Mechanosensory responses of osteocytes to physiological forces occur along processes and not cell body and require α_Vβ₃ integrin

Significance

Although polarized mechanotransduction in osteocytes has been established, participation of α_Vβ₃ integrin still remains to be conclusively demonstrated. We addressed this issue using the novel Stokesian fluid stimulus probe to discretely stimulate the osteocyte processes and cell body with physiologically relevant hydrodynamic forces in the piconewton range while imaging intracellular Ca²⁺ signals. We demonstrate that osteocyte cell processes but not the cell bodies are mechanosensitive through discrete attachment sites provided by α_Vβ₃ integrin, thereby revealing that integrin attachment sites provide the substrate for polarized osteocyte mechanosignaling and mechanotransduction.

Abstract

Osteocytes in the lacunar–canalicular system of the bone are thought to be the cells that sense mechanical loading and transduce mechanical strain into biomechanical responses. The goal of this study was to evaluate the extent to which focal mechanical stimulation of osteocyte cell body and process led to activation of the cells, and determine whether integrin attachments play a role in osteocyte activation. We use a novel Stokesian fluid stimulus probe to hydrodynamically load osteocyte processes vs. cell bodies in murine long bone osteocyte Y4 (MLO-Y4) cells with physiological-level forces <10 pN without probe contact, and measured intracellular Ca²⁺ responses. Our results indicate that osteocyte processes are extremely responsive to piconewton-level mechanical loading, whereas the osteocyte cell body and processes with no local attachment sites are not. Ca²⁺ signals generated at stimulated sites spread within the processes with average velocity of 5.6 μm/s. Using the near-infrared fluorescence probe IntegriSense 750, we demonstrated that inhibition of α_Vβ₃ integrin attachment sites compromises the response to probe stimulation. Moreover, using apyrase, an extracellular ATP scavenger, we showed that Ca²⁺ signaling from the osteocyte process to the cell body was greatly diminished, and thus dependent on ATP-mediated autocrine signaling. These findings are consistent with the hypothesis that osteocytes in situ are highly polarized cells, where mechanotransduction occurs at substrate attachment sites along the processes at force levels predicted to occur at integrin attachment sites in vivo. We also demonstrate the essential role of α_Vβ₃ integrin in osteocyte-polarized mechanosensing and mechanotransduction.

A dynamic and complex structure such as bone has the ability to adapt and to adjust to changes in its functional environment. Emerging evidence suggests that osteocytes play an important role in regulating bone mechanoadaptation and in adjusting and maintaining bone mass (1–6). Osteocytes, which account for ∼90% of the entire bone cell population, are viewed as the main mechanosensing cells that detect whole tissue mechanical loading due to their unique distribution throughout the mineralized matrix and to their connection to the neighboring osteocytes and osteoblasts via gap junctions (7).

To sense and respond to mechanical loading and thus maintain bone homeostasis, the osteocyte must be properly anchored to its extracellular surroundings (reviewed in ref. 8). In vivo, the osteocyte cell processes in the lacunar–canalicular system (LCS) are surrounded by transverse tethering elements (9–11) and are directly connected to the canalicular wall at discrete attachment sites (12, 13) that contain β₃ integrin (12). Integrins are transmembrane proteins that link the cell’s cytoskeleton to the extracellular matrix and are recognized for their key roles in mechanosensory transduction (14, 15). All bone cells express integrins, and osteocytes in particular express both β₁ and α_Vβ₃ integrins (16–19).

It has been proposed that the interstitial fluid flow in the LCS results in focally high membrane strain around both the tethering and discrete attachment sites along the process (9, 10) but not at the osteocyte cell body where such attachments are not observed (11–13). This unique local structural arrangement, and its strain amplification effects, make the osteocyte process a strong candidate for highly specialized mechanosensation and mechanotransduction (8). Such a role is supported by in vitro findings; for example, observations that dendritic osteocyte processes are far more sensitive to mechanical deformation than the cell body (20) and that osteocytes respond in a polarized manner when their processes and cell body are discretely stimulated (21). Although osteocyte conductance changes when stimulated at the focal attachment sites (22), it remains to be demonstrated that the signals generated at these discrete sites along the osteocyte cell processes are sufficient to activate the osteocyte mechanotransduction and signaling pathways. Furthermore, although the presence of integrins at the attachment sites of the osteocyte’s process with the canalicular wall suggest their participation in osteocyte mechanoactivation, such a role still remains to be conclusively demonstrated.

