INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

THE ACTIN CYTOSKELETON and the integrin family of cell adhesion molecules have been shown to play important roles in mechanotransduction (23, 39). Integrins are heterodimeric adhesion molecules composed of an -subunit and a -subunit that physically link the extracellular matrix (ECM) and the actin cytoskeleton by interacting with ECM proteins outside the cell and with bundles of actin filaments (stress fibers) inside the cell (5, 19). The organization of actin and myosin filaments into contractile stress fibers is thought to increase internal tension in cells and is regulated, in part, by the GTPase Rho (5, 32). Growing evidence suggests that the development of internal tension by actin and myosin plays a central role in signal transduction from the ECM to the nucleus to regulate gene expression (21, 34, 41). Ligand binding stimulates the clustering of integrins in the membrane and causes the attachment of stress fibers to integrins at specialized sites of cell-ECM attachment called focal adhesions (5). Several linker proteins anchor actin filaments to integrins at focal adhesions, including -actinin, vinculin, and talin (16, 24, 28, 33). In this study, we investigated the roles Rho-mediated stress fiber development and the anchoring of stress fibers to integrins had in the transmission of fluid shear forces into intracellular signals that upregulate gene expression and stimulate new bone formation.

Mechanically induced bone formation is preceded by expression of the transcription factor c-Fos and prostaglandin production (8, 38). The inducible isoform of the enzyme cyclooxygenase (COX-2) is important in mechanotransduction, since selective inhibition of COX-2 eliminates mechanically induced bone formation in the rat tibia (12). Mechanical forces appear to be transduced to bone cells by fluid flow-induced shear stress in the canaliculi and canals within the bone tissue (25, 35). The organization of the actin cytoskeleton and its attachments to integrins may be part of a sensing apparatus used by cells to detect and respond to mechanical signals. The term autobaric has been proposed to describe a process in which cells use the cytoskeleton to apply an internal load to themselves that is part of an intracellular signaling mechanism (4). In this study, we found that application of fluid shear to cultured osteoblasts induced cytoskeletal changes that were consistent with an increase in internal load. These changes in response to fluid shear included development of stress fibers and focal adhesions and recruitment of ₁-integrins and -actinin into focal adhesions. Fluid shear also increased expression of both c-Fos and COX-2 in osteoblasts. Three lines of evidence support a critical role for actin cytoskeleton reorganization, attachment of actin stress fibers to integrins in focal adhesions, and the development of internal tension in mediating fluid shear-induced mechanochemical signal transduction in osteoblasts. First, inhibiting fluid shear-induced cytoskeletal reorganization by treatment with cytochalasin D blocked the expression of COX-2. Second, microinjection of single cells with a fragment of -actinin that prevents fluid shear-induced stress fiber development by competing with endogenous -actinin for binding to integrin cytoplasmic tails (26) blocked COX-2 and c-Fos expression. Third, expression of COX-2 and c-Fos was inhibited in cells transfected with a dominant negative mutant of the GTPase Rho, which blocks fluid shear-induced stress fiber and focal adhesion formation. Together, these results suggest that the development of stress fibers and their anchorage to the membrane at focal adhesions in response to fluid shear play a critical role in transducing mechanical signals applied at the cell surface into intracellular signals that are necessary for increased gene expression and new bone formation.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Cell culture and fluid flow. The mouse osteoblast cell line MC3T3-E1 was cultured in -MEM containing 10% FCS and maintained in 5% CO₂ at 37°C. The MC3T3-E1 cell line has a reproducible phenotype with a distinct proliferative stage and, on reaching confluence, a differentiated stage in which osteoblastic markers such as alkaline phosphatase are expressed. For microscopy, cells were grown on glass slides coated with 10 µg/ml fibronectin (Sigma Chemical). Fluid flow was performed in chambers using the flow loop designed by Frangos et al. (14) and marketed by Cytodyne (San Diego, CA). This system produces laminar flow over a cell monolayer. The system was maintained at 37°C, and the medium was bubbled with 5% CO₂-95% air during flow experiments. Cells were exposed to fluid flow in -MEM containing 1% serum. A flow sensor (SWF-5 flowmeter, Zepeda Instruments, Seattle, WA) incorporated into the flow loop was used to monitor flow rate. The flow rate was 18 ml/min, which yielded a shear stress of 12 dyn/cm². Control cells were placed in -MEM containing 1% serum but not subjected to flow. For experiments using cytochalasin D, cells were pretreated for 1 h with 10 µM cytochalasin D in -MEM with 10% serum and then subjected to flow in -MEM containing 10 µM cytochalasin D with 1% serum.

