The osmorespiratory compromise is a physiological trade-off between the characteristics of the gill that promote respiratory gas exchange and those that limit passive flux of ions and water with the environment. In hypoxia, changes in gill blood flow patterns and functional surface area that increase gas transfer can promote an exacerbation in ion and water flux. Our goal was to determine whether the osmorespiratory compromise is flexible, depending on environmental salinity (fresh, isosmotic and sea water) and oxygen levels (hypoxia) in euryhaline killifish, Fundulus heteroclitus. Plasma ion concentrations were minimally affected by hypoxia, indicating a maintenance of osmoregulatory homeostasis. In freshwater killifish, hypoxia exposure reduced branchial Na+/K+-ATPase and NEM-sensitive ATPase activities, as well as diffusive water flux rates. Unidirectional Na+ influx and Na+ efflux decreased during hypoxia in freshwater, but net Na+ flux remained unchanged. Net loss rates of Cl−, K+ and ammonia were also attenuated in hypoxia, suggesting both transcellular and paracellular reductions in permeability. These reductions appeared to be regulated phenomena as fluxes were restored immediately in normoxia. Na+ flux rates increased during hypoxia in 11 ppt, but decreased in 35 ppt, the latter suggesting a similar response to hypoxia to that in freshwater. In summary, freshwater and seawater killifish experience a reduction in gill permeability, as seen in other hypoxia-tolerant species. Fish acclimated to isosmotic salinity increased Na+ influx and efflux rates, as well as paracellular permeability in hypoxia, responses in accord with the predictions of the classic osmorespiratory compromise.
The teleost fish gill is a multifunctional organ that is the primary site for ion and gas exchange in addition to playing a central role in acid–base balance and nitrogenous waste excretion (Evans, 2005). When needed, branchial oxygen exchange can be improved by increasing water flow across the gills (ventilation) and also changing the rate and pattern of gill blood flow (perfusion) (Perry et al., 2009). A reduction in heart rate and increase in cardiac stroke volume can lead to a recruitment of otherwise unperfused lamellae (Booth, 1978), thereby increasing the functional surface area and reducing the blood-to-water diffusion distance for oxygen. However, an exacerbation of the unfavorable ion and water flux can occur. As stated in the classic first paper on this topic (Randall et al., 1972) ‘Any change in the pattern of blood and water flow which reduces diffusion distance or increases the functional exchange area between blood and water will increase the rate of ion and water diffusion, unless there are compensatory reductions in the permeability of the gill epithelium’. This functional trade-off in optimal gill function has been termed the osmorespiratory compromise (Randall et al., 1972; Gonzalez and McDonald, 1992).
To date, the osmorespiratory compromise has been relatively well characterized in response to exercise. In the highly aerobic freshwater rainbow trout (Oncorhynchus mykiss), increases in swimming activity led to elevations in ion and water flux (Wood and Randall, 1973a,b,c; Gonzalez and McDonald, 1992; Postlethwaite and McDonald, 1995; Onukwufor and Wood, 2018).
However, fewer studies have looked at this phenomenon in response to the reduction of oxygen in the water. In hypoxia, fish improve the conditions for oxygen uptake by ventilating more water (Perry et al., 2009), and by changing blood flow patterns at the lamellae so as to increase functional surface area and decrease diffusion distance for exchange (Booth, 1978; Nilsson et al., 1995; Sundin and Nilsson, 1998). A likely consequence of these mechanisms is again a disturbance of ion and water balance as a result of the osmorespiratory compromise. Indeed, when exposed to hypoxia, rainbow trout experienced an increase in ion loss (Thomas et al., 1986; Iftikar et al., 2010; Robertson et al., 2015a), an effect that was paralleled by modifications in the gill epithelium, where ionocytes (mitochondria-rich cells) became more exposed, thereby increasing in surface area (Iftikar et al., 2010; Matey et al., 2011). In contrast, the Amazonian oscar (Astronotus ocellatus), a hypoxia-tolerant fish, exhibited a decrease in branchial ion and water flux in hypoxia (Wood et al., 2009; De Boeck et al., 2013; Robertson et al., 2015b). In addition, a reduction in the exposed surface area of ionocytes, due to pavement cell modifications, was observed at the gills (Wood et al., 2009; Matey et al., 2011; De Boeck et al., 2013). These responses occurred despite the fact that the oscar increased ventilation and the O2 transfer factor of the gills (an index of effective gill O2 permeability) during hypoxia (Scott et al., 2008a; Wood et al., 2009). These data suggest that hypoxia-tolerant fishes may circumvent or decouple the osmorespiratory compromise, by recruiting specific mechanisms at the level of the gill, preventing a disruption in the ionic balance, while gas exchange in hypoxia remains unaffected.
A few studies have examined the osmorespiratory compromise in seawater (SW) (Sardella and Brauner, 2007). Fish in SW are hypo-osmotic, contrary to what is seen in freshwater (FW). Therefore, it has always been thought that fish would face a net ion loading and water loss during hypoxia and exercise, given the reversed nature of the ionic gradients between fish and environment (Gonzalez, 2011). The few experimental studies addressing the osmorespiratory compromise in SW have supported this conclusion (Farmer and Beamish, 1969; Stevens, 1972; Wood et al., 2019). Additionally, there has been no research on this phenomenon at the isosmotic salinity (the salinity at which the concentration gradient between the body fluids and external water is minimal). In theory, the osmorespiratory compromise at this salinity should be negligible, as no net movement of ions and water between the plasma and the environment should occur. A goal of the present study was to address these knowledge gaps.
Our overall goal was to increase our understanding of the ionoregulatory aspects of the osmorespiratory compromise during hypoxia across a range of salinities. We chose the Atlantic killifish, also known as the mummichog, Fundulus heteroclitus, as our model, because of its outstanding euryhalinity (Marshall, 2013) and its exceptional tolerance of hypoxia (Cochran and Burnett, 1996; Richards et al., 2008; McBryan et al., 2016). Killifish are native to estuaries and salt marshes but often enter rivers and creeks. In these environments, they can encounter variations in salinity and dissolved oxygen. In recent studies, we found that salinity affected killifish oxygen consumption and hypoxia tolerance (Giacomin et al., 2019) and also water exchange rates (Wood et al., 2019). Key findings from these two studies were that killifish increased gill ventilation during hypoxia, and exhibited the classic hypoxic bradycardia. Acclimation to FW led to reductions in hypoxia tolerance, likely due to a decrease in gill respiratory surface area and the presence of a greater interlamellar cell mass. Acclimation to 11 ppt (i.e. the isosmotic salinity for killifish) proved to be the most advantageous for respiratory gas exchange.
