Gene expression of hypoxia-inducible factor (HIF), HIF regulators, and putative HIF targets in ventricle and telencephalon of Trachemys scripta acclimated to 21 °C or 5 °C and exposed to normoxia, anoxia or reoxygenation

Abstract

In anoxia-sensitive mammals, hypoxia inducible factor (HIF) promotes cellular survival in hypoxia, but also tumorigenesis. By comparison, anoxia-tolerant vertebrates likely need to circumvent a prolonged upregulation of HIF to survive long-term anoxia, making them attractive biomedical models for investigating HIF regulation. To lend insight into the role of HIF in anoxic Trachemys scripta ventricle and telencephalon, 21 °C- and 5 °C-acclimated turtles were exposed to normoxia, anoxia (24 h at 21 °C; 24 h or 14 d at 5 °C) or anoxia + reoxygenation and the gene expression of HIF-1α (hif1a) and HIF-2α (hif2a), two regulators of HIF, and eleven putative downstream targets of HIF quantified by qPCR. Changes in gene expression with anoxia at 21 °C differentially aligned with a circumvention of HIF activity. Whereas hif1a and hif2a expression was unaffected in ventricle and telencephalon, and BCL2 interacting protein 3 gene expression reduced by 30% in telencephalon, gene expression of vascular endothelial growth factor-A increased in ventricle (4.5-fold) and telencephalon (1.5-fold), and hexokinase 1 (2-fold) and hexokinase 2 (3-fold) gene expression increased in ventricle. At 5 °C, the pattern of gene expression in ventricle or telencephalon was unaltered with oxygenation state. However, cold acclimation in normoxia induced downregulation of HIF-1α, HIF-2α, and HIF target gene expression in telencephalon. Overall, the findings lend support to the postulation that prolonged activation of HIF is counterproductive for long-term anoxia survival. Nevertheless, quantification of the effect of anoxia and acclimation temperature on HIF binding activity and regulation at the protein level are needed to provide a strong scientific framework whereby new strategies for oxygen related pathologies can be developed.

Graphical abstract

The findings lend support to the postulation that prolonged activation of HIF in heart and brain of the anoxia-tolerant turtle is counterproductive for long-term anoxia survival.

1.0. Introduction

The ability to detect and respond to oxygen deprivation is critical for all forms of aerobic life due to the critical role of oxygen in cellular metabolism. In vertebrates, hypoxia is signaled across multiple levels of biological organization (Lutz and Prentice, 2002). Central and peripheral chemoreceptors sense acute changes in oxygen availability and initiate retaliatory cardiorespiratory adjustments to maintain oxygen uptake and delivery to tissues. Intracellular molecular oxygen sensors trigger transcriptional changes in hypoxia-responsive genes involved in erythropoiesis, angiogenesis, metabolic programming, cell-cycle regulation, and tumorigenesis (Samanta and Semenza, 2018; Semenza, 2014b). Finally, a fall in cellular ATP concentration and accumulation of adenosine signals downregulation of ATP consuming processes (Pek and Lutz, 1997).

The transcriptional response to hypoxia is coordinated by hypoxia-inducible factor (HIF), of which three distinct members exist in most vertebrates, HIF-1, HIF-2, and HIF-3(Semenza, 1999). All three HIFs are heterodimers, consisting of a unique, oxygen-sensitive α-subunit and a common, oxygen-insensitive β-subunit (Duan, 2016). HIF-1α and HIF-2α are the most studied and best understood, whereas less is known about HIF-3α, owing to the existence of numerous HIF-3α variants (Duan, 2016). In normoxia, the HIF-1α and HIF-2α subunits, although continuously transcribed and translated, are rapidly targeted for proteasomal degradation. HIF-1α and HIF-2α undergo oxygen-dependent hydroxylation by HIF prolyl-hydroxylases, including hypoxia-inducible factor prolyl-hydroxylase 2 (PHD2; also known as Egl-9 family hypoxia-inducible factor 1, EGLN1) and are subsequently targeted for degradation via the von Hippel-Lindau ubiquitylation complex (Ratcliffe et al., 2017). Conversely, in hypoxia, HIF-1α and HIF-2α subunits are not hydroxylated, leading to their stabilization, translocation to the nucleus, dimerization with HIF-1β, and binding to hypoxia response elements (HREs) in target genes, which spurs transcription (Semenza, 1999). Currently, the number of HIF target genes exceeds 2500, which are unique to each HIF and cell type (Samanta and Semenza, 2018), but the assemblage includes vascular endothelial growth factor (VEGF), to stimulate angiogenesis (Ikeda et al., 2006), virtually all of the glycolytic enzymes, to alter metabolism away from oxidative phosphorylation (Semenza, 2014a), and BCL2 interacting protein 3 (BNIP3), which triggers mitochondrial-selective autophagy and apoptosis in an effort to prevent oxidative damage to cells (Bellot et al., 2009; Zhang et al., 2008).

Given that hypoxia is associated with a plethora of pathophysiological conditions in vertebrates sensitive to oxygen deprivation, including cardiovascular disease, ischemia, pulmonary artery hypertension, pressure-overload heart failure, arthritis, and inflammation, among others, HIF has been the subject of intense and extensive biomedical research (Semenza, 2014a, b). There is especially interest in developing anticancer therapies that selectively target HIF activation and regulation (Albadari et al., 2019; Jin et al., 2020; Liu and Simon, 2004). The tumor microenvironment is intrinsically hypoxic (Vaupel et al., 2001), and HIF plays a pivotal role in cancer cell survival, progression, malignancy, metastasis, and immune evasion (Ratcliffe, 2013; Samanta and Semenza, 2018; Semenza, 2013; Vaupel, 2004).

By comparison, considerably less is known about the role of HIF in the anoxia survival strategy of anoxia-tolerant vertebrates, which can survive hours, days, weeks, and even months of anoxia (Bickler and Buck, 2007; Jackson, 2000; Stecyk et al., 2008; Stecyk, 2017; Vornanen et al., 2009). Yet, anoxia-tolerant vertebrates are attractive model organisms in which novel insights regarding HIF activation and regulation that are relevant to the biomedical community promise to be discovered. Theoretically, in contrast to the beneficial effects of HIF to mediating local responses to tissue hypoxia in anoxia-sensitive vertebrates, well-adapted facultative anaerobes likely need to circumvent a prolonged upregulation of HIF and HIF responses to survive long-term global anoxia (Bickler and Buck, 2007; Milton, 2019; Storey, 2006, 2007). To allow HIF to stimulate glycolysis, angiogenesis and apoptosis in tissues would be counterproductive since the energetic cost would be too high and energy store depletion too rapid.

Indeed, key to the remarkable anoxia survival capability of freshwater turtles of the genus Chrysemys and Trachemys (Ultsch, 1985; Warren et al., 2006) is their ability to enter a hypometabolic state via the induction of an array of physiological, biochemical and molecular responses that serve to balance cellular ATP supply and demand (Bickler and Buck, 2007; Biggar et al., 2019; Bundgaard et al., 2020; Couturier et al., 2019; Jackson, 1968, 2000, 2002; Melleby et al., 2020; Stecyk et al., 2008). Notably, the response includes early activation, but then marked suppression of glycolytic enzyme activities in tissues, including the brain, heart and skeletal muscle (Brooks and Storey, 1988; Kelly and Storey, 1988; Mehrani and Storey, 1995; Smith et al., 2015). Moreover, through a process termed inverse (i.e., noncompensatory) thermal acclimation, cold acclimation in normoxia primes physiological processes for winter anoxia exposure (Hochachka, 1986; Jackson, 2000). Cold acclimation induces whole-body metabolic depression (Jackson and Ultsch, 2010; Ultsch, 1985), reduced cardiac function (Hicks and Farrell, 2000; Stecyk and Farrell, 2007; Stecyk et al., 2008; Stecyk et al., 2007; Stecyk et al., 2021), altered cardiac gene expression (Melleby et al., 2020; Stecyk et al., 2012), and brain function remodeling (Hogg et al., 2014; Lari and Buck, 2021), including the widespread downregulation of genes involved in excitatory and inhibitory neurotransmission pathways in Trachemys scripta telencephalon (Couturier et al., 2019).

