Functional MRI (fMRI) studies have shown that the blood oxygenation level-dependent (BOLD) contrast mechanism can be sensitive to small changes in oxygenation state and blood volume (1). While this method is now widely used to study functionality in the brain, and is beginning to be used in tumors (2–6), only a few studies have focused on fMRI of skeletal muscle (7–9). Measuring the dynamics of blood flow and oxygenation in skeletal muscle during exercise could be important for elucidating aspects of muscle physiology and pathology.
In the last few years, the feasibility of observing time-resolved BOLD effects in exercising human muscles has been demonstrated in several studies (10–12). These studies showed that transient increases in the signal intensity (SI) in skeletal muscle occur after single, short-duration contractions, and that this effect is prolonged for several tens of seconds (as opposed to the on/off kinetics in the brain). This effect coincided with the time course of transient increases in muscle hemoglobin saturation, as shown by near infrared spectroscopy (11), and was further demonstrated to be field-dependent (12). These observations suggest that the transients are due to a positive BOLD effect.
Other investigators have studied the relaxation behavior of skeletal muscle in vivo using T2 relaxation times. It is now well established that the T2 of skeletal muscle water increases during exercise (13, 14). Although the underlying mechanism is not fully understood, it has been used to map the location and relative intensity of muscle recruitment during various motor tasks (15, 16). The assumed mechanism is that the T2 increase is related to osmotically driven shifts of fluid between intra- and extracellular compartments (13, 17).
The usefulness of fMRI in muscle will depend on our ability to identify the factors that contribute to the BOLD signal fluctuations. This issue was first investigated in the brain (18), and later in tumors (4, 6, 19). The aim of the present study was to characterize the parameters that can contribute to the BOLD contrast in exercising mouse skeletal muscle. For this purpose, BOLD imaging was performed during and after application of an electrical stimulation protocol to the sciatic nerve. In addition, hemodynamic parameters (such as partial oxygen pressure (pO2) and blood flow) were monitored. We also investigated whether the T2 parameter of skeletal muscle could be modified by this exercise protocol. To discriminate between these factors, we compared control mice with mice lacking endothelial nitric oxide synthase (eNOS−/−). Indeed, NO is able to inhibit oxygen consumption in muscle, and eNOS−/− mice may have a different sensitivity to the exercise protocol in terms of oxygenation (20, 21).
MATERIALS AND METHODS
Male C57/BL6J wild-type (WT) (Elevage Janvier, Le Genest-St-Isle, France) and age-matched eNOS knockout mice (eNOS−/−) were studied. C57BL/6J eNOS−/− mice were originally from the Jackson Laboratory (JAX® GEMM® Strain) and were inbred at the FATH Laboratory (UCL, Brussels, Belgium). Each procedure was approved by the local authorities according to national animal-care regulations.
Anesthesia was induced with an i.p. injection of ketamine (80 mg/kg)/xylazine (8 mg/kg), and maintained with ketamine alone (30 mg/kg). The body temperature of the mice was maintained at 37°C by warm air blown into the magnet for the MR experiments, or by a homeothermic blanket for the flow and oxygenation measurements. Platinum electrodes were placed around the right sciatic nerve. A 15-min exercise protocol was achieved by electrical stimulation at 5 Hz, with 0.2-ms duration pulses (22). The pulse voltage was 0.5 V in the moderate exercise protocol generally applied in this study, and ∼3 V in a more intense protocol used in some individual cases (see Discussion).
MRI acquisition was performed with a 4.7 Tesla (200 MHz, 1H) bore system (40-cm inner diameter; Bruker Biospec, Ettlingen, Germany). A birdcage radiofrequency (RF) coil with an inner diameter of 70 mm was used. For both MR techniques, a preliminary anatomical T2 image (rapid acquisition with relaxation enhancement (RARE) sequence: TR = 1500 ms, TE = 11.5 ms, six averages, echo train length = 8, slice thickness = 2 mm, and acquisition time = 2 min 32 s) was acquired to define a first region of interest (ROI) encompassing the muscle of the stimulated leg (about 40 pixels), and a second ROI encompassing the nonstimulated muscle (control).
