Ionic exchanges of turtle shell in vitro and their relevance to shell function in the anoxic turtle

The shell of the freshwater turtle Chrysemys picta bellii plays an important role in enabling this animal to adapt to the acid–base consequences of prolonged anoxia. Two general functions have been ascribed to the shell in this context: (1) a demineralizion involving the release of calcium and magnesium buffers that supplement normal extracellular buffering capacity and permit the turtle to tolerate lactic acid loads far exceeding the intrinsic capacity of the extracellular fluid (Jackson and Heisler, 1982); (2) an uptake of lactate during anoxia that removes it from the general circulation, and a subsequent release of lactate into the circulation for reutilization during normoxic recovery (Jackson et al., 1996; Jackson, 1997). Together, these two mechanisms may account, under extreme anoxic conditions, for up to two-thirds of the total body lactic acid buffering (Jackson, 1997).

To some extent, these mechanisms are based on direct observations from the intact animal, but much of the detail is inferential. Plasma calcium and magnesium concentrations rise consistently during anoxia and presumably derive from the shell and skeleton, since these are by far the largest reserves of these elements in the body. Significant changes in shell concentration, however, have only been observed with respect to magnesium; the shell calcium levels are so large that the relatively small loss to the extracellular fluid is not easily detected. Total CO₂ concentration falls significantly in the shell during anoxia, but the chemical form of the mobile CO₂, whether carbonate or bicarbonate, is uncertain. Phosphate levels do not increase in the plasma of anoxic turtles, although this is the most abundant anion in the shell, and phosphate levels are elevated in the plasma of snakes following exhaustive exercise (Ruben, 1983). Shell lactate levels rise during anoxia, but it is not certain what ion movements balance the lactate, an issue that is crucial for understanding the acid–base significance of the lactate uptake. To resolve some of these uncertainties, we have studied the acid–base properties of the turtle shell in a reduced system that consists of just shell exposed to a defined salt solution. Our goals were to quantify exchanges more precisely and to resolve some of the uncertainties of the intact system. Our challenge, however, was to establish whether this simplified preparation preserves essential features operating in the whole animal.

Shells were collected from western painted turtles Chrysemyspicta bellii Gray obtained commercially. The turtles were decapitated, and the shells were thoroughly cleaned of adherent muscle and connective tissue. To prepare for study, pieces of shell (sampled randomly from both carapace and plastron) were ground to powder at the temperature of liquid nitrogen using a Freezer Mill (Spex Certiprep; model 6700). Powdered samples were stored frozen in a sealed container prior to study. Most of the measurements reported were performed on shell samples from six turtles that had been in a standard normoxic state prior to shell collection. For comparison, however, additional experiments were conducted on powdered shell with a high concentration of lactate obtained from a turtle that had been anoxic at 3 °C for 3 months.

A diagram of the experimental arrangement for studying turtle shell in vitro is shown in Fig. 1. Prior to addition of the shell, 11 ml of solution was equilibrated with a CO₂-containing gas (2.5 % or 15 %) until a stable pH was reached. The gas was supplied by a compressed gas cylinder, and the flow rate was monitored using a Labcrest rotameter and quantified more accurately at the end of each experiment using a Brooks Vol-U-Meter. The pH of the solution depended on the composition of the buffer (see below) and was 6.0, 6.5, 7.0, 7.5 or 8.0. Once full equilibration had been achieved, a 1 ml sample of solution (pre-incubation) was collected for ion analysis, leaving a final volume of 10 ml in the flask. During the equilibration period, powdered shell (1.5–2.5 g) resided in a bulbous side-arm of the equilibration flask that had been weighed prior to attachment to the flask both with and without the added shell. The side-arm was connected to the flask by a ground-glass joint, and when the time came to add the shell powder to the buffer solution, the side arm was rotated by 180 ° and the powder flowed into the solution. At the end of the experiment, the side-arm was again weighed to establish exactly how much shell had been added. This closed-system arrangement permitted the shell and buffer to mix without contamination by room air. Once added, the powder and buffer were mixed thoroughly and continuously with a magnetic stirrer, and this usually caused an immediate alkalization of the solution and a release of CO₂ into the solution and the ventilating gas.

