- Email: [email protected]
Change in phase synchronization of local field potentials in anesthetized rats after chronic dopamine depletion
Neuroscience Research 49 (2004) 179–184
Change in phase synchronization of local field potentials in anesthetized rats after chronic dopamine depletion Kyoung-Min Lee a,∗ , Tae-Beom Ahn a , Beom S. Jeon a , Dong Gyu Kim b a
Department of Neurology, Seoul National University, 28 Yongon-Dong Chongno-Gu, Seoul 110-744, South Korea b Department of Neurosurgery, Seoul National University, Seoul, South Korea Received 29 October 2003; accepted 16 February 2004 Available online 9 April 2004
Abstract In order to investigate temporal characteristics of oscillatory neural activity and the effect of chronic dopamine depletion on it, local field potentials were measured in anesthetized rats without or with a 6-OHDA lesion at either the ventral tegmental area or the substantia nigra compacta, using a pair of electrodes that were separated by 120 m. Coupling of neural activity in this mesoscopic scale was measured by a synchronization index that quantified the distribution of differences in instantaneous phase between the two potentials recorded. Phase synchrony was significantly stronger at deep basal ganglia sites, more so than at cortical sites, over a ␥ range (30–120 Hz) in normal rats. After chronic dopamine depletion, this synchrony was no longer observable, especially after a substantia nigra lesion, while there was an increase in phase synchrony in the cortex at a lower frequency band (10 Hz). These findings are consistent with previously reported findings that the effect of dopaminergic innervation on oscillatory neural activity varies from synchronizing to desynchronizing, depending on the structure innervated and the frequency of the oscillation. © 2004 Elsevier Ireland Ltd and The Japan Neuroscience Society. All rights reserved. Keywords: Phase synchrony; Oscillatory neural activity; Dopamine depletion; Parkinson’s disease; Rat brain
1. Introduction Alteration in oscillatory activity of neurons or neuronal assembles has been implicated in the pathogenesis of Parkinson’s disease (PD). It has been demonstrated that periodic, oscillatory bursting in neurons at the internal segment of the globus pallidus (GPi) of MPTP-treated monkeys and patients with PD is synchronized (Nini et al., 1995; Bergman et al., 1998; Hurtado et al., 1999; Raz et al., 2000). Similarly, high-frequency oscillations were reported in the subthalamic nucleus (STN) of PD patients with tremor (Levy et al., 2000), and enhanced synchronized population bursting was observed in the primary motor cortex of the MPTP primate model (Goldberg et al., 2002). The origin of such burst firing, oscillations, or neuronal synchronization is not entirely clear. They can arise through the properties of individual cells and/or network properties of neuronal ensembles. On the one hand, synchronized ac∗ Corresponding author. Tel.: +1-82-2-760-2985; fax: +1-82-2-3672-7553. E-mail addresses: [email protected], [email protected] (K.-M. Lee).