Antibodies. Antibodies were obtained as follows: one integrin antibody (no. 763) was generated by immunizing rabbits with a peptide corresponding to residues 758-774 of the human ₁-integrin cytoplasmic tail, -actinin monoclonal antibody BM75 was purchased from Sigma, COX-2 and c-Fos antibodies were purchased from Santa Cruz, and all fluorescently labeled secondary antibodies were purchased from Jackson Immunoresearch.

Western blot analysis. For Western blotting, cells on glass slides were lysed in 1 ml of Tris-buffered saline containing 1% Triton X-100, 1% deoxycholate, and 0.5% SDS, along with the protease inhibitors phenylmethylsulfonyl fluoride, aprotinin, and leupeptin. Cells were scraped from the substrate after 5 min in lysis buffer, and the insoluble material was removed by centrifugation at 12,000 g for 15 min. The supernatant was transferred to a fresh tube, and the protein concentration was determined using bicinchoninic acid reagent (Pierce Chemical); 10-µg samples were loaded onto 10% SDS-PAGE gels for separation and transferred to nitrocellulose for immunoblot analysis. Equal protein loadings were confirmed by Coomassie blue staining of gels run in parallel. Each experiment was carried out a minimum of four times, scanning densitometry of bands was performed with the Bio-Rad Molecular Analyst program, and statistical analysis was made by Kruskal-Wallis one-way ANOVA.

Northern Blot analysis. Total RNA was extracted from the cells using TRIzol reagent (GIBCO). Ten micrograms of total RNA were run on a 1% agarose-0.44 M formaldehyde gel, transferred to a Hybond nylon membrane (Amersham) by capillary blotting, and fixed to the membrane by ultraviolet irradiation. Filters were hybridized for 2 h at 65°C in a rapid hybridization buffer (Amersham) containing radiolabeled cDNA probes. Rat COX-2, c-Fos, and glyceraldehyde-3-phosphate dehydrogenase cDNA probes were labeled with [³²P]dCTP (specific activity > 3,000 Ci/mmol; NEN), using a random prime labeling kit (Boehringer Mannheim). Filters were washed and exposed to Fuji Rx film at 80°C. Northern blot analysis was carried out in four separate experiments, bands were analyzed by scanning densitometry, and the statistical significance of differences in expression was confirmed by ANOVA.

Microinjection and fluorescence microscopy. For microinjection, cells grown on glass slides were placed in a 100-mm dish with DMEM containing 15 mM HEPES (pH 7.3). Microinjection was performed at room temperature using a Leitz micromanipulator with needles prepared from glass capillaries pulled on a two-stage needle puller, as previously described (26). After injections, cells were returned to the incubator at 37°C for 1 h before being subjected to fluid shear. At the end of the flow period, cells were fixed in 4% paraformaldehyde and processed for immunofluorescence microscopy as previously described (29), using appropriate primary and fluorescently labeled secondary antibodies, rhodamine-phalloidin or FITC-phalloidin (Molecular Probes). Images were recorded on Tmax 400 film (Kodak) using a Nikon Optiphot II microscope through either ×60 or ×100 planapo objectives (1.4 numerical aperture). The effect of -actinin fragment microinjection on cytoskeletal organization and gene expression following fluid shear was confirmed by analysis of ~35 injected cells stained with phalloidin and 60 injected cells stained with anti-c-Fos or anti-COX-2.

