Effect of Metabolic Acidosis on the Potassium Content of Bone


  • David A. Bushinsky,

    Corresponding author
    1. Nephrology Unit, Department of Medicine, University of Rochester School of Medicine, Rochester, New York, U.S.A.
    • Prof. of Medicine and of Pharmacology and Physiology, University of Rochester School of Medicine and Dentistry, Chief, Nephrology Unit, Strong Memorial Hospital, 601 Elmwood Avenue, Box 675, Rochester, NY 14642 U.S.A.
    Search for more papers by this author
  • Konstantin Gavrilov,

    1. Enrico Fermi Institute, Department of Physics, University of Chicago, Chicago, Illinois, U.S.A.
    Search for more papers by this author
  • Jan M. Chabala,

    1. Enrico Fermi Institute, Department of Physics, University of Chicago, Chicago, Illinois, U.S.A.
    Search for more papers by this author
  • John D. B. Featherstone,

    1. Department of Restorative Dentistry, University of California at San Francisco, San Francisco, California, U.S.A.
    Search for more papers by this author
  • Riccardo Levi-Setti

    1. Enrico Fermi Institute, Department of Physics, University of Chicago, Chicago, Illinois, U.S.A.
    Search for more papers by this author


Metabolic acidosis induces resorption of cultured bone, resulting in a net efflux of calcium (Ca) from the bone and an apparent loss of mineral potassium (K). However, in these organ cultures, there is diffusion of K between the medium and the crystal lattice, causing difficulty in interpretation of the acid-induced changes in mineral ion composition. To determine the effects of acidosis on bone mineral K, we injected 4-day-old neonatal mice with pure stable isotope41K, equal to ∼5% of their total body K. Calvariae were dissected 24 h later and then cultured for 24 h in medium without added41K, either at pH ∼7.4 (Ctl) or at pH ∼7.1 (Ac), with or without the osteoclastic inhibitor calcitonin (3 × 10−9 M, CT). The bone isotopic ion content was determined with a high-resolution scanning ion microprobe utilizing secondary ion mass spectrometry.41K is present in nature at 6.7% of total K. The injected41K raised the ratio of bone41K/(39K+41K) to 9.8 ± 0.5% on the surface (ratios of counts per second of detected secondary ions, mean ±95% confidence interval) but did not alter the ratio in the interior (6.9 ± 0.4%), indicating biological incorporation of the41K into the mineral surface. The ratios of41K/40Ca on the surface of Ctl calvariae was 14.4 ± 1.2, indicating that bone mineral surface is rich in K compared with Ca. Compared with Ctl, Ac caused a marked increase in the net Ca efflux from bone that was blocked by CT. Ac also induced a marked fall in the ratio of41K/40Ca on the surface of the calvariae (4.3 ± 0.5, p < 0.01 vs. Ctl), which was partially blocked by CT (8.2 ± 0.9, p < 0.01 vs. Ctl and vs. Ac), indicating that Ac causes a greater release of bone mineral K than Ca which is partially blocked by CT. Thus, bone mineral surface is rich in K relative to Ca, acidosis induces a greater release of surface mineral K than Ca, and osteoclastic function is necessary to support the enriched levels of surface mineral K in the presence of acidosis.


IN MAN AND OTHER MAMMALS, metabolic acidosis leads to an increase in urine calcium (Ca) excretion without significantly increasing intestinal Ca absorption, resulting in a net loss of body Ca.1–3 Because the vast majority of total body Ca is located within the mineral stores of bone, the negative Ca balance implies depletion of bone mineral.4 In vivo studies have shown that metabolic acidosis, induced by ammonium chloride, leads to a loss of bone mineral.5

An in vitro model of metabolic acidosis, produced by a decrement in medium [HCO3], induces a marked efflux of Ca from cultured neonatal mouse calvariae6–11 while metabolic alkalosis induces an influx of Ca into bone.12 During short-term, 3 h cultures, the acid-induced Ca efflux appears to be secondary to physiochemical bone mineral dissolution.6,7 However, over longer time periods, >24 h, the Ca efflux from bone appears to be predominantly secondary to cell-mediated bone resorption.8–11 We have shown that metabolic, but not respiratory, acidosis leads to an increase in osteoclastic β-glucuronidase activity and a decrease in osteoblastic collagen synthesis.9,11,13

