Oxygenation properties of hemoglobin and the evolutionary origins of isoform multiplicity in an amphibious air-breathing fish, the blue-spotted mudskipper (Boleophthalmus pectinirostris)

Facultative air-breathing fishes have always been of interest to comparative physiologists because of the versatility required to switch between aquatic and aerial respiration (Graham, 1997; Graham and Wegner, 2010; Bayley et al., 2019). Moreover, the physiological features that distinguish air-breathing fishes from their obligate water-breathing kin may provide clues about the types of phenotypic changes that facilitated the invasion of land by the shallow-water progenitors of modern tetrapods. In addition to evolutionary changes in branchial and cardiovascular function in fish that have evolved the capacity for aerial respiration, there has been much interest in associated changes in hemoglobin (Hb) function and respiratory gas transport (Damsgaard et al., 2014, 2015; Bayley et al., 2019).

Regardless of breathing mode, the physiologically optimal Hb–O₂ affinity is dictated by the trade-off between O₂ loading at the respiratory surfaces (gills, lungs or other air-breathing structures) and O₂ unloading to respiring tissues (Brauner and Wang, 1997; Wang and Malte, 2011; Storz, 2019), and there are reasons to expect that this trade-off may be especially profound in facultative air-breathing fishes. Because air has a much higher O₂ content than water, aerial breathing should generally place a lower premium on O₂ uptake at the respiratory surfaces (as arterial O₂ saturation is typically not a limiting factor in tissue O₂ delivery). This has led some authors to suggest that the evolutionary transition from water to air breathing generally entailed a reduction in Hb–O₂ affinity that permitted a concomitant increase in circulatory O₂ delivery and aerobic metabolism (Johansen and Lenfant, 1972; Johansen et al., 1978; Powers et al., 1979). However, among amphibious fish that alternate between breathing modes, a high Hb–O₂ affinity retains much utility under conditions of aquatic hypoxia because it improves O₂ uptake at the gills, thereby enhancing blood O₂ capacitance [the quantity of O₂ unloaded to the tissues for a given difference in arterial and venous partial pressures of O₂ (P_O2)]. Indeed, among water-breathing and facultative air-breathing fish alike, increased Hb–O₂ affinities are generally associated with adaptation to aquatic hypoxia (Powers, 1980; Jensen, 2004; Mandic et al., 2009; Wells, 2009; Fago, 2017; Harter and Brauner, 2017).

In addition to considerations related to differences in O₂ capacitance between air and water and the challenges posed by aquatic hypoxia, the optimal Hb–O₂ affinity in air-breathing fish may also relate to the diffusive conductance of air-breathing structures and design features of the cardiovascular system. Among most air-breathing teleosts, extrabranchial gas exchange occurs across the vascularized epithelia of the buccopharyngeal cavity, esophagus, intestine, stomach or swim bladder (Graham, 1997; Graham and Wegner, 2010). Such air-breathing structures generally have much lower diffusion capacities and lower gas exchange efficiencies than normal gills, so an elevated Hb–O₂ affinity can promote O₂ uptake by maintaining a steep P_O2 gradient across the respiratory epithelium (Hlastala and Berger, 2001; Damsgaard et al., 2014). Finally, aerial breathing generally increases the P_O2 of arterial blood in the ventral aorta that flows towards the gills and – in the absence of a non-respiratory branchial shunt – O₂-rich arterial blood could desaturate during branchial passage (Randall et al., 1981; Olson, 1994; Ishimatsu, 2012; Bayley et al., 2019). Thus, a high Hb–O₂ affinity may also be beneficial in facultative air breathers because it helps prevent transbranchial O₂ loss, thereby ensuring adequate O₂ delivery to metabolizing tissues. For this reason, Damsgaard et al. (2014) suggested that facultative air-breathing fish species that do not possess transbranchial shunts could be expected to have evolved higher Hb–O₂ affinities than obligate air breathers with reduced gills.

