Oxygen Consumption of the Isolated Heart of Octopus: Effects of Power Output and Hypoxia

The systemic heart of Octopus vulgaris consists of a ventricle and two auricles and is capable of generating pressures above 100cmH₂O (1 cmH₂O = 98 ·1 Pa) in the dorsal aorta (Wells, 1979). The cephalopod heart appears to be a highly aerobic tissue. Ghiretti-Magaldi, Guiditta & Ghiretti (1958) found that the metabolic rate of octopod cardiac muscle was higher than that of any other octopod tissue measured. The systemic hearts of squids have been shown to contain mitochondria capable of high rates of oxidation (Ballantyne, Hochachka & Mommsen, 1981; Mommsen & Hochachka, 1981) and slices of the hearts have respiratory rates exceeding those of rat heart (Barron, Sights & Wilder, 1953). Unlike fish hearts, the systemic heart of Octopus receives oxygenated blood directly from the gills and therefore would seem to be well sited in terms of oxygen supply.

However, there have been some reports which indicate that the heart muscle receives a separate coronary supply. Smith & Boyle (1983) describe a coronary arterial supply to the heart of the octopod Eledone and Smith (1985) mentions that the thick octopod ventricle has well-developed ventricular arteries. Wells & Wells (1986) state that the aorta, after leaving the heart, gives off a pair of small vessels that provide the coronary circulation for the systemic and branchial hearts. In contrast, Foti, Trara Genoino & Agnisola (1985) report that with each contraction of the heart a proportion of the total cardiac output passes through the walls of the heart, and is collected in coronary veins which lead back to the vena cava and the gills. They also provide evidence for a linkage between the coronary efflux and the cardiac output in the heart in vitro. In the absence of any previous measurements of the oxygen consumption of the working ventricle of Octopus the present study aims to measure the oxygen consumption of the isolated heart, differentiating between the oxygen consumed from the aortic output and from the coronary output.

The regulation of cardiac output by the isolated heart of Octopus has been shown to be primarily effected by changes in stroke volume, with perfusion pressures having little effect on heart rate (Smith, 1981 ; Foti et al. 1985). Smith (1981), using input pressures between 10 and 40cmH₂O with the in vitro heart of Eledone cirrhosa, found large increases in aortic output, while Foti et al. (1985) found maximum cardiac output values with an input pressure of 20 cmH₂O in the in vitro heart of Octopus. In the present study variations in input pressure have been used to drive the work output of the heart by increasing aortic output. The aim was to increase the aortic output by at least two-fold, simulating the 2·5-fold increase in cardiac output which occurs during exercise (Houlihan et al. 1986). The hearts have also been made to work against variable output pressures within the range 25–50cmH₂O (in vivo aortic pressures range from 15 to 40cmH₂O at rest; Wells, 1983), while maintaining constant aortic output.

Octopus can continue to extract oxygen from sea water at a constant rate as the ambient falls to values of around 70 Torr (Maginnis & Wells, 1969). Measurements in vivo showed that the blood falls in parallel with the ambient but that changes in blood pH ensure that the haemocyanin remains saturated (Houlihan, Innes, Wells & Wells, 1982; Smith, Duthie, Wells & Houlihan, 1985). In acute hypoxia, heart rate falls and the rate of blood flow may be reduced (Wells, 1983). In view of the oxygen demand of the heart, we decided to investigate the heart’s response to hypoxic perfusates and the regulatory mechanisms permitting continue cardiac function in hypoxia.

The experiments were carried out at the Stazione Zoologica, Naples, Italy. The animals, of both sexes, were maintained in a circulating seawater aquarium (20°C) and used within 7 days of capture. Seventy-one animals were used for the study. For the investigation of the effects of body mass on heart mass, cardiac output and myocardial oxygen consumption, a size range of 0·19–1·50 kg (N = 39) was used. In the investigation of the effects of volume loading, pressure loading and hypoxia, a size range of 0·450–0·980kg was used (0·661 ± 0·042kg, mean ± s.E., N = 32).

In Octopus vulgaris, blood enters the heart through two auricles and leaves the ventricle through three arteries: the major aorta and the abdominal and gonadial arteries. Over the surface of the heart there is a network of coronary veins which run to the vena cava. The arrangement of the coronary flow was determined by cannulating in situ the efferent branchial vessel leading to an auricle and perfusing the heart with Mercox (Japan Vilene Company Ltd) or with methylene blue in sea water.

