Fluid Flow Induction of Cyclo-Oxygenase 2 Gene Expression in Osteoblasts Is Dependent on an Extracellular Signal-Regulated Kinase Signaling Pathway†‡
Presented in part at the 22nd annual meeting of the American Society for Bone and Mineral Research, Toronto, Ontario, Canada, September 25, 2000
The authors have no conflict of interest
Abstract
Mechanical loading of bone may be transmitted to osteocytes and osteoblasts via shear stresses at cell surfaces generated by the flow of interstitial fluid. The stimulated production of prostaglandins, which mediates some effects of mechanical loading on bone, is dependent on inducible cyclo-oxygenase 2 (COX-2) in bone cells. We examined the fluid shear stress (FSS) induction of COX-2 gene expression in immortalized MC3T3-E1 osteoblastic cells stably transfected with −371/+70 base pairs (bp) of the COX-2 5′-flanking DNA (Pluc371) and in primary osteoblasts (POBs) from calvaria of mice transgenic for Pluc371. Cells were plated on collagen-coated glass slides and subjected to steady laminar FSS in a parallel plate flow chamber. FSS, from 0.14 to10 dynes/cm2, induced COX-2 messenger RNA (mRNA) and protein. FSS (10 dynes/cm2) induced COX-2 mRNA within 30 minutes, with peak effects at 4 h in MC3T3-E1 cells and at ≥8 h in POBs. An inhibitor of new protein synthesis puromycin blocked the peak induction of COX-2 mRNA by FSS. COX-2 promoter activity, measured as luciferase activity, correlated with COX-2 mRNA expression in both MC3T3-E1 and POB cells. FSS induced phosphorylation of extracellular signal-regulated kinase (ERK) in MC3T3-E1 cells, with peak effects at 5 minutes. Inhibiting ERK phosphorylation with the specific inhibitor PD98059 inhibited FSS induction of COX-2 mRNA by 55-70% and FSS stimulation of luciferase activity by ≥80% in both MC3T3-E1 and POB cells. We conclude that FSS transcriptionally induces COX-2 gene expression in osteoblasts, that the maximum induction requires new protein synthesis, and that induction occurs largely via an ERK signaling pathway.
INTRODUCTION
Mechanical loading of bone is essential for maintaining bone mass and integrity. External forces applied to bone can stimulate new bone formation.1-8 Loss of loading, as during immobilization9 or spaceflight,10 can result in bone loss. One current hypothesis to explain these findings is that mechanical loading causes the release of mediators that promote the recruitment and differentiation of osteoblastic progenitor cells.1, 11 Mechanical loading also may inhibit the expression of the receptor activator of NF-κB ligand (RANKL) by osteoblasts, thereby causing a reduction in osteoclastogenesis.12 However, the mechanisms by which mechanical loads are converted into biochemical signals remain unclear.
Mechanical loading produces strains in the mineralized matrix of bone that are thought to generate interstitial fluid flow through the lacunar/canalicular spaces.13 This fluid flow exerts a shear stress at surfaces of osteocytes and osteoblasts lining these spaces and the shear stress generates biochemical signals. Fluid flow through mineralized bone has been shown using tracers.14, 15 Support for flow-generated shear stress as a means of transducing mechanical loading comes from bending studies. A compressive force on one side of compact bone drives interstitial fluid toward the other side, and the velocity with which the fluid flows is related to the rate at which the force is applied. Studies on bending of tibias in rats show that bone formation increases with higher bending frequencies, consistent with higher fluid-generated shear stresses.16
Prostaglandins may mediate the stimulatory effects of mechanical loading on bone formation.17-19 The limiting enzyme in the conversion of membrane-released arachidonic acid to prostaglandins is cyclo-oxygenase (COX). There are two isoforms of COX. COX-1 is constitutively expressed and COX-2 is inducible.20 In bone cells, the prostaglandin elevation in response to a variety of stimuli is dependent on the induction of COX-2.21 Fluid shear stress (FSS) on osteoblastic cells has been shown to induce prostaglandin production22, 23 and COX-2 messenger RNA (mRNA) expression.24, 25 Furthermore, a selective inhibitor to COX-2 can block mechanically induced bone formation in vivo.26
Multiple signaling pathways have been reported to be involved in FSS induction of COX-2 in osteoblastic cells. They include cytoskeletal integrin rearrangement,25, 27 intracellular Ca2+,27 and phospholipase C.27 FSS activation of the mitogen-activated protein (MAP) kinase signaling pathway has been reported in endothelial cells28-30 but not in osteoblastic cells. The MAP kinase signaling pathway has been shown to be involved in anabolic responses to growth factors in osteoblastic cells.31, 32 Because mechanical loading also can stimulate anabolic responses in bone, we hypothesized that the MAP kinase signaling pathway also would mediate some responses to FSS.