Transfection of MC3T3-E1 cells. A pEXV expression vector containing N19 Rho was purified by CsCl density gradient centrifugation. MC3T3-E1 cells were washed in serum-free DMEM and transiently transfected using Lipofectin (GIBCO). Cells were returned to -MEM containing 10% FCS and cultured for 48 h before use in experiments. Transfected cells were identified using the myc epitope tag antibody 9E10. Control cells were transfected with a vector expressing the myc epitope tag only.

Purification and labeling of the 53-kDa -actinin fragment and intact -actinin. -Actinin was purified from chicken gizzard smooth muscle as previously described (11). Purified -actinin was digested with thermolysin, and the 53-kDa integrin-binding fragment was purified on a fast protein liquid chromatography MonoQ anion exchange column (26). Intact -actinin was conjugated to iodoacetaminotetramethyl rhodamine (Molecular Probes), and the purified 53-kDa fragment was conjugated to rhodamine isothiocyanate. Both proteins were dialyzed extensively against microinjection buffer (75 mM KCL, 10 mM KHPO₄, and 0.1% -mercaptoethanol) and concentrated to 5 mg/ml by ultrafiltration before use for microinjection experiments.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Application of fluid shear to MC3T3-E1 osteoblasts induces development of actin stress fibers and formation of focal adhesions containing ₁-integrins and -actinin. Figure 1 illustrates the dramatic changes in actin filament and focal adhesion organization in MC3T3-E1 osteoblasts that occur following fluid shear. Before fluid shear, actin filaments, visualized by staining with FITC-phalloidin, were poorly organized in MC3T3-E1 cells, with only a few thin stress fibers being detectable (Fig. 1, A and E). After the application of fluid shear for 60 min at 12 dyn/cm², actin filaments became organized into stress fibers that were thicker and more abundant than in nonflowed cells (Fig 1, B and F). Immunostaining with antisera directed against the integrin ₁-subunit demonstrated that, before fluid shear, ₁-integrins were diffusely distributed over the surface of cells (Fig. 1C). However, after fluid shear, ₁-integrins became concentrated in the focal adhesions that formed at the termini of the stress fibers (Fig. 1D). Similarly, -actinin (Fig. 1, G and H) was localized with a periodic distribution along the stress fibers and was also present in the focal adhesions at stress fiber termini (Fig. 1H) after fluid shear.

View larger version (128K): [in this window] [in a new window]   Fig. 1.   MC3T3-E1 osteoblasts respond to fluid shear by developing prominent stress fibers and recruiting 1-integrins to focal adhesions. Cells that were not subjected to fluid shear (A and E) or were subjected to fluid shear at 12 dyn/cm2 for 60 min (B and F) were stained with rhodamine-phalloidin to visualize F-actin. Fluid shear induced development of prominent F-actin stress fibers that were oriented roughly parallel to long axis of cell, whereas stress fibers in cells not subjected to flow were smaller and randomly oriented. Cells were also labeled with antibodies against 1-integrin (C and D) or -actinin (G and H). Before fluid flow, each of these proteins were diffusely localized in cells. However, after fluid flow, 1-integrins (D) were recruited to focal adhesions and -actinin (H) became localized periodically along stress fibers and in focal adhesions located at termini of prominent stress fibers. Scale bar, 20 µm.

Osteoblasts upregulate COX-2 and c-Fos expression in response to fluid shear. Application of fluid shear (60 min at 12 dyn/cm²) to MC3T3-E1 osteoblasts also induced a dramatic increase in the intensity of immunostaining for the enzyme COX-2 (Fig. 2, A and B) and the nuclear protein c-Fos (Fig. 2, C and D). Western blotting of protein from Triton X-100 extracts (Fig. 3A) and Northern blot analysis of mRNA (Fig. 3B) from nonflowed and flowed cells confirmed that fluid flow (1 h of flow at 12 dyn/cm²) significantly increased expression of COX-2 and c-Fos protein and mRNA levels (P < 0.005). In contrast to COX-2 and c-Fos, levels of the cytoskeletal proteins -actinin, vinculin, and talin were not significantly different after fluid flow compared with nonflowed controls (Fig. 3). Figure 4 shows fields of cells at lower magnification than in Figs. 1 and 2 and demonstrates 1) that fluid shear induced changes in actin organization (increased stress fiber and focal adhesion development) in essentially all the cells in the field, 2) that fluid shear induced an elongated cell shape compared with nonflowed controls, and 3) that essentially all cells respond to fluid flow by increasing c-Fos expression.

