Osteopontin gene regulation by oscillatory fluid flow via intracellular calcium mobilization and activation of mitogen-activated protein kinase in MC3T3-E1 osteoblasts.

Recently fluid flow has been shown to be a potent physical stimulus in the regulation of bone cell metabolism. However, most investigators have applied steady or pulsing flow profiles rather than oscillatory fluid flow, which occurs in vivo because of mechanical loading. Here oscillatory fluid flow was demonstrated to be a potentially important physical signal for loading-induced changes in bone cell metabolism. We selected three well known biological response variables including intracellular calcium (Ca(2+)i), mitogen-activated protein kinase (MAPK) activity, and osteopontin (OPN) mRNA levels to examine the response of MC3T3-E1 osteoblastic cells to oscillatory fluid flow with shear stresses ranging from 2 to -2 Newtons/m(2) at 1 Hz, which is in the range expected to occur during routine physical activities. Our results showed that within 1 min, oscillatory flow induced cell Ca(2+)i mobilization, whereas two MAPKs (ERK and p38) were activated over a 2-h time frame. However, there was no activation of JNK. Furthermore 2 h of oscillatory fluid flow increased steady-state OPN mRNA expression levels by approximately 4-fold, 24 h after exposure to fluid flow. The presence of both ERK and p38 inhibitors and thapsigargin completely abolished the effect of oscillatory flow on steady-state OPN mRNA levels. In addition, experiments using a variety of pharmacological agents suggest that oscillatory flow induces Ca(2+)i mobilization via the L-type voltage-operated calcium channel and the inositol 1,4,5-trisphosphate pathway.

Mechanical loading plays an important role in regulating bone metabolism. Increased mechanical loading increases bone formation and decreases bone resorption (1). The absence of mechanical stimulation causes reduced bone matrix protein production, mineral content, and bone formation, as well as an increase in bone resorption (2). However, the mechanism by which bone cells sense and respond to their physical environment is still poorly understood. In this study we examine a novel physical stimulus and loading-induced oscillatory fluid flow and demonstrate that when applied to cultured osteoblastic cells at levels expected to occur in vivo it regulates mRNA levels for an important bone matrix protein, osteopontin (OPN). 1 Furthermore, this regulation occurs via an increase in intracellular calcium (Ca 2ϩ i ) and mitogen-activated protein kinases (MAPKs).
The sensitivity of bone tissue to mechanical loading has been proposed to involve a variety of cellular biophysical signals including loading-induced electric fields, matrix strain, and fluid flow. The latter effect of loading, originally described by Piekarski et al. (3), has recently been proposed to directly regulate bone cell metabolism in vivo (4,5). Furthermore, relative to other loading-induced biophysical signals applied to cells in vitro, fluid flow appears to be significantly more potent at physiological levels (6 -10). The origin of loading-induced fluid flow is a consequence of the fact that a significant component of bone tissue is unbound fluid. Bone tissue contains an extracellular fluid compartment that has been demonstrated to communicate with the vascular compartment, and mechanical loading has been shown to enhance fluid exchange between the two spaces (11).
When bone is exposed to mechanical loading fluid in the matrix is pressurized and tends to flow into haversian canals. As loading is removed (e.g. during the gait cycle) the pressure gradients, and consequently the direction of fluid flow, are reversed resulting in a flow-time history experienced by the cells that is oscillatory in nature. In vitro experiments have shown fluid flow to have a number of effects on bone cells including Ca 2ϩ i mobilization (12), production of nitric oxide and prostaglandin E 2 (8,13), and regulation of the expression of genes for OPN, Cyclooxygenase-2, and c-Fos (9,14,15). However, it is important to note that only one study to date utilized a reversing flow profile and found significantly different results when contrasted with nonreversing flow (16). Thus, the aim of this study is to detail important aspects of the biochemical response pathway including immediate, interme-diate, and long term effects of oscillatory fluid flow on bone cells, as well as on their inter-relationships.
