Abstract
White seabass (Atractoscion nobilis) are increasingly experiencing periods of low oxygen (O2; hypoxia) and high carbon dioxide (CO2, hypercapnia) due to climate change and eutrophication of the coastal waters of California. Haemoglobin (Hb) is the principal O2 carrier in the blood and in many teleost fishes Hb-O2 binding is compromised at low pH. However, Hb is contained within red blood cells (RBC) that, in some species, regulate intracellular pH with adrenergically-stimulated sodium-proton-exchangers (β-NHE). We hypothesised that white seabass have RBC β-NHEs that protect the blood O2-carrying capacity during hypoxia and hypercapnia. In a series of in vitro experiments, we determined the O2-binding characteristics of white seabass blood, the response of RBCs to adrenergic stimulation, and quantified the protective effect of β-NHE activity on Hb-O2 saturation during a hypercapnic acidosis in normoxia and hypoxia. White seabass had typical teleost Hb characteristics, with a moderate O2 affinity that was highly pH-sensitive. Functional, molecular and bioinformatic data confirmed that white seabass have RBC β-NHEs, and super-resolution imaging revealed, for the first time, the subcellular location of β-NHE protein in intracellular vesicles and on the RBC membrane. The activation of RBC β-NHEs increased Hb-O2 saturation by ∼8% in normoxia at 1% PCO2, and by ∼20% in hypoxia at arterial PCO2 (0.3%), but the protective effects decreased at higher PCO2. Combined, these data indicate that RBC β-NHE activity in white seabass can safeguard arterial O2 transport and the mechanism likely plays an important role in the fishes’ physiological response to environmental hypoxia and hypercapnia.
Summary Statement White seabass have highly pH-sensitive haemoglobins, but their red blood cells can actively protect oxygen transport during hypoxia and hypercapnia, conditions that occur more frequently due to a changing climate.
Introduction
White seabass (Atractoscion nobilis), a teleost fish species endemic to the coastal waters of California, are apex-predators with ecological significance, sought-after targets of recreational and commercial fisheries and are gaining importance in aquaculture. Their natural habitat along the kelp forests of the north-eastern Pacific is subject to strong seasonal fluctuations in water chemistry, due to the upwelling of deeper waters that are often depleted of oxygen (O2; hypoxia), have a high carbon dioxide tension (CO2; hypercapnia) and thus, a low pH (Frieder et al., 2012). In addition, a steadily warming climate and growing anthropogenic nutrient loading are increasing the frequency of large algae blooms (“red tides”) in California’s coastal waters (Van Dolah, F M, 2000). When these algae blooms wane, the microbial decomposition of their biomass, consumes O2 and produces CO2 and other toxic and acidic by-products of biological decay, such as hydrogen sulfide, altogether creating large hypoxic and hypercapnic zones (Diaz and Rosenberg, 2008). Species that are highly mobile may be able to avoid these areas, but for many sedentary species, survival will depend on enduring these conditions. Climate change at large is also leading to more hypoxic and acidic oceans and, over generations, some species may adapt to cope with their altered habitats (Harley et al., 2006). However, the reoccurrence of upwelling and severe algae blooms may acutely expose animals to conditions that far exceed worst-case predictions for the end of the century, creating strong selective pressures for hypoxia and hypercapnia tolerance and perhaps overwhelming the rates at which some species can adapt to climate change.
The most recent severe red tide in Southern California occurred in April-May of 2020, when the water measurements at the pier of the Scripps Institution of Oceanography (SIO) in La Jolla (CA, USA) revealed average daily dissolved O2 levels of <2 mg l-1 and pH as low as 7.06 (Clements et al, 2020). At a water temperature of 17°C and 35 ppt salinity, these values correspond to 5.3% PO2 (or 40 mmHg; Boutilier et al., 1984) and 1.16% PCO2 (CO2SYS software; Lewis and Wallace, 1998). Equally alarming was the prolonged duration of hypoxia, where for nine consecutive days, water PO2 was below the threshold (4.6 mg l-1) that is considered lethal for 90% of marine life (Vaquer-Sunyer and Duarte, 2008). The SIO aquatics facility is supplied with water taken in at the pier, which resulted in hypoxic and hypercapnic exposures of all research animals, despite every effort to aerate the tanks. However unfortunate, this natural experiment revealed a remarkable tolerance of white seabass to these adverse water conditions, despite being deprived the behavioural avoidance of hypoxia that may be recruited in the wild. Therefore, the aim of the present study was to explore the O2-transport capacity of white seabass with a focus on the physiological mechanisms at the level of the red blood cell (RBC) that may allow these fish to endure severe hypoxia and hypercapnia.
For obligate aerobic animals, the challenge to surviving unavoidable environmental hypoxia is balancing the uptake and delivery of O2 with its consumption in the mitochondria (Hughes, 1973). Haemoglobin (Hb) is the principal O2 carrier in the blood and therefore the cardiovascular O2-carrying capacity is largely determined by the O2-binding characteristics of Hb. As such, a higher Hb-O2 affinity will favour the extraction of O2 from hypoxic waters and thus, a lower Hb P50 (the partial pressure of O2 at which Hb is 50% saturated) is typically associated with hypoxia tolerance in fishes (Jensen and Weber, 1982; Mandic et al., 2009); however, whether white seabass have high-affinity Hbs that would confer some hypoxia tolerance is currently unknown.
Hb-O2 binding in teleost fishes is highly pH-sensitive, where a reduction in pH decreases Hb-O2 affinity via the Bohr effect (Bohr et al., 1904), and the Root effect prevents Hb from becoming fully O2-saturated at low pH, even at super-atmospheric PO2 (Root, 1931; Scholander and Van Dam, 1954). The reduction in Hb-O2 carrying capacity due to the Root effect is physiologically significant, as it enhances the unloading of O2 at the eyes and the swimbladder of teleosts, where blood is acidified locally (Damsgaard et al., 2020; Pelster, 1997; Wittenberg and Wittenberg, 1962). In contrast, during a systemic blood acidosis that may occur during exercise or hypoxia, the pH-sensitive Hbs of teleosts may fail to become fully oxygenated at the gills, decreasing the O2-carrying capacity of arterial blood and leading to hypoxemia at the tissues. Thus, a combined hypoxic and hypercapnic exposure may be especially dangerous for teleosts, as a reduced availability of O2 in the environment is paired with the simultaneous reduction of Hb-O2 affinity via the Bohr effect at low pH; however, whether white seabass have pH-sensitive Hbs is currently unknown.
Hb is housed within RBCs that, in teleosts, may prevent systemic hypoxemia by actively regulating their intracellular pH (pHi) to protect Hb-O2 binding during a reduction in extracellular pH (pHe). In brief, a decrease in arterial PO2 or pH leads to the release of catecholamines into the blood (Randall and Perry, 1992; Tetens and Lykkeboe, 1988), which bind to a β-adrenergic receptor on the RBC membrane and activate a sodium-proton-exchanger (β-NHE, Slc9a1b,) via the cyclic adenosine monophosphate (cAMP) pathway (Mahe et al., 1985). The extrusion of H+ by the β-NHE raises pH above the equilibrium condition, which increases Hb-O2 affinity and will promote the extraction of O2 from hypoxic waters (Nikinmaa, 1992). The adrenergic stimulation of RBCs also causes an influx of Na+ and Cl- that leads to osmotic swelling and that has been used as a marker to determine the presence of RBC β-NHEs in fish species (Rummer et al., 2010; Shu et al., 2017). A broader phylogenetic analysis indicates that most teleosts, but not other fishes, have RBC β-NHEs (Berenbrink et al., 2005); however, whether white seabass RBCs have β-NHE activity is currently unknown.
Based on these considerations, we hypothesised that white seabass RBCs have β-NHE activity that will protect the blood O2-carrying capacity during hypoxic and hypercapnic conditions. We tested this hypothesis in a series of in vitro experiments and predicted that: i) white seabass have a high Hb-O2 affinity to maintain O2 uptake under hypoxic conditions, which was addressed by generating oxygen equilibrium curves (OEC) over a range of PO2; ii) white seabass display the large Bohr and Root effects that are typical of teleosts, which was addressed by generating OECs over a range of PCO2, and measuring pHe and RBC pHi; iii) white seabass have a RBC β-NHE, which was addressed using molecular, bioinformatic and immunocytochemical techniques to establish its presence and localisation, and by measuring RBC swelling after adrenergic stimulation and the inhibition of NHEs with amiloride; and finally iv) the adrenergic stimulation of RBC β-NHEs protects the blood O2-carrying capacity during an acidosis and this effect was quantified by measuring Hb-O2 saturation at increasing levels a hypercapnia in normoxia and hypoxia.
