Influence of Hypoxia and Adrenaline Administration on Coronary Blood Flow and Cardiac Performance in Seawater Rainbow Trout (Oncorhynchus Mykiss)

In fish, unlike mammals, the presence of a coronary circulation is species-dependent and closely related to activity level (Davie and Farrell, 1991a). This relationship implies that a coronary blood supply is only essential for maintaining cardiac function in fish capable of high cardiac work (i.e. high myocardial oxygen demand). Early experimental evidence for this generalization comes from the survival of coronary-ablated fish (Daxboeck, 1982; Farrell and Steffensen, 1987) and from the 35% reduction in the maximum swimming speed of coronary-ablated trout (Farrell and Steffensen, 1987).

In recent years, the role of the coronary circulation in determining fish cardiac performance has been defined further. Gamperl et al. (1994) have shown that adrenaline-stimulated cardiac performance in trout during normoxic conditions is not affected by coronary ablation. In addition, Davie and Farrell (1991b) and Davie et al. (1992) have shown that perfusion of the coronary circulation with oxygenated red cell suspensions only restored the cardiac power output of hypoxic dogfish (Squalus acanthias) and eel (Anguilla dieffenbechii) hearts to 50–75% of levels measured during normoxia. These experiments suggest that oxygen delivery from the luminal blood to the myocardium is not diffusion-limited during normoxic conditions in fish with a small percentage of compact myocardium and that increases in coronary oxygen delivery are unlikely to provide for near-maximal cardiac performance when venous oxygen content is diminished. Axelsson and Farrell (1993) measured coronary blood flow in the coho salmon (Oncorhynchus mykiss) during hypoxia, following the injection of adrenaline and during spontaneous activity. However, while these authors were able to show that resting coronary blood flow was approximately 1% of cardiac output and that coronary blood flow increased upon exposure to all three experimental conditions, their study did not address the question of whether elevations in coronary blood flow are capable of supporting near-maximal cardiac performance during hypoxia.

Farrell (1987) using an in vitro perfused rainbow trout heart revealed that arterial pressure, adrenoreceptors, extravascular compression and cardiac metabolism are involved in the determination of coronary blood flow in fish. In addition, in vivo studies on coho salmon (Axelsson and Farrell, 1993) showed that α-adrenergic constriction, β-adrenergic vasodilation, dorsal aortic pressure and hypoxic exposure can all affect coronary blood flow. However, no studies have investigated the interactive effects of these variables on coronary flow in vivo.

We measured in vivo coronary blood flow, cardiac output and dorsal aortic pressure in trout (Oncorhynchus mykiss) exposed to adrenergic stimulation under conditions of normoxia and hypoxia. The goals of this study were twofold: (1) to investigate whether cardiac performance is depressed under the combined conditions of hypoxaemia and high output pressures, irrespective of alterations in coronary blood flow; and (2) to determine whether hypoxia (hypoxaemia) alters adrenoreceptor-mediated control of coronary blood flow. Adrenergic stimulation is an appropriate model for studying the in vivo relationship between cardiac performance and coronary blood flow because adrenaline increases both cardiac output and systemic vascular resistance; the concomitant increase in cardiac power output increases oxygen demand by the myocardium (Graham and Farrell, 1990).

Seawater-adapted rainbow trout [Oncorhynchus mykiss (Walbaum)] (900–1500 g) were obtained from Merlin Fish Farms (Wentworth, Nova Scotia) and held in tanks (1 m×1 m×1.5 m) for at least 1 month prior to experimentation. Fish were fed daily, to satiation, on a diet of commercially available feed pellets, but were fasted for 48 h prior to surgery. Photoperiod was 12 h:12 h light:dark. Experiments were conducted between 1 May and 1 August 1992.

Dorsal and ventral aortic cannulae were prepared from PE 50 polyethylene tubing (Clay Adams Intramedic), total length 0.8 m. Bubbles were made 5.0 cm from the tip of the dorsal cannula and 2.5 and 5.0 cm from the tip of the ventral aortic cannula. Trout (1260±102 g) were anaesthetized (0.1 g l⁻¹ tricaine methane sulphonate, MS 222) and placed supine in a wetted chamois leather sling. The fish were fitted with a dorsal aortic cannula (Smith and Bell, 1964), after which intermittent retrograde irrigation with sea water containing anaesthetic was begun. Ventral aortic cannulation was subsequently performed using a new technique (Fig. 1). For ventral aortic cannulation, the trout was held upright in the sling, a hole was made at the side of the mouth with a 13 gauge steel needle and a short piece of heated-flared PE 160 tubing was exteriorized through the resultant hole. A 21 gauge needle was inserted into the cartilage of the tongue (while slowly rotating the needle), just prior to the junction of the second gill arches at an angle of between 30 and 40°. Once resistance ceased, the needle was removed and a PE 50 cannula, with indwelling stainless-steel wire, was advanced fully into the resultant hole. The steel wire was then withdrawn and the position of the PE 50 tubing was adjusted until blood flowed freely into the cannula. After the cannula had been positioned to maximize blood flow, a wire staple (0.75 cm long, 0.4 cm wide; fashioned from a paperclip) was used to secure the cannula to the tongue at a position just posterior to the first bubble. Finally, the remaining length of the cannula was threaded through the side of the mouth until the second bubble rested firmly against the flared PE 160 tubing, and a constricting knot was tied around the PE 160 tubing to prevent cannula movement.

