Mechanical Stressing of Integrin Receptors Induces Enhanced Tyrosine Phosphorylation of Cytoskeletally Anchored Proteins*

,

Physical forces play a fundamental role in the regulation of cell function in many tissues, but little is known about how cells are able to sense mechanical loads and realize signal transduction. Adhesion receptors like integrins are candidates for mechanotransducers. We used a magnetic drag force device to apply forces on integrin receptors in an osteoblastic cell line and studied the effect on tyrosine phosphorylation as a biochemical event in signal transduction. Mechanical stressing of both the ␤1 and the ␣2 integrin subunit induced an enhanced tyrosine phosphorylation of proteins compared with integrin clustering. Application of cyclic forces with a frequency of 1 Hz was more effective than a continuous stress. Using Triton X-100 for cell extraction, we found that tyrosine-phosphorylated proteins became physically anchored to the cytoskeleton due to mechanical integrin loading. This cytoskeletal linkage was dependent on intracellular calcium. To see if mechanical integrin stressing induced further downstream signaling, we analyzed the activation of mitogen-activated protein (MAP) kinases and found an increased phosphorylation of MAP kinases due to mechanical stress. We conclude that integrins sense physical forces that control gene expression by activation of the MAP kinase pathway. The cytoskeleton may play a key role in the physical anchorage of activated signaling molecules, which enables the switch of physical forces to biochemical signaling events.
Application of physical forces to cells induces gene expression and proliferation in a variety of cell types (1)(2)(3)(4)(5)9). Therefore, mechanical forces are a fundamental physiological factor in regulating structure and function in many tissues. In bone, mechanical loading stimulates the increase of bone mass (6 -8) and plays an important role in the therapy of osteoporosis. The cellular mechanisms of mechanically induced signal transduction are largely unknown. Above all, it has remained elusive how cells are able to sense physical forces. Indications exist that integrin receptors may serve as mechanotransducers (10 -14). Integrins are heterodimeric transmembrane molecules by which cells adhere to the extracellular matrix (15). The ␤ subunit is combined with one of the different ␣ subunits, and both extracellular domains are involved in ligand binding. Engagement and clustering of these receptors induce signal trans-duction, which involves integrin linkage to the cytoskeleton and the generation of second messengers and biochemical events (13,16,17). Activation of protein tyrosine kinases appears to play a central role in integrin-mediated signaling (18 -20).
Because mechanical strain may act in different frequencies and strength, which appears to have relevance in regulating cell physiology (21)(22)(23), an important question is whether perception of physical forces by integrin receptors induces a differential intracellular signal transduction. Recently, we developed a method to mechanically stress cell surface receptors (14). Using magnetic beads, drag forces in defined strength and frequency can be applied to receptors of cells in a monolayer. The method enables the application of physical forces to defined integrins, which allows the evaluation of the relevance of specific integrin subunits in mechanical signal transduction. Herein, we report that in the osteosarcoma cell line U-2 OS, mechanical stressing of integrins induces an increased tyrosine phosphorylation of proteins, including the MAP 1 kinase, compared with integrin clustering and depending on whether a permanent or intermittent stress is applied. We also observed an increased physical anchorage of tyrosine-phosphorylated proteins at the cytoskeleton, which suggests that the cytoskeleton may serve as a structure where mechanical signals can switch into a chemical-signaling pathway.

EXPERIMENTAL PROCEDURES
Materials-The osteosarcoma U-2 OS cell line was obtained from American Type Culture Collection (Rockville, MD). 96-well Fluoro Nunc modules from Nunc A/S (Roskilde/Denmark) were used for plating the cells. Paramagnetic microbeads (size 2.8 m, coated with streptavidin) were purchased from Dynal (Hamburg, Germany). For coating of the microbeads, biotinylated anti-␤1 (clone 2A4) and anti-␣2 (clone AK7) integrin antibodies were from Southern Biotechnology Associates, Inc. (Birmingham, AL); biotinylated anti-CD71 (transferrin receptor) antibody was from Oncogene Science, Inc. (Uniondale, NY). For immunoprecipitation of MAP kinases, anti-ERK-1 (p44) antibody (clone C-16), which also reacts with ERK-2 (p42), was used from Santa Cruz Biotechnology. Protein A-agarose was also purchased from Santa Cruz Biotechnology. Recombinant anti-phosphotyrosine antibody (clone RC-20) conjugated with alkaline phosphatase was from Transduction Laboratories. CDP-star for chemiluminescence was obtained from Boehringer Mannheim.
