To provide evidence of active accumulation of K+ in bone extracellular fluid (BECF), electric currents driven by damaged living metatarsal bones of weanling mice, immersed in physiological media at different [K+], in the presence of blockers of the K+ channels or of the Na+-K+-ATPase inhibitor, were measured by means of a voltage-sensitive two-dimensional vibrating probe. At 4 mM extracellular K+concentration ([K+]o), an inward steady current density (7.85–38.53 μA/cm2) was recorded at the damage site, which was significantly dependent on [K+]o. At [K+]o equal to that of BECF (25 mM), current density was reduced by 76%. At [K+]o of 0 mM, the current density showed an increase, which was hindered by tetraethylammonium (TEA). Basal current density was reduced significantly after exposure to TEA or BaCl2 and was unchanged after long- term exposure to ouabain. By changing control medium with a chloride-free medium, current density was reversed. The results support the view that K+ excess in bone is maintained by a biologically active cellular system. Because the osteocyte-bone lining cell syncytium was at the origin of the current in bone, it is likely that this system controls the ionic composition of BECF.
- ionic current
- bone extracellular fluid
bone has an extracellular fluid (BECF) that flows inside the osteocyte lacunae and the rich network of connecting canalicula. BECF has a distinctly different ionic composition from systemic extracellular fluid (ECF) and, in particular, a higher concentration of potassium (K+; 25 mM vs. 4 mM, respectively) (10, 25, 27, 35). The high content of K+ in BECF was demonstrated many years ago (10, 16, 25, 27, 35), but neither the mechanisms underlying K+ accumulation in bone nor the physiological relevance of the K+ excess has been explained. Bone K+ might participate in the maintenance of acid-base equilibrium, being rapidly removed from the bone in response to an acute acid load (7, 17). K+ concentrations might control bone resorption (9) and matrix mineralization. In fact, K+ reduces the incorporation of and alkali anions into octacalcium phosphate, a transient intermediate in the precipitation of the thermodynamically more stable hydroxyapatite (11), thus preventing the mineralization excess (31).
Passive physicochemical equilibrium (26) cannot explain the high K+ levels. Because K+ is not bound to any degree in the crystal hydration shells and does not interact significantly with any of the solid phase constituents of formed bone (10, 16), the elevated levels must be due to a metabolically active partition (35). Several lines of evidence support this view. 1)BECF K+ flows out of the bone surface after bone death (6).2) Reduced medium pH depletes surface K+ of cultured calvariae with respect to calcium (8). 3) The block of the Na+- K+-ATPase alters the transport of85Sr and 86Rb from BECF to ECF (22). 4)The potential difference between BECF and ECF is consistent with an energy-dependent accumulation of K+ within BECF (39, 40).5) Finally, bone cells are polarized (1, 17, 38).
Despite the clear presence of a pronounced K+ gradient, direct anatomic evidence for a barrier compartmentalizing the ionic composition of BECF from ECF is lacking. Recent observations, however, point to the existence of a bone-lining cell-osteocyte syncitium (29) that could perform such a function (17, 23, 30, 34). The presence of gap junctions linking cells (29) and ion transport structures in cells of osteoblast lineage (1, 20, 41) are in agreement with this view.
Classical electrophysiological methods have not been successfully applied to the characterization of ion fluxes at the BECF-ECF interface. The hard structure of bone hinders the placement of microelectrodes within the endocanalicular space. However, the noninvasive, voltage-sensitive, vibrating probe system bypasses this limitation because it measures that part of the ionic current loop flowing outside the bone in the extracellular milieu (28). The probe, therefore, can provide direct information on ion fluxes that occur through bones immersed in physiological media. It has already been observed that mechanically unloaded, living bone drives a steady ionic current through a puncture site penetrating the diaphyseal cortex (4,33, 34). Because the transcortical damage has the immediate effect of exposing the BECF to ECF, ions are free to move down their electrochemical gradients at the damaged site (leak), and the pumps, devoted to the maintenance of the gradient, are activated. The activation of a putative pump-leak system, first proposed by Borgens (4), generates a detectable electric signal at the damage site (leak), giving reliable information on the ionic exchanges occurring over the intact bone surface. Because of its stability over time (4, 33, 34), its temperature dependence (4), and its absence in dead bone (4, 33,34) and after solubilization of the cell membrane (34), the injury current was ascribed to a biologically active cell system. A further study gave the first convincing evidence that the osteocyte-lining cell system provides the driving force for the current (34). Because the current is dependent on specific ion species, such as sodium, chloride (4), and bicarbonate (33), which are known to display a concentration gradient between BECF and ECF (25, 27, 32), a specific ion transport mechanism at the BECF-ECF interface should be operative.