Materials and Methods
Animals and husbandry
White seabass (A. nobilis, Ayres 1860) were obtained from the Hubbs Sea World Research Institute (HSWRI, Carlsbad, USA) and were held indoors at the SIO aquatics facility for several months before experiments. Photoperiod was set to a 12:12 h light-dark cycle and fish were housed in large fiberglass tanks (∼3.5-10 m3) supplied with flow-through seawater from an inshore intake; the average water temperature at the time of experiments was 17°C. Aeration was provided to ensure normoxic conditions in all tanks (>90% air saturation of O2) and these water parameters were monitored every day. All fish were fed twice a week with commercial dry pellets (Skretting; Classic Bass 9.5 mm; Stavanger, Norway) and feeding was suspended 48 h before blood sampling. The white seabass used for the determination of blood O2-binding characteristics had an average weight of 1146±96 g (N = 8), while those used for the β-NHE experiments had an average weight of 357±27 g (N = 6). Animal husbandry and all experimental procedures were in strict compliance with the guidelines by the Institutional Animal Care and Use Committee (IACUC) and approved by the Animal Care Program at the University of California San Diego (Protocol no. S10320).
Blood sampling
White seabass were moved individually into darkened boxes supplied with air and flow-through seawater, 24 h prior to blood sampling. The next day the water supply was shut off and the fish were anaesthetized by carefully pouring a diluted benzocaine solution (Fisher Scientific, Acros 150785000; Waltham, USA; concentrated stock made up in ethanol) into the box without disturbing the fish, for a final concentrations of 70 mg l-1 benzocaine (<0.001% ethanol). After visible loss of equilibrium, fish were transferred to a surgery table, positioned ventral-side-up and their gills were perfused with water containing a maintenance dose of anaesthetic (30 mg l-1 benzocaine). Blood sampling was by caudal puncture and 3 ml of blood were collected into a heparinized syringe. This procedure ensured minimal disturbance of the fish (Montgomery et al., 2019), which can decrease blood pH due to air-exposure (respiratory acidosis) and due to anaerobic muscle contractions during struggling (metabolic acidosis). After sampling, the fish were recovered and returned to their holding tank, and each individual was only sampled once. In the lab, the blood was centrifuged at 500 g for 3 min to separate the plasma from the blood cells. The plasma was collected in a bullet tube and stored over-night at 4°C. To remove any catecholamines released during sampling, the blood cells were rinsed three times in cold Cortland’s saline (in mM: NaCl 147, KCl 5.1, CaCl 1.6, MgSO4 0.9, NaHCO3 11.9, NaH2PO4 3, glucose 5.6; adjusted to the measured plasma osmolality in white seabass of 345 mOsm and pH 7.8) and the buffy coat was aspirated generously to remove white blood cells and platelets. Finally, the pellet was re-suspended in 10 volumes of fresh saline to allow the RBCs to return to a resting state and stored aerobically on a tilt-shaker, over-night, at 17°C (Caldwell et al., 2006).
Blood O2-binding characteristics
The next day, RBCs were rinsed with saline three times and re-suspended in their native plasma at a haematocrit of 5%; this value was chosen based on preliminary trials and yields an optic density that allows for spectrophotometric measurements of Hb-O2 binding characteristics (∼0.6 mM Hb). A volume of 1.4 ml of blood was loaded into a glass tonometer at 17°C and equilibrated to arterial gas tensions (21% PO2, 0.3% PCO2 in N2) from a custom-mixed gas cylinder (Praxair; Danbury, USA). After one hour, 2 µl of blood were removed from the tonometer and loaded into the diffusion chamber of a spectrophotometric blood analyser (BOBS, Loligo Systems; Viborg, Denmark). The samples were equilibrated to increasing PO2 tensions (0.5, 1, 2, 4, 8, 16 and 21% PO2) from a gas mixing system (GMS, Loligo), in two-minute equilibration steps and the absorbance was recorded once every second at 190-885 nm. At the beginning and end of each run, the sample was equilibrated to high (99.7% O2, 0.3% CO2 in N2 for 8 min) and low (0% O2, 0.3% CO2 in N2 for 8 min) PO2 conditions; for the calculation of Hb-O2 saturation from raw absorbance values, it was assumed that Hb was fully oxygenated or deoxygenated under the two conditions, respectively. A PCO2 of 0.3% was maintained throughout these trials to prevent RBC pHi from increasing above physiologically relevant levels and this value was chosen to match that measured in the arterial blood of rainbow trout (Oncorhynchus mykiss) in vivo (Brauner et al., 2000a). All custom gas mixtures were validated by measuring PO2 with an FC-2 Oxzilla and PCO2 with a CA-10 CO2 analyser (Sable Systems, North Las Vegas, USA) that were calibrated daily against high purity N2, air, or 5% CO2 in air.
An additional 250 µl of blood were removed from the tonometer to measure blood parameters as follows. Haematocrit (Hct) was measured in triplicate in microcapillary tubes (Drummond Microcaps, 15 µl; Parkway, USA), after centrifuging at 10,000 g for 3 min. Hb was measured in triplicated using the cyano-methaemoglobin method (Sigma-Aldrich Drabkin’s D5941; St. Louis, USA) and an extinction coefficient of 10.99 mmol cm-1 (van Kampen and Zijlstra, 1983). Blood pH was measured with a thermostatted microcapillary electrode at 17°C (Fisher Accumet 13-620-850; Hampton, USA; with Denver Instruments UB-10 meter; Bohemia, USA), calibrated daily against precision pH buffers (Radiometer S11M007, S1M004 and S11M002; Copenhagen, Denmark). Thereafter, the blood was centrifuged to separate plasma and RBCs and total CO2 content (TCO2) of the plasma was measured in triplicate with a Corning 965 (Midland, USA). The RBCs in the pellet were lysed by three freeze-thaw cycles in liquid nitrogen and pHi was measured in the lysate as described for pHe (Zeidler and Kim, 1977). After completing these measurements, the PCO2 in the tonometer was increased in steps from 0.3 to 2.5% and, each time, OECs and blood parameters were measured as described above.
RBC swelling after β-adrenergic stimulation
After storage of blood samples over-night in saline, the RBCs were rinsed three times in fresh saline and re-suspended in their native plasma at a Hct of 25%. A volume of 1.8 ml was loaded into a tonometer and equilibrated to 3% PO2 and 1% PCO2 in N2 at 17°C for one hour; similar hypoxic and acidotic conditions have been shown to promote β-NHE activity in other teleost species (Motais et al., 1987; Salama and Nikinmaa, 1989). After one hour, an initial subsample of blood was taken and Hct, Hb, pHe and pHi were measured as described above. Thereafter, the blood was split into aliquots of 600 µl that were loaded into individual tonometers and treated with either: i) a carrier control (0.25% dimethyl sulfoxide, DMSO; VWR BDH 1115; Radnor, USA), ii) the β-adrenergic agonist isoproterenol (ISO; Sigma I6504; 10 µM final concentration, which stimulates maximal β-NHE activity in rainbow trout; Tetens and Lykkeboe, 1988), or iii) ISO plus the NHE inhibitor amiloride (ISO+Am; Sigma A7410; 1 mM, according to Borgese et al., 1992). These treatments were staggered so that samples from each tonometer could be taken for the measurements of blood parameters at 10, 30 and 60 mins after drug additions.
To collect RBC samples for immunocytochemistry, the above tonometry trial was repeated (N = 2) with RBCs that were suspended in saline instead of plasma; this step was necessary as initial trials showed that plasma proteins interfered with the quality of cell fixations. Subsamples were removed from individual tonometers at the initial and 60 min time points. A volume of 60 µl was immediately re-suspended in 1.5 ml ice-cold fixative (3% paraformaldehyde, 0.175 % glutaraldehyde in 0.6 x phosphate buffered saline with 0.05 M sodium cacodylate buffer; made up from Electron Microscopy Sciences RT15949, Hatfield, USA) and incubated for 60 min on a revolver rotator at 4°C. After fixation, cells were washed three times in 1 x phosphate Buffered Saline (PBS, Corning 46-013-CM, Corning, USA) and stored at 4°C for processing. An additional subsample of 100 µl was removed from the tonometers and centrifuged to remove the saline. The RBC pellet was re-suspended in 5 volumes of lysis buffer containing 1 mM DL-Dithiothreitol (DTT; Thermo Fisher R0861; Waltham, USA), 1 mM phenylmethylsulfonyl fluoride (PMSF; Sigma P7626) and 10 mM benzamidine hydrochloride hydrate (BHH; Sigma B6506) in PBS. The RBCs were lysed by three cycles of freeze-thawing in liquid nitrogen, the lysate was centrifuged at 500 g for 10 min at 4°C and the supernatant was frozen at -80°C for Western blot processing.
