INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

REFLEX MECHANISMS mediated via thin myelinated and unmyelinated sensory nerve fibers have been shown to be of importance for cardiovascular and neuroendocrine responses and for the mobilization of extramuscular fuel stores in exercise (1, 2, 8, 19, 22, 23, 26, 27, 33, 36), whereas regulation of breathing most likely is coupled to motor center activity (14). The stimulus eliciting these reflexes is supposed to be intimately associated with metabolic changes in muscle; for example, the postexercise ischemic pressor response is diminished by blockade of afferent nerves (2, 8, 9). When exercise is performed during hypoxic conditions, metabolic changes in contracting muscle are exaggerated compared with those in normoxic conditions (15). This is accompanied by enhanced responses of cardiac output, heart rate (HR), ventilation, and sympathoadrenergic activity, as well as by enhanced release of glucoregulatory hormones during exercise in hypoxic compared with normoxic conditions (16, 30, 35). These findings are in accordance with the view that the enhancement by hypoxia of extramuscular responses to exercise may be mediated by muscle reflexes. However, during exercise the cardiovascular, respiratory, and neuroendocrine systems may also be influenced from the motor cortex (central command) and by blood-borne mechanisms (11, 20, 21, 24). It cannot be excluded that such effects are enhanced during hypoxia. On this background, the present study was carried out to evaluate if the enhancement by hypoxia of cardiovascular and neuroendocrine responses, as well as of the mobilization of extramuscular fuels during exercise, is influenced by neural feedback from working muscle. Healthy subjects performed leg exercise during both hypoxic and normoxic conditions and both with and without lumbar epidural anesthesia (EA), which impairs afferent nervous activity.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Subjects and maximal oxygen uptake test. Seven healthy, nonsmoking, young males (mean age 25 yr, range 20-32 yr; mean weight 78 kg, range 71-89 kg; mean height 184 cm, range 176-187 cm) gave their informed consent to participate in the study, which was approved by the Ethical Committee of Copenhagen. Subjects abstained from alcohol and exercise on the day before experiments and were encouraged to eat more than 250 g carbohydrates/day for 3 days before each experiment. Each individual appeared at three occasions: first day for testing, second day for control study, and third day for EA experiment. Exercise was performed on a Krogh cycle ergometer in the semisupine (45° angle) position, sitting on a couch behind the ergometer. The feet were placed in shoes fastened to the pedals. On the first of three experimental days, the relationship between oxygen uptake (O₂), HR, and work rate was established by means of graded exercise to exhaustion, and a work load corresponding to ~50% of maximal oxygen uptake (O_{2 max}) was calculated. The mean O_{2 max} was 4.2 l/min (range, 3.9-4.8 l/min), and mean maximal HR was 194 beats/min (range, 180-206 beats/min), obtained at a work load of 255 ± 12 W.

Experimental protocol. The experimental design of the study is outlined in Fig. 1. On the days of control (2nd day) or EA (3rd day) experiments, subjects arrived postabsorptive (after 10-h fast) at 8:00 AM. Catheters were inserted into the brachial artery of the nondominant arm (1.0 mm ID) and into the left femoral artery and vein (1.2 mm ID) 2 cm distal to the inguinal ligament and were advanced 10 cm proximally.

View larger version (10K): [in this window] [in a new window]   Fig. 1.   Subjects rested for 120 min before leg cycling for 20 min at 46% maximal O2 uptake (O2 max) during normoxia (EX1; 20.9% O2). Individuals then rested for 120 min before exercising at similar absolute work load for 20 min during moderate hypoxia (EX2; 11.5% O2); after another 60 min rest individuals cycled to exhaustion (3.4 ± 1.0 min) at severe hypoxia (EX3; 7.8% O2). Before each hypoxic exercise period, hypoxic gas was inspired for 5 min at rest. Experiments were carried out without or with administration of epidermal anesthesia (EA) to impair afferent nervous activity. In EA experiments, anesthesia was administered during rest period before exercise, and after 30 min dysesthesia and leg strength were tested and found to be reduced and stable. All basal measurements were then performed, and exercise protocol was undertaken. Blood samples (Blood) were taken at 5-min intervals during exercise, and muscle biopsy (Biopsy) was obtained at rest after EX1 and EX2. Estimated hepatosplanchnic blood flow (HBF), cardiac output (CO), leg blood flow (LBF), and rating of perceived exertion (RPE) were determined at rest and during exercise as indicated.