The surface of bone mineral appears rich in potassium (K) and sodium (Na) relative to Ca.6,8,14–19 The surplus of K appears to be maintained by a biologically active “bone membrane” which effectively separates the mineral from the extracellular fluid.17 When bone cells are killed, the excess K rapidly leaves the bone mineral surface in conjunction with a marked influx of Ca.17 We have shown that with metabolic acidosis the release of Ca appears to be associated with a greater release of mineral K.15,19 However, medium K and Na may rapidly exchange between the medium and the crystal lattice, making interpretation of the ion fluxes induced by acidosis difficult.20

Understanding ion fluxes relative to bone is an important part of understanding the effects of metabolic acidosis on the bone mineral. Clinically, metabolic acidosis is observed during renal insufficiency and failure and may be a contributing factor in renal osteodystrophy.21 Endogenous acid production has been associated with bone turnover and Ca loss in elderly women5; K bicarbonate therapy to neutralize this acid production appears beneficial in reducing bone turnover and Ca loss.22

To determine the effects of acidosis on mineral K content, we labeled neonatal mouse calvariae in vivo with the stable isotope41K.41K is present in nature at 6.7% of the most prevalent stable isotope39K.23 We then cultured the calvariae in control and acidic medium without additional41K. We found that we were able to enrich substantially the mineral surface with41K, that metabolic acidosis results in a greater loss of K relative to Ca, and that osteoclastic function is necessary to support the high levels of surface mineral K.


In vivo41K labeling

To determine total body K content, neonatal (4- to 6-day-old) CD-1 mice were weighed, killed, and dissolved in a 35% nitric acid and 35% perchloric acid (by volume) solution for 24 h. The K content of 15 mice (3.65 ± 0.06 g of body weight [bw]) was found to be 2.02 ± 0.07 mg K/g of bw.

Some neonatal mice were then injected subcutaneously with41K (batch #149492, Oak Ridge National Laboratory, Oak Ridge, TN, U.S.A.), which contains 99.04%41K, 0.015%40K, and 0.95%39K as KCl to equal 5% of total body K. Other mice were injected with a similar amount of standard KCl which contains 6.73%41K, 0.01%40K, and 93.26%39K.23 Twenty-four hours later, all the mice were killed and their calvariae removed by dissection and cultured as described below. In a previous study, we determined that by ∼24 h, the subcutaneous distention at the injection site completely resolved.16

Calvarial culture procedures

The41K-labeled and nonlabeled neonatal CD-1 mice were killed by cervical dislocation, and their calvariae (frontal and parietal bones of the skull) removed by dissection and the adherent cartilaginous material trimmed.6–11 The periosteum was left intact. Exactly 2.8 ml of culture medium (Dulbecco's modified Eagle's medium [DMEM] with 4.5 g/l of glucose; Whittaker M.A. Bioproducts, Walkersville, MD, U.S.A.) containing heat-inactivated (1 h at 56°C) horse serum (15%), Na heparin (10 U/ml), and K penicillin (100 U/ml) was preincubated at a fixed chosen Pco2, 37°C, for 3 h in 35-mm Petri dishes. One milliliter was then removed to determine preincubation medium pH, Pco2, and Ca concentration, and two calvariae were placed in each dish on a stainless steel wire grid. Total bone content in each culture dish was controlled by using mouse pups that were the same age and size, by using a standardized dissection procedure, and by placing two bones in each dish. Experimental and control cultures were performed in parallel and in random order. Bones were incubated for 24 h at 37°C, at which time a second sample of culture medium was obtained and similarly analyzed. The incubator (Forma model 3154, Marietta, OH, U.S.A.) maintains a constant temperature (±0.02°C) and a constant Pco2 (±0.1%) at an ambient O2 concentration of 21%.

