Pulsatile Urea Excretion In Gulf Toadfish (Opsanus Beta): Evidence For Activation Of A Specific Facilitated Diffusion Transport System

The gulf toadfish (Opsanus beta) is unusual amongst teleost fish in expressing a full complement of ornithine–urea cycle enzymes in its liver as an adult (Read, 1971; Mommsen and Walsh, 1989; Anderson and Walsh, 1995). The toadfish is normally ammoniotelic when first collected from the wild, but becomes facultatively ureotelic (>70 % urea-N, <30 % ammonia-N excretion) when subjected to crowding and/or confinement in the laboratory (Walsh et al. 1994a; Walsh and Milligan, 1995; Hopkins et al. 1995). Under these conditions, the toadfish demonstrates a remarkable pattern of urea-N excretion, with over 90 % appearing in pulses of less than 3 h duration which emanate from the head region, presumably from the gills (Wood et al. 1995a). These pulses are irregular in time and amongst individuals, but occur on average approximately once per day. The phenomenon is not due to pulsatile production by the liver, which actually synthesizes urea at a steady rate (Wood et al. 1997). Rather, urea-N pulses are due to the periodic activation of an excretion mechanism that rapidly clears urea from the blood plasma, thereby lowering stored levels throughout the whole body (Wood et al. 1997). The adaptive significance of this behaviour remains unknown, but its existence presents a fascinating physiological problem as to the mechanism and control of pulsatile excretion. Classically, urea has been viewed as an unreactive molecule which moves passively across biological membranes by simple diffusion. Indeed, the branchial fluxes of urea have often been used as a general index of gill surface area or permeability (e.g. Bergman et al. 1974; Haywood et al. 1977). However, recent research in higher vertebrates has suggested a much more complex picture, with a variety of facilitated diffusion and secondary active transport mechanisms possible (e.g. Marsh and Knepper, 1992; Dytko et al. 1993; Gillin and Sands, 1993). For example, in the mammalian kidney, a hormonally controlled facilitated diffusion carrier has been pharmacologically characterized (Chou and Knepper, 1989; Knepper and Star, 1990; Knepper and Chou, 1995), cloned in two species (You et al. 1993; Smith et al. 1995) and localized subcellularly in the tubule cells of the terminal inner medullary collecting duct (IMCD; Nielsen et al. 1996). Strong evidence also exists for the presence of a Na⁺-coupled active urea transporter which may be experimentally induced in the initial portion of the IMCD by dietary protein restriction (Isozaki et al. 1993, 1994a,b). There is precedent for such transporters in fish. Active, Na⁺-coupled urea reabsorption has long been recognized in the elasmobranch kidney (Schmidt-Nielsen and Rabinowitz, 1964; Schmidt-Nielsen et al. 1972; Hays et al. 1977). Recently, pharmacological evidence has been obtained for a facilitated-diffusion-type transporter in hepatocytes of the toadfish itself, which apparently serves to equilibrate urea rapidly between the liver and blood (Walsh et al. 1994b; Walsh and Wood, 1996).

With this background in mind, the goal of the present study was to test three general hypotheses about the mechanism of pulsatile urea excretion in the toadfish. The first hypothesis (Fig. 1A) was that there is a reabsorptive ‘back-transporter’ in the gills, normally pumping urea back into the blood as it continually leaks across the gill cells. Periodic inhibition of such a transporter would lead to pulsatile excretion. This idea was tested by placing putative inhibitors of such a transporter in the external water. The second hypothesis (Fig. 1B) was that pulsatile urea excretion occurs because of a periodic generalized increase in gill permeability, i.e. a relatively non-specific effect such as gill vasodilation, whereby permeability is increased not just to urea but to a wide range of other molecules as well. This model was evaluated by the use of agents thought to increase general gill surface area or permeability and by testing for possible changes in the permeability of the gills to other substances during natural pulse events. The third hypothesis (Fig. 1C) was that pulsatile excretion occurs because of periodic insertion or activation of urea transport proteins in the gills. This explanation, which proved to be the most probable, was evaluated by pharmacological studies and by testing whether bidirectional urea transport capability, a key property of facilitated diffusion urea carriers, could be seen at the time of natural pulses.

Sexually mature gulf toadfish (Opsanus beta Goode and Bean; 108±7 g, range 47–202 g, N=45) were captured with a roller trawl by commercial shrimp fisherman in Biscayne Bay, Florida, USA, and held in the laboratory at 25±1 °C, 31±2 ‰ salinity, under natural photoperiod in April and May 1996. During the first 7–10 days, they were maintained in groups of 3–4 fish in 45 l aquaria supplied with a bed of beach sand, flowing sea water and individual polyvinylchloride (PVC) pipes for shelter (‘non-crowded’ conditions, less than 12 g fish l⁻¹ seawater). During this period, the fish were subjected to two treatments with Malachite Green plus formalin as prophylaxis against the ciliate Cryptocaryon irritans (see Wood et al. 1997) and were fed live shrimp on alternate days.

A standardized crowding and confinement protocol was used to induce and maintain ureotelic behaviour (see Walsh et al. 1994a; Wood et al. 1995a, 1997; Hopkins et al. 1995). In brief, toadfish together with their PVC shelters, were transferred in groups of 3–6 into covered 6 l tubs supplied with flowing sea water (‘crowded conditions’, greater than 80 g fish l⁻¹ seawater) for 48–72 h prior to surgery. They were not fed during this period. Following cannulation, fish were placed individually into covered rectangular 3 l containers made of opaque plastic and allowed to recover for 24 h prior to the start of experiments. Each container was supplied with a PVC shelter, aeration and flowing sea water,

Surgery was performed under MS-222 anaesthesia (0.67 g l⁻¹, adjusted to pH 8.0 with NaOH). Each fish was fitted with a single catheter (Clay-Adams PE50) filled with toadfish saline containing sodium heparin (100 i.u. ml⁻¹). Catheters were placed, via the haemal arch, into either the caudal artery or the caudal vein exactly as described by Wood et al. (1997). The arterial or venous location of the catheter had no detectable influence on the results.

