Aquatic surface respiration improves survival during hypoxia in zebrafish (Danio rerio) lacking hypoxia-inducible factor 1-α

Abstract

Hypoxia-inducible factor 1-α (Hif-1α), an important transcription factor regulating cellular responses to reductions in O₂, previously was shown to improve hypoxia tolerance in zebrafish (Danio rerio). Here, we examined the contribution of Hif-1α to hypoxic survival, focusing on the benefit of aquatic surface respiration (ASR). Wild-type and Hif-1α knockout lines of adult zebrafish were exposed to two levels (moderate or severe) of intermittent hypoxia. Survival was significantly compromised in Hif-1α knockout zebrafish prevented from accessing the surface during severe (16 mmHg) but not moderate (23 mmHg) hypoxia. When allowed access to the surface in severe hypoxia, survival times did not differ between wild-type and Hif-1α knockouts. Performing ASR mitigated the negative effects of the loss of Hif-1α with the knockouts initiating ASR at a higher PO₂ threshold and performing ASR for longer than wild-types. The loss of Hif-1α had little impact on survival in fish between 1 and 5 days post-fertilization, but as the larvae aged, their reliance on Hif-1α increased. Similar to adult fish, ASR compensated for the loss of Hif-1α on survival. Together, these results demonstrate that age, hypoxia severity and, in particular, the ability to perform ASR significantly modulate the impact of Hif-1α on survival in hypoxic zebrafish.

1. Introduction

Low environmental O₂ (hypoxia) is common and widespread in aquatic ecosystems [1,2]. Not surprisingly, numerous fish species are capable of tolerating and surviving periodic hypoxia. One such species is the zebrafish (Danio rerio), a freshwater cyprinid native to flood plains of the Indian subcontinent where it inhabits slow-moving streams or standing waters of rice fields, which can become hypoxic diurnally [3,4]. Zebrafish are most vulnerable to the effects of hypoxia as larvae. Critical O₂ tension (P_crit; the partial pressure of O₂ (PO₂) at which O₂ consumption can no longer be maintained) is highest in larvae, indicating lower hypoxia tolerance [5,6], and declines over development and into adulthood [6,7]. Adult zebrafish can survive for at least 2 days in water PO₂ of 15 mmHg [8], a severe hypoxic exposure that is well below their P_crit of 20 mmHg [6]. Moreover, survival increases with previous acclimation to hypoxia [8].

As in other species, the hypoxia tolerance of zebrafish is attributed to the integration of traits across several levels of biological organization. Depending on the degree and duration of hypoxia, zebrafish increase ventilation frequency (f_V) [9], cardiac output and vascularization [10], myoglobin and neuroglobin transcript and protein abundances [11], protein expression of glycolytic enzymes [12], mitochondrial function [13] and haemoglobin–O₂ binding affinity [14]. Together, these responses enhance aerobic capacity by increasing O₂ uptake, transport and delivery. Many of these hypoxia-triggered responses are thought to be regulated at the transcriptional level by hypoxia-inducible factor 1-α (Hif-1α), a well-known global regulator of the transcriptional response to hypoxia in vertebrates [15–17]. As cellular O₂ falls, Hif-1α accumulates in the cell, moves to the nucleus and dimerizes with Hif-1β. The Hif heterodimer, along with its co-activators p300 and CREB (cAMP-response-element-binding protein), binds to hypoxia responsive elements of target genes, altering their expression patterns during hypoxia [18,19].

In mammals, knockout of HIF-1α results in embryonic lethality [20], whereas in zebrafish, complete knockout of both paralogues of Hif-1α (Hif-1α^−/−: signifies knockout of both Hif-1aa and Hif-1ab paralogues) produces a viable mutant [21]. However, the loss of Hif-1α causes a significant decrease in the time to loss of equilibrium in adult zebrafish exposed to severe hypoxia [22,23] and an increase in P_crit in larvae [23]. Although these studies provide evidence of an integral role of Hif-1α in dictating hypoxia tolerance, the degree to which compromised tolerance ultimately affects survival, particularly when behavioural responses to hypoxia are taken into consideration, remains unknown.