With this background in mind, our overall hypothesis was that there is a trade-off between efficient gas exchange and ion/water balance at the gills (osmorespiratory compromise) and the nature of this trade-off is flexible, depending on environmental salinity and oxygen levels. This hypothesis led to several predictions. Our first prediction was that killifish would reduce gill ion permeability in acute hypoxia, minimizing the flux of ions and water in both FW and SW, as seen in other hypoxia-tolerant species. Second, we predicted that in FW and SW, killifish would decrease the exposed surface area of ionocytes when exposed to hypoxia, so that respiratory surface area could be maintained while ion permeability was reduced, thereby avoiding the negative osmoregulatory effects of the osmorespiratory compromise. Lastly, we posited that in killifish acclimated to the isosmotic salinity, the effects of the osmorespiratory compromise would be minimal so that permeability could be increased without incurring negative osmoregulatory effects of the osmorespiratory compromise.
MATERIALS AND METHODS
The experiments were performed throughout the year, from multiple field collections (April to October), between 2014 and 2018. All procedures were approved by the University of British Columbia Animal Care Committee (AUP A14-0251) and complied with the guidelines of the Canada Council for Animal Care. Fundulus heteroclitus macrolepidotus (Walbaum 1792) (northern subspecies) were collected by beach seine from a brackish estuary near Hampton, NH, USA, by Aquatic Research Organisms (Inc.) and air-shipped to the University of British Columbia, where they were held initially in 20 ppt water, as used by the supplier. They were later directly transferred to their experimental salinities, and held there for at least 6 weeks in each salinity prior to experimentation. There were no mortalities. Fish were held in 120 l glass tanks, containing FW (0 ppt: recirculating, charcoal-filtered dechlorinated Vancouver tap water containing mmol l−1: [Na+]=0.09, [Cl−]=0.10, [Ca2+]=0.13, [Mg2+]=0.01) or FW mixed with Instant Ocean Aquarium Salt (Instant Ocean, Spectrum Brands, Blacksburg, VA, USA) up to 11 ppt (mmol l−1: [Na+]=146, [Cl−]=196, [Ca2+]=4.3, [Mg2+]=21.0) or SW (35 ppt; mmol l−1: [Na+]=459, [Cl−]=663, [Ca2+]=12.4, [Mg2+]=53.6). Fish were fed daily with commercial flakes (Nutrafin Max Tropical Flakes) and fasted for at least 24 h prior to experiments. Salinity in the tanks and in the water prepared for experiments was monitored with a conductivity meter (Cond 3310, WTW, Xylem Analytics, Weilhem, Germany). The photoperiod was 12 h light:12 h dark, and the temperature during both acclimation and experimentation was 18–20°C. Water PO2 was monitored during acclimation and throughout all experiments using a handheld oxygen meter (Accumet AP84, Fisher Scientific, Toronto, ON, Canada).
Influence of hypoxia and salinity on plasma ion composition and gill ionoregulatory enzymes
Fish (n=8 at each salinity and O2 level) were held in individual plastic containers whose sides were replaced with a plastic mesh for free water flow. These were placed in 30 l darkened Plexiglas tanks overnight, one for each acclimation salinity (FW, 11 ppt and SW). The water in each 30 l tank was aerated and continuously recirculated by a small submersible water pump. Approximately 80% of the water was replaced prior to the start of the trial to avoid nitrogenous-waste build-up. PO2 was either maintained at normoxia (∼155 Torr=20.7 kPa) or reduced from normoxia to hypoxia (15±2 Torr=2 kPa) over a 30 min period and maintained at a constant level with bubbling of N2 or air. Fish were kept in hypoxia for 3 h, and at the end of the experiment were immediately killed with an overdose of neutralized MS-222 (Syndel Laboratories, Nanaimo, BC, Canada). Normoxic fish underwent the same protocol, but PO2 was kept at normoxia. Blood samples (∼100 µl) were quickly drawn by caudal puncture using a modified gas-tight Hamilton 100 µl syringe (Reno, NV, USA). Two 5 µl micro-hematocrit tubes were filled with blood and centrifuged at 10,000 g for 10 min. The remainder of the blood was also centrifuged (5 min, 10,000 g), and the separated plasma was then flash-frozen in liquid N2. Additionally, the left-hand side of the gill basket was excised and also flash-frozen in liquid N2. Plasma and tissue samples were kept at −80°C until analysis of plasma ion concentrations and osmolality, plasma lactate concentration, and gill Na+/K+-ATPase and H+-ATPase activity (described below). The right-hand side of the gill basket was immediately fixed in Karnovsky's fixative for 24 h at 4°C. The samples were transferred to 0.1 mol l−1 sodium cacodylate and stored at 4°C until use. For scanning electron microscopy (SEM), sections of the center part of the second gill arch were progressively dehydrated in ethanol, critically point dried with CO2, and then sputter-coated with 9 nm of gold–palladium. Samples were imaged on a Hitachi S-4700 Field Emission Scanning Electron Microscope (Tokyo, Japan) at 2000× magnification.
Influence of hypoxia and salinity on diffusive water flux rates
Separate groups of fish (n=12 at each salinity and O2 level) were used to measure diffusive water flux rates in animals acclimated to FW, 11 ppt and SW, in normoxia and exposed to hypoxia. Groups of n=6 fish were placed in darkened 1 l containers, at the acclimation salinity and dosed with 20 µCi l−1 of 3H2O (Amersham Pharmacia Biotech, Little Chalfont, UK). This loading period was performed overnight, while the chambers were constantly aerated, and temperature controlled to 20°C. Preliminary trials demonstrated that 3H2O concentration in the fish's body water pool and in the external water of the loading bath had reached an equilibrium within 6 h. After the loading period, the fish were quickly rinsed in non-radioactive water, and then transferred to individual darkened 250 ml containers, each filled with water at the appropriate salinity and PO2, which served as the experimental chambers. Water PO2 was set to normoxia (155 Torr) or hypoxia (15±2 Torr) and was kept at these levels for the following 60 min. Water samples (5 ml) were taken every 5 min for 60 min. Normoxia was restored and a final water sample was taken at 8 h, by which time the radioactivity in the fish had completely equilibrated with the external water. Scintillation fluor (Optiphase, Perkin Elmer, Waltham, MA, USA) was added to the water samples in a 2:1 (fluor:water) ratio, and samples were stored in the dark for a minimum of 2 h to eliminate chemiluminescence prior to scintillation counting for beta emissions (Beckman Coulter, LS6500, Wellesley, MA, USA). Tests showed that quench was constant.
