New insights into the allosteric effects of CO₂ and bicarbonate on crocodilian hemoglobin

The oxygenation properties of vertebrate hemoglobins (Hbs) are sensitive to a number of small-molecule effectors present in red blood cells (RBCs), including protons, CO₂, chloride ions and organic phosphates. Allosteric regulation of Hb–O₂ affinity by such effectors helps keep blood O₂ supply in balance with tissue O₂ consumption. These effectors bind preferentially to the low-affinity (deoxy) T-state of the Hb tetramer, thereby reducing Hb–O₂ affinity by shifting the T-R equilibrium towards the T quaternary state (Perutz, 1976; Storz, 2019; Weber and Fago, 2004). Hence, increases in blood CO₂ levels (as produced by increased metabolic activity) cause a reduction in Hb–O₂ affinity, thereby facilitating O₂ unloading to tissues. The underlying mechanism is the reaction of CO₂ with N-terminal amino groups (-NH₂) of the α- and β-chain subunits of the Hb tetramer (Kilmartin and Rossi-Bernardi, 1971), resulting in the formation of a carbamino product (-NH-COO⁻), which preferentially stabilizes the low-affinity T-state of Hb via interactions with positively charged amino acid residues. CO₂ also exerts an indirect allosteric effect on Hb–O₂ affinity by reducing RBC pH. Upon entry into the RBC, CO₂ reacts with water to form bicarbonate ions and protons, a reaction catalyzed by the enzyme carbonic anhydrase. Protons bind to Hb and promote O₂ unloading via the Bohr effect, while bicarbonate ions leave the RBC in exchange for chloride ions.

Among vertebrate Hbs, those of crocodilians are reportedly unique in that their O₂ affinity is allosterically regulated by binding of bicarbonate ions to the Hb protein (Bauer et al., 1981; Perutz et al., 1981). This unusual bicarbonate sensitivity of crocodilian Hb is thought to support aerobic metabolism when blood bicarbonate levels are elevated. This occurs during breath-hold diving (Bautista et al., 2021; Hicks and White, 1992), and during the postprandial ‘alkaline tide’ where gastric acid secretion causes a pronounced metabolic alkalosis (Busk et al., 2000). In support of these interpretations, we recently found that the concentration of Hb-bound bicarbonate increased in the blood of diving caimans and correlated with the degree of O₂ depletion (Bautista et al., 2021), providing strong in vivo evidence of the regulatory effect of bicarbonate ions on Hb oxygenation.

The bicarbonate sensitivity of crocodilian Hbs was first described in vitro in the spectacled caiman (Caiman crocodilus) (Bauer et al., 1981; Perutz et al., 1981), but the allosteric effect was inferred without direct experimental verification of bicarbonate binding to Hb. Subsequent studies on Hbs from other crocodilian species documented a direct allosteric effect of CO₂ (Fago et al., 2020; Jensen et al., 1998; Weber et al., 2013), but the direct contribution of bicarbonate binding has been neither confirmed in these studies nor quantified. The reason is likely that it is difficult to experimentally distinguish the allosteric effects of CO₂ from those of bicarbonate and protons because at equilibrium all these three compounds are present in solution. Moreover, previous studies (Bauer et al., 1981) used different experimental and buffer conditions and rarely accounted for the potential effect of other anions, such as chloride ions, which are potent allosteric modulators of crocodilian Hb–O₂ affinity (Fago et al., 2020; Storz, 2019; Weber et al., 2017).

Here, we used kinetic and equilibrium approaches at a constant pH and in the absence of chloride ions or organic phosphates to separate the allosteric effects of CO₂ and bicarbonate on crocodilian Hb oxygenation. We provide a detailed functional analysis on the allosteric regulation of the Hb of the spectacled caiman, the same species that was originally studied by Bauer and coworkers (Bauer et al., 1981; Bauer and Jelkmann, 1977; Jelkmann and Bauer, 1980). We also confirm the same mode of allosteric regulation in the Hbs of 13 other caiman, alligator and crocodile species, indicating that this unique property evolved in the common ancestor of extant crocodilians.