To achieve this goal, we first investigated three well known biological osteogenic response variables. Ca 2ϩ i , a known second messenger transducing extracellular signals to the cell interior, was our immediate response variable. Activity of MAPKs is important for regulating cell differentiation and apoptosis by transmitting extracellular signals to the nucleus (17,18) and was our intermediate response variable. OPN is characterized as one of the predominant noncollagenous proteins that accumulate in the extracellular matrix of bone (19,20) and is also believed to be an important factor associated with bone remodeling caused by mechanical stress in vivo (21). Recently, strong evidence suggests that OPN is an important factor in loading induced bone cell metabolism (22)(23)(24). Furthermore, the role of osteopontin in extracellular matrix is more than structural. It has been shown to be involved in regulating bone cell attachment, osteoclast function, and mineralization, suggesting a central role in both the initiation and regulation of bone remodeling (25,26). Therefore, we quantified steady-state OPN mRNA levels as a long term response to oscillatory flow.
Recently MAPK family members including extracellular signal-regulated kinase (ERK), c-Jun N-terminal kinase (JNK), and p38 MAP kinase have been shown to be important signaling components linking mechanical stimuli to cellular responses, including cell growth, differentiation, and metabolic regulation, in endothelial cells, smooth muscle cells, and myocytes (27)(28)(29)(30). However, the role of MAPKs in bone cell mechanotransduction has not been determined. Moreover the role of Ca 2ϩ i in osteogenic gene transcription is unclear, especially in the case of oscillatory fluid flow. Therefore, the second goal of this study is to elucidate the roles of Ca 2ϩ i and the three major MAPKs in bone cell osteopontin gene expression induced by oscillatory flow.
Finally, the mechanism responsible for fluid-flow-induced Ca 2ϩ i mobilization has not been fully established, particularly for the oscillatory flow profiles expected to occur in vivo. Yellowley et al. (31) demonstrated that the steady flow-induced Ca 2ϩ i responses in bovine articular chondrocytes involved both influx of external Ca 2ϩ and release of internal Ca 2ϩ from IP 3sensitive stores and that the mechanism is G-protein-activated. Similar results were observed in bone cells stimulated by steady fluid flow (14,32). However there is evidence to suggest that steady and oscillatory fluid flow may have different biophysical effects on bone cells (16). Therefore, the third goal of this study is to elucidate the mechanism contributing to oscillatory flow-induced Ca 2ϩ i mobilization in bone cells. Steady flow (32), substrate stretch (33), and whole bone loading experiments (34) suggest that either stretch-activated mechanosensitive channels and/or L-type voltage-operated calcium channels (L-type VOCCs) may be involved. Additionally, it is not known whether the involvement of the IP 3 -sensitive stores is as important in the response to oscillatory fluid flow or whether other internal pathways (the ryanodine-sensitive pathway) may be involved in Ca 2ϩ i mobilization.

MATERIALS AND METHODS
Cell Culture-The mouse osteoblastic cell line MC3T3-E1 was cultured in minimal essential medium (MEM-␣; Life Technologies, Inc.) containing 10% fetal bovine serum (FBS; Hyclone, Logan, UT), 1% penicillin and streptomycin (Life Technologies, Inc.) and maintained in a humidified incubator at 37°C with 5% CO 2 . All cells were subcultured on glass slides for 2 days prior to experiments, with the exception of cells cultured for Ca 2ϩ i studies, for which quartz slides were used, for UV transparency. 3 ϫ 10 5 cells were seeded on the glass slides (75 ϫ 38 ϫ 1.0 mm), and 0.85 ϫ 10 5 cells were seeded on the quartz slides (76 ϫ 26 ϫ 1.6 mm). There are no significant differences observed in the behavior of MC3T3-E1 cells grown on normal glass slides versus quartz slides. 2 It is important to note that under these conditions the cells had not reached confluency nor did the medium (which did not include ascorbic acid or ␤-glycerophosphate) or time in culture (2 days) support differentiation or mineralization (35). Cells were exposed to oscillatory fluid flow in MEM-␣ and 2% FBS for calcium imaging experiments, and in MEM-␣ and 10% FBS for long term 2-h experiments.
