelial cells were first allowed to bind beads coated with a synthetic peptide containing the Arg-Gly-Asp (RGD) sequence that is a known ligand for fibronectin receptors, such as integrin beta sub 1 alpha sub 5, which these cells express on their surface (13). Efficient transmembrane force transfer was observed in cells bound to RGD beads; the cells became stiffer and increased their resistance to mechanical deformation (bead twisting) at higher levels of applied stress, such that angular strain only reached a bead rotation of approximately 25 degrees (Figure 2). To demonstrate the specificity of transmembrane force transfer, we included a soluble synthetic peptide, Gly-Arg-Gly-Asp-Ser-Pro (GRGDSP) (1 mg/ml), in the culture medium as a competitor (14). This fibronectin peptide inhibited CSK stiffening (Figure 2), whereas a control hexapeptide with a single amino acid substitution (Gly-! Arg-Gly-Glu-Ser-Pro) had no inhibitory effect. Beads coated with antibodies directed against integrin beta sub 1 receptor subunits produced a similar stiffening response (Figure 2). In contrast, surface-bound beads coated with nonspecific cell attachment ligands, such as acetylated-low density lipoprotein (AcLDL) (15) or bovine serum albumin (BSA), were not nearly as restricted in their rotation (Figure 2).

*Figure 2. Stress-strain relation measured with magnetic microbeads attached to the surfaces of living cells. Applied stress was determined by a calibration technique in which the same beads were twisted in a standard solution of known viscosity (22). Angular strain (bead rotation) was calculated as the arc cosine of the ratio of remanent field after 1 min of twist to the field at time 0. Angular strain is plotted here as degrees. Bead coatings were as follows: RGD, Arg-Gly-Asp-containing synthetic peptide; Ab-beta sub 1, antibodies against integrin beta sub 1; AcLDL, acetylated-low density lipoprotein; BSA, bovine serum albumin; GRGDSP, soluble fibronectin peptide (1 mg/ml added for 10 min); Cyt, cytochalasin D (0.1 mug/ml). Measurements analyzing the effects of different bead coatings with or without GRGDSP were made at stresses from 0 to 40 dyne/cm sup 2; for clarity, intermediate data points are shown only for Ab-beta sub 1 and RGD beads that exhibit i! ntegrin-dependent stiffening. The effects of cytochalasin D were measured only at the highest stress. Error bars equals SEM.*

Figure 3. Immunofluorescence micrographs of cells bound to beads coated with RGD or AcLDL and immunostained for the focal adhesion proteins talin, vinculin, and alpha-actinin.

To confirm that applied mechanical loads were indeed transmitted to the CSK, we measured the mechanical properties of cells bound to RGD beads before and after disrupting microfilament lattice integrity with a low concentration of cytochalasin D (0.1 mu g/ml), which had minimal effects on cell shape. Angular strain increased after exposure to cytochalasin for only 15 min (Figure 2). Efficient force transfer and associated CSK stiffening also correlated with focal adhesion formation, as defined by the recruitment of talin, vinculin, and alpha-actinin to the site of bead binding (Figure 3). These focal adhesion proteins, which appeared along the surface of RGD beads but not AcLDL beads, form the molecular bridge that physically interlinks integrins with actin microfilaments (9). Recruitment of talin also appears to be required for cell spreading on ECM (16).

Importantly, disruption of microfilament lattice integrity with cytochalasin D did not completely suppress CSK stiffening (Figure 4A), suggesting that other filament systems may also contribute to the CSK response to force. Disruption of microtubules or intermediate filaments with nocodazole (10 mu g/ml) or acrylamide (4 mM; 17), respectively, inhibited the stiffening response by approximately 25% (Figure 4A), and no additive effect was observed when they were combined. Combination of cytochalasin D with acrylamide reduced stress-induced CSK stiffening by more than 85%, and combination with nocodazole resulted in complete suppression (Figure 4A). Thus, although integrins may initially transmit forces to microfilaments within focal adhesions, higher order structural interactions among all three CSK filament systems appear to be responsible for efficient transduction of the mechanical stimulus into a cellular response. The! finding that actin microfilaments contribute the most to cell stiffness is consistent with recent data which shown that networks of purified actin polymers exhibit a higher shear modulus than networks containing microtubules or intermediate filaments (18).