Most teleost fish co-express multiple, structurally distinct Hb isoforms (isoHbs) in adult red blood cells, which are conventionally classified as ‘anodic’ or ‘cathodic’ on the basis of electrophoretic mobility (Weber, 1990, 2000). Anodic isoHbs tend to have relatively low O₂ affinities and a large Bohr effect (reduced Hb–O₂ affinity at low pH), whereas cathodic isoHbs tend to have relatively high O₂ affinities, a heightened sensitivity to the affinity-reducing effects of organic phosphates (e.g. ATP and GTP), and a negligible Bohr effect in the presence of organic phosphates (Weber and Jensen, 1988; Weber, 1990, 2000; Jensen et al., 1998; Weber et al., 2000; Wells, 2009). According to the classification scheme of Weber (1990), ‘class I’ species such as plaice and carp express multiple anodal isoHbs with very similar O₂-binding properties, whereas ‘class II’ species such as anguillid eels, salmonids and catfish additionally express one or more cathodal isoHbs that broaden the spectrum of O₂ affinities and allosteric regulatory capacities. The isoHb multiplicity in class II species may permit a physiological division of labor whereby pH-insensitive cathodal isoHbs provide a reserve capacity for blood–O₂ transport under conditions of environmental hypoxia or metabolic acidosis (Weber, 1990, 2000). In some teleost species, evidence suggests that regulatory switches in isoHb expression may play a role in the acclimatization response to environmental hypoxia (Rutjes et al., 2007), and in amphibious fish such as mudskippers it seems plausible that a physiological division of labor between isoHbs with different oxygenation properties could contribute to the versatility required to reversibly switch between aquatic and aerial respiration.

Here, we examine the structure and function of Hbs in the blue-spotted mudskipper (Boleophthalmus pectinirostris), an amphibious, facultative air-breathing fish that is distributed along the coastlines of China, Taiwan, the Korean peninsula and Japan in the northwest Pacific. This species routinely switches between breathing modes when shuttling between tidepools and mudflats in the intertidal zone (Martin and Bridges, 1999) and it uses an extensive respiratory surface for aerial breathing, including the buccal, pharyngeal, branchial and opercular cavities, in addition to the gills and integument (Graham, 1997). We also report sequence data for the full complement of α- and β-type globin genes from B. pectinirostris, we characterize the isoHb composition of adult red cells, and we integrate comparative genomic and phylogenetic analyses to shed light on the evolutionary origins of isoHb multiplicity in mudskippers.

We annotated the full complement of α- and β-type globin genes in the genome assembly of the blue-spotted mudskipper, Boleophthalmus pectinirostris (Linnaeus 1758) (GenBank accession number: JACK00000000.1). For analyses of conserved synteny and phylogenetic relationships, we used genomic sequence data from a diverse set of bony fish and cartilaginous fish, including globin gene sequences that we annotated previously (Hoffmann et al., 2012; Opazo et al., 2013, 2015).

Protein coding sequences were translated and aligned using L-INS-i strategies in Mafft v7.3 (Katoh and Standley, 2013) and were then reverse-translated so that coding sequences were codon-aligned. We annotated several genes with truncated coding sequences [Hbb3-MN in platyfish (Xiphophorus maculatus), Hbb2-MN and Hbb6-MN in Nile tilapia (Oreochromis niloticus) and Hba1 in spotted gar (Lepisosteus oculatus)], but we did not include such sequences in the phylogenetic analyses. We reconstructed separate phylogenies for the α- and β-globin genes using maximum likelihood (ML). After partitioning the alignment into codon positions, we performed the ML analyses in RAxML v 8.1.3 (Stamatakis, 2006) using the GTRGAMMA substitution model. Support for each node in the best tree was based on 1000 bootstrap replicates.

Prior to blood sampling, the fish were maintained in plastic aquaria with artificial seawater (15% salinity) at 27°C for 24 h. Two individual specimens of B. pectinirostris were anesthetized with 0.01% MS222 (Sigma-Aldrich, St Louis, MO, USA) and blood was collected via caudal venipuncture with a heparinized syringe. The animals were handled in accordance with a protocol approved by the Animal Ethics Committee and the Institutional Review Board on Bioethics and Biosafety of BGI (certification number FT15029).