All protocols of animal handling and sampling were performed in accordance with the Danish Law for Animal Experimentation and were approved by the Danish Animal Experiments Inspectorate, under permit number 2018-15-0201-01507.

Approximately 5 ml blood was drawn from the post-occipital sinus into heparinized syringes from three non-anesthetized adult spectacled caimans (Caiman crocodilus Linnaeus 1758; around 1.5 kg each) of undetermined sex. The procedure lasted less than 5 min. Blood was immediately centrifuged at 4°C (5000 g, 5 min) to remove plasma and RBCs were then washed 5 times in 0.9% NaCl, 0.5 mmol l⁻¹ EDTA and stored at −80°C until further processing. As described previously (Fago et al., 2020), we sampled blood and prepared hemolysates from 13 other crocodilian species [American alligator, Alligator mississippiensis (Daudin 1802); Chinese alligator, Alligator sinensis (Fauvel 1879); broad-snouted caiman, Caiman latirostris Daudin 1802; yacaré caiman, Caiman yacare Daudin 1802; black caiman, Melanosuchus niger (Spix 1825); smooth-fronted caiman, Paleosuchus trigonatus (Schneider 1801); American crocodile, Crocodylus acutus Cuvier 1807; Philippine crocodile, Crocodylus mindorensis Schmidt 1935; Nile crocodile, Crocodylus niloticus Laurenti 1768; New Guinea crocodile, Crocodylus novaeguineae (Schmidt 1928); saltwater crocodile, Crocodylus porosus Schneider 1801; Cuban crocodile, Crocodylus rhombifer Cuvier 1807; and Siamese crocodile, Crocodylus siamensis Schneider 1801].

All protocols of animal handling and sampling were performed in accordance with the Danish Law for Animal Experimentation and were approved by the Danish Animal Experiments Inspectorate, under permit number 2018-15-0201-01507.

For each individual spectacled caiman (N=3), the hemolysate was prepared by dilution (1:5) of washed RBCs in chloride-free 10 mmol l⁻¹ Hepes pH 7.6, 0.5 mmol l⁻¹ EDTA, and incubated on ice for 30 min. The hemolysate was then centrifuged at 4°C (12000 g, 20 min) to remove cell debris and then passed (‘stripped’) on a PD-10 desalting column (GE Healthcare) equilibrated with 10 mmol l⁻¹ Hepes pH 7.6, 0.5 mmol l⁻¹ EDTA to remove small molecular weight molecules, including allosteric effectors of Hb. Chloride-free Hepes buffers were used throughout. The absence of chloride in solutions was verified by a model 926S Mark II chloride analyzer (Sherwood Scientific Ltd, Cambridge, UK).

To discriminate between the effects of CO₂ and bicarbonate on Hb oxygenation, we measured the time course of changes in O₂ saturation of stripped hemolysate and gel-filtration purified Hb (see ‘Hb purification and mass spectrometry analysis’, below) after addition and removal of 1% CO₂ gas at a fixed P_O2. These experiments were performed in the presence and absence of 1 mmol l⁻¹ acetazolamide, a specific inhibitor of carbonic anhydrase contained in the hemolysate, in order to manipulate the rate of bicarbonate formation from CO₂ and distinguish between effects due to bicarbonate or CO₂ on Hb oxygenation (Fago et al., 1999). Experiments were performed at 25°C using a thin-layer modified diffusion chamber technique (Cadiz et al., 2019; Natarajan et al., 2018; Storz et al., 2020; Weber et al., 2017). In brief, 4 µl samples were placed in the temperature-controlled gas diffusion chamber connected to a Cary 60 UV-Vis spectrophotometer equipped with fiber optic probes (Agilent Technologies), and O₂ saturation was determined from the absorption trace at 415 nm. Discrete values of P_O2 and P_CO2 within the chamber were obtained by mixing O₂ and CO₂ with ultrapure N₂ gas using a programmable Gas Mixing System (GMS, Loligo Systems, Viborg, Denmark), with the gas mixture constantly flowing above the sample surface. The experiments were performed at a fixed P_O2 corresponding approximately to 50% O₂ saturation, under the same conditions as in our previous study on crocodilian Hbs (0.3 mmol l⁻¹ heme, 0.1 mol l⁻¹ Hepes pH 7.2, 0.5 mmol l⁻¹ EDTA) (Fago et al., 2020). Heme concentration of samples was measured spectrophotometrically using published extinction coefficients at 577 and 542 nm (Van Assendelft, 1970).

