A critical analysis of carbonic anhydrase function, respiratory gas exchange, and the acid-base control of secretion in the rectal gland of Squalus acanthias

We compared in vivo responses of rectal gland secretion to carbonic anhydrase (CA) inhibition (10^-4 mol l^-1acetazolamide) in volume-loaded dogfish with in vitro responses in an isolated-perfused gland stimulated with 5×10^-6 mol l^-1 forskolin and removed from systemic influences. We also measured respiratory gas exchange in the perfused gland, described the acid-base status of the secreted fluid, and determined the relative importance of various extracellular and intracellular acid-base parameters in controlling rectal gland secretion in vitro. In vivo, acetazolamide inhibited Cl^- secretion and decreased pHi in the rectal gland, but interpretation was confounded by an accompanying systemic respiratory acidosis, which would also have contributed to the inhibition. In the perfused gland,

\({\dot{M}}_{\mathrm{CO}_{2}}\)

and

\({\dot{M}}_{\mathrm{O}_{2}}\)

increased in linear relation to increases in Cl^- secretion rate. CA inhibition (10^-4 mol l^-1 acetazolamide) had no effect on Cl^- secretion rate or pHi in the perfused gland, in contrast to in vivo, but caused a transitory 30% inhibition of

\({\dot{M}}_{\mathrm{CO}_{2}}\)

(relative to stable

\({\dot{M}}_{\mathrm{O}_{2}}\)

⁠) and elevation in secretion P_CO2 effects, which peaked at 2 h and attenuated by 3.5-4 h. Secretion was inhibited by acidosis and stimulated by alkalosis; the relationship between relative Cl^-secretion rate and pHe was almost identical to that seen in vivo. Experimental manipulations of perfusate pH, P_CO2 and HCO₃^-concentration, together with measurements of pHi, demonstrated that these responses were most strongly correlated with changes in pHe, and were not related to changes in P_CO2, extracellular HCO₃^-, or intracellular HCO₃^-levels, though changes in pHi may also have played a role. The acid-base status of the secreted fluid varied with that of the perfusate, secretion pH remaining about 0.3-0.5 units lower, and changing in concert with pHe rather than pHi; secretion HCO₃^- concentrations remained low,even in the face of greatly elevated perfusate HCO₃^-concentrations. We conclude that pH effects on rectal gland secretion rate are adaptive, that CA functions to catalyze the hydration of CO₂,thereby maintaining a gradient for diffusive efflux of CO₂ from the working cells, and that differences in response to CA inhibition likely reflect the higher perfusion-to-secretion ratio in vitro than in vivo.

The rectal gland is a small finger-shaped organ in the posterior intestine of elasmobranchs. Its ionoregulatory and volume-regulatory functions were first characterized by Burger and Hess(Burger and Hess, 1960; Burger, 1962), who demonstrated that the rectal gland of the intact dogfish shark (Squalus acanthias)secreted an almost pure NaCl solution at a concentration of approximately 500 mmol l^-1, almost twice the NaCl level in blood plasma, but close to isosmotic with seawater and the urea-rich plasma. Subsequently, the gland has been used at many levels as a powerful in vitro model for understanding the mechanism and control of NaCl secretion. These analyses have included the use of in situ preparations in pithed, artificially ventilated dogfish, isolated-perfused whole gland preparations, perfused rectal gland tubules, cultured tubular epithelial cells, and membrane vesicles of the tubules (reviewed by Epstein et al.,1983; Shuttleworth,1988; Riordan et al.,1994; Silva et al.,1990; Silva et al.,1996; Silva et al.,1997; Olson, 1999; Hazon et al., 2003). The isolated-perfused rectal gland preparation, first developed by Hayslett et al.(Hayslett et al., 1974) and subsequently improved by many others, has been a particularly useful approach,in which secretion can be directly stimulated using forskolin(Moran and Valentich, 1991) to activate the adenyl cyclase pathway (e.g. Forrest, Jr et al., 1997; Walsh et al., 2006).

The following model has emerged. The oral intake of seawater or salty food during feeding (Mackenzie et al.,2002) is assumed to cause volume expansion in vivo(Solomon et al., 1984a; Solomon et al., 1984b; Solomon et al., 1985). The latter is known to mobilize C-type natriuretic peptide (CNP) from the heart(Silva et al., 1987), which stimulates increased rectal gland blood flow and secretion, by both direct and indirect mechanisms. The direct action seems to involve activation of both protein kinase C and guanyl cyclase signalling pathways(Silva et al., 1999). However,at least in Squalus acanthias, there is also a very potent indirect action, by which CNP causes the release of vasoactive intestinal polypeptide(VIP) from nerve endings in the rectal gland, and VIP in turn activates the adenyl cyclase pathway (Stoff et al.,1979; Stoff et al.,1988; Silva et al.,1987). A cascade of events (reviewed by Silva et al., 1997) is initiated by the various intracellular signals, in which apical CFTR-like channels are activated for Cl^- extrusion, together with basolateral Na⁺,K⁺,2Cl^- co-transporters (NKCC),Na⁺,K⁺-ATPase and K⁺ channels. The Na⁺ electrochemical gradient powers the NKCC-mediated entry of both Cl^- and K⁺. These ions exit via their respective apical and basolateral channels, while Na⁺ is secreted in equimolar amounts to Cl^- through a paracellular pathway, driven by the negative potential created by apical Cl^- extrusion. The net NaCl transport into the tubule entrains an isosmotic flux of water.

The rectal gland contains high carbonic anhydrase (CA) activity(Maren, 1967; Lacy, 1983) but its function has never been satisfactorily integrated into this model. Several in vitro studies with isolated-perfused glands have reported a complete lack of effect of CA inhibition on secretory performance(Siegel et al., 1975; Silva et al., 1977; Swenson and Maren, 1984). However, using an in situ preparation in a pithed, artificially ventilated whole dogfish, Swenson and Maren(Swenson and Maren, 1984)reported that CA inhibition reduced rectal gland secretion. They speculated that the normal role of CA is to facilitate the diffusive excretion of CO₂ from the metabolically active gland cells, in accord with the pioneering work of Gros in muscle tissue(Gros et al., 1976; Gros, 1991), but their results may have been confounded by the loss of neural control and depressed circulation in this preparation, as well as systemic effects of CA inhibition.

Recently we developed a method for assessing rectal gland performance in intact, unanaesthetized dogfish subjected to volume loading(Wood et al., 2006). Therefore, our first objective was to use this approach to further investigate the role of CA in the gland by examining the effect of CA blockade with acetazolamide in the intact animal on secretory performance, secretion acid-base status, and intracellular pH in the rectal gland. We followed this up with comparable studies on the effects of CA inhibition using the isolated-perfused rectal gland preparation to separate gland-specific effects from systemic effects of the CA blockade.

One advantage of the isolated-perfused preparation is that it allows the direct collection of not only the secreted fluid itself, but also the outflow through the rectal gland vein. Respiratory gas exchange by the secreting cells can therefore be directly measured, though to our knowledge, this has only been done for O₂ uptake (e.g. Silva et al., 1980), and not for CO₂ excretion. Our second objective was therefore to measure both O₂ and CO₂ exchange of the perfused gland, as well as the acid-base status of the secreted fluid, at rest, after activation with forskolin, under various acid-base manipulations, and during blockade of CA function with acetazolamide, to assess whether CA is involved in facilitating the diffusive excretion of CO₂, as suggested by Swenson and Maren(Swenson and Maren, 1984).