Experimental groups

Calvariae were studied in the following four groups: nonlabeled (NL), control (Ctl), acidosis (Ac), and acidosis + calcitonin (3 × 10−9 M final concentration, Ac+CT). We have previously shown that this concentration of the osteoclastic inhibitor calcitonin is effective in suppressing cell-mediated bone resorption.6,11 The NL group contained only calvariae from animals injected with standard KCl and the Ctl, Ac, and Ac+Ctl contained only calvariae from animals injected with41KCl. In the NL and Ctl groups calvariae were cultured in unaltered medium (pH ≈ 7.40, Pco2 ≈ 40 mm Hg, [HCO3] ≈ 25 meq/l). In the Ac and Ac+CT groups the medium pH was decreased by the addition of concentrated HCl (pH ≈ 7.12, Pco2 ≈ 40, [HCO3] ≈ 12).

Scanning ion microprobe

Calvariae were removed from the incubation medium, rapidly washed three times with deionized distilled water, rapidly frozen in an acetone/dry ice bath (−77°C) and then lyophilized while frozen until dry (at least 6 h).6,8,14–19 The dry calvariae were mounted on aluminum supports with conductive glue and coated with a thin layer (∼5 nm) of gold. This layer, which is rapidly sputtered away from the area being scanned, prevents artifact-inducing electrical charging.

The scanning ion microprobe utilized for these studies was conceptualized and built at the University of Chicago and employs a 40 keV gallium beam focused to a spot 40 nm in diameter.6,8,14–19,24,25 The beam is scanned across a sample surface in a controlled sequence resulting in the emission of secondary electrons and atoms. These secondary particles originate within, and consequently carry information about, the most superficial 1–2 nm of the sample and can be used to generate images of the surface topography of a sample similar to those obtained using a scanning electron microscope. The particles can also be collected and analyzed by secondary ion mass spectroscopy (SIMS), a technique that separates the sputtered ions according to their mass-to-charge ratio.

Microanalysis of the calvariae by SIMS was performed in four distinct modes. In the imaging SIMS mode, the mass spectrometer is tuned to transmit a single ion species while the probe is scanned over a square region of the sample surface. The result is a monoisotopic elemental distribution image, called a SIMS map. Because the mass-resolved counting rates during the image acquisition are recorded, these SIMS maps provide both a local quantitative record as well as a two-dimensional representation of the distribution of a particular element on the surface being scanned. In the nonimaging SIMS mode, the spectrometer is rapidly and sequentially returned to filter several chosen ions species. At the same time, the probe is quickly scanned over a square area, so that the measured signals are secondary ion intensities averaged over the entire field of view. Using this “peak-switching” technique, the relative concentrations of several elements can be acquired simultaneously from one area and at one sample depth. In the mass analysis mode, the spectrometer mass tuning can be systematically varied (as in a conventional mass spectrometer) to yield mass spectra. If the spectra data are corrected by element-dependent sensitivity factors, they provide quantitative relative abundance measurements for a given area of sample. In the fourth microanalysis mode, the gallium beam is used to erode the surface, continually exposing deeper layers for analysis. As in the nonimaging SIMS mode, the “peak switching” procedure is used to analyze simultaneously several ion species. As deeper layers are exposed, a depth profile can be created that characterizes composition as a function of depth. Erosion depth is estimated from the primary ion current, the average density of bone, the duration and area of the scan, and a secondary particle sputter yield of 10 which is typical of these materials.6,8,15–19,26–28

For SIMS analysis, the secondary ions emerging from the sample are transported through a high transmission optical system containing an electrostatic energy analyzer and a magnetic sector mass spectrometer (mass resolution of ∼0.07 AMU measured at 40 AMU). The secondary ions are accelerated to 5000 eV energy for mass separation and detection by a secondary electron detector operated in pulse mode (each collected ion yields one digital pulse). Mass spectra are accumulated with a multichannel scaler which counts each detected ion by ramping the magnetic field of the mass spectrometer to scan a preselected mass region of the spectrum while the probe is scanning an area of arbitrary dimensions. The choice of the scanned area determines the depth over which the target composition is sampled for a given probe current and time. Elemental maps are constructed by recording the individual detected pulses in a 1024 × 1024 array of computer memory, each element of the array corresponding to a position of the probe on the sample. The imaged areas were 40 × 40 μm2. Images are printed on Polaroid film. For the most abundant elements (Na, K, and Ca), counting rates as high as 4.0 × 105 cps/pA of primary current were observed. In such cases statistically significant elemental images could be obtained in scan times of less than 30 s with probe currents of a few pA.

