Characterization of the hypoxia-inducible factor-1 pathway in hearts of Antarctic notothenioid fishes

Abstract

The ability of Antarctic notothenioid fishes to mount a robust molecular response to hypoxia is largely unknown. The transcription factor, hypoxia-inducible factor-1 (HIF-1), a heterodimer of HIF-1α and HIF-1β subunits, is the master regulator of oxygen homeostasis in most metazoans. We sought to determine if, in the hearts of Antarctic notothenioids, HIF-1 is activated and functional in response to either an acute heat stress or hypoxia. The red-blooded Notothenia coriiceps and the hemoglobinless icefish, Chaenocephalus aceratus, were exposed to their critical thermal maximum (CT_MAX) or hypoxia (5.0 ± 0.3 mg of O₂ L⁻¹) for 2 hrs. Additionally, N. coriiceps was exposed to 2.3 ± 0.3 mg of O₂ L⁻¹ for 12 hrs, and red-blooded Gobionotothen gibberifrons was exposed to both levels of hypoxia. Levels of HIF-1α were quantified in nuclei isolated from heart ventricles using western blotting. Transcript levels of genes involved in anaerobic metabolism, and known to be regulated by HIF-1, were quantified by real-time PCR, and lactate levels were measured in heart ventricles. Protein levels of HIF-1α increase in nuclei of hearts of N. coriiceps and C. aceratus in response to exposure to CT_MAX and in hearts of N. coriiceps exposed to severe hypoxia, yet mRNA levels of anaerobic metabolic genes do not increase in any species, nor do lactate levels increase, suggesting that HIF-1 does not stimulate metabolic remodeling in hearts of notothenioids under these conditions. Together, these data suggest that Antarctic notothenioids may be vulnerable to hypoxic events, which are likely to increase with climate warming.

Graphical Abstract

1. Introduction

The Western Antarctic Peninsula region is warming faster than all other areas in the southern hemisphere, resulting in widespread alterations in the marine ecosystem (Clarke et al., 2007; King, 1994; Meredith and King, 2005). Antarctic notothenioid fishes, which dominate the fish fauna of the Southern Ocean, are well adapted to life in their icy environment where temperatures have been less than 5°C for over 12 million years, and with little seasonal variation (Beers and Jayasundara, 2015; Eastman, 1993). Yet, evolution in these thermally stable waters has decreased the resilience of notothenioids to warming compared to temperate fish species (e.g., Bilyk and DeVries, 2011; Joyce et al., 2018a). Notably, the classical heat shock response and ability to induce the expression of heat shock proteins is markedly diminished in some species (Shin et al., 2014) and absent in most (Bilyk et al., 2018; Hofmann et al., 2000), which may be attributable, in part, to accelerated evolution, relaxed purifying selection, and exaptation within the cis-regulatory regions that bind Hsf1, the master transcriptional regulator of the heat shock response (Bogan and Place, 2019).

While many studies have characterized the responses of Antarctic notothenioid fishes to warming (e.g., Beers and Sidell, 2011; Bilyk and DeVries, 2011; Bilyk et al., 2018; Franklin et al., 2007; Joyce et al., 2018a; Rebelein et al., 2018) and ocean acidification (e.g., Davis et al., 2018; Davis et al., 2016; Enzor et al., 2013; Flynn et al., 2015; Strobel et al., 2013a; Strobel et al., 2013b), much less is known about their ability to endure hypoxia that often accompanies warming. This is a concern as the oxygen content in the world’s oceans declines (Schmidtko et al., 2017), along with habitats capable of sustaining the aerobic metabolic demands of marine organisms (Deutsch et al., 2015). It is conceivable that evolution in an oxygen-rich environment, such as the Southern Ocean, has weakened selective pressure to maintain a robust response to hypoxia in Antarctic fishes, similar to the loss of a heat shock response. Undoubtedly, hypoxia will be a formidable challenge for members of the hemoglobinless Channichthyidae family of Antarctic icefishes, with an oxygen-carrying capacity only 1/10^th that of red-blooded notothenioids (Ruud, 1954).

Temperature and hypoxia have interactive and synergistic effects (McBryan et al., 2013). In the water column above 1,000 m, ~ 50% of the decline in the world’s oxygen levels in the marine environment is attributable to the warming-induced decrease in oxygen solubility (Schmidtko et al., 2017). At the organismal level, as temperature increases, metabolic rate and oxygen demand increase in fishes, potentially outpacing oxygen supply and leading to hypoxic conditions in some tissues (Pörtner et al., 2007). Not surprisingly therefore, hypoxia and thermal tolerance are often positively correlated in fishes. For example, the critical thermal maximum (CT_MAX) of Atlantic salmon is positively correlated with hypoxia tolerance, and there is a positive correlation between CT_MAX and heart size, as well as between CT_MAX and cardiac myoglobin levels, suggesting that the ability to deliver and store oxygen within the heart limits thermal tolerance (Anttila et al., 2013). Similarly, cardiac failure occurs at CT_MAX in Antarctic notothenioids (Joyce et al., 2018b). However, supplemental oxygen does not increase CT_MAX in Antarctic notothenioids, suggesting that oxygen alone does not constrain thermal tolerance, but may be a contributing factor (Devor et al., 2016). In support of this, while hyperoxia does not increase CT_MAX in the killifish, Fundulus heteroclitus, CT_MAX is lower under hypoxic conditions (Healy and Schulte, 2012).

Fishes employ multiple behavioral, physiological, and biochemical strategies to cope with hypoxia and increase oxygen uptake and delivery to tissues, enhance the activity of anaerobic metabolic pathways, and suppress oxygen demand when possible (Farrell and Richards, 2009). Oxygen levels, sensed by chemoreceptors predominantly localized to the gill, stimulate an immediate increase in ventilation volume through an increase in ventilation rate and/or amplitude (Perry et al., 2009). However, many of the physiological and biochemical adjustments to hypoxia require changes in gene expression that are driven by the master transcriptional regulator of oxygen homeostasis in the majority of metazoans, hypoxia-inducible factor-1 (HIF-1) (Semenza, 1998), a heterodimer of HIF-1α and HIF-1β subunits (Wang and Semenza, 1995). Both HIF-1α and HIF-1β are members of the basic helix-loop-helix-Per-Arnt-Sim (bHLH-PAS) family of proteins that, in addition to hypoxia, regulate responses to xenobiotics and circadian rhythms (McIntosh et al., 2010). HIF-1β, also known as the aryl hydrocarbon receptor nuclear translocator protein (ARNT), is constitutively expressed, stable at all oxygen tensions, and heterodimerizes with several other members of the bHLH-PAS family (Wang et al., 1995). In contrast, while HIF-1α is transcribed and translated under normoxia, the protein is hydroxylated at two proline residues in an oxygen-dependent reaction catalyzed by prolyl hydroxylase, which targets the protein for polyubiquitination by the ubiquitin ligase, von Hippel Lindau protein, and degradation by the 26S proteasome (reviewed by Semenza, 2007). Under hypoxic conditions, HIF-1α accumulates and translocates into the nucleus, dimerizes with HIF-1β, and binds to hypoxia response elements (HREs), transactivating the expression of a myriad of genes that mediate, among other processes, angiogenesis, glycolysis, cell growth, and erythropoiesis (Semenza, 1998).

