Temperature independence of haemoglobin–oxygen affinity in smalleye Pacific opah (Lampris incognitus) and swordfish (Xiphias gladius)

The swordfish (Xiphias gladius) and the smalleye Pacific opah (Lampris incognitus) are both large, mesopelagic species that can conserve metabolic heat and maintain the temperature of select tissues or organs warmer than the surrounding water (i.e. regional heterothermy). Opah can maintain the cranial region and entire body core, including the heart, warmer than ambient water, which is unique among fishes because the hearts of all other fishes remain near ambient temperature (Runcie et al., 2009; Wegner et al., 2015). In swordfish, thermogenic extraocular muscles heat the eyes and brain (Carey, 1982b), and recent work shows that their red slow-twitch swimming muscles can maintain some degree of temperature elevation relative to the environment (Carey, 1990; Carey and Gibson, 1987; Stoehr et al., 2018, 2020). Only a few other fishes are also capable of heating their eyes and brain [i.e. billfishes (family Istiophoridae), tunas (family Scombridae), lamnid sharks (family Lamnidae) and the butterfly mackerel (Gasterochisma melampus)] or the red swimming muscles [i.e. tunas, lamnids and the common thresher shark (Alopias vulpinus)] warmer than ambient water (Bernal and Sepulveda, 2005; Block, 1986; Carey and Teal, 1966, 1969a,b; Carey et al., 1985; Sepulveda et al., 2008). The ability to maintain a warm tissue temperature while residing in cold water provides regionally heterothermic fishes with certain physiological advantages over their ectothermic competitors and ectothermic prey, such as greater power output by the warm swimming muscles, and superior vision as a result of enhanced temporal resolution in the heated retinas (Altringham and Block, 1997; Fritsches et al., 2005; Stoehr et al., 2020).

Regionally heterothermic fishes are able to conserve metabolic heat in select body regions via vascular specializations (retia mirabilia) that form countercurrent heat exchangers, enabling cold arterial blood flowing from the gills to be warmed by venous blood (except in the opah) that is returning to the gills (Bernal et al., 2001; Block, 1986; Carey, 1982a; Carey and Teal, 1966,a; Carey et al., 1985; Fudge and Stevens, 1996; Patterson et al., 2011; Runcie et al., 2009; Wegner et al., 2015). This heat transfer diminishes convective heat loss, as the venous blood cools to near ambient temperature before reaching the gill lamellae, where blood and water reach thermal equilibrium (Brill et al., 1994; Stevens and Sutterlin, 1976). Opah and swordfish have heat exchanging retia in the orbital circulation (Carey, 1982b; Runcie et al., 2009), and warmed blood perfuses the opah's entire body, including the heart, as a result of a series of unique afferent–efferent arterial heat exchanging retia within the gill arches, each of which is relatively thick and insulated with fat (Wegner et al., 2015). The swordfish also has putative heat exchanging retia that supply blood to the medially located red muscle; however, little is known about the degree of control swordfish have over red muscle temperature fluctuation during extended exposure to cold environmental temperatures (Carey, 1990; Stoehr et al., 2018). A consequence of efficient vascular heat exchange is that blood O₂ uptake at the gills occurs at quite different temperatures to O₂ offloading in the tissues that are thermally isolated by a rete, and these internal temperature gradients are exacerbated when opah and swordfish dive into cold water below the thermocline (Carey, 1990; Wegner et al., 2015).

During daily sojourns between the warm upper mixed layer above the thermocline (e.g. >20°C) and deeper colder waters below the thermocline (e.g. <5°C), swordfish may experience large and rapid changes in water temperature in addition to possibly low environmental oxygen levels near the oxygen minimum layer (Carey, 1990; Carey and Robinson, 1981; Dewar et al., 2011; Sepulveda et al., 2010), whereas below the thermocline, swordfish cranial temperature is relatively constant and can be elevated as much as 12°C above ambient water temperature (Carey, 1990). Although opah occasionally swim into the warm upper mixed layer, they tend to remain in cooler water below the thermocline where cranial temperatures are relatively constant and elevated at least 6°C above the surrounding water, while body and heart temperature are elevated at least 3–5°C above the surrounding water (Wegner et al., 2015). Therefore, blood O₂ uptake at the gills must occur over the range of environmental temperatures encountered by opah and swordfish, but the blood must also transport O₂ over steep internal temperature gradients, and O₂ offloading must occur from the lowest temperature (e.g. at the gills) to the highest body temperatures (e.g. at the ocular muscles). Moreover, as blood flows through a heat exchanging rete, it is subjected to what has been described as ‘closed-system’ temperature changes as the blood P_O2 and P_CO2 can vary with temperature, but the content of blood gases remains essentially constant as a result of the size and thickness of rete vessels, diminishing diffusion of gases out of the blood (Brill and Bushnell, 1991; Cech et al., 1984; Stevens et al., 1974). The O₂ affinity of the blood can, thus, be directly affected by closed-system temperature changes as well as by variable environmental and body temperatures.

The O₂ affinity of most jawed vertebrate haemoglobins (Hbs) typically decreases with increasing temperature. This is because the heat of O₂ binding to the haem groups (⁠⁠) is intrinsically exothermic, so the overall enthalpy of oxygenation (ΔH′) is usually also exothermic (i.e. numerically negative). However, temperature-independent Hb–O₂ affinity and even reverse temperature dependence (i.e. increasing temperature increases Hb–O₂ affinity) have been reported in several species of ectothermic and regionally heterothermic fishes, including tuna, billfish and lamnid sharks (Andersen et al., 1973; Barlow et al., 2017; Bernal et al., 2018; Brill and Bushnell, 1991, 2006; Carey and Gibson, 1977, 1983; Cech et al., 1984; Clark et al., 2008, 2010; Graham, 1973; Larsen et al., 2003; Lilly et al., 2015; Lowe et al., 2000; Sharp, 1975; Weber et al., 2010). Among fishes, reductions and reversals in the temperature dependence of Hb–O₂ affinity appear to stem predominantly from oxygenation-linked dissociation of allosteric effectors, such as hydrogen ions (i.e. Bohr protons) and organic phosphates (e.g. adenosine triphosphate, ATP), which contribute endothermically to ΔH′, reducing the overall effect of temperature on Hb–O₂ affinity (Carey and Gibson, 1977; Dickinson and Gibson, 1981; Ikeda-Saito et al., 1983; Larsen et al., 2003; Morris and Gibson, 1982). The main effector of tuna Hb is protons, whereas ATP is the primary effector of lamnid shark Hb, and in billfishes it is pH-dependent binding of ATP (Ikeda-Saito et al., 1983; Larsen et al., 2003; Weber et al., 2010). In previous experiments on swordfish Hb, increasing temperature considerably decreased Hb–O₂ affinity (i.e. a normal temperature dependence), although it is not clear whether those experiments were conducted on whole blood, erythrolysates or stripped haemolysates (Andersen et al., 1973). Therefore, it is not known whether swordfish Hb exhibits an ATP-induced temperature independence in a manner like that of the closely related istiophorid billfishes. We are not aware of any O₂ equilibria studies on Hb or blood from any of the opah species.

