ATP-induced reverse temperature effect in isohemoglobins from the endothermic porbeagle shark (Lamna nasus).

The evolutionary convergence of endothermic tunas and lamnid sharks is unique. Their heat exchanger-mediated endothermy represents an interesting example of the evolutionary pressure associated with this specific characteristic. To assess the implications of endothermy for gas transport and the possible contribution of hemoglobin (Hb), we investigated the effect of temperature on the oxygen equilibria of purified isohemoglobin components V and III from the porbeagle shark (Lamna nasus). In the absence of ATP the effect of temperature on oxygen affinity is normal in both Hb III (P50 = 0.9 and 2.2 torr at 10 and 26 degrees C, respectively) and Hb V (P50 = 1.5 and 2.5 torr at 10 and 26 degrees C, respectively). In the presence of this effector P50 decreases with increasing temperature in both components (P50 at 10 and 26 degrees C = 9.9 and 8.4 torr (Hb III), respectively, and 9.6 and 7.4 torr (Hb V), respectively. The reverse temperature effect in the presence of ATP will reduce the risk of oxygen loss from the arterial to the venous blood by lowering the oxygen tension gradient between the blood vessels. The mechanism behind the reverse temperature effect resembles that found in the bluefin tuna (Thunnus thynnus), an endothermic teleost, thus evidencing further convergent evolution.

True tunas and lamnid sharks are the only fish that have developed endothermy. The fact that this characteristic has evolved independently in two such different families (Scombridae and Lamnidae, respectively) makes these species choice subjects for comparison of the traits associated with endothermy (1)(2)(3)(4). Endothermic fishes maintain body temperatures of up to 20°C above ambient water temperatures, depending on species, water temperature, and activity level (5)(6)(7). The consequences of high core body temperatures for gas transport, ion balance, oxygen consumption, aerobic capacity, and muscle performance have been the subject of extensive studies in endothermic tuna species (for review, see Refs. 8 -11); however, little is known about these relationships in elasmobranchs. The porbeagle shark (Lamna nasus) has a core body temperature of 8 -10°C above ambient water temperature (12) and has one of the highest body temperatures among the Lamnidae (13)(14)(15). As in tunas its endothermy is brought about by a number of countercurrent heat exchangers, rete mirabilia, placed in series with the heat-producing organ (16 -18). The lateral heat exchangers are placed between the surroundings and the internally located red muscle, and the major blood supply to the red muscle is directed through these via large cutaneous arteries and veins. By minimizing the heat loss via the blood, metabolically produced heat is retained within the active organ (5,6).
One consequence of heat exchanger-mediated endothermy is that factors other than heat may be exchanged. The smaller the arteries and veins constituting the heat exchanger are, the more efficient the heat retention and the warmer the fish can be relative to the surrounding water. However, decreased dimensions of the heat exchanger increase the risk of oxygen diffusing from the cold arterial blood to the warm venous blood.
Already in 1960 Rossi-Fanelli and Antonini (19) showed that temperature has virtually no effect on oxygen binding of Rstate crystalline Hb 1 from the endothermic bluefin tuna (Thunnus thynnus) at pH values between 6.45 and 8.7. Further studies revealed that although temperature does not influence P 50 in the 10 -30°C temperature range, it does alter the shape of the oxygen equilibrium curve (20,21), resulting from a normal temperature effect at low oxygen saturation and a reversed one at higher saturations.
The heat exchanger-mediated endothermy represents a curious incidence of convergent evolution. In this context it is interesting to investigate whether endothermy in the porbeagle shark is associated with adaptations in the temperature sensitivities of Hb, as documented in Hb I of bluefin tuna (19 -21). This paper reports the effects of temperature, ATP, and pH on isolated Hb components III and V of the porbeagle shark. Previous studies on hemolysate indicate that the oxygen affinity of porbeagle shark blood is temperature-insensitive (22,23); however, the allosteric mechanism behind the temperature effect remains to be investigated in purified isohemoglobins. Moreover the contribution of organic phosphates needs to be investigated in light of the observation that ATP induces a reverse temperature effect in striped marlin (Tetrapturus audax) Hb (24).