tivity was not observed in the GPi, the external segment of the globus pallidus (GPe), or the substantia nigra pars reticulata (SNr) of non-tremulous PD patients (Levy et al., 2002), suggesting that overt neuronal synchronization is not intrinsic to the parkinsonian basal ganglia. It has been postulated that the coupled STN–GP network constitutes a central pattern generator that is able to maintain synchronized burst discharges (Plenz and Kital, 1999), upon which the cortex provides an oscillatory drive (Magill et al., 2000). After dopamine depletion, oscillatory cortical activity is expressed more powerfully by the STN–GP network, and the manner in which cortical information is processed by the network is altered profoundly (Magill et al., 2001). On the other hand, a small proportion of STN and GP cells show oscillation in the absence of cortical input, suggesting that either a local network of STN–GP neurons confined within the basal ganglia or intrinsic neuronal properties are able to sustain such activity (Magill et al., 2001). In a recent experiment where whole-cell patch-clamp techniques were used to record from single and pairs of GP cells, it was observed that type A neurons of the GP independently fire at a frequency determined by intrinsic subthreshold membrane
0168-0102/$ – see front matter © 2004 Elsevier Ireland Ltd and The Japan Neuroscience Society. All rights reserved. doi:10.1016/j.neures.2004.02.008
180
K.-M. Lee et al. / Neuroscience Research 49 (2004) 179–184
oscillations, while GABA inhibitory postsynaptic potentials can produce rebound spiking that promotes coherent firing in multiple GP neurons through a phase realignment of this intrinsic oscillatory activity (Stanford, 2003). In view of these issues on the synchronization of neuronal activity in Parkinson’s disease, the main objectives of this study were: (1) To examine the temporal dynamics of phase synchronization at various structures in the brain, especially focusing on the synaptic input activity of each region. While spikes mainly represent the output activity of neural elements (Legatt et al., 1980; Freeman, 1975), local field potentials (LFPs) reflect more of input synaptic activity (Mitzdorf, 1987; Juergens et al., 1999). Therefore, we chose to examine LFPs and measure their phase synchronization. For computing the degree of phase synchrony from bivariate data obtained using a pair of electrodes, we employed an approach developed by Tass et al. (1998). (2) To assess the role of dopaminergic innervation on phase synchronization comparing normal brains and those with chronic dopamine depletion, we recorded also from a rat PD model where a 6-OHDA lesion was made either at the ventral tegmental area (VTA) or at the substantia nigra pars compacta (SNc).
2. Methods Experiments were carried out on 18 adult male Sprague– Dawley rats (weight ranging from 180 to 200 g; seven in the normal control group, five in the VTA-lesion group, and six in the SNc-lesion group) and were conducted in accordance with the institutional guidelines on animal research. 2.1. Unilateral lesion of dopaminergic neurons Anesthesia was induced with chloral hydrate (400 mg/ 100 g wt.). Twenty-five minutes before the injection of 6-OHDA, all animals received a pre-treatment with desipramine (12.5 mg/100 g wt.) to minimize the uptake of 6-OHDA by noradrenergic neurons, and five minutes before the OHDA injection, pargyline (25 mg/100 g wt.) was administered to maximize the toxic effects on dopaminergic neurons. Using a stereotaxic frame (David Kopf Instruments, Tujunga, CA, USA), 6-OHDA (8 g/4 l in 0.2% ascorbic acid solution) was injected into the VTA (at A-P 3.4 mm, left 0.8 mm, and D-V 2.2 mm from the interaural line, aimed at 10◦ angle from the vertical) or the SNc (at AP 4.2 mm, left 1.8 mm, and D-V 2.0 mm from the interaural line). The lesions were verified by tyrosine hydroxylase immunohistochemistry after the animals were sacrificed at the end of a recording session. 2.2. Recording the local field potentials Electrophysiological recordings were performed using tungsten stereotrodes (WPI, Sarasota, FL, USA, Model