Several hypotheses have been proposed for the functional significance of reduced and reverse temperature-dependent Hb–O₂ affinity in regionally heterothermic fishes. In most, the importance is attributed to either a decreased influence of temperature on blood O₂ transport across large internal temperature gradients or the energetic savings of an increased ΔH′ as less energy would be required to bind and unload O₂ (Carey and Gibson, 1977; Clark et al., 2008; Giardina et al., 1989; Graham, 1973; Weber and Campbell, 2011; Weber et al., 2010). Information on the effect of temperature on Hb–O₂ affinity in understudied regionally heterothermic fishes, such as swordfish and opah, can further broaden our understanding of the potential functional significance of reduced and reverse temperature-dependent Hb–O₂ affinity in regionally heterothermic fishes. To that end, we investigated the effect of temperature on Hb–O₂ affinity in blood and haemolysates from the swordfish and the smalleye Pacific opah.

As reductions in the temperature sensitivity of Hb have been reported in other regional heterotherms, including billfishes, we hypothesized that both swordfish and opah Hb would have a similar reduced temperature sensitivity. Furthermore, as swordfish and istiophorid billfishes are closely related and their specialized extraocular muscles and heat-exchanging retia are very similar (Block, 1986, 1991), we expected that swordfish Hb would exhibit an ATP-induced temperature independence like previously studied billfish Hbs. We assessed the temperature sensitivity of whole blood from these species by constructing oxygen equilibrium curves (OECs) and quantifying P₅₀ (the P_O2 at 50% Hb–O₂ saturation) at different temperatures, as well as by measuring blood P_O2 during closed-system temperature changes in an experimental system meant to mimic the temperature changes that the blood experiences in a heat exchanging rete. We also constructed OECs and determined P₅₀ in stripped haemolysates buffered at different pH levels, and in the absence and presence of ATP. Experiments were also conducted on haemolysates of Atlantic bluefin tuna that were opportunistically sampled, which allowed us to evaluate the enthalpic contributions of oxygenation-linked effector dissociation among the different lineages of regionally heterothermic teleosts. Collectively, this information gives insight into the evolution and functional significance of reduced and reverse temperature-dependent Hb–O₂ affinity among disparate lineages of regionally heterothermic teleosts.

All capture, handling and experimental procedures followed guidelines approved by the University of Massachusetts (animal care protocol no. 13-06), the California Department of Fish and Wildlife (Scientific Collection permit nos SC-2471, SC-12372) and the University of British Columbia (UBC) Animal Care Committee (animal care no. A11-0235 and A15-0266). All partial pressures and P₅₀ values are reported in mmHg (1 mmHg=0.133 kPa).

Swordfish, Xiphias gladius Linnaeus 1758 (n=7), were captured by deep-set buoy gear (Sepulveda et al., 2014), and opah, Lampris incognitus Underkoffler, Luers, Hyde and Craig 2018 (n=4), were captured by deep-set buoy gear or short set pelagic long-line in the coastal waters off Southern California (i.e. the Southern California Bight). Atlantic bluefin tuna, Thunnus thynnus Linnaeus 1758 (n=2), were captured by hook and line off Massachusetts (fork lengths are reported in Table 1). Blood was drawn by caudal puncture into heparinized syringes. Blood samples were kept on ice and shipped by courier to the Department of Zoology, UBC Vancouver campus, where experiments on whole blood were conducted within 1–4 days after the blood was collected. Preliminary experiments with swordfish blood (n=3) that were kept under refrigeration (4°C) for up to 6 days post-collection showed no changes in Hb concentration, haematocrit (Hct; the proportion of red blood cells in blood) and plasma pH, and there was no evidence of red blood cell (RBC) lysis (Morrison, 2020). However, during this time period, RBC intracellular nucleoside triphosphate (NTP) levels likely changed from levels typical of freshly sampled blood, possibly causing Hb–O₂ affinity to increase over time. Although we were not able to construct OECs on freshly drawn blood (i.e. <1 day post-collection), we found that whole-blood P₅₀ at 15°C and 3.8 mmHg CO₂ was relatively unchanged from 4 to 8 days post-collection in two swordfish (P₅₀ ranged between 24 and 30 mmHg), and blood P_50­ of a third swordfish was unchanged from 5 to 8 days post-collection (23–24 mmHg). Moreover, in a previous study of chub mackerel (Scomber japonicus) blood–O₂ affinity, experiments were also conducted at the UBC Vancouver campus on blood collected from mackerel captured off Southern California, and it was concluded that mackerel blood was viable for up to 6 days when stored at 4°C (Clark et al., 2010).

Immediately after blood samples arrived at UBC, Hb concentration and Hct were measured, and subsamples of blood were centrifuged to separate the plasma from the RBCs for measurement of plasma osmolality. The packed RBCs and remaining plasma were frozen at −80°C for determination of RBC intracellular ATP concentration and plasma lactate concentration. Whole-blood OECs were constructed by quantifying the relative Hb–O₂ saturation at a range of equilibration P_O2 values at two CO₂ levels (1.9 and 7.6 mmHg), and at 10 and 25°C for swordfish, and 10, 15 and 20°C for opah. Blood pH and P_O2 were measured in subsamples of blood equilibrated with gas mixes at each of the OEC temperature treatments. After completion of the whole-blood experiments, RBCs were separated from blood plasma by centrifugation. The RBCs were then thrice rinsed in ice-cold marine teleost saline (Hoar and Hickman, 1983) and frozen at −80°C for experiments on stripped haemolysates. OECs were constructed for stripped haemolysates in the presence and absence of effector ions at 10 and 25°C for swordfish, 10 and 20°C for opah, and 15 and 25°C for bluefin tuna.