Behavioural responses provide an effective strategy for combating the effects of hypoxia. As water becomes hypoxic, some fishes move to the surface to ventilate at the water–air interface, where water O₂ tension may be higher than bulk water PO₂ [24]. This behaviour, termed aquatic surface respiration (ASR), is widespread in fishes of freshwater and marine habitats where hypoxia occurs [25–28]. The use of ASR maximizes O₂ extraction and increases arterial blood O₂ content [24], thereby reducing the impact of environmental hypoxia and significantly increasing the likelihood of survival [25,27]. Adult zebrafish exposed to hypoxia perform ASR, and this behaviour occurs in larvae as young as 5 days post-fertilization (dpf) [29]. In the studies that demonstrated severely compromised hypoxia tolerance in zebrafish with knockout of Hif-1α [22,23], fish were not allowed access to the surface to perform ASR. Thus, it remains unknown whether fish, given the opportunity to perform ASR, can use this behavioural response to mitigate the effects of loss of Hif-1α in decreasing hypoxia tolerance.

In the current study, we tested the hypothesis that the loss of Hif-1α affects the survival of zebrafish during intermittent hypoxia and asked whether survival was improved by the use of ASR. We assessed survival in wild-type (WT) and Hif-1α^−/− adult zebrafish exposed to intermittent hypoxia for 5 days, with individuals either allowed to access or prevented from accessing the water surface. Intermittent hypoxia was chosen to mimic the natural habitat of zebrafish (see above). Further experiments quantified the time spent performing ASR and the PO₂ threshold at which ASR was initiated. Because the putative peripheral O₂ chemoreceptors in fish, the neuroepithelial cells (NECs; reviewed by the authors in [30,31]), have been linked to the initiation of ASR [29], the number of NECs was quantified in gills of adult fish. Similar experiments were conducted in larvae to assess the impact of Hif-1α^−/− on the less tolerant life-history stages. We predicted greater mortality in larval and adult Hif-1α^−/− during exposure to hypoxia and that the decrease in survival caused by the loss of Hif-1α would be alleviated by the use of ASR.

2. Material and methods

(a) . Experimental animals

Adult zebrafish were housed in 10 l acrylic tanks in recirculating aquatic housing systems (Aquatic Habitats, Apopka, USA) at the University of Ottawa aquatic care facility. Fish were maintained at 28°C under a 14 h : 10 h light : dark cycle in aerated (PO₂ = 153 mmHg) dechloraminated city of Ottawa tap water (‘system’ water) and were fed twice a day to satiation with GEMMA 300 fish feed (Skretting USA). All experiments were conducted on WT zebrafish from in-house stocks or on an established Hif-1α^−/− (double knockout of Hif-1aa and Hif-1ab paralogues) mutant line originally generated via CRISPR/Cas9 gene editing at the Max Planck Institute for Heart and Lung Research [21]. Embryos were obtained from controlled breeding events. Briefly, one male and two females were placed overnight in a 2 l breeding tank and separated by a transparent divider. The following morning, water was replaced, and the dividers were removed for breeding. Embryos were collected after 30 min and were pooled across different breeding tanks to generate randomized samples of either genotype. Embryos were reared in an incubator maintained at 28°C in 50 ml Petri dishes at a density of 40 per dish, in system water containing 0.05% methylene blue, an antifungal agent. The water in the Petri dishes was replaced daily. At 5 dpf, larvae were transferred to static 2 l tanks and fed twice daily to satiation with GEMMA Micro 75 fish feed (Skretting USA). The water in the static tanks was replaced every second day. Larvae were raised in the static tanks until 16 dpf. Fish were not fed 24 h prior to experimentation.

(b) . Survival rate during intermittent hypoxia

Adult, male zebrafish were transferred from holding conditions (described above) to experimental chambers and exposed to intermittent hypoxia, with water PO₂ set to 23 mmHg or 16 mmHg between 8.00 and 12.00, for 5 consecutive days. We have previously shown that sex does not contribute to the significant difference in hypoxia tolerance between WT and Hif-1α mutants [23]; however, given the possibility of sex-related differences in survival within a genotype, only male fish were used, to avoid potential confounding effects of sex. Adult fish of each genotype were allocated randomly into one of two treatments; they were either allowed to access or prevented from accessing the surface for the duration of the experiment (three trials with three individuals per trial at PO₂ of 23 mmHg and nine trials with three individuals per trial at PO₂ of 16 mmHg). Fish without access to the surface were held in 300 ml cylindrical chambers submerged in a 30 l tank supplied with flowing, aerated system water. The tops and bottoms of the chambers were composed of fine mesh to allow for mixing of water via a submersible pump within the holding tank. Fish with access to the surface were placed in similar cylindrical chambers with mesh bottoms within a 30 l tank. In this case, the chambers were partially submerged to allow fish access to the surface. Following a 24 h recovery period from handling, the 30 l tanks were supplied with hypoxic water from a water equilibration column that was gassed with appropriate mixtures of air and N₂ to yield a PO₂ level of 23 or 16 mmHg. The 4 h exposure period began once water PO₂ in the tank stabilized at 23 or 16 mmHg, approximately 30 min after the gas mixture was applied to the water equilibration column. A custom gas mixer fabricated at the University of Ottawa provided the gas mixtures, and O₂ levels were monitored with a hand-held O₂ meter (Handy Polariz O₂, OxyGuard, Birkerod, Denmark). After the 4 h hypoxic period, the water equilibration column was gassed with air, and PO₂ quickly rose to normoxic levels (PO₂ = 153 mmHg) and remained there until the subsequent bout of hypoxia on the following day. Adult fish mortality was assessed after every bout of hypoxia.

