Carbonic anhydrase expression and CO₂ excretion during early development in zebrafish Danio rerio

Carbonic anhydrase (CA) is the zinc metalloenzyme that catalyzes the reversible reactions of CO₂, i.e. CO₂+H₂O↔H⁺+HCO₃⁻. Fish, like mammals (reviewed by Chegwidden and Carter, 2000; Hewett-Emmett, 2000; Sly, 2000; Tashian et al., 2000; Purkerson and Schwartz, 2007; Hilvo et al., 2008), possess an abundance of CA isoforms that differ in molecular sequence, kinetic properties and susceptibility to inhibitors, tissue distribution and subcellular localization (reviewed by Henry and Swenson, 2000; Esbaugh and Tufts, 2006; Gilmour and Perry, 2009). In adult fish, several of these isoforms have been implicated in a specific fashion in physiological processes ranging from CO₂ excretion (Henry and Swenson, 2000; Perry and Gilmour, 2006; Esbaugh and Tufts, 2006), through acid—base balance and ionic regulation (Perry and Gilmour, 2006; Gilmour and Perry, 2009), to ureagenesis and metabolism (Henry, 1996). However, much less is known about the expression or function of CA isoforms during early development in fish. The objective of the present study was to investigate CA expression and function during early development in zebrafish (Danio rerio), with a particular focus on the main intracellular CA isoforms and their potential contribution to CO₂ excretion.

Work carried out on zebrafish and rainbow trout (Oncorhynchus mykiss) suggests that adult teleost fish possess two closely related cytosolic CA isoforms that differ in tissue distribution and kinetic properties (Rahim et al., 1988; Esbaugh et al., 2004; Esbaugh et al., 2005; Lin et al., 2008) (reviewed by Gilmour and Perry, 2009). One, referred to here as the ‘b’ isoform, is expressed predominantly in the blood and has a higher catalytic efficiency than the second, ‘c’, isoform, which is more widely distributed, with high expression in the gills, lower expression in the kidney and little or none in red blood cells (RBCs) (Rahim et al., 1988; Esbaugh et al., 2005; Lin et al., 2008). RBC CA plays a key role in CO₂ excretion by catalyzing the hydration of CO₂ to HCO₃⁻ to load CO₂ into the blood in the tissues, and by catalyzing the reverse process in the gills, where HCO₃⁻ is dehydrated to CO₂ that then diffuses out of the blood and into the ventilatory water (see Perry, 1986; Tufts and Perry, 1998; Perry and Gilmour, 2006) (reviewed by Esbaugh and Tufts, 2006). When the availability of RBC CA is reduced, through either severe anaemia (Wood et al., 1982; Gilmour and Perry, 1996; Gilmour and MacNeill, 2003) or treatment with a permeant CA inhibitor such as acetazolamide (e.g. Hoffert and Fromm, 1973; Gilmour et al., 2001), CO₂ excretion is impaired as evidenced by increases in the partial pressure of CO₂ (P_CO2) in the arterial blood and the appearance of a respiratory acidosis. By contrast, branchial cytosolic CA contributes to ionic regulation and acid—base balance, but not CO₂ excretion. Cytosolic CA in the branchial epithelium catalyzes the hydration of CO₂ to H⁺ and HCO₃⁻ that can then be used as counter-ions for, respectively Na⁺ and Cl⁻ uptake, transport processes that are critical for both ionic regulation and acid—base balance (reviewed by Evans et al., 2005; Perry and Gilmour, 2006; Tresguerres et al., 2006; Hwang and Lee, 2007; Gilmour and Perry, 2009). Correspondingly, inhibition of branchial CA reduces Na⁺ and/or Cl⁻ uptake (Maetz, 1956; Maetz and Garcia-Romeu, 1964; Kerstetter et al., 1970; Payan et al., 1975; Boisen et al., 2003; Chang and Hwang, 2004) as well as branchial net acid excretion (Georgalis et al., 2006).

Zebrafish embryos and larvae express both zCAb and zCAc (Lin et al., 2008), but the relative expression of zCAb versus zCAc as a function of developmental stage has not yet been reported. Using in situ hybridization, zCAc mRNA expression was localized to a subtype of ionocytes in the skin and gills that is enriched in vacuolar-type H⁺-ATPase (HR cell), whereas zCAb mRNA expression was detected in blood cells (Lin et al., 2008). A role for zCAc in Na⁺ uptake and acid excretion has been proposed on the basis of data showing reduced Na⁺ uptake and/or acid excretion in larvae treated with the CA inhibitor ethoxzolamide (Esaki et al., 2007) and in embryos and larvae in which zCAc has been knocked down using antisense morpholino oligonucleotides (Lin et al., 2008). The function of zCAb has not yet been investigated. Moreover, several studies have assessed O₂ uptake in embryonic or larval fish (see reviews by Rombough, 1988; Rombough, 2004), but corresponding data for CO₂ excretion do not appear to be available.

