The purpose of this study was to quantify how shivering activity would be affected by large changes in fuel metabolism (see Haman F, Peronnet F, Kenny GP, Doucet E, Massicotte D, Lavoie C, and Weber J-M, J Appl Physiol 96: 000–000, 2004). Adult men were exposed to 10°C for 2 h after a low-carbohydrate diet and exercise (Lo) and after high-carbohydrate diet without exercise (Hi). Using simultaneous metabolic and electromyographic (EMG) measurements, we quantified the effects of changes in fuel selection on the shivering activity of eight large muscles representing >90% of total shivering muscle mass. Contrary to expectation, drastic changes in fuel metabolism [carbohydrates 28 vs. 65% of total heat production (Ḣprod), lipids 53 vs. 23% Ḣprod, and proteins 19 vs. 12% Ḣprod for Lo and Hi, respectively] are achieved without altering the EMG signature of shivering muscles. Results show that total shivering activity and the specific contribution of each muscle to total shivering activity are not affected by large changes in fuel selection. In addition, we found that changes in burst shivering rate (∼4 bursts/min), relative contribution of burst activity to total shivering (∼10% of total shivering activity), and burst shivering intensity (∼12% of maximal voluntary contraction) are the same between Lo and Hi. Spectral analysis of EMG signals also reveals that mean frequencies of the power spectrum remained the same under all conditions (whole body average of 78 ± 5 Hz for Lo and 83 ± 7 Hz for Hi). During low-intensity shivering, humans are therefore able to sustain the same thermogenic rate by oxidizing widely different fuel mixtures within the same muscle fibers.
- energy metabolism
- shivering thermogenesis
- shivering pattern
- muscle fiber recruitment
- glycogen reserves
shivering is an involuntary process generating heat through rhythmic, asynchronous muscle contractions. To sustain shivering thermogenesis over prolonged periods of cold exposure, muscle recruitment and fuel metabolism must be tightly coordinated. Previous research in this field falls into two broad categories dealing either with muscle metabolism (8, 13) or with electrophysiological aspects of muscle recruitment (2, 22, 30). These two complementary perspectives on the same problem have not been traditionally integrated, and this study is a first attempt at doing so. In the first part (9), we have shown that the size of glycogen reserves has a considerable effect on fuel selection. When carbohydrate (CHO) reserves were low (Lo), total heat production (Ḣprod) was partitioned between CHO (28% Ḣprod), lipids (53% Ḣprod), and proteins (19% Ḣprod), but the fuel selection pattern was strikingly different when CHO reserves were high (Hi): CHO (65% Ḣprod), lipids (23% Ḣprod), and proteins (12% Ḣprod). In particular, intramuscular glycogen was used at very different rates for Lo and Hi (168 ± 43 vs. 380 ± 58 mg/min), but the effects of such large metabolic changes on shivering activity and muscle fiber recruitment have never been addressed. Fuel selection of contracting muscles can be modified in two ways: 1) mobilizing different metabolic pathways within the same fibers or 2) recruiting distinct fiber populations specialized for different fuels.
Over the last several decades, mechanisms of motor unit (MU) recruitment during voluntary contractions have received a lot of attention (see Refs. 6, 19, 33 for review), but very few studies have investigated this process during shivering (20–22, 26). Previous work shows that all fiber types can be involved in shivering (13, 22). For example, Jacobs et al. (13) reported decreased glycogen concentration in all fiber types of the vastus lateralis after cold exposure. In addition, electromyography (EMG) recordings reveal two distinct patterns of shivering: 1) thermoregulatory muscle tone, or continuous, low-intensity shivering (at 4–8 Hz) and 2) bursts of high-intensity shivering occurring at much lower frequencies (0.1–0.2 Hz or 8–16 times/min) (12, 22). These two patterns are associated with the recruitment of specific MU (20, 22, 26). Whereas continuous, low-intensity shivering is linked to low-threshold MU (type I, slow-oxidative, fatigue-resistant fibers), shivering bursts are associated with high-threshold MU (type II, fast-glycolytic, more fatigable fibers). Human type II fibers (IIa and IIx) show lower activities for oxidative enzymes than type I fibers (-60%) and much higher activities for glycolytic enzymes and creatine kinase (+300–400%) (6). Because of these large biochemical differences, the two shivering patterns observed may reflect the use of distinct metabolic substrates, with type I being mostly geared toward lipid use and type II toward CHO use. Together, these observations suggest that fast-glycolytic type II fibers, and therefore burst shivering activity, may be significantly affected by changes in glycogen reserves.
