Supraspinal fatigue, defined as an exercise-induced decline in force caused by suboptimal output from the motor cortex, accounts for over one-quarter of the force loss after fatiguing contractions of the knee extensors in normoxia. We tested the hypothesis that the relative contribution of supraspinal fatigue would be elevated with increasing severities of acute hypoxia. On separate days, 11 healthy men performed sets of intermittent, isometric, quadriceps contractions at 60% maximal voluntary contraction to task failure in normoxia (inspired O2 fraction/arterial O2 saturation = 0.21/98%), mild hypoxia (0.16/93%), moderate hypoxia (0.13/85%), and severe hypoxia (0.10/74%). Electrical stimulation of the femoral nerve was performed to assess neuromuscular transmission and contractile properties of muscle fibers. Transcranial magnetic stimulation was delivered to the motor cortex to quantify corticospinal excitability and voluntary activation. After 10 min of breathing the test gas, neuromuscular function and cortical voluntary activation prefatigue were unaffected in any condition. The fatigue protocol resulted in ∼30% declines in maximal voluntary contraction force in all conditions, despite differences in time-to-task failure (24.7 min in normoxia vs. 15.9 min in severe hypoxia, P < 0.05). Potentiated quadriceps twitch force declined in all conditions, but the decline in severe hypoxia was less than that in normoxia (P < 0.05). Cortical voluntary activation also declined in all conditions, but the deficit in severe hypoxia exceeded that in normoxia (P < 0.05). The additional central fatigue in severe hypoxia was not due to altered corticospinal excitability, as electromyographic responses to transcranial magnetic stimulation were unchanged. Results indicate that peripheral mechanisms of fatigue contribute relatively more to the reduction in force-generating capacity of the knee extensors following submaximal intermittent isometric contractions in normoxia and mild to moderate hypoxia, whereas supraspinal fatigue plays a greater role in severe hypoxia.
- central fatigue
- peripheral fatigue
- transcranial magnetic stimulation
whole body exercise performance in aerobic activities is impaired in hypoxia (62). The mechanisms underpinning this impairment in performance are not fully understood. In mild to moderate hypoxia, the decline in performance is associated with an increased metabolic disturbance (6, 10, 41, 42). This hypoxia-induced metabolic disturbance evokes a rise in discharge frequency of group III/IV muscle afferents (24) that may affect exercise performance through increased inhibitory influences on central motor drive (5, 9). In more severe hypoxia, afferent feedback from locomotor muscles may contribute less to exercise termination, as demonstrated by the prolongation of exercise time with reversal of arterial O2 desaturation via hyperoxygenation at task failure (10, 13, 31, 52). The effect that hyperoxygenation has on exercise performance at task failure occurs too quickly to be mediated by a reduction in accumulating metabolites. Therefore, the decrease in whole body exercise performance in severe hypoxia has been suggested to result directly from a reduction in motor command from the hypoxic central nervous system (CNS) (7).
Testing the effect of hypoxia on fatigue mechanisms during whole body exercise is complicated because there is not only a decrease in arterial O2 saturation (SaO2), but also an increase in cardiorespiratory requirements. The reduction in O2 delivery during whole body exercise in hypoxia precipitates a reduction in peak work rate and maximal O2 uptake, resulting in a shift of a given absolute workload to a higher relative exercise intensity (4). This increase in the relative exercise intensity increases the rate of accumulation of metabolites (23, 25) and thus the rate of fatigue of limb locomotor muscles (6, 56). The work of breathing also contributes significantly to the rate of development of limb locomotor muscle fatigue during whole body exercise in acute hypoxia by reducing blood flow, and hence O2 delivery, to the working limb (8, 22). When a small muscle mass is activated, a given absolute force output is carried out at the same relative exercise intensity (29, 30), and the cardiorespiratory requirements are reduced (14). Thus small muscle mass exercise is a suitable model for investigating the independent effects of SaO2 on muscle fatigue.
Acute hypoxia has been shown to accelerate the decline in voluntary force output with intermittent maximal and submaximal isometric contractions (16, 20, 29, 30, 36). Such reductions in voluntary force are associated with an increased rate of development of peripheral fatigue, as demonstrated by an accelerated decline in the responses evoked by peripheral stimuli and an increased rate of rise of electromyographic (EMG) signals during the contractions (20, 29, 30). In addition to a failure of contractile mechanisms (i.e., peripheral fatigue), the reduction in voluntary force output during intermittent isometric contractions in hypoxia may result from a progressive failure of voluntary muscle activation (i.e., central fatigue) (20, 29). This conclusion has been inferred from deficits in voluntary activation estimated by interpolation of a single stimulus to the motor nerve during a maximal voluntary contraction (i.e., “twitch interpolation”) (35). The twitch interpolation method, however, does not enable the exact site of fatigue to be determined, because decrements in voluntary activation could be mediated at any site proximal to the motoneurons, including reflex, spinal, brain stem, or supraspinal circuits (18). More specific information regarding the site of central fatigue can be discerned using transcranial magnetic stimulation (TMS) of the motor cortex. Presence of a superimposed twitch (SIT) evoked by TMS during maximal contraction implies that voluntary output from the motor cortex is insufficient to drive the motoneuron pool optimally. A progressive failure of drive from the motor cortex indicates that central fatigue has a supraspinal component (59). Supraspinal fatigue has been demonstrated after maximal and submaximal isometric contractions in normoxia (26, 50). However, the site(s) for the failure of voluntary drive during hypoxia is unknown.