On both control and EA days, after a 2-h supine rest for equilibration of infused [³H]glucose (see Procedures) and 10 min of rest in the semisupine position, subjects cycled for 20 min (EX1), aiming at the previously defined work rate corresponding to 50% of individual O_{2 max} in this position. A metronome gave a guiding pedalling frequency of 60 rpm, and the number of revolutions was multiplied with the ergometer braking force to obtain work output. After 2 h of recovery, the subjects breathed air with a fractional concentration of oxygen (FI_O2) of 11.5%. After a further 5 min (acclimatization at rest), the subjects cycled for 20 min (EX2) at the same work rate as in EX1. Finally, after 1 h recovery in ambient air, subjects breathed air with 7.8% O₂. After five min of acclimatization, the subjects cycled to exhaustion, aiming at the same work rate as in the previous bouts of exercise (EX3). Before and immediately after the end of EX1 and EX2, three subjects had a needle biopsy taken from the vastus lateralis muscle after a small incision was made under local anesthesia in the skin and the fascia (2 ml 0.5% lidocaine).

Administration of hypoxia. Hypoxic air was administered from premixed air tanks through a mouth piece. Expired air was collected in Douglas bags, and O₂ and CO₂ concentrations were determined by gas meters (Beckman Instruments). Spirometry was used for determination of ventilation, and O₂ was calculated. Because of technical problems, only spirometry results from EX1 were obtained. An Ergo-Oxyscreen apparatus (Erich Jaeger, Heidelberg, Germany) was used to determine O_{2 max}.

EA. For EA experiments (10-25 days after control), a 20-gauge spinal needle was inserted in the epidural space through vertebral interspace L3-L4. Anesthesia was induced by 20 ml of 2.5% bupivacain (Marcain; Astra). The level of sensory dysesthesia was determined by pin prick and light touch on both sides of the body before and after exercise. A strain gauge used to detect leg strength was placed 35 cm distal to the head of the fibula while the subject was sitting semi-supine with his lower leg vertical. To evaluate loss of muscle strength, maximal voluntary isometric knee extension was performed as the best of three attempts both before and during sensory blockade.

Procedures. Intensity of effort was quantified as the perceived exertion, in which 6 represents rest, 10 represents somewhat hard exercise, and 20 represents very, very intense exercise (3). Mean arterial blood pressure (MAP) was measured by a Bentley transducer positioned at heart level and connected to a Simonsen & Weel amplifier (Copenhagen, Denmark), and HR was obtained from precordial electrodes. Cardiac output was determined by indocyanine green dye (IGC; Cardiogreen; Becton Dickinson, Franklin Lakes, NJ) at rest, after 15 min during the first two exercise bouts, and close to exhaustion in EX3. A bolus of 4 mg of dye (0.8 ml of 5 mg/ml) was injected rapidly into the femoral vein and immediately followed by a flush with 10 ml of isotonic saline. Blood was withdrawn from the femoral artery by a pump (Harvard Apparatus) at 30 ml/min for measurement of dye concentration with a linear densitometer (Waters, Milford, MA). Plasma ICG was read spectrophotometrically at 805 nm. After each experiment, the dye concentration in blood was calibrated. Leg blood flow was measured by the reverse dye technique between the 17th and 19th min (EX1 and EX2) or close to exhaustion (EX3). Two milligrams of ICG were injected into the femoral artery, and blood was withdrawn from the femoral vein. The hepatosplanchnic plasma flow rate was assessed by a bolus injection of 12.5 mg (2.5 ml of 5 mg/ml) of dye into the femoral vein at rest and after 5 min of EX1 and EX2 and by sampling of arterial blood 5, 7.5, 10, and 15 min later. The plasma ICG fractional disappearance rate (k = 0.692, t_1/2) was determined from the regression line for the log plasma concentrations vs. time, and values obtained during exercise were expressed as a percentage of rest. Before any given injection, plasma concentration of ICG was never higher than 5% of the increase expected by the injection.

