Skip to main content
NCBI home page
The Journal of Physiology logoLink to The Journal of Physiology
. 2007 Feb 8;580(Pt 3):801–814. doi: 10.1113/jphysiol.2006.124503

Network oscillations and intrinsic spiking rhythmicity do not covary in monkey sensorimotor areas

Claire L Witham 1, Stuart N Baker 1
PMCID: PMC1891004  PMID: 17289787

Abstract

We investigated the relationship between local field potential (LFP) oscillations and intrinsic spiking rhythmicity in the sensorimotor system, because intrinsic rhythmicity has the potential to enhance network oscillations. LFPs and 918 single units were recorded from primary motor cortex (M1), primary somatosensory cortex (S1, areas 3a and 2), posterior parietal cortex (area 5) and the deep cerebellar nuclei (DCN). Some cells were antidromically identified as pyramidal tract neurons (PTNs). In each area the power of ∼20 Hz LFP oscillations was assessed during periods of steady holding, when such oscillations have previously been shown to be maximal in M1. Oscillations were strongest in area 5 and weakest in the DCN. Using a previously developed method, the postspike distance-to-threshold trajectory was determined from the interspike interval histogram for each cell. Many cells had significant peaks, suggesting an intrinsic tendency towards rhythmic firing. Surprisingly, trajectory peaks were most common for M1 PTNs (115/146 cells) and rarest for area 5 neurons (12/82 cells). The extent of intrinsic spiking rhythmicity is not therefore simply related to the strength of 20 Hz oscillations in the sensorimotor system. These results suggest that intrinsic rhythmicity is not required for the generation and maintenance of oscillatory activity.

Research into sensorimotor oscillations has concentrated predominantly on the motor side of the system. The beta-frequency oscillations (15–30 Hz) in motor cortex have been studied in both humans and animals (Murthy & Fetz, 1992; Sanes & Donoghue, 1993; Salmelin & Hari, 1994; Baker et al. 1997; Kilner et al. 1999). They are coherent with muscle and cerebellar activity (Conway et al. 1995; Baker et al. 1997; Salenius et al. 1997; Courtemanche & Lamarre, 2005; Soteropoulos & Baker, 2006). However, a detailed specification of the function of these oscillations remains elusive.

There is increasing evidence for the involvement of oscillations in the somatosensory system. Early on, Murthy & Fetz (1992) showed oscillatory synchrony between pre- and postcentral sites in monkey, and this was confirmed by Brovelli et al. (2004). Using Granger causality methods, Brovelli et al. (2004) also demonstrated that the information flow in the beta-band is strongest in the direction from area S1 to M1. Additionally, oscillations are present in posterior parietal cortex (MacKay & Mendonca, 1995; Brovelli et al. 2004) and in peripheral afferents (Baker et al. 2006). Corticomuscular coherence may involve both feedback (sensory) as well as feedforward (motor) pathways (Riddle & Baker, 2005), although this is controversial (Gerloff et al. 2006).

In computer simulations, stable oscillatory activity is an emergent property of small networks of inhibitory (GABAergic) interneurons (Wang & Buzsáki, 1996; Pauluis et al. 1999; Traub et al. 1999). Enhancing GABAA receptor transmission by administering diazepam increases the power of 20 Hz EEG oscillations in humans (Baker & Baker, 2003), supporting the hypothesis that inhibitory networks form the basis of oscillatory rhythmogenesis. At the single-cell level, rhythmic discharge at a particular frequency could be due to regular synaptic input (e.g. from local inhibitory networks), and/or to intrinsic properties of the cell. Neurons with an intrinsic tendency to rhythmic firing at a particular frequency could augment synchronous network oscillations (Gray & McCormick, 1996). Such cells have been discovered in M1 (Wetmore & Baker, 2004; Chen & Fetz, 2005), S1 (Lebedev & Nelson, 1995) and the deep cerebellar nuclei (DCN; Thach, 1968; Jahnsen, 1986).

If cells with an intrinsic tendency to rhythmic firing are important in assisting the generation of oscillatory activity, they might be found preferentially in brain areas with large amplitude oscillations. In this study, firstly we compare the magnitude of local field potential (LFP) oscillations in several areas of sensorimotor cortex, as well as the deep cerebellar nuclei. Using invasive recordings with penetrating microelectrodes in monkeys that were awake, we are able to provide definitive comparisons of oscillatory power, free of the possible confounding effects of signal spread which affect non-invasive work in humans. Secondly, we apply an interspike interval statistical analysis method pioneered by Matthews (1996), and refined by Wetmore & Baker (2004), to determine whether cells in each of these areas show a propensity to rhythmic firing. Surprisingly, we find that whereas oscillations are strongest in somatosensory areas, intrinsic rhythmicity is most prevalent for the pyramidal tract output cells in M1.

Methods

Behavioural task

Four female rhesus macaques (M. mulatta) were trained to perform behavioural tasks for food reward. Two monkeys (E and T) were trained on a bimanual precision grip task (for full details see Wetmore & Baker, 2004; Soteropoulos & Baker, 2006). Two animals (M and L) performed an index finger flexion task. The finger was inserted into a narrow tube, which splinted the finger and constrained movement to the metacarpo-phalangeal (MCP) joint. The tube was mounted on a lever that rotated on an axis aligned to the MCP joint. Lever movement was sensed by an optical encoder, and a motor exerted torque in a direction to oppose flexion. This was programmed to act like a spring (initial torque, 48 mN m). The task required movement into target (between 6 and 24 deg flexion) and holding for 2 s (torque required at target either 64 or 128 mN m). Motor torque then rose, and the animal released the lever to obtain its reward. The majority of the analysis reported here focuses on the hold period of both the precision grip and finger flexion tasks, as this has previously been shown to contain the strongest beta-band activity (Baker et al. 1997, 2001).