Influence of hypoxia and salinity on gill paracellular permeability and drinking rate
Separate groups of fish (n=8 at each salinity and O2 level) were used to measure gill paracellular permeability and drinking rate, using radiolabeled polyethylene glycol ([3H]PEG-4000) following methods developed by Robertson and Wood (2014). Briefly, the appearance of radioactivity in the carcass was used as a measurement of gill paracellular permeability, while the appearance of radioactivity in the gut was used to measure drinking rate. Groups of fish (n=8) were exposed for 3 h in 1 l of water at the acclimation salinity labeled with 25 µCi l−1 [3H]PEG-4000 (Amersham). Preliminary experiments demonstrated that there was no rectal excretion of [3H]PEG-4000 in that time. Fish were allowed to settle in the aerated light-shielded experimental container for 2 h in normoxia prior to the addition of radioactivity. After that, PO2 was either maintained at 155 Torr by air bubbling or quickly reduced to 15±2 Torr by N2 bubbling, and when PO2 reached hypoxic levels, the radioisotope was added. Water samples (5 ml) were taken at 0, 1.5 and 3 h for scintillation counting. At the end of 3 h, all fish were killed by an overdose of neutralized MS-222, rinsed in radioisotope-free water, blotted dry and kept on ice until dissection. The gastrointestinal tract was tied at both ends (to avoid content loss) and excised. The gut and remaining carcass were weighed separately, and digested in sealed vials with 2 mol l−1 HNO3 for 48 h at 60°C, with periodic agitation. The digests were then centrifuged (5000 g for 5 min) and the clear supernatant was aliquoted for scintillation counting. Water and carcass digest samples (2 ml), and gut digest samples (0.2 ml) were dosed with 10 ml of scintillation cocktail (Ultima Gold AB, Perkin Elmer) and read on a scintillation counter (LS6500, Beckman Coulter). Radioactivity of the tissue digests was quench-corrected to the same counting efficiency as water samples based on a quench curve constructed previously with different amounts of tissue digest.
Unidirectional and net ion flux rates in killifish acclimated to FW, 11 ppt and SW
Previous studies (Wood and Laurent, 2003; Wood, 2011) have demonstrated that it is necessary to use different methods to measure unidirectional flux of Na+ in saline waters versus FW, and we adopted those methods in the present study. All techniques described below were employed to meet the experimental criteria outlined in Wood (2011). Briefly, the specific activity (ratio of radioisotopic Na+ to ‘cold’ Na+) in the compartment where radioactivity appearance was being monitored, either water or fish, needs to remain less than 10% of the specific activity in the compartment to which radioactivity was added, so as to prevent radio-isotopic back flux. The difficulty of meeting this criterion when measuring 22Na flux in 11 ppt and full-strength SW constrained our ability to replicate all the experiments performed in FW at those salinities. A limitation of these approaches is that because the methods differ for influx and efflux determinations, and must be performed on different fish over different time frames (0.5 h for influx, 3 h for efflux), influx and efflux rate measurements may not be quantitatively comparable to one another. Therefore comparisons should only be made of influx rates across treatments, and efflux rates across treatments, and not of efflux rates versus influx rates within treatments. Therefore, our investigation of the unidirectional ion flux responses to hypoxia is more detailed in FW than in 11 ppt or SW. Furthermore, our previous study (Giacomin et al., 2019) indicated that the impacts of the hypoxia-induced osmorespiratory compromise are most intense in FW killifish. All 22Na measurements were made with a gamma counter (1470 Wallac-Wizard, Perkin Elmer, Waltham, MA, USA).
Separate groups of fish (n=15 at each salinity and O2 level) were used for each of the experimental series described below. For all experimental flux measurements, fish were held in individual 250 ml plastic containers, served with aeration tubing, darkened with black plastic and partially submerged in a wet table to ensure temperature control. The water surface was covered in plastic so as to avoid oxygen diffusion from the air, and to prevent the fish from performing aquatic surface respiration.
Ion flux in FW-acclimated killifish
In FW, Na+ influx (JNa+influx) was measured by monitoring the disappearance of radioactivity from the water. To that end, fish were transferred to the containers described above and allowed to acclimate overnight under flow-through conditions. Prior to the start of the experiment, flow-through ceased, the water level was set at 250 ml, and 5 µCi 22Na was added. Water samples (5 ml) were collected hourly for 3 h and stored for measurement of radioactivity, total Na+, Cl−, K+ and ammonia concentrations.
Cl−, K+ and total ammonia (NH3+NH4+) net flux rates (nmol g−1 h−1) were determined from the concentrations in the water (nmol l−1) and calculated using an equation analogous to Eqn 5.
Unidirectional Na+ influx and efflux in 11 ppt- and SW-acclimated killifish
Ammonia net flux rate (Jammnet flux: nmol g−1 h−1) was determined from the measured total concentrations of ammonia (NH3 and NH4+) in the water and calculated as described for the FW trials above. Net flux rates of Na+, K+ and Cl− were not measured in 11 ppt and SW because of the background ionic concentrations being too high to reliably detect fish-induced changes.
Water Na+ and K+ concentrations (nmol l−1), and plasma Na+ concentrations (mmol l−1) were determined by atomic absorption (1275 Atomic Absorption Spectrophotometer, Varian, Mulgrave, VIC, Australia) using certified commercial standards (Fluka Analytical, Sigma-Aldrich, St Louis, MO, USA). Plasma osmolality (mOsm kg−1) was measured using a Wescor vapor pressure osmometer and standards (Wescor 5100C, Logan, UT, USA). Water (µmol l−1) and plasma (mmol l−1) Cl− were determined colorimetrically using the mercury–thiocyanate method described by Zall et al. (1956). Ammonia concentration (µmol l−1) in the water was measured using the colorimetric method of Verdouw et al. (1978).
Plasma lactate concentration (mmol l−1) was determined as described by Bergmeyer (1983). Briefly, plasma samples were deproteinized with 8% perchloric acid, centrifuged briefly to precipitate proteins (8 min, 5000 g), and assayed in triplicate on a microplate spectrophotometer (SpectraMAX Plus; Molecular Devices, Menlo Park, CA, USA) with a reaction buffer solution (0.18 mol l−1 glycine, 0.15 mol l−1 hydrazine, 0.015 βNAD+, 15 U ml−1 lactate dehydrogenase, pH 9.4). The reaction mixture was incubated for 60 min at 37°C, and lactate concentration was determined against a standard curve.
Na+/K+-ATPase and NEM-sensitive ATPase activity (µmol ADP mg−1 protein h−1) were measured in gill tissues according to protocols described in McCormick (1993) and Lin and Randall (1993), respectively. Tissues were homogenized (1:10 w/v) in ice-cold buffer solution [50 mmol l−1 imidazole, 125 mmol l−1 sucrose and 5 mmol l−1 EGTA (pH 7.3)], centrifuged (5000 g, 3 min, 4°C), and supernatants kept on ice throughout the analysis. All samples were assayed within 2 h of thawing and homogenization. Ouabain (0.05 mmol l−1) was used as an inhibitor of Na+/K+-ATPase with sodium azide (NaN3, 5.15 mmol l−1) as an inhibitor of ATP synthase. N-Ethylmaleimide (NEM) at 1.03 mmol l−1 was used as an inhibitor of H+-ATPase. An important caveat to note is that at this high concentration, NEM may not be a specific inhibitor of only V-type H+-ATPase and might inhibit other ATPases (Forgac, 1989; Lin and Randall, 1993). For this reason, in accord with Tipsmark et al. (2004), we describe these results as NEM-sensitive ATPase throughout; all the tissue samples were analyzed using the same method, so the comparisons between treatments remain valid. Activity was calculated by the difference between the amounts of adenosine diphosphate (ADP) produced in control and inhibited reactions. Absorbance was monitored at 10 s intervals over 5 min (25°C) on the same microplate spectrophotometer. Protein concentration in homogenates was determined using the Bradford reagent (Sigma-Aldrich).