To address these points, we used the novel Stokesian fluid stimulus probe (SFSP) (21) to achieve the spatial resolution required to discretely stimulate the osteocyte processes and cell body with physiologically relevant hydrodynamic forces at piconewton levels. We imaged changes in intracellular Ca²⁺ levels to assess osteocyte mechanoactivation because Ca²⁺ changes are among the first events observed following mechanical stimulation of bone cells. In addition, intracellular Ca²⁺ signaling not only participates in the release of mechanosignaling molecules, such as ATP and prostaglandin E₂ (PGE₂), but has also been proposed to functionally couple and coordinate cellular Ca²⁺ and electrical activity within the osteocyte network (22, 23).

Results

Polarized Mechanotransduction in Osteocytes.

Previous studies using unidirectional and pulsatile fluid flow mechanical stimulation have shown that osteocytes respond with global increase in intracellular calcium concentration ([Ca²⁺]_i) (4, 23). This form of stimulation, however, does not allow discrimination of which regions of the cell initiate the response. Using the SFSP to focally stimulate the cells while recording intracellular Ca²⁺, we have characterized the functional polarity of osteocyte mechanoresponsiveness in individual murine long bone osteocyte Y4 (MLO-Y4) cells. As shown in Fig. 1, solitary osteocytes responded to the SFSP with increase in [Ca²⁺]_i in a polarized manner. For example, when we focally stimulated a MLO-Y4 cell process, we detected a significant increase in [Ca²⁺]_i at the point of stimulation (denoted as “0 μm”; see phase image in Fig. 1) and the generated Ca²⁺ signal spread along the cell process and into the cell body with an average speed of 5.6 ± 0.5 μm/s (n = 6), as illustrated in graphical representations (Fig. 1A and Movie S1). In contrast, cell body stimulation of the same cell (point of stimulation denoted as “0 μm”) produced only a small response (Fig. 1B and Movie S2). This is an example of the polarized osteocyte mechanotransduction behavior we typically observed when both cell process and body were firmly attached to the underlying substrate.

Fig. 1.

(Upper) Cell process rather than the body of osteocyte responds to SFSP stimulation with increase in intracellular Ca²⁺ levels. Sparsely plated MLO-Y4 cells were stimulated with SFSP along the process (A) or cell body (B) at an average distance of 3–4 μm from cell surface. Note at 25 s increased [Ca²⁺]_i in the cell body after process stimulation (A) whereas no change in [Ca²⁺]_i in the cell body after body stimulation (B). (Lower) The 3D graphical representations of changes in [Ca²⁺]_i in response to SFSP stimulation from regions of interest (ROIs, black dots corresponding to various distances from point of stimulation in Upper) placed along either the cell process or on the cell body were plotted using Microsoft Excel software (Lower; point of process stimulation: red trace). Arrows indicate moment of SFSP stimulation. (Scale bar, 10 μm.)

The Role of Firm Attachment Sites in Osteocyte Response to SFSP Stimulation.

To determine responsive regions of osteocytes, we sequentially stimulated sites along the cell processes and cell bodies of MLO-Y4 cells. Focal stimulation with SFSP on a cell process at a firm attachment site (as illustrated in Fig. 1A) resulted in a significant localized increase in [Ca²⁺]_i that spread along the cell process (Fig. 2A and Movie S3). Stimulation of regions lacking local attachment sites did not result in Ca²⁺ responses, but created transient artifactual signals due to mechanical buckling (Fig. 2B and Movie S4). To determine whether stimulated regions were firmly attached, we localized the site of Ca²⁺ signal initiation relative to that of the SFSP application (Fig. S1). Our calculations revealed that stimulated and response initiation regions coincided when firm attachments were present, whereas response lagged stimulation when local attachments were absent in that region of the cell. Our findings indicate that 50% of the responses originated from firm attachment regions (53 responses to 107 stimulations), whereas the remaining 22% of the responses were initiated from nearby firm attachment sties (ranging from 2 to 18 μm away) (Fig. 2 C and D).

