Hypoxia inducible factor-1α knockout does not impair acute thermal tolerance or heat hardening in zebrafish

Abstract

The rapid increase in critical thermal maximum (CT_max) in fish (or other animals) previously exposed to critically high temperature is termed ‘heat hardening’, which likely represents a key strategy to cope with increasingly extreme environments. The physiological mechanisms that determine acute thermal tolerance, and the underlying pathways facilitating heat hardening, remain debated. It has been posited, however, that exposure to high temperature is associated with tissue hypoxia and may be associated with the increased expression of hypoxia-inducible factor-1 (Hif-1). We studied acute thermal tolerance in zebrafish (Danio rerio) lacking functional Hif-1α paralogs (Hif-1aa and Hif-1ab double knockout; Hif-1α^−/−), which are known to exhibit markedly reduced hypoxia tolerance. We hypothesized that Hif-1α^−/− zebrafish would suffer reduced acute thermal tolerance relative to wild type and that the heat hardening ability would be lost. However, on the contrary, we observed that Hif-1α^−/− and wild-type fish did not differ in CT_max, and both genotypes exhibited heat hardening of a similar degree when CT_max was re-tested 48 h later. Despite exhibiting impaired hypoxia tolerance, Hif-1α^−/− zebrafish display unaltered thermal tolerance, suggesting that these traits are not necessarily functionally associated. Hif-1α is accordingly not required for short-term acclimation in the form of heat hardening.

1. Introduction

As global temperatures, as well as the incidence of aquatic hypoxia [1,2], are on the rise, the tolerance of fishes to these stressors is expected to dictate future species distribution [3–5]. The physiological responses to increased temperature and hypoxia are interconnected from the cellular to the whole-organismal level [6–9] and it has been suggested, albeit not without controversy [10–12], that oxygen supply capacity limits thermal tolerance (OCLTT) in fishes [13–16]. An underlying principle is that, at high temperature, a mismatch of oxygen supply and demand results in tissue hypoxia [12,14]. As a corollary, the OCLTT model predicts that hypoxia and thermal tolerance are correlated [17], which could allow natural selection to favour phenotypes with enhanced tolerance to both stressors [18]. The molecular links between thermal and hypoxia tolerance remain under acute focus.

Hypoxia-inducible factor 1 (Hif-1), which controls the expression of genes involved in metabolism and oxygen supply, plays a renowned role in orchestrating the cellular response to hypoxia [19–22] and also may be important during exposure to high temperature, as suggested in the OCLTT model [13,14,23]. Increased Hif-1α (the hypoxia-sensitive Hif-1 subunit) expression has been reported following exposure to high temperature in diverse species, from oyster [24] to mammals [25,26]. Results obtained from fishes have been equivocal. In an Antarctic species, Nothenia coriiceps, Beers and Sidell [27] reported increased Hif-1α mRNA expression in the heart following acute heating to critical temperatures, but this was not replicated in a similar study on the same species [28]. In viviparous eelpout, Zoarces viviparus, Hif-1 DNA binding activity was increased following exposure to moderate but not critically high temperature [29].

In the roundworm, Caenorhabditis elegans, 18 h pre-exposure to moderately high temperature increased survival during acute exposure to critical temperature, but this improvement was absent in Hif-1 loss-of-function mutants [30]. No analogous studies have been performed in fish, but are critical to establish a mechanistic link between Hif-1a expression and thermal tolerance. Recently, it was demonstrated that Hif-1α^−/− zebrafish (Danio rerio) exhibit vastly reduced whole-animal hypoxia tolerance based on a reduced time to loss of equilibrium during acute hypoxia [31]. If oxygen tolerance underlies thermal tolerance [14,17], we hypothesized that the hypoxia-sensitive Hif-1α^−/− line would exhibit reduced thermal tolerance.

As a form of short-term acclimation, ‘heat hardening' describes the increase in critical thermal maximum (CT_max) in individuals exposed consecutively to critical temperature two or more times, which is thought to allow rapid response to environmental extremes [32–34]. As part of the OCLTT model, it was suggested that Hif-1α is involved in thermal acclimation [13,14,23], and as such we hypothesized that the heat hardening response would be absent or attenuated in Hif-1α^−/− zebrafish.