The first objective of this paper was to quantify the effects of changes in glycogen stores on whole body shivering activity. More specifically, we monitored the recruitment levels of eight large muscles representing >90% of total shivering muscle mass in men with low- or high-CHO reserves during sustained cold exposure. The second objective was to characterize the detailed shivering pattern and spectral parameters of individual muscles in an attempt to uncover more subtle effects of CHO availability on the shivering response. The chosen experimental design allowed us to investigate whether changes in fuel selection were achieved by recruiting “fuel-specific fibers,” as previously suggested (27). According to this idea, the recruitment of fast-glycolytic MU would be directly correlated with CHO utilization. At the whole organism level, we predicted that total shivering activity would not be affected by changes in CHO availability, although the relative contributions of individual muscles could be altered to maintain heat production. Within each muscle, we anticipated that the shivering pattern and EMG spectral parameters would change with CHO availability because distinct fuel-specific fiber populations would be recruited.
Subjects. Six healthy and physically active men participated in this study conducted with the approval of the Health Sciences Ethics Committee of the University of Ottawa. Description of volunteers was given in part I of this study (9).
Experimental procedures. Five to 6 days before the experiments, a 1-h session was held to familiarize the subjects with the equipment and the level of cold exposure faced in the experiments. For the actual experiments, subjects were exposed to the cold on two separate occasions after following 1) a diet low in CHO and heavy exercise bouts (Lo; last glycogen-depletion exercises were conducted at least 20 h before the experiment), and 2) a diet high in CHO without exercise bouts (Hi; no exercise was performed at least 2.5 days before the experiment). A detailed description of the diet and exercise regimen has been given in Haman et al. (9). On their arrival in the laboratory (8:00 AM; 12 h postabsorptive), subjects were instrumented with thermal probes (esophageal and skin thermocouples, heat-flux transducers), EMG electrodes, and an indwelling catheter (18 G, 32 mm, Medical, Arlington, TX). They were then fitted with a liquid-conditioned suit (three piece Delta Temax, Pembroke, ON). After voiding the bladder (time T = 0 min), subjects remained seated comfortably for the next 2 h at 23.2 ± 0.01°C (758 ± 2 mmHg, 39.8 ± 3.6% relative humidity). After this period, they were transferred to an environmental chamber (10.5 ± 0.01°C, 755 ± 3 mmHg, 61 ± 2% relative humidity), and a 10°C water perfusion was started through the liquid-conditioned suit by using a temperature-controlled circulation bath (Endocal, NESLAB and model 200-00, Micropump, Vancouver, WA). Average environmental conditions in laboratory and thermal chamber were the same for Lo and Hi experiments. Thermal response was monitored continuously at 23°C and during the subsequent 2-h cold exposure by use of a Hewlett-Packard data-acquisition and control unit (model 3497A). Metabolic and thermal comfort data were collected periodically throughout the experimental session.
Shivering EMG signals were recorded from eight muscles located on the right side of the body: trapezius (TR), latissimus dorsi (LA), pectoralis major (PE), rectus abdominis (RA), vastus lateralis (VL), rectus femoris (RF), vastus medialis (VM), and gastrocnemius (GA). These muscles were selected to represent the largest possible fraction of total muscle mass involved in shivering (>90%) based on a previous study by Bell et al. (2). Surface electrodes (Blue Sensor, Medicotest) were positioned 2 cm apart over the bellies of each muscle, and their exact positions were identified with an indelible skin marker to allow consistent placement between measurement sessions [i.e., cold-exposure experiments and maximal voluntary contraction (MVC) measurements]. Surface electrodes were connected by using preamplified and grounded EMG wires (375X) to a ME-3000 Professional EMG system (Mega Electronics, Kuopio, Finland). Raw EMG signals were collected at 1,000 Hz and downloaded via fiber optic to a desktop computer. Shivering activity of the eight individual muscles was monitored 30 min before and six times during cold exposures of 5–20 min, 25–40 min, 45–60 min, 65–80 min, 85–100 min, and 105–120 min (see Fig. 1A). Intervals between sampling periods (5 min) were used to save the recorded data to the computer's hard drive. Voluntary muscle activity was minimized as much as possible throughout cold exposure by asking subjects to avoid voluntary movements during recording periods.