Arterial and venous blood was sampled at rest and during exercise at 5- to 10-min intervals for determination of hemoglobin, oxygen saturation, PO₂, PCO₂, bicarbonate (HCO₃), pH, and base excess (BE) with an ABL4 and OSM3-hemoxymeter (Radiometer, Copenhagen, Denmark; results only provided for extreme hypoxia conditions). Arterial blood for catecholamine analysis was collected in chilled tubes containing EGTA and reduced glutathione and was placed on ice. Samples were centrifuged promptly at 4°C, and plasma was subsequently stored at 80°. Catecholamine concentrations were determined by a single isotope radioenzymatic method as previously evaluated (21). The concentrations of insulin, pancreatic glucagon, growth hormone (GH), ACTH, cortisol, and plasma renin activity were determined with RIA (10). Free fatty acids (FFA), glycerol, and lactate were determined by enzymatic fluorometric methods, and plasma glucose was determined electrochemically with an automated glucose analyzer (Yellow Springs Instruments). Hematocrit (Hct) was measured by the microhematocrit method.

Glucose appearance and disappearance rates were estimated by a primed (24 µCi) continuous (0.24 µCi/min) infusion of 3-[³H]glucose (2.1 mCi/mol; Isotopapoteket, Denmark) begun at 120 min and continued during rest and the first two exercise periods (18). From measurements of plasma glucose and determination of specific activity of radiolabeled glucose in the preexercise period, individual distribution volumes for glucose (glucose pool/plasma glucose concentration) were determined (34). A pool fraction of 0.65, in which rapid changes in glucose turnover take place, was used (12).

Net uptake, or release over the exercising leg, was calculated from leg plasma flow [LPF = (1  Hct) · leg blood flow] and arterial and venous plasma concentrations of hormones and metabolites. The rate of NE spillover into plasma was calculated as [(Cv  Ca) + Ca(Ee)] · LPF from arterial (Ca) and venous (Cv) plasma concentrations of norepinephrine (NE), the fractional extraction of epinephrine over the exercising leg (Ee), and LPF (32). A good correlation has been found (r = 0.88) between NE extraction determined during infusion of [³H]NE and the epinephrine extraction. In steady state, the extraction of [³H]NE was lower than that of epinephrine (68 ± 0.1% of Ee) (32), perhaps reflecting recirculation of tritiated NE released from sympathetic nerve endings after prior uptake.

Muscle biopsies were frozen immediately in liquid nitrogen and stored at 80°C. The samples were then freeze-dried, freed from blood and connective tissue, and extracted in 0.5 M perchloric acid, which was neutralized with KHCO₃ (2.2 M). Phosphocreatine (PCr), creatine, lactate, glucose-6-phosphate, citrate, malate, fumarate, carnitine, acetylcarnitine, and ATP were measured by enzymatic fluorometric methods (25). ADP, AMP, and inosine 5'-monophosphate (IMP) were measured by HPLC (31).

The Friedman test was used to evaluate whether significant changes occurred with time. If so, the Wilcoxon ranking test for paired data was used to evaluate differences between hypoxia and normoxia, as well as between EA and control experiments. P < 0.05 (2-tailed testing) was considered significant. Data are given as means ± SE or as range. Data from muscle biopsies (n = 3) were analyzed by a t-test.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Effects of hypoxia. Exposure to 11.5% O₂ increased HR, whereas arterial PO₂ (Pa_O2) and arterial O₂ saturation decreased (P < 0.05) (data not shown). In addition, plasma concentrations of epinephrine, FFA, and glycerol increased (P < 0.05; see Fig. 4 and Table 3).