#TST33A20KT, outer diameter 0.365 mm) which have two electrode tips, separated by 120 m, each with an impedance of 2.0 M. In the case of lesioned animals, the recording was performed 3–4 weeks after a lesion was made. After anesthesia was established using ketamine (bolus injection of 100 mg/kg i.p. and supplementary injection of 50 mg/kg i.m.), an animal was positioned in the same stereotaxic frame as used for lesion-making (David Kopf Instruments, Tujunga, CA, USA). A craniotomy was done and the dura was opened above the aimed recording site (at A-P −0.6 mm, lateral 3.0 mm from the bregma). After an electrode was approximately positioned above the exposed surface of the brain, it was slowly advanced downward using a hydraulic microdrive (Stoelting Co., Illinois, USA). A contact to the brain tissue was noted visually and at the same time by monitoring a sudden change in the noise pattern. The contact point was marked as a reference for the depth of subsequent recordings. The electrode was then slowly advanced downward and stopped at each millimeter up to 6 mm in depth. At least 3 min of rest was allowed for the brain tissue to settle before LFPs were recorded for half a minute. According to a rat brain atlas (Paxinos and Watson, 1986), our recorded sites were located at the frontoparietal motor cortex (CTX, for sites 1 and 2 mm deep from the surface), the caudate-putamen (CP, for sites 3, 4, and probably 5 mm deep from the surface), and the globus pallidus (GP, for sites possibly 5 and 6 mm deep from the surface) (Fig. 1). However, these localizations were approximate at best, considering expected anatomical variability across individual animals. LFP was sampled and stored at 100 kHz using TDT System 2 (Tucker-Davis Technologies, FL, USA), and then sub-sampled to 1 kHz for further offline analysis by averaging over 100 data points. The dual recordings were both effectively monopolar since a common reference lead from the pre-amplifier was connected to the subcutaneous tissue around the craniotomy site with an impedance negligible compared to that of the recording electrodes. 2.3. Computation of phase synchrony Quantification of the degree of phase synchrony between signals that were recorded simultaneously from the two electrodes of a stereotrode was performed using the following steps (Fig. 2): (1) LFPs were bandpassfiltered at 10 Hz steps from 10 to 150 Hz with a bandwidth of 5 Hz using a linear-phase FIR filter designed by constrained least-squares (Selesnick et al., 1996) implemented in MATLAB (The Mathworks Inc., Natick, MA, USA). (2) Instantaneous phases of each filtered signal were computed by the analytic signal approach using the Hilbert transform. (3) The distribution of the relative phase, Ψ = (Φ modulo 2π), between the two signals was obtained. Since our data was not necessarily stationary, we adopted a sliding window approach in which the distribu-
K.-M. Lee et al. / Neuroscience Research 49 (2004) 179–184
181
CTX CP GP
(A) 1mm Normal VTA-lesion SNc-lesion Recording Depth Channel1 Channel2 Channel1 Channel2 Channel1 Channel2 1mm
2mm
3mm
4mm
5mm
6mm 0.2 mV
(B)
0.5 sec
Fig. 1. (A) Presumed recording sites schematically depicted over a coronal plane of the rat brain at −0.6 mm from the bregma. The background was hand-drawn based on the dimensions provided in Paxinos and Watson (1986). (B) Examples of data from one animal in each of the normal, VTA-lesion, and SNc-lesion groups at each recording depth. Shown is a pair of half-second segments of data simultaneously obtained from the two electrodes of a stereotrode.
tion function of Ψ (t) for each time point t was computed from the time window of (t − 100 ms, t + 100 ms). (4) We then computed the synchronization indexfor each t, SI(t) = (Hmax –H(t))/Hmax , where H(t) = − pk ln(pk ), Shannon’s entropy of the distribution of Ψ with pk denoting the relative frequency of finding Ψ within the kth bin, k = 1, . . . , 20. Hmax was the maximum entropy obtainable and equaled to ln(P), where P was the number of partitions (n = 20), in the case of a uniform distribution of Ψ . The degree of phase synchrony reflected by SI increases monotonically in a range from 0 (no synchronization) to 1 (perfect synchrony). (5) Since we do not know the distribution of SI, we used a bootstrap method to assess the statistical significance of the result (Efron and Tibshrani, 1993). Five bootstrap samples were generated by pairing two recording data randomly chosen from the pool of all animals (n = 18). Each bootstrap sample was subjected to the same steps of data analysis as described above for the calculation of Ψ boot , SIboot and the mean of five SIboot values. This bootstrap sampling procedure was replicated 200 times to yield a bootstrap distribution of mean SIboot . We took the 99th percentile of the distribution (SIboot 99%) as a confidence limit. (6) For each animal group, a group mean SI at each recording depth was compared with SIboot 99%
Fig. 2. (A) Steps in the calculation of phase synchrony are shown; see Section 2 for details; (B) steps of data analysis are shown with an example of raw data. The data are the same as those shown for VTA-lesion group at 6 mm recording depth in Fig. 1B. Only 30 and 60 Hz filtered data are shown for clarity.