Several early studies reported that acidic perfusates inhibited secretion in the isolated-perfused gland in vitro(Siegel et al., 1975; Silva et al., 1992), and Swenson and Maren demonstrated that severe metabolic or respiratory acidosis inhibited secretion of their in situ gland preparation(Swenson and Maren, 1984). Recently, we have shown that the NaCl secretion rate of the rectal gland in intact unanaesthetized dogfish is very responsive to blood acid-base status in vivo, being stimulated by metabolic alkalosis, and inhibited by both respiratory acidosis and metabolic acidosis in volume-loaded animals(Wood et al., 2006). The nature of the acid-base stimulus (i.e. extracellular or intracellular pH,HCO₃^- level or P_CO2) was not determined in any of these investigations. Our final objective was therefore to use the forskolin-stimulated isolated-perfused preparation together with independent manipulations of perfusate P_CO2, HCO₃^- level and pH to analyze the precise nature of the extracellular and/or intracellular acid-base stimuli controlling secretion rate in the rectal gland.

Pacific spiny dogfish (Squalus acanthias L., 0.8-5.0 kg) were collected over the course of three summers from Barkley Sound, British Columbia, Canada. The fish were caught as bycatch in the trawls of commercial shrimp fishing. At Bamfield Marine Sciences Centre, they were held for a period of at least 1 week and up to 4 weeks prior to experimentation. Food(herring) was offered, but the fish would not feed in captivity. The outdoor holding tank (3000 l) was served with running seawater at the experimental temperature (11±1°C), salinity (29±2%thou) and pH(7.9±0.15).

In order to economize on radioisotopes, the smallest animals (0.3-1.0 kg)were selected for intracellular pH measurements, while larger dogfish (1.0-5.0 kg) proved more suitable for cannulation of the secretory duct of the rectal gland. Dogfish were anaesthetized with MS-222 (0.1-0.2 g l^-1;Syndel Labs., Vancouver, BC, Canada), weighed, and irrigated on an operating table. Dogfish intended for intracellular pH measurements (N=20) were fitted with only an indwelling mesenteric artery catheter (PE50 polyethylene tubing; Clay-Adams, Parsipanny, NJ, USA) by the method of Graham et al.(Graham et al., 1990), whereas those intended for secretion collections (N=13) were also fitted with a rectal gland duct catheter (PE50 or PE90 depending on animal size) by the method described (Wood et al.,2006). The arterial catheter was filled with dogfish saline[recipe as in Pärt et al. (Pärt et al., 1998), but with the omission of PVP-40 and albumin]containing 50 i.u. ml^-1 sodium heparin (Sigma-Aldrich, St Louis,MO, USA), and the rectal gland catheter with 500 mmol l^-1 NaCl to duplicate normal secretion composition. The animal was then revived and placed in an individual Plexiglas™ chamber served with aeration and a constant flow (>0.5 l min^-1) of fresh seawater. Dogfish were allowed to recover for at least 36 h before experiments commenced.

Rectal gland secretion flow proved to be negligible in these unfed resting dogfish. Therefore the animals were infused via the arterial catheter with 500 mmol l^-1 NaCl at a rate of 15 ml kg^-1h^-1 using a Minipuls peristaltic pump (Gilson, Middleton, WI, USA),France). This created a brisk, stable secretory flow from the gland, reaching a stable plateau from 1.5 to 6 h (cf. Wood et al., 2006), against which the effects of carbonic anhydrase inhibition could be assessed. The 1.5-3.0 h period served as the pre-treatment control, and the experiment was terminated at 6 h. Control animals(N=7) simply received the infusion alone for 6 h, whereas the experimental animals (N=6) received 20 mg kg^-1 of acetazolamide (as the HCl salt, Wyeth-Lederle, Madison, NJ, USA) in 0.5 ml kg^-1 dogfish saline via the arterial catheter at 3 h, and the experiment continued with 500 mmol l^-1 NaCl infusion up to 6 h. This acetazolomide dose was calculated to produce an in vivocirculating concentration of about 10^-4 mol l^-1,representing the level generally recognized as the maximum safe dose needed to achieve significant CA inhibition while avoiding non-specific effects(Maren, 1977). This concentration has been commonly used to produce effective CA inhibition in previous shark studies (Gilmour et al.,1997; Perry et al.,1999; Wilson et al.,2000; Gilmour and Perry,2004). Rectal gland secretion flow was collected over successive 0.5 h periods, quantified gravimetrically, analysed immediately for acid-base status (pH and total CO₂) and frozen at -20°C for later ionic analysis. Blood samples (0.6 ml) were drawn from the arterial catheter at 1 h intervals, and analysed immediately for arterial blood gases and pH, while plasma was separated (2 min at 9000 g) and frozen (-20°C)for later ion measurements. The red cell pellet was resuspended in dogfish saline, combined with blood recovered from the electrodes, and the original volume re-infused, so the impact of blood sampling on hematocrit was minimal.

Intracellular pH was measured in the rectal gland as well as in white muscle (as a point of reference), together with extracellular acid-base status and fluid volume distribution in these tissues, in three treatments. These were unstimulated control (N=8), stimulated control (N=6),and stimulated plus acetazolamide-treated (N=6). In the first treatment, no infusion was given. The stimulated control (infusion with 500 mmol l^-1 NaCl for 6 h) and stimulated acetazolamide treatments(infusion with 500 mmol l^-1 NaCl for 6 h with administration of 20 mg kg^-1 of acetazolamide at 3 h) were identical to the comparable treatments in the secretion measurement experiments.

The DMO (5,5-dimethyloxazolidine-2,4-dione) method(Waddell and Butler, 1959) was employed, using mannitol as the extracellular fluid volume marker. The two radiolabelled compounds - [³H]mannitol (28 μCi kg^-1,specific activity 27.4 mCi mmol^-1) and [¹⁴C]DMO (7μCi kg^-1, specific activity 50.0 mCi mmol^-1), both from Dupont-NEN, Boston, MA, USA - were administered via the arterial catheter in 1 ml kg^-1 of 500 mmol l^-1 NaCl approximately 12 h prior to sampling to allow adequate equilibration time(Munger et al., 1991). Immediately prior to sacrifice, an arterial blood sample (1.0 ml) was drawn for blood gas and pH measurements, with a portion of plasma immediately separated by centrifugation (2 min at 9000 g) and divided into aliquots for radioactivity measurements. The dogfish was then killed by a sharp cephalic blow and spinal section. The rectal gland and a section of epaxial muscle were immediately removed, blotted, weighed and processed for radioactivity measurements as described below. The capsule was removed from the rectal gland prior to processing.

Isolated-perfused rectal gland preparations (N=75) were made from dogfish in the 1.5-3.5 kg range. Each dogfish was anaesthetized with MS-222(0.2 g l^-1), injected with 5000 i.u kg^-1 of sodium heparin (Sigma-Aldrich, St Louis, MO, USA) via the caudal haemal arch, then killed by pithing with a steel wire inserted via the rostrum through the brain and spinal cord. The rectal gland artery, vein and duct were cannulated in situ as described(Shuttleworth, 1983) using PE50 polyethylene tubing (Clay-Adams, Parsipanny, NJ, USA). The arterial and venous catheters were filled with standard dogfish saline, with the following composition: 257 mmol l^-1 NaCl, 7 mmol l^-1Na₂SO₄, 6 mmol l^-1 NaHCO₃, 4 mmol l^-1 KCl, 3 mmol l^-1 MgSO₄.7H₂O, 2 mmol l^-1 CaCl₂.2H₂O, 0.1 mmol l^-1Na₂HPO₄, 30 mmol l^-1 glucose, 20 mmol l^-1 trimethylamine oxide, 400 mmol l^-1 urea. The saline was passed through a 0.45 μm filter and equilibrated with a precision gas mixture (0.256% CO₂, 99.744% O₂) so as to achieve control acid-base conditions (see below). The rectal gland duct catheter was filled with 500 mmol l^-1 NaCl. The preparation was then dissected free and placed on a thermostatted platform.