Work rate was lower during EX2 (145 ± 7 W) vs. EX1 (154 ± 9, P < 0.05; Fig. 2). However, the increase in HR, cardiac output [7.0 ± 1.8 to 18.3 ± 1.4 (EX2) vs. 6.2 ± 1.0 to 16.3 ± 2.3 l/min (EX1); see Table 2], leg blood flow [to 6.6 ± 1.7 (EX2) vs. to 6.2 ± 1.1 l/min (EX1)], and perceived exertion (Fig. 2), as well as the decrease in hepatosplanchnic blood flow (see Table 2) was exaggerated in hypoxia. In contrast to normoxic conditions, Pa_O2, arterial PCO₂ (Pa_CO2), and pH decreased during exercise in hypoxia (data not shown).

View larger version (17K): [in this window] [in a new window]   Fig. 2.   Work output and rating of perceived exertion (RPE) during leg cycling at 46% O2 max for 20 min during both normoxia (EX1; 20.9% O2) and moderate hypoxia (EX2; 11.5% O2) and to exhaustion at severe hypoxia (EX3; 7.8% O2). Experiments were carried out without (, solid lines) and with (, dotted lines) EA, which impaired afferent nervous activity. Before exercise in hypoxia, gas mixtures with low O2 content were inhaled for 5 min at rest (thin bars). Exercise duration is indicated by thick bars. Vales are means ± SE (n = 7 subjects). Between EX1 and EX2 a rest period of 120 min was allowed, and between EX2 and EX3 a 30-min recovery period was allowed.

The exercise-induced increases in plasma GH, ACTH, cortisol, renin, epinephrine, and NE, as well as spillover of NE from the leg and decrease in plasma insulin, were exaggerated in hypoxia compared with normoxia (see Fig. 4 and Tables 3 and 5). Also, the increases in glucose production and disappearance of plasma from and glucose uptake in the leg were enhanced (see Fig. 5 and Table 5), as were the release of lactate from the exercising legs and the arterial concentrations of lactate and glycerol (see Fig. 6 and Table 5). The decrease in plasma concentration of FFA with exercise was enhanced by hypoxia, and it became similar to that obtained in normoxia (see Table 3).

In muscle, concentrations of lactate, creatine, and IMP were higher, and those of of PCr were lower (P < 0.05) at the end of exercise in hypoxic compared with normoxic conditions (see Table 4). The exercise-induced increase in MAP was not affected by hypoxia (Fig. 3), and plasma glucagon did not change significantly (see Table 3).

View larger version (15K): [in this window] [in a new window]   Fig. 3.   Intra-arterial mean blood pressure during EX1, EX2, or EX3 without (, solid lines) or with (, dotted lines) EA. Values are means ± SE. At all time points during exercise, values from epidural experiments were lower (P < 0.05) compared with control experiments without EA. For further details see Fig. 1.

In response to marked hypoxia (7.8% O₂) at rest, MAP decreased (Fig. 3; P < 0.05) and HR increased more than during moderate hypoxia (Table 1), as did the plasma catecholamine concentrations (Fig. 4; see Table 6). The subjects exercised at a work load of 112 ± 9 W for 3.7 ± 0.8 min (Fig. 2), and Pa_O2 was lower than in moderate hypoxia (P < 0.05; Table 1). The increase in HR during marked hypoxia was higher than that during moderate hypoxia. The MAP response during marked hypoxia was similar to that during moderate hypoxia; however, it was significantly higher than in normoxia (Figs. 2 and 3; Table 1). Glucose turnover (Fig. 5; see Table 5) and plasma epinephrine (Fig. 4) were further enhanced by severe compared with moderate hypoxia, and plasma ACTH and the renin activity increased during exercise in severe hypoxia (see Table 6). These and other responses could not be compared with those at moderate hypoxia because of differences in sampling times.

View larger version (9K): [in this window] [in a new window]   Fig. 4.   Arterial plasma concentration of norepinephrine (top) and epinephrine (bottom) during leg cycling in EX1, Ex2, or EX3 without (, solid lines) or with (, dotted lines) administration of EA. Values are means ± SE. For further details see Fig. 1.