Surgical preparation

Following behavioural training, each monkey was implanted under general anaesthesia and aseptic conditions with a headpiece (to allow head fixation) and a recording chamber placed over the central sulcus (Lemon, 1984; Baker et al. 1999). The anaesthesia consisted of either 2–2.5% isoflurane inhalation in 50% N2O–50% O2 (monkeys E and T) or 3.0–5.0% sevoflurane inhalation in 100% O2 supplemented with a continuous infusion of intravenous alfentanil (0.025 mg kg−1 h−1; monkeys M and L). A full programme of postoperative analgesia (10 μg kg−1 buprenorphine (Vetergesic), Reckitt and Colman Products; 5 mg kg−1 carprofen (Rimadyl), Pfizer) and antibiotic care (coamoxyiclav 140/35, 1.75 mg kg−1 clavulanic acid, 7 mg kg−1 amoxycillin (Synulox), Pfizer; 10 mg kg−1 cefalexin (Ceporex), Schering-Plough Animal Health or 15 mg kg−1 amoxycillin (Clamoxyl LA), Pfizer) followed surgery. Two stimulating electrodes placed in the pyramidal tract permitted antidromic identification of M1 pyramidal tract neurons (PTNs) (Lemon, 1984). In a further surgical procedure a chamber was implanted, centred over stereotaxic coordinates P8.5, ML4, to allow access to the deep cerebellar nuclei (DCN). All procedures were carried out under the authority of licences issued by the UK Home Office under the Animals (Scientific Procedures) Act 1986.

Recording

In daily experiments, a 16-channel microdrive, loaded with either glass-insulated platinum electrodes (M1) or tetrodes (area 3a, area 2, area 5 and DCN), was used to record single units and LFPs (Eckhorn & Thomas, 1993). For the DCN, six sharpened guide tubes penetrated the dura in order to avoid electrode deviation (Soteropoulos & Baker, 2006). Spike waveforms (300 Hz–10 kHz bandpass) were sampled continuously at 25 kHz, and saved to hard disk together with lever position, LFPs (bandpass, 1–100 Hz; sampling rate, 500 Hz) and task behavioural markers. Spike occurrence times were discriminated off-line using custom-written cluster cutting software (Getspike, S.N. Baker; SpikeLab, Dyball & Bhumbra, 2003). Only clean single units with consistent wave shapes and no interspike intervals below 1 ms were used for subsequent analysis.

The different brain areas were identified by a clinical examination of unit receptive fields and by noting the motor responses to microstimulation. Neurons were identified as PTNs when they showed a constant latency antidromic response to electrical stimulation of the pyramidal tract. Identification was confirmed using a collision test (Lemon, 1984).

Power spectral analysis

To compute one-sided power spectra for LFPs, data were first split into 1.024 s long (512 sample point) non-overlapping sections taken from the hold period of the task. Analysis used only monkeys M and L; these performed the finger flexion task, where one trial yielded two sections as the hold period was 2 s long. Denoting the Fourier transform of the ith section at frequency f as Fi(f), the power spectrum P(f) was estimated by:

graphic file with name tjp0580-0801-m1.jpg (1)

Using this normalization, P(f) has units of μV2 (Press et al. 1989), permitting direct comparison of recordings from different areas.

Measure of spike train irregularity

Davies et al. (2006) have recently introduced a measure of spike train irregularity. This has advantages over alternatives such as the coefficient of variation of the interspike intervals (CV), because it is resistant to changes in firing rate over the length of data analysed. We used this measure as an initial means of assessing spiking regularity. Given a series of interspike intervals Ii, the measure IR was calculated as:

graphic file with name tjp0580-0801-m2.jpg (2)

Davies et al. (2006) showed that IR can be modulated in a task-dependent way. Because the aim was to compare cells from different areas, only intervals falling in the hold phase of the task were used in the calculation. This provided a single summary statistic of the spiking irregularity for that cell under standardized behavioural conditions.

Calculation of distance-to-threshold trajectory

Figure 1 shows the steps involved in calculating the distance-to-threshold trajectory using methods described in more detail by Wetmore & Baker (2004). First the population interspike interval (ISI) histogram (Fig. 1A, binwidth 1 ms) is used to calculate the population death rate curve (Fig. 1B). The death rate is the probability that an interval will end during a particular time bin after the previous spike, given that it has not already ended by the start of the bin. The death rate for a Poisson process, for example, is constant. The death rate curve is then transformed into the distance-to-threshold trajectory (Fig. 1C), using a monotonic transform determined on the basis of simulations of an integrate-and-fire model neuron. This transform allows the distance-to-threshold trajectory to be scaled in terms of the standard deviation of the membrane voltage noise (‘noise units’). It reflects the excitability of the cell membrane as a function of time after the spike.

Figure 1. Steps involved in distance-to-threshold calculation for an example cell from area 2.

Figure 1

Open in a new tab

A, interspike interval histogram for all recorded spikes. B, death rate curve, calculated from A, showing the fraction of surviving intervals that are terminated by a spike within each bin. Thin lines indicate 95% confidence limits. C, distance-to-threshold trajectory, calculated from B using a transform derived from a computational model. D, composite distance-to-threshold trajectory, calculated by combining multiple trajectories from intervals occurring at different firing rates. E, firing rate histogram for all recorded spikes. Shaded area indicates spikes used in composite analysis. F, interspike interval histograms for three subpopulations of spikes at different estimated firing rates, G, death rate curves and H, distance-to-threshold trajectories, calculated from the subpopulation interspike interval histograms shown in F.