Throughout this study, the pH of Hb samples equilibrated with the same CO₂ of the experiments was routinely measured using a pH electrode (InLab^® Micro, Mettler Toledo) and it did not change appreciably (range 7.16–7.21).

Hb was purified from endogenous carbonic anhydrase by gel filtration using fast protein liquid chromatography (FPLC) on an ÄKTA Pure Protein Purification System equipped with a Superdex 75 10/300 GL gel-filtration column and a 100 µl loop (GE Healthcare Life Sciences). Buffer was 10 mmol l⁻¹ Hepes pH 7.6, 0.5 mmol l⁻¹ EDTA, 0.15 mol l⁻¹ NaCl, and flow rate was 0.5 ml min⁻¹. Absorbance was measured at 280 nm. The column was calibrated with standard proteins (Sigma Aldrich) of known molecular mass: horse myoglobin (17.6 kDa), bovine carbonic anhydrase II (30 kDa) and human adult Hb (64.6 kDa). The Hb fractions purified by gel filtration were concentrated by ultracentrifugation and dialyzed against 10 mmol l⁻¹ Hepes pH 7.5, 0.5 mmol l⁻¹ EDTA to remove NaCl. After dialysis, the absence of Cl⁻ in the samples was confirmed by using a MK II Chloride Analyzer 926S (Sherwood Scientific Ltd). Hb fractions were analyzed for the presence or absence of intermolecular disulfide bonds by SDS polyacrylamide gel electrophoresis (range 4–12%) using NuPAGE^® Novex^® Bis-Tris Pre-Cast Gels and stained with Coomassie Blue, in the presence and absence of the reducing agent DTT.

To establish whether carbonic anhydrase (∼30 kDa) co-eluted with Hb by FPLC, we ran the purified Hb fractions on a mini-protean precast 4–20% SDS PAGE gel (Bio-Rad, Hercules, CA, USA), which we then stained with Coomassie Brilliant Blue-G. We conducted tandem mass spectrometry (MS/MS) analyses on the bands ranging from ∼25 to 50 kDa in size (Fig. S3). The stained bands were excised and processed for in-gel tryptic digestion (Shevchenko et al., 2006) and the eluted peptides were then analyzed using a Thermo Orbitrap Fusion Lumos Tribrid (Thermo Scientific™) mass spectrometer in data-dependent acquisition mode. Peptides were identified by searching MS/MS data against the UniprotKB/Swiss-Prot database. The search was set up for full tryptic peptides with a maximum of two missed cleavage sites. The precursor mass tolerance threshold was set as 10 ppm and maximum fragment mass error was set at 0.02 Da. Qualitative analysis was performed using PEAKS X software. The significance threshold of the ion score was calculated based on a false discovery rate of ≤1%.

To establish whether the purified Hb fractions had different sensitivity to CO₂, we determined O₂ equilibrium curves for each gel-filtration purified Hb fraction at 25°C (0.3 mmol l⁻¹ heme, 0.1 mol l⁻¹ Hepes pH 7.2, 0.5 mmol l⁻¹ EDTA) at discrete CO₂ tensions in the gas mixture, and calculated P₅₀ and n at each P_CO2 as described above (see ‘Measuring O₂ equilibrium curves’).