The isolated systemic heart was prepared as described by Foti et al. (1985). Both the auricles and the dorsal aorta were cannulated and the abdominal and gonadial aortae ligatured at their bases. The heart was then installed in the perfusion apparatus which was similar to that described by Foti et al. (1985), but was modified as a respirometer for the aortic and coronary circuits (Fig. 1). Approximately 10 min elapsed between killing the animal and placing the heart in the chamber.

The two auricles received perfusion fluid (sea water + 0·05 % glucose maintained at 20°C, pH 8·0) at the same controlled input pressures and its was measured throughout the experiment. Contraction of the heart resulted in a proportion (the aortic output) of the input perfusion fluid passing out of the cannulated aorta and the sealed chamber: we measured the flow and of this output. With each contraction of the heart, fluid also passed through the walls of the heart into the cut coronary veins and out into the chamber (the coronary output). The chamber was initially filled with aerated saline and the output flow and from the coronary output was measured throughout the experiment. The chamber had a volume of 56 ml; the fluid in it was stirred continuously (Fig. 1) and the turnover time (volume of chamber/ coronary output) was between 8 and 10 min. The temperature in the chamber was 20°C. Because of the chamber arrangement, the heart was bathed in the coronary output and there was the possibility of oxygen consumption occurring over the surface of the heart. Experiments were therefore carried out on hearts with no input pressure, and the chamber was initially filled with aerated saline and sealed. Repeated sampling and measurement of the of the chamber fluid gave an estimate of the surface oxygen consumption of the heart and the microbial oxygen consumption in the saline. Values obtained were less than 3 % of the total oxygen consumption of the heart and were not included in the calculations. Because of the coronary flow it is not possible to measure surface oxygen consumption in working hearts. We have assumed that the gradient between the chamber fluid and the heart is maximal, even in the resting heart, and that surface oxygen consumption would therefore not increase.

The heights of the input and output tubes were measured with a ruler to within 1 mm with reference to the levels of the input cannulae and the aortic output, respectively. The coronary output tube was always fixed to the same level as the input tubes to the heart, i.e. the difference between input and output on the coronary flow was equal to the input pressure to the heart. Input and output pressures and coronary output pressures were also measured through saline-filled sidearms.

Following the cannulation procedures and placing the heart in the chamber, it was supplied with aerated saline at an input pressure of 20cmH₂O and the aortic output pressure was set at 40cmH₂O (Δpressure = 40–20cmH₂O). These values were chosen as the input value gave the maximum cardiac output in the in vitro heart (Foti et al. 1985) and the output was similar to in vivo aortic values (see Introduction). In these conditions, aortic output was 16·5 ± 1·0mlg⁻¹ min⁻¹. Over the course of 10 min, samples of cardiac output and coronary outputs were collected for measurements of flow rates and ⁠. If these variables had not stabilized after 10 min, the heart was discarded. Experiments were run for no more than 40 min after setting up the heart; returning input and output pressures to control values up to this time resulted in the heart being able to return to initial values.

These experiments (N – 12) examined the effects of changing power output of the heart while changing O₂ delivery to the heart. Varying input pressure, while keeping the Δpressure at a constant value of 20cmH₂O, resulted in variable aortic output; the of the input perfusate remained at around 150 Torr. The input pressure was reduced in steps of 5 cmH₂O to decrease aortic output from the maximum value, and perfusate ⁠, aortic and coronary output flows and were measured over 10 min. By this time, repeated measurements gave the same values and sufficient time had elapsed for the coronary efflux to have stabilized relative to the chamber volume. It has previously been demonstrated that the range 5–20cmH₂O input pressure gave the greatest range of cardiac outputs (Foti et al. 1985).

These experiments (N = 5) examined the effects of changing the power output of the heart, while maintaining O₂ delivery (maximum aortic output and ⁠) constant. The input and output pressures were arranged to give a Δpressure of 30cmH₂O (20cmH₂O input, 50cmH₂O output pressure). After a 10-min stabilization period and the measurement of variables, the Δpressure was changed to 25cmH₂O while keeping aortic output constant. Measurement of variables was followed by continued step changes of 5cmH₂O in Δpressure at constant aortic output. Foti et al. (1985) describe the relevant input and output pressure necessary to achieve constant aortic output.

