Mechanisms of acid–base regulation in the African lungfish Protopterus annectens

The regulation of metabolic acid–base disturbances can occur through the control of ventilation and/or the direct transfer of acid–base equivalents between the animal and its external environment. Increased pulmonary ventilation during a metabolic acidosis reduces arterial P_CO2, thereby raising arterial pH, while CO₂retained via hypoventilation can compensate for a metabolic alkalosis. Such respiratory compensation is a mainstay of the strategy used by tetrapods to regulate acid–base disturbances(Heisler, 1986; Swenson, 2000; McNamara and Worthley, 2001). In water-breathing fish, on the other hand, respiratory compensation is of negligible importance and acid–base balance is restored metabolically(for reviews, see Goss et al.,1998; Claiborne et al.,2002; Hirose et al.,2003; Perry et al.,2003b; Evans et al.,2005; Perry and Gilmour,2006). A systemic acidosis is corrected by the increased output of acidic equivalents (thereby raising the plasma HCO ^–₃ concentration), whereas the response to alkalosis consists of decreased net efflux of acid (equivalent to increased base efflux). In fish, the gill accounts for 90% or more of such direct transfer of acid–base equivalents(Claiborne et al., 2002). Adjustment of renal acid secretion plays an important complementary role to the gills (Wood et al., 1999; Georgalis et al., 2006a) (for reviews, see Perry and Fryer,1997; Perry et al.,2003b; Perry and Gilmour,2006). Tetrapods also employ metabolic mechanisms to compensate for acid–base disturbances, with the kidney being the main regulatory site (Heisler, 1986; Swenson, 2000; McNamara and Worthley, 2001). Aquatic amphibians, however, parallel fish in utilizing extra-renal transfer of acid–base equivalents, in their case, via the skin(Stiffler and Bachoura, 1991; Stiffler, 1991; Talbot and Stiffler, 1992). Regardless of the tissue site, work on a variety of organisms suggests that similar cellular and molecular mechanisms of acid–base equivalent transfer are employed (for reviews, see Claiborne et al., 2002; Kirschner, 2004; Wall, 2005; Perry and Gilmour, 2006).

With this background in mind, the present study focused on characterizing strategies for acid–base regulation in the African lungfish Protopterus annectens. Mechanisms of acid–base compensation in lungfish have received little attention to date(Sanchez et al., 2005), but several factors argue compellingly for their investigation. African lungfish(Protopterus sp.) are obligate air breathers that possess true lungs and cannot survive without access to air, but gills are also present,permitting bimodal breathing (reviewed by Burggren and Johansen, 1986). The lung is the primary site of O₂ uptake in all four Protopterus species, but the majority of CO₂ excretion occurs across the gill (and/or the skin)(Lenfant and Johansen, 1968; Lahiri et al., 1970; McMahon, 1970), except in P. dolloi, where the lung appears to be a major route of both O₂ and CO₂ transfer(Perry et al., 2005). Because most O₂ uptake occurs from the air, branchial ventilation in lungfish need not be constrained by convection requirements for O₂uptake. Consequently, the gill is ventilated by low volumes of water(Jesse et al., 1967) and this factor, coupled with the low surface area and high blood-to-water diffusion distances of the lungfish gill (Sturla et al., 2001), results in arterial P_CO2 values that are unusually high for fish (∼20–30 mmHg)(Lenfant and Johansen, 1968; Perry et al., 2005). Thus, the respiratory status of lungfish is more suitable for using ventilation to regulate acid–base disturbances than is the case for water-breathing fish. At the same time, however, the gills, although reduced in surface area relative to those of water-breathing fish, are covered by a heterogeneous epithelium containing several cell types, including pavement cells,mitochondria-rich (MR) cells, and sensory cells(Sturla et al., 2001). MR cells are thought to be responsible for the excretion of acid–base equivalents by the gills of freshwater fish (for a review, see Perry and Gilmour, 2006), and the presence of MR cells therefore implies the potential for the lungfish gill to function in metabolic acid–base compensation.

Two hypotheses were tested in the present study. First, the hypothesis that ventilatory adjustments are used by P. annectens to regulate acid–base disturbances was tested. It was predicted that lungfish would increase pulmonary and/or branchial ventilation in response to metabolic acidosis, but hypoventilate under conditions of metabolic alkalosis. Second,the transfer of acid–base equivalents across the gill (and/or skin) was hypothesized to play a significant role in recovery from acid–base disturbances. Under these conditions, net acid excretion to the water would be predicted to increase during metabolic acidosis and decrease during metabolic alkalosis. Finally, the molecular mechanisms underlying the transfer of acid–base equivalents at the gill and kidney were investigated by cloning fragments of lungfish V-type H⁺-ATPase, Na⁺/HCO ^–₃ cotransporter (NBC) and carbonic anhydrase(CA), and examining the mRNA expression of these genes in the gill and kidney prior to and following acid–base disturbances.

Adult specimens of either sex of the African lungfish Protopterus annectens Owen (143±6 g; mean mass ± s.e.m.; N=46)were imported from central Africa to Singapore through a local fish farm. Lungfish were then shipped to Ottawa by air cargo. During transit, lungfish were placed in partially filled bags of oxygenated water held in insulated containers and upon arrival, the fish were transferred into individual covered plastic aquaria filled with 2–3 l of dechloraminated city of Ottawa tapwater warmed to 25°C. The tank water was changed on alternate days or sooner if obvious fouling of the water occurred. Lungfish were held in a room in which the air temperature was maintained at 25°C, under an artificial photoperiod of 10 h:14 h light:dark. Fish were fed on alternate days with frozen blood worms or pieces of rainbow trout flesh, and were allowed to acclimate to these conditions for at least 1 month prior to experimentation.