To explore the hypothesis that there is a partition system responsible for K+ fluxes at the BECF-ECF interface, we measured the electric current driven by living, mechanically unloaded, damaged bone. The bone response to changes in the physiological media mimicking ECF and containing different K+ concentrations and/or specific K+ channels and Na+- K+-ATPase blockers was tested.
MATERIALS AND METHODS
The medium containing ECF K+ concentration ([K+]o = 4 mM) was defined as the control medium, and media containing different K+concentrations ([K+]o = 0 to 27 mM), different Cl− concentrations ([Cl−]o = 0 to 50 mM), and/or drugs were defined as the test media (Tables1–3). Osmolarity was measured by an Osmostat Os 6020 pressure osmometer (Damchi, Kyoto, Japan). Resistivity was measured by a HI 9033 multi-range conductivity meter (PBI International, Milano, Italy). Reagents were purchased from Sigma (St. Louis, MO).
Weanling mice (Swiss; Charles River, Italy) were killed with CO2 in a gas chamber (Tecniplast, Varese, Italy). The back limbs were amputated at the distal tibia epiphysis and immersed in an excision medium with the composition of the control medium (Table 1) except that bicarbonate was substituted with Na+-isethionate to assure the stability of pH during surgical manipulations. The metatarsal bones were dissected intact from the digit, with care taken to avoid damage to the bone surface. All manipulations were carried out on samples immersed in the medium at room temperature with an M3 surgical microscope (Wild, Zurich, Switzerland). After the bone was freed of soft tissue ensheathments, a 50-μm-diameter hole, penetrating into the marrow cavity through the diaphyseal cortex, was made with a thin stainless steel dental drill (Mani, Matsutani Seisakusho, Ken, Japan). This involved a penetration to a depth of ∼200 μm. The animal use was approved by the local Institutional Animal Care and Use Committee (Protocol no. TS 9501).
Experimental setup and data acquisition.
The experimental setup has been previously described (33, 34). Briefly, all equipment (chamber, probe electrode, and microscope) was placed on an M-TS 23 anti-vibration platform (Newport, Fountain Valley, CA). The bone was held at the bottom of a specifically built Petri dish and placed in an aluminum container over a Peltier heating chip kept at 37°C. The dish was filled with the appropriate prewarmed (37°C) medium and was then covered by a light, white, mineral oil (Mineral Oil, Sigma). Medium stability was monitored by a pH-Po2-Pco2 automatic analyzer (Instrumentation Laboratory, Lexington, MA) on aliquots taken at intervals during readings. Temperature was monitored by a T801 thermoprobe (Radiometer, Copenhagen, Denmark). A video camera (TK S200, JVC, Tokyo, Japan) was connected through a BH Olympus microscope (Tokyo, Japan) to an RGB monitor (EUM1491A, Mitsubishi, Tokyo, Japan). The area including the vibrating electrode and the bone surface was viewed at an 18.23-mm working distance using a D-achromat A4 × 4 lens (Olympus, Tokyo, Japan). Light was provided through fiber-optic cables (Olympus) connected to a light source (Intralux 5000, Volpi, Zurich, Switzerland).
