|Home | About | Journals | Submit | Contact Us | Français|
Performance of a unimanual motor task often induces involuntary mirror electromyographic (EMG) activity in the opposite, resting hand. In spite of the ubiquitous presence of mirroring, little is known regarding the underlying cortical contributions. Here, we used functional magnetic resonance imaging (fMRI) to study brain regions activated in association with parametric increases in right isometric wrist flexion force (10%, 20%, 30%, and 70%) in 12 healthy volunteers. During scanning, EMG activity was recorded bilaterally from flexor carpi radialis (FCR), extensor carpi radialis (ECR), biceps brachii (BB), and triceps brachii (TB). Mirror EMG was observed in left FCR during 20%, 30%, and 70% of force. Left ECR, BB, and TB showed mirror EMG only at 70% of force. Increasing force was associated with a linear increase of blood-oxygen-level–dependent (BOLD) signal in bilateral primary motor cortex (M1), supplementary motor area (SMA), caudal cingulate, and cerebellum. Mirroring in the left FCR correlated with activity in bilateral M1, SMA, and the cerebellum. Overall, our results suggest that activity in these regions might reflect sensorimotor processes operating in association with mirroring and suggest caution when interpreting fMRI activity in studies that involve unilateral force generation tasks in the absence of simultaneous bilateral EMG/kinematics measurements.
During unimanual motor tasks, muscle activity may not be restricted to the contracting arm, but has also been reported in the contralateral resting arm, referred to as mirror electromyographic (EMG) activity (Mayston et al. 1999; Leocani et al. 2000; Hoy et al. 2004; Carson 2005; Giovannelli et al. 2006; Cincotta and Ziemann 2008). Mirror EMG activity has been reported during performance of simple (Uttner et al. 2007; Ottaviani et al. 2008) and complex (Armatas et al. 1994; Cincotta et al. 2006) motor tasks as well as during strong unimanual voluntary contractions (Muellbacher et al. 2000; Zijdewind et al. 2006) in healthy individuals and in patients with motor disorders.
Previous studies demonstrated a possible role for the primary motor cortex (M1) ipsilateral to a voluntary contraction in the generation of mirror EMG activity (Mayston et al. 1999; Hoy et al. 2004, Carson 2005, Zijdewind et al. 2006). Accordingly, functional magnetic resonance imaging (fMRI) studies have shown an increase in activity in M1 ipsilateral to a contracting arm during strong unimanual force generation (Dettmers et al. 1995; Thickbroom et al. 1998; Dai et al. 2001; van Duinen et al. 2008), when mirror EMG activity has been reported. Furthermore, studies in patients with movement disorders have demonstrated a decrease in intracortical inhibition in the M1 ipsilateral to the moving arm in the presence of mirror EMG activity (Cincotta et al. 2006; Cincotta and Ziemann 2008) supporting the view that mirroring is related to functional changes in corticospinal projections originating in the M1 ipsilateral. Although these studies characterized detailed physiological and blood-oxygen-level–dependent (BOLD) changes in ipsilateral M1 (Dettmers et al. 1995; Thickbroom et al. 1998; Dai et al. 2001), others have evaluated simultaneously changes in BOLD signal and EMG activity (Post et al. 2008), suggesting that simultaneous recording of BOLD signal and EMG would be necessary to identify the possible involvement of brain regions other than M1 in mirroring.
To address this issue, we used fMRI to study brain regions activated during performance of a parametric unimanual force generation task with simultaneous EMG recordings from 8 muscles in both arms. We chose this combined fMRI/EMG design to be able to relate accurately activation in different cortical areas engaged in force generation with mirror EMG activity. The main finding of our study was that during performance of a unimanual force generation task, in addition to M1 ipsilateral to the active hand, mirroring correlated with increased activity in the contralateral M1 and the medial premotor structures including SMA and the cerebellum.