For the measurement of hematocrit (Hct), micro-hematocrit tubes (2×5 µl) were centrifuged (10,000 g, 10 min) and Hct was measured as the percentage of packed red blood cells (%RBC) within the total blood volume.
Morphometric quantification of exposed ionocyte surface area and density (number of ionocytes per mm2) was performed on randomly selected areas of the afferent surface of filaments near the respiratory lamellae. The individual exposed surface area of approximately 50 ionocytes was measured over randomly selected areas of the epithelium for each fish at 2000× magnification. The surface area of individual pavement cells (PVCs) was also measured on randomly selected cells from each of five fish on the same micrographs. Microridge density on the surface of PVCs was calculated on 20 random cells from each of five fish as the total number of intercepts (for the whole cell) of microridge profiles on a lattice grid composed of 2.82×2.82 µm squares superimposed on the micrograph. This result (total number of intercepts per individual cell) was then divided by the individual PVC surface area (µm2), yielding intercepts per µm2 of PVC surface area. The exposed ionocyte fractional surface area was calculated by multiplying ionocyte surface area (µm2 per cell) by ionocyte density (cells µm−2). All measurements were performed using ImageJ v152 (Public domain).
All data are shown as means±s.e.m. (n=number of animals). Assumptions of parametric statistics (data normality and homogeneity of variances) were checked, and if not met, data were transformed using a log transformation. Data were analyzed using one-way ANOVA, two-way ANOVA or repeated measures ANOVA as appropriate. Detailed results of the statistical tests and post hoc analysis are shown in figure captions. When applied, one-way ANOVA was followed by a Tukey post hoc test. Two-way ANOVA was followed by a Bonferroni post hoc test. Mean values were considered significantly different when P<0.05.
Plasma ions and enzyme activity
Salinity acclimation exerted a significant overall effect on plasma ions and osmolality (Fig. 1). Under normoxia, plasma [Na+] was significantly higher in SW-acclimated fish (∼150 mmol l−1) than in FW-acclimated fish (∼135 mmol l−1) (Fig. 1A). Plasma Cl− followed a similar trend, where 11 ppt-acclimated fish showed an intermediate value between those of FW- and SW-acclimated fish (∼110 and ∼140 mmol l−1, respectively; Fig. 1B). Neither plasma [Na+] (Fig. 1A) nor plasma [Cl−] (Fig. 1B) was significantly affected by hypoxia at any salinity. There was no significant interaction between salinity and oxygen for plasma [Na+] and [Cl−] (Fig. 1A,B). Plasma osmolality was similar in fish acclimated to 11 ppt and SW (35 ppt) (∼375 mOsm kg−1) and both were significantly higher than in FW-acclimated fish (∼340 mOsm kg−1; Fig. 1C). There was no significant effect of hypoxia and no significant interaction between oxygen and salinity on plasma osmolality (Fig. 1C).
Hypoxia exposure had a significant overall effect on both gill Na+/K+-ATPase and NEM-sensitive ATPase activity, with lower activity in hypoxia than normoxia (Fig. 2). Two-way ANOVA did not detect a significant effect of salinity acclimation on either Na+/K+-ATPase or NEM-sensitive ATPase activity, and there was no significant interaction between salinity and oxygen.
Blood Hct and plasma lactate
Hypoxia exposure had a significant overall effect on blood Hct, while salinity acclimation did not (Fig. 3A). At all salinities, hypoxia exposure caused an increase in blood Hct, which was significantly different from normoxic values in 11 ppt and SW (35 ppt) (Fig. 3A).
There were significant overall effects of both salinity and hypoxia on plasma lactate concentration (Fig. 3B); however, plasma lactate was significantly elevated only in FW- and 11 ppt-acclimated fish. Fish in 11 ppt showed the largest difference in plasma lactate between normoxia and hypoxia exposure (60% increase), followed by fish acclimated to FW (46% increase; Fig. 3B).
Diffusive water flux rates, gill paracellular permeability and drinking rate
Salinity acclimation had a significant overall effect on the diffusive water flux rate (Fig. 4A). Water flux rate was the highest (∼94% of the body water pool h−1) in fish acclimated to FW, and significantly different from that of fish acclimated to 11 ppt and SW, which had similar rates (∼57–47% h−1; Fig. 4A). There was a significant decrease (30%) with hypoxia exposure in FW-acclimated fish, but no effect of hypoxia in fish acclimated to 11 ppt and SW (Fig. 4A).
There was an overall effect of salinity on gill clearance rates of [3H]PEG-4000, an index of gill paracellular permeability, while under normoxia, FW- and SW-acclimated fish exhibited very similar rates and 11 ppt-acclimated fish had a significantly lower rate (∼1.0 µl g−1 h−1) (Fig. 4B). While hypoxia exposure did not have an overall effect on the [3H]PEG-4000 gill clearance rate (Fig. 4B), at 11 ppt gill clearance rate was elevated (Fig. 4B).
Both salinity acclimation and hypoxia exposure had significant overall effects on drinking rate (Fig. 4C). Under normoxia, drinking rate was the lowest in FW (∼0.15 µl g−1 h−1), and approximately 40-fold higher (∼6.1 µl g−1 h−1) in SW (Fig. 4C). Drinking rate was decreased significantly with hypoxia exposure (88%) in SW (Fig. 4C).
Unidirectional flux rates of Na+, and net flux rates of Cl−, K+ and ammonia
In FW, there were no significant changes in unidirectional flux rates of Na+ or the net flux rates of Cl−, K+ and ammonia in the fish subjected to the normoxic control sham procedures (data not shown). FW-acclimated killifish exhibited an average JNa+influx of 310 nmol g−1 h−1 in normoxia (Fig. 5A). In the first hour of hypoxia, JNa+influx did not change, but it decreased significantly by about 55% for the remainder of the hypoxia exposure (Fig. 5A). JNa+influx returned to pre-exposure levels as soon as fish were returned to normoxia (Fig. 5A). JNa+efflux showed a similar response pattern to that for influx, with an overall reduction in the second and third hours of hypoxia, and recovery to close to control levels upon re-instatement of normoxia (Fig. 5B). In normoxia, the average JNa+efflux was about −300 nmol g−1 h−1, and during hypoxia, it decreased significantly by 50% at the second hour, and by 36% at the third hour (Fig. 5B). The resulting net flux rates (∼−25 nmol g−1 h−1) were therefore essentially unchanged with no significant differences among normoxia, hypoxia and recovery (Fig. 5C).