Following recovery, each toadfish was first monitored for 24–48 h to ensure that it was exhibiting ureotelic ‘pulsing’ behaviour. The water flow was shut off, the container volume was set to 2.0 l, the aeration was set to provide good mixing and a peristaltic pump plus fraction collector system started to continually collect hourly 5 ml water samples for colorimetric assay of urea-N and ammonia-N. In some cases, hourly water samples were drawn by hand because of a shortage of fraction collectors. Throughout this initial monitoring period and the subsequent experiments, the water was changed at 24 h intervals or more frequently, depending on the experimental design. The containers were bathed externally with flowing sea water on a wet-table which maintained the experimental temperature at 25±1 °C.

In many of the experiments, it was necessary to measure urea-N excretion rate under conditions in which the chemistry of the external water was altered (e.g. by addition of urea analogues, high levels of urea itself) in such a way as to prevent accurate colorimetric analysis of urea by the normal diacetyl monoxime assay (Rahmatullah and Boyde, 1980; Price and Harrison, 1987). In these experiments, the procedure developed by Wood et al. (1997) was employed, in which the fish was loaded with [¹⁴C]urea, the specific activity of urea in the plasma was measured periodically and urea-N excretion was determined by hourly monitoring of the appearance of ¹⁴C radioactivity in the water. Wood et al. (1997) demonstrated that more than 95 % of the ¹⁴C counts in plasma and water remained as urea (i.e. no degradation). More importantly, there was a 1:1 relationship between urea-N pulses measured colorimetrically and urea-N pulses measured radioactively, demonstrating that urea-N pulses originate from the plasma or from a compartment in equilibrium with plasma. Typically, fish in these experiments were injected via the catheter with 400 μCi kg⁻¹ of [¹⁴C]urea (New England Nuclear, specific activity 8.60 mCi mmol⁻¹; 1 mCi=3×10⁷ Bq) in 2 ml kg⁻¹ saline, washed in with a further 2 ml kg⁻¹ toadfish saline. After a 3 h equilibration period, the experiment was started. In longer-term experiments, the plasma specific activity was ‘topped up’ at 48 h intervals by additional injections of 200 μCi kg⁻¹ [¹⁴C]urea.

Periodic blood sampling was used not only to monitor plasma specific activity, but also to follow changes in absolute plasma urea-N concentrations accompanying pulsatile excretion events. Long catheters allowed blood sampling without disturbance. Blood samples were drawn at 2 h or 4 h intervals with some breaks, depending on the experimental design; 200 μl was first drawn to clear the catheter, then 50 μl was taken as the true sample, and then the first 200 μl was returned, together with sufficient saline to flush the catheter and replace the sampled volume. Approximately 25 blood samples were drawn from each fish (maximum 36). Haematocrits typically declined from 25–30 % to 5–12 %, but this had no detectable influence on the excretion patterns or plasma levels of urea-N.

Na⁺-free sea water was used to block possible Na⁺-coupled urea transport (Schmidt-Nielsen et al. 1972; Isozaki et al. 1994a), while thiourea and acetamide were chosen as urea analogues which are known to act as competitive inhibitors of carrier-mediated urea transport in many systems (Marsh and Knepper, 1992). A potential limitation of this approach should be noted which relates to the location of putative ‘back-transporters’. If they are located on the apical membranes of the gill cells, then these agents applied in the external water would undoubtedly have access to them. However, if they are located on the basolateral membranes, low permeability of the apical membranes could impede access.

The rationale behind these experiments was to wait until a natural urea-N pulse had occurred and finished, and then to change the external water to a treatment medium containing the inhibitor of the putative back-transport mechanism for a further 3 h period. On the basis of past experience (Wood et al. 1995a, 1997), we knew that a second natural pulse would rarely occur within 8 or 9 h of the first one. Therefore, a significant appearance of urea-N into the external medium during this 3 h period would indicate that the reabsorptive back-transporter had been pharmacologically inhibited, allowing urea-N to leak out from the blood. Finally, the external medium was changed back to ordinary sea water (‘washout’) for a further 3 h. This was performed as a safeguard against non-specific stress effects of the treatment, which would probably be manifested as a continuation of urea-N leakage after removal of the inhibitor.

The [¹⁴C]urea technique was used. Hourly water samples were analysed immediately following collection. Once the occurrence and completion of a natural pulse had been detected, the external sea water was changed to (a) synthetic Na⁺-free sea water (N=6); (b) sea water containing 30 mmol l⁻¹ thiourea (N=5); or (c) sea water containing 30 mmol l⁻¹ acetamide (N=9). There was a time lag of approximately 2 h between the end of the natural pulse and the changeover. In (a), the container and fish were first flushed four times with distilled water to ensure complete removal of Na⁺, following which the synthetic Na⁺-free sea water was added: 400 mmol l⁻¹N-methyl-D-glucamine, 10 mmol l⁻¹ CaCl₂.2H₂O, 33 mmol l⁻¹ MgSO₄.7H₂O, 20 mmol l⁻¹ MgCl₂.6H₂O, 2.5 mmol l⁻¹ KHCO₃, 6.8 mmol l⁻¹ KCl, 0.5 mmol l⁻¹ KBr, adjusted to pH 8.0 by titration with HCl. Measured mean Na⁺ concentrations (Varian 1275 atomic absorption spectrophotometer) were 0.35±0.08 and 0.68±0.10 mmol l⁻¹ at the start and end of the 3 h period, respectively, relative to a normal seawater concentration of approximately 450 mmol l⁻¹. In (b) and (c), 30 mmol l⁻¹ thiourea and acetamide were employed so as to give a level approximately three times higher than urea levels (approximately 10 mmol l⁻¹=20 000 μmol-N l⁻¹) in the body fluids (Wood et al. 1995a, 1997).