Fifteen healthy volunteers (6 women, 9 male; 29 ± 9.5 years) participated in the study. All subjects gave their informed consent to the experimental procedure, which was approved by the National Institutes of Neurological Disorders and Stroke (NINDS) ethics committee. The study was performed in accordance with the Declaration of Helsinki. All the subjects were right handed as tested by the Edinburgh handedness inventory (Oldfield 1971). All subjects participated in a single session carried out in the MRI scanner. In this session, subjects performed 10%, 20%, 30%, and 70% of their maximal right wrist isometric flexion force in a pseudo-randomized order with their right arm. Two subjects were excluded from data analysis because of excessive head movement (>5 mm) and 1 for not respecting the behavioral instructions during scanning. The other subjects showed no significant movements during the different levels of right wrist flexion force (>2 mm; range 0.1–2.6 mm, mean = 0.84 ± 0.67 mm). Therefore, data from 12 subjects were included in the analysis.
During testing, subjects lay in the MRI scanner. A custom 6-axis load cell (35-E15A; Woodland, CA) was attached to the right arm to measure wrist flexion force (Fig. 1A). Custom software was written to acquire signals from the load cell and to display visual feedback (Fig. 1B) corresponding to rest, 10%, 20%, 30%, and 70% of each subject maximal right wrist flexion force in real-time (Matlab R14SP3, Mathworks, Natick, MA; Hidler et al. 2006). Subjects were instructed to respond to the GO (target) signal presented on a computer monitor by moving a cursor to a target box located at different distances according to their maximal right wrist flexion force (i.e., larger distances at larger levels of force generation) and to maintain the cursor in the target box for 2 s by performing isometric right wrist flexion force (Fig. 1B). The visual target was displayed on a rear-projection screen using a color liquid crystal display (LCD) projector. The subject viewed the display through a mirror fixed to the top of the head coil. The instruction for the subject was “When you see the GO signal bend the right wrist.” The maximal right wrist flexion force was measured 3 times at the beginning of each session and the measurements were averaged. The experiment was conducted in an event-related design using all force conditions (10%, 20%, 30%, and 70%). The gap between movements was 6 and 14 s. The stimulus-onset-asynchrony between trials was jittered with an average of 9 s (range = 6–14 s). Subjects performed 2 experimental runs. Each run consisted of 64 trials with 16 trials for each force level. EMG measurements (see below EMG recordings and analysis) were acquired from both arms at rest and during 10%, 20%, 30%, and 70% of maximal right wrist flexion force (Fig. 1C). During testing, the left arm was oriented parallel to the trunk and supported from the forearm to the radial side of the hand with cushions. It is important to consider that during force generation in the lying position, stabilization of the trunk might be achieved by contracting the other extremity or other segments of the same arm. To minimize this possibility, the segments proximal to the wrist were supported by 4 padded adjustable bumpers. The contact point of the bumpers along the forearm and the width of the bumpers were adjusted to accommodate forearms of different subjects. The configuration of the bumpers secured form closure, which prevented the forearm from rotating and also minimized the forces generated at the hand and wrist from propagating up the arm and trunk (Hidler et al. 2006).