View larger version (105K): [in this window] [in a new window]   Fig. 2.   Fluid shear induces expression of COX-2 and c-Fos in osteoblasts. Immunofluorescence microscopy of MC3T3-E1 osteoblasts using antibodies against COX-2 (A and B) or c-Fos (C and D) demonstrates increased expression of both COX-2 and c-Fos proteins following fluid shear for 1 h at 12 dyn/cm2 (B and D) compared with cells not subjected to flow (A and C). Scale bar, 20 µm.

View larger version (66K): [in this window] [in a new window]   Fig. 3.   Fluid shear induces expression of COX-2 and c-Fos protein and mRNA in osteoblasts. A: immunoblot analysis of protein extracts from cells not subjected to fluid shear (No flow) or cells subjected to fluid shear for 1 h at 12 dyn/cm2 (1 h flow) confirms increased expression of both c-Fos and COX-2 protein following fluid shear. Expression of cytoskeletal proteins -actinin, vinculin, and talin, in contrast, were not altered by fluid shear. Ten micrograms of protein were loaded onto each lane of gel. B: Northern blot analysis of mRNA from control cells not subjected to fluid shear or from cells subjected to fluid shear at 12 dyn/cm2 for 1 or 3 h demonstrated increased COX-2 and c-Fos mRNA expression following fluid shear. A probe to glyceraldehyde-3-phosphate dehydrogenase (GAPDH) confirmed equivalent mRNA loadings on gel.

View larger version (133K): [in this window] [in a new window]   Fig. 4.   Increased expression of c-Fos correlates with stress fiber development following fluid shear. A and B: viewed at low magnification, F-actin staining with rhodamine-phalloidin demonstrates high proportion of cells that develop stress fiber bundles following fluid flow (B; flow was from right to left), whereas nonflowed cells have randomly oriented F-actin structures that are not organized in bundles (A). C and D: overlaid F-actin and c-Fos immunofluorescence images demonstrate that a high proportion of cells that develop stress fibers in response to fluid shear (D) also express higher levels of c-Fos than nonflowed control cells (C). Scale bar, 40 µm.

Disruption of microfilaments with cytochalasin D inhibits fluid shear-induced increases in COX-2 protein expression. To evaluate the role of the actin cytoskeleton in fluid shear-induced upregulation of gene expression, cells were pretreated with 10 µM cytochalasin D for 1 h and then subjected to fluid shear for 30 min or 1 h in the presence of 10 µM cytochalasin D. Western blot analysis of protein extracts from untreated control cells and cytochalasin D-treated cells demonstrated that microfilament disassembly by cytochalasin D resulted in decreased COX-2 expression compared with untreated controls (Fig. 5). Analysis by scanning densitometry of four separate experiments demonstrated that COX-2 expression increased an average of 2.7-fold following 1 h of fluid shear compared with nonflowed controls and was significant as determined by ANOVA (P < 0.05). Pretreatment of cells with cytochalasin D caused COX-2 expression following 1 h of fluid shear to not be significantly different from nonflowed controls. Immunofluorescence microscopy confirmed the inability of fluid shear to induce reorganization of the actin cytoskeleton (Fig. 6, A and B) or to increase the intensity of immunostaining for COX-2 (Fig. 6, C and D) when cells were treated with cytochalasin D. Unlike COX-2, c-Fos protein expression was relatively unaffected by treatment with cytochalasin D (not shown).