EXPERIMENTAL PROCEDURES
Heparinized blood from a porbeagle shark (L. nasus) was supplied from North Sea fishermen. Red blood cells were lysed by addition of three volumes of 0.1 M Tris buffer, pH 7.00 (25°C). Following centrifugation for 15 min at 14,000 rpm to remove cell debris, the Hb stock solution was stored at Ϫ80°C.
Three Hb components (Hb V, IV, and III) accounted by far for most of the Hb (Fig. 1). Their isoelectric points at 16°C were 7.58, 7.62, and * This work was supported by the Oticon Foundation and the Danish Natural Science Research Council. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
‡ To whom correspondence should be addressed. 7.68, respectively. Given the small differences in isoelectric points and the fact that fish Hbs with similar electrophoretic mobility exhibit basically similar functional properties (25), further investigations were focused on Hb V and III.
Oxygen binding equilibria were generated at 10, 16, 21, and 26°C in the absence or presence of ATP (ATP/Hb 4 ratio, Ͼ30) using a modified gas-diffusion chamber allowing stepwise increments of oxygen partial pressure (26). The chamber was coupled to two cascaded Wösthoff pumps (type M201a-f and M301) for admixture of atmospheric air, O 2 , and pure N 2 (99.998%). Based on close similarity of the temperature sensitivities of Hbs III and V, analysis of the interactive effects of pH in the absence and presence of ATP was limited to Hb III.
Data Analysis-Extended Hill plots were analyzed using the MWC (two-state) model (27) where S denotes saturation, P the partial pressure of oxygen, L the allosteric constant, K T and K R the association constants for the low affinity (T, tense) and the high affinity (R, relaxed) forms, respectively, and q the number of interacting binding sites. Based on the results of Andersen et al. (23), who performed ultracentrifugation on hemolysate of porbeagle shark blood and found the slope of the absorbance versus radius squared to be consistent with a homogenous solution of tetrameric hemoglobins, the Hb was assumed tetrameric, and a fixed value of q ϭ 4 was imposed at all times. Non-linear least squares curve fitting was performed using the software package Mathematica® (Wolfram Research Inc., Cambridge, MA) employing the Levenberg-Marquardt method. Estimates of the standard errors of the fitted parameters were obtained from the diagonal elements of the curvature matrix associated with the fit (26). To minimize the errors introduced by incomplete saturation or desaturation, when equilibrating with pure oxygen or nitrogen, respectively, the absorbance values at zero and full saturation were extrapolated from the data in the fitting procedure as described by Fago et al. (26). P 50 , the oxygen tension at half-saturation of the Hb, was calculated as the PO 2 value at log(S/(1 Ϫ S)) ϭ 0, and n 50 was calculated as Ѩlog(S/(1 Ϫ S))/ѨlogPO 2 at that PO 2 . The median oxygen tension P m was calculated from where c ϭ K T /K R (28,29). The free energy of heme-heme interaction ⌬G was calculated according to Wyman et al. (30).
The difference in bond energies in the R and T states between the absence and presence of ATP was calculated as ⌬G ϭ RTln The heat of oxygenation was calculated according to the following relationship where R is the gas constant, X is P 50 , 1/L, 1/K T , or 1/K R , and T is the absolute temperature. ⌬H app is ⌬H at P 50 and includes the heat of release of effectors and solution of ligands and the heat of oxygen binding.

Effects of Temperature and ATP-Extended
Hill plots of the oxygen equilibria of Hbs III and V at pH 7.3 and at four temperatures (10, 16, 21, and 26°C) each in the presence or absence of ATP are shown (Fig. 2). The MWC model was successfully fitted to the data. In all instances P 50 of the fitted curves was similar to P m , and the fitted curves were symmetrical around P 50 .
Both Hbs exhibit high intrinsic oxygen affinities (low P 50 values) (Table I) and small, normal temperature effects at half-saturation (⌬H ϭ Ϫ40.1 and Ϫ20.7 kJ⅐mol Ϫ1 in Hbs III and V, respectively) at pH 7.3 (Table II). Increasing temperature from 10 to 26°C increases P 50 from 0.9 to 2.2 torr (Hb III) and from 1.5 to 2.5 torr (Hb V) ( Table I). The addition of ATP reduces the oxygen affinity in Hbs III and V, but less so at higher temperatures, resulting in a small reverse temperature effect in both Hbs (⌬H ϭ 12.2 and 11.6 kJ⅐mol Ϫ1 in Hbs III and V, respectively) (Table II). Fig. 3 shows the temperature dependence of P 50 (the van't Hoff plot) and n 50 with and without ATP present. A reverse temperature effect in the presence of ATP is seen. ATP additionally increases cooperativity expressed as n 50 (Fig. 3, lower panel), particularly at high temperature. As evident, Hb III and V show very similar oxygenation characteristics.