In this study, we characterized FSS-induced COX-2 expression in osteoblastic cells. Using MC3T3-E1 cells and calvarial osteoblastic cells carrying COX-2 promoter-luciferase reporter DNA constructs, we showed that FSS transcriptionally induced COX-2 and that this induction occurs largely via an extracellular signal-regulated kinase (ERK) signaling pathway.
MATERIALS AND METHODS
Materials
Murine COX-2 complementary DNA (cDNA) and DNA constructs consisting of −371 to +70 base pairs (bp) of the COX-2 promoter fused to a luciferase reporter gene in pXp-2 vector (Pluc371) were the kind gift of Harvey Herschman (University of California, Los Angeles [UCLA], Los Angeles, CA, USA) and have been described previously.20, 33 cDNA for glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was amplified by polymerase chain reaction (PCR) with a control amplifier set from Clontech (Palo Alto, CA, USA). PD98059 and U-0126, specific inhibitors of mitogen-activated protein kinase kinase (MEK-1), and puromycin were purchased from Biomol (Plymouth Meeting, PA, USA). Phosphospecific monoclonal ERK-1/2 and polyclonal ERK-1/2 primary antibodies were purchased from New England Biolabs (Beverly, MA, USA). Phorbol myristate acetate (PMA) and other chemicals were purchased from Sigma (St. Louis, MO, USA).
Fluid flow chamber
The parallel plate fluid flow chamber was designed by M.A.F. Epstein and D.R. Peterson.34 It is shown schematically in Fig. 1. This chamber generated a uniform flow field in which magnitude and direction of the velocity vector are constant under steady flow conditions. There is sufficient entry length in the chamber before reaching the testing section to ensure uniform and fully developed flow. The flow field was generated by a roller-type pump with a console drive (Masterflex Laboratory digital variable-speed console drive, model 7523-30; Masterflex Laboratory, Vernon Hills, IL, USA). Fluid medium was pumped through the flow channel from a fluid reservoir, and fluid exiting the pump returned to the reservoir to complete the flow loop. The flow chamber has a test area of 13.5 cm × 8.5 cm and accommodates two glass slides of 83 cm2 separated by a spacer or four standard microscope slides (75 mm × 25 mm). The flow medium was DulbeccO's modified Eagle's medium (DMEM) without phenol red (Sigma) containing 0.05% heat-inactivated fetal calf serum (FCS; Gibco BRL, Grand Island, NY, USA). The flow medium was maintained at 37°C. A gassing system saturated the medium with 5% CO2. For steady laminar flow conditions, the wall shear stress (τ) that acts on the test section is directly proportional to the flow rate: τ = 6Q v/w h2, where Q is the volumetric flow rate, v is fluid viscosity, w is the flow channel width (7.5 cm), and h is the distance between parallel plates (0.5 mm).