Haemoglobin concentration, expressed as tetrameric Hb ([Hb₄], in mmol l⁻¹), was measured by the cyanmethaemoglobin method using Drabkin's reagent and a haem-based extinction coefficient of 11 mmol⁻¹ cm⁻¹ at a wavelength of 540 nm (Völkel and Berenbrink, 2000). Hct was measured as the percentage of packed RBCs relative to total blood volume after centrifuging samples in glass microcapillary tubes at approximately 13,000 RCF for 5 min. Mean corpuscular haemoglobin concentration (MCHC, in mmol l⁻¹) was calculated by dividing [Hb] by Hct. Plasma osmolality (mOsm kg⁻¹) was measured in 10 µl of undiluted plasma with a vapour pressure osmometer (VAPRO 5520, Wescor, Logan, UT, USA). ATP was assayed with an ATP colorimetric assay kit (MAK190, Sigma-Aldrich Co. LLC, St Louis, MO, USA), and plasma lactate was measured spectrophotometrically using the LDH-catalysed reaction converting lactate to pyruvate, where the reduction of NAD⁺ to NADH was measured at 340 nm (Bergmeyer et al., 1983).

Oxygen equilibria experiments were conducted at 10 and 25°C for swordfish, and 10, 15 and 20°C for opah. The coldest experimental temperature was 10°C for both species as this is close to the coldest water temperature regularly encountered by both swordfish and opah (Sepulveda et al., 2010; Wegner et al., 2015). The warmer experimental temperatures were chosen because 25°C is near to the warmest cranial and water temperatures for swordfish, and 15 and 20°C are near the warmest body and cranial/water temperatures, respectively, for opah (Carey, 1990; Sepulveda et al., 2010; Wegner et al., 2015). At each temperature treatment, experiments were conducted at two physiologically relevant CO₂ levels, low (1.9 mmHg) and high (7.6 mmHg), to quantify the Bohr coefficient (i.e. ΔP₅₀/ΔpH).

The relationship between Hb–O₂ saturation and P_O2 (i.e. an OEC) was assessed on 2–3 replicate samples using a custom microplate-based, parallel assay, multi-cuvette tonometry cell as described by Lilly et al. (2013). Cuvettes were formed by sandwiching blood samples (∼3 μl) between two sheets of low-density polyethylene (Glad^® ClingWrap) that were secured on an aluminium ring with two plastic O-rings, which were then placed in a gas-tight tonometry cell designed to fit into a SpectraMax 190 microplate reader (Molecular Devices, Sunnyvale, CA, USA). Optical density (OD) was measured every 20–30 s at 390 nm (near an isosbestic point between oxygenated and deoxygenated Hb, where OD is independent of Hb–O₂ saturation), and at 430 and 436 nm (near the peak absorption for deoxygenated Hb), wavelengths commonly used in thin-film optical methods for measuring Hb–O₂ binding (e.g. Clark et al., 2008; Reeves, 1980; Weber et al., 2010). Initially, blood was equilibrated with pure N₂ for a minimum of 30 min until OD at 430 and 436 nm was stable, which was assumed to indicate full Hb deoxygenation. After deoxygenation, the Hb–O₂ saturation was increased with at least nine stepwise increments of the O₂ tension, balanced with N₂, up to a P_O2 of 159.6 mmHg (i.e. approximate atmospheric P_O2 at sea level). In a tenth and final increment, the P_O2 was increased to 228 mmHg in the absence of CO₂ to achieve full Hb–O₂ saturation. OEC experiments lasted between 1 and 2 h. In preliminary experiments, absorption spectra were recorded following the last oxygenation step (i.e. full Hb–O₂ saturation), and these spectra showed no evident peak at 630 nm, an absorption maximum for metHb (e.g. Völkel and Berenbrink, 2000). Moreover, the OD at 630 nm was not unusually high compared with an absorption maxima for oxygenated Hb (OD630 nm/OD575 nm<0.04), and α/β peak OD ratios (OD575 nm/OD540 nm) were around 1.06, which are both typical of oxygenated Hb with no significant metHb fraction (Völkel and Berenbrink, 2000; Weber et al., 2010; Zijlstra and Buursma, 1997). Therefore, in the remainder of the experiments, we did not consider metHb formation to have significantly affected our Hb–O₂ saturation measurements, and full Hb–O₂ saturation was assumed at the final oxygenation step. Gas mixtures of O₂, CO₂ and N₂ were obtained using a Wösthoff DIGAMIX^® gas mixing pump (H. Wösthoff Messtechnik, Bochum, Germany). At each equilibration step, the difference in OD (ΔOD) between 390 nm and 430 nm or 436 nm (ΔOD=430 nm or 436 nm–390 nm) was used to calculate the fractional Hb–O₂ saturation ([Hb–O₂]/[Hb]) from the change in ΔOD from full deoxygenation, relative to that between full deoxygenation (pure N₂) and full oxygenation (P_O2 of 228 mmHg and no CO₂). OECs constructed using 430 nm were identical to those constructed using 436 nm.

Whole-blood pH was measured in approximately 500 μl of blood equilibrated for 1 h with either 1.9 mmHg or 7.6 mmHg CO₂ and a range of O₂ tensions between 7.6 mmHg and 159.6 mmHg (balanced with N₂), in rotating glass tonometers thermostatically set to 10, 15, 20, or 25°C. The gas mixtures were humidified at the experimental temperature prior to entering the tonometers. Blood was drawn into a gas-tight syringe pre-flushed with the gas mixture, and pH was measured by drawing the blood through a Microelectrodes 16-705 flow-thru pH electrode in combination with a 16-702 flow-thru reference electrode (Microelectrodes Inc., Bedford, NH, USA) thermostatically set to the experimental temperature.