Mechanisms of the Temperature Effect-The MWC parameters L, K R , and K T provide a mechanistic basis for the effects of temperature on the Hbs (Fig. 4 and Table I). At increasing temperature L decreases, indicating a destabilization of the T (tense) state relative to the R (relaxed) state (Fig. 4a). ATP stabilizes the T state (increases L) but only slightly affects the temperature sensitivity.
Increasing temperature decreases oxygen affinity of the R and T states in Hb III and V, but more so in the R state (Fig. 4,  b and c). ATP reduces the oxygen affinities of both the T and R states and their temperature sensitivities, making K T of both Hbs insensitive to temperature change, whereas K R is still somewhat reduced by increasing temperature. The effect of ATP on K R decreases with increasing temperatures, indicating that ATP also binds to the R state, particularly at low temperatures. The T state becomes less stable at high temperatures (L decreases), and the overall oxygen binding to the T state accordingly decreases. The reverse temperature effect in the presence of ATP thus stems from the combined effects of the reduction of L with increasing temperature and reduced temperature sensitivity of the T state and especially of the R state. These effects are reflected by the temperature invariance of the lower asymptotes of the extended Hill plots in Fig. 2 (constant K T values) and a change in the slope of the curve around log[S/(1 Ϫ S)] ϭ 0 as L falls at higher temperatures. The change in K R is evident from the greater stability of the upper asymptotes of the curves in the presence of ATP than in its absence.
Free Energies and Enthalpies of Reaction-In the absence of ATP the free energy of heme-heme interaction, ⌬G, falls with increasing temperature (Table I). By stabilizing the T state, as evidenced by the temperature invariance of K T and the increase in L (Fig. 4), ATP increases the free energy of cooperativity but does so more at high than at low temperature, thus resulting in a low ATP sensitivity at high temperature. The change in ⌬G mirrors the effect of temperature on L and n 50 and reflects the ATP-induced increased cooperativity.
In both Hb III and V, ATP increases the apparent (overall) heat of oxygenation, ⌬H app (⌬H at P 50 ), which becomes positive (Table II). This effect is consistent with the elevation of ⌬H L , ⌬H T , and ⌬H R by ATP (Table II).
The displacement of the lower asymptote of the Hill plots in the presence of ATP reflects the additional bond energies that stabilize the T state in the presence of the polyanionic phosphate anion (Fig. 2). The ATP-induced bond energy differences are greater in the T state than in the R state (4.8 -5.0 kJ⅐(mol    Table III). Additionally, the bond energy difference decreases with increasing temperature in both the T and R states. The difference in bond energy, ⌬G T and ⌬G R , with and without ATP is presented in Table III. Effect of Changes in pH-Hill plots of oxygen equilibria of Hb III at 10 and 26°C with and without ATP and at different pH values were fitted using the MWC model (Figs. 5 and 6). The pH sensitivities of P 50 and n 50 are shown in Fig. 7. At 10°C Hb III shows a reverse Bohr effect between pH 8.3 and 7.5 ( ϳ 0.5) that becomes normal at pH Ͻ7.5 ( ϳ Ϫ0.6). At 26°C the phosphate-free Hb show no Bohr effect in the experimental range of pH values. In the presence of ATP the Bohr effect becomes normal at both temperatures. Noticeably, the reverse temperature dependence with ATP present is marked at low pH and phases out at high pH (Fig. 7). The Bohr effect at pH ϳ7.0 -7.3 in the presence of ATP is larger at 10°C ( ϳ Ϫ0.76) than at 26°C ( ϳ Ϫ0.3).
The cooperativity coefficient n 50 is negatively affected by lowering pH at low temperature, whereas it is positively affected at high temperature (Fig. 7) and invariably higher in the presence of ATP than in phosphate-free solution. The n 50 values at 10°C exceed those at 26°C at all pH values when ATP is absent and above pH 7.3 with ATP.