However, the population trajectory shows a negative slope at longer latencies, which is a sign of the population intervals being drawn from a mixture of underlying firing rates (Matthews, 1996; Wetmore & Baker, 2004). To prevent this, subpopulations of intervals with similar firing rates were selected. The instantaneous firing frequency of each interval was calculated from the mean of the four preceding and four following intervals; the distribution of this firing frequency estimate is shown in Fig. 1E. Intervals that fell in the tails of the firing rate distribution (unshaded part of curve in Fig. 1E) were excluded from further analysis. The central part of the rate distribution (shaded black in Fig. 1E) was divided into between six and 50 subpopulations. The number of subpopulations, and the cut off point for the tails, were chosen to maximize the number of subpopulations while ensuring that each contained at least 500 intervals. For each subpopulation, the ISI histogram, death rate curve and distance-to-threshold trajectory were calculated (Fig. 1F–H for three subpopulations). These distance-to-threshold trajectories were combined to form the composite trajectory (Fig. 1D). Full details of this combination and the calculation of confidence limits on the trajectory are given by Wetmore & Baker (2004).

Significance testing of peaks

Several distance-to-threshold trajectories exhibited peaks (Wetmore & Baker, 2004). The significance of these peaks was tested by comparing the peak to the surrounding points (usually the 3rd to 10th points before and after the peak). Denoting trajectory estimates and their 95% confidence limits at two postspike latencies as x ± σx and y ± σy, the points were considered significantly different (P < 0.05) if:

graphic file with name tjp0580-0801-m3.jpg (3)

To correct for multiple comparisons, a binomial distribution with the probability of single trial success P(hit) = 0.05 was used to calculate the minimum number of significant points needed before the peak was considered statistically significant.

Histology

At the end of experiments, monkeys were deeply anaesthetized (60 mg kg−1 pentobarbitone) and perfused through the heart with phosphate-buffered saline (pH 7.2) followed by 4% formal saline fixative. For monkey M, 50 μm sagital sections of the sensorimotor cortex were cut and stained with cressyl violet. These were used to check the location of the different cortical areas and to estimate the depths of the cortical layers in area 2. In three animals, the location of pyramidal tract stimulating electrodes was confirmed histologically.

Results

LFP oscillations

Figure 2 illustrates the differences in LFP oscillations between the five areas recorded for monkey M. Representative traces from each of the areas are shown in Fig. 2A. These traces have been centred on the task-hold period, because this has the strongest ∼20 Hz oscillations (Baker et al. 1997). Even in these raw data, clear differences are evident in the amplitude of oscillations between different brain regions.

Figure 2. Local field potential (LFPs) recorded from different areas.

Figure 2

Open in a new tab

A, examples of raw LFPs. Arrow indicates end of hold period. B, averaged LFP power spectra for different areas (monkey M only). C, box plots of ∼20 Hz power in different areas recorded from monkey M (thick lines) and monkey L (thin lines). Whiskers indicate minimum and maximum 20 Hz power. D, box plots of 20 Hz power in area 2 grouped based on estimated cortical layer (monkey M only).

These differences are quantified in Fig. 2B. Power spectra were computed using only recordings from the hold phase of the task; the figure shows spectra averaged over all available recordings from each area in monkey M. All cortical LFPs show a prominent peak at 20 Hz, which is not seen in the cerebellar recordings. However, within the cortical areas there are considerable differences in the size of this peak. It is surprising that although much previous work has concentrated on oscillations in M1, this has the smallest beta-band cortical oscillations.

In order to compare the different areas, power in the 17.5–22.5 Hz range was summed for each recording site. Figure 2C presents the distribution of this power as a box plot; results for monkeys M and L are shown side by side. The 20 Hz oscillations were stronger in the postcentral areas 2 and 5 than in the precentral M1 (Kruskal–Wallis test, P < 0.05, Tukey–Kramer post hoc test, P < 0.05).

It is of considerable interest to know how oscillation strength varies with cortical layer. Unfortunately, the folded nature of the cortex in this region makes it difficult to reconstruct, in our chronic recordings, the cortical layer with any confidence. Small surface deviations of the electrodes could lead to considerable differences in placement at the depth of a sulcus, and the fine nature of the electrodes precludes histological reconstruction of individual tracks (Mountcastle et al. 1991). However, area 2 lies on the brain surface, mid-way between central and intraparietal sulci, so that it is possible to make approximate assignments of cortical laminae to the recording sites. To do this, we used the depth of electrodes below the cortical surface noted during the experiment, and the thickness of layers measured in post mortem histology. Figure 2D shows the distributions of ∼20 Hz power for the different layers. There were significant differences between layers (Kruskal–Wallis test, P < 0.05). Oscillations in layer V had significantly higher amplitude than those in layers II/III (Tukey–Kramer post hoc test, P < 0.05).

Unit firing properties

A total of 370 M1 cells (146 PTNs and 224 unidentified neurons; UIDs) and 230 DCN cells were recorded from monkeys E, T, M and L. A further 144 area 3a cells, 92 area 2 cells and 82 area 5 cells were recorded from monkeys M and L. The mean number of spikes per cell was 89 777 (range, 10 065–584 678) and the mean recording duration per cell was 4011 s (range, 515–10 259 s). PTNs and UIDs from M1 were treated separately in the following analyses.

Figure 3A shows the distribution of the mean firing rate during the task-hold phase, for the different cell classes. In all of the cortical areas recorded, neurons fired close to 20 Hz. The similarity of the firing rate to the beta-band frequency may make it easier for cells to phase-lock to the synchronous oscillations. Cells in the DCN fired at significantly higher rates (Kruskal–Wallis test, P < 0.05; Tukey–Kramer post hoc test, P < 0.05).

Figure 3. Differences in firing rate and irregularity between different cell classes.

Figure 3

Open in a new tab

A, box plots showing the population distribution of the mean firing rate during the task-hold period. Dotted line indicates 20 Hz. B, box plots of the mean irregularity (calculated as described in the text) during the hold period. Whiskers indicate extrema in both panels. Results for M1 pyramidal tract neurons (PTNs) and unidentified cells (UIDs) are presented separately. DCN, deep cerebellar nuclei.