To mimic the closed-system temperature changes that blood may experience in the arterioles and venules of a heat exchanging rete, approximately 500 μl blood samples equilibrated at 10, 15, 20 or 25°C were injected into a Radiometer E5046 P_O2 electrode thermostatically set at the equilibration temperature as well as another P_O2 electrode thermostatically set to a warmer or cooler experimental temperature according to Cech et al. (1984) and Brill and Bushnell (1991). Swordfish blood temperature was changed between 10 and 25°C, whereas opah blood temperature was changed either between 10 and 15°C or between 15 and 20°C. Although the blood was static within the electrode chamber, it was rapidly heated or cooled in a system where there is minimal exchange of gases and ions between the blood and another medium. Prior to injecting the blood, each P_O2 electrode was flushed with the experimental gas mixture to minimize electrode response time to the respective P_O2. Temperature-induced changes in P_O2 were monitored using Acqknowledge^® Data Acquisition Software (v.3.7.3, BIOPAC Systems, Inc.) by viewing traces of P_O2, and when it appeared that the traces had stabilized, the respective values of each were recorded.

Frozen and packed RBCs were placed on ice and put in a refrigerator (4°C) where they thawed over 24 h. Thawed RBCs were then mixed with an equal volume of cold 0.1 mmol l⁻¹ Hepes buffer (pH 8.0) and centrifuged at 10,000 RCF to remove cell debris. The resulting erythrolysates were stripped of endogenous ionic effectors by passage through mixed-bed ion exchange resin (Amberlite^® MB-20). MetHb (Hb⁺) levels were assessed by oxygenating 10–20 µl of the haemolysates in 1000 µl of 100 mmol l⁻¹ Hepes buffer (pH 7.4) that was bubbled with 100% O₂, and a spectral scan was made from 500 to 700 nm (i.e. an oxyHb spectrum). If there was an evident peak or unusually high absorbance at 630 nm, an absorption maximum for metHb (Völkel and Berenbrink, 2000; Zijlstra and Buursma, 1997), then the Hb⁺ was reduced by adding a molar excess of sodium dithionite to the haemolysate, followed by passage through mixed-bed ion exchange resin. If spectral scans of reduced oxyHb showed α/β peak OD ratios (OD575 nm/OD540 nm) around 1.06 (see above: Weber et al., 2010; Zijlstra and Buursma, 1997), and no evident peak at 630 nm relative to the OD at wavelengths either side of 630 nm (∼600–660 nm), then the haemolysates were concentrated with centrifugal filters (30 kDa exclusion size limit) and used for OEC experiments. Oxygen equilibria were determined in 100 mmol l⁻¹ Hepes buffer (pH 6.9–7.8 and no Cl⁻) at a Hb concentration of 0.6 mmol l⁻¹ in the absence and presence of saturating levels of ATP (ratio of the concentration of ATP/Hb=30). Although guanosine triphosphate (GTP), a potent allosteric effector of teleost Hbs, is also present in fish RBCs, ATP was used in these experiments as ATP concentrations are around sixfold higher than GTP concentrations in billfish RBCs (Weber et al., 2010). ATP wasadded from a 500 mmol l⁻¹ stock solution of ATP di-sodium salt (Sigma-Aldrich A7699) in water. OECs were generated at 10 and 20°C for opah haemolysates, 10 and 25°C for swordfish haemolysates, and 15 and 25° bluefin tuna haemolysates following the procedures described above, except without CO₂, and the final O₂ equilibration step (i.e. full saturation) was with 100% O₂. Experiments on bluefin tuna haemolysates were not conducted in the presence of ATP as others have reported that organic phosphates negligibly affect tuna Hb–O₂ affinity (Andersen et al., 1973; Ikeda-Saito et al., 1983). The pH of the haemolysate solutions was measured at the experimental temperature with a thermostatically controlled glass pH electrode (InLab^® Micro, Mettler-Toledo LLC, Columbus, OH, USA).

Blood parameters and fork length are summarized in Table 1. Whole-blood OECs (Figs 1 and 2) were successfully constructed for four opah and five swordfish, and the blood pH levels measured in the OEC experiments ranged from pH 7.23 to 8.04 for opah, and pH 7.33 to 7.94 for swordfish (Fig. 3). Whole-blood P₅₀ values at pH 7.7 (an approximation of arterial blood pH at 25°C from measurements reported from swimming yellowfin tuna; Korsmeyer et al., 1997), n_H values and Bohr coefficients are reported in Table 2. Measured whole-blood OECs are shown in Figs 1 and 2, the pH dependence of blood–O₂ affinity (P₅₀) is shown in Fig. 3, and modelled OECs at different temperatures and blood pH are presented in Fig. 4.

The aim of our analysis was to assess the effects of temperature and pH on P_S, the blood P_O2 corresponding to specific Hb–O₂ saturations (e.g. P₅₀). In opah blood, temperature was an important model factor for only the P₉₀ regression (χ²=10.101 d.f.=4, P=0.039), where a reverse temperature dependence was evident in the corresponding Bohr plot (additional Bohr plots are available in Fig. S1). At lower Hb–O₂ saturations, temperature did not significantly predict blood–O₂ affinity, which is evident by the overlapping data and regression lines presented in the Bohr plots (Fig. 3A). In swordfish blood, temperature was an important model factor from P₅₀ to P₉₅ (χ²=7.284–18.790, d.f.=2, P≤0.026), where increasing temperature decreased blood–O₂ affinity (Fig. 4E–H). The effect of temperature on swordfish blood–O₂ affinity showed a pH dependence, with temperature having a reduced effect on blood–O₂ affinity with declining pH (Fig. 4E–H), as is evident by the merging or crossing of the Bohr plot regression lines (Fig. 3C). The random intercept, id, accounted for 27–58% of the total variance for the swordfish regression models, and 28–80% of the total variance for the opah regression models.

OECs of both opah and swordfish blood exhibited considerable Bohr effects, evident as a right-shift (i.e. decreased blood–O₂ affinity) with declining pH (Fig. 4; additional modelled OECs are shown in Fig. S2). Opah blood pH was a significant predictor of logP_S from 40% to 95% saturation (i.e. P₄₀–P₉₅) at all temperatures (β≤−0.38, P≤0.025). From 10% to 30% saturation (i.e. P₁₀–P₃₀), blood pH was not a significant predictor of logP_S, and an interaction between pH and temperature was not a significant predictor of logP_S at any saturation. Swordfish blood pH was a significant predictor of logP_S­ at all Hb–O₂ saturation levels and at both temperatures (β≤−1.11, P≤0.036), but blood–O₂ affinity was more sensitive to pH changes at 10°C, as is evident by larger Bohr coefficients at 10°C than at 25°C (Table 2 and Fig. 3C). An interaction between pH and temperature was a significant predictor of logP_S for swordfish logP₅₀–logP₉₀ (β≥0.74, P≤0.049). The pH dependency of the effect of temperature on whole blood-O₂ affinity is evident in plots of ΔH'_WB (Fig. 5).