Normalization of EMG amplitude. Three to 4 days before the experiments, maximal EMG signals of each muscle (RMSmvc) were determined from maximal isometric contractions (MVC) performed with the use of a muscle testing and training system (KIN COM 500H, Chattecx, Chattanooga, TN). MVC protocols comprised a ramp-and-hold protocol that involved gradually increasing force from baseline to maximum over 2–3 s and then holding the maximum force for an additional 2–3 s (15). Participants were verbally encouraged to achieve maximum force during the three trials and were given at least 30 s rest between each MVC measurement. The following procedures were used to determine RMSmvc of individual muscle: 1) For TR, MVC was measured while subjects were standing up grasping the load cell with the right hand and arm in full extension. Subjects were then asked to raise the shoulder with maximal force while keeping the arm as close to the body as possible. 2) For PE and LA, subjects were asked to sit upright in a secured chair and measurements were performed with the arm extended away from the body at a 90° angle. For PE, the load cell was placed on the front, slightly over the elbow, and subjects were asked to push maximally forward against it. For LA, the load cell was positioned under the elbow and subjects pushed downward against it with maximal force. 3) For RA, while subjects sat upright in a secured chair, the load cell was attached to a strap located around the subjects' chests. Subjects then curled forward using their abdominal muscle while keeping their backs straight and applying maximal force against the strap. 4) For VL, RF, and VM, MVC measurements were performed with subjects sitting securely in an upright position and the load cell was attached slightly above the ankle. The load cell arm was placed at a 45° angle from its vertical position, and subjects performed a maximal knee extension pushing against the load cell. 5) For GA, MVCs were measured with subjects lying down with the load cell attached on the front of the foot and the leg secured to the table with straps, while subjects performed a maximal ankle flexion against the load cell.
Determination of shivering intensity and pattern. Raw EMG signals were analyzed with the use of custom-designed MATLAB algorithms (Mathworks, Natick, MA). EMG signals were filtered to remove spectral components below 20 Hz and above 500 Hz, as well as 60-Hz contamination (and associated harmonics).
Shivering intensity of individual muscles (, where the index m identifies the muscle) was determined from root-mean-square values (RMS) calculated from EMG signals using a 50-ms overlapping window (50%). Baseline RMS values (RMSbaseline: 15 min RMS average measured before cold exposure) were subtracted from RMS shivering (RMSshiv) as well as RMSmvc values. was normalized to RMSmvc by using the following equation 1 In addition, shivering activity of individual muscles () was determined from the area under the curve by using the Riemann sum 2 where the summation is taken over all samples and change in time (Δt) = 1 ms. The relative contribution of muscle m to shivering activity was determined by dividing by the sum of shivering activities for all muscles.
and were subsequently quantified with 11 outcome variables corresponding to changes in overall shivering, continuous low-intensity shivering, and burst shivering. The 11 outcome variables (determined for the three latter EMG components unless indicated otherwise) are 1) total shivering time, total amount of time for which was >0% MVC (i.e., above RMSbaseline, Eq. 1) over the entire recording time (∼90 min); 2) summed area (in %MVC·min); 3) mean intensity (in %MVC); 4) percent shivering time (in min), summed area of individual EMG components as a percentage of total recording time (%RT); 5) steady-state intensity (in %MVC), average overall intensity reached in the last 15 min of cold exposure; 6) summed duration (in min), total duration continuous low-intensity shivering and burst shivering; 7) number of bursts, total number of bursts during the recording period; 8) mean burst duration (in s), mean duration of all bursts; 9) burst rate (in bursts/min), mean number of bursts per minute; 10) percent contribution to total whole body shivering (%Shivtotal), relative contribution of continuous low-intensity shivering and burst shivering to total shivering activity calculated based on whole body index (see below); and 11) percentage of burst shivering activity to continuous low-intensity shivering activity, percentage of burst summed area (see variable 1) to total continuous low-intensity shivering area (%EMGburst/EMGcont). A shivering burst was arbitrarily defined in the present study as an EMG interval with a duration >0.2 s, an interburst interval >0.75 s, and an amplitude higher than the intensity threshold at each recording period (5–20, 25–40, 45–60, 65–80, 85–100, or 105–120 min). Intensity threshold was determined by first calculating the average shivering intensity () over the entire recording period. Remaining values above ĀEMG were then averaged again (B̄EMG), and the intensity threshold was set at B̄EMG (see Fig. 5).
