A Protein Titration Hypothesis for the Temperature-Dependence of Tissue CO₂ Content in Reptiles and Amphibians

Previously we reported changes in whole-body CO₂ stores, estimated from transients in CO₂ elimination, with temperature changes (ΔCO_2WB/ΔT ) in three species of reptiles and one amphibian species (Stinner and Wardle, 1988; Stinner et al. 1994a; Stinner, 1982). The ΔCO_2WB/ΔT coefficient varied from −0.18 mmol kg−1 °C−1 in the black racer snake Coluber constrictor to −0.05 mmol kg−1 °C−1 in the cane toad Bufo marinus. The absolute value of the plasma ΔpH/ΔT coefficient also differs among these species and is low in C. constrictor compared with B. marinus (−0.003 units °C−1 versus −0.015 units °C−1 between 30 and 10 °C, respectively; Stinner et al. 1996; Boutilier et al. 1987). This inverse relationship between the ΔCO_2WB/ΔT and ΔpHpl/ΔT coefficients suggests an explanation for the thermal dependence of CO₂ stores. When temperature is changed, the change in [CO₂] is related to the nonbicarbonate buffer value (β_NB), the change in the ionization constant of the nonbicarbonate buffers (ΔpK) and ΔpH by the expression Δ[HCO₃−]=β_NB(ΔpK−ΔpH) (Heisler and Neuman, 1980; Heisler, 1986a). Thus, if β_NB and ΔpK/ΔT are similar among species, Δ[HCO₃−] (and hence the change in total [CO₂]) will be greatest in species exhibiting the smallest change in pH. For example, the ΔpK/ΔT coefficient for 4-methyl-imidazole is −0.023 units °C−1 (Figge et al. 1991a) and so, when body temperature changes, the absolute value of ΔpKim−ΔpH is much greater in C. constrictor than in B. marinus (0.019 versus 0.007 units °C−1). Consequently, there would be considerably more nonbicarbonate buffering of CO₂, with concomitant changes in net imidazole charge, in C. constrictor than in B. marinus.

This study tests the hypothesis that the increase in whole-body CO₂ stores with cooling results from titration of proteins by carbonic acid. We estimated Δ[CO₂]_WB/ΔT coefficients in eight additional species of reptiles and one additional amphibian species from measurements of the respiratory exchange ratio (rate of CO₂ elimination/rate of O₂ uptake), and compared Δ[CO₂]_WB/ΔT coefficients with values from the literature for ΔpHpl/ΔT coefficients. In addition, we estimated net plasma protein charge in C. constrictor at 30 and 10 °C from the principle of electroneutrality. Strong ions (Na+, K+, Ca2+, Mg2+, Cl−, SO₄²⁻;, lactate), inorganic phosphate, protein concentration, total [CO₂], and pH were measured in plasma at the two temperatures in resting chronically cannulated black racer snakes under steady-state conditions. The interspecific ΔCO_2WB/ΔT coefficients were found to be strongly correlated with the interspecific ΔpHpl/ΔT coefficients, and net plasma protein charge in C. constrictor fell by 21 %. Furthermore, between 30 and 10 °C, [CO₂] of skeletal muscle in C. constrictor increased, without corresponding changes in intracellular [Na+], [K+] or [Cl−]. The findings of this study support the hypothesis that the thermal dependence of whole-body CO₂ stores in reptiles and amphibians results from the titration of protein by CO₂.

Animals were purchased from commercial suppliers, except for painted turtles (Chrysemys picta) and snapping turtles (Chelydra serpentina) which were collected locally. They were housed in the animal facility at The University of Akron, where room temperature was maintained at 25–30 °C. The cages contained light bulbs set to a 12 h:12 h L:D cycle that provided basking sites at 35–40 °C. Approximately once weekly, the snakes, American alligators (Alligator mississippiensis), savannah monitors (Varanus exanthematicus), snapping turtles and bullfrogs (Rana catesbeiana) were fed on mice and rats. The desert iguanas (Dipsosaurus dorsalis), painted turtles and Mediterranean spur-thighed tortoises (Testudo graeca) were fed on canned dog food, fruit and vegetables several times each week. All animals used in this study appeared to be in excellent health. The animals were starved for 1 week prior to experimentation in order to avoid metabolic, acid–base and electrolyte changes associated with the digestion and absorption of food.