To allow periodic blood sampling as well as acid or base infusion, a cannula (Clay-Adams PE50 polyethylene tubing; VWR, Montreal, QC, Canada) was inserted into the dorsal aorta of all fish, as described by Perry et al.(Perry et al., 2005). Surgery was carried out on lungfish that were first anaesthetized by immersion in a solution of MS-222 (ethyl-p-aminobenzoate; 0.67 g l^–1) adjusted to neutral pH with NaHCO₃ (1.3 g l^–1); anaesthesia was maintained during surgery by wrapping the fish in paper towels soaked with the anaesthetic solution. A subset of lungfish (N=7) was fitted with an external urinary catheter according to the procedure of Curtis and Wood(Curtis and Wood, 1991). The use of an external urinary catheter allows the collection of urine as it is naturally discharged from the urogenital papilla. Following surgery, lungfish were returned to their containers for a 24 h recovery period. Cannulae were flushed daily with Cortland saline (Wolf,1963).

Fish (N=19) that previously had been fitted with a dorsal aortic cannula were placed into customized respirometry chambers approximately 2 h before an experiment was initiated. The cylindrical chambers were filled with water (∼1 l) apart from an adjustable air space (60 ml maximum volume;typically set to ∼30 ml) at one end of the respirometer; exact water and air volumes were noted for each fish. Lungfish quickly became aware of the presence of the air space, and normally would begin breathing air at regular intervals, adopting a position in the respirometer where their head was just below the airspace. The occurrence of each air breath was captured by a custom-built device that detected interruptions in two infrared light beams aimed across the surface of the water. Both compartments were provided with continually flowing media (water or humidified air, at 25°C) except during measurements of the rates of O₂ consumption(Ṁ_O2),CO₂ excretion(Ṁ_CO2), and net acid excretion into the water. During these measurements, the water was re-circulated using a peristaltic pump, and the air chamber was sealed. Fibre-optic O₂ electrodes (Ocean Optics AL300, Dunedin, FL, USA)sealed into the air chamber and into the tubing through which water was re-circulating were used to measure aerial and aquatic P_O2, respectively. Aerial P_CO2 was measured using a CO₂ electrode (Analytical Sensors E201, Sugarland,TW, USA) inserted into the air chamber, while measurement of aquatic P_CO2 involved using a small peristaltic pump to move water past a CO₂ electrode (Analytical Sensors E201) housed within a thermostatted (25°C) cuvette (Radiometer) before returning it to the respirometer. These respirometers were used in a previous study(Perry et al., 2005) during which the extent of gas transfer across the air–water interface was quantified and found to be negligible.

Once fish were breathing air at regular intervals and had been doing so for at least 1 h, baseline (`pre') measurements of Ṁ_O2, Ṁ_CO2, ventilation frequency, net acid excretion and blood acid–base status were carried out over the course of 1 h. The water and air chambers were sealed after stopping the flow of media, and the re-circulating pump was started to provide mixing. Changes in aerial P_O2 and P_CO2 were monitored for two 30-min periods, between which the air within the chamber was refreshed. Changes in aquatic P_O2 and P_CO2 were monitored for approximately 30 min or until stable rates of CO₂ accumulation and O₂ depletion were achieved. In addition to the automatic detection of air breaths,water-breathing frequency was determined by visually counting water breaths at periodic intervals. To measure net acid excretion, 20 ml water samples were withdrawn at the beginning and end of the 1–h flux period. A 0.5 ml blood sample was withdrawn at the end of the measurement period. The blood sample was centrifuged and plasma was removed for immediate analysis of pH and total CO₂ content (CCO₂). The remaining red blood cells were re-suspended in heparinized (50 i.u. ml^–1 ammonium heparin; Sigma, Oakville, ON, Canada) Cortland saline and re-injected into the fish. At the end of this measurement period, the respirometer was returned to a flow-through condition and acid or base infusion was initiated.

Lungfish were infused with acid (0.9 mol l^–1NH₄Cl; N=11) or base (0.9 mol l^–1NaHCO₃; N=8) at a rate of 3 mmol kg^–1h^–1 for 1 h. A syringe pump (SAGE instruments 355) was used to infuse acid or base via the dorsal aortic cannula. Once acid or base infusion had been initiated, the water and air chambers were sealed and the series of measurements described above was repeated, culminating with the withdrawal of a blood sample immediately following the infusion period. Blood samples were also withdrawn at 3, 6 and 18 h post-infusion, and were preceded in each case by the same series of measurements of Ṁ_O2, Ṁ_CO2, ventilation frequency and net acid excretion. After withdrawal of the 18 h blood sample,lungfish were anaesthetized in situ by intra-arterial injection of 400 mg kg^–1 of sodium pentobarbital (Somnotol), and then sacrificed by an intra-arterial injection of saturated KCl. A final measurement period for Ṁ_O2, Ṁ_CO2 and net acid excretion was carried out over 45 min to estimate background levels of gas transfer and net acid excretion. These background values were assumed to reflect cutaneous metabolism (which they may underestimate) and/or metabolism arising from microorganisms in the water or on the surface of the fish(Perry et al., 2005). For each fish, the values for Ṁ_O2, Ṁ_CO2 and net acid excretion were corrected for background metabolism by subtracting the rates determined in this final respirometry/flux period.

Lungfish (N=7) were placed into the customized respirometry chambers and experiments were not initiated until at least 1 h of breathing air at regular intervals had occurred. Fish used in these experiments were fitted with both a dorsal aortic cannula (for base infusion; blood samples were not collected in these experiments) and an external urinary catheter. Urine was collected continuously throughout the experimental period by allowing the urinary catheter to drain, by gravity, into a vial located outside the respirometer and held approximately 5 cm below water level. The urinary catheter was checked for leaks by raising the catheter 5 cm above water level. Under these conditions, a fall of the urine level in the catheter indicated a leak, and the experiment was terminated. First, a flux period was carried out to determine baseline (`pre') levels of net acid excretion into water and urine, as well as urine flow rate (UFR), pH and CCO₂. Water flow to the respirometer was halted, the chamber was sealed and the re-circulating pump was started to provide mixing. Water samples (20 ml) were collected at the beginning and at 3 h of the 4 h flux period. To ensure that the dead space volume within the urinary catheter was cleared, urine collected during the initial ∼60 min of the flux period was discarded. At the end of the flux period, water flow to the respirometer was re-established and base infusion was initiated. Base (0.9 mol l^–1 NaHCO₃)was infused into the dorsal aortic cannula, using a syringe pump, at a rate of 3 mmol kg^–1 h^–1 for 1 h. Flux periods were carried out over the initial 4 h following base infusion, and from 17–21 h post base infusion.