Oscillatory Fluid Flow Device-Two different parallel plate flow chamber sizes were utilized. Larger chambers with a rectangular fluid volume of 56 ϫ 24 ϫ 0.28 mm were employed for long term flow to accommodate the larger glass slides. This size of slide was necessary to obtain adequate amounts of cell protein and mRNA. The smaller chamber design, fluid volume 38 ϫ 10 ϫ 0.28 mm, was employed in the calcium imaging studies where total cell number is not an issue. The oscillatory flow device was described in our previous study (16). Briefly, a Hamilton glass syringe was mounted in a small servopneumatic loading frame (EnduraTec, Eden Prairie, MN). The flow rate was monitored with an ultrasonic flowmeter with a 100-Hz frequency response (Transonic Systems Inc., Ithaca, NY).
Calcium Imaging-Intracellular calcium ion concentration ([Ca 2ϩ ] i ) was quantified with the fluorescent dye fura-2. fura-2 exhibits a shift in absorption when bound to Ca 2ϩ such that the emission intensity when illuminated with ultraviolet light increases with calcium concentration at a wavelength of 340 nm and decreases with calcium concentration at 380 nm. The ratio of light intensity between the two wavelengths corresponds to calcium concentration. A calibration curve of intensity ratio and calcium concentration was obtained using fura-2 in buffered calcium standards supplied by the manufacturer (Molecular Probes, Inc., Eugene, OR). Preconfluent (80%) cells were washed with MEM-␣ and 2% FBS at 37°C, incubated with 10 M fura-2-acetoxymethyl ester (Molecular Probes, Inc., Eugene, OR) solution for 30 min at 37°C, then washed again with fresh MEM-␣ and 2% FBS prior to experiments.
Cell ensembles were illuminated at wavelengths of 340 and 380 nm in turn. Emitted light was passed through a 510-nm interference filter and detected with an intensifier charge coupled device camera (International Ltd., Sterling, VA). Images were recorded, one every 2 s, and analyzed using image analysis software (Metafluor; Universal Imaging, West Chester, PA). Basal [Ca 2ϩ ] i was sampled for 3 min and followed by 3 min of oscillatory fluid flow (peak shear stress 2 N/m 2 , 1 Hz).
MAPK Activity Assay-There are three major MAPKs, p38 MAPK, ERK, and JNK. 100 g of lysate protein from either control or flowed cells was immunoprecipitated with anti-p38 MAPK, anti-ERK1/2, or anti-JNK antibody (Santa Cruz Biotechnology, Inc., Santa Cruz, CA) overnight. Following addition of 15 l of protein A/G for 2 h, the immunocomplex was collected by centrifugation, and the kinase reaction was then conducted in a kinase reaction buffer containing substrates myelin basic protein (for p38 MAPK or ERK) or c-Jun glutathione S-transferase (for JNK) in the presence of [␥-32 P]ATP as described before (36). The reaction mix was subjected to SDS polyacrylamide gel electrophoresis, and phosphorylation of substrates was determined by autoradiography.
Osteopontin mRNA Analysis-The steady-state osteopontin mRNA level was quantified by quantitative real time reverse transcription polymerase chain reaction (QRT RT-PCR) (9). Briefly, this technique is based on the detection of a fluorescent signal produced by an OPNspecific oligonucleotide probe during PCR primer extension (Prism 7700 sequence detection system; Applied Biosystems, Frost City, CA). The RNeasy mini kit (Qiagen Inc., Valencia, CA) was used to extract total RNA after lysis and homogenization with the QIAshredder mini column system (Qiagen Inc., Valencia, CA). Mouse osteopontin cDNA primers and probes were designed using sequence data from Miyazaki et al. (37) (GenBank TM accession number X51834) and the QRT RT-PCR probe/ primer design software Primer Express (version 1.0; Applied Biosystems, Frost City, CA). The fluorogenic oligonucleotide probe for mouse osteopontin was 5Ј-CGG TGA AAG TGA CTG ATT CTG GCA GCT C-3Ј (Synthetic Genetics, San Diego, CA). The forward and reverse PCR primers were 5Ј-GGC ATT GCC TCC TCC CTC-3Ј, and 5Ј-GCA GGC TGT AAA GCT TCT CC-3Ј, respectively. These sequences were synthesized, and PCR conditions were optimized with respect to concentrations of Mg 2ϩ , probe, and both primers. Relative changes in the levels of OPN mRNA and 18 S rRNA were quantified 24 h after mechanical stimulation.