Net changes in whole-body CO₂ stores caused by changes in body temperature were assessed indirectly from measurements of the .respiratory exchange ratio (RE)..Rates of O₂ consumption and CO₂ production in undisturbed animals were determined in a sealed recirculating apparatus housed inside a darkened temperature-controlled cabinet (Precision Scientific, model 815, ±1 °C). The details of this apparatus and the procedure followed have been described previously (Stinner et al. 1994a). Briefly, animals were allowed to adjust to the experimental apparatus for 24 h. The chamber was then sealed, and air was recirculated at 2.1–2.5 l min−1. A portion of the recirculated air was continuously passed through a Drierite column and O₂ and CO₂ analyzers (Ametek S-3A and CD-3A; Ametek, Pittsburgh, PA, USA). The analyzers were calibrated with dry, CO₂-free room air and a 5 % CO₂ (balance N₂) certified gas mixture. The metabolism chamber was flushed with room air when the O₂ levels had fallen to no lower than 19 % an.d/or the CO₂ levels had risen to a maximum of 1.0 %, and were calculated using the equations for closed-system respirometry given by Stinner et al. (1994a). Net changes in whole-body CO₂ stores caused by changing the temperature were estimated from the differences between nonsteady-state and steady-state (control) RE values (see Stinner, 1982, for details of the integration procedure). The animals remained inside the metabolism apparatus for up to 3 weeks. They appeared to be in excellent condition at the end of these studies. The temperatures used and masses of each species are listed in Table 1.

The dorsal aorta of 49 C. constrictor (mean mass 217±87 g) was occlusively cannulated (PE 50 or PE 10) as described by Stinner et al. (1996). After a recovery period of at least 72 h, the snakes were placed inside individual chambers within a darkened Precision temperature-controlled cabinet. The trailing ends of the cannulas extended outside the cabinets so that blood could be sampled without disturbing the snakes. Additional details of this apparatus and the blood sampling procedure (see below) are given in Stinner et al. (1996). The snakes were left undisturbed for 24 h at 30 °C and 20 °C and for 48 h at 10 °C before blood sampling began to allow them to achieve steady state with respect to whole-body CO₂ stores.

For ⁠, pH and [CO₂] measurements, a train of four heparinized microhematocrit tubes was connected to a cannula and filled with a total volume of approximately 300 μl of blood. The capillary tube train was then immediately sealed and immersed in iced water. All blood sample analyses of ⁠, pH and [CO₂] were completed within 1 h of collection. and pH were measured on blood from two and one of the microcapillary tubes, respectively, using Radiometer BMS-3 electrodes thermostatted to the snake’s body temperature and PHM 72 acid–base analyzers (Radiometer, Copenhagen, Denmark). The electrode was calibrated using a 1.39 % CO₂ mixture supplied by mass-flow controllers (Tylan Corp., Torrance, CA, USA). The pH electrode was calibrated using Radiometer precision buffers. Plasma [CO₂] was determined in duplicate on the fourth microcapillary tube using a Capni-con 3a CO₂ analyzer (Cameron Instrument Co., Port Aransas, TX, USA). pK₁′ was calculated for each blood sample using the solubility coefficients given by Severinghaus (1965) (see Table 3). Snakes at 10 °C were sampled once each day for four consecutive days and snakes at 30 °C were sampled once each day for 2 days. Replicate values for each snake were then averaged before calculating group means ± S.D.

