The effects of sustained exercise and hypoxia upon oxygen tensions in the red muscle of rainbow trout

The anatomy of the skeletal musculature in teleost fish differs significantly from that of the tetrapod vertebrates. Teleosts possess distinct blocks of highly vascularised oxidative slow-twitch fibres (`red' muscle, RM),arranged alongside blocks of less vascularised glycolytic fast-twitch fibres(`white' muscle, WM), whereas tetrapod muscles all comprise a mixture of oxidative slow-twitch and glycolytic fast-twitch fibres(Bone, 1978; Young, 1981). The different anatomical arrangement of the oxidative RM of teleost fish has led to the suggestion that the partial pressures of oxygen(P_O2) in their red muscle fibres(Prm_O2) may be significantly higher than typically found in skeletal muscles of mammals(Egginton, 2002). Recent measurements of the P_O2 in various skeletal muscles of mammals never seem to exceed approximately 40 mmHg under resting conditions in normoxia (Hutter et al.,1999; Jung et al.,1999; Behnke et al.,2001; Suttner et al.,2002). Although arterial and venous blood Prm_O2 values are known for teleost fishes such as the rainbow trout Onchorhynchus mykiss(Holeton and Randall, 1967; Stevens and Randall, 1967; Kiceniuk and Jones, 1977; Thomas and Hughes, 1982; Thomas et al., 1987; Farrell and Clutterham, 2003),we are unaware of any Prm_O2measurements that have tested this prediction. In fact, measurements of the P_O2 of muscle appear to be limited to those of Jankowsky (1966), who reported a very low value of <5 mmHg in the glycolytic WM of eels (Anguillasp.).

Measurements of O₂ tensions in the skeletal musculature of teleost fish would be particularly informative for two other reasons. One of these is to investigate the extent to which convective O₂ supply might be a limiting factor in the performance of sustained aerobic exercise. During sustained exercise in tetrapods, increased muscle O₂ demand relative to rates of supply causes a reduction in muscle P_O2 (Jung et al., 1999; Behnke et al.,2001), and fatigue is associated with a severe decline in intramuscular O₂ tension(Molé et al., 1999; Howlett and Hogan, 2001). Fish support sustained swimming activity with their RM, while WM powers the faster,unsteady sprint and burst swimming activities(Bone, 1978). Therefore,measurements of Prm_O2 during swimming would provide insight into whether RM O₂ supply is a factor limiting the performance of sustained swimming, that is, whether exhaustion is associated with a profound decline in Prm_O2.

Another reason why Prm_O2 of teleosts might be particularly interesting relates to a unique characteristic of some teleost haemoglobins, the Root effect(Root, 1931). When blood pH drops, haemoglobins with a Root effect exhibit a markedly reduced capacity to bind O₂, and hence will release bound O₂(Root, 1931; Randall, 1998; Pelster and Randall, 1998). A well established physiological role for the Root effect is found in specialised vascular beds (retes), where high rates of lactic acid and CO₂ production by specialised cells generate low pH, resulting in localised P_O2 values that are considerably higher than in arterial blood leaving the gills, due to unloading of O₂ from haemoglobin. In particular, the choroid rete ensures that photoreceptors in the retina are well oxygenated, while the rete mirabilis provides O₂ to inflate the swimbladder and maintain buoyancy as fish descend in the water column (Jensen et al., 1998; Pelster and Randall, 1998).

Theoretically, the Root effect may also promote the release of O₂ from haemoglobin at other respiring tissues. This is because in vitro evidence shows that diffusion of respiratory CO₂into the teleost erythrocyte, and its carbonic anhydrase-catalysed hydration to HCO₃^– and H⁺, occurs more rapidly than diffusional release of O₂ from haemoglobin in response to a P_O2 gradient(Brauner and Randall, 1998; Pelster and Randall, 1998). Consequently, a transient drop in erythrocyte pH following the catalysed hydration of CO₂ could elicit a Root effect and generate high P_O2 values in well-vascularised aerobic tissues. The presence and extent of this effect in tissues other than retes,such as RM, has not been studied. If measurements of Prm_O2 revealed that it was higher than the P_O2 of arterial blood leaving the gills (Pa_O2), then this would be dramatic evidence that the Root effect influences O₂ tensions in aerobic tissues.