To determine the effect of bicarbonate on Hb oxygenation, we equilibrated samples of each gel-filtration purified Hb fraction (0.05 mmol l⁻¹ heme, 0.1 mol l⁻¹ Hepes buffer pH 7.2, 0.5 mmol l⁻¹ EDTA) in Eschweiler tonometers for at least 40 min at 25°C with a gas mixture of 2% air and 98% N₂ generated by a gas-mixing pump (Wösthoff, Bochum, Germany). The same gas mixture (corresponding to a P_O2 close to the P₅₀ of the Hb) was also used to equilibrate a freshly prepared stock solution of sodium bicarbonate (NaHCO₃) dissolved in MilliQ H₂O in a separate tonometer at the same temperature. After equilibration, the Hb sample was transferred with a gastight syringe into a 2.5 ml Tucker chamber (Tucker, 1967), previously flushed with 2% air and 98% N_2. The chamber was equipped with a fiber optic O₂ probe connected to a FireSting-O2 (FSO2-1) oxygen and temperature meter and controlled by FireSting logger V.2.365 software for P_O2 signal recording (PyroScience, Germany). Following stabilization of the P_O2 signal of the Hb sample in the chamber, a small volume (<60 µl) of the NaHCO₃ solution was injected into the Tucker chamber to achieve discrete NaHCO₃ concentrations (2, 4, 7.7, 15.4, 23.1 and 30 mmol l⁻¹), and the subsequent increase in P_O2 in the chamber (due to bicarbonate-induced O₂ unloading by the Hb) was recorded. The bicarbonate concentration of the stock solution was 0.5, 1, 1.2 or 1.3 mol l⁻¹, chosen based on the desired final concentration within the chamber, to maintain a similar injection volume (<60 µl) in all samples. A new Hb sample was used for each bicarbonate addition and separate experiments were performed for the Hb fractions of each of the three individuals. Two-point calibration of the oxygen probe was performed immediately before recording each trace by flushing the chamber with water-saturated air or N₂ and taking into account the barometric pressure on the day. The pH of the bicarbonate-containing Hb solution was measured at the end of each experiment using a pH electrode (InLab^® Micro, Mettler Toledo) and no significant change was observed (range 7.19–7.23). The accuracy of the Tucker chamber technique was investigated in technical triplicates, yielding standard error of final mean P_O2 of 0.5% (Table S2).

Statistical analysis and curve fitting were performed using SigmaPlot version 12 (SigmaStat) and Prism V.8 (GraphPad). The obtained parameters S_O2,min and K_CO2 were statistically compared by two-way ANOVA considering CO₂ and Hb fraction as factors. The K_CO2 values obtained for the Hb fractions were statistically compared by t-test. Similarly, the ΔP_O2,max and K_bic values obtained for the Hb fractions were statistically compared by t-test. Statistical significance was considered at α<0.05. Data are given as means±1 s.e.m. (corresponding 95% confidence intervals, CI, are provided in Tables S1 and S2).