Hb-O2 binding after β adrenergic stimulation
An additional aliquot of RBCs was re-suspended in native plasma at a Hct of 5%. Volumes of 300 µl were loaded into one of four tonometers and equilibrated to arterial gas tensions at 17°C (as described previously) and treated with either: i) a carrier control (DMSO; 0.25%), ii) ISO (10 µM), or iii) ISO+Am (1 mM). These treatments were staggered to allow for standardised measurements at 60 min after drug additions, when 2 µl of blood were removed from the tonometer and loaded into the BOBS for real-time measurements of Hb-O2 saturation during a respiratory acidosis. Therefore, the blood was exposed to stepwise increases in PCO2 (0.3, 0.5, 1, 1.5, 2 and 3%) allowing for two minutes of equilibration at each step; preliminary trials showed full equilibration to the new PCO2 after ∼1 min and absorbance remained constant thereafter. As described previously, this protocol also included initial and final calibration steps, at which the sample was fully O2-saturated and then desaturated. A first trial was performed in normoxia (21% PO2) and then a second sample was loaded form the same tonometer for an additional run under hypoxic conditions (3% PO2). The PO2 value in these hypoxic runs was chosen to yield a Hb-O2 saturation close to P50 and was informed from the previous measurements of Hb-O2 binding characteristics. Finally, 250 µl of blood were removed from the tonometer for the measurement of blood parameters, as described previously.
Subcellular localisation of RBC β-NHE
Fixed RBCs were permeabilized in 1.5 ml 0.1% triton-X100 (VWR Amresco 1421C243) in PBS for 15 min at room temperature on a revolver rotator. Thereafter, the RBCs were blocked for auto-fluorescence in 100 mM glycine in PBS for 15 min, after which the cells were rinsed three times in PBS. For immunocytochemistry, 200 µl of these fixed RBCs were re-suspended in a blocking buffer containing 3% bovine serum albumin (VWR 0332) and 1% normal goat serum (Lampire Biological Laboratories 7332500; Pipersville, USA) in PBS and incubated for six hours on a rotator. Primary antibodies were added directly into the blocking buffer and incubated on a rotator over-night at 4°C. A monoclonal mouse anti-Tetrahymena α tubulin antibody (deposited by Frankel, J. and Nelsen, E.M at the Developmental Studies Hybridoma Bank, DSHB12G10; Iowa City, USA) was used at 0.24 µg ml-1 and a custom, polyclonal, affinity-purified, rabbit anti-rainbow trout β (epitope: MERRVSVMERRMSH) was used at 0.02 µg ml-1. After primary incubations the RBCs were washed three times in PBS and incubated for three hours on a rotator at room temperature in blocking buffer containing secondary antibodies: 1:500 goat anti-mouse (AlexaFlour 568; Thermo Fisher Life Technologies A-11031), 1:500 goat anti-rabbit (AlexaFlour 488; A-11008) and 1:1000 4′,6-diamidino-2-phenylindole (DAPI; Roche 10236276001; Basel, Switzerland). After secondary incubations, RBCs were washed three times and were re-suspended in PBS. To validate the β-NHE antibody, controls were performed by leaving out the primary antibody and by pre-absorbing the primary antibody with its pre-immune-peptide. All images were acquired with a confocal laser-scanning fluorescence microscope (Zeiss Observer Z1 with LSM 800, Oberkochen, Germany) and ZEN blue edition software v.2.6. For super-resolution imaging the cells were re-suspended in PBS with a mounting medium (Thermo Fisher Invitrogen ProLong P36980) and acquisition was with the Zeiss AiryScan detector system. To ensure that images were comparable, the acquisition settings were kept the same between the different treatments and between treatments and the controls. Optical sectioning and three-dimensional (3D) reconstructions of single RBCs from the different treatments were processed with the Imaris software v.9.0. (Oxford Instruments, Abingdon, UK) and rendered into movies.
Molecular β-NHE characterisation
For Western blotting, RBC crude homogenates were thawed and centrifuged at 16,000 g for one hour at 4°C to obtain a supernatant containing the cytoplasmic fraction and a pellet containing a membrane-enriched fraction that was re-suspended in 100 µl of lysis buffer. The protein concentration of all three fractions was measured with the Bradford’s assay (BioRad 5000006; Hercules, USA). Samples were mixed 1:1 with Laemmli’s sample buffer (BioRad 161-0737) containing 10% 2-Mercaptoethanol (Sigma M3148) and were heated to 75°C for 15 min. Sample loading was at 5 µg protein from each fraction for the detection of β-NHE and 60 µg protein from crude homogenate for the detection of α and 10% separating polyacrylamide gel. The proteins were separated at 60 V for 30 min and 150 V until the Hb fraction (∼16 kDa) ran out the bottom of the gel (∼60 min); previous trials had shown that the high Hb content of these lysates may bind some antibodies non-specifically. The proteins were transferred onto a Immun-Blot polyvinylidene difluoride membrane (PVDF; BioRad) using a semi-dry transfer cell (Bio-Rad Trans-Blot SD) over-night, at 90 mA and 4°C. PVDF membranes were blocked over-night, on a shaker at 4°C in tris-buffered saline with 1% tween 20 (TBS-T; VWR Amresco ProPure M147) and 0.1 g ml-1 skim milk powder (Kroger; Cincinnati, USA). Primary antibodies were made up in blocking buffer and mixed on a shaker at 4°C, over-night, before applying to the membranes. The anti-α tubulin antibody was used at 4.7 ng ml-1, the anti-β-NHE antibody at 0.42 ng ml-1 and controls at a peptide concentration exceeding that of primary antibody by 10:1. Primary incubations were for four hours on a shaker at room temperature and membranes were rinsed three times in TBS-T for 5 min. Secondary incubations were with either an anti-rabbit or mouse, horse-radish peroxidase conjugated secondary antibody (BioRad 1706515 and 1706516) for three hours on a shaker at room temperature. Finally, the membranes were rinsed three times in TBS-T for 5 min and the proteins were visualized by enhanced chemiluminescense (BioRad, Clarity 1705061) in a BioRad Universal III Hood with Image Lab software v.6.0.1. Protein sizes were determined relative to a precision dual-colour protein ladder (BioRad 1610374).
The white seabass β-NHE sequence was obtained by transcriptomics analysis of gill samples that were not perfused to remove the blood and these combined gill and RBC tissue samples were stored in RNA later for processing. Approximately 50 µg of sample were transferred into 1 ml of Trizol reagent (Thermo Fisher 15596026) and were homogenized on ice with a handheld motorized mortar and pestle (Kimble Kontes, Dusseldorf, Germany). These crude homogenates were centrifuged at 1000 g for 1 min and the supernatant was collected for further processing. RNA was extracted in RNA spin columns (RNAEasy Mini; Qiagen, Hilden, Germany) and treated with DNAse I (ezDNase; Thermo Fisher, 11766051) to remove traces of genomic DNA. RNA quantity was determined by spectrophotometry (Nanodrop 2000; Thermo Fisher) and RNA integrity was determined with an Agilent 2100 Bioanalyzer (Agilent; Santa Clara, USA). Poly-A enriched complementary DNA (cDNA) libraries were constructed using the TruSeq RNA Sample Preparation Kit (Illumina; San Diego, USA). Briefly, mRNA was selected against total RNA using Oligo(dt) magnetic beads and the retained RNA was chemically sheared into short fragments in a fragmentation buffer, followed by first- and second-stand cDNA synthesis using random hexamer primers. Illumina adaptor primers (Forward P5-Adaptor, 5’AATGATACGGCGACCACCGAGA3’; Reverse P7-Adaptor 5’ CGTATGCCGTCTTCTGCTTG 3’) were then ligated to the synthesized fragments and subjected to end-repair processing. After agarose gel electrophoresis, 200-300 bp insert fragments were selected and used as templates for downstream PCR amplification and cDNA library preparation. The combined gill and RBC samples (1 µg RNA) were sent for RNAseq Poly-A sequencing with the Illumina NovoSeq™ 6000 platform (Novogene; Beijing, China) and raw reads are made available on NCBI (PRJNA722314).