Larvae were transferred from holding conditions to experimental chambers and exposed to intermittent hypoxia (16 mmHg) for 4 h d⁻¹ for 5 consecutive days beginning at 1, 5 or 12 dpf. Three treatment groups were used; the first group was kept in normoxia for the entire 5 day experiment, the second group was exposed to intermittent hypoxia and prevented from accessing to the surface, and the third group was exposed to intermittent hypoxia and allowed access the surface. Sample size was six trials with 25 larvae per trial for each treatment. Because larval zebrafish begin to perform ASR at 5 dpf [29], the 1–5 dpf age group was not prevented from accessing the surface during hypoxia. The hypoxia protocol was identical to that of the adults, and mortality was assessed after each hypoxic exposure or at the equivalent time for fish that remained in normoxia. Larvae were held in similar cylindrical chambers to the adult fish.

(c) . Aquatic surface respiration

For adult fish, ASR was measured in conjunction with the survival rate for hypoxia trials in which fish had access to the surface. Fish were video recorded using an iPhone for 30 min during the daily hypoxic exposure, at the beginning of the third hour of hypoxia. The time that each fish spent performing ASR was scored from the videos by an observer who used event-logging software (BORIS, https://www.boris.unito.it/) and expressed as a percentage of the 30 min observation period. The observer was unaware of the genotype of the fish being scored. The sample size was 18 for WT and 15 for Hif-1α^−/− individuals.

The water PO₂ threshold at which ASR was initiated in adult fish was measured in a separate experiment. WT and Hif-1α^−/− fish were placed individually into adjacent experimental chambers (2.2 l) supplied with flowing, aerated system water for an overnight recovery period (sample size was 10 adult individuals per genotype). The following morning, fish were video recorded (Canon Vixia HF R400) during exposure to graded hypoxia. Water PO₂ (measured using a FireStingO₂, PyroScience, Aachen, Germany) was lowered from 153 mmHg (normoxia) to 10 mmHg over 1 h by gassing the water equilibration column supplying the tanks with N₂ using a precision mass flow controller (Sierra C100 L Smarttrak, SRB Controls, Markham, Ontario, Canada). Water PO₂ at the first occurrence of ASR was determined from the videos by an observer who was unaware of fish genotype (for similar protocol, see [32]).

Time spent performing ASR during hypoxia also was evaluated for 5, 9, 12 and 16 dpf larvae. Experimental chambers were made from 50 ml cell culture flasks (Fisherbrand, Thermo Fisher Scientific, Ottawa, Ontario, Canada) by replacing the small end of the flask (opposite the cap) with mesh. Water PO₂ was adjusted to 16 mmHg prior to the trial by gassing the equilibration column supplying the tank holding the chambers with N₂ as described above, and water PO₂ in the tank was checked immediately prior to and following each trial (Handy Polariz O₂). Larvae were video recorded using an iPhone for 5 min during the second hour of hypoxic exposure. Time spent performing ASR was scored from the videos (by an observer who used BORIS) and expressed as a percentage of the observation period. A larva was considered to be performing ASR when it adopted a horizontal position at the surface of the water. The observer did not know the genotype of the fish being scored.