Perfusion with control saline was initiated using the peristaltic pump/overflow system (Shuttleworth,1983) to maintain an inflow pressure of ∼20 mmHg. A windkessel reduced pressure pulsatility to ∼3 mmHg and served as a bubble trap. The perfusion reservoirs were fitted with gassing ports and a combination pH electrode (GK2401C, Radiometer-Copenhagen, Copenhagen, Denmark) for monitoring of gas equilibration. The reservoirs and the perfusion platform were jacketed with flowing seawater and the perfusion lines were immersed so as to maintain the experimental temperature at 11±1°C. Perfusion pressure was monitored by a P23 dB pressure transducer (Statham, Hato Rey, Puerto Rico)connected via a sidearm to the perfusion line immediately proximal to the gland. This T-junction also served as an arterial sampling point. A second T-junction just distal to the gland served as the venous sampling point. The venous outflow and the secretion outflow from the rectal gland duct were set to 0 mmHg relative to the gland, and drained through infrared drop-sensors connected to custom-built microprocessor-controlled digital flowmeters, so as to provide continuous flow records. The venous outflow (which is always less than the arterial inflow) represents the actual perfusion flow through the secretory parenchyma of the gland, whereas various non-secretory shunt pathways drain diffusely (Kent and Olsen,1982). Therefore determinations of O₂ consumption(⁠

⁠) and CO₂ production(⁠

\({\dot{M}}_{\mathrm{CO}_{2}}\)

⁠)rates by the perfused gland employed the Fick principle using the measured venous outflow rate and simultaneous arterial and venous measurements of total O₂ or CO₂ concentrations (see below).

After a series of pilot experiments, the following general protocol was established. The preparation was first perfused at ∼20 mmHg (1 mmHg=133.3 Pa) with standard saline equilibrated with the 0.256% CO₂, 99.744%O₂ gas mixture so as to establish control acid-base conditions of P_CO2 ∼1.9 mmHg, pH ∼7.8,HCO₃^- ∼5.3 mmol l^-1 and a P_O2 >400 mmHg. After 0.75 h, samples of the rectal gland secretion, venous perfusate and arterial perfusate were collected for gas and acid-base analyses, and the secretion was frozen (-20°C) for the later analysis of ions. These sampling procedures took 15 min, and then at 1 h, 5×10^-6 mol l^-1 forskolin (Sigma-Aldrich) was added to the perfusion reservoir to stimulate secretion(Moran and Valentich, 1991; Forrest, Jr et al., 1997; Walsh et al., 2006). After a further 0.75-1 h, the sampling and analyses were repeated, and then the perfusion pressure was lowered to ∼12 mmHg, a procedure chosen to decrease the perfusion flow through the gland without altering secretion flow so as to increase the precision of measurement of arterial-venous differences (see Results). A third set of sampling and analyses was performed after a further 0.75-1 h; these represented the pre-treatment control measurements. The experimental treatment was then implemented, while maintaining stimulation with 5×10^-6 mol l^-1 forskolin. In later preparations, only the pre-treatment control measurements were performed so the first two sets of analyses were omitted, but the gland was put through the same protocol so as to maintain consistency with earlier preparations.

using the solubility of carbon dioxide (α_CO2)and the apparent pK (pK_app) for dogfish at the experimental temperature (Boutilier et al.,1984). When NaHCO₃ was experimentally elevated, NaCl was reduced on an equimolar basis, and vice versa so as to maintain osmolality. Alterations in P_CO2 were achieved by gassing the perfusate with the output from a 301-af gas mixing pump(Wosthof, Bochum, Germany), in which the mixing pistons were fed from various certified mixtures of CO₂ in O₂.

Experimental treatments lasted routinely 2 h, except for acetazolamide where the experimental period was 4 h. Sampling and analysis of rectal gland secretion, venous perfusate, and arterial perfusate were performed at 1 h intervals, except in the acetazolamide treatment where they were performed at 0.5 h intervals. In two series (control acid-base conditions, ∼12 mmHg,5×10^-6 mol l^-1 forskolin; 10^-4 mol l^-1 acetazolamide, ∼12 mmHg, 5×10^-6 mol l^-1 forskolin), at the end of the experiment, a portion of the gland tissue was immediately freeze-clamped in liquid N₂ and stored at -80°C for later determination of tissue buffer capacity. In addition,in three experiments (glands from three different series), samples of the secreted fluid and venous effluent were immediately fixed in two volumes of ice-cold 8% HClO₃ for the assay of lactate, and the glands were freeze-clamped and stored as above. This tissue was later homogenized in ten volumes of ice-cold 8% HClO₃ for lactate analysis.

The DMO method (Waddell and Butler,1959), with mannitol as the extracellular fluid volume marker, was again employed, and measurements were made in each of the acid-base series,and in the acetazolamide treatment. Additional series (N=5-6) were run to measure fluid volume distribution and intracellular pH in the two treatments at the start of the perfusion protocol prior to lowering of the perfusion pressure: (i) non-stimulated preparations (control acid-base conditions, ∼20 mmHg, no forskolin); and (ii) stimulated preparations(control acid-base conditions, ∼20 mmHg, 5×10^-6 mol l^-1 forskolin).

[¹⁴C]DMO (12.5 μCi l^-1) and[³H]mannitol (50 μCi l^-1), both from NEN-Dupont,Boston, MA, USA, were added to the perfusion reservoirs 2 h before terminal sampling - i.e. at the start of most experimental treatments or after 2 h of acetazolamide treatment. Pilot experiments demonstrated that a time period of 2 h was sufficient to achieve full equilibration. At termination, immediately after the final set of perfusion analyses, the rectal gland was blotted and weighed, and the capsule was removed prior to processing for radioactivity and water content measurements.

The same methods were used for both in vivo and in vitrosamples. Notably, the response time of this [¹⁴C]DMO technique is less than 15 min, even in poorly perfused trout trunk muscle(Milligan and Wood, 1985),whereas a minimum of 2 h in vitro and 3 h in vivo was allowed for label redistribution in the present study on the much better perfused rectal gland. Triplicate samples (50-150 mg each) of the tissue were used for radioactivity measurements, and the remaining tissue was dried to a constant mass at 70°C in order to determine its total water content. Plasma water content was determined by refractometry using a Goldberg refractometer (American Optical TS meter, Buffalo, NY, USA) recalibrated for dogfish plasma. For radioactivity measurements, samples were added to 2 ml NCS(Amersham, Piscataway, NJ, USA) digest medium and incubated at 35-40°C for 12 h in sealed glass scintillation vials. The clear digests were then neutralized with glacial acetic acid, diluted with 10 ml OCS fluor (Amersham),and stored in the dark overnight to reduce chemiluminescence prior to scintillation counting for [¹⁴C]DMO and [³H]mannitol d.p.m. on an LKB Rackbeta 1217 (LKB-Wallac, Turku, Finland). Triplicate plasma samples from in vivo experiments, or inflow and outflow perfusate samples from in vitro experiments (100 μl each) were similarly processed and counted. Separation of ³H and ¹⁴C d.p.m. was achieved using on-board dual label quench-correction programs in the Rackbeta 1217, calibrated by the external standard ratio method. The coordinates of the programs were experimentally generated using a range of quenched samples from perfusate, plasma, rectal gland, and muscle tissue in the NCS/OCS system. Recovery of d.p.m. was regularly checked by spiking with known amounts of [³H]mannitol and [¹⁴C]DMO (internal standardization).