To obtain a whole body index of shivering activity (ShivWBI) that can be related to Ḣprod in the cold (measured by indirect calorimetry), shivering activities of all muscles were summed, taking into account the relative mass of the body region they represent (2, 30) 3 where , , , and are upper trunk (UT; average of TR, LA, and PE), lower trunk (LT; RA), upper leg (UL; average of VL, RF, and VM), and lower leg (LL; GA) shivering activities, and t is time. The coefficients fUT (0.34), fLT (0.19), fUL (0.29), and fLL (0.085) correspond to the relative muscle masses of each body region (UT, LT, UL, and LL) to total muscle mass (2, 30). This whole body index represents ∼91% of total muscle mass and excludes upper limbs, head, and feet, which contribute minimally to Ḣprod (2). Following the approach proposed by Bell et al. (2) and Tikuisis et al. (30), we have verified the linearity of the EMG vs. force relationship by performing submaximal contractions at 10, 25, 50 and 75% MVC using a KIN COM 500H. This relationship was linear for all muscles selected in this study (r2 = 0.85–0.98). Therefore, the calibration factor developed by these authors (2) to account for nonlinearity under 20% MVC was not used.
EMG spectral analysis. Mean frequency of the EMG power spectrum (MPF) as a function of time was estimated by applying a short-time Fourier transform to the raw data. A Hanning window consisting of 4,096 samples (or 4 s wide for a spectral resolution of 0.25 Hz) was applied with a 50% overlap, thereby producing a spectral estimate every 2 s. Consistent with the computations for the intensity, the spectral components below 20 Hz and above 500 Hz were removed. The raw data were also slightly contaminated with 60-Hz noise; the corresponding samples and associated harmonics were removed (set to zero) from the discrete spectrum. A MPF was then obtained by taking the mean of each spectral estimate. Finally, MPF data were smoothed by using a running average with a seven-sample-wide (14-s) window.
Statistical analyses. Changes in shivering intensity and MPF as a function of time for individual muscles as well as for UT, LT, UL, and LL were assessed by two-way analysis of variance with replication. Differences in MPF (UT, LT, UL, LL) between Lo and Hi were determined by using a paired t-test. Statistical differences were considered significant when P ≤ 0.05. All values presented are means ±SE (n = 6), unless indicated otherwise.
Changes in heat loss, Ḣprod, body temperature, and mean skin temperature before and during cold exposure for Lo and Hi have been reported in Haman et al. (9). A total of 90 min of EMG signal was collected from eight large muscles corresponding to four regions of the body (UT, LT, UL, and LL) and representing >80% of total muscles mass involved in shivering (2). An example of raw EMG signal (Fig. 1A) and calculated shivering intensities (Fig. 1B) and mean power frequencies (Fig. 1C) are shown in Fig. 1.
Whole body shivering. Respective contributions of CHO (%CHOox), lipid, and protein oxidation to Ḣprod under Lo and Hi conditions have been presented in detail in part I of this study (9). In the last 30 min of cold exposure, CHO, lipids, and proteins contributed, respectively, 28 ± 5, 53 ± 5, and 19 ± 2% Ḣprod for Lo and 65 ± 5, 23 ± 5, and 12 ± 5% Ḣprod for Hi. Changes in %CHOox, Ḣprod, and ShivWBI during cold exposure are shown in Fig. 2. Whereas %CHOox was 114% higher for Hi than for Lo (71.4 ± 5.3 vs. 33.4 ± 5.1% Ḣprod) (Fig. 2A), Ḣprod (Fig. 2B) and ShivWBI (Fig. 2C) were not different between Lo and Hi. Relationships between changes in ShivWBI and Ḣprod for Lo and Hi are presented in Fig. 3 and were not different between treatments. Ḣprod increased 2.2-fold from control values before cold exposure, reaching a maximum of 210.0 ± 10.4 W for Lo and 186.9 ± 11.0 W for Hi by the end of cold exposure. ShivWBI also increased continuously from an average area of 75 ± 17 and 254 ± 38%MVC·s in the first sampling interval (t = 5–20 min) to 112 ± 40 and 214 ± 47%MVC/s by the end of cold exposure (t = 105–120 min) for Lo and Hi, respectively.