RNAseq data was used to generate a de novo transcriptome assembly which was mined for white seabass isoforms of the Slc9a1 protein family using methods previously described (Clifford et al., 2017). Briefly, raw reads were analysed, trimmed of adaptor sequences, and processed with the OpenGene/fastp software (Chen et al., 2018), to remove reads: i) of low quality (PHRED quality score < 20), ii) containing >50% unqualified bases (base quality < 5), and iii) with >10 unknown bases. Any remaining unpaired reads were discarded from downstream analysis and quality control metrics were carried out before and after trimming (raw reads 80.07 x 106; raw bases 12.01 Gb; clean reads 79.44 x 106, clean bases 11.84 Gb, clean reads Q30 95.26%; GC content 46.67%). Thereafter, fastq files were merged into a single data set, normalized, and used for de novo construction of a combined gill and RBC transcriptome using the Trinity v2.6.6 software. Normalization and assembly were performed using the NCGAS (National Centre for Genome Analysis Support) de novo transcriptome assembly pipeline (github.com/NCGAS/de-novo-transcriptome-assembly-pipeline/tree/master/Project_Carbonate_v4) on the Carbonate High Performance Computing cluster housed at Indiana University. For assembly, minimum kmer coverage was set to three and the minimum number of reads needed to glue two inchworm contigs together, was set to four (Grabherr et al., 2011). The resulting nucleotide FASTA file was translated into six protein reading frames using BBMap (Bushnell, 2014), which were mined for the NHE-like proteins using HMMER3 v.3.0 (hmmer.org) by querying the de novo assembly against a hidden markov model (HHM) homology matrix generated from 132 aligned protein sequences of the vertebrate NHE family (slc9a1 – slc9a9; see Table S1 for accession numbers). Sequences were aligned using MUSCLE (Edgar, 2004) in SeaView (Galtier et al., 1996; Gouy et al., 2010), with NHE2 from Caenorhabditis elegans as an outgroup and results were refined using GBlocks (Castresana, 2000) according to the parameters specified previously (Talavera and Castresana, 2007). Phylogenetic analysis was conducted on the Cyberinfrastructure for Phylogenetic Research (CIPRES) Science Gateway (Miller et al., 2010) using the RAxML software v.8.2.12 (Stamatakis, 2014) with the LG evolutionary and GTRGAMMA models (Le and Gascuel, 2008). Branch support was estimated by bootstraps with 450 replications and the constructed tree was edited in FigTree v.1.4.4. Finally, the open reading frame of the white seabass β-NHE sequence (predicted 747 amino acids), was analysed for the presence of the Kozak nucleic acid motifs (5’- (gcc)gccRccAUGG-3’; Kozak, 1987) immediately upstream of putative start codons, using the ATGpr software (Nishikawa et al., 2000).
To confirm the expression of the β-NHE in the RBCs of white seabass, an additional blood sample of 1 ml was collected and processed as described previously. RBCs were lysed by repeatedly passing them through a 23G needle and RNA extraction was on 50 µg of RBCs by standard Trizol and chloroform extraction following the kit instructions. Isolated RNA was treated with DNAse I and 1 µg RNA was used to synthesize first-strand cDNA using SuperScript IV reverse transcriptase (Thermo Fisher 18090010). Full-length cDNA sequences were obtained in 35 cycles of PCR reactions with Phusion DNA polymerase (New England Biolabs, Ipswich, USA; MO531L) and specific primers designed against the sequence of the phylogenetically characterized white seabass β-NHE obtained from the combined gill and RBC transcriptome (Integrated DNA Technologies, Coralville, Iowa; F: 5’TCC CGT ACT ATC CTC ATC TTC A-3’ R: 5’-CCT CTG CTC TCT GAA CTG TAA AT-3’). Amplicons were analysed by gel electrophoresis on a ChemiDoc imaging system (Bio-Rad) that confirmed the presence of a single band (2372 bp). A-overhangs were added to Phusion products with one unit Taq polymerase (New England Biolabs; MO267S) followed by 10 min incubation at 72°C. Products were cloned (TOPO TA Cloning Kit/pCR 2.1-TOPO Vector; Invitrogen; K4500) and the ligated product was transformed into TOP10 chemically competent E. coli cells (Invitrogen; K457501) according to manufacturer specifications. Following over-night incubation at 37°C, single colonies of transformants were grown in Luria-Bertani (LB) broth over-night on a shaking incubator (37°C, 220 rpm; Barnstead MaxQ 4000). Plasmid DNA was isolated (PureLink Quick Plasmid Miniprep kit; Invitrogen K210010) according to manufacturer specifications and inserts were sequenced to confirm their identity and uploaded to NCBI (MW962257).
Calculations and statistical analysis
All data were analysed with R v.4.0.4 (RCoreTeam, 2020) in RStudio v.1.4.1106 (RStudioTeam, 2021) and figures were created with the ggplot2 package (Wickham, 2009). Normality of the residuals was tested with the Shapiro-Wilk test (stat.desc function in R) and homogeneity of variances was confirmed with the Levene’s test (leveneTest function in R). Deviations from these parametric assumptions were corrected by transforming the raw data.
To determine the blood O2-binding characteristics, Hb P50 and nH values were those determined with the BOBS software v.1.0.20 (Loligo) and oxygen equilibrium curves (OEC) were generated by fitting a two parameter Hill function to the mean P50 and nH for 8 individual fish. The main effects of PCO2 on P50 and nH were analysed with ACOVA (P < 0.05, N = 8). Plasma [HCO3-] was calculated from TCO2 by subtracting the molar [CO2] calculated from the dissociation constant and solubility coefficients in plasma at 17°C and the corresponding sample pH (Boutilier et al., 1984). The Bohr effects relative to pHi and pHe, the relationship between RBC pHi and pHe and the non-bicarbonate buffer capacity of whole blood were determined by linear regression analysis, results of which are shown in detail in the supplement (Fig. S1A-D). The average values for these blood characteristics are shown in the main text and were calculated as the average slopes across all individuals.
In the RBC swelling trial, mean cell Hb was calculated as [Hb] divided by Hct as a decimal. Since Hb is a membrane impermeable solute, MCHC is used as a common indicator of RBC swelling. Main effects of drugs (DMSO, ISO and ISO+Am), time (10, 30 and 60 min) and their interaction (drug×time) on Hct, [Hb], MCHC, pHe and pHi were determined by two-way ANOVAs (lm and Anova functions in R; N = 5-6; P < 0.05) and multiple comparisons were conducted with t-tests (pairwise.t.test function in R) and controlling the false detection likelihood (FDR) with a Benjamini-Hochberg correction (p.adjust function in R).
The effect of RBC β-adrenergic stimulation on Hb-O2 binding was assessed by analysing the absorbance data from the BOBS with a custom R script (github.com/tillharter/White-Seabass-beta-NHE). In brief, the absorbances recorded at 430 nm were divided by the isosbestic wavelength of 390 nm (where absorbance is independent of Hb-O2 binding), and these ratios were used as the raw data for subsequent analyses. For each trace, the ten final absorbance ratios at each equilibration step were averaged (i.e. 10 s) and Hb-O2 saturation was calculated relative to the absorbance at the two initial calibration conditions (i.e. high: 99.7% PO2, 0.3% PCO2; and low 0% PO2, 0.3% PCO2), assuming full saturation or desaturation of Hb, respectively. These calibration values were measured again at the end of each trial and a linear correction of drift was performed for each sample. The resulting values for Hb-O2 saturation were plotted against PCO2 and several non-linear models were fit to the data (Michaelis-Menten, Exponential and Hill). The model with the best fit (lowest AIC) was a three-parameter Hill function that was applied to each individual trace. The parameter estimates from this model yielded the maximal reduction in Hb-O2 saturation (Max. ΔHb-O2 sat.; %) and the PCO2 at which this response was half-maximal (EC50PCO2; %). The main effects of drugs (DMSO, ISO and ISO+Am), O2 (normoxia and hypoxia) and their interaction (drug×O2) on the parameter estimates from the Hill functions were determined by two-way ANOVAs (lm and Anova functions in R; N = 6; P < 0.05). When significant main effects were detected, multiple comparisons were conducted with t-tests (pairwise.t.test function in R) and controlling the false detection likelihood (FDR) with a Benjamini-Hochberg correction (p.adjust function in R). The Root effect was determined as the Max. ΔHb-O2 sat. of the control treatment (DMSO). To quantify the relative changes in Hb-O2 saturation (ΔHb-O2 sat.) due to drug treatments, data were expressed relative to the paired measurements in the DMSO treatment for each individual fish and relative to the initial Hb-O2 saturation at 0.3% PCO2 (i.e. 95.6 and 55.0% Hb-O2 saturation in normoxia and hypoxia, respectively). All data are presented as means±s.e.m.