Alkalization was counteracted, however, and solution pH was held at its initial value by titration with 1 mol l⁻¹ HCl using a pH-stat system (Radiometer; model TTT60 titrator, ABU11 Autoburette, PHM 62 pH meter, and pH electrode). Only in one circumstance did pH fall after addition of the shell to the buffer, and pH was controlled in this case by titration with 1 mol l⁻¹ NaOH. Gas flowing out of the equilibration chamber was dried and analyzed for percentage CO₂ using an infrared CO₂ analyzer (Applied Electrochemistry; model CD-3A) connected to a Kipp and Zonen (model BD 41) chart recorder. At the end of the experiment, a final 1 ml sample (post-incubation) was taken from the flask and centrifuged, and the supernatant was stored for subsequent analysis. A portion of the wet powder was blotted dry, weighed, dried in an oven to constant mass, and processed for lactate analysis. The water content of pre-incubation shell powder was also determined.

All solutions utilized had the following ion concentrations in common (values in mmol l⁻¹): 100 Na⁺; 2.5 K⁺; 2.0 Ca²⁺; 1.0 Mg²⁺. To achieve pH values of 6.0, 6.5, 7.0, 7.5 and 8.0 when equilibrated with 2.5 % CO₂, solutions had [HCO₃⁻] values of 0.6, 2.0, 6.0, 20.0 and 59.0 mmol l⁻¹, respectively. Solutions contained either a high lactate concentration (50 mmol l⁻¹) or zero lactate, and in all cases the remaining negative charge was supplied by Cl⁻. In high- experiments, 15 % CO₂ was used and [HCO₃⁻] levels were raised accordingly to give solution pH values of 6.5, 7.0 and 7.5.

The standard protocol used a high-[lactate] Ringer’s solution equilibrated with 2.5 % CO₂ at pH values of 6.0, 6.5, 7.0, 7.5 and 8.0. Cumulative titration volumes were recorded at 5, 10, 15, 30, 45, 60, 75, 90, 105 and 120 min. Several variations on this basic design were investigated. First, to test for full equilibration, shell powder was incubated at pH 6.5, 7.0 or 7.5 in the high-[lactate] solution for 20 h. Second, to evaluate the effect of lactate on exchange processes, shell was incubated in Ringer’s solution at pH 6.5 or 7.5 without lactate. Third, to determine whether the nature of the acidosis affected exchange, shell was incubated in lactate-free solution equilibrated with 15 % CO₂ at pH 6.5 and 7.0. Fourth, to test for the effects of prior anoxia, powdered shell containing high lactate levels from a turtle after 3 months of anoxic submergence at 3 °C was studied at pH 6.5, 7.0 and 7.5 in solution without lactate.

The initial and final solution samples were analyzed immediately for lactate concentration using a YSI (model 2300) stat plus analyzer. From these measurements and the solution volume, the amount of lactate leaving the solution and entering the shell was calculated. The remaining solutions were stored frozen in sealed Eppendorf tubes. Prior to analysis of other ionic components, the samples were thawed and vortexed thoroughly. Sodium and potassium were measured using a flame photometer (IL; model 943), calcium and magnesium using an atomic absorption spectrophotometer (Perkin-Elmer; model 280), chloride using a chloride titrator (Radiometer; model CTM 10) and phosphate using the Fiske and SubbaRow method with Sigma reagents and a spectrophotometer (Milton Roy; Spectronic model 601). The level of bicarbonate present was known from the preparation procedure and was assumed to remain unchanged during incubation at constant pH and ⁠.

These measurements were intended to provide an essentially complete assessment of changes in levels of the major ions present. To verify that no unmeasured ion contributed significantly, ion balance was calculated for both the pre-incubation and post-incubation solutions.

Lactate concentration of shell powder, both fresh and post-incubation, was determined using the YSI analyzer following extraction in 8 % perchloric acid (7 ml g⁻¹ dry mass). This analysis was intended to provide a verification of the entry of lactate into the shell (or in the case of the anoxic turtle shell, the movement of lactate out of the shell), and also to permit an analysis of the conservation of lactate in the total system. Comparisons with changes of lactate levels in the incubation solution, however, suggested that our extraction of lactate from the shell was incomplete. Post-incubation shell powder was blotted with filter paper, weighed and dried to constant mass to determine water content. We assumed that the lactate in the sample prior to drying was in two compartments: first, in the water phase at the same concentration as in the bulk incubation solution; and second, associated with the non-aqueous phase at an unknown concentration. We also assumed that lactate from both compartments remained in the dry powder that was extracted and analyzed. Our final calculation subtracted out the aqueous component, and shell lactate concentration is expressed as μmol g⁻¹ dry mass.