The two-dimensional vibrating probe system has been described in detail (15, 21, 28, 36, 37). Briefly, the reference electrode is a stationary platinum black wire, and the probe is an insulated stainless steel electrode (SS 300305A, Microprobe, Clarksburg, MD) with a platinum black tip. This electroplated ball has a final diameter of ∼3–5 μm (28). The probe is vibrated between two positions along thex- and y-axes by means of a π-shaped linkage of three rectangular piezoelectric bimorphs to which sinusoidal signals are applied. The probe is vibrated at frequencies of 200–300 Hz for the x-axis and 400–600 Hz for the y-axis over distances twice the diameter of the platinum black ball. The instrumentation measures the electrical potential differences between the extremes of the probe excursion, along the directions of vibration, with a lock-in amplifier. The signal is transferred via an analog-digital board (DAS8, Keithley Metrabyte, Taunton, MA) to a computer (PC CompaQ 3.86) for data acquisition, analysis, and storage. The probe position is recorded, and software analyzes the two orthogonal components of the measured signal, expressing them as an average vector (Software lock-in amplifiers and vibrational assemblies were developed at the National Vibrating Probe Facility, Marine Biological Laboratory, Woods Hole, MA). The vectors are overlaid on a digitized image of the preparation (PCVision Plus, Imaging Technology, Bedford, MA). Each vector length represents the density of the current at the measurement point, with directionality corresponding to the arrowhead. By convention, the direction of the current flow is depicted as normal to a net cation flux. If the current is carried by an anion, the polarity must be interpreted as the reverse of that depicted by the vector. The analog outputs of the system were also recorded on a four-channel chart recorder (BD101, Kipp & Zonen, Delft, Holland).
To calibrate the probe, a 50 nA current was delivered into the control medium through a glass micropipette (1 × 90 mm, GD-1, Narishige, Tokyo, Japan) obtained with a micropipette puller (PB-7, Narishige) and filled with 3 M KCl. The current source was placed (manipulator M-152, Narishige) at two locations 150 μm away from the probe, in mutually orthogonal directions. During experimentation when significant signals out of phase with the probe vibration (quadrature output) developed, the results were discarded because this usually indicated an artifact caused, for example, by contact with the tissue.
Typically, vibrating probes were calibrated each day. The background value was first measured by placing the probe far from the bone (>3 mm). Electrodes giving background values in excess of 0.5 μA/cm2 were rejected. The probe was moved using a 3D micromanipulator (MO-203, Narishige) to the recording site, ∼35–50 μm above the cortical hole, to map current densities. The final current density was obtained by subtracting the background value from the reading. The coefficient of variation, defined as SD/average ×100 of the measurements obtained when the location was kept constant, was 7%. Because the voltage probe technique is based on a version of Ohm's Law (21), the current-dependent voltage will be affected by the resistivity of the medium. The software corrects for differences in resistivity with experimental treatment. It assumes that the resistivity at the recording site is the same as the bulk medium; given the lissajous movement of the probe, and the mixing that results at a vibrational frequency of 200–400 Hz, this is a valid assumption. Resistivity was measured at 37°C with a conductivity meter (HI 9033 PBI International).
After testing for the spatial distribution of current density over the injury site, the probe was located at the point of maximal density. This point was generally found over the center of the injury. Two series of experiments were then performed. In the first series, the current magnitude was first recorded in control medium (basal steady current) and then in test media either with different K+concentrations (Table1) and/or in the presence of blockers for K+ channels, [TEA (35mM, 50mM; Table 3) and BaCl2 (2mM)] or in the presence of the Na+-K+-ATPase inhibitor ouabain (1 mM). In the second series, the current magnitude was first recorded in control medium (basal steady current) and then in test media with different Cl− concentrations (Table 2). The probe electrode was positioned at the same location before and after media substitutions.
All experiments were performed at a controlled temperature (37°C), pH (7.33 ± 0.06 at 37°C), and osmolarity (343 ± 11 mosM). Total elapsed time for the K+ concentration experiments was ∼1 h, whereas total elapsed time for the experiments with blockers was ∼2 h.
Data were compared by a two-tailed Student's t-test for paired observations. Data obtained from the experiments with different K+ concentrations were analyzed by means of ANOVA. Differences were considered significant for P < 0.05.