View larger version (100K): [in this window] [in a new window]   Fig. 5.   Inhibition of actin cytoskeleton with cytochalasin D inhibits fluid shear-induced upregulation of COX-2 protein. Protein extracts were prepared from cells not subjected to fluid shear (No flow) or from cells subjected to fluid shear at 12 dyn/cm2 for 30 min (30-min flow) or 60 min (60-min flow) in either absence () or presence (+) of 10 µM cytochalasin D, as described in MATERIALS AND METHODS. Equal amounts of protein were loaded onto SDS-PAGE gels, transferred to nitrocellulose, and blotted for COX-2. Top: Coomassie blue staining of a gel to confirm equivalent protein loadings. MM, molecular mass. Bottom: an immunoblot for COX-2 that demonstrates decreased levels of expression in cells subjected to either 30 or 60 min of fluid flow in presence of cytochalasin D compared with controls not treated with drug.

View larger version (111K): [in this window] [in a new window]   Fig. 6.   Immunofluorescence microscopy of cells subjected to fluid shear in presence of cytochalasin D fail to develop stress fibers or upregulate COX-2 protein expression. Staining with rhodamine-phalloidin confirms that, in absence (A and C) or presence (B and D) of fluid flow over cells treated with 10 µM cytochalasin D, stress fibers fail to develop (A and B) and cells also fail to upregulate COX-2 protein expression (C and D). Scale bar, 20 µm.

Microinjection of the integrin-binding domain of -actinin inhibits linkage of actin filaments to integrins and blocks fluid shear-induced cytoskeletal reorganization and gene expression. We next used an alternative approach to inhibit reorganization of the actin cytoskeleton that avoids the potential pharmacologic complications frequently associated with use of cytochalasin D. For these experiments, we microinjected a proteolytic fragment of the microfilament-integrin linker protein -actinin into cells to inhibit the function of endogenous -actinin. -Actinin can be cleaved by proteolysis into major fragments of 27 and 53 kDa using the enzyme thermolysin. The 27-kDa fragment contains an actin-binding site, and the 53-kDa fragment binds to the cytoplasmic domain of ₁-, ₂-, and ₃-integrins. When microinjected into cells such as fibroblasts that normally contain well-developed stress fibers, the 53-kDa integrin-binding fragment of -actinin causes disassembly of existing stress fibers by competitively displacing the endogenous -actinin from focal adhesion and causing detachment of stress fibers from the focal adhesions (26). Figure 7 illustrates that microinjection of the 53-kDa integrin-binding fragment of -actinin completely blocked the fluid shear-induced development of stress fibers in osteoblasts. Observation of ~45 microinjected cells stained with phalloidin showed that stress fibers never formed, strongly suggesting that -actinin plays a critical role in mediating the anchorage of actin filaments to integrins in focal adhesions that is required for stress fiber formation to occur.

View larger version (87K): [in this window] [in a new window]   Fig. 7.   Microinjection of 53-kDa -actinin fragment into osteoblasts blocks fluid shear-induced development of stress fibers. Cells microinjected with rhodamine-labeled 53-kDa fragment (A and C) were returned to incubator for 1 h before being subjected to fluid shear at 12 dyn/cm2 for 1 h and then fixed and stained with FITC-phalloidin to visualize F-actin (B and D). Cells injected with 53-kDa fragment failed to develop stress fibers in response to fluid shear. Scale bar, 20 µm.

View larger version (96K): [in this window] [in a new window]   Fig. 8.   Microinjection of 53-kDa -actinin fragment blocks fluid shear-induced COX-2 protein expression in osteoblasts. Cells microinjected with rhodamine-labeled 53-kDa fragment (A and C) were returned to incubator for 1 h before being subjected to fluid shear at 12 dyn/cm2 for 1 h and then fixed and stained with an antibody against COX-2, followed by a FITC-labeled secondary antibody (B and D). Microinjected cells, which are outlined, failed to upregulate COX-2 protein in response to fluid shear. Scale bar, 20 µm.

View larger version (123K): [in this window] [in a new window]   Fig. 9.   Microinjection of 53-kDa -actinin fragment blocks fluid shear-induced c-Fos protein expression. Cells microinjected with rhodamine-labeled 53-kDa fragment (A and C) were returned to incubator for 1 h before being subjected to fluid shear at 12 dyn/cm2 for 1 h and then fixed and stained with an antibody against c-Fos, followed by a FITC-labeled secondary antibody (B and D). Microinjected cells failed to upregulate c-Fos protein in response to fluid shear. Scale bar, 20 µm.