The aim of the present study, therefore, was to further understand the mechanisms and sites for the reduction in force-generating capacity of human knee extensors in response to submaximal, intermittent, isometric contractions of the quadriceps under varying fractions of inspired oxygen (FiO2), ranging from normoxia to severe hypoxia. The relative contributions of peripheral and central mechanisms of fatigue to the reduction in force-generating capacity were assessed using motor nerve stimulation and TMS, respectively. We hypothesized that the peripheral contribution to fatigue would predominate in conditions of normoxia to moderate hypoxia (FiO2 0.21–0.13), whereas supraspinal fatigue would become more important in severe hypoxia (FiO2 0.10).
Eleven healthy, recreationally active male volunteers participated in the study (mean ± SD age 23.5 ± 2.8 yr, stature 1.77 ± 0.06 m, body mass 77.9 ± 10.4 kg). The participants were asked to avoid vigorous exercise for 24 h, caffeine for 12 h, and food for 2 h before the trials. All participants gave written, informed consent before the commencement of the study once the experimental procedures, associated risks, and potential benefits of participation had been explained. The study was approved by the Brunel University Research Ethics Committee. All procedures conformed to the Declaration of Helsinki.
Each participant completed a familiarization session and four experimental trials during which they breathed normoxic (FiO2 0.21) or hypoxic (FiO2 0.16 mild, 0.13 moderate, and 0.10 severe) gas mixtures. The trials were randomized, separated by at least 5 days, and performed at the same time of day under consistent laboratory conditions (temperature 22 ± 1°C, humidity 50 ± 10%, barometric pressure 757 ± 4 mmHg). During each trial, electrical stimulation of the femoral nerve was performed at resting baseline to assess neuromuscular transmission and contractile properties. In addition, TMS was delivered to the motor cortex during voluntary contractions to determine cortical voluntary activation. Baseline measurements were repeated after a 10 min wash-in of the test gas, during a fatigue protocol, then immediately (<2 min) after the fatigue protocol, while participants continued to breathe the test gas, and at 15, 30, and 45 min into recovery while breathing ambient air. Arterial and cerebral/muscle oxygenation were measured at rest and throughout the fatigue protocol using pulse oximetry and near-infrared spectroscopy (NIRS), respectively. Participants were blinded to the FiO2 and kept naive to the expected outcomes of the study.
Force and EMG recordings.
Knee-extensor force during voluntary and evoked contractions was measured using a calibrated load cell (model ABA Ergo Meter, Globus Italia, Codogne, Italy) connected to a noncompliant strap attached around the participant's right leg, just superior to the ankle malleoli. The load cell was fixed to a custom-built chair and adjusted to a height that was in the direct line of applied force for each participant. Participants sat upright in the chair with the hips and right knee at 1.57 rad (90°) of flexion. EMG activity of the knee extensors and flexors was recorded from the right vastus lateralis and biceps femoris, respectively. After the skin was shaved and swabbed with isopropyl 70% alcohol, self-adhesive electrodes (Kendall H59P, Tyco Healthcare Group, Mansfield, MA) were placed 2 cm apart over the muscle bellies, and a reference electrode was placed over the patella. The positions of the EMG electrodes were marked with indelible ink to ensure that they were placed in the same location at subsequent visits. The electrodes were used to record the electrically evoked compound muscle action potential (M-wave) and the motor evoked potential (MEP) elicited by TMS. During the fatigue protocol, EMG was quantified as the average root-mean-square amplitude for each set of submaximal and maximal contractions (EMGrms; time constant = 0.25 s). All signals were amplified (gain 1,000; 1902, Cambridge Electronic Design, Cambridge, UK), band-pass filtered (EMG only: 20–2,000 Hz), digitized (4 kHz; micro 1401, Cambridge Electronic Design), and finally acquired for post hoc analysis (Spike 2 version 5.20, Cambridge Electronic Design).
Force and EMG parameters were assessed before and up to 45 min after the fatigue protocol. At the beginning of each assessment period, maximal voluntary contraction (MVC) force was determined from three control contractions. Femoral nerve stimulation was delivered during each MVC, and an additional stimulus was delivered at rest, ∼2 s after the superimposed stimulus, to determine the potentiated quadriceps twitch (Qtw,pot) and hence peripheral voluntary activation (see Data analysis below). TMS was delivered during brief (∼5 s) voluntary contractions at 100, 75, and 50% MVC separated by ∼5 s of rest, to determine cortical voluntary activation (Fig. 1). The three contraction intensities were repeated three times with 15 s between each set. Participants received visual feedback of the target force on a computer monitor.