Opah n_H values ranged from 1.49 to 3.00 at 10°C, from 1.51 to 2.97 at 15°C, and from 1.69 to 2.49 at 20°C (Fig. 3B), and swordfish n_H values ranged from 1.06 to 2.11 at 10°C, and from 0.93 to 1.46 at 25°C (Fig. 3D). Model-predicted n_H values are presented in Table 2. Blood pH was not a significant predictor of n_H for either species, temperature was also not an important predictor of opah n_H values, but temperature was an important predictor of swordfish n_H values (β=−0.44, t_13.47=−4.21, P=0.0009), with higher values at 10°C than at 25°C.

The effects of closed-system temperature changes on blood P_O2 are shown in Fig. 6, with modelled temperature-induced changes in plasma P_O2 predicted by Henry's law (i.e. increasing temperature will increase P_O2 in a closed-system as a result of a reduction in plasma O₂ solubility and vice versa) using O₂ solubilities at different temperatures from Boutilier et al. (1984). Closed-system warming of opah blood generally decreased blood P_O2, presumably as a result of increased Hb–O₂ affinity with increasing temperature (i.e. a reverse temperature dependence). Closed-system cooling of opah blood tended to increase blood P_O2, but this effect was more variable. In contrast, closed-system temperature changes of swordfish blood changed blood P_O2 beyond the predicted temperature-induced change in plasma P_O2.

Haemolysate experiments were conducted on samples from three opah, six swordfish and the two bluefin tuna. Stripped haemolysate P₅₀ values at pH 7.4 (an approximation of RBC intracellular pH; Weber et al., 2010), Bohr coefficients and ΔH′ values are summarized in Table 3. At 10°C, stripped haemolysate P₅₀ of swordfish (1.8 mmHg) was lower than that of opah (5.9 mmHg), and at 25°C, swordfish P₅₀ (4.5 mmHg) was lower than that of bluefin tuna (13.7 mmHg). The addition of ATP increased haemolysate P₅₀ (i.e. decreased Hb–O₂ affinity) for both opah and swordfish (Fig. 7 and Table 3). The relationships between haemolysate pH and logP­₅₀, with associated n_H values, are presented in Fig. 7.

Opah stripped haemolysates in the absence of ATP showed a significant effect of pH on logP₅₀ (β=−0.527, P=0.005), and P₅₀ increased with increasing temperature (ΔHʹ=−15 kJ mol⁻¹), but logP₅₀ values at pH 7.4 were not significantly different between 10 and 25°C (overlapping 95% confidence intervals; Table 3) and temperature was not a significant predictor of logP₅₀ (β=0.083, P=0.211). In the presence of ATP, temperature was a significant predictor of logP₅₀ (β=−0.506, P<<0.001), but opah haemolysate P₅₀ exhibited a reverse temperature dependence (ΔHʹ=+84 kJ mol⁻¹) with significantly lower haemolysate P₅₀ (pH 7.4) values at 25°C than at 10°C (Table 3 and Fig. 7A). Opah Bohr coefficients were increased with the addition of ATP (Table 3). For opah stripped haemolysates, temperature and pH were not significant predictors of n_H, both in the absence and presence of ATP.

Swordfish stripped haemolysates in the absence of ATP showed no significant effect of pH on logP₅₀ (β=−0.381, P=0.700), as indicated by Bohr coefficients that were not significantly different from zero (Table 3). However, temperature was a significant predictor of logP₅₀ (β=0.389, P<<0.001), with haemolysate P₅₀ increasing with increasing temperature (ΔHʹ=−42 kJ mol⁻¹). In the presence of ATP, pH was a significant predictor of swordfish logP₅₀ (β=−1.180, P<<0.001), but temperature was not (β=0.073, P=0.061). The addition of ATP to swordfish stripped haemolysates significantly increased the Bohr coefficients at both 10 and 25°C (−1.32 at 10°C, and −1.06 at 25°C), and the effect of temperature on swordfish P₅₀ was reduced (ΔHʹ=−9 kJ mol⁻¹) compared with when ATP was absent (ΔHʹ=−42 kJ mol⁻¹; Table 3). For swordfish stripped haemolysates, pH was not a significant predictor of n_H, but temperature was a significant predictor of n_H, in both the absence (β=−0.804, P=0.012) and presence of ATP (β=0.339, P<<0.001).

For bluefin tuna stripped haemolysate (no ATP), pH was a significant predictor of logP₅₀ (β=−0.670, P<<0.001), as was temperature (β=−0.169, P=0.004). However, tuna haemolysate P₅₀ exhibited a reverse temperature dependence (ΔHʹ=+29 kJ mol⁻¹; Fig. 7E), with significantly lower P₅₀ (pH 7.4) values at 25°C than at 15°C (Table 3). Temperature was not a significant predictor of tuna stripped haemolysate n_H values, and although pH did not influence n_H­ at 15°C, pH was a significant predictor of n_H at 25°C (β=0.989, P=0.039).

The purpose of this study was to investigate the temperature dependence of Hb–O₂ affinity in two regionally heterothermic teleosts, the smalleye Pacific opah and the swordfish. As reductions in the temperature sensitivity of Hb seem to be associated with regional heterothermy in vertebrates (e.g. Weber and Campbell, 2011), and reduced and reverse temperature effects have been reported from regionally heterothermic fishes, including istiophorid billfishes, we expected both swordfish and opah Hb to have a similar reduced temperature sensitivity. The results show temperature-independent and reverse temperature dependence of Hb–O₂ affinity in blood and stripped haemolysates of opah, and a pH- and saturation-dependent effect of temperature on Hb–O₂ affinity in blood and stripped haemolysates of swordfish (Figs 3 and 4, 5 and 7).