The main advantages of this technique are: (1) a double cannulation can be completed in 5–7 min and therefore only limited irrigation of the gills is necessary; (2) there is no leakage from the ventral aorta because the cartilaginous tissue of the tongue closes tightly around the cannula; and (3) pressure records and blood samples can be obtained for periods in excess of a week. Although blood can be withdrawn from all ventral aortic cannulae, satisfactory pulsatile pressure records can only be obtained from approximately 50% of fish cannulated using the above technique. Once surgery had been completed, fish were placed into black Perspex boxes (11 cm×16 cm×75 cm) to recover. Boxes were supplied with aerated sea water (8.6±0.5°C) at a flow rate of 4 l min⁻¹.

After 24–48 h of recovery, to determine a point where fish were experiencing hypoxaemia but not bradycardia, venous ⁠, arterial ⁠, heart rate (f_H) and dorsal aortic pressure (P_DA) were monitored in trout (N=7) as the sea water was gradually reduced from 19.3 to 8.7 kPa. ⁠, f_H and P_DA were measured in trout after every 2 kPa drop in ⁠; time between blood samples (0.25 ml) ranged from 25 to 30 min. Seawater oxygen content was manipulated by bubbling a controlled mixture of air and nitrogen through a gas exchange column (8 cm in diameter, 45 cm in length) and was monitored with a YSI oxygen meter (model 50). Water oxygen content (mg l⁻¹) was converted to partial pressure on the basis of calibrations obtained with a thermostatted Radiometer O₂ electrode (model 5046-0). If fish struggled during the experiment, a 10 min ‘recovery’ period was allowed before any further samples were taken. Trout that struggled repeatedly were discarded.

Trout (1400±47 g) (N=9) were anaesthetized (0.1 g l⁻¹ MS 222) and placed supine in a wetted chamois leather sling. The fish were quickly fitted (approximately 45 s) with a dorsal aortic cannula (PE 50), after which retrograde irrigation with sea water containing anaesthetic (0.05 g l⁻¹ MS 222) was begun. A 3–4 cm incision was made through the skin and muscle at a position overlying the ventral aorta and anterior aspect of the pericardium. After cutting through the pectoral girdle and expanding the resultant cavity with tissue spreaders, the anterior portion of the pericardium was cut to expose the ventral aorta and coronary artery. The connective tissue from the anterior portion of the ventral aorta, and that attaching the coronary artery to the ventral aorta and the anterior bulbus, was subsequently removed to facilitate the placement of the Doppler flow probes (Fig. 2). Once the flow probes were in place, the musculature and skin were closed using continuous silk sutures. The Doppler probe leads were secured to the skin at the anterior apex of the incision, and at a position just posterior to the pectoral fins. The operation generally took between 1 and 2 h, and the bleeding was usually minimal. The integrity of the pericardium was not restored by suturing because of time constraints. Once surgery had been completed, the fish were placed into black Perspex boxes to recover. During recovery and subsequent experiments, the boxes were supplied with aerated sea water (10.9±0.9°C) at a flow rate of 4 l min⁻¹.

Flow probes (Fig. 2) were constructed by implanting piezoelectric crystals with 80 cm leads (Crystal Biotech, Hopkinton MA) into a 1 cm length of Tygon tubing (i.d. 2.2 mm) and into a 3–4 mm length of PE 50 or PE 60 tubing; these probes were utilised for the measurement of ventral aortic flow (cardiac output, Q̇) and coronary flow (q̇_cor), respectively. Both sections of tubing were split to facilitate placement on their respective vessels and had small notches to allow for crystal attachment. Piezoelectric crystals were secured in the notches with cyanoacrylate cement after thoroughly roughening the area with a scalpel blade. A notch at the anterior end of the ventral aortic cuff, and a shallow groove along the length of the ventral aortic cuff (for coronary cuff placement), ensured that the coronary artery was not bent or deformed as it passed over the ventral aortic cuff and through the coronary artery cuff. Probes were fitted with tie-strings to prevent tube diameter from increasing at elevated blood pressures.