Tissue intracellular pHi, [HCO₃^-]_i,extracellular fluid volume (ECFV) and intracellular fluid volume (ICFV) were calculated from measurements of extracellular acid-base status, water contents of tissue, plasma or perfusate, and [¹⁴C]DMO and[³H]mannitol radioactivities in tissue, and plasma or perfusate,using equations given by Wright et al.(Wright et al., 1988). For in vivo experiments, arterial measurements of all parameters were used. For in vitro experiments, extracellular pH and P_CO2 were taken as the average of arterial and venous values, and inflow and outflow [¹⁴C]DMO and[³H]mannitol d.p.m. were similarly averaged for these calculations. The pK of DMO at the experimental temperature was interpolated from the measurements of Heisler et al. (Heisler et al., 1976) at comparable ionic strength.

In two series, rectal gland tissue buffer capacity was measured by an acid titration of tissue homogenate using the protocol described(Wood et al., 1990). The slope of the curve relating mmol HCl added versus pH was linear over the pH range 7.0-7.8 (the operative pHi range in vivo) and taken as the buffer capacity of the tissue in mmol pH unit^-1 kg^-1. This was then converted to mmol pH unit^-1 l^-1intracellular water (slykes), taking into account the measured intracellular fluid volume. This technique measures total physico-chemical buffer capacity(i.e. non-HCO₃^- plus HCO₃^-buffering). However, as the original HCO₃^- content of the tissues is low, and the P_CO2 is close to zero during titration, the HCO₃^- component is negligible, and the results in essence yield an estimate of the non-HCO₃^- buffer capacity.

For in vivo analyses, blood samples were withdrawn viathe arterial catheter into ice-cold, gas-tight Hamilton syringes (Reno, NV,USA). Arterial blood pH (pHa) and oxygen tension(Pa_O2) were measured using Radiometer micro-electrodes (Radiometer-Copenhagen, Copenhagen, Denmark) kept at the experimental temperature with water jackets. Electrode outputs were displayed on Radiometer pHM 71 and pHM 72 acid-base analysers. True plasma CO₂ was measured on plasma obtained from blood samples centrifuged in sealed tubes. In some experiments, the measurements were made using a Corning 965 total CO₂ analyzer (Loughborough, UK) and in others the method of Cameron (Cameron,1971); the two yielded identical results, though the former was more rapid. The same analytical methods (pH, total CO₂) were used for rectal gland secretions. Because of the low buffer capacity of the secretions, it was necessary to flush the pH electrode capillary 3 times with the sample over a 2 min period followed by a further 2 min of equilibration to achieve stable, reproducible values. Carbon dioxide tensions(P_CO2) and bicarbonate concentrations([HCO₃^-]) were calculated using the solubility of carbon dioxide (α_CO2), the apparent pK (pK_app)for dogfish at the experimental temperature, and rearrangements of the Henderson-Hasselbalch equation (Boutilier et al., 1984).

For in vitro analyses, venous followed by arterial perfusate samples (typically 300 μl) were obtained anaerobically from the outflow and inflow ports of the perfusion set-up, by gentle withdrawal into ice-cold,gas-tight Hamilton syringes. The rectal gland secretion was diverted from the drop counter into the needle of a Hamilton syringe; the secretion was allowed to overflow directly through the open syringe barrel for 10 min, which was then sealed to yield a sample with negligible air exposure. Arterial (pHa),venous (pHv), and secretion pH, arterial (Pa_O2)and venous oxygen tensions (Pv_O2), and arterial(Ca_CO2) and venous(Cv_CO2) total CO₂ concentrations,were measured using the same equipment as for the in vivo samples. Again, because of the low buffer capacity of the perfusate and secretion fluid, it was necessary to pre-condition the pH capillary electrode when reading these samples, as outlined above. Pa_O2and Pv_O2 values were converted to total O₂ concentrations (Ca_O2, Cv_O2) using tabulated solubility coefficients(Boutilier et al., 1984).

Na⁺ and Cl^- in blood plasma and rectal gland secretions were measured by atomic absorption spectrophotometry (Varian 1275 AA, Mulgrave, Victoria, Australia) and coulometric titration (Radiometer CMT-10, Radiometer-Copenhagen, Copenhagen, Denmark), respectively. For lactate analyses of fluid and gland samples, the deproteinized HClO₃extracts were neutralized and assayed enzymatically by the l-lactate dehydrogenase method(Bergmeyer, 1983) using reagents from Sigma-Aldrich.

Data are reported as means ± 1 s.e.m. (N). When a paired experimental design was used, single paired t-tests, or Dunnett's paired multiple comparison test, as appropriate, were used to compare treatment values back to pre-treatment measurements in the same glands or the same animals, or measurements of two different parameters in the same preparation. Specific differences between treatments were evaluated using Student's unpaired two-tailed t-test, with the Bonferroni correction applied when multiple comparisons were made. For comparisons across a range of treatments, data were first tested for normality and homogeneity of variances,and then analyzed by one-way ANOVA, followed by Tukey's honestly significant difference test to detect specific differences. Regression lines were fitted by linear and exponential models, and the significance of the correlation coefficient assessed. A significance level of 0.05 was used throughout.

Relative to a very stable rectal gland secretion rate in control infused dogfish, injection of 20 mg kg^-1 of acetazolamide at 3 h caused a progressive decrease in secretion output of the rectal gland, which became significant by 4.5-5 h (Fig. 1A). However, there was also an accompanying fall in pHa(Fig. 1B) and rise in Pa_CO2 (Fig. 1C) and plasma HCO₃^- levels(Fig. 1D), all of which became significant by 4 h. By 6 h, pHa had decreased by ∼0.35 units to ∼7.4 due to a tripling of Pa_CO2 to ∼4 mmHg, and rectal gland secretion rate had fallen by 53%. Na⁺ and Cl^- concentrations in the secretion (both ∼520 mmol l^-1) were not altered, but its pH was significantly depressed(6.85±0.16 at 5.5-6 h versus 7.30±0.07 in the control fish) and its P_CO2 elevated (4.38±2.01 versus 0.90±0.10 mmHg) by a comparable amount to the elevation in Pa_CO2. Secretion HCO₃^-concentration was only marginally increased (0.99±0.19 versus0.77±0.18 mmol l^-1 in the control treatment). These changes in secretion P_CO2 and HCO₃^- concentration accompanying acetazolamide treatment were significant relative to the pre-treatment period in the same animals, but not relative to the control group. Plasma Na⁺ and Cl^-levels (both ∼320 mmol l^-1) and Pa_O2 (∼90 mmHg) were unaffected (data not shown).

Intracellular pHi and fluid volume distribution were measured in three experimental treatments in vivo: unstimulated (non-infused) control,stimulated control (infused), and stimulated (infused) plus acetazolamide-treated (Table 1). In these smaller dogfish, infusion alone caused a small decrease in pHa, probably associated with dilution of plasma HCO₃^- (i.e. slight metabolic acidosis). Acetazolamide treatment caused the expected respiratory acidosis in the extracellular fluid. Intracellular pHi in the rectal gland and white muscle tended to fall with infusion, though the difference was only significant for the latter(Table 1). However the combination of infusion plus acetazolamide caused a significant intracellular acidosis in the gland tissue. Intracellular HCO₃^-concentration in the rectal gland rose significantly with acetazolamide treatment of volume-loaded fish, whereas there were no differences relative to non-infused fish. Similar trends were seen with intracellular HCO₃^- concentrations in the white muscle. There was also an increase in total rectal gland water content and ECFV in both treatments,as well as an increase in rectal gland ICFV in the infusion + acetazolamide treatment. Muscle water and ICFV increased in the infusion + acetazolamide treatment.