of the same frequency band, and taken as significant (P < 0.01) if the ratio of SI over SIboot 99% was greater than 1. (7) For testing the significance of a change in lesion groups compared with the normal, a distribution of SI differences was obtained from the 200 bootstrap replications described above. If the absolute value of an SI difference between a lesion group and the normal was greater than that of 99% of the distribution, then it was taken as significant (P < 0.01). With these steps we were essentially calculating a synchrony between instantaneous phases of two signals. The advantage of such a procedure over more conventional measures, such as frequency coherence or magnitude squared coherence, has previously been pointed out (Lachaux et al., 1999). First, coherence is a measure of the linear covariance between two spectra. As such, it requires that data are stationary in time in continuous measurements, or across trials in repeated measurements. Since this assumption can rarely be validated, methods using instantaneous phase calculation and thus not requiring stationarity are preferred. Second, coherence is influenced by amplitude covariance as well as phase covariance. The relative contribution of amplitude or phase to covariance is not cleanly separable. In situations
182
K.-M. Lee et al. / Neuroscience Research 49 (2004) 179–184
Fig. 4. The differences in mean SI between VTA-lesion group and normal control (A), and between SNC-lesion group and normal control (B) are shown. The SI difference ratio, or the ratio of the observed SI difference with respect to the 99th percentile level of absolute SI difference in bootstrap samples are coded as indicated by the gray scale at the bottom of the figure. Only cells with an absolute SI difference ratio greater than one (i.e., P < 0.01) are shown for clarity.
Fig. 3. Synchronization index (SI) ratio, or the ratio of SI relative to the 99th percentile of SI in bootstrap samples, is shown for each recording depth and at each frequency analyzed for phase synchronization. Only the cells with SI ratio higher than 1 (i.e., P < 0.01) are coded as indicated by the gray-to-black scale bar at the bottom of the figure, for normal controls (A), VTA-lesion group (B), and SNC-lesion group (C).
where one wants to investigate the phase relationship between two signals independent of their relative amplitudes, comparison of instantaneous phase is better.
3. Results Histological examination showed that no more than 1% of dopaminergic cells survived to be stained in the SNc of the injected side compared to the opposite side in all animals of SNc-lesion group. Removing dopaminergic neurons in the target area was similarly successful in all animals of VTAlesion group, but we noted additional loss of dopaminergic neurons in SNc of the same side in varying degrees (ranging from 30 to 80%). Fig. 3A–C depicts the average SI at each recording site and frequency band for each experimental group. For consistency across cells in the plot, the mean SI was grayscalecoded after normalizing to the 99th percentile value from surrogate samples, i.e., SIboot 99% of each cell. Brighter color denotes higher phase synchronization, while darker color indicates a level of phase synchrony not significantly different from that expected from bootstrap samples (see Section 2). As shown in Fig. 3A, normal animals showed higher phase synchrony at deep basal ganglia sites than at cor-
tical sites over a ␥ range (20–100 Hz). More specifically, strong phase synchronization was observed at 60 Hz at 3 mm depth, and around 20–30, 50, and 80 Hz at depths of 5 and 6 mm. Similar plots of the mean SI of the VTA-lesion and the SNc-lesion groups are shown in Fig. 3B and C, respectively. The most notable differences observed between the VTAlesion group and the normal were that the former showed higher SI in the cortex at 10 and 100 Hz and, in contrast, lower SI in the deep GP at 30 Hz (Fig. 4A). In addition, lower SI for the VTA group was observed at 60 Hz at 3 mm depth. In the case of the SNc-lesion group (Fig. 4B), SI was overall lower than the normal in the basal ganglia in broad bands of frequency ranging from 20 to 110 Hz. This decrease of SI was more pronounced than that observed in VTA group, and was most conspicuous at a high ␥ range (60–80 Hz at the depths of 3 and 6 mm). In contrast, there was an increase in SI observed at 3 mm depth at the low 10 Hz band. The differences noted between SNc-to-normal and VTA-to-normal comparisons were not statistically significant in a direct comparison between SNc-lesion and VTA-lesion groups. In an effort to examine the possibility that the observed changes in SI were related to other characteristics of the recorded data, a power spectral density (PSD) was estimated using the Thompson’s multitaper method (Perceval and Walden, 1993). The PSDs obtained from lesioned animals were within the 95% confidence intervals of the PSDs of control animals at all frequency bands and recording sites. Nor was there a systematic difference in PSD across recording sites that could be correlated with the difference of observed SI changes between cortical and subcortical sites.
