Cell Physiology


The strong correlation between a bone's architectural properties and the mechanical forces that it experiences has long been attributed to the existence of a cell that not only detects mechanical load but also structurally adapts the bone matrix to counter it. One of the most likely cellular candidates for such a “mechanostat” is the osteocyte, which resides within the mineralized bone matrix and is perfectly situated to detect mechanically induced signals. However, as osteocytes can neither form nor resorb bone, it has been hypothesized that they orchestrate mechanically induced bone remodeling by coordinating the actions of cells residing on the bone surface, such as osteoblasts. To investigate this hypothesis, we developed a novel osteocyte-osteoblast coculture model that mimics in vivo systems by permitting us to expose osteocytes to physiological levels of fluid shear while shielding osteoblasts from it. Our results show that osteocytes exposed to a fluid shear rate of 4.4 dyn/cm2 rapidly increase the alkaline phosphatase activity of the shielded osteoblasts and that osteocytic-osteoblastic physical contact is a prerequisite. Furthermore, both functional gap junctional intercellular communication and the mitogen-activated protein kinase, extracellular signal-regulated kinase 1/2 signaling pathway are essential components in the osteoblastic response to osteocyte communicated mechanical signals. By utilizing other nonosteocytic coculture models, we also show that the ability to mediate osteoblastic alkaline phosphatase levels in response to the application of fluid shear is a phenomena unique to osteocytes and is not reproduced by other mesenchymal cell types.

  • osteocyte
  • osteoblast
  • fluid-flow
  • coculture
  • mechanical stimulation
  • gap junction
  • intercellular communication

one of the major tenets of bone biology is that mechanical load perturbs interstitial fluid and that, upon detecting these perturbations, osteocytes embedded deep within mineralized bone remotely coordinate the adaptive response by directing the actions of effector cells such as bone-forming osteoblasts and bone-resorbing osteoclasts (4, 12). Although this theory is widely accepted, there is little experimental evidence that, in response to fluid shear, embedded osteocytes directly alter the cellular behavior of surface-residing osteoblasts or osteoclasts. In vivo, osteocytes pass dendritic-like cellular processes through channels in the mineralized matrix (canaliculi) and form physical connections, specifically gap junctions, with both neighboring osteocytes and surface-residing bone cells (11, 23). We hypothesize that these intercellular gap junctions provide the means by which mechanically induced fluid shear signals may be communicated from osteocytes to effector cells such as osteoblasts and osteoclasts.

There is abundant evidence suggesting that gap junctional intercellular communication (GJIC) contributes to mechanotransduction in the musculoskeletal system. Banes et al. (3) demonstrated that equibiaxial strain upregulates connexin 43 (Cx43) expression in tendon cells in vitro, and cyclic stretch has been shown to increase both the expression of Cx43, the predominate gap junction protein in bone, and GJIC in osteoblastic cells in vitro (46). We (1) and others (5–7) demonstrated that fluid flow increases gap junction expression and function in osteocytic mouse long bone osteocyte (MLO-Y4) cells; however, Thi et al. (41) demonstrated that fluid flow decreases gap junction function and expression in MLO-Y4, emphasizing the need for additional examination of this issue. In vivo, Lozupone et al. (30) demonstrated that mechanical loading of rat metatarsal bones increased the incidence of osteocytic gap junctions, and we demonstrated that expression of Cx43 by osteocytes is increased in areas of bone exposed to tension relative to areas exposed to compression or to control bone (39).

Although accumulating evidence suggests that GJIC may have important roles in transducing mechanical signals throughout bone cell networks, most of this evidence is from experiments that examined osteoblastic cells in two-dimensional monoculture. However, the majority of cells in bone are osteocytes, and these cells are the best candidates for detecting and coordinating responsiveness to mechanical signals and communicating these signals to osteoblastic cells to affect their behavior (10, 12). Recent evidence suggests that osteocytes regulate osteoclastic cell activity and mesenchymal stem cell (MSC) differentiation (16, 17). Additionally, in response to mechanical signals, osteocytic cells have been shown to extend dendritic processes (45), and release a number of soluble factors, including PGE2, ATP, and nitric oxide, which could potentially regulate osteoblastic behavior (5, 8, 26). We have demonstrated that GJIC contributes to mechanically induced Ca2+ wave propagation from osteocytic MLO-Y4 cells to osteoblastic MC3T3-E1 cells (42), but, surprisingly, no studies have examined the physiological consequence of mechanical signals detected by osteocytes and communicated to osteoblasts.

To address this issue, we developed a three-dimensional coculture system that includes both osteocytic and osteoblastic cells and examined the hypothesis that osteocytic cells detect mechanical signals and communicate this, via GJIC, to osteoblastic cells, thus activating them. By utilizing commercially available cell culture inserts perforated with 1-μm pores (BD Biosciences), we are able to coculture osteocytes and osteoblasts simultaneously under conditions that enable GJIC but prevent osteocytes and osteoblasts from mixing (see Figs. 1A, 2, and 3A). To this system, we added a rotating disc mechanism that allows a defined field of fluid shear to be imparted to the osteocyte network while they are in physical contact with osteoblasts that are shielded from the fluid movement.