These experiments (N= 11) initially examined the effects of a reduction in the perfusate with constant input and output pressures of 20 and 40cmH₂O, respectively. In addition to the control (air-saturated) perfusate, four other levels were examined: 120, 100, 80 and 50Torr. The of the perfusate was constant and as controlled by the mixing of N₂ and air streams. After 10 min with air-saturated saline, the was reduced to the first level of hypoxia, and after 8 min of continuously recording cardiac variables the was reduced again. It was found that after 8 min at any particular Δ the flow rates and values had stabilized. In a further series of experiments with five animals, after reducing the to 70 Torr for 15 min after normoxia, the Δpressure was reduced and cardiac variables were monitored.

Throughout the experiments, the aortic output and coronary output were collected over 1 min, weighed, values were corrected for temperature and salinity and expressed as a volume measurement. Mean values for input and output pressures were obtained from the directly measured heights of the input and output reservoirs and from pressure measurements made with a Statham P23 pressure transducer coupled to a chart recorder (Gould, Cleveland). Pressures were referenced to the level of saline in the bath. Pressure drops across the input and output cannulae were measured at variable flows and corrections applied where necessary. Separate samples were taken anaerobically by syringe of the input and output flows for the measurement of ⁠. Heart rate was measured by counting contractions over 1 min.

Samples of input perfusate, aortic output and coronary output were withdrawn anaerobically in the intervals between the measurements of fluid flow, and was measured with Radiometer oxygen sensors mounted in thermostatted cuvettes and connected to oxygen meters (Strathkelvin Instruments). The sensors were calibrated with a sodium sulphite/borax solution for zero and using air of calculated -Normally, the heart received aerated perfusion medium, but in a series of experiments on the effects of hypoxia, N₂ and air were bubbled into the input saline at controlled rates to give the required perfusate ⁠.

All pressures are expressed in cmH₂0. Aortic output (ml g⁻¹ min⁻¹) = heart rate X stroke volume. Power output of the heart from the aortic output was calculated as (mWg⁻¹) = (mean output pressure – mean input pressure) (980/60) × 10⁻⁴ (Farrell, Wood, Hart & Driedzic, 1985). The power output from the coronary flow (mWg⁻¹) was calculated as the product of the aortic pressure and the coronary flow (V_c); coronary flow power output = aortic pressure × × (980/60) ·10⁻⁴. Total power output of the heart was calculated from the sum of the aortic and coronary power outputs. Total oxygen uptake of the heart ⁠, where a is the solubility coefficient of oxygen determined from its solubility in aerated sea water plus glucose by the method of Strickland & Parsons (1972). Mechanical efficiency of the heart (%) = (100) [power (mW) × 0·0498] [( μl O₂ min⁻¹)]. The ventricle was weighed at the end of each experiment, and all the results are expressed per gram ventricle mass.

The results were compared using Student’s t-test and, in the case of the scaling investigation, analysis of covariance was used. The 5 % level of confidence has been employed throughout.

As described by Foti et al. (1985), contraction of the isolated ventricle was accompanied by the loss of some fluid from the cut coronary veins. Observations of isolated contracting hearts clearly revealed the streams of coronary output from the cut coronary vessels on the anterior lateral edges of the heart. Injection of methylene blue into a cannulated efferent branchial vessel in situ revealed that contraction of the undamaged heart is accompanied by the passage of dye through the walls of the ventricle, and its collection in coronary veins which empty into the lateral vena cava. Inspection of the in situ Mercox-injected (Fig. 2) heart revealed a network of fine capillaries over the surface of the ventricle which were connected to the coronary veins and led to the vena cava and hence to the gills. In the isolated heart, there was no leakage of fluid out of the heart except through the cut coronary veins; contraction of the ventricle resulted in flow of perfusion fluid through the cannulated aorta and a coronary output. The latter only occurred during ventricular contraction.

The in vitro heart preparation generated its own rhythm. Generally the hearts did not perform well after 50 min in the apparatus; after this there was a notable decline in cardiac output, particularly at the higher input pressures. After several trials, a protocol was developed for the volume-loading experiments, as shown in Fig. 3. In initial experiments, the hearts were started with an input pressure of 5cmH₂O, but under these conditions steady-state conditions were not reached and the heart performance steadily declined.