Femoral nerve stimulation.
Single electrical stimuli of 200-μs duration were delivered to the right femoral nerve via 32-mm-diameter surface electrodes (CF3200, Nidd Valley Medical, North Yorkshire, UK) using a constant-current stimulator (DS7AH, Digitimer, Welwyn Garden City, Hertfordshire, UK). The cathode was positioned over the nerve high in the femoral triangle, and the anode was placed midway between the greater trochanter and the iliac crest (49). The site of stimulation that produced the largest quadriceps twitch amplitude (Qtw) and maximal M-wave (Mmax) at rest was located. Single stimuli were delivered during an incremental protocol, beginning at 100 mA and increasing by 20 mA until plateaus were evident in Qtw and Mmax. To ensure supramaximal stimulation, the stimulation intensity was then increased by 30%. The stimulation intensity was kept constant throughout each of the experimental trials (mean current = 291 ± 71 mA). Muscle contractility was assessed for each peripherally derived resting twitch as maximum rate of force development (MRFO), contraction time (CT), maximum relaxation rate (MRR), and one-half relaxation time (RT0.5). Membrane excitability was determined by measuring the peak-to-peak amplitude and area of the electrically evoked Mmax.
Single magnetic stimuli of 1-ms duration were applied over the left motor cortex (postero-anterior intracranial current flow) using a monopulse magnetic stimulator (Magstim 200, The Magstim Company, Whitland, UK) with a concave double-cone coil (110 mm diameter; maximum output 1.4 T). The coil was placed such that a large MEP was elicited in the vastus lateralis with only a small MEP in the antagonist (biceps femoris). This optimal coil position (1.5 ± 0.6 cm lateral to the vertex) was marked on the scalp with indelible ink to ensure reproducibility of the stimulation conditions for each participant throughout the entire experiment. Resting motor threshold for the knee extensors was identified by constructing a stimulus-response curve. Stimulator output was decreased in 5% increments from 80% until the MEP response was below 0.05 mV in more than one-half of eight stimuli (47). Resting motor threshold for the knee extensors occurred at 56 ± 5% of maximum stimulator output. During each of the experimental trials, TMS was delivered at 130% of resting motor threshold (73 ± 7% maximum stimulator output). This stimulation intensity elicited a large MEP in the vastus lateralis (area between 60 and 100% Mmax during knee-extensor contractions ≥50% MVC; Fig. 2), indicating that the TMS stimulus activated a high proportion of knee-extensor motor units, and a small MEP in the biceps femoris (amplitude ∼20% knee-extensor MEP during knee-extensor contractions).
Ventilatory and pulmonary gas exchange were measured using an online system (Quark b2, Cosmed, Rome, Italy). SaO2 was estimated using an oximeter with finger sensor (SpO2) (Model 2500A, Nonin Medical, Plymouth, MN). In a separate group of participants (n = 6), we found excellent agreement between directly measured SaO2 (ABL800 FLEX, Radiometer, Copenhagen, Denmark) and estimated values over the 60–100% range (mean coefficient of variation = 3.2%; intraclass correlation coefficient = 0.83; y = 0.79x + 0.23, where y = %SpO2 and x = %SaO2, SE of the estimate = 1.1%). Hemoglobin concentration ([Hb]) was determined at baseline from an earlobe capillary blood sample (HemoCue, Ängelholm, Sweden). Arterial O2 content (CaO2) was estimated using the measured [Hb] and an assumed alveolar [estimated via end-tidal partial pressure of O2 (PetO2)] to arterial O2 difference of 10 Torr (29): Heart rate was measured using a short-range telemetry system (Polar Electro Oy, Finland). Ratings of perceived exertion for dyspnea and limb discomfort were obtained at baseline, after wash-in of the test gas, and after every MVC during the fatigue protocol using Borg's modified CR10 scale (11).
Participants were instrumented with three near-infrared sensors (SomaSensor, Somanetics, Troy, MI) to monitor absorption of light across cerebral and muscle tissue (INVOS 5100C, Somanetics) (40). Two near-infrared sensors were placed over the left and right frontal lobe region of the forehead; these signals were averaged to determine cerebral oxygenation. A third sensor was affixed over the belly of the right vastus lateralis muscle, ∼15 cm above the proximal border of the patella and 5 cm lateral to the midline of the thigh. The sensors alternately emit two wavelengths of near-infrared light (730 and 810 nm). The sensors also contain two detectors located at 3 and 4 cm from the emitting source that detect oxygenated and deoxygenated states of hemoglobin to estimate an index of regional O2 saturation based on internal microprocessing algorithms. Each sensor was secured to the skin using adhesive tape and shielded from ambient light using elastic bandages. The position of the sensor on the vastus lateralis was marked with indelible ink for consistent application during subsequent visits. Statistical analyses were performed on the absolute values, whereas plotted data were normalized to reflect changes from the mean value attained during the baseline rest period.