This experiment (N=5) was based on findings in parallel experiments on an isolated, perfused toadfish head preparation (P. Part, unpublished results) that high levels of total ammonia (T_Amm) in the perfusate would greatly elevate the permeability of the gills to urea. Hourly urea-N excretion was monitored until a natural urea-N pulse had occurred and finished as in series 1, except that measurements were performed colorimetrically. Rates of ammonia-N and urea-N excretion were then monitored for a further 3 h prior to and 5 h following infusion via the catheter with 2000 μmol kg⁻¹T_Amm (11.4 ml kg⁻¹ of 175 mmol l⁻¹ NH₄Cl, washed in with a further 2 ml kg⁻¹, delivered over 20 min). Blood samples for the measurement of plasma T_Amm were drawn 1 h before and at 1, 3 and 5 h after the infusion.

The synthetic β-adrenergic agonist L-isoprenaline, a powerful branchial vasodilating agent (Wood, 1974), was administered to induce a generalized increase in gill permeability. High circulating levels of catecholamines appear to elevate functional gill surface area in all teleosts including Opsanus beta (Oduleye and Evans, 1982; Oduleye et al. 1982). An identical protocol to that for series 2 was followed, except that plasma T_Amm was not measured and toadfish (N=6) were injected with a dose of L-isoproterenol-D-bitartate calculated to yield an extracellular concentration of approximately 10⁻⁵ mol l⁻¹ (0.3 ml kg⁻¹ of a 10 mmol l⁻¹ stock, stabilized with 10⁻³ mol l⁻¹ ascorbic acid in saline, and washed in with a further 2 ml kg⁻¹ of saline).

This series examined whether the increases in urea-N permeability that occurred during natural pulse events were correlated with simultaneous increases in the permeability to other molecules. Substances tested were [¹⁴C]polyethylene glycol, M_r=4000 (PEG-4000, a paracellular diffusion marker; Wood and Part, 1997), tritiated water (³H₂O, a transcellular diffusion or water-channel marker; Oduleye and Evans, 1982; Part et al. 1998) and [¹⁴C]thiourea (a urea analogue that is not only a competitive blocker but also a substrate transported by certain facilitated diffusion carriers; Marsh and Knepper, 1992). Radiotracers allowed the use of trace amounts so that blocking or non-specific effects would be negligible.

Toadfish were injected via the catheter with (a) 400 μCi kg⁻¹ [¹⁴C]PEG-4000 (NEN, specific activity 0.80 mCi mmol⁻¹) in 4.8 ml kg⁻¹ saline (N=6); (b) 800 μCi kg^{−1 3}H₂O (NEN, specific activity 1 mCi g⁻¹), i.e. 0.8 ml kg⁻¹ tritiated water (N=4); or (c) 400 μCi kg⁻¹ [¹⁴C]thiourea (NEN, specific activity 58.0 mCi mmol⁻¹) in 1.8 ml kg⁻¹ saline (N=6). In each case, the injection was washed in with a further 2 ml kg⁻¹ saline. Six different fish were used for each of the [¹⁴C]PEG-4000 and [¹⁴C]thiourea experiments, whereas two each of the fish injected with these two radiolabels were also injected with ³H₂O. After a 3 h post-injection equilibration period, water samples were taken hourly and blood samples every 2 h over the following 20–36 h, with some breaks, for the measurement of plasma and water radioactivities and urea-N concentrations. Urea excretion was measured colorimetrically. These measurements allowed calculation of urea, PEG-4000, tritiated water and thiourea permeabilities in common units (cm s⁻¹, see below) during natural pulse events and the intervening non-pulse periods.

This experiment (N=8) evaluated whether pulsatile urea-N excretion could be inhibited by systemic administration of phloretin, a potent competitive blocker of many facilitated diffusion urea transporters in higher vertebrates (Marsh and Knepper, 1992). Because phloretin must be solubilized in dimethylsulphoxide (DMSO), the possible effects of this vehicle alone were evaluated on each fish during the initial 24 h period prior to the administration of phloretin.

During this initial 24 h monitoring period, DMSO (1 ml kg⁻¹dissolved in 4 ml kg⁻¹ saline and washed in with a further 2 ml kg⁻¹ saline) was injected at 0, 8 and 16 h. Water samples were taken hourly for colorimetric assay of urea-N and ammonia-N, and blood samples every 4 h for assay of plasma urea-N concentrations. At 24 h, the containers were flushed, 250 μmol kg⁻¹ phloretin in the above vehicle was injected, and sampling commenced again. Further doses of 100 μmol kg⁻¹ phloretin were injected at 8 h and 16 h after the initial injection, and the experiment was terminated at 24 h.

An important property of most facilitated diffusion urea carriers is that they are bidirectional, transporting urea in either direction with equal efficiency in proportion to the relevant concentrations (Chou et al. 1990). We therefore reasoned that, if a pulsatile urea excretion event occurred because of activation/insertion of such a carrier in the gills, then net inward urea transport should take place at this time if the external water contained a high level of urea. The efflux component could be measured by the appearance of [¹⁴C]urea radioactivity in the water, and the net flux could be calculated from the change in plasma urea-N levels (see below). The influx component could then be calculated by difference.