Nevertheless, the response of HIF in tissues of anoxia-tolerant turtles during long term anoxia exposure remains unresolved. For anoxia exposure at warm temperature, studies on brain of T. scripta report equivocal results. HIF-1α mRNA expression remained unchanged in whole brain with 4 h of anoxia exposure and 4 h of reoxygenation at 21 °C (Prentice et al., 2003), but increased in hindbrain with 24 h of anoxia exposure (Lutz and Prentice, 2002). By comparison, brain total HIF-1α protein levels have been reported to remain unchanged with 24 h anoxia and reoxygenation, whereas nuclear levels of HIF-1α protein and HIF-1α DNA binding activity decreased, suggesting suppression of HIF in anoxia (Milton et al., 2010). Moreover, the response of HIF to anoxia at warm temperature in other tissues, such as the heart, remains unknown. For anoxia exposure at cold temperature (3-5 °C), the total and nuclear fractions of HIF-1α protein increased 2.4- and 2.0-fold, respectively, in T. scripta heart following 20 h of anoxia exposure at 4 °C(Biggar et al., 2011). However, no study has examined the effects of long-term anoxia exposure (i.e., weeks) at cold temperature on HIF expression in turtle tissues, when whole animal metabolic rate, cardiac performance, and cardiac energetic status transitions to a new, reduced ‘steady-state’ (Farrell and Stecyk, 2007; Herbert and Jackson, 1985; Stecyk et al., 2008). Finally, no study has factored in the effect of cold acclimation in normoxia on HIF expression in the anoxia-tolerant turtle.

To fill these information gaps and further advance our understanding of the HIF response in tissues of the anoxia-tolerant turtle during long-term anoxia exposure, we quantified by qPCR the gene expression of HIF-1α (hif1a) and HIF-2α (hif2a) in ventricle and telencephalon of T. scripta acclimated to 21 °C or 5 °C and exposed to normoxia, anoxia (24 h at 21 °C; 1 and 14 days at 5 °C) or anoxia followed by reoxygenation (24 h anoxia +24 h reoxygenation at 21 °C; 14 days of anoxia + 13 days of reoxygenation at 5 °C). The ventricle and telencephalon were selected for study because of their relatively high and continuous ATP demand compared to other tissues (Buck and Pamenter, 2018) and to align with past studies on HIF-1α expression (Biggar et al., 2011; Lutz and Prentice, 2002; Milton, 2019; Prentice et al., 2003) and gene expression (Couturier et al., 2019; Melleby et al., 2020; Stecyk et al., 2012) in T. scripta heart and brain. Additionally, to lend insight into HIF regulation under the various oxygenation states, we quantified the gene expression of HIF-1α inhibitor (hif1an) and hypoxia-inducible factor prolyl-hydroxylase 2 (phd2). Further, because HIF is regulated at multiple levels in addition to at the mRNA level, including protein synthesis, stability, nuclear translocation, and trans-activation (Semenza et al., 1994), we indirectly assessed the transcriptional effects of HIF during long term anoxia exposure by quantifying the effect of oxygenation state on the gene expression of eleven putative downstream targets of HIF (Table 1). Finally, given the important role of cold acclimation in priming physiological processes for winter anoxia exposure, we measured the effect of cold acclimation on gene expression. Compared to normoxia and reoxygenation, we hypothesized increased HIF-1α, HIF-2α and HIF target gene expression during the transition period into anoxia exposure at 5 °C (i.e., at 1 day of anoxia exposure), but decreased gene expression with long term anoxia exposure at 5 °C (i.e., at 14 days of anoxia exposure) and 21 °C (i.e., at 24 h of anoxia exposure). Conversely, we hypothesized hif1an and phd2 expression to be decreased during the transition period into anoxia but increased with long term anoxia exposure. With acclimation from 21 °C to 5 °C in normoxia, we hypothesized HIF-1α, HIF-2α and HIF target gene expression to be reduced.

Table 1.

Gene targets, the source/Accession # of the sequence the qPCR primers were designed from, and the primer sequences, efficiencies, and Cq values of the qPCR primers averaged from all qPCR reactions for telencephalon and ventricle.

Ge ne	Encoded protein	qPCR Primer Design Sequence Source/Accession #		Primers for qPCR	Tis su e	Eff	Cq	Am plic on Len gth
bnip1	BCL2 interacting protein 3	Reptilian-Transcriptome v1.0	F	TGCTACCAAGCATTACAGCAGAT	Tel	1.91 ± 0.06	25.1 ± 0.8	195
			R	CAAATGGTGAATGGCAGGAACTT	Vent	1.91 ± 0.02	24.4 ± 0.7
eno1	alpha-enolase	AF072588.1	F	GCGGCTCAGATACTCAAAATGTC	Tel	2.06 ± 0.13	23.1 ± 0.7	185
			R	CGAGTCTTGTCATTGTCACGAAG	Vent	1.97 ± 0.09	23.2 ± 0.6
gck(hk4)	glucokinase (hexokinase 4)	XM_024099615.1	F	CGATGACTGGAAACAGACCTACA	Tel	1.91 ± 0.03	33.3 ± 1.1	248
			R	CATGGAACTGGTCTTTGAAGCTG	Vent	1.90 ± 0.02	29.8 ± 1.0
hif1a	hypoxia-inducible transcription factor-1α	Reptilian-Transcriptome v1.0	F	GCTGCAATGGATGAAGAACTGAA	Tel	1.93 ± 0.04	25.3 ± 0.6	176
			R	CACCCTTTTACACGTTTCCAACA	Vent	1.99 ± 0.09	23.0 ± 0.9
hif1an (fih1)	hypoxia-inducible factor-1α inhibitor (factor inhibiting hypoxia-inducible factor-1α)	XM_005285544.2	F	GGATTAACAAGCAGCAAGGGAAG	Tel	1.93 ± 0.07	27.5 ± 1.6	162
			R	GATCTGGAGGGAACAGGATACAC	Vent	1.92 ± 0.02	25.2 ± 0.7
hif2a (epas1)	hypoxia-inducible transcription factor-2α (endothelial PAS domain-containing protein 1)	Reptilian-Transcriptome v1.0	F	ACTTCAAAAAGGCCACTATTTGGT	Tel	1.91 ± 0.05	25.0 ± 0.08	159
			R	GTAACCTTGCACTATTCCTTGGC	Vent	2.05 ± 0.12	22.3 ± 1.0
hk1	hexokinase 1	Reptilian-Transcriptome v1.0	F	ATCTGGATATGCAGCTTGCTCTT	Tel	2.01 ± 0.12	22.4 ± 0.8	196
			R	CCAGTTAGGCATGCAATTCTTCA	Vent	1.92 ± 0.02	25.8 ± 0.07
hk2	hexokinase 2	XM_005285230.3	F	TCTCCCCATCACTTCGCTTTTTA	Tel	1.89 ± 0.04	31.6 ± 1.3	205
			R	TTCACGTGTACATCACCAGACAT	Vent	1.89 ± 0.03	31.6 ± 1.3
phd2 (egln1)	hypoxia-inducible factor prolyl-hydroxylase 2 (Egl-9 family hypoxia-inducible factor 1)	Reptilian-Transcriptome v1.0	F	TCCTGCCGTTGATTTTGTAGTTG	Tel	1.93 ± 0.05	21.7 ± 1.1	160
			R	GAAGAGCGACTCCTCCAAGG	Vent	1.92 ± 0.03	26.6 ± 0.6
pfkp	phosphofrucktokinase, platelet	Reptilian-Transcriptome v1.0	F	GACCGAAAAGATGAAAACCAGCA	Tel	2.09 ± 0.13	21.6 ± 0.7	195
			R	GTCCCAAAGTTCCTGTCAAATG	Vent	1.99 ± 0.10	22.9 ± 0.9
pkm	pyruvate kinase M1/2	Reptilian-Transcriptome v1.0	F	TAGTCTGCACCTTTCCCCTTAAC	Tel	2.00 ± 0.11	23.5 ± 1.0	189
			R	AGCAGAGGTGGAACTTAAGAAGG	Vent	2.11 ± 0.11	21.4 ± 0.6
pygl	glycogen phosphorylase L	Reptilian-Transcriptome v1.0	F	GAATCCCAACTGCACAACTCAA	Tel	1.94 ± 0.06	21.0 ± 0.6	217
			R	CTTTTATTCCAGCACCGAAAGCA	Vent	1.91 ± 0.02	25.8 ± 0.8
vegfa	vascular endothelial growth factor A	Reptilian-Transcriptome v1.0	F	TTCATGATTCCTGGAGCAGGTC	Tel	1.92 ± 0.04	28.7 ± 0.8	183
			R	GAAACCCGAGACTGAAGAGTTCC	Vent	1.95 ± 0.10	23.8 ± 1.0
vegfc	vascular endothelial growth factor C	XM_005304171.2	F	TTTGGAGCAACAACAAACACCTT	Tel	1.92 ± 0.04	30.4 ± 0.9	170
			R	GTTACAGGTTTAGGGCCATGAGA	Vent	1.92 ± 0.02	31.2 ± 1.1
vegfd	vascular endothelial growth factor D	XM_005287790.2	F	CAGCTGCTATTTGTGTAGGGAGA	Tel	1.92 ± 0.04	31.8 ± 1.6	191
			R	CTTTGCTTTGTGACCCATGAGAG	Vent	1.92 ± 0.02	28.7 ± 0.9
mw2060	external RNA control from the cyanobacterium Microcystis cf. wesenbergi	DQ075244	F	GTGCTGACCATCCGAG	Tel	2.06 ± 0.11	22.0 ± 0.7	235
			R	GCTTGTCCGGTATAACT	Vent	2.11 ± 0.13	20.3 ± 0.8

F: forward primer; R: reverse primer.; Tel: telencephalon; Vent: ventricle; Eff: priming efficiency; Cq: quantification cycle. Values are means ± SD.