K.-M. Lee et al. / Neuroscience Research 49 (2004) 179–184
4. Discussion 4.1. Sources of observed phase synchrony Phase synchrony between two electrode recordings arises when there is synchronized activity in two neuronal populations recorded independently by the electrodes, or due to volume conduction of the same activity reaching the electrodes simultaneously. Although we cannot completely rule out the latter, we believe that the former situation is more likely the case for our recordings for the following reasons. A metal electrode of greater than 1 M impedance can detect signals from large pyramidal cells up to 300 m from the electrode tip (Humphrey and Corrie, 1978), but most recordings are confined to within 200 m (Humphrey and Corrie, 1978; Stoney et al., 1968). Since the radius of the recording field is negatively related to electrode impedance (Legatt et al., 1980), the receptive volume of our 2 M electrode is further restricted. A simple geometric calculation shows that the overlap volume is at maximum 20.8% of the total record volume if 100 m is taken as the radius of signal reception for a 2 M electrode. (The effective volume of overlap where the neural activity can be recorded above a noise level by two electrodes could be much less than the calculated value, because an electrical signal gets weaker as a function of the radius squared. The exact function relating the impedance and the distance of effective reception by an electrode is not known. We assumed a linear inverse relationship, which was inferred from a finding by Tehovnik and Slocum (2003, their Fig. 4F)). Therefore it is likely that each electrode received a signal predominantly from its own neighborhood, and a much smaller contribution from neural elements in an overlapping volume. Furthermore, the observed specificity of SI values may be taken as an indication that the phase synchrony we observed was biological, rather than physical, in nature. If we had been recording from the same neural elements by volume conduction, all the SI would have been significantly much higher than that obtained from our surrogate data analysis, regardless of recording depths, frequency bands, and experimental groups. This is contrary to what we found, namely a systematic variation of SIs across recording depths and frequencies in the normal group, which was then no longer observable after dopamine depletion. We believe these findings together provide further evidence against the volume conduction hypothesis. Neuronal populations are believed to work as ensembles on a mesoscopic scale (Freeman, 2000) of around a few hundred microns, comparable to the separation distance between our two electrodes. Phase synchrony on this scale may arise as a result of a common input to neural elements of the ensemble from outside, or from synchronized oscillation of the networked elements within the ensemble. A feature of our data is compatible with the former view: The observed pattern of phase synchrony was very focal in space and frequency. For example, in normal animals, there was a strong synchronization at 60 Hz which was observed only
183
at sites within the CP and similarly at 80 Hz only in deep GP sites. This is more compatible with the within-ensemble synchrony view, at least in an anesthetized preparation because otherwise one would have expected to find synchronization of the same frequency over many structures. On the other hand, one may wonder if the 80 Hz phase synchrony observed at the GP is driven by the oscillatory activity of similar frequency that was previously recorded from the subthalamic nucleus of the alert healthy rat (Brown et al., 2002). In this case, the phase synchronization may have come from a synchronized input to the recorded ensemble. 