After the initial 24 h monitoring period, toadfish (N=14) were loaded with [¹⁴C]urea, and the external sea water was replaced with sea water containing 30 mmol l⁻¹ urea (60 000 μmol l⁻¹ urea-N, i.e. approximately three times higher than blood levels). After a 3 h equilibration period, water samples were taken hourly and blood samples every 2 h for the measurement of plasma and water radioactivities and urea-N concentrations. Sampling was continued for a further 20–30 h (with some breaks) or until at least one natural urea-N pulse event occurred.

As the results confirmed the occurrence of bidirectional transport during pulsatile events, the experiment was extended in some of the toadfish (N=7) to test whether the influx component in the presence of 30 mmol l⁻¹ urea could be inhibited by the simultaneous presence of a high concentration of external thiourea (60 mmol l⁻¹), a competitive inhibitor in many facilitated diffusion transport systems (Chou and Knepper, 1989; Marsh and Knepper, 1992). At the end of the monitoring period in 30 mmol l⁻¹ urea alone, the containers were flushed and refilled with normal sea water for approximately 12 h. The plasma specific activity was then ‘topped up’ by an additional injection of 200 μCi kg⁻¹ [¹⁴C]urea, and the external medium was changed to one containing 30 mmol l⁻¹ urea (60 000 μmol urea-N l⁻¹) plus 60 mmol l⁻¹ thiourea. After a 3 h equilibration period, hourly water sampling and blood sampling every second hour were resumed for another 30 h, with some breaks, to examine the unidirectional flux components during natural pulses in the presence of the putative inhibitor.

Water and plasma samples for analysis of urea-N, ammonia-N or radioactivity were assayed immediately, stored at 4 °C for no more than 24 h prior to assay, or frozen at −20 °C pending later analysis. Standard chemical colorimetric assays on a micro-plate reader (Thermomax; Molecular Devices) were employed for urea-N (Rahmatullah and Boyde, 1980; Price and Harrison, 1987) and ammonia-N (Ivancic and Deggobis, 1984) in sea water, while ammonia-N (T_Amm) in plasma was measured enzymatically using a Sigma kit (Mondzac et al. 1965). For analysis of ¹⁴C and ³H activity, 1 ml of radioactive sea water, or 10 μl of radioactive plasma plus 1 ml of ‘cold’ sea water, was added to 10 ml of ICN Ecolume. Single-labelled samples were counted on a TmAnalytic 6895 BetaTrac liquid scintillation counter, while dual-labelled samples were counted on a Beckman LS1801 liquid scintillation counter, employing inbuilt quench correction programs.

All urea and ammonia concentrations and fluxes have been expressed in units of nitrogen for comparative purposes and consistency with previous studies (Wood et al. 1995a, 1997). Pulses of urea-N excretion were identified from stepwise increases in seawater urea-N concentration or [¹⁴C]urea radioactivity recorded in hourly water samples. When colorimetric methods were employed, excretion rates (in μmol-N kg⁻¹ h⁻¹) of urea-N and ammonia-N were calculated from changes in concentration in the external sea water (μmol-N l⁻¹) multiplied by the volume (l) and factored by time (h) and mass (kg). When the [¹⁴C]urea method was used, the total radioactivity appearing in the water in each hour (cts min⁻¹) was divided by the measured specific activity (cts min⁻¹ μmol⁻¹ urea-N) of the preceding plasma sample, again factored by time (h) and mass (kg). Total urea-N pulse sizes were calculated in the same way, but not factored by time.

In experiments designed to measure bidirectional urea-N fluxes with 60 000 μmol urea-N l⁻¹ in the external water, the efflux component was calculated exactly as outlined above from the appearance of [¹⁴C]urea radioactivity in the water. The net flux component was calculated from the measured change in plasma urea-N concentration accompanying the pulse event, divided by 0.75. Wood et al. (1997) demonstrated that, on average, the change in plasma urea-N concentration (μmol l⁻¹) was 75 % of the measured urea-N pulse (μmol kg⁻¹). The influx was then calculated as the difference between the net flux and the efflux. Influx and efflux per unit concentration were calculated by dividing the influx (μmol-N kg⁻¹ pulse⁻¹) by the external concentration (60 000 μmol-N l⁻¹) and the efflux (μmol-N kg⁻¹ pulse⁻¹) by the measured plasma concentration (μmol-N l⁻¹).

Data have been expressed as means ± 1 S.E.M. (N) where N represents either the number of fish or the number of pulse events, as specified. Regression lines were fitted by the method of least squares, and the significance of Pearson’s correlation coefficient r was assessed. Student’s two-tailed t-test, unpaired or paired as appropriate, was employed to evaluate the significance of differences between means. For multiple comparisons, the t value was adjusted using the Bonferroni procedure (Nemenyi et al. 1977). A fiducial limit of P⩽ ;0.05 was employed throughout.