Fig. 2.

Firm attachment sites along the processes of MLO-Y4 play a key role in SFSP stimulation. Graphical representations of changes in [Ca²⁺]_i in response to SFSP stimulation from firm attachment sites (A) and no local attachment sites (B) on osteocyte cell processes. Ca²⁺ response traces corresponding to various distances from point of stimulation (0 μm) were plotted with respect to amplitude vs. time vs. distance (matching ROI: black dots; see corresponding Insets). Arrows indicate SFSP stimulation time point. (Scale bar, 10 μm.) (C). Statistics of number of responses vs. Ca²⁺ signal initiation point relative to point of stimulation. Firm attachments (square dots) indicate that Ca²⁺ signal initiation point is the same as point of stimulation, whereas no local attachments (diamond-shaped dots) represent otherwise. (D). Pie chart representing the manner in which MLO-Y4 cell processes respond to SFSP stimulation.

The SFSP generates local forces in the 1–5 pN range as evidenced in part from measured local process deflections (21). In our studies using cells visualized with Ca²⁺ indicator, we could not reliably measure the extent of deflection because of the very small deflections that occur when SFSP is positioned in the vicinity of an attachment site. However, in our previous studies using phase contrast microscopy, such deflections were consistently observed where there was no local attachment site at the point of SFSP stimulation (21). To estimate the magnitude of the forces activating osteocytes when stimuli were applied to regions lacking firm attachments, we assumed the force F acting at the midpoint between two firm attachment sites to be in the same range of 1–2 pN as predicted by Wu at al. (21) and calculated maximum process deflection (δ_m). Using F = 48EIδ_m/L³, where EI (flexural rigidity of the central actin filament bundle in the process) = 2.2 × 10⁸ pN ⋅ nm², allows calculation of δ_m. Using the results from Fig. 2C, we measured the length (L) between two firm attachment sites on the process to be between 4 and 38 µm. Our calculation indicated that δ_m is similar to the values from table 2 in Wu et al. (21) in the range of 3–5 µm for L ≥ 30 µm (Table S1). For lengths shorter than 30 µm, δ_m varies from 6 nm to 3 µm (Table S1) for SFSP forces of 1–5 pN. For distances less than 6 µm, deflections in the absence of local attachments are calculated to be so low that direct activation by the SFSP is likely [figure 5 c and d and figure S3 in Wu et al. (21)].

As shown in Fig. 1, cell body response to SFSP stimulation was minimal when the body appeared to be firmly attached, i.e., no distinct lateral deflection was observed upon stimulation. However, SFSP stimulation of the cell body elicited a response when the cell body appeared to be less firmly attached, i.e., a distinct horizontal deflection was observed upon stimulation (Fig. 3A and Movie S5). In cases where lateral displacements of the cell body were observed, the Ca²⁺ response was initiated on the cell process rather than on the cell body, as evidenced by the earlier rise in [Ca²⁺]_i in the cell process (solid trace in Fig. 3B) and delayed increase in [Ca²⁺]_i in the cell body (dashed trace in Fig. 3B). This finding indicates that the Ca²⁺ signal is triggered in response to tensile forces that are applied from the cell body to the closest anchoring sites on the cell processes (Fig. 3B). A polar plot showing the locus of initiation of each response on the cell process relative to the cell body location revealed that such initiation regions were on average about 16 μm from the center of the cell body (Fig. 3D). These indirect responses to SFSP were observed in 59% of the stimulated cell bodies (n = 19), whereas no responses were observed when the cell body did not exhibit a lateral displacement (n = 12) (Fig. 3C).

Fig. 3.

