The Effect of Perfusion Flow Rate and Adrenergic Stimulation on Oxygen Transfer in the Isolated, Saline-Perfused Head of Rainbow Trout (Salmo Gairdneri)

There is considerable controversy about whether oxygen transfer across fish gills is perfusion or diffusion limited. Results of recent studies using isolated, saline-perfused head preparations (Pettersson & Johansen, 1982; Pärt, Tuurala & Soivio, 1982; Pettersson, 1983) have shown that epinephrine enhances gill oxygen transfer. These observations have confirmed earlier speculation that in vivo injections of epinephrine in the eel (Anguilla anguilla) elevate arterial by increasing gill O₂ diffusing capacity (Steen & Kruysse, 1964; Peyraud-Waitzenegger, 1979). The effect of epinephrine on gill O₂ diffusing capacity is due to increased functional surface area (Booth, 1979) and/or increased epithelial permeability to oxygen (Pettersson, 1983). These results suggest that the fish gill is diffusion limited with respect to O₂ transfer.

A similar conclusion was reached by Fisher, Cobum & Forster (1969) from studies on carbon monoxide transfer across the gills of the bullhead catfish (Ameiurus nebulosas). Another method for investigating diffusion/perfusion limitations in gills is to alter flow (Q̇in) in perfused preparations while monitoring With this approach, Daxboeck, Davie, Perry & Randall (1982), using a spontaneously ventilating, blood-perfused trout preparation, and P. Pärt (personal communication), using a saline-perfused trout head preparation, have concluded that fish gills are primarily perfusion limited with respect to O₂ transfer. This conclusion was based on the observation that increasing Q̇in did not reduce ⁠. These results may not be representative of the situation in intact, resting fish because of the high concentration of epinephrine employed by Pärt (10⁻⁶moll⁻¹) and the probability of high levels of circulating catecholamines in the highly-stressed, blood-perfused trout preparation (experiments commenced following only 2–3 h of recovery from extensive surgery). In other words, it is not surprising that the gills are perfusion limited under these conditions, considering the known effect of epinephrine on increasing gill O₂ diffusing capacity.

In this study, we have used a novel approach whereby the gills were irrigated with hyperoxic water in order to obtain a stable, saline-perfused preparation in the absence of catecholamines. With this technique we have attempted to clarify the diffusion/perfusion limitations in perfused fish gills by modifying Q̇ and monitoring Moreover, we have investigated the action of epinephrine on gill O₂ transfer both at ‘normal’ Q̇and elevated Q̇ to test the hypothesis that epinephrine has a greater effect at high flow rates, due to increased diffusion limitations. Additionally, we have investigated the nature of the adrenergic receptors which control oxygen transfer by perfusing with a variety of alpha- and beta-agonists and antagonists.

Adult rainbow trout (Salmo gairdneri Richardson) of either sex, weighing between 200 and 300g, were obtained from Sun Valley Trout Farm (Mission, BC). They were kept outdoors in large, circular fibreglass tanks and supplied with aerated, dechlorinated Vancouver tap water (Na⁺ = 40/μquivl⁻¹, Cl⁻ = 20μquivl⁻¹, hardness = 12p.p.m. CaCO₃), at ambient temperature and photoperiod. Fish were fed daily with a commercial trout diet (Moore-Clark Co.). They were not fed for 48 h prior to experimentation.