In the unstimulated rectal gland preparation perfused initially at an inflow pressure of 20 mmHg, there was no detectable secretion flow (<2μl g^-1 gland min^-1). After stimulation with 5×10^-6 mol l^-1 forskolin, perfusion flow increased modestly (by 18%), and secretion flow was activated at a brisk rate (25-30μl g^-1 gland min^-1). This was accompanied by increases in the gradients of pH, P_CO2, total O₂ and total CO₂ between arterial inflow and venous outflow, achieved almost entirely by changes in the venous outflow values(Table 2). The activation of secretion was therefore accompanied by 3-4-fold increases in

and

\({\dot{M}}_{\mathrm{O}_{2}}\)

(Table 2). Notably

\({\dot{M}}_{\mathrm{CO}_{2}}\)

was approximately twice as large as

\({\dot{M}}_{\mathrm{O}_{2}}\)

in both treatment conditions, but the measurements of

\({\dot{M}}_{\mathrm{CO}_{2}}\)

were of lower precision, because of the difficulty of resolving small arterial-venous differences against the background total CO₂concentrations in the perfusate.

To increase the precision of measurement of arterial-venous differences in the stimulated gland, the perfusion pressure was decreased to approximately 12 mmHg, which reduced the perfusion flow rate by 34%(Table 2). There was no significant decrease in rectal gland secretion rate, but the arterial-venous gradients in pH, P_CO2, total O₂, and total CO₂ all increased. Again, these were achieved largely by changes in the venous outflow values, though there were also some small differences in arterial inflow values.

and

\({\dot{M}}_{\mathrm{O}_{2}}\)

did not change significantly, but the discrepancy between them was reduced(Table 2). Notably, while the pH of the secretion was well below the venous outflow level, its P_CO2 remained fairly close to arterial rather than venous values. Secretion total CO₂ concentrations were only about 20% of those in the perfusate. In all cases, HCO₃^-concentrations (data not shown) remained similar to but slightly smaller than total CO₂ concentrations in perfusate and secretion samples. Pilot experiments demonstrated that all parameters remained approximately stable for up to 6 h of perfusion under these conditions.

Intracellular pHi (terminal measurements) and[HCO₃^-]i, together with fractional fluid volume distributions, were determined in separate groups of glands under the three perfusion conditions (Table 3). Total O₂ and CO₂ concentrations, as well as acid-base characteristics of both inflowing and outflowing perfusate (data not shown),did not differ significantly from the values for the respective treatments in Table 2. Intracellular pHi was about 0.3 units below mean extracellular pHe at rest, and did not change significantly when the glands were stimulated with 5×10^-6 mol l^-1 forskolin, or when inflow pressure was reduced in the continued presence of forskolin. However, all these treatments exhibited a significantly greater [HCO₃^-]i relative to the unstimulated control. The stability of pHi was seen despite the observed decreases in pHe(Table 3) under these conditions due to the falls in pHv (cf. Table 2). Fractional water content remained unchanged but was clearly re-distributed, with ECFV increasing, and ICFV decreasing reciprocally after stimulation with forskolin,a situation which persisted after reduction of the inflow pressure.

There was a very strong linear relationship (r=0.99, N=127, P<0.0001) between secretion flow rate and Cl^- secretion rate in the perfused, forskolin-stimulated rectal glands, over a wide secretory range (Fig. 2A). The slope of the relationship indicated a mean Cl^-concentration of 520 mmol l^-1 in the secretion, which may be compared with 270 mmol l^-1 in the perfusate. While data were initially analysed according to whether the gland was at high pressure(∼20 mmHg), reduced pressure (∼12 mmHg), or subjected to a particular experimental treatment, it became clear that a single linear relationship fitted all the data. In those samples (N=32) where Na⁺concentration was monitored simultaneously, it was indistinguishable from Cl^- concentration (both ∼520 mmol l^-1). Because of this consistency, Cl^- secretion rate was used as the indicator of secretory performance in all further analyses, as in vivo.

Again using all the data points, including those obtained in the absence of forskolin, there was a strong linear relationship (r=0.73, N=139, P<0.0001) between Cl^- secretion rate and the simultaneous rate of oxygen consumption(⁠

⁠) of the perfused glands (Fig. 2B). The relationship was not altered or improved by considering any of the perfusion conditions or experimental treatments separately. The intercept at zero Cl^- secretion was positive, indicating that unstimulated glands consume O₂ at a rate of about 0.11 μmol g^-1gland min^-1, in accord with the mean measured value (0.11 μmol g^-1 gland min^-1) of Table 2.

The rate of carbon dioxide excretion(⁠

⁠) by the perfused glands was more difficult to measure, and essentially undetectable when high HCO₃^- salines were used because it was impossible to resolve small arterial-venous differences in total CO₂ concentrations. The analysis of

\({\dot{M}}_{\mathrm{CO}_{2}}\)

(versus Cl^- secretion rate) therefore uses the same set of data points as for the analysis of

\({\dot{M}}_{\mathrm{O}_{2}}\)

⁠, but without those obtained with high HCO₃^- saline, or under acetazolamide treatment (which could potentially inhibit CO₂excretion). The relationship between Cl^- secretion rate and

\({\dot{M}}_{\mathrm{CO}_{2}}\)

was much more variable than for

\({\dot{M}}_{\mathrm{O}_{2}}\)

⁠, but still highly significant (r=0.48, N=117, P<0.0001; Fig. 2B). The intercept at zero Cl^- secretion was 0.22 mol g^-1gland min^-1, comparable to the mean measured value (0.20 mol g^-1 gland min^-1) for unstimulated glands in Table 2. However, the two regression lines were not statistically different, and there was no significant difference when all simultaneous measurements of

\({\dot{M}}_{\mathrm{CO}_{2}}\)

versus

\({\dot{M}}_{\mathrm{O}_{2}}\)

were compared (N=117; paired t-test).

Lactate concentrations measured in three perfused glands were 7.15±1.19 mmol kg^-1; lactate concentrations in the secreted fluid and venous effluent were below the resolution of the assay (<0.06 mmol l^-1).

Addition of 10^-4 mol l^-1 acetazolamide to the perfusate had no significant effect on Cl^- secretion rate(Fig. 3A),

(Fig. 3B), or perfusion rate(data not shown) in the stimulated rectal gland. However

\({\dot{M}}_{\mathrm{CO}_{2}}\)

exhibited a transient decline of about 30%, which became significant at 2 h,2.5 h and 3 h, but recovered by 4 h (Fig. 3C). The gas exchange ratio(⁠

\(\mathrm{R}={\dot{M}}_{\mathrm{CO}_{2}}{/}{\dot{M}}_{\mathrm{O}_{2}}\)

⁠)declined significantly from 1.41±0.06 under control conditions to 1.04±0.17 at 2 h and 1.01±0.22 at 2.5 h, and then recovered to 1.30±0.08 by 4 h. Pa_CO2 (controlled experimentally) and Pv_CO2 did not change significantly, but the P_CO2 of the secretion was significantly elevated by about 1.5 mmHg at most times from 1.0 to 3.5 h of acetazolamide treatment (Fig. 4). There were no significant changes in pHa, pHv,[HCO₃^-]a, or secretion pH, whereas[HCO₃^-]v rose in mirror image to the transient decline in

\({\dot{M}}_{\mathrm{CO}_{2}}\)

⁠,and secretion HCO₃^- levels increased slightly in accord with increases in secretion P_CO2 (data not shown).