For analyses of ion concentrations, approximately 450 μl of whole blood was collected directly into heparinized microcapillary tubes and immediately centrifuged at 2000 g for 5 min in an Eppendorf microcentrifuge. For lactate analysis, 100 μl of the plasma was immediately deproteinized by adding 200 μl of a cold 8 % perchloric acid solution. The mixture was centrifuged at 15 000 g for 10 min, and the supernatant was then frozen for no more than 1 week before plasma [lactate] analysis using an enzymatic test kit (Sigma no. 726-uv/826-uv) and Coleman Jr II model 6/20 spectrophotometer. The remaining plasma was frozen for future determination of ion concentrations.

Plasma [Na+] and [K+] were analyzed in duplicate using flame photometry (Coleman model 51; Bacharach, Inc., Pittsburgh, PA, USA). Plasma [Cl−] was measured (six replicates) using a Digital chloridometer (Buchler Instr., Lenexa, KS, USA). Total plasma osmolality was determined in duplicate using a model 5500 vapor pressure osmometer (Wescor, Inc, Logan, UT, USA). Means of the replicates were used to calculate group means ± S.D. for tabulation. Total plasma Ca2+, Mg2+, inorganic phosphate and protein concentrations were determined using Sigma diagnostic test kits (procedure nos 587, 595, 360-uv and 690, respectively; Sigma Diagnostics, St Louis, MO, USA). Calibration curves were constructed for [protein] from Sigma protein standards provided with the diagnostic kit, for [Ca2+] and [phosphate] from Sigma standards no. 360-11, and for [Mg2+] from Sigma standards no. 595-11. [SO₄²⁻;] was determined using high-pressure liquid chromatography (Dionex HPLC 2010i; Dionex Corp., Sunnyvale, CA, USA).

Skeletal muscle [CO₂] was measured using the technique described by Stinner et al. (1994a). Twelve black racer snakes (mean mass 158±93 g) and 42 bullfrogs (mean mass 353±84 g) were individually housed inside Precision temperature cabinets at 30 or 10 °C, as described above. The bullfrogs had access to water (10–15 cm deep) as well as dry areas. After at least 48 h at 30 °C and 96 h at 10 °C, animals were gently taken by hand from the temperature cabinets and decapitated. The spinal cord of the bullfrogs was immediately pithed, and their deep body temperature was taken using a Schultheis quick-reading cloacal thermometer. Duplicate pieces of skeletal muscle weighing 0.2–0.3 g were swiftly harvested and added to 0.7 ml of 0.1 mol l−1 NaOH for soaking overnight in sealed bottles. [CO₂] of the solution was measured for each bottle in duplicate using a Capni-con 3a CO₂ analyzer. The four [CO₂] measurements for each muscle were then averaged before calculation of group means ± S.D.

Analysis of intracellular pH (pHi) in 19 black racer snakes (mean mass 170±88 g) was performed on muscle homogenates after metabolic inhibition according to the method of Pörtner et al. (1990). The snakes were housed inside Precision temperature cabinets for at least 48 h at 30 °C and 96 h at 10 °C. The cabinets were then briefly opened, and gauze soaked in isoflurane was dropped into the chambers. The chambers were then sealed for 15–20 min. The anesthetized snakes were removed and immediately decapitated. Four muscle samples (a combination of the transversospinalis, longissimus dorsi and iliolumbaris) were quickly taken from the mid-region of each snake, freeze-clamped and stored in liquid N₂ for up to 2 h. Additional muscle samples were taken for electrolyte and water content determinations (see below). Frozen muscle samples were rapidly transferred to a porcelain mortar half-filled with liquid N₂ inside an ice chest and ground to a fine powder using a pre-cooled pestle. Grinding typically took 1 min. Approximately 150 mg of the powder was rapidly strained under the liquid N₂ atmosphere of the mortar and transferred to a 500 μl Eppendorf microcentrifuge tube containing 200 μl of ice-cold inhibitor solution (150 mmol l−1 potassium fluoride, 5 mmol l−1 nitrilotriacetic acid; pH approximately 7.00). The transferred powder was quickly covered with the inhibitor solution and mixed with a needle in order to release the N₂ bubbles from the thawed tissue powder. The microcentrifuge tube was then filled with the inhibitor solution and capped. The mixture was agitated for 3 s using a vortex mixer, centrifuged for 20 s at maximal speed in an Eppendorf centrifuge, and stored in ice for no more than 1 h. The supernatant pH was repeatedly measured using a Radiometer BMS-3 pH electrode thermostatted to the snake’s body temperature and a PHM 72 acid–base analyzer. At least three pH measurements from each centrifuge tube were averaged, and the pH of the four muscle samples from each snake was averaged before calculation of group means ± S.D. In two snakes held at 10 °C and three snakes held at 30 °C, additional samples of the muscle powder were prepared and analyzed as described above, except that pH was measured at 30 °C and at 10 °C, respectively. The difference in supernatant pH between the 10 and 30 °C samples taken from the same ground muscle powder was used to determine the passive component of the temperature-dependent adjustment of pHi.