Results
Blood O2-binding characteristics
The blood O2-binding characteristics of white seabass are summarised in Figure 1. When PCO2 was increased from arterial tension (0.3%) to severe hypercapnia (2.5%) Hb P50 increased significantly from 21.4±0.5 to 88.6±2.6 mmHg. At the same time, the cooperativity of Hb-O2 binding, expressed by nH, decreased significantly from 1.52±0.04 to 0.84±0.03, which was reflected in a change in shape of the OECs from sigmoidal to hyperbolic. Over the tested range of PCO2, white seabass displayed a Bohr coefficient of -0.92±0.13 when expressed relative to the change in pHe and -1.13±0.11 when expressed relative to the change in RBC pHi (Fig. S1A and B). In addition to the reduction in Hb-O2 affinity at elevated PCO2, white seabass blood also displayed a large Root effect with a maximal reduction of Hb-O2 carrying capacity of 52.4±1.8%. The relationship between pHi and pHe had a slope of 0.67±0.07 (Fig. S1C), reflecting the higher buffer capacity of the intracellular space. The non-bicarbonate buffer capacity of white seabass whole blood was -2.43±0.56 mmol l-1 pHe-1 at a Hct of 5% (Fig. S1D). By correcting this value for the higher Hct in vivo, according to Wood et al. (1982), white seabass with a Hct of 25% are expected to have a whole blood non-bicarbonate buffer capacity of -9.68 mmol l-1 pHe-1.
Oxygen equilibrium curves showing haemoglobin-oxygen saturation (Hb-O2 sat.; %) as a function of the partial pressure of oxygen (PO2; mmHg, where 7.5 mmHg equals 1%) at five partial pressures of carbon dioxide (PCO2). The PO2 that yields 50% Hb-O2 saturation (P50) and the cooperativity coefficient of Hb-O2 binding (Hill coefficient, nH) are shown for each curve. The main effects of PCO2 on P50 and nH were analysed with ACOVA (P < 0.05, N = 8). The Bohr coefficients (B) relative to extracellular (pHe) and intracellular (pHi), the relationship between pHe and pHi and the non-bicarbonate buffer capacity of the blood (at 5% Hct) were determined by linear regressions (see Fig. S1). The Root effect (R) was determined at 21% PO2 from the model parameters shown for the DMSO treatment in Fig. 6B. All data are means±s.e.m.
RBC swelling after β-adrenergic stimulation
The β-adrenergic stimulation of white seabass blood with ISO induced changes in the measured blood parameters (Fig. 2). Significant main effects of drug, time and their interaction (drug×time), indicate that Hct was affected by the experimental treatments (Fig. 2A). A large increase in Hct was observed in ISO-treated blood that was absent in ISO+Am and DMSO-treated RBCs. In addition, a main effect of drug treatment on MCHC indicated that the increase in Hct after ISO addition was due to swelling of the RBCs, whereas [Hb] was not affected by the treatments (Fig. 2B). Significant main effects of drug and time were also detected for pHe, where a large extracellular acidification was observed in ISO-treated blood, relative to the DMSO and ISO+Am treatments (Fig. 2C). No significant main effect of drug or time were observed on RBC pHi, but multiple comparisons indicated a trend for a higher pHi in ISO compared to DMSO treated blood (P = 0.081; Fig. 2D). Differential interference contrast (DIC) images confirmed a normal morphology of the RBCs at the beginning and the end of the trials, thus validating the fixation procedure. Swelling was observed in ISO-treated RBCs, relative to initial measurements or DMSO and ISO+Am-treated cells (Fig. 2E-H). The swelling of ISO-treated RBCs occurred largely along the z-axis of the cells (indicated by the arrows), whereas no visible distortion was observed in the x-y directions.
A) Haematocrit (%), B) mean cell haemoglobin content (MCHC; mM haemoglobin l-1 red blood cells), C) extracellular pH (pHe) and D) intracellular pH (pHi). Blood was equilibrated in tonometers at 3% PO2 and 1% PCO2 and treated with either: i) a carrier control (DMSO; 0.25%), ii) the β-adrenergic agonist isoproterenol (ISO; 10 µM) or iii) ISO plus amiloride (ISO+Am; 1 mM), an inhibitor of sodium-proton exchangers (NHE). The dotted line indicates initial values for each parameter and changes were recorded over 60 min. The main effects of drug, time, and their interaction term (drug×time) were analysed with a two-way ANOVA (P < 0.05, N = 6 and N = 5 for ISO+Am). There were no significant changes in haemoglobin concentration ([Hb]; mM) throughout the trials. Multiple comparisons were with paired t-tests with a Benjamini-Hochberg correction and superscript letters that differ indicate significant differences between treatments at 60 min. Individual datapoints and means±s.e.m. Inserts E-H) differential interference contrast (DIC, 60x) images of red blood cells fixed at the beginning (ini) and the end of the trial (T60). Cell swelling was visually confirmed in ISO-treated cells, but not in other treatments, and mostly along the z-axis of the cells (arrows), while the x-y-axis seemed largely unaffected.
Molecular β-NHE characterisation
The combined gill and RBC de novo transcriptome of white seabass contained nine Slc9a1 transcripts and phylogenetic analysis placed these sequences within well-supported clades of the NHE family tree (Fig. 3). Importantly, one of these white seabass transcripts grouped with the Slc9a1b sequences from other teleost fishes, supporting its classification as a white seabass β-NHE. Results from RT-PCR, cloning and sequencing confirmed the expression of β-NHE mRNA in isolated white seabass RBCs. A search for the Kozak nucleic acid motif in the open reading frame of the white seabass β-NHE sequence yielded the five most likely potential start codons, including one that would produce a 66 kDa protein (Table S2). This size closely matched the single band that was specifically recognized by the polyclonal β-NHE antibody in Western blots with crude homogenate, cytosolic and membrane-enriched fractions of a white seabass RBC lysate (Fig. S2A); whereas no immunoreactivity was observed in lanes where the antibody had been incubated with its pre-immune peptide. The anti-α-tubulin antibody detected a single band in the RBC crude homogenate, at the predicted size of 54 kDa (Fig. S2B). Finally, the white seabass β-NHE protein sequence shared seven consecutive amino acids with the peptide used to raise the polyclonal antibodies (Fig. S2C), which is sufficient for specific antibody binding (Dunn et al., 1999). More importantly, the antigen peptide sequence was absent in the other eight white seabass NHE isoforms, ruling out non-specific antibody recognition of these NHEs.
Novel white seabass sequences are highlighted in blue and the β-NHE in red (Slc9a1b). Shadings delineate sub-families of the Slc9a1 gene family. The tree was rooted against the NHE2 from Caenorhabditis elegans in orange. Accession numbers for all species are those reported in Table S1.
Subcellular localisation of RBC β-NHE
The subcellular location of β-NHE protein in white seabass RBCs was determined by immunofluorescence cytochemistry and super-resolution confocal microscopy (Fig. 4). In DMSO-treated RBCs, the β-NHE immunolabelling was most intense in intracellular vesicle-like structures, and weaker at the plasma membrane. There was substantial heterogeneity in the staining pattern for β-NHE in these control cells, with varying amounts of intracellular and membrane staining. In ISO-treated RBCs, the staining pattern for β-NHE was more homogeneous and most cells showed intense immunoreactivity for β-NHE in the membrane that co-localised with α tubulin in the marginal band, and the intracellular, vesicle-like staining that was observed in the control cells was reduced (see Fig. S3 for an overview image with more cells). In contrast, ISO-treated cells that were incubated without the primary antibody or where the antibody was pre-absorbed with its pre-immune-peptide showed no immunoreactivity for β-NHE (Fig. S4). Finally, optical sectioning and three-dimensional reconstruction of these RBCs confirmed that the membrane staining for β-NHE occurred in a single plane and co-localised with α-tubulin in the marginal band (see 3D Movies S1 and S2).
Blood was equilibrated in tonometers at 3% PO2 and 1% PCO2 for 60 mins (see Fig. 2 for details) in the presence of either: A-D) a carrier control #x03B2;-adrenergic agonist isoproterenol (ISO; 10 µM). Fixed cells were immuno-stained with a monoclonal α-tubulin antibody to visualize the marginal band (red), with DAPI to visualize the cell nuclei (A and E), and with a polyclonal anti-β-NHE antibody (green, B and F). D and H) Magnified view of the insets in the merged images, where arrows indicate weak or absent β-NHE immunoreactivity on the membrane of Ctrl cells and intense staining in ISO-treated cells(for a larger sample size of these representative patters see Fig. S3).