In normoxia, the average net Cl− flux rate was −39 nmol g−1 h−1, and did not change in the first hour of hypoxia, but was significantly reversed to about +100 nmol g−1 h−1 in the second and third hours (Fig. 6A). Cl− net flux rate returned to pre-exposure levels once normoxia was restored. In normoxia, the net K+ flux rate was about −20 nmol g−1 h−1, and during hypoxia, it followed the same pattern as seen for unidirectional Na+ flux, and net Cl− flux, where the rates were unchanged in the first hour, but decreased substantially (by about 70%) in the second and third hours (Fig. 6B). In recovery, the K+ flux rate was restored to levels similar to original in normoxia (Fig. 6B). The ammonia excretion rate was about −350 nmol g−1 h−1 in normoxia, and it significantly decreased by about 60% immediately upon hypoxia exposure (Fig. 6C). Ammonia excretion did not change during subsequent hypoxia, but increased immediately back to control levels once normoxia was restored (Fig. 6C).
In SW, only single values were obtained for undirectional Na+ flux rates during normoxia and hypoxia, and net flux rates of Na+, K+ and Cl− were not measured. In Fig. 7, we have shown the overall time-averaged unidirectional Na+ flux rates for normoxia and hypoxia of FW-acclimated killifish (from Fig. 5A,B) in order to make comparisons with fish acclimated to 11 ppt and SW. Note that the scale is broken to accommodate the wide range of values.
On average, JNa+influx in FW-acclimated fish was +310 nmol g−1 h−1, while in hypoxia this decreased to +210 nmol g−1 h−1 (Fig. 7A). The average JNa+efflux was −300 nmol g−1 h−1, while in hypoxia it decreased to −250 nmol g−1 h−1 (Fig. 7B). Fish acclimated to 11 ppt in normoxia showed a severalfold higher JNa+influx of +6400 nmol g−1 h−1, while in hypoxia this increased by 150% to +16,000 nmol g−1 h−1 (Fig. 7A). There was also a 92% increase in JNa+efflux in 11 ppt fish exposed to hypoxia (−5400 nmol g−1 h−1) in comparison to normoxia (−2800 nmol g−1 h−1; Fig. 7B). The JNa+influx in normoxia of SW-acclimated fish was about +22,000 nmol g−1 h−1, which decreased significantly by 13% when fish were exposed to hypoxia (Fig. 7A). In normoxia, JNa+efflux was approximately −12,000 nmol g−1 h−1, and when SW-acclimated fish were exposed to hypoxia, it was not significantly different at about −10,000 nmol g−1 h−1 (Fig. 7B). Although the absolute unidirectional Na+ flux rates in SW were more than an order of magnitude higher than in fish acclimated to FW, the overall patterns of response to hypoxia were similar to those in fish acclimated to FW (Fig. 7). However, the patterns were different from those in fish acclimated to the isosmotic salinity (11 ppt), where flux increased rather than decreased during hypoxia. Note that the discrepancies in absolute values between unidirectional Na+ influx and efflux rates in 11 ppt- and SW-acclimated fish (Fig. 7) were likely of technical origin (see Materials and Methods), but the relative differences between normoxia and hypoxia remain comparable as they were determined by the same techniques within a salinity.
Salinity acclimation had a significant overall effect on Jammnet flux, with elevated rates in both normoxia and hypoxia at 11 ppt relative to FW (Fig. 8). Hypoxia also had a significant overall effect, reducing rates at all salinities by 33–54%.
Gill epithelial morphology and morphometrics
The shape of ionocyte apical openings varied greatly among salinities. Fish acclimated to FW showed ionocytes that were circular, oval, triangular and roughly trapezoidal, with large exposed surfaces (Fig. 9A), compared with fish acclimated to 11 ppt and SW, which showed much smaller round and/or oval apical openings (Fig. 9B,C). The surfaces also varied between showing protruding and flat microvilli. These were smaller in fish acclimated to SW (Fig. 9C). Hypoxia-exposed fish (Fig. 9D,E,F) had PVCs which were less complex in surface, with less-developed microridges and flatter surfaces throughout. Salinity acclimation had a significant effect on the exposed surface area of individual ionocytes, while hypoxia exposure had no overall effect (Fig. 10A). In FW, ionocyte surface area was higher than in SW, while at 11 ppt, fish exhibited intermediate-sized ionocytes (Fig. 10A). At 11 ppt, ionocyte surface area was significantly elevated in hypoxia. Both hypoxia and salinity had significant effects on individual PVC surface area (Fig. 10B). PVC surface area was significantly higher in hypoxia at 0 and 35 ppt, while no effects of hypoxia were seen at 11 ppt (Fig. 10B). Ionocyte density was significantly affected by salinity, where FW-acclimated fish had a lower density than SW-acclimated fish. There was no significant effect of hypoxia on ionocyte density (Fig. 10C). PVC microridge density was significantly affected by hypoxia, independent of salinity acclimation (Fig. 10D). In fish exposed to hypoxia, microridge density (intercepts per unit surface area) decreased by 42% in FW-acclimated fish, and by 23% in 11 ppt-acclimated fish (Fig. 10D). There was a significant effect of salinity on ionocyte fractional surface area, where FW-acclimated fish had higher fractional surface area than 11 ppt- or SW-acclimated fish (Fig. 10E).
Overall, our findings lend support to our hypothesis that in the killifish there is a trade-off between efficient gas exchange and ion/water balance at the gills (osmorespiratory compromise) and the nature of this trade-off is flexible, depending on environmental salinity and oxygen levels. The gas exchange components of this trade-off have been addressed in Giacomin et al. (2019), and the ion and water exchange components in the present investigation. This hypothesis led to several predictions with respect to osmoregulation. Our first prediction, that killifish would decrease gill permeability during hypoxia exposure, reducing ion and water flux in both FW and SW, was generally supported. In FW-acclimated fish, active Na+ influx rate, gill Na+/K+-ATPase and NEM-sensitive ATPase activity (see Materials and Methods), ammonia, K+ and Cl− efflux rates, as well as diffusive water flux rates were all reduced in hypoxia. The responses of this suite of markers suggest that both transcellular and paracellular permeability were reduced during hypoxia as part of a regulated process. In SW, where Na+ gradients are reversed, and Na+ flux is complicated by the presence of a large exchange diffusion component, both Na+ influx and efflux declined in accord with our prediction.
Contrary to our second prediction, in FW and SW, killifish did not show a decrease in the exposed surface area of ionocytes when exposed to hypoxia, but there was an increase in the surface area of PVCs. Nevertheless, at 11 ppt, killifish showed a significantly elevated surface area of individual ionocytes, in accord with the increased flux rates at this salinity.