EMG recordings were acquired using the BrainAmp MR Plus recorder and software (Brainproducts GmbH, Munich, Germany; van Duinen et al. 2005). Surface electrodes were positioned bilaterally on the skin overlying the flexor carpis radialis (FCR), extensor carpi radialis (ECR), biceps brachii (BB), and triceps brachii (TB) muscles in a bipolar montage (interelectrode distance, 2 cm). To avoid EMG movement artifact, the EMG leads were secured with adhesive tapes on the electrodes and cohesive bandages around the arms. The EMG signals were filtered (band-pass, 10–400 Hz), sampled at 5 kHz, and stored on a PC for off-line analysis. Brain Vision Analyzer software was used to remove scanner artifacts according to the method described by Allen et al. (2000). After the scanner artifact was removed, EMG data (ASCII format) were imported into Spike2 (Cambridge Electronic Design, Cambridge, UK) for further analysis. EMG signals were rectified, and the mean EMG activity was obtained, trial-by-trial, from all recording muscles. The onset of each EMG burst was defined as the time point when the mean EMG activity exceeded the baseline activity (BL) by 3 standard deviations (SD) of the baseline (BL ± 3SD). The offset of each EMG burst was defined as the time point when the EMG signal fell below this value. EMG data were expressed as percent of baseline (BL) EMG. A total of 32 trials (trials from 2 fMRI runs combined) per condition were averaged. Kolmogorov–Smirnov and Mauchly's tests were initially used to characterize the distribution and sphericity of data, respectively. Repeated measures analysis of variance (ANOVARM) was used to determine the effect of FORCE (Baseline, 10%, 20%, 30%, and 70%) on each muscle EMG. Tukey post hoc test was used for multiple comparisons (SigmaStat, Version 2.03, Systat Software Inc., San Jose, CA). To assess the presence of mirror EMG activity, we tested the effect of FORCE (Baseline, 10%, 20%, 30%, and 70%) on EMG activity on each muscle. Significance was set at P< 0.05. Variance is expressed as mean ± SD.
The fMRI scans were performed using a GE 3-T scanner (GE Medical, Waukesha, WI) with an 8-channel receiving head coil. The functional images were acquired in 2 separated runs (274 volumes each) using a T2*-weighted interleaved echo-planar imaging (EPI) sequence covering the whole brain (TR = 2 s, TE = 30 ms, flip angle = 90°, NEX = 1, field of view = 220 × 220, number of slices = 32, slice thickness = 4 mm, gap = 0, and matrix size = 64 × 64). T1-weighted structural images with 1 × 1 × 4-mm resolution were also acquired for each subject.
fMRI data preprocessing and analysis were performed with SPM2 and SPM5 (for second level group analysis) software (the Wellcome Department of Imaging Neuroscience, University College London, London, UK). The first 5 volumes (scans) of images were automatically discarded prior to data processing. All images of each time series (i.e., a scan run) were first slice-time corrected and then realigned to the first image of the first run with a 6-parameter rigid-body transformation for each subject. Resulting spatially realigned images were normalized to the template MNI brain. Therefore, the coordinates used to report the locations of activation are all in the MNI space. The image volumes were subsequently spatially smoothed (FWHM = 8 mm3) and high-pass filtered with a cut-off frequency of 1/128 s to remove low frequency drift. Statistical analyses were carried out with the SPM2 and SPM5 software using the general linear model (Friston et al. 1995; Turner et al. 1998). The SPM matrix modeled an implicit baseline that included the resting baseline between trials in the fMRI design.
For the first level fixed-effects analysis, statistical contrasts (t-tests) of MR signals for each level of force (10%, 20%, 30%, and 70%) relative to the baseline were performed for each individual subject. Anatomical identification was carefully performed by superimposing activation foci on the MNI brain and on the normalized structural T1-weighted images of each subject. The resulting contrast images were used as input for the random-effects analysis at the group level using SPM5. Task-related changes in BOLD signal between and across force levels were examined using 1-sample t-tests over the voxels in the whole brain. A voxel is considered to be significantly different from the baseline or in a contrast if it survived the False Discovery Rate (FDR) correction for multiple comparisons at the threshold of P < 0.05 or otherwise specified. To characterize changes in BOLD signal across conditions as the force level increased, a linear trend analysis was also