Summary of the manner in which the MLO-Y4 cell body responds to SFSP stimulation. (A) An example of a cell body responding to SFSP stimulation. Traces of the Ca²⁺ responses recorded at various points along the osteocyte (depicted by black dots in the Inset) were plotted with respect to amplitude vs. time vs. distance from the point of stimulation (0 μm). (B) Concurrent Ca²⁺ response traces at 0 μm (cell body; black solid trace) and −11 μm (cell process; dashed trace) showing that the Ca²⁺ signal originated from the cell process and not from the cell body. The arrow indicates the time of SFSP stimulation. Vertical bars indicate the time at the half-maximal increase in [Ca²⁺]_i recorded at 0 μm and −11 μm, from which a delay of ∼4 s between the Ca²⁺ signal initiation on the cell process and the cell body response can be estimated. (C). Pie chart summarizing how MLO-Y4 cell bodies respond to SFSP stimulation. (D) Polar plot of Ca²⁺ signal initiation points on cell process (black dots) relative to the point of stimulation on the cell body (white dot).

Regardless of the initiation sites, the average amplitudes of Ca²⁺ responses at the point of origin were similar (∼50 nM) (Fig. S2A). This finding suggests that local forces acting in these circumstances are also most likely comparable in magnitude (1–5 pN, see Table S1).

The Role of ATP in Osteocyte-Polarized Mechanotransduction.

Although the average velocity of the Ca²⁺ signal spread along the osteocyte process was uniform, we routinely observed amplification of the Ca²⁺ signal as it spread from the stimulated process toward the cell body (Figs. 1A and 2A). Because ATP is one of main signaling molecules released by bone cells in response to mechanical stimulation (24), we speculated that ATP might play a role in Ca²⁺ signal transmission and amplification. To test this hypothesis, we used apyrase, an enzyme that hydrolyses ATP to AMP and thereby functions as an extracellular ATP scavenger. As such, apyrase diminishes autocrine/paracrine ATP signaling and reduces Ca²⁺ responses mediated by activation of the ATP cell surface purinergic (P2) receptors. In the presence of apyrase, the amplitude of the Ca²⁺ responses induced by SFSP stimulation significantly reduced and in the majority of the cases was below 20 nM (our established threshold response), but the response was recovered upon apyrase washout (Fig. 4 and Fig. S3). Thus, our data suggest that ATP not only plays a critical role in SFSP-induced Ca²⁺ signal transmission and amplification, but also that piconewton-level forces are sufficient to evoke release of mechanosignaling molecules.

Fig. 4.

Apyrase treatment significantly reduces SFSP induced Ca²⁺ response in osteocytes. MLO-Y4 cells were stimulated on the process and average change in the amplitude of the Ca²⁺ response was measured from processes and cell bodies under control (Ctrl) conditions, in the presence of apyrase (50 U/mL) and after washing. All data are presented as mean ± SEM, n = 41 (Ctrl), n = 24 (apyrase), and n = 11 (wash) (**P < 0.005, ****P < 0.00005). Note that the amplitude of the majority of the Ca²⁺ responses under apyrase treatment were below the response threshold of 20 nM.

The Role of α_Vβ₃ Integrin Attachment Sites in Osteocyte-Polarized Mechanotransduction.

It has been proposed that the integrin plaques along osteocyte processes play a key role in osteocyte mechanosensing and mechanotransduction by providing discrete attachment sites (12, 13). To test this role in cultured bone cells, we used the fluorescence near-infrared (NIR) probe IntegriSense 750, a potent and selective nonpeptide small molecule α_Vβ₃ integrin antagonist, to compromise the attachment sites. This probe allowed simultaneous localization of α_Vβ₃ distribution and Ca²⁺ imaging as shown in Fig. 5A and Movie S6. The observed distribution of α_Vβ₃ along both MLO-Y4 cell processes and body is consistent with its reported localization in cultured rat osteocytes (17), but it contrasts with the discrete distribution along the osteocyte processes seen in vivo (12). Treatment with IntegriSense for 30 min completely abolished Ca²⁺ responses to cell process stimulation; the small transient Ca²⁺ signals detected were the movement artifacts resulting from compromised attachment sites. Treatment with IntegriSense for 15, 30, and 45 min demonstrated that the disruption of integrin attachments was time dependent (Fig. 5B).

Fig. 5.