Muscle samples weighing approximately 0.25 g were analyzed for [Na+], [K+] and [Cl−] according to the method of Boutilier et al. (1986). Four samples from each of 25 C. constrictor (mean mass 167±80 g) were placed in tared vials containing 4 ml of 0.1 mol l−1 HNO₃, weighed, and gently shaken for 24 h at room temperature. Dissolved [Na+], [K+] and [Cl−] were determined as described above for plasma. Four additional muscle samples from 21 of the snakes were weighed and dried overnight to a constant mass at 60 °C.

Comparisons between values were performed using paired t-tests, Student’s t-tests and least-squares regression. Significance was assumed when P<0.05. Means ± S.D. are reported for group values.

Metabolic rates for the nine species studied are listed in Table 1. These rates are for animals kept for at least 24 h at the specified body temperatures. They do not necessarily represent minimal resting values because the animals were occasionally active inside the metabolic chambers, especially at 30–40 °C. This probably accounts for the high thermal quotients (Q₁₀>3) evident in Table 1 (e.g. Q₁₀ for C. picta of 3.6).

Cooling and heating the animals produced transient decreases and increases, respectively, in the respiratory exchange ratio. In eight of the nine species studied, steady-state RE values were not significantly affected by temperature (P>0.05). However, in T. graeca, mean steady-state RE at 30 °C (0.78±0.08) was significantly higher than mean RE measured after 150 h at 10 °C (0.61±0.06), suggesting that the animals still had not returned to steady state. Estimates of changes in whole-body CO₂ stores in T. graeca at 10 °C were made using the 30 °C steady-state RE values. As expected, there were large differences among species, as illustrated by comparing T. graeca with A. mississippiensis (Figs 1, 2). Cooling T. graeca from 30 to 10 °C reduced the RE to approximately 0.3–0.5, which was then followed by a gradual increase in RE (Fig. 1A). Heating the tortoises back to 30 °C resulted in a sharp increase in RE to well above 1.0, and even above 2.0 in one tortoise, and then a return to steady state by approximately 12 h (Fig. 2A). The effect of temperature changes upon the RE was smaller in A. mississippiensis. Lowering body temperature from 30 to 10 °C lowered the RE to approximately 0.6–0.7, and steady state returned within approximately 20 h (Fig. 1B). Raising the temperature back to 30 °C increased the mean RE to a little above 1.0, and RE then returned to steady state within 5 h (Fig. 2B).

Calculated changes in whole-body CO₂ stores are listed in Table 2, and range from −0.21±0.10 mmol kg−1 °C−1 in T. graeca to −0.02±0.02 mmol kg−1 °C−1 in R. catesbeiana. In addition to the nine species measured in this study, Table 2 contains ΔCO₂/ΔT coefficients for four species taken from earlier reports and ΔpHpl/ΔT coefficients available for 10 of the species. There is a significant correlation (r2=0.87, P<0.01) between ΔCO_2WB/ΔT and ΔpHpl/ΔT (Fig. 3). Least-squares linear regression yields ΔCO₂/ΔT=−0.18−8.24ΔpHpl/ΔT, where ΔCO₂/ΔT is in mmol kg−1 °C−1 and ΔpH/ΔT is in pH units °C−1. The equation predicts that a ΔpHpl/ΔT=−0.022 units °C−1 will result in no changes in whole-body CO₂ stores with changes in temperature (x-intercept of the line) and that a ΔpHpl/ΔT=0.0 units °C−1 will result in a whole-body change in CO₂ stores of −0.18 mmol kg−1 °C−1 (y-intercept of the line).

