Effects of temperature on acid–base regulation, gill ventilation and air breathing in the clown knifefish, Chitala ornata

The vast majority of ectothermic vertebrates reduce arterial pH (pHa) as body temperature increases (Amin-Naves et al., 2007; Damsgaard et al., 2018; Fobian et al., 2014; Heisler, 1984; Rahn et al., 1971; Reeves, 1977; Smatresk and Cameron, 1982; Thinh et al., 2018; Truchot, 1987; Ultsch and Jackson, 1996; Wang and Jackson, 2016). Aquatic water-breathing vertebrates are constrained to achieve pHa adjustments through modulation of plasma HCO₃⁻ concentration ([HCO₃⁻]) because their low arterial P_CO2 (Pa_CO2) demands large ventilatory changes to adjust pHa, potentially compromising oxygen uptake given the low oxygen solubility in water (Bayley et al., 2019; Dejours, 1981; Rahn, 1966). In contrast, terrestrial vertebrates with their higher Pa_CO2 (Rahn, 1966) can rapidly regulate blood pH across temperatures through small changes in pulmonary ventilation (Glass and Wood, 1983; Jackson, 1989). This has the additional advantage of avoiding the osmotic changes that can ensue from the epithelial ion exchange adopted by aquatic animals (Austin et al., 1927; Burton, 2002). The transition from water to air breathing was also accompanied by a shift from primarily regulating arterial oxygen levels in water breathers by virtue of peripheral receptors to central monitoring of CO₂ and pH in air breathers (Bayley et al., 2019; Milsom, 2002, 2010; Milsom and Burleson, 2007).

There are approximately 400 species of bimodally breathing fish, which are extant evidence of the vertebrate transition from aquatic to terrestrial life, and air breathing has independently arisen on >80 occasions throughout the vertebrate phylogeny (Damsgaard et al., 2019). Air-breathing species give fascinating insight into the transitional states and are divided into obligate air breathers, which drown if prevented access to air, and facultative air breathers, which thrive in normoxic water without air breathing. Pa_CO2 in these two groups reflects their breathing habits (Rahn, 1966), with Pa_CO2 of facultative fish being similar to that of water breathers, whereas the higher Pa_CO2 of obligate air-breathing fish resembles that of tetrapods (Bayley et al., 2019). It might therefore be hypothesized that the modality used to adjust pHa across environmental temperature should follow the animal's partitioning pattern for gas exchange across the air–water interface. Temperature itself influences partitioning, with an increased reliance on the air phase as temperature increases associated with a concomitant elevation of Pa_CO2 in air-breathing garfishes (Lepisosteus oseus and Lepisosteus oculatus) (Rahn et al., 1971; Smatresk and Cameron, 1982). Furthermore, the facultative air-breathing striped catfish (Pangasionodon hypophthalmus) shows the aquatic vertebrate-type temperature adjustment by modulating plasma [HCO₃⁻ ] when in normoxic water (Damsgaard et al., 2018), but modulates Pa_CO2 across temperature in hypoxic water, where this species relies on aerial gas exchange. Finally, the obligate swamp eel (Monopterus albus) with vestigial gills and a tetrapod-like Pa_CO2 seems to regulate pHa by ventilatory adjustments while plasma [HCO₃⁻ ] is unaffected by temperature (Thinh et al., 2018).

The air-breathing clown knifefish (Chitala ornata Gray 1831) is commercially grown across South-East Asia, where it is favoured both as an ornamental fish and as a source of protein for human consumption. It regulates pHa when confronted with aquatic hypercapnia (Gam et al., 2018) – something that has been considered unusual in air-breathing fish, which typically show a reduced gill surface area, argued to limit transepithelial ion exchange and therefore branchial capacity for extracellular acid–base regulation (Brauner and Baker, 2009; Shartau and Brauner, 2014). Further, it has recently found that, unique among teleosts, it possesses a strong ventilatory response to changes in Pa_CO2 and is thus argued to have internally orientated CO₂/H⁺ sensors (Tuong et al., 2018a; Tuong et al., 2019). However, it is a facultative air breather (Tuong et al., 2018b) with a Pa_CO2 of 5–8 mmHg in normoxic normocapnic water (Gam et al., 2018). These aspects make it an interesting object to study the regulation of pHa with temperature changes, something that has also become timely with global warming scenarios, where median temperature in its habitat is expected to increase from the current 27–29°C (Li et al., 2013) to as much as 33°C within the next century (IPCC, 2014).