To investigate the mode of allosteric regulation of crocodilian Hb, we first tested whether the CO₂ effect on O₂ saturation of Hb could be attributed exclusively to bicarbonate, as originally proposed by Bauer and co-workers (1981). In these experiments, we used stripped hemolysates of spectacled caiman (N=3 individuals) and 13 other crocodilian species representing the two most speciose families, Alligatoridae and Crocodylidae, which together capture the deepest evolutionary split among all living crocodilians. We used a kinetic approach originally developed by us to demonstrate that hagfish Hb is sensitive to bicarbonate, but insensitive to CO₂ (Fago et al., 1999). This method takes advantage of the presence of active endogenous carbonic anhydrase in the hemolysate that rapidly converts added CO₂ to bicarbonate, and of the carbonic anhydrase inhibitor acetazolamide that specifically inhibit this fast enzymatic reaction. Thus, with active carbonic anhydrase, a fast decrease of Hb O₂ saturation upon addition of CO₂ means that either CO₂ or bicarbonate can act as allosteric effectors, whereas with carbonic anhydrase inhibited by acetazolamide, a slower decrease of O₂ saturation upon addition of CO₂ indicates that CO₂ must first be converted to bicarbonate (via the slow non-catalyzed reaction) to be able to affect oxygenation and that bicarbonate – not CO₂ – is the true allosteric effector (Fago et al., 1999). As shown in Fig. 1, at a fixed P_O2 (yielding ∼50% O₂ saturation), addition of 1% CO₂ in the gas mixture caused a rapid and strong decrease in the Hb O₂ saturation down to ∼10% in the stripped hemolysate of spectacled caiman (Fig. 1A). This CO₂-mediated decrease in Hb O₂ saturation was fully reversible, and 50% saturation was restored when CO₂ was removed from the gas mixture (Fig. 1A). Remarkably, the decrease in Hb-O₂ saturation upon addition of CO₂ was equally fast when endogenous carbonic anhydrase was inhibited by acetazolamide (Fig. 1B). We also observed the same response in the hemolysates of other crocodilian species (Fig. S1).

The fact that the recovery to ∼50% O₂ saturation upon removal of CO₂ was slower in the presence of acetazolamide (Fig. 1; Fig. S1) was puzzling and could indicate (1) that carbonic anhydrase was involved in the dissociation of Hb-bound CO₂, perhaps by direct interaction with Hb, or (2) that acetazolamide could somehow interfere with the dissociation of CO₂ from Hb. The next set of experiments was aimed to test these possibilities. The addition of bovine carbonic anhydrase II to spectacled caiman hemolysate produced similar traces to those of Fig. 1 (Fig. S2), indicating that endogenous carbonic anhydrase was indeed present in the hemolysate.

In order to determine whether endogenous carbonic anhydrase was somehow involved in the reversibility of the CO₂-mediated effect on Hb oxygenation, we passed spectacled caiman hemolysate through a gel-filtration column to remove carbonic anhydrase. Spectacled caiman expresses a single adult Hb isoform (Jelkmann and Bauer, 1980), similar to all other crocodilians examined (Fago et al., 2020; Hoffmann et al., 2018). Results from gel filtration showed differences in the apparent molecular mass of Hb, with some Hb assembled into a higher-order aggregate – presumably an octamer – with an apparent molecular mass of 95.7 kDa, while most Hb was present as a regular tetramer with an apparent molecular mass of 62.7 kDa (Fig. 2A,B), as reported earlier in the same species (Bauer and Jelkmann, 1977). We named these Hb fractions Hb1 and Hb2, respectively. Analysis of Hb1 and Hb2 by SDS-PAGE under reducing and non-reducing conditions (Fig. 2C and Fig. S3, respectively) showed identical patterns and intensity of bands, indicating that high-order aggregates are not formed via disulfide bonds as in the Hb aggregates of other crocodilians (Weber et al., 2013) and non-avian reptiles (Petersen et al., 2017; Storz et al., 2015), but via some other interaction.

SDS-PAGE of Hb1 and Hb2 revealed not only the α- and β-type Hb chains but also less intense bands of higher molecular mass that could correspond to carbonic anhydrase, having a molecular mass of 30 kDa (Fig. 2C). The identity of these bands, corresponding to a size range of ∼25–50 kDa, was established by mass spectrometry. The ∼30 kDa band corresponded to stable intact heterodimers (Hb α- and β-chains), possibly associated with aggregate formation of Hb, and the ∼42 kDa band corresponded to actin protein. Thus, carbonic anhydrase had been effectively removed from both Hb1 and Hb2 by gel filtration.