In the current study, novel O₂-sensitive optical fibre sensors(`micro-optodes') were used to measure the Prm_O2 of conscious free-swimming rainbow trout, a species with a pronounced Root effect(Binotti et al., 1971). Measurements were made under three regimes: during normoxia, to compare with data reported for mammalian skeletal muscles(Hutter et al., 1999; Jung et al., 1999; Suttner et al., 2002; Behnke et al., 2001) and to investigate whether the Root effect contributes to elevated Prm_O2; during graded sustained exercise, to gain insights into RM O₂ supply during an increase in demand, and also during mild hypoxia, to investigate whether reducing the arterial-to-tissue P_O2 gradient would expose an impact of the Root effect upon Prm_O2.

Rainbow trout Oncorhynchus mykiss Walbaum with a mean (± s.d.) mass of 697±152 g and fork length of 36±2 cm, were transported from Sun Valley Trout Farm (Mission, BC,Canada) to Simon Fraser University, where they were held outside in 1000 l circular fibreglass tanks provided with a flow of fresh water at seasonal temperatures of 13–15°C (mean temperature 13.8±0.4°C). Fish were acclimated to these conditions for at least 2 weeks, and fed daily. Individual trout were starved for 24 h prior to surgery.

Fish were anaesthetised in 0.1 mg l^–1 MS-222 buffered with 0.1 mg l^–1 NaHCO₃, and then transferred to an operating table where their gills were irrigated with aerated water containing diluted anaesthetic (0.05 mg l^–1 MS-222 and NaHCO₃). A small incision was made in the skin just dorsal to the lateral line to reveal the underlying RM sheet. A blunted surgical needle (15G Terumo, Leuven, Belgium) was then advanced under the skin for approximately 1 cm, with the foremost end of the blunted needle bevel against the underside of the skin. Great care was taken to avoid penetrating the underlying musculature. An oxygen-sensitive optical chemical fibre sensor (PreSens;Precision Sensing GmbH, Regensburg, Germany), with a tapered Teflon-coated tip(diameter <10 μm), was inserted into the bore of the needle and advanced until the tip reached the end of the needle. The needle was then angled at approximately 45° to the skin such that the bevelled end rested flat against the musculature and the tip of the optode advanced gently, at the prevailing angle of 45°, for approximately 3 mm into the underlying sheet of RM. The needle was then withdrawn along the optode lead, and the optode secured in position with sutures to the skin. Trout were then cannulated in the dorsal aorta (DA) using the technique described by Soivio et al.(1975).

While the trout were still under anaesthesia, the optode was connected to a Microx 1 oxygen meter (PreSens), connected in turn via a serial port to a PC with dedicated software, which displayed Prm_O2 at the optode tip every 1 s and saved a measure of Prm_O2 every 1 min in an ASCII file. Prior to surgery, each optode was calibrated in oxygen-free and air-saturated water, and the tip soaked for 10 min in 100 i.u. ml^–1 heparin (Farrell and Clutterham, 2003). The position of the probe in the RM was confirmed post-mortem by careful dissection under a binocular microscope. Data are reported only for those experiments where the probe could be recalibrated, post-mortem, to correct for any drift, according to the manufacturer's instructions. In one case where blood clotting and tissue damage were visible around the tip of the probe, the results were disregarded.

Fish were recovered for approximately 42 h in normoxic water while swimming gently at a speed equivalent to 0.5 body lengths s^–1(BL s^–1) in the Brett-type swimming respirometer described in Gallaugher et al.(1995), and Prm_O2 was measured every 1 min throughout. The DA cannula was flushed every 24 h with heparinised (10 i.u. ml^–1) teleost saline. Measurements of control normoxic Prm_O2 values were made while the animals were swimming gently so as to establish a constant level of muscular work and consequent O₂ demand and to reduce spontaneous changes in activity level, thus minimising variability in Prm_O2 (see Fig. 1).