View larger version (139K): [in this window] [in a new window]   Fig. 10.   Microinjection of intact -actinin does not inhibit fluid shear-induced stress fiber formation or COX-2 or c-Fos protein expression. Cells microinjected with rhodamine isothiocyanate-labeled intact -actinin (A, C, and E) were returned to incubator for 1 h before being subjected to fluid shear for 1 h at 12 dyn/cm2 and then fixed. Cells were stained with FITC-phalloidin (B), COX-2 antibody (D), or c-Fos antibody (F), followed by an FITC-labeled secondary antibody. Scale bar, 20 µm.

Expression of a dominant negative Rho mutant blocks stress fiber formation and decreases expression of COX-2 and c-Fos protein in osteoblasts. To further investigate the relationship between fluid shear-induced stress fiber/focal adhesion formation and gene expression in osteoblasts, cells were transfected with a dominant negative Rho mutant (N19 Rho) as an alternate method of inhibiting stress fiber formation. Cells expressing N19 Rho were detected by staining with an antibody directed against a myc epitope tag engineered into the expressed protein. After fluid shear for 60 min at 12 dyn/cm², cells transfected with N19 Rho failed to develop stress fibers (Fig. 11, A and B) and did not increase expression of COX-2 (Fig. 11, C and D) or c-Fos (Fig. 11, E and F). Thus Rho activity is required for flow-induced stress fiber development and increased expression of both COX-2 and c-Fos proteins. Control cells transfected with the myc-tagged vector alone did not affect stress fiber formation or gene expression.

View larger version (93K): [in this window] [in a new window]   Fig. 11.   Expression of a dominant negative Rho mutant (N19 Rho) inhibits fluid shear-induced stress fiber formation and blocks fluid shear-induced COX-2 and c-Fos protein expression. Cells were transfected with N19 Rho, cultured for 48 h, subjected to fluid shear at 12 dyn/cm2 for 1 h, and then fixed. Transfected cells were detected using an antibody against a myc epitope tag (A, C, and E). Cells were stained with FITC-phalloidin (B), an antibody against COX-2 (D), or an antibody against c-Fos (F). Scale bars (in A for A-D, in E for E and F), 20 µm.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

In this study we showed that MC3T3-E1 osteoblasts respond to fluid shear by increasing expression of the enzyme COX-2 and the early response gene product c-Fos. Increased expression of these proteins correlated with changes in the organization of the actin cytoskeleton, including the development of contractile actin stress fibers and increased formation of focal adhesions that contain ₁-integrins. Three distinct approaches to inhibiting cytoskeletal function, namely, treatment with cytochalasin D, microinjection of cells with the 53-kDa fragment of -actinin, and transfections of cells with a dominant negative Rho, each blocked the increased expression of COX-2 following fluid shear, whereas c-Fos expression was inhibited by the 53-kDa -actinin fragment and dominant negative Rho. These results suggest that the actin cytoskeleton plays a critical role in the transmission of fluid shear-mediated mechanical signals that lead to increased COX-2 and c-Fos expression.

Because -actinin is an actin-binding protein that can cross-link actin filaments and interact directly with the cytoplasmic domain of integrin subunits, -actinin appears to be capable of dual functions: bundling actin filaments into stress fibers and linking stress fibers to integrins at the cytoplasmic face of the plasma membrane in focal adhesions. The intracellular distribution of -actinin is consistent with these dual functions. When microinjected into cells, fluorescently labeled intact -actinin is incorporated both into the focal adhesions and along stress fibers with a periodic distribution (26, 29). After microinjection into fibroblasts, the 53-kDa integrin-binding fragment of -actinin, however, initially concentrates in focal adhesions and then rapidly causes the disassembly of preexisting stress fibers and focal adhesions (26).