Here, we tested the hypothesis that C. ornata, with its ventilatory response to Pa_CO2, follows the pattern of acid–base regulation across temperature seen in terrestrial amniotes, with constant [HCO₃⁻ ] and modulated Pa_CO2. We assessed this by measuring arterial blood gases and relevant plasma ions in chronically catheterized animals across relevant environmental temperatures as well as measuring gill- and air-breathing ventilation rates from 24 to 36°C. Further, we (Thinh et al., 2018) previously proposed that a way of confirming the deliberate temperature-induced regulation of plasma pH would be to expose the animal to environmental hypercapnia at the temperature extremes and determine whether the animal defended the temperature-specific pH set-point. Therefore, we also exposed animals acclimated to 25 and 33°C to aquatic hypercapnia (21 mmHg CO₂) and followed acid–base status over the subsequent 72 h.

Clown knifefish (C. ornate, mean±s.e.m. body mass 503±55 g) were purchased from a local aquaculture farm, transported to Can Tho University and held in 4 m³ tanks at 27°C with constant aeration (dissolved oxygen >90%, P_CO2<0.7 mmHg and pH 7.75–7.85) for 2 weeks prior to experimentation. Fish were fed commercial shrimp feed with 38% protein (Tomboy Aquafeed Company, Vietnam). Water was changed (30%) every second day and total ammonia and nitrite were always below 40 µmol l⁻¹ and 1 µmol l⁻¹, respectively. Fish were fasted for 24 h prior to cannulation and throughout experimentation. All experiments were conducted in accordance with national guidelines on animal welfare in Vietnam.

Fish were anaesthetized in 0.05 g l⁻¹ benzocaine and moved to a surgical table where the gills were irrigated with water containing 0.025 g l⁻¹ benzocaine while a polyethylene PE40 catheter was inserted into the dorsal aorta through the dorsal surface of the buccal cavity (Soivio et al., 1975). The catheter was filled with heparinized saline (50 IU per 1 ml saline) and fixed with several stitches on the mouth, nose and near the dorsal fin. Fish were allowed to recover in well-aerated water for 24 h at 27°C before experimentation to normalize blood gasses and lactate levels (Phuong et al., 2017a).

Additional fish were anaesthetized as described above for attachment of impedance leads to the outer surface of both the left and right opercula and secured in place with nylon suture. This procedure lasted a maximum of 10 min. Fish were allowed to recover in well-aerated 27°C water for 24 h prior to experimentation, at which time the leads were connected to the impedance converter via an acquisition system (Dataq DI-720, DATAQ Instruments, Inc., Akron, OH, USA) recording at 125 Hz for ventilatory measurements.

Four temperature levels were tested in the experiment (25, 30, 33 and 36°C), each replicated 6 times. In addition, as the entire procedure was identical to one used previously (Gam et al., 2018), we included the dataset for 27°C from that study in the present analysis. Following surgical recovery at the standard laboratory temperature at Can Tho of 27°C, each replicate was moved to separate 120 l tanks containing water circulated from a central 500 l tank. Temperature was elevated (or decreased) at 0.5°C h⁻¹ to the target temperature and left for 48 h. Thereafter, blood was drawn at 0, 24, 48 and 72 h from the dorsal aorta catheter into heparinized syringes. Acid–base parameters [pHa, Pa_CO2, total CO₂ of plasma (TCO₂)] were measured immediately. These values were used to calculate the temperature-specific acid dissociation constant (pK′). In the same blood samples, haemoglobin concentration ([Hb]) and haematocrit (Hct) were measured. Remaining blood was centrifuged and the plasma stored at −80°C for subsequent analysis of ions (Cl⁻, Na⁺, K⁺) and osmolality.