Fig. 1.

In vitro coculture model. A: cell culture inserts are seeded on the apical surface with 2 × 104 mouse long bone osteocytes (MLO-Y4; Ocy) and on the basal surface with 4 × 104 human fetal osteoblastic hFOB1.19 cells (hFOB; Ob). The 1-μm perforations in the polyethylene terephthalate (PET) membrane permit the cells on adjacent sides of the membrane to establish osteocyte-osteoblast gap junctions (GJ) but do not permit cell migration (Figs. 2 and 3B). A rotating disk located above the osteocytes is used to introduce fluid flow in the system. B: theoretical (-) and computational fluid dynamics-modeled (□) shear induced by rotating a disk at 255 revolutions/min plotted as a function of membrane radius.

Fig. 2.

Environmental-scanning electron microscopy images of apical osteocytic MLO-Y4 and basal osteoblastic hFOB cocultured on porous PET membranes. A: cocultured osteocytic MLO-Y4 form an interconnected network. B: high magnification reveals that dendritic processes from osteocytic MLO-Y4 form close association with membrane pores (white arrows). C: cocultured osteoblastic hFOB form confluent monolayers covering the majority of the membrane. D: at high magnification, both uncovered pores (white arrows) and pores covered by the hFOB cell body (black arrow) can be visualized. Scale bar = 20 μm in A and B and 5 μm in C and D.

Fig. 3.

Osteocytic MLO-Y4 and osteoblastic hFOB in adjacent cocultures display gap junction intercellular communication (GJIC). A: dimensions of the cellular separation in adjacent and remote cocultures. In adjacent cocultures, osteocytic MLO-Y4 and osteoblastic hFOB are separated by the insert membrane (a), which is equal to 10–20 μm. In remote cocultures, osteoblastic hFOB are located on the bottom of the tissue culture dish and are separated from the osteocytic MLO-Y4 on the membrane by 900 μm (b). B: osteoblastic hFOB in adjacent cocultures with osteocytic MLO-Y4 take up significantly greater amounts of calcein compared with remotely cocultured hFOB cells (###P < 0.001, adjacent hFOB vs. remote hFOB). The level of calcein uptake in osteocytic MLO-Y4 cells was not significantly different between adjacent and remote cocultures. The dye transfer to all MLO-Y4 cells, and to osteoblastic hFOB in adjacent cocultures, was significantly inhibited by the application of 30 μM 18α-glycyrrhetinic acid (αGA) (***P < 0.001, control vs. αGA treated), but αGA had no significant effect on remotely cocultured hFOB cells. NS, not significant. Bars represent means ± SE; n = 9 experiments.

Osteoblasts cocultured with osteocytes exposed to a fluid shear inducing a τmax of 4.4 dyn/cm2 showed a rapid and highly significant increase in osteoblastic activity, as measured by increases in alkaline phosphatase (AP) activity. This increase was not observed when monocultures of human fetal osteoblastic hFOB1.19 cells (hFOB) were directly exposed to fluid shear, nor when we used other nonosteocytic cell lines. Interestingly, within our coculture model, fluid shear did not increase the proliferation of osteoblastic cells; however, monocultures of osteoblastic hFOB exposed directly to a τmax of 4.4 dyn/cm2 did show significant increases in cellular proliferation. Taken together, these data highlight significant differences between the osteoblastic response to fluid shear when it is detected directly vs. indirectly through osteocytes and imply that the eventual outcome of mechanically loading bone is dependent on the cellular detector.


All reagents were from Invitrogen, unless otherwise stated.

Cell culture.

Osteocytic (MLO-Y4), osteoblastic (hFOB 1.19 and MC3T3-E1), and fibroblastic (BJ-FIB) cell lines were maintained as previously described (15, 25, 36). Basal medium for coculture experiments was comprised of DMEM-F-12 supplemented with 15 mM HEPES buffer, 1% (vol/vol) penicillin/streptomycin, and 10% (vol/vol) FBS (Hyclone). Osteocyte and osteoblast cocultures were established on commercially available Falcon cell culture inserts (BD Biosciences) comprised of a polyethylene terephthalate (PET) membrane perforated with 1-μm pores in a random orientation.