Schematic of the parallel plate flow chamber. This chamber generates a uniform flow field in which magnitude and direction of the velocity vector are constant under steady flow conditions. The flow medium is maintained at 37°C by submerging the section of the tubing in a heating bath. A standard regulator set at 15 psi connected to a 95% air 5% CO2 tank saturates the medium in the reservoir with CO2.
Cell culture
All cells were cultured at 37°C in a humidified atmosphere of 5% CO2 in air. MC3T3-E1 cells were plated at 5000/cm2 in phenol red-free DMEM with 10% heat-inactivated FCS, penicillin (100 U/ml), and streptomycin (50 μg/ml) and grown to confluence over 4-6 days. To make primary calvarial cells, calvariae were excised from neonatal mice, dissected free of loose connective tissue, and washed with phosphate-buffered saline (PBS). Calvaria were treated with crude collagenase P (Boehringer Mannheim, Indianapolis, IN, USA) and trypsin at 37°C for 10 minutes to release fractions 1-4 and for 20 minutes to release fraction 5. Released cells were removed, and the reaction was stopped with DMEM + 10% FCS. A single cell suspension was formed by filtering the cells through a Nitex membrane. Populations 2-5 were pooled and cells were grown to confluence in the same medium as MC3T3-E1 cells before replating for an experiment. MC3T3-E1 cells and primary osteoblasts (POBs) were plated on type I collagen-coated (rat tail collagen; Collaborative Biomedical Products, Bedford, MA, USA) glass slides or plates at 5000/cm2 and 15,000/cm2, respectively, and cultured for 5 days before FSS experiments. Slides and plates were coated according to the instructions of the manufacturer with 5 μg/cm2 collagen. The vehicle for PD98059 was dimethylsulfoxide (DMSO) and was added at the same concentration (0.17%) to control cultures.
Stable transfection
Constructs were purified by CsCl banding, cotransfected with pSV-2 neo into MC3T3-E1 cells, and cultured as described previously to 50-80% confluency in 6-well dishes. Cells in each well were rinsed twice with serum-free medium and incubated with 1 μg of promoter-reporter DNA, 0.067 μg of pSV2-neo DNA, and 8 μl of Lipofectamine reagent (Gibco BRL) in 1 ml of serum-free medium without antibiotics. After 5 h of incubation, a second milliliter of medium with 20% FCS was added, and 19 h later the medium was replaced with fresh complete medium. After 48 h, cells were split 1:10 into 100-mm dishes and placed under selection with 400 μg/ml of G418 for 2 weeks. After selection, more than 200 colonies were pooled and cells continued in culture medium containing 200 μg/ml of G418.
Luciferase assay
Luciferase activity was measured in soluble cell extracts prepared with a kit from Promega (Madison, WI, USA) using an automatic injection luminometer (Berthold Lumat, Perkin Elmer Life Sciences, Boston, MA, USA). Activity in counts per second (cps) was normalized to total protein measured with a bicinchoninic acid (BCA) protein assay kit (Pierce, Rockford, IL, USA). For each experiment, three to four glass slides plated with cells that were subjected to FSS simultaneously were analyzed per treatment group.
Extraction of RNA and Northern blot analysis
Total RNA was extracted using Tri-Reagent (Molecular Research Center, Cincinnati, OH, USA) following the manufacturer's instructions. Ten to 20 μg of total RNA was run on a 1% agarose-2.2 M of formaldehyde gel, transferred to a nylon membrane by capillary pressure and fixed to the membrane by UV irradiation. After 3 h of prehybridization in a 50% formamide solution at 42°C, filters were hybridized overnight in a similar solution in rotating cylinders at the same temperature with a random [32P]deoxycytosine triphosphate (dCTP)-labeled cDNA probe. Filters were washed once in a 1× SSC, 1% sodium dodecyl sulfate (SDS) solution at room temperature, once in 0.1× SSC, 0.1% SDS solution at 65°C, and three more times in the latter solution at room temperature. After washing, the filters were exposed to Kodak XAR-5 film (Eastman Kodak Co., Rochester, NY, USA) at −70°C. Bands were scanned into the computer and the band density was quantitated by the National Institutes of Health (NIH) Image 1.61 software (free software available from the NIH).