Although mean plasma pH in C. constrictor was higher by 0.056 units at 10 versus 30 °C (Table 3), the difference was not significant (P<0.1). was 7.1 mmHg lower, and total plasma [CO₂] was 3.2 mmol l−1 higher, at 10 versus 30 °C. The values for and pH at 30 °C listed in Table 3 also include measurements for five snakes from previous work in this laboratory (Stinner and Wardle, 1988). Plasma [HCO₃−] and dissolved [CO₂] were estimated from mean total [CO₂] and mean using the solubility coefficients listed in Table 3. Between 30 and 10 °C, [HCO₃−] increased by 3.1 mmol l−1 and dissolved [CO₂] remained virtually unchanged. pK₁′ was slightly but significantly higher (0.111 units) at 10 °C.

Inspection of Table 4 reveals that plasma electrolyte concentrations were largely unaffected by cooling the snakes from 30 to 10 °C. Only [phosphate] changed significantly, increasing by 0.7 mmol l−1. However, at 20 °C, the concentrations of several electrolytes (Na+, K+, Mg2+ and phosphate) were significantly different from those at 10 and/or 30 °C. The higher [Na+] at 20 °C appeared to have been accompanied by a higher [Cl−]. There was no effect of temperature upon [protein]. Protein charge (in mequiv 1−1) was calculated as (the sum of the concentrations of Na+, K+, 2×Ca2+ and 2×Mg2+) minus (the sum of the concentrations of Cl−, lactate, 2×SO₄²⁻;, 2×HPO₄2− and H₂PO₄−). The concentrations of HPO₄2− and H₂PO₄− were estimated using the Henderson–Hasselbalch equation, mean pH values from Table 3 and a pK of 6.791 at 37 °C (Siggaard-Anderson, 1974). The van’t Hoff equation [pK=(ΔH°/4.576)(1/T₂−1/T₁)] was used to correct pK to 30 and 10 °C, where ΔH°=982 cal mol−1 (Edsall and Wyman, 1958). Protein charge decreased from −0.48 mequiv g−1 protein at 30 °C to −0.38 mequiv g−1 protein at 10 °C (Fig. 4), largely because of the increases in total [CO₂] (carbonic acid) and inorganic [phosphate].

Between 30 and 10 °C, [CO₂] increased significantly by approximately 2.5 mmol kg−1 in the three muscles studied in C. constrictor (Table 5). This contrasts sharply with the pattern found in R. catesbeiana (Table 6): in 14 skeletal muscles, only one exhibited a significant increase in [CO₂] with cooling. Intracellular muscle pH in C. constrictor increased significantly with cooling (ΔpHi/ΔT=−0.009 units °C−1; Table 7). The passive component was ΔpHi/ΔT=−0.013± 0.001 units °C−1. Muscle [Na+] and [Cl−] each decreased significantly by approximately 12 mequiv l−1, but there were no significant differences in [K+] or in percentage water content in snakes at 30 and 10 °C (Table 7).