Our third prediction was that because killifish acclimated to the isosmotic salinity (11 ppt) were able to prioritize gas exchange during hypoxia, as shown by Giacomin et al. (2019), they would increase permeability while exhibiting minimal disturbance of ionoregulatory homeostasis. Both unidirectional Na+ influx and efflux rates increased as well as [3H]PEG-4000 clearance rate, in accord with our hypothesis. However, diffusive water flux rate was not altered by hypoxia. At all three salinities, plasma ions and osmolality were unaffected by hypoxia exposure, indicating that despite disturbance of ion and water flux, F. heteroclitus are able to maintain osmoregulatory homeostasis under hypoxia in diverse environments, via a flexible strategy with respect to the osmorespiratory compromise.
Effects of salinity and hypoxia on plasma ion homeostasis
Salinity acclimation had only a small effect on the plasma concentrations of Na+ and Cl− and on plasma osmolality (Fig. 1), in accord with previous studies (Wood and Marshall, 1994). In their natural habitat, F. heteroclitus can encounter transient as well as seasonal fluctuations in salinity; therefore, the ability to maintain ionoregulatory homeostasis is likely an adaptation for this unpredictable environment. As a result of this flexible strategy with respect to the osmorespiratory compromise, plasma ion concentration and osmolality were not significantly affected by hypoxia in killifish at any of the three salinities tested (Fig. 1), indicating a remarkable homeostatic regulation of these parameters, despite diverse changes in associated parameters such as flux rates and transport enzyme activity. In previous studies, an increase in plasma [Na+] with hypoxia was observed for two contrasting species, the hypoxia-tolerant Amazonian oscar (Astronotus ocellatus) (Wood et al., 2007) and the hypoxia-intolerant rainbow trout (Oncorhynchus mykiss) (Thomas et al., 1986). In both studies, the slight increase in plasma [Na+] was attributed to a fluid shift towards the intracellular compartment as well as a movement of Na+ to balance the increased lactate in the plasma (Thomas et al., 1986). In our study, we observed an increase in plasma lactate concentration in response to hypoxia in FW- and 11 ppt-acclimated fish (Fig. 3B). Normoxic plasma lactate values at all salinities are comparable to those in previous studies in F. heteroclitus that employed similar blood-sampling techniques (Richards et al., 2008; Healy and Schulte, 2012). However, they are above resting values reported for other larger fish where blood sampling by indwelling catheters was possible (e.g. Milligan and Wood, 1986). Although blood acid–base status was not measured in the present study, Giacomin et al. (2019) reported that a comparable hypoxia exposure caused a reduction in blood pH. Therefore, it is likely that killifish are able to deal with a hypoxia-induced acidosis by employing mechanisms that are independent of the modulation of plasma ions.
Effects of salinity and hypoxia on branchial enzyme activity
A fundamental problem imposed by hypoxia exposure is the maintenance of metabolic energy balance, as a decrease in blood O2 content directly impacts mitochondrial energy production in the form of ATP (Boutilier and St-Pierre, 2000). Hochachka et al. (1996) suggested that maintaining transmembrane electrical gradients through Na+/K+-ATPase is amongst the most energetically costly processes performed by a cell. Using an isolated gill perfusion preparation, Morgan and Iwama (1999) found that Na+/K+-ATPase and H+-ATPase contributed to 35% of the total gill ṀO2. Therefore, the reduction in both of these enzymes in FW-acclimated fish exposed to hypoxia (Fig. 2A,B) was in accord with the reduction of active Na+ influx (Fig. 5A) and would have a clear benefit in terms of energy conservation. A similar rapid (3–4 h) down-regulation of Na+/K+-ATPase activity was observed in the Amazonian oscar, where a lack of change in protein abundance raised the hypothesis that post-translational modifications may have been responsible for the reduction in enzyme activity (Richards et al., 2007; Wood et al., 2007). Notably, killifish in 11 ppt and SW did not show this response.
FW-acclimated killifish have a reduced respiratory capacity in hypoxia relative to other salinities (Giacomin et al., 2019), which could account for the different strategy that they employed in dealing with the osmorespiratory compromise. We suggest that the downregulation in enzyme activity in the FW killifish could be a reflection of the compromised respiratory ability in these animals, as a strategy to conserve energy, possibly in concert with a simultaneous downregulation of gill permeability. This way, plasma ion concentrations were maintained and whole-animal ion balance remained unaffected, in accord with our overall hypothesis of flexible responses dependent upon salinity.
Effects of salinity and hypoxia on diffusive water flux rates, paracellular permeability and drinking rates
In confirmation of our initial prediction that killifish would decrease gill ion and water permeability when exposed to hypoxia, we observed a reduction in the flux of 3H2O across the gills, but this effect was salinity dependent, occurring only in FW (Fig. 4A). 3H2O flux represents unidirectional diffusive water flux, and is approximately 100-fold larger than osmotic water flux, which is net water flux usually estimated from body mass changes and urine flow (Stevens, 1972; Isaia, 1984). Diffusive water flux is thought to occur mainly by the transcellular route (Evans 1969; Isaia, 1984), very likely through aquaporins (Evans et al., 2005), whereas osmotic water flux may occur by both paracellular and transcellular routes. Additional explanation of the significance of diffusive water flux measurements has been provided by Onukwufor and Wood (2018) and Wood et al. (2019). In general, SW teleosts have lower gill permeability to water than FW teleosts (Evans, 1969), and the water permeability of euryhaline fishes decreases with transfer to SW (Potts and Fleming, 1970). This appears to correlate with lower expression of levels of aquaporins at higher salinities (reviewed by Cutler et al., 2007; Madsen et al., 2015; Breves et al., 2016). Our data agree with this, where in normoxia, fish that were acclimated to FW showed higher water permeability than their SW-acclimated counterparts. The decrease in diffusive water flux in FW during hypoxia but not at the other salinities may suggest the involvement of aquaporins in this response.
Our findings of reduced diffusive water flux during hypoxia in FW killifish are comparable to observations on the Amazonian oscar, which was able to reduce both diffusive and osmotic water flux during hypoxia (Wood et al., 2009), while maintaining high branchial oxygen permeability (Scott et al., 2008a). The explanation offered by these authors was that there was an adaptive reduction in transcellular permeability, due to an observed paving over of ionocytes by PVCs, in addition to channel (e.g. aquaporin) closure (Wood et al., 2009; Matey et al., 2011). Thus, we speculate that this ability to similarly downregulate gill water permeability in the FW-acclimated killifish, an effective decoupling of the classic osmorespiratory compromise, could be an adaptation to frequent variations in O2 in their natural environments (Layman et al., 2000; Smith and Able, 2003). In contrast, gill permeability and consequently diffusive water flux rates increased markedly in both goldfish (Carassius auratus) and rainbow trout during exposure to moderate hypoxia – the classic trade-off of the osmorespiratory compromise response (Loretz, 1979; Onukwufor and Wood, 2018). When rainbow trout were exposed to continuing severe hypoxia, water flux rates returned to control levels, indicating some regulatory ability, perhaps involving aquaporins (Onukwufor and Wood, 2018).