Following recovery from surgery, lungfish fitted with a dorsal aortic cannula were infused intra-arterially with saline (Cortland saline, 0.5 ml h^–1; N=3), acid (0.9 mol l^–1NH₄Cl; N=8) or base (0.9 mol l^–1NaHCO₃; N=7) at a rate of 3 mmol kg^–1h^–1 for 1 h using a syringe pump. Lungfish were sacrificed 4 h (N=8) or 8 h (N=7) after the acid or base infusion period,and gill and kidney tissue samples were collected, flash frozen in liquid N₂ and stored at –80°C for later analysis of gene expression. Saline-infused fish were sampled 8 h after the infusion period to provide control data.

For respirometry, P_CO2 electrodes were connected to a blood gas analyzer (Cameron Instruments BGM200, Port Aransas, TX, USA) that was customized to accept two CO₂ inputs. Output from the blood gas analyzer and the infrared air-breath detector was converted to digital data and stored by interfacing with a data acquisition system (Biopac Systems Inc.,Harvard Apparatus Canada, Saint-Laurent, QC, Canada) using Acknowledge™data acquisition software (sampling rate set at 30 Hz) and a PC. Output from the fibre-optic O₂ electrodes was collected using Ocean Optics software running on the same PC. These data were compiled as text files for later importation into spreadsheet software for storage and analysis. To calibrate the fibre-optic O₂ electrode used for water P_O2 measurements, the electrode was immersed in zero solution (2 g l^–1 sodium sulphite) or air-saturated water until stable readings were recorded. The fibre-optic O₂ electrode used for air P_O2 measurements was calibrated in situ in the respirometer by flowing humidified N₂ gas (zero)or air continuously through the air chamber. Both CO₂ electrodes were calibrated in situ using mixtures of 0.5% and 1.0%CO₂ in air that were provided by a gas mixing flowmeter (Cameron Instruments GF-3/MP). Humidified gas mixtures flowing through the air chamber of the respirometer were used to calibrate the air P_CO2 electrode, whereas water equilibrated with the gas mixtures and pumped through the cuvette was used to calibrate the CO₂ electrode used for water P_CO2 measurements.

Rates of aerial gas transfer were determined from the slopes of the relationships between inspired gas tensions and time, over the period that the air chamber was sealed, taking into account air chamber volume, fish mass, and the solubility coefficients of O₂ and CO₂ in air at 25°C (Boutilier et al.,1984). Similarly, aquatic gas transfer was calculated using the slopes of the relationships between water gas tensions and time over the interval that the water in the respirometer was re-circulated, taking into account water chamber volume, and fish mass. For O₂, the solubility coefficient in fresh water at 25°C was obtained from Boutilier et al.(Boutilier et al., 1984). For CO₂, the capacitance coefficient in dechloraminated city of Ottawa tap water at 25°C was determined experimentally to be 0.041 μmol l^–1 mmHg^–1 by measuring the total CO₂ concentrations (Cameron Instruments Capni-Con 5) of water samples equilibrated to the range of P_CO2 values encountered in respirometry trials.

Blood samples were centrifuged (∼10 000 g for 1 min)immediately following withdrawal to yield plasma. Plasma total CO₂concentration was determined in duplicate on 50 μl samples (Cameron Instruments Capni-Con 5). Plasma pH was measured using a pH electrode and calomel reference (Analytical Sensors E301 glass pH electrode) that were housed in a temperature-controlled, low-volume pH chamber (Cameron Instruments) and connected to a PHM 72 acid–base analyzer (Radiometer). The arterial blood P_CO2 (Pa_CO2) and bicarbonate concentration ([HCO ^–₃]) were then calculated from the Henderson–Hasselbalch equation using appropriate values for αCO₂ and pK′(Boutilier et al., 1984).

Net acid excretion (J_netH⁺) was determined from measurements of titratable net acid flux (J_netTA) and the change in ammonia concentration in the water samples collected at the beginning and end of a flux period. Water J_netTA and ammonia concentration were assessed within 24 h of sample collection and the remainder of the water sample was frozen for later analysis of ion concentrations. J_netTA was determined by titrating (using a Gilmont precision microburet) 5 ml water samples from the beginning and end of each flux period to pH 4.00 with 0.02 mol l^–1 HCl. Samples were continuously aerated prior to and during titration to ensure mixing and removal of CO₂. A micro-modification of the salicylate–hypochlorite colorimetric assay of Verdouw et al.(Verdouw et al., 1978) was used to measure total ammonia levels in water samples. J_netH⁺ was then calculated as the sum of J_netTA and the ammonia flux(J_netNH₃), signs considered, as described by McDonald and Wood (McDonald and Wood,1981).

Net renal acid excretion was also determined as the sum of titratable net acid flux and ammonia efflux (Wood and Caldwell, 1978). Urinary J_netTA was measured by lowering the pH of a 200 μl aliquot of urine below 5.0 through the addition of a known volume of 0.02 mol l^–1 HCl. The sample was then aerated for 20 min to remove CO₂. While continuing to aerate, the pH of the urine sample was titrated back to the pH of blood representative of the particular sampling period through the addition of 0.02 mol l^–1 NaOH using a precision microburet (Gilmont). The titratable component of net renal acid excretion is given by the difference in the quantities of acid and base added to the urine. Urine flow rates were determined gravimetrically. As with water samples, urine J_netTA and ammonia concentration were assessed within 24 h of sample collection together with urine pH and CCO₂. Urine ammonia concentration, pH and total CO₂ concentration were assessed using the procedures described above for water or blood. For the calculation of urine [HCO ^–₃], constants derived for freshwater were employed (Boutilier et al.,1984).