(d) . Quantification of branchial neuroepithelial cell abundance

Adult zebrafish were euthanized with an overdose of MS-222. Both opercula were removed, and the head of the zebrafish was fixed overnight at 4°C in 4% paraformaldehyde prepared in phosphate-buffered saline (PBS). The following day, the fixed head was rinsed twice in PBS containing 0.1% tween-20 (PBST). The first gill arch from the left side of the fish was dissected out and rinsed three times in PBST. Gill arches were placed in a blocking solution containing 0.8% Triton-X 100, 3% bovine serum albumin and 2% normal goat serum dissolved in PBST for 1 h at room temperature. After blocking, gills were incubated overnight at 4°C with rabbit anti 5-HT (Sigma Aldrich, cat no. S5545) primary antibody at 1 : 500 dilution, followed by a 1 h incubation with 1 : 500 dilution of goat anti-rabbit Alexa Fluor 568 (Abcam, cat no. ab175471) secondary antibody at room temperature, with five rinses in PBST after each antibody incubation. As NECs contain the neurotransmitter serotonin (5-HT), the 5-HT antibody is a standard marker for quantifying NECs in fish (e.g. [30]). Gill arches were mounted on depression slides and imaged with a Nikon A1R MP confocal using a 10×/0.30w lens.

For each gill arch imaged, the lengths of two to three filaments were measured, and NECs on the filament were manually counted for the first 30% of the filament lengthwise beginning from the distal tip. The density of NECs was determined as the number of NECs per mm of filament length and averaged across the number of filaments analysed for a specific gill. All density measurements were conducted without knowledge of the genotype of the gill arch in the image analysed.

(e) . Statistical analysis

All statistical analyses were performed in R (https://www.R-project.org/).

In adult zebrafish, the effects of genotype and day of hypoxic exposure on time spent performing ASR were tested using two-way repeated-measures analysis of variance (RM ANOVA) in the car package [33]. The data were tested for sphericity using Mauchly's test. Effects of genotype on ASR threshold in adults were tested using Student's t-test, and effects of genotype on per cent time spent performing ASR in larvae were tested using Welch's t-tests (unequal sample size) or, if data violated assumptions, Mann–Whitney non-parametric tests. All remaining datasets were analysed using two-way ANOVA and Tukey's post hoc test was used if a significant interaction was detected. All data were tested for normality (Shapiro–Wilk test) and equal variance (Levene's test). If normality or equal variance assumptions were violated, data were transformed. If the data could not be transformed to meet the assumptions, then aligned rank transform for non-parametric factorial ANOVAs was performed using ARTool package [34], followed by post hoc pairwise comparisons [35]. Significance was set at p < 0.05.

3. Results

(a) . Survival rate of adult zebrafish during intermittent hypoxia

There was no significant effect of intermittent hypoxia at PO₂ of 23 mmHg on survival in adults (figure 1a–c). The more severe hypoxic exposure (16 mmHg PO₂) lowered survival rates for both genotypes, in particular for individuals without access to the surface (figure 1d–f). Mortality occurred exclusively after the first exposure, with no further mortality during or after subsequent bouts of hypoxia. Survival of WTs did not differ significantly with or without access to the surface (figure 1f). Survival of Hif-1α^−/− mutants allowed access to the surface was 88% and did not differ significantly from that of the WTs. However, survival decreased to 37% in Hif-1α^−/− fish prevented from accessing the surface, which was significantly lower than in WTs and knockouts allowed access to the surface (figure 1f).

Figure 1.

Percentage survival of wild-type and Hif-1α^−/− adult zebrafish during exposure to intermittent hypoxia (4 h d⁻¹ for 5 days). (a,b) Survival was assessed in zebrafish exposed to 23 mmHg hypoxia without access to the surface (a) or with access to the surface (b). (c) Comparisons between genotypes were made on day 5 of the exposure to 23 mmHg. (d,e) In a separate trial, survival was determined in zebrafish exposed to 16 mmHg hypoxia without access to the surface (d) or with access to the surface (e). (f) Comparisons between genotypes were made on day 5. There was no effect of genotype or surface access in individuals exposed to 23 mmHg hypoxia (two-way aligned rank ANOVA; genotype: F = 1.3, p = 0.29; surface access: F = 1.3, p = 0.29; genotype × surface access: F = 1.3, p = 0.29; n = 3 trials with three individuals per trial). There was a significant interaction of genotype and surface access in individuals exposed to 16 mmHg hypoxia (two-way aligned rank ANOVA; genotype: F = 18.5, p < 0.01; surface access: F = 16.8, p < 0.01; genotype × surface access: F = 9.8, p < 0.01; n = 9 trials with three individuals per trial). Different upper case letters indicate significant (p < 0.05) differences between genotypes without access to the surface, different lower case letters indicate significant differences between genotypes allowed access to the surface and an asterisk indicates a significant difference between treatments within a genotype. Wild-type and Hif-1α^−/− data are superimposed in instances where 100% survival occurred in both genotypes (e.g. b). Data are presented as means ± s.e.m.