Each blood sample was incubated for 5 min in a 5-fold volume of ice-cold buffer (10 mmol l⁻¹ Hepes, 0.5 mmol l⁻¹ EDTA, pH 7.75). Hemolysates were centrifuged to remove cell debris (30 min at 15,000 g, 4°C) and KCl was then added to the clear supernatant to a final concentration of 0.2 mol l⁻¹ (Jendroszek et al., 2018). The sample was then passed through a 5 ml PD-10 (GE Healthcare) desalting column equilibrated with 10 mmol l⁻¹ Hepes, 0.5 mmol l⁻¹ EDTA, pH 7.7. This desalting step removes endogenous organic phosphates, which are mainly ATP and GTP in fish red cells.

We purified individual isoHbs by means of anion-exchange fast-protein liquid chromatography (FPLC) using an Äkta Pure system (GE Healthcare). Stripped hemolysate was loaded on the prepacked HiTrap Q-HP column (GE Healthcare) and equilibrated with 10 mmol l⁻¹ Hepes buffer (0.5 mmol l⁻¹ EDTA, pH 7.66), and separation of individual isoHbs was achieved using a linear gradient of 0–80% of 0.2 mol l⁻¹ NaCl, with a flow rate of 1 ml min⁻¹. Fractions were concentrated using Amicon Ultra centrifugal filters (10 kDa) and subsequently dialyzed against 10 mmol l⁻¹ Hepes, 0.5 mmol l⁻¹ EDTA, pH 7.7, at 4°C, to remove NaCl. Purity of the isoHb fractions was verified by means of native polyacrylamide gel electrophoresis (PAGE) using a PhastSystem (GE Healthcare) in 10–15% gradient precast gels. Purified samples were stored at −80°C at a 1 mmol l⁻¹ heme concentration.

The subunit composition of each eluted Hb fraction was determined by tandem mass spectrometry (MS/MS). For the MS/MS analysis, eluted fractions were separated on a 20% SDS-PAGE gel and stained with Coomassie Brilliant Blue. The individual α- and β-chain subunits of the purified isoHbs were excised and digested with trypsin. Following previously described protocols (Storz et al., 2011; Revsbech et al., 2013; Storz et al., 2015), peptide mass fingerprints derived from MS/MS were queried with a custom database that included amino acid sequences from all embryonic and adult-expressed α- and β-type globin genes that we annotated from the B. pectinirostris genome assembly. The queries were conducted with the Mascot data search system (version 1.9.0, Matrix Science, London, UK). Search parameters included no restriction on protein molecular weight or isoelectric point, and methionine oxidation was allowed as a variable peptide modification. Mass accuracy settings were 0.15 Da for peptide mass and 0.12 Da for fragment ion masses. We identified all significant protein hits that matched more than one peptide with P<0.05.

We measured O₂-equilibrium curves of purified isoHbs at 25°C in 0.1 mol l⁻¹ Hepes buffer (pH 7.40) in the absence (stripped) and presence of 0.1 mol l⁻¹ KCl and ATP (0.75 mmol l⁻¹), and in the simultaneous presence of KCl and ATP. The 0.75 mmol l⁻¹ ATP concentration falls within the range of values estimated for teleost red cells (Nikinmaa, 1982; Cadiz et al., 2019) and is therefore relevant to in vivo conditions. Heme concentration for each Hb solution was 0.3 mmol l⁻¹, yielding a 10-fold molar excess of ATP to tetrameric Hb. This ATP:Hb₄ ratio permits an assessment of the maximal possible effect of ATP on Hb–O₂ affinity. To measure the Bohr effect, we took replicate measurements of stripped isoHbs under identical conditions at pH 6.7 and 7.4. The final pH of Hb solutions used in each experiment was measured at 25°C using an InLab Micro pH electrode (Mettler Toledo). We measured O₂ equilibria of 3 µl thin-film samples in a modified diffusion chamber where absorption at 436 nm was monitored during stepwise changes in the equilibration of N₂/O₂ mixtures generated by precision Wősthoff gas-mixing pumps (Weber, 1992; Grispo et al., 2012; Weber et al., 2013; Natarajan et al., 2015, 2016). We estimated values of P₅₀ and n₅₀ (Hill's cooperativity coefficient at half-saturation) by fitting the Hill equation Y = Pⁿ_O2/(Pⁿ₅₀ + Pⁿ_O2) to the experimental O₂ saturation data by means of nonlinear regression (where Y is the fractional O₂ saturation and n is the cooperativity coefficient). The nonlinear fitting was based on four to nine equilibration steps between 30% and 70% oxygenation, and standard errors are reported for the resultant estimates of P₅₀ and n₅₀. Free Cl⁻ concentrations were measured with a model 926S Mark II chloride analyzer (Sherwood Scientific Ltd, Cambridge, UK).