Western blot analysis
Cells were washed with PBS, lysed with SDS buffer (62.5 mM of Tris-HCl, pH 6.8, 2% wt/vol SDS, 10% glycerol, 50 mM of dithiothreitol [DTT], and 0.1% wt/vol bromphenol blue), and scraped into a microcentrifuge tube. The extraction mixture was centrifuged at 14,000g for 30 minutes. The supernatants were sonicated 10-15 s, heated to 95°C for 5 minutes, cooled on ice, and centrifuged for 5 minutes. Equal amounts of protein, determined by BCA assay, were run on a SDS-polyacrylamide gel electrophoresis (PAGE) gel and electrotransferred to a nitrocellulose membrane. Membranes were washed with Tris-buffered saline (TBS; pH 7.6), blocked with blocking buffer (1× TBS, 0.1% Tween-20 with 5% wt/vol nonfat dry milk), and incubated with primary antibody in blocking buffer. After washing with TBS with 0.1% Tween-20 (TBST), the membrane was incubated with horseradish peroxidase (HRP)-conjugated secondary antibody and HRP-conjugated antibiotin antibody. After washing with TBST, the signal was detected with LumiGLO chemiluminescent reagent (New England Biolabs).
Transgenic mice
Mice carrying −371/+70 bp of the 5′-flanking murine COX-2 DNA fused to a luciferase reporter (Pluc371) were developed in a CD-1 background by the Transgenic Animal Facility at the University of Connecticut Health Center under the direction of Dr. Steve Clark. Initially, four separate lines were studied; no differences in luciferase responses were noted and two lines are maintained as breeding colonies. All animal protocols were approved by the Animal Care and Use Committees of the University of Connecticut Health Center.
Statistical analysis
Statistical significance of differences among means was determined by analysis of variance (ANOVA) with post hoc comparison of more than two means by the Bonferroni method or the Mann-Whitney rank sum test for nonparametric populations using SigmaStat (Jandel Scientific, San Rafael, CA, USA).
RESULTS
FSS induction of COX-2 mRNA and protein in MC3T3-E1 cells
In stationary cultures, COX-2 mRNA and protein levels generally were undetectable by Northern or Western blotting (Figs. 2, 3, and 4). To decrease the potential effects of growth factors in serum, the culture medium was changed to 0.05% serum for all FSS experiments. FSS of 10 dynes/cm2 induced COX-2 mRNA expression within 30 minutes, and mRNA levels reached a plateau 4 h after the initiation of FSS. New COX-2 protein was evident at 4 h after the initiation of FSS (Fig. 2). COX-2 mRNA and protein levels were increased in a dose-related manner by an FSS of 0.14-10 dynes/cm2 applied for 4 h (Fig. 3). To assess the effects of the manipulations involved in placing cells in the chamber, some experiments included slides with cells that were “sham-loaded” (Sh in Fig. 3A)—sealed briefly in the flow chamber (without flow) and then returned to stationary culture (Fig. 3A).
Time course for FSS induction of COX-2 mRNA and protein expression in MC3T3-E1 cells. Cultures were stationary (left in incubator) or subjected to FSS (10 dynes/cm2) for the times indicated. (A) Northern blot analysis for COX-2 and GAPDH) mRNA. The ratio of COX-2 mRNA band density to GAPDH band density is indicated below bands. (B) Western analysis for COX-2 protein (triplicate samples).
Dose response for FSS induction of COX-2 mRNA and protein expression in MC3T3-E1 cells. Cells were subjected to no flow (stationary culture in incubator), sham-loading (Sh; sealed in flow chamber with no flow and then returned to stationary culture), or FSS for 4 h. (A) Northern blot analysis for COX-2 and GAPDH mRNA. (B) Western blot analysis for COX-2 protein.