While swimming gently at a speed of 0.5 BL s^–1,the trout were exposed to two levels of mild hypoxia, comprising 30 min at 100 mmHg, followed by 30 min at 75 mmHg, followed by 1 h recovery to normoxia (140 mmHg). The Prm_O2 was monitored every 1 min throughout the exposure protocol. Water P_O2 was monitored continually, and water entering the respirometer made hypoxic by passing it counter-current to a flow of compressed 100% N₂ in a gas-exchange column. The Pw_O2 recording was used to measure Ṁ_O2 in the sealed respirometer for 30 min in normoxia, for 30 min at both levels of hypoxia, and for 30 min centred upon 1 h recovery to normoxia. A measurement of Pa_O2 was made every 5 min by gently withdrawing blood along the DA catheter and into the O₂ electrode cuvette. Samples of arterial blood (300 μl, replaced immediately with an equal volume of saline) were collected from the DA cannula in normoxia, at 30 minexposure to each level of hypoxia, and following 1 h recovery to normoxia,to measure Ca_O2 and pHa.

One-way analysis of variance (ANOVA) for repeated measures was used to reveal effects of exercise or hypoxia on any single variable. A two-way repeated-measures ANOVA was used to assess the effects of progressive hypoxia on Pa_O2versus Prm_O2. Where changes in P_O2 were expressed as a percentage of the normoxic value, data were arc-sine transformed prior to analysis by ANOVA. In all cases, Bonferroni post-hoc tests were used to identify where significant differences lay. A probability of less than 5%(P<0.05) was taken as the fiducial level for statistical significance.

In all fish, Prm_O2 was close to zero under anaesthesia, but gradually rose over a few hours during recovery(Fig. 1) and, at full recovery,was approximately 60 mmHg (Table 1).Under steady state normoxia, mean Prm_O2 was significantly lower than both mean Pw_O2 and Pa_O2 (Table 1). There was variability in normoxic Prm_O2 among fish, ranging from a low of 39 mmHg to a high of 101 mmHg, but no fish exhibited a higher P_O2 in their muscle than in either their inspired water or arterial blood. Therefore, there was no dramatic evidence that the Root effect influenced Prm_O2. In contrast to the stable Prm_O2 observed while fish were swimming steadily during normoxia, sharp reductions in Prm_O2 occurred if the animal struggled in the respirometer, followed by a gradual return to the previous P_O2 (Fig. 1), but Prm_O2 never decreased below ∼40 mmHg.

As expected, exercise caused an exponential increase in O₂uptake, and the maximum rate of Ṁ_O2 was observed at the maximum speed which the fish were able to sustain for a complete 30 min interval (Fig. 2). Exercise also caused a decline in pHa, particularly at U_crit(Fig. 2), which is evidence of a switch to glycolytic metabolism prior to exhaustion. Nevertheless, trout in the current study did not exercise exceptionally well, reaching a U_crit of 1.38±0.16 BL s^–1(N=5), which is lower than the U_crit of approximately 2.0 BL s^–1 reported earlier for chronically instrumented rainbow trout at the same temperature (e.g. Shingles et al., 2001; Farrell and Clutterham, 2003). Arterial blood P_O2 also showed a significant reduction during sustained exercise and at exhaustion, but recovered rapidly,unlike both Ṁ_O2and pHa, which slowly returned towards control values during 2 h of recovery. Despite the arterial acidosis, Ca_O2 was unchanged throughout exercise and at exhaustion(Fig. 2).

During sustained exercise, Prm_O2showed a significant drop and although the mean value remained above 40 mmHg, Prm_O2 never exceeded Pa_O2. Moreover, Prm_O2 rose significantly at the moment of exhaustion to a level that was not statistically different from the control, and remained thus for the ensuing 2 h recovery period. The partial pressure gradient between arterial blood and the RM dropped as exercise intensity increased, to a low at fatigue, but then returned rapidly to control values during recovery (Fig. 2). Given that the partial pressure gradient dropped as Ṁ_O2 increased during exercise, resolution of the Fick equation revealed that O₂delivery, hence blood flow, to the red muscle would have to increase by a factor of 4.6±1.1 times (mean ± s.e.m., N=5) between swimming speeds of 0.5 BL s^–1 and 1.38 BLs^–1.

