Part of this work was presented at the Bone and Tooth Society, Manchester, U.K., April 1996.
Bone cells respond to mechanical stimuli, but the transduction mechanisms responsible are not fully understood. Integrins, a family of heterodimeric transmembrane glycoproteins, which link components of the extracellular matrix with the actin cytoskeleton, have been implicated as mechanoreceptors. We have assessed the roles of integrins in the transduction of cyclical mechanical stimuli to human bone cells (HBCs), which results in changes in membrane potential. HBC showed membrane depolarization following 0.104 Hz mechanical stimulation and membrane hyperpolarization following stimulation at 0.33 Hz. The membrane depolarization response involved tetrodotoxin-sensitive sodium channels and could be inhibited by antibodies against αV, β1, and β5 integrins. In contrast, the hyperpolarization response was inhibited by gadolinium and antibodies to the integrin-associated protein (CD47), α5 and β1 integrin. Both responses could be abrogated by Arg-Gly-Asp (RGD)-containing peptides, inhibition of tyrosine kinase activity, and disruption of the cytoskeleton. These results demonstrate differential electrophysiological responses of HBC to different frequencies of mechanical strain. Furthermore, they suggest that integrins act as HBC mechanoreceptors with distinct signaling pathways being activated by different frequencies of mechanical stimuli.
Noninvasive loading of a variety of bones in a number of animal species results in strain-related modeling responses.(1) In vitro and in vivo studies have shown that cyclical loading is associated with a variety of physiological and biological responses in bone cells. These include opening of membrane ion channels and changes in membrane potential,(2) alteration in proliferative activity,(3) production of a variety of molecules of importance in the regulation of bone formation and turnover including type I collagen,(4) prostaglandins,(5) nitric oxide,(6) and cytokines.(7) However, at the present time the transduction mechanisms responsible for changes in bone cell metabolism following mechanical stimulation are not fully understood.
Advances in the understanding of the cellular anatomy of specialized force receptors suggest that mechanical forces are transmitted to cells via the extracellular matrix (ECM).(8) Recent evidence has suggested that ECM-cell surface receptors, in particular integrins,(9) may act as mechanoreceptors.(10) Integrins, a family of heterodimeric (α and β chain) transmembrane glycoproteins which form specific receptors for a variety of ECM proteins(9) are expressed by cells in all tissues including bone.(11) Data from in vivo and cell culture systems demonstrate that β1 integrin cell adhesion molecules transduce signals that regulate cell function.(9) As well as acting as adhesive molecules, integrins have been shown to act as signaling molecules.(12) Thus, by inducing either cytoskeletal organization or via activation of second messenger systems, integrin-mediated cell ECM can influence gene expression and control cell growth. Endothelial cell integrins have been shown to act as mechanoreceptors supporting a force-dependent stiffening response,(13) but virtually nothing is known of the ability of integrins to transduce mechanical stimuli in other cell types. We have therefore undertaken a series of experiments to investigate the role of integrin molecules in the transduction of mechanical stimuli that results in changes in the membrane potential of bone cells.
MATERIALS AND METHODS
Human bone cell culture
Human bone cell cultures were established from trabecular bone explants as described previously.(14) Bone was collected at surgery from children under the age of 12 years undergoing corrective osteotomy. Trabecular bone was cut into small pieces, washed in phosphate buffered saline (PBS; pH 7.4), and cultured in minimal essential medium (MEM; Gibco BRL, Paisley, U.K.) containing 10% fetal calf serum (FCS; Gibco), 2 mM glutamine, and antibiotics (0.1 mg/ml streptomycin, 100 U/ml penicillin, 2.5 μg/ml fungizone) in a humidified 5% CO2 atmosphere at 37°C. Cultures were initially grown in petri dishes, and once established were passaged into tissue culture flasks. Cell populations grown from the bone showed osteoblast-like characteristics (type I collagen, osteocalcin synthesis, and high alkaline phosphatase activity increased by 1,25-dihydroxyvitamin D3 [1,25(OH)2D3] stimulation) at the passages used (1st–5th).