Inserts were inverted, and the basal side of the membrane was seeded with 4 × 104 osteoblastic hFOB (950/cm2) in 500 μl basal medium and incubated for 2 h at 37°C to permit cellular adhesion. Inserts were then reverted into six-well tissue culture plates with 2 ml basal medium and 2 × 104 (470/cm2) MLO-Y4, hFOB, MC3T3-E1, or BJ-FIB cells in 2 ml basal medium applied to the apical side of the membrane (Fig. 1A). “Remote” cocultures were set up by seeding 4 × 104 osteoblastic hFOB directly on the surface of six-well tissue culture plates and were subsequently cultured with inserts seeded, on the apical surface, with 2 × 104 MLO-Y4 cells (see Fig. 3A). For monoculture experiments, where osteoblasts were directly exposed to fluid shear, 4 × 104 osteoblastic hFOB were seeded directly on the apical side of inserts. Before experimentation, cultures were maintained for 72 h in basal medium at 37°C with 5% (vol/vol) CO2. The cellular viability of cocultured cells pre- and postflow was assessed microscopically using a commercially available kit (Live/Dead; Invitrogen). We used environmental-scanning electron microscopy (ESEM) to examine the cellular morphology of osteocytes and osteoblasts cocultured on PET membranes. Briefly, cocultures were fixed for 150 min in 2.5% (vol/vol) glutaraldehyde in 0.2 M sodium cacodylate buffer and then washed two times in 0.2 M sodium cacodylate buffer. Samples (5 mm in diameter) were excised from the PET membrane, mounted to scanning electron microscopy stubs with either the apical or basal side of the membrane uppermost, and imaged within a vacuum (2.05–2.65 Torr) on a FEI Quanta 200 ESEM.

Theoretical and computational fluid dynamic modeling of fluid shear.

To apply fluid flow across the apical side of cocultures, we modified the experimental apparatus previously described by Sill et al. (35). Briefly, cell culture inserts were placed in a holder, and a silicon gasket was applied. This complete unit was set inside a precision-machined Lexan chamber and then butted against the rotation apparatus. Once assembled, the compression ring was tightened, causing the expansion of the silicon gasket and separating the basal from the apical chamber. The rotational apparatus consists of a cylindrical stainless steel disk with a variable positioning device that can be set at a predetermined height above the apical side of the membrane. The drive shaft of the disc is connected in series to an identical rotation apparatus, and both are rotated by a motor-driven belt to produce a defined radial shear stress.

Theoretically, the shear stress experienced by cells on the apical side of the membrane ranged radially from zero at the center of the membrane to a maximum at the edge, with the average value being equal to two-thirds that of the maximum (35). Maximum shear stress (τmax) is calculated by τmax = (μωr)/h, where both μ (media coefficient of friction) and r (radius of disc) are constant, and h (height of disc from cells) and ω (rotational frequency) can be varied. A commercial computational fluid dynamics (CFD) program (Fluent) was used to calculate the shear stress using the actual geometry and dimensions of our chamber.

Application of fluid shear.

Cocultures were exposed for 1 h to a τmax of 4.4 dyn/cm2 in flow medium [basal media with 2% (vol/vol) FBS] at room temperature, with control experiments performed in parallel using an identical setup in which the disc cannot rotate. After exposure to fluid shear or control conditions, cocultures were placed in fresh flow medium and postincubated at 37°C with 5% (vol/vol) CO2 in air for the desired time.

Inhibitors of apical and apical-basal GJIC [30 μM 18α-glycyrrhetinic acid (αGA), Sigma; see Ref. 9], mitogen/extracellular signal-regulated kinase (MEK; 10 μM U-0126; Cell Signaling Technology), and appropriate vehicle controls were applied to cocultures 30 min before flow and for the duration of the experiment.

To examine the role of flow-induced secreted factors, osteocytic MLO-Y4 were subjected for 1 h to a τmax of 4.4 dyn/cm2 after which the fluid flow or control conditioned media were collected and directly applied to osteoblastic hFOB for 2 h. Remote cocultures were used to examine the role of factors secreted by osteocytic MLO-Y4 postflow (see Fig. 3A).

Quantification of GJIC.

GJIC between apical MLO-Y4, hFOB, MC3T3-E1, or BJ-FIB cells and basal osteoblastic hFOB was assessed by the transfer of cytosolic calcein (42). Briefly, donor cells (hFOB) were loaded for 25 min with 10 μM calcein AM and 10 μM 1,1′-dioctadecyl-3,3,3′,3′-tetramethylindocarbocyanine perchlorate (DiI) and trypsinized to yield a single cell suspension. Labeled donor cells (9 × 103) were dropped on to the apical side of cocultures or remote cocultures. Donors were incubated with cocultures or remote cocultures for 75 min and washed free of unassociated donors. Cells from both the apical and basal side of the membrane were independently trypsinized and analyzed via fluorescence-activated cell sorter analysis (FACS; FACScan; BD-Pharmingen). Transfer of cytosolic calcein from donors to MLO-Y4 (apical) and MLO-Y4 to hFOB (apical-basal) was quantified by analyzing the number of cells falling within the “acceptor” gate. The acceptor gate was set up using fluorescence intensity and set to exclude 99% of negative cells and 99.25% of donors. Apical and apical-basal GJIC were inhibited in cocultures by the application of 30 μM αGA for the duration of the experiment. Independent monoculture experiments confirmed that 30 μM αGA, in the presence of 2% (vol/vol) FBS, was sufficient to inhibit GJIC between MLO-Y4 and hFOB cells (data not shown).

Transmembrane migration.