Recently we identified a polyglutamine/glutamic acid (polyQ/E) insertion mutation in HIF-1α of notothenioids, whose length varies with phylogeny and is longest (16 – 34 amino acids) in members of the crown family of notothenioids, the Channichthyidae (Rix et al., 2017). The impact of the polyQ/E insertion mutation on HIF-1 function is unclear. PolyQ regions can cause protein aggregation (Adegbuyiro et al., 2017), interfere with gene transcription and protein degradation (Dunah et al., 2002; Steffan et al., 2000), and disrupt calcium homeostasis and mitochondrial function (De Mario et al., 2016; Papsdorf et al., 2015). Yet polyQ regions are not uncommon in proteins, especially in those that regulate gene transcription, chromatin maintenance and signaling, and in proteins that have multiple binding partners, where polyQ regions stabilize protein-protein interactions (Schaefer et al., 2012).

We sought to determine if Antarctic notothenioids possess a functional HIF-1 pathway by exposing animals to either an acute heat stress (CT_MAX) or hypoxia to probe the activity of the HIF-1 pathway. Our studies focused on the heart ventricles because in notothenioids, as in many teleost fishes, the ventricle lacks a coronary circulation, rendering it particularly vulnerable to hypoxia, and because of the apparent role of the heart in delimiting thermal tolerance. The red-blooded Notothenia coriiceps and hemoglobinless icefish, Chaenocephalus aceratus, were exposed to their CT_MAX or mild hypoxia (5.0 ± 0.3 mg of O₂ L⁻¹ for 2 hours). N. coriiceps was also exposed to a more severe hypoxia of 2.3 ± 0.3 mg of O₂ L⁻¹ for 12 hours. Protein levels of HIF-1α were quantified in nuclei isolated from heart ventricles. Transcript levels of genes known to be regulated by HIF-1 and lactate levels were measured in heart ventricles. Additionally, the red-blooded notothenioid, Gobionotothen gibberifrons, was exposed to both the mild and severe hypoxia treatments and mRNA levels of hypoxia-sensitive genes quantified.

2. Materials and methods

2.1. Animal collection

Animals were captured in the austral fall and winter of 2015 for characterizing the activity of the HIF-1 pathway in response to exposure to CT_MAX, and in 2017 for characterizing the activity of the HIF-1 pathway in response to hypoxia. C. aceratus (Lönnberg 1906; average body mass ± SD = 1342 ± 556 g), N. coriiceps (Richardson 1844; 1420 ± 616 g), and G. gibberifrons (Lönnberg 1906; 848 ± 464 g) were captured using otter trawls. N. coriiceps and G. gibberifrons were also captured with baited pots. Fishing gear was deployed from the ARSV Laurence M. Gould off the southwestern shore of Low Island (63°30′S, 62°42′W) and in Dallmann Bay (64°08′S, 62°40′W). Animals were held in flow-through seawater tanks and transported to the U.S. Antarctic research station, Palmer Station, where they were held in circulating seawater tanks at 0±1 °C. Animals were allowed to recover from the stress of handling and transport for at least 3 days before experiments began. N. coriiceps were fed chopped fish every 2–3 days. C. aceratus and G. gibberifrons did not feed in captivity but were used within 14 days of capture. All procedures were approved by the University of Alaska, Fairbanks Institutional Animal Care Committee (570217–9).

2.2. Critical Thermal Maximum (CT_MAX)

Measurements of CT_MAX in N. coriiceps and C. aceratus were conducted in 2015 and tissues from some individuals were also used previous studies (Biederman et al., 2019b; Joyce et al., 2018b; O’Brien et al., 2018). CT_MAX was determined as the temperature at which fish lost the ability to right themselves (Beers and Sidell, 2011; Devor et al., 2016) or cardiac function failed when heated at a rate of 4 °C per hour (O’Brien et al., 2018). There was no significant difference in the CT_MAX within a species using these two criteria (O’Brien et al., 2018).

2.3. Hypoxia Exposure

N. coriiceps, C. aceratus and G. gibberifrons were exposed to a mild hypoxia of 5.0 ± 0.3 mg O₂ L⁻¹ for two hours. N. coriiceps and G. gibberifrons were also exposed to a more severe hypoxia of 2.3 ± 0.3 mg O₂ L⁻¹ for 12 hours. Levels of hypoxia were based on several prior studies, including one in which N. coriiceps survived oxygen tensions as low as 1 mg L⁻¹, whereas C. aceratus ceased ventilating between 3 and 3.5 mg L⁻¹ (Holeton, 1970). Other studies have shown that oxygen consumption rates significantly decrease in C. aceratus at 3.7 mg O₂ L⁻¹ (Hemmingsen and Douglas, 1969), and at 1.5 mg O₂ L⁻¹ (Hemmingsen et al., 1969) or 2.3 mg O₂ L⁻¹ (Heise and Abele, 2008) in N. coriiceps.

Two animals at a time were transferred into a flow-through, 56-gallon tank equipped with a submersible pump to circulate seawater. Fishes were allowed to recover from handling for two hours prior to beginning experiments. The tank was covered with clear plastic and oxygen levels were decreased at a rate of 3–4 mg O₂ L⁻¹ seawater per hour by bubbling in nitrogen. The total length of time for each experiment, including the time to decrease O₂ levels, was 3.5 hours for the mild hypoxia experiment and 14 hours for the severe hypoxia experiment. Oxygen levels were monitored with a galvanic oxygen probe connected to a dissolved oxygen process controller (Hanna Instruments HI8410) to regulate the flow of nitrogen. The oxygen probe was calibrated daily using oxygen-saturated water. Control animals were handled in the same fashion as those exposed to hypoxia except oxygen levels were maintained in the tank at 10 ± 1.0 mg O₂ L⁻¹ with flow-through seawater and air diffusers.

At the end of the CT_MAX and hypoxia exposures, animals were euthanized by a sharp blow to the head followed by spinal cord transection. Ventricles were quickly excised and allowed to contract in ice-cold notothenioid Ringer’s solution (260 mmol l⁻¹ NaCl, 2.5 mmol l⁻¹ MgCl₂, 5 mmol l⁻¹ KCl, 2.5 mmol l⁻¹ NaHCO₃, 5 mmol l⁻¹ NaH₂PO₄, pH 8.0). Nuclei were isolated as described below or hearts were flash frozen in liquid nitrogen and stored at −80 °C until use.