Aliases are presented in parentheses.

Reptilian Transcriptome v1.0 (Tzika et al., 2011).

2. Materials and Methods

2.1. Experimental animals and exposure groups

The University of Alaska Anchorage (UAA) Institutional Animal Care and Use Committee approved all animal procedures (852436, 852437). Sixty red-eared slider turtles (T. scripta) of both sexes and with body masses ranging between 110 and 659 g (314 ± 103 g; mean ± SD) were utilized. Turtles were obtained from Niles Biological (Sacramento, CA, USA) and transported overnight by air to UAA. Following a 2-week quarantine period at 21 °C, turtles were assigned to one of seven experimental exposure conditions: 21 °C Normoxia (21N), 21 °C Anoxia (21A), 21 °C Reoxygenated (21R), 5 °C Normoxia (5N), 5 °C 1 day Anoxia (5A1), 5 °C 14 days Anoxia (5A14), and 5°C Reoxygenated (5R). The anoxia, reoxygenation, and cold acclimation in normoxia exposure times were chosen to be consistent with previous studies examining the anoxic turtle brain at high and low temperature (Bickler, 1998; Bickler et al., 2000; Couturier et al., 2019; Hylland et al., 1997; Keenan et al., 2015; Kesaraju et al., 2009; Krivoruchko and Storey, 2010; Lutz and Leone-Kabler, 1995; Stecyk et al., 2012; Stecyk et al., 2017; Warren and Jackson, 2007) and followed protocols previously described (Couturier et al., 2019; Melleby et al., 2020; Stecyk and Farrell, 2007; Stecyk et al., 2004; Stecyk et al., 2007; Stecyk et al., 2021; Stecyk et al., 2009; Stecyk et al., 2012; Stecyk et al., 2017; Stecyk et al., 2010).

21N turtles were held indoors in 437 L polyethylene aquaria under a 12 h: 12 h light: dark photoperiod, had free access to basking platforms and diving water, and were fed commercial aquatic turtle food pellets three times a week. 21A turtles were submerged in anoxic water for 24 h by being placed in water-filled plastic chambers with a tight-fitting lid and that contained a submerged metal grate to prevent air-breathing. In addition, the chamber water was continuously bubbled with 100% N₂ to deplete the water of oxygen. 21R turtles were allowed access to atmospheric air for 24 h following anoxia exposure.

5N turtles were acclimated to cold temperature in plastic containers containing 3-4 cm of water that were placed in a commercial refrigerator (GDM-47-LD, True Manufacturing Co., O’Fallon, MO, USA). The cold acclimation period occurred in the autumn or winter months (September-December) and was 5-6 weeks to ensure adequate acclimation of the cardiovascular system to cold temperature (Hicks and Farrell, 2000). Turtles were fasted during this time.

5A1, 5A14 and 5R turtles were sub-sets of the 5N turtles. Following the cold acclimation period, 5A1 and 5A14 individuals were exposed to anoxia at 5 °C for 24 h and 14 days, respectively, using the same apparatus as at high temperature. 5R turtles were allowed access to atmospheric air for 13 days following the 14-day anoxia exposure.

At both acclimation temperatures, anoxic conditions (0.1 mg O₂ 1⁻¹; = 0.16 kPa) were confirmed at the conclusion of the anoxia exposure periods with a fibre optic FDO 925 oxygen probe and Multi 3410 meter (WTW, Weilheim, Germany).

2.2. Tissue sampling

At each sampling time, turtles were quickly removed from the exposure tank and killed by decapitation. Within 30 s of the initiation of animal handling, the telencephalon was removed and freeze-clamped in liquid N₂. Within another minute, the heart was accessed by removing the plastron with a bone-saw, excised, and rinsed in ice-cold saline (9‰ NaCl) to eliminate any residual blood. The ventricle was sub-sampled by slicing the chamber along the coronal plane at a thickness of 2 mm and a medial section containing portions of each the cavum pulmonale, cavum venosum, and cavum arteriosum was freeze-clamped in liquid N₂. All tissues were stored at −80 °C.

2.3. Quantification of gene expression

Total RNA was extracted, and cDNA synthesized from telencephalon and ventricle in accordance with protocols previously outlined in detail (Couturier et al., 2019; Ellefsen et al., 2008a; Ellefsen et al., 2008b; Melleby et al., 2020; Stecyk et al., 2020; Stecyk et al., 2012; Wilson et al., 2013). Briefly, total RNA was extracted using TRIzol reagent (Invitrogen, Carlsbad, CA, USA). An external RNA control, mw2060, a 2060-bp-long mRNA species from the cyanobacterium Microcystis cf. wesenbergi that shows no sequence homology to known vertebrate mRNA species, was added to the tissue on a per unit weight basis immediately prior to tissue homogenization to provide an external reference for qPCR quantification. The use of an external RNA control has been argued to be the most accurate method for the normalization of qPCR data (Huggett et al., 2005) and mw2060 has been shown to be a more reliable method for the normalization of qPCR data from tissues of species tolerant of low oxygen conditions than other commonly employed normalization techniques (Ellefsen et al., 2008b; Stecyk et al., 2012). One microgram of total RNA from each sample was treated with DNase I (DNA-free; Invitrogen) and subsequently reverse transcribed (iScript cDNA Synthesis Kit; Bio-Rad, Hercules, CA, USA) according to the manufactures’ protocol. Duplicate cDNA syntheses were performed on all RNA samples to allow detection of potential technical errors. Care was taken to avoid systematic errors introduced by sample processing during RNA extraction and cDNA synthesis. All samples were handled without intermission and in a systematic, yet random order. Samples were processed in groups, wherein each group consisted of samples representing each of the seven acclimation conditions. The samples within each group were processed at random.

qPCR primer pairs were designed from published T. scripta or Chrysemys picta belli nucleotide coding domain sequences (Table 1). Priority was given to designing qPCR primers against T. scripta sequences identified in Reptilian-Transcriptome v1.0 (Tzika et al., 2011) and qPCR primers were tested as previously detailed (Couturier et al., 2019; Ellefsen et al., 2008a; Ellefsen et al., 2008b; Melleby et al., 2020; Stecyk et al., 2020; Stecyk et al., 2012; Wilson et al., 2013). Briefly, Primer3 was used for primer design, and forward and reverse primers were targeted to either side of an exon-exon overlap when possible as a further precaution against amplifying genomic DNA (i.e., in addition to the extraction of total RNA and DNAse treatment). A minimum of 3 primer pairs were tested for each transcript, and the primer pair that displayed a single melting curve (CFX Manager™ software; Bio-Rad), the highest efficiency and lowest quantification cycle (Cq) values was selected for use. Amplification of the desired cDNA species by the selected primer pair was also verified by sequencing of the primer pair products (Eurofins Genomics, DNA Sequencing Services, Louisville, KY, USA) following their purification with ExoSap-IT (ThermoFisher Scientific, Waltham, MA, USA). The sequences utilized to design the qPCR primers, as well as the efficiencies and average Cq values obtained for the primer pairs utilized for the qPCR measurements are summarized in Table 1.

qPCR was performed using a CFX96 Touch™ Real-Time PCR Detection System (Bio-Rad). All qPCR reactions were performed in a reaction volume of 10 μl that contained 5 μl of SsoAdvanced™ Universal SYBR® Green Supermix (Bio-Rad), 3 μl of 1:30 diluted cDNA as the template, 1 μl of 5 mM gene-specific forward primer and 1 μl of 5 mM gene-specific reverse primer (i.e., final primer concentrations of 1 mM). The following qPCR program was used: 95 °C for 10 min, 95 °C for 10 sec, 60 °C for 30 sec. Steps 2-3 were repeated 42x. A total of four qPCR reactions were performed on each transcript for each sample of total RNA (two qPCR reactions per transcript per cDNA synthesis).