Effects of puromycin on FSS induction of COX-2 mRNA in MC3T3-E1 cells. Northern blot analysis was performed for COX-2 and GAPDH mRNA. Cells were subjected to FSS (10 dynes/cm2) with and without puromycin (10 μg/ml) for (A) 1 h and (B) 4 h.
To determine if new protein synthesis was necessary for the FSS induction of COX-2 mRNA, MC3T3-E1 cultures were treated with the protein synthesis inhibitor puromycin (10 μg/ml). The small induction of COX-2 mRNA induced by FSS (10 dynes/cm2) at 1 h was not inhibited by puromycin (Fig. 4). At 4 h, the time of peak induction of COX-2 mRNA by FSS treatment with puromycin blocked the COX-2 response to FSS (Fig. 4B). Similar results were obtained with cycloheximide, another inhibitor of protein synthesis, but cycloheximide alone caused a much larger induction of COX-2 mRNA than did puromycin alone, making results difficult to interpret (data not shown).
Transcriptional induction of COX-2 by FSS in MC3T3-E1 cells
To determine if the induction of COX-2 gene expression was transcriptionally mediated, MC3T3-E1 cells stably transfected with Pluc371were subjected to FSS. FSS (10 dynes/cm2) stimulated luciferase activity, which peaked at 4.5 h (Fig. 5A). There was a dose-dependent increase in luciferase activity with FSS, 0.14-10 dynes/cm2, applied for 4.5 h (Fig. 5B). FSS of 0.14 dynes/cm2 applied for 4.5 h significantly (p < 0.01) stimulated luciferase activity 1.8- and 2.2-fold, relative to no flow in two independent experiments. In each of seven independent experiments, FSS of 10 dynes/cm2 applied for 4-4.5 h significantly (p < 0.01) increased luciferase activity relative to no flow or sham-loaded controls, the mean increase (±SE) being 6.1 ± 1.0-fold.
Stimulation of luciferase activity by FSS in MC3T3-E1 cells stably transfected with Pluc371. Luciferase activity is expressed as counts per second per microgram of total protein in the sample. (A) Cells were subjected to no flow (stationary culture), sham-loading (Sh; sealed in flow chamber with no flow and then returned to stationary culture), or FSS for 4.5 h. Each point is the mean and SEM for n = 4 slides of cells.aSignificant effect of FSS, p < 0.01;bp < 0.05. (B) Cells were subjected to an FSS of 10 dynes/cm2 for the indicated times or sham-loaded. Each point is the mean and SEM for n = 3 slides of cells.aSignificant effect of FSS, p < 0.01.
FSS induction of COX-2 expression and luciferase activity in primary osteoblasts
FSS also induced COX-2 and luciferase activity in POBs derived from the calvaria of mice transgenic for Pluc371 (Fig. 6). The time course for induction was delayed relative to that obtained in the clonal MC3T3-E1 cell line, a difference we have observed for multiple COX-2 agonists (C. C. Pilbeam, unpublished data, 2001). FSS (10 dynes/cm2) induced COX-2 mRNA by 1 h and mRNA levels continued to rise at 8 h (Fig. 6A). FSS (10 dynes/cm2) also increased luciferase activity, with levels continuing to increase at 8 h (Fig. 6B). In three independent experiments, FSS (10 dynes/cm2) at 4-4.5 h significantly (p < 0.01) increased luciferase activity relative to no flow, the mean increase (±SE) being 2.7 ± 0.2.
Induction of COX-2 mRNA and luciferase activity by FSS in primary calvarial osteoblasts from Pluc 371 transgenic mice. Cells were subjected to an FSS of 10 dynes/cm2 for the times indicated. (A) Northern blot analysis for COX-2 and GAPDH mRNA levels. (B) Luciferase activity in counts per second normalized to total protein (μg) in the sample. Points are the mean and SEM for n = 4 slides of cells.aSignificant effect of FSS, p < 0.01.