2.4. Isolation of Nuclei from Ventricles

Nuclei were isolated from heart ventricles of N. coriiceps and C. aceratus as described by Small et al., (2003) with some modifications. Ventricles were diced on an ice-cold stage and homogenized on ice using a Dounce homogenizer (Wheaton, Millville, NJ) with four or seven strokes with a loose fitting pestle followed by two or three strokes with a tight fitting pestle in 15 – 30 mL of homogenization buffer [300 mmol l⁻¹ sucrose, 10 mmol l⁻¹ Hepes, 5 mmol l⁻¹ KCl, 0.75 mmol l⁻¹ spermidine, 0.15 mmol l⁻¹ spermine, 0.1 mmol l⁻¹ EDTA, 0.1 mmol l⁻¹ EGTA, 0.5 mmol l⁻¹ phenylmethanesulfonyl fluoride (PMSF) and 2 μg ml⁻¹ final concentration each of aprotinin, leupeptin and pepstatin A, pH 7.5]. The crude homogenate was then filtered through eight layers of cheesecloth or sterile gauze pre-wetted with 50% homogenization buffer and 50% cushion buffer [1.8M sucrose for C. aceratus; 1.82 or 1.87M sucrose for N. coriiceps and 10 mmol l⁻¹ Hepes, 5 mmol l⁻¹ KCl, 1 mmol l⁻¹ dithiothreitol (DTT), 0.1 mmol l⁻¹ EDTA, 0.1 mmol l⁻¹ EGTA, 0.75 mmol l⁻¹ spermidine, 0.15 mmol l⁻¹ spermine, pH 7.5] into ultraclear centrifuge tubes (Beckman Coulter, Brea, CA) containing 10 mL of cushion buffer. Nuclei were collected by centrifugation at 60,000 g for 60 minutes at 4 °C in a Beckman Ultracentrifuge using a SW32 Ti swinging bucket rotor. Nuclear pellets were resuspended in storage buffer (25% glycerol, 50 mmol l⁻¹ Hepes, 3 mmol l⁻¹ MgCl₂, 0.1 mmol l⁻¹ EDTA, 1 mmol l⁻¹ DTT, 0.1 mmol l⁻¹ PMSF, and 2 mg ml⁻¹ each of aprotinin, leupeptin, and pepstatin A, pH 7.5), flash frozen in liquid nitrogen, and stored at −80 °C.

2.5. Western blotting

A polyclonal antibody for notothenioid HIF-1α was developed against a conserved 14 amino acid-length peptide antigen corresponding to residues 594–607 in C. aceratus and 570–583 in N. coriiceps (Thermo Fisher Scientific, Rockford, IL). The predicted molecular weight of HIF-1α is 84.45 kDa in N. coriiceps and 87.55 kDa in C. aceratus (Rix et al., 2017). Antibody specificity was determined using a blocking peptide and day 0 serum with western blotting (Fig. 1 supplemental data).

The protein concentration of nuclear extracts was determined using a Bradford assay (Bradford, 1976). Twenty μg of protein, along with molecular weight markers (Thermo Fisher Scientific) were separated on 10% denaturing polyacrylamide gels. Proteins were transferred to membranes (0.45 μm PVDF for CT_MAX experiments and 0.22 μm nitrocellulose for hypoxia experiments) using a semi-dry transfer TE77 electrophoresis unit (Hoefer Inc., Holliston, MA, USA). Membranes were rinsed twice quickly and once for 5 minutes in TBS-T (150 mmol L⁻¹ NaCl, 20 mmol L⁻¹ Tris, 0.05% Tween 20, pH 7.5).

To quantify total protein transferred onto the membrane, membranes were stained in Ponceau S [0.1% (w/v) Ponceau S, 5% (v/v) acetic acid] for 40 minutes, rinsed three times quickly then once for 1 minute in Milli-Q water, and scanned using an Amersham Biosciences image scanner (UTA-1100, GE Healthcare). Membranes were destained by rinsing three times quickly then once for 15 minutes in Milli-Q water, twice for 5 minutes in 0.1 mmol L⁻¹ NaOH, followed by three times for 5 minutes in Milli-Q water, twice for 5 minutes in TBS-T, and then incubated for at least one hour in Blotto [5% non-fat milk powder (Bio-Rad, Hercules, CA) in TBS-T] at room temperature. Membranes were then incubated in primary antibody, rabbit anti-notothenioid HIF-1α (custom antibody produced by Thermo Fisher Scientific, 1:500 in Blotto) for 14.5 hours at 4 °C. Membranes were washed (two quick rinses, one 15 minute rinse, followed by two 5 minute rinses) in TBS-T, incubated for one hour in horseradish peroxidase-conjugated goat anti-rabbit IgG secondary antibody (1708241, Bio-Rad, 1:15,000 in Blotto) at room temperature, and washed as described above in TBS-T.

Antibody binding was visualized using ECL Select Western Blotting Detection Reagent (GE Healthcare) and a ChemiDoc-It2 Imager (UVP). Signal intensity was quantified using ImageQuant TL v8.1 (Cytiva, Marlborough, MA, USA). The molecular weight (MW) of HIF-1α was also determined using ImageQuant TL v8.1 on western blots for CT_MAX experiments to confirm that the appropriate band for HIF-1α was quantified. The identity of HIF-1α on western blots of hypoxia experiments was confirmed by loading a sample of HIF-1α synthesized in vitro (described below). Each sample was run in duplicate (at a minimum) on separate gels (n= 4–6 individuals per species and treatment). HIF-1α protein levels were normalized to total protein as determined by Ponceau S staining.

2.6. In vitro synthesis of HIF-1α

HIF-1α was synthesized in vitro to verify antibody specificity. Full length HIF-1α sequence for N. coriiceps was generated using Phusion DNA polymerase (ThermoFisher Scientific) and primers designed from the HIF-1α sequence of N. coriiceps (KX950829.1). The resultant PCR product was inserted into a pTnT vector (Promega). E. coli were transformed with plasmids, and plasmid DNA was isolated using the QIAprep Miniprep Kit (Qiagen) and quantified using a NanoDrop ND-1000 (Thermo Fisher Scientific). The presence of the HIF-1α was verified with a restriction digest, and plasmids with inserts were sequenced by GeneWiz.

HIF-1α was expressed using TnT® Quick Coupled Transcription/Translation System (Promega) following the manufacturer’s protocol with 750 ng of DNA.