performed at the second level group analysis. The trend analysis was done using the output of a within-subject ANOVA in SPM5 by assigning linear weights to the force condition (using a “Flexible Factorial” design with force level as a within-subject factor). The purpose of the linear trend analysis was to identify cortical regions that showed a significant linear increase in BOLD signal with increasing levels of force. The activated areas were defined with a probabilistic cytoarchitectonic map (Eickhoff et al. 2005). We used the atlas of Schmahmann et al. (1999) for a specific neuroanatomical differentiation of cerebellar activations and the nomenclature of Larsell and Jansen (1972) to label cerebellar lobules. The SPM anatomy toolbox (Eickhoff et al. 2005) was used to estimate cytoarchitectonic probabilities when possible. In order to evaluate the relationship between EMG activity and BOLD signal, Pearson correlation analysis was performed between mean EMG signals and the activity of the peak voxel of the areas showing a significant linear increase across force levels depicted in Figure 3 using SPM5. In addition, we performed multiple regression analyses of BOLD signal change at the group level using 2 regressors: 1) right FCR EMG and 2) left FCR EMG. The 2 EMG regressors were orthogonalized to the mean EMG of the 2 muscles (i.e., the constant) in 2 regression models using the Gram–Schimdt orthogonalization algorithm. With the orthogonalized regressors, the 2 models accounted for (i.e., subtracted) the mean effect of force/task from the EMG regressors and allowed estimation of BOLD signal change uniquely associated with the right and the left FCR EMG across force levels (van Duinen at al. 2008; van Rootselaar et al. 2008). In model 1, right FCR EMG was entered as the first regressor and left FCR as the second regressor. This model provides estimate for the BOLD signal change associated with the right FCR and of the variance of the activation uniquely associated with the left FCR. In model 2, left FCR EMG was entered as the first regressor and right FCR as the second regressor. This model provides estimate for the BOLD signal change associated with the left FCR and of the variance of the activation uniquely associated with the right FCR. All significant correlations survived the threshold of P = 0.05 (FDR corrected).
Figure 2 (upper traces) illustrates the right FCR EMG activity recorded in a single subject during 10%, 20%, 30%, and 70% of force. The group data are shown in Figure 2A–D. We found a significant effect of FORCE on EMG activity (% of baseline) in all muscles tested in the right arm (FCR, F(4,44) = 76.9, P ≤ 0.001, Fig. 2A; ECR, F(4,44) = 22.0, P ≤ 0.001, Fig. 2B; BB, F(4,44) = 16.9, P ≤ 0.001, Fig. 2C; TB, F(4,28) = 34.1, P ≤ 0.001, Fig. 2D). In the right FCR, EMG activity was increased at 70% (P ≤ 0.001), 30% (P ≤ 0.001), 20% (P ≤ 0.001), and 10% (P = 0.036) of force compared with baseline. In the right ECR, EMG activity was increased at 70% (P ≤ 0.001), 30% (P ≤ 0.01), and 20% (P ≤ 0.02) of force compared with baseline. In the right BB, increased EMG activity was observed at 70% (P≤ 0.001) but not at 30% (P = 0.2), 20% (P = 0.5), and 10% (P = 0.9) of force compared with baseline. In the right TB, increased EMG activity was observed at 70% (P ≤ 0.001) and 30% (P ≤ 0.01) but not at 20% (P = 0.08) and 10% (P = 0.3) of force compared with baseline. Similar effect of FORCE was observed on raw EMG data in all muscles tested in the right arm (FCR, F(4,44) = 27.1, P ≤ 0.001; ECR, F(4,44) = 42.8, P ≤ 0.001; BB, F(4,44) = 6.6, P ≤ 0.001; TB, F(4,28) = 6.2, P ≤ 0.001).
Figure 2 (lower traces) illustrates the left FCR EMG activity recorded in a single subject while the right FCR was performing 10%, 20%, 30%, and 70% of force. The group data are shown in Figure 2E–H. ANOVARM showed a significant effect of FORCE on EMG activity (% of baseline) in all muscles tested in the left arm (FCR, F(4,44) = 11.64, P ≤ 0.001, Fig. 2E; ECR, F(4,44) = 10.23, P ≤ 0.001, Fig. 2F; BB, F(4,44) = 11.38, P ≤ 0.001, Fig. 2G; TB, F(4,28) = 16.77, P≤ 0.001, Fig. 2H). In the left FCR, post hoc testing showed an increase in EMG activity at 70% (P ≤ 0.001), 30% (P = 0.02), 20% (P = 0.02) but not at 10% (P = 0.09) of force compared with baseline, indicating the presence of mirror activity. Mirror EMG activity was also observed at 70% in the left ECR (P ≤ 0.001), left BB (P ≤ 0.001) and TB (P ≤ 0.001). Similar effect of FORCE was observed on raw EMG data in all muscles tested in the left arm (FCR, F(4,44) = 7.6, P ≤ 0.001; ECR, F(4,44) = 7.7, P < 0.001; BB, F(4,44)=11.4, P < 0.001; TB, F(4,28)=12, P ≤ 0.001).