With this background in mind, the present study examined the hypothesis that, as rates of CO₂ production increase during early development in zebrafish, CA becomes necessary for effective CO₂ excretion, and that the pattern of CA expression during early development reflects this transition. That is, increased zCAb mRNA expression was expected to precede or coincide with a developmental time point at which CA becomes beneficial for CO₂ excretion, as demonstrated by significant reductions in CO₂ excretion when CA activity, and specifically CAb activity, is reduced by selective gene knockdown. Relative expression of zCAc and zCAb mRNAs was evaluated using real-time RT-PCR. CO₂ excretion was determined relative to O₂ uptake using closed system respirometry. CA activity was reduced by treatment with acetazolamide, and because acetazolamide inhibits all CA activity, the effects of selective knockdown using antisense morpholino oligonucleotides of zCAc and zCAb were also examined.

Mature zebrafish (Danio rerio Hamilton) obtained from a commercial supplier (Big Al's, Montreal, Canada) were transported to the University of Ottawa Aquatic Care Facility where they were maintained in fibreglass tanks (4 l) held under a constant 10 h:14 h L:D photoperiod and supplied with aerated, dechloraminated city of Ottawa tap water at 28°C (water ion concentrations (in mmol l⁻¹): Na⁺ 0.68, Cl⁻ 0.13, K⁺ 0.02, Ca²⁺ 0.23; water pH 7.2-7.7). Newly spawned (within 30 min) eggs were collected from breeding tanks containing equal numbers of male and female fish. Fertilized eggs were incubated in rearing tanks at 28°C until the desired developmental stage. Hatching normally occurred around 48 h post fertilization (h.p.f.), but larvae were not fed; the oldest larvae used in the present study were 120 h.p.f. All procedures for animal use were carried out in accordance with the animal care guidelines of the University of Ottawa and the Canadian Council on Animal Care.

Real-time RT-PCR was used to assess whole body mRNA expression of the two main intracellular CA isoforms as a function of developmental stage for embryos and larvae ranging from 3 h.p.f. to 120 h.p.f. In addition, the relative mRNA expression of these two CA isoforms in RBCs from adult zebrafish was compared. Adult zebrafish were killed by a blow to the head and blood was collected into microcapillary tubes following caudal severance. Blood samples collected from four to eight zebrafish were pooled to form a single sample.

Primers (Table 1) were designed using Genetool software (BioTools Inc., Jupiter, FL, USA) to amplify an ~100 base pair (bp) section of NM_131110 [termed zCAb in this study or zCA2-like b by Lin et al. (Lin et al., 2008) and of NM_199215 (termed zCAc in this study but also known as zCA 2-like a) (Lin et al., 2008)]. On the basis of phylogenetic (see Esbaugh and Tufts, 2006; Gilmour and Perry, 2009) and functional (see Lin et al., 2008) analyses, zCAb was expected to be expressed at high levels in RBCs whereas a broader distribution across tissues with little expression in RBCs was expected for zCAc. All primers were selected for optimal amplification at an annealing temperature of 58°C. The specificity of individual primer pairs was confirmed by sequencing of amplicons.

Total RNA was extracted from homogenized samples (<100 mg, representing ~40-60 embryos or larvae, or blood from four to eight adult zebrafish) using TRIzol reagent (Invitrogen, Carlsbad, CA, USA) according to the manufacturer's instructions. Embryos, larvae or blood samples were homogenized using a 21 gauge needle attached to a 3 ml syringe until the mixture could pass easily through the needle. Following quantification of RNA by spectrophotometry at 260 nm (Spectramax 340 PC, Molecular Devices, Sunnyvale, CA, USA or ND-1000 NanoDrop, ThermoFisher Scientific, Ottawa, ON, Canada), samples were treated according to the manufacturer's instructions with amplification grade DNase I (Invitrogen) to eliminate any genomic DNA contamination. First-strand cDNA synthesis was then performed on 2 μg of DNase-treated total RNA using random primers (200 ng per reaction) and RevertAid H-minus M-MulV reverse transcriptase (Fermentas, Burlington, ON, Canada) according to the manufacturer's instructions. The final cDNA product was diluted with an equal volume of autoclaved water.