FACS studies to assess the rate of transmembrane migration were performed by labeling osteoblastic hFOB with the lipophilic dye tracer DiI. Before cocultures were established, osteoblastic hFOB were incubated for 20 min in a sterile solution of HBSS containing 10 μM DiI and 20 mg/ml BSA. Cells were washed free of unassociated DiI and cultured under normal conditions for 4 days. Cocultures were set up as described previously, but, to increase the probability of transmembrane migration, the number of DiI-labeled hFOB and unlabeled MLO-Y4 cells seeded was increased to 2 × 105 (4,700/cm2). To increase the probability of transmembrane migration, the DiI-labeled cocultures were maintained for 7 days, after which cells were trypsinized from the basal and apical sides of the membrane. Recovered cells were resuspended in PBS containing 0.2% (vol/vol) BSA and 5 mM EDTA and analyzed via FACS. Gates were set up with negative and DiI positively stained hFOB/MLO-Y4, and the percentage of DiI-labeled cells was calculated for apical and basal samples.


Cellular proliferation was assessed as described previously (33) using a commercially available bromodeoxyuridine (BrdU) kit (BD-Pharmingen). Before cells were exposed to fluid shear (24 h), serum concentrations were reduced to 0.5% (vol/vol) to synchronize cells within the cell cycle. Cocultures were subjected for 1 h to a τmax of 4.4 dyn/cm2 in basal medium with 0.5% (vol/vol) serum and subsequently postincubated for 24 h, after which proliferating cells were labeled for 60 min with 10 μM BrdU. Cocultures were washed with PBS, and cells from both the apical and basal side of the membrane were independently trypsinized and fixed in 2% (vol/vol) paraformaldehyde. After subsequent washing and permeabilization steps, cells were incubated for 1 h with DNase to expose the BrdU antigen and incubated with an anti-BrdU antibody conjugated to fluorescein. The percentages of proliferating cells were determined by FACS using gates set up to exclude 99% of a negatively stained cellular population. Positive controls treated with 20% (vol/vol) FBS were set up in parallel to all proliferation experiments (data not shown).

AP activity.

AP activity was determined by the colorimetric conversion of p-nitrophenol phosphate to p-nitrophenol (Sigma) and normalized to total protein (BCA; Pierce) (19). Briefly, inserts were washed with PBS, and osteoblastic hFOB from the basal side of the membrane were independently trypsinized. Osteoblastic hFOB were subsequently lysed with 0.1% (vol/vol) Triton X-100 supplemented with a cocktail of broad-range protease inhibitors (Calbiochem), subjected to two freeze-thaw cycles, and cleared via centrifugation. Lysates were incubated with 3 mg/ml p-nitrophenol phosphate in an alkaline buffer (pH 8) for 30 min at 37°C and read at 405 nM (MRX; Dynex Technologies). The enzymatic activity of AP was determined by comparison with known p-nitrophenol standards (Sigma).

Effects of PGE2 and ATP.

Osteoblastic hFOB and cocultures of osteocytic MLO-Y4 and hFOB were treated directly with flow medium containing vehicle or 5 μM PGE2 or ATP (both from Sigma). Cells were incubated for 3 h, and lysates were collected for quantification of AP activity as previously described.

Statistical analysis.

Data are expressed as means ± SE (n = 6–20) and are from a minimum of three independent experimental runs. Statistical significance was assessed by one-way ANOVA followed by a Student-Newman-Keuls post hoc test, when appropriate, using GraphPad Prism 4 (GraphPad Software). P < 0.05 was considered significant.


Theoretical and CFD modeling of fluid shear.

The theoretical τmax predicted at the outer radius of the disk is 4.991 dyn/cm2 (0.4991 N/m2), where disk radius (r) = 1.108 cm, media coefficient of friction (μ) = 8.434 × 10−4 N·s·m−2, height of disk above cells (h) = 500 μm, and rotations rate (ω) = 255 revolutions/min (26.7 rad/s; Fig. 1B). The CFD model predicts fluid shear to be steady, laminar, and axisymmetric (Figs. 1B and supplemental Fig. 1) and agreed well with the theoretical model except at the outer radius of the disk, where the shear stresses drop off rapidly.1 The CFD model predicts that cells located on the apical surface of the coculture experience a τmax of 4.408 dyn/cm2 (0.4408 N/m2). Although this value is 12% lower than that predicted by the theoretical model, it is still within the range of shear stresses in which osteocytes are predicted to experience in vivo (31).

Characterization of osteocytic and osteoblastic cocultures.

ESEM showed that that cocultured osteocytic MLO-Y4 formed an interconnected cellular network and confirmed that they maintain their classical dendritic phenotype (Fig. 2A ) (25). Moreover, at higher magnification, ESEM revealed that these dendritic extensions form a close association with the pores in the PET membrane (Fig. 2B). In contrast, cocultured osteoblastic hFOB grew to form a confluent monolayer, effectively covering the majority of the pores on the basal surface of the PET membrane (Fig. 2C). At higher magnification, both covered (as viewed through the cellular monolayer) and uncovered pores can be visualized on the basal surface (Fig. 2D).