Opah and swordfish both have elevated Hb concentrations, suggesting high blood O₂ carrying capacity when compared with other less active teleosts (Bernal et al., 2001; Brill and Bushnell, 2006; Gallaugher and Farrell, 1998). The Hct values (mean±s.e.m.) for swordfish (46.3±4.7%), opah (59.1±2.5%) and the two bluefin tuna (54% and 58%) were higher than values reported for resting tuna (35–44%) (Brill and Bushnell, 1991, 2006; Lowe et al., 2000). These relatively high Hct values are, however, within the range of values reported for capture-stressed marlins (43–55%) and tunas (75–83%) (Dobson et al., 1986; Wells et al., 1986), and our opah Hct was similar to a published value (53.5±14.1% mean±s.d.) (Wegner et al., 2015). Capture stress can cause high Hct in fish, usually as a result of adrenergic release of RBCs stored in the spleen (i.e. splenic contraction), β-adrenergic receptor stimulated RBC swelling, or a combination of the two (e.g. Wells et al., 1986; Wendelaar Bonga, 1997). The Hb concentrations reported in this study (1.8–2.1 mmol l⁻¹; Table 1) are within the range of values reported for resting tunas (1.9–2.3 mmol l⁻¹), while MCHCs (3.7–4.4 mmol l⁻¹; Table 1) are lower than those of resting tunas (5.1–5.7 mmol l⁻¹) but are within the range of values reported for capture-stressed tunas and marlins (3.8–4.8 mmol l⁻¹) (Brill and Bushnell, 1991, 2006; Lowe et al., 2000; Wells et al., 1986). Therefore, RBC swelling in response to capture stress likely contributed to the high Hct that we measured (Wells et al., 1986).

In this study, blood was withdrawn from swordfish and opah shortly after capture at sea, because it was unrealistic to sample resting and cannulated fish. Consequently, the fish probably experienced varying levels of respiratory and metabolic acidosis concomitant with burst swimming and capture-induced fatigue, which is indicated by relatively high plasma lactate and osmolality levels (Table 1) that are comparable to values reported for capture-stressed tunas and marlins (Dobson et al., 1986; Wells et al., 1986). Even so, we were able to model OECs over a range of pH levels that are comparable to published studies on marlins and tunas, because the blood pH levels that we achieved with the CO₂ exposures (pH 7.23–8.04 for opah, and pH 7.33–7.94 for swordfish) overlapped with the range of pH levels reported in O₂ equilibria studies of blood from both capture-stressed marlins (25°C) and resting tunas (15–30°C) (Brill and Bushnell, 1991, 2006; Dobson et al., 1986; Lowe et al., 2000; Wells and Davie, 1985). However, the RBC ATP levels were low relative to the RBC Hb concentrations (Table 1), and although they are lower than ATP levels reported for most teleosts under non-stressed conditions, they are within the range of values reported for capture-stressed tunas and marlins (Filho et al., 1992; Wells and Davie, 1985; Wells et al., 1986). Given the findings of our haemolysate experiments and those of previous studies, the amount of ATP present in the RBCs will undoubtably affect Hb–O₂ affinity and its temperature dependence (e.g. Larsen et al., 2003; Nelson et al., 2019; Weber et al., 2010). Therefore, fresh blood from unstressed opah and swordfish may have higher ATP levels, potentially causing different temperature dependencies and lower O₂ binding affinities than we measured (i.e. higher P₅₀ values than reported here). The effects of temperature that we observed in our experiments on whole blood were qualitatively like those for our experiments on stripped haemolysates, in that swordfish Hb showed a reduced temperature dependence that was dependent on pH in both blood and stripped haemolysates (in the presence of ATP), and the reduced and reverse temperature dependencies evident in opah blood were also observed in stripped haemolysates either in the absence of ATP (reduced temperature dependence) or in the presence of ATP (reverse temperature dependence) (Figs 3 and 7).

Opah whole blood–O₂ affinity showed both reduced and reversed temperature dependencies that were dependent on O₂ saturation, while the temperature dependence of swordfish blood–O₂ affinity was dependent on both O₂ saturation and blood pH (Figs 4 and 5). Saturation- and pH-dependent effects of temperature on Hb–O₂ affinity have also been reported in studies of blood, stripped haemolysates or isolated Hb components from some other regionally heterothermic fishes (Andersen et al., 1973; Carey and Gibson, 1977, 1983; Ikeda-Saito et al., 1983; Sharp, 1975; Weber et al., 2010). In teleosts, the saturation dependence of the effect of temperature is largely due to the non-linear release of Bohr protons, with most proton dissociation occurring between 50% and 100% Hb–O₂ saturation (Brauner et al., 1996; Ikeda-Saito et al., 1983; Lowe et al., 1998). Therefore, the results of this study exemplify the importance of evaluating how temperature influences the shape and position of the entire OEC, and that generalized conclusions can be misleading if only P₅₀ values are evaluated (Sharp, 1975).

Among regionally heterothermic fishes, reductions in the thermal sensitivity of Hb–O₂ affinity have been attributed to oxygenation-linked effector dissociation that contributes endothermically to ΔH′ (e.g. Ikeda-Saito et al., 1983; Larsen et al., 2003; Weber and Campbell, 2011; Weber et al., 2010). In stripped haemolysate from three species of istiophorid billfishes, oxygenation-linked dissociation of ATP with secondary contributions from additional Bohr proton dissociation greatly reduced or reversed the temperature dependence of Hb–O₂ affinity (Weber et al., 2010). We observed similar effects of ATP and Bohr protons on the temperature dependence of Hb–O₂ affinity in stripped haemolysates of swordfish, where the addition of ATP induced a Bohr effect and increased ΔH′ (pH 7.4) from −42 to −9 kJ mol⁻¹ O₂ in the absence of ATP (Fig. 7C and Table 3). In opah stripped haemolysates in the absence of ATP, temperature had a relatively reduced effect on Hb–O₂ affinity with a ΔH′ of −15 kJ mol⁻¹ O₂ at pH 7.4, but the addition of ATP caused a reverse temperature dependence, increasing ΔH′ to +84 kJ mol⁻¹ O₂ (Fig. 7A and Table 3). In the absence of ATP, opah stripped haemolysate–O₂ affinity may decrease with increasing temperature at high pH (extrapolated from regression lines in Fig. 7A), indicating that Bohr proton dissociation potentially contributes to the low ΔH′, although other processes such as Hb conformational changes (i.e. T→R state transitions) may also have significant enthalpic contributions. We observed reverse temperature-dependent Hb–O₂ affinity in haemolysates (no ATP) of Atlantic bluefin tuna (Fig. 7E and Table 3). This finding is similar to previous work on bluefin tuna Hb showing that proton dissociation from many Bohr groups, concomitant with cooperative O₂ binding, underlies the reversed effect of temperature on Hb–O₂ affinity (Carey and Gibson, 1977; Ikeda-Saito et al., 1983; Morris and Gibson, 1982).