Mechanisms of the Bohr Effect-The allosteric control mechanisms of the Bohr effect are illustrated in Fig. 8. At 10°C without ATP L decreases from Ͼ10 4 at pH 8.0 to ϳ10 2 at a pH of 7.6, indicating a destabilization of the T state at low pH. The reverse Bohr effect at 10°C in the absence of ATP described previously concurs with the large fall in L as pH is lowered, whereas K T and K R remain unchanged (Fig. 8). With ATP present this effect is shifted to a lower pH (6.7); the pH effect becomes normal, K T decreases, and K R drops drastically as pH falls below 7.5. At 26°C the effect of pH on L, K T , and K R is negligible in the absence of ATP. At 26°C ATP elevates L by stabilizing the T state, in part because of a stabilizing effect of additional proton binding to the T state, its effect being larger at low pH. Concurrently, K T becomes lower and slightly de-creases at lower pH values, and K R remains low and stable irrespective of the presence ATP (Fig. 8).
Free Energies of Reaction-The free energy of heme-heme interaction (⌬G) is slightly higher at 10°C than at 26°C over the entire pH range and in the presence of ATP than in stripped Hb (Table IV). In the absence of ATP, ⌬G varies only  Table I.  (Table IV). DISCUSSION The MWC model, which assumes functional homogeneity between the subunits comprising the tetrameric Hb, successfully describes the oxygen binding characteristics of porbeagle shark Hb. This indicates either that the chain heterogeneity found in porbeagle shark hemolysate (23) is not found in the purified Hb components investigated in this study or that it does not influence oxygen binding.
The reverse temperature effect on isolated Hb III and V from porbeagle shark resembles that in bluefin tuna Hb I in phosphate buffer (21). In Hbs from both species low temperature stabilizes the T state, and a reverse temperature effect on K R is seen at low pH. The difference between the two species is qualitative. In the bluefin tuna the stabilization of the T state at low temperature (high L) abolishes the T 3 R transition.
Therefore, oxygen binds in the T state even at high oxygen saturation levels, revealing the presence of two functionally distinct subunits in the T state (21). In porbeagle shark the T 3 R transition does occur, and we find no evidence for differences in the two subunits comprising the Hb. Our results thus agree with those of Andersen and co-workers (23), who found a reverse temperature effect in hemolysate in the 5-15°C range and a decrease in L with rising temperature. Thus it appears that the evolutionary convergence of endothermic tunas and lamnid sharks includes changes in the functional mechanisms of their main Hb components.
In porbeagle shark Hb the reverse temperature effect is only seen at pH values below 7.5 and in the presence of ATP (Fig. 7). We find that the reverse temperature effect is caused largely by an increase in the allosteric constant, L, with decreasing temperature (Fig. 4). Although acting in the opposite direction, the concurrent increase in K R at pH 7.3 in the presence of ATP is insufficient to counteract the effect of L. At lower pH values the temperature effect on K R becomes reversed, and this magnifies the overall reverse temperature effect. The effect of ATP on K R is most prominent at low pH, where it increases the magnitude of the reverse temperature effect, and more or less vanishes at pH values Ն7.7 (Fig. 8). The lack of an effect of ATP at 26°C indicates that no ATP is allosterically bound to the oxyhemoglobin at this temperature. The bond energy per salt bridge is 4 -8 kJ⅐mol Ϫ1 (31), which suggests that an additional salt bridge per heme group constrains the T state when ATP is present (Table III). The small changes in the differential bond energy in the R state further indicate the preferential binding of ATP to the T state relative to the R state, although additional weaker bonds may be implicated, particularly at high temperature, where the difference bond energy is below 4 kJ⅐mol Ϫ1 . At pH 7.3 (which likely falls in the physiological intraerythrocytic range) the overall ⌬H app is negative (oxygenation is exothermic) in the absence of ATP but becomes positive (oxygenation is endothermic) with ATP (Table II). The increase is due to the positive heat of ATP and proton release from the Hb during the T 3 R transition. The fact that ⌬H T is much less negative (or positive) than ⌬H R is due to a greater ATP and proton release in the T state. Thus the addition of ATP obliterates the temperature effect in the T state, whereas it persists in the R state at high pH and is reversed at pH below 7.3 (Figs. 2, 4, and 7 and Table I). In bluefin tuna the temperature effect is normal in the T state (⌬H 1 is Ϫ8.0 kJ⅐mol Ϫ1 , and ⌬H 2 is Ϫ11.0 kJ⅐mol Ϫ1 ) and reversed in the R state (⌬H 3 and ⌬H 4 are 10.4 and 21.3 kJ⅐mol Ϫ1 , respectively) at pH 7.0 (21). In contrast, in trout Hb I, the effect of temperature on CO binding shows opposite dependence on ligand saturation; ⌬H T is 6 kJ⅐mol Ϫ1 , and ⌬H R is Ϫ11.3 kJ⅐mol Ϫ1 (21,30). In tench (Tinca tinca) the addition of ATP increases ⌬H T (Ϫ55.8 to Ϫ30.9 kJ⅐mol Ϫ1 ) and slightly decreases ⌬H R (Ϫ60.5 to Ϫ68.0 kJ⅐mol Ϫ1 ) at pH 7.35, indicating that the proton release occurs gradually and that ATP does not bind to the R state (32).