For each calvariae the surface and subsurface concentration of39K,41K, and40Ca were measured. We randomly selected several calvariae, scanned them in their entirety, and made simultaneous ion determinations of39K,41K, and40Ca. The analyses were repeated on representative areas of calvariae. Surface data were recorded after the erosion of the ∼5 nm of material; this procedure ensured the removal of any superficial surface contamination and permitted the acquisition of highly reproducible surface measurements and is consistent with our previously published studies.6,8,14–19 Representative elemental images of39K,41K, and40Ca, were then made for each sample. Given the beam current of 30 pA and image acquisition time not exceeding 524 s, the depth of erosion for these topographic and elemental measurements never exceeded ∼5 nm which did not result in significant sample depletion as the calvariae are thicker than 10 μm. Finally, the gallium beam was used to sputter-erode the samples while the spectrometer was adjusted to filter, using the “peak-switching” capability of the instrument, the39K,41K, and40Ca secondary ions. In this manner, the relative subsurface concentrations of these ions were determined as a function of depth.

Correction methods similar to those that we have previously reported6,8,14–19 were applied to the observed mass-resolved counting rates to obtain secondary ion yields proportional to the elemental concentration in the sample. Corrections are necessary because of the species-dependent sputtering and ionization probabilities of the emitted atoms. Relative to Ca taken as standard, the corrected yields for K are obtained by dividing the observed counting rates by the sensitivity factor 1.9 for apatites, assumed to apply to other bone crystals as well.

The total ion counts in a micrograph are a function not only of the emission properties of ions from a sample but also of the fraction of the field of view occupied by the sample, which in the case of the calvariae may have physical holes. In addition, the detected ion yields are dependent on the degree of sample surface roughness. Because of these considerations, we express our results in terms of the ratios of counts obtained for the same area of a sample. Such ratios are independent of the fraction of the field of view occupied by the sample and of the surface topography.

Mean ion ratio calculations

Mean ion ratios were calculated as the difference between the log of the counts in the numerator and the log of the counts in the denominator. The mean of the differences in each group was then obtained and analysis of variance (ANOVA) calculated. Values expressed are the antilog of the mean of the differences in each group plus the upper 95% confidence limit. The confidence limit was obtained by taking the mean of the differences in each group and adding the product of 2.24 times the standard error of this mean and then calculating the antilog of the resulting value. The number 2.24 is the T score for a one-tailed p < 0.025; so that 2.5% of the data lies above the upper confidence limit.6,8,14–19

Conventional measurements

pH and Pco2 were measured with a blood gas analyzer (Radiometer BMS 2 Mk III, Copenhagen, Denmark). Medium [HCO3] was calculated from the pH and Pco2 using the Henderson-Hasselbalch equation.6,9 The medium total Ca concentration was measured by automatic fluorometric titration, which we have shown to give results similar to those obtained using atomic absorption spectroscopy.7

Net Ca flux was calculated as Vm × (Caf − Cai) where Vm is the medium volume (1.8 ml) and Caf and Cai are the final and initial Ca concentrations.6,9 A positive value indicates movement of the ion from the bone into the medium and a negative value indicates movement from the medium into the bone. Results are expressed as nanomoles per bone per 24 h.


All tests of significance were calculated using ANOVA with a Bonferroni correction for multiple comparisons (BMDP, University of California at Los Angeles, CA, U.S.A.) on digital computer. Values are expressed as mean plus the upper 95% confidence limit for ion ratios; p < 0.05 was considered significant.