2.4. Calculations

Gene expression was quantified in relation to the mRNA external standard mw2060. Cq values were obtained for each reaction using CFX Manager™ Software (Bio-Rad) and were computed with the built-in proprietary multivariable, nonlinear regression model. Priming efficiencies were calculated for each qPCR reaction using LinRegPCR software (Ruijter et al., 2009), but in the final calculations, average priming efficiencies (E_mean) were used, calculated separately for each tissue and primer pair from all qPCR reactions (Table 1) (Cikos et al., 2007). Then, E_mean^Cq was calculated for every reaction, as well as the ratio (R1) between _mw2060E_mean^Cq and _TarE_mean^Cq to normalize mRNA expression to the expression of the external RNA control mw2060 (where Tar = target gene, E = priming efficiency and Cq = quantification cycle). To compare the expression of each gene among the different oxygen regimes for each temperature, the ratio R1 was referenced to the mean gene expression of the control normoxic turtles. Likewise, to compare the effect of 5 °C acclimation on the expression of each gene, the ratio R1 at 5 °C in normoxia was referenced to the mean gene expression in normoxia at 21 °C. Additionally, gene-family profiling analysis was conducted for the sets of genes that share base-line properties and/or fulfil similar physiological roles (i.e., hif1a and hif2a; hk1, hk2, and gck; and vegfa, vegfc, and vegfd; see Table 1) (Ellefsen and Stensløkken, 2010). For each gene family, the relative abundance (i.e., profiling) of each family member was calculated as a percentage of overall gene-family mRNA abundance.

2.5. Univariate statistical analysis

To evaluate whether the different oxygen and temperature regimes influenced gene expression, statistical analyses were performed on log 10 transformed (Hellemans and Vandesompele, 2011) target gene expression using SigmaPlot 14.5 (Aspire Software International, Ashburn, USA). One-way ANOVA followed by a Student Newman-Keuls post hoc test was used to determine statistically significant effects of anoxia and reoxygenation on gene expression at each acclimation temperature, whereas an effect of acclimation temperature on gene expression was assessed by comparing the control normoxic groups at 5 °C and 21 °C with a Student’s t-test (Couturier et al., 2019; Ellefsen et al., 2008a; Ellefsen et al., 2009; Melleby et al., 2020; Stecyk et al., 2020; Wilson et al., 2013). Gene-family profiles expressed as percentages were arcsine-transformed prior to statistical analyses. In all instances, significance was accepted when P<0.05 and P values between 0.1 and 0.05 are commented upon as tendencies.

2.6. Multivariate statistical analysis

A pairwise one-way permutational multivariate analysis of variance (PERMANOVA) was conducted using PAST 4.08 (Hammer et al., 2001) to determine statistically significant (P<0.05) differences in the pattern of gene expression in telencephalon and ventricle among the seven acclimation conditions. Principal component analysis (PCA) was conducted to visualize gene expression patterns among the different oxygen and temperature regimes. Principal components were calculated using the Factoextra package (Kassambara, 2015) in R 4.1.2 (R Development Core Team, 2021). Corresponding factor maps were created to demonstrate the contribution of the response variables (i.e., the 15 gene targets) to the first two principal components.

3. Results

3.1. HIF-1α and HIF-2α gene expression in normoxia was tissue-specific

Gene expression of HIF-1α (hif1a) and HIF-2α (hif2a) was present in normoxia (Fig. 1A and B). Regardless of acclimation temperature, the relative expression of hif1a and hif2a was approximately equal in ventricle, whereas hif2a expression dominated in telencephalon (Fig. 2A).

Fig. 1. Comparison of gene expression in ventricle and telencephalon of 21°C-acclimated and 5°C-acclimated normoxic turtles.

Gene expression of (A) hypoxia-inducible transcription factor-1α (hif1a), (B) hypoxia-inducible transcription factor-2α ( hif2a; also known as endothelial PAS domain-containing protein 1, epas1), (C) hypoxia-inducible factor-1α inhibitor (hif1an), (D) hypoxia-inducible factor prolyl-hydroxylase 2 (phd2; also known as Egl-9 family hypoxia-inducible factor 1, egln1), (E) phosphofructokinase, platelet (pfkp), (F) pyruvate kinase M1/2 (pkm), (G) hexokinase 1 (hk1), (H) hexokinase 2 (hk2), (I) glucokinase (gck; also known as hexokinase 4, hk4), (J) enolase 1 (eno1), (K) glycogen phosphorylase L (pygl), (L) BCL2 interacting protein 3 (bnip3), (M) vascular endothelial growth factor A (vegfa), (N) vascular endothelial growth factor C (vegfc), and (O) vascular endothelial growth factor D (vegfd) in ventricle (Vent) and telencephalon (Tel) of 21 °C-acclimated (21N) and 5 °C-acclimated (5N) normoxic turtles. Data sets are normalized to the external RNA control mw2060. A statistically significant difference (P < 0.05) in the expression of a transcript between ventricle and telencephalon within an acclimation temperature is indicated with an asterisk (Student’s t-test). Statistically significant differences (P < 0.05) among isoforms (i.e., hif1a vs. hif2a in panels A and B; hk1 vs. hk2 vs. gck in panels G, H and I; vegfa vs. vegfc vs. vegfd in panels M, N and O) within a tissue at each acclimation temperature are demarcated by dissimilar lowercase letters that are colour-matched to the bars (Student’s t-test or one-way ANOVA and Student Newman-Keuls post hoc test). Values are means ± SEM. N = 6-9 per exposure group.

Fig. 2. Gene-family profiling analysis for the sets of genes that share base-line properties and/or fulfil similar physiological roles.

Comparison of the relative gene expression of (A) hypoxia-inducible transcription factor-1α (hif1a) and hypoxia-inducible transcription factor-2α hif2a; also known as endothelial PAS domain-containing protein 1, epas1), (B) hexokinase 1 (hk1), hexokinase 2 (hk2), and glucokinase (gck; also known as hexokinase 4, hk4), and (C) vascular endothelial growth factor A (vegfa), vascular endothelial growth factor C (vegfc) and, vascular endothelial growth factor D (vegfd) in ventricle and telencephalon of 21°C-acclimated and 5°C-acclimated normoxic turtles. Statistical analysis: Student’s t-test or one-way ANOVA and Student Newman-Keuls post hoc test on arcsine-transformed data. Statistically significant differences (P < 0.05) between isoforms within a tissue and acclimation temperature are detailed in sections 3.1 through 3.3 or the Results. Values are means ± SEM. N = 6-9 per exposure group.

3.2. Gene expression of the regulators of HIF in normoxia was tissue-specific

In general, the gene expression of the regulators of HIF, HIF-1α inhibitor (hif1an) and hypoxia-inducible factor prolyl-hydroxylase 2 (phd2), was less than that of HIF-1α and HIF-2α in tissues of 21N and 5N turtles (Fig. 1C and D). However, regardless of acclimation temperature, hif1an expression was 2.6- to 7.7-fold greater in ventricle than telencephalon (Fig. 1C) , whereas phd2 expression was 2.0- to 2.7-fold greater in telencephalon than ventricle (Fig. 1D).

3.3. Gene expression of the putative downstream targets of HIF in normoxia was tissue- and isoform-specific

The expression of pfkp (Fig. 1E) and eno1 (Fig. 1J) did not differ between ventricle and telencephalon of normoxic turtles, regardless of acclimation temperature. By comparison, expression of gck (Fig. 1I), vegfa (Fig. 1M) and vegfd (Fig. 1O) was 7.2-, 9.1- and 2.9-times greater, respectively, in ventricle than telencephalon in 21N and 5N turtles. In contrast, pgyl (Fig. 1K) and bnip3 (Fig. 1L) expression was 9.2- and 1.9-times higher, respectively, in telencephalon than ventricle of normoxia-acclimated turtles regardless of acclimation temperature. Similarly, in 21N, but not 5N turtles, pkm (Fig. 1F), hk2 (Fig. 1H) and vegfc (Fig. 1N) expression was 2.1-, 2.6- and 4.7-fold greater, respectively, in telencephalon than ventricle.

Gene profiling revealed hk1 (Fig. 1G, H and I; Fig. 2B) and vegfa (Fig. 1M, N and O; Fig. 2C) to be the predominantly expressed hexokinase and VEGF isoform, respectively, in T. scripta ventricle and telencephalon. The expression of hk1 was 7- to 355-fold greater than hk2 and gck expression in ventricle and telencephalon, regardless of acclimation temperature (Fig. 1G-I; Fig. 2B). The expression of vefga was 3- to 110-fold greater than vegfc and vegfd expression in ventricle and telencephalon, regardless of acclimation temperature (Fig. 1M-O; Fig. 2C).

3.4. Gene expression of HIF-1α, HIF-2α, as well as the regulators of HIF, was largely unaffected by anoxia exposure and reoxygenation

Except for a 35% decrease in hif1an expression in telencephalon of 21A turtles, the expression of hif1a, hif2a, hif1an, and phd2 was unaltered by anoxia exposure and reoxygenation in ventricle and telencephalon of 21 °C-acclimated (Figs. 3A-D and 4A-D) and 5 °C-acclimated turtles (Figs. 5A-D and 6A-D). The decreased hif1an expression that occurred with 24 h of anoxia exposure in telencephalon of 21 °C-acclimated turtles returned to a level not statistically different than the control normoxic level with reoxygenation (Fig. 3C).