Figure 3B presents the distribution of the spike train irregularity, calculated as described in the Methods. Both M1 PTNs and M1 UIDs showed significantly lower irregularity than the other areas (Kruskal–Wallis test, P < 0.05; Tukey–Kramer post hoc test, P < 0.05).

Distance-to-threshold trajectories

Figure 4 shows the ISI histograms, death rate curves and composite trajectories for six example cells; one from each area. These illustrate two of the three trajectory categories described by Wetmore & Baker (2004): significant peaks (Fig. 4A–C) and exponentially increasing trajectories (Fig. 4D–F). Cells with significant peaks have a tendency to fire rhythmically at a particular frequency. These two categories accounted for over 90% of all the cells analysed. A few cells in each area had exponentially decreasing trajectories (29/928 cells) or did not fit any of these patterns (62/928 cells). The area 3a cell illustrated had a large number of short intervals (1–4 ms; Fig. 4Ca) suggesting a tendency to burst. This is a common feature of S1 cells (see Baker et al. 2003a). However, the bursting was taken account of by the distance-to-threshold analysis, which yielded a trajectory with larger values just after time zero than a little later (Fig. 4Cc). This represents the increased excitability at short times after the previous spike, presumably caused by intrinsic depolarizing inward currents.

Figure 4. Example distance-to-threshold trajectories for cells in different classes.

Figure 4

Open in a new tab

A–F, results from single cells, recorded from the site as labelled on the left. a, whole population interspike interval histograms, b, corresponding population death rate curves and c, composite distance-to-threshold trajectories. Thin lines in b and c indicate 95% confidence limits. Cells A–C have significant peaks in their trajectories.

The proportion of each trajectory type varied significantly between the six recorded regions (χ2 test, P < 0.001). Almost 80% of M1 PTNs had significant peaks in their composite trajectories (115/146 cells) compared with approximately 40% of M1 UIDs, area 3a cells and DCN cells (104/224, 58/144 and 92/230 cells, respectively) and less than 20% of area 2 and area 5 cells (15/92 and 12/82 cells, respectively).

Peak height

For all cells, peak height was measured from the composite distance-to-threshold trajectory. This was calculated as the greatest height difference between the highest point of the trajectory, and the region 3–10 ms later. Figure 5 shows, for each brain region, the distribution of peak heights. Filled bars indicate those cells with a significant peak, assessed as described in the Methods. Most of the significant peaks were of moderate height (0.284 ± 0.013 noise units; median ± standard error of the median) although six cells did have peaks larger than 1 noise unit (one in area 3a, one in area 2 and four in DCN). The distribution for M1 PTNs was clearly shifted to the right compared to the other regions (median peak height across all cells: M1 PTNs, 0.282 ± 0.015; M1 UIDs, 0.151 ± 0.011; area 3a, 0.168 ± 0.026; area 2, 0.075 ± 0.019; area 5, 0.084 ± 0.038; DCN, 0.181 ± 0.021 noise units). M1 PTNs had significantly larger peaks than all other cell classes (Kruskal–Wallis test, P < 0.05; Tukey–Kramer post hoc test, P < 0.05).

Figure 5. Histograms showing the distribution of peak height in the composite distance-to-threshold trajectories for different cell classes.

Figure 5

Open in a new tab

Open bars show distribution of peak height of all cells, filled bars show cells with peaks that reached statistical significance.

Peak latency

In agreement with the work of Wetmore & Baker (2004), the peak latency showed considerable variability across the recorded cell population (Fig. 6A). M1 cells (both PTNs and UIDs) had significantly longer peak latency (PTN, 37.1 ± 0.8 ms; UID, 36.6 ± 1.1 ms) than area 3a cells (30.9 ± 1.5 ms) and DCN cells (19.0 ± 0.9 ms). Area 3a cells also had significantly longer peak latency than DCN cells (Kruskal–Wallis test, P < 0.05; Tukey–Kramer post hoc test, P < 0.05).

Figure 6. Variation in the latency of peaks in the composite distance-to-threshold trajectories.

Figure 6

Open in a new tab

A, box plot of peak latencies for the different cell classes (only those with significant peaks in their distance-to-threshold trajectories were included). Whiskers represent extrema. B, scatter plot of peak latency versus median interspike interval. Different cell classes have been indicated with different symbols. Thick line, regression fit; thin line, identity.

In order to determine whether the peak latency influenced the preferred firing rate of the cell, we correlated the peak latency with the median interspike interval measured over all recorded spikes (Fig. 6B). This showed a strong correlation (regression slope, 0.54 ± 0.07, 95% confidence limit). However, the slope was significantly different from unity, indicating that peak latency increased more steeply than median interspike interval. Wetmore & Baker (2004) reported a similar correlation for their smaller dataset of M1 neurons.

Dependence of trajectory peaks on network oscillations

It is attractive to assume that the peaks in distance-to-threshold trajectories reflect the action of conductances intrinsic to the cells. However, our recordings contained robust network oscillations, at a frequency similar to the periodicity implied by the peak latency (Fig. 2B cf. Fig. 6A). Trajectory peaks could therefore simply represent cells synchronizing to the local beta-frequency oscillations (Wetmore & Baker, 2004). To investigate this possibility, we calculated the trajectories separately for task periods with high (hold period) and low (movement phase) beta-power in the LFP (Fig. 7A and B). Only cells that previously showed significant peaks were used.

Figure 7. Changes in the distance-to-threshold trajectories in periods of low and high 20 Hz network oscillations.