Real-time RT-PCR reactions were performed using a SYBR green master mix kit (Stratagene, Agilent Technologies, Cedar Creek, TX, USA) and Mx3000P Real-Time PCR System with associated software (Stratagene). The thermocycler settings were those suggested by the manufacturer, as was the composition of the reaction, with the exception that the reaction volume was scaled to 12.5 μl from 25 μl. Specificity of primer pairs and the occurrence of primer—dimer formation were monitored by creating and evaluating a dissociation curve for each reaction. To confirm that products did not arise from amplification of genomic DNA, ‘no reverse transcriptase’ control templates, generated by omission of reverse transcriptase during cDNA synthesis, were included in every plate. Standard curves were used to assess reaction efficiency. Templates for standard curves were generated by serial dilution (1:5 using RNase/DNase-free molecular grade water; ThermoFisher Scientific) of embryo cDNA. Reactions were then carried out using 2 μl of template for six diluted samples and a linear regression was calculated for the relationship between cycle threshold (Ct) and relative template concentration. Primer pair efficiencies of 85-115% with an R² of at least 0.98 were deemed to be acceptable. Relative mRNA expression of the gene of interest against the reference 18S gene (template diluted 1000-fold) was determined according to the modified Δ-ΔCt method of Pfaffl (Pfaffl, 2001). For purposes of comparison across the developmental time course, mRNA expression of both zCAb and zCAc was calculated relative to the expression of zCAc mRNA at 1 h.p.f. Similarly, expression of zCAc mRNA in adult zebrafish RBCs was calculated relative to that of zCAb.

O₂ consumption and CO₂ excretion were determined for groups of 40-60 zebrafish embryos or larvae using two closed-cell glass micro-respirometry chambers (2.94 ml volume; Loligo Systems, Tjele, Denmark). Micro-respirometry chambers were used in conjunction with a mini magnetic stirring system (Loligo Systems); each chamber contained a magnetic stirrer bar that was separated from the embryos or larvae by a stainless steel mesh. The chambers were connected individually to a supply of aerated, 28°C water and the entire system was submerged in a water bath held at 28°C. Water P_O2 was monitored using fibre-optic O₂ electrodes (FOXY AL-300; Ocean Optics, Dunedin, FL, USA) and a PC running Ocean Optics software. Electrodes were calibrated immediately before each experimental run by immersing them first in zero solution (2 g l⁻¹ sodium sulphite) and then in air-saturated water until stable readings were recorded. Water total CO₂ concentration was measured at the beginning and end of each experimental run. Water samples (500 μl) were withdrawn from the chamber into gas-tight Hamilton syringes and total CO₂ content was measured in triplicate on 100 μl samples using a Capni-Con 5 (Cameron Instruments, Port Aransas, TX, USA).

An experimental run commenced once stable P_O2 readings were obtained following the addition of a group of embryos or larvae to the micro-respirometry chamber. An initial water sample was withdrawn, the chamber was sealed, and P_O2 was monitored as oxygen levels declined owing to consumption by the fish. The experimental run was terminated when water P_O2 reached 6666 Pa (~20-90 min depending on the age of the embryos/larvae) and the final water sample was withdrawn. O₂ consumption () was calculated from the rate of change of P_O2 taking into account the volume of chamber, mass of embryos/larvae and the solubility coefficient of O₂ in water at 28°C (Boutilier et al., 1984). CO₂ excretion () was calculated from the difference in total CO₂ content between the initial and final water samples, taking into account the duration of the experimental run and the mass of embryos/larvae. To examine CO₂ excretion relative to O₂ uptake, the respiratory exchange ratio (RE) was calculated for each experimental run as /.

For purposes of comparison, O₂ uptake and CO₂ excretion were also determined for adult zebrafish. The experimental approach used was essentially the same as that detailed above, but a single fish was used rather than a group, and the micro-respirometry chamber was replaced by a respirometer fashioned from a 10 ml plastic syringe. Water circulation during the period of closed-system respirometry was achieved using a peristaltic pump that passed water through a short tubing loop; the total volume of the respirometer was approximately 7 ml. Fish were acclimated to the respirometer for 1 h before commencing an experimental run.