Because both flow and control experiments were performed outside of the incubator, we sought to quantify changes in media temperature and pH. Because of the presence of HEPES buffer in the media over the duration of the experimental period (1 h), there was a negligible increase in media pH (0.11 pH units); however, temperature was observed to steadily decline under both flow and control conditions by an average 21.4 + 2.7% (data not shown). Additionally Live/Dead staining revealed that, during the duration of the experiment, apical osteocytic MLO-Y4 and basal osteoblastic hFOB cocultured on PET membranes remained viable (data not shown).

Quantification of GJIC.

To confirm the presence of functional heterotypic gap junctions between cocultured osteocytes and osteoblasts, we used a calcein ester conjugate that, when subjected to intracellular esterase activity, becomes a membrane-impermeable fluorescent dye with a molecular mass of 0.66 kDa. Because gap junctions permit the cellular movement of molecules <1 kDa, calcein introduced in apical osteocytic MLO-Y4 can pass, via gap junctions, to osteoblastic hFOB on the opposite side of the insert. Thus, by assessing the level of dye transferred from apical MLO-Y4 to basal hFOB with FACS, we can examine heterotypic osteocyte-osteoblast GJIC within our coculture model. FACS analysis of adjacent cocultures showed high levels of dye transfer from the osteocytic MLO-Y4 to cocultured osteoblastic hFOB cells (Fig. 3B). There was a significantly greater amount of dye transferred to hFOB cells cocultured adjacent to osteocytic MLO-Y4 relative to those cocultured with remote MLO-Y4 (P < 0.001, Fig. 3B). There was no significant difference between the levels of calcein taken up by MLO-Y4 cells adjacent to hFOB or MLO-Y4 cells in remote coculture with hFOB. Inhibition of GJIC by the application of 30 μM αGA, a pharmacological GJIC inhibitor, significantly reduced the apical-basal transfer of calcein in adjacent cocultures (Fig. 3B, P < 0.001) but had no significant effect in remote cocultures. GJIC was also detected between cocultured hFOB, MC3T3-E1, or BJ-FIB and hFOB osteoblastic cells (data not shown).

Transmembrane migration.

Transmigration of cells from the basal to the apical surface in adjacent cocultures was negligible, such that unlabeled osteocytic MLO-Y4 and DiI-labeled osteoblastic hFOB remain spatially isolated even after extended periods of culture (Fig. 4A).

Fig. 4.

Cocultured osteoblastic hFOB do not migrate through cell culture inserts and respond to fluid shear differently than hFOB alone. A: osteoblastic hFOB prelabeled with the lipophilic cell tracer 1,1′-dioctadecyl-3,3,3′,3′- tetramethylindocarbocyanine perchlorate (DiI) were cocultured for 7 days with unlabeled osteocytic MLO-Y4. Cells collected from both the apical and basal surfaces of the PET membrane were analyzed via fluorescence-activated cell sorter analysis for DiI. Bars represent means ± SE; n = 12 experiments. B: cocultured osteoblastic hFOB respond differently than isolated hFOB to a τmax of 4.4 dyn/cm2. Osteoblastic hFOB exposed directly for 1 h to a τmax of 4.4 dyn/cm2 fluid shear show significantly increased proliferation after 24 h (*P < 0.05 flow vs. control). Fluid shear had no significant effect on cellular proliferation in cocultures. Bars represent means ± SE; n = 6–8 experiments.


In accordance with previous reports, we found that directly subjecting osteoblastic hFOB cells to a physiological level of fluid shear stress for 1 h (τmax of 4.4 dyn/cm2) significantly increased cellular proliferation after 24 h (P < 0.05; Fig. 4B) (24, 32, and 38). Because osteocytes in vivo are encapsulated within bone matrix, we would predict that applying fluid shear to cocultured MLO-Y4 would not induce a proliferative response, and our data supported this hypothesis (data not shown). We did not observe any significant difference between the rates of proliferation of osteoblastic hFOB cells cocultured with fluid shear-exposed MLO-Y4 vs. those cocultured with control treated MLO-Y4 (Fig. 4B).

AP activity.

Osteoblastic hFOB cells directly subjected for 1 h to a τmax of 4.4 dyn/cm2 and postincubated for 2 h showed no change in AP activity when compared with unflowed controls (Fig. 5A). Conversely, when cocultured in direct physical contact and in GJIC with osteocytic MLO-Y4 cells that were subjected to a τmax of 4.4 dyn/cm2, the osteoblastic hFOB cells, which were not exposed to fluid flow, showed a rapid and highly significant increase in AP activity (P < 0.01; Fig. 5A). To address the possibility that increases in hFOB AP activity are mediated through the osteocytic release of soluble factors, we treated hFOB with media conditioned by flow-exposed osteocytic MLO-Y4 and examined the effects of culturing hFOB in remote cocultures with flowed MLO-Y4 (Fig. 3A). By remotely culturing osteoblastic hFOB on the bottom of tissue culture plates instead of on the inserts, we were able to reproduce the coculture model in a manner that prevented osteocytes and osteoblasts from being in direct contact but permits the diffusion of secreted factors. Neither hFOB cells exposed to media conditioned by flowed MLO-Y4 (data not shown) nor those in remote coculture with MLO-Y4 exposed to flow displayed increases in AP activity (Fig. 5A).