Intracellular pH (pHi) and [HCO₃^-]i of the rectal gland, measured at the end of the 4 h acetazolamide treatment, remained unchanged relative to the control value, and fractional water content, ECFV and ICFV were also unaffected (Table 3).

The non-HCO₃^- buffer capacity of the forskolin-activated gland perfused under control conditions (normal perfusate P_CO2 and HCO₃^-, 12 mmHg)was 35.28±2.10 (6) slykes per unit intracellular fluid[16.83±0.79 (6) mmol pH unit^-1 kg^-1 on a whole tissue basis] and this value remained unchanged at 33.97±1.03 (6)slykes [16.87±0.32 (6) mmol pH unit^-1 kg^-1] after treatment with 10^-4 mol l^-1 acetazolamide for 4 h.

P_CO2 and [HCO₃^-] in the inflowing perfusate were experimentally manipulated in order to discern the relative importance of extracellular and intracellular pH,[HCO₃^-], and/or P_CO2, in controlling the rate of secretion in the stimulated rectal gland. The measured acid-base status of the inflowing and outflowing perfusate and estimated[HCO₃^-]i for all treatments are detailed in Table 4, and the secretory response of the rectal gland relative to pHe or pHi is illustrated in Fig. 5. The control glands(control P_CO2, control[HCO₃^-]) were simply put through the same manipulations but without change in the acid-base status of the inflowing perfusate, and exhibited a 14% decline in secretion rate. There were no significant changes in fractional water content, ECFV or ICFV of the glands in any of the treatments (data not shown), mean values staying close to the control data of Table 3.

Approximately equal depressions of both pHe (by ∼0.3 units) and pHi (by about 0.15 units) were accomplished in one group by raising P_CO2 at control [HCO₃^-],and in another by lowering [HCO₃^-] at control P_CO2. Both treatments resulted in comparable marked inhibitions (by ∼50%) of rectal gland secretion, despite very different changes in intracellular and extracellular[HCO₃^-] and P_CO2,suggesting that pH rather than P_CO2 or[HCO₃^-] was the important factor(Fig. 5, Table 4). This conclusion was reinforced by a treatment where both P_CO2 and[HCO₃^-] were lowered so as to maintain pHe and pHi unaltered. Rectal gland secretion rate remained unchanged relative to the control treatment despite the changes in intracellular and extracellular[HCO₃^-] and P_CO2. Further support for this conclusion and some distinction between the importance of pHi and pHe were obtained by a treatment where both P_CO2 and [HCO₃^-] were raised. Rectal gland secretion rate again remained unchanged despite a marked rise in [HCO₃^-]i and a marked fall in pHi, the latter equal to that seen in the two treatments that had caused about a 50%inhibition of rectal gland secretion rate(Fig. 5, Table 4). This suggests that pHe, rather than pHi, may be the key controlling factor. Notably, however, pHe was significantly elevated in this treatment. When pHe was further raised to about 1.0 unit above control pHe, by increasing [HCO₃^-]at control P_CO2, a treatment that did not significantly alter pHi or [HCO₃^-]i, there was a dramatic stimulation (2.3-fold) of rectal gland secretion rate(Fig. 5, Table 4). A smaller increase in pHe (∼0.6 units) achieved by reduction of P_CO2 at control [HCO₃^-]resulted in a very similar pHi, but a reduced [HCO₃^-]i,and a secretion rate which was slightly elevated but not-significantly different from the control. Overall, these results suggest that pHe, rather than P_CO2, or intracellular or extracellular[HCO₃^-], may be the dominant influence on rectal gland secretion rate, but do not eliminate a role for pHi (see Discussion).

As expected from the relationship in Fig. 2B, changes in rectal gland

varied in parallel to changes in rectal gland secretion rate in the various experimental treatments (data not shown). However, on a relative basis they tended to be smaller - e.g. ∼24% versus ∼50% inhibitions in the two extracellular acidosis treatments, 1.9-fold versus 2.3-fold stimulation in the extracellular alkalosis treatment, and only the latter change in

\({\dot{M}}_{\mathrm{O}_{2}}\)

was statistically significant. This reflected the fact that a portion of rectal gland

\({\dot{M}}_{\mathrm{O}_{2}}\)

is devoted to routine metabolism and occurs even in the absence of secretion,as well as the greater variability in

\({\dot{M}}_{\mathrm{O}_{2}}\)

measurements (Table 2, Fig. 2B).

The acid-base characteristics of the secretions from the rectal glands subjected to the various acid-base manipulations are summarized in Table 5. In general, secretion pH was about 0.3-0.5 units below pHa and a more variable amount (0.1-0.5 units) below pHv. Secretion pH declined in those treatments where the extracellular fluid was acidotic, and increased in those treatments where it was alkalotic, though only some of these changes were statistically significant. Thus secretion pH correlated with changes in pHe rather than pHi:note for example, the unchanged secretion pH associated with the high P_CO2, high [HCO₃^-]treatment (where pHi was depressed), and the greatly elevated secretion pH associated with the control P_CO2, high[HCO₃^-] treatment where pHi was unchanged. Changes in secretion P_CO2 reflected changes in the Pa_CO2 of inflowing perfusate, but were generally smaller than those in Pa_CO2. In all cases (Table 5), secretion P_CO2 remained well below Pv_CO2 (Table 4). Secretion HCO₃^- concentrations reflected changes in perfusate HCO₃^- concentrations, but remained only a small fraction (8-42%) of the latter.

Systemic treatment with 20 mg kg^-1 acetazolamide (calculated to produce an in vivo concentration of about 10^-4 mol l^-1) caused a profound inhibition of rectal gland secretion (by 53±6% after 3 h) in the intact, unanaesthetized dogfish(Fig. 1A). In general, this result is similar to that of Swenson and Maren(Swenson and Maren, 1984) on the in situ gland preparation in pithed dogfish subjected to a similar volume-loading protocol, where the degree of inhibition by another CA blocker (methazolamide) was about 50%. However, key points differ. In the present study, there was also a significant fall in rectal gland pHi(Table 1) accompanying CA inhibition, a point to which we return later in our discussion of role of carbonic anhydrase in rectal gland function. Swenson and Maren(Swenson and Maren, 1984) did not measure pHi, but reported a paradoxical 40% decrease in rectal gland CO₂ content after CA inhibition (from 18 to 11 mmol kg^-1), for which the mechanism remained unexplained. In the present study, total gland CO₂ content was not measured directly, but[HCO₃^-]i in the rectal gland increased significantly(from 2.08 to 3.15 mmol l^-1 ICF; Table 1), as might be expected during respiratory acidosis. The gross discrepancy in absolute values and opposite direction of the changes between the two studies suggests that the in situ preparation (Swenson and Maren, 1984) may have been severely perfusion-limited due to a depressed circulation resulting in CO₂ retention, such that methazolamide treatment of the whole animal put the gland into metabolic depression or metabolic acidosis. Swenson and Maren attempted a more specific inhibition of gland CA alone using a low dose of benzolamide(Swenson and Maren, 1984), but a systemic P_CO2 rise still occurred; they reported a small fall in secretion rate and a large rise in secretion P_CO2. While the present results on the intact animal appear more coherent, in neither study can we attribute the decreases in secretion rate specifically to an effect on rectal gland CA. This is because the accompanying systemic respiratory acidosis, which has been demonstrated in previous studies on CA inhibition in Squalus acanthias (e.g. Swenson and Maren,1984; Swenson and Maren,1987; Gilmour et al.,1997; Gilmour et al.,2001; Perry et al.,1999; Gilmour and Perry,2004), could have contributed a major portion of the effect. Indeed Swenson and Maren (Swenson and Maren, 1984) demonstrated inhibitory effects on secretion of severe respiratory and metabolic acidoses alone (without CA inhibition), and fully recognized this complication. Recently, we have established a relationship between pHa and secretion rate in intact Squalus acanthias subjected to an identical volume-loading protocol, and in which acid-base status was experimentally manipulated without using acetazolamide(Wood et al., 2006). Based on this regression, the fall in pHa (to 7.455±0.034; Fig. 1B) alone caused by acetazolamide would have caused a 31% inhibition of secretion, whereas the actual inhibition by acetazolamide in the intact animal was 53±6%,right at the 95% confidence limit of the relationship. We therefore turned to the in vitro perfused gland preparation to more clearly dissect these influences.