Hb-O2 binding after β adrenergic stimulation
To study whether the activation of RBC β-NHEs in white seabass has a protective effect on Hb-O2 binding, blood samples were first equilibrated to 21% PO2 and 0.3% PCO2 in tonometers and no significant effects of drug treatment (DMSO, ISO or ISO+Am) were observed on any of the measured blood parameters and average values were: Hct 5.20±0.14% (P = 0.095), [Hb] 0.178±0.006 mM (P =0.889), MCHC 3.46±0.15 mM L-1 RBC (P = 0.490), pHe 7.848±0.018 (P = 0.576), pHi 7.464±0.021 (P = 0.241). Thereafter, blood was loaded into the BOBS, where Hb-O2 saturation was measured spectrophotometrically at increasing levels of a respiratory acidosis in normoxia (21% PO2). As expected from the pH-sensitivity of Hb-O2 binding in white seabass, an increase in PCO2 from 0.3 to 3% caused a severe reduction in Hb-O2 saturation in all treatments via the Root effect (Fig. 5). The raw data were analysed by fitting a three-parameter Hill model to the individual observations within each treatment and significant differences were observed in the parameter estimates that describe these models. EC50PCO2 was affected by the experimental treatments, as shown in a significant main effect of drug (Fig. 6A). Multiple comparisons confirmed significant differences in EC50PCO2, which was 0.85±0.06% in DMSO, 0.91±0.06% in ISO+Am and 1.08±0.06% in ISO-stimulated blood. In contrast, the magnitude of the responses, Max. ΔHb-O2 sat., was not affected by the experimental treatments and no significant main effect of drug was detected; the average Max. ΔHb-O2 sat. across treatments was -51.1±0.7% (Fig. 6B).
Haematocrit was set to 5%, blood was equilibrated in tonometers at 21% PO2 and 0.3% PCO2 and treated with either: i) a carrier control (DMSO; 0.25%), ii) the β-adrenergic agonist isoproterenol (ISO; 10 µM), or iii) ISO plus amiloride (ISO+Am; 1 mM), an inhibitor of sodium-proton exchangers (NHE). For each sample, runs were performed in normoxia (21% PO2; solid symbols) or hypoxia (3% PO2; open symbols). Individual datapoints and means±s.e.m. (N = 6).
A) The PCO2 that elicits a half-maximal reduction in Hb-O2 saturation (EC50PCO2; %). and B) the maximal reduction in Hb-O2 saturation due to acidification (Max. ΔHb-O2 sat.; %). Treatments were: i) a carrier control (DMSO; 0.25%), ii) the β-adrenergic agonist isoproterenol (ISO; 10 µM), or iii) ISO plus amiloride (ISO+Am; 1 mM) an inhibitor of sodium-proton exchangers (NHE). For each sample, runs were performed in normoxia (21% PO2; solid symbols) or hypoxia (3% PO2; open symbols). The main effects of drug treatments (drug), oxygen (O2) and their interaction term (drug×O2) were analysed with a two-way ANOVA (P < 0.05, N = 6). Multiple comparisons were with paired t-tests and a Benjamini-Hochberg correction and superscript letters that differ indicate significant differences between treatments for each O2 tension. Individual datapoints and means±s.e.m. (N = 6).
When the same experiment was repeated under hypoxic conditions (3% PO2), an increase in PCO2 likewise caused a severe reduction in Hb-O2 saturation, indicating that in white seabass, a Root effect can also be expressed at saturations around P50 (Fig. 5). A significant main effect of O2 indicated that cells in the hypoxic condition required a lower EC50PCO2 to achieve Max. ΔHb-O2 sat., compared to the normoxic condition (Fig. 6A). There was also a significant main effect of drug on EC50PCO2 and multiple comparisons indicated a similar pattern in the individual drug effects as in the normoxic experiment, which was further confirmed by the absence of a significant drug×O2 interaction. Finally, a significant effect of O2 on Max. ΔHb-O2 sat. (Fig. 6B) indicated a larger response magnitude in hypoxic blood, but that was unaffected by drug treatments, and the average Max. ΔHb-O2 sat. across treatments was -74.1±0.1%.
To quantify the benefit of β-adrenergic stimulation during a hypercapnic acidosis, Hb-O2 saturation was expressed relative to the paired measurements in the DMSO treatment for each individual fish and relative to the initial Hb-O2 saturation at 0.3% PCO2 (i.e. 95.6 and 55.0% Hb-O2 saturation in normoxia and hypoxia, respectively). In normoxia, the benefit of β-NHE stimulation with ISO showed a bell-shaped relationship with a maximal ΔHb-O2 sat. of 7.8±0.02% at 1% PCO2 (Fig. 7A). When NHEs were inhibited in ISO+Am blood, ΔHb-O2 sat. was only 1.9±0.4% at 1% PCO2 and significantly lower compared to the other treatments; at higher PCO2 the 95% confidence intervals overlapped with the DMSO values, indicating no difference from controls. In hypoxic blood, the ISO treatment had the largest effects on ΔHb-O2 sat. at 0.3% PCO2, with maximal values of 19.2±0.0% that decreased towards higher PCO2 (Fig. 7B). Whereas, in the ISO+Am treatment, ΔHb-O2 sat. was 6.4±0.0% at 0.3% PCO2, and significant differences to the DMSO controls were only observed at PCO2 below 1.5%.
A) in normoxia (21% PO2; solid symbols) or B) in hypoxia (3% PO2; open symbols). Treatments were: i) a carrier control β-adrenergic agonist isoproterenol (ISO; 10 µM), or iii) ISO plus amiloride (ISO+Am; 1 mM), an inhibitor of sodium-proton exchangers (NHE). Individual datapoints, means±s.e.m. and 95% confidence intervals (N = 6).
Discussion
In line with our initial hypothesis, white seabass RBCs had β-NHE activity that protected the blood O2-carrying capacity during hypoxic and hypercapnic conditions. However, not all predictions were met as expected: white seabass did not have an unusually high Hb-O2 affinity and thus, other aspects of their physiology are likely more important in determining their tolerance to hypoxia. Like other teleosts, white seabass had highly pH-sensitive Hbs, where a reduction in pH decreased both Hb-O2 affinity via the Bohr effect and Hb-O2 carrying capacity via the Root effect. Several lines of evidence corroborated the presence of a RBC β-NHE in white seabass and super-resolution imaging revealed, for the first time, the subcellular location of β-NHE protein in intracellular vesicles and on the RBC membrane. Furthermore, adrenergic stimulation induced changes in the intracellular distribution of the β-NHE that may indicate a role of protein translocation in regulating β-NHE activity. The activation of RBC β-NHEs significantly increased Hb-O2 saturation during in-vivo relevant conditions of hypercapnic acidification. In fact, the largest benefit of β-NHE activation in normoxia was observed at ∼1% PCO2, where Hb-O2 saturation increased by ∼8%. Whereas in hypoxia (3% PO2), β-NHE activity had its largest effect at arterial PCO2 (0.3%) and enhanced Hb-O2 saturation by ∼20%. However, the benefits of β-NHE activation in hypoxia decrease rapidly at higher PCO2, revealing a potential vulnerability of white seabass to combinations of these stressors.
Many hypoxia tolerant vertebrates have evolved Hbs with a high affinity for O2 (low Hb P50 values), which helps them extract the gas from the respiratory medium (Hughes, 1973; Mairbäurl, 1994; Tenney, 1995). White seabass in the present study had a Hb P50 of 21.4±0.5 mmHg (Fig.1), which is higher than the values typically found in hypoxia tolerant fishes, such carp (Cyprinus carpio) that have Hb P50 values as low as 3.6 mmHg (Brauner et al., 2001). In fact, the Hb P50 of white seabass resemble more closely the values in the well-studied rainbow trout, of 24.8 mmHg (Rummer and Brauner, 2015), a cold-stream salmonid, of no noteworthy hypoxia tolerance. However, the O2-binding affinity of Hb is a trade-off that must strike a balance between loading O2 at the gas exchange surface and unloading O2 at the tissues (Brauner and Wang, 1997). Everything else being equal, a higher Hb P50 can sustain a higher PO2 at the tissue capillaries, enhancing the diffusion gradient of O2 to the mitochondria, which is of particular benefit to those species with a high scope for exercise (Mairbäurl and Weber, 2012). Thus, it seems that a high Hb-O2 affinity is not part of the physiological mechanism that facilitates hypoxia tolerance in white seabass, but instead, a high tissue PO2 may be important to sustain exercise performance in these active piscivours.