Figure 7

Open in a new tab

A, overlain lever displacement for 25 trials performed while recording in area 3a. Note the relatively constant displacement during the hold phase (−2 to 0 s). B, time-resolved local field potential power recorded in same session. A and B are aligned to the end of the hold phase (time 0 s). C, distance-to-threshold trajectories for spikes occurring during periods of high 20 Hz power (red) and low 20 Hz power (blue) from a simultaneously recorded cell. The high-power period was chosen to encompass the 2 s long hold phase. The low-power period ended at the start of the hold period; its duration was chosen to include the same number of spikes as recorded during the high-power period. Thin lines indicate 95% confidence limits. D, results for all cells from M1, area 3a and DCN which had significant peaks in their distance-to-threshold trajectories computed using all spikes, and > 10 000 spikes in both movement and hold phases. Coloured bars show the fraction of cells with significant peaks during only the period with low 20 Hz power (blue) or only for high 20 Hz power (red). Grey bars show fraction of cells with a larger peak during the period with greatest 20 Hz power (dark grey) or least 20 Hz power (light grey).

One problem in such an analysis is that the number of spikes used to calculate the trajectory substantially affects the statistical significance of the peak. Because units modulate their activity with task performance, this will lead to different statistical power to detect trajectory peaks in different task phases. To overcome this, the length of the task period with most spikes was shortened until the number of spikes matched those in the phase with lower firing rate. Figure 7C shows an example of trajectories at high and low beta-power for an area 3a cell. The two trajectories have very similar-sized peaks, which are significant in both cases.

Figure 7D shows the results across a population of M1 PTNs, M1 UIDs, area 3a cells and DCN cells. Area 2 and area 5 cells were not included in this analysis because of the low number of cells with significant peaks. Very few cells showed peaks in one task period and not the other (blue or red bars in Fig. 7D). Among the cortical cells, there was a slight bias towards larger peaks during the periods of low beta-power – the opposite of what would be expected if trajectory peaks were caused solely by network oscillations. This may be because larger peaks occur at high firing rates, as shown by Wetmore & Baker (2004), and rates were usually higher during the movement phase of the task. This tendency was less obvious in the DCN, where there was both less modulation in ∼20 Hz LFP power, and less modulation in firing rates, between movement and hold phases of the task.

Becaue there was no trend for peaks to be larger during periods of high ∼20 Hz network oscillations, trajectory peaks are probably the result of processes intrinsic to the single neuron.

Identification of neuron type by spike width

Lebedev & Nelson (1995) described a subset of S1 neurons that fired rhythmically during rest periods; they proposed that these were inhibitory interneurons. It is possible to distinguish some groups of inhibitory interneurons (fast spiking) from other groups of cortical cells (pyramidal cells and spiny stellate cells) by measuring the duration of the extracellularly recorded action potential (Swadlow, 1989, 1995). A suggested dividing line between the two distributions is 0.6 ms, although there is considerable overlap. We used this approach to see whether any of the unidentified cortical cells that we recorded were likely to be interneurons. Spike width was measured as illustrated in Fig. 8A; the distributions for different cortical areas are shown in Fig. 8B. The values for M1 PTNs (top panel in Fig. 8B) serve as a useful standard, because we know that these antidromically identified output neurons are pyramidal cells. As expected, all of these neurons have a spike width > 0.6 ms. For other regions, the distributions were centred at spike durations > 1.0 ms, overlapping with the M1 PTNs. This implies that the majority of cells were either pyramidal cells or spiny stellate neurons. There was no significant difference between the spike duration for cells with and without peaked trajectories (two-way ANOVA, P > 0.05). All the cells with peaked trajectories had durations above the 0.6 ms boundary, making it unlikely that they are fast-spiking inhibitory interneurons.

Figure 8. Spike width for cortical cells.

Figure 8

Open in a new tab

A, schematic illustration of the measurement of spike width, taken as the sum of the duration of positive and negative phases of the action potential. B, distributions of spike widths for different groups of cortical cells. Open bars represent spike widths of all cells, filled bars represent cells with significant peaks in their distance-to-threshold trajectories. Dotted line represents the 0.6 ms boundary commonly used to separate pyramidal cells from cortical interneurons.

Analysis of S1 PTNs

M1 PTNs have a higher tendency for rhythmic firing than other cortical cells (Fig. 5). It is possible that this is a property of all PTNs, including those in S1, and not just those in M1. PTNs are substantially harder to locate and record in S1 than M1. There are only a few reports of S1 PTN activity in primates (Fromm & Evarts, 1982), even though parietal cortex accounts for up to 40% of the cells of origin of the pyramidal tract (Philips & Porter, 1977). In our recordings we were fortunate to record two S1 PTNs, one from area 3a and one from area 2. Figure 9 presents data from these two neurons, which were activated at fixed latency from the stimulating electrodes in the pyramidal tract (Fig. 9Aa and Ba). Spontaneous spikes were capable of colliding with evoked spikes if the stimulus was appropriately timed (Fig. 9Ab and Bb), confirming the antidromic nature of the activation and the identification as PTNs.

Figure 9. Pyramidal tract neurons (PTNs) recorded from S1.

Figure 9

Open in a new tab

A and B, PTNs recorded from area 3a (A) and area 2 (B). a, overlain traces showing antidromic response to pyramidal tract stimulation with fixed latency. b, collision test. Upper sweep shows collision of the antidromic response with a spontaneous spike; lower sweep shows the failure of this collision when the spontaneous spike-to-stimulus interval was increased by 0.1 ms c, composite distance-to-threshold trajectories. Thin lines indicate 95% confidence limits.

Figure 9Ac and Bc shows the composite distance-to-threshold trajectories for these two neurons. The trajectory of the area 3a PTN had a significant peak, but the height (0.21 noise units) was low compared to the average peak height of M1 PTNs (0.34 ± 0.03 noise units). The area 2 PTN showed no significant peak. Although these data are clearly limited, they suggest that S1 PTNs show trajectories similar to adjacent UIDs from the same cortical area, rather than behaving like M1 PTNs.