The present results confirm that secretion in the isolated-perfused rectal gland is very sensitive to acid-base status, in agreement with several early studies on the perfused gland in vitro(Siegel et al., 1975; Silva et al., 1992), on the in situ gland preparation in the pithed, artificially ventilated dogfish (Swenson and Maren,1984), and our recent investigation on the intact, unanaesthetized dogfish (Wood et al., 2006). However our data extend these studies by dissecting out the influences of various extracellular and intracellular acid-base parameters. As noted in Results, the results overall point to pHe as the dominant influence, but pHi may also be involved. Thus the inhibitory action associated with low pHe occurred to the same extent regardless of whether pHe was reduced by raising P_CO2 at control [HCO₃^-]e and [HCO₃^-]i (respiratory acidosis) or by lowering[HCO₃^-]e and [HCO₃^-]i at control P_CO2 (metabolic acidosis)(Fig. 5). Furthermore, lowering both P_CO2 and [HCO₃^-]e(and [HCO₃^-]i) so as to leave pHe unchanged had no effect, as did lowering only intracellular pHi (and not pHe) by raising both P_CO2 and [HCO₃^-]e (and[HCO₃^-]i). Raising pHe (and not pHi or[HCO₃^-]i) by high [HCO₃^-]e treatment at normal P_CO2 (metabolic alkalosis)markedly stimulated secretion, whereas simultaneous treatment with both high[HCO₃^-]e (and [HCO₃^-]i) and high P_CO2 had no effect.

To further test this conclusion, we regressed the extent of secretion inhibition against each of the acid-base treatment means separately using both linear and exponential models. The relationships were uniformly non-significant for P_CO2,[HCO₃^-]i, and [HCO₃^-]e, but significant for pHi (r=0.73, P<0.05, linear model only)and strongest for pHe (r=0.91, P<0.002 by the linear model; r=0.94, P<0.001 by the exponential model). Thus while pHe appears to be the dominant influence, we cannot eliminate a role for pHi as well. Indeed, in the intact animal, it is likely that pHe and pHi may often covary. An additional note of caution is that the [¹⁴C]-DMO technique measures mean tissue pHi. While the non-secretory capsule was removed, it remains possible that differences in pHi may have occurred between secreting and non-secreting cells, which would not have been detected.

Fig. 6 integrates the in vitro data of the present study with the responses of the rectal gland in vivo from Wood et al. (Wood et al., 2006) where only extracellular acid-base parameters were measured. Overall, there was good agreement with the in vivo data reinforcing the present in vitro data, such that the relationship between pHe and relative secretion rate did not change, but became more significant due to the greater number of treatment means (r=0.94, P<0.0001; exponential model). Again there was no significant relationship with [HCO₃^-]e. These responses of rectal gland secretion to pHe are probably adaptive in the context of the intact animal because feeding will cause both a metabolic alkalosis (`alkaline tide'; Wood et al., 2005) and volume loading due to the ingestion of seawater and/or salty food(Mackenzie et al., 2002),necessitating activation of rectal gland secretion(Walsh et al., 2006). The highest pHe of about 8.6 from the in vitro experiments is probably outside the physiological range and may explain the non-linearity of response in this range (Fig. 6). However significant stimulation did occur in vivo at a pHe of 8.15, which is within the physiological range (Wood et al., 2005). Conversely inhibition of rectal gland output by acidosis would be useful, because metabolic acidosis normally occurs after severe exercise, a time when volume is contracted due to a shift of extracellular fluid into white muscle because of the high intracellular lactate load (Holeton and Heisler,1983; Richards et al.,2003).

As the effects in vivo are similar to those in the forskolin-stimulated gland completely removed from the dogfish(Fig. 6), these data place the site of acid-base action at the level of the gland itself, rather than at the volume detection, neuroendocrine communication, or surface receptor activation level. One possible site of action might be basolateral K⁺channels, which are known to be activated as part of the secretory mechanism(Valentich and Forrest, Jr,1991; Greiger et al.,1999), and which are known to be very sensitive to pHi in the appropriate fashion (Kerst et al.,2001). Other possibilities might include the apical Cl^-channels, the basolateral Na⁺,K⁺,2Cl^-co-transporter (NKCC) or the basolateral Na⁺,K⁺-ATPase;work at the cellular level will be needed to separate these possibilities.

The present measurements of

in relation to secretion rate are in reasonable agreement with the only other detailed study on this topic in the isolated-perfused rectal gland. Using very similar methodology but a higher perfusion pressure (40 mmHg), Silva et al.(Silva et al., 1980) reported that the

\({\dot{M}}_{\mathrm{O}_{2}}\)

of the unstimulated gland was 0.30 μmol g^-1 gland min^-1, but that extrapolation of the regression of

\({\dot{M}}_{\mathrm{O}_{2}}\)

on Cl^- secretion rate back to zero Cl^- secretion gave a much lower value, 0.14 μmol g^-1 gland min^-1. Our values were 0.11 μmol g^-1 gland min^-1 by both techniques (Table 2, Fig. 2B). From the slope of their regression, Silva et al. calculated that the cost of transporting 1μmol Cl^- was about 0.030 μmol O₂(Silva et al., 1980), whereas our value indicated slightly greater efficiency, 0.022 μmol O₂per μmol Cl^- (Fig. 2B).

\({\dot{M}}_{\mathrm{CO}_{2}}\)

is a more difficult measurement, and we are aware of no previous data in the isolated-perfused rectal gland. Our determinations of

\({\dot{M}}_{\mathrm{CO}_{2}}\)

were generally higher than simultaneous measurements of

\({\dot{M}}_{\mathrm{O}_{2}}\)

⁠. For example, resting

\({\dot{M}}_{\mathrm{CO}_{2}}\)

was 0.20-0.22 μmol g^-1 gland min^-1 by direct measurement or extrapolation of the regression relationship(Table 2, Fig. 2B) and the cost of Cl^- transport from the slope was 0.036 μmol O₂ perμmol Cl^-. However, these differences were not statistically significant. While lactate concentrations in the secreted fluid and venous effluent were below detection [as also reported by Silva et al.(Silva et al., 1980)], lactate levels in the gland tissue itself were about 7 mmol l^-1, comparable to those in the white muscle of intact Squalus acanthias or an isolated-perfused muscle preparation from this species(Richards et al., 2003). Thus a glycolytic component to the metabolism of the perfused rectal gland cannot be excluded; indeed, high lactate dehydrogenase activities have recently been identified in the rectal gland, which increase after feeding(Walsh et al., 2006).