After a 24–48 h recovery period, cardiovascular variables (Q̇, f_H, P_DA and q̇_cor) were measured in trout before and after the injection of 1.0 μg kg⁻¹ adrenaline (Sigma Chemical Co., St Louis, MO) under three conditions: (1) normoxia; (2) following 30 min of graded hypoxia (final 12 kPa); and (3) after recovery from hypoxia. Treatments were separated by 2.5 h. During the normoxic treatments, ‘resting’ cardiovascular variables and were measured 5 min prior to epinephrine injection. In the hypoxic treatment, cardiovascular variables were continuously measured during the graded hypoxia, and ‘resting’ Q̇, f_H, P_DA, q̇_cor and were measured 7–10 min after had stabilized at 12 kPa. In all treatments, Q̇, f_H, P_DA and q̇_cor were recorded for 20 min following adrenaline injection. Adrenaline was administered through the dorsal cannula in 0.2–0.4 ml of carrier of saline. The second normoxic period was used to provide some preliminary information on the duration of the recovery period required to restore the scope for adrenergically stimulated cardiovascular performance and to ensure that hypoxia-induced differences in adrenaline-stimulated cardiovascular performance were not solely due to repeated adrenaline injection.

P_DA was measured by attaching the dorsal aortic cannula to a Gould Statham (model P23-10) pressure transducer connected to a amplifier recorder (Asea Brown and Boveri; model SE-120). Pressure was calibrated daily against a static water column.

Mean ventral aortic flow (Q̇) was measured by connecting a pulsed Doppler flowmeter (model 545c-4; Bioengineering, University of Iowa) to an RC integrator and an amplifier recorder. Coronary flow (q̇_cor) was measured by connecting the Doppler flowmeter to an amplifier recorder. In order to determine absolute flow rates (ml min⁻¹) an in situ post-mortem calibration of the ventral aortic flow probe and an in vitro calibration of the coronary flow probe were performed using a peristaltic pump (Gilson, MinipulsII) and human blood (approximate haematocrit 20%). To calibrate the ventral aortic flow probe, the sinus venosus and atrium were removed, the ventricle was bisected laterally and the peristaltic pump outflow tubing (PE 160) was tied into the ventricular lumen (see Gamperl et al. 1994, for further details). To calibrate the coronary flow probe, the coronary artery was cut and the flow probes (with coronary artery in place) were placed in a Petri dish filled with teleost saline (Hoar and Hickman, 1983). After tying a piece of 27 gauge needle, with an attached length of PE 10 tubing, into each end of the segments of coronary artery, the coronary flow probe was calibrated at flow rates between 0.05 and 0.55 ml min⁻¹.

and were measured using a thermostatted Radiometer O₂ electrode (E 5046-0, Denmark) connected to an amplifier recorder. The electrode was calibrated with humidified N₂ and air prior to each experiment, and the calibration was rechecked with air prior to each sample.

Resting f_H, or f_H at a particular time post-injection, was determined by measuring the number of systolic peaks on the P_DA record during a 30 s interval, the interval being 15 s on either side of the desired time. Mean P_DA was calculated as [systolic pressure + 2(diastolic pressure)]/3 (Burton, 1972; Wood et al. 1979). Stroke volume (V_S) was calculated from Q̇/f_H. Coronary vascular resistance (R_cor) was calculated as P_DA/q̇_cor (Axelsson and Farrell, 1993). Statistical differences (P<0.05) between treatments (normoxia, hypoxia, normoxia II) for post-injection cardiovascular variables were determined using a repeated-measures analysis of variance (Proc GLM; SAS Institute Inc.) with multiple contrasts. Comparisons between cardiovascular variables at normoxia (⁠ 19.3 kPa) and at various levels of hypoxia (⁠ 17.3–12 kPa) were made using the same analysis. Tests for between-subject effects (i.e. type effects, see Results) indicated that response type had no effect on resting or post-injection cardiovascular variables.

As was gradually lowered from 19.3 to 8.7 kPa, both arterial and venous decreased (Fig. 3). In all fish, fell linearly (0.92<r²<0.99), while the decline appeared to be curvilinear. As fell, decreased more (by 10.7–12 kPa) than did (1.5–3.3 kPa). P_DA and f_H remained at the normoxic value until was reduced below approximately 10.7–11.3 kPa. However, a severe bradycardia was seen in five of nine trout as dropped below 10.7 kPa. Because severe bradycardia would have complicated the comparison of cardiovascular performance and q̇_cor between hypoxic and normoxic fish, a of 12 kPa was chosen for subsequent experiments. At this ⁠, and were 4.7–6.0 kPa and 2.0–3.3 kPa, respectively; levels approximately 40–50% of those seen during normoxia.