BOLD-sensitive images were acquired with the use of a regular gradient-echo (GE) pulse sequence. The raw data acquisition parameters were TR = 200 ms, TE = 15 ms, α = 45 °, FOV = 6 cm, matrix = 128 × 128, six averages, and slice thickness = 2 mm. The total data acquisition time was 2 min 33 s. We first acquired five BOLD images to establish the baseline, and then acquired five images during the 15-min exercise protocol. The acquisitions were continued every 3 min for 30 min following stimulation. To avoid effects from external sources of variation, we normalized the relative change in BOLD SI in the stimulated leg with the values from the control leg.
Multi-echo T2 images were also acquired to follow the evolution of the T2 parameter in stimulated and nonstimulated muscles. The imaging parameters were TR = 3000 ms, TE = 6.65 ms, matrix size = 64 × 64, FOV = 6 cm, one average, slice thickness = 2 mm, 30 echoes, and total acquisition time = 6 min 49 s. Acquisitions were performed every 7 min before, during, and after the exercise protocol was applied. We analyzed the T2 decay using a biexponential fit in order to determine the fast (T2) and short (T) T2 components of the skeletal muscle. An arbitrary noise threshold was fixed at three times the SNR, below which the values were not taken into account in the fitting process.
pO2 and Flow Measurements
Local muscle oxygenation measurements were carried out with the use of two independent techniques: electron paramagnetic resonance (EPR) oximetry, and fiber-optic probes (OxyLite™). We used OxyFlo™ probes to assess the blood flow inside the muscle.
EPR spectra were recorded with an EPR spectrometer (Magnettech, Germany) with a low-frequency microwave bridge operating at 1.2 GHz, and an extended loop resonator. Charcoal (CX0670-1; EM Science, Gibbstown, NJ) was used as the oxygen-sensitive probe in all of the experiments. We obtained the calibration curves by measuring the EPR line width as a function of pO2. The charcoal suspension (100 mg/ml, 50 μl injected, 1–25 μ particle size) was injected into the mice in the center of the gastrocnemius muscle. The EPR measurements were initiated 2 days after the injection. We placed the muscle under study in the center of the extended loop resonator (the sensitive volume of which extended 1 cm into the muscle mass) using a previously described protocol (23, 24).
We used Oxylite™ in conjunction with Oxyflo™ (Oxford Optronix, Oxford, UK) to simultaneously and continuously monitor tissue blood flow, oxygenation, and temperature at the same location. Fiber-optic microprobes combining a laser Doppler system, an oxygen-sensor, and a thermocouple were inserted into the muscle (25, 26). We inserted the probes with a 26G needle in order to make a track inside the tissue. We performed back-scatter measurements to validate that movement artifacts did not influence the flow measurements. A baseline of 15 min of stable readings was obtained before the stimulation protocol was initiated. Data were collected continuously at a sampling frequency of 20 Hz before, during, and after electrical stimulation.
Data are reported as mean ± SEM. We used a Student's t-test or analysis of variance (ANOVA) with a Tukey's multiple-comparison post-hoc test when appropriate.
The relative BOLD SI in mouse muscle increased directly when the stimulation protocol began. During the stimulation period, the average SI showed a mean (±SEM) increase of 6.4% ± 0.7% (N = 4, P < 0.01) for control C57BL6 mice, and 3.3% ± 0.4% (N = 4, P < 0.01) for eNOS knockout mice (Fig. 1a). While the SI remained high in the post-stimulation period for control mice (4.9% ± 0.6%, P < 0.01), it slowly decreased in eNOS−/− mice (1.4% ± 0.6%, P > .05) and returned to the basal level after 15 min (Fig. 1a).