2. Material and methods

Zebrafish are important model organisms for understanding the thermal physiology of fish [34–36]. We studied wild-type and Hif-1A^−/−B^−/− (Hif-1a double knockout; Hif-1α^−/−) zebrafish (D. rerio), which were generated previously [37]. The fish were maintained at 27.5–28°C, which is within the normal thermal range in their natural environment (24.5–34.9°C [35]) in 3 or 10 l tanks in a recirculating system, where they were fed a commercial diet daily. The fish used in the present study were 15 months of age and of either sex. All procedures for animal use and experimentation were carried out in compliance with the University of Ottawa Animal Care and Veterinary Service guidelines (Protocol BL-226) and followed the recommendations for animal use provided by the Canadian Council for Animal Care.

Critical thermal maximum (CT_max) was determined using a protocol similar to that previously described in zebrafish [34]. A 10 l polycarbonate tank was filled with 3.9 l dechloraminated tap water and partitioned with a mesh divider. In one half, a submersible pump flushed water through a heating coil (125 W) into the other half of the tank, where 6–7 fish were housed. The water was aerated and replaced for each group of fish. Temperature was measured, in the half of the tank holding the fish, to the nearest 0.1°C, and was homogeneous throughout this region. During the CT_max trial, temperature was increased at 0.3°C min⁻¹, the standard rate recommended for zebrafish [34,35,38,39] and other freshwater fishes [40–43].

One day (24–30 h) before the first CT_max trial (CT_max1), fish were lightly anaesthetized (120 mg l⁻¹ buffered tricaine methanesulfonate, less than 1 min to induce anaesthesia; normal behaviour resumed within less than 1 min after being returned to the tank [44]) to allow them to be photographed. After each experimental trial (described below), the fish were again photographed and each individual could be identified based on unique patterns on the anal and/or caudal fin. Three Hif-1α^−/− and three (or in one group, four) wild-type fish were thereafter housed together in a 3 l holding tank at 27.5–28°C. During holding periods, the fish were fed commercial food daily, but were fasted for 24 h prior to each CT_max or hypoxia tolerance trial. The next day, CT_max1 was determined in mixed-genotype trials (i.e. each group of 6–7 fish housed together). Fish were allowed 45 min to acclimate to the experimental tank before heating started. CT_max was determined for each individual as the temperature of loss of equilibrium for at least 3 s. At the end of the trial, the fish were individually returned to aerated water (at 27.5–28°C) and soon after re-anaesthetized, photographed and returned to their group holding tank. No mortality was observed after the CT_max1.

Fish were allowed to recover for 48 h before the CT_max determination was repeated (CT_max2). In two other freshwater species, heat hardening was demonstrated to subside after 24 h [33], although in zebrafish, Morgan et al. [34] demonstrated an increase in CT_max in trials one week apart, thus we adopted the 48 h interval as a compromise. Fish from one mixed-genotype group were euthanized (immersion in ice-cold water and pithing) after CT_max2 (and body mass was recorded after photographing to confirm individual identity). The other two groups were returned to holding tanks to determine hypoxia tolerance 5 days later, thus allowing the correlation between thermal tolerance and hypoxia tolerance within individuals to be evaluated. One wild-type fish died overnight following CT_max2, so hypoxia tolerance was not measured in this individual, but CT_max measurements were retained.

Hypoxia tolerance was measured in 2 l closed polycarbonate tanks. Each mixed-genotype group (3 wild-type and 3 Hif-1α^−/− individuals) was tested separately. The fish were allowed 1 h to acclimate to the tank before hypoxia was initiated. Hypoxia was induced by flushing the tank with 28–29°C water pre-equilibrated with 98% N₂ and 2% air in a plexiglass column. Oxygen partial pressure (PO₂) was recorded at 1 Hz with a Firesting optical sensor (PyroScience GmbH, Aachen, Germany) connected to a personal computer. Water PO₂ was rapidly reduced from 153 mm Hg (normoxia) to less than 14 mmHg after approximately 15 min and less than 12 mmHg after 20 min, thereafter plateauing between 12 and 10.5 mmHg, the target range for loss of equilibrium [31]. The fish were removed when they lost equilibrium for at least 3 s and the time from the start of hypoxia induction was recorded. At the end of the protocol, the fish were killed and body mass was recorded.