Participants wore a nose clip and breathed through a mouthpiece connected to a two-way non-rebreathing valve (Salford HPL). The inspiratory port was connected via wide-bore tubing to a Douglas bag containing the test gas (BOC Gases, Surrey, UK). The gas was humidified by heating water in the bottom of the bag using a ceramic hotplate (Bibby HB500, Wolf Laboratories, York, UK). Sets of five submaximal isometric knee-extensor contractions (target force 60% of initial MVC), followed by one MVC maneuver, were performed; each contraction was held for 5 s, followed by 5-s relaxation. The sets of contractions were separated by 15 s and were performed until the participant failed to reach the 60% target force on three occasions in one set (i.e., task failure). Participants received continuous visual feedback of force on a computer monitor and were verbally encouraged to maintain force during the contractions. A computer-controlled audible metronome was used to ensure maintenance of the correct rhythm. The fatigue protocol was repeated at the same workload in all four conditions (mean target force 318 ± 34 N).
Peripheral voluntary activation was assessed using twitch interpolation (35). Briefly, the force produced during a superimposed single twitch delivered within 0.5 s of peak force being attained early during the MVC was compared with the force produced by the single twitch delivered during relaxation ∼2 s after the MVC. Cortical voluntary activation was quantified by measurement of the force responses to motor cortex stimulation. Motor cortex and spinal cord excitability increase during voluntary contraction (43); therefore, it was necessary to estimate rather than measure directly the amplitude of the resting twitch evoked by motor-cortex stimulation (60, 61). The mean SIT amplitude at each contraction strength (100, 75, and 50% MVC) was calculated, and the y-intercept of the linear regression between the mean SITs and voluntary force was used to quantify the estimated resting twitch (ERT) (21, 60, 61). Cortical voluntary activation (%) was quantified using the equation: [1 − (SIT/ERT)] × 100. The reliability of the TMS protocol for the determination of voluntary activation and ERT for the knee extensors has been established in our laboratory (21) and elsewhere (48).
The peak-to-peak amplitude and area of MEPs and Mmax were measured offline. The area of vastus lateralis MEPs was normalized to that of Mmax elicited during the MVC at the beginning of each assessment period to ensure the motor cortex stimulus was activating a high proportion of the knee-extensor motor units (19, 57). The area of biceps femoris MEPs was not normalized to Mmax because of the discomfort associated with maximally stimulating the sciatic nerve (48). The duration of the cortical silent period evoked by TMS delivered during the MVC was determined as the interval from stimulation to the time at which poststimulus EMG exceeded ±2 SD of prestimulus EMG for at least 100 ms.
Two-way ANOVA with repeated measures on condition (FiO2 0.21, 0.16, 0.13, and 0.10) and time (baseline, wash-in, and 0, 15, 30, and 45 min postfatigue) was used to test for within-group differences in evoked and voluntary force and EMG measures, cardiorespiratory and perceptual measures, and cerebral and muscle oxygenation. Least squares linear regression analysis was used to evaluate the rate of change in measures of muscle function (MVC, Qtw,pot, EMGrms, SIT) during the fatigue protocol relative to the number of sets. For each participant, the regression equations were determined over the time equivalent to that during the shortest condition. Goodness of fit was determined by calculating the group mean coefficient of determination (r2) and SE of the estimate for each variable. One-way ANOVA with repeated measures on condition was used to test for within-group differences in the slope and y-intercept values for each variable. When ANOVA revealed a significant main effect, pairwise comparisons were made using the Bonferroni method. Pearson product-moment correlation was used to determine the relationships between selected physiological measures. Data are presented as means ± SD within the text and displayed as means ± SE in Figs. 2⇓⇓–5. Statistical analyses were performed using SPSS (version 15.0, Chicago, IL), and statistical significance was set at P < 0.05.
There was a progressive effect of FiO2 on cardiorespiratory function at rest and during the fatigue protocol, as shown by the decreases in SpO2, CaO2, and PetO2 (Table 1). During the fatigue protocol, minute ventilation in severe hypoxia was elevated compared with normoxia and mild hypoxia, but O2 uptake and CO2 production were similar across conditions. Thus the ventilatory equivalents for O2 and CO2 were higher, whereas the PetO2 and end-tidal partial pressure of CO2 (PetCO2) were lower in severe hypoxia than in the other conditions. At end-exercise, leg discomfort was close to maximum in all conditions. Time-to-task failure ranged from 24.7 ± 5.5 min in normoxia to 15.9 ± 5.4 min in severe hypoxia [F(1,10) = 7.1, P = 0.004 vs. normoxia; F(1,10) = 7.1, P = 0.002 vs. mild hypoxia].