O₂ consumption and CO₂ excretion were determined for groups of zebrafish embryos or larvae at developmental stages ranging from 3 h.p.f. to 120 h.p.f. To determine whether changes in CO₂ excretion at hatching (~48 h.p.f.) were associated with loss of the egg capsule, the egg capsule was manually removed under a light microscope using fine forceps. The egg capsule functions to protect the developing embryo, but forms a barrier to O₂ diffusion from the bulk water to the embryo (Ciuhandu et al., 2007) and might therefore also impede CO₂ diffusion in the opposite direction. Egg capsule removal was carried out at ~22 h.p.f., and O₂ consumption and CO₂ excretion were then measured at 24 h.p.f. or following an additional 24 h of incubation without the egg capsule, at 48 h.p.f. The contribution of CA to CO₂ excretion was then examined at 48 h.p.f. using one of two approaches, widespread inhibition of CA activity using the permeating sulphonamide inhibitor acetazolamide (Az), or selective knockdown of cytosolic CA isoforms using antisense morpholino oligonucleotides (MOs). In either case, whole-body CA activity was assayed to confirm CA inhibition (see below). Larvae were treated with acetazolamide by incubating them for 15 min in a solution of 2×10⁻⁴ mol l⁻¹ Az. Acetazolamide was dissolved in water in which the fish were to incubated by raising the pH to ~11 with NaOH. Water pH was brought back to approximately neutral (pH 7.2) by titration with HCl prior to the addition of larvae. To selectively inhibit the translation of zCAc, zCAb, or both zCAc and zCAb, specific morpholino oligonucleotides were purchased from Gene Tools LLC (Philomath, OR, USA). The morpholinos (Table 1) were designed to be complementary to the translational start site of zCAb (NM_131110) and zCAc (NM_199215) based on the annotated sequences. Morpholinos were prepared in 1× Danieau buffer (58 mmol l⁻¹ NaCl, 0.7 mmol l⁻¹ KCl, 0.4 mmol l⁻¹ MgSO₄, 0.6 mmol l⁻¹ Ca(NO₃)₂ and 5 mmol l⁻¹ Hepes, pH 7.6) containing 0.1% Phenol Red as a visible indicator of successful injection, and were injected into zebrafish embryos at the one- to two-cell stage using a microinjector system (IM 300 programmable microinjector; Narishige, East Meadow, NY, USA) with injection needles pulled from filamented 1.0-mm borosilicate glass (Sutter Instrument, Novato, CA, USA). Embryos were injected with either morpholino singly (4 ng embryo⁻¹), or with both morpholinos together (2 ng embryo⁻¹ of each morpholino). Control for the effects of morpholino injection itself was achieved by injecting a separate group of embryos with a nonsense morpholino (4 ng embryo⁻¹; standard control morpholino oligonucleotide; Gene Tools LLC). To confirm the specificity and effectiveness of the morpholinos, fusion constructs were made in which the zCAc and zCAb morpholino target sequences were separately introduced upstream of and in frame with the red fluorescent protein dTomato coding sequence (Shaner et al., 2004) (a kind gift from Marc Ekker). Fusion constructs were generated via PCR and directionally ligated into the pCS2+ vector (Turner and Weintraub, 1994) (a kind gift from Marc Ekker) using T4 DNA ligase (Fermentas). Capped and poly(A)⁺ tailed mRNAs were subsequently generated using the SP6 mMESSAGE mMACHINE in vitro transcription system (Ambion; Applied Biosystems Canada, Streetsville, ON, Canada), according to the manufacturer's specifications. Capped mRNAs (100 pg embryo⁻¹) were then injected alone or with each CA isoform morpholino (4 ng embryo⁻¹) at the one- to two-cell stage. The occurrence of dTomato fluorescence was then examined in larvae at 6 days post fertilization (d.p.f.) using a UV assembly attached to a SMZ-1500 microscope (Nikon; Mississauga, ON, Canada) and photographed using a DXM 1200C digital camera (Nikon).