An electric (ionic) current entering the site of damage was recorded in all metatarsal bones tested (n = 98). Maximal current density vector was normal to the bone surface, whereas moving along the bone, current density decreased and current direction became progressively parallel to the bone longitudinal axis (Fig.1). The current appeared inward by the convention we have discussed. In control medium, maximal current density became steady after an initial slow decay. Control experiments confirmed that the current remained stable for ≥2 h (Fig.2, n = 4). At steady state, maximal current density averaged 20.16 μA/cm2 and ranged from 7.85 to 38.53 μA/cm2 (n = 89; basal steady current). When the control medium was exchanged with test media containing different K+ concentrations, current density changed. At [K+]o equal to (27 mM,n = 2; 25 mM n = 8, P < 0.01; Fig3, A and B) or lower than (22 mM, n = 4, P < 0.01; 20 mM, n = 4,P < 0.01; Fig. 3, C and D) that of BECF (25 mM), the current density decreased significantly after an initial, transient reversal. At 10 mM [K+]o,current density decreased without an initial reversal (n = 4,P < 0.01; Fig. 3 E), whereas at 0 mM [K+]o, the current density increased significantly (n = 7; P < 0.01; Fig.3 F) and remained steady for ≥70 min (data not shown). Current density returned to the basal steady values after exchange with control medium. Current density changes on media exchange, expressed as a ratio of current density measured in the test media at different K+ concentrations to that measured in control medium, were significantly (Fisher's F = 32.14; P < 0.0001; Fig. 4) related to the medium K+ concentration. By applying Ohm's law on the current density values for any given resistivity of the media at each different K+ concentration, the calculated potential differences were linearly related to the log [K+]o (Fig.5).
Potassium channel blockers affected current density. When basal injury current density was considered, the effect of K+ channels blockers was inhibitory. In fact, the addition of 50 mM TEA, 35 mM TEA, or 2 mM BaCl2 to the control medium significantly (P < 0.001) reduced basal steady current density by 60% (n = 5), 40% (n = 3), and 30% (n = 4), respectively. The reduction of the current density was reversible, because basal steady current fully recovered after drug removal (Fig.6, A–C). The inhibitory effect of TEA was different at different [K+]o. By increasing [K+]o to 20 mM, current density was reduced by 50% and was further reduced by 70% by the addition of 35 mM TEA (it did not matter when the drug was added) (Fig.7, A and B). When TEA was present in both control and test media, no increased current density was observed by removing K+ from the external medium, whereas when the TEA addition was coupled to the test medium with [K+]o = 0 mM, current density was reduced by 58% (Fig. 7, A and B).
When the bones were exposed to 1 mM ouabain for short times (∼30 min), there was an apparent decrease of current density (Fig.8 A, n = 4) that was not confirmed by extending the exposition time to over 90 min (Fig.8 B, n = 3). In fact, after a long exposition time that was chosen to assure the complete permeation of the Na+-K+-ATPase inhibitor into the bone, current density remained stable for ≥2 h (Fig. 8 B), as it did under control conditions (Fig. 2).
When the control medium was changed with test media containing different Cl−concentrations, current density changed. When [Cl−]o was lower than ECF (100 mM), the current density decreased in a nonconsistent manner. At [Cl−]o equal to 50 mM (n = 5) or lower (25 mM, n = 6; 10 mM, n = 6; 4 mM, n = 7), current decreased after a transient reversal that did not occur in all bones tested (Fig. 9,A–D). At 0 mM [Cl−]o, current density reversed in all bones (n = 5) (Fig. 9 E). By reconstituting the control conditions ([Cl−]o =100 mM), current density returned to the basal steady values after a transient increase. Current density changes on media exchange, expressed as the ratio of current density measured in the test media at different Cl− concentrations to that measured in control, were not significantly related to the medium Cl−concentration (data not shown).
This study confirmed previous observations (4, 33, 34) that damaged bone generates a steady electric (ionic) inward current at the damage site. According to the general interpretative model of electric currents at an injury site mapped with a vibrating probe (3), the damage of the cortex creates a “point sink” of the current, partially shorting out the potential difference between BECF and ECF at the site of damage. Thus ions are free to move along their electrochemical gradients through this low-resistance pathway. As a consequence, a driving force is activated to maintain the ionic composition of the BECF. The activation of this system generates the detectable electric current at the damage site with an inward direction. The injury current is sustained over time by a driving force provided by a cellular battery (4, 33, 34). The cell lineage that accomplishes the task of maintaining the current over time is the osteocyte-lining cell system (34) that appears to compartmentalize BECF from ECF (6, 17, 23, 24). At the intact bone surfaces, the fluxes of anions and cations exchanged between BECF and ECF are electrically neutral, and no current densities are detectable (4, 33, 34).