2.7. Quantitative real-time PCR (qPCR)

For quantifying transcript levels of genes for which there were no published sequences (aldolase a, ALD-A; lactate dehydrogenase-B, LDH-B; caspase-3, CASP-3 in all species and 18S and TATA-binding protein, TBP in G. gibberifrons, degenerate primers were designed (Table 1) using iCODEHOP (v1.0; Boyce et al., 2009) to amplify partial cDNA sequences so that gene-specific primers could be designed for qPCR. Degenerate primers for ALD-A were designed based on sequences for N. coriiceps (XP_010782556.1), Danio rerio (AAN04476.1), Epinephelus coioides (ADG29126.1), Oryzias latipes (NP_001278765.1), and Gasterosteus aculeatus (ENSGACP00000010267). Degenerate primers for LDH-B were designed based on sequences from N. coriiceps (XP_010787599.1), Gasterosteus aculeatus (ENSGACP00000001426), Larimichthys crocea (KKF17557.1), Lates niloticus (ACR55895.1), Plectropomus leopardus (ACR55893.1), and Plectropomus laevis (ACR55897.1). Degenerate primers for HO-2 were designed based on sequences from N. coriiceps (XP_010786435.1), Larimichthys crocea (KKF30521.1), Megalobrama amblycephala (ANQ45532.1), Danio rerio (NP_001096609.1), Scleropages formosus (KPP77677.1), and Cyprinus carpio (KTG38401.1). Degenerate primers for CASP-3 were designed using sequences from N. coriiceps (XP_010794822.1), Gasterosteus aculeatus (ENSGACP00000022402), Larimichthys crocea (NP_001290322.1), Dicentrarchus labrax (ABC70996.1), and Sebastes schlegelii (ANQ45635.1). Degenerate primers for TBP and 18S were based on those published in Urschel and O’Brien (2008). All primers and oligonucleotides were purchased from Life Technologies Corporation (Carlsbad, CA).

Table 1.

Primers used for sequencing cDNA

Primer Name	Primer Type	Sequence
ALD-A F1	Degenerate	5’-CCTACCCTTTCCTGACACCTgarcaraaraa-3’
ALD-A F2	Degenerate	5’-GCCTGTACGAGCGGTGCGCCcartayaaraa-3’
ALD-A R1	Degenerate	5’-CGCTTGATGAACTCCTCCTGrcangcytt-3’
ALD-A R2	Degenerate	5’-CGGTGCAGAGGGCACTGGTTcatngcrtt-3’
CASP-3 F1	Degenerate	5’-TCGGCCAGTGCATCATCATcaayaayaaraa-3’
CASP-3 F2	Degenerate	5’-GGACGCCAAGCCCAGCGCCCA-3’
CASP-3 R1	Degenerate	5’-ACCTTGTGGTTCACCCGGGTcatdatrtgytg-3’
CASP-3 R2	Degenerate	5’-GAGATCATGTCGCACAGGGAytgcatraacca-3’
HO-2 F 1	Degenerate	5’-CCGAGAACACCCAGTTCGTGaargaytt-3’
HO-2 F 2	Degenerate	5’-TGGCCCGGGACCTGGAGTACttytaygg-3’
HO-2 R 1	Degenerate	5’-TTCATCCGGGACCGGTACAGytgyttraa-3’
HO-2 R 2	Degenerate	5’-CATCTTGGCGGCGTArtanggrca-3’
LDH-B F 1	Degenerate	5’-CCATCCTGCTGCGGGACCTGtgygayga-3’
LHD-B F 2	Degenerate	5’-ACGTGATGGAGGACCGGCTGaarggngarat-3’
LDH-B R 1	Degenerate	5’-AGGTCCTTCAGGTCCTTCTGdatncccca-3’
LDH-B R 2	Degenerate	5’-AGGAACACCTCCTCGCCGATnccrtacatrtc-3’
18S F	Degenerate	5’-ACTGTGGYAATYCYAGAGCTAATACATGC-3’
18S R	Degenerate	5’-TRYRCTCATTCCRATTACAGGGCC-3’
TBP F	Degenerate	5’-GGAGGAGCAGCAGCGACArcarcarca-3’
TBP R	Degenerate	5’-GGATCCCACCATGTTCTGGATyttraartc-3’
HIF-1A 5’	Sequencing	5’-GGACACCAAGACATTTCTCAGC-3’
HIF-1A Start	Plasmid	5’-GTTTGCCCTCGAGGACATGGACACAGGAACTGTAC-3’
HIF-1A Stop	Plasmid	5’-TACTCTAGTCTAGATTAGATGACATTGACATGGTCCA-3’
SP6	Sequencing	5’-GATTTAGGTGACACTATAG-3’
M13F	Sequencing	5’-GGTTTTCCCAGTCACGAC-3’
M13R	Sequencing	5’-CAGGAAACAGCTATGAC-3’

Primers were named as follows: aldolase-A (ALD-A), caspase-3 (CASP-3), heme oxygenase-2 (HO-2), lactate dehydrogenase-B (LDH-B), 18S ribosomal RNA (18S), TATA-binding protein (TBP), and hypoxia inducible factor-1α (HIF-1A). The degenerate nucleotides are in lowercase letters. TBP and 18S degenerate primers were designed by Urschel and O’Brien (2008). Sequencing primers were used for Sanger sequencing of cDNA inserted into plasmids, and plasmid primers were used to amplify the HIF-1α cDNA sequence and insert it into a pTNT vector for synthesizing the protein using an in vitro transcription- translation coupled system.

Total RNA from ventricles of two individuals of N. coriiceps, C. aceratus and G. gibberifrons was isolated using the RNeasy fibrous tissue mini kit (Qiagen). cDNA was synthesized using random hexamers, ligated into pGEM-T (Promega, Madison, WI), cloned into E. coli, and sequenced with M13F and M13R primers (Table 1) by GeneWiz (South Plainfield, NJ). Sequences were aligned with Genome Compiler (version 2.2.88, Los Altos, CA), and consensus sequences were generated for each species. Only sequences with a read length > 150 base pairs and a QC score > 30 were used for generating consensus sequences.

Gene-specific primers for qPCR were designed using Primer-BLAST (National Center for Biotechnology Information (NCBI), Bethesda MD) and published sequences with at least one primer spanning a splice site (Table 2). LDH-A primers were designed using the following published sequences: N. coriiceps (GenBank accession number AF079822.1), C. aceratus (AF079819.1), and G. gibberifrons (AF079823.1). HIF-1α primers were designed using the following sequences: C. aceratus (KX950828.1), N. coriiceps (KX950829.1), and G. gibberifrons (GU362091.1). PGK-1 primers were designed based on the sequence for N. coriiceps (XM_010795477.1).

Table 2.

Primers for quantitative real-time PCR (qPCR).