Cardiac output, q̇_cor, P_DA, f_H, V_S and R_cor during the initial normoxic period averaged 18 ml min^₋₁ kg⁻¹, 0.14 ml min⁻¹ kg⁻¹, 2.8 kPa, 64 beats min⁻¹, 0.29 ml kg⁻¹ and 22.5 kPa min kg ml⁻¹, respectively. During graded hypoxia, the threshold for hypoxia-induced alterations in cardiovascular variables (q̇_cor, Q̇, f_H, V_S) appeared to be between 13 and 14.5 kPa (Fig. 4). After 30 min of exposure to graded hypoxia and at a of 12 kPa, changes in Q̇, q̇_cor, V_S and R_cor, but not P_DA and f_H, were significant (P<0.05) (Table 1). Q̇and V_S increased by 3.1 ml min⁻¹ kg⁻¹ (17%) and 0.07 ml kg⁻¹ (25%), respectively, while R_cor decreased by 6.3 kPa min kg ml⁻¹ (28%). Absolute q̇_cor increased significantly (0.052 ml min⁻¹ kg⁻¹; 36%) following exposure to graded hypoxia. Although the percentage increase in q̇_cor was greater than that for Q̇, the percentage of Q̇delivered to the coronary circulation (q̇_cor/ Q̇ ×100) was not increased significantly (0.10>P>0.05). fell from 11.7 to 5.1 kPa during the 30 min hypoxic exposure.

Two and a half hours after the trout had been returned to normoxia , values for most resting variables (Q̇, q̇_cor, f_H, V_S and R_cor) were intermediate between those measured during the initial normoxic period and those recorded during the hypoxic exposure (Table 1). A notable exception was P_DA, which was marginally lower than those of the normoxic and hypoxic groups (P=0.07). during the second normoxic period was 10.7 kPa, a level significantly lower than that measured during the initial normoxic period (Table 1).

Adrenergic stimulation during normoxia or hypoxia resulted in two types of cardiovascular responses (Figs 5, 6). Type 1 was characterised by an initial post-injection increase in Q̇, which reached a maximum value within 4–6 min. Type 2 was characterised by an initial drop in Q̇, followed by a steady increase until maximum values were reached at approximately 10–14 min. As shown in Gamperl et al. (1994), these two types of response to adrenaline injection were mediated by type-specific differences in the temporal pattern of f_H and V_S responses. Type 2 trout showed a greater post-injection bradycardia and a 2–4 min delay in the increase in V_S.

Differences in the temporal pattern of changes in coronary flow were also evident between fish exhibiting type 1 and type 2 responses. In type 1 fish, q̇_cor increased immediately following adrenaline injection and reached a maximum within 6–8 min (Figs 5, 7). In type 2 fish, increases in q̇_cor were delayed by approximately 4 min and maximum q̇_cor values were not reached until 8–12 min post-injection (Figs 6, 8). In type 2 fish, q̇_cor, measured as a percentage of Q̇, increased immediately following adrenaline injection and remained elevated until 6 min post-injection. In contrast, type 1 fish displayed no post-injection alterations in the percentage of Q̇delivered to the coronary circulation (see Fig. 10).

The degree to which hypoxic exposure affected adrenergically stimulated cardiovascular variables and q̇_cor was quantitatively similar in both response types (Figs 5, 6). Adrenaline injection into normoxic trout increased Q̇ by 45%, q̇_cor by 66% and P_DA by 100% (Figs 7, 8; Table 2). However, adrenaline injection into hypoxic fish resulted in a 50% smaller increase in Q̇compared with that in fish during the initial normoxic treatment, an effect that was evident despite similar increases in q̇_cor and P_DA (Figs 5–8; Table 2). The smaller increase in adrenaline-stimulated Q̇ during hypoxia, compared with that during normoxia, was due to differences in both V_S and f_H. For example, in type 1 normoxic fish, V_S increased by approximately 0.12 ml kg⁻¹ while f_H returned to pre-injection levels by 4–6 min. However, in hypoxic fish, V_S only increased by approximately 0.08 ml kg⁻¹ and f_H remained below pre-injection levels for at least 15 min (Fig. 9). Although hypoxic fish had a 50% smaller increase in Q̇, post-injection maximum Q̇ in hypoxic fish (25.05±1.72 ml min⁻¹ kg⁻¹) was not significantly different from that measured in normoxic fish (26.17±1.3 ml min⁻¹ kg⁻¹) because hypoxic fish had a higher pre-injection Q̇.

Although maximum post-injection q̇_cor (measured as% Q̇ or ml min⁻¹ kg⁻¹) was significantly greater during hypoxia than during normoxia (Table 2; Fig. 10), this difference was related to discrepancies that existed prior to adrenaline injection. Post-injection increases in q̇_cor were not significantly different between hypoxic and normoxic fish (Table 2). Although R_cor increased following adrenaline injection, differences in R_cor between resting hypoxic and normoxic fish were maintained (Figs 7 and 8). R_cor was 39% higher at rest and 27% higher after adrenaline injection in normoxic fish compared with values in hypoxic fish (Tables 1, 2).