The temperature-dependence of whole-body CO₂ stores is strongly correlated with changes in plasma pH per unit change in temperature in reptiles and amphibians (Table 2; Fig. 3): species exhibiting relatively small absolute ΔpH/ΔT coefficients experience greater changes in whole-body CO₂ stores. An explanation for this relationship is that, with differences between ΔpK and ΔpH, there is titration of nonbicarbonate buffers by CO₂ (see Introduction). Assuming that the ΔpHpl/ΔT coefficients correspond to whole-body ΔpH/ΔT coefficients, as reported by Bickler (1982) and by Cameron and Kormanik (1982), the absolute value of the slope in Fig. 3 (8.24 mmol kg−1 pH unit−1) then represents a whole-body nonbicarbonate buffer value and the x-intercept (−0.022 units °C−1) represents the change in the mean whole-body ionization constant. Hence, ΔCO_2WB/ΔT=8.24(−0.022−ΔpHpl/ΔT), which fits the general expression Δ[HCO₃−]=β_NB(ΔpK−ΔpH), but with the addition of dissolved CO₂ (the effect of which is minor; see Table 3). While this explanation is speculative, it is interesting to note that the mean temperature coefficient for biological imidazole-like buffers is ΔpK/ΔT≈−0.02l units °C−1 (Edsall and Wyman, 1958), which is very close to the −0.022 units °C−1 obtained in the present study. This implies that the postulated nonbicarbonate buffering is dominated by proteins (as opposed to phosphates, where ΔpK/ΔT=−0.0025 units °C−1; Edsall and Wyman, 1958).

Increases in whole-body CO₂ stores with cooling in reptiles and amphibians obviously represents CO₂ (and hence carbonic acid) retention. However, within a given tissue compartment (e.g. extracellular fluid), a rise in [HCO₃−] may not be accompanied by increases in [H+] because the protons can be removed by ion-exchange mechanisms (e.g. by the kidney, stomach and skeletal muscle cells). If this were to occur, then there must be adjustments in the levels of other ions in order to preserve electroneutrality with the rise in [HCO₃−]. The data contained in Tables 3 and 4 and Fig. 4 for plasma in C. constrictor clearly show that ion-exchange mechanisms were not involved. There were no changes in strong ion levels (Na+, K+, Mg2+, Cl−, Ca2+ SO₄²⁻; and lactate) or protein concentration, and phosphate concentration (a weak acid) actually increased with cooling. Thus, the 3 mmol l−1 increase in [HCO₃−] was accompanied by an equivalent increase in [H+]pl that was buffered by the plasma proteins. The rise in phosphate concentration also titrated the plasma proteins. Consequently, net protein charge fell by 21 %, from −0.48 mequiv g−1 protein at 30 °C to −0.38 mequiv g−1 protein at 10 °C. Plasma protein charge at 37 °C and pH 7.4 in humans is −0.6 mequiv g−1 protein (Figge et al. 1991b). Plasma strong ion levels have been shown to be independent of temperature in other air-breathing ectotherm species (Bickler, 1984; Herbert and Jackson, 1985; Douse and Mitchell, 1991).

The data in Table 3 also show that the change in [CO₂]_pl cannot be explained by simple increases in the solubility coefficients with cooling. Depending upon the in vivo steady-state and ⁠, the increase in and could potentially result in an accumulation of dissolved O₂ and dissolved CO₂. This would have the effect of adding to the pulmonary O₂ uptake and reducing the pulmonary CO₂ elimination with cooling. Hence, the depression in RE and elevation in CO₂ stores could be interpreted on the basis of the solubilities of the respiratory gases. However, inspection of Table 3 reveals that between 30 and 10 °C there was no change in dissolved [CO₂]_pl and dissolved [O₂]_pl decreased, because of the reductions in and ⁠.