Fig. 5.

Effects of exposure to a τmax of 4.4 dyn/cm2 on osteoblastic hFOB are different when mediated through osteocytic MLO-Y4 and are reliant on functional GJIC and extracellular signal-regulated kinase (ERK) signaling. A: when osteoblastic hFOB cells are cocultured adjacent to osteocytic MLO-Y4 exposed for 1 h to a τmax of 4.4 dyn/cm2, a highly significant increase in alkaline phosphatase (AP) activity is observed 2 h postflow (**P < 0.01 flow vs. control). No significant effect of fluid shear on AP activity was observed when hFOB cells were directly exposed or cocultured indirectly with flowed osteocytic MLO-Y4. Application of 30 μM of the GJIC inhibitor αGA abolished the fluid shear-stimulated increase in AP activity, as did inhibiting ERK1/2 phosphorylation with the mitogen/extracellular signal-regulated kinase inhibitor U-0126. Bars represent means ± SE; n = 10–20 experiments. B: there was no significant effect on the AP activity of osteoblastic hFOB cells when cocultured adjacent to either fibroblasts (BJ-FIB) or osteoblasts (MC3T3-E1 and hFOB) themselves exposed for 1 h to a τmax of 4.4 dyn/cm2. Bars represent means ± SE; n = 10 experiments.

To examine the role of GJIC in the osteocytic communication of mechanical signals to remote osteoblasts, we repeated our experiments in the presence of the GJIC inhibitor αGA (9). Application of 30 μM αGA completely abolished the fluid shear-induced increase in osteoblastic AP activity in MLO-Y4 hFOB cocultures (Fig. 5A). Our laboratory has shown a relationship between extracellular signal-regulated kinase (ERK) 1/2 phosphorylation and increases in GJIC in osteocytic MLO-Y4 (1). By treating our cocultures with the MEK inhibitor U-0126, thus preventing ERK1/2 phosphorylation, we could block the fluid shear-induced, MLO-Y4-mediated increase in osteoblastic AP activity (Fig. 5A).

To determine if the flow-induced increase in osteoblastic AP activity is attributed to the nature of the coculture system or dependent on the presence of osteocytes, we replaced the apical osteocytic MLO-Y4 with either a fibroblastic (BJ-FIB) or osteoblastic (MC3T3-E1 or hFOB) cell line. We observed no significant effect on the AP activity of osteoblastic hFOB when they were cocultured with either BJ-FIB, MC3T3-E1, or hFOB cells exposed for 1 h to τmax of 4.4 dyn/cm2 (Fig. 5B).

Effects of PGE2 and ATP.

It has been recently demonstrated that the protein components of gap junctions can also exist in unopposed membranes where they form functional channels (21) and, importantly, that these “hemichannels,” like gap junctions, are susceptible to inhibition with αGA. Within osteocytic MLO-Y4, it has been shown that fluid shear activates hemichannels facilitating the release of both ATP (26) and PGE2 (8, 20) in the extracellular environment. To assess the potential role of ATP or PGE2 released through fluid shear-activated hemichannels, we treated both cocultures and hFOB monocultures with exogenous 5 μM ATP or PGE2. Neither PGE2 nor ATP treatment had a significant effect on osteoblastic AP activity in cocultures (Fig. 6). PGE2 did not have an affect on AP activity of hFOB monocultures; however, 5 μM ATP significantly reduced AP activity in monocultures of hFOB (P < 0.05; Fig. 6).

Fig. 6.

Effects of exposure to a τmax of 4.4 dyn/cm2 on osteoblastic hFOB are not mediated directly through the actions of PGE2 or ATP. Treating cocultures with either 5 μM PGE2 or ATP did not alter the AP activity of osteoblastic hFOB. hFOB monocultures treated with 5 μM ATP showed a significant decrease in AP activity when compared with untreated controls (*P < 0.05, ATP vs. control). Bars represent means ± SE; n = 11–22 experiments.


We hypothesized that GJIC provides the means by which mechanical signals are transmitted through the osteocytic network and by which they direct the actions of remote osteoblasts. To examine our hypothesis, we designed a novel in vitro coculture model to better mimic the three-dimensional and multicellular architecture of bone in vivo. Unlike traditional parallel plate systems where monocultures of osteocytes, osteoblasts, or MSCs are exposed to fluid shear, our system enables us to apply fluid shear to osteocytes while they are in direct contact with effector cells, such as osteoblasts, while not exposing osteoblasts to fluid shear (Fig. 1A). Additionally, by utilizing porous PET membranes in our cocultures, we have the advantage over traditional monocultures of being able to model in vivo intercellular communication between embedded osteocytes and surface-residing osteoblasts, thus enabling us to study the role of both secreted and direct (physical) intercellular communication in mechanotransduction.