Induction of cyclical mechanical strain
The technique used for inducing strain in cells in monolayer was the same as that which has been previously described in detail.(15,16) Flexible, plastic 55 mm tissue culture dishes (Nunc, Naperville, IL, U.S.A.) were placed in a sealed chamber with inlet and outlet ports. The culture dish was supported on six horizontal pins in a brass cylinder attached to the base of the chamber. A tight seal was present between the top of the brass cylinder and the culture dish. A space of volume 8 ml was present between the base of the culture dish and the base of the chamber. Gas entered this space from the main chamber (volume 89 ml) via 18 holes each 2.2 mm in diameter in the brass cylinder. The chamber was pressurized with nitrogen gas from a cylinder, the frequency being dictated by an electronic timer controlling the inlet and outlet valves. Cyclical pressurization of this system induces deformation and strain on the base of the plastic tissue culture dish and its adherent cells(16) due to a differential rate of rise of pressure above and below the culture dish. Under these experimental conditions, adherent cells are subjected to 16 kPa above atmospheric pressure. Previous studies in which cells have been subjected to this increased pressure without associated deformation of the dishes to which they are adherent have shown conclusively that this degree of pressure itself has no effect on membrane potential and that the electrophysiological response is wholly dependent on deformation of the dish base and adherent cells.(16)
Strain on the base dishes was measured using a mild steel rosette guage (Radiospares, Corby, Northants, U.K.) and a purpose-built linear DC amplifier whose output was fed into a digital display calibrated in microstrains.(16) The strain, produced by the 16 kPa, was 15 microstrain (μstr).(16)
Membrane potentials of bone cells were recorded using a single electrode bridge circuit and calibrator as previously described in detail.(15,16) Microelectrodes having tip resistances of between 40 and 60 MΩ and tip potentials of approximately 3 mV were used to impale the cells. Results from a cell were accepted if on impalement there was a rapid change in voltage to the membrane potential level of the cell and this voltage remained constant for at least 60 s.
Experiments were performed using subconfluent cultures of human bone cells. The day before experiments were carried out, media containing FCS was replaced by FCS-free media. In each set of experiments, a control dish of bone cells was examined whose culture medium did not contain any agent to be tested. Prior to mechanical stimulation, membrane potentials of 5–10 cells were measured. Bone cells were then subjected to the cyclical strain for 20 minutes, following which membrane potentials were recorded on a further 5–10 cells.
To investigate possible ion channels involved and roles for integrins in the transduction processes, a panel of pharmacological inhibitors of channel ion activity, tyrosine phosphorylation, cytoskeletal integrity, RGD-containing peptides, and antibodies against integrin subunits were used. Inhibitors of ion channel activity included: tetrodotoxin, which blocks Na+ ion channels(17); apamin and quinidine, which block Ca-activated K channels(18,19); gadolinium, which blocks stretch activated ion channels(20); and Bric 126 (IBGRL, Birmingham, U.K.), an antibody against the integrin associated protein (CD47) believed to be important for integrin-activated Ca fluxes.(21,22) Genistein(23) and cytochalsin D(24) were used to inhibit tyrosine protein kinase activity and disrupt the actin cytoskeleton, respectively. GRGDSP and control peptide GRADSP were obtained from Gibco. Fully characterized anti-integrin antibodies used were gifts or obtained from commercial sources. They included: P4C10 (anti-β1, Gibco), TS2/7 (anti-α1, Dr. F. Sanchez-Madrid), Gi9 (anti-α2, Immunotech/Coulter Electronics Ltd., Luton, Bedfordshire, U.K.), M KID2 (anti-α3, Immunotech/Coulter), CLB-705 (anti-α5, Chemicon, Temecula, CA, U.S.A.), AMF-7 (anti-αV, Immunotech), 23C6 (anti-αVβ3, Prof. M. Horton), and PIF6 (anti-αVβ5, Gibco). All antibodies recognize epitopes in extracellular domains. All, except TS2/7, have been shown to have function-blocking activity.