Mild hypoxia had no significant effects on Ṁ_O2 and Ca_O2 (Table 2), so it could be assumed that rates of tissue O₂demand and blood O₂ transport capacity did not change significantly. Furthermore, the absence of any changes in pHa indicates that there was no major increase in the release of lactic acid and CO₂from the tissues (Table 2). Thus, the most significant effect of mild hypoxia was a decrease in the P_O2 of arterial blood as it left the gills(Table 1). Correspondingly, the Prm_O2 showed close temporal sensitivity to changes in Pw_O2 and Pa_O2 during exposure to hypoxia and the return to normoxia (Fig. 3). Red muscle P_O2 was significantly reduced from normoxic values at both levels of hypoxia(Pw_O2=100 mmHg and 75 mmHg), but changes in Prm_O2 were significantly less than those in Pa_O2, and so the arterial to RM P_O2 gradient declined as hypoxia deepened(Fig. 3, Table 1). Proportional (%)changes in P_O2, relative to normoxic values,were much more pronounced than the net changes in the RM but, nonetheless,were significantly smaller in the RM than in the arterial blood(Fig. 3, Table 1). At no time during either hypoxia or recovery did any fish exhibit a higher Prm_O2 than their Pa_O2. In fact, the estimates of apparent O₂ unloading did not decrease as hypoxia deepened but, rather,increased slightly (Table 2).

Farrell and Clutterham(2003) used the same micro-optodes to measure mixed venous P_O2(Pv_O2) in the ductus Cuvier of rainbow trout at a similar temperature. They found that Pv_O2declined to 20 mmHg immediately after surgery, and recovered to a steady-state value of approximately 35 mmHg within 30 min. These measurements of Pv_O2 are very different from our measurements for RM, where Prm_O2 declined almost to zero during surgery and, although it then rose rapidly during the first few hours of recovery, it did not achieve a steady-state value for approximately 20 h. This result clearly shows that the RM muscle can become severely hypoxic during deep anaesthesia and the slow recovery of Prm_O2 may reflect reduced distribution of blood to the RM, as a consequence of post-surgical cardiac depression and decreased total peripheral resistance, and/or increased muscle O₂ demand, perhaps to metabolise the anaesthetic or repay an O₂ debt.

Farrell and Clutterham(2003) also found that Pv_O2 dropped precipitously to around 20 mmHg whenever the fish struggled, and attributed this to sudden increases in muscle O₂ extraction. The sharp reductions in Prm_O2 that were observed when fish struggled could be due either to increased O₂ demand and extraction, or to a decrease in local blood flow associated with struggling behaviours. The latter may be the main contributing factor, as struggling behaviours are associated with bradycardia and reduced cardiac output(Stevens et al., 1972; Farrell, 1982; Farrell and Jones, 1992), and would also result in hypoperfusion if increased intramuscular pressure compresses the supplying segmental arteries. Even so, the fact that mean Prm_O2 did not decrease below ∼40 mmHg is a novel finding indicating that RM remained well supplied with oxygen during spontaneous struggling behaviours in rainbow trout. These observations suggest that the previously observed precipitous decrease in Pv_O2(Farrell and Clutterham, 2003)during struggling is driven by tissues in addition to RM, the most likely candidate being the WM.

Another novel finding of the present study is that the Prm_O2 of approximately 60 mmHg measured in the free-swimming normoxic trout is appreciably higher than the P_O2 of 2–4 mmHg measured with microelectrodes in the WM of eels(Jankowsky, 1966). Unfortunately, this earlier study does not detail exactly how the probes were implanted and whether the eels were conscious during subsequent measurements(Jankowsky, 1966). Therefore,given our observation of a low Prm_O2during anaesthesia, further studies with WM are warranted to confirm this difference between RM and WM. In contrast, it is very clear that the normoxic Prm_O2 in rainbow trout is significantly higher than in the skeletal muscle of mammals, where P_O2 values measured with implanted microelectrodes range from 25 mmHg to 35 mmHg in conscious humans Homo sapiens (Jung et al.,1999; Suttner et al.,2002) and dogs Canis canis(Hutter et al., 1999). Similar values were obtained in anaesthetised rats Rattus norvegicus, using phosphorescence quenching techniques(Behnke et al., 2001). In view of this difference, we provide the first direct evidence to support the earlier suggestions by Egginton(2002) that the anatomy and physiology of RM in teleost fish could lead to an elevated P_O2 compared with mammals.