As in other teleosts, especially those in the highly-derived group of perciformes, Hb-O2 binding in white seabass was highly pH-sensitive. An increase in PCO2 from arterial levels (0.3%) to severe hypercapnia (2.5%), caused a significant right-shift of the OEC (Fig. 1) via the Bohr effect, increasing P50 to 88.6±2.6 mmHg. When considering the corresponding changes in pHe (from 8.1 to 7.2 over the range of tested PCO2; see Fig. S1A) the Bohr coefficient in white seabass was -0.92, and slightly higher, at -1.13, when considering the changes in RBC pHi (from
7.7 to 7.0: Fig. S1B). Again, these Hb-O2 binding characteristics resemble closely those of rainbow trout, where P50 increases to 75 mmHg at 2% PCO2, yielding a Bohr coefficient (relative to pHe) of -0.87 (Rummer and Brauner, 2015). A normoxic increase in PCO2 caused a significant reduction in Hb-O2 saturation via the Root effect, and at PCO2 above 3% the O2-carrying capacity of white seabass Hb was reduced by 52.4±1.8%. These results are in line with those of other teleosts, such as rainbow trout (∼55%), tench (Tinca tinca; ∼50%) and the European perch (Perca fluviatilis; ∼70%), where the larger Root effect values may reflect the higher final PCO2 used during those trials (Berenbrink et al., 2011).
The Root effect is part of a specialized system of O2 supply to the eye and the swimbladder of teleosts, where blood is acidified in a counter-current exchanger (the rete mirabile) to produce high PO2 that bridge the large diffusion distances to the avascular retina of teleosts and inflate the swimbladder against large hydrostatic pressures (Pelster and Randall, 1998). In the course of teleost evolution there have been numerous secondary losses of the choroid and swimbladder retes. While their presence has not directly been determined in white seabass, an ancestral state reconstruction predicts no secondary loss of either rete on the teleost branch leading up to the perciformes, which include the white seabass (Berenbrink et al., 2005). In addition, all of the five independent losses of the choroid rete have coincided with a reduction of the Root effect below 40% (Berenbrink, 2007; Berenbrink et al., 2005). Thus, the large Root effect of white seabass is consistent with the presence of a choroid rete and likely critical for maintaining a high ocular PO2 that facilitates the visual acuity in these active predators (Damsgaard et al., 2020).
Vertebrate Hbs are intracellular proteins and, as such, are affected by the microenvironment within the RBC cytoplasm. Teleost β-NHEs can actively modulate Hb-O2 binding by controlling RBC pHi and several lines of evidence in our study indicate the presence of functional β-NHEs in white seabass RBCs. A combined gill and RBC transcriptome detected nine sequences belonging to the vertebrate NHE (Slc9a1) family and phylogenetic analysis classified the white seabass Slc9a1b transcript as belonging to the larger group of teleost RBC β-NHEs (Fig. 3). These findings were supported by the results from RT-PCR confirming the expression of the β-NHE in white seabass RBCs. Western blots with a polyclonal anti-trout β-NHE antibody recognized a single band at 66 kDa in white seabass RBC homogenates (Fig. S2A), which is smaller than the 84 kDa predicted based on the longest possible mRNA transcript (Table S2). However, a search for Kozak motifs revealed the five most likely potential start codons in the open reading frame of the white seabass β-NHE mRNA sequence, one of which predicted a protein size of 66 kDa that matches the protein size detected in Western blots. This predicted β-NHE isoform lacks 158 amino acids on the N-terminus, which, according to structural models for NHE proteins (Landau et al., 2007), are not essential for the transporter’s activity, but may determine a differential sensitivity to inhibitors (Landau et al., 2007; Lee et al., 2011). NHE isoforms from other teleosts have been shown to separate in Western blotting with a similar size discrepancy (Chen et al., 2017), and also show differential sensitivity to amiloride and its derivatives compared to mammalian NHEs (Blair et al., 2021).
Adrenergic stimulation of white seabass RBCs with ISO caused a ∼25% volume increase in the course of 60 min, whereas no changes in RBC volume were detected in DMSO treated cells. The swelling response was corroborated by a significant reduction in MCHC and by visually confirming the increase in cell volume under a microscope; these results closely match previous reports of RBC swelling after adrenergic stimulation in other teleosts (Nikinmaa and Huestis, 1984; Shu et al., 2017; Weaver et al., 1999). In addition, the ISO-induced swelling was completely abolished by the inhibition of NHEs in ISO+Am-treated RBCs, providing additional pharmacological support for the presence of a RBC β-NHE in white seabass. In ISO-treated RBCs, but not those treated with DMSO or ISO+Am, we observed a decrease in pHe, which is the direct result of H+ excretion by NHE activity. Corresponding changes in RBC pHi are typically smaller, due to the higher buffer capacity of the intracellular space (pHi = 0.67±0.07×pHe; Fig. 1) and the freeze-thaw method is typically associated with a larger error compared to pHe measurements. Consequently, we were not able to resolve significant treatment effects on RBC pHi, but a non-significant trend may point towards a small increase in RBC pHi. Another interesting observation in these RBC swelling trials were the changes in cell morphology due to adrenergic stimulation. The increase in cell volume was largely due to an expansion along the z-axis of the cells, whereas the dimensions in the x-y axis apparently remained unaffected. The nucleated RBCs of non-mammalian vertebrates, including fish, have a marginal band, a structural component of their cytoskeleton formed by strands of α ulin that maintains their elliptical shape in the face of shear and osmotic disturbances (Joseph-Silverstein and Cohen, 1984). The stiffness of this marginal band (Dmitrieff et al., 2017) may be a major impediment to swelling along the x-y axis forcing the cells to widen in the z-direction.
Fixed RBCs from these swelling trials were studied in more detail by super-resolution microscopy and by immunolabelling β and α tubulin (Fig. 4). All RBCs showed β immunoreactivity, corroborating the presence of β-NHE protein in these cells. In control RBCs, β-NHE protein was detected intracellularly and appeared to be confined to vesicles, while weaker staining was detected on the plasma membrane (Fig. 4C, D and S3). A similar staining pattern has been described for a NHE1-like protein in the RBCs of winter flounder (Pseudopleuronectes americanus; Pedersen et al., 2003). However, the immunolabelling of this NHE was with polyclonal antibodies raised against a region of the human NHE1 sequence (aa 631-746) that is highly conserved with both the teleost Slc9a1a and Slc9a1b isoforms. Therefore, these previous results likely include staining of several NHE isoforms including the flounder β-NHE. The antibody used in the present study showed a high specificity for the white seabass β-NHE (Fig. S2) and a confounding detection of other RBC NHE isoforms is unlikely.
Strikingly, the intracellular localization of β-NHE protein changed after incubating RBCs with ISO for 60 min. In these stimulated RBCs the staining pattern for β-NHE was more homogeneous compared to controls, with strong signal at the plasma membrane, and weaker intracellular signal (Fig. 4G, H and S3). Optical sectioning and 3D reconstructions of these cells clearly showed that the intense membrane staining for β-NHE was confined to a single plane, colocalizing with α-tubulin in the marginal band, and that this staining was mostly absent in DMSO-treated cells (Movies S1 and S2). Furthermore, the use of super-resolution microscopy allowed us to discern the subcellular orientation the β-NHE signal, which was extracellular relative to α-tubulin (Fig. 4H), thus, indicating a direct contact with the blood plasma that is essential for regulating pHi via NHE activity. Combined, these observations may point towards an adrenergically-induced translocation of β-NHE protein from the cytoplasm into the membrane of white seabass RBCs. Intracellular translocation of NHEs in response to various stimuli has been reported in other systems, such as the gills of acid infused hagfish (Parks et al., 2007), insulin-treated rat cardiomyocytes (Lawrence et al., 2010), isolated mammalian cells after acidification (Gens et al., 2007) or the initiation of Na+-glucose co-transport in intestinal epithelial cells (Zhao et al., 2004). Our results may indicate the presence of a similar mechanism for the teleost RBC β-NHE, where further research is needed to quantify the translocation of β-NHE protein, resolve its time-course, and characterise the underlying cellular mechanisms. If substantiated, these findings may open new avenues in the research of RBC pHi regulation in teleosts and perhaps other vertebrates.
The activation of RBC β-NHEs has been shown to raise pHi above the equilibrium condition and plays an important role in protecting Hb-O2 binding in teleosts (Nikinmaa, 1992). Previous work has characterised the resulting left-shift in the OEC (Jensen et al., 1983; Nikinmaa, 1983; Nikinmaa and Soivio, 1979) and the changes in arterial O2-carrying capacity due to adrenergic stimulation of the RBCs (Cossins and Richardson, 1985; Ferguson et al., 1989; Nikinmaa et al., 1984). In the present study we quantified the protective effect of β-NHE activation on Hb-O2 saturation in white seabass under environmentally relevant levels of hypercapnia and hypoxia. As expected, Hb-O2 saturation decreased significantly, due to the Root effect, when PCO2 was increased from 0.3-3% (Fig. 5). Adrenergic stimulation of the RBCs with ISO significantly delayed the reduction in Hb-O2 saturation to higher EC50PCO2 that were 1.08±0.06% in ISO compared to 0.85±0.06% in DMSO-treated blood (Fig. 6A). In ISO+Am-treated RBC, the EC50PCO2 decreased significantly to 0.91±0.06%, compared to ISO-treated cells, corroborating the involvement of the RBC β-NHE in the response. However, the EC50PCO2 of ISO+Am-treated RBCs was still significantly higher compared to DMSO controls, perhaps indicating that 1 mM amiloride did not lead to a full inhibition of the β-NHE under the tested conditions, or that other, amiloride insensitive transporters, play a role in elevating RBC pHi after adrenergic stimulation.