In previous reports, increases in extracellular CO₂ stores at lower temperatures were attributed to ion-exchange mechanisms generating HCO₃− (which is equivalent to removal of H+) (Stinner and Wardle, 1988; Stinner et al. 1994a, 1996). This view now appears to be incorrect. The suggestion of ion exchange was based upon the lengthy time courses for accumulation of CO₂ in C. constrictor (35 h to reach steady state when cooled from 30 to 10 °C, and 60 h to reach steady state when cooled from 30 to 5 °C) and the relatively large changes in [CO₂]_pl which could not be accounted for by estimated extracellular fluid nonbicarbonate buffering (Stinner et al. 1996). However, the estimates of [HCO₃−] generated from extracellular fluid nonbicarbonate buffers did not include ΔpK/ΔT, and it was assumed that interstitial fluid is free of proteins. This assumption is not consistent with the high protein concentrations reported for interstitial fluid and lymph in ectotherms (30–100 % of plasma protein concentration; Hargens et al. 1974; Hillman et al. 1987). In mammals, the interstitial colloid osmotic pressure in loose connective tissue is approximately 50 % of the plasma colloid osmotic pressure (Aukland and Reed, 1993). Thus, the estimated extracellular fluid β_NB was too low. Extracellular β_NB for C. constrictor can be obtained from the [HCO₃−]_pl and pHpl values listed in Table 3 and an assumed ΔpK/ΔT of −0.022 units °C−1, where β_NB=Δ[HCO₃−]/(ΔpK−ΔpH). The resulting calculation yields an extracellular fluid β_NB of 8.1 mmol l−1 pH unit−1, which is close to the value of 6.2 mmol l−1 pH unit−1 measured in the toad B. marinus (Boutilier et al. 1979). Hence, we conclude that nonbicarbonate buffering and ΔpKpr/ΔT account for the rise in [HCO₃−] with cooling in C. constrictor.

A similar pattern for CO₂ was found in skeletal muscle tissue of C. constrictor cooled from 30 to 10 °C. [CO₂] increased by approximately 2.5 mmol kg−1, which is close to that measured in plasma. Skeletal muscle [Na+] and [Cl−] each decreased significantly by approximately 12 mequiv l−1, but the small increase in [K+] was not significant (P>0.4; Table 7). These changes in whole-muscle electrolyte levels suggest that there was a decrease in the fraction of extracellular fluid volume relative to intracellular fluid volume with cooling, as reported for B. marinus (Stinner et al. 1994a). Extracellular fluid, relative to intracellular fluid, is high in Na+ and Cl− but low in K+. Hence, a decrease in the fraction of extracellular fluid will produce a decrease in whole-muscle [Na+] and [Cl−], and an increase in whole-muscle [K+]. We conclude that the increase in [CO₂] within the muscle resulted from an increase in [H₂CO₃] rather than from active ion exchange across the sarcolemma. The absolute value of the skeletal muscle ΔpHi/ΔT coefficient was greater than the ΔpHpl/ΔT coefficient (0.009 versus 0.003 units °C−1, respectively). In terms of our protein titration hypothesis, the higher absolute ΔpH/ΔT coefficient in skeletal muscle resulted from its higher β_NB (e.g. 19.8 mmol kg−1 fresh mass pH unit−1 in B. marinus skeletal muscle; Pörtner, 1990) compared with β_NB for extracellular fluid (approximately 8 mmol kg−1 pH unit−1 in C. constrictor). This higher buffering capacity is, however, partly offset by a lower absolute ΔpK/ΔT coefficient. In C. constrictor, the passive ΔpH/ΔT coefficient (and hence ΔpK/ΔT) is −0.013 pH units °C−1 compared with a plasma passive ΔpH/ΔT coefficient of approximately −0.019 pH units °C−1 (Reeves, 1976). The result in skeletal muscle, relative to extracellular fluid, is that when cooled the passive component elevates pHi less but then the superimposed CO₂ retention lowers pHi very little so that the net ΔpHi/ΔT coefficient is greater than the ΔpHpl/ΔT coefficient.