The results of the effect of temperature on whole-blood–O₂ affinity were paralleled during closed-system temperature changes. Opah whole-blood–O₂ affinity showed evidence of a reverse temperature dependence during the closed-system experiments, in which warming tended to cause a reduction in blood P_O2 and cooling tended to increase blood P_O2 (Fig. 6A). In contrast, closed-system warming and cooling of swordfish blood caused blood P_O2 to increase and decrease, respectively (Fig. 6B). The changes to swordfish blood P_O2 were greater than predicted from the temperature dependence of the O₂ solubility of blood plasma, which indicates that closed-system warming probably caused Hb–O₂ offloading as a result of decreased Hb–O₂ affinity and, vice versa, cooling probably caused Hb–O₂ binding as a result of an increased Hb–O₂ affinity.

Early experiments on closed-system temperature changes on blood P_O2 from regionally heterothermic fishes tested the hypothesis that temperature-independent Hb–O₂ affinity prevents premature Hb–O₂ offloading during closed-system warming in a heat exchanging rete and the surrounding muscle (Carey and Gibson, 1977; Cech et al., 1984; Graham, 1973). It was thought that if temperature-induced Hb–O₂ offloading occurred as the blood was warmed in an artery or arteriole, then O₂ may be lost from the arterial blood, potentially by arterio-venous O₂ diffusion in a heat exchanging rete, which could reduce arterial blood O₂ levels prior to perfusing the muscle capillaries (Carey and Gibson, 1977, 1983; Graham, 1973). In addition to our results for opah, temperature-independent and reverse temperature-dependent blood–O₂ affinity during closed-system warming has been reported for several species of tuna and regionally heterothermic sharks (Bernal et al., 2018; Brill and Bushnell, 1991, 2006; Cech et al., 1984; P.R.M., D.B., C.A.S. and C.J.B., unpublished data). However, if warming of arterial blood in a heat exchanging rete causes premature Hb–O₂ offloading that is detrimental to circulatory O₂ delivery, then it would be expected that temperature-independent blood–O₂ affinity would be evident during closed-system temperature changes for all regionally heterothermic fishes. This is not the case as closed-system warming greatly increased the P_O2 of swordfish blood (Fig. 6B), and similar results have been reported for bigeye tuna (Thunnus obesus) and kawakawa (Euthynnus affinis) (Jones et al., 1986; Lowe et al., 2000). It seems unlikely that O₂ would be lost from the arterial vessels upstream of the capillaries, especially given the arterioles and venules of vertebrate heat-exchanging retia, as well as the diffusion distance between them, are about an order of magnitude greater than the vessels of a rete specialized for gas exchange, the swim bladder rete (Carey et al., 1985; Clark et al., 2008; Graham and Dickson, 2001; Lemons et al., 1987; Stevens et al., 1974). As the anatomy of heat-exchanging retia likely precludes significant arterio-venous O₂ diffusion, reduced and reverse temperature-dependent Hb–O₂ affinity is probably not needed to prevent premature Hb–O₂ offloading. It is probably more important that Hb–O₂ affinity is not too different in the capillaries of warm and cold tissues.

In most animals, endothermic Hb O₂ offloading in the tissues and exothermic Hb–O₂ binding at the gas-exchange surface causes the outward transport and loss of metabolic heat to the environment, which contributes to ectothermy in fishes (Jensen et al., 1998; Stevens and Sutterlin, 1976; Weber and Fago, 2004). However, in the opah, most tunas and regionally heterothermic sharks, ΔH′_WB values are near zero or positive (i.e. temperature-independent and reverse temperature-dependent Hb–O₂ affinity), potentially eliminating any outward heat transport linked to Hb O₂ offloading and Hb–O₂ binding at the tissues and gills, respectively, which could potentially save up to 13% of the heat produced during glucose metabolism, assuming an intrinsic enthalpy of haem oxygenation of −62 kJ mol⁻¹ and an oxycalorific equivalent for glucose metabolism of 473 kJ mol⁻¹ (Weber and Wells, 1989; Weber et al., 2010). This may increase the efficiency of heat retention in the warm red muscles of the opah and species that inhabit cold temperate or polar waters (e.g. some tunas and lamnid sharks). However, it is not clear whether Hb oxygenation enthalpy significantly contributes to thermoconservation, as both swordfish and bigeye tuna ΔH′_WB values were negative (i.e. exothermic), yet swordfish undertake prolonged (up to 12 h) foraging excursions below the thermocline, where they remain within cold water (6–12°C) while maintaining cranial and red muscle temperatures elevated up to 10°C or more above the surrounding water (Carey, 1990; Lowe et al., 2000; Sepulveda et al., 2010; D.B. and C.A.S., unpublished data).

Reduced and reverse temperature-dependent Hb–O₂ affinity should prevent excessive temperature-induced shifts to the physiological OEC (i.e. in vivo Hb–O₂ affinity), and thus blood P_O2, as peripheral tissue temperature changes with environmental temperature, and as blood flows from the gills to warmer tissues. In other words, Hb–O₂ affinity would not be greatly affected by temperature changes in the body or environment of a regionally heterothermic fish such as the opah. A rightward shifted OEC at colder temperatures should ensure that Hb–O₂ affinity is not high enough to impair Hb O₂ offloading to the cold peripheral tissues, or possibly even enhance O₂ offloading compared with Hb with a ‘normal’ temperature dependence (Clark et al., 2008; Giardina et al., 1989; Weber and Campbell, 2011). If the physiological OEC is not too different in the capillaries of the warm and cold tissues, then metabolically produced CO₂ and protons will predominantly promote Hb–O₂ offloading and maintain matching between O₂ supply and O₂ demand to all the tissues and organs even though they vary in temperature.