Cerebral and muscle oxygenation.
There was a dose response between the severity of hypoxia and the decrease in cerebral oxygenation below baseline during the wash-in, with greater reductions occurring in severe hypoxia (72.5 ± 8.1 vs. 53.2 ± 7.8%, P < 0.001) than in mild hypoxia (73.0 ± 7.7 vs. 69.6 ± 7.7%, P < 0.001) (Fig. 3A). A similar but less exaggerated response was observed for muscle oxygenation, with greater reductions occurring in severe hypoxia (77.1 ± 7.5 vs. 67.2 ± 9.7%, P < 0.001) than in moderate hypoxia (76.1 ± 8.7 vs. 73.0 ± 7.7%, P < 0.001) (Fig. 3B).
Compared with values attained during the wash-in, the fatigue protocol elicited an increase in cerebral oxygenation in normoxia (72.8 ± 7.4 vs. 79.2 ± 8.3%, P = 0.002), mild hypoxia (69.6 ± 7.7 vs. 76.2 ± 8.3%, P < 0.001), and moderate hypoxia (63.1 ± 7.7 vs. 68.7 ± 7.3%, P < 0.001); a similar trend was found in severe hypoxia, but the final minute value was not different from that attained after the wash-in (53.2 ± 7.8 vs. 57.1 ± 7.5%, P = 0.091) (Fig. 3A). During the final minute of the fatigue protocol, cerebral oxygenation remained below baseline in moderate and severe hypoxia (P < 0.05; Fig. 3A). There was a sudden and sustained decrease in muscle oxygenation during the fatigue protocol in normoxia (76.0 ± 8.6 vs. 42.9 ± 19.8%, P < 0.01), mild hypoxia (75.2 ± 9.9 vs. 44.4 ± 14.3%, P < 0.001), moderate hypoxia (73.0 ± 7.7 vs. 38.9 ± 16.6%, P < 0.001), and severe hypoxia (67.2 ± 9.7 vs. 36.6 ± 13.2%, P < 0.001) (Fig. 3B). During the final minute of the fatigue protocol, muscle oxygenation in severe hypoxia was significantly reduced compared with that in normoxia and mild hypoxia (P < 0.05).
Voluntary and evoked measures of muscle function did not differ between conditions at baseline and were unaffected by wash-in of the test gas (Fig. 4 and Table 2). MVC force was reduced below baseline immediately after the fatigue protocol in normoxia (−30.1 ± 7.9%), mild hypoxia (−31.1 ± 9.3%), moderate hypoxia (−31.5 ± 6.7%), and severe hypoxia (−30.8 ± 7.8%); these reductions were not different between conditions [F(3,30) = 0.5, P = 0.69] (Fig. 4A). Motor nerve estimates of voluntary activation immediately after the fatigue protocol fell in normoxia (93.4 ± 3.3 vs. 83.5 ± 9.2%; P = 0.003), mild hypoxia (94.9 ± 2.7 vs. 82.0 ± 9.4%; P = 0.001), moderate hypoxia (91.8 ± 4.6 vs. 79.3 ± 11.9%; P < 0.001), and severe hypoxia (94.2 ± 3.2 vs. 82.7 ± 7.6%; P < 0.001); these reductions were also not different between conditions [F(2,21) = 0.5, P = 0.63]. The fatigue protocol resulted in a decline in Qtw,pot in normoxia (−34 ± 17%), mild hypoxia [F(3,30) = 5.3, P = 1.00 vs. normoxia; −37 ± 23%], moderate hypoxia [F(3,30) = 5.3, P = 0.144 vs. normoxia; −27 ± 17%], and severe hypoxia [F(3,30) = 5.3, P = 0.022 vs. normoxia; −20 ± 20%] (Fig. 4B). The fatigue protocol elicited decreases in peripherally derived measures of muscle contractility (i.e., maximum rate of force development, contraction time, maximum relaxation rate, and one-half relaxation time; see Table 2). M-wave properties did not change in response to the fatigue protocol in any condition (Table 2). In addition, M-waves were not observed in the biceps femoris during femoral nerve stimulation. During the fatigue protocol, there was a linear relationship between selected measures of muscle function and set number, as demonstrated by the moderate-to-high r2 and low SE of the estimate values for the respective regression equations in each condition (Table 3). Specifically, there were progressive decreases in MVC (force and EMGrms) and Qtw,pot, and a progressive increase in EMGrms during the submaximal contractions. There was a tendency for the rate of change to increase with the severity of hypoxia, but the slope for each measure was not significantly different between conditions (Table 4). The EMGrms during the last set of submaximal contractions also did not differ between conditions (36 vs. 49% above baseline in normoxia vs. severe hypoxia).