In the discussion above, Fig. 3 was viewed as a whole-body nonbicarbonate buffer curve. Support for this conclusion in C. constrictor comes from the nearly identical value for the change in plasma [CO₂] per unit change in plasma pH (−8.3 mmol l−1 pH −1; Table 3) and the Δ[CO₂]_WB/ΔpHpl coefficient in Fig. 3 (−8.24 mmol l−1 pH unit−1). The Δ[CO₂]_pl/ΔT in C. constrictor was not associated with ion exchange, but was instead apparently the result of the change in pK of the protein buffers (physicochemical buffering) and the change in (ventilatory regulation). The same was also apparently true for the intracellular fluid of skeletal muscle. When cooled, reptiles and amphibians increase their air convection requirement and thus lower blood (Ultsch and Jackson, 1996). CO₂ is lipid-soluble, it readily diffuses across cell membranes, and intracellular fluid decreases in a parallel fashion. The Δ[CO₂]/ΔT coefficients of skeletal muscle in C. constrictor and in R. catesbeiana parallel the Δ[CO₂]/ΔT coefficients of plasma ([CO₂]_pl is independent of temperature in R. catesbeiana; Reeves, 1972). Also, changes in extracellular pH appear to correspond to changes in whole-body pH (Bickler, 1982; Cameron and Kormanik, 1982). Consequently, it is reasonable that the Δ[CO]_2WB/ΔpHpl coefficient of Fig. 3 is similar in magnitude to the Δ[CO₂]pl/ΔpHpl coefficient measured in C. constrictor (Table 3). Alternatively, the close agreement between the slope of Fig. 3 and β_NB estimated from Table 3 may simply be coincidental.

The absence of changes in strong ion levels refers to steady-state conditions, i.e. after the change in whole-body CO₂ stores has been completed. Presumably, the ΔpK occurs virtually instantaneously with the ΔT so that, during the nonsteady-state period, there are transient adjustments in plasma and intracellular fluid strong ion levels, such as the rise in plasma lactate concentration observed in C. constrictor during cooling (Stinner et al. 1996). However, when comparing steady-state conditions, the ΔpK/ΔT accounts for CO₂ retention in plasma and cells, despite hyperventilation and a lowering of with cooling. According to this hypothesis, individual differences in the Δ[CO₂]/ΔT coefficients among tissues, as found in B. marinus (Stinner et al. 1994a), result from differences in β_NB and/or differences in (ΔpK−ΔpH).

In summary, on the basis of the findings of this study, titration of proteins with cooling occurs because the rising air convection requirement does not lower sufficiently to maintain a constant (ΔpK−ΔpH)/ΔT, i.e −ΔpK is greater than −ΔpH. C. constrictor at 10 °C compared with 30 °C maintains a lower by hyperventilating, but [CO₂] is higher (Tables 3, 5) because of the thermally induced upward shift in pK. This protein titration hypothesis does not support general theories of acid–base regulation in ectotherms that are based upon preserving the same net protein charge irrespective of body temperature (i.e. constant relative alkalinity, alphastat and z-stat; Austin et al. 1927; Howell et al. 1970; Reeves, 1972; Cameron, 1989). In the present study, only bullfrogs appear to conform closely to the alphastat model, which predicts constant tissue CO₂ content. What effect, if any, CO₂ retention and changes in protein charge have upon enzyme activities and the physiology of these animals is unknown (Heisler, 1986b). There is a growing body of evidence that CO₂ retention (acidification) and low (see Table 3) are potent metabolic inhibitors in ectotherms (Boutilier et al. 1997; Guppy et al. 1994). Furthermore, mammalian hibernators exhibit a transient reduction in RE when entering torpor and a transient elevation in RE during arousal (Nestler, 1990). C. constrictor is native to all 48 contiguous North American states and extends into southern Canada and northern Mexico. Black racer snakes have a wide temperature range and are even somewhat active in hibernacula below 5 °C (Stinner, 1987; Sexton and Hunt, 1980). By contrast, bullfrogs (R. catesbeiana) show minimal changes in whole-body CO₂ stores, but they also occupy a wide range in North America and are active at near-freezing temperatures in ice-covered ponds during winter (Stinner et al. 1994b). The possible relationship between CO₂ retention, protein charge and behavior is an important topic that deserves careful study in the future.

The authors wish to thank Dr H. O. Pörtner for providing encouragement and valuable technical advice on the measurement of intracellular pH.