We used a genome assembly of the blue-spotted mudskipper, B. pectinirostris, to characterize the physical organization of the globin gene clusters, and we integrated analyses of conserved synteny with phylogenetic reconstructions to unravel the duplicative history of the mudskipper α- and β-type globin genes. We identified a total of four α-type globin genes and three β-type globin genes in the genome of B. pectinirostris. As in other teleosts examined to date, these genes were distributed between two unlinked clusters, designated ‘LA’ and ‘MN’ (the abbreviations refer to flanking genes that demarcate the 5′ and 3′ boundaries of each set of tandemly linked globin genes) (Fig. 1). The LA and MN gene clusters represent the paralogous products of the teleost-specific whole-genome duplication (TGD) (Opazo et al., 2013) that occurred ∼350 million years ago (Meyer and Van de Peer, 2005). The LA cluster of B. pectinirostris contains two α-type globin genes (Hba1-LA and Hba2-LA, from 5′ to 3′) with two β-type globin genes interleaved between them in the opposite orientation (Hbb2-LA and Hbb1-LA), whereas the MN cluster contains three α-type globin genes (Hba1-MN, Hba2-MN and Hba3-MN) (Fig. 2).

Red cell lysates from B. pectinirostris were subjected to anion-exchange FPLC, which revealed two major adult isoHbs, HbI and HbII, in approximately equal quantities and with similar anodic mobilities on native polyacrylamide gels. MS/MS analysis of the FPLC-purified isoHbs revealed that each of them incorporate the same β-chain (product of the Hbb1-LA gene) but different α-chains, as HbI incorporates the product of Hba1-LA and HbII incorporates the product of Hba1-MN (Fig. 3A). The two α-globin genes, Hba1-LA and Hba1-MN, are members of the two separate TGD-derived gene clusters (Fig. 2). Given the antiquity of the TGD, it is not surprising that the two α-globin paralogs are highly divergent at the amino acid sequence level, differing at 54 of 141 residue positions (Fig. 3B).

Consistent with the comparative genomic analysis of conserved synteny, phylogenetic analysis revealed that α- and β-type globin genes from a taxonomically diverse set of teleost species clustered into discrete LA and MN clades (Fig. 4). Within each main clade of LA- and MN-associated globin genes, the adult-expressed mudskipper globin genes were in some cases nested within subclades of genes from other species that are annotated as early-expressed embryonic globins. For example, within ‘LA Hba clade 2’ in the phylogeny of α-type globins (Fig. 4A), the adult-expressed Hba1-LA gene of mudskipper forms a subclade with the adult-expressed Hba1-LA gene from medaka (Oryzias latipes) and the early-expressed Hba2-LA gene from Nile tilapia (O. niloticus). Likewise, within the ‘LA Hbb clade 1’ in the phylogeny of β-type globins (Fig. 4B), the adult-expressed Hbb1-LA of mudskipper forms a subclade with the adult-expressed Hbb1-LA of medaka and the early-expressed Hbb1-LA of Nile tilapia, as well as the early- and late-expressed Hbb1-LA genes of green-spotted puffer (Dichotomyctere nigroviridis) and fugu (Takifugu rubripes).