Primer Name	Sequence	Average Efficiency	Amplicon Length (bp)
ALD-A F	5’-GTCACTGGCATTACCTTCCT-3’	1.01	70
ALD-A R	5’-TCATGGCGTTCAGATTGACAGA-3’
EF-1a F	5’-CTGGAAGCCAGTGAAAAGATGAC-3’	1.02	51
EF-1a R	5’-ACGCTCAACCTTCCATCCC-3’
HIF-1a F	5’-ATCCAACACCCCTCCAACAT-3’	1.01	105
HIF-1a R	5’-CTCAGTGATCCTCTCGTCACA-3’
LDH-A F	5’- CAAGCTGAAGGGTGAGGTCAT-3’	0.98	83
LDH-A R	5’- CACTGTAGTCTTTGTCTCCCACAA-3’
PGK-1 F	5’-CATGGCCTTCACCTTCCTCAAA-3’	0.85	49
PGK-1 R	5’-AGGTGCCGATCTCCATGTTGT-3’
TBP-F	5’-GAGGAGCAGTCGAGGTTAGC-3’	0.92	77
TBP-R	5’-TCCAAGAACTTAGCAGGAAAGC-3’
18S R	5’-CCGAGTCGGGAGTGGGTAAT-3’	0.83	51
18S F	5’-ACCACATCCAAGGAAGGCAG-3’
CASP-3 F	5’-TTGAAAGGAGAACAGGCATGAA-3’	1.04	59
CASP-3 R	5’-TCGCATTGGCTGCATCTACA-3’
HO-2 F	5’-CTTTTCAAGCTTGGTGCCGT-3’	0.99	81
HO-2 R	5’-GGGGTGGTCCTTGTTCCTTT-3’
LDH-B F	5’-ACGTCAACGTGTTCAAGTCCA-3’	n/a	89
LDH-B R	5’-CACGTCCACAGGGTTGGAGA-3’

The genes are abbreviated as follows: aldolase-A (ALD-A), elongation factor-1α (EF-1a), hypoxia inducible factor-1α (HIF-1a), lactate dehydrogenase-A (LDH-A), phosphoglycerate kinase-1 (PGK-1), TATA-binding protein (TBP), 18S ribosomal RNA (18S), caspase-3 (CASP-3), heme oxygenase-2 (HO-2), and lactate dehydrogenase-B (LDH-B). EF-1α primers were designed by Urschel and O’Brien (2008).

Total RNA was extracted from heart ventricles using the RNeasy fibrous tissue mini kit (Qiagen) according to the manufacturer’s protocol with one modification: each sample was treated twice with DNase I, first for 25 minutes and then for 20 minutes. RNA concentration was determined using a NanoDrop ND-1000 (Thermo Fisher Scientific) and diluted to approximately 500 ng μL⁻¹. RNA integrity was determined using a 2100 Bioanalyzer (Agilent, Santa Clara, CA) for the RNA from the CT_MAX and mild hypoxia experiments and a 4200 TapeStation (Agilent) for the RNA from the severe hypoxia experiment. Only samples with RNA integrity numbers (RIN) greater than 7.3 were used in subsequent steps. Final RNA concentrations were determined using Qubit 2.0 (Invitrogen, Carlsbad, CA). The cDNA was synthesized using TaqMan Reverse Transcriptase with random hexamers (Applied Biosystems, Foster City, CA). Samples lacking the reverse transcriptase (-RT) were prepared simultaneously as a control to ensure that genomic DNA was not amplified.

Transcript abundance was quantified using an ABI PRISM 7900HT Sequence Detection System (Applied Biosystems). Each reaction mixture contained 1X Power-Up SYBR Green PCR Master Mix (Applied Biosystems), 300 nmol L⁻¹ of each primer (except HO-2 which used 500 nmol L⁻¹) and 5 ng of cDNA (except 18S which used 50 pg). The primer concentrations were determined by evaluating primer efficiency using primer concentrations of 150 nmol L⁻¹, 300 nmol L⁻¹, and 500 nmol L⁻¹. A standard curve, prepared by serially diluting cDNA pooled from at least six individuals per species and treatment (within each experiment), was used to quantify relative transcript abundance. All measurements were made in triplicate. Two negative controls were used: one using the -RT reaction and a second in which water was used in place of cDNA. The cDNA was amplified for 40 cycles of 95 °C for 15 s and 60 °C for 1 min. Dissociation curves were analyzed to ensure that a single product was amplified. Target genes were normalized to elongation factor 1α (EF-1α) for the CT_MAX and severe hypoxia experiments, and to the geometric mean of TATA-binding protein (TBP) and 18S for the mild hypoxia experiment. The selection of the housekeeping gene used for the CT_MAX measurement was based on a prior study (Mueller et al., 2012), and for the hypoxia experiments, the housekeeping genes TBP, 18S, and EF-1α were evaluated with BestKeeper (v1; Pfaffl et al., 2004) as described previously (Orczewska et al., 2010) (n= 4–6). While levels of housekeeping genes did not change in response to experimental treatments, they were not equivalent between species and so no species comparisons were made.

2.8. Lactate quantification

Ventricles were homogenized in ice-cold 0.5 M Tris (pH 8.3) on ice using a Tissumizer (Tekmar, Cincinnati, OH) and then centrifuged for 15 min at 16,000 g at 4 °C. Supernatant was filtered through an Amicon Ultra centrifugal filter (10kDa cut-off; Millipore, Billerica, MA) for 40 min at 14,000 g at 4 °C, and filtrates were stored at −80 °C until use.

Lactate levels were quantified using a modified microplate assay as described previously (Devor et al., 2016). Filtrates were diluted (8-fold for N. coriiceps and 6-fold for C. aceratus) in 0.5 M Tris (pH 8.3). Diluted filtrate (50 μL) was mixed with 30 μL of assay buffer solution (0.5 mM NAD+ and WST-1 reagent (1:50 dilution of stock; Roche) in 0.5 M Tris, pH 8.3. Background absorbance at 440 nm was measured after 3 min using a SpectraMax Plus384 plate reader (Molecular Devices, Sunnyvale, CA) with SoftWare Softmax Pro v.6.3 (Molecular Devices). 20 μL of lactate dehydrogenase solution (0.083 U μl⁻¹ in assay buffer) was then added to the wells and final absorbance was measured after 80 min. Each sample was run in triplicate. Lactate levels were determined by subtracting background absorbance from the final absorbance and using a standard curve of known sodium-L-lactate concentrations run on every plate.

All chemicals were purchased from Sigma-Aldrich (St. Louis, MO), unless otherwise noted.

2.9. Statistical analyses

Significant differences in levels of HIF-1α in response to heat stress or mild hypoxia, and lactate levels in response to mild hypoxia, were determined using a two-way ANOVA followed by post-hoc Tukey’s test. Data were transformed as necessary to meet assumptions of normality determined using a Shapiro-Wilk test, and equal variance was determined using Levene’s test. If the data were not normally distributed or lacked equal variance following transformation, a Mann-Whitney Rank Sum Test was used to determine significant differences. Significant difference in levels of HIF-1α and lactate in hearts of N. coriiceps in response to severe hypoxia, and in transcript levels for all experiments, were determined within each species using two-sided t-tests. Data were analyzed with Sigma Plot 11.0 (Systat Software, San Jose, CA, USA) and significance was set at p<0.05. All data are presented as the mean ± sem.