After they had recovered from the insertion of impedance electrodes, fish were moved individually to the temperature system described above. Water temperature was progressively increased as in series 1. After 48 h at the target temperature, fish were moved to a 50 l glass tank containing fresh target temperature water and left for 1 h to reach resting levels; gill ventilation and air-breathing frequency were then measured for 60 min.

A total of 24 fish were used. After recovery from cannulation, fish were transferred to the system described in series 1. An Oxyguard Pacific system coupled with a G10 ps CO₂ probe and a K01 svpld pH probe (Oxyguard International A/S, Farum, Denmark) and a solenoid valve was used to regulate water P_CO2 at 21 mmHg. In the normocapnic treatment, P_CO2 was kept below 0.7 mmHg, the lower detection limit of the system, by vigorous aeration. This experiment was performed at both 25 and 33°C. Temperature was adjusted as in series 1 and blood subsequently sampled at 0, 24, 48 and 72 h for measurement of the same blood parameters.

The non-bicarbonate buffer line and temperature-specific pK′ values were calculated from data from series 1. pHa and Pa_CO2 were measured using an iSTAT analyser (iSTAT Corporation, Princeton, NJ, USA) with CG3+ cartridges and temperature corrected using the iSTAT manual equation as validated previously (Gam et al., 2018). TCO₂ was measured using a Cameron chamber, where a blood aliquot is added to a known acid volume and the resulting elevation in P_CO2 reflects the total CO₂ content of the sample (Cameron, 1971). [HCO₃⁻] was calculated using the equation: [HCO₃⁻]=TCO₂−Pa_CO2×αCO₂, where αCO₂ is CO₂ solubility in trout plasma (Boutilier et al., 1984).

Plasma Cl⁻ was measured using a chloride titrator (Sherwood, model 926S MK II Chloride analyser, Sherwood Science Ltd, Cambridge, UK) and plasma Na⁺ and K⁺ with flame photometry (Sherwood model 420). Total plasma osmolality was measured on a Fiske One-Ten Micro Osmometer (Fiske*Associates, Two Technology Way, Norwood, MA, USA).

Hct was measured as a fractional volume of packed red blood cells after centrifugation at 12,000 g for 3 min. [Hb] was spectrophotometrically measured at 540 nm after conversion to cyano-methaemoglobin using Drabkin's solution (Zijlstra et al., 1983). Mean corpuscular haemoglobin concentration (MCHC) was calculated from the ratio between [Hb] and Hct.

All figures were made in Sigma plot 12.5 and data were analysed with PASW statistics (SPSS18.0). A one-way ANOVA was used to evaluate the effects of temperature on arterial acid–base parameters (series 1). A two-way ANOVA (the Holm–Šidák multiple comparison method, pair-wise comparison) was used to identify differences between sampling time and treatments to evaluate the effects of temperature on gill ventilation and air-breathing frequency and the effects of temperature and hypercapnia on pHa regulation (series 2 and 3). A P-value of less than 5% (P<0.05) was considered significant. All data are shown as means±s.e.m.

The pattern of pHa adjustment across environmental temperature shown by C. ornata resembles that of terrestrial vertebrates with large changes in Pa_CO2, but without the typical piscine adjustments in plasma [HCO₃⁻] (Fig. 1). The slope of the pHa change of −0.024 pH units °C⁻¹ is higher than that in most other vertebrates, but by no means unique. The air-breathing rice eel, M. albus, showed a similar slope of −0.025 pH units °C⁻¹ (Thinh et al., 2018) and examination of values across vertebrates reveals several other similar examples (Ultsch and Jackson, 1996). Plasma [Na⁺], [K⁺], [Cl⁻] and osmolality remained largely unaffected across temperature (Table S1). A slight but significant reduction in [Hb] and Hct was seen with temperature elevation, due probably to repeated blood sampling, while MCHC remained unchanged (Table S2). A similar pattern with elevated Pa_CO2 has been observed in two other air-breathing teleosts, P. hypophthalmus and M. albus (Damsgaard et al., 2018; Thinh et al., 2018). Both P. hypophthalmus and C. ornata are facultative air breathers, relying predominantly on their gills for oxygen uptake in normoxic water (Lefevre et al., 2011a, 2013; Tuong et al., 2018b; Ultsch and Jackson, 1996). While P. hypophthalmus only showed this pattern in hypoxic water, resembling a classic water breather in normoxia, the obligate air breather M. albus modulated Pa_CO2 even in normoxic water (Thinh et al., 2018).