Cell Culture-U-2 OS cells were cultured in Dulbecco's modified Eagle's medium and supplemented with 10% fetal calf serum at 37°C and in 5% CO 2 atmosphere. For the experiments, 100 l of cells in complete medium containing 10 5 cells were seeded into wells of a 96-well culture module and grown to near confluence. 2 h before mechanical strain was applied, the cells were depleted of serum.
Mechanical Receptor Stressing-The procedure to strain integrin receptors was described in detail elsewhere (14). In brief, the cell monolayer was incubated with paramagnetic microbeads coated with anti-␤1 or anti-␣2 antibodies. This is termed here as clustering. In average, five beads bound at the surface of one cell. To apply mechanical * This work was supported by a Grant from Deutsche Forschungsgemeinschaft, GK-Br 1255/4-1 (to C. S.) and Grants from the Ministry of Research FKZ 01ZZ9601 (to H. P. and F. D.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
‡ To whom correspondence should be addressed: Universitä t Rostock, Klinik fü r Innere Medizin, Ernst-Heydemann-Str. 6 drag forces, each of the wells was then placed between the poles of the magnetic device, which induces an inhomogeneous magnetic field (14). The forces subjected to one bead were adjusted to 10 dyne/cm 2 . Analyses revealed that this strength was exerted to most of the beads, whereas a smaller part of the beads in the vicinity of the flat pole was subjected to lower forces. The forces were applied either with a frequency of 1 Hz if not otherwise mentioned or permanently during the indicated time. In general, the following experimental designs were compared: 1) control cells without any treatment (Ϫ); 2) clustered samples (c), i.e. the cell monolayer was incubated with anti-integrin antibody-coated beads for 50 min; 3) mechanically stressed samples, i.e. the cell monolayer was incubated (in a control experiment with anti-CD71) with anti-integrin antibody-coated beads for 20 min followed by application of forces for 30 min.
Cell Lysis and Immunoblotting-After integrin stimulation as described above, adherent cells in each well were washed in phosphatebuffered saline and lysed in 10 l of SDS sample buffer. Lysates of several wells were pooled and then boiled. The samples were subjected to a 7.5% SDS-polyacrylamide gel electrophoresis. For immunoblotting, the proteins were transferred to polyvinylidene difluoride membranes. To block nonspecific binding, membranes were incubated with 5% milk powder in buffer containing 0.01% Tween 20, 100 mM Tris/HCl, pH 9.0, 155 mM NaCl for 30 min at 37°C. Immunoblotting was performed with alkaline phosphatase-labeled anti-phosphotyrosine antibody at a dilution of 1:20,000 and then visualized with chemiluminescence (CDP star).
Immunoprecipitation-To analyze activation of MAP kinases, cells were lysed in precipitation buffer containing protease inhibitor, SDS, and Triton X-100 (0.1%). After centrifugation, the supernatant was incubated with 100 l of anti-ERK antibody for 1 h at room temperature followed by adding 50 l of protein A-agarose. After centrifugation, the pellet was washed in precipitation buffer and analyzed for phosphotyrosine as described above.
Preparation of Cytoskeletal Fractions-The cell monolayer was incubated with cell extraction buffer containing 1% Triton X-100, 20 mM imidazole, 2 mM MgCl 2 , 80 mM KCl, 2 mM EGTA for 5 min at 4°C. The Triton nonsoluble fractions were then collected in SDS sample buffer and subjected to gel electrophoresis as described above.
Treatment with Cytochalasin and Calcium Chelator-To disrupt the actin filaments of the cytoskeleton, the cell monolayer was treated with 25 nM cytochalasin D for 20 min at 37°C. The cells were then washed and incubated with the microbeads to perform the procedure for mechanical loading.