FSS stimulation of ERK phosphorylation in MC3T3-E1 cells
To examine the FSS-induced phosphorylation of ERK, Western blot analysis was performed and probed for phosphospecific and total ERK-1/2 (Fig. 7A). FSS induced phosphorylation of ERK-1/2, which peaked at 5 minutes and returned to baseline at 30 minutes after the initiation of flow. Treatment for 10 minutes with PMA (1 μM), a known inducer of ERK phosphorylation, was included as a positive control (Fig. 7A). Treatment with the specific MEK-1 inhibitor PD98059 (50 μM) completely blocked the FSS-induced phosphorylation of ERK-1/2 at 5 minutes (Fig. 7B).
Effects of FSS on phosphorylation of ERK-1 and ERK-2 in MC3T3-E1 cells. Western analysis was done with a monoclonal antibody specific for doubly phosphorylated ERK-1 and ERK-2 or antisera detecting both phosphorylated and unphosphorylated kinase levels. (A) Cells were subjected to FSS (10 dynes/cm2) for 0-30 minutes or treated with PMA (1 μM) in stationary culture for 10 minutes. (B) Cells were subjected to FSS (10 dynes/cm2) for 5 minutes or left in stationary culture with and without PD98059 (50 μM).
Effects of PD98059 on the FSS induction of COX-2 mRNA and luciferase activity
In two independent experiments in MC3T3-E1 cells at 4 h, PD98059 (40 μM) inhibited the FSS (10 dynes/cm2) induction of COX-2 mRNA on Northern analysis by 55% (Fig. 8A) and 72% (data not shown). In a similar experiment in POB cells, PD98059 inhibited the FSS induction of COX-2 mRNA by 65% (data not shown). The smaller FSS induction of COX-2 mRNA at 1 h was inhibited by 50% (data not shown). Inhibition was calculated after normalization of COX-2 mRNA levels to the corresponding GAPDH mRNA levels.
Effects of PD98059 on FSS induction of COX-2 mRNA and luciferase activity. Cells were subjected to FSS (10 dynes/cm2) with and without PD98059 (40 μM). (A) Northern analysis for COX-2 and GAPDH mRNA in MC3T3-E1 cells subjected to FSS for 4 h. (B) Luciferase activity in MC3T3-E1 cells stably transfected with Pluc371, expressed as counts per second per microgram of total protein in the sample. Each point is the mean and SEM for n = 4 slides of cells.aSignificant effect of FSS, p < 0.01;bsignificant effect of PD98059, p < 0.01. (C and D) Luciferase activity in primary calvarial cells obtained from mice transgenic with Pluc371, expressed as counts per second (cps) per microgram of total protein in the sample. Each point is the mean and SEM for n = 4 slides of cells.aSignificant effect of FSS, p < 0.01;bsignificant effect of U-0126, p < 0.01.
In MC3T3-E1 cells stably transfected with Pluc371, 4.5 h of FSS (10 dynes/cm2) stimulated luciferase activity 4.2-fold and this stimulation was reduced 88% by PD98059 (40 μM; Fig. 8B). In a second similar experiment, FSS induced an 8.8-fold increase in luciferase activity, which was decreased 98% by treatment with PD98059 (data not shown). In POB cells from Pluc371 mice, FSS (10 dynes/cm2) for 4.5 h stimulated luciferase activity 3.2-fold, and this stimulation was reduced 80% by treatment with PD98059 (40 μM; Fig. 8C). In a second similar experiment, FSS stimulated a 2.5-fold increase in luciferase activity that was inhibited 99% by treatment with PD98059 (data not shown). In no experiment did FSS stimulate a statistically significant increase in luciferase activity in the presence of PD98059.