The results of the linear trend analysis showed a significant (FDR corrected P < 0.05) linear trend of the BOLD signal increase primarily in 6 regions as the force level increased (Fig. 3). The 6 cortical areas that exhibited a linear increasing trend in BOLD activity with force were bilateral M1 (Fig. 3A,B), SMA (Fig. 3C), caudal cingulate (Fig. 3D), and 2 cerebellar areas (left hemisphere, lobule IV and right hemisphere, lobules IV–V: Fig. 3E,F). A second order quadratic trend analysis showed no voxels activated in these regions. The bar graphs of Figure 3 show beta values of the contrast estimates (relative to the baseline) of the peak voxels of each condition. Additional activation, relative to the baseline, was also observed in the occipital cortex (BA 17 and BA 18) and bilateral somatosensory cortex (BA3b and BA1).
In addition to the linear trend analysis, second level (group) random-effects analysis showed significant effect of FORCE. One-sample t-tests showed significant (P <0.05, FDR corrected) differences between 70% and 30% relative to 10% of force. There was no significant difference between the higher force levels (70% and 30%), and between the lower force conditions (20% and the 10%). A separate 1-sample t-test, combining the higher force levels (70% and 30%) to contrast the 10% (lower force), showed additional activation at the higher force level (P < 0.005, uncorrected) in the SMA, bilateral M1, somatosensory, caudal cingulate, caudate nucleus, and bilateral cerebellum (Table 1). Reverse contrasts (t-tests comparing lower force level minus higher force level) did not show any significant differences except for the contrast (1-tailed t-test) 10–20%, which showed significantly more activation in the right inferior parietal cortex (BA40).
The results of the multiple regression analyses with orthogonalized regressors showed that when the mean effect of force was removed significant activation was observed in the bilateral M1, SMA, somatosensory cortex, and the cerebellum (Fig. 6 and Table 2). These areas are consistent with the clusters seen in the linear trend analysis (Fig. 3).
EMG activity in the right FCR correlated with BOLD signal increases in the following regions integrating the network in Figure 3: left M1 (r = 0.36, P = 0.01; Fig. 4A), right M1 (r = 0.42; P ≤ 0.01, Fig. 4B), SMA (r = 0.40, P ≤ 0.01; Fig. 4C), caudal cingulate (r = 0.46, P ≤ 0.01; Fig. 4D) and the cerebellum (left cerebellum, r = 0.39, P ≤ 0.01, Fig. 4E; right cerebellum, r = 0.49, P ≤ 0.001, Fig. 4F). The results of model 1 whole-brain regression analysis with the right FCR EMG as the first regressor also showed bilateral M1, SMA, somatosensory cortex, caudal cingulate, and cerebellum (Table 2). The regression model 2 with the right FCR as the second regressor that allowed estimation of a unique association between the right FCR EMG and the BOLD signal change (i.e., above and beyond the activation accounted for by the mean effect of force and the left FCR) revealed again significant activation in these brain regions (Fig. 6A, Table 2). EMG activity in the left FCR (mirror EMG activity) showed a significant correlation with BOLD signal change in left M1 (r = 0.47, P≤ 0.001; Fig. 5A), right M1 (r = 0.59, P ≤ 0.0001; Fig. 5B), SMA (r = 0.29, P = 0.04, Fig. 5C), caudal cingulate (r = 0.33, P = 0.02; Fig. 5D) and left cerebellum (r = 0.64, P ≤ 0.0001; Fig. 5E). A correlational trend was observed between mirror EMG and activity in the right cerebellum (r = 0.27, P = 0.06; Fig. 5F). Consistently, the results of model 1 whole-brain regression analysis with the left FCR EMG as the second regressor showed activation in bilateral M1, SMA, and cerebellum (Fig. 6B, Table 2). No correlations were found between mirror EMG activity at the high levels of force in the ECR, BB, and TB and the regions involved in mirror EMG activity in the left FCR.