Discussion

In this paper we report two principal findings. Firstly, LFP oscillations were actually stronger in S1 and posterior parietal cortex than in M1. Secondly, the proportion of neurons with peaked distance-to-threshold trajectories varied considerably between the different areas, with a particularly high proportion of peaks in M1 PTNs.

Functional significance of oscillations in somatosensory cortex

The present findings extend and support recent work on beta-band oscillations in the somatosensory system. Riddle & Baker (2005) argued that corticomuscular coherence is likely to be generated by sensory feedback pathways as well as descending motor tracts. Oscillations synchronized with those in EMG appear in peripheral afferents (Baker et al. 2006). In addition, S1 oscillations show strong Granger causal influences over those in M1 (Brovelli et al. 2004). Oscillatory networks therefore form a fully connected loop from cortex to the periphery and back. Such a system would be well suited to some form of proprioceptive processing, as suggested by Riddle & Baker (2006). It is interesting that lower frequency (∼10 Hz) oscillations have been previously suggested to play a role in decoding tactile information in both monkey hand (Ahissar & Vaadia, 1990) and rodent whisker systems (Ahissar et al. 1997).

Dissociation of 20 Hz oscillations and intrinsic rhythmicity

Surprisingly, we have shown a clear difference between the areas with large beta-oscillations and those with a high proportion of peaked distance-to-threshold trajectories. The lack of peaks in areas 2 and 5 may be due to preferentially sampling the upper layers (II–IV) of the cortex rather than layer V. However, the one area 2 PTN recorded had no trajectory peak, and cells with significant peaks were evenly distributed throughout the layers (results not shown). The small proportion of neurons in areas 2 and 5 with peaks therefore suggests that intrinsic rhythmicity is not necessary for the generation and maintenance of strong oscillations. This agrees with much previous work suggesting that oscillations are generated by local networks of inhibitory interneurons (Wang & Buzsáki, 1996; Pauluis et al. 1999; Traub et al. 1999).

Peaks in PTN distance-to-threshold trajectories

Although significantly peaked distance-to-threshold trajectories were observed in all areas recorded, they were substantially more common among the motor cortical PTNs than all other cell classes. The distribution of peak heights for PTNs (Fig. 5) was unimodal, suggesting that intrinsic rhythmicity was a common feature even for cells where peaks failed to reach statistical significance. The absence of a peak in the single area 2 PTN recorded implies that this may be a feature particularly related to motor cortical PTNs. This could reflect the different functions of M1 and S1 PTNs; however, a larger database of S1 PTNs is needed before definitive conclusions can be drawn.

In a previous study we reported that PTN spiking was coherent with LFP oscillations in M1; however, the size of this coherence was low (Baker et al. 2003b). Using a computational model, we showed (Baker et al. 2003b) that the non-linearity inherent in spike generation markedly impairs the representation of oscillations in neural spiking, and therefore attenuates coherence. Even if 20% of the inputs to a simple integrate-and-fire model neuron are coherent with LFP oscillations, the coherence with output spiking is as low as 0.05. If PTNs have a tendency towards rhythmic discharge at the same frequency as the local network oscillations, this will increase the reliability with which these oscillations can be represented in their spike output. As PTNs are the output neurons of M1, transmission of the oscillatory signal to the spinal motoneurons will consequently be improved, boosting corticomuscular coherence. The preferential association of peaked trajectories with M1 PTNs therefore provides indirect evidence that the transmission of oscillations to the periphery has a functional role in motor control. By contrast, the processing of ascending oscillatory information by somatosensory areas does not appear to require an intrinsic tendency towards rhythmicity.

Rhythmicity and DCN neurons

It has previously been shown, both in vivo and in vitro, that cells in DCN fire spontaneously in a rhythmic pattern (Thach, 1968; Jahnsen, 1986). This is reflected by the relatively high proportion of DCN cells with peaked distance-to-threshold trajectories in the current dataset (92/230 cells). Beta-frequency oscillations have previously been reported in the cerebellar cortex and DCN (Aumann & Fetz, 2004; Courtemanche & Lamarre, 2005; Soteropoulos & Baker, 2006). However, the short latency of the trajectory peaks (19.0 ± 0.9 ms; significantly shorter than the other areas) makes it difficult to link them to the oscillations. The major input to the DCN, from the cerebellar cortex, is inhibitory. The function of intrinsic conductances that produce spontaneous, rhythmic firing may instead be to allow synaptic inputs to modulate the activity in either direction (Jahnsen, 1986).

Conclusions

The present results support the idea that beta-band oscillations have a role in sensorimotor integration, rather than purely in motor control (Riddle & Baker, 2006). Oscillations were present in both motor and somatosensory cortices, but stronger in the latter. The generation of these oscillations clearly does not require the intrinsic tendency to rhythmic firing at beta-band frequencies which we observed as peaks in the distance-to-threshold trajectories. However, the high incidence of such peaks in M1 PTNs will improve the fidelity of transmission of oscillations to the periphery. This argues that oscillatory coupling between cortex and motoneurons is functionally important, rather than a simple side effect of a primarily cortical oscillatory phenomenon. We also concluded previously (Baker & Baker, 2003), from a very different study, that corticomuscular coherence was likely to be playing a functional role over and above any assigned to cortical oscillations. The exact nature of the oscillatory interplay between cortex and periphery, and the specific advantage that it confers on sensorimotor processing, remains to be elucidated.

Acknowledgments

We would like to thank Karen Fisher, Terri Jackson, Rosie Pyper, Shelley Rhodes, Demetris Soteropoulos, Minyan Wang and Daniel Wetmore for assistance with experiments and Daniel Wetmore for providing the code for the trajectory analysis. This work was funded by The Wellcome Trust.