While β-NHE activity shifted the reduction in Hb-O2 saturation to a higher PCO2, the magnitude of the Root effect was not affected by adrenergic stimulation (Fig. 6B). No significant differences were observed in Max. ΔHb-O2 sat. in any of the tested treatments and therefore, a severe acidosis generated by high PCO2 can overwhelm the physiological capacity of the β to maintain an elevated RBC pHi. The H+ extrusion by the β-NHE is secondarily active and driven by the trans-membrane Na+ gradient created by the RBC Na+-K+-ATPase (NKA). While both NKA activity (Bourne and Cossins, 1982) and the RBC rate of O2 consumption (O2) ML increase after adrenergic stimulation (Boutilier and Ferguson, 1989), it is possible that the capacity of the NKA to maintain the larger Na+ gradients required to compensate for a greater reduction in pHi is limited, as could be the availability of ATP to fuel the exchange. In addition, H+ that are extruded by the -NHE will react with HCO3 in the plasma to form CO2 that can, once again, diffuse into the cells. This re-acidification of the cells via CO2 is part of the Jacobs-Stewart cycle and typically rate-limited by the formation of CO2 in the plasma of teleosts (Jacobs and Stewart, 1942). However, as pHe decreases, the pool of plasma H2CO3 becomes larger (Motais et al., 1989), accelerating the Jacobs-Stewart cycle and the re-acidification of the cells, which may explain, in part, why β-NHE activity is ineffective at very high PCO2.
The benefit of β activity on Hb-O2 saturation was non-linear over the range of PCO2 tested, and in normoxia the bell-shaped response had a maximum at ∼1% PCO2 (Fig. 7A). The observed relationship is likely dependent on the sigmoidal shape of the OEC, where a left-shift due to β-NHE activity has only marginal effects when Hb-O2 saturation is high and the curve is flat (Kobayashi et al., 1994). In addition, β-NHE activity in many teleosts is stimulated by high intracellular [H+] and inhibited by higher extracellular [H+] as pHe decreases, yielding a bell-shaped relationship between β activity and pH (Borgese et al., 1987). The ecological implications are noteworthy, as the protective effect of β activity on Hb-O2 binding is greatest over the range of PCO2 that white seabass are currently experiencing during severe red-tide or upwelling events. The increase in Hb-O2 saturation at these PCO2 is ∼8%, and the effect can be harnessed continuously with every pass of the RBCs through the gills. Everything else being equal, an increase in arterial O2 content can sustain a proportionally higher ML O2, increasing the scope for activity or reducing the requirements for anaerobic pathways of ATP production that can lead to a toxic accumulation of metabolic by-products, such as lactate and H+. Thus, for fish that experience a potentially life-threatening surge in PCO2, an 8% increase in arterial O2 content could make the difference between escaping into less-noxious waters or perishing in the attempt.
In the hypoxic trials, DMSO treated blood at arterial PCO2 (0.3%) had a Hb-O2 saturation of 55.0±3.3%, which was close to the target value around Hb P50 (Fig. 5). As in normoxia, an increase in PCO2 caused a significant reduction in Hb-O2 saturation, indicating the presence of a Root effect in hypoxia, which further decreased Hb-O2 saturation, even below the level of the maximal normoxic Root effect. Consequently, H+ binding to Hb must occur over nearly the entire range of the OEC, which stands in contrast to previous findings in rainbow trout where the Bohr effect and H+ binding to Hb occurred largely in the upper half of the OEC (Brauner et al., 1996; Brauner et al., 2000b). The possibility of inter-specific differences in the interaction between Hb-O2 and H+ binding cannot be resolved from the present data. However, it seems more likely that the kinetics of H+ binding that induce the Bohr effect are different from those of the Root effect, which is supported by previous work indicating different molecular mechanisms for the two effects (Brittain, 1987; Perutz and Brunori, 1982). The interacting kinetics of O2 and H+ binding to the Root effect Hbs of teleosts remain a worthwhile avenue for future research and studying a broader range of environmental and metabolic scenarios, in more species, may strengthen the important ecological implications of the present work.
As in normoxic blood, adrenergic activation of the β-NHE in hypoxia, increased Hb-O2 saturation during a hypercapnic acidosis. This protective effect of the β-NHE was reflected in a significantly higher EC50PCO2 in ISO-compared to DMSO or ISO+Am-treated RBCs (Fig. 6A). A significant main effect of O2 on EC50PCO2, would indicate that in hypoxic blood a lower PCO2 is required to desaturate Hb, compared to normoxic blood. However, this parameter estimate is influenced by the combined effects of PO2 and PCO2 on Hb-O2 saturation (by taking into account the full scale from 0-100%), which is not easily untangled statistically. Importantly, there was no drug×O2 interaction, indicating that the effect of the drugs was similar under normoxia and hypoxia, highlighting the benefit of β-NHE activation under both conditions.
In hypoxia, the benefit of β-NHE activation on Hb-O2 saturation was also non-linear over the tested range of PCO2 (Fig. 7B). β-NHE activation in hypoxic blood caused the largest increase in Hb-O2 saturation at arterial PCO2 (0.3%) and the benefits decreased markedly towards higher levels of hypercapnia; likely due to the flattening of the OEC at low Hb-O2 saturations and perhaps some inhibition of the transporter by the increasing extracellular [H+]. In hypoxic blood, β-NHE activity had a larger effect on Hb-O2 binding compared to normoxic blood, and at 0.3% PCO2, increased Hb-O2 saturation by 11±0.4%. This effect is even greater when considering that the available O2-carrying capacity is lower in hypoxia and when expressed relative to the available Hb-O2 saturation (55% in DMSO treated blood), the relative benefit of β-NHE activity was 19.2±0.0%. Many teleost β-NHEs are O2-sensitive (Gibson et al., 2000) and the larger effects of β-NHE activity on Hb-O2 saturation in hypoxia may be related to a partial inhibition of the transporter in normoxia; whereas, the effect appears to be less severe in white seabass compared to other species (Motais et al., 1987; Salama and Nikinmaa, 1988; Weaver et al., 1999). The nearly 20% increase in Hb-O2 saturation due to β-NHE activity is of great ecological significance and could be a principal pathway to safeguard arterial O2 transport and facilitate the hypoxia tolerance of white seabass in the wild. However, the present data also indicate a diminishing benefit of the β-NHE response when PCO2 increases; thus, revealing a potential vulnerability of white seabass to the combined stressors of hypoxia and hypercapnia; surviving these conditions likely requires additional behavioural and metabolic adjustments, that are yet to be determined.
Conclusion
The present results provide a thorough characterisation of the Hb-O2 binding system of white seabass, a non-model marine teleost with great ecological and economic importance in Southern California. Several lines of evidence confirmed the presence of a RBC β-NHE and super-resolution microscopy may point towards the regulation of the transporter’s activity via intracellular translocation, a potentially novel pathway that deserves a more thorough investigation. In white seabass, the activity of the RBC β-NHE provides significant protection of Hb-O2 binding during hypercapnic conditions with maximal benefits around the ecologically relevant level of ∼1% PCO2. Large benefits of β-NHE activation were also observed in hypoxia, however, with a greater sensitivity to increases in PCO2. Combined, these data indicate that RBC function plays a critical role in modulating the O2-binding characteristics of the pH-sensitive Hbs in white seabass and is likely part of the suite of physiological responses that determines their hypoxia and hypercapnia tolerance. Finally, these results also highlight a potential vulnerability of white seabass to combinations of these stressors and further research is needed to study the implications for wild fish conservation along the steadily warming and eutrophicated Californian coast and in high density aquaculture.
Competing interests
The authors declare no competing interests.
Funding
The present study was funded by a National Science Foundation (NSF) grant to MT (award no. 1754994) and AMC was supported by a SIO Postdoctoral Scholar fellowship.
Data Availability
All data are made available through github (github.com/tillharter/White-Seabass-beta-NHE) for R source codes and raw data files and NCBI for sequence data (see detailed accession numbers in manuscript and supplement).
Acknowledgements
Thanks are due to Mark Drawbridge at the Hubbs SeaWorld Research Institute (HSWRI) for generously providing the white seabass and Phil Zerofski, Jessica Hallisey, Garfield Kwan and Daniel Jio for their help with animal care.