Pharmacological Agents-The following series of pharmacological agents was used to examine the mechanism of calcium mobilization: thapsigargin (50 nM), gadolinium chloride ( Thapsigargin is an inhibitor of the ATP-dependent Ca 2ϩ pump of intracellular Ca 2ϩ stores that causes Ca 2ϩ discharge (38) and was used to empty the intracellular calcium stores. Gadolinium chloride (10 M) (Aldrich) is a putative stretchactivated channel blocker (39). Nifedipine is a blocker of the L-type VOCC (40). Ryanodine, which affects ryanodine-sensitive channels in intracellular calcium stores, was used in two concentrations, 1 M, which is expected to hold the channel open, and 20 M, which is expected to block the channel (41,42). U73122 inhibits the action of phospholipase C and possibly phospholipase A 2 and thereby the production of IP 3 . Thus, it inhibits the release of calcium through IP 3sensitive intracellular calcium stores (14,43). U73343, an isoform of U73122 that does not inhibit IP 3 production, was used as a control. 2APB is a specific inhibitor of the IP 3 receptor and does not affect ryanodine-sensitive or membrane calcium channels (44,45). Cells were pretreated with medium containing the required drug for 30 -60 min prior to flow, and the drug remained present during the flow experiments. Nifedipine, ryanodine, and 2APB were dissolved in 100% ethanol to give a final concentration of ethanol in the flow medium of 0.1% (v/v), and vehicle controls were conducted with the same concentration of ethanol. Thapsigargin, U73122, and U73343 were dissolved in Me 2 SO to give a final concentration of Me 2 SO in the flow medium of 0.0032, 0.17, and 0.17% (v/v), respectively. Gadolinium chloride was directly dissolved in medium.
Thapsigargin and gadolinium chloride were also used in long term flow experiments to examine the role of [Ca 2ϩ ] i in downstream responses. For the MAPK investigations, cells were incubated with MAPK inhibitor for 2 h before the fluid flow experiments were performed. The p38 inhibitor SB203580 (10 M) or the ERK inhibitor PD98059 (10 M; Calbiochem-Novabiochem) was also present in the flow medium. All pharmacological agents were from Sigma unless indicated.
Data Analysis-We used a numerical procedure from mechanical analysis, known as Rainflow cycle counting, to identify calcium oscillations (46). Briefly, this technique identifies complete cycles or oscillations in the time history data even when they are superimposed upon each other and therefore can be used to distinguish and quantify [Ca 2ϩ ] i responses from background noise. We defined a response as an oscillation in [Ca 2ϩ ] i at least 2-fold greater than that of the average baseline level of nontreated cells. Baseline [Ca 2ϩ ] i data were recorded for each slide for 3 min prior to the application of oscillatory fluid flow.
Data were expressed as mean Ϯ S.E. To compare observations from no flow and flow responses a two-sample Student's t test was used in which sample variance was not assumed to be equal. To compare observations from more than two groups, a one-way analysis of variance was employed followed by a Bonferroni selected pairs multiple comparisons test (Instat; GraphPad Software Inc., San Diego, CA). p Ͻ 0.05 was considered statistically significant. For calcium experiments all controls were combined as no effect of vehicles was found (one-way analysis of variance).

Ca 2ϩ
i Responses to Oscillatory Fluid Flow-Typical cell Ca 2ϩ i responses are shown in Fig. 1A. The fraction of MC3T3-E1 cells responding with an increase in Ca 2ϩ i to oscil-latory fluid flow (peak shear stress 2 N/m 2 , 1 Hz) is shown in Fig. 1B Fig. 2. At each time point cells from two slides were combined to yield sufficient protein for the MAPK activity assay. In the absence of flow there was minimal p38, ERK1/2, and JNK activity. However, dramatic responses for p38 and ERK1/2 activities were observed beginning 15 min after applying oscillatory flow. p38 activity reached a maximum at 30 min and returned to initial levels 90 min after the onset of oscillatory flow (Fig. 2). ERK1/2 activity reached a maximum at 60 min and returned to its pre-flow value at 90 min. In contrast, there was no change in JNK activity during a 90-min flow period, indicating a selective activation of p38/ERKs in response to flow.