Fig. 3. Effect of anoxia and reoxygenation at 21°C on gene expression in ventricle.

Gene expression of (A) hypoxia-inducible transcription factor-1α (hif1a), (B) hypoxia-inducible transcription factor-2α ( hif2a; also known as endothelial PAS domain-containing protein 1, epas1), (C) hypoxia-inducible factor-1α inhibitor (hif1an), (D) hypoxia-inducible factor prolyl-hydroxylase 2 (phd2; also known as Egl-9 family hypoxia-inducible factor 1, egln1), (E) phosphofructokinase, platelet (pfkp), (F) pyruvate kinase M1/2 (pkm), (G) hexokinase 1 (hk1), (H) hexokinase 2 (hk2), (I) glucokinase (gck; also known as hexokinase 4, hk4), (J) enolase 1 (eno1), (K) glycogen phosphorylase L (pygl), (L) BCL2 interacting protein 3 (bnip3), (M) vascular endothelial growth factor A (vegfa), (N) vascular endothelial growth factor C (vegfc), and (O) vascular endothelial growth factor D (vegfd) in ventricle of 21 °C-acclimated turtles exposed to normoxia (21N), anoxia (21A) and reoxygenation (21R). Data sets are normalized to the external RNA control mw2060 and referenced to the control normoxic turtles. Statistical analysis: one-way ANOVA (P value given) and Student Newman-Keuls post hoc test. Dissimilar letters indicate statistically significant differences (P < 0.05) among exposure groups for each transcript. Shaded quadrants highlight where a statistically significant change occurred. Box plots show the 10^th, 25^th, 75^th and 90^th percentiles and median (black) and mean (white) values of gene expression. N = 6-8 per exposure group.

Fig. 4. Effect of anoxia and reoxygenation at 21°C on gene expression in telencephalon.

Gene expression of (A) hypoxia-inducible transcription factor-1α (hif1a), (B) hypoxia-inducible transcription factor-2α ( hif2a; also known as endothelial PAS domain-containing protein 1, epas1), (C) hypoxia-inducible factor-1α inhibitor (hif1an), (D) hypoxia-inducible factor prolyl-hydroxylase 2 (phd2; also known as Egl-9 family hypoxia-inducible factor 1, egln1), (E) phosphofructokinase, platelet (pfkp), (F) pyruvate kinase M1/2 (pkm), (G) hexokinase 1 (hk1), (H) hexokinase 2 (hk2), (I) glucokinase (gck; also known as hexokinase 4, hk4), (J) enolase 1 (eno1), (K) glycogen phosphorylase L (pygl), (L) BCL2 interacting protein 3 (bnip3), (M) vascular endothelial growth factor A (vegfa), (N) vascular endothelial growth factor C (vegfc), and (O) vascular endothelial growth factor D (vegfd) in telencephalon of 21 °C-acclimated turtles exposed to normoxia (21N), anoxia (21A) and reoxygenation (21R). Data sets are normalized to the external RNA control mw2060 and referenced to the control normoxic turtles. Statistical analysis: one-way ANOVA (P value given) and Student Newman-Keuls post hoc test. Dissimilar letters indicate statistically significant differences (P < 0.05) among exposure groups for each transcript. Shaded quadrants highlight where a statistically significant change occurred. Box plots show the 10^th, 25^th, 75^th and 90^th percentiles and median (black) and mean (white) values of gene expression. N = 6-8 per exposure group.

Fig. 5. Effect of anoxia and reoxygenation at 5°C on gene expression in ventricle.

Gene expression of (A) hypoxia-inducible transcription factor-1α (hif1a), (B) hypoxia-inducible transcription factor-2α ( hif2a; also known as endothelial PAS domain-containing protein 1, epas1), (C) hypoxia-inducible factor-1α inhibitor (hif1an), (D) hypoxia-inducible factor prolyl-hydroxylase 2 (phd2; also known as Egl-9 family hypoxia-inducible factor 1, egln1), (E) phosphofructokinase, platelet (pfkp), (F) pyruvate kinase M1/2 (pkm), (G) hexokinase 1 (hk1), (H) hexokinase 2 (hk2), (I) glucokinase (gck; also known as hexokinase 4, hk4), (J) enolase 1 (eno1), (K) glycogen phosphorylase L (pygl), (L) BCL2 interacting protein 3 (bnip3), (M) vascular endothelial growth factor A (vegfa), (N) vascular endothelial growth factor C (vegfc), and (O) vascular endothelial growth factor D (vegfd) in ventricle of 5 °C-acclimated turtles exposed to normoxia (5N), anoxia (5A1 and 5A14) and reoxygenation (5R). Data sets are normalized to the external RNA control mw2060 and referenced to the control normoxic turtles. Statistical analysis: one-way ANOVA (P value given) and Student Newman-Keuls post hoc test. Dissimilar letters indicate statistically significant differences (P < 0.05) among exposure groups for each transcript. Shaded quadrants highlight where a statistically significant change occurred. Box plots show the 10^th, 25^th, 75^th and 90^th percentiles and median (black) and mean (white) values of gene expression. N = 8-10 per exposure group.

Fig. 6. Effect of anoxia and reoxygenation at 5°C on gene expression in telencephalon.

Gene expression of (A) hypoxia-inducible transcription factor-1α (hif1a), (B) hypoxia-inducible transcription factor-2α ( hif2a; also known as endothelial PAS domain-containing protein 1, epas1), (C) hypoxia-inducible factor-1α inhibitor (hif1an), (D) hypoxia-inducible factor prolyl-hydroxylase 2 (phd2; also known as Egl-9 family hypoxia-inducible factor 1, egln1), (E) phosphofructokinase, platelet (pfkp), (F) pyruvate kinase M1/2 (pkm), (G) hexokinase 1 (hk1), (H) hexokinase 2 (hk2), (I) glucokinase (gck; also known as hexokinase 4, hk4), (J) enolase 1 (eno1), (K) glycogen phosphorylase L (pygl), (L) BCL2 interacting protein 3 (bnip3), (M) vascular endothelial growth factor A (vegfa), (N) vascular endothelial growth factor C (vegfc), and (O) vascular endothelial growth factor D (vegfd) in telencephalon of 5 °C-acclimated turtles exposed to normoxia (5N), anoxia (5A1 and 5A14) and reoxygenation (5R). Data sets are normalized to the external RNA control mw2060 and referenced to the control normoxic turtles. Statistical analysis: one-way ANOVA (P value given) and Student Newman-Keuls post hoc test. Dissimilar letters indicate statistically significant differences (P < 0.05) among exposure groups for each transcript. Shaded quadrants highlight where a statistically significant change occurred. Box plots show the 10^th, 25^th, 75^th and 90^th percentiles and median (black) and mean (white) values of gene expression. N = 8-10 per exposure group.

3.5. The effect of oxygenation state on the gene expression of the putative downstream targets of HIF was temperature- and tissue-specific

Much like the largely unchanged expression of hif1a, hif2a, hif1an and phd2 with oxygenation state at 21 °C and 5 °C, few changes in the gene expression of the putative downstream targets of HIF occurred with oxygenation state in ventricle and telencephalon of 21 °C- or 5 °C-acclimated turtles. Of the eleven putative HIF targets that were assessed, only four in 21 °C ventricle (hk1, hk2, vegfa, and vegfc; Fig. 3G, H, M and N), two in 21 °C telencephalon (bnip3 and vegfa; Fig. 4L and M), two in 5 °C ventricle (hk1 andpygl; Fig. 5G and K), and one in 5 °C telencephalon (pfkp; Fig. 6E) displayed a statistically significant alteration in gene expression with oxygenation state.

In 21 °C ventricle, relative to 21N, hk1 expression doubled (Fig. 3G), hk2 expression tripled (Fig. 3H) and vegfa expression increased 4.5-fold with 24 h of anoxia exposure (Fig. 3M). At 24 h of reoxygenation, hk1 and vegfa expression returned to levels not statistically different than in 21N turtles, whereas hk2 remained elevated from control normoxia. Expression of vegfc was unchanged by anoxia, but it increased 2.5-fold with reoxygenation (Fig. 3N).

In 21 °C telencephalon, bnip3 expression decreased by 30% with 24 h of anoxia exposure (Fig. 4L). By comparison, the effect of anoxia exposure and reoxygenation on vegfa expression mirrored the response of vegfa in 21 °C ventricle. VEGFA gene expression (vegfa) increased by 1.5-fold at 24 h of anoxia exposure and subsequently returned to control normoxic levels with reoxygenation (Fig. 4M).