41K labeling

After injection of41K equal to 5% of the pup's total body K, there was an increase in the proportion of41K to total bone K; the ratio of bone41K/(39K+41K) increased to 9.8 ± 0.5 (ratio of counts per second of detected secondary ions ±95% confidence interval) on the surface of41K-labeled control (Ctl, n = 23) calvariae compared with the ratio of 6.7 ± 0.5% in nonlabeled (NL, n = 19) calvariae (p < 0.001). There was increased incorporation of41K only on the surface, but not in the interior (6.9 ± 0.4%, n = 12), of the Ctl calvariae (p < 0.001 vs. Ctl surface, p > 0.05 vs. NL surface) indicating that over the 24 h from injection to dissection, the excess41K is localized on the bone surface but not in the interior. The contribution of40K, which is present at 0.01% of total K in nature, is ignored for these calculations.23

Initial medium pH, Pco2, [HCO3], and net calcium flux

Compared with Ctl, the initial medium pH and [HCO3] were each significantly lower in both the Ac and Ac+CT medium, whereas initial medium Pco2 was unchanged (Table 1). There was no difference in initial medium pH, Pco2, or [HCO3] between Ac and Ac+CT.

original image

Compared with calvariae incubated in Ctl medium (n = 6), incubation in Ac medium (n = 6) led to a marked increase in net Ca efflux from bone (120 ± 3 nmol/bone/24 h vs. 619 ± 56, Ctl vs. Ac, respectively, p < 0.01). Incubation in Ac+CT medium (n = 6) resulted in less Ca efflux (29 ± 2 nmol/bone/24 h) than either Ctl or Ac (p < 0.01).

Effects of acidosis on bone41K/40Ca

Compared with calvariae incubated in control medium (Ctl), incubation in acidic medium (Ac) caused a marked decrease in the ratio of41K/40Ca on the surface and in the interior of the calvariae (Fig. 1). Compared to Ctl, addition of CT to the acidic medium (Ac+Ct) resulted in a fall in41K/40Ca on the surface that was not as great as that caused by Ac alone; however, the fall in41K/40Ca in the interior was not different from that caused by acidosis alone.

Figure FIG. 1.

Ratio of41K/40Ca (mean ratio of counts per second of detected secondary ions + 95% confidence limit) on the surface (top) and interior (bottom) of neonatal mouse calvariae cultured in control (Ctl: n = 23, surface; n = 12, interior) or acidic medium without (Ac: n = 28, surface; n = 13, interior) or with the osteoclast inhibitor calcitonin (Ac+CT, n = 17, surface; n = 11, interior) after injection of41K. *p < 0.01 vs. Ctl; +p < 0.01 vs. Ac.

Ionic depth profile

The gallium beam was utilized to erode the surface of the41K-enriched bone. Compared with Ctl, Ac caused an abrupt fall in the ratio of41K/40Ca after the first few nanometers from the bone surface and remained lower until about 150 nm (Fig. 2, top). Ac+CT appeared to have an intermediate effect. There was no difference in the ratio of41K/(39K+41K) over the depth profile studied in Ctl, Ac or Ac+Ctl (Fig. 2, bottom).

Figure FIG. 2.

Relationship between ratios of41K/40Ca and depth (top) and41K/(39K+41K) expressed as % and depth (nm) (bottom) for neonatal mouse calvariae cultured in control (Ctl) or acidic medium without (Ac) or with the osteoclast inhibitor calcitonin (Ac+CT) after injection of41K.

Relationship of41K/40Ca to net calcium flux

Compared with Ctl, Ac caused a marked increase in net Ca efflux and a fall in the ratio of41K/40Ca, indicating that acidosis must cause a greater efflux of K than Ca from the bone mineral (Fig. 3). Compared with Ctl, Ac+CT caused no increase in net Ca flux and a fall in the ratio of41K/40Ca, indicating that acidosis, in the presence of the osteoclastic inhibitor calcitonin, results in a marked efflux of K in the absence of Ca efflux.

Figure FIG. 3.

Relationship between the surface41K/40Ca (mean ratio of counts/s of detected secondary ions + 95% confidence limit) and net Ca flux (mean ± SE) for neonatal calvariae cultured in control (Ctl, circle) or acidic medium without (Ac, square) or with the osteoclast inhibitor calcitonin (Ac+CT, diamond) after injection of41K. Differences in41K/40Ca as in Fig. 1 and differences in net Ca flux as in RESULTS. Absence of error bars indicates that symbol is larger than the error.