Our results illustrate the complex relationship between pH and ⌬H app (Fig. 7). In the presence of ATP at pH values below 7.7, ⌬H app becomes increasingly positive with falling pH because of an increasing ⌬H L that eventually becomes positive at pH Ͻ6.7 and an increasing ⌬H R that becomes positive at pH Ͻ7.1 (Fig. 8, a and c). In spiny dogfish (Squalus acanthias), the ⌬H app of stripped Hb also is pH-dependent with a value of Ϫ35 kJ⅐mol Ϫ1 at pH 7.0 and Ϫ44 kJ⅐mol Ϫ1 at pH 7.9 (33).
The free energies of cooperativity, ⌬G, at 10 and 26°C with ATP (Hb III) are comparable to those found in tench (6.1 kJ⅐(mol of Hb) Ϫ1 ) (34). Human Hb has a higher ⌬G at 25°C (8.8 kJ⅐(mol of Hb) Ϫ1 at pH 7.4), close to values found at 10°C with ATP at high pH (Hb III, Table IV) and at 16°C with ATP (Hb V, Table I) (35).
The Bohr effect in porbeagle shark Hb at 10°C with ATP ( ϳ Ϫ0.76) is large and similar to values for other active fishes. Albacore tuna (Thunnus alalunga) and striped marlin   39), where it also disappears when ATP is present. This suggests that upon oxygenation alkaline Bohr groups change pK value in the reverse direction from normal. Possible Implications for Oxygen Transport-Allosteric cofactors such as organic phosphates and protons lower the intrinsically high oxygen affinity of Hb to values that ensure oxygen unloading at oxygen tensions that are sufficiently high for physiological needs. Compared with small Bohr effects reported in sluggish elasmobranch species (40), active ones, like the porbeagle and mako sharks, have substantial Bohr effects at low temperature and in the presence of ATP (23). The pH insensitivity on K R has been interpreted as advantageous under changing environmental oxygen tensions (34). By maintaining a constant high affinity for oxygen in the R state, loading is protected during hypoxia and during exercise when the blood may become acidotic. In porbeagle shark K R falls at low temperature when pH is low, an effect not present at high temperature when environmental, dissolved oxygen concentration can be assumed to be lower and loading may become compromised.
The concentration of ATP and other organic phosphates in nucleated fish red cells decreases under conditions of decreased oxygen availability, forming a regulatory mechanism that favors oxygen loading through the resulting increase in oxygen affinity. During exercise the affinity gain may, however, be offset by the decreased blood pH (Bohr effect). In elasmobranchs the presence of high urea levels may hinder oxygen affinity modulation by dampening the ATP effect as seen in stripped hemolysate of spiny dogfish (33).
The presence of countercurrent heat exchangers in the porbeagle shark imposes a risk of oxygen loss from the arterial to the venous blood. However, as in the endothermic bluefin tuna, the presence of a reverse temperature effect in isolated Hb indicates that this risk is reduced. The increased oxygen affinity with rising temperature reduces the oxygen tension gradient between the arterial and venous blood. This adaptation to an efficient heat conservation mediated by heat exchangers may only be necessary in cases in which the heat exchangers are very efficient. As such the bluefin tuna and the porbeagle shark are the among the warmest species in their distantly related families, respectively, each representing very efficient countercurrent heat exchangers (6,(13)(14)(15)(16)(41)(42)(43).