The evolution of CO₂ from the flask during incubation was calculated as mmol of CO₂ for each of the four 30 min periods, and the total CO₂ produced (in mmol) during the 2 h period was then calculated and compared with the volume of 1 mol l⁻¹ HCl titrated (1 ml=1 mmol). The CO₂ analyzer was calibrated using standard gases, and percentage CO₂ change was determined from the chart recording by planimetry. Gas flow was converted to ^STPD, and ml(STPD) of CO₂ was converted to mmol by dividing by 22.26 ml mmol⁻¹ (Cameron, 1989).

Differences between treatment groups were compared by analysis of variance (ANOVA) using SigmaStat software.

The volume of titrant required to maintain constant pH varied inversely with pH in the high-[lactate] solutions (Fig. 2). At pH 6.0, approximately four times as much acid titration was required as at pH 7.5. At pH 8.0, essentially no titration occurred. The titration of acid was associated with an evolution of CO₂ from the shell that, in mmol, averaged nearly 50 % of the HCl titrated (Fig. 3).

The lactate concentration of the solution fell significantly (P<0.001) during the incubation, and the magnitude of this decrease increased as solution pH decreased (Fig. 4). On the basis of this fall in the concentration of lactate in the solution and the amount of shell powder present, we deduced that shell uptake rose significantly as the solution became more acidic (P<0.01; ANOVA). Direct determination of shell lactate content produced a similar result (Fig. 4), but a lack of agreement with the changes in lactate level in the solution became pronounced at higher pH, possibly as a result of incomplete extraction of lactate from the shell.

Changes were observed in levels of all the inorganic ions measured in the incubating solutions. Calcium and magnesium levels both increased as the solution acidity increased, but the level of phosphate, the major anion of the shell, changed only slightly over the pH range studied (Fig. 5). Chloride level increased markedly, but this could all be accounted for by the titrated HCl and no release from the shell appeared to occur. Potassium concentration increased significantly in all conditions, but the increase bore no relationship to the pH of the solution. Sodium concentration increased irregularly, the largest and most consistent increases occurring in shell powder that had a low water content prior to incubation. Cation and anion concentrations (in mequiv l⁻¹) agreed closely in the pre- and post-incubation samples, as illustrated for the pH 6.5 experiments in Fig. 6.

Increasing the incubation time from 2 to 20 h brought the solutions at pH 6.5 and 7.0 closer to full equilibration with the shell powder (Fig. 7), whereas at pH 7.5 equilibration was already essentially complete by 2 h. At pH 6.5 and 7.0, the full course of equilibration was best fitted by a double-exponential curve, suggesting that demineralization had a fast phase with a half-time of approximately 9 min and a slow phase with a half-time of approximately 2–3 h.

Incubation of shell in the absence of lactate with either 2.5 % or 15 % CO₂ required similar volumes of acid titration (per gram dry mass of shell) to hold the solution at pH 6.5 to those required with the high-[lactate] solution (Fig. 8). The total amount of calcium (and other cations) released from the shell relative to chloride added to the solution, however, was significantly lower in the high-[lactate] solution (results not shown). This indicates that the entry of lactate into the shell alkalized the incubation solution and contributed to the requirement for acid titration.

Post-anoxic shell powder released lactate to the solution as a function of solution pH. At pH 6.5, the concentration of lactate in the solution at the end of a 2 h incubation was20.8 mmol l⁻¹,whereas at pH 7.5 the final concentration of lactate in the solution was 26.8 mmol l⁻¹. The volume of acid titrant required for pH-stat control of anoxic shell was substantially reduced at all pH values tested (Fig. 9). At pH 7.0, an initial fall in the pH of the solution was followed by a gradual increase and a delayed initiation of acid titration. At pH 7.5, the pH of the solution tended to fall throughout the 2 h incubation period, and titration with 1 mol l⁻¹ NaOH was necessary to hold the pH constant. This was the only condition studied in which titration with base was required. The diminished acid titration in these experiments reveals that the release of lactate was associated with solution acidification; in other words, as the lactate left the shell, it was, in effect, accompanied by protons.