Isolated, saline-perfused head preparations were obtained according to Payan & Matty (1975) with minor modifications. Fish were injected with sodium heparin (2500 USP units fish⁻¹) either directly into the caudal vein/artery or intraperitoneally. Following a 20 min period, the fish was decapitated just behind the pectoral fins. The head was placed onto an operating table and the gills were irrigated with water via a tube placed into the mouth. As quickly as possible the pericardium was cut and the ventricle severed. Next, the head was pithed by inserting a metal probe along the neural arch into the cranium (Oduleye, Claiborne & Evans, 1982; S. F. Perry & C. Daxboeck, in preparation). Pithing has been shown to cause no adverse effects on gill oxygen transfer or haemodynamics (S. F. Perry & C. Daxboeck, in preparation) and simplifies surgical procedures by eliminating head movement. Viscera were removed from the body cavity and a catheter (Clay, Adams, PE 90) was inserted into the bulbus arteriosus through the severed ventricle and tied in place. The gills were cleared of blood by perfusing with filtered (0·45 μm, Millipore) saline (Payan & Matty, 1975; HCO₃⁻ concentration was changed from 26 to 13 mmol 1⁻¹, a concentration more representative of in vivo values Perry, Haswell, Randall & Farrell, 1981) at a constant pressure of 50cmH₂O (Perry, Payan & Girard, 1984). A tightly fitting semi-circular plastic collar was placed into the abdominal cavity and sutured into position, thus making the body wall rigid. Finally, a catheter (Clay Adams, PE 200) was inserted into the dorsal aorta for collection of post-branchial saline. On average, the surgery was completed in approximately 8 min and the gills were ischaemic for periods never exceeding 2 min. At this stage, the gills were inspected visually to determine efficiency of clearing. Gills which displayed less than approximately 80% clearance of blood cells were discarded (less than 10% of all preparations).

The head was removed from the operating table and placed into a cylindrical plastic container and held in place by a thin rubber membrane which prevented leakage of the recirculating fresh water contained inside. The head was irrigated at about 500mlmin⁻¹ with aerated (mean Torr± 1·1s.E.) or hyperoxic (mean Torr ± 1·4S.E.) dechlorinated tap water (maintained at 9°C) by a tube placed into the mouth. The gills were perfused using constant pulsatile flow (∼1·5 ml min⁻¹100g⁻¹ body weight; pump frequency = 40 strokes min⁻¹) from a double reservoir, with gas-equilibrated saline (0·3% CO₂, 5% O₂, remainder N₂; Torr, Torr) using a cardiac pump (Harvard, model 1405) as modified according to Davie & Daxboeck (1983). Dorsal aortic pressure (Pda) was kept at zero in all experiments. Gas mixtures were supplied by gas mixing pumps (WÖsthoff). Input pressure (Pin) was monitored from a T-junction in the input catheter connected to a Statham P23Db pressure transducer and displayed on a chart recorder (Gould). Catheter resistance was measured upon termination of each experiment and Pin was corrected accordingly. Isolated heads which exhibited Pin greater than 80 cmH₂0 were indicative of trapped air bubbles in the gill vasculature and were discarded (approximately 10% of all preparations). Heads were perfused for a period of 10–15 min (or until Pin had stabilized) before experiments were started (time = zero). All experiments commenced with zero catecholamine content in the perfusate except where otherwise stated.

Four separate series of experiments were performed. In the first series, a comparison of oxygen transfer and haemodynamic variables was made between heads perfused with saline containing 10⁻⁷moll⁻¹ norepinephrine and irrigated with nor-moxic water (⁠ Torr) and heads perfused without norepinephrine but irrigated with hyperoxic water (⁠ Torr). The objective of these comparisons was to try and develop a relatively stable short-term preparation in the absence of high levels of catecholamines such that subsequent manipulations of perfusion flow rate and adrenergic stimulation could be performed without catecholamines in the perfusate. Indeed, all further experiments were performed using hyperoxic water. The second series consisted of manipulations of the perfusion flow rate (Q̇), to determine what effects these had on oxygen uptake ⁠, Pin, branchial vascular resistance to flow (Rbv) and perfusate flow distribution (Q̇in = Q̇da + Q̇av; where Q̇in is input perfusion flow rate set by the pulsatile pump, Q̇da is dorsal aortic (arterial) perfusate flow and Q̇av is venous drainage from the general head circulation as well as venous outflow from the central filamental sinus of gill filaments). Q̇in was adjusted by varying stroke volume while leaving pumping frequency constant. The third series of experiments was designed to study the effect of epinephrine (10⁻⁶ and 10⁻⁷moll⁻¹) on ⁠, Pin, Rbv and perfusate flow distribution. These effects were documented for preparations perfused at ‘normal’ flow rates (approximately 4 ml min⁻¹ ) and at increased flow rate (approximately 6 ml min⁻¹ ). The final series of experiments was designed to determine the nature of the receptor(s) (alpha or beta) responsible for the stimulatory effect of epinephrine on gill O₂ transfer. Experiments consisted of monitoring ⁠, ⁠, Pin, Rbv and perfusate flow distribution when isoproterenol (isoprenaline, beta agonist), phenylephrine (alpha agonist) and propranolol (beta antagonist) were added to final concentrations of 10⁻⁶mmoll⁻¹ in the perfusate.