Post-injection changes in q̇ _cor (as% Q̇ or ml min⁻¹ kg⁻¹) and P_DA during the second normoxic period were comparable to those observed during the initial normoxic period (Table 2; Figs 5–8, 10). In contrast, although the maximum increase in Q̇ (6.3 ml min⁻¹ kg⁻¹) was higher than that measured during hypoxia (3.9 ml min⁻¹ kg⁻¹), it was significantly lower than that recorded during the initial normoxic treatment (8.2 ml min⁻¹ kg⁻¹). Post-injection maximum Q̇ during the second normoxic period was not significantly different from that measured in either of the other two treatments. The magnitudes of changes in R_cor, measured during the second normoxic period, were intermediate between those obtained in the initial normoxic and the hypoxic treatment.

During the 30 min of graded hypoxia, V_S and Q̇ increased, while f_H decreased (Fig. 4). The drop in f_H that was concomitant with hypoxia is thought to be caused by chemoreceptor-mediated inhibitory activity in the efferent cholinergic fibres of the vagus nerve (Randall and Smith, 1967; Wood and Shelton, 1980b; Smith and Davie, 1984; Fritsche and Nilsson, 1989). However, although an increase in V_S which results in the maintenance or elevation of Q̇ is often observed during hypoxic bradycardia (Holeton and Randall, 1967; Cech et al. 1977; Wood and Shelton, 1980a; Fritshe and Nilsson, 1989), the mechanisms which mediate this effect are unclear. Numerous factors are known to increase V_Sin vivo or in vitro, including elevations in the levels of circulating catecholamines (Wood and Shelton, 1980a; Farrell, 1984; Gamperl et al. 1994), increases in cardiac filling time ( i.e. decreases in f_H; Farrell et al. 1989) and increases in cardiac filling pressure (venous pressure) (Farrell, 1984; Farrell and Jones, 1992). In rainbow trout exposed to gradual hypoxia, increases in plasma adrenaline and noradrenaline levels are not detected until $(PWO2)$ falls below 6.7 kPa (approximate $(PWO2)$ 2.8 kPa) (Perry and Reid, 1992). Because increases in V_S were detected at levels between 12 and 13.3 kPa (approximate 4.7–6.0 kPa), humoral adrenergic stimulation of the heart is highly unlikely. While a decrease in f_H did occur during hypoxic exposure, the magnitude of this change was relatively minor (Fig. 4). At a of approximately 13.3 kPa, f_H had only fallen by 2–3 beats min⁻¹ (5%), while V_S had increased by about 15% (from 0.15 to 0.17 ml kg⁻¹). In addition, at the final of 12 kPa, f_H had decreased by approximately 5 beats min⁻¹ (8%) while V_S had increased by 25%. Although a decrease in f_H will increase the time available for atrial and ventricular filling (Farrell, 1984), it is probable that some other factor, in combination with the small change in f_H, mediated the observed change in V_S. Data from Farrell et al. (1982) and Farrell (1984) for in situ perfused hearts have shown that preloads (venous pressures) of 0.03–0.05 kPa (0.3–0.5 cmH₂O) are sufficient to generate Q̇ levels comparable to those seen in resting fish, and that small increases in preload can substantially increase V_S and Q̇. Hypoxic exposure in fish causes ventilation volume to increase (Holeton and Randall, 1967; Nonnette et al. 1993) and results in a synchrony between f_H and breathing (Randall and Smith, 1967). Because both of these adjustments have been suggested as possible mechanisms that increase venous return (Farrell, 1984), it is feasible that a hypoxia-induced increase in venous return partially mediated the elevated V_S that was concomitant with hypoxia. Although this is an attractive hypothesis for explaining the hypoxia-induced increase in V_S, Cech et al. (1977) found that V_S increased in the flounder despite non-significant changes in f_H or venous pressure. It is clear that the mechanisms mediating increases in V_S during hypoxia require further investigation.

Adrenaline injection into normoxic trout resulted in two distinct types of cardiovascular response; a type 1 response characterised by a gradual increase in post-injection Q̇, and a type 2 response characterised by an initial fall in Q̇, followed by a gradual increase to a similar peak Q̇as for type 1 fish (see Figs 5–8). This confirms the results of Gamperl et al. (1994), who identified two response types in rainbow trout acclimated to 5°C. Hypoxic exposure failed to alter the response type of any individual fish (see Figs 5–8), a result suggesting that response type is not influenced by blood oxygen tension and/or content.

Adrenaline injection (1.0 μg kg⁻¹) during normoxia increased Q̇ and P_DA by 45% and 100%, respectively. Because these values are comparable to those obtained by Gamperl et al. (1994) for coronary-ablated trout (Q̇48%; P_DA 100%), the results of the present study provide evidence that oxygen supplied by the venous (luminal) blood is sufficient to support elevated cardiac work during normoxia even though q̇_cor normally increases. As in the study of Gamperl et al. (1994), post-injection increases in Q̇ were mediated by elevations in V_S that more than compensated for the effects of the pressor-stimulated bradycardia. Although the decrease in f_H associated with adrenaline injection may have increased the time available for atrial filling by vis-a-tergo mechanisms (Farrell et al. 1989), it is probable that direct adrenergic stimulation of the heart was the predominant factor mediating increases in V_S. This conclusion is based on (1) the observation that relatively small decreases in f_H (2–4 beats min⁻¹) were concordant with maximum post-injection Q̇ (e.g. Fig. 9); and (2) evidence showing that adrenergic stimulation increases myocardial contractility and the heart’s sensitivity to filling pressure (preload) (Farrell, 1984; Farrell et al. 1986; Franklin and Davie, 1992).