RESULTS
The osteosarcoma cell line expressed the ␤1 as well as the ␣2 integrin subunits on the cell surface (14). Therefore, we examined the effect of mechanical stress applied to both integrin subunits on tyrosine phosphorylation of proteins as a mechanism in integrin-mediated signal transduction. First, we were interested in the time course of tyrosine phosphorylation due to clustering of the ␤1 integrin subunit by incubation of the cells with anti-␤1-coated microbeads. We observed an increase of phosphorylation during the time of incubation, which reached the maximum after 60 min (Fig. 1). Based on this finding, mechanical stress was applied to integrins for 30 min after an incubation time of 20 min to bind the beads to the receptors. Application of forces to the ␤1 as well as to the ␣2 subunit induced an increased tyrosine phosphorylation of proteins compared with integrin clustering alone (Fig. 2). Stressing the ␤1 chain, the effect was more pronounced than with ␣2. To prove whether the mechanically induced cellular reactions are specific for integrins, we stressed the transferrin receptor (CD71) for comparison. Although a slightly increased tyrosine phosphorylation was observed compared with untreated cells, the effect was distinctly lower than after stressing an integrin receptor (Fig. 3). Next we compared the effect of permanent mechanical loading with an intermittent stress of 1 Hz on tyrosine phosphorylation. Application of a stress with a frequency of 1 Hz induced a more profound phosphorylation than permanent drag forces (Fig. 2). To exclude the possibility that the influence of different modes of magnetic field application alone and not the mechanical receptor stress provoked the differences in tyrosine phosphorylation, we compared controls in which pure cells were subjected to a permanent and a cyclic magnetic field. This experiment clearly demonstrated that the magnetic field alone did not influence tyrosine phosphorylation, independent of the mode of the magnetic field (Fig. 3).
To evaluate the role of the cytoskeleton in mechanically induced tyrosine phosphorylation, we examined whether tyrosine-phosphorylated proteins are anchored to the cytoskeleton. After clustering and mechanically loading of the integrins, cells were extracted with Triton X-100, and the detergent insoluble fraction analyzed for tyrosine-phosphorylated proteins (Fig. 4). Both clustering and additional stress induced a linkage of tyrosine-phosphorylated proteins. Again, mechanical load was more effective than clustering. This observation concerned the ␤1 subunit, whereas the linkage of ␣2 to the cytoskeleton was similar comparing the effect of clustering and additional mechanical load. Furthermore, cytoskeletally anchored phosphorylated proteins were preferably detected in the higher molecular weight range. The anchorage of phosphorylated proteins to the cytoskeleton was further studied by disruption of Cells in a monolayer were incubated with microbeads to bind at the ␤1 or ␣2 integrin subunit followed by application of drag forces (s). These samples are compared with clustering of the integrins by incubation with the microbeads without subsequent mechanical loading (c), and untreated control cells (Ϫ). Total cell lysates were electrophoresed and blotted for anti-phosphotyrosine. Most pronounced differences are observed in the 40-kDa region. In all cases, physical forces induced a significantly enhanced tyrosine phosphorylation compared with clustering and controls. Mechanical stressing of ␤1 induced a higher response than loading the ␣2 subunit (compare ␤1/1 Hz/s versus ␣2/1 Hz/s; or ␤1/P/s versus ␣2/P/s). A cyclic stress with a frequency of 1 Hz was more effective than a permanent stress in both integrin subunits (compare ␤1/1 Hz/s versus ␤1/P/s; and ␣2/1 Hz/s versus ␣2/P/s). The results are representative of four independent experiments. the actin filaments by cytochalasin D (Fig. 5). Treatment with cytochalasin D abolished the cytoskeletal linkage of tyrosinephosphorylated proteins indicating the requirement of actin polymerization.
Because intracellular calcium is induced by integrin stimulation, we determined whether intracellular calcium plays a role in the association of tyrosine-phosphorylated proteins to the cytoskeleton. During mechanical loading of integrins, cells were treated with the intracellular calcium chelator BAPTA-AM. The following analysis of cytoskeletally anchored proteins, which are tyrosine-phosphorylated, revealed that BAPTA significantly reduced the physical anchorage of these proteins to the cytoskeleton (Fig. 6). This indicates that calcium regulates the linkage of activated proteins to the cytoskeleton during integrin-mediated mechanically induced signal transduction.