The critical components of the injury current are the ionic species for which a concentration gradient exists between the two extracellular ionic compartments, i.e., BECF and ECF (25). Ionic substitution experiments demonstrated in fact that chloride (4), sodium (4, 34), and bicarbonate (33) are carriers of the current. This study made progress in determining the ionic species carrying the current, and it was specifically addressed to K+ , for which the different composition between BECF and ECF has been a puzzle for many years (10, 25-27). By demonstrating that injury current varied by modifying the [K+] gradient between the BECF and ECF, this study gives convincing proof that an ion transport system is operative in bone to control the K+content of bone microenvironment. By also demonstrating that injury steady current was reduced after the bone exposure to specific K+channel blockers, this study suggests that K+channels should be involved in fine-tuning bone K+balance.
At the present state of knowledge, an interpretative model of ion fluxes between bone and plasma that could describe current density changes as a function of K+ concentration in the external medium could be only speculative. It is based on the following observations:
• a potential difference exists between the ECF and BECF; this potential difference is negative with respect to ECF (39); and the polarity of the potential difference is consistent with an active transport system for K+ into BECF (39), as postulated by Neuman (25, 27);
• the osteoblast appears to behave like an epithelial cell, with a Na+-K+-ATPase and Na+/H+exchanger at the basolateral membrane facing the systemic circulation, and Cl−/ and Cl− and K+ channels facing the mineral side; by allowing K+ and Cl− exit through the apical membrane to the BECF (17), these exchangers determine the alkalinization of the bone microenvironment, thus favoring the process of mineralization that is dependent on alkaline pH;
• although the cells of the osteogenic lineage constitute a functional syncytium that extends up to the vascular endothelium (29), they present gaps large enough to allow the passage of macromolecules (horseradish peroxidase, antimony compounds, and lanthanum nitrate) (19, 24); in particular, the lining cells that cover endosteal, periosteal, and haversian surfaces and that are considered to function as a selective barrier between BECF and ECF (17, 23) by sealing up the endocanalicular space, display paracellular spaces and could even retract in the presence of parathyroid hormone. Ions could therefore leave BECF through a paracellular pathway along a concentration gradient.
The model, as shown in Fig. 10, represents the cell of the osteogenic lineage, the distribution of the channels and transporters, and the source and direction of the measured extracellular net current. Intracellular currents were not considered in the model because they are beyond the scope of the present study.
According to the general distribution of electric currents at an injury site mapped with a vibrating probe (3), the model assumes that the damage tends to short out the potential difference at the damage site (the point sink of the current) and that the K+ current loop originates from the intact portions of bone where the K+ concentration gradient drives K+out from BECF through the paracellular spaces. The return pathway of the current loop into the bone occurs at the point sink where the electrical potential difference (39) between BECF and ECF has been shorted out by the damage. Because of the potential difference of BECF (negative) with respect to ECF, K+ is driven back into BECF, thus closing the K+ current loop. As far as the osteocyte-lining cell system is viable, the loop can be maintained, and the relative current can be measured (4, 33, 34). Current density was in fact increased by removing K+ from the bath, and it was reduced by placing the ECF K+ concentration equal or proximal to the BECF one. By removing K+ from the bath, the outward driving force for K+ exit at the intact portions of bone is amplified (origin of the current loop) with subsequent enhancement of the cation return pathway and associated current at the point sink. By placing external K+ concentration equal to the BECF one, the outward driving force for K+ exit at the intact portions of bone is reduced, with subsequent decrease of the return pathway and associated current at the point sink. Because current density was also reduced by blockers of the K+channels such as BaCl2 and TEA, K+ should leave the osteocyte-lining cell system and accumulate into the endocanalicular space and lacunae. Because of the polarity of the cells from the osteogenic lineage, the block of K+ channels facing the mineral side (BECF) prevents the transcellular exchange of K+ between BECF and ECF and the associated return pathway of the current at the point sink. When TEA was already present in control medium, the model assumes that K+ concentration in BECF is reduced, as well as the BECF-ECF concentration gradient for K+. K+ could in fact only leak from bone both at the damage site and through the paracellular spaces, as it is not transported back into the BECF by the cells because of the block of the K+ channels. By blocking the channels, it is likely that K+ accumulates in the cell, thus blocking the still undefined active transporters. Under this nonphysiological condition, BECF content of K+ would change, acquiring the same concentration as that of the infinite bath (ECF), because the bone tissue is unable to refill its subdiaphyseal cortex reservoir of K+. The reduction of the K+ content into the bone under TEA exposure could therefore explain why no increased current density was observed when K+ was removed from the bath. Of course, when TEA was already present in control medium, no further reduction of the basal injury current could be observed.