We used EPR oximetry, an in vivo non-oxygen-consuming technique, to monitor fluctuations in muscle tissue oxygenation. Muscle oxygenation increased during the exercise protocol (Fig. 1b). The mean increases in pO2 were 41.8% ± 3.9% (N = 3, P < 0.01) for C57BL6 mice, and 42.2% ± 2.1% (N = 4, P < 0.01) for eNOS−/− mice. This effect was prolonged during the post-stimulation period in WT mice; however, in mice lacking eNOS, muscle oxygenation significantly decreased after the end of the exercise protocol (down to 20.0% ± 2.9% of the increase). These results were confirmed by OxyLite™ probes (see typical result in Fig. 2a). OxyLite measurements were acquired simultaneously with Doppler measurements (see below), and hence could be compared with changes in blood flow.
Relative changes in muscle blood flow were monitored by laser Doppler probes (OxyFlo™), continuously before, during, and after the stimulation protocol. The OxyFlo™ technique allows relative measurements of blood flow in arbitrary units (blood perfusion unit) to be obtained. Both groups presented a similar pattern, namely an immediate increase (an approximately twofold increase) in muscle blood flow at the onset of the exercise protocol, and a relatively fast decrease after electrical stimulation stopped. The level of increase in the blood flow was quite variable from one mouse to another. The mean changes in muscle blood flow are presented in Fig. 1c, and a typical monitoring is shown in Fig. 2b.
Finally, the T2 relaxation time was evaluated with the use of a regular spin-echo (multi-echo, 6.7 ms echo-spacing) sequence. First, we measured T2 relaxation times of muscle from C57BL6 mice before and after intramuscular injection of a saline solution (0.9% NaCl) in order to validate the feasibility of this study. Osmotic effects are known to modulate T2 decay of muscle in vivo by elevating the extracellular water component in muscle (13, 17). Two relaxation components (a fast one (T2) and a slower one (T)), which are assigned to intra- and extracellular water protons, are usually observed in skeletal muscle. This setting allowed us to observe osmotically induced changes in T2 in C57BL6 mice (data not shown). The same acquisitions were repeated before, during, and after electrical stimulation of the sciatic nerve in eNOS−/− and WT mice. When the decay was analyzed with a biexponential fit, a significant increase in the slow T2 component (T) was observed in the stimulated leg, whereas no changes were found in the control muscle (Fig. 1d, Table 1). Interestingly, both the eNOS−/− and WT mice showed a sustained increase in T after stimulation. The fast T2 component was not modified by the exercise protocol in any of the groups (Table 1).
Table 1. T2 Relaxation Times of the Fast (T2) and Slow (T) Component Measured in Mouse Skeletal Muscle, Before, During, and After Application of an Exercise Protocol
Comparison between wild type (WT) and eNOS knockout mice (eNOS−/−).
We document here that the BOLD SI increases in response to an exercise protocol achieved by electrical stimulation of the sciatic nerve. Both eNOS-knockout and WT mice were responsive during the stimulation period. This increase was maintained in the post-stimulation period in WT mice, but not in mice lacking eNOS. Those observations correlate with the evolution of muscle oxygenation: pO2 increased during electrical stimulation, and progressively returned to initial levels in eNOS-knockout mice (contrary to the WT mice). We observed an immediate increase in blood flow at the onset of the exercise protocol, which explains the initial modification in muscle oxygenation. Muscle blood flow decreased rapidly in the post-stimulation period. Accordingly, the persistence of a high BOLD SI in the post-stimulation period has to be related to oxygenation, rather than to blood flow. However, the increase in BOLD SI at the onset of exercise could be related to both oxygenation and blood flow.
Our flow data are in accord with previous reports that blood flow in human muscles rapidly increases at the onset of exercise, and linearly in relation to work intensity (21, 27–31). Therefore, blood flow is precisely regulated to match the oxygen demand and the local metabolic requirements of skeletal muscle. The control mechanisms that are engaged in exercise-induced skeletal muscle vascular regulation include muscle mechanical, metabolic, myogenic, neural, and endothelial-derived relaxing factors (e.g., NO). It has been documented that NO is formed in skeletal muscle in response to contraction (21, 29).