The micro-method of Brion et al. (Brion et al., 1988) was used to assay the CA activity present in 48 h.p.f. control larvae and those treated with acetazolamide, as well as the CA activity of 48 h.p.f. larvae that had been micro-injected with morpholino oligonucleotides. Groups (40-60) of larvae were weighed and then homogenized in 20 volumes of distilled water using a motor-driven Teflon homogenizer. As described by Brion et al. (Brion et al., 1988), the reaction vessel for the assay consisted of a 10×75 mm glass tube held in an ice-water bath and fitted with a rubber stopper containing 18 and 20 gauge needles for sample delivery as well as gas entry (CO₂ at 150 ml min⁻¹) and exhaust. An aliquot (500 μl) of water (uncatalyzed rate) or sample was added to the reaction chamber and gassed with CO₂. The reaction was initiated by the addition to the chamber of 500 μl of indicator buffer (20 mmol l⁻¹ imidazole, 5 mmol l⁻¹ Tris and 0.2 mmol l⁻¹p-nitrophenol) and timed manually until the initially yellow solution turned colourless. To prevent the occurrence of foaming in the reaction medium, 1 μl of octanol (Sigma-Aldrich; Oakville, ON, Canada) was added to all reactions. CA activity was calculated from the ratio of the times for colour disappearance measured for water versus active sample, corrected for dilution factor and sample mass, as described by Brion et al. (Brion et al., 1988).

Data are reported as means ± 1 standard error (s.e.m.). The effects of developmental stage were analyzed statistically using one-way analysis of variance (ANOVA). The statistical significance of treatment effects was assessed by Student's t-tests or one-way ANOVA, as appropriate. Where one-way ANOVA indicated that significant differences existed, the Holm-Sidak method was applied for post-hoc multiple comparisons. Equivalent non-parametric tests were substituted where assumptions of normality or equal variance were violated. Statistical analyses were carried out using SPSS SigmaStat v3.0 (Systat Software, Chicago, IL, USA) with the fiducial limit of significance set to 0.05.

Real-time RT-PCR evaluation of the relative mRNA expression of zCAb and zCAc revealed that both were detectable as early as 0 h.p.f. and increased markedly during early development (Fig. 1). The mRNA expression of zCAb was generally higher than that of zCAc, particularly during the earliest stages of development (≤48 h.p.f.). Whereas zCAc mRNA expression remained low during the early stages of development, becoming significantly elevated over the 0 h.p.f. value only at 72, 96 and 120 h.p.f., zCAb mRNA expression was distinctly elevated at 8 h and remained relatively high (although not significantly so) thereafter, with the relative mRNA expression at 72 and 96 h.p.f. also being significantly higher than at 0 h.p.f. (Fig. 1). The relative mRNA expression of the two isoforms was also examined in RBCs sampled from mature zebrafish, where zCAb mRNA expression, at 24.3±39.0 (N=8) was found to be significantly higher (paired Student's t-test, P=0.008) than that of zCAc, 0.001±0.001 (N=8; relative to zCAb).

Rates of O₂ consumption and CO₂ excretion were also determined as a function of developmental stage. To provide context, values for adult zebrafish are presented in Fig. 2, but these data were not included in statistical analyses. Both and were strikingly and significantly (one-way ANOVA, P<0.001 in both cases) higher at 48 to 120 h.p.f. than at the two earlier stages examined, 3 and 24 h.p.f. (Fig. 2A). However, calculation of RE ratios uncovered differences in the relative increases in and . In particular, RE ratios at 48, 96 and 120 h.p.f. were significantly higher (one-way ANOVA, P<0.001) than those at 3 and 24 h.p.f. (Fig. 2B), a difference that is indicative of a relative increase in CO₂ excretion between 24 and 48 h.p.f. Because hatching occurs over this period, the effect of egg capsule removal on , and RE ratio was examined. Egg capsules were removed at 22 h.p.f. and embryos were assessed at 24 or 48 h.p.f., the rationale being that any impact of hatching alone would be eliminated in embryos subjected to egg capsule removal. Embryos from which the egg capsule had been removed exhibited the same pattern as untreated embryos with the RE ratio being significantly higher (Student's t-test, P=0.014) at 48 h.p.f. (1.15±0.04, N=5) than at 24 h.p.f. (0.82±0.10, N=4), a finding that suggests that the relative increase in CO₂ excretion at 48 h.p.f. is not due to hatching.