During the initial 24–48 h monitoring period, only two of the 45 toadfish studied did not exhibit pulsatile urea-N excretion; they were rejected. Excluding the fish treated with DMSO for series 5 (see below), the other 35 toadfish were predominantly ureotelic, excreting 77 % urea-N (220.2±23.9 μmol-N kg⁻¹ h⁻¹) and only 23 % ammonia-N (66.2±11.0 μmol-N kg⁻¹ h ⁻¹). Of the total urea-N excretion, 89.7±1.7 % occurred in discrete pulse events of 3 h or less. Mean urea-N pulse size was 3150±391 μmol-N kg⁻¹, and mean frequency was 1.67±0.12 pulses day⁻¹. Ammonia-N excretion was never pulsatile. Mean plasma urea-N concentration was 19 562± 2714 μmol-N l⁻¹ (N=35) and mean plasma ammonia-N (T_Amm) concentration, measured only in the fish of series 2, was 124±22 μmol-N l⁻¹ (N=5).

Averaged over 3 h, urea-N excretion rate during natural pulse events in series 1 was 600–1000 μmol-N kg⁻¹ h⁻¹ (Fig. 2). After completion of the pulses, the external sea water was exchanged for a medium containing one of three potential inhibitors of a possible reabsorptive back-transport mechanism in the gills: Na⁺-free sea water (Fig. 2A), 30 mmol l⁻¹ thiourea in normal sea water (Fig. 2B) or 30 mmol l⁻¹ acetamide in normal sea water (Fig. 2C). None of these agents resulted in any significant ‘leakage’ of urea-N into the external media above that normally seen (approximately 20 μmol-N kg⁻¹ h⁻¹), despite the fact that measured blood plasma urea-N levels remained in the range 15 000–26 000 μmol-N l⁻¹ in these fish. There was also no significant leakage upon return to normal sea water (‘washout’). These results argue against the hypothesis that pulsatile urea-N excretion occurs because of periodic inhibition of a back-transport mechanism that normally prevents urea-N leakage.

When toadfish of series 2 were infused with 2000 μmol kg⁻¹ NH₄Cl a few hours after completion of a natural pulse, plasma T_Amm levels were significantly elevated approximately 3.5-fold to 425 μmol-N l⁻¹, measured at 1 h post-infusion (Fig. 3A). Although declining thereafter, plasma T_Amm remained significantly elevated at 5 h. This treatment did not cause any stimulation of urea-N excretion, which remained at the normal basal rate of approximately 20 μmol-N kg⁻¹ h⁻¹ for the next 5 h (Fig. 3C). Interestingly, there was a marked stimulation of ammonia-N excretion, significant for the first 2 h after infusion, but only approximately 50 % (1000 μmol kg⁻¹) of the infused ammonia load was excreted before ammonia-N excretion returned to control levels (Fig. 3B).

In series 3, a similar protocol demonstrated that injection of the powerful branchial vasodilator L-isoprenaline had no effect on urea-N excretion over the following 5 h (Fig. 4B). However, isoprenaline caused an elevation of ammonia-N excretion that was significant at 1 h and 3 h post-injection (Fig. 4A).

In series 4, the permeability of toadfish to several radiolabelled compounds was compared with that for urea during pulsing and non-pulsing periods; a logarithmic scale has been employed in Fig. 5 to encompass the wide absolute range of permeabilities measured. Urea permeability during non-pulsing periods was approximately 3×10⁻⁷ cm s⁻¹ and increased 25-to 50-fold to approximately 10⁻⁵ cm s⁻¹ (minimum estimate; see Discussion) during natural pulses. Permeability to the paracellular permeability marker PEG-4000 (approximately 10⁻⁸ cm s⁻¹ ) was more than one order of magnitude lower than that to urea during non-pulsing periods and exhibited no increase during natural pulses (Fig. 5A). Permeability to ³H₂O (approximately 1.5×10⁻⁵ cm s⁻¹) was almost two orders of magnitude higher than that to urea during non-pulsing periods, but again did not increase during natural pulses (Fig. 5B). During non-pulsing periods, permeability to the urea analogue thiourea (approximately 1.5×10⁻⁷ cm s⁻¹) was similar to, but slightly lower than, that to urea (difference not significant). Unlike the other markers, permeability to thiourea increased significantly during urea pulses, but the elevation was only fivefold in contrast to the 35-fold increase in urea permeability in this group of fish (Fig. 5C). Thiourea permeability was therefore significantly lower than urea permeability during pulses. Fig. 6 shows an example of an original record of the concentrations of urea-N and [¹⁴C]thiourea (cts min⁻¹) during two natural pulse events, illustrating the correspondence in time between thiourea and urea pulses.

Taken together, the ammonia, L-isoprenaline, PEG-4000 and ³H₂O data argue strongly against any generalized (i.e. non-specific) increase in permeability as the mechanism of pulsatile urea-N excretion. The thiourea data suggest the activation of a selective transport system which recognizes another molecule similar to urea.

As a control in series 5, prior to phloretin administration, the DMSO/saline vehicle was administered to the same fish at 8 h intervals over the initial 24 h monitoring period (Table 1). This caused subtle changes in the pattern of nitrogen waste excretion, with somewhat lower urea-N excretion rates, and a lower percentage (approximately 56 %) of nitrogen excretion occurring in the form of urea relative to non-injected animals (see above). However, urea-N excretion remained pulsatile at approximately the same frequency, 90 % of total urea-N excretion still occurred in discrete pulses, and plasma urea-N concentration fell in the predicted manner coincident with each pulse event (cf. Figs 2, 3 of Wood et al. 1997).

Phloretin, a potent blocker of many facilitated diffusion systems, proved to be highly toxic when administered systemically. Two toadfish (omitted from the data set of Table 1) died almost immediately after the first injection (250 μmol kg⁻¹). Table 1 summarizes the data for the remaining six fish for the 0–16 h period. All survived past 17 h (i.e. past the second and third maintenance injections of 100 μmol kg⁻¹), but only one remained alive at 24 h and survived indefinitely. Phloretin treatment had no significant effect on the absolute rates of urea-N or ammonia-N excretion. However, phloretin abolished the pulsatile excretion pattern, such that pulse frequency dropped to zero and urea simply accumulated in the external water at a steady rate from all fish. Interestingly, despite this continuous excretion, plasma urea-N concentration increased steadily and significantly during phloretin treatment (Table 1).