3. Results

3.1. HIF-1α protein accumulates in the nucleus in response to either an acute heat stress or severe hypoxia

The CT_MAX of N. coriiceps was 16.0 ± 0.6 °C (n = 14) and 13.0 ± 0.5 °C (n= 13) for C. aceratus. Although the higher CT_MAX of N. coriiceps compared to C. aceratus resulted in a longer period of warming (~ 4 hrs versus ~ 3 hrs, respectively), HIF-1α protein levels increased in nuclei of heart ventricles to a greater extent in C. aceratus (6-fold) compared to N. coriiceps (4-fold) in response to exposure to CT_MAX (Fig. 1). HIF-1α protein levels did not increase in hearts of either C. aceratus or N. coriiceps in response to mild hypoxia (5.0 ± 0.3 mg O₂ L⁻¹ for 2 hours) but increased 5 -fold in hearts of N. coriiceps in response to severe hypoxia (2.3 ± 0.3 mg of O₂ L⁻¹ for 12 hours) (Fig. 1). The average MW of HIF-1α on the CT_MAX western blots, calculated from the MW standard, was 85.68 ± 3.42 kDa for N. coriiceps and 88.91 ± 4.46 kDa for C. aceratus (predicted MW = 84.45 kDa in N. coriiceps and 87.55 kDa in C. aceratus).

Figure 1. Protein levels of HIF-1α in nuclei of hearts from notothenioids.

Representative western blots of HIF-1α protein levels in nuclei from ventricles of C. aceratus and N. coriiceps exposed to ambient temperature and their CT_MAX (A), N. coriiceps exposed to normoxia and 2.3 mg O₂ L⁻¹ 12 hrs (C) and C. aceratus exposed 5.0 mg O₂ L⁻¹ for 2 hrs (E). Below each western blot is shown the membrane stained with Ponceau S for normalizing and quantifying HIF-1α protein levels (B, D). Values are means +/− s.e.m. The asterisk indicates a significant difference within a species and between treatments (p< 0.05). Each circle represents an individual sample (n=4–6). COR = N. coriiceps; ACE = C. aceratus.

3.2. Transcript levels of genes regulated by HIF-1 do not increase in response to an acute heat stress or hypoxia

The mRNA levels of HIF-1α and genes known to possess HREs were quantified. These included genes encoding the anaerobic metabolic enzymes LDH-A, ALD-A, and phosphoglycerate kinase-1 (PGK-1). Transcript levels of HIF-1α were also quantified in hearts of animals exposed to hypoxia. In addition, we attempted to quantify levels of LDH-B, CASP-3 and HO-2 in hearts of animals exposed to hypoxia, but mRNA levels of these genes were below the limits of detection.

HIF-1α mRNA levels did not increase in any species in response to mild or severe hypoxia but decreased in hearts of G. gibberifrons (Fig. 2B,C). Despite an increase in protein levels of HIF-1α in hearts of N. coriiceps and C. aceratus in response to an acute heat stress, and in N. coriiceps exposed to severe hypoxia, mRNA levels of known HIF-1 target genes (LDH-A, ALD-A, and PGK-1) did not change (Fig. 2A,C).

Figure 2. Transcript levels of hypoxia responsive genes.

Relative transcript levels of hypoxia responsive genes in heart ventricles of N. coriiceps (left), C. aceratus (center) and G. gibberifrons (right) held at ambient temperature or exposed to their CT_MAX (N. coriiceps and C. aceratus only) (A; n=7–10); normoxia or 5.0 mg O₂ L⁻¹ for 2 hrs (B, n=5–6) or 2.3 mg O₂ L⁻¹ 12 hrs (N. coriiceps and G. gibberifrons only) (C; n=6–7). Transcript levels were normalized to EF-1α for the CT_MAX and severe hypoxia experiments and to the geometric mean of TBP and 18S rRNA for the mild hypoxia experiments. Values are means +/− s.e.m. Each circle represents an individual. ALD-A, aldolase-A; LDH-A, lactate dehydrogenase-A; HIF-1a, hypoxia-inducible factor-1α; PGK-1, phosphoglycerate kinase-1. The asterisk indicates a significant difference (p< 0.05).

3.3. Lactate levels do not change in response to hypoxia

Lactate levels did not change in response to hypoxia but were were 1.8-fold higher in hearts N. coriiceps (23.73 ± 1.13 μmol g⁻¹) compared to C. aceratus (13.01 ± 0.95 μmol g⁻¹, Fig. 3).

Figure 3. Lactate levels in hearts of C. aceratus and N. coriiceps exposed to hypoxia.

Values are means +/− s.e.m. Each circle represents an individual (n=6). The asterisk indicates a significant difference between species (p< 0.05).

4. Discussion

Our results indicate that the HIF-1 pathway does not stimulate metabolic remodeling in response to either an acute heat stress or hypoxia in hearts of Antarctic notothenioid fishes. Although the HIF-1α protein accumulates in the nuclei of cardiac myocytes in response to these stressors, the mRNA levels of anaerobic metabolic genes known to be regulated by HIF-1 do not increase, suggesting that Antarctic notothenioids may have a limited capacity to endure hypoxia.

4.1. HIF-1α protein accumulates in hearts of notothenioids in response to an acute heat stress, but there is little evidence of a shift towards anaerobic metabolism

HIF-1α protein accumulates in response to an acute heat stress, indicative of cross-talk between pathways mediating the responses to hypoxia and thermal stress, as has been shown in other fishes (McBryan et al., 2013). For example, acclimation of channel catfish to hypoxia increases their CT_MAX (Burleson and Silva, 2011), and in zebrafish exposure to hypoxia stimulates the transcription of both HIF-1 regulated genes and heat shock proteins (Levesque et al., 2019).

Despite an increase in HIF-1α protein levels in hearts of N. coriiceps and C. aceratus in response to exposure to CT_MAX, transcript levels of anaerobic metabolic genes regulated by HIF-1 do not increase. This is consistent with prior results from our laboratory showing that the activity of LDH does not increase in hearts of either species in response to exposure to CT_MAX, although lactate levels increase (O’Brien et al., 2018). Lactate may be taken up from the blood plasma, although this is unlikely for notothenioids since their hearts, unlike most temperate fishes, have LDH isoenzyme type M4 (LDH-A) rather than LDH isoenzyme type H4 (LDH-B) (Feller et al., 1991; Fitch, 1988), which favors the reduction of pyruvate over lactate oxidation (Gesser and Poupa, 1973; Hansen and Sidell, 1983; Yang et al., 1992). Consistent with these results, LDH-B transcript levels were below the limits of detection in hearts of both N. coriiceps and C. aceratus. Constitutive LDH activity, which is higher in hearts of the hemoglobinless C. aceratus compared to N. coriiceps, may be sufficiently high to maintain anaerobic metabolic activity even if oxygen levels decline at elevated temperature (O’Brien et al., 2018). Moreover, levels of aerobic metabolic enzymes, such as citrate synthase, increase in hearts of N. coriiceps in response to exposure to CT_MAX, which may support the increase in heart rate that accompanies warming (Joyce et al., 2018b; O’Brien et al., 2018). Similarly, in response to warm acclimation, protein levels of aerobic metabolic enzymes increase in hearts of the goby, Gillichthys mirabilis (Jayasundara et al., 2015).