References

  1. Ahissar E, Haidarliu S, Zacksenhouse M. Decoding temporally encoded sensory input by cortical oscillations and thalamic phase comparators. Proc Natl Acad Sci U S A. 1997;94:11633–11638. doi: 10.1073/pnas.94.21.11633. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Ahissar E, Vaadia E. Oscillatory activity of single units in a somatosensory cortex of an awake monkey and their possible role in texture analysis. Proc Natl Acad Ssci U S A. 1990;87:8935–8939. doi: 10.1073/pnas.87.22.8935. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Aumann TD, Fetz EE. Oscillatory activity in forelimb muscles of behaving monkeys evoked by microstimulation in the cerebellar nuclei. Neurosci Lett. 2004;361:106–110. doi: 10.1016/j.neulet.2003.12.091. [DOI] [PubMed] [Google Scholar]
  4. Baker MR, Baker SN. The effect of diazepam on motor cortical oscillations and corticomuscular coherence studied in man. J Physiol. 2003;546:931–942. doi: 10.1113/jphysiol.2002.029553. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Baker SN, Chiu M, Fetz EE. Afferent encoding of central oscillations in the monkey arm. J Neurophysiol. 2006;95:3904–3910. doi: 10.1152/jn.01106.2005. [DOI] [PubMed] [Google Scholar]
  6. Baker SN, Curio G, Lemon RN. EEG oscillations at 600 Hz are macroscopic markers for cortical spike bursts. J Physiol. 2003a;550:529–534. doi: 10.1113/jphysiol.2003.045674. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Baker SN, Olivier E, Lemon RN. Coherent oscillations in monkey motor cortex and hand muscle EMG show task-dependent modulation. J Physiol. 1997;501:225–241. doi: 10.1111/j.1469-7793.1997.225bo.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Baker SN, Philbin N, Spinks R, Pinches EM, Wolpert DM, MacManus DG, Pauluis Q, Lemon RN. Multiple single unit recording in the cortex of monkeys using independently moveable microelectrodes. J Neurosci Methods. 1999;94:5–17. doi: 10.1016/s0165-0270(99)00121-1. [DOI] [PubMed] [Google Scholar]
  9. Baker SN, Pinches EM, Lemon RN. Synchronization in monkey motor cortex during a precision grip task. II. Effect of oscillatory activity on corticospinal output. J Neurophysiol. 2003b;89:1941–1953. doi: 10.1152/jn.00832.2002. [DOI] [PubMed] [Google Scholar]
  10. Baker SN, Spinks R, Jackson A, Lemon RN. Synchronization in monkey motor cortex during a precision grip task. I. Task-dependent modulation in single-unit synchrony. J Neurophysiol. 2001;85:869–885. doi: 10.1152/jn.2001.85.2.869. [DOI] [PubMed] [Google Scholar]
  11. Brovelli A, Ding M, Ledberg A, Chen Y, Nakamura R, Bressler SL. Beta oscillations in a large-scale sensorimotor cortical network: directional influences revealed by Granger causality. Proc Natl Acad Sci U S A. 2004;101:9849–9854. doi: 10.1073/pnas.0308538101. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Chen D, Fetz EE. Characteristic membrane potential trajectories in primate sensorimotor cortex neurons recorded in vivo. J Neurophysiol. 2005;94:2713–2725. doi: 10.1152/jn.00024.2005. [DOI] [PubMed] [Google Scholar]
  13. Conway BA, Halliday DM, Farmer SF, Shahani U, Maas P, Weir AI, Rosenberg JR. Synchronization between motor cortex and spinal motoneuronal pool during the performance of a maintained motor task in man. J Physiol. 1995;489:917–924. doi: 10.1113/jphysiol.1995.sp021104. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Courtemanche R, Lamarre Y. Local field potential oscillations in primate cerebellar cortex: synchronization with cerebral cortex during active and passive expectancy. J Neurophysiol. 2005;93:2039–2052. doi: 10.1152/jn.00080.2004. [DOI] [PubMed] [Google Scholar]
  15. Davies RM, Gerstein GL, Baker SN. Measurement of time-dependent changes in the irregularity of neural spiking. J Neurophysiol. 2006;96:906–918. doi: 10.1152/jn.01030.2005. [DOI] [PubMed] [Google Scholar]
  16. Dyball REJ, Bhumbra GS. Digital spike discrimination combining size and shape elements. J Physiol. 2003;547.P:D9. [Google Scholar]
  17. Eckhorn R, Thomas U. A new method for the insertion of multiple microprobes into neural and muscular tissue, including fiber electrodes, fine wires, needles and microsensors. J Neurosci Methods. 1993;49:175–179. doi: 10.1016/0165-0270(93)90121-7. [DOI] [PubMed] [Google Scholar]
  18. Fromm C, Evarts EV. Pyramidal tract neurons in somatosensory cortex: central and peripheral inputs during voluntary movement. Brain Res. 1982;238:186–191. doi: 10.1016/0006-8993(82)90781-8. [DOI] [PubMed] [Google Scholar]
  19. Gerloff C, Braun C, Staudt M, Hegner YL, Dichgans J, Krageloh-Mann I. Coherent corticomuscular oscillations originate from primary motor cortex: evidence from patients with early brain lesions. Hum Brain Mapp. 2006;27:789–798. doi: 10.1002/hbm.20220. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Gray CM, McCormick DA. Chattering cells: superficial pyramidal neurons contributing to the generation of synchronous oscillations in the visual cortex. Science. 1996;274:109–113. doi: 10.1126/science.274.5284.109. [DOI] [PubMed] [Google Scholar]