MVC force remained below baseline up to 45 min after the fatigue protocol in all conditions, but there was a tendency for MVC force to recover quicker in severe hypoxia compared with normoxia (Fig. 4A). Qtw,pot also remained below baseline up to 45 min into recovery in all conditions, but the nadir occurred at ∼15 min after fatigue (Fig. 4B). The decrease in Qtw,pot was greater at 0 and 15 min after fatigue in severe hypoxia vs. normoxia [F(3,30) = 7.9, P = 0.048], but not at 30 min [F(3,30) = 4.1, P = 0.10] or 45 min [F(3,30) = 4.8, P = 0.084]. Peripherally measured voluntary activation remained below baseline at 15 min after fatigue in all conditions (P < 0.05), but was not different from baseline at 30 min after fatigue (data not shown).
Cortical voluntary activation, MEP characteristics, and cortical silent period after the wash-in did not differ from baseline. Although the relationship between SIT and set number during the fatigue protocol was less strong than for other measures of muscle function (Table 3), there was a tendency for the rate of change to increase in line with the severity of hypoxia (Table 4). The aforementioned decline in peripheral force-generating capacity immediately after the fatigue protocol was accompanied by an increase in the force produced by the TMS stimulus superimposed onto voluntary contractions. Thus the average SIT evoked at baseline (pooled across all conditions) was 1.4 ± 1.3% of MVC, whereas during the final MVC the SIT was 2.9 ± 2.4% in normoxia (P = 0.030), 2.8 ± 2.6% in mild hypoxia (P = 0.002), 4.1 ± 3.3% in moderate hypoxia (P = 0.011), and 5.3 ± 3.9% in severe hypoxia (P = 0.004). In turn, the ERT immediately after fatigue was reduced below baseline in all conditions [F(3,30) = 3.1, P = 0.041; Table 2], such that cortical voluntary activation was also reduced in normoxia (86.8 ± 10.9 vs. 94.8 ± 4.8%, P = 0.005), mild hypoxia (86.0 ± 10.7 vs. 96.7 ± 3.6%, P < 0.001), moderate hypoxia (81.6 ± 12.1 vs. 94.1 ± 4.8%, P < 0.001), and severe hypoxia (70.7 ± 24.3 vs. 94.5 ± 4.7%, P = 0.005) (Fig. 5). The decrease in cortical voluntary activation in severe hypoxia was greater than in normoxia [F(1,13) = 4.1, P = 0.048]. There was a strong correlation for group mean values between cortical voluntary activation and cerebral oxygenation obtained immediately postexercise (r = 0.93, P = 0.026). The SIT evoked by TMS returned to baseline by 15 min after the fatigue protocol in all conditions [F(2,16) = 1.1, P = 0.42]. Consequently, cortical voluntary activation was not different from baseline at 15 min after the fatigue protocol in any condition [F(2,16) = 1.2, P = 0.81; Fig. 5]. During voluntary contractions, neither MEP characteristics, nor cortical silent period changed across time in any condition (pooled average silent period: 221 ± 78 ms at baseline vs. 233 ± 100 ms postfatigue).
This study assessed the mechanisms and sites for the reduction in force-generating capacity with submaximal, intermittent, isometric contractions of human knee extensors under varying conditions of FiO2, ranging from normoxia to severe hypoxia. In line with our hypothesis, the peripheral contribution to fatigue predominated in conditions of normoxia to moderate hypoxia, whereas failure of drive from the motor cortex (supraspinal fatigue) became more important in severe hypoxia. Another novel finding was that acute hypoxia had no effect on prefatigue cortical function.
Prefatigue neuromuscular function.
The MVC, Qtw,pot, and M-wave characteristics were not influenced by acute hypoxia. These data support existing evidence that hypoxia has minimal effect on voluntary force, contractile properties, or neuromuscular transmission (7, 39). We extend these previous findings by showing also that cortical voluntary activation of knee extensors is unaffected by acute hypoxia. In the only other study to assess directly the influence of hypoxia on motor cortical function in unfatigued healthy humans, Szubski et al. (54) used TMS and EMG recordings from the first dorsal interosseus muscle to compare corticospinal excitability in normoxia and acute hypoxia (FiO2 = 0.12, SpO2 = 75%). In agreement with the present study, those authors found no effect of hypoxia on MEP amplitude. In contrast, they reported a significant shortening of the cortical silent period in hypoxia. The disparity may stem from a difference in the muscle group tested, the length of exposure to hypoxia, or the method used to determine the silent period. Collectively, our findings indicate that reduced arterial oxygenation per se has no effect on either neuromuscular properties or cortical voluntary activation of the knee extensors.