OPN Responses to Oscillatory Fluid Flow-The long time frame biological response, steady-state OPN mRNA level, was quantified in response to oscillatory fluid flow at 1 Hz, resulting in a wall shear stress of 2 N/m 2 , utilizing QRT RT-PCR. The cells that experienced oscillatory flow or no flow for 2 h were then incubated for an additional 24 h prior to collection. Our results show oscillatory fluid flow increased steady-state osteopontin mRNA levels by 3.96 Ϯ 0.76-fold over no flow control (see Fig. 3; p Ͻ 0.05).
Role of Ca 2ϩ i and MAPK Activities in the OPN mRNA Response to Oscillatory Fluid Flow-To assess the role of Ca 2ϩ i in the OPN mRNA response to oscillatory fluid flow, cells were subjected to oscillatory fluid flow in the presence of 50 nM thapsigargin. Interestingly thapsigargin completely blocked the oscillatory flow effect on steady-state OPN mRNA levels (0.93 Ϯ 0.08 ϫ no flow levels), which were not statistically different from those for no flow period (Fig. 3). However, gadolinium chloride (GdCl 3 ; 10 M) did not attenuate the flow effect on steady-state OPN mRNA levels (4.19 Ϯ 0.21 ϫ no flow levels), which were not statistically different from those for flow control case.
Based on the MAPK activation results, two MAPK inhibitors were employed to block the activity of ERK1/2 and p38. Cells On application of thapsigargin, which emptied intracellular stores, there was a significant (p Ͻ 0.05) decrease in the percentage of cells responding to oscillatory flow. U73122, which inhibits production of IP 3 via the phospholipase C pathway and does not affect membrane channels, reduced the number of cells responding to 18.0 Ϯ 9.0%, compared with 48.0 Ϯ 9.5% in the control group. 2APB, which acts directly on IP 3 receptors (45) rather than on IP 3 catalysis, blocked the response completely. Ryanodine at 1 M had a small but statistically significant effect on the number of cells responding, which was reduced to 41.6 Ϯ 9.9%. It also had a small effect on the mean increase in [Ca 2ϩ ] i over baseline, which was reduced to 37.4 Ϯ 5.5 nM. At 20 M, at which concentration the ryanodine-sensitive channel should have been blocked, there was a very small and not statistically significant reduction in the number of cells responding, to 55.1 Ϯ 10.4%. Although there were some differences in the mean response amplitudes (Fig. 4B), these were not found to be statistically significant. Some drug-treated cells (nifedipine and ryanodine) showed higher responses in the no flow period compared with controls (data not shown), because the drugs caused increased spontaneous calcium oscillations and increased drift in the baseline levels. DISCUSSION Although a large number of in vitro studies have been aimed at discovering the regulatory effect of mechanical loading in bone adaptation, little consensus can be found in the literature regarding the appropriate biophysical signals. For example, bone cells have been shown to respond with metabolic changes to deformation induced by stretching of the substrate to which they are attached (22,(47)(48)(49). However, these studies employed either hyperphysiologic levels of strain or systems known to induce mechanical effects other than pure strain (50). More recent studies have suggested that bone cells are more responsive to the fluid flow induced by mechanical strain than directly to the strain in the tissue (7-9). However, loadinginduced fluid flow in vivo involves a reversal of flow direction associated with the cyclical unloading that occurs in the vast majority of physical activities. To date, the ability of the resulting oscillatory flow profiles to regulate bone cell behavior in vitro has not been investigated beyond its ability to mobilize cytosolic calcium (16). In this study a novel oscillatory fluid flow system was designed to demonstrate that oscillatory fluid flow is capable of regulating bone cell gene expression via ERK and p38 MAPK activity and intracellular calcium signaling involving IP 3 -mediated calcium release. Additionally, we were able to demonstrate some potentially important differences in the characteristics of the response of bone cells to oscillatory flow when contrasted with published experiments on steady/ pulsatile flow. The study of the effects of oscillatory fluid flow on bone cells will allow us to more accurately understand the mechanism of mechanotransduction in bone cells in vivo, for which the other in vitro systems may not be as suitable.