Measurements of the acid-base status of the secreted fluid from the isolated-perfused gland (Tables 2, 5) were in excellent agreement with those from intact dogfish (Wood et al., 2006). Thus in both data sets, secretion pH was about 0.3-0.5 units below pHa, while secretion P_CO2 was very close to Pa_CO2, and secretion HCO₃^- concentration was well below that in the arterial samples. Also in both data sets, changes in secretion acid-base status tended to track those in extracellular acid-base status during experimental disturbances, but were less pronounced. There is substantial disagreement of our values with the sparse earlier measurements(Burger and Hess, 1960; Siegel et al., 1975; Swenson and Maren, 1984). Possibly, earlier workers may not have realized the need to pre-condition the electrodes to the poorly buffered secretion fluid, which would have caused an underestimation of its pH and overestimation of its P_CO2. The present data demonstrate that the secretion barrier has low permeability to HCO₃^- but allows approximate equilibration of Pa_CO2 with the secreted fluid. The barrier function was particularly apparent in the high HCO₃^- treatments, where secretion HCO₃^- levels rose to only 3-6 mmol l^-1(Table 5), in the face of 40 mmol l^-1 levels in the perfusate(Table 4). In agreement with studies on the in situ gland preparation(Swenson and Maren, 1984) and the intact dogfish (Wood et al.,2006), the present results with the isolated-perfused gland provide no evidence that this organ serves as a significant site of systemic acid-base regulation, but do show that it is effective in minimizing passive HCO₃^- losses. Similarly, the gland is effective in minimizing passive urea losses (Zeidel et al., 2005).

The present results clarify the formerly uncertain role of carbonic anhydrase (CA) in the rectal gland by showing that blockade of the enzyme with acetazolamide (10^-4 mol l^-1) affects the dynamics of CO₂ excretion in the isolated-perfused rectal gland. Thus

was depressed by about 30% for several hours in the actively secreting gland(Fig. 3C) without significant change in

\({\dot{M}}_{\mathrm{O}_{2}}\)

(Fig. 3B) or Cl^-secretion rate (Fig. 3A). P_CO2 in the secreted fluid was also significantly elevated over a similar time course(Fig. 4), as also seen in vivo (present study) and in the in situ preparation(Swenson and Maren, 1984). These data suggest that CA is not involved in the secretory mechanism itself,but rather serves to facilitate the diffusive efflux of CO₂ into the bloodstream (or perfusate) by ensuring that it is converted into HCO₃^- at the extracellular boundary, thereby maintaining an outwardly directed P_CO2 gradient. This might include the recycling of metabolic protons and imported HCO₃^- back into the bloodstream. Indeed the original localization of CA by Lacy (Lacy,1983) fits well with this interpretation. Lacy concluded that CA appears to lie in the `intercellular spaces' (i.e. in the extracellular compartment) along the basolateral membranes of the gland cells, close to Na⁺,K⁺-ATPase and the underlying mitochondria which are the actual sites of CO₂ generation in the working cells(Lacy, 1983). Lacy reported a second site of concentrated CA activity, along the epithelium of the central canal draining the secreted fluid (Lacy,1983). Blockade at this site would therefore explain the increase in P_CO2 of the fluid(Fig. 4).

Our interpretation is therefore similar to the explanation offered by Swenson and Maren based on their experiments with an in situ gland preparation in the pithed, artificially ventilated dogfish(Swenson and Maren, 1984). However, there is one important difference: Swenson and Maren interpreted CA as acting in the intracellular compartment as a true facilitator of CO₂ diffusion [i.e. via the co-diffusion of HCO₃^- and buffered H⁺, which is the model of Gros and coworkers (Gros et al.,1976; Gros,1991)], whereas we propose that CA serves to catalyze the CO₂ hydration reaction in the extracellular compartment, so as to maintain the P_CO2 gradient from cell to blood/perfusate. This interpretation is in accord with Lacy's original localization of CA (Lacy,1983). As such, its role would be identical to that of extracellular CA in trout muscle where there is a need to accelerate CO₂ excretion after exercise(Henry et al., 1997). An added benefit of this role is that it would ensure that acid-base reactions in the extracellular compartment next to the working cells would be at equilibrium. Therefore any change in systemic acid-base status important in controlling rectal gland function would be instantly transmitted to the secreting cells. However, one note of caution must be raised. Given the concentration of acetazolamide used (10^-4 mol l^-1), a very potent inhibition of extracellular carbonic anhydrase would be expected to occur rapidly, whereas the reduction in CO₂ excretion of the perfused gland did not occur until 1.5-2 h of treatment. This is more in line with slow permeation of acetazolamide into the intracellular compartment.

Regardless of the site of acetazolamide action, if the CO₂excretion function of CA is important, then a pertinent question is why CA inhibition did not inhibit Cl^- secretion in the isolated-perfused gland in this (Fig. 3A) or earlier studies (Siegel et al.,1975; Silva et al.,1977; Swenson and Maren,1984). This question was first raised by Swenson and Maren(Swenson and Maren, 1984), and the explanation they offered now seems very reasonable in light of the current data. Swenson and Maren speculated that in vivo and in situperfusion flows relative to secretory rates are much lower than in maximally vasodilated in vitro preparations(Swenson and Maren, 1984). In the present study, the ratio of perfusion flow to secretion flow in the isolated-perfused gland was about 67 (Table 1), whereas Swenson and Maren estimated that the ratio might be tenfold lower in vivo, making perfusive limitations on CO₂excretion much more severe in the intact animal(Swenson and Maren, 1984). Thus CA inhibition decreased secretory output in vivo(Fig. 1A) and in situ(Swenson and Maren, 1984),though as earlier noted, interpretation in both studies is confounded by the accompanying systemic acidosis.

Further support for these ideas is provided by the pHi data. Absolute pHi values in the rectal gland were very similar in vivo and in vitro (both about 7.5), but pHi in the secreting cells was not reduced by CA inhibition in the isolated-perfused gland(Table 3), whereas it was in vivo (Table 1). We were initially concerned that this apparent lack of effect was an artifact since acetazolamide itself could act as a buffer if it accumulated to high concentration in the tissue, but in fact measurements demonstrated there was no change in the buffer capacity of the gland. Notably, activation of secretion did not lower pHi in the perfused gland(Table 3), whereas it did in vivo (Table 1),again suggesting that CO₂ excretion is a much more serious problem in vivo. Acid-base regulatory mechanisms are critical in maintaining a suitable intracellular pHi for secretion in analogous NaCl-secreting systems such as the bird salt gland (Shuttleworth and Wood, 1992; Shuttleworth and Hildebrandt, 1999). Similar Na⁺-dependent and HCO₃^--dependent pHi-regulatory mechanisms have been identified in shark rectal gland cells, but only at rest in perfused tubules(Bleich et al., 1998). Clearly there is a need to understand how effective these are in the working rectal gland in vivo.

List of abbreviations
 
• CA

  carbonic anhydrase

 
• CNP

  C-type natriuretic peptide

 
• ECFV

  extracellular fluid volume

 
• ICF

  intracellular fluid

 
• ICFV

  intracellular fluid volume

 
• NKCC

  Na⁺,K⁺,2Cl^- co-transporter

 
• VIP

  vasoactive intestinal polypeptide

Supported by an NSERC Discovery Grant to C.M.W. C.M.W. is supported by the Canada Research Chair Program. We thank the director and staff of Bamfield Marine Sciences Centre for their help. Animals were cared for in accord with the principles of the Canadian Council for Animal Care, and protocols were approved by institutional animal care committees.