Under identical experimental conditions, O₂-equilibrium curves revealed that the two isoHbs exhibited virtually identical O₂ affinities (as measured by P₅₀, the P_O2 at which Hb is 50% saturated) and cooperativities (as measured by n₅₀, Hill's cooperativity coefficient at half-saturation) (Fig. 5, Table 1). O₂ affinities of the two isoHbs were reduced (i.e. P₅₀s were increased) in the presence of ATP but not in the presence of Cl⁻ ions alone (added as 0.1 mol l⁻¹ KCl) (Table 1, Fig. 5). Curves for both isoHbs measured in the joint presence of KCl and ATP revealed a slight decrease in P₅₀ compared with curves measured with ATP alone, indicating that the binding of monovalent Cl⁻ ions interferes with the allosteric binding of ATP in the central cavity of T-state deoxyHb. The O₂ affinities of both isoHbs exhibited pronounced sensitivities to changes in pH: Bohr factors (ΔlogP₅₀/ΔpH) were −0.79 and −0.81 for HbI and HbII, respectively.

Boleophthalmus pectinirostris expresses multiple anodic isoHbs, each with a relatively high intrinsic O₂ affinity, moderate sensitivity to organic phosphates, and a large Bohr effect – a ‘class I’ profile, according to the classification scheme of Weber (1990). This same basic pattern of isoHb multiplicity has been well documented in other teleosts, including facultative air breathers and obligate water breathers alike. The Hb system of B. pectinirostris appears functionally similar to that of the swamp eel (Monopterus albus), a facultative air breather in which gas exchange is largely restricted to the buccopharyngeal cavity (Damsgaard et al., 2014, 2015). Similar to B. pectinirostris, M. albus expresses multiple anodic isoHbs and no cathodic components. The two major isoHbs of M. albus exhibit high intrinsic O₂ affinities [P_50(stripped)=4.8−5.2 torr (25°C, pH 7.7)], similar to values measured for the two isoHbs of B. pectinirostris (Table 1), but the mudskipper isoHbs exhibited a higher sensitivity to ATP and estimated Bohr coefficients were 2- to 3-fold larger when measured under identical buffer conditions. The mudskipper isoHbs have P₅₀ values similar to those reported for the anodic isoHbs of hypoxia-tolerant, water-breathing teleosts such as carp (Cyprinus carpio) (Gillen and Riggs, 1972; Weber and Lykkeboe, 1978; Jensen et al., 2017) and European eel (Anguilla anguilla) (Fago et al., 1997). In contrast to the mudskippers, another facultative air-breathing fish, the Amazonian armored catfish (Hoplosternum littorale), exhibits a far more pronounced level of isoHb differentiation, with anodic and cathodic isoHbs that differ substantially in intrinsic O₂ affinity, phosphate sensitivity and Bohr effect (Weber et al., 2000).

The high Hb–O₂ affinity of the mudskipper should promote branchial O₂ uptake under conditions of aquatic hypoxia and – during terrestrial excursions – transepithelial O₂ uptake in the pharyngeal air-breathing structures. The potential drawback of an increased Hb–O₂ affinity is that it hinders O₂ unloading in the tissue capillaries, but this can be mitigated to some extent by the Bohr effect. The large Bohr effect of B. pectinirostris and some other facultative air-breathing fish may enhance O₂ delivery to tissues during metabolic or respiratory acidosis upon switching from aquatic to aerial respiration, where O₂ uptake may be less impaired. Likewise, liberation of Bohr protons upon Hb oxygenation would greatly facilitate CO₂ excretion. Measurements on whole blood have revealed numerous facultative air-breathing fish species that exhibit pronounced Bohr effects, but there is extensive interspecific variation (Shartau and Brauner, 2014; Bayley et al., 2019; Mendez-Sanchez and Burggren, 2019). Given the especially large Bohr effects of the mudskipper isoHbs, it is worth noting that the β-chain C-termini (HC3) have His (Fig. 3B), which is known to play a major role in deoxygenation-linked proton-binding, whereas the cathodic isoHbs of other teleosts such as Hoplosternum and Anguilla (which exhibit little or no Bohr effect, or even a reverse Bohr effect in the absence of organic phosphates) have non-ionizable Phe at the same residue position (Fago et al., 1995; Weber et al., 2000).