Series 6 employed an alternative approach to test for a facilitated diffusion transport system. Urea-N concentration in the water was elevated to 60 000 μmol-N l⁻¹ (30 mmol l⁻¹ urea) in order to detect whether bidirectional transport occurred at the time of natural pulse events. Efflux pulse events were quantified by the appearance of [¹⁴C]urea radioactivity in the external water. Fig. 7 depicts a typical record. Under these conditions, plasma urea-N concentration increased in coincidence with each large pulse event, such as the first and third pulses in Fig. 7, or remained more or less unchanged after small pulses, such as the second one in Fig. 7. Fig. 8 summarizes mean data for nine efflux pulses (1350–5730 μmol-N kg⁻¹) in eight toadfish chosen on the basis of the joint criteria of typical pulse size (1000–6000 μmol-N kg⁻¹) plus the availability of continuous plasma data for at least 4 h prior to and 10 h after a pulse event. The patterns depicted in Figs 7 and 8 contrast markedly with the patterns reported in Figs 2 and 3 of Wood et al. (1997) for normal pulse events of toadfish in the absence of high external urea levels. Thus, toadfish exposed to high external concentrations became loaded with urea-N by inward transport every time the efflux mechanism was activated, confirming the presence of a bidirectional transport system.

In Fig. 8, note that the mean rise in plasma urea concentration (in the 3 h period coincident with a pulse event, starting at the arrow) was approximately 6000 μmol-N l⁻¹, or 1.5 times the mean efflux pulse of approximately 4000 μmol-N kg⁻¹. A net pulse flux (uptake) of approximately 8000 μmol-N kg⁻¹ is obtained by dividing this change in plasma urea-N concentration by 0.75 (see Materials and methods) and, by difference, the mean influx pulse must have been approximately 12 000 μmol-N kg⁻¹. Thus, influx was approximately three times greater than efflux. This is exactly the situation predicted for a bidirectional facilitated diffusion transporter with equal efficiency in both directions, because the water concentration (60 000 μmol-N l⁻¹) was almost three times greater than mean plasma urea-N concentration (21 830±2835 μmol-N l⁻¹, N=9) at the time of these pulses.

A more extensive analysis of 21 pulse events from 14 toadfish confirmed this conclusion. Over a wide range of efflux pulse sizes (620–15 870 μmol-N kg⁻¹), there was a highly significant, proportional 1:1 relationship between influx per unit external concentration and efflux per unit internal (i.e. plasma) concentration (Fig. 9).

Interestingly, during exposure to high external urea-N concentrations, the size of urea-N efflux pulses increased significantly compared with the control day (5084±926 versus 3450±674 μmol kg⁻¹) in the same fish (N=14), but there was no change in pulse frequency (1.31±0.31 versus 1.41±0.19 pulses day⁻¹). This 47 % increase in pulse size probably reflected the proportional 45 % increase in plasma urea-N concentration, which rose significantly during exposure to high urea-N levels (23 960±2219 μmol-N l⁻¹ during the experimental period versus 16 520±1795 μmol-N l⁻¹, N=14, on the control day).

The experiment was extended in seven toadfish to evaluate the potential inhibitory effect of high external thiourea concentration. The ratio of urea-N influx per unit external concentration to urea-N efflux per unit internal concentration was significantly reduced by 73 % in the presence of 30 mmol l⁻¹ urea plus 60 mmol l⁻¹ thiourea (11 pulses) compared with the ratio in the presence of 30 mmol l⁻¹ urea alone (11 pulses) in these same fish (Fig. 10).

The overall pattern of N-waste excretion in these ureotelic toadfish was similar but not identical to that reported previously (Wood et al. 1995a, 1997). As in the earlier studies, the present toadfish excreted urea-N in discrete pulses which accounted for approximately 90 % of total urea-N excretion. However, they exhibited somewhat higher absolute rates of both urea-N and ammonia-N excretion, slightly larger urea-N pulse size and frequency, and a slightly lower overall percentage of total-N excretion in the form of urea. These differences were probably the result of two factors: (i) the present experiments were run at a slightly higher temperature (25 °C versus 20–24 °C); (ii) the present toadfish were fed during the initial holding period in the laboratory under ‘non-crowded’ conditions, a factor that is known to elevate overall N-excretion and favour the persistence of significant ammonia-N excretion in predominantly ureotelic fish (Walsh and Milligan, 1995).

An assumption throughout the present paper is that the gills are the site of pulsatile urea-N excretion, whereas our previous study had localized the phenomenon only to the head region (Wood et al. 1995a). In addition to the molecular and ultrastructural evidence for branchial involvement summarized below, parallel studies have now detected urea-N pulses using an opercular catheter to siphon water coming directly off the gills (K. M. Gilmour, unpublished results) and an absence of urea movement through cephalic skin (P. Part, unpublished results). However, it remains possible that the well-vascularized opercular epithelium may also make a contribution.

This study evaluated three hypotheses to explain the mechanism of pulsatile urea-N excretion (Fig. 1). The results obtained argue against the first two (inactivation of a reabsorptive ‘back-transporter’ mechanism, or a generalized permeability increase) and strongly favour the third (insertion/activation of a facilitated diffusion carrier).