As a next step, we investigated the reversibility of CO₂ binding on O₂ saturation for the purified Hb1 and Hb2 fractions in the absence and presence of acetazolamide (Fig. S4). As with the stripped hemolysate (Fig. 1), return to the original O₂ saturation upon removal of CO₂ was slower in the presence of acetazolamide. This result suggests that acetazolamide somehow interferes with Hb–CO₂ dissociation, but not with Hb–CO₂ binding.

To establish whether purified Hb1 and Hb2 had different O₂ binding affinities and CO₂ sensitivities, we measured O₂ equilibrium curves under identical buffer conditions at different P_CO2 values. These experiments yielded similar P₅₀ at a range of P_CO2 values for the two Hb fractions (Fig. 2D) and similar cooperativity coefficients (n ∼1.0–1.8), indicating that the assembly of Hb into higher-order aggregates did not alter the oxygenation properties and CO₂ sensitivity.

We used the diffusion chamber technique to determine the apparent CO₂ affinity of the purified Hb1 and Hb2 fractions (Fig. 3A–C). These experiments measured the decrease in O₂ saturation at different P_CO2 values, while maintaining a fixed P_O2 roughly corresponding to the P₅₀ of each Hb measured in the absence of CO₂ (Fig. 3A). The Hb1 and Hb2 fractions exhibited changes in O₂ saturation upon addition of CO₂ that were not significantly different from each other (two-way ANOVA, P=0.543), so the data were pooled. These data showed a single hyperbolic relationship (Fig. 3B,C), indicating the presence of a single type of binding site for CO₂ in each Hb, with apparent affinities (expressed as P_CO2) corresponding to 1.48±0.22 Torr (95% CI, 1.10–1.96 Torr; Table S1; where 1 Torr=0.133 kPa) and 1.32±0.15 Torr (95% CI, 1.06–1.66 Torr; Table S1) for Hb1 and Hb2, respectively, that were not significantly different from each other (t-test, P=0.570). The use of a double hyperbola equation did not narrow the confidence interval of the fitted parameters (Table S1), so the data support the existence of a single type of binding site for CO₂ on caiman Hb.

To verify experimentally the direct allosteric effect of bicarbonate, we used purified Hb1 and Hb2 fractions of spectacled caiman in a Tucker chamber, equipped with an O₂ probe. In these experiments, we kept the Hb solution at a fixed P_O2 (close to the P₅₀) and measured the increase in P_O2 in the chamber following the injection of a measured small volume of a bicarbonate stock solution that was maintained at the same P_O2 (Fig. 3D–F). The results of these experiments show that bicarbonate had a dose-dependent effect on O₂ release from Hb following a simple hyperbolic function, indicating the existence of a single type of allosteric binding site. Injection of bicarbonate solutions in buffer alone did not increase P_O2 in the chamber (Fig. 3D). From the bicarbonate binding curves, the apparent bicarbonate affinities were 3.31±0.35 mmol l⁻¹ (95% CI, 1.89–5.49 mmol l⁻¹; Table S2) and 2.48±0.16 mmol l⁻¹ (95% CI, 1.72–3.45 mmol l⁻¹; Table S2) for Hb1 and Hb2, respectively, which are not significantly different from each other (P=0.094). Interestingly, the apparent affinity for bicarbonate of spectacled caiman Hb in the deoxy T-state reported by Bauer and co-workers (1981) was ∼0.7 mmol l⁻¹, close to our estimated values at the same pH and temperature.

This study provides the first experimental evidence that bicarbonate binds directly to crocodilian Hb and promotes O₂ delivery (Fig. 3), independently of CO₂. In addition, this study demonstrates that CO₂ binds to crocodilian Hb directly without being converted to bicarbonate, as shown by the fast and acetazolamide-independent decrease in Hb O₂ saturation upon addition of CO₂ (Fig. 1; Fig. S1). This indicates that CO₂ is a potent allosteric effector of crocodilian Hbs in its own right, in contrast with the original study by Bauer and coworkers (1981) performed under different experimental conditions, including the use of chloride-containing solutions. Also, the fact that the CO₂-induced decrease in Hb O₂ saturation was of the same magnitude in the presence and absence of acetazolamide (Fig. 1; Fig. S1) indicates that there is no additional binding of bicarbonate (which would form rapidly without acetazolamide) elsewhere on the protein. Thus, the allosteric effect of bicarbonate occurs via the binding of bicarbonate to the same protein binding site for CO₂.