In the present study, we combined fMRI/EMG to investigate BOLD changes associated with mirror EMG activity during parametric increase of unimanual force generation. We found mirror EMG activity in the left FCR during 20%, 30%, and 70% of right wrist force. Left ECR, BB, and TB showed mirror EMG activity only at 70% of right wrist force. Mirror EMG activity in the left FCR correlated with BOLD changes in bilateral M1, SMA, and the cerebellum.
We found a linear increase in BOLD signal activity in bilateral M1, SMA, caudal cingulate, and the cerebellum with increasing levels of right wrist force consistent with previous results (Dettmers et al. 1995; Thickbroom et al. 1998; Dai et al. 2001; Peck et al. 2001; Cramer et al. 2002; Halder et al. 2007; van Duinen et al. 2008). EMG activity in all muscles tested in the right arm linearly increased with increasing right force generation. We also observed a linear relationship between force and BOLD activation, as reported in previous studies (Dettmers et al. 1995; Ward and Frackowiak 2003; van Duinen et al. 2008). A number of studies have addressed simultaneously the relation between muscle output (EMG) and fMRI-measured brain activity (Liu et al. 2000; Dai et al. 2001; Oga et al. 2002; van Duinen et al. 2005, 2008) without a detailed description of the brain regions that show activity associated to mirroring in different muscle groups in both arms, which was a goal of our study. Here, using an artifact correction paradigm similar to that of van Duinen et al. 2005, we found increasing activity in bilateral M1, SMA, caudal cingulate, and cerebellum correlated with increasing EMG activity of the right FCR (primary mover), consistent with the direct involvement of these brain regions in the force generation task (see Fig. 4) and in agreement with previous studies (Dettmers et al. 1995; van Duinen et al. 2008). It is less likely that the increased activity on these areas is related to muscle fatigue, because we observed a parallel increase in EMG and brain activity, whereas during muscle fatigue brain activity increases and EMG activity decreases (Post et al. 2008). The use of visual targets in the experimental setup (Sestieri et al. 2007) and visual feedback (Hadler et al. 2007) could have contributed to the significant activation identified in the visual cortex (Fig. 3).
We found mirror EMG activity in all muscles tested in the left arm during 70% of right wrist force, in agreement with previous results in more proximal and distal muscles (Zijdewind and Kernell 2001; Carson 2005; Zijdewind et al. 2006; van Duinen et al. 2008). Mirror EMG in the left FCR was also present during 20% and 30% of right wrist force. These results are in line with previous studies that demonstrated the presence of mirror activity during similar low levels of force generation by the contralateral hand (Armatas et al. 1994; Mayston et al. 1999; Shinohara et al. 2003). The main novel finding in our study was the correlation between changes in BOLD signal in bilateral M1, SMA, and the cerebellum and mirror EMG in the left FCR when using force as a covariate. The unique involvement of bilateral M1, SMA, and the cerebellum in mirroring was further supported by the results of the whole-brain multiple regression analyses in which the mean effect of force was accounted for and the right and left EMG regressors were orthogonalized. The finding that BOLD signal activity in the CG correlated with mirror EMG activity in the left FCR only when force was not accounted for, does not allow us to exclude the possibility that the linear increase in activation in the CG (reported in Fig. 3) was partially related to the force level.