  21. Jahnsen H. Electrophysiological characteristics of neurons in the guinea-pig deep cerebellar nuclei in vitro. J Physiol. 1986;372:129–147. doi: 10.1113/jphysiol.1986.sp016001. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Kilner JM, Baker SN, Salenius S, Jousmaki V, Hari R, Lemon RN. Task-dependent modulation of 15–30 Hz coherence between rectified EMGs from human hand and forearm muscles. J Physiol. 1999;516:559–570. doi: 10.1111/j.1469-7793.1999.0559v.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Lebedev MA, Nelson RJ. Rhythmically firing (20-50 Hz) neurons in monkey primary somatosensory cortex: activity patterns during initiation of vibratory-cued hand movements. J Comput Neurosci. 1995;2:313–334. doi: 10.1007/BF00961443. [DOI] [PubMed] [Google Scholar]
  24. Lemon RN. Methods for neuronal recording in conscious animals. In IBRO. In: Smith AD, editor. Handbook Series: Methods in Neuroscience 4. London, UK: Wiley; 1984. pp. 1–162. [Google Scholar]
  25. MacKay WA, Mendonca AJ. Field potential oscillatory bursts in parietal cortex before and during reach. Brain Res. 1995;704:167–174. doi: 10.1016/0006-8993(95)01109-9. [DOI] [PubMed] [Google Scholar]
  26. Matthews PB. Relationship of firing intervals of human motor units to the trajectory of post-spike after-hyperpolarization and synaptic noise. J Physiol. 1996;492:597–628. doi: 10.1113/jphysiol.1996.sp021332. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Mountcastle VB, Reitboeck HJ, Poggio GF, Steinmetz MA. Adaptation of the Reitboeck method of multiple microelectrode recording to the neocortex of the waking monkey. J Neurosci Methods. 1991;36:77–84. doi: 10.1016/0165-0270(91)90140-u. [DOI] [PubMed] [Google Scholar]
  28. Murthy VN, Fetz EE. Coherent 25- to 35-Hz oscillations in the sensorimotor cortex of awake behaving monkeys. Proc Natl Acad Sci U S A. 1992;89:5670–5674. doi: 10.1073/pnas.89.12.5670. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Pauluis Q, Baker SN, Olivier E. Emergent oscillations in a realistic network: the role of inhibition and the effect of the spatiotemporal distribution of the input. J Comput Neurosci. 1999;6:27–48. doi: 10.1023/a:1008804916112. [DOI] [PubMed] [Google Scholar]
  30. Philips CG, Porter R. Corticospinal Neurons: Their Role in Movement. London: Academic Press; 1977. [Google Scholar]
  31. Press WH, Flannery BP, Teukolsky SA, Vetterling WT. Numerical Recipies in Pascal. Cambridge: Cambridge University Press; 1989. [Google Scholar]
  32. Riddle CN, Baker SN. Manipulation of peripheral neural feedback loops alters human corticomuscular coherence. J Physiol. 2005;566:625–639. doi: 10.1113/jphysiol.2005.089607. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Riddle CN, Baker SN. Digit displacement, not object compliance, underlies task dependent modulations in human corticomuscular coherence. Neuroimage. 2006;33:618–627. doi: 10.1016/j.neuroimage.2006.07.027. [DOI] [PubMed] [Google Scholar]
  34. Salenius S, Portin K, Kajola M, Salmelin R, Hari R. Cortical control of human motoneuron firing during isometric contraction. J Neurophysiol. 1997;77:3401–3405. doi: 10.1152/jn.1997.77.6.3401. [DOI] [PubMed] [Google Scholar]
  35. Salmelin R, Hari R. Spatiotemporal characteristics of sensorimotor neuromagnetic rhythms related to thumb movement. Neuroscience. 1994;60:537–550. doi: 10.1016/0306-4522(94)90263-1. [DOI] [PubMed] [Google Scholar]
  36. Sanes JN, Donoghue JP. Oscillations in local field potentials of the primate motor cortex during voluntary movement. Proc Natl Acad Sci U S A. 1993;90:4470–4474. doi: 10.1073/pnas.90.10.4470. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Soteropoulos DS, Baker SN. Cortico-cerebellar coherence during a precision grip task in the monkey. J Neurophysiol. 2006;95:1194–1206. doi: 10.1152/jn.00935.2005. [DOI] [PubMed] [Google Scholar]
  38. Swadlow HA. Efferent neurons and suspected interneurons in S-1 vibrissa cortex of the awake rabbit: receptive fields and axonal properties. J Neurophysiol. 1989;62:288–308. doi: 10.1152/jn.1989.62.1.288. [DOI] [PubMed] [Google Scholar]
  39. Swadlow HA. Influence of VPM afferents on putative inhibitory interneurons in S1 of the awake rabbit: evidence from cross-correlation, microstimulation, and latencies to peripheral sensory stimulation. J Neurophysiol. 1995;73:1584–1599. doi: 10.1152/jn.1995.73.4.1584. [DOI] [PubMed] [Google Scholar]
  40. Thach WT. Discharge of Purkinje and cerebellar nuclear neurons during rapidly alternating arm movements in the monkey. J Neurophysiol. 1968;31:785–797. doi: 10.1152/jn.1968.31.5.785. [DOI] [PubMed] [Google Scholar]
  41. Traub RD, Jeffreys JGR, Whittington MA. Fast Oscillations in Cortical Circuits. Cambridge, MA: MIT Press; 1999. [Google Scholar]
  42. Wang X-J, Buzsáki G. Gamma oscillation by synaptic inhibition in a hippocampal interneuronal network model. J Neurosci. 1996;16:6402–6413. doi: 10.1523/JNEUROSCI.16-20-06402.1996. [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Wetmore DZ, Baker SN. Post-spike distance-to-threshold trajectories of neurons in monkey motor cortex. J Physiol. 2004;555:831–850. doi: 10.1113/jphysiol.2003.048918. [DOI] [PMC free article] [PubMed] [Google Scholar]

Articles from The Journal of Physiology are provided here courtesy of The Physiological Society

RESOURCES