Numerous studies have examined GJIC between bone cells and its requirement in bone cell differentiation (1, 13, 14, 22, 34, 37). Doty (11) provided morphological evidence that in vivo heterotypic gap junctions form between osteocytes and osteoblasts, and, more recently, in vitro evidence has demonstrated the functionality of gap junctions forming between osteocytes and osteoblasts (42). ESEM imaging of our cocultures revealed that cellular processes from osteocytic MLO-Y4 form close association with the pores present in our PET membrane (Fig. 2B), implying that, within our system, osteocytes and osteoblasts are in physical contact and may be forming gap junctions. We demonstrated the functionality of osteocytic-osteoblastic gap junctions across the PET membrane by following the transfer of calcein from the apical osteocytic MLO-Y4 to the basally located osteoblastic hFOB and were able to significantly block dye transfer by treating cocultures with αGA, a pharmacological GJIC inhibitor (Fig. 3B). Importantly, cell tracer studies that used PET membranes with randomly orientated 1-μm pores showed that we were able to maintain cellular segregation even after extended periods of coculture (Fig. 4A).

An important experimental difference between ours and previously described studies is the profile of fluid shear applied. Studies utilizing parallel plate chambers are specifically designed to impart a uniform level of fluid shear across the entire cellular monolayer, but, by utilizing a rotating disc in our system, our cocultured osteocytes were exposed to different levels of fluid shear ranging from 0 dyn/cm2 at the center to 4.4 dyn/cm2 at the outermost edge of the membrane. By utilizing CFD, we were able to model the fluid shear produced by our disc and confirmed that it was laminar (Fig. 1B). We suggest that, by applying a gradient of fluid shear across the osteocytic monolayer, rather than more uniform flow, our system exposes cells to a flow pattern more representative of that experienced by osteocytic networks in vivo. A limitation of our coculture model is that the system is located outside of a CO2-gassed incubator. However, by treating flow and control cocultures identically and by including 15 mM HEPES in our flow media, we can minimize any effects stemming from temperature and pH variations. Our coculture system enables us to apply fluid shear to osteocytes while they are in physical contact with osteoblasts, thus enabling us to apply a mechanical signal and in real time observe the downstream cellular consequences in cells not themselves exposed to the mechanical signal (Fig. 1A).

The most reproducible downstream consequence of applying fluid shear directly to potential effector cells in vitro is to increase cellular proliferation (24, 28, 29, 32, 33). While many studies suggest fluid shear increases bone cell proliferation, studies examining the effect of fluid shear on bone cell differentiation have produced less consistent results. In vitro studies examining the role of fluid shear in modulating osteoblastic differentiation commonly assess changes in events associated with bone deposition (2), such as production of matrix proteins and increases in AP activity. Although fluid shear has been demonstrated in MC3T3-E1 osteoblasts to upregulate mRNA levels of the bone matrix protein osteopontin (43), others have found that osteopontin levels, and indeed the capacity to produce a mineralized bone matrix, remain unaffected or downregulated by fluid shear (18, 32). Studies examining the role of fluid shear in modulating AP activity have also been conflicting. Kapur et al. (24) and Liegibel et al. (29) demonstrated that primary human osteoblasts respond to physiologically high and low levels of fluid shear with a significant increase in AP activity. However, Hillsley and Frangos (18), using primary rat osteoblasts, demonstrated that shear stresses of 5 dyn/cm2 resulted in the rapid downregulation of both AP activity and mRNA. Furthermore, Nauman et al. (32) detected no significant effect of fluid shear on the AP activity of MSCs derived from rat marrow. We propose that these experimental inconsistencies may also be attributed to the need for an osteocyte to act as a “cellular interpreter” that detects perturbations in fluid shear and directs the actions of effector cells. By using osteoblastic monocultures along with our osteocytic-osteoblastic cocultures, we demonstrate that osteoblasts respond in different ways to the same level of fluid shear and that the resultant behavior depends on whether they perceive fluid shear directly or indirectly. Osteoblasts exposed directly to a fluid shear of 4.4 dyn/cm2 respond with a significant increase in cellular proliferation, as previously reported (Fig. 4B and Ref. 24, 28, 29, 32, 33). However, osteoblastic hFOB cocultured with osteocytes, which were exposed to the same level of fluid shear, showed no significant alteration in cellular proliferation (Fig. 4B) but rather showed a rapid and significant increase in AP activity, consistent with increased osteoblastic activity (Fig. 5A).

Interestingly, the inability of other mesenchymal cell types, including fibroblasts, MC3T3-E1, or hFOB themselves, to modulate osteoblastic activity in response to flow may be a result of differing levels of GJIC within these cell types (Fig. 5B). When calcein was introduced into monocultures of osteocytic MLO-Y4, it was transferred to 82% of the population in 75 min; however, in identical monoculture experiments, calcein was transferred to 40.9% of hFOB; 44.6% of MC3T3-E1 and 46.7% of the BJ-FIB population. These data may explain why, within our coculture, MLO-Y4 cells, which have a higher level of GJIC, are unique in their ability to act as cellular interpreters of fluid shear.