Tissues for the cloning of lungfish CA, NBC and H⁺ V-ATPase were collected from fish that were terminally anaesthetized by immersion in a solution of MS-222 (see above). Total RNA was isolated from a given tissue using Trizol (Invitrogen, Burlington, ON, Canada) according to the manufacturer's instructions. RNA quantity and quality were verified by spectrophotometry (Eppendorf BioPhotometer) and/or by inspection of an RNA gel. First strand cDNA was synthesized from RNA using Superscript reverse transcriptase (Invitrogen) and an oligo(dT) primer (Sigma). Gene-specific primers (Table 1) were then used to amplify cDNA fragments by PCR. PCR was performed using 1 μl of cDNA template in 25 μl reaction mixture containing 3.5 mmol l^–1MgCl₂, 200 μmol l^–1 of each dNTP, 250 nmol l^–1 each of forward and reverse primers, and 1 i.u. Taq polymerase (New England Biolabs, Toronto, ON, Canada) in PCR buffer supplied with the enzyme. All PCR reactions involved an initial denaturation at 94°C for 3 min followed by 39 cycles of 94°C for 30 s, annealing temperature for 30 s, 72°C for 60 s, and ending with a final extension for 10 min at 72°C. PCR products obtained in this manner were gel-purified,cloned into pCR2.0-TOPO vector (TOPO TA cloning kit, Invitrogen) and sequenced. The resulting sequences were compared against those in the GenBank database using BLASTX to identify the gene, and also to design primers (where needed) for 3′- and 5′-rapid amplification of cDNA ends (RACE) to extend the amplified sequence lengths. For 3′-RACE, cDNA was synthesized using a 3′-RACE adapter primer (Invitrogen) and Superscript II reverse transcriptase (Invitrogen). Semi-nested PCR was performed on the cDNA using abridged universal amplification primers (AUAP; Invitrogen) and gene-specific 3′-RACE primers (Table 1). For 5′-RACE, cDNA was synthesized using a gene-specific primer (Table 1) and Superscript II reverse transcriptase (Invitrogen), and then purified using a PCR purification kit (Sigma). The newly purified cDNA was tailed with dCTP using a recombinant terminal transferase TdT (Invitrogen) with final reaction conditions; 10 mmol l^–1 Tris–HCl (pH 8.4), 25 mmol l^–1 KCl, 1.5 mmol l^–1 MgCl₂, 200μmol l^–1 dCTP, 1 μl cDNA, and 1 μl rTdT. The tailed cDNA was then used for two rounds of PCR using a gene-specific forward primer(Table 1) and a 5′abridged anchor primer (5AP; Invitrogen) for the first round, and a semi-nested gene-specific forward primer(Table 1) and AUAP (Invitrogen)for the second round. PCR products were cloned into the pCR2.0-TOPO vector using TOPO TA cloning kits (Invitrogen). All RACE product sequences were confirmed by overlap with the initial fragment of cDNA. After repeated bi-directional sequencing of both RACE products, a consensus sequence was created by multiple sequence alignment using DNAMAN (v4.0, Lynnon Biosoft,Vaudreuil, QC, Canada).

Using red blood cell cDNA as template, a 315 base pair (bp) internal segment of a CA coding region (denoted lfCAb) was PCR amplified at an annealing temperature of 46°C using the primer pair CA(Table 1). These degenerate primers were designed based on alignments of vertebrate cytoplasmic CA isoforms exhibiting a high degree of amino acid conservation. The sequence was then extended by three rounds of 3′-RACE using the gene-specific primers CA3R1, CA3R2 and CA3R3.

A 665 bp internal segment of the coding region of NBC was PCR amplified at an annealing temperature of 57°C using cDNA derived from lungfish gill tissue and the primer pair NBC. These primers corresponded to the nucleotides at positions 1492–1512 and 2183–2203 of the tiger salamander(Ambystoma tigrinum) kidney NBC (GenBank accession no. AF001958)(Romero et al., 1997). The resultant clone, lfNBC, was deemed of sufficient length for the purposes of the present study and therefore no attempt was made to extend the sequence length using RACE.

Using lungfish gill tissue, an internal segment of the coding region of the B subunit of V-type H⁺-ATPase (denoted lfH⁺ V-ATPase)was PCR amplified using the primer pair HV1 at an annealing temperature of 50°C. These primers, which were used successfully by Perry et al.(Perry et al., 2000) to clone the rainbow trout V-type H⁺-ATPase B subunit, correspond to the nucleotides at positions 441–461 and 1395–1415 of the bovine kidney V-type H⁺-ATPase B subunit sequence (GenBank accession no. M88691) (Nelson et al., 1992). A second round of PCR was required to obtain an 810 bp fragment. This PCR round used the primer pair HV2 at an annealing temperature of 49°C. The new forward primer corresponded to the nucleotides at positions 606–626 of the bovine kidney H⁺ V-ATPase. The sequence length was then further extended by two rounds of 3′-RACE using gene-specific primers HV3R1 and HV3R2, and by 5′-RACE. For 5′-RACE, the gene-specific primer HV5R1 was used for cDNA synthesis. Two rounds of 5′-RACE were then carried out using gene-specific primers HV5R2 and HV5R3.

Lungfish deduced amino acid sequences were aligned with GenBank sequences with which high identity was exhibited (determined by BLASTX searching of GenBank databases) using ClustalX version 1.83(Thompson et al., 1997) with penalties for gap opening and gap extension set to 30 and 0.75, respectively,for pairwise alignments, and 15 and 0.3, respectively, for multiple alignments. Note that full-length sequences from other vertebrates were truncated appropriately so as to compare only regions overlapping with the lungfish sequences. The PHYLIP package was then used to carry out neighbour-joining phylogenetic analysis(Saitou and Nei, 1987) on a matrix of mean character distances, with a bootstrapping resampling option to assess the support for nodes (100 pseudoreplicates). In general, default parameters were used except that the outgroup was specified (as Drosophila) and the input order of species was randomized where possible. The accession numbers for sequences used in alignments or phylogenetic analysis are presented in the appropriate figure legends.