A comparison of oxygen transfer and other perfusion variables in heads perfused with 10⁻⁷moll⁻¹ norepinephrine under conditions of external normoxia (⁠ Torr) with heads perfused without any catecholamine under conditions of external hyperoxia (⁠ Torr) is shown in Table 1. Not only were and increased in hyperoxic preparations, as one would predict, but so was the overall stability. In normoxic heads (with 10⁻⁷moll⁻¹ norepinephrine) there was considerable deterioration after 20 min as indicated by significant increases of Pin and Rbv as well as significant decreases in and ⁠. In contrast, there were no significant changes in any perfusion variable after 20 min during conditions of external hyperoxia. It is important to note that at the onset of perfusion, all haemodynamic variables (Q̇da, Q̇av, Pin, Rbv) were similar in both types of preparations. The stabilizing effect of external hyperoxia on Pin, in the absence of norepinephrine, is illustrated in Fig. 1. During 30 min perfusion with normoxic water (zero norepinephrine), Pin increased by approximately 15cmH₂O. In contrast, perfusing under conditions of external hyperoxia resulted in constant Pin throughout the 30min period.

Increasing perfusion flow rate from the ‘normal’ rate of 3·84 ml min⁻¹ to 5·59 ml min⁻¹ resulted in significant reductions in and (Table 2). was elevated (although not significantly) simply due to the increased flow rate. Input pressure to the branchial circulation was not increased significantly but Rbv was significantly decreased by 28%. No changes were observed in the perfusate flow distribution (in both cases, Q̇da represented approximately 87% of Q̇in). Control preparations displayed no changes in any of the measured variables following 7·5 min perfusion (Table 2). Following 15 min perfusion, Q̇in was decreased to a level below ‘normal’ (2·2 ml min⁻¹). Although in this situation, and were elevated to levels above those recorded at 7·5 min (increased flow), the overall changes were no different from those observed in controls after 15 min of perfusion. Thus, we conclude that reduced Q̇in did not affect or ⁠. Although Pin was significantly lower and Rbv higher than that recorded during either of the perfusion flow rates used, the perfusate flow distribution remained unchanged over this time period (Q̇da = 87% of Q̇in).

When the isolated head preparation was perfused for 20 min at ‘normal’ flow rate with physiological saline containing no epinephrine, there were progressive decreases in and as a consequence in (Table 3). With the addition of 10⁻⁷moll⁻¹ epinephrine, all measured variables remained stable throughout the experimental period, with the exception of Rbv, which decreased slightly (10%− Table 3). The addition of 10⁻⁶moll⁻¹ epinephrine resulted in significant increases in ⁠, and which endured throughout the 20-min period. At 10 min, Q̇da was significantly elevated while Q̇av was significantly reduced. Pin and Rbv declined throughout the experiment, although Pin was significantly lower than the pre-epinephrine value only after 20 min.

Table 4 summarizes the results of identical experiments performed on preparations perfused at higher Q̇in (approximately 1·5 ×’normal’). In the absence of epinephrine, there were more pronounced decreases in and hence with time. The proportional distribution of flow (Q̇da/Q̇av) decreased significantly during 20 min while Pin and Rbv remained constant. Addition of 10⁻⁷moll⁻¹ epinephrine under high flow conditions again had a stabilizing effect upon the measured variables over the experimental time course. Addition of 10⁻⁶moll⁻¹ epinephrine resulted in marked increases in ⁠, Q̇da and while Q̇av, Pin and Rbv were decreased significantly. Selected data from Tables 3 and 4 are presented graphically in Fig. 2.