As maximal voluntary force fell during the fatigue protocol, there was an increase in knee-extensor force elicited by TMS, signifying the development of supraspinal fatigue. Participants became unable to activate the knee-extensor muscles fully with continuing maximal voluntary efforts. The increase in force elicited with TMS was largest during exercise in severe hypoxia. Previous studies have been unable to discern the etiology of hypoxia-induced central fatigue, because estimates of voluntary activation were made using motor nerve stimulation (see Introduction). In the present study, both TMS and motor nerve stimulation were used to monitor cortical and peripherally derived voluntary activation of the knee extensors, respectively. Cortical voluntary activation was affected most after exercise in severe hypoxia and least in normoxia, supporting the suggestion that a larger contribution from central mechanisms of fatigue is apparent as the severity of hypoxia increases (10). Although the deficits in voluntary activation were reduced with motor nerve stimulation vs. TMS, it is problematic to compare voluntary activation measured using these methods (59). The shape of the voluntary force vs. superimposed-twitch relationship differs between motor nerve and motor cortical methods of stimulation (60); the relationship becomes nonlinear when using motor nerve stimulation at high contraction strengths, with the implication that changes in voluntary force elicit minimal changes in SIT size (2, 60). Regardless, data from the present study indicate that a large proportion of the reduction in voluntary drive after exercise during increasing severities of hypoxia occurs because of suboptimal output from the motor cortex.
Because the relationship between force output and cortical voluntary activation of the knee extensors is linear (21, 48), it is possible to estimate the contribution of supraspinal fatigue to the total force loss. The MVC force decreased to ∼70% of baseline immediately after exercise in each condition, whereas cortical voluntary activation decreased from ∼95% at baseline to 87% in normoxia, 86% in mild hypoxia, 82% in moderate hypoxia, and 71% in severe hypoxia. By assuming that cortical voluntary activation remained at prefatigue levels, it is possible to determine the magnitude of voluntary force loss postexercise (50). Using this approach, supraspinal fatigue in normoxia accounted for ∼18% of the decrease in voluntary force, whereas in mild-to-moderate hypoxia the supraspinal component of fatigue was ∼25%. The greatest role of supraspinal fatigue occurred in severe hypoxia, where 54% of the drop in voluntary force was due to processes at or above the level of motor cortical output. Thus increasing severities of hypoxia exacerbate the supraspinal component of fatigue and reduce the relative contribution of peripheral fatigue to the total force loss.
Cortical activation was significantly reduced immediately after exercise in all conditions, suggesting that the mechanisms of central fatigue acted upstream of the motor cortex to impair voluntary descending drive (17, 59). Firing of fatigue-sensitive muscle afferents exerts an inhibitory influence on motor cortical cells (33, 34). Furthermore, multiple ascending afferent pathways have been described that affect higher centers in the brain, including the prefrontal cortex, the thalamus, and the cerebellum (3). The net discharge frequency of group III/IV afferents is higher during muscle contractions in hypoxia vs. normoxia; this increased net discharge results from a higher baseline firing frequency plus an additional increase in firing frequency evoked by the hypoxia-induced accumulation of muscle metabolites (24). Persistent impairments in cortical voluntary activation in response to locomotor exercise have been associated with long-term disturbances in metabolic homeostasis (49). Thus the additional supraspinal fatigue in severe hypoxia noted in the present study may have been due, in part, to the elevated inhibitory influences on central motor drive mediated by metabosensitive muscle afferents.
The MEP evoked by TMS during a voluntary contraction is influenced by corticospinal cell and motoneuron responsiveness and can be inferred as a measure of corticospinal excitability (58). The MEP is followed by a period of EMG silence, the initial part of which has been attributed to spinal mechanisms (28), whereas the later period (>100 ms) may represent increased cortical inhibition (15, 28, 58). In the present study, repetitive isometric contractions did not alter either the MEP characteristics or the cortical silent period in any condition. In the only other study to assess the influence of hypoxia on cortical function during fatiguing exercise, Szubski et al. (55) reported an increased MEP amplitude and silent period with sustained contraction of the first dorsal interosseus muscle. In line with the present study, however, the cortical responses were not different in normoxia vs. acute hypoxia (FiO2 = 0.12, SpO2 = 75%). Together, these results suggest that fatiguing exercise in hypoxia does not impair the responsiveness of the neurons involved in motor cortical output to muscle.