Figure 4. Continuum mechanics analysis of living cells and a three-dimensional tensegrity model. (A) Stiffness of the CSK of living cells was defined as the ratio of stress to strain (in radians) at 1 min of twisting. Noc, nocodazole (10 mu g/ml); Acr, acrylamide (4 mM); Cyt, cytochalasin D (0.1 mu g/ml). (B) A tensegrity cell model under different mechanical loads. This model consisted of a geodesic spherical array of wood dowels (0.3 cm by 15 cm) and thin elastic threads (0.06 cm by 6 cm). The model was suspended from above and loaded, from left to right, with 0-, 20-, 50-, 100-, or 200-g weights on a single strut at its lower end. (C) Stiffness of the stick and string tensegrity model was defined as the ratio of applied stress to strain (linear deformation of the entire structure). Similar measurements were carried out with an isolated tension element, that is, a single thin elastic thread of a size similar to that found in the model.

How could a “solid” lattice composed of interconnected microfilaments, microtubules, and intermediate filaments (19) respond dynamically as a single integrated unit? Consider the observation that CSK stiffness increased in direct proportion to the stress applied to integrins (the slope of the curve in Figure 4A is linear). This type of mechanical behavior is not commonly seen in man-made materials, but it is often observed in biological tissues (20). This mechanical response cannot be explained by current theories (20). We have proposed that in the construction of cells a building system may be used that was first described by the architect-inventor Buckminster Fuller and that depends on tensional integrity (tensegrity) rather than compressional continuity (21). Tensegrity cell models that incorporate isolated rigid struts interconnected by a continuous series of elastic te! nsile threads predict cell shape changes and mimic specific structural patterns that are observed within the CSK of living cells (21).

To explore whether cells might use tensegrity to mediate mechanotransduction within the CSK, we carried out stress-strain measurements with a stick and elastic string tensegrity model. When increasing force (metal weights) was applied to these models, the mechanically interdependent structural elements rearranged without topological disruption or loss of tensional continuity (Figure 4B). A plot of stiffness versus applied stress (force) based on these models (Figure 4C) mimicked the linear response exhibited by the CSK of living cells (Figure 4A) as well as by intact biological tissues (20). This linear response was in direct contrast to the behavior exhibited by nonprestressed tensile filaments taken from the same structure (Figure 4C). Stiffness of the compression-resistant struts was essentially infinite over the range of forces applied. Viewed in this light, the CSK response to! applied stress appears to be a property of the integrated system and not a characteristic of any one of its individual parts. Gels containing purified CSK filaments (for example, F-actin) that lack structural continuity and internal tension (prestress) either do not exhibit force-induced stiffening or, if they do, the response is nonlinear (18,22) and appears similar to that exhibited by a non-prestressed tensile filament (Figure 4C).

Thus, our experimental data are consistent with the possibility that the CSK is organized as a tensegrity network. In living cells, contractile microfilaments generate and distribute tension to all CSK filament systems (23). In addition, microfilaments resist compression locally when either cross-linked within large bundles or contracted to their shortest length (21). Microtubules also resist compression in cells (21,24), possibly because they are stabilized against buckling by lateral interconnections with tensionally stiffened intermediate filaments (25). Our finding that a combination of acrylamide and nocodazole did not further reduce CSK stiffness supports this possibility that intermediate filaments and microtubules resist compression as a paired unit. The tensegrity paradigm therefore provides a novel mechanism for CSK integration (21) as well as ! a plausible explanation for why the CSK stiffening response is linear in cells (Figure 4A) and tissues (20). It also could explain how a local stress, induced by ligation of a subset of CSK-associated membrane receptors, can result in global modulation (immobilization) of receptors over the entire cell surface (26).

On a more general level, our findings suggest that the balance of mechanical forces that preexists within the CSK before an external mechanical load is applied (that is, prestress) may be a critical determinant of the subsequent cellular response. This result may have direct implications for understanding specialized mechanosensory mechanisms (6) as well as coupling between cell shape and function (27). For example, the change in the level of CSK prestress that accompanies changes in cell shape may provide regulatory information to the cell (21,27). Prestress of the CSK also may play a critical role in the cellular mechanism of aging, given that the load-bearing properties of any structural support element would be expected to weaken over time, if continually stressed.