The inability of osteoblasts that were microinjected with the integrin-binding 53-kDa fragment of -actinin to develop stress fibers or focal adhesions or to upregulate COX-2 or c-Fos protein suggests that -actinin plays a critical role in transmitting mechanical signals across the membrane to the cytoskeleton. Experiments with the dominant negative Rho mutant (N19 Rho) further suggest that stress fiber formation mediated by Rho and linkage of stress fibers to integrins at the cytoplasmic face of focal adhesions are necessary for the transmission of mechanical signals received at the cell surface into intracellular signals necessary for increased gene expression and new bone formation. This suggests that osteoblasts may experience an increase in intracellular tension resulting from the formation of actin/myosin-containing stress fibers. The linkage of stress fiber to integrins, in response to fluid shear, may play a critical role in transducing mechanical signals into intracellular signals that precede increased gene expression and new bone formation. Additionally, it is possible that actin filaments anchored to the cell membrane may function to facilitate the trafficking of signaling molecules from the membrane to the nucleus.

Mechanotransduction in bone tissue involves several steps: 1) mechanochemical transduction of the signal (the subject of this study), 2) cell-to-cell signaling, and 3) increased number and activity of osteoblasts (10). Cell-to-cell signaling after a mechanical stimulus involves prostaglandins, especially those produced by the inducible isoform of cyclooxygenase (COX-2) (8, 12), and nitric oxide (13, 38). Prostaglandins induce new bone formation by promoting both proliferation and differentiation of osteoprogenitor cells. The selective inhibition of COX-2 in the rat tibial four-point bending model eliminates mechanically induced bone formation (12). A role for the actin cytoskeleton and ₁-integrins in the metabolic response of osteoblasts and osteosarcoma cells to mechanical strain has also been observed (6). Thus the direct linkage between contractile actin stress fibers and mechanically induced COX-2 and c-Fos expression observed in the present study is relevant to mechanically induced bone formation in vivo. It should be noted that these experiments were carried out under conditions of steady fluid flow. Bearing in mind that cells in vivo may experience fluid movements that are pulsatile and/or oscillatory in nature, the response of cells in our in vitro system will also need to be examined under conditions of nonsteady fluid flow.

There is now evidence of at least two mechanochemical transduction pathways within osteoblasts. One clearly involves the actin cytoskeleton. For example, expression of osteopontin in embryonic chick osteoblasts, which is upregulated in response to mechanical stimuli, is dependent on an intact actin cytoskeleton, since treatment with the drug cytochalasin D blocks osteopontin expression (34). Furthermore, focal adhesion kinase (p125^FAK) is tyrosine phosphorylated and becomes associated with the actin cytoskeleton in response to mechanical stimuli. This strongly suggests that recruitment of focal adhesion proteins to sites of integrin aggregation may play a role in promoting mechanically induced cytoskeletal reorganization. These results are consistent with the existence of an internal loading mechanism that detects and responds to external mechanical signals as previously proposed (4, 39). Electron microscopy of alveolar bone has revealed actin filaments in situ that are organized into bundles that appear to be similar to the stress fibers of cultured cells, suggesting that stress fibers are relevant in vivo (40).

A second pathway involving G protein-linked mechanosensors in the cell membrane, calcium, and the cytoskeleton may also be involved in mechanical signaling (1, 9, 18, 19). In bone cell culture, fluid shear increases production of prostaglandins and nitric oxide within minutes (20, 30). This stimulation of prostaglandins can be 70-80% blocked by the G protein inhibitors guanosine 5'-O-(2-thiodiphosphate) and pertussis toxin, indicating that a G protein-associated mechanotransducer attached to the cell membrane may be responsible for prostaglandin production (31). Constitutive isoforms of cyclooxygenase (COX or prostaglandin synthase) and nitric oxide synthase are typically bound to the cell membrane and thus would be available for mechanochemical transduction involving G proteins (36, 37). It is possible that the cytoskeletal and G protein-linked mechanosensor pathways are tightly coupled; our experiments do not yet address the interrelationships between these pathways.