(b) . Aquatic surface respiration in adult zebrafish during intermittent hypoxia

For a given genotype, time spent performing ASR at 16 mmHg PO₂ was similar on days 1 and 5 of the intermittent hypoxia exposure (figure 2a). However, the genotypes differed significantly in the duration of ASR, with WTs performing ASR for approximately 28% of the observation period versus approximately 80% for Hif-1α^−/− zebrafish (figure 2a). The water PO₂ at which ASR was initiated also differed significantly between genotypes; the PO₂ threshold for ASR in WT zebrafish was 28.1 ± 3.5 mmHg, which was significantly lower than the 45.3 ± 5.8 mmHg threshold for Hif-1α^−/− zebrafish (figure 2b). The abundance of NECs in the gills did not differ between WT and Hif-1α^−/− fish (electronic supplementary material, figure S1; WT: 117 ± 13 mm⁻¹; Hif-1α^−/−: 148 ± 18 mm⁻¹; Welch's two-sample t-test: t = −1.41, p = 0.19).

Figure 2.

ASR in wild-type and Hif-1α^−/− adult zebrafish. (a) Percentage time spent performing ASR was assessed on days 1 and 5 of intermittent exposure to 16 mmHg hypoxia. (b) In separate trials, the PO₂ at which ASR was initiated was assesed in individuals of both genotypes. There was a significant effect of genotype but not day of exposure on percentage time spent performing ASR (two-way RM ANOVA; genotype: F = 54.9, p < 0.01; day: F = 0.02, p = 0.9; genotype × day: F = 0.08, p = 0.8, n = 18 for wild-type and 15 for Hif-1α^−/− individuals). There was a significant effect of genotype on the PO₂ threshold for ASR (Student's t-test; t = −2.5, p = 0.02; n = 10). An asterisk indicates a significant difference between genotypes. Data are presented as means ± s.e.m.

(c) . Survival rate in larvae during normoxia and intermittent hypoxia

Larvae were exposed to normoxia or intermittent hypoxia at PO₂ of 16 mmHg from 1 to 5, 5 to 9 or 12 to 16 dpf, and detailed survival data are presented in electronic supplementary material, figure S2. The effects of genotype and hypoxic exposure on growth and hatch rate were also assessed (electronic supplementary material, figure S3, tables S1 and S2). There was little mortality in fish raised in normoxia at any stage of development (figure 3), with the exception of Hif-1α^−/− larvae at 1–5 dpf, where survival was lower (figure 3a). With intermittent exposure to hypoxia, survival rate did not differ between WT and Hif-1α^−/− larvae at 5 dpf (figure 3a). However, in 9 dpf larvae, there was a significant interaction of genotype and hypoxic exposure on survival (figure 3b). Hypoxia had the strongest impact on larvae that did not have access to the surface; mortality was high (greater than 60%) after the first exposure at 5 dpf with 0% survival by 7 dpf regardless of genotype (electronic supplementary material, figure S2D). In individuals allowed access to the surface during hypoxia, the survival rate was 82% in WTs, which was not significantly different from that of WTs in normoxia (figure 3b). By contrast, survival in Hif-1α^−/− zebrafish under these conditions decreased from 69% following the first exposure at 5 dpf to 49% following the last exposure at 9 dpf (electronic supplementary material, figure S2E), at which point survival was significantly lower than in Hif-1α^−/− larvae maintained in normoxia (figure 3b).

Figure 3.