Similar experiments were performed with another MEK-1 inhibitor U-0126. On Northern analysis of MC3T3-E1 cells subjected to FSS (10 dynes/cm2) for 4 h, U-0126 (20 μM) inhibited the induction of COX-2 mRNA normalized to GAPDH by 40% (data not shown). In POB cells from Pluc371 mice, 4.5 h of FSS (10 dynes/cm2) stimulated luciferase activity 4.4-fold and this stimulation was reduced 80% by U-0126 (20 μM; Fig. 8D).
To confirm that the PD98059 inhibition of luciferase activity was not a nonspecific effect, we examined the effects of PD98059 on luciferase activity stimulated by forskolin, a cyclic adenosine monophosphate (cAMP) agonist shown to stimulate COX-2 expression in osteoblasts.35 In three independent experiments in MC3T3-E1 and POB cells, 3-4.5 h of forskolin (10 μM) stimulated 2- to 20-fold increases in luciferase activity and addition of PD98059 either had no effect or increased the stimulation seen with forskolin (data not shown). Similar to PD98059, U-0126 (20 μM) had no effect on forskolin (10 μM)-stimulated luciferase activity in MC3T3-E1 cells stably transfected with Pluc371 (data not shown).
DISCUSSION
FSS induced COX-2 mRNA and protein expression in both osteoblastic MC3T3-E1 cells and primary calvarial osteoblasts. In MC3T3-E1 cells, FSS (10 dynes/cm2) induced COX-2 mRNA within 30 minutes and levels peaked at 4-5 h. The induction of COX-2 gene expression by FSS was paralleled by the stimulation of luciferase activity in MC3T3-E1 cells stably transfected with Pluc371 and in POBs from Pluc371 transgenic mice, indicating that the FSS induction of COX-2 gene expression was largely transcriptionally mediated. Generally, COX-2 mRNA was undetectable in unstressed cells and, hence, the fold induction stimulated by FSS could not be calculated. On the other hand, luciferase activity was always detectable in unstressed cultures, perhaps because luciferase mRNA was more stable than COX-2 mRNA or because the COX-2 5′-flanking region that determined the luciferase transcription rate was not subject to all the regulatory constraints acting on the full-length COX-2 promoter. Preliminary screening in our laboratory (data not shown) found no difference in the maximal luciferase response to FSS between cells stably transfected with 4 kilobases (kb) of the COX-2 promoter and cells carrying Pluc371.
Osteocytes, terminally differentiated osteoblasts housed in mineralized lacunae and communicating with each other via processes extending through narrow canaliculi, are considered to form the major strain-sensing network in bone.36 A theoretical model for flow-generated shear stresses in lacunar-canicular spaces developed by Weinbaum et al.37 predicts physiological fluid-induced shear stresses of 8-30 dynes/cm2 in the proteoglycan-filled fluid annuli around osteocyte processes. Generally, it is assumed that the marrow sinusoids enclosing osteoblasts are much too wide to generate meaningful levels of shear stress during physiological loading. Because osteocytes presumably do not replicate, because antibodies that can be used to isolate osteocytes have only been identified for chicken osteocytes38 and because osteocytes and osteoblasts are closely related, osteoblasts have been used frequently in FSS studies. We and others39 have found that COX-2 gene expression in osteoblasts can be induced by FSS as low as 0.1-0.2 dynes/cm2, suggesting that osteoblasts are not just a convenient model for study but may have a role in the mechanosensing process.