If the P_O2 in respiring tissues is determined by the rate at which O₂ is supplied in the blood, the distance and speed it diffuses, and the rate at which the tissue consumes it(Egginton, 2002), then a comparison of these variables between teleost RM and skeletal muscles of mammals might provide insight into why Prm_O2 is so high. Mass-specific blood flow rates to trout RM may be up to twice the level reported for mammalian skeletal muscles at rest, although they are similar during exercise(Egginton, 1987, 2002; Taylor et al., 1996). Rainbow trout haemoglobin has a similar affinity for O₂ to that of, for example, humans and rats (Wilmer et al.,2000). Egginton(2002) calculated and compared the mean geometric supply area (domain of influence) as well as the mean Krogh's diffusion distance for capillaries in the tibialis anterior (TA) of both rats and Syrian hamsters Mesocritus auratus versus those in the RM of both rainbow trout and striped bass Morone saxatilis. In these two teleosts, domains and diffusion distances were approximately 20% smaller than in the hamster, whereas rat domains were approximately eightfold larger and diffusion distances approximately twofold larger than in the other three species (Egginton, 2002). Thus,the higher blood flow and smaller capillary domains clearly favour a higher P_O2 in the red muscle of the rainbow trout relative to the skeletal muscles of the rat(Behnke et al., 2001). The lower body temperature of the fish, however, will lead to a significant reduction in O₂ diffusivity, an effect that is only partially offset by a concurrent reduction in tissue O₂ consumption(Taylor et al., 1997; Egginton, 2002).

Our measurements of Prm_O2 in fish during graded exercise, at exhaustion and during recovery are also novel. Furthermore, it is evident that they contrast with results in exercising mammals. In both conscious humans and the anaesthetised rat, sustained exercise reduces intramuscular P_O2 from around 30 mmHg to below 20 mmHg, a change that is attributed to increased rates of O₂ extraction by the working muscle(Jung et al., 1999; Behnke et al., 2001). The significant decrease in Prm_O2 during sustained exercise in the current study presumably occurred for the same reason. However, Prm_O2 declined to only 45 mmHg at the maximum rates of exercise performance and Ṁ_O2, which is considerably higher than the P_O2 values observed in mammals (Jung et al.,1999; Behnke et al.,2001). Egginton et al.(2000), using morphological data and analysis of the resulting physico-chemical conditions for O₂ diffusion, estimated that the P_O2gradient between capillaries and the centre of a red muscle fibre may be less than 4 mmHg in trout at maximum sustained exercise. Thus, the high Prm_O2 suggests that rainbow trout RM may not become hypoxic at high levels of sustained exercise, a suggestion that is supported by two other lines of evidence. First, resolution of the Fick equation revealed that the reduction in the arterial to RM P_O2 gradient that occurred between a swimming speed of 0.5 BL s^–1 and maximal exercise would have required an approximately fivefold increase in blood supply to meet the measured increase in Ṁ_O2. This increase compares favourably with the eightfold increase in blood supply to RM measured with microspheres during maximum sustained exercise in rainbow trout(Taylor et al., 1996). Second,at exhaustion Prm_O2 increased rather than decreased. This contrasts with tetrapod skeletal muscles, where fatigue is associated with a profound decline in P_O2 to below 50% of resting values (Moléet al., 1999; Howlett and Hogan, 2001). This suggests that when WM is recruited to power swimming speeds above 70% of U_crit(Burgetz et al., 1998; Lee et al., 2003) subsequent exhaustion is not linked to major reductions in RM O₂ supply. Consequently, convective O₂ supply to the RM seems not to be a limiting factor for maximum aerobic performance in rainbow trout. Prolonged exercise at 90% of U_crit leads to depletion of oxidative substrates in trout RM (Richards et al.,2002), so this may be the cause of fatigue. Alternatively, RM may simply reduce its activity when WM is recruited during incremental exercise, a gait transition representing an orderly and necessary transition to a muscle that can generate the required increase in tailbeat frequency and muscular power output (see Jones and Randall,1978). One consequence of this gait transition is that Prm_O2 remains high, higher than Pv_O2 at fatigue(Farrell and Clutterham,2003). Further research into this area is clearly required, not least to determine the validity of an incremental graded exercise protocol in investigating factors limiting maximum rates of aerobic metabolism and performance in fish.