The exchange characteristics of powdered turtle shell observed in this study resemble in important respects the behavior of shell in vivo during long-term anoxic acidosis. As in the in vivo situation, calcium and magnesium, but not phosphate, built up progressively in the solution exchanging with the shell as acidosis developed (Fig. 5; Jackson and Ultsch, 1982; Jackson and Heisler, 1982; Herbert and Jackson, 1985). Associated with the release of these divalent cations, an alkalization of the solution and a discharge of CO₂ occurred that mirrored the enhanced extracellular buffering capacity and loss of shell CO₂ observed in vivo (Fig. 3; Jackson and Heisler, 1982; Warburton and Jackson, 1995). In addition, lactate moved into the shell from a high bathing solution concentration and moved out of the shell when the gradient was reversed, just as in vivo lactate is stored in the shell during anoxic acidosis and is released during normoxic recovery (Fig. 4; Jackson et al., 1996; Jackson, 1997).

The release of calcium and magnesium from the shell in vivo increases during the course of anoxic submergence, but it is not certain whether this is a passive process or whether it is under cellular or hormonal control. Our in vitro results suggest that release can be a direct effect of acidosis on the chemical equilibrium between the shell and the solution bathing it. In this regard, Bushinsky and Lechleider (1987) found that the medium bathing cultured mouse calvariae was in chemical equilibrium with the calcium carbonate component of the bone, at least in the acute phase. Acidity promotes bone demineralization in mammalian bone (Bushinsky, 1994), and lactic acid in particular is implicated in dissolution of tooth enamel in the production of dental caries (Arends, 1982). Cellular and hormonal effects contribute to demineralization in mammalian bone (Bushinksy et al., 1985), but these effects appear to be pH-insensitive whereas the passive effect, as in the present study, is pH-dependent. Little evidence exists for a hormonal contribution to demineralization of turtle shell (Warburton and Jackson, 1995).

The present results confirm that calcium and magnesium are released from turtle shell as carbonates. Previous work has shown that shell CO₂ levels fall during long-term anoxic submergence (Warburton and Jackson, 1995) and that the level of phosphate, the major anion of shell, does not increase in the plasma of anoxic turtles (Herbert and Jackson, 1985). In the present study, only negligible amounts of phosphate were released from shell over the pH range we studied, and the release of CO₂ in the form of carbonate could account, in terms of charge balance, for the observed changes in ion levels in the solution. In mammalian bone, it is thought that CO₂ exists both as carbonate in the apatite crystal, where it can substitute for either hydroxyl or phosphate, and as bicarbonate in the hydration layer at the crystal surface (Bushinsky, 1994). Poyart et al. (1975) concluded that the labile fraction of bone CO₂, presumably the hydration layer component, was predominantly in the form of bicarbonate. In turtle shell, in contrast, bicarbonate does not appear to play a role based on the stoichiometry we observed. The lack of a significant release of phosphate from the turtle shell during acidosis contrasts with results from snakes (Ruben, 1983), in which an increase in blood calcium level following exhaustive exercise was accompanied by a parallel increase in blood phosphate level. This phosphate may not have originated in bone, however.

Recent studies in vivo have demonstrated an important role for the turtle shell in sequestering lactate during long-term anoxic submergence (Jackson et al., 1996; Jackson, 1997). After 9 days of anoxia at 10 °C and 3 months at 3 °C, shell lactate accounted for over 40 % of total body lactate. The assumption from these studies was that the entry of lactate into the shell represents a special case of acid buffering by the shell, although the associated ion exchanges (satisfying charge balance) could only be inferred. The present in vitro observations confirm this capacity of shell and, in addition, reveal several important features of this phenomenon.

First, our present data reveal unambiguously that the entry of lactate into shell is balanced electrically by an acid–base-relevant ion, either cotransported protons or countertransported carbonates. This conclusion is based on the observation that a portion of the titrated HCl was required to balance the loss of lactate from the incubating solution. Had lactate entry been coupled to another ‘strong’ ion, such as cotransport with sodium or countertransport with chloride, the uptake of lactate by the shell would have been acid–base neutral and would not have contributed to alkalization of the solution. Furthermore, when shell with a high initial lactate level was incubated, the solution required less titration of acid; i.e. the release of lactate acidified the solution, indicating that movement in this direction was also associated with an acid–base-relevant cotransport.

A second feature revealed by the present study is that shell lactate uptake is pH-dependent (Fig. 4), an observation with clear functional significance. At low pH, a condition that exists in vivo when lactate is being produced anaerobically, shell sequestration of lactate is favored, whereas at high pH, a condition associated in vivo with recovery from anoxic submergence, lactate binding by the shell is reduced and release into the extracellular fluid is favored. One aspect not clarified by these experiments is the state of lactate within the shell, although our assumption is that it is bound to calcium in the mineral phase. We know from previous studies (Jackson andHeisler, 1982) that calcium and lactate form a complex within the plasma, especially when their concentrations are high.