Inhibition of α_Vβ₃ integrin attachment sites, using a NIR fluorescence probe IntegriSense 750, compromises localized responses to SFSP stimulations. (A) An example of α_Vβ₃ integrin distribution (Upper Left) and corresponding phase image (Upper Right) with ROIs before SFSP stimulation. (Lower) Graphical representations of cell process Ca²⁺ responses to SFSP stimulation after 30 min incubation with IntegriSense 750. Ca²⁺ response traces relating to various distances from point of stimulation (0 μm, Upper Right) were plotted with respect to amplitude vs. time vs. distance. Arrows indicate time of SFSP stimulation (Scale bar, 10 μm.) (B) Percent of cell processes responding to SFSP stimulation after 15, 30, and 45 min incubation with IntegriSense 750.

To summarize, in cells firmly attached to the substrate, we observed minimal Ca²⁺ responses when cell body was stimulated, compared with robust responses of cell processes. Our interpretation is that the cell processes are much stiffer than the cell body (9), so low forces are immediately transmitted across the process to pull on the α_Vβ₃s. The cell bodies are soft, so piconewton forces applied to the cell body may deform the cell membrane (like a finger pushing into a balloon), but they are not easily transmitted across the cell body relative to α_Vβ₃s at its base.

Piconewton-Level Forces Initiate Ca²⁺ Signaling in Osteocyte Networks.

Transmission of intercellular Ca²⁺ signals between osteocytes has been proposed to play a key role in coordinating activity within the osteocytic network (23, 25, 26). Models of bone loading predict sensitivity to very low applied forces (13), and we have observed that piconewton-level forces produced by the SFSP can initiate signaling both within the stimulated cell and extending to others beyond it (Fig. 6 and Movie S7). This type of response was commonly detected in networks consisting of 2 to 15 cells and likely involved both direct gap junction coupling and activation of P2 receptors (22, 27, 28).

Fig. 6.

SFSP stimulation can initiate response in a network of MLO-Y4 osteocytic cells. The 3D graphical representation of the Ca²⁺ signal spread in a network of eight cells responding to SFSP stimulation is shown. Ca²⁺ response from stimulated cell 1 is labeled in black. Representative Ca²⁺ responses from other cells were denoted with a gradient of colors from back to white with corresponding cell number (Inset). The arrow indicates time of SFSP stimulation. (Scale bar, 10 μm.)

Discussion

In this study, we demonstrate that low piconewton hydrodynamic forces can discretely trigger Ca²⁺ responses, and that this signal generation is polarized and requires firm attachment sites containing integrin. Similarly polarized electrophysiological responses were observed in SFSP-stimulated MLO-Y4 cells, which also required firm attachments (21, 22). However, the latency for Ca²⁺ response to SFSP stimulation observed in the present studies is three orders of magnitude longer than the electrical signaling response time reported in our previous study (21). The different response times may reflect different sensitivities of the two techniques: detection of [Ca²⁺]_i changes in long, slender osteocyte processes is limited by the Ca²⁺ indicator concentration, whereas even very small changes in cell conductance can be immediately observed (21).

The experimental studies reported here (Figs. 1–3) and previously (21, 22) validate the prediction that attached osteocyte processes are extremely powerful force transducers, whereas the cell processes lacking attachment sites and the cell body are not. Other supporting evidence for a role of cell processes in specialized mechanosensation and mechanotransduction includes the demonstration that dendritic processes of rat osteocytes are much more responsive than the cell body to mechanical deformations caused by precise displacements of coated microparticles (20). However, in that study, the applied force was predicted to be three orders of magnitude greater than the piconewton range predicted to occur in vivo and tested in the current studies.