This study did not succeed in demonstrating any involvement of the ouabain-sensitive Na+-K+-ATPase. The small decay in current density observed after brief exposure could be aspecific, as it was not confirmed by the long exposure to the inhibitor. It is therefore likely that the Na+-K+-ATPase pump is not part of the homeostatic mechanism responsible for the generation of the ionic gradient between BECF and ECF, as early ouabain experiments (35) as well as the more recent ones demonstrated (40).
The proposed model is confirmed by Cl− substitution experiments. According to the model just reported, the reversal of the injury current in Cl− free medium is due to magnification of the Cl− concentration gradient between BECF and ECF that drives Cl− out from BECF through the paracellular spaces at the intact surfaces. The reversal of the polarity of BECF with respect to ECF could drive Cl− back into the endocanalicular fluid compartment at the point sink, thus closing the Cl− current loop (return pathway). The prevailing contribution of the Cl− current loop leads to the observed net current reversal at the point sink, because the direction of current flow is defined, by convention, as the direction in which positive ions move (see Experimental set-up and data acquisition). Because current density was not linearly related to Cl−concentration, it is likely that several transporters that are devoted to the maintenance of the physiological ionic environment of the endocanalicular space are activated.
The meaning of the current in bone.
The ionic current described in this study could be involved in the site-directed bone remodeling and repair processes. In fact, it can be speculated that a point sink of current could be induced by the osteoclast resorption of the bone matrix as well as by the damage of the bone cortex. In intact bone, it can be hypothesized that osteoclast activity induces a microinjury current at the resorption lacuna by exposing BECF to ECF. The current would be sustained by the osteocyte-lining cell system of the surrounding bone until the opening of the endocanalicular space to the plasma ceased to be sealed by the subsequent osteoblast matrix deposition. Under the influence of specific ionic gradients associated with the current, a directed migration of the osteoblasts might occur. In fact, osteoblasts exhibit cathodal galvanotaxis in vitro by migrating along the axis of an electrical field produced by an exogenously applied constant current of the same order of magnitude as that endogenously generated (14). Moreover, by exposing BECF to ECF in the resorption lacunae, osteoclast activity would locally reduce the K+ content. It is therefore likely that the drop of external K+ would cause an increment of the membrane potential difference of the osteocyte-lining cell system that would activate Cl−channels and induce a consequent Cl− efflux from the cell cytoplasm. Cl− exit should in turn increase in the lacunae and stimulate mineralization. In damaged bone, the phenomena described above are most likely reproduced in a larger scale.
This study confirms that a complex ion transport system is operative in bone to control the ionic composition of the bone microenvironment. This study provides evidence that K+ is accumulated in BECF and that K+ channels are involved in fine-tuning bone K+ balance. Although a working model of transcellular ionic movement in bone is, as yet, incomplete and speculative, this study strengthens the view that bone is an active ion-exchanging system that participates in mineral homeostasis and acid-base equilibrium maintenance (2). The endogenous generation of an ionic current in bone as soon as BECF is exposed to ECF by both osteoclast activity and damage could be part of the mechanisms of site-directed bone remodeling and repair, respectively.
We thank Peter J. S. Smith, Biocurrent Research Center, Marine Biological Laboratory (Woods Hole, MA) for criticism and editorial comments. We are greatly indebted to Gastone Marotti, Department of Anatomy, University of Modena, and to Antonio Zaza, Department of Physiology, University of Milano, for helpful discussion and scientific support in the development of the interpretative model of ion fluxes in bone.
Address for reprint requests and other correspondence: A. Rubinacci, Bone Metabolic Unit, Scientific Institute H, San Raffaele,Via Olgettina 60, 20132 Milano, Italy (E-mail:).
This work was supported in part (40%) by the Italian Ministry of University and Scientific Research MURST.
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. §1734 solely to indicate this fact.
- Copyright © 2000 the American Physiological Society