The increase in HIF-1α in hearts in response to exposure to CT_MAX suggests that cardiac performance may be somewhat oxygen-limited at elevated temperature. Cardiac performance contributes to thermal tolerance in several fish species, including Antarctic notothenioids (Joyce et al., 2018b). The red-blooded N. coriiceps has a CT_MAX that is ~ 3°C higher compared to the hemoglobinless C. aceratus (Joyce et al., 2018b), correlated with a greater capacity for cardiac work at elevated temperature and greater scope of cardiac output (Egginton et al., 2019). To date, the molecular underpinnings of thermal tolerance in notothenioids are not entirely known but may be attributable to membrane integrity and/or mitochondrial function and the ability to maintain ATP levels (Biederman et al., 2019a, b; O’Brien et al., 2018). While the addition of supplemental oxygen does not increase CT_MAX in either N. coriiceps or C. aceratus, suggesting that oxygen alone does not limit thermal tolerance in notothenioids, lactate levels are lower in hearts of both species exposed to CT_MAX under hyperoxia compared to those exposed to CT_MAX under normoxia, suggesting that supplemental oxygen may minimize the need for anaerobic metabolism to fuel cardiac work (Devor et al., 2016).

4.2. Despite an increase in HIF-1α in response to hypoxia, anaerobic metabolism does not increase

Similar to other fish species, HIF-1α protein accumulates in nuclei of N. coriiceps hearts in response to severe hypoxia (2.3 mg O₂ L⁻¹, 20% O₂ for 12 hr). Increases in HIF-1α protein have been observed in nuclei of rainbow trout gonad (RTG) and chinook salmon embryonic cells (CHSE) in response to hypoxia, with the greatest increase in response to 5% O₂ for 1 hr (Soitamo et al., 2001). HIF-1α levels also increase in muscle of F. heteroclitus in response to 12 hr at 0.8 mg L⁻¹ O₂ (~7.5% O₂) (Borowiec et al., 2018), and in liver, heart, gills, and kidney of C. carassius in response to 0.7 mg L⁻¹ O₂ (6–8% O₂) for 6–48 hr, although the effect was more pronounced in most tissues at low temperature (8°C versus 18°C and 26°C) (Rissanen et al., 2006), and in liver and brain of Pelteobargrus vachelli held at 0.7 mg L⁻¹ O₂ (6.5 % O₂) for 6.5 hr (Zhang et al., 2017).

Few studies in teleosts, however, have coupled measurements of changes in HIF-1α protein levels with measurements of gene expression to characterize the importance of HIF-1 in mediating biochemical and physiological responses to hypoxia in fishes. One of the few was in the catfish, P. vachelli, where an increase in HIF-1α protein levels in brain and liver in response to 0.7 mg l⁻¹ O₂ for 6.5 hr was correlated with an increase in activity of anaerobic metabolic enzymes, including LDH, as well as an increase in the HIF-1 regulated genes, prolyl hydroxylase −1 and −2 (PHD1 and PHD-2); the ubiquitin ligase, von Hippel Lindau; as well as HIF-1α, HIF-2α, and HIF-3α (Zhang et al., 2017). Transcriptomic studies also provide support for the role of HIF-1 in mediating changes in gene expression in response to hypoxia in fishes. In the goby, G. mirabilis; the medaka, Oryzias latipes (Ju et al., 2007); and in several species of sculpins (Mandic et al., 2018), transcript levels of glycolytic metabolic genes increase in liver in response to hypoxia, as well as other genes known to be regulated by HIF-1. In hearts of flounder (Platichthys flesus) and ruffe (Gymnocephalus cernua), mRNA levels of several putative hypoxia response genes, including LDH-A, increase in response to hypoxia (Tiedke et al., 2014).

As evidenced from transcriptomic studies, enhanced reliance on anaerobic metabolism to fuel the work of the heart under hypoxic conditions is common among teleost fishes and is likely driven by HIF-1 (Gamperl and Driedzic, 2009). In many fishes, especially those that are hypoxia tolerant, levels of glucose transporters and the activity of anaerobic metabolic enzymes increase as glycogen levels decrease, and the uptake of glucose, released from the liver, increases (Gamperl and Driedzic, 2009). Hearts of Antarctic notothenioids, however, preferentially oxidize fatty acids and have a lower anaerobic metabolic capacity compared to temperate teleosts, which may be a result of the greater thermal sensitivity of glycolytic enzymes compared to aerobic metabolic enzymes (Crockett and Sidell, 1990; Sidell et al., 1995). Moreover, plasma glucose levels are low (1.1–1.8 mM) in Antarctic notothenioids (Magnoni, 2002), which has been shown to limit anaerobic metabolism in some teleosts hearts under hypoxic conditions (Clow et al., 2017). Consistent with the lack of increase in mRNA levels of glycolytic enzymes, levels of lactate do not change in the heart in response to hypoxia, nor do glycogen levels in hearts of N. coriiceps exposed to severe hypoxia (2.3 mg L⁻¹ O₂, 20% O₂ for 12 hr, unpublished data). Taken together, the lack of metabolic remodeling in hearts of notothenioids in response to hypoxia suggests they may be vulnerable to hypoxia unless alternative strategies to minimize ATP demand such as bradycardia, metabolic suppression, and/or channel arrest are employed (Stecyk, 2017).

4.3. HIF-1α is post-transcriptionally regulated in notothenioids

Similar to mammals, HIF-1α appears to be post-transcriptionally regulated in notothenioids. HIF-1α protein levels increase following exposure to CT_MAX and severe hypoxia but mRNA levels do not (this study and Devor et al., 2016). This is not surprising given that the oxygen-dependent domain (ODD) that regulates HIF-1α protein stability is highly conserved in notothenioids (Rix et al., 2017). Hydroxylation of the two proline residues within the ODD likely diminishes in response to exposure to CT_MAX and hypoxia, thereby stabilizing the protein and allowing it to accumulate and translocate into the nucleus.

In addition to the hydroxylation of two proline residues within the ODD, the stability of HIF-1α is regulated by the redox state of the cell with reducing conditions favoring its stabilization in both fish and mammals (Lando et al., 2000; Nikinmaa et al., 2004). The addition of antioxidants increases HIF-1α protein levels and DNA binding in RTG-2 and CHSE cells under normoxia and hypoxia (Nikinmaa et al., 2004). Four cysteine residues within and near the C-terminal transactivation domain (CTAD) are thought to play a key role in the redox sensitivity of HIF-1α (Nikinmaa et al., 2004); both N. coriiceps and C. aceratus have three cysteine residues within, and one upstream of, the CTAD that may contribute to the low constitutive level of HIF-1α protein under normoxic, ambient conditions. This is similar to the low constitutive levels of expression of HIF-1α protein in other teleost tissues, including livers of the crucian carp, Carassius carassius (Rissanen et al., 2006) and North Sea eelpout, Zoarces viviparous (Heise et al., 2006), as well as in muscle of Fundulus heteroclitus (Borowiec et al., 2018) and rainbow trout gonad cells (RTG-2) (Soitamo et al., 2001). The more oxygen-variable environment of fishes compared to mammals may require greater constitutive expression of HIF-1α in teleosts.