Our results show that activity in both M1s was associated with mirror EMG activity in the left FCR. The finding of a linear correlation between BOLD changes in right M1 and mirror EMG supports the proposal that descending pathways originated in the right M1 are at least partially involved in the generation of mirror EMG activity (Carson 2005). Consistent with this view, excitability in the M1 ipsilateral to a moving arm increases in the presence of mirror EMG in healthy subjects (Zijdewind et al. 2006) and in patients with motor disorders (Cincotta et al. 2006). It has been proposed that by increasing the functional demands of the motor task (in this case force) transcallosal pathways originating in the voluntarily active M1 exert a disinhibitory effect on the excitability of intracortical circuits measured in the resting hemisphere, which may partially contribute to control excitability in the ipsilateral M1 (Li 2007; Perez and Cohen 2008). Therefore, our results of a correlation between activity in the left M1 and mirror EMG suggest that the contribution of the left M1 to mirror might theoretically occur throughout ipsilateral descending pathways or disinhibitory influences on the right M1.Because previous evidence suggest that ipsilateral descending pathways play a less fundamental contribution to the generation of mirror EMG activity (Cincotta and Ziemann 2008), we would like to suggest that their involvement may occur to some extent through the right M1. However, it is important to keep in mind that from our experimental design it is not possible to determine if the contribution of the left M1 to mirroring occurs through their inputs on the right M1, direct descending ipsilateral pathways or both. Although the pattern of BOLD signal activity observed in the right M1 was similar to previous studies (Dettmers et al. 1995; van Duinen et al. 2008), at 20% and 30% of force, very small increases in BOLD signal were observed in the presence of mirror EMG activity in the left FCR. A possible explanation is that lower levels of force do not induce significant change in the magnitude of BOLD signal activity (van Duinen et al. 2008). This result might also indicate that the right M1 is more likely to contribute to the mirror EMG at the higher levels of force generation. Furthermore, we cannot exclude the possibility that afferent input contributed partially to changes in BOLD signal in the right M1 (Naito et al. 2002; Bestmann et al. 2004).
Our results also indicate that the SMA is active in association with left FCR mirror EMG activity. Studies in monkeys have shown that neuronal activity in the SMA encodes movement dynamics (such as force) during movement execution (Smith 1979; Padoa-Schioppa et al. 2004). In addition, neuroimaging studies in humans found force-related BOLD signal activity changes in the SMA (Dettmers et al. 1995; Cramer et al. 2002). The involvement of the SMA in mirror EMG activity is supported by their role in bimanual coordination (Wenderoth et al. 2005), movement lateralization (Brinkman 1984; Chan and Ross 1988), and in the performance of mirror motor tasks (Kraft et al. 2007). It is possible that the SMA contributes to modulate activity in the M1 ipsilateral to the moving hand during the presence of mirroring. This proposal is also supported by the strong connections between SMA and M1 (Civardi et al. 2001; Liu et al. 2003) and by the involvement of the SMA in online control of voluntary movements (Tanji and Shima 1994). Alternatively, the SMA, which projects directly to interneurons in the spinal cord (Picard and Strick 1996), could theoretically influence EMG activity in the mirror hand through the recruitment of direct descending corticospinal projections. This is in agreement with the view that, in addition to the classical serial hierarchical organization of the motor system, there is also a parallel involvement of primary and secondary motor areas in the control of mirroring (Cincotta and Ziemann 2008).
The BOLD activity in the caudal cingulate was not correlated with the mirror EMG activity in the left FCR when the variability in the BOLD signal was accounted for by the right EMG and the mean effect of force (Fig. 6B). Therefore, we cannot exclude the possibility that the linear increase in activation in the CG (reported in Fig. 3) was partially related to the generation of wrist force in general. Previous studies have demonstrated that cingulate motor areas play a substantial role in different aspects of motor behaviors, such as in the preparation, execution, and error detection of movements (Shima et al. 1991; Shima and Tanji 1998; O'Connell et al. 2007). It is also known that activity in the caudal cingulate is modulated by force production in humans (Dettmers et al. 1995; Ehrsson et al. 2007) and monkeys (Cadoret and Smith 1995, 1997; Richardson et al. 2008). Therefore, a possibility is that caudal cingulate contributes to force adaptation-related changes in directional tuning (Richardson et al. 2008). Because neurons in the caudal cingulate project to M1s and the SMA (Picard and Strick 1996), these connections may provide possible pathways for this contribution. On the other hand, studies in monkeys have shown substantial projections from caudal cingulate neurons to the spinal cord as the SMA (Picard and Strick 1996), which may provide another potential pathway for the observed effects.