Real-time PCR was used to assess the expression of the various lungfish mRNAs in gill and kidney tissue under control conditions and following acid or base infusion. Total RNA was extracted using Trizol (Invitrogen) from tissues that were homogenized to powder under liquid N₂ with a mortar and pestle, and then passed through a needle using a syringe. The extracted total RNA was treated with amplification grade DNase I (RNase free; Invitrogen)according to the manufacturer's instructions for 15 min at room temperature,followed by 10 min at 70°C, to remove any remaining genomic DNA. The quality of the RNA was verified by spectrophotometry (Molecular Devices,Sunnyvale, CA, USA) and cDNA was then synthesized from 2 μg of RNA using random hexamer primers (IDT DNA) and RevertAid M-MuLV H-minus (Fermentas,Burlington, ON, Canada) according to the manufacturer's instructions. A Brilliant SYBR Green QPCR Master Mix kit (Stratagene, Cedarlane Laboratories,Hornby, ON, Canada) and Stratagene MX-4000 multiplex quantitative PCR system were then used to carry out real-time PCR, with ROX (Stratagene) as the reference dye. The PCR conditions (final reaction volume 25 μl) were as follows: 1 μl cDNA template (note that for 18S, the cDNA was diluted 1000 fold), 100 nmol l^–1 forward and reverse primers, 12.5 μl of 2× Master Mix, 0.375 μl of 1:1500 ROX final dilution. The annealing and extension temperatures over 40 cycles were, respectively,58°C (30 s) and 72°C (30 s). Primer pairs for each lungfish gene and lungfish 18S were designed using Primer3 software and are listed in Table 1 (pairs designated `Q'were used for real-time PCR). A 1150 bp segment of lungfish 18S was cloned from gill tissue using the methods described above and the primer pair 18S(Table 1; annealing temperature of 52°C), which consisted of primers designed against regions of high amino acid conservation identified by the examination of an alignment of 18S sequences from a range of vertebrates. The sequence was then used to design the real-time PCR 18S primers listed in Table 1. The specificity of each primer pair was verified by the cloning (TOPO TA cloning kit; Invitrogen)and sequencing of the amplified product. SYBR green dissociation curves were constructed after the completion of 40 PCR cycles and revealed the presence of single amplicons for each primer pair. The omission of reverse transcriptase during cDNA synthesis was used as a control to ensure that residual genomic DNA was not being amplified. For comparisons among treatments (control, acid infused, base infused), mRNA expression relative to the control group was calculated using the modified delta-delta Ct method(Pfaffl, 2001) with lungfish 18S mRNA expression as a normalizing factor.

Data are reported as mean values ± 1 s.e.m. The statistical significance of treatment effects was assessed by one-way repeated measures analysis of variance (RM-ANOVA). Where one-way RM-ANOVA indicated that significant differences existed, a post hoc multiple comparisons test(Holm–Sidak method) was applied. Where assumptions of normality or equal variance were violated, equivalent non-parametric tests were used. Statistical analyses were carried out using SPSS SigmaStat v3.0 (Systat Software) software with a fiducial limit of significance in all tests of 0.05.

In acid-infused lungfish, arterial blood pH (pHa) fell by 0.24 units but had recovered to a level not significantly different from the control by 6 h(Fig. 1). The fall in pHa was accompanied by a significant 20% reduction in the plasma HCO ^–₃ concentration ([HCO ^–₃]) and a more transient 1.4-fold elevation of arterial CO₂ tension (Pa_CO2). Base infusion caused a metabolic alkalosis in the arterial blood, characterized by a 0.29 unit increase in pHa associated with a 1.7-fold increase in plasma [HCO ^–₃] and unchanged PaCO₂. By 3 h, blood acid–base status was not significantly different from the control situation (Fig. 1).

Rapid correction of the acid–base disturbances was accomplished, at least in part, through adjustment of pulmonary and branchial ventilation frequencies (Figs 2 and 3). Acid-infused lungfish(Fig. 2) approximately doubled both air (1.8-fold) and water (1.7-fold) breathing frequencies, thereby increasing total CO₂ excretion(Ṁ_CO2) 2.4-fold. This increase in total Ṁ_CO2 was accomplished through a doubling of CO₂ transfer to water together with a 4.8-fold increase in aerial CO₂ excretion, illustrating the fact that P. annectens is capable of effective CO₂excretion via the lungs, even though under control conditions aquatic CO₂ transfer accounted for 78.2±4.6% (N=18) of total Ṁ_CO2. By contrast with the enhancement of CO₂ excretion, O₂uptake (Ṁ_O2) was unaffected by acid infusion, causing the respiratory exchange ratio (RER) to increase, although not significantly (Table 2). Air-breathing frequency and consequently aerial CO₂transfer were unaffected by the infusion of a base load(Fig. 3), whereas water-breathing frequency in base-infused lungfish was halved(Fig. 3), accounting for the 2.6-fold decrease in total Ṁ_CO2 and driving the RER value down significantly (Table 2).

Adjustment of the net excretion of acidic equivalents into the water(J_netH⁺) was also important in correcting acid–base disturbances, particularly following the infusion of a base load (Fig. 4). Acid infusion resulted in a significant 5.7-fold increase in net ammonia excretion(J_netNH₃). However, owing to more variable titratable acid net flux (J_netTA) data in which only seven of 11 fish increased J_netTA in response to acid infusion,there was no significant increase in the net excretion of acidic equivalents. By contrast, both J_netH⁺ and J_netTA values were strongly negative following base infusion, indicating that net excretion of base to the water occurred; net ammonia excretion was unaffected by base infusion.