Comparison of percentage survival between wild-type and Hif-1α^−/− larval zebrafish exposed to normoxia or intermittent hypoxia (16 mmHg, 4 h d⁻¹) with and without access to the surface at (a) 5 days post-fertilization (dpf), (b) 9 dpf and (c) 16 dpf. Data are derived from electronic supplementary material, figure S2. Larvae of either genotype exposed to intermittent hypoxia without access to the surface did not survive past 7 dpf (b) or 14 dpf (c). Larval zebrafish from 1 to 5 dpf were not exposed to hypoxia without access to the surface (ND, no data). There was a significant interaction between genotype and exposure in 5 dpf larvae (two-way aligned rank ANOVA; genotype: F = 14.2, p < 0.01; exposure: F = 3.8, p = 0.07; genotype × exposure: F = 7.1, p = 0.01), in 9 dpf larvae (two-way aligned rank ANOVA; genotype: F = 12.2, p < 0.01; exposure: F = 39.1, p < 0.01; genotype × exposure: F = 4.6, p = 0.01) and in 16 dpf larvae (two-way aligned rank ANOVA; genotype: F = 13.1, p < 0.01; exposure: F = 57.7, p < 0.01; genotype × exposure: F = 5.5, p < 0.01). Sample size was six trials with 25 individuals per trial. Different upper case letters indicate significant differences among exposure treatments in wild-type larvae, different lower case letters indicate differences among exposure treatments in Hif-1α knockout larvae and an asterisk indicates significance between exposure treatments across genotypes. Data are presented as means ± s.e.m.

The interaction of hypoxic exposure and genotype was significant in 16 dpf larvae (figure 3c). The impact of hypoxia was greatest for larvae without access to the surface, where survival was less than 10% in both genotypes following the first hypoxic exposure, with 0% survival by 14 dpf (electronic supplementary material, figure S2G). In larvae that had access to the surface, the survival of WTs was more-or-less constant at approximately 70% (electronic supplementary material, figure S2H). During hypoxia, the survival of WT larvae with access to the surface did not differ significantly from the normoxic group (figure 3c). The survival of Hif-1α^−/− larvae with access to the surface was significantly lower than in the corresponding WT group, with 24% survival at 16 dpf compared to 68% in the WTs (figure 3c).

(d) . Aquatic surface respiration in larvae

There was no effect of genotype on time larvae spent performing ASR during hypoxia at 5, 9 or 12 dpf (figure 4a–c). However, at 16 dpf, Hif-1α^−/− larvae performed ASR for a significantly greater proportion of the observation period than WT larvae (59.3 ± 3.5% versus 30 ± 8.1; figure 4d).

Figure 4.

Percentage time spent performing ASR during exposure to 16 mmHg hypoxia was assessed in (a) 5, (b) 9, (c) 12 and (d) 16 dpf larvae. There was no effect on genotype on per cent time spent performing ASR in 5 dpf larvae (Welch's t-test: t = 0.6, p = 0.6, n = 4 for wild-type and n = 3 for Hif-1α^−/−), 9 dpf larvae (Mann–Whitney non-parametric t-test: W = 21, p = 0.6, n = 4 for wild-type and n = 13 for Hif-1α^−/−) and 12 dpf larvae (Welch's t-test: t = −0.79, p = 0.5, n = 7 for wild-type and n = 5 for Hif-1α^−/−). There was a significant effect of genotype, indicated by the asterisk, in 16 dpf larvae (Mann–Whitney non-parametric t-test: W = 46, p < 0.01, n = 8 for wild-type and n = 6 for Hif-1α^−/−). Data are presented as means ± s.e.m.

4. Discussion

The present study aimed to determine first whether Hif-1α affects survival during exposure to hypoxia in larval and adult zebrafish, and second, whether the behavioural response of ASR mitigated the loss of Hif-1α. The survival rate of adult zebrafish subjected to severe intermittent hypoxia was significantly reduced by the loss of Hif-1α but only when fish were prevented from accessing the surface to perform ASR. When allowed to perform ASR, survival significantly improved in Hif-1α knockouts. For early stage larvae (1–5 dpf), the loss of Hif-1α had little impact on survival. As fish aged, the role of Hif-1α in hypoxic survival increased with the greatest impact of Hif-1α knockout in the older larvae (12–16 dpf). As in adults, ASR appeared to compensate for the loss of Hif-1α on survival. Overall, the results suggest that the importance of Hif-1α for hypoxic survival is dependent on life-history stage and the ability to perform ASR.