[3H]PEG-4000 (MW-4000) is a classic gill paracellular marker (Wood and Part, 1997) and its use in whole-animal experiments in the influx direction has been validated (Robertson and Wood, 2014). In normoxia, 11 ppt fish had lower gill paracellular permeability, in accordance with the theory that at the isosmotic salinity, the need for active Na+ excretion would be reduced, as gradients between fish and environment are minimal, and maintenance of high paracellular permeability is unnecessary (Wood and Marshall, 1994). Paracellular permeability increased in killifish acclimated to the isosmotic medium, opposite to what was seen in fish acclimated to FW and SW (Fig. 4B). These results coupled with increased ionocyte surface area at 11 ppt (Fig. 10A) may account for the elevation in unidirectional Na+ flux (Fig. 7A,B), suggesting that despite the maintenance of respiratory gas exchange (Giacomin et al., 2019), no compromise in ion and osmoregulation occurred.
In hypoxia, no change in [3H]PEG-4000 clearance rate was seen for the Amazonian oscar, and all reductions in gill permeability during hypoxia appeared to be transcellular in nature (Wood et al., 2009), contrary to our data (Fig. 4B). Noteworthy, the absolute branchial clearance rate in the FW oscar was less than 50% of that of the killifish in FW or SW. This suggests that the oscar may more tightly regulate gill permeability in its ion-poor environment, while the FW-acclimated killifish with normally higher paracellular permeability may employ reductions in both paracellular and transcellular permeability during hypoxia as part of its flexible osmorespiratory strategy (discussed below).
Drinking rates in SW were much higher than in FW (Fig. 4C), in accord with several previous studies on F. heteroclitus (Potts and Evans, 1967; Scott et al., 2006; Wood et al., 2019), reflecting the fact that marine teleosts must ingest water to compensate for net osmotic water losses across the gills (Takei and Balment, 2009). SW killifish reduced drinking in hypoxia, indicating that dealing with the absorption and excretion of excess salts gained upon imbibing SW must be an energy-demanding process. In normoxia, the stimulation of drinking in SW and its cessation in FW are both thought to be mediated by the renin–angiotensin system (Malvin et al., 1980). Possibly, the reduction of drinking in hypoxia is mediated by the same pathway; further studies on this present an interesting area for future investigations.
Effects of salinity and hypoxia on ammonia excretion
Ammonia net flux rates were inhibited by hypoxia at all three salinities (Fig. 8). In FW, Jammnet flux decreased immediately upon hypoxia exposure (Fig. 6C). It is widely accepted now that in FW fishes the majority of ammonia excretion occurs transcellularly, most likely via Rh proteins (Wright and Wood, 2009; Weihrauch et al., 2009). Therefore, the reduction in Jammnet flux is likely another indication of a reduction in transcellular permeability in response to hypoxia exposure in FW-acclimated killifish. In addition to transcellular ammonia transport, paracellular ammonia flux is also probably an important pathway in SW-acclimated fishes, given the leakier nature of junctions in SW gills (Wilkie, 1997, 2002). Giacomin et al. (2019) have shown that at low PO2, F. heteroclitus depresses ṀO2 regardless of salinity acclimation, and therefore a reduction in aerobic metabolism could be an additional cause of decreased ammonia excretion.
Effects of salinity and hypoxia on unidirectional and net ionic flux rates
Fish that were acclimated to 11 ppt or SW had more than one order of magnitude higher Na+ flux than in FW-acclimated fish (Fig. 7A), in agreement with previous studies (Motais et al., 1966; Potts and Evans, 1967; Maetz et al., 1967; Wood and Laurent, 2003; Wood, 2011). Flux in 11 ppt was 20-fold higher than that in FW, but 3.5-fold lower than that in SW (Fig. 7A). This is explained by the fact that, in contrast to FW acclimation, at higher salinities, killifish exhibit a large percentage contribution of exchange diffusion to unidirectional Na+ flux (reviewed by Wood and Marshall, 1994). This component is proportional to external [Na+] (Wood and Marshall, 1994), and not present in FW-acclimated animals (Patrick et al., 1997).
In our previous work (Giacomin et al., 2019), we showed that in comparison with SW- and 11 ppt-acclimated fish, FW-acclimated animals showed the greatest limitation in respiratory capacity when exposed to hypoxia. Therefore, we expected that the largest compensatory responses in order to maintain ionoregulatory balance would be seen in FW-acclimated fish. Based on the responses of the Amazonian oscar (Wood et al., 2007, 2009; De Boeck et al., 2013; Robertson et al., 2015a), we had predicted that the killifish would show a hypoxia response opposite to the classic osmorespiratory compromise in both FW and SW, decreasing overall gill permeability and ion flux. Indeed, the FW-acclimated F. heteroclitus showed reduced JNa+influx and JNa+efflux during hypoxia, so that no changes were seen in net Na+ flux (Fig. 5C). Given the lack of an exchange diffusion component in the FW-acclimated killifish (Patrick et al., 1997), the very similar reductions of JNa+influx and JNa+efflux during hypoxia were probably independent of each other. The reduction in both unidirectional fluxes appear to be regulated phenomena, as they were significantly inhibited only at the second hour of exposure, remaining depressed until normoxia was restored (Fig. 5A,B). This represents an effective decoupling of the classic osmorespiratory compromise as part of the flexible strategy. Interestingly, net K+ flux was reduced with an identical time-dependent profile (Fig. 6B). As K+ concentrations are much higher inside cells than in blood plasma, K+ flux is often considered a marker of transcellular permeability (Laurén and McDonald, 1985; Wood et al., 2007, 2009; Robertson et al., 2015b), so a role for the transcellular pathway in Na+ flux reduction is implicated, as in the oscar (Wood et al., 2009). This would also be in accord with the transcellular origins of the reductions in diffusive water permeability and ammonia excretion (Fig. 6C), both discussed earlier.
FW-acclimated killifish do not take up Cl− at the gills (Patrick et al., 1997; Patrick and Wood, 1999; Wood, 2011), only at the opercular epithelium (Wood and Laurent, 2003), so the apparent reversal of net Cl− loss during hypoxia (Fig. 6A) suggests that the diffusive loss pathway at the gills for Cl− was also down-regulated. However, it should be noted that none of the positive values of Cl− flux during hypoxia (Fig. 6A) were significantly different from zero (one sample t-test). However, our results for F. heteroclitus point to a response different from the typical osmorespiratory compromise seen in salmonids, where ion and water flux are exacerbated when gas transfer needs to be upregulated (Wood and Randall, 1973b,c; Iftikar et al., 2010).