Membrane potentials of at least 5–10 cells were recorded prior to and 10 minutes after the addition of the agent being assessed to the culture medium. The cells were then subjected to the standard regimes of cyclical strain, following which the membrane potentials of at least another 5–10 cells were measured, with the agent under investigation being present in the culture medium throughout. For the experiments involving RGD-containing peptides, anti-integrin, and anti-CD47 antibodies, the experimental protocol was similar, with the exception that prior to the second series of measurements of membrane potentials, cells were incubated in the presence of peptide or antibody for 30 minutes. Different dishes were used for each reagent tested. At least two experiments with different cells on different days were performed with each reagent.
Electrophysiological response of human bone cells to cyclical strain
Human bone cells were subjected to cyclical strain over a range of frequencies and their membrane potential measured. The frequencies were altered by extending the quiescent phase but keeping the 2 s pressure pulse constant. The results (Table 1) show that the electrophysiological response varied with the frequency (and hence length of the quiescent phase). At 0.104 and 0.125 Hz, cells showed a depolarization response. In contrast, at 0.21 and 0.33 Hz, human bone cells showed a hyperpolarization response. There were no significant changes in membrane potential when cells were stimulated at 0.146 Hz.
Table Table 1. Effect of Frequency on Human Bone Cell Response to Cyclical Strain
Two cycle frequencies of 0.104 Hz (2 s strain, 7.4 s no strain) and 0.33 Hz (2 s strain, 1 s no strain) were taken as the standard regimes for further study. At these frequencies, the waveforms of the cyclical strain differ only in the length of the quiescent phase between pulses. The rate of change of strain, peak strain, and rate of decrease of strain over the 2 s pressure pulse at both frequencies are virtually identical (Fig. 1).
Effect of ion channel blockade on the electrophysiological response of human bone cells to cyclical strain
Following cyclical strain at 0.104 Hz, cells showed a membrane depolarization response (Table 2). In contrast, when stimulated at 0.33 Hz, bone cells show a hyperpolarization response. The degree of response to both frequencies of stimuli was similar between experiments using bone cells from individuals at different ages and at different passage stages. Gadolinium, apamin, quinidine, and tetrodotoxin and antibodies to CD47 (integrin-associated protein) were used in an attempt to establish the probable membrane ion channels involved in the production of the changes in membrane potential (Table 2). None of these agents had effects on resting membrane potential. Ten-micromolar gadolinium, an inhibitor of stretch-activated ion channels, completely abolished the hyperpolarization response to 0.33 Hz strain but had no effect on the depolarization response to 0.104 Hz strain. Similarly, both quinidine and apamin, which block Ca2+-activated K+ channels, inhibited the hyperpolarization response to 0.33 Hz cyclical strain. Bric 126 (anti-CD47) also inhibited the hyperpolarization response but had no effect on the response of bone cells at 0.104 Hz stimulation. In contrast, tetrodotoxin inhibited the membrane depolarization following 0.104 Hz stimulation, suggesting a role for Na+ channels in this response but had no effect on membrane hyperpolarization at 0.33 Hz stimulation.
Table Table 2. The Effect of Ion Channel Blockade on Human Bone Cell Membrane Potential Response to Cyclical Strain
Effects of RGD peptides
A number of integrins recognize and interact with extracellular matrix molecules via RGD sequences. Experiments were therefore carried out to see whether RGD containing oligopeptides known to interfere with integrin-mediated cell binding influenced human bone cell response to cyclical strain. The results are shown in Table 3. At a concentration of 100 μg/ml, GRGDSP inhibited both the depolarization response to 0.104 Hz cyclical strain and the hyperpolarization response to 0.33 Hz cyclical strain, suggesting that RGD-mediated interactions are involved in the transduction of strain to bone cells at both frequencies. The control peptide GRADSP had no effect on responses.