Changing the respiratory medium significantly impacts blood gases and a shift to air breathing brings about a rise in blood Pa_CO2 through the lowering of ventilation rates when fish transition to air breathing (Rahn et al., 1971). This elevated Pa_CO2 allows for ventilatory regulation of plasma pH, because at high Pa_CO2, small changes in ventilation give rise to large changes in pH, whereas at the low Pa_CO2 levels of water breathers, elevation of Pa_CO2 requires significant ventilatory changes that are likely to impact oxygen uptake (Bayley et al., 2019). Indeed, there is one water-breathing fish that illustrates this dilemma (Piaractus mesopotamicus) because it regulates arterial pH across temperature by increasing Pa_CO2 at the expense of arterial P_O2 (Pa_O2) (Soncini and Glass, 1997). The authors argue for this species that the drop in Pa_O2 is of little consequence, because a right shift in the haemoglobin–oxygen binding curve facilitates tissue unloading and the reduced Pa_O2 remains on the upper flat part of the blood equilibrium curve, hence only marginally affecting saturation. Air-breathing fish in general show Pa_CO2 values that clearly reflect the partitioning of CO₂ excretion, with those predominantly using their gills showing levels similar to those of pure water breathers, while those reliant on air breathing showing Pa_CO2 levels between 14 and 28 mmHg (Bayley et al., 2019; Thinh et al., 2018). Hence, the ventilatory adjustments required to alter pHa in response to temperature in the obligate M. albus (Thinh et al., 2018) are relatively minor. Pangasionodon hypophthalmus, in contrast, is a facultative air breather with little or no aerial gas exchange in normoxic water (Lefevre et al., 2011a, 2013), low Pa_CO2 levels (Damsgaard et al., 2015) and no apparent ventilatory response to even very severe aquatic hypercapnia (Thomsen et al., 2017). Temperature-induced adjustments of pHa in normoxic water thus follow the standard aquatic organism pattern, with transmembrane ion exchange and little change in Pa_CO2, whereas in hypoxia, Pa_CO2 is elevated as a result of increased air breathing, leaving only minor adjustments required through branchial HCO₃⁻/Cl⁻ exchange (Damsgaard et al., 2018).

In C. ornata, increased temperature stimulated both air breathing and gill ventilation (Fig. 2), but the Q₁₀ for gill ventilation was considerably lower than that for air breathing (1.4 and 3.8, respectively), indicating a progressive transition from aquatic to aerial gas exchange as temperature rose. While C. ornata is a facultative air breather, a mere 86% of its oxygen is covered by uptake over its small branchial surfaces in normoxic water at 27°C (Tuong et al., 2018b), whereas P. hypophthalmus derives more than 95% of its oxygen uptake across the very large gills (Phuong et al., 2017b) in the same conditions (Lefevre et al., 2011a). This probably explains why C. ornata has a higher Pa_CO2 (7 mmHg, at 27°C; Fig. 1C) than P. hypophthalmus (3 mmHg) at the same temperature (Damsgaard et al., 2018). Chitala ornata modulates air breathing, but not branchial ventilation, in response to changes in Pa_CO2 (Tuong et al., 2019), and the persistence of this ventilatory response following branchial denervation indicates internal orientation of the sensors (Tuong et al., 2019). While in a water-breathing fish, the required elevation in Pa_CO2 with elevated temperature from its low starting level requires a severe depression of ventilation relative to metabolism and hence reduced oxygen uptake (see Soncini and Glass, 1997), a bimodal breather can achieve the necessary increase in Pa_CO2 without compromising oxygen uptake by simply changing its partitioning towards the air phase. Interestingly, this is exactly what is observed in two species of gar (Rahn et al., 1971; Smatresk and Cameron, 1982), and in toads, which show a similar change in the partitioning from cutaneous to pulmonary gas exchange as temperature increases, with a similar attendant rise in Pa_CO2 (Wang et al., 1998). We therefore interpret both the temperature-associated elevation in Pa_CO2 and the accompanying change in ventilatory frequency in the present study as reflecting altered gas exchange partitioning for both oxygen and CO₂.