We earlier showed that acclimation to 35 ppt (SW) also placed constraints on respiratory performance during hypoxia, though not as severe as those at 0 ppt (FW) (Giacomin et al., 2019). At 35 ppt, fish are hypo-osmotic, so the classical osmorespiratory compromise would predict increased Na+ loading. This did not occur here – both Na+ influx and efflux rates were reduced in hypoxia by about 15%, though only the change in the former was significant, and there was no significant change in the [3H]PEG-4000 clearance rate. While the Na+ efflux component includes the active pathway through leaky paracellular junctions (Evans, 2005, 2008), it must be remembered that the major components of both unidirectional Na+ fluxes are exchange diffusion fluxes (Motais et al., 1966; Potts and Evans, 1967; Maetz et al., 1967; Wood and Marshall, 1994). Overall, these observations suggest that there was a decrease in branchial permeability during hypoxia in SW, as in FW, again providing evidence for a flexible response that is salinity dependent. Wood and Grosell (2015) have reported that acute hypoxia exposure reduced branchial transepithelial potential in F. heteroclitus acclimated to both 0 and 35 ppt, providing additional evidence of decreased ion flux at both salinities.
Killifish acclimated to isosmotic salinity (11 ppt) have the greatest capacity to maintain oxygen consumption in hypoxia (Giacomin et al., 2019), in accord with the idea that the osmorespiratory compromise at the isosmotic salinity is of minimal consequence. The observed increases in both Na+ influx and efflux rates coupled with the increase in the paracellular permeability likely reflected an increase in functional surface area of the gills for respiratory gas exchange. Additionally, we observed an increase in the exposure of ionocytes in hypoxia. However, the unchanged diffusive water flux rate (Fig. 4A) and the greatest lactate accumulation (Fig. 3B) at this salinity argue against our prediction. Unfortunately, no data are available for transepithelial potential in 11 ppt-acclimated fish exposed to hypoxia, which might clarify the responses seen. Perhaps there are other constraints on gill function at the isosmotic salinity of which we are currently unaware.
Salinity- and hypoxia-induced changes in gill morphology
We observed a well-established effect of salinity on ionocyte surface area (Fig. 10A), where FW-acclimated fish showed large, oval and sometimes trapezoid-shaped ionocytes (Fig. 9A), with protruding microvilli, whereas SW-acclimated fish had ionocytes with small round deep openings, called apical crypts (Fig. 9C). These salinity-specific differences in ionocyte morphology are in accordance with previous studies on FW- and SW-acclimated killifish, where morphological differences were similarly noted (Katoh and Kaneko, 2002, 2003; Scott et al., 2008b).
Rainbow trout increase the density and exposed surface area of ionocytes shortly after hypoxia exposure (Iftikar et al., 2010; Matey et al., 2011), in accord with the typical osmorespiratory compromise (increased Na+ flux rates) shown by this species. In the Amazonian oscar, large reductions in Na+ flux rates during hypoxia were paralleled by an 80% reduction in fractional ionocyte surface area, which was largely due to covering by PVCs (Wood et al., 2009). A simultaneous increase in functional gill surface area (due to increased perfusion) and decrease in exposed ionocyte surface area was believed to maintain O2 transfer in hypoxia (Scott et al., 2008a), while concomitantly reducing gill permeability to ions. Therefore, our second prediction was that a similar mechanism would occur in hypoxia-exposed killifish. Specifically, we posited that ionocyte surface area would be reduced through a decrease in exposed ionocyte density and/or by PVCs stretching over ionocytes, thereby decreasing their exposed surface. Contrary to our prediction, hypoxia exposure in FW and SW did not change the fractional surface area of the filaments occupied by exposed ionocytes (Fig. 10E). However, individual ionocyte surface area increased in 11 ppt-acclimated fish in hypoxia (Fig. 10A). This is in accord with increased gill ion flux and paracellular permeability seen in 11 ppt-acclimated fish (discussed above). We also observed an increase in PVC surface area in FW- and SW-acclimated fish with hypoxia exposure, which could be explained by the PVCs increasing in size and spreading so as to cover up the ionocytes. This would be similar to the mechanism previously seen in oscar exposed to hypoxia (Matey et al., 2008, 2011; Wood et al., 2009). Overall, fish exposed to hypoxia show a less-complex PVC surface, where, contrary to normoxic fish, the well-defined microridge structure disappeared. The lower microridge density in hypoxia-exposed FW-acclimated fish (Fig. 10D) is in accord with this idea.
As the prevalence and severity of hypoxia in the world's aquatic environments increase, a comprehensive understanding of the physiological responses to hypoxia has become extremely important. Collectively, the present data elucidate the nature of the osmorespiratory compromise following acute hypoxia exposure in F. heteroclitus at different salinities. Salinity had a strong influence on how killifish responded to hypoxia as measured by a wide range of parameters. This was in general accord with our overall hypothesis of a trade-off between efficient gas exchange and ion/water balance at the gills that is flexible and dependent on salinity and oxygen levels. Our results show that both FW- and SW-acclimated killifish are similar to the Amazonian oscar model of response to hypoxia, inasmuch as active and passive components of gill Na+ transport were reduced, along with a decrease in water flux rates, pointing to an overall reduction in gill transcellular and paracellular permeability, particularly in FW, so that ionoregulatory homeostasis can be decoupled from the needs of respiratory gas exchange. In contrast, 11 ppt-acclimated fish increased both Na+ influx and efflux rates in hypoxia, as well as paracellular permeability, responses that follow the predictions for the classical osmorespiratory compromise. As, hypothetically, osmotic gradients are minimal at the isosmotic point, this increase in gill permeability should not be of great physiological concern. These results complement our previous study (Giacomin et al., 2019), which demonstrated that the greatest respiratory challenges for killifish during hypoxia occur in FW, and the least at the isosmotic salinity (11 ppt). Future studies should examine how these adjustments in gill permeability are achieved at cellular and molecular levels.
We thank Derrick Horne, Garnet Martens and Miki Fujita of the UBC Bioimaging Facility for skillful assistance with sample preparation and SEM imaging, and Victoria Matey for advice on image analysis. We would also like to thank both reviewers, whose comments have greatly improved our manuscript.
Conceptualization: M.G., P.M.S., C.M.W.; Methodology: M.G., J.O.O.; Formal analysis: M.G., J.O.O.; Investigation: M.G., J.O.O.; Resources: C.M.W.; Writing - original draft: M.G.; Writing - review & editing: M.G., J.O.O., P.M.S., C.M.W.; Visualization: P.M.S., C.M.W.; Supervision: P.M.S., C.M.W.; Project administration: C.M.W.; Funding acquisition: P.M.S., C.M.W.
This work was supported by Natural Sciences and Engineering Research Council of Canada (NSERC) Discovery Grants to C.M.W. (RGPIN-2017-03843, RGPIN/473-2012) and P.M.S. (RGPIN-203189). M.G. was supported by a 4 year graduate fellowship from the University of British Columbia.
The authors declare no competing or financial interests.