Utilizing a high-resolution scanning ion microprobe in conjunction with secondary ion mass spectroscopy of bone labeled with the stable isotope41K, we have shown that bone mineral surface is rich in K relative to Ca. Incubation in acidic medium produced by a decrease in medium [HCO3], a model of metabolic acidosis, causes a marked fall in the ratio of41K/40Ca on the bone surface. The fall in this ratio in concert with an efflux of Ca from bone indicates that acidosis causes a greater loss of bone mineral K than Ca. When osteoclastic function is inhibited with calcitonin, the acid-induced loss of mineral Ca is abolished, yet the surface ion ratio of41K/40Ca falls, indicating that osteoclastic function is necessary to support the enriched levels of mineral surface K in the presence of acidosis.

That acidosis causes a greater release of K than Ca suggests that acidosis can alter the metabolically active membrane that separates the mineral from the extracellular fluid. We have previously shown that the bone mineral is separated from the fluid bathing the bone by a metabolically active partition.17 This so-called “bone membrane” appears necessary to support the high K to Ca ratio found on the surface of the bone mineral. When the bone cells are killed, there is a large influx of Ca into the bone in conjunction with an efflux of K.17 In this study, we have shown that acidosis causes an efflux of both K and Ca, indicating that the bone cells remain viable yet the ability of this membrane to exclude K is altered. Acidosis has been shown to decrease osteoblastic collagen synthesis and increase osteoclastic β-glucuronidase activity.9–11,13 Perhaps the changes in bone cell activity are responsible for the acid-induced alteration of the function of this bone membrane.

It is necessary to do these studies with labeled K to exclude that exchange of medium and mineral K was not responsible for either the elevated levels of K found on the bone surface or the acid-induced loss of mineral K relative to Ca.6,8,14–19 When slices of bovine cortical bone are incubated in similar culture medium to that utilized in this experiment, there is an increase over unity (2.3 ± 2.2 counts/s of detected secondary ions, mean ±95% confidence interval) in the ratio of K to Ca8; however, this increase is far less than that found in the calvariae in the current study (14.4 ± 1.2). Thus the marked increase in surface mineral K found in this study is not simply a function of medium K diffusing into the bone mineral but must be a function of an active metabolic process.

In vivo labeling with the stable isotope41K almost certainly represents the distribution of K present during recent normal bone formation. The K is injected subcutaneously into the mouse, where it is absorbed and enters the systemic circulation to be utilized during mineralization. Because41K is normally present in nature and is found in the vehicle injected (nonlabeled) mouse calvariae in the same ratio as normally found in nature, there is no reason to believe that41K would be handled any differently than39K in the mouse. It is not surprising that after 24 h the41K label localized to the surface, but not the interior, of this actively growing bone or that the ratio of41K/(39K+41K) fell rapidly from the surface to the interior of the calvariae (Fig. 2, bottom). With a longer interval between injection and dissection, the label might be expected to become increasingly incorporated into the interior of the bone. In a previous study, there was far greater localization of the stable isotope of Ca,44Ca, on the surface compared with the interior, of similarly injected mice.16 As expected, there was no difference in the ratio of41K/(39K+41K) among the Ctl, Ac, or Ac+CT groups. Neither acidosis nor calcitonin would be expected to effect41K differently than39K. Another isotope of K,40K, is present at 0.01% of total K in nature23; however, its extremely low abundance makes enrichment difficult and it is currently only available enriched to approximately 3% of total K (Oak Ridge National Laboratory).