Maximum post-injection Q̇was the same in hypoxic and normoxic trout. However, hypoxic trout had a diminished capacity to increase Q̇ over pre-injection levels compared with normoxic trout (Table 2), despite similar increases in q̇_cor (ml min⁻¹ kg⁻¹). One possible explanation for the failure of hypoxic hearts to increase Q̇ above levels measured in normoxic trout is that post-injection V_S during normoxia was already approaching a maximum limit, and that the inability to increase V_S further was related to a reduction in cardiac emptying in the face of elevated output pressures (P_DA, and possibly P_VA). Kiceniuk and Jones (1977) showed that swimming rainbow trout can still elevate Q̇and V_S at values of P_VA greater than 6.7 kPa. Because our estimated value of P_VA at Q̇ _max (5.3 kPa) is less than those in Kiceniuk and Jones (1977), these data suggest that the elevated blood pressures associated with adrenaline injection did not limit V_S in hypoxic trout. In contrast, Gamperl et al. (1994) suggested that most of the scope for adrenaline-stimulated increases in V_S (Q̇) occurs below 0.5 μg kg⁻¹ and that the large increases in systemic vascular resistance (output pressure) that accompany doses greater than 1.0 μg kg⁻¹ limit the ability of adrenaline to increase V_S (systolic emptying).

A second explanation for the inability of hypoxic fish to increase Q̇ above levels achieved during normoxia is that increases in q̇_cor (coronary oxygen delivery) during hypoxia were not sufficient to support further increases in myocardial power output. In vitro, the maximum tetanic force developed by cardiac muscle is reduced by hypoxia (Gesser et al. 1982; Gesser, 1985) and this effect is seen as a reduction in the ability of hypoxic perfused preparations to generate pressure (Farrell et al. 1989). In addition, while perfusion of the coronary circulation with air-saturated red cell suspensions (10% haematocrit) partially restored the power output of severely hypoxic dogfish (Davie and Farrell, 1991b) and eel (Davie et al. 1992) hearts , the levels of cardiac power output achieved were 25–45% lower than those measured during normoxia.

In our study, myocardial oxygen delivery through the coronary artery is estimated to be 10% lower in hypoxic trout than in normoxic trout, at maximum cardiac output. This estimate is based on (1) the reported q̇_cor measurements, (2) the assumption that adrenergic contraction of the spleen leads to similar post-injection levels of haematocrit in hypoxic and normoxic fish (Pearson and Stevens, 1991), and (3) the in vivo oxygen dissociation curves for rainbow trout blood at comparable temperatures (Perry and Reid, 1992). Although this deficit in coronary O₂ delivery seems unlikely to explain the 50% decrease in the scope for Q̇ increases, it must be remembered that the coronary O₂ supply in salmonids supplements rather than replaces the luminal O₂ supply (Farrell and Jones, 1992).

Although the existence of a maximum adrenaline-stimulated V_S and a limitation in the ability of hypoxic trout to increase q̇_cor are the two most probable explanations for the failure of hypoxic trout to increase Q̇ to the same extent as normoxic trout, other possible explanations do exist. These include constraints on V_S imposed by the lack of vis-a-fronte filling (i.e. the absence of an intact pericardium), hypoxia-related differences in the regulation of venous return to the heart, differences in adrenergic tone and a hypoxia-induced decrease in myocardial adrenoreceptor density and/or affinity.

Trout that had been allowed to recover from hypoxia for 2.5 h also had a maximum Q̇similar to that of normoxic trout and a diminished scope for adrenaline-stimulated increases in Q̇. However, because it is clear that trout during the second normoxic period had not fully recovered from the hypoxic exposure, this result does not assist in resolving why hypoxic trout had a reduced increase in adrenaline-stimulated Q̇. Resting trout, in the second normoxic period, had an elevated Q̇ (V_S), a lower , and a diminished P_DA compared with trout during the initial normoxic period. Because the lower probably indicates that was also depressed during the second normoxic period, and because Steffensen and Farrell (1994) found that the homeometric lowering of P_VA by coronary-ablated coho salmon was concomitant with a diminished ability of the heart to maintain cardiac power output, it is probable that the power-generating ability of the hearts in our ‘recovered’ trout was compromised prior to adrenaline injection.