View larger version (12K): [in this window] [in a new window]   Fig. 5.   Hepatic glucose production in 7 subjects measured by infusion of radiolabeled glucose (3-[3H]glucose) at rest and during leg cycling in EX1, EX2, or EX3 without (, solid lines) or with (, dotted lines) administration of EA. Values are means ± SE. For further details see Fig. 1.

Effects of EA. Thirty-minute administration of EA caused dysesthesia below Th10-Th12, and leg strength was reduced to 71 ± 9% of the control value both before and after the first exercise bout. After the second exercise bout the reduction was to 77 ± 3% of the basal value. Cardiovascular, hormonal, and metabolic variables were not affected significantly by EA.

In normoxia, work load was not influenced significantly by EA [145 ± 13 (EA) vs. 154 ± 9 W (control); Fig. 2], and O₂ was similar with or without EA [1.68 ± 0.09 (EA) vs. 1.88 ± 0.12 l/min (control)], corresponding to 46% of O_{2 max} obtained without EA. EA slightly reduced work load during moderate hypoxia (Fig. 2), and perceived exertion was higher in experiments with EA than in control experiments (P < 0.05; Fig. 2).

Comparison between the two experimental days showed that during normoxia epidural blockade reduced the exercise-induced increase in MAP (P < 0.05) and enhanced the increase in HR (Fig. 3; Table 1). EA also diminished leg NE spillover (P < 0.05; see Table 5) and splanchnic blood flow (P < 0.05), and it increased leg release and the arterial concentration of lactate in normoxic exercise (Fig. 6 and Table 3). During normoxia, remaining variables, e.g., cardiac output and leg blood flow responses to exercise, were not influenced by EA (Table 2). Comparison between the two experimental days revealed that EA did not blunt the influence of moderate hypoxia on any responses to exercise (Figs. 3-6; Tables 2-5). However, interaction between EA and moderate hypoxia elicited the highest values (P < 0.05) for HR, GH, ACTH, renin activity, glucose production, and arterial concentration of lactate and glucose, and the lowest value (P < 0.05) for splanchnic blood flow seen in the first and second exercise period on the two experimental days.

View larger version (11K): [in this window] [in a new window]   Fig. 6.   Lactate exchange over exercising leg rest and during leg cycling in EX1, EX2, or EX3 without (, solid lines) or with (, dotted lines) administration of EA. Values are means ± SE. For further details see Fig. 1.

During 7.8% O₂ hypoxia, subjects performed similar amounts of work [111 ± 10 (EA) vs. 112 ± 9 W (control)], and the time to exhaustion [3.2 ± 1.2 (EA) vs. 3.7 ± 0.8 min (control); Fig. 2] was also similar in experiments with and without EA, respectively. Before exercise, the hypoxia-induced decrease in MAP was enhanced by EA (P < 0.05; Fig. 3). Furthermore, the exercise-induced increase in MAP was blunted by EA (P < 0.05), and the reduction was greater during severe hypoxia compared with normoxic conditions (P < 0.05; Fig. 3). Leg NE spillover during exercise was reduced by EA (P < 0.05; Table 5). EA did not blunt other responses to exercise in severe hypoxia (Figs. 4-6; Tables 1, 5, and 6). In contrast, epinephrine and ACTH responses were enhanced by EA (Fig. 4; Table 6). Cardiac output [to 13.3 ± 3.6 (control) and to 14.1 ± 3.0 l/min (EA)] and leg blood flow [to 4.6 ± 0.8 (control) and to 5.0 ± 0.7 l/min (EA)] increased similarly during exercise in severe hypoxia with and without EA, respectively.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

The present study has confirmed that during exercise cardiovascular and neuroendocrine responses, as well as mobilization of extramuscular fuels, are enhanced by hypoxia (4, 6, 16, 30, 35). Furthermore, as expected, metabolic changes in contracting muscle were also exaggerated by hypoxia (Table 4) (15). The major conclusion of the study is, however, that in essence the hypoxia-induced enhancement of systemic adaptations to exercise is not mediated by neural feedback from working muscle (13). Thus, apart from the blood pressure response to severe hypoxia, hypoxia-induced increases of systemic cardiovascular and neuroendocrine responses and mobilization of extramuscular fuels were not blunted by EA, which markedly impaired spinal conduction of afferent nerve impulses from the exercising legs.