To further investigate metabolic compensation of an infused base load in African lungfish, a separate group of base-infused lungfish was fitted with urinary catheters, allowing the contribution of the kidneys (basic equivalents appearing in urine) to net base excretion to be distinguished from that of the gills and/or skin (basic equivalents appearing in water). Responses of total(i.e. the sum of branchial and renal) J_netH⁺, J_netTA and J_netNH₃ of these fish to base infusion were qualitatively and quantitatively comparable to those of the lungfish not fitted with urinary catheters(Fig. 5; Table 3). Moreover, branchial and renal J_netH⁺, J_netTA and J_netNH₃ responses to base infusion were qualitatively similar, with net ammonia excretion into water and urine remaining constant and the transition to negative values of branchial and renal J_netH⁺ being driven by increasingly negative J_netTA values. Consistent with the loss of basic equivalents in the urine, urine [HCO ^–₃] increased 3.6-fold and urine pH rose by 0.68 units in response to base infusion(Fig. 6); urine flow rate over the experimental period did not differ significantly from the control value of 5.9±0.7 ml kg^–1 h^–1(N=6). Quantitatively, however, J_netH⁺ was dominated by branchial (and/or cutaneous) contributions to the net excretion of acidic equivalents, which accounted for >90% of J_netH⁺ under control conditions(Fig. 5B). During metabolic compensation of an acid–base disturbance, the renal contribution to the transfer of acid–base equivalents appeared to increase, although the effect was not statistically significant.

Because the responses of lungfish to acid–base disturbances included changes in the net excretion of acid–base equivalents via the gills and kidneys, the expression of mRNA of three genes thought to be involved in both renal and branchial mechanisms of acid–base excretion was assessed. Using homology cloning strategies, a 683 base pair cDNA was assembled from lungfish gill tissue. This cDNA included a deduced protein fragment of 226 amino acid residues. A BLAST search of the GenBank protein database determined that this lungfish gene shared high sequence identity with several vertebrate NBCs, and in particular, NBC1. Protein alignment revealed that the lungfish NBC fragment (dubbed lfNBC) corresponded to residues 498–723 of the human NBC (GenBank acc. no. AAG47773), and was 73–77% identical to amphibian and mammalian NBC1s, and 65–67%identical to NBCs from other fish (Fig. 7A). A second cDNA, of 921 base pairs, was assembled from lungfish gill tissue and included a deduced protein fragment of 307 amino acid residues that was found to share high sequence identity with the B subunit,particularly the type 2 B subunit, of the V-type H⁺-ATPase from several vertebrates. Using protein alignment, the lungfish H⁺V-ATPase (lfVATPase) corresponded to residues 195–511 of the human B2 subunit (GenBank acc. no. CAA44721) and was ⩾95% identical and ⩾98%similar to B subunits from a range of vertebrates including fish, amphibians,birds and mammals (Fig. 7B). Finally, degenerate primers were used to obtain a 663 base pair cDNA from lungfish blood (Fig. 8). This cDNA encoded a deduced protein fragment of 221 amino acid residues. The collective evidence of a BLAST search of the GenBank protein database, protein alignment and phylogenetic analysis suggests that this lungfish gene is a cytoplasmic carbonic anhydrase (CA) that, based on its tissue of origin, was labelled lfCAb. Phylogenetic analysis grouped lfCAb with fish cytoplasmic(blood-specific and general cytoplasmic) isoforms as well as mammalian CA I,II, III and XIII. Similarly, lfCAb was found to be 61–65% identical and 77% similar to blood CA isoforms from zebrafish, trout, carp and gar (GenBank protein acc. no. NP_571185, AAP73748, AAZ83748 and AAM94169, respectively) as well as 60% identical and 78% similar to mouse and rat CA XIII (NP_078771 and XP_574890, respectively). Examination of the mRNA expression for these three lungfish genes, lfNBC, lfVATPase and lfCAb, in gill and kidney tissue 4 h and 8 h following a 1 h infusion of acid or base did not reveal any statistically significant changes (Fig. 9).

The results of the present study demonstrate that the African lungfish P. annectens utilizes both respiratory and metabolic compensation to recover from metabolic acid–base disturbances. In modulating ventilation to meet the demands of acid–base regulation, P. annectensadopts what is considered to be a primarily tetrapod strategy (for a review,see Perry and Gilmour, 2006). By contrast, metabolic compensation of pH imbalances in P. annectensrelies predominantly on the exchange of acid–base equivalents across the gills (and/or skin, although with a possible role for the intestine) rather than on urinary excretion of acid–base equivalents, a strategy that is distinctly fish-like (for a review, see Perry and Gilmour, 2006). Because CO₂ excretion and net H⁺ excretion were not monitored continuously following acid or base infusion, it was not possible to accurately quantify the relative contributions of respiratory and metabolic mechanisms to acid–base compensation. However, estimates can be obtained by examination of the 1 h periods of greatest excretion (`peak' values),providing at least a snapshot of the relative contributions of the different mechanisms. Lungfish received an acid or base load of 3 mmol kg^–1. Following acid infusion, CO₂ excretion into air and water peaked at 0.62 mmol kg^–1 above the pre-infusion value, while net H⁺ excretion peaked at 0.18 mmol kg^–1 in excess of the pre-infusion value. Thus, peak CO₂ and H⁺ excretion (for a 1 h period) amounted to a loss of 27% of the total acid load, with respiratory compensation accounting for 77% of this value, and metabolic compensation, 23%. Respiratory and metabolic contributions were more balanced following a base load, accounting for 46% and 54%, respectively, of the 0.91 mmol kg^–1 of the base load that was lost during the 1 h period of greatest excretion (30% of the total base load).