Our experimental data have shown that oscillatory fluid flow induced three biological responses that are believed to be important in the response of bone tissue to mechanical load. In the short term, within 2 min of the start of oscillatory flow, 59.1 Ϯ 4.6% of cells increased [Ca 2ϩ ] i , which was significantly different from the no flow period. This is consistent with prior observations of the Ca 2ϩ i response of bovine aortic endothelial cells (51), articular chondrocytes (52), and bone cells (12,16) to steady/pulsatile fluid flow. However, oscillatory flow appears to be significantly less stimulatory than steady/pulsatile flow for bone cells in terms of the Ca 2ϩ i response (16). This suggests that the mechanotransduction pathways induced by oscillatory flow could be different in part or in whole from those activated by steady/pulsatile flow.
Recently MAPK activity has been shown to be modulated by various external stimuli such as growth factors, cytokines, and physical stresses (ultraviolet radiation, hyperosmolarity, hypoxia, and fluid flow shear stress) (17,18,53) and is known to play a pivotal role in a variety of cell functions. Our results are the first to examine the regulation of MAPK activity in response to biophysical stimulation in bone cells. We show that fluid flow induces an increase in the activity of two of three major MAPKs (ERKs and p38) over a period of 2 h for bone cells. p38 activity started to increase at 15 min and reached a maximum at 30 min and then returned to initial levels 90 min after the onset of oscillatory flow. A similar pattern was observed for ERK1/2 activity with some delay, at 60 min it reached a peak and returned to its pre-flow value at 90 min. JNK activity was unchanged during 90 min of oscillatory fluid flow stimulus. Our biphasic time course ERK1/2 and p38 activity results are consistent with previous studies in endothelial cells and smooth muscle cells (27,28). However, our time to peak ERK1/2 activity (60 min) is slower than observed for FIG. 2. Time courses for p38, ERK1/2, and JNK activation during oscillatory flow (2 N/m 2 , 1 Hz). At each time point the cells from two slides were combined to yield sufficient protein for the MAPK activity assay. Kinase activity was assayed by incubating lysates with [␥-32 P]ATP and myelin basic protein (for p38 MAPK and ERK1/2) or c-Jun glutathione S-transferase (for JNK). The reaction mix was subjected to SDS polyacrylamide gel electrophoresis, and phosphorylation of substrates was determined by autoradiography. A representative autoradiograph is shown. The experiments were repeated with similar results. The appearance of double bands is an accepted occurrence with this assay (64,65) and is because of impurity, phosphorylation, or degradation of the myelin basic protein (MBP) substrate but cannot be ascribed to differential ERK1/2 activity.

Oscillatory Flow Activated Ca 2ϩ
i , MAP Kinases, and Osteopontin steady fluid flow (5 min), possibly because of the different mechanical stimuli. Another difference is that oscillatory fluid flow did not induce JNK activity in bone cells; however steady fluid flow is capable of activating JNK in endothelial cells within 60 min (27). Although possibly because of differences between the cell types, it may be a result of differences in the effects of the physical signals applied. This is consistent with the possibility that oscillatory fluid shear stress may stimulate different mechanotransduction pathways from steady/pulsatile fluid shear stress. It was suggested previously that the ERK pathway is involved in the regulation of cell proliferation and differentiation whereas p38 and JNK are important signaling pathways in the regulation of cell apoptosis (54). However, recent information demonstrated that p38 MAPK may also play a critical role in the regulation of differentiation (55,56). In this study, both the p38 inhibitor SB and the ERK1/2 inhibitor PD were applied to determine whether the increased MAPK activity we observed was required for the effect of oscillating flow on steady-state OPN mRNA levels. Either MAPK inhibitor alone was found to greatly attenuate (80%) the flow effect on steady-state OPN mRNA, whereas the presence of both inhibitors (SB ϩ PD) completely abolished the effect of flow on steady-state OPN mRNA levels. This indicates that oscillatory flow-induced OPN expression involves both ERK and p38 MAPK activity with mild redundancy but does not require JNK activity. It is interesting to note that JNK activity has been observed in endothelial cells in response to the steady flow associated with apoptosis (27). In contrast, bone cells experiencing more moderate oscillatory shear stress exhibit increased ERK1/2 activity associated with proliferation and differentiation but no change in JNK activity. These findings support the view that oscillatory fluid flow may be a potent cellular physical signal in bone remodeling in vivo.