We found that the cerebellum is active in association with left FCR mirror EMG activity. What was striking in our results is that the correlation between mirroring and BOLD signal activity in the cerebellum was stronger compared with the SMA. The involvement of the cerebellum in mirroring is in agreement with previous evidence showing that cerebellar neurons encode force in a linear fashion (Smith and Bourbonnais 1981; Townsend et al. 2006), but note that this correlation remained present when using force as a covariate. More specifically, activity in the cerebellum has been related to force prediction (Kawato et al. 2003; Boecker et al. 2005). One of the major projections from the cerebellum is to the M1, most corticopontine and cerebellar projections to the cerebral cortex are contralateral (Middleton and Strick 1998). Therefore, a possibility is that one of the pathways mediating cerebellar influence on mirror EMG activity involves the right M1. Neuroimaging studies have suggested that the cerebellum is involved in the storage (Imamizu et al. 2000), retrieval (Kawato et al. 2003), and switching (Imamizu et al. 2003) of internal models. Because internal models are considered to be mechanisms that can mimic the input–output in the motor system (Kawato et al. 2003), it is possible that the cerebellum contributes to monitor the involuntary output (mirror EMG) in the left resting arm. Our results support the hypothesis of the “generalization capability” of the cerebellum (Boecker et al. 2005), which may be one of the features involved in the generation of mirror activity in the resting arm during movement of the contralateral arm.
The presence of mirror EMG activity has been reported in healthy individuals and in patients with motor disorders (for reviews see Carson 2005; Cincotta and Ziemann 2008). Our results provide novel physiological insights suggesting that activation in nonprimary brain regions might reflect sensorimotor processes operating in association with mirror EMG activity. It should also be kept in mind that activation in these regions might point to distant influences from other interconnected areas (Picard and Strick 1996) and reflect either excitatory or inhibitory processes (Waldvogel et al. 2000; Almeida and Stetter 2002) engaged in mirroring. In addition, a spinal contribution to mirroring cannot be ruled out because patients with lesion of the corpus callosum show some level of mirror motor activity to a contralateral resting limb (Meyer et al. 1995).
On the one hand, our results suggest the need for monitoring of EMG activity from multiple muscle groups in both upper limbs of stroke patients performing a motor task to control for the possibility that abnormal contralesional activation observed in M1 ipsilateral (Ward et al. 2006) might be related to the presence of mirror EMG activity. It is tempting to speculate that modulation of excitability in any of these activated areas (i.e., left and right M1, SMA, and cerebellum) by noninvasive brain stimulation might contribute to the balance of unwanted mirroring during performance of unimanual motor tasks in patients with motor disorders. On the other hand, our results might explain previous findings of activation in SMA, ipsilateral M1, and/or the cerebellum in association with performance of unilateral hand movements when EMG activity in the resting hand was not monitored, emphasizing the need for careful control of bilateral upper extremity EMG activity in the scanner.
We found that mirror EMG activity in the left FCR during parametric increases in right isometric force was associated with increased activity in both M1s, the SMA and the cerebellum. Additionally, our results suggest caution when interpreting fMRI activity in studies that involve unilateral force generation tasks in the absence of simultaneous bilateral EMG/kinematics measurements.
Intramural Research Program of the NIH, NINDS; Alfried Krupp von Bohlen und Halbach-Foundation (to Bernhard Sehm).
Conflict of Interest: None declared.