Taken together, these results indicate that at least one type of transmembrane ECM receptor, integrin beta sub 1, can act as a mechanoreceptor in that it can transfer mechanical signals to the CSK by way of a specific molecular pathway. A cell’s sensitivity to a mechanical stimulus therefore may be altered by changing ECM receptor number, location, or adhesion strength or by modulating focal adhesion formation. Other types of transmembrane molecules that interconnect with CSK filaments (for example, different integrin subunits, cadherins, or cell surface proteoglycans) may also transfer external mechanical signals to the CSK. The magnetic twisting device provides a simple method to directly address this possibility. In addition, these results suggest that transfer of force from integrins to the CSK may represent a proximal step in an intracellular mechanical signaling cascade that leads to global CSK rearrangements and simultaneous mechanotransduction events at multiple locati! ons inside the cell (21,28). If cells use a tensegrity-based transduction system, then mechanical signal transfer throughout the entire cell would be essentially instantaneous and thus more rapid than any diffusion-based signaling system.

REFERENCES AND NOTES

1. Y. C. Fung, Biomechanics: Mechanical Properties of Living Tissues (Springer-Verlag, New York, 1988) D. W. Thompson, On Growth and Form (Cambridge Univ. Press, New York, 1977) E. S. Russell, Form and Function (Univ. of Chicago Press, Chicago, 1982) J. Koch, Am. J. Anat. 21, 177 (1917).

2. F. Sachs, in Cell Shape: Determinants, Regulation, and Regulatory Role, W. D. Stein and F. Bronner, Eds. (Academic Press, San Diego, CA, 1989), pp. 63-92.

size=”-1”>3. I. Komuro et al., J. Biol. Chem. 266, 1265 (1991) R. P. De Groot et al., J. Cell Sci. 97, 33 (1990) S.-P. Olesen et al., Nature 331, 168 (1988) J. B. Lansman et al., ibid. 325, 811 (1987).

4. P. Watson, J. Biol. Chem. 265, 6569 (1990)

5. K. G. Murti, K. Kaur, R. M. Goorha, Exp. Cell Res. 202, 36 (1992).

6. B. Kacahr, P. Parakkal, J. Fex, Hear. Res. 45, 179 (1990) L. J. Wilson and D. H. Paul, J. Comp. Neurol. 296, 343 (1990).

7. L. F. Cipriano, Physiologist 34, 72 (1991) T. J. Dennerll et al., J. Cell Biol. 107, 665 (1988) J. Kolega, ibid. 102, 1400 (1986) R.-P. Franke et al., Nature 307, 648 (1984).

8. V. Wagner et al., Proc. Natl. Acad. Sci. U.S.A. 89, 3644 (1992) R. Wayne, M. P. Staves, A. C. Leopold, J. Cell Sci. 101, 611 (1992) D. E. Ingber, Curr. Opin. Cell Biol. 3, 841 (1991).

9. S. M. Albelda and C. A. Buck, FASEB J. 4, 2868 (1990) K. Burridge et al., Annu. Rev. Cell Biol. 4, 487 (1988).

10. H. R. W. Wirtz and L. G. Dobbs, Science 250, 1266 (1990) J. L. Samuel and H. H. Vandenburgh, In Vitro Cell. Dev. Biol. 26, 905 (1990) R. C. Harris, M. A. Haralson, K. L. Badr, Lab. Invest. 66, 548 (1992) B. E. Sumpio et al., Arch. Surg. (Chicago) 123, 1233 (1988) L. Terracio et al., In Vitro Cell Dev. Biol. 24, 53 (1988).

11. C. L. Bizal, J. P. Butler, P. A. Valberg, J. Leukocyte Biol. 50, 240 (1991) P. A. Valberg and J. P. Butler, Biophys. J. 52, 537 (1987) P. A. Valberg and D. F. Albertini, J. Cell Biol. 101, 130 (1985) P. A. Valberg, Science 224, 513 (1984). [Medl! ine Link]

12. Capillary endothelial cells were plated (3 times 10 sup 4 cells per well) on fibronectin-coated bacteriological plastic (96- well Removawells, Immunolon II) and cultured for 6 to 10 hours in chemically defined medium (27) before bead addition. Spherical ferromagnetic beads (5.5 mu m in diameter, 1 mg/ml) [W. Moller, C. Roth, W. Stahlhofen, J. Aerosol Sci. 21, S657 (1990)] were coated with RGD-containing peptide (Peptite 2000), AcLDL, BSA, or anti-integrin beta sub 1 antibodies as previously described (all at 50 mug/ml) (13).