(a) . Survival of adult zebrafish in hypoxia

In WT zebrafish, survival rate was high during the more severe exposure to hypoxia (16 mmHg) regardless of access to the surface. These results are consistent with previous findings that most individuals do not lose equilibrium within 5 h of exposure to 16 mmHg hypoxia [23] and can survive for 2 days in 15 mmHg hypoxia [8]. By contrast, the loss of Hif-1α reduced survival during severe hypoxia when access to the surface was prevented. The low survival was consistent with reduced tolerance in Hif-1α knockout adult zebrafish as measured by a significantly shorter time to loss of equilibrium [23]. Interestingly, 37% of individuals lacking Hif-1α were able to survive the hypoxic exposure in the current study, which is roughly similar to the 25% of Hif-1α^−/− individuals that maintained equilibrium during hypoxia for at least twice as long as other individuals of the same genotype [23]. Thus, hypoxia tolerance, even among fish lacking Hif-1α, varies substantially. The mechanisms underlying intraspecific differences in hypoxic survival are largely unknown. Although sex and seasonality have been implicated [8], neither varied in the current study. Because only males were used, future investigation should determine whether similar survival patterns are noted in females. It is worth noting, however, that sex does not contribute to significant differences in hypoxia tolerance between WT and Hif-1α^−/− zebrafish [23].

Mortality in both genotypes occurred exclusively during the first bout of hypoxia. It is likely that individuals with low tolerance did not survive the first bout, while more tolerant individuals were able to survive a single hypoxic event and tolerated repeated exposure to similar levels of hypoxia. These results suggest that the 20 h normoxic period following hypoxic exposure was sufficient for the fish to recover and remain resilient to subsequent, similar bouts of hypoxia.

For Hif-1α^−/− fish allowed access to the surface to perform ASR, survival rate was 88% during severe intermittent hypoxia. Therefore, the behavioural response of ASR reduced the detrimental impact of Hif-1α loss, and little difference in survival rate was found between genotypes. In goldfish (Carassius auratus), ASR increased arterial blood O₂ content and improved O₂ transport [24], probably bolstering oxygen consumption in severe hypoxia. Hif-1α-deficient fish, however, spent more time performing ASR and initiated ASR at a significantly higher PO₂ than WTs. Therefore, although survival during severe intermittent hypoxia was not compromised in Hif-1α knockout adults allowed access to the surface, the substantially greater reliance on ASR suggests that Hif-1α loss compromised the hypoxic response. Increased use of ASR is not without consequence, as there are potential trade-offs such as increased energy use required for continued swimming near the surface of the water and increased risk of predation [36,37].

The mechanisms underlying the control of ASR behaviour remain largely unknown, although previous work revealed a role for peripheral chemoreceptors in triggering ASR [29,36]; in zebrafish, these are presumed to be NECs. The earlier onset and greater use of ASR in Hif-1α knockout fish may signal either quantitative or functional differences between NECs in the two genotypes. NEC abundance was similar between WT and Hif-1α knockout individuals, suggesting that effects of Hif-1α loss on functional properties of NECs should be examined. Currently, it remains unclear how Hif-1α knockout zebrafish adjust their ASR behaviour, and future studies are required to elucidate the underlying pathways.

Exposure to moderate hypoxia (23 mmHg) did not result in mortality in WT or Hif-1α^−/− fish. These results suggest that in moderate environmental hypoxia, non-Hif-1α-related mechanisms are sufficient to compensate for the loss of Hif-1α. It is difficult to speculate on what these mechanisms could be given how little is known in fish about which hypoxia responsive genes are under the regulation of Hif-1α and which are Hif-1α-independent. Alternatively, it is possible that moderate hypoxia in adult fish may not decrease cellular O₂ sufficiently to initiate Hif-1α hypoxic signalling. However, this scenario is less likely because Hif-1α knockout adult fish exposed to water PO₂ of 30 or 90 mmHg exhibit altered hypoxic ventilatory responses [38,39]. In general, owing to challenges with Hif-1α antibodies, few studies in adult fish have measured Hif-1α protein levels during hypoxia, creating difficulties in determining the level and duration of hypoxia needed to activate the Hif-1α pathway [40]. It is unknown whether the moderate level of hypoxia used in the current study was sufficient to activate the Hif-1α pathway because protein levels were not quantified owing to repeated failed attempts to produce reliable, paralog-specific zebrafish Hif-1α polyclonal antibodies.

(b) . Survival of larval zebrafish in hypoxia

A stark difference in survival was apparent between the early and later stages of larval development in WT fish during exposure to severe intermittent hypoxia. Embryonic and early stage larvae (1–5 dpf) experienced little mortality, whereas larvae without access to the surface exhibited 100% mortality within 3 days in the 5–12 dpf exposure group. These observations agree with previous findings that early embryonic life stages of zebrafish are extremely tolerant to hypoxia, with limited mortality during exposure to little or no O₂ [41,42].