Statistical analysis was performed in GraphPad Prism (v. 8.4.0). The effect of genotype (wild type versus HIF-1a^−/−) and trial (CT_max1 versus CT_max2) on CT_max was assessed with a two-way repeated-measures analysis of variance. To assess individual repeatability, a linear regression was performed to compare individual scores in CT_max1 and CT_max2. An unpaired Student's t-test was used to compare the time to loss of equilibrium in the hypoxia test between wild-type and Hif-1α^−/− fish. An unpaired t-test was used to compare body mass between wild-type and Hif-1α^−/− fish. Linear regression was performed to assess the relationship between body mass and thermal or hypoxia tolerance for each genotype. Linear regression was used to investigate the relationship between hypoxia tolerance and thermal tolerance (CT_max2). Statistical significance was assigned when p < 0.05. All data are presented as mean ± standard error of the mean (s.e.m.).

3. Results and discussion

There was no effect (F_1,17 = 1.10, p = 0.31) of genotype on CT_max (figure 1a; CT_max1 in wild type, 40.9 ± 0.1°C; in Hif-1α^−/−, 41.1 ± 0.04°C). Thus, we must reject our first hypothesis that Hif-1α^−/− fish would suffer decreased upper thermal tolerance. Although in certain circumstances [27], Hif-1α expression may increase at critical temperatures in some species, it remains to be investigated if Hif-1α expression is increased at high temperature in zebrafish specifically. Presuming that Hif-1α expression is increased, our functional data suggest it does not act to extend the thermal tolerance range. Conversely, if it were not induced during acute heating, it would further consolidate our conclusion that Hif-1α does not play an important role at the upper thermal limit in fishes.

Figure 1.

Critical thermal maximum (CT_max) is similar in wild-type and HIF-1α^−/− zebrafish. CT_max2 was conducted 48 h after CT_max1 under identical conditions. (a) Two-way ANOVA revealed a significant effect of CT_max trial (p < 0.0001) but no effect of genotype (p = 0.31). (b) There was a significant correlation between individual values for CT_max1 and CT_max2, confirming the trait is repeatable. Given that there was no difference in CT_max between genotypes, the linear regression was performed on pooled data from all fish. The dotted line represents that line of identity, which the majority of data points fall above, indicating the increase in CT_max. N = 10 for wild type, N = 9 for HIF-1α^−/− (note that 3 HIF-1α^−/− individuals shared identical CT_max with other fish in both trials and are therefore hidden behind other data points on figures). Data are individual values or means ± s.e.m.

CT_max increased in the second trial (CT_max1 versus CT_max2; F_1,17 = 29.29, p < 0.0001) by a similar magnitude in both genotypes (figure 1a; wild type, +0.4°C; Hif-1α^−/−, +0.3°C; trial × genotype interaction: F_1,17 = 0.09, p = 0.77). In our study, baseline CT_max1 and the increase in CT_max2 in both genotype groups were virtually identical to that independently reported in the same species [34]. Accordingly, we must reject our hypothesis that HIF-1a is required for heat hardening. This contrasts with the finding in C. elegans that loss of HIF-1a eliminates the rapid acclimation of thermal tolerance [30]. The molecular basis of the Hif-1α-independent pathway that underscores heat hardening remains an interesting avenue for further exploration, which could benefit from similar experiments using genetically manipulated zebrafish. Heat shock proteins, which are also induced at high temperatures in fish [45,46], remain likely candidates. It is also plausible that Hif-2, the expression of which is also increased by environmental hypoxia in fish [47,48], could be involved.