There was no evidence that the Root effect influenced tissue O₂tension enough to raise Prm_O2 above Pa_O2 in the rainbow trout. In fact, the opposite was always true, both in normoxia and in mild hypoxia, when the Pa_O2 to Prm_O2 gradient was reduced. Indeed, Prm_O2 was sensitive to changes in Pa_O2 and although the proportional changes in Prm_O2 during hypoxia were significantly less than the changes in Pa_O2,this could be attributed to the sigmoid shape of the trout Hb–O₂ dissociation curve. Thus, as Pa_O2 declined, the arterial to RM P_O2 difference shifted left towards the steep portion of the dissociation curve, such that a smaller drop in P_O2 was required to elicit the same degree of O₂ unloading.

In addition to Prm_O2 being elevated compared with measurements in mammalian muscles, it is interesting that Prm_O2 was also consistently higher than published values for mixed Pv_O2 in the trout, both in normoxia and at comparable degrees of hypoxia(Holeton and Randall, 1967; Farrell and Clutterham, 2003). Indeed, the measured values for Prm_O2 lie almost exactly midway between published values for Pa_O2 and Pv_O2 at the appropriate water P_O2(Holeton and Randall, 1967). In contrast, the reported range for mammalian intramuscular P_O2 (Hutter et al., 1999; Jung et al.,1999; Behnke et al.,2001; Suttner et al.,2002) is consistently lower than that of mixed Pv_O2, which is typically around 40 mmHg(Hutter et al., 1999; Wilmer et al., 2000). The higher Prm_O2 relative to Pv_O2 in the trout can be interpreted in one of two ways. One possibility is that venous return from RM is a relatively small contribution to mixed venous blood. The other possibility is that the high Prm_O2 of rainbow trout relative to mammals could be, at least in part, a consequence of a Root effect in blood perfusing the RM. The Root effect would be engendered by transient changes in erythrocyte pH caused by the faster rates of CO₂ diffusion than O₂ diffusion, and the strong coupling of O₂ and CO₂ movements that are known to exist in trout blood(Brauner and Randall, 1998; Brauner et al., 2000). That is,when arterial blood enters the RM of trout, rapid diffusion of metabolic CO₂ into the erythrocyte would cause a transient drop in pH and cause a Root `off-shift', driving O<sub>2</sub> off the haemoglobin and raising P<sub>O2</sub>. The deoxygenated haemoglobin would, however, then bind protons (the Haldane effect) and cause blood pH to rise again, eliciting a Root `on-shift' that binds O₂ back onto the haemoglobin and lowers P_O2 in the venous blood leaving the tissue.

While such a role of the Root effect is conjecture at this time, we can eliminate the possibility that we measured an artefact of mixed arterial,tissue and venous P_O2 values rather than intramuscular P_O2 If this had been the case, Prm_O2 should have varied directly with Pa_O2 during hypoxia and exercise, which it did not. Furthermore, similar concerns would, presumably, exist for mammalian studies of intramuscular P_O2 that involved the implantation of microelectrodes (Hutter et al., 1999; Jung et al.,1999; Suttner et al.,2002). Thus, in addition to the anatomical reasons for elevated Prm_O2 that have been raised by Egginton (2002), the current study has not eliminated the possibility that O₂ tensions are also influenced by the action of the Root effect within the muscle vasculature. Future investigations should perhaps be aimed at experimental manipulation of the Root effect to investigate how this change influences Prm_O2 relative to Pa_O2 and mixed Pv_O2.

The results show that the P_O2 prevailing in the RM of rainbow trout is higher than that reported for skeletal muscles of rats and humans. While there was a significant decrease in Prm_O2 during sustained exercise, it did not decline below 40 mmHg and increased slightly at exhaustion. These observations are taken as a strong indication that O₂ supply to the RM does not become limiting either at the moment of recruitment of WM or at exhaustion. We found no dramatic evidence that the Root effect raises O₂ tensions in red muscle because Prm_O2 remained almost exactly midway between previously published values of Pa_O2 and Pv_O2 for rainbow trout and was sensitive to reductions in Pa_O2 during mild hypoxia. Further work is needed to explain the higher P_O2 in the RM relative to mixed venous blood because, while this may reflect a limited contribution from RM to mixed venous return, the phenomenon might also be a consequence of a transient Root effect in the RM vasculature.

This study was supported by NSERC grants to A.P.F. and D.J.R. D.J.M. was employed by a research project within the 5th Framework of the CEC.