In addition to the hypothesis that obligate and facultative air-breathing fishes have evolved reduced Hb–O₂ affinities compared with obligate water breathers (Johansen and Lenfant, 1972; Johansen et al., 1978; Powers et al., 1979), it has been suggested that facultative air-breathing fishes may have evolved Hbs with reduced Bohr effects to cope with CO₂ retention and/or exposure to hypercarbic water (Carter, 1931). Available data do not appear to support either hypothesis. Among ray-finned fish, lobe-finned fish and amphibians, facultative air and water breathers tend to have higher blood–O₂ affinities than obligate air breathers, and facultative air breathers tend to have slightly larger Bohr effects than members of the other two groups (Bayley et al., 2019). The high intrinsic O₂ affinities and large Bohr effects that we measured for the purified mudskipper isoHbs appear to be consistent with both trends.

Given that the two adult-expressed isoHbs of B. pectinirostris are present at approximately equimolar concentrations in the red cell, it seemed reasonable to expect that the two distinct α-chain subunits would be encoded by a tandemly linked pair of genes under the transcriptional control of the same cis-acting regulatory elements. Such a linkage arrangement could also help explain why the two isoHbs exhibit such similar O₂-binding properties, because tandem duplicates often have similar coding sequences owing to interparalog gene conversion (Hoffmann et al., 2008 a,b; Runck et al., 2009; Storz et al., 2010; Gaudry et al., 2014; Natarajan et al., 2015; Signore et al., 2019). Contrary to this seemingly reasonable expectation, the analysis of conserved synteny revealed that the genes that encode the α-type subunits of HbI and HbII are members of unlinked, paralogous gene clusters that represent the duplicated products of the teleost-specific genome duplication (Figs 1 and 2).

In the phylogenies of α- and β-type globins, the fact that adult-expressed mudskipper genes clustered with some embryonically expressed globin genes of other teleosts is not too surprising, as evolutionary changes in stage-specific expression during ontogeny have been well documented in teleosts (Opazo et al., 2013) and in many tetrapods as well (Opazo et al., 2008; Hoffmann et al., 2010, 2018; Storz, 2016, 2019).

In the case of mudskippers, there is no appreciable difference in the functional properties of co-expressed isoHbs in spite of extensive amino acid divergence between the alternative α-chain subunit isoforms (Fig. 3B). It therefore appears that the ability to switch between aquatic and aerial respiration does not necessarily require a division of labor between functionally distinct isoHbs that are specialized for O₂ transport under different conditions. The adult-expressed isoHbs of many obligate water-breathing fish are known to exhibit much higher levels of functional differentiation than those of mudskippers, including quantitative differences in O₂ affinity and qualitative differences in the mode of allosteric regulation, especially in relation to the magnitude of the Bohr effect and sensitivity to organic phosphates (Weber, 1990, 2000). In many teleost species, the presence of specific isoHbs with functional specializations such as the Root effect (an extreme form of pH sensitivity that plays a key role in oxygen secretion and general tissue O₂ delivery) exert an important influence on physiological capacities (Berenbrink, 2007; Rummer et al., 2013; Randall et al., 2014). However, the absence of functional isoHb differentiation in amphibious mudskippers and other facultative air breathers, and the often pronounced functional heterogeneity in the isoHbs of obligate water breathers, supports the conclusions of several authors (Fyhn et al., 1979; Ingermann, 1997; Wells, 2009; Storz, 2019) that the overall diversity of co-expressed isoHbs in fish red cells is not generally a strong determinant of physiological versatility or ecological niche breadth. In mudskippers, any changes in blood–O₂ affinity that are associated with transitions between aerial and aquatic breathing are likely caused by changes in red cell pH and/or red cell concentrations of nucleotide triphosphates.

We thank Elin E. Petersen for assistance in the lab and two anonymous reviewers for helpful comments.