The two-way ANOVA revealed a strong effect of individual (F_17,17 = 3.88, p < 0.005), which was further explored with a linear regression to compare CT_max in the two trials (figure 1b). Individual score in CT_max1 was closely correlated with that in CT_max2 (r² = 0.40, p < 0.005), confirming that CT_max is highly repeatable [34] and therefore a valid reflection of an individual's thermal tolerance. Notably, the Hif-1α^−/− individual with the standout decrease in CT_max (41.1 to 40.8°C) suffered minor caudal fin damage between trials. While the decrease in thermal tolerance may have been owing to the direct impairment this injury may have had on performance, it could also be indicative of social stress. Biting from dominant individuals is typical during the establishment of social hierarchies in zebrafish [49], and in rainbow trout (Oncorhynchus mykiss), subordinate individuals exhibit a significantly lower CT_max than dominant fish [42].

In a subset of individuals, we replicated the recent result of Mandic et al. [31] in demonstrating that Hif-1α knockout reduces hypoxia tolerance (figure 2). Hif-1α^−/− fish rapidly succumbed once PO₂ fell below 15 mmHg (figure 2a) and the time to loss of equilibrium was significantly (t₁₀ = 5.48, p < 0.001) shorter than in wild types (figure 2b). Accordingly, there was no correlation (r² = 0.005, p = 0.82) between hypoxia tolerance and thermal tolerance among individuals (electronic supplementary material, figure S1). This discordance suggests thermal and hypoxia tolerance are not necessarily mechanistically linked, even though previous studies in other fish species have suggested that they may be correlated [17,18]. Our data, nevertheless, supplement the growing body of evidence [12] that impaired oxygen transport (e.g. experimental anaemia) has little or no effect on CT_max in fish [50,51], contradicting the OCLTT model [14].

Figure 2.

Hypoxia tolerance is reduced in HIF-1α^−/− relative to wild-type zebrafish. (a) Loss of equilibrium of each individual during the decrease in partial oxygen pressure (PO₂). (b) Time to loss of equilibrium was significantly (unpaired t-test, p < 0.001) lower in HIF-1α^−/− relative to wild type (indicated by asterisk). N = 6 for both genotypes. Data are individual values or mean ± s.e.m.

Despite being of the same age and having been maintained under similar housing conditions, Hif-1α^−/− fish were significantly (t₁₆ = 2.99, p < 0.01) smaller than wild type (wild-type body mass, 0.44 ± 0.05 g; in HIF-1a^−/−, 0.29 ± 0.02 g). The reason for this difference in body mass is unknown, but it was recently shown that conditional Hif-1α knockout also reduces body mass in mice [52], suggesting that Hif-1α may play a conserved role in determining body mass across vertebrates. Nevertheless, within each genotype there were no significant (p > 0.35) correlations between body mass and hypoxia or thermal tolerance, and we regard it as highly unlikely that the differences and similarities we report between the groups were attributable to the difference in body mass.

In summary, despite substantially compromising hypoxia tolerance, the loss of Hif-1α did not affect acute thermal tolerance (CT_max), nor did it impair the rapid increase in CT_max in the second of two consecutive trials 48 h apart (heat hardening) in zebrafish, in contrast with predictions based on the OCLTT model [14,17,18]. Acute thermal tolerance, as measured in our study, is relevant to the predicted impacts of extreme weather conditions, such as those expected to worsen with climate change [39]. Our approach of using a genetic knockout model reveals a dissociation between hypoxia and thermal tolerance, suggesting that tolerance to these stressors may not necessarily be expected to co-evolve [18], despite their frequent co-occurrence.

Data accessibility

Raw data have been uploaded to Dryad: doi:10.5061/dryad.j6q573nb0.

Authors' contributions

W.J. and S.F.P. designed the experiments; W.J. performed the experiments and conducted the statistical analyses, W.J. wrote the initial draft of the manuscript, which received input from S.F.P. W.J. and S.F.P are both accountable for the content and approved the final version of the manuscript.

Competing interests

We declare we have no competing interests.

Funding

This study was supported by Natural Sciences and Engineering Research Council of Canada Discovery with grant no. 286723 to S.F.P.