Last, we were interested in whether the increased tyrosine phosphorylation of proteins due to mechanical integrin stressing may lead to downstream signaling, which could be relevant for gene expression. Therefore, we examined whether among the tyrosine-phosphorylated proteins the MAP kinases are also activated. Immunoprecipitation of MAP kinases and analysis of tyrosine phosphorylation demonstrated that mechanical loading of both the ␤1 and the ␣2 integrin subunits induced a distinctly higher degree of activation of the MAP kinases compared with integrin clustering (Fig. 7). This suggests that regulation of gene expression by physical forces is controlled by differential activation of MAP kinases and mediated by integrins. DISCUSSION Tyrosine phosphorylation of several cellular proteins appears to play an essential role in integrin-mediated signal transduction because its inhibition blocks gene expression (24). The mechanisms by which extracellular interactions of integrins regulate tyrosine phosphorylation remains elusive. We demonstrate that application of physical forces to integrin receptors enhanced the tyrosine phosphorylation of proteins com- The cells were then extracted with Triton X-100 to obtain the insoluble cytoskeletal fraction. This fraction was then processed for anti-phosphotyrosine immunoblotting. Cytochalasin D dramatically blocked the cytoskeletal linkage of tyrosine-phosphorylated proteins. In the absence of cytochalasin D, mechanical stressing of integrins induced a distinctly enhanced anchorage of phosphorylated proteins. This represents the typical results of four independent experiments. (To obtain a background in the cytochalasin-treated samples, the blots in these experiments were exposed longer to the film.) FIG. 6. Influence of the calcium chelator BAPTA-AM on the anchorage of tyrosine-phosphorylated proteins. Cells were treated by mechanical stressing of the ␤1 integrin subunit (1 Hz) (s), clustering of ␤1 (c), or without treatment (Ϫ) in the presence of BAPTA-AM (ϩBAPTA) to chelate intracellular calcium. The cells were then extracted with Triton X-100, and the insoluble fraction was processed for anti-phosphotyrosine immunoblotting. BAPTA-AM significantly inhibited the anchorage of tyrosine-phosphorylated proteins to the cytoskeleton, which is most obvious in the higher molecular weight range. For comparison, in the absence of BAPTA-AM (ϪBAPTA), the most profound tyrosine phosphorylation of cytoskeletally linked proteins was found due to mechanical stressing of the ␤1 integrin subunit. The results are representative of four independent experiments. pared with integrin clustering alone. These results support the idea that integrin receptors function as mechanosensors (11,12,25). It indicates that physical forces applied to integrin receptors determine the degree of tyrosine phosphorylation that may have consequences in gene expression. Furthermore, tyrosine phosphorylation was also quantitatively influenced by the mode of receptor stressing. Intermittent forces are more effective than a permanent stress. Several reports have shown that application of an intermittent mechanical load on bones is more effective to stimulate growth than a permanent strain (26,27). Therefore, increased tyrosine phosphorylation due to cyclic integrin stressing possibly provides a cellular-signaling mechanism for the differential effect of mechanical loads on bones. One of the initial effects of a mechanical stress to integrins is the induction of their cytoskeletal anchorage. This has been demonstrated in migrating cells by integrin cross-linking or mechanical twisting of the integrins (11,28,29). It is an intriguing aspect that the strength of this linkage is variable and depends on external forces. Using an optical trap to restrain integrin bound beads resulted in a proportional strengthening of the integrin cytoskeletal linkage (30). In the line of these findings, we suggest that in our experiments mechanical forces induced an increased strength of the integrin cytoskeletal linkage. Other experiments have revealed that the characteristics of the ligand and receptor-ligand interaction control components of the integrin-mediated signal transduction resulting in functional consequences (31)(32)(33). For example, integrin clustering induced a restricted accumulation of signaling molecules into the cytoskeletal complex compared with additional occupancy of the functional receptor epitope, which led to a model of a hierarchy in signal transduction (34). Furthermore, the physical characteristics of the collagen matrix was affecting integrin activation and signal transduction, which inhibited proliferation (32). This clearly suggests that how integrins interact with a ligand is of importance for the induced signal transduction. Although the ␤ and ␣ subunits of integrins form a dimer, we have found a difference in mechanically induced tyrosine phosphorylation between the two integrin subunits. Mechanical load on the ␤ subunit provoked a more profound tyrosine phosphorylation than stressing the ␣ subunit. We cannot exclude that this finding was due to a higher expression of ␤1 compared with ␣2 that we have found in these cells (14), but it could reflect a different mechanism of the intracellular interaction of the cytoplasmic domains of both integrin subunits. The ␤ chain can directly interact with the cytoskeletal proteins ␣ actinin and talin (35,36). In addition, the integrin-linked kinase can associate with the ␤1 cytoplasmic domain and mediate further downstream signaling (37).