The compensatory strategies of P. annectens reflect the unique opportunities available to bimodally breathing animals that inhabit an aquatic environment. Yet despite the opportunity afforded by such animals to understand acid–base strategies, few studies have investigated, and none appear to have attempted to quantify, the relative involvement of respiratory versus metabolic compensatory mechanisms in bimodal breathers(reviewed by Truchot, 1987). The involvement of air breathing in the recovery from acid–base disturbances elicited by exhaustive exercise was assessed for two fish species that are facultative air-breathers, bowfin Amia calva and spotted gar Lepisosteus oculatus (Burleson et al., 1998; Gonzalez et al.,2001). Both studies concluded that air breathing delayed compensation of the post-exercise acid–base disturbance because branchial ventilation was not increased to the extent that occurs in unimodal water breathers, thereby hindering branchial exchange of acid–base equivalents and CO₂ elimination, processes critical for the restoration of acid–base balance(Burleson et al., 1998; Gonzalez et al., 2001). Neither study, however, assessed metabolic compensation post-exercise. The blood acid–base responses of lungfish in the present study were very comparable to those of channel catfish, a unimodal water-breather, subjected to virtually the same infusion protocol(Cameron and Kormanik, 1982),implying that air-breathing in lungfish did not appear to delay recovery. Indeed, CO₂ loss via the lungs of P. annectensshould have aided recovery from an infused acid load, a difference from bowfin and gar that may reflect the obligate nature of pulmonary gas transfer in lungfish versus the facultative role it plays in bowfin and gar.

The best-studied bimodal breathers are amphibians, among which responses to acid–base disturbances reflect the degree of terrestrialization of the species (reviewed by Toews and Boutilier,1986; Truchot,1987; Toews and Stiffler,1989). Amphibians that are predominantly aquatic rely on the cutaneous exchange of acid–base equivalents to adjust extracellular pH(reviewed by Stiffler, 1989). By contrast, semi-terrestrial amphibians with well-developed lung ventilation,such as adult anurans, exhibit respiratory compensation supported to a variable extent by the exchange of acid–base equivalents viathe kidney/bladder or skin (e.g. Boutilier et al., 1979; Tufts and Toews,1985; Boutilier and Heisler,1988; Stiffler,1991).

The use of ventilatory adjustments to compensate for acid–base disturbances by semi-terrestrial but not predominantly aquatic amphibians parallels the situation in fish – water-breathing fish do not appear to modulate ventilation to correct pH imbalances (for a review, see Perry and Gilmour, 2006),whereas P. annectens clearly utilized respiratory compensation (Figs 2 and 3). The implication of this finding is that P. annectens, like tetrapod vertebrates, possesses CO₂ and/or pH-sensitive chemoreceptors that monitor the status of the body fluids. In tetrapods, central and peripheral CO₂/pH-sensitive chemoreceptors initiate ventilatory adjustments for acid–base regulation (reviewed by Gonzalez et al., 1994; Milsom, 1995; Milsom, 2002; Taylor et al., 1999). By contrast, the balance of evidence suggests that strictly water-breathing fish lack internally oriented CO₂/pH chemoreceptors – central chemoreceptors appear to be absent, and ventilatory chemoreflexes are dominated by branchial chemoreceptors that respond primarily to changes in water CO₂ (reviewed by Gilmour and Perry, 2007). Although the locations of CO₂/pH chemoreceptors in P. annectens remain to be determined, work on the South American lungfish Lepidosiren paradoxa suggests the existence of both central and peripheral CO₂ and/or pH chemoreceptors(Sanchez et al., 2001; Amin-Naves et al., 2007). In Lepidosiren, pulmonary ventilation increased as cerebrospinal fluid(CSF) pH was lowered by perfusing the fourth cerebral ventricle with mock CSF,providing evidence of central CO₂/pH chemosensitivity(Sanchez et al., 2001). Similar manipulation of CSF pH moderated but did not eliminate pulmonary ventilation responses to (combined) aerial and aquatic hypercapnia, a result that supports the existence of peripheral CO₂/pH chemoreceptors in Lepidosiren but without providing information on their orientation(water/air versus blood)(Amin-Naves et al., 2007). Interestingly, the data for P. annectens indicate that branchial as well as pulmonary ventilation is sensitive to internal acid–base status,a possibility that was not examined in the studies on Lepidosiren. Similarities (e.g. in acid–infused lungfish; Fig. 2) and differences (e.g. in base-infused lungfish; Fig. 3) between pulmonary and branchial ventilatory responses to acid–base disturbances suggest that both shared and independent control systems are present.

Although the modulation of ventilation to correct acid–base disturbances, and the underlying implication that internal CO₂/pH sensors must be present evokes a tetrapod scenario of acid–base regulation for P. annectens, the primary reliance on branchial(and/or cutaneous) rather than renal routes for the exchange of acid–base equivalents is typical of fish and predominantly aquatic amphibians. In P. annectens, partitioning of the peak excretion of a base load into branchial/cutaneous versus renal contributions revealed that 62% was eliminated into the water, and 38% into urine. Urine is similarly a minor route of net acid efflux during acid–base disturbances in fish (Kobayashi and Wood,1980; Wood and Jackson,1980; Cameron and Kormanik,1982; Wood, 1991; Curtis and Wood, 1992) and aquatic amphibians (Stiffler and Bachoura,1991; Stiffler,1991; Talbot and Stiffler,1992), with extra-renal excretion of acid–base equivalents playing a dominant role. In fish, up to 90% or more of acid–base movements occur across the branchial epithelium(Claiborne et al., 2002; Evans et al., 2005), whereas cutaneous acid–base excretion appears to predominate in aquatic amphibians (Stiffler, 1989). These widely accepted views are based primarily on morphological considerations rather than on explicit attempts to partition extra-renal net acid efflux into branchial versus cutaneous components. For example,the role of the gill in correcting pH imbalances in fish is strongly supported by the abundance of ion-transporting MR cells in the branchial epithelium coupled with remodelling of this epithelium in response to acid–base disturbances (reviewed by Goss et al.,1992; Goss et al.,1994; Goss et al.,1995; Laurent and Perry,1995; Perry and Gilmour,2006). MR cells are found in both the branchial epithelium and skin of P. annectens (Sturla et al., 2001), indicating that either surface could contribute to the movement of acid–base equivalents, but the gills may be more likely to play a larger role. Whereas the skin is covered by a protective coating of scales, the gills are actively ventilated and involved in CO₂excretion, as evidenced by the relationship between ventilation frequency and Ṁ_CO2 (Figs 2 and 3), and the hydration of CO₂ provides H⁺ and HCO ^–₃for exchange with the environment (see below). However, whether the branchial epithelium of P. annectens is a more important contributor to metabolic acid–base regulation than the skin requires empirical confirmation.