Because 48 h.p.f. was the first developmental stage at which higher RE ratios were detected, the potential contribution of CA to CO₂ excretion was examined at this time point. Whole-body CA activity in 48 h.p.f. larvae treated with acetazolamide was 38% of the value for untreated larvae (Fig. 3C), a difference that was significant (Student's t-test, P=0.002). Acetazolamide-treated larvae exhibited significantly lower than control larvae in the absence of any impact on (Student's t-tests, P=0.001 for and 0.644 for ), an effect that resulted in a correspondingly significant difference (Mann—Whitney rank sum test, P=0.011) in RE ratio between control and acetazolamide-treated larvae (Fig. 3). Acetazolamide inhibits most or even all CA isoforms (although with differing sensitivity) (see Hilvo et al., 2008), and therefore the effects of selective knock down of zCAb or zCAc were examined using antisense morpholino oligonucleotides. Injection of embryos with one or both CA MOs resulted in significantly lower (one-way ANOVA, P=0.001) whole-body CA activity in 48 h.p.f. larvae than in control larvae at the same stage (Fig. 4C); morphant CA activity ranged from 39% to 69% of the control value. Similarly, RE ratios for zCAb, zCAc and zCAb+zCAc morphants at 48 h.p.f. were significantly lower (one-way ANOVA, P<0.001) than values for control larvae (Fig. 4B). Interestingly, was unaffected by zCAc and zCAb knockdown (one-way ANOVA on ranks, P=0.551), with the differences in RE ratio reflecting a trend for higher in morphant larvae, a trend that was significant for the double morphant (one-way ANOVA, P=0.01; Fig. 4A). It is also noteworthy that no differences were detected between zCAb and zCAc morphants. Morpholino specificity was confirmed by examination of fluorescence in zebrafish larvae (6 dpf) that were injected as one- or two-cell embryos with an RNA construct, consisting of zCAb or zCAc together with a fluorescent reporter (dTomato), either alone or in conjunction with one of the two morpholinos. Fluorescence was eliminated in larvae that had been co-injected as embryos with both RNA and morpholino for the same CA isoform, but remained in larvae that had been co-injected with RNA for one CA isoform and the morpholino against the other CA isoform (Fig. 5).

Rates of O₂ uptake have been measured for early developmental stages in a variety of fish species (e.g. Rombough, 1988), and values measured in the present study (3.8-53.4 μmol g⁻¹ h⁻¹ for zebrafish of 3-120 h.p.f.; Fig. 2) were comparable with those reported in other studies carried out on zebrafish of similar developmental stage and at similar temperatures (Pelster and Burggren, 1996; Barrionuevo and Burggren, 1999; Bagatto et al., 2001; Grillitsch et al., 2005; Rombough and Drader, 2009). By contrast, the present study appears to be the first to report measurements of CO₂ excretion during early development in a fish. Rates of CO₂ excretion (2.2-60.3 μmol g⁻¹ h⁻¹ for zebrafish of 3-120 h.p.f.; Fig. 2) were comparable with those of O₂ uptake, but calculation of RE ratios revealed significant differences in the rate of CO₂ excretion relative to O₂ uptake with developmental stage. Under steady state conditions, the RE ratio is expected to equal the respiratory quotient (the ratio of CO₂ production to O₂ consumption at the cellular level) with values between ~0.7 and 1.0 depending on the metabolic substrate (Burggren and Roberts, 1991). As endogenous feeders that rely on yolk lipids and amino acids (Kamler, 2008), zebrafish embryos and larvae would be expected to exhibit RE ratios or respiratory quotients of 0.7 (lipid) to 0.8 (amino acids). The RE ratio will deviate from the respiratory quotient when CO₂ is stored in the tissues (resulting in RE values <0.7) or when stored CO₂ is eliminated (resulting in RE values >1.0). In the present study, RE ratios prior to hatching were 0.5-0.7, whereas RE ratios for larvae were 1.2-1.4 with the exception of 72 h.p.f. larvae, for which the RE ratio was 0.8. This pattern indicates that CO₂ excretion during embryonic development did not match O₂ uptake, suggesting storage of CO₂ in the tissues, whereas conditions post-hatch favoured CO₂ excretion in excess of O₂ uptake.