In 5 °C-acclimated turtles, none of the putative targets of HIF exhibited an increase in gene expression with long term anoxia exposure. Rather, and in contrast to the increased hk1 expression that occurred in the ventricle with anoxia exposure at 21°C, hk1 expression decreased by 45% in 5 °C ventricle after 14 days of anoxia (Fig. 5G). The reduced hk1 expression persisted following reoxygenation (Fig. 5G). Similarly, ventricular pygl expression in 5A1 and 5A14 turtles was only 52% and 40%, respectively, of its expression in 5N turtles, and pygl expression remained depressed upon reoxygenation (Fig. 5K). In 5 °C telencephalon, pfkp expression decreased by 40% following 14 days of anoxia exposure and returned to control normoxic levels with reoxygenation (Fig. 6E).

Overall, despite the few significant changes in gene expression based on the univariate statistical analysis, the pattern of gene expression in 21A and 21R ventricle and telencephalon, as assessed by multivariate statistical analysis, was distinct from 21N (Figs. 7 and 8A; Tables 2 and 3). Notably, the loading plots reveal that in ventricle, hif1a expression was strongly correlated with PC1 and pointed in the same direction as the downstream targets of HIF-1 and HIF-2 (Fi. 7A), whereas in telencephalon, the changes in hif1a and hif1an expression influenced PC2 and were generally orthogonal to the other gene targets (Fig. 8A).

Fig. 7. Effect of cold acclimation in normoxia on gene expression in ventricle.

Gene expression of (A) hypoxia-inducible transcription factor-1α (hif1a), (B) hypoxia-inducible transcription factor-2α ( hif2a; also known as endothelial PAS domain-containing protein 1, epas1), (C) hypoxia-inducible factor-1α inhibitor (hif1an), (D) hypoxia-inducible factor prolyl-hydroxylase 2 (phd2; also known as Egl-9 family hypoxia-inducible factor 1, egln1), (E) phosphofructokinase, platelet (pfkp), (F) pyruvate kinase M1/2 (pkm), (G) hexokinase 1 (hk1), (H) hexokinase 2 (hk2), (I) glucokinase (gck; also known as hexokinase 4, hk4), (J) enolase 1 (eno1), (K) glycogen phosphorylase L (pygl), (L) BCL2 interacting protein 3 (bnip3), (M) vascular endothelial growth factor A (vegfa), (N) vascular endothelial growth factor C (vegfc), and (O) vascular endothelial growth factor D (vegfd) in ventricle of 21 °C-acclimated (21N) and 5 °C-acclimated (5N) normoxic turtles. Data sets are normalized to the external RNA control mw2060 and referenced to the 21°C turtles. Statistical analysis: Student’s t-test (P value given). * Statistically significant difference (P < 0.05) between acclimation temperatures. Shaded quadrants highlight where a statistically significant change occurred. Box plots show the 10^th, 25^th, 75^th and 90^th percentiles and median (black) and mean (white) values of gene expression. N = 6-9 per exposure group.

Fig. 8. Effect of cold acclimation in normoxia on gene expression in telencephalon.

Gene expression of (A) hypoxia-inducible transcription factor-1α (hif1a), (B) hypoxia-inducible transcription factor-2α ( hif2a; also known as endothelial PAS domain-containing protein 1, epas1), (C) hypoxia-inducible factor-1α inhibitor (hif1an), (D) hypoxia-inducible factor prolyl-hydroxylase 2 (phd2; also known as Egl-9 family hypoxia-inducible factor 1, egln1), (E) phosphofructokinase, platelet (pfkp), (F) pyruvate kinase M1/2 (pkm), (G) hexokinase 1 (hk1), (H) hexokinase 2 (hk2), (I) glucokinase (gck; also known as hexokinase 4, hk4), (J) enolase 1 (eno1), (K) glycogen phosphorylase L (pygl), (L) BCL2 interacting protein 3 (bnip3), (M) vascular endothelial growth factor A (vegfa), (N) vascular endothelial growth factor C (vegfc), and (O) vascular endothelial growth factor D (vegfd) in telencephalon of 21 °C-acclimated (21N) and 5 °C-acclimated (5N) normoxic turtles. Data sets are normalized to the external RNA control mw2060 and referenced to the 21°C turtles. Statistical analysis: Student’s t-test (P value given). * Statistically significant difference (P < 0.05) between acclimation temperatures. Shaded quadrants highlight where a statistically significant change occurred. Box plots show the 10^th, 25^th, 75^th and 90^th percentiles and median (black) and mean (white) values of gene expression. N = 6-9 per exposure group.

Table 2.

P-values (top right) and F-statistics (bottom left) of the pairwise one-way permutational multivariate analysis of variance (PERMANOVA) utilized to assess differences in the pattern of gene expression of the 15 target genes among the seven exposure conditions in T. scripta ventricle.

Exposure Condition	21N	21A	21R	5N	5A1	5A14	5R
21N	-	0.0043	0.0066	0.0337	0.0037	0.0612	0.0031
21A	5.855	-	0.1346	0.0022	0.0002	0.0004	0.0002
21R	4.157	1.774	-	0.0569	0.0069	0.0168	0.0003
5N	3.858	5.67	2.573	-	0.2022	0.093	0.0375
5A1	3.172	7.047	3.81	1.464	-	0.238	0.3722
5A14	2.177	6.699	3.362	2.243	1.259	-	0.7152
5R	2.81	8.323	4.673	3.246	1.079	0.6804	-

Permutation N: 9999, Total sum of squares: 36.93, Within-group sum of squares: 25.27, F: 3.765, P (same): 0.0001.

Significant P-values are marked in bold.

21N, 21A and 21R: normoxia, anoxia and anoxia followed by reoxygenation, respectively, at 21 °C.

5N, 5A1, 5A14 and 5R: normoxia, anoxia for 1 day, anoxia for 14 days, anoxia for 14 days followed by reoxygenation, respectively, at 5 °C.

Table 3.

P-values (top right) and F-statistics (bottom left) of the pairwise one-way permutational multivariate analysis of variance (PERMANOVA) utilized to assess differences in the pattern of gene expression of the 15 target genes among the seven exposure conditions in T. scripta telencephalon.

Exposure Condition	21N	21A	21R	5N	5A1	5A14	5R
21N	-	0.0004	0.0004	0.0001	0.0001	0.0004	0.0019
21A	4.314	-	0.0574	0.0002	0.0001	0.0003	0.0001
21R	3.636	1.982	-	0.0001	0.0003	0.0001	0.0006
5N	11.9	15.55	14.87	-	0.5863	0.4921	0.1364
5A1	9.975	13.33	11.47	0.7342	-	0.2436	0.2537
5A14	12.11	13.68	13.49	0.8281	1.334	-	0.065
5R	5.068	8.186	6.808	1.702	1.278	2.446	-

Permutation N: 9999, Total sum of squares: 31.62, Within-group sum of squares: 17.54, F: 6.951, P (same): 0.0001.

Significant P-values are marked in bold.

21N, 21A and 21R: normoxia, anoxia and anoxia followed by reoxygenation, respectively, at 21 °C.

5N, 5A1, 5A14 and 5R: normoxia, anoxia for 1 day, anoxia for 14 days, anoxia for 14 days followed by reoxygenation, respectively, at 5 °C.

In contrast to 21 °C-acclimated turtles, the overall pattern of gene expression in ventricle and telencephalon of 5 °C-acclimated turtles did not differ from 5N at 24 h or 14 days of anoxia exposure (Figs. 7 and 8B; Tables 2 and 3). However, the pattern of gene expression in ventricle was distinct from 5N with reoxygenation (Fig. 7B; Table 2).

3.6. The effect of cold acclimation in normoxia on gene expression differed between ventricle and telencephalon

In ventricle, three genes exhibited an increased expression with acclimation to 5 °C from 21 °C in normoxia. The expression of hk1, pygl and vegfc increased 2.0-, 3.2- and 1.9-fold, respectively (Fig. 9G, K and N). In contrast, in telencephalon, none of the gene targets investigated exhibited an increase in expression with cold acclimation in normoxia. Rather, 8 of the 15 genes investigated exhibited decreased gene expression (Fig. 10). These included hif1a (−42%), hif2a (−39%), hif1an (−31%), hk1 (−38%), hk2 (−54%), bnip3 (−31%), vegfc (−36%) and vegfd (−78%). Overall, cold acclimation in normoxia altered the pattern of gene expression in ventricle (Fig. 7C; Table 2) and telencephalon (Fig. 8C; Table 3). Indeed, except for 5A14 ventricle compared to 21N ventricle, all 5 °C exposure groups displayed a gene expression pattern that was distinct from the 21 °C exposure groups (Tables 2 and 3).

Fig. 9. Effect of exposure condition on the pattern of gene expression in ventricle.