The exposure to 21 mmHg of aquatic hypercapnia at 25 and 33°C was designed to confirm the hypothesis that C. ornata defends the correct temperature-dependent pHa set point. To this end, the experiment was only partially successful. Hypercapnia enforced a respiratory acidosis at both temperatures (25 and 33°C; Fig. 3), and the resulting pHa disturbance measured at 24 h was greater at 25 than at 33°C (Fig. 3C,D), reflecting the lower initial Pa_CO2 at the lower temperature (Fig. 3E,F). The recovery of pHa was more complete after 72 h at 33°C (not significantly different from controls) than at 25°C (Fig. 3A–D). Further, the regulatory rise in [HCO₃⁻] at the two temperatures revealed similar responses (Fig. 3G,H), and did not appear to be slowed at the termination of the experiment. At 72 h, the increase in plasma [HCO₃⁻] was accompanied by a greater than expected decrease in plasma [Cl⁻], and an additional drop in plasma [Na⁺] and osmolality (Table S3). The change in [Na⁺] is opposite to that expected from Na⁺/H⁺ exchange, but probably reflects the osmolality change. Osmolality was reduced by ca. 10 mOsm, indicating a drop of ca. 5 mmol l⁻¹ for both [Na⁺] and [Cl⁻]. [Cl⁻] drops by 10 mmol l⁻¹, half of which is explained by Cl⁻/HCO₃⁻ exchange and the other half by the change in osmolality. A decrease in osmolality with aquatic hypercapnia has been observed previously in this species (Gam et al., 2018) but presently remains unexplained. The high temperature group requires less HCO₃⁻ for full pH compensation than the low temperature group, and in our view more time would probably increase [HCO₃⁻] and pH compensation further. A slow pH compensation is seen in other freshwater fish, and the tench (Tinca tinca) takes up to 2 weeks to elevate plasma HCO₃⁻ to the almost 40 mmol l⁻¹ required for complete pH compensation at 10 mmHg CO₂ (Jensen et al., 1993). Furthermore, the water concentrations of calcium, bicarbonate and chloride are important in determining both the speed and completeness of acid–base regulation during hypercapnia in fish (Larsen and Jensen, 1997). In the present case, the water levels of these ions were moderate [e.g. water [Cl⁻] ∼0.3 mmol l⁻¹ (Lefevre et al., 2011b); [Na⁺] ∼236±12 µmol l⁻¹, [K⁺] ∼34±1; and [Ca²⁺] ∼342±3 µmol l⁻¹ (Yanagitsuru et al., 2019)], pointing to a somewhat delayed acid–base regulation.

In summary, we have demonstrated that this facultative air-breathing fish, which at low temperatures shows a Pa_CO2 level only slightly elevated above that of water breathers, reduces pHa with increasing temperature by elevating Pa_CO2 under constant [HCO₃⁻]. We argue that this is a reflection of a regulated shift in the partitioning of gas exchange, which is probably common among bimodal breathers. Bayley et al. (2019) proposed that the elevated Pa_CO2 is a pre-requisite for the utility of respiratory pH regulation because it radically reduces the problem of hypoventilation forming a constraint on oxygen uptake. These findings suggest that air breathing, and hence the partitioning of gas exchange, is to some extent regulated by acid–base status in air-breathing fish and suggest that these bimodal breathers will be increasingly likely to adopt respiratory pH control as temperature increases: an interesting avenue for future research.