13. M. A. Schwartz, C. Lechene, D. E. Ingber, Proc. Natl. Acad. Sci. U.S.A. 88, 7849 (1991).

14. E. G. Hayman, M. D. Pierschbacher, E. Ruoslahti, J. Cell Biol. 100, 1948 (1985).

15. AcLDL binds to specific transmembrane receptors on the endothelial cell surface however, it does not normally play a role in cell adhesion. The small but statistically significant resistance to deformation that these beads did exhibit (deviation from 90 degrees) may be due to generalized distortion of elements of the submembranous CSK that are known to be highly deformable [N. O. Petersen, W. B. McConnaughey, E. L. Elson, Proc. Natl. Acad. Sci. U.S.A. 79, 5327 (1982)]. Local nonspecific CSK deformation was also observed when this cell magnetometry system was used with ingested ferromagnetic particles that were uncoated, irregularly shaped, and contained within intracellular lysosomes (11). Thus, establishment of a specific molecular path for force transmission appears to be required for efficient signal transfer as well as an effective CSK response.

16. G. H. Nuckolls, L. H. Romer, K. Burridge, J. Cell Sci. 102, 753 (1992).

17. M. Hay and U. De Boni, Cell Motility Cytoskeleton 18, 63 (1991). [Medline Link]

18. P. A. Janmey, U. Euteneur, P. Traub, M. Schliwa, J. Cell Biol. 113, 155 (1991).

19. E. G. Fey et al., ibid. 98, 1973 (1984) J. J. Wolosewick and K. R. Porter, ibid. 82, 114 (1979).

20. D. Stamenovic, Physiol. Rev. 70, 1117 (1990) T. A. McMahon, Muscles, Reflexes, and Locomotion (Princeton Univ. Press, Princeton, NJ, 1984) I. M. Ward, Mechanical Properties of Solid Polymers (Wiley, New York, 1983) J. G. Pinto and Y. C. Fung, J. Biomech. 6, 597 (1973) Y. C. Fung, Am. J. Physiol. 213, 1532 (1967) I. S. Sokolnikoff, Mathematical Theory of Elasticity (McGraw-Hill, New York, 1956). !

21. D. E. Ingber, J. Cell Sci. 104, 613 (1993) D. E. Ingber et al., in Physical Forces and the Mammalian Cell, J. A. Frangos and C. L. Ives, Eds. (Academic Press, San Diego, CA, 1993), pp. 61-78 D. E. Ingber and J. Folkman, in Cell Shape: Determinants, Regulation, and Regulatory Role, W. Stein and F. Bronner, Eds. (Academic Press, San Diego, CA, 1989), pp. 3-31 D. E. Ingber and J. D. Jamieson, in Gene Expression During Normal and Malignant Differentiation, L. C. Andersson, C. G. Gahmberg, P. Ekblom, Eds. (Academic Press, Orlando, FL, 1985), pp. 13-32 B. Fuller, Portfolio Art News Annu. 4, 112 (1961).

22. K. S. Zaner and P. A. Valberg, J. Cell Biol. 109, 2233 (1989).

23. J. Sims, S. Karp, D. E. Ingber, J. Cell Sci. 103, 1215 (1992).

24. R. E. Buxbaum and S. R. Heidemann, J. Theor. Biol. 134, 379 (1988) H. C. Joshi et al., J. Cell Biol. 101, 697 (1985).

25. G. W. Brodland and R. Gordon, J. Biomech. Eng. 112, 319 (1991).

26. I. Yahara and G. M. Edelman, Proc. Natl. Acad. Sci. U.S.A. 69, 608 (1972).

27. D. E. Ingber, ibid., p. 3579 _____ and J. Folkman, J. Cell Biol. 109, 317 (1989) Cell 58, 803 (1989).

28. K. J. Pienta and D. S. Coffey, Med. Hypotheses 34, 88 (1991).

29. We thank W. Moller and W. Stahlhofen for providing ferromagnetic microbeads, G. Plopper for assistance in the immunostaining studies, J. Fredberg and P. Valberg for helpful discussions, and J. Folkman for reviewing the manuscript. This work was supported by grants from the National Institutes of Health (HL-33009 and CA45548) and the Space Biology Program at the National Aeronautics and Space Administration (NAG-9-430) and by a Faculty Research Award from the American Cancer Society (D.E.I.).

unostaining studies, J. Fredberg and P. Valberg for helpful discussions, and J. Folkman for reviewing the manuscript. This work was supported by grants from the National Institutes of Health (HL-33009 and CA45548) and the Space Biology Program at the National Aeronautics and Space Administration (NAG-9-430) and by a Faculty Research Award from the American Cancer Society (D.E.I.).