The molecular mechanisms involved in the movement of acid–base equivalents across the gills of freshwater fish remain a subject of debate. Recent models (reviewed by Perry and Gilmour, 2006) suggest that acid secretion reflects the combined actions of apical membrane H⁺ efflux, probably via a V-type H⁺-ATPase in freshwater fish, and basolateral membrane HCO ^–₃ efflux by means of Cl^–/HCO ^–₃ exchange and/or NBC. Mechanisms for base excretion are probably located in a different cell type, and involve apical membrane HCO ^–₃ efflux viaCl^–/HCO ^–₃ exchange coupled to basolateral H⁺ efflux by a V-type H⁺-ATPase. In both cases, the hydration of molecular CO₂ catalyzed by a high activity cytosolic CA provides the necessary protons and HCO ^–₃ ions. Several components of these mechanisms,specifically NBC, V-type H⁺-ATPase, and cytosolic CA, were cloned from lungfish tissues (Figs 7and 8), but no evidence of changes in mRNA expression was detected in the gills following acid or base infusion (Fig. 9). This result was somewhat surprisingly given the considerable support for transcriptional regulation of branchial proteins during acid–base disturbances in other fish (Galvez et al., 2002; Hirata et al., 2003; Perry et al., 2003a; Perry et al., 2003b; Georgalis et al., 2006b). It is possible that the acid–base challenges were not sufficient to activate transcriptional regulation, that isolation of discrete cell types(acid versus base excreting) is necessary to detect such changes, or that post-translational mechanisms are involved, as recently reported for dogfish (Tresguerres et al.,2005; Tresguerres et al.,2006). Alternatively, the sampling times (4 and 8 h post infusion)and/or animal numbers may not have been appropriate or sufficient to detect changes that did occur; unfortunately, these were constrained by animal availability.

The kidney of P. annectens, although quantitatively less important than the gill/skin, nevertheless played a significant role in eliminating an infused base load, a finding that is true for other freshwater fish(Cameron and Kormanik, 1982; Curtis and Wood, 1992; Wood et al., 1999). Moreover,the freshwater teleost kidney exhibits a flexibility of response that is comparable to that of the mammalian kidney(Wood et al., 1999), a statement that arguably can also be applied to the lungfish. Urine flow rate in P. annectens in the present study, at ∼6 ml kg^–1 h^–1, was within the range of values reported previously for Protopterus sp., ∼2–6 ml kg^–1 h^–1(Sawyer, 1965; Sawyer, 1970; Babiker and Rankin, 1979),although somewhat higher than values for rainbow trout (Oncorhynchus mykiss) or channel catfish (Ictalurus punctatus), ∼2–4 ml kg^–1 h^–1(Cameron and Kormanik, 1982; Curtis and Wood, 1992; Wood et al., 1999). This result is surprising in view of the reduced gill surface area of lungfish, and may indicate a degree of stress-induced diuresis(Babiker and Rankin, 1979). A high control value may also have masked any tendency for metabolic alkalosis to increase urine flow rate, as observed previously in trout and catfish(Cameron and Kormanik, 1982; Curtis and Wood, 1992; Wood et al., 1999). As in other freshwater fish as well as mammals(Cameron and Kormanik, 1982; Curtis and Wood, 1992; Wood et al., 1999), the renal response of P. annectens to a metabolic alkalosis was characterized by greatly elevated urinary HCO ^–₃ loss(Fig. 6), presumably reflecting increased HCO ^–₃ filtration that was not matched by a corresponding increase in tubular HCO ^–₃reabsorption.

The mechanisms involved in renal HCO₃^–reabsorption appear to be similar in freshwater fish and mammals(Perry et al., 2003b; Perry and Gilmour, 2006; Georgalis et al., 2006a). In mammals, filtered HCO ^–₃ ions combine with protons secreted via apically localized Na⁺/H⁺exchangers (NHE3) and/or V-type H⁺-ATPase. Molecular CO₂formation is catalyzed by membrane-associated CA IV; CO₂ diffuses into the tubule cell and is hydrated in the presence of cytosolic CA. While protons are recycled into the tubule lumen, HCO ^–₃ions move across the basolateral membrane via NBC1, resulting in a net transfer of HCO ^–₃ ions from filtrate to blood. In P. annectens, mRNA expression of NBC1, V-type H⁺-ATPase and cytosolic CA was detected in kidney tissue,suggesting that these components of the HCO ^–₃reabsorption mechanism are present, but expression was not affected by acid or base loading (Fig. 9). As discussed above for the branchial mRNA expression of acid–base relevant proteins, the apparent lack of transcriptional regulation in lungfish was surprising in view of responses in both fish and mammals(Tsuruoka et al., 1998; Amlal et al., 2001; Perry et al., 2003a; Perry et al., 2003b; Georgalis et al., 2006a), and may reflect the selection of inappropriate sampling times, insufficient sample number, that increased activities of these proteins are not required to respond to the acid–base disturbance, or the involvement of posttranslational modifications, as reported for mammalian renal tubules(Amlal et al., 2001; Nakhoul and Hamm, 2002).

Thanks are extended to Branka Vulesevic and Cheryl Szebedinszky for technical assistance, to Bill Fletcher for caring for the lungfish, and to Ian Myers for development of the lungfish air breath detector. Financial support for this study was provided by Natural Sciences and Engineering Research Council (NSERC) of Canada grants to S.F.P. and K.M.G., NUS R-154-000-226-112 grant to Y.K.I., and by Canada Foundation for Innovation support of the Centre for Advanced Research in Environmental Genomics (CAREG). Graduate student support for L.K. was provided through the Premier's Research Excellence Award(PREA) program (K.M.G.). A.J.E. was supported by an NSERC post-doctoral fellowship.