There are numerous signaling candidates that may be employed by osteocytes in communicating mechanical signals to osteoblasts; however, current literature seems to implicate the actions of osteocytic-secreted factors (8, 16, 26, 27, 44). For example, a recent study by Heino et al. (16) implies that even unstimulated osteocytes secrete osteotropic factors, since osteoblasts exposed for extended periods of time to media conditioned by osteocytic MLO-Y4 display an increased osteogenic potential with elevated levels of AP activity. Thus, within our coculture model, we sought to clarify the potential role of factors secreted by osteocytic MLO-Y4 in response to fluid shear. We found that, compared with controls, conditioned media taken from osteocytic MLO-Y4 exposed to fluid shear and applied directly to osteoblastic hFOB had no significant effect on their AP activity (Fig. 5A). Similarly, there was no significant change in the AP activity of osteoblastic hFOB that were cocultured remotely with fluid flow-exposed osteocytes when compared with unflowed controls (Fig. 5A). These data suggest that osteocytes are required to be in physical contact with osteoblasts to communicate mechanical signals to them and implicate a role for GJIC. Indeed, when we repeated our fluid shear experiments in the presence of 30 μM αGA, we were able to significantly inhibit the fluid shear-induced, MLO-Y4-mediated increase in osteoblastic AP activity (Fig. 5A).

Intriguingly, Jiang and Cherian (20) recently identified fluid shear-sensitive hemichannels in osteocytic MLO-Y4. Hemichannels are formed of connexins, the same protein components of gap junctions, and appear to behave like classical gap junctions; however, they function without the need for an opposing cell, realizing signaling molecules in the extracellular environment. In MLO-Y4 cells, hemichannels pass small molecules such as ATP (26) and PGE2 directly in the extracellular environment (8, 20). However, neither PGE2 nor ATP treatment had a significant effect on osteoblastic AP activity in cocultures (Fig. 6). In hFOB monocultures, PGE2 did not affect AP activity; however, 5 μM ATP did significantly reduce the AP activity of hFOB monocultures (P < 0.05, Fig. 6). These data, together with conditioned media and remote coculture studies, indicate that the fluid shear-induced, MLO-Y4-mediated increases in osteoblastic activity are not a direct result of the actions of PGE2 or ATP. However, they do not rule out a potential codependent role for these or other unidentified factors in the transduction of mechanical signals. Indeed, Cheng et al. (5) demonstrated that fluid shear-induced PGE2 release mediates GJIC in osteocytic MLO-Y4. Thus it remains possible that, within our cocultures and in response to fluid shear, a soluble factor is released and that this, along with GJIC and other cellular triggers, enables osteocytes to modulate osteoblast activity.

The concept that osteocytes communicate physiological signals to osteoblasts via GJIC correlates well with several recent studies that report fluid shear can directly affect GJIC and the membrane localization of connexins in both osteocytic MLO-Y4 (5) and a murine osteoblast model (41). Furthermore, we have shown that physiological levels of fluid shear act to increase the number of gap junctions formed in a network of osteocytic MLO-Y4 in a manner dependent on the activation of ERK1/2 (1). Although not definitive, by utilizing the MEK inhibitor U-0126 in our coculture studies, we could block flow-induced increases in AP activity (Fig. 5A). Thus functional ERK1/2 signaling is a prerequisite for osteoblasts to respond to osteocytic-communicated mechanical signals, although the exact role of mitogen-activated protein kinase signaling remains undetermined.

One of the more surprising outcomes of these studies is the speed at which osteoblastic activity was modulated in response to fluid shear. In as little as 2 h postflow, we gained robust and significant increases in AP activity. Typically, increases in the level of osteoblastic activity are reported following days of treatment with osteogenic agents such as 1,25-dihydroxyvitamin D3 (15) and bone morphogenic protein-2 (40). Therefore, our data suggest that osteocytic cells not only detect and communicate biophysical signals to osteoblasts but, remarkably, enhance the signal enabling osteoblasts to rapidly respond to a signal they cannot themselves detect. As such, these data provide valuable insight into the mechanisms by which the osteocyte may transmit biophysical signals to effector cells, such as the osteoblast, in vivo. Our novel coculture model provides for the first time direct evidence that a biophysical signal can induce very different cellular responses from an effector cell and that the resultant outcome is dependent on the cell that interprets the stimulus.


These studies were funded by National Institute on Aging Grants AG-13087–09 and K25AG-022464 and by the Whitaker Foundation.


We thank Nate Sheaffer of the Flow Cytometry Core Facility at the Pennsylvania State University College of Medicine for assistance with FACS analysis, Mark Angelone at the Materials Characterization Laboratory at the Pennsylvania State University for assistance with ESEM, Don Whitehaus of Whitehaus Precision Machine for assistance in the design of custom equipment, and Dr. J. Bidwell for critically reviewing this manuscript.


  • 1 The online version of this article contains supplemental data.

  • The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.


View Abstract