Resting q̇_cor, during normoxia, was 0.143 ml min⁻¹ kg⁻¹ or 0.84% of Q̇, an estimate very comparable to the value of 1.1% Q̇measured by Axelsson and Farrell (1993) in coho salmon (Oncorhynchus kisutch). Taken together, these results indicate that q̇_cor in salmonids is approximately 1% of resting Q̇. The only other direct measurement of q̇_cor in fish was recently provided by Davie and Franklin (1993). Mean q̇_cor in one of the paired coronary arteries of an anaesthetized school shark (Galeorhinus australis) was estimated to be 0.103 ml min⁻¹ kg⁻¹.

q̇_cor in our trout increased as the fell below 13.3 kPa and was 36% (0.052 ml min⁻¹ kg⁻¹) greater in hypoxic trout than in normoxic trout prior to adrenaline injection (Table 1). Although an increase in q̇_cor was also observed in hypoxic coho salmon (Axelsson and Farrell, 1993), the magnitude of the increase (60%) was greater than that reported here. The difference between the two studies may be due to three factors. First, Axelsson and Farrell (1993) only used two fish to investigate the effect of hypoxia on coronary flow. Second, the increase in P_DA (60%), and presumably P_VA, associated with hypoxia in the coho salmon would have increased coronary driving pressure and myocardial power output (i.e. oxygen demand) (Graham and Farrell, 1990). In our study, no change in P_DA was associated with hypoxic exposure. Third, the level of hypoxia used by Axelsson and Farrell (1993) (8–10 kPa) very probably lowered to a greater degree than in the present study (see Fig. 3). Because oxygen delivered to the heart by the venous blood appears to be limited by the partial pressure of oxygen (Davie and Farrell, 1991a), it is likely that Q̇and V_S in the coho salmon were more dependent upon O₂ supplied by the coronary artery. Taken together, the results of Axelsson and Farrell (1993) and those in the present study suggest that elevations in q̇_cor are important for determining cardiac performance during environmental hypoxia. For example, in our study a 36% increase in q̇_cor was associated with the 25% increase in myocardial power output (estimated from Q̇ ×P_DA) that was concomitant with hypoxic exposure.

Increases in q̇ _cor during experimental hypoxia were associated with a significant decrease in coronary resistance (28%) but no significant increase in P_DA (Table 1), indicating that the observed increase in q̇_cor was mediated by coronary vasodilation and/or the decrease in f_H. This finding confirms the results of Axelsson and Farrell (1993), who found that an important component of the increase in q̇_cor was not dependent upon α-adrenergic, cholinergic or physical (arterial pressure) mechanisms. Although changes in P_DA were not important for determining q̇_cor in resting hypoxic fish, it is clear that increases in P_DA associated with adrenergic stimulation (see below) or more severe hypoxia (Axelsson and Farrell, 1993) may mediate elevations in coronary flow.

Axelsson and Farrell (1993) indicated that -adrenergic mechanisms and/or increases in cardiac metabolism can potentially alter q̇_cor to the myocardium. However, -adrenergic control of q̇_cor was probably not the predominant factor mediating the increased q̇_cor in hypoxic trout since (1) Perry and Reid (1992) have shown that endogenous catecholamine release does not occur in rainbow trout until falls below 6.7 kPa; and (2) perfusion of the coronary circulation (coronary artery and associated arterioles) with adrenaline in vitro, and injection of adrenaline in vivo, indicates that α-adrenoreceptor effects dominate β-adrenoreceptor effects, i.e. that vasoconstriction predominates (Farrell and Graham, 1986; Farrell, 1987; Axelsson and Farrell, 1993). Results from Farrell (1987) and Axelsson and Farrell (1993) suggest that there is a small tonic α-adrenergic constriction of the coronary vasculature and that metabolism-related vasodilation of the coronary circulation, as occurs in mammals (Feigl, 1983), could override direct sympathetic vasoconstriction. Therefore, it is possible that the increase in q̇_cor and decrease in R_cor during hypoxia were due to a metabolically related coronary vasodilation.

f_H decreased by approximately 5 beats min⁻¹ during hypoxia, an effect which would have increased the heart’s diastolic period and reduced the fraction of the cardiac cycle occupied by systole. Because an increase in the diastolic period would reduce vascular compression, and thus R_cor, the decrease in f_H that was associated with hypoxia may have contributed to the increase in q̇_cor. However, because the reduction in f_H was only 10% (5 beats min⁻¹), it is doubtful that the decrease in f_H was the main factor mediating the 37% increase in q̇_cor. Although vasodilation of the coronary vasculature due to local changes in metabolism is the most plausible explanation for the decrease in R_cor associated with hypoxia, other mechanisms cannot be ruled out. These include neural vasodilatory reserve, alterations in a-adrenergic tone and the release of vasoactive substances from the myocardium.