Principal component analysis (PCA) plots (PC1 versus PC2) of gene expression (15 target genes) in ventricle of Trachemys scripta exposed to (A) normoxia (21N), anoxia (21A) or anoxia followed by reoxygenation (21R) at 21 °C, (B) normoxia (5N), anoxia for 1 (5A1) or 14 days (5A14) or anoxia followed by reoxygenation (5R) at 5 °C and (C) acclimated to 21 °C or 5 °C in normoxia (21N and 5N). Accompanying factor maps demonstrate the contribution of the response variables (i.e., the 15 gene targets symbolized by arrows) to the principal components. The length of the arrow is directional proportional with the contribution of variance of each gene to the total variability. The color gradient highlights the most important genes in explaining the variation (contribution %) retained by the principal components. The scree plots of eigenvalues (insets) depict the proportion of variance of the first six principal components.

Fig. 10. Effect of exposure condition on the pattern of gene expression in telencephalon.

Principal component analysis (PCA) plots (PC1 versus PC2) of gene expression (15 target genes) in telencephalon of Trachemys scripta exposed to (A) normoxia (21N), anoxia (21A) or anoxia followed by reoxygenation (21R) at 21 °C, (B) normoxia (5N), anoxia for 1 (5A1) or 14 days (5A14) or anoxia followed by reoxygenation (5R) at 5 °C and (C) acclimated to 21 °C or 5 °C in normoxia (21N and 5N). Accompanying factor maps demonstrate the contribution of the response variables (i.e., the 15 gene targets symbolized by arrows) to the principal components. The length of the arrow is directional proportional with the contribution of variance of each gene to the total variability. The color gradient highlights the most important genes in explaining the variation (contribution %) retained by the principal components. The scree plots of eigenvalues (insets) depict the proportion of variance of the first six principal components.

4. Discussion

Based on the physiological, cellular and molecular responses that endow anoxia-tolerant turtles the ability to enter a severe hypometabolic state and successfully navigate prolonged oxygen exposure (Alderman et al., 2021; Bickler and Buck, 2007; Biggar et al., 2019; Bundgaard et al., 2020; Fanter et al., 2020; Jackson, 2000, 2002; Stecyk et al., 2008; Storey, 2004, 2007), coupled with the rationale that allowing HIF to stimulate glycolysis, angiogenesis and apoptosis in tissues would be counterproductive for long term anoxia survival in the turtle since the energetic cost would be too high and energy store depletion too rapid (Bickler and Buck, 2007; Milton, 2019; Storey, 2006, 2007), we hypothesized that T. scripta exposed to long-term anoxia would exhibit downregulated gene expression of HIF-1α, HIF-2α, as well as the putative targets of HIF, but upregulated gene expression of the inhibitors of HIF in ventricle and telencephalon. Our findings were more nuanced, with the results revealing temperature- and tissue-specific effects of oxygenation state on the gene expression of HIF-1α, HIF-2α, two regulators of HIF, and eleven putative downstream targets of HIF.

4.1. Changes in gene expression with oxygenation state in 21 °C-acclimated turtles

The changes in gene expression that occurred in ventricle and telencephalon of 21 °C-acclimated T. scripta exposed to prolonged anoxia exposure differentially align with the notion that anoxia-tolerant vertebrates such as T. scripta need to circumvent a prolonged upregulation of HIF to survive long-term anoxia. Changes in gene expression that suggest a circumvention of a HIF response in T. scripta tissues with anoxia exposure include the unchanged hif1a, hif2a and phd2 expression in ventricle and telencephalon, the absence of increased gene expression of most of the putative downstream targes of HIF that were assessed, and the 30% reduction of bnip3 expression in telencephalon. On the other hand, the 25% decrease in hif1an expression in telencephalon, 1.5-fold increase in vegfa expression in telencephalon, and 4.5-, 2.0- and 3.0-fold increases in vegfa, hk1, and hk2 expression, respectively, in ventricle at 24 h of anoxia exposure, coupled with the distinct pattern of gene expression in ventricle and telencephalon in anoxia compared to normoxia, point towards a role of HIF in the response to anoxia at warm acclimation temperature in T. scripta ventricle and telencephalon.

The unchanged hif1a and hif2a expression in telencephalon despite altered oxygenation state is consistent with prior reports of unchanged HIF-1α mRNA (Prentice et al., 2003) and HIF-1α total protein (Milton et al., 2010) expression in T. scripta brain with anoxia exposure and reoxygenation at warm temperature. However, the findings contrast with the increased HIF-1α and HIF-1β mRNA expression reported for turtle hindbrain, as well as the increased HIF-1β mRNA expression reported for turtle forebrain with 24 h of anoxia exposure at warm temperature (Lutz and Prentice, 2002). The present findings also contrast with the decreased nuclear levels of HIF-1α protein reported for T. scripta brain with 24 h of anoxia exposure at warm temperature (Milton et al., 2010). The discrepancies among studies could reflect differences in anoxia exposure design (i.e., turtles exposed to 100% N₂ vs. turtles submerged in anoxic water), regional differences in gene expression (i.e., forebrain vs. hindbrain), differences in measurement technique (semiquantitative assessment of RT-PCR products via gel electrophoresis vs. qPCR), and/or that post-translational regulation of HIF, rather than transcriptional control of HIF, may predominate in T. scripta tissues (Biggar et al., 2011; Prentice et al., 2003).

The 1.5-fold increase in vegfa expression in telencephalon with anoxia at 21 °C closely aligns with the 2-fold increase in VEGF transcript and protein levels previously reported in brain of T. scripta exposed to 24 h of anoxia at warm acclimation temperature (Milton et al., 2010). Combined, the past and present findings suggest that angiogenesis may be stimulated during anoxia exposure at warm temperature in turtle heart and brain, which aligns with the maintenance of relative systemic blood flow distribution and absolute blood flow to the heart and brain in vivo during anoxia exposure at 21 °C (Stecyk et al., 2004). Although, anoxia exposure of adult T. scripta also triggers enhanced expression of SERPINC1 (also known as PEDF) in heart (Storey, 2006), which antagonizes VEGF and opposes vascular growth (Duh et al., 2002).

The elevated hk1 and hk2 expression in T. scripta ventricle with anoxia at 21 °C is consistent with the upregulation of glycolysis that occurs in turtle tissues at the onset of anoxia exposure, as evidenced by the quick depletion of brain, heart and skeletal muscle glycogen stores (Daw et al., 1967; Lutz and Nilsson, 1993; Wasser et al., 1991), but it is inconsistent with the suppression of glycolytic enzyme activity (Brooks and Storey, 1988; Kelly and Storey, 1988) and protein expression (Smith et al., 2015) that occurs with extended anoxia exposure. The contrariety could reflect that changes in gene expression persist longer than changes to protein regulation and expression, and/or that RNA degradation is reduced in turtle heart during prolonged anoxia exposure as an energy conserving mechanism. Indeed, total RNA (Douglas et al., 1994; Stecyk et al., 2012) and poly(A)+ RNA (Douglas et al., 1994) levels are unaffected by anoxia exposure or aerobic recovery in T. scripta ventricle, despite changes in specific genes (Stecyk et al., 2012) and the suppression of protein synthesis (Bailey and Driedzic, 1997). Clearly, the inclusion of an additional sampling time at a shorter duration of anoxia exposure at 21 °C would have been informative in elucidating the time course of the alterations in gene expression.

The 30% decrease in bnip3 expression in telencephalon with anoxia exposure at 21 °C is consistent with the prominent role that a coordinated enhancement of anti-apoptotic pathways plays in the brain during mammalian hibernation and prolonged torpor in vertebrates to aid cellular preservation (Rouble et al., 2013; Smith et al., 2015). The reduced bnip3 expression, as opposed to the HIF-1 mediated increase in bnip3 expression that occurs in mammalian tissues during times of oxygen limitation to minimize oxidative phosphorylation and ROS production (Semenza, 2013), could reflect that T. scripta is endowed with impressive ROS production management strategies to defend tissues against oxidative reperfusion injuries during and following prolonged bouts of anoxia (Bundgaard et al., 2019a; Bundgaard et al., 2018; Bundgaard et al., 2019b; Bundgaard et al., 2020; Hermes-Lima et al., 2001; Milton et al., 2007).

With reoxygenation, hk1 and vegfa expression returned to levels not statistically different than in 21N turtles, whereas hk2 remained elevated from control normoxia and vegfc expression increased. In mammals, VEGF-C mRNA abundance responds to several growth stimulatory agents such as platelet-derived growth factor (PDGF), endothelial growth factor (EGF) and tumor growth factor β (TGF-β), indicating that VEGF-C protein may be induced in response to tissue injury and repair (Enholm et al., 1997). Assuming VEGF-C protein expression reflects gene expression, the increased vegfc expression in T. scripta ventricle upon reoxygenation at 21 °C could signify that the turtle may experience some degree of anoxia-induced cellular damage in ventricle upon reoxygenation at warm acclimation temperature, and that VEGF-C may be involved in the repair of this damage. Indeed, the anoxia-tolerant crucian carp (Carassius carassius) is not able to fully protect its brain from the cellular insult that arises from reoxygenation post-prolonged anoxia exposure and the fish initiates neurogenesis and cellular mechanisms to repair the damage (Lefevre et al., 2017). Similarly, the persistent elevation of hk2 expression in the ventricle could reflect a role of hexokinase 2 in protecting ventricular tissue from ROS damage upon reoxygenation. In mammals, hexokinase 2 facilitates protective cellular autophagy (Roberts et al., 2014) and minimizes ROS accumulation within cardiac and other tissues (McCommis et al.; Roberts et al., 2013; Sun et al., 2008).