Our results also suggest that the biochemical mechanism of Ca 2ϩ i mobilization is different between nonreversing steady/ pulsatile fluid flow and oscillatory flow. The results of the calcium experiment using nifedipine show that the L-type VOCC membrane channel is involved in the calcium response to oscillatory flow in contrast to steady flow experiments in primary bone cells (32) and in the same cell line (14). However our data are in agreement with substrate stretch experiments on primary osteoblasts in which the calcium response was inhibited by nifedipine (33). In those experiments fluid flow may have been induced in the system, as well as substrate stretch (50). Thus, it is possible that the Ca 2ϩ i response that the investigators observed was because of the pathway we describe here in response to oscillatory flow. Furthermore, the nitric oxide and prostaglandin E 2 response of loaded whole rat bones in an in vivo model has been shown to be eliminated by nifedipine (34). This is again consistent with the view that oscillatory flow, rather than steady flow, is the cellular physical signal that regulates the adaption of bone to mechanical load in vivo.
Our data also suggest that the stretch-activated membrane channel, blocked using gadolinium chloride, is not important for response to oscillatory fluid flow. This is in contrast to data for steady flow where calcium responses were inhibited by blocking this channel (14,32). However, our finding that GdCl 3 did not influence steady-state OPN gene mRNA is consistent with the results of Chen et al. (14) that showed that the effect of steady flow on cytoskeletal reorganization and Cyclooxygenase-2 mRNA involved IP 3 -mediated intracellular calcium release but not extracellular calcium. One interpretation is that both oscillatory and steady flow activate an IP 3 cascade that is important in bone adaptation; however steady flow also stimulates an GdCl 3sensitive calcium influx whereas oscillatory flow does not.
Our finding that thapsigargin completely blocked the calcium response to oscillatory flow demonstrated that the source of Ca 2ϩ is release from intracellular stores. The next series of experiments were designed to further elucidate the mechanism of this release. The combination of the U73122 and the 2APB data strongly suggest that the IP 3 pathway is involved. We achieved a partial block of the calcium response using U73122. This may be because there are other pathways to the formation of IP 3 besides the phospholipase C and phospholipase A 2 pathways blocked by U73122. However the effect of U73122 is shown to be maximal at 10 M, and we used only 4 -5 M, because we found that the concentration of solvent (Me 2 SO) necessary to achieve the higher concentration of the drug induced cell toxicity. 2APB, a novel IP 3 blocker that is specific to the IP 3 channel (44), resulted in a total block of the Ca 2ϩ response. This finding is consistent with our previous observations using neomycin sulfate (57) and published results from other laboratories supporting the involvement of this second messenger in fluid flow responses in various cell types and flow regimes (13,14,31,32).
The role of ryanodine-sensitive internal stores in mechanotransduction has received less attention, although they have been shown to be present in osteoblasts (58,59). Our finding that low concentration ryanodine inhibited the response of MC3T3-E1 cells to oscillatory flow confirms the presence of ryanodine-sensitive calcium channels. In our experiments the opening of the ryanodine-sensitive stores, which would cause calcium to leave the stores before the flow was applied, in a similar way to thapsigargin, did have an inhibitory effect on the response, though less than that of thapsigargin. However the blocking of the ryanodine-sensitive channel with high concentration ryanodine had no significant effect on the calcium response. This suggests that ryanodine-sensitive Ca 2ϩ stores, which can also be mobilized by the IP 3 pathway (60), were partially depleted by the low concentration ryanodine but that ryanodine-sensitive channels were not affected by oscillatory fluid flow.