A decrease in input pressure with constant Δpressure produced a decrease in the maximum aortic output and coronary output (Fig. 3). The coronary output amounted to 41·5 ± 3·6% of the aortic output (30% of the total of cardiac output) over the range of pressures used. Aortic output was linearly related to coronary output where coronary output = 0·37 × aortic output + 0·98 (r=0·83, df = 24, P< 0·001) (Fig. 4). The of the aortic output was only 10Torr less than the perfusion (145·8 ±0·6 and 156 ± 0·1 Torr, respectively), and this and the percentage utilization (mean 6·7 ±0·5%) did not change significantly with aortic output (Fig. 3).

The tissue supplied via the network of coronary vessels was clearly the major site of oxygen consumption, with the coronary output accounting for 80 ± 0·5 % of the total oxygen consumption of the heart over the range of input pressures used. Both aortic output and coronary output were linearly related to the total oxygen consumption of the heart. Fig. 5 shows aortic output and the total oxygen consumption of the heart, where total oxygen consumption = 1·49 × aortic output + 1·25 (r=0·81, df = 22, P<0·001). The total oxygen consumption showed a 2·8-fold increase with a 2·2-fold increase in aortic output.

The total power output of the heart was almost equally divided between the aortic flow and overcoming the resistance of the coronary vessels. Total oxygen consumption was linearly related to total power output (Fig. 6A), where total oxygen consumption = 28·46 × power – 0·87 (r=0·89, df = 25, P<0·01). Mechanical efficiency of the heart was 12·4 ± 0·7 % at the maximum and 12·4 ± 1·9% at the minimum aortic outputs, an insignificant difference.

There was no significant change in heart rate with increase in input pressure; mean values were 31·0 ± 1·13 and 30·0 ± 0·7 beats min⁻¹ at 20 and 5 cml₂O, respectively; the increases in aortic output were due to increases in stroke volume. In three hearts, the coronary output pressure was increased by 5 cmH₂0 while the hearts were receiving 20cmH₂O input pressure and 40cmH₂O output pressure. There was no significant change in the coronary output flow rate.

Thus, when total power output was increased by increasing aortic output and coronary output, there was an increase in total oxygen consumption due to as increase in coronary volume flow. There was a negligible change in the oxygen removal from the perfusate and a direct linear linkage between aortic output and coronary output.

Changing Δpressure from 30 to 5 cmH₂O without significantly changing aortic output resulted in a fall in coronary output (Fig. 7). The utilization of oxygen in the aortic and coronary outputs had mean values of 6·5 ±0·4% and 56 ±2·3%, respectively, not significantly different from values obtained in the volume-loading experiments. The total oxygen consumption of the heart declined as Δpressure fell, but in this case total oxygen consumption correlated with coronary output and not aortic output (Fig. 7). Total oxygen consumption varied by 2·3-fold with a 3·5-fold increase in coronary output. The coronary output accounted for 80% of the total oxygen consumption, as with the volume-loaded hearts.

Total oxygen consumption was linearly related to total power output (Fig. 6B), where total oxygen consumption = 16·34 × power + 9·29 (r=0·95, df = 10, P <0·01). This regression equation is significantly different in slope from that obtained with the volume-loaded hearts (P<0·01) but not significantly different in elevation. Mechanical efficiency increased with pressure loading; across the range of power outputs efficiency improved from 2·7 % to 13·7 %.

Heart rate did not change significantly with pressure loading. Thus, when power output is increased with a constant O₂ delivery (aortic output and constant) an increase in is achieved through improved coronary perfusion with little change in O₂ removal from the perfusate.

In the series of experiments on hypoxia the mean values for the cardiac variables of the normoxic hearts were not significantly different from those obtained in the previous experiments at input and output pressures of 20 and 40cmH₂O, respectively. There was a significant decline in aortic output with hypoxia with a cessation of aortic output (but not heart rate) at 50 Torr (Fig. 8). Coronary output rose during hypoxia whereas aortic output fell. Plotting the aortic output and coronary output data for hypoxia below a perfusate of 100 Torr gives a linear relationship where coronary output = –0·10 × aortic output + 11·99 (r=0·51, df=11, P<0·05) (Fig. 4). This line is significantly different from that obtained in the volume-loading experiments where coronary output fell as aortic output fell (Fig. 4).