In endothelial cells, activation of ERK by FSS has been well documented.28-30, 40, 41 Activation of the ERK pathway is involved also in the FSS induction of Egr-1 transcription in endothelial and epithelial cells.42 Little is known about the involvement of MAP kinase pathways in the response of osteoblasts to FSS. In a recent study, oscillatory fluid flow applied over 1 h caused sustained ERK activity in MC3T3-E1 osteoblastic cells.43 In contrast, we found that FSS (10 dynes/cm2) caused rapid and transient phosphorylation of ERK in MC3T3-E1 cells. ERK phosphorylation peaked at 5 minutes and returned to basal levels after 30 minutes. These results could suggest that osteoblasts respond differently to different types of flow, continuous versus oscillatory. Interestingly, it has been suggested that transient activation of ERK is associated with a mitogenic response whereas sustained ERK activity is associated with a differentiation response.44
Inhibiting ERK phosphorylation with the specific inhibitors PD98059 and U-0126 inhibited the FSS induction of COX-2 mRNA at 4 h in both MC3T3-E1 and POB cells by 40-70%. PD98059 and U-0126 also inhibited the FSS stimulation of luciferase activity in both MC3T3-E1 and POB cells by ≥80% but did not inhibit stimulation of luciferase activity by forskolin. Hence, the transcriptional induction of COX-2 gene expression by FSS occurred largely via an ERK signaling pathway. The greater inhibition of stimulated luciferase activity compared with mRNA induction might be caused by the less accurate quantitation of mRNA changes or to the prolongation of COX-2 mRNA half-life by FSS.
Protein synthesis inhibitors blocked the peak induction of COX-2 mRNA at 4 h but not the small early induction of mRNA at 1 h. These observations suggest that the early COX-2 response to FSS is the result of the direct activation of a transacting factor, and new protein synthesis is required for maximal induction of COX-2. One factor that might be involved in both the early protein-independent and the late protein-dependent responses is c-Fos. We have identified a functional activator protein 1 (AP-1; c-Fos/c-Jun) binding site in the COX-2 promoter, which mediates the induction of COX-2 by PMA,45 and a recent study showed that mutation of this site reduced the FSS-induced luciferase activity in MC3T3-E1 cells by 48%.39 The AP-1 complex can be activated by activation of the ERK pathway.46 In addition, multiple studies have shown that mechanical loading can increase expression of c-Fos in bone cells25, 27, 47 and inhibition of the ERK pathway by PD98059 has been shown to inhibit impulse flow-induced c-Fos mRNA expression in human endothelial cells.47 However, an additional FSS-activated signaling pathway is likely to be involved in the FSS induction of COX-2 because inhibition of ERK did not completely abolish the FSS induction of COX-2 mRNA expression.
Another candidate for the de novo protein involved in the peak COX-2 response to FSS might be COX-2 itself because prostaglandins can induce COX-2.48 However, in preliminary experiments, indomethacin, an inhibitor of COX activity, had no effect on the FSS induction of COX-2 (data not shown).
A study by Duncan's group has shown that FSS induction of COX-2 in osteoblastic cells is dependent on cytoskeleton-integrin interactions and intracellular calcium release.27 ERK activation by FSS has been shown to be dependent on focal adhesion kinase (FAK) in endothelial cells.40 Therefore, it seems likely that cytoskeletal-integrin interactions also may mediate the FSS activation of ERK in osteoblasts. ERK activation also might be dependent on intracellular calcium release. However, a recent study in bovine articular chondrocytes49 showed that ERK activation by fluid flow did not require calcium mobilization.
The induction of COX-2 in osteoblastic cells by FSS is believed to mediate the anabolic effects of mechanical loading in bone.26 Recent studies in our laboratory have shown that osteoblastic expression of COX-2 is necessary for maximal osteoblastic differentiation and function in vitro and that disruption of the COX-2 gene in vivo may decrease bone formation more than resorption.50 Hence, the FSS induction of COX-2 may contribute to maintaining skeletal integrity. The identification of signaling pathways involved in the FSS induction of COX-2 expression could aid in future development of pharmacologic therapies to increase bone formation and prevent the bone loss that occurs with mechanical unloading.
Acknowledgements
We appreciate the excellent technical assistance of Cynthia Alander. This work was supported by an NIH grant (DK48361; to C.C.P.) and by an American Association of Orthodontics Foundation research award (to S.W.).