Changes in shivering intensity for each muscle (%MVC) in Lo and Hi are presented in Fig. 4. Shivering intensity increased continuously during cold exposure, but no difference was observed between Lo and Hi. Maximal intensities reached by the end of cold exposure averaged 3.5 ± 0.9, 6.3 ± 1.1, 5.4 ± 1.5, 3.8 ± 1.3, 4.0 ± 1.6, 3.8 ± 0.7, 4.1 ± 1.6, and 4.2 ± 1.4%MVC for Lo and 3.2 ± 0.8, 5.8 ± 1.0, 3.8 ± 0.6, 3.5 ± 0.7, 4.2 ± 1.7, 5.6 ± 2.0, 3.6 ± 1.6, and 8.7 ± 2.4%MVC for Hi (for TR, LA, PE, RA, VL, RF, VM, and GA, respectively). The relative contributions of individual muscles to total shivering activity were also found to be the same between Lo and Hi. Shivering intensity frequency distributions were calculated for each muscle and each subject separately, over the entire period of cold exposure. Although minor differences between Lo and Hi were detected for individual muscles in some subjects, no significant differences between mean values could be established.
Shivering pattern and EMG spectral characteristics. The method used to distinguish burst shivering from continuous, low-intensity shivering is schematized in Fig. 5. Separating these two patterns was achieved on the basis of large differences in intensity (2–5 vs. 7–15% MVC) and rate of occurrence (8–10 vs. 0.1–0.2 Hz). Table 1 summarizes all parameters calculated for burst shivering and continuous, low-intensity shivering for UT, LT, UL, and LL muscles. Results show that the two patterns were the same between Lo and Hi for all body regions. The effect of changes in fuel selection on burst shivering activity is emphasized in Fig. 6. Burst shivering intensity, burst shivering rate, and relative contribution of burst shivering to total recording time were not different between Lo and Hi. Burst shivering intensity increased from 7.0 ± 0.8%MVC in the first recording interval to 12.1 ± 1.1%MVC by the end of cold exposure for Lo and from 7.0 ± 1.7 to 11.8 ± 2.4%MVC for Hi. Burst shivering rate increased from 3.1 ± 0.5 to 4.6 ± 0.5 bursts/min for Lo and from 3.0 ± 0.5 to 4.5 ± 0.5 bursts/min for Hi, and the relative contribution of burst shivering to total recording time increased from 7.0 ± 1.6 to 10.7 ± 0.8%RT for Lo and from 7.0 ± 1.4 to 11.2 ± 0.6%RT for Hi.
Changes in MPF for each muscle for Lo and Hi are presented in Fig. 7. MPF remained constant in cold exposure averaging and was not different between Lo and Hi. As for shivering intensities, frequency distributions of MPF were calculated for each muscle and each subject separately, over the entire period of cold exposure. Minor differences between Lo and Hi were detected for individual muscles in some subjects, but no significant differences between mean values could be established.
Figure 8 summarizes the effect of a very large change in fuel selection on changes in spectral characteristics of the EMG signal (MPF). Averages were calculated over the last 15 min of cold exposure. Although the relative contributions of CHO, lipids, and proteins for Lo were -58, +132, and +60% higher, respectively, for Lo than for Hi, no difference in MPF for UT, LT, UL, and LL was found between Lo and Hi. The effect of a very large change in CHOox on mean whole body MPF is emphasized in Fig. 8C. Although there was a 2.4-fold difference in the relative contribution of CHOox to Ḣprod between Lo (27.7 ± 5.2% Ḣprod) and Hi (65.3 ± 5.4% Ḣprod), MPF was identical for both groups (78.4 ± 5.2 vs. 82.5 ± 6.9 Hz).