The back-transport hypothesis was laid out in detail by Wood et al. (1997). In brief, the idea is based on findings of secondary active Na⁺-coupled urea-N reabsorption in the kidneys of both elasmobranchs (Schmidt-Nielsen et al. 1972; Hays et al. 1977) and mammals (initial inner medullary collecting ducts of protein-restricted rats; Isozaki et al. 1993, 1994a,b), various other types of urea-N reabsorptive systems in amphibian skin (Dytko et al. 1993) and circumstantial evidence for urea-N reabsorption in the elasmobranch gill (Wood et al. 1995b; Part et al. 1998). Na⁺-free conditions were tested for obvious reasons, whereas high levels of thiourea and acetamide were tested because both analogues appear to be effective in the elasmobranch gill (Wood et al. 1995b), while only the latter is effective in elasmobranch kidney (Schmidt-Nielsen and Rabinowitz, 1964) and the skin of the toad Bufo viridis (Garcia-Romeu et al. 1981; Katz et al. 1981). The complete failure of all three treatments, when placed in the external environment of the toadfish, to cause any ‘leakage’ of urea-N (Fig. 2) argues against the presence of a back-transport system in the gills of Opsanus beta. However, one reservation must be kept in mind, as noted earlier. While these results logically eliminate an apical ‘back-transporter’, a basolaterally located mechanism remains possible, if apical permeability to these externally applied blocking agents was limiting.

All approaches employed to test the idea that a generalized permeability increase constitutes the mechanism for urea-N pulses yielded results that opposed the hypothesis. NH₄Cl infusions were performed because of the results of experiments on an isolated perfused toadfish head preparation (P. Part, personal communication). These in vitro tests demonstrated that raising perfusate T_Amm from 100 to 400–1600 μmol l⁻¹ immediately increased the urea permeability of the gills four-to 32-fold, with minimal effect on ³H₂O or ammonia-N permeability itself. While the mechanism is unknown, the idea is attractive because it might help coordinate pulsatile events with the internal build-up of a more dangerous N-waste product, which could subsequently be detoxified by conversion to urea. However, an increase in plasma T_Ammin vivo to 425 μmol l⁻¹ (measured at 1 h, undoubtedly higher earlier) caused by exogenous NH₄Cl had no effect on urea-N excretion (Fig. 3). Presumably the in vitro response is in some way associated with the special conditions of perfusion. In vivo, approximately 50 % of the exogenous ammonia-N load was not excreted and was probably detoxified through glutamine and/or urea synthesis (Walsh et al. 1990; Barber and Walsh, 1993).

L-Isoprenaline is a powerful synthetic β-adrenergic agonist; at the dose chosen (circulating levels of approximately 10⁻⁵ mol l⁻¹), it would undoubtedly have caused a maximal vasodilation and increase in surface area of the gills (Wood, 1974; Oduleye and Evans, 1982; Oduleye et al. 1982), together with probable increases in cardiac output and ventilatory flow (Randall and Perry, 1992). Marked vasodepressor and cardio-acceleratory actions of L-isoprenaline in vivo in Opsanus beta have been recorded (S. F. Perry, unpublished results). In toadfish perfused gill preparations, β-adrenergic stimulation has been shown to increase ³H₂O permeability (Oduleye et al. 1982; P. Part, unpublished results), and in perfused gills from trout (Oncorhynchus mykiss), increases in [¹⁴C]urea fluxes have even been reported (Bergman et al. 1974; Haywood et al. 1977). The absence of any detectable effect of isoprenaline on urea-N excretion in the present toadfish, even though there was substantial stimulation of ammonia-N excretion (Fig. 4), is a strong argument against the generalized permeability increase hypothesis.

Permeability calculations assumed a pulse period of 3 h, the minimum resolution time of the recording system. Recent experiments employing a monitoring system with a much finer time resolution recorded a mean pulse duration of only 62 min (K. M. Gilmour, unpublished results), but the mean size of the urea-N pulses in that study was only approximately 25 % of the size measured in the toadfish used in the present study. Permeability calculations also assumed a gill area equivalent to that measured in the closely related Opsanus tau (Hughes and Gray, 1972). While this choice is probably not a significant source of error, the assumption that fluxes occur only through the gills for all the measured labels is problematic, because the skin of Opsanus beta lacks scales and is well-vascularized. Fortunately, at least for urea and ³H₂O, skin permeabilities and gill permeabilities measured separately in vitro were almost identical to one another (P. Part, unpublished results) and were also very similar to the values calculated for the gills of intact, non-pulsing toadfish in the present study (Fig. 5).

Urea permeabilities during non-pulse periods in toadfish (3×10⁻⁷ cm s⁻¹) were approximately one order of magnitude lower than in other teleosts (see summaries in Isaia, 1984; Part et al. 1998), confirming that urea-N excretion is tightly restricted in this species. Paracellular permeability (10⁻⁸ cm s⁻¹), as typified by PEG-4000 fluxes, appeared to be similarly low (see summary in Wood and Part, 1997), while ³H₂O permeability (1.5×10⁻⁵ cm s⁻¹) was fairly typical of other teleosts (Isaia, 1984; Part et al. 1998). The key point, however, is that neither PEG-4000 nor ³H₂O permeabilities increased during natural urea-N pulses, at a time when urea-N permeability increased at least 35-fold (or up to 100-fold if pulse duration was really 1 h rather than 3 h). These results rule out generalized increases in either paracellular or transcellular/water channel permeability as the explanation for natural urea-N pulse events. The relative increase in urea permeability of the gills at these times in the present study was comparable to or greater than that seen in toad bladder (Maffly et al. 1960; Eggena, 1973; Levine et al. 1973) or mammalian IMCD (Sands et al. 1987; Knepper and Star, 1990; Ashkar et al. 1995) when neurohypophyseal peptides activate facilitated diffusion carriers in these tissues.