4.4. A muted stress response to hypoxia?

The lack of change in gene expression of canonical HIF-1 genes in hearts of notothenioids in response to an acute heat stress or hypoxia may be due to the presence of a polyQ/E insert in HIF-1α disrupting the functionality of HIF-1 (Rix et al., 2017). In addition to nuclear localization and dimerization with HIF-1β, transactivation of gene expression by HIF-1 requires binding of the co-transcriptional activator, CREB-binding protein (CBP) or its paralog, p300 (Arany et al., 1996; Kallio et al., 1998), a histone acetyltransferase (Bannister and Kouzarides, 1996). The polyQ/E insert may impair binding of CBP/p300 or HIF-1β, potentially diminishing HIF-1 transcriptional activity. There is precedence for this conjecture: the co-transcriptional activator, PGC-1α, the master regulator of metabolic homeostasis in mammals, possesses a poly serine and Q insertion in teleosts that is thought to disrupt binding to its partner, the transcription factor NRF-1, impairing the ability of PGC-1α to drive metabolic adjustments in response to temperature and exercise in teleosts (LeMoine et al., 2010). Another potentially confounding factor is that ~ 60% of the amino acid sequence of CBP/p300 is comprised of intrinsically disordered regions, thought to be critical for its promiscuous interactions with over 300 binding partners (Dyson and Wright, 2016). In N. coriiceps, this disordered region includes multiple polyQ tracts, with the longest one containing 46 amino acids (XP_010795471.1). In contrast, HIF-1β does not contain a polyQ region [ie; the icefish Pseudochaenichthys georgianus (XP_033957812.1, isoform X1; XP_033957812.1, isoform X2) and the red-blooded dragonfish, Gymnodraco acuticeps (XP_034063890.1, isoform X1; XP_034063891.1, isoform X2; XP_034063892.1, isoform X3; 1 XP_034063893.1, isoform X4; XP_034063894.1 isoform X5)]. The extensive polyQ regions in both HIF-1α and CBP/p300 may render them incompatible binding partners.

Alternatively, it is possible that evolution in the oxygen-rich Southern Ocean has relaxed selective pressure on cis-regulatory regions that bind HIF-1, the HREs. Mutations in cis-regulatory regions causing differential gene expression drive phenotypic variation and speciation more often than mutations in protein coding regions (Zheng et al., 2011). Consistent with this, a recent study showed that the lack of an inducible heat shock response in Antarctic notothenioids is associated with relaxed selective pressure on heat shock elements (HSEs) that bind the transcription factor, Hsf-1, and within the cis-regulatory regions of chaperone genes (Bogan and Place, 2019). In addition, Antarctic notothenioids have fewer HSEs and higher mutation rates within chaperone promoters compared to subpolar notothens and other Perciformes (Bogan and Place, 2019).

We cannot rule out the possibility that we did not sample a sufficient number of genes to fully characterize the functionality of the HIF-1 pathway. We attempted to quantify transcript abundance of both CASP-3 and HO-2, two genes with HREs in N. coriiceps known to be regulated by HIF-1 (Rashid et al., 2017), however, both were below the limits of detection even in hearts of N. coriiceps exposed to severe hypoxia. Changes in gene expression in response to environmental stressors are well known to be transient (Gasch et al., 2000), and so it is also possible that we harvested tissues at an unfavorable time point following exposure to hypoxia or CT_MAX and missed changes in gene expression. A previous study investigating changes in gene expression in response to exposure to CT_MAX in notothenioids allowed the animals to recover for 2 hrs at ambient temperature to maximize changes in gene expression (Bilyk et al., 2018), whereas we harvested animals immediately following exposure to CT_MAX. Future studies aimed at characterizing genome-wide changes in gene expression in response to hypoxia will help to elucidate the functionality of HIF-1.

4.5. Conclusions and perspectives

Previous studies have shown that cardiac performance under low oxygen is associated with whole-organismal hypoxia tolerance (Joyce et al., 2016). The lack of an increase in expression of anaerobic metabolic genes in hearts of Antarctic notothenioids in response to hypoxia or an acute heat stress is consistent with their limited glycolytic capacity (Crockett and Sidell, 1990; Sidell et al., 1995) and suggests that Antarctic notothenioids may be vulnerable to hypoxic events.

Much remains to be learned about the presence of disordered polyQ regions in proteins. A preliminary screening of genomes from N. coriiceps and other temperate and tropical fishes indicates that polyQ regions of 10 amino acids or longer are more prevalent in N. coriiceps (unpublished data). Disordered regions enhance protein flexibility in the cold, and cold-adapted species, in general, tend to have more disorganized surface loop features to enhance protein-solvent interactions and decrease compactness (Feller and Gerday, 1997). However, expansion of polyQ regions are known to disrupt protein function, and the question remains whether expansion of the polyQ/E region in HIF-1α, especially in the hypoxia-sensitive icefishes (Holeton, 1970) where it is 34 amino acids long in some species, may disrupt hypoxia signaling and render notothenioids vulnerable to hypoxia as their habitat warms.

Supplementary Material

1

Supplementary Figure 1. Western blot verifying the specificity of HIF-α notothenioid primary antibody. The validity of the primary notothenioid anti-HIF-1α antibody was tested using as the primary antibody (1) the antibody mixed with the HIF-1α peptide used to generate the antibody (left panel), day 0 serum from the rabbit used to generate the antibody (center panel) and purified antibody (right panel). Samples loaded on the gel included HIF-1α synthesized using an in-vitro transcription-translation coupled system with cDNA from N. coriicep (in-vitro COR HIF-1α) and nuclear extract proteins from N. coriiceps exposed to 2.3 mg O₂ L⁻¹ for 12 hours (COR hyp2). Western blots were treated as described in the materials and methods section except that a slightly lower concentration of secondary antibody was used. NS = non-specific.

Highlights:

• Quantification of hypoxia-inducible factor −1a (HIF-1a) in Antarctic fish hearts

• HIF-1a protein increases in response to heat stress and hypoxia

• Downstream anaerobic metabolic genes are not activated

Abbreviations:

ALD-A

aldolase-A

ARNT

aryl hydrocarbon receptor nuclear translocator protein

bHLH-PAS

basic helix-loop-helix-Per-Arnt-Sim

CASP-3

caspase-3

CHSE

chinook salmon embryonic cells

CBP

CREB-binding protein

CT_MAX

critical thermal maximum

CTAD

C-terminal transactivation domain

EF-1α

elongation factor −1α

HSEs

heat shock elements

HO-2

heme oxygenase-2

HIF-1

hypoxia-inducible factor-1

HREs

hypoxia response elements

LDH

lactate dehydrogenase

ODD

oxygen-dependent domain

PMSF

phenylmethanesulfonyl fluoride

PGK-1

phosphoglycerate kinase-1

polyQ/E

polyglutamine/glutamic acid

PHD

prolyl hydroxylase

qPCR

quantitative real-time PCR

RTG

rainbow trout gonad

TBP

TATA-binding protein