Contrary to expectation, this study shows that drastic changes in oxidative fuel selection are achieved without modifying the EMG signature of shivering muscles. Results support the notion that the same population of muscle fibers is responsible for oxidizing widely different fuel mixtures, but they are not compatible with the hypothesis that changing fuel selection is realized by recruiting fuel-specific MU. EMG signal analyses were performed at two levels to assess potential differences in 1) whole body shivering activity (monitoring >90% of total shivering muscle mass) based on the recruitment intensity of eight large muscles, and 2) detailed EMG activity of individual muscles based on their shivering pattern and spectral characteristics.
Whole body shivering. Total shivering activity (ShivWBI, Fig. 2) and the respective contributions of specific muscles to ShivWBI were not affected by large changes in fuel selection. Consequently, the contribution of each muscle to Ḣprod was the same for Lo and Hi. The normalized EMG amplitudes found here (%MVC) were small and consistent with the only two other studies reporting this parameter for low-intensity shivering (3–10%) (2, 30). Interindividual and intertrial differences in the shivering response are mainly responsible for the high variability observed at each time point in this and previous studies. Because such temporal variability could mask subtle differences between treatments, we have eliminated temporal effects by calculating frequency distributions of shivering intensities for each muscle and each subject separately over the entire period of cold exposure. Although minor differences between Lo and Hi were detected for individual muscles in some subjects, no significant differences between mean values could be established. This additional analysis strengthens the conclusion that shivering intensity is not modified by changes in fuel selection.
Shivering activity (EMG) and heat production (indirect calorimetry) are two different estimates of thermogenesis, and their divergence over time could be used to detect possible contributions from nonshivering thermogenesis (brown adipose tissue, substrate cycles) (25). In this study, changes in shivering activity and heat production were closely correlated for Lo and Hi (Fig. 3), and the same observation was previously made for individuals with normal CHO reserves (2). These results were somewhat expected because nonshivering thermogenesis is generally assumed to be negligible in adult humans (11). However, the tight correlation between shivering activity and heat production suggests that nonshivering thermogenesis probably remains unimportant for thermoregulation, even when fuel selection is drastically modified.
Shivering pattern and EMG spectral characteristics. A detailed characterization of shivering pattern and spectral parameters was performed for each muscle to try detecting subtle effects of changes in fuel selection. The EMG signal was analyzed by separating the patterns for “burst shivering” and “continuous, low-intensity shivering” (Table 1). Distinguishing these two patterns was based on large differences in intensity (2–5 vs. 7–15% MVC) and rate of occurrence (8–10 vs. 0.1–0.2 Hz) (Fig. 5). Results clearly show that both shivering patterns remained unaffected by large changes in fuel selection. In addition, we found that burst shivering rate, relative contribution of burst activity to total shivering, and burst shivering intensity were the same between Lo and Hi (Fig. 6). Previous studies have associated continuous low-intensity shivering with the recruitment of slow-oxidative MU (type 1) and burst shivering with fast-glycolytic MU (type 2) (20, 22, 26). The lack of difference in shivering pattern between Lo and Hi suggests that MU recruitment is the same under both conditions, and, more importantly, that the recruitment of fast-glycolytic fibers is not affected by the size of CHO reserves. Very little is known about the relative importance of continuous, low-intensity shivering and of burst shivering to total heat generation. How they are partitioned may have important implications on cold endurance, and additional research is needed to determine the physiological significance of this dual pattern.
The second part of our detailed analysis dealt with spectral parameters of the EMG signal. Over the last few decades, changes in EMG frequency spectra have been used extensively to investigate MU recruitment and muscle fatigue (1, 3–5, 7, 16, 17, 23, 32, 33). Using surface EMG, these studies have shown that the recruitment of fast-glycolytic fibers results in higher myoelectric frequencies than slow-oxidative fibers (33). During shivering, however, information on changes in spectral parameters is scarce, and it was only obtained from a limited number of muscles (2, 24). Here, time-frequency analysis was used to quantify potential changes in the spectral signature of eight muscles (Figs. 7 and 8). MPF remained the same under all conditions, showing that the frequency spectrum was unaffected by the size of CHO stores. Furthermore, the frequency distributions of MPF over the entire cold exposure period were identical for Lo and Hi, as we had found for shivering intensity. Therefore, detailed spectral analysis of EMG signals also shows that the pattern of MU recruitment is not modified.