Mechanical stress to the ␤ subunit provoked a significant anchorage of tyrosine-phosphorylated proteins to the cytoskeleton, which was increased compared with integrin clustering. Tyrosine phosphorylation of cytoskeletally anchored proteins could be a prerequisite to form the cytoskeletal complex (38), and a higher degree of phosphorylation may be a prerequisite for the higher strengthening between receptors and cytoskeleton. Regarding the factors that determine the association of activated signaling molecules to the cytoskeleton, we have found that intracellular calcium is obviously an important regulator of the immobilization of proteins to the cytoskeleton. This concerns not only intracellular-signaling proteins but also the cytoskeletal anchorage of integrins to the cytoskeleton (28). The role of calcium for a mechanically induced signal transduction is also stressed by data that have shown that intracellular calcium concentrations correlated with increasing force levels applied to integrins (39). However, our previous experiments suggest that the differential cytoskeletal anchorage of tyrosinephosphorylated proteins and integrin subunits due to stimulation of ␤1 compared with the ␣ subunit is not controlled by differences in the magnitude of the calcium response. Incubation of cells with anti-integrin antibodies prior to mechanical stimulation of the cells (13), as well as preliminary results concerning the comparison of the calcium responses due to mechanical stress applied with magnetic beads to ␤1 and ␣2, revealed no quantitative differences in calcium signaling.
Concerning downstream signaling, we argue that the cytoskeleton could represent a structure where physical forces are transformed into a biochemical signal pathway. The differential anchorage of tyrosine-phosphorylated proteins due to physically stimulated integrins may regulate downstream intracellular-signaling events.
One of these events is the activation of MAP kinases as a key mechanism to control the activation of transcription factors, which therefore mediates gene expression. The involvement of this pathway in integrin signaling has been established (40,41). We found that activation of the MAP kinases was significantly increased due to physical forces compared with integrin clustering. Due to the key role of the MAP kinases, our finding emphasizes that physical forces transduced by integrins differentially regulate cell proliferation and the expression of genes through the MAP kinase cascade. The fact that activation of MAP kinase by integrins depends on an intact cytoskeleton (41,42) and the involvement of cytoskeletally associated signaling molecules like focal adhesion kinase (43) highlights the significance of a controlled cytoskeletal anchorage of tyrosine-phosphorylated proteins for consequences in cell behavior. Because the integrin-mediated MAP kinase pathway converges with growth factor-induced pathways (44), our result suggests a synergistic effect of mechanical forces and cytokines in the regulation of cell function.
In conclusion, integrins mediate physical forces and may regulate physiological consequences in the cell by a well tuned induction of the degree of tyrosine phosphorylation of proteins. A significant aspect is the cytoskeletal anchorage of activated signaling proteins, which depends on the mobilization of intracellular calcium. The functional relevance of these mechanisms is supported by the result of an enhanced activation of MAP kinases due to mechanical integrin stimulation. FIG. 7. Induction of the activation of MAP kinases by mechanical stressing of integrin receptors. Cells were treated by mechanical stressing (1 Hz) of the ␤1 or the ␣2 integrin subunits (s), clustering of the subunits (c), or without treatment (Ϫ). After cell lysis, MAP kinases were immunoprecipitated with anti-ERK antibody. After separation in the electrophoresis, the samples were probed with antiphosphotyrosine antibody. Mechanical stressing of both the ␤1 and the ␣2 integrin subunit induced an enhanced tyrosine phosphorylation of MAP kinases p42 (ERK-2) and p44 (ERK-1) compared with controls and clustering of the integrins. The results are representative of three independent experiments.