Rather than inhibiting systemic responses to exercise in hypoxia, during moderate hypoxia EA in fact further enhanced heart rate concentrations of GH and ACTH, renin activity, and production and concentration of glucose, and it further diminished splanchnic blood flow. During severe hypoxia EA enhanced epinephrine and ACTH responses (Figs. 5 and 6; Tables 2, 3, 5, and 6). These unexpected findings may at least partly result from an enhanced motor center activity (central command) during exercise with EA (23). In accordance with this view, EA reduced muscle strength and increased perceived exertion. Furthermore, during exercise both with and without hypoxia, EA increased lactate release from the legs and arterial concentrations of lactate despite no significant influence on leg blood flow and femoral venous oxygen tension (Fig. 6; Tables 2 and 5). These findings would be in line with recruitment of more glycolytic muscle fibers as a result of augmented central command.

It cannot be excluded that during EA an exaggerated central command was counterbalanced by blockade of stimulatory effects mediated by afferent nervous activity from muscle. Accordingly, the present study does not completely rule out some role of the latter activity for hypoxia-induced enhancement of responses to exercise. On the other hand, it is reasonable to expect that the influence of hypoxia on muscle also leads to augmented central command and that the latter, accordingly, explains exaggerated systemic responses to exercise during hypoxia, when epidural blockade is not applied. Alternatively, the enhancement by hypoxia of responses to exercise is elicited from sensors depending on oxygen availability (30). However, regarding neuroendocrine responses such sensors remain putative.

At rest MAP was reduced by severe hypoxia, and the reduction was enhanced by EA (Fig. 3). These findings may reflect that hypoxia causes a vasodilation, which is partially compensated by sympathetic vasoconstriction. Correspondingly, during exercise in moderate hypoxia, in contrast with normoxic conditions, overall systemic vascular resistance was reduced in the face of enhanced sympathetic nervous activity, as indicated by an unaltered blood pressure increase despite enhanced increases in cardiac output, leg spillover, and arterial plasma concentration of NE. In line with these findings, blood pressure responses to exercise have previously been found to be either unchanged (30) or somewhat lowered (35) by moderate hypoxia.

EA reduced vascular resistance in exercise during both moderate hypoxia and normoxia, as evidenced by the fact that blood pressure responses were reduced, whereas increases in cardiac output and leg blood flow were unchanged in experiments with epidural blockade compared with experiments without (Fig. 3; Table 2). The effect of EA on blood pressure was more marked during severe compared with moderate hypoxia. During exercise with EA, the decrease in vascular resistance was accompanied by a reduced NE spillover in the legs (Table 5). This may reflect that EA impaired a muscle reflex recruiting sympathetic nervous activity. Alternatively, however, the EA directly blocked the first neuron sympathetic fibers feeding the lower part of the sympathetic chain.

Hepatosplanchnic blood flow was estimated indirectly from ICG plasma clearance, assuming that extraction of ICG in the liver was complete. During normoxia, hepatosplanchnic blood flow was slightly underestimated both at rest and in exercise because hepatosplanchnic ICG extraction has been shown to be <100%. In the study by Rowell et al. (29), hepatosplanchnic extraction was found to range from 87 to 100% at rest, and during both moderate and more intense exercise (using a broad range of exercise work loads from 26 up to 97%) O_{2 max} was reduced to 88-94% (mean 92%) (29), provided a period of 4-5 min was allowed between infusion of ICG and the beginning of blood sampling to allow major volume shifts to occur (29). Still, this inaccuracy regarding ICG in the present experiment cannot influence the conclusion that hepatosplanchnic blood flow decreases during exercise (Table 2). In contrast, the finding that during exercise hepatosplanchnic blood flow was lower in hypoxia compared with normoxia may be accounted for by more marked underestimation of blood flow in the former than in the latter condition. During moderate exercise, ICG extraction is reduced by 25% if 11% O₂ is inspired rather than normal air (30). On the basis of determinations of ICG extraction, hypoxia has previously been found not to influence the decrease in hepatosplanchnic blood flow during exercise (30). Unchanged hepatosplanchnic blood flow and resistance during exercise at moderate hypoxia despite exaggerated sympathetic nervous activity and plasma renin levels are compatible with compensating vasodilating effects of hypoxia and augmented plasma epinephrine concentrations (28).