During sustained isometric contractions, the onset of fatigue is known to be associated with a decrease in mean or median frequency of the surface EMG power spectrum (3). However, the MPF of all the muscles selected in our study remained constant throughout cold exposure, suggesting that no fatigue occurs. Using a similar approach, two other studies have investigated whether prolonged shivering could cause fatigue. Although no downward shift in the EMG power spectrum was found for masseter, rectus abdominis, biceps brachii, brachioradialis, rectus femoris, and gastrocnemius, a decrease in the median frequency of pectoralis major was reported (2, 24). It is worth noting here that a decrease in MPF cannot be unequivocally interpreted as muscle fatigue, because a clear link between these two parameters has not been established for shivering. For example, a shift in MPF can occur when changes in recruitment take place in the absence of fatigue (33). Although some authors have provided evidence supporting the idea that prolonged shivering can lead to fatigue (31), further research is needed to clarify this issue.
It could be argued that the surface EMG method used here provides information on the recruitment of superficial MU and may not detect changes taking place in deeper regions. However, such a scenario is very unlikely because the surface EMG signal provides action potential information from a large number of MU (17), and all fiber types are represented in superfi-cial regions (28). For the muscles in which superficial-to-deep gradients in fiber composition have been reported (vastus lateralis, tibialis anterior), more fast-glycolytic fibers were found in the periphery (10, 18). In our context, such a gradient would improve the ability to detect potential changes in the recruitment of fast-glycolytic fibers. In addition, to minimize the effects of variability in fiber composition between individuals and between locations along the muscles, the same subjects and the same EMG sampling sites were used for Lo and Hi. In the absence of measurable changes in shivering intensity (Figs. 2, 3, 4, Table 1), shivering pattern (Table 1, Fig. 6), and EMG spectral parameters (Figs. 7 and 8), failing to detect differences in MU recruitment between Lo and Hi is, therefore, highly improbable.
Fuel selection within the same muscle fibers. We had anticipated that changing fuel selection would be achieved by recruiting different populations of muscle fibers. However, this hypothesis can be rejected for sustained shivering because no change in fiber recruitment was observed between Lo and Hi. It appears that most of the heat is generated within type I fibers, because burst shivering represents only <10% of total shivering activity (Table 1). What are the mechanisms responsible for regulating such a large change in fuel mixture within these same fibers? Fuel selection has not been investigated in shivering muscle, but a lot of information is available on the reciprocal regulation of CHO and lipid metabolism during exercise (14, 29). A combination of many factors is responsible for controlling fuel oxidation. They include fuel availability (size of glycogen reserves), circulating hormones (e.g., catecholamines), transmembrane transporters (GLUT and fatty acid transporters), and a large series of intracellular metabolites (acetyl-CoA, malonyl-CoA, Ca2+, ADP, AMP, Pi, and AMP-activated protein kinase 1). The relative importance of these factors is far from being understood for exercise, and fuel selection may be regulated differently during shivering.
Conclusion. Together with the companion paper (9), this study provides the first integrated analysis of fuel metabolism and muscle fiber recruitment during shivering. It shows that shivering intensity, shivering pattern, and EMG spectral parameters are not modified by extreme changes in oxidative fuel selection. During low-intensity shivering, humans are able to sustain thermogenic rate by oxidizing widely different fuel mixtures within the same muscle fibers. Whether this selection strategy is also used during high-intensity shivering, rather than recruiting different fuel-specific fibers, remains to be established.
We thank M. Lamontagne (Human Kinetics, University of Ottawa), P. F. Gardiner (Health, Leisure and Human Performance Research Institute, University of Manitoba), and D. Parry (Cellular and Molecular Medicine, University of Ottawa) for excellent feedback.
This project was funded through a Natural Sciences and Engineering Research Council of Canada (NSERC) grant to J.-M. Weber. F. Haman was the recipient of a NSERC scholarship.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2004 the American Physiological Society