4.2. Change in phase synchrony after chronic dopamine depletion We observed that, after chronic dopamine depletion, phase synchrony increased in the cortex at frequency bands lower than 20 Hz, whereas it decreased in the basal ganglia at higher frequency bands. The decrease of phase synchrony of deep basal ganglia sites suggests that dopaminergic innervation may serve as a phase synchronizing input to this area, consistent with a previous observation that the spectral peak in the upper ␥ band of local field potentials recorded from the STN in rats was increased by systemic injection of a D2 dopamine agonist (Brown et al., 2002). At the same time, dopaminergic inputs seem to have a desynchronizing influence on the cerebral cortex, resulting in an increase of SI in this structure at lower ␥ frequencies when removed. Previous studies similarly have reported that effects of dopamine depletion on temporal dynamics of neural activity vary across brain structures and frequency ranges. The power within the STN and the coherence between the STN and the GPi were shown to be dominated by activity with a frequency of less than 30 Hz in the absence of dopaminergic medication in patients with PD, whereas it was dominated by activity at 70–85 Hz in the presence of exogenous dopaminergic stimulation (Brown et al., 2001). It was also shown that, while the lower frequency coherence was attenuated by movement, the coherence in the higher frequency range was increased with movement, indicating that the dynamic coordination of oscillatory synchronization in the basal ganglia and the subcortical–cortical motor loops varies in the frequency domain (Cassidy et al., 2002). Even though some differences in SI changes were noted between the two lesion models when they were separately compared to the normal, none of these differences were statistically significant when a direct comparison was made between the two lesion groups. Given the fact that VTA and SNc differ in their predominant projection areas, namely the cortex and the basal ganglia, one would have expected to find a systematic difference in the changes of SI. One of the reasons for our failure to observe this difference was that there was some loss of dopaminergic neurons in SNc, in addition to the loss in VTA, in the VTA-lesion group. Therefore, whether the changes in SI would differ after VTA- and SNclesions could not be answered definitely by our investigation.
184
K.-M. Lee et al. / Neuroscience Research 49 (2004) 179–184
The importance of synchronized activity in basal ganglia and its connection to the cortex in the control of motor behavior is emphasized in a hypothesis by Brown and Marsden (1998). They proposed that the basal ganglia (including those dopaminergic source areas such as SNc and VTA) act to release selected cortical elements from idling alpha rhythms so that they may become coherent (synchronized) in the ␥ range, which makes those channels of motor processing necessary for achieving a given movement favored and bound together. The failure of this mechanism is suggested to contribute to the bradykinesia found in Parkinson’s disease. The current study adds to this line of thinking in that the effect of dopaminergic innervation may vary from synchronizing to desynchronizing, depending on the structure innervated and the frequency of oscillation.
Acknowledgements We thank Dr. Joydeep Bhattacharya for his suggestion on data analysis methods. We also thank Hejin Koh, Minseok Kang, Jinsook Roh, Kyongmi Kim, and Dr. Seon-Ha Baek for their assistance in conducting experiments. Supported by Seoul National University Hospital Research Fund (Grant No. 04-2000-018).
References Bergman, H., Raz, A., Feingold, A., Nini, A., Nelken, I., Hansel, D., Ben Pazi, H., Reches, A., 1998. Physiology of MPTP tremor. Mov. Disord. 13 (Suppl. 3), 29–34. Brown, P., Kupsch, A., Magill, P.J., Sharott, A., Harnack, D., Meissner, W., 2002. Oscillatory local field potentials recorded from the subthalamic nucleus of the alert rat. Exp. Neurol. 177, 581–585. Brown, P., Marsden, C.D., 1998. What do the basal ganglia do. Lancet 351, 1801–1804. Brown, P., Oliviero, A., Mazzone, P., Insola, A., Tonali, P., Di, L.V., 2001. Dopamine dependency of oscillations between subthalamic nucleus and pallidum in Parkinson’s disease. J. Neurosci. 21, 1033– 1038. Cassidy, M., Mazzone, P., Oliviero, A., Insola, A., Tonali, P., Di, L.V., Brown, P., 2002. Movement-related changes in synchronization in the human basal ganglia. Brain 125, 1235–1246. Efron, B., Tibshrani, R.J., 1993. An Introduction to the Bootstrap. Chapman & Hall, New York. Freeman, W.J., 1975. Mass Action in the Nervous System. Academic Press, New York. Freeman, W.J., 2000. How Brains Make Up their Minds. Columbia University Press, New York, NY. Goldberg, J.A., Boraud, T., Maraton, S., Haber, S.N., Vaadia, E., Bergman, H., 2002. Enhanced synchrony among primary motor cortex neurons in the 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine primate model of Parkinson’s disease. J. Neurosci. 22, 4639–4653.