In the present study, hypoxia was shown to enhance tissue glucose uptake during exercise both by whole body glucose turnover measurements and by direct measurements in the exercising legs (Table 5). Similar findings have previously appeared from studies using one or the other of these methods (4, 6). The hypoxia-induced increase in leg glucose uptake reflected increased blood flow and increased glucose extraction (data not shown) in the leg. The increase in extraction agrees with the fact that hypoxia has been shown to increase glucose transport in muscle in vitro (5). In accordance with previous findings (4, 6), during exercise glucose production was even more enhanced by hypoxia compared with uptake from plasma (4, 6). This effect may at least partly be caused by a direct action of hypoxia on the liver and by exaggerated changes in insulin and epinephrine concentrations (17, 18).

The present study reports data on exercise in extreme normobaric hypoxia. During these conditions subjects were able to perform at a work load of 110 W for 3-4 min. Exhaustion reflected central fatigue as a result of cerebral hypoxia, indicated by waning consciousness. MAP always increased during the exercise, and even during extreme hypoxia the increase in cardiac output is relatively higher than the decrease in peripheral vascular resistance. Interestingly, the fact that endurance and symptoms of cerebral hypoxia were similar in experiments with and without EA, in the face of similar arterial oxygen content but lower blood pressure in the former experiments, indicates that during the prevailing conditions cerebral blood flow and, in turn, oxygen delivery were not critically depending on arterial blood pressure.

The fact that, in response to severe hypoxia, cerebral symptoms occurred during exercise but not at rest is explained by the finding that Pa_O2 decreased from rest to exercise. Breathing air with an O₂ concentration of 7.8%, as in the present study, corresponds to breathing at an altitude of 7,500 m. In agreement with this, during exercise we found values for Pa_O2 (3.2 kPa) and saturation (53%) just as low as previously seen in the simulated high-altitude expedition Operation Everest II (37). In the present study, Pa_O2 also decreased during exercise in moderate hypoxia. The decrease in Pa_O2 in response to exercise was accompanied by a decrease in Pa_CO2 during moderate hypoxia and by no change in Pa_CO2 from resting levels during severe hypoxia. These findings may reflect that, albeit exaggerated, ventilation during exercise in hypoxia was not sufficient to maintain alveolar O₂ tension at resting levels. Alternatively, diffusion capacity limited O₂ transport during exercise in hypoxia (38). This was the case during 120-W exercise in the Operation Everest II study, in which an inspiratory PO₂ of 8.3 kPa was accompanied by an alveolar PO₂ of 5.3 kPa and a Pa_O2 of 3.7 kPa. In that study, the O_{2 max} was reduced to ~50% of values at sea level, and at 120-W dynamic exercise, the O₂ tension in mixed venous blood was 1.8 kPa. In the present study, O₂ tension in venous blood draining the exercising muscle was only 1.2 kPa during severe hypoxia. However, leg O₂ uptake and venous pH and BE values were not significantly different from values during normoxic exercise. These facts add to the view that fatigue was of central rather than of muscular origin.

In conclusion, the present study has demonstrated that, apart from the blood pressure response to severe hypoxia, hypoxia-induced increases in systemic cardiovascular and neuroendocrine responses and mobilization of extramuscular fuels during exercise are not blunted by EA impairing afferent nerve impulses from the exercising legs. Thus in essence, the hypoxia-induced enhancement of systemic adaptations to exercise is not mediated by neural feedback from working muscle.

Perspectives