Self-paced movements induce high-frequency gamma oscillations in primary motor cortex

Abstract

There has been increasing interest in the functional role of high-frequency (> 30 Hz) cortical oscillations accompanying various sensorimotor and cognitive tasks in humans. Similar “high gamma” activity has been observed in the motor cortex, although the role of this activity in motor control is unknown. Using whole-head MEG recordings combined with advanced source localization methods, we identified high-frequency (65 to 80 Hz) gamma oscillations in the primary motor cortex during self-paced movements of the upper and lower limbs. Brief bursts of gamma activity were localized to the contralateral precentral gyrus (MI) during self-paced index finger abductions, elbow flexions and foot dorsiflexions. In comparison to lower frequency (10–30 Hz) sensorimotor rhythms that are bilaterally suppressed prior to and during movement (Jurkiewicz et al., 2006), high gamma activity increased only during movement, reaching maximal increase 100 to 250 ms following EMG onset, and was lateralized to contralateral MI, similar to findings from intracranial EEG studies. Peak frequency of gamma activity was significantly lower during foot dorsiflexion (67.4 ± 5.2 Hz) than during finger abduction (75.3 ± 4.4 Hz) and elbow flexion (73.9 ± 3.7 Hz) although markedly similar for left and right movements of the same body part within subjects, suggesting activation of a common underlying network for gamma oscillations in the left and right motor cortex. These findings demonstrate that voluntary movements elicit high-frequency gamma oscillations in the primary motor cortex that are effector specific, and possibly reflect the activation of cortico-subcortical networks involved in the feedback control of discrete movements.

Introduction

In recent years, there has been a growing interest in the role of neural oscillations in motor control (Baker et al., 1999, MacKay, 1997, Murthy and Fetz, 1996). It has long been known that movements elicit frequency specific changes in the EEG (Chatrian et al., 1958, Jasper and Penfield, 1949) and changes in spectral power in the mu (8–14 Hz) and beta (15–30 Hz) frequency bands can be observed during both voluntary (Feige et al., 1996, Jurkiewicz et al., 2006, Pfurtscheller and Aranibar, 1977, Salenius et al., 1997) and passive movements (Alegre et al., 2002, Cassim et al., 2001). Although the majority of neuroimaging studies have focused on these lower frequency “sensorimotor” rhythms, there has been increasing interest in the presence of high-frequency (> 30 Hz) gamma oscillatory activity within the human motor system. Gamma band activity has been implicated in perceptual binding of multiple inputs within sensory systems (Engel and Singer, 2001) and may also reflect functional couplings within distributed cortical networks that underlie perceptual or cognitive processes (Tallon-Baudry and Bertrand, 1999). It has been suggested that sensorimotor rhythms may serve a similar role within the motor system (Marsden et al., 2000, McAuley and Marsden, 2000) and frequency specific brain activity arising from motor areas has been proposed by a number of investigators as a potential control parameter for brain–machine interfaces (Schalk et al., 2007, Waldert et al., 2008).

Crone et al. (Crone et al., 1998) reported increased gamma activity in the electrocorticogram (ECoG) in awake patients performing sustained muscle contractions, noting that high-frequency (75–100 Hz) gamma oscillations showed greater somatopic organization over the sensorimotor areas and tended to be more time-locked to movement onset in comparison to lower frequency mu and beta rhythms. More recent studies have confirmed movement-related increases in the low (30–60 Hz) and high (60–100 Hz) gamma band in intracerebral depth electrode (Szurhaj et al., 2006) and ECoG (Brovelli et al., 2005, Pfurtscheller et al., 2003) recordings from motor and premotor areas. These results were recently confirmed in a study of 22 epilepsy patients with implanted subdural electrode grids using movements of various body parts (based on grid location), showing somatotopically specific increases in the 76–100 Hz frequency band during continuous movements in comparison to resting state (Miller et al., 2007). ECoG studies have also shown evidence of corticomuscular coherence within the gamma band including frequencies above 60 Hz (Marsden et al., 2000). Thus, invasive recordings in humans have shown frequent evidence of gamma oscillations in cortical motor areas, although the precise localization of the generators of this activity has been limited to information derived from depth electrodes or subdural recordings in or near motor areas in surgical patients. Movement-related gamma activity has been reported in scalp EEG recordings (Shibata et al., 1999) although EEG studies of gamma activity are hampered by the limited ability to localize the underlying generators, and the potential contamination of high-frequency EEG signals by muscle activity. A number of recent MEG studies have also reported gamma band activity in the 30 to 120 Hz range during various cognitive and sensorimotor tasks. Increases in corticomuscular coherence in the 40 to 60 Hz band were observed in MEG signals overlying the motor cortex during preparation to respond to a visual stimulus (Schoffelen et al., 2005). Increased high (61 to 90 Hz) gamma activity during a maintained isometric contraction task was localized to the hand motor cortex using a novel MEG source modeling approach (Tecchio et al., 2008) and a recent study reported increased power in the 62 to 87 Hz frequency band in MEG sensors overlying the contralateral sensorimotor cortex when subjects moved a joystick in self-selected directions to a visual cue (Waldert et al., 2008).

Although a number of studies using non-invasive EEG and MEG methods have reported both changes power and coherence in the higher gamma frequency range within or overlying cortical motor areas during tasks that involve visuomotor control or sensorimotor coordination, there have been relatively few studies examining changes in the high gamma band during simple self-paced movements. Previous studies have reported changes in either spectral power, or in corticomuscular coherence (with or without concomitant increases in spectral power) over variable duration time windows encompassing either periods of continuous muscular contraction, motor preparation, or both. In addition, the terms ‘gamma’ and ‘high gamma’ are used somewhat inconsistently with regard to the specific frequency range of analysis, or choice of on-line filter settings, making it difficult to compare results across studies. In contrast, intracranial EEG studies have shown increases in spectral power the 70 to 80 Hz range, and specifically during movement execution rather than movement preparation (Brovelli et al., 2005, Pfurtscheller et al., 2003). This raises questions regarding the role of somatosensory or proprioceptive feedback in eliciting high gamma band oscillations in sensorimotor regions during tasks that involve motor responses, and their relationship to frequency changes in other brain areas as the result of sensory input. For example, Bauer and colleagues reported increased gamma activity in the primary somatosensory cortex during simple tactile stimulation (Bauer et al., 2006). Taken together, these findings would predict the presence of high-frequency gamma oscillations in either motor or somatosensory areas during any task involving simple transient movements, although the question remains whether such high-frequency gamma activity reflects cognitive aspects of motor preparation or control, or may be simply related to the preparation or execution of discrete movements.

Neuromagnetic recordings combined with recently developed spatial filtering methods based on beamforming algorithms provide a new non-invasive tool for the precise localization of oscillatory brain activity (Hillebrand and Barnes, 2005). We recently demonstrated the localization of movement-related beta and mu band activity to specific regions of the precentral and postcentral cortex in humans using these methods (Jurkiewicz et al., 2006). In the present study, we demonstrate that simple, self-paced movements of the upper and lower limbs in humans are accompanied by a characteristic narrow band burst of high-frequency (65 to 80 Hz) gamma activity and that this activity can be localized to somatotopically specific regions of the primary motor cortex using whole-head MEG recordings and beamformer source analysis.