An important question is how reasonable the observed in vitro exchanges are in relation to events occurring in vivo. Clearly, the conditions for exchange are far more favorable in vitro owing to the large exposed shell surface and the ready access to a large volume of well-mixed solution. However, the in vivo exchanges took place over far longer periods (9 days at 10 °C and 12 weeks at 3 °C) so that, even though shell perfusion is slow and diffusion exchange parameters are less favorable, similar equilibria may be achieved in both circumstances. This is borne out by several direct comparisons. In the case of magnesium and total CO₂, we calculated the percentage of the estimated total amounts of these substances lost from the shell during the in vitro incubations over the pH range studied (Fig. 10). The in vivo loss of magnesium from the shell after 12 weeks of anoxia at 3 °C (Warburton and Jackson, 1995) was 19 %, and the loss of CO₂ over the same period was 12.5 %. These values fall in the in vitro range near the pH 7.0 value characteristic of anoxic turtles. For these substances, therefore, the in vivo and in vitro losses were quite similar. The in vitro percentage loss of calcium from the shell is also shown in Fig. 10, and the low values (<4 %) help to explain why significant losses are difficult to detect (Warburton and Jackson, 1995). The concentrations of lactate found in the shell in vivo after prolonged anoxia at 3 °C and 10 °C (Jackson, 1997), calculated in the same way as in the present study and expressed in μmol g⁻¹ dry mass, were approximately the same as the plasma concentration (in mequiv l⁻¹). This condition is seen in vitro (Fig. 4) at a pH of approximately 7.3, which is not far from the in vivo pH. Recall, however, that we may have underestimated shell lactate content using this method, so that the published in vivo estimates may be low as well. In these important respects, therefore, the in vitro preparation behaves quite similarly to the far more complex situation of the intact anoxic turtle.

In some other respects the in vitro system is less in agreement with, or reveals phenomena not apparent in, the intact animal. For example, calcium is lost to a greater extent relative to magnesium in vitro than in vivo. The increase in plasma magnesium level during anoxia in vivo is approximately 40–50 % of the increase in calcium level, whereas the magnesium level only increased by 15–30 % as much as the calcium level in vitro. Because the total shell magnesium content is only 2 % of the total calcium content, magnesium is clearly released preferentially both in vivo and in vitro, but the exchange circumstances in vitro may potentiate the loss of calcium. Both sodium and potassium moved out of the shell into the incubating medium in vitro, but neither event was pH-dependent. Potassium release was quite consistent across the pH range and produced elevations in the potassium concentration of the incubation solution not unlike the changes observed in plasma potassium level in vivo. It is not clear, however, whether the in vivo hyperkalemia is shell-derived, although the in vitro result supports this intriguing possibility. An observation in conflict with this interpretation, however, is the similar release of potassium from incubated post-anoxic shell, since a loss of shell potassium during the 3 months of submergence should have severely depleted the available potassium. Sodium loss was less consistent and most pronounced when dried shell powder was incubated. Preliminary results on intact turtles indicate a loss of shell sodium (Jackson, 1997), although past studies have not reported increases in plasma sodium concentration during anoxia (Herbert and Jackson, 1985; Jackson and Heisler, 1982). The loss of mammalian bone sodium during chronic acidosis is well known (Burnell, 1971).

In conclusion, this study provides support for the view that the shell of the turtle is of crucial importance to the remarkable capacity of this animal to tolerate prolonged anoxic acidosis. The dual roles of the shell, releasing carbonate buffers to the extracellular fluid and sequestering lactic acid, have been verified in our in vitro preparation. Both mechanisms represent effective buffering of lactic acid and confirm our earlier suggestion (Jackson, 1997) that the shell can account for some two-thirds of the lactic acid buffering during long-term anoxia. The similarity between our observations in this reduced preparation and previous observations on turtles in vivo suggests that this approach may be fruitful for understanding shell and bone acid–base function in these animals.

This work was supported by US National Science Foundation grant IBN 97-28794. We thank Mr David King of the Brown University Chemistry Department for his skillful glassblowing and Dr Carlos Crocker for helpful comments on the manuscript.