Utilization of oxygen in the aortic output was the same as in the volume-loaded experiments and did not change significantly during hypoxia. The oxygen utilization in the coronary output decreased significantly (P< 0 ·01) at the lowest perfusate ⁠.

Total oxygen consumption declined linearly as aortic output fell (Fig. 5), where total oxygen consumption = 0·77 × aortic output + 8·59 (r = 0·81, df = 15, P <0·01). This regression equation is not significantly different from that obtained from the volume-loading experiments, but in this case the fall in aortic output is accompanied by an increase in coronary output. At 50 Torr, all the oxygen consumption was due to the flow through the coronary output. Combining all the data on total cardiac oxygen consumption and aortic output (i.e. from the volumeloading and hypoxia experiments, Fig. 5) gave the relationship: total oxygen consumption = 1·08 × aortic output + 6·35 (r = 0· 79, df =39, P < 0·001).

Total power output and total oxygen consumption were linearly related, where total oxygen consumption = 24·5 × power output – 8·42 (r=0 ·86, df = 13, P< 0·01). The elevation of this line is significantly lower than that for the volume-loaded normoxic hearts (P< 0·001) (Fig. 6A) and for any given power output in hypoxia, oxygen consumption was approximately 50% less. Total power output in hypoxia is mainly due to the coronary output and we may be overestimating the pressure fall across the coronary vessels when the aortic pressure is used in these calculations. Because of these low values, mechanical efficiency appears to be high in hypoxia (over 30 %).

Heart rate was 32·36 ± 1·77 in normoxia and 29·80 ± 4·05 beats min⁻¹ at 50 Torr, an insignificant decrease. Thus, the hypoxia-induced reduction in aortic output was brought about by a decrease in stroke volume until, at 50 Torr, there was no aortic output but only a coronary output.

Fig. 9 shows the mean results from five hearts in which after a progressive reduction in perfusate to 70 Torr there was an almost complete cessation of the aortic output. The output pressure was then reduced from 40 to 30 cmH₂0 (constant 20cmH₂O input pressure). Aortic output immediately increased with an insignificant increase in total oxygen consumption. Calculation of the total power output showed an insignificant change. However, with the reduction in output pressure the power output to the aortic flow increased and that to the coronary flow decreased. Thus the work done by the heart had been transferred from pressure work to volume work.

Allometric equations describing the relationships between systemic heart mass, in vitro aortic output, in vitro coronary output and in vitro total myocardial oxygen consumption are given in Table 1. In each case the in vitro exponents are significantly less than one.

Total oxygen consumption of resting Octopus cyanea has been found to be proportional to mass^{0· 833} (Maginnis & Wells, 1969). Testing the exponent for the total myocardial oxygen consumption against 0·833 revealed no significant difference.

The in vitro heart in the present experiments performed in a similar way to previous reports for Octopus vulgaris and Eledone cirrhosa hearts in terms of the responsiveness of the aortic output to input pressures (Smith, 1981 ; Fotiet al. 1985). Heart rate has been found to be affected by input pressures when these ranged up to 40cmH₂O (Smith, 1981; Foti et al. 1985); this was not observed in the present experiments, where a smaller range of input pressures was used. The calculated range of power output is above that for the in vitro heart of Eledone cirrhosa working at 10°C (Smith, 1985).

As in previous experiments with isolated octopod hearts, the heart continued to beat regularly for several hours after isolation (Smith, 1981; Foti et al. 1985). However, input pressures of at least 5 cmH₂O are needed to ensure regular contractions [as also found by Smith (1981) for Eledone], and in the present experiments the ventricle displayed maximum contractility at an input pressure of 20cmH₂O. Because the heart showed a marked decline in performance after 40min, it was necessary to start all experiments with the highest levels of performance. Clearly there are a number of problems with interpreting the performance of the in vitro heart in terms of the likely conditions in vivo.