4.2. Changes in gene expression with oxygenation state in 5 °C-acclimated turtles

In contrast to 21 °C-acclimated turtles, gene expression of HIF-1α, HIF-2α and the regulators of HIF remained unchanged from normoxia in ventricle and telencephalon of 5 °C-acclimated T. scripta with short-term (1 day) and long-term (14 days) anoxia exposure, as well as upon reoxygenation. Moreover, anoxia exposure at 5 °C did not induce a distinct pattern of gene expression in ventricle or telencephalon. Rather, anoxia exposure at 5 °C resulted in reduced hk1 (−45%) and pygl (−60%) expression in ventricle, and reduced pfkp (−30%) expression in telencephalon. Combined, the findings are compatible with the postulation that prolonged activation of HIF is counterproductive for long-term anoxia survival. Indeed, the downregulated hk1, pygl, and pfkp expression is consistent with the extreme hypometabolism (>90%) and suppression of cardiac and neuronal activity displayed by cold-acclimated, anoxic turtles (Bickler and Buck, 2007; Buck and Pamenter, 2018; Couturier et al., 2019; Farrell and Stecyk, 2007; Herbert and Jackson, 1985; Stecyk et al., 2008). Notably, the 30% reduction of pfkp gene expression in telencephalon is in accord with the 30% reduction of PFK enzyme activity in T. scripta brain following 20 h of anoxic submergence at 7 °C (Willmore et al., 2001). Further, the predominantly unaltered gene expression of the metabolic-related gene targets with prolonged anoxia exposure at 5 °C is consistent with the previous finding that the activity of enzymes associated with intermediary metabolic pathways is largely unchanged with anoxia exposure at cold temperature (Willmore et al., 2001). Specifically, of 21 enzymes assessed in liver, kidney, heart, brain, and red and white skeletal muscle of T. scripta, only 13 were altered with 20 h of anoxia at 7 °C (Willmore et al., 2001).

The unchanged hif1a and hif2a expression in T. scripta ventricle with anoxia exposure at 5 °C contrasts with increased total and nuclear fractions of HIF-1α protein following 20 h of anoxia exposure at cold temperature (Biggar et al., 2011). Thus, like for 21 °C-acclimated turtles, in aggregate, the past and present results indicate that post-translational regulation of HIF-1, rather than the transcriptional control of HIF-1 may predominate in tissues of cold-acclimated T. scripta.

4.3. Ventricle and telencephalon displayed differential responses to cold acclimation in normoxia

Cold acclimation is instrumental for priming anoxia-tolerant turtles for ensuing winter anoxia (Eliason and Stecyk, 2020; Herbert and Jackson, 1985; Jackson, 2000; Stecyk et al., 2008; Stecyk, 2017). However, recent studies have revealed that non-compensatory inverse thermal compensation is not a universal response in the anoxia-tolerant turtle, but that it is tissue- and physiological process-specific. For instance, the effect of cold acclimation on peak current density of ionic currents involved in cardiac contraction is ion channel specific (Stecyk et al., 2007). Additionally, whereas the density of Na⁺-K⁺-ATPase pumps is reduced by 22% in turtle ventricle with acclimation to 5 °C from 25 °C in normoxia (Overgaard et al., 2005), Na⁺-K⁺-ATPase activity is unaltered with cold acclimation in normoxia in turtle brain (Stecyk et al., 2017). The differential response of gene expression in turtle ventricle and telencephalon in response to cold acclimation supports the premise that inverse thermal compensation in turtle is tissue- and physiological process-dependent.

The upregulated expression of hk1, pygl, and vegfc in turtle ventricle with cold acclimation could reflect a means to overcome Q₁₀ effects on enzyme activity that may be required to prime the heart to successfully navigate the initial onset of anoxic submergence, prior to their subsequent downregulation with anoxia exposure. Similarly, the upregulation of vegfc could reflect a priming of the turtle heart against the cellular insult arising from cyclical periods of anoxic submergence and reoxygenation that turtles are likely to experience in nature until ice-cover becomes permanent. Indeed, markedly increased gene expression of heat shock proteins with cold acclimation in the turtle cardiac chambers may be important for protecting the heart during winter anoxia (Stecyk et al., 2012). The broad decrease of gene expression in the telencephalon with cold acclimation aligns with the cold acclimation induced remodeling of brain function in anoxia-tolerant turtles (Hogg et al., 2014; Lari and Buck, 2021) and the reduced gene expression of cellular components of excitatory and inhibitory neurotransmission (Couturier et al., 2019).

In other organisms, cold temperature, in addition to oxygen deprivation, is an important factor modulating HIF-mediated gene expression. In the anoxia-tolerant crucian carp, HIF-1α protein, but not mRNA levels, increased with cold acclimation in liver, gills, and heart (Rissanen et al., 2006). Additionally, HIF-1 DNA binding activity increased with cold acclimation in heart, gills, and kidney (Rissanen et al., 2006). In the freeze-tolerant goldenrod gall fly (Eurosta solidaginis), cold exposure (3 °C) and freezing (−16 °C) significantly increased both HIF-1α mRNA and protein levels relative to 15 °C control (Morin et al., 2005). In the present study, hif1a expression was not altered in response to cold acclimation in turtle ventricle, but some putative downstream targets of HIF-1α (hk1 and pygl) were upregulated or tended towards upregulation and (pfkp). Thus, transcription factors other than HIF-1 may be involved in regulating the changes in gene expression. By comparison, in telencephalon, the decreased hif1a and hif2a expression was mimicked by 5 of the 11 putative downstream targets of HIF that were examined. These changes could reflect an oxygen-independent influence of cold temperature on HIF activity.

4.4. Conclusions, biomedical perspectives, study limitations and future directions

Our findings lend support to the postulation that activation of HIF and HIF-mediated responses is unproductive and energetically wasteful during long-term anoxia exposure in the anoxia-tolerant red-eared slider turtle. In addition, our results uncovered the novel finding that cold acclimation in normoxia is an important factor modulating HIF-related gene expression in T. scripta, but that the ventricle and telencephalon exhibit contrasting responses to cold acclimation in normoxia in terms of putative HIF-mediated gene expression. Combined, the findings reinforce that anoxia-tolerant turtles present an excellent model system to explore biomedically relevant questions (Bundgaard et al., 2019a; Bundgaard et al., 2020; Milton, 2019; Smith et al., 2015), including elucidating the cellular mechanisms of HIF regulation to provide a foundation for new ideas and novel therapies for oxygen related pathologies. Nevertheless, an important caveat for the present results is that the measurements focused solely on gene expression, whereas HIF-1 protein synthesis, stability, and nuclear translocation are also regulated in addition to mRNA expression (Semenza et al., 1994). Indeed, the present findings of stable HIF-1α gene expression with anoxia exposure at 21 °C and 5 °C contrasted with the response of nuclear fractions of HIF-1α protein during anoxia exposure (Biggar et al., 2011; Milton et al., 2010). In this regard, quantification of the protein levels of HIF, its associated coactivating protein CREB-binding protein (CBP) and it paralog p300 (Arany et al., 1996), as well as that of HIF inhibitors such as hypoxia-inducible factor-1α inhibitor (HIF1AN, or FIH-1)(Lando et al., 2002; Mahon et al., 2001), HACE1 (HECT domain and ankyrin repeat-containing E3 ubiquitin-protein ligase) (Turgu et al., 2021) and inhibitory PAS (Per/Arnt/Sim) domain protein IPAS (Makino et al., 2001) is required to draw conclusions regarding HIF protein function in T. scripta tissues during long-term anoxia exposure and cold acclimation in normoxia. Moreover, future investigation assessing HIF binding to various transcriptional start sites is required to functionally appraise how long-term anoxia and cold acclimation in normoxia affect HIF activity.

Highlights.

• HIF-1α and HIF-2α gene expression was not upregulated during anoxia exposure.

• HIF-related gene expression was also not broadly upregulated with anoxia.

• Increased vegfa expression with anoxia at 21°C suggests angiogenesis

• Decreased hk1, pygl and pfkp expression with anoxia 5 °C reflects hypometabolism.

• Cold acclimation decreased gene expression in turtle brain, but not heart.