Injection of adrenaline into the dorsal aorta during normoxia increased q̇_cor, P_DA, Q̇and R_cor. It is difficult, in an in vivo model, to determine to what extent a particular cardiovascular variable contributes to the observed elevations in q̇_cor. However, the existence of two response types with different patterns of R_cor, Q̇ and P_DA elevation (Figs 7, 8) provides indirect evidence that increases in R_cor (α-vasoconstriction of the coronary circulation), P_DA and myocardial oxygen demand (metabolic coronary vasodilation) can all mediate changes in q̇_cor in fish. This information complements the results of Axelsson and Farrell (1993), who used pharmacological agents to investigate the control of q̇_cor in coho salmon in vivo.

In type 2 fish, Q̇ (f_H) fell by 40%, R_cor increased by 55%, P_DA increased by 55% and q̇_cor increased by 5–10% shortly (2 min) after adrenaline injection. Because the resultant cardiac power output (P_DA × Q̇) was slightly lower than the value recorded before drug injection, it is probable that the increase in q̇_cor was mediated by the increase in P_DA (Axelsson and Farrell, 1993) or the lowered f_H (increased diastolic blood flow) (Farrell, 1987), and not by increased myocardial oxygen demand. Farrell (1987) using perfused in vitro hearts concluded that, in swimming trout, an increase in P_DA of 18% would increase q̇_cor by approximately 30%. In our study, because an increase in P_DA of 55% only increased q̇_cor by 5–10%, it must be concluded that the adrenergically mediated vasoconstriction of the coronary vascular bed (Farrell, 1987) severely limited, but did not preclude, increases in q̇_cor associated with elevations in P_DA. Axelsson and Farrell (1993) found that adrenaline injection (1.8 μg kg⁻¹) resulted in an increase in P_DA of 60% and an increase in q̇_cor of 60%. Although the reason for the difference between the two studies with regard to the relationship between q̇_cor and P_DA is not known, the results of Axelsson and Farrell (1993) support the conclusion that increases in R_cor, associated with constriction of the coronary circulation, can reduce but not obviate the effect of P_DA on q̇_cor.

In type 1 fish, where myocardial power output (P_DA × Q̇) constantly increased during the first minutes post-injection and reached a maximum value at approximately 6 min post-injection, elevations in q̇_cor mirrored alterations in Q̇ and increases in R_cor were short-lived (<6 min). However, in type 2 fish, where myocardial power output was decreased initially (through a severe drop in Q̇) and failed to reach a maximum value until 10–15 min post-injection, initial increases in q̇_cor were minimal and the increase in R_cor was prolonged (8–10 min) (Figs 7, 8). These results suggest that increases in myocardial oxygen demand, correlated with increases in cardiac power output (Graham and Farrell, 1990), mediated changes in q̇_cor and R_cor following adrenaline injection. Farrell (1987) suggested that sympathetic stimulation of the heart could increase q̇ _corvia metabolically related vasodilation, and Axelsson and Farrell (1993) suggested that increases in q̇_cor associated with activity may be achieved via a metabolically mediated vasodilation. When the results of Farrell (1987) and Axelsson and Farrell (1993) are considered, it appears that a metabolically related dilation of the coronary vasculature (i.e. decrease in R_cor) is the most probable explanation for the apparent relationship between myocardial power output and q̇_cor during normoxia.

In our trout, the coronary vasodilatory reserve [measured as (maximum post-injection hypoxic q̇_cor minus resting normoxic q̇_cor)/resting normoxic q̇_cor] was 88%. Because Axelsson and Farrell (1993) calculated that the coronary vasodilatory reserve in coho salmon is at least 200%, the results of the present study suggest that adrenergically mediated vasoconstriction of the coronary circulation (Farrell, 1987) may have limited the ability of the heart maximally to dilate the coronary vasculature (i.e. to increase myocardial oxygen delivery).

In summary, the in vivo estimate of resting coronary blood flow in this study (0.85% Q̇) compares well with that obtained by Axelsson and Farrell (1993) (1.1% Q̇) and indicates that coronary blood flow in salmonids is approximately 1% of resting Q̇. Increased q̇_cor, probably mediated by metabolically related coronary vasodilation, is associated with hypoxia-induced alterations in resting Q̇. Maximal dilation of the coronary vasculature (i.e. increase myocardial oxygen delivery) may be limited under conditions where an adrenergically mediated vasoconstriction is also present. Exposure of fish to moderate environmental hypoxia reduces the scope for adrenergically mediated increases in Q̇by approximately 50%. A recovery period greater than 2.5 h is required to restore the cardiovascular performance of trout to levels seen during normoxia.

The authors wish to thank Jessica Meijer for assistance with data analysis, Wade Blanchard for assistance with statistical analysis, Kathy Clow for assistance with preparation of the manuscript, and the Canadian Red Cross Society for supplying human blood. This work was made possible by Natural Sciences and Engineering Research Council of Canada operating grants to R.G.B. and A.W.P. A.K.G. was supported by an NSERC postgraduate scholarship and a Dalhousie University Fellowship. R.R.G. was supported by an NSERC Summer Research Scholarship. We thank Dr A. P. Farrell for reading the manuscript and providing valuable input.