First, the cardiac output in vitro is below that in vivo. In vivo, the cardiac output calculated from pre- and postbranchial blood gas measurements and allowing for cutaneous oxygen consumption is approximately 44 ml kg⁻¹ min⁻¹ for resting animals (Houlihan et al. 1986). As a 1-kg Octopus has a heart mass of approximately 1 g, this gives a cardiac output of 44 ml g⁻¹ min⁻¹ and a stroke volume of 0·6mlbeat⁻¹ (Houlihan et al. 1986; Wells, 1979). In vitro, the minimum heart output, combining aortic output and coronary output, was 11·0mlg⁻¹ min⁻¹ for a 765-g animal (Fig. 3) with a stroke volume of 0·4 ml beat⁻¹. The similarities in stroke volume between the in vivo and m vitro hearts are deceptive as the in vitro heart is only beating at 69% of the in vivo rate (in vivo values from Wells, 1979). The total maximum cardiac output values for the in vitro heart (24 ml g⁻¹ min⁻¹, stroke volume 0·8 ml beat⁻¹) are approximately 55% of the resting in vivo values. However, it has been calculated that the cardiac output in vivo shows a 2·5-fold increase when the animals become active, mainly due to an increase in stroke volume (Wells, 1983; Houlihan et al. 1986). Thus, the maximum total cardiac output values of the in vitro heart as reported here and by Foti et al. (1985) are approximately 22 % of the likely maximum values in vivo. The in vitro heart could also show a 2·2-fold increase in total cardiac output but only from the lowered baseline values.

Second, the input pressures used here (5–20cmH₂O) and in other in vitro experiments (Smith, 1981; 5–40cmH₂O) are much higher than the likely pressures in the efferent branchial vessels returning blood to the heart (0·5–3·0cmH₂O, Smith, 1985). One exception are the in vivo input pressures recorded for Octopus dofleini (10– 20cmH₂O; Johansen & Martin, 1962). Output pressures (25– 40cmH₂O) compare with aortic pressures of 35cmH₂O (systole) and 15cmH₂O (diastole) reported to be typical of Octopus vulgaris at rest (Wells, 1983). Following exercise, the pressures may rise to between two- and three-fold above those recorded at rest (Wells, 1983).

Third, we have no data on the pressures in the coronary veins in vivo or the likely extent of the coronary flow. Wells & Wells (1986) have reported that 70% of the total systemic flow passes down the aorta to the head and arms leaving the rest for the gonad, mantle, gut and coronary circulation. Therefore, it seems likely that the coronary flows reported here are higher than those in vivo, possibly due to the necessity for cutting the coronary veins, the absence of any venous back pressure and the lower oxygen content of the perfusion fluid compared with blood; the perfusate only delivers 0·22 vol% with a 50-Torr fall in whereas the blood under similar conditions would deliver 2·8vol%. Nevertheless, the experiments using increased coronary output pressures indicated that coronary flow was rather insensitive to back pressure in vitro.

Fourth, the relatively poor performance of the in vitro heart in terms of total cardiac output is also evident in its susceptibility to changes in -In vivo the heart receives arterial blood with a of 78 Torr (Houlihan et al. 1982) but in vitro aortic output was reduced when the perfusate fell below 120Torr (Fig. 8). Also, in vivo the of the arterial blood may fall below 30 Torr during environmental hypoxia (Houlihan et al. 1982; Smith et al. 1985), but in vitro at a perfusate of 10Torr the heart beat generated no aortic output and only a coronary output. It is likely that the reduced oxygen content of the perfusate compared with that of blood is responsible for the poor performance in hypoxia. It is possible that the heart has some anaerobic capacity (Pritchard, Huston & Martin, 1963 ; Driedzic, 1985), but we have no information on anaerobic metabolism in vitro. Also, in vivo the heart will be subject to a range of neurotransmitters and circulating cardioactive substances (Wells, 1983) which will presumably be absent in vitro.

This is the first report of the myocardial oxygen consumption of an intact cephalopod heart. Based on the present observations, maximum myocardial oxygen consumption is 15μO₂min⁻¹ for a 600-g Octopus with a 0·6-g ventricle. This represents approximately 3 % of the total oxygen consumption of the resting animal for a tissue that occupies 0·1 % of the total body mass. Clearly, as cardiac output and cardiac oxygen consumption scale in proportion to total oxygen consumption (Table 1) comparisons must be made for defined sizes of animal. Allowing for these scaling relationships, the rates of oxygen consumption, power output and mechanical efficiency of the in vitro Octopus heart are comparable with values for the isolated teleost heart (Driedzic, Scott & Farrell, 1983), but are slightly lower than the values from an in situ teleost heart preparation (Farrell et al. 1985). In both the teleost and Octopus heart, oxygen consumption is linearly correlated with power output, and in both groups pressure work and volume work have the same metabolic costs.