Several factors could account for, or contribute to, the relative enhancement of CO₂ excretion post-hatch. Prior to hatching, the developing embryo is surrounded by the egg capsule, a multilayer, acellular, protective shell that forms a barrier to O₂ diffusion from the bulk water to the embryo (Ciuhandu et al., 2007; see also Rombough, 1988) and could also therefore limit CO₂ diffusion away from the embryo. However, embryos subjected to egg capsule removal at 22 h.p.f. and tested at 24 or 48 h.p.f. exhibited the same pattern of enhanced CO₂ excretion at 48 h.p.f. as intact embryos allowed to hatch normally, indicating that the relative suppression of CO₂ excretion prior to hatching was not the result of CO₂ retention by the egg capsule. An alternative possibility is that CO₂ retention during embryonic development reflects a requirement for HCO₃⁻ for urea synthesis. Zebrafish embryos excrete ~80% of their nitrogenous waste as urea, but switch from ureotely to ammonotely at hatching, a transition that is associated with increases in the expression of ammonia transporters (Rh glycoproteins) (Braun et al., 2009). Thus, ureotelism during early development may be a strategy to prevent ammonia toxicity under conditions where the rate of ammonia excretion is limited by a lack of ammonia transporters (Braun et al., 2009). Urea is formed in embryonic and larval teleost fish from glutamine (as the nitrogen donor) and HCO₃⁻via the ornithine—urea cycle (e.g. Wright et al., 1995; Chadwick and Wright, 1999; Barimo et al., 2004), and at least in adult toadfish, the HCO₃⁻ for urea synthesis is supplied by CA-catalyzed hydration of metabolic CO₂ (Walsh et al., 1989). Diversion of CO₂ into urea synthesis during embryonic development would lower CO₂ excretion relative to O₂ uptake. Quantitatively, however, this effect is not large enough to account for the discrepancy in rates of CO₂ excretion versus O₂ uptake prior to hatching, since rates of urea excretion at 0-24 h.p.f. are 0.01-0.02 μmol urea g⁻¹ h⁻¹ (Braun et al., 2009) whereas over the same period O₂ uptake exceeds CO₂ excretion by 1.6-3.4 μmol g⁻¹ h⁻¹ (Fig. 2). Bone formation might also result in CO₂ storage as carbonate (Pellegrino and Biltz, 1965), although bone formation extends well beyond hatching. It is likely that a number of factors, including urea synthesis and bone formation, contributed to the RE values <0.7 detected prior to hatching in the present study.

RE values >1.0 occurred post-hatching, a time when not only O₂ uptake and CO₂ excretion increased markedly, but also when ammonia excretion exceeds urea excretion (Braun et al., 2009). CO₂ excretion at this time clearly benefited from CA activity as evidenced by the significantly lower rates of CO₂ excretion by larvae in which CA activity was inhibited using acetazolamide. In adult teleost fish, the contribution of CA to CO₂ excretion is tied to the RBC localization of CA activity. RBC CA aids in loading CO₂ into the blood at the tissues by catalyzing the hydration of CO₂ to H⁺, which is buffered by haemoglobin, and HCO₃⁻, which exits the RBC via band 3 anion exchange; at the gill, this process is reversed to unload CO₂ from the blood with RBC CA catalyzing the dehydration reaction (Perry, 1986; Tufts and Perry, 1998). When RBC CA is inhibited by acetazolamide, falls and a respiratory acidosis (elevated arterial P_CO2 in conjunction with lowered arterial pH) develops [e.g. see figure 9 in Gilmour et al. (Gilmour et al., 2001)]. The elevation of arterial P_CO2 is expected ultimately to re-establish CO₂ excretion by increasing the partial pressure gradient for CO₂ diffusion from plasma to water. Proof-of-principle experiments using adult zebrafish exposed to waterborne acetazolamide confirmed this pattern; fell significantly by 25 min of exposure but showed evidence of recovery by 105 min (data not shown). The acetazolamide-induced lowering of CO₂ excretion in zebrafish larvae implies a significant role for CA in CO₂ excretion.

RBC CA expression in trout is primarily as the CAb isoform (Rahim et al., 1988; Esbaugh et al., 2005) and analysis of RBCs from adult fish indicated that the same is true of zebrafish. Examination of whole body mRNA expression profiles for zCAb and zCAc as a function of developmental stage indicated that zCAb is the more highly expressed isoform during early development, particularly before 48 h.p.f. Differential tissue distribution of zCAb and zCAc in larval zebrafish was suggested by in situ hybridization results, in which zCAc appeared in specific cells of the skin and gill as well as spinal cord neurons and the pronephric duct, whereas zCAb appeared in blood cells of the yolk and tail (Lin et al., 2008). In the present study, the use of a morpholino oligonucleotide targeted to zCAb revealed that zCAb knock down resulted in a significantly lower RE ratio, a finding indicative of the inhibition of CO₂ excretion relative to O₂ uptake. Given localization of zCAb to blood cells and by analogy with the model of CO₂ excretion for adult fish, the relative inhibition of CO₂ excretion by zCAb knock down suggests that the RBC is involved in CO₂ excretion at this early stage of development, although convective delivery of CO₂ for elimination is likely to the body surface as a whole and/or the yolk sac surface rather than exclusively the gills (Rombough, 1988; Burggren and Pinder, 1991; Wells and Pinder, 1996; Rombough, 2004). By contrast, several studies using a variety of experimental approaches have demonstrated that O₂ uptake in developing zebrafish is not dependent on haemoglobin O₂ transport (via RBCs) until at least 12-14 d.p.f. (Pelster and Burggren, 1996; Weinstein et al., 1996; Jacob et al., 2002; Rombough and Drader, 2009; see review by Rombough, 2004) even though erythrogenesis is initiated around 15 h.p.f. and RBCs containing haemoglobin are present in the circulation by 24 h.p.f. (Weinstein et al., 1996; Brownlie et al., 2003). It is not clear why CO₂ excretion would rely on convective gas transport at a time when O₂ delivery does not. However, the apparent requirement for CA and possibly RBCs for effective CO₂ excretion in larvae may account for the appearance of haemoglobin-containing RBCs long before they are needed for O₂ delivery.