Fig. 3 and Table 5 illustrate the effects of adrenergic agonists and antagonists on and O₂ extraction effectiveness while perfusing at ‘normal’ flow rate. In terms of effectiveness in increasing O₂ uptake; 10⁻⁶moll⁻¹ isoproterenol > 10⁻⁶ mol 1⁻¹ epinephrine > 10⁻⁷mol1¹⁻ epinephrine > control > 10⁻⁷moll⁻¹ epinephrine + propanolol > 10⁻⁶mol 1⁻¹ phenylephrine. Although deterioration of and O₂ ext. eff., in the presence of phenylephrine, was greater than that observed in controls (Fig. 3; Table 5), these changes were not significantly different from controls.

As well as increasing ⁠, and ⁠, the addition of 10⁻⁶moll⁻¹ isoproterenol caused significant reductions in Pin and Rbv as well as an increase in Q̇da and a decrease in Q̇av (Table 6). In contrast, the addition of 10⁻⁶moll⁻¹ phenylephrine to the perfusate resulted in significant decreases in ⁠, and - Despite the fact that under this condition, Pin and Rbv were significantly increased, there was no change in perfusate flow distribution. Epinephrine, in the presence of propranolol, resulted in a less stable preparation with respect to and compared to preparations perfused only with 10⁻⁷moll⁻¹ epinephrine (Table 3) and in this respect closely resembled the results of control perfusions (Table 3). The increases in Pin and Rbv during perfusion with epinephrine plus propranolol reflect the alpha-constrictory component of epinephrine.

The data in Tables 3–6 are somewhat complicated because of occasional significant differences between the various groups in their starting levels of ⁠, Pin and Rbv. However, due to the paired design of the experiments (i.e. each fish acting as its own control) we are confident that these differences have not affected the interpretation of the results.

Irrigation of the gills with moderately hyperoxic water (⁠ Torr) allowed us to achieve a stable, short-term (20min), saline-perfused head preparation in the absence of epinephrine. It is important to note that external hyperoxia did not cause an increase in branchial vascular resistance to flow, and we therefore believe that hyperoxic vasoconstriction was not occurring in the perfused head preparation. In contrast, Wood & Jackson (1980) have suggested that exposure of trout to hyperoxic water results in a decrease in functional surface area of the gill, presumably as a result of branchial vasoconstriction. A possible explanation for the discrepancy is the much higher external (350–650 Torr) employed by Wood & Jackson (1980) compared with the present study (250Torr). The stabilizing effect of hyperoxic water suggests that the progressive increase in Rbv and decrease in during normoxia is due to hyperoxic water does not simulate in vivo conditions yet we have used this technique to obviate the need for excessive levels of catecholamines in the perfusate in order to achieve a stable preparation. We believe the conditions for O₂ diffusion to be comparable in hyperoxic versus normoxic heads as indicated by similar values for Q̇da, Q̇av, Pin, Rbv and O₂ extraction effectiveness (≃10% greater in hyperoxic preparations). Using this technique, we were able to demonstrate that increasing Q̇in causes a reduction of This result indicates that perfusate transit time through the gill vasculature is an important factor determining oxygen extraction efficiency and that, clearly, there are diffusion limitations to oxygen transfer in the perfused head preparation, over the physiological range of Q̇ values employed. The inhibitory effect of increased Q̇in on was evident even though gill surface area was probably increased, and percentage utilization of saline O₂ by metabolism at post-lamellar sites (Daxboeck et al. 1982) was probably reduced during elevated Q̇in. The decrease of Rbv at high Q̇in reflects the compliant nature of the gill vasculature (Farrell, Daxboeck & Randall, 1979). The lack of an effect of reduced Q̇in on may be due to greater percentage utilization of saline O₂ by metabolizing tissue (transit time limitation), thereby obscuring any increase in gill O₂ diffusion. Alternatively, O₂ diffusion limitations may be absent at flow rates below normal. It is difficult to extrapolate results from perfused preparations to the intact animal, especially when saline is used as the perfusion medium. Problems associated with using saline as a perfusate include its low O₂-carrying capacity and, as a result, low ⁠, a linear O₂ dissociation curve and the possibility of severe and irreversible lamellar oedema (Ellis & Smith, 1983). However, results of a recent study (Perry, Lauren & Booth, 1984) have shown lack of oedema or other structural abnormalities in gills of saline-perfused trout heads. This finding, together with the stable Rbv observed in the present study, indicates that no additional structural barriers to O₂ diffusion are forming during the course of experiments. Other limitations of the perfused head are artificial and sometimes elevated gill ventilation (approximately five times the normal in vivo value in the present investigation) and absence of Pda which probably affect, respectively, the gradient profile across the gill and the pattern of flow through the branchial vasculature. Nevertheless, we believe the results of this study may be representative of the situation in intact fish because, as in in vivo experiments (Holeton & Randall, 1967; Cameron & Davis, 1970; Randall, 1970), was approximately 75% of under control conditions (zero catecholamine content, ‘normal’ Q̇). Moreover, the use of a low O₂ capacitance perfusate serves to decrease diffusion limitations to gill O₂ transfer, and therefore one might predict a greater effect of increased Q̇ on if blood was perfusing the gills. Results of previous perfusion studies (Daxboeck et al. 1982; P. Part, personal communication) have shown to be insensitive to flow adjustments and as such disagree with the results of this study. The lack of diffusion limitations in the previous studies was probably due to the high level of catecholamines present. P. Pärt (personal communication) used 10⁻⁶ mol 1⁻¹ epinephrine in the perfusate while the preparation of Daxboeck et al. (1982) may have contained high levels of circulating catecholamines due to the stressed condition of the animal. Catecholamines are known to enhance gill oxygen diffusing capacity (Wood, McMahon & McDonald, 1978; Pettersson & Johansen, 1982; Pärt et al. 1982; Pettersson, 1983), so it is not surprising that O₂ transfer was perfusion limited and therefore insensitive to adjustments of Q̇in in these experiments.