In analogy with other Hbs, we propose that CO₂ chemically binds to the crocodilian Hb protein to form a carbamino compound (-NH-COO⁻) at a non-protonated amino group (-NH₂) (see reaction scheme in Fig. 4A). The negatively charged carbamino group would then be stabilized by nearby positively charged amino acid residues on a specific T-state binding site (Fig. 4B), thereby causing a decrease in the O₂ affinity of Hb and promoting O₂ unloading. As pH does not change upon addition of CO₂, as reported here and earlier (Jensen et al., 1998), the proton generated in carbamino formation (Fig. 4A) is picked up by the protein molecule, likely contributing to further electrostatic stabilization of the negative carbamino derivative. We note that there is very little structural difference between covalent CO₂ binding via carbamino formation and non-covalent bicarbonate binding to the Hb protein (Fig. 4B), consistent with the inference from our results that the two allosteric effectors bind with high affinity (Fig. 3) to the same site (Fig. 1). Thus, the degree of site occupancy by either CO₂ or bicarbonate depends on their relative concentrations. In addition, protein binding of CO₂ but not of bicarbonate requires an unprotonated amino group (Fig. 4B), and should be favored by an increase in pH (which, however, increases bicarbonate concentration). These aspects should be investigated further once the binding site is known. It is possible that the evolution of a particularly strong interaction with CO₂ in crocodilian Hbs could have caused the strong interaction with bicarbonate ions, which characterize these Hbs. Hagfish Hb, whose function is not regulated by T–R allosteric transitions (Fago et al., 2001), is thus the only one among chordates to be sensitive to bicarbonate but not CO₂ (Fago et al., 1999).

Consistent with previous studies (Bauer et al., 1981; Perutz et al., 1981), our results (Fig. 3) suggest the existence of a single physical binding site for CO₂ and bicarbonate in the T-state of crocodilian Hb or a pair of identical binding sites due to the two-fold symmetry of the Hb tetramer. We propose that this binding site (or pair of identical sites) comprises a reactive non-protonated amino group close to positively charged residues able to chemically bind CO₂. Carbamino formation in human Hb occurs at the α- and β-subunit N-termini (Bauer and Schröder, 1972; Kilmartin and Rossi-Bernardi, 1971; Perrella et al., 1975), but in crocodilian Hbs, these groups are acetylated to a variable extent without affecting the strong sensitivity to CO₂ (Fago et al., 2020; Natarajan et al., 2020). In the original structural model proposed by Perutz and co-workers (1981), two bicarbonate ions bind two identical sites lining the central cavity of T-state crocodilian Hb, where organic phosphates normally bind, comprising Lys82β, Glu144β and the N-terminus of the α subunit. However, the smooth-fronted caiman (Paleosuchus trigonatus) possesses Gln at 82β rather than positively charged Lys (Fago et al., 2020), and yet the Hb of this species exhibits the same CO₂/bicarbonate effect as that of other crocodilians (Fig. S1). Moreover, acetylation of the α-chain N-termini of crocodilian Hbs does not appear to affect CO₂ sensitivity (Natarajan et al., 2020). Based on site-directed mutagenesis of recombinant human Hb, other studies (Komiyama et al., 1996, 1995) have proposed a more complex mechanism, involving numerous residues at the α₁β₂ interface of the Hb tetramer, in particular, Lys38β, which is present in all crocodilian Hbs. Molecular modeling has revealed numerous positively charged residues in the central cavity of alligator Hb, including Arg135β involved in several interactions (Fago et al., 2020). In the vicinity of other positively charged residues, the amino group of Lys or Arg side chains, which normally have high pK_a values and are protonated, would be largely in the non-protonated form at physiological pH values and would therefore be available for carbamino formation. Mutagenesis studies using recombinant crocodile Hb, rather than human Hb, are underway in our lab to resolve these possibilities. Although the residues involved remain to be identified, it appears from our study that there is great diversity among chordate Hbs in the allosteric control of Hb function by CO₂ and bicarbonate, with Hbs sensitive to only CO₂ (e.g. human Hb), only bicarbonate (hagfish) or a combination of the two (crocodile).