Table Table 3. The Effect of RGD-Containing Peptides on Human Bone Cell Membrane Potential Response to Cyclical Strain
Effect of anti-integrin antibodies
To examine the role of specific integrin molecules in the transduction process, a series of experiments were undertaken using antibodies against integrin subunits known to be expressed by human bone cells.(11,25,26) The results are shown in Table 4. Antibodies had no effect on resting membrane potential but showed different effects on the bone cell membrane potential response to different frequencies of strain. P4C10, an antibody against β1 integrin, inhibited both the depolarization response to 0.104 Hz stimulation and the hyperpolarization response to 0.33 Hz stimulation. P1F6 (anti-αVβ5) reduced the response at 0.104 Hz but had no effect on the 0.33 Hz response. 23C6 (anti-αVβ3) had no effect on the membrane potential response at either frequency of stimulation. Of the antibodies used against α subunits expressed by bone cells, only two showed significant effects. The anti-αV antibody, AMF-7, inhibited the response to 0.104 Hz but had no effect on the 0.33 Hz hyperpolarization response. In contrast, CLB-705, an antibody against α5 integrin, inhibited the 0.33 Hz hyperpolarization response but had no effect on the 0.104 Hz depolarization response. Antibodies to α1, α2, and α3 integrins did not inhibit the membrane potential responses of bone cells to either frequency of cyclical strain.
Table Table 4. The Effect of Anti-integrin Antibodies (1μg/ml) on Human Bone Cell Membrane Potential Response to Cyclical Strain
Effect of genistein and cytochalasin D on bone cell responses to cyclical strain
Integrin signaling involves the actin cytoskeleton and tyrosine phosphorylation events.(12) Experiments were therefore undertaken to see whether the inhibition of tyrosine phosphorylation and/or disruption of the actin cytoskeleton affected the response of bone cells to mechanical strain. The results are shown in Table 5. Genistein, at a concentration of 40 μM, and cytochalasin D, at a concentration of 1 μM, prevented the membrane potential responses of human bone cells at both 0.33 and 0.104 Hz cyclical strain.
Table Table 5. The Effect of Disruption of the Actin Cytoskeleton and Inhibition of Tyrosine Kinase Activity on Human Bone Cell Membrane Potential Response to Cyclical Strain
We have shown that human bone cells show changes in membrane potential in response to cyclical mechanical strain and that the response varies with different frequencies of stimulus. This relatively early response of bone cells, being seen within 20 minutes of applied stimulation, has been used to study the signaling pathways involved in the transduction of cyclical mechanical stimuli to human bone cells in vitro. The results suggest that integrin-dependent pathways are involved in the transduction of mechanical stimuli to produce changes in membrane potential at both frequencies. β1 and α5 integrins are involved in the transduction of strain at 0.33 Hz, which results in a hyperpolarization response via a Ca2+-activated K+ channel. In contrast, the depolarization response mediated by a tetrodotoxin-sensitive Na+ channel(17) seen at 0.104 Hz involves αV, β1, and αVβ5 integrins.
Using in vitro systems similar to that described here, several groups of workers have found a link between mechanical deformation of bone cells and cell metabolism. Somjen et al.(27) showed a rise in both cyclic adenosine monophosphate (cAMP) and prostaglandin E2 (PGE2) synthesis, which reached a maximum after 15–20 minutes of deformation. Subsequently, others have shown increases in ornithine decarboxylase and Na+, K+ ATPase activity,(28) thymidine incorporation,(29) and in the synthesis of collagenous and noncollagenous protein.(30) Distortion of the plasma membrane is thought to be important in transduction of the mechanical strain, but the pathways involved have not been elucidated.