Knowing that increased flow rate reduces gill O₂ extraction efficiency or imposes additional diffusion limitations, one might predict that epinephrine would have a greater effect on stimulating O₂ transfer at higher perfusion flow rates. Indeed, this is the exact result observed in the present study (Fig. 2). This phenomenon may be of physiological significance during exercise, when cardiac output is increased (see Jones & Randall, 1978). Thus, epinephrine released into the blood may enable the exercising fish to maintain O₂ transfer and arterial constant (Stevens & Randall, 1967; Kiceniuk & Jones, 1977), even though transit time through the gill is reduced. Clearly, possible diffusion limitations imposed by increased cardiac output are offset by the effect of epinephrine on increasing gill O₂ diffusing capacity. The levels of epinephrine used in the present study (10⁻⁶–10⁻⁷mol 1⁻¹) are similar to levels measured in stressed fish (Nakano & Tomlinson, 1967; Mazeaud, Mazeaud & Donaldson, 1977) and resemble levels in exercising fish (D. J. Randall, personal communication).

As well as enhancing branchial O₂ uptake, epinephrine is known to affect gill haemodynamics (Keys & Bateman, 1932; Ostlund & Fange, 1962; Richards & Fromm, 1969; Bergman, Olson & Fromm, 1974; Wood, 1975; Girard & Payan, 1976; Payan & Girard, 1977; Booth, 1979; Claiborne & Evans, 1980; Pettersson, 1983), water flux (Isaia, Payan & Girard, 1979) and ionic transport (see Girard & Payan, 1980). The haemodynamic effects of epinephrine are bimodal, primarily composed of an alpha constrictory component acting at the level of the arterio-venous anastomoses (AVAs) and a beta dilatory component acting at the level of the afferent lamellar arterioles (ALAs; see review by Laurent, 1982). The net result is a decrease in branchial vascular resistance and an increase in Q̇da (Tables 3,4). More importantly, the haemodynamic effects of epinephrine cause a greater surface area of the gill to be perfused via lamellar recruitment (Holbert, Boland & Olson, 1979; Booth, 1979). Thus, the observed effects of epinephrine on oxygen, ion and water transport may be a result of lamellar recruitment. An alternative explanation is that epinephrine increases the permeability of the gill epithelium to small lipophilic and water-soluble substances as proposed by Haywood, Isaia & Maetz (1977). In an attempt to differentiate between these two possibilities, we examined the effects of adrenergic agonists and antagonists on oxygen uptake. Unlike the results of Pettersson (1983), which demonstrated stimulation of following both alpha- and beta-adrenergic stimulation (primarily alpha) in the cod head, the results of the present study show that, in the trout head is enhanced only by beta-receptor stimulation whereas alpha-receptor stimulation (both by addition of phenylephrine and beta blockade) had no significant effect on Beta-receptor stimulation is believed to cause dilation of ALAs (Girard & Payan, 1976; Claiborne & Evans, 1980), which is consistent with lamellar recruitment. However, beta-adrenergic stimulation also increases gill epithelial permeability to water and perhaps oxygen (see review by Isaia, 1984). Thus, the observed stimulation of following beta-adrenergic stimulation could be due to either of these two possibilities. Alpha-receptor