Interestingly, our finding that the source of the calcium response is IP 3 -mediated release from intracellular stores seems to be contradicted by our finding that the L-type VOCC is also involved. If the VOCC is important to a calcium response mechanism one might expect to observe some residual calcium increase of extracellular origin, even in the presence of blockers of intracellular stored calcium. However, in our study both 2APB and thapsigargin totally abolished the calcium response to oscillatory fluid flow. Similar results were found in the Walker et al. substrate stretch study (33) in which thapsigargin inhibited the calcium response more than would be expected if only the residual calcium released after nifedipine treatment was from intracellular stores sensitive to thapsigargin. An explanation for these results may be that the IP 3 receptor on the endoplasmic reticulum has been shown to be coregulated by cytosolic calcium concentration. Thus, the VOCC could potentate the fluid flow calcium response by regulating the local calcium concentration surrounding the endoplasmic reticulum IP 3 receptor but at levels that are not detectable with our imaging system. This mechanism would require that the VOCC and endoplasmic reticulum IP 3 receptor are in close association. Such an arrangement has previously been described in muscle cells between the VOCC and ryanodine-sensitive channels (61).
In the final phase of our investigation we related these intracellular signaling pathways to the regulation of gene expression. OPN has been implicated as an important factor in triggering bone remodeling caused by mechanical stress in vivo (21). Our OPN data are consistent with the in vitro results of Owan et al. (7). Steady-state OPN mRNA levels increased almost 4-fold within 24 h after 2-h oscillatory fluid flow. To elucidate the role of Ca 2ϩ i in bone cell mechanotransduction and OPN gene regulation, thapsigargin was employed to empty Ca 2ϩ i stores, which prevents Ca 2ϩ i from being available to the cells during the oscillatory flow period. Thapsigargin completely abolished the increase in steady-state OPN mRNA levels that occurred on application of fluid flow. This finding combined with the role of OPN in mechanically mediated remodeling suggests a prominent role of cytosolic calcium mobilization in the adaptation of bone to mechanical loading.
Although our results suggest that both Ca 2ϩ i and MAPK are involved in the mechanical stress-induced OPN expression in bone cells via oscillatory flow, the relationships between Ca 2ϩ i and MAPK are still unclear. Our results show that Ca 2ϩ i is required for OPN expression induced by oscillatory flow. However some investigators demonstrated that steady flow in chondrocytes activated ERK1/2 in a way that did not require Ca 2ϩ i , and Ca 2ϩ i alone was not sufficient for MAPK activation by steady flow (62). Therefore the role of Ca 2ϩ i in MAPK activation under oscillatory flow remains to be determined. However, both IP 3

and Ca 2ϩ
i have been shown to be a necessary step in G-protein-mediated MAPK activation in smooth muscle cells (63). Little is known about the signaling pathways between MAPK and target genes, although some investigations have shown that MAPK phosphatase-1 may act as a mediator to regulate target gene expression in vascular smooth muscle cells  (29). Further investigation of the whole cascade of mechanotransduction in bone cells is necessary.
In summary, our study demonstrates that oscillatory fluid flow is a potent physiological stimulator that induces Ca 2ϩ i release and OPN gene expression via ERK1/2 and p38 activation but not JNK. OPN gene expression required Ca 2ϩ i mobilization. Ca 2ϩ i is mobilized using primarily the IP 3 pathway, with the L-type VOCC membrane channel also playing a role. Although we did not compare oscillatory fluid flow directly to steady/pulsatile flow in this study, when compared with previously published studies on steady/pulsatile flow, our findings suggest that there are some potentially important differences in the response of bone cells to these two stimuli. This contrast indicates that there may exist multiple mechanotransduction pathways in bone cells that are activated depending on stimulus type and that determining an appropriate cellular mechanical stimulus is critical in understanding the role of mechanical loading in the regulation of bone.