Integrins have been shown to act as mechanoreceptors for endothelial cells,(13,31) and it is likely that many other mechanosensitive cells will use a similar mechanism to recognize and transmit mechanical signals. Using RGD and anti-β1 integrin antibody-coated magnetic microbeads, Wang and coworkers(13) have shown that mechanical stresses applied to capillary endothelial cells result in focal adhesion formation and an increase in cytoskeletal stiffness, suggesting that integrins may transmit mechanical signals via the cytoskeleton. Cytoskeletal changes may be transduced directly into a chemical response by activating stretch-activated membrane channels on the cell surface.(32) Our results support such a view and suggest that RGD-mediated cell–matrix interactions are involved in bone cell mechanotransduction. RGD-dependent integrin binding has been shown for a number of ECM proteins, such as fibronectin, osteopontin, thrombospondin, bone sialoprotein, type I collagen, and vitronectin,(9,26,33,34) and may be involved in osteoblast adhesion and transduction of mechanical signals. Of integrins expressed by bone cells(11,25,26) it appears that α5β1 integrin, the classical fibronectin receptor, αVβ5, which has been shown to act as a receptor for vitronectin, fibronectin, and osteopontin,(35,36) and potentially another β1 integrin, may act as bone cell mechanoreceptors. Furthermore, the membrane responses elicited by the two frequencies of cyclical strain used in our studies appear to be transduced via different integrin molecules.
Integrins act as a transmembrane link among the ECM, cytoskeletal proteins, and actin filaments. Mechanical stimuli may activate intracellular signaling as a result of deformation of the ECM and induce changes in the conformation or clustering of integrins following mechanical stimulation.(10,32,37) Integrin signaling involves a complex series of events including organization of the actin cytoskeleton, localization, and phosphorylation of associated signaling molecules, including pp125FAK,(38) paxillin,(39,40) tensin, and actin binding protein(41) and stimulation of signaling pathways, including phosphotidylinositol metabolism.(12) The roles of integrins in the opening of ion channels necessary for changes in membrane potential as a response of mechanical strain remain to be elucidated fully. Our results suggest that both genistein-inhibited (tyrosine) phosphorylation events and an intact cytoskeleton are required for both hyperpolarization and depolarization responses. There is a recognized variation in the effects of cytoskeletal disruption and inhibition of tyrosine kinase activity on the accumulation of intracellular signaling molecules following integrin activation.(12) Tyrosine kinase–mediated phosphorylation and actin cytoskeletal integrity are required for integrin-induced accumulation of some cytoskeletal proteins, including F-actin, paxillin, and filamin, but not others,(42) raising the possibility of a critical role for these molecules in the mechanotransduction by integrins in bone cells. Indeed phosphorylation of paxillin but not pp125FAK appears to be involved in the intracellular mechanotransduction of blood flow forces in endothelial cells.(31) Integrin signaling, however, involves a diverse range of tyrosine kinase signaling cascades,(12) and the detailed nature of the intracellular pathways activated by mechanical stresses in bone cells remains to be defined.
The results with anti-integrin antibodies, however, suggest that the changes in membrane potential as a result of the different stimuli used in our study involve different integrins. Evidence is accumulating for distinctive types of signaling via different integrins and for the regulation of function by one integrin of another.(43) Furthermore, responses to integrin activation, including calcium influx, appear to be dependent on integrin, ligand, and cell type,(12) suggesting that integrin expression represents specific cellular requirements for specific signals as well as ligand binding. For instance, αV integrins but not α5β1 mediate a rise of intracellular calcium when endothelial cells spread on fibronectin, although they have only a minor role in cell adhesion.(44) Integrins appear to play an important, but as yet incompletely understood, role in bone formation.(45,46) Signaling of mechanical stimuli to bone cells via different integrins may involve specific phosphorylation events(47) and initiation of diverse metabolic pathways of potential importance for activation of bone modeling or remodeling cycles.