Potassium is one of the most abundant trace elements in the mineral of bones and teeth, occurring normally in the range of 0.02–0.05%.29 The mineral of teeth and bone is a highly substituted carbonated hydroxyapatite.30 In bone, carbonate ions are present at approximately 6% by weight, and substitute for about one in five phosphate ions in the hydroxyapatite [Ca10(PO4)6(OH)2] crystal structure.30 Carbonated hydroxyapatite is much more soluble in acid than pure hydroxyapatite.31,32 Divalent ions such as Mg and Sr are readily incorporated into the apatite structure together with carbonate at levels found in biological mineral.30,33,34 These cations substitute for Ca, and although they have the same charge, they are, respectively, smaller and larger than Ca (ionic radii: Mg = 0.66; Ca = 0.99; Sr = 1.12 Å) causing distortions in the crystal lattice somewhat offset by the smaller carbonate ion. Magnesium and carbonate are preferentially lost from carbonated hydroxyapatite and from biological mineral as it dissolves in acid.30,33,35 Strontium, when incorporated with carbonate and fluoride, reduces the solubility by offsetting the size differences in the ions.34 The monovalent ions Na and K are also readily incorporated into hydroxyapatite and into biological mineral.30,36,37 Sodium has the same ionic radius as Ca and readily substitutes in the crystal structure together with carbonate, is found in biological mineral in the range of 0.5–1.0%, and is comparable to the range found for Mg. More Na is incorporated as the carbonate content is increased. Since Na is monovalent, however, this causes a charge imbalance which is offset by carbonate substitution for phosphate.30,36,37 Such carbonated apatites are highly reactive to acid dissolution. Potassium has not been studied as extensively as other trace elements such as Na, Mg, and Sr, but has been shown to be incorporated into the carbonated apatite structure and is present in biological mineral at similar levels to Sr.30 Potassium would be expected to be incorporated in a similar fashion to Na because of its related electronic configuration. However, the ionic radius of K (1.33 = Å) is much larger than Ca and Na. Potassium incorporation into carbonated hydroxyapatite would therefore be expected to be related to carbonate inclusion and to increase the crystalline disorder leading to a highly reactive mineral. Although this has not been demonstrated experimentally in synthetic systems, the results of the present study can be interpreted based on the above considerations. During the addition of new bone mineral to the surface of existing crystals in the manner that occurred in the present study, it appears that the stable isotope41K was readily incorporated into a poorly formed mineral. This mineral readily dissolved under conditions of acidosis and preferentially lost carbonate and K, leaving relatively more Ca behind than K (Fig. 3). Further, in the presence of the calcitonin, the dissolution seemed to become almost exclusively pH driven. That is, with a lower pH, highly soluble fractions of the mineral will continue to dissolve simply by a physicochemical mechanism, not requiring cell-mediated processes. The K appeared to be part of, or perhaps the major reason, why part of the mineral so readily dissolved. This is consistent with defects in the crystal structure related to inclusion of K, Na, and carbonate.

In this study, we did not attempt to measure changes in medium K induced by acidosis. In previous studies we were not able to detect consistent changes in the medium monovalent ion composition secondary to various stimuli.6,8,15–19 The inability to detect changes in medium K may be due to insensitivity of measuring change in medium ion composition as a reflection of net ion flux to the relatively large medium K concentration (5.6 mM) which makes small changes in K difficult to detect. Changes in medium Ca concentration are easier to detect perhaps because Ca is present at a lower concentration in the culture medium.

The results of this study are in accordance with that of our previous studies. As in this study with41K, we have previously shown that calvariae can be labeled with44Ca, another stable isotope.16 We have also shown that the surface of the bone mineral is rich in K relative to Ca6,8,14–19 and that inhibition of bone cell function leads to a loss of K from the surface of the bone mineral.17 We have shown that acidosis causes a net efflux of Ca from the bone mineral and that this occurs in conjunction with a greater efflux of mineral K.6,8,15,19 However, in previous studies we did not use stable isotope labeling of41K and thus were unable to determine that the exchange of medium and mineral K were not responsible for either the elevated levels of K found on the bone surface or the acid-induced loss of mineral K relative to Ca.

Thus, utilizing the stable isotope41K to label the bone mineral with detection by high resolution scanning ion microscopy with secondary ion mass spectroscopy, we have demonstrated that bone can be labeled with this stable isotope and that the mineral surface is rich in K relative to Ca. Incubation in acid medium causes a fall in the surface ratio of K to Ca in conjunction with Ca efflux indicating a greater release of mineral K than Ca.


This study was supported by grants AR 39906 and DK 47631 (D.A.B.) from the National Institutes of Health.