Of the molecules tested, only the analogue thiourea caused a permeability increase during natural pulse events (Fig. 6) and the elevation was only approximately 15 % compared with that for urea permeability (Fig. 5C). This result provided the first evidence that a specific facilitated diffusion carrier might be involved in the toadfish gill because, in three such transport systems for urea which have been characterized (toad bladder carrier, Levine et al. 1973; mammalian erythrocyte carrier, Mayrand and Levitt, 1983; mammalian kidney vasopressin-regulated carrier, Chou et al. 1990), thiourea is similarly transported with much lower efficiency (<25 % of urea permeability).

Phloretin was evaluated because it is a potent inhibitor of all these carriers and, indeed, sensitivity to phloretin has become a standard test for the involvement of facilitated diffusion in urea transport (Marsh and Knepper, 1992). Unfortunately, the toxicity and non-specificity of phloretin are well recognized: it also inhibits glucose, chloride and other transport systems (Levine et al. 1973; Knepper and Star, 1990). In the present study, systemic phloretin administration at the commonly used extracellular level (250 μmol l⁻¹) was highly toxic, but the results, while equivocal, did not oppose the facilitated diffusion hypothesis (Table 1). Phloretin did eliminate pulsatile urea-N excretion and caused an internal accumulation of urea-N, but did not reduce the overall excretion rate of urea-N, which became continuous. We have demonstrated previously that, in highly stressed toadfish, urea production is elevated and urea-N leaks across the general body surface in a continuous fashion (Wood et al. 1995a), which provides a likely explanation for our observations. In the light of the results discussed below, it might be useful in the future to apply phloretin in the external sea water to achieve a more specific effect, although such experiments will be extremely expensive.

The strongest evidence in favour of facilitated diffusion as the mechanism of urea-N pulses was provided by the elevation of urea and thiourea concentrations in the external sea water. When external urea-N was set to 60 000 μmol-N l⁻¹, approximately three times plasma levels, each natural efflux pulse event was accompanied by an influx of urea-N (Figs 7, 8). Influx pulses per unit external concentration were exactly equal to efflux pulses per unit internal concentration (Fig. 9). This bidirectional transport capability, moving urea in both directions with equal efficiency in proportion to the relevant concentrations, is a key property of facilitated diffusion urea carriers such as UT-2 in the mammalian kidney (Chou et al. 1990). It is questionable whether such inward transport would ever occur across the gills of toadfish in nature, but the results clearly show that the capability is there.

A twofold higher concentration (60 mmol l⁻¹) of thiourea relative to urea (30 mmol l⁻¹) in the external sea water blocked the influx component by 73 % (Fig. 10). This inhibition by thiourea is comparable to that seen for the toad bladder carrier (Eggena, 1973), the mammalian erythrocyte carrier (Mayrand and Levitt, 1983; Olives et al. 1994), and the mammalian kidney vasopressin-regulated carrier (Chou and Knepper, 1989; You et al. 1993) and, indeed, appears to be a common property of facilitated diffusion urea transporters (Knepper and Star, 1990; Marsh and Knepper, 1992).

Four other pieces of evidence acquired in parallel studies from our laboratory also argue in favour of the hypothesis that urea-N pulses are due to the insertion or activation of a facilitated diffusion transport protein in the gills of the toadfish. First, using northern blot analysis, Walsh (1997) and colleagues have detected a 2 kilobase mRNA which hybridizes with the cDNA for the rat vasopressin-regulated transporter (Smith et al. 1995). Using rt-PCR with degenerate primers based on the mammalian sequence, they have amplified a 484 base pair fragment with 65 % homology to the rat carrier at the amino acid level. Second, injection of physiological doses of arginine vasotocin (AVT) activates pulsatile urea-N excretion in the toadfish gill (S. F. Perry, unpublished results). This parallels the way in which AVT increases urea-N permeability via the facilitated diffusion transporter in the toad bladder and the way in which the mammalian homologue arginine vasopressin increases urea-N permeability via the transporter in the mammalian IMCD (see above). Third, ultrastructural studies of toadfish gill pavement cells have shown that, during both natural pulse events and AVT-induced pulses, vesicles emanating from the Golgi bodies appear to fuse with the apical membranes, and the normally smooth apical surface is thrown into ridges (P. Laurent, unpublished results). This is clearly reminiscent of the situation in the ICMD, where vasopressin treatment causes the insertion of vesicles containing the transport protein into the apical membranes (Nielsen et al. 1996). Finally, treatment of toadfish with colchicine, an inhibitor of microtubule assembly and therefore of vesicle traffic, completely blocks pulsatile urea-N excretion (K. M. Gilmour, unpublished results). Converging physiological, pharmacological, molecular and ultrastructural evidence therefore all strongly supports the facilitated diffusion hypothesis (Fig. 1C).

This work was upported by NSF grant IBN-9507239 to P.J.W., NSERC research grants to C.M.W. and S.F.P., CNRS support and an NSERC International Fellowship to P.L., grants from NFR (Sweden) and the Magn. Bergwall Foundation to P.P. and travel grants from the Carnegie Trust for the Universities of Scotland and the Royal Society of Edinburgh (D. S. MacLagan Travel Grant) to K.M.G. P.P.’s travel was partially supported by a grant from NFR (Sweden) and partially by NIH grant ES05705. We thank Jimbo Luznar for collection of toadfish and Jean Paupe for excellent technical assistance.