However, our data are valid only for a moderate exercise protocol, where the oxygen supply (improved by the increase in blood flow) largely compensates for the oxygen demand during exercise. Meyer et al. (11) also observed a significant BOLD response in muscle during “light” exercise. In experiments with higher-intensity exercise (3-V pulses instead of 0.5-V pulses), a drop in pO2 and BOLD SI were observed (Fig. 3). Muscle deoxygenation during intense exercise has often been described (32, 33). These results reinforce the notion that oxygenation is related to the changes observed in T-GRE imaging of skeletal muscle.
We observed that muscle oxygenation remained high even when the blood flow decreased to basal levels in the post-stimulation period. This effect is likely due to a decrease in oxygen consumption, mediated by NO. Indeed, there is much evidence that NO regulates mitochondrial respiration of muscle cells by virtue of reversible interactions with cytochrome c oxidase (reversible inhibition of mitochondrial respiration by NO competing with O2 for the active site of cytochrome oxidase) (20, 21, 34–36). Higher NO levels or more prolonged exposure to NO may also involve the permanent inhibition of complexes I and II. NO donors have been shown to decrease oxygen consumption of tissue slices from canine limb muscle and inhibition of NOS increased oxygen consumption (21). It has been demonstrated that skeletal muscle can produce NO, and that endogenous NO modulates muscle function. The physiological importance of endogenous NO as a mitochondrial modulator in intact muscle remains controversial, however. In this regard, we observed here that mice lacking eNOS showed a decrease in muscle oxygenation after the exercise protocol ended, contrary to the WT mice. This finding further supports the hypothesis that NO is produced in skeletal muscle in response to exercise, and can inhibit mitochondrial respiration and hence the oxygen consumption rate in muscle cells, which can then maintain a high pO2. In the absence of eNOS, muscle oxygenation decreases in the post-stimulation period, concomitant with blood flow. As stated above, this hypothesis is only applicable in the case of a moderate exercise protocol, since we observed that pO2 was decreased during exercise of higher intensity.
In addition to hemoglobin, the rate of myoglobin saturation/desaturation with oxygen could also account for changes in BOLD SI. Nevertheless, at physiological muscular pO2, myoglobin is always saturated (33), and accordingly is not involved in the changes observed with fMRI.
In order to discriminate between an effect of oxygenation and an involvement of T2 changes on the BOLD SI, we evaluated T2 before, during, and after exercise in control and eNOS−/− mice. Muscle T2 has been reported to be modified during exercise (13–17), and may therefore be involved in BOLD SI changes. In this study, the slow component (T) of T2 decay was shown to be modified by the exercise protocol. The observed increase in T is in good agreement with previous studies (13, 14, 16). Nevertheless, T remained high in the post-stimulation period in both C57Bl6 (WT) and eNOS−/− mice. Consequently, the evolution of the BOLD SI is better correlated with muscle oxygenation than with T changes, since only pO2 decreased in the post-stimulation period for eNOS−/− mice, similarly to the BOLD intensity.
This study documents that increases in the BOLD SI can be observed in mouse skeletal muscle during a moderate and prolonged (15-min) exercise protocol. We demonstrated that the BOLD response likely depends on oxygenation and blood flow effects at the onset of exercise, and on the single effect of oxygenation after electrical stimulation stops. Blood flow and T2 changes were not correlated with the sustained increase in the BOLD SI after exercise, in contrast to muscle pO2. These data present new possibilities for elucidating physiological and pathological phenomena in skeletal muscle by fMRI.
This study was supported in part by grants from the Belgian National Fund for Scientific Research to Bénédicte Jordan (postdoctoral researcher) and Olivier Feron and Chantal Dessy (research associates).