stimulation is believed to cause constriction of AVAs (Claiborne & Evans, 1980). In our study, addition of phenylephrine resulted in an increase in Pin and Rbv with no change in Q̇da or Q̇av. These results are not consistent with constriction of AVAs but imply a site of action proximal to the AVAs. Pettersson (1983) has suggested the possibility of alpha-adrenergic constriction of efferent lamellar arterioles (ELAs) to explain this result. Constriction of ELAs would lead to lamellar recruitment (Pettersson, 1983) and result in increased but this was not observed in the present study (Fig. 3). The lack of an effect of phenylephrine on might be a result of reduced gill epithelial permeability to oxygen. Thus, even if lamellar recruitment were occurring following alpha-adrenergic stimulation, clearly the decrease in epithelial permeability must negate this effect so that the net result is no change in That this result was not observed by Pettersson (1983) may reflect species differences or the high diffusion limitations in his preparation, as suggested by low O₂ extraction effectiveness (23% without epinephrine, 43% with 10⁻⁶moll⁻¹ epinephrine). In other words, the stimulatory effect of lamellar recruitment on may have predominated since O₂ uptake was already severely limited.

The overall effect of epinephrine addition is an increase in gill oxygen uptake and is the result of both alpha- and beta-receptor stimulation. This effect can be accounted for by beta-adrenergic increase in surface area and/or increase in gill permeability plus alpha-adrenergic increase in surface area (if it occurs) minus alpha-adrenergic decrease in gill permeability (if it occurs). Clearly, the beta effects must predominate to explain the net result. This is similar to the argument which is used to explain the decrease in gill resistance following epinephrine application, in which the overall result is the sum of opposing alpha and beta effects, with the beta dilatory effects predominating (Payan & Girard, 1977).

In summary, the results presented above demonstrate that in the saline-perfused trout head preparation, and perhaps in the intact animal, there are diffusion limitations to branchial O₂ transfer which increase with increasing Q̇ Epinephrine enhances by increasing the number of lamellae perfused (increase in gill surface area) and/or by increasing gill epithelial permeability to oxygen. These effects are mediated primarily by beta-receptors. Alpha-receptor stimulation did not affect although our results suggest alpha-receptor-mediated constriction of ELAs (increased Rbv, no effect on Q̇da/Q̇av) which would cause lamellar recruitment. Thus, we speculate that alpha-adrenergic stimulation may decrease gill permeability to oxygen. Presumably, during epinephrine treatment the beta effects dominate so that the overall result is an increase in gill oxygen uptake.

This study was performed while SFP and CD were postdoctoral fellows at the University of British Columbia. SFP was supported by an NSERC Postdoctoral Scholarship. GPD wishes to thank Dr P. W. Hochachka for allowing him to get away from the blenders for a few weeks. Financial support for this study was provided by an NSERC operating grant to Dr D. J. Randall. We appreciate helpful comments provided by him even after several devastating floods.