It has been demonstrated that integrin attachments play a key role in modulating cation and stretch-activated channels (29, 30) in addition to their well-known role in activating focal adhesion kinase (FAK) and mitogen-activated protein kinase (MAPK) signaling pathways (14, 30). Moreover, it has been shown that both α₅β₁ and α_Vβ₃ integrins not only interact with focal adhesion complexes but also regulate L-type Ca²⁺ channels in brain and other excitable cells (31). In osteocytes, α_Vβ₃ integrin has been shown to regulate mechanosensitive cation channels (17), whereas flow-activated α₅β₁ or β₁ integrins have been shown to induce dye uptake, interpreted as opening of Cx43 hemichannels (32, 33), and the release of PGE₂ (33, 34). Here we demonstrate that inhibition of α_Vβ₃ integrin attachment sites compromises localized Ca²⁺ response to SFSP stimulation (Fig. 5). This finding of functional participation of integrins in osteocyte mechanosensation/transduction strongly validates the central role of α_Vβ₃ integrin in structural polarization of the osteocyte in response to flow-induced piconewton-level forces predicted to occur in vivo in the LCS. These findings are consistent with the mechanotransduction hypothesis formulated on the basis of structural evidence for integrin attachment of osteocyte processes to canalicular protrusions in situ (12, 13).

Our studies also provide evidence that ATP is a major autocrine/paracrine mechanosignaling messenger that amplifies spread of Ca²⁺ signals from cell process to cell body and from one osteocyte to another in the network (Figs. 4 and 6). As a consequence of the signal amplification that ATP provides, piconewton forces applied to discrete attachment sites of osteocyte processes can activate the entire cell and initiate intercellular signaling in the osteocyte network. Although previous studies have provided evidence that both ATP and gap junctions provide mechanisms for intercellular signaling throughout the network, stimulation has generally been achieved using fluid shear stress, mechanical stretch or indentation methods (4, 17, 20, 24, 26, 32). The fluid shear stimulus does not determine the site of signal initiation, whereas indentation methods produce forces much larger than those predicted to occur in vivo. Through the use of the SFSP stimulus, we have been able to demonstrate that osteocytes in vitro are highly polarized cells, where mechanosensory transduction resides at α_Vβ₃ integrin attachment sites along the process and that the excitation force can be as little as 1–5 pN.

Materials and Methods

Cell Culture.

MLO-Y4 (35) osteocytic cells (passages 18–21) were cultured in α-MEM (Invitrogen) supplemented with 5% (vol/vol) FBS (FBS, Invitrogen), 2.5% (vol/vol) bovine serum (BS, Invitrogen), and 1% (vol/vol) Penicillin–Streptomycin. Cells were plated at 100 cells per cm² seeding density for 7 d on collagen type I-coated (0.1 mg/mL, BD Biosciences) glass-bottom imaging dishes (MatTek Corporation).

Ca²⁺ Imaging and SFSP Stimulation.

Cells were loaded with the ratiometric Ca²⁺ indicator Fura-2 acetoxymethyl ester (10 μM, Invitrogen) at 37 °C for 45 min, then washed and maintained in phenol-free α-MEM media supplemented with 1% FBS and 20 mM Hepes (Sigma) throughout the experiment. Detailed Ca²⁺ imaging procedures are described in SI Materials and Methods. Protocol for stimulation of single sparsely plated cells with the SFSP was previously described (21) and is provided in SI Materials and Methods.

IntegriSense 750 Treatment.

Cells were treated with IntegriSense 750 (PerkinElmer), a fluorescent imaging agent that combines a selective antagonist for α_Vβ₃ integrin with a near-infrared (NIR) fluorochrome, at a concentration of 20 fmol/μL for 15, 30, and 45 min. The difference in spectral properties of IntegriSense 750 and Fura-2 allows the localization of α_Vβ₃ integrin in MLO-Y4 cells while imaging changes in intracellular Ca²⁺ in response to SFSP stimulation.

Apyrase Treatment.

To determine the involvement of ATP in the amplification of the Ca²⁺ signals generated by SFSP stimulation, the cells were first stimulated under control conditions in phenol-free α-MEM supplemented with 2% (vol/vol) FBS and 20 mM Hepes to establish baseline responses. Cells were then treated with apyrase (50 U/mL, Sigma), an ATP scavenger, for 10 min and stimulated again with the SFSP.

Image Analysis.

The properties of SFSP induced in [Ca²⁺]_i were analyzed using Excel (Microsoft Office), Prism 6 (GraphPad), and ImageJ [National Institutes of Health (NIH)] as described in SI Materials and Methods.

Statistical Analysis.

The detailed analysis is described in SI Materials and Methods.