Osteoblast-like cells contain a number of mechanosensitive ion channels, nonselective cation, and K+-selective channels, which could cause membrane hyperpolarization or depolarization in response to mechanical stimulation.(48–50) In addition, a gadolinium-sensitive stretch-activated ion channel has been identified that may be modulated by parathyroid hormone and chronic, intermittent load-bearing,(51) suggesting a role for this channel in the regulation of bone response to physical strain. The hyperpolarization response to cyclical strain appears to involve the opening of gadolinium-sensitive stretch-activated ion channels,(16) and our results are consistent with a role for a gadolinium-sensitive ion channel in human bone cell responses to mechanical strain. The nature of this channel remains to be elucidated, however, because gadolinium, as well as being able to block stretch activated ion channels in a variety of cell types, has been shown to block voltage-activated ion currents in cardiac and skeletal muscle.(52)
Spreading endothelial cells on fibronectin triggers the elevation of intracellular-free calcium from the extracellular space via integrin-associated voltage-independent membrane channels.(53,54) Consistent with the results of our studies, soluble antibodies against integrin molecules and RGD peptides have no effect on the Ca flux in resting endothelial cells,(53) suggesting that the cytoskeleton plays a role in the opening of calcium channels in the plasma membrane. Furthermore, activation of the calcium signaling pathway resides primarily in the integrin α subunit.(44) Calcium influx associated with endothelial cell spreading requires a 50 kD integrin-associated protein (IAP), recently identified as CD47.(21,22) Evidence exists for CD47 being an integrin-associated, transmembrane voltage-independent ion channel.(53,54) Inhibition of the hyperpolarization response of bone cells to a 0.33 Hz cyclical strain with the anti-CD47 antibody Bric 126 would support a role for integrin-regulated calcium entry in this response to mechanical strain. Further work is necessary to establish the relationship between different integrins expressed by bone cells, CD47, and gadolinium stretch-activated ion channels.
The physiological significance of changes in bone cell membrane potential to mechanical strain are still unclear. The strains and strain rates used in our study are much lower than those measured in vivo at the surface of bone. In bone, fracture occurs at 25,000 μstr, although normal maximum strain is only 3000, and maintenance strain is 300 μstr.(55) It is unknown, however, how much of this strain is transmitted to bone cells, and it is likely that much of these forces do not directly reach the cells. In vitro, where cells are relatively unprotected, application of smaller strains directly appear to be adequate to elicit physiological responses. Patch clamped osteoblasts respond at 65 μstr.(55)
The frequencies of stimuli we have used, although producing opposite effects on membrane potential, are relatively similar. The results, however, suggest that human bone cells, at least in vitro, can rapidly identify the form of the mechanical stimulus and respond electrophysiologically in different ways. The frequencies of strain used in our study are less than that of normal walking, which ranges from 0.68 to 1.06 Hz.(56) Physiological strain frequencies will vary considerably with activity throughout the day, and cells in vivo are likely to be subjected to, and potentially respond to, a wide range of frequencies and waveforms of mechanical stimuli. Changes in membrane potential may result in a cell becoming refractory to further stimulation.(57) It is possible that in vivo bone cells respond rapidly to a frequency or waveform of mechanical stimulus and then become refractory to that or other forms of mechanical stimulation for a period of time. Indeed Lanyon and colleagues have shown that only extremely short periods of dynamic loading are sufficient to reduce bone resorption, increase bone formation, and activate quiescent periosteum.(58–60) Furthermore, the osteogenic stimulus saturates rapidly, and additional stimuli have no significant additional osteogenic responses.(58)
Studies from our group have shown that human chondrocytes and fibroblasts also show changes of membrane potential in response to cyclical strain.(15,16) Fibroblasts, however, unlike bone and cartilage cells, show membrane depolarization following stimulation at 0.33 Hz, suggesting cell type–specific responses to mechanical strain. The hyperpolarization response is a result of activation of small conductance Ca2+-activated K+ channels,(15,16) leading to efflux of K+ ions from the cell and cell hyperpolarization. Endothelial cells also undergo membrane hyperpolarization in response to mechanical deformation in the form of flow shear stresses as a result of opening of quinine and apamin sensitive to small conductance Ca2+-activated K+ channels.(61) Hyperpolarization of endothelial cells, as well as regulating further calcium influx and a variety of second messenger signaling systems, also leads to the release of nitric oxide.(62) The opposite effect, depolarization, attenuates cellular functions, such as nitric oxide release, which rely on calcium influx.(57,62) It is possible that similar pathways are involved in the induction of bone cell nitric oxide release by mechanical strain in vitro and in vivo.
We thank those individuals listed in the Materials and Methods section for kind gifts of antibodies. This work was supported by a grant from the Sir Stanley and Lady Davidson Fund of Edinburgh University.