With this scenario, impairment of CO₂ excretion in larval zebrafish would be predicted to occur following selective inhibition of zCAb but not zCAc. Surprisingly, however, the use of a morpholino oligonucleotide targeted to zCAc also resulted in a significantly lower RE ratio, i.e. relative inhibition of CO₂ excretion. An explanation for this result is not obvious. The specificity of each morpholino oligonucleotide for the cytosolic CA isoform against which it was designed was confirmed experimentally. In addition, significantly lower levels of CA activity were measured for CA-morphant larvae than for controls, indicating that the morpholino oligonucleotides were effective in lowering CA activity. More specific confirmation of lowered protein levels for each isoform must await development of isoform-specific zCA antibodies. zCAc has been localized to the HR (proton pump-rich) cells in the skin and gills of larvae, where it is postulated to catalyze the hydration of CO₂ to produce protons that are pumped out across the apical membrane as part of the Na⁺ uptake mechanism (Hwang and Lee, 2007; Lin et al., 2008). Knock down of zCAc in this location would be expected to inhibit proton excretion, for which there is experimental support (Lin et al., 2008), but might if anything be expected to increase CO₂ excretion by reducing the rate at which molecular CO₂ is converted within the cell to protons and HCO₃⁻. It is important to note, however, that knock down of either cytosolic CA isoform is likely to affect the expression profiles of genes associated with CA isoform function, a phenomenon that can lead to confusing results. For example, Lin et al. (Lin et al., 2008) reported increased (rather than reduced) Na⁺ uptake by 96 h.p.f. larvae in which zCAc was knocked down, and attributed this apparently contradictory effect at least in part to zCAc knockdown-induced changes in the mRNA expression of several other genes involved in H⁺ excretion/Na⁺ uptake (a membrane-bound CA isoform, V-type H⁺-ATPase, and a Na⁺/H⁺ exchanger). This type of effect might also account for the interesting observation of elevated in CA morphants. Detailed and quantitative information on the patterns of CA isoform expression in the RBCs of larval zebrafish is also required. Characterization of globin genes in embryonic and larval zebrafish has revealed complex patterns of globin gene expression and globin switching through development (Brownlie et al., 2003). Whether analogous changes in RBC CA isoform expression occur during zebrafish development is unknown.

In conclusion, the results of the present study implicate CA in CO₂ excretion during early development in zebrafish. Whereas CO₂ excretion in embryos is low relative to O₂ uptake, possibly owing to conversion of CO₂ to HCO₃⁻ for use in urea synthesis and/or bone formation, hatching is associated with a relative enhancement of CO₂ excretion that is reliant, at least in part, on CA activity. The involvement of CA in CO₂ excretion in larval zebrafish may imply a role for the RBC and could help to explain the appearance of RBCs early in development. In particular, the selective advantage of early haemoglobin synthesis has not been obvious given that it is not needed for O₂ transport (Rombough and Drader, 2009). Based on the findings of the present study, i.e. that CA and possibly RBCs are involved in CO₂ excretion during early development in zebrafish, it is tempting to speculate that haemoglobin synthesis early in development may be driven by the role it plays in buffering H⁺ used in CO₂ hydration—dehydration reactions.

Luc Poitras and Marc Ekker are thanked for their kind gifts of the dTomato construct and pCS2+ vector. The assistance of Shelby Steele and Tammy Rodela in removing the egg capsule from embryos was appreciated. A.J.E. was the recipient of an NSERC postdoctoral fellowship, while K.T. received awards from the University of Ottawa and the NSERC USRA program.