With a cardiac output of 44 ml kg⁻¹ min⁻¹ and a blood oxygen content of 3 vol% (Houlihan et al. 1982) the oxygen consumption of the heart would make a 1 % difference to the oxygen content of the arterial blood as it passed through the heart. As this is not likely to be significant, it seems probable that the value of the coronary circulation is in overcoming the problem of oxygen diffusion through the thick ventricular muscle.

The coronary circulation of the Octopus heart accomplishes the important task of perfusing the organ which generates the driving pressure for the systemic circulation. The Octopus arrangement of the coronary perfusion provides oxygenated blood from the gills which flows directly from the ventricle through the walls of the heart. This is in contrast to the arrangement in fish, where the inner spongy myocardium derives oxygen from the venous blood pumped through the ventricular chamber, and the outer compact layer of the ventricle, where it is present (Santer & Greer Walker, 1980), receives a coronary supply from arterial blood. Blood flow through the coronary vessels of Octopus must be determined by the driving pressure and the resistance of the coronary vascular bed. The driving pressure across the bed is considered to be the difference between the intraventricular pressure and the downstream pressure of the coronary veins. In vitro it appears that the coronary flow is rather insensitive to the output back pressures. The coronary resistance will depend upon the number, size and calibre of the coronary vessels.

Interestingly, the coronary perfusion only seemed to occur during systole; For et al. (1985) have estimated that the diastolic efflux was only 3 % of the systolic value.In the mammalian heart there is a systolic obstruction of ventricular coronary blood flow and only 16–25 % of the stroke flow enters the coronary vasculature in systole. As diastole becomes abbreviated during exercise, the proportion of the flow occurring during systole increases to 32–50% (Barnard, Duncan, Livesay & Buckberg, 1977; Sanders, White, Peterson & Bloor, 1978). In mammals and fish, the arterial blood pressure is the main driving force for coronary perfusion and coronary resistance is the main regulator of cardiac blood flow. However, in Octopus it appears that the critical opening pressure for the coronary vessels is only reached during systole.

The in vitro heart of Octopus demonstrated remarkable abilities in the control of the coronary circulation. The coronary output was found to vary in three different ways with respect to aortic output in the present experiments, and any explanation of how the aortic output is varying must also take into account the coronary output. The aortic output and coronary output may be modelled as two sets of resistances in parallel. First, volume-loaded hearts are filled to varying amounts by the end diastolic pressure under the influence of input pressure. The increased input pressure increases systolic pressure as aortic output is increased. Thus coronary output can increase with both output resistances remaining constant, while there is constant oxygen utilization and increased oxygen delivery to the cardiac muscle. Thus, the degree of contraction of the cardiac muscle under the influence of volume loading appears to supply the cardiac muscle automatically with the appropriate level of coronary perfusion.

During pressure loading at constant aortic output, increased systolic pressure produces increased coronary flow (Fig. 7), with the resistance of the coronary circuit presumably remaining constant. Thus, the increased power output of the heart due to pressure loading results in increased coronary flow, and with constant oxygen utilization, increased oxygen supply to the heart. Overall, the power output/oxygen consumption relationship is very similar in the volume-loaded and pressure-loaded hearts.

During hypoxia with constant input and output pressures, aortic output, power output and, presumably, systolic pressure declined, but we must postulate a fall in coronary output resistance to account for the increased coronary perfusion. Coronary output remained almost constant as aortic output fell below a perfusion of 100 Torr, implying that the coronary vessels were almost maximally open at these values. This ensured the best possible oxygen supply to the hypoxic cardiac muscle.

How coronary output is regulated in vitro is not clear. In mammals, hypoxia is the most potent stimulator of a reduction in coronary resistance and it is likely that the metabolite adenosine is responsible for this (Berne & Rubio, 1969). As we have demonstrated that the coronary output is increased during hypoxia, presumably through dilation of the coronary vessels, it is possible that adenosine formed from the Y-nucleotidase found in Octopus heart (Sciurba, Agnisola, Foti & Trara Genoino, 1985) is also involved in the cardiac muscle’s response to hypoxia.

DFH received support from the Royal Society, The Carnegie Trust and Aberdeen University. We are grateful for the helpful comments from Dr A. P. Farrell on an early version of the manuscript.