Swordfish Hb, however, has a ‘normal’ temperature dependence at high O₂ saturations and high blood pH, but at low pH, Hb–O₂ affinity becomes temperature independent. The possible benefits of temperature-independent Hb–O₂ affinity in billfishes have been generally attributed to a relatively left-shifted OEC at high temperatures, potentially preventing large decreases to Hb–O₂ affinity as the blood is rapidly warmed (Weber et al., 2010). However, rather than preventing a right-shifted OEC in warm tissues (i.e. decreased Hb–O₂ affinity), the pH dependency of the effect of temperature on swordfish and billfish Hb may ensure a right-shifted OEC (i.e. decreased Hb–O₂ affinity), in cold tissues, which combined with a larger Bohr coefficient at low temperatures (Table 2), may promote Hb O₂ offloading in the capillaries of cold tissues and organs with a high metabolic demand (Clark et al., 2008). This may have the greatest physiological relevance to swordfish, in which organs such as the heart operate at cold ambient water temperatures (6–12°C) during foraging excursions below the thermocline, but the cardiac and swimming muscles of swordfish exhibit relatively greater tolerance to functioning in cold water, potentially elevating the O₂ demand of these muscles compared with those of other teleosts at cold temperatures (Galli et al., 2009; Stoehr et al., 2018, 2020).

Reductions in the thermal sensitivity of Hb may also be influenced by the species’ thermal behaviour as reduced temperature-dependent Hb–O₂ affinity is not a trait that is exclusive to regional heterotherms, and has been reported for several ectothermic elasmobranchs and teleosts (e.g. Barlow et al., 2017; Bernal et al., 2018; Cech et al., 1994; Clark et al., 2010; Hopkins and Cech, 1994; Weber et al., 1976). Temperature-independent Hb–O₂ affinity was originally proposed to potentially enable O₂ uptake as Atlantic bluefin tuna rapidly swim through waters of varying temperatures (Rossi Fanelli and Antonini, 1960). Ectothermic fishes could plausibly gain the same advantage from Hb with a reduced thermal sensitivity; however, it is not clear whether temperature-independent Hb–O₂ affinity in ectothermic fishes is a consequence of effector binding that was selected for reasons unrelated to the temperature sensitivity of Hb.

Opah and swordfish whole blood P₅₀ values at pH 7.7 and 10°C were around 18 and 12 mmHg, respectively (Table 2), suggestive of relatively high blood–O₂ affinity in these species compared with other marine teleosts, including some tunas (e.g. Brill and Bushnell, 1991; Brill and Bushnell, 2006; Clark et al., 2008; Harter et al., 2022; Lowe et al., 2000). Low blood P₅₀ values in this study were possibly linked to the low RBC ATP levels that may have been caused by capture stress and the blood storage duration (1–4 days), although the stripped haemolysate experiments revealed high intrinsic Hb–O₂ affinity in both opah and swordfish (Table 3), with P₅₀ values (pH 7.4) for swordfish (1.8 and 4.5 mmHg at 10 and 25°C, respectively) being lower than those of opah (5.9 and 7.3 mmHg at 10 and 20°C, respectively) and other istiophorid billfishes (3.5–4.0 mmHg at 10°C, and 10.5–14.8 mmHg at 25°C; Weber et al., 2010). Low swordfish P₅₀ values were also reported by Andersen et al. (1973), around 4 and 12 mmHg at 5 and 25°C, respectively, although they did not report the pH and it is unclear whether their experiments were conducted on blood or haemolysates. Furthermore, in two capture-stressed istiophorid billfishes, blue marlin (Makaira nigricans) and striped marlin (Kajikia audax), blood P₅₀ values at pH 7.6 and 25°C were around 13 and 16 mmHg, respectively, and the striped marlin ATP:Hb ratio was around 0.22, lower than the mean for the swordfish in this study (Dobson et al., 1986; Wells and Davie, 1985). Although a high blood–O₂ affinity in the swordfish may be due to the relatively low RBC ATP levels, we suspect that the high intrinsic Hb–O₂ affinity of the swordfish causes a high whole-blood–O₂ affinity that might help swordfish to exploit relatively deep and potentially low oxygenated waters (Carey and Robinson, 1981; Dewar et al., 2011; Sepulveda et al., 2010).

Swordfish diurnal movements follow the vertical migrations of the organisms that form the deep scattering layer, thus exposing swordfish to depths (during the daytime) that are typically proximal to the upper reaches of the oxygen minimum layer (e.g. Sepulveda et al., 2010). Consequently, swordfish likely encounter low environmental oxygen levels during the day. In addition to a potentially high blood–O₂ affinity, swordfish also have a relatively large gill surface area that is comparable to tunas, but above that of istiophorid billfishes as well as most other marine teleosts (Wegner et al., 2010). A large gill surface area combined with a high blood-O₂ affinity should enable adequate O₂ extraction from oxygen poor water and allow swordfish to exploit the upper reaches of the oxygen minimum layer during the day.

The results presented here show temperature-independent and reversed temperature-dependent Hb–O₂ affinity in blood and haemolysates of opah, and temperature-independence at low pH in blood and haemolysates of swordfish. In regionally heterothermic fishes such as opah and swordfish, an increased ΔH′ might conserve heat-energy and possibly prevents Hb–O₂ affinity from being too different from the cold to the warm tissues and organs. The latter effect should promote O₂ unloading uniformly to all tissues despite differences in tissue temperature. Whole blood-O₂ affinities were relatively low for both opah and swordfish, which may have resulted from RBC ATP/Hb ratios that were relatively low for teleosts. RBC ATP/Hb ratios were, however, similar to those of capture-stressed tunas and marlins, and swordfish stripped haemolysates had high intrinsic O₂ affinities. A potentially high blood-O₂ affinity in the swordfish, may enable oxygen uptake from the water of the upper reaches of the oxygen minimum layer where swordfish tend to forage during daylight hours. Among regionally heterothermic teleosts, the relative contributions of the different effector ions that underlie the molecular mechanism of modulations to the enthalpy of Hb–O₂ binding differ among the opah, billfishes, and tunas, which all differ from lamnid sharks (Larsen et al., 2003; Weber and Campbell, 2011). It does not seem to be a coincidence that Hbs with reduced temperature sensitivity are present in all lineages of regionally heterothermic fishes investigated to date, yet this trait is shared with some ectothermic fishes (e.g. Clark et al., 2010; Nelson et al., 2019). Further studies on the temperature-dependence of Hb–O₂ affinity in regionally heterothermic fishes, including other opah and billfish species, and closely related ectothermic species are warranted to provide further insight into the evolution and functional significance of reduced and reverse temperature-dependent Hb–O₂ affinity.

The authors thank Anne Kim for help with the OEC experiments, John Hyde for help with opah sample acquisition, and Jennifer Thirkell for logistical assistance. We also thank Dr Richard Brill and Dr Peter Bushnell for their assistance developing parts of this work.