This study provides direct evidence that linkage of actin stress fibers to integrins in focal adhesions via -actinin is a critical link in the mechanical signal transduction pathway in osteoblasts, since blocking this linkage prevents fluid shear-induced COX-2 and c-Fos expression. The selective inhibition of -actinin-integrin interaction by the 53-kDa -actinin fragment is a useful approach for assaying the role of stress fiber formation and increased intracellular tension in mechanically induced gene induction. Interpreting the results of experiments using pharmacologic inhibition of actin filament integrity or kinase activity in mechanically induced signal transduction is complicated by possible unintended side effects of the drugs on signaling pathways not involving the cytoskeleton. Competitive inhibition of -actinin-mediated stress fiber linkage to integrins, however, is less likely to have unanticipated effects on noncytoskeletal intracellular signaling pathways. Interpreting the results of the present studies is therefore simplified, since -actinin fragment microinjection is unlikely to directly affect this or other non-cytoskeleton-mediated signaling pathways.

The Ras-like GTPase, Rho, has been shown to control stress fiber and focal adhesion formation in fibroblasts (32) in response to growth factor and Ras signaling. We have shown that N19 Rho, a dominant negative mutant of Rho, blocks stress fiber and focal adhesion formation in response to shear stress. The Rho effector Rho-kinase has been shown to mediate Rho-dependent stress fiber formation, probably by increasing phosphorylation of the myosin II regulatory light chain (2, 3, 22). Moreover, Rho-kinase induces transcriptional activation of the c-Fos serum response element (7). Rho-kinase is therefore a likely candidate for mediation of shear stress-induced stress fiber and focal adhesion formation in osteoblasts.

How then does Rho activity affect COX-2 and c-Fos expression? It is possible that Rho regulates stress fiber formation, on the one hand, and expression of COX-2 and c-Fos, on the other, via distinct pathways, perhaps both involving Rho-kinase. In fibroblasts, disruption of the actin cytoskeleton by treatment with cytochalasin D (15) did not inhibit the ability of Rho-mediated signaling, stimulated by lysophosphatidic acid (LPA), to induce c-Fos transcription, suggesting that LPA-stimulated Rho signaling to the nucleus is not mediated by intact cytoskeletal structures. The rapid response of osteoblasts to fluid shear also argues that c-Fos and COX-2 expression in response to mechanical stimuli may be mediated by a regulatory pathway that is distinct from LPA-mediated signaling. In our studies, fluid shear-induced expression of COX-2 and c-Fos was inhibited by both N19 Rho and by injection of the 53-kDa -actinin fragment, suggesting that Rho could influence expression of the genes for COX-2 and c-Fos via stress fiber formation in response to a mechanical stimulus. However, although cytochalasin D inhibited fluid shear-induced COX-2 expression, surprisingly, this drug did not inhibit c-Fos expression. Although we cannot yet explain the different response of these two genes to cytochalasin D, it is possible that the mechanical signaling pathways leading to increased expression of COX-2 and c-Fos are distinct. Indeed, other pharmacologic effects of cytochalasin D, unrelated to the actin cytoskeleton, may stimulate c-Fos expression through a mechanism that is independent of a mechanical stimulus. Further experiments will be required to clarify these points.

Recent evidence of linkages between the ECM and nucleoplasm involving the cytoskeleton (23) predict that mechanotransduction through integrin-cytoskeleton complexes may regulate gene expression. The studies described here provide direct evidence to support this prediction. Specifically, our results indicate that Rho-dependent stress fiber formation in response to fluid shear is involved in the mechanically induced upregulation of COX-2 and c-Fos in MC3T3-E1 osteoblasts. Furthermore, the increased intracellular tension that results from stress fiber formation is likely to be an important part of this signaling pathway, since the selective inhibition of -actinin-mediated stress fiber-integrin linkage blocked fluid shear-induced COX-2 and c-Fos protein expression. It is also possible that changes in intracellular contractility alter the conformation of the nuclear matrix and activate matrix-associated transcription factors. Future studies should aim to determine the mechanisms of transcriptional control that are affected by the contractile actin cytoskeleton in mechanically responsive cells such as osteoblasts.