Maximum force-generating capacity (MVC force) declined progressively over time and reached ∼70% of baseline immediately after the fatigue protocol in all conditions, indicating that the exercise induced substantial fatigue. That exercise was terminated in each condition with a similar decrease in MVC force, despite significant differences in exercise time, suggests that a critical threshold of muscle fatigue existed at the point of task failure. This finding is in line with previous studies using intermittent isometric knee-extensor contractions (36) and locomotor exercise involving the knee extensors (5, 6). Maximum force-generating capacity tended to recover quickest after exercise in severe hypoxia (Fig. 4A), which may denote a blunted metabolic disturbance in response to exercise in this condition. In conjunction with the impairment in maximum force-generating capacity, there was a progressive increase in neural drive over time, as reflected by EMGrms during submaximal contractions, presumably to compensate for fatiguing muscle fibers. In addition, there was a progressive reduction over time in the force response to motor nerve stimulation (Qtw,pot), as well as pre- to postexercise declines in Qtw,pot and the cortically derived ERT in all conditions. Preservation of neuromuscular transmission in all conditions, as judged by maintenance of the M-waves, places the site of fatigue beyond the sarcolemma. The most likely explanation for the exercise-induced reductions in Qtw,pot is excitation-contraction uncoupling via diminished Ca2+ release and disruption of the contractile apparatus (1). The dominance of peripheral effects on task failure was demonstrated at end-exercise in normoxia and mild hypoxia. The level of end-exercise peripheral fatigue (change in Qtw,pot, before vs. after exercise) was substantial and similar at task failure in normoxia and mild hypoxia. In moderate hypoxia, the amount of peripheral fatigue at task failure tended to be less than in normoxia, and in severe hypoxia the end-exercise level of peripheral fatigue was significantly less than in normoxia (Fig. 4B). These results are in line with previous studies (10, 52) and suggest that peripheral fatigue and the associated inhibitory afferent feedback to the CNS might play an important role in the decision to stop exercising in conditions of normoxia to moderate hypoxia, whereas cerebral oxygenation may become the dominant regulated variable in severe hypoxia.
Hypoxia during resting conditions had a profound effect on cerebral oxygenation (Fig. 3A), despite no changes in MVC force or cortical voluntary activation. These results suggest that cerebral hypoxia has negligible effect on maximal central motor drive under such conditions. Our measure of cerebral oxygenation using NIRS over the frontal lobes has been shown to reflect changes in oxygenation of the motor cortex during exercise in hypoxia (53). Previous research suggests that the decrease in cerebral oxygenation may hasten the decision to stop exercising by supraspinal mechanisms (53). Throughout exercise in all conditions, cerebral oxygenation rose, reflecting increases in CaO2 and cerebral blood flow (12). In normoxia and mild hypoxia, cerebral oxygenation rose above baseline, whereas in moderate and especially in severe hypoxia, cerebral oxygenation remained below baseline, in line with the significant reductions in CaO2. Furthermore, the largest decrease in cortical voluntary activation occurred in parallel with the lowest cerebral oxygenation at task failure. This latter finding, in combination with the attenuated peripheral fatigue in severe hypoxia, suggests that a significant reduction in O2 transport may have a direct influence on CNS activity. In vitro studies show clearly that hypoxia affects neuronal function (37), and studies in humans have linked cerebral deoxygenation in acute hypoxia with increased activity of the prefrontal regions of the cortex (45), regions that have been suggested to affect processes involved in supraspinal modulation of muscular performance (44). Collectively, these findings suggest that at least part of the impairment in central drive in severe hypoxia was mediated by cerebral deoxygenation.
To produce a state of hypoxia encountered in the field or clinical setting, no effort was made to control the partial pressure of arterial CO2, despite its known influence on the regulation of cerebral blood flow (38). Compared with all other conditions, we found that severe hypoxia yielded a significant level of hyperventilation-induced hypocapnia, as reflected by the lower PetCO2 during rest (33 Torr) and exercise (25 Torr). Hypocapnia causes cerebral vasoconstriction (27) and attenuates the normal increase in cerebral blood flow in hypoxia (12). Severe hypocapnia (PetCO2 ≤15 Torr) increases the excitability of the motor cortex (46, 51). In the present study, however, cortical excitability was unaffected by the differing levels of hypocapnia; this finding is in agreement with what has been reported previously for similar levels of hypocapnia (PetCO2 ∼22 Torr) (32). Thus we believe it unlikely that the levels of hypocapnia observed in the present study had a significant influence on the exercise-induced changes in CNS function.
In conclusion, we have confirmed that acute hypoxia has no effect on either neuromuscular properties or cortical function during resting conditions. In addition, we have shown for the first time that peripheral mechanisms of fatigue contribute relatively more to the reduction in force-generating capacity of the knee extensors after submaximal intermittent isometric contractions in normoxia and acute mild-to-moderate hypoxia (SpO2 ≥ 85%), whereas supraspinal fatigue plays a greater role in the force reduction in acute severe hypoxia (SpO2 < 80%). We argue that this transition from a predominantly peripheral origin of fatigue to an O2-sensitive source of inhibition of central motor drive within the CNS is mediated, at least in part, by cerebral hypoxia. The findings could have important implications for understanding the processes that determine exercise intolerance in people who live at, or ascend to, high altitude, as well as patients who are challenged by reductions in convective O2 transport.
No conflicts of interest, financial or otherwise, are declared by the author(s).
We thank Daniel Phillips for assistance during data collection and analysis. Mr. Phillips was sponsored by a Vacation Studentship from The Physiological Society.
- Copyright © 2010 the American Physiological Society