The presence of at least one paralogue of Hif-1α protein can be detected during normoxic development and in response to hypoxia as early as 1 dpf in embryonic zebrafish [43,44]. Although Hif-1α cellular responses initiated at 1 dpf improved the regulation and maintenance of O₂ uptake in hypoxia in 4 dpf fish [44], no difference in survival was detected in the current study between WT and Hif-1α^−/− fish from 1 to 5 dpf of hypoxia exposure. These results indicate that embryonic tolerance of hypoxia is not dependent on Hif-1α-related mechanisms. Regardless of genotype, older larvae prevented from performing ASR succumbed to severe hypoxia within 3 days, suggesting that 16 mmHg hypoxia was too severe for survival even with Hif-1α-induced responses.

Access to the surface significantly improved survival in larvae exposed to hypoxia after 5 dpf. As in adults, larvae relied on ASR presumably to take advantage of the better-oxygenated surface layer of water. By 9 dpf, WT larvae spent 30 ± 11% of the observation period performing ASR, similar to the adult WT fish (27 ± 7%). The survival of WT larvae in hypoxia was not significantly different from that of normoxic individuals, indicating the effectiveness of ASR in the hypoxic response. Similarly, survival significantly improved in older (greater than 5dpf) knockout larvae performing ASR during hypoxia. However, unlike in WT larvae, survival of Hif-1α knockouts allowed access to the surface during hypoxia remained significantly lower than that of normoxic individuals. This result suggests that Hif-1α-mediated responses to hypoxia are important for larval survival in hypoxia and that engaging in ASR is not sufficient to compensate for the loss of Hif-1α. Time spent performing ASR increased with age in developing Hif-1α^−/− larvae, reaching 59 ± 4% of the observation period by 16 dpf. However, adult Hif-1α knockout zebrafish engaged in ASR for 80 ± 4% of the observation period, a value well above that of larvae. The lower survival rate of larval zebrafish even with access to the surface suggests a limitation either in the efficiency of ASR or in the time spent performing ASR when compared with that of the adults. It is possible that the energy expenditure required to maintain the fish at the surface is more costly in larvae than in adults, limiting the time larvae are capable of engaging in ASR.

5. Conclusion

Knockout of Hif-1α impacted survival in adults exposed to severe but not moderate hypoxia, indicating that under certain hypoxic conditions, Hif-1α has direct effects on organismal survival. When given the opportunity to perform ASR, survival in severe hypoxia improved in adults lacking Hif-1α and survival of Hif-1α knockout fish did not differ significantly from that of WTs. The Hif-1α knockout fish, however, spent a greater proportion of time performing ASR and initiated ASR at a higher PO₂ than WT fish, indicating a greater reliance on this behaviour during exposure to hypoxia. Although severe hypoxia did not elicit mortality in early larvae (less than 5 dpf) during the exposure period, older larvae exhibited 100% mortality by the third day of intermittent hypoxia. As in the adult fish, engaging in ASR improved survival in severe hypoxia in larvae, although to a greater extent in WTs than in Hif-1α knockouts. Overall, the present study links Hif-1α with hypoxic survival and highlights the complexity of the role Hif-1α plays in the hypoxic response, as influenced by life-history stage, severity of exposure and use of complex behaviours.

Ethics

All procedures for animal use and experimentation were carried out in compliance with the University of Ottawa Animal Care and Veterinary Service guidelines under protocols BL-1700 and BL- 2118 and adhered to the recommendations for animal use in research and teaching provided by the Canadian Council for Animal Care.

Data accessibility

All data in the study are presented in figures as individual data points as well as means with standard errors. The data are provided in electronic supplementary material [45].

Authors' contributions

M.M.: conceptualization, formal analysis, methodology, supervision, writing—original draft, writing—review and editing; K.F.: investigation, writing—review and editing; P.Q.: investigation, writing—review and editing; Y.K.P.: investigation, writing—review and editing; S.F.P.: conceptualization, funding acquisition, writing—review and editing; K.M.G.: conceptualization, funding acquisition, project administration, resources, supervision, writing—review and editing. All authors gave final approval for publication and agreed to be held accountable for the work performed therein.

Competing interests

We declare we have no competing interests.

Funding

This work was supported by Natural Sciences and Engineering Research Council (NSERC) of Canada Discovery grants to K.M.G. (RGPIN-2017–05487) and S.F.P. (RGPIN-2017-05545), and University of Ottawa Undergraduate Research Opportunity and University Research Scholarship to K.F.