This study aimed to unravel CO₂ and bicarbonate binding to caiman Hb under in vitro controlled buffer conditions without the confounding effect of organic phosphates and chloride ions, which are normally present in crocodile RBCs. As an intended consequence, the experimental conditions in our study did not mimic the in vivo conditions within intact RBCs. In a recent study on caiman blood (Bautista et al., 2021), we found progressive accumulation of CO₂ and bicarbonate in RBCs during diving, and that most of the RBC bicarbonate was bound to Hb rather than exported to the plasma. Most importantly, Hb-bound bicarbonate was inversely correlated with the degree of Hb O₂ saturation in the blood (Bautista et al., 2021), which is fully consistent with the binding of bicarbonate to the T-state of the Hb described in the present study. Although the direct CO₂ binding to Hb was not measured (Bautista et al., 2021), the progressive increase of CO₂ in RBCs during diving is strongly consistent with our present demonstration that Hb acts as a sink for CO₂ as Hb O₂ saturation decreases. Physiologically, the strong effect of blood CO₂ and bicarbonate on Hb oxygenation during diving is to facilitate O₂ delivery to sustain aerobic metabolism of tissues. At the same time, Hb desaturation facilitates CO₂ transport in the venous blood as protein-bound CO₂ and bicarbonate. A study on blood properties from digesting alligators (Busk et al., 2000) proposed that the increase in plasma bicarbonate (the ‘alkaline tide’) should safeguard tissue O₂ delivery by allosterically reducing Hb–O₂ affinity, thereby helping to sustain the large increase in metabolic rate during digestion. However, whether CO₂ and bicarbonate content within RBCs also increases during digestion remains to be investigated. In addition, the possibility of right-to-left shunts in the unique crocodilian heart adds further complexity in these situations through their influence O₂ and CO₂/bicarbonate concentrations in the blood (Conner et al., 2019; Malte et al., 2016).

This is the first study of crocodilian Hb to quantify the direct allosteric binding of both CO₂ and bicarbonate. Both effectors bind to T-state Hb with high affinity at the same positively charged site involving a reactive amino group. Physiologically, the strong Hb sensitivity to CO₂ and bicarbonate provides the basis for a unique linkage between O₂ and CO₂ transport in the blood of crocodilians, where the build-up of CO₂ and bicarbonate ions during breath-hold diving or digestion facilitates O₂ delivery to tissues to sustain aerobic metabolism. These distinct effects of CO₂ and bicarbonate on Hb–O₂ affinity are shared by a phylogenetically diverse set of caiman, alligator and crocodile species, indicating that these unique regulatory properties evolved prior to the diversification of extant crocodilians >80–100 million years ago in the late Cretaceous, but after the stem lineage of crocodilians diverged from the ancestors of dinosaurs and modern birds.

We thank Elin E. Petersen and Marie Skou Pedersen for their technical assistance. We also would like to thank Rene Hedegaard and the staff from the Krokodille Zoo (Eskilstrup, Denmark) for assistance in blood sampling. We also thank Vikas Kumar (Mass Spectrometry and Proteomics Core Facility, University of Nebraska Medical Center) for assistance with the MS/MS data analysis.