Humphrey, D.R., Corrie, W.S., 1978. Properties of pyramidal tract neuron system within a functionally defined subregion of primate motor cortex. J. Neurophysiol. 41, 216–243. Hurtado, J.M., Gray, C.M., Tamas, L.B., Sigvardt, K.A., 1999. Dynamics of tremor-related oscillations in the human globus pallidus: a single case study. Proc. Natl. Acad. Sci. U.S.A. 96, 1674–1679. Juergens, E., Guettler, A., Eckhorn, R., 1999. Visual stimulation elicits locked and induced gamma oscillations in monkey intracortical- and EEG-potentials, but not in human EEG. Exp. Brain Res. 129, 247–259. Lachaux, J.P., Rodriguez, E., Martinerie, J., Varela, F.J., 1999. Measuring phase synchrony in brain signals. Human Brain Map. 8, 194–208. Legatt, A.D., Arezzo, J., Vaughan, J., 1980. Averaged multiple unit activity as an estimate of phasic changes in local neuronal activity: effects of volume-conducted potentials. J. Neurosci. Meth. 2, 203–217. Levy, R., Hutchison, W.D., Lozano, A.M., Dostrovsky, J.O., 2000. Highfrequency synchronization of neuronal activity in the subthalamic nucleus of parkinsonian patients with limb tremor. J. Neurosci. 20, 7766–7775. Levy, R., Hutchison, W.D., Lozano, A.M., Dostrovsky, J.O., 2002. Synchronized neuronal discharge in the basal ganglia of parkinsonian patients is limited to oscillatory activity. J. Neurosci. 22, 2855–2861. Magill, P.J., Bolam, J.P., Bevan, M.D., 2000. Relationship of activity in the subthalamic nucleus–globus pallidus network to cortical electroencephalogram. J. Neurosci. 20, 820–833. Magill, P.J., Bolam, J.P., Bevan, M.D., 2001. Dopamine regulates the impact of the cerebral cortex on the subthalamic nucleus–globus pallidus network. Neuroscience 106, 313–330. Mitzdorf, U., 1987. Properties of the evoked potential generators: current source-density analysis of visually evoked potentials in the cat cortex. Int. J. Neurosci. 33, 33–59. Nini, A., Feingold, A., Slovin, H., Bergman, H., 1995. Neurons in the globus pallidus do not show correlated activity in the normal monkey, but phase-locked oscillations appear in the MPTP model of parkinsonism. J. Neurophysiol. 74, 1800–1805. Paxinos, G., Watson, C., 1986. The Rat Brain in Stereotaxic Coordinates. Academic Press, New York. Perceval, D.B., Walden, A.T., 1993. Spectral Analysis for Physical Applications: Multitaper and Conventional Univariate Techniques. Cambridge University Press, Cambridge, UK. Plenz, D., Kital, S.T., 1999. A basal ganglia pacemaker formed by the subthalamic nucleus and external globus pallidus. Nature 400, 677– 682. Raz, A., Vaadia, E., Bergman, H., 2000. Firing patterns and correlations of spontaneous discharge of pallidal neurons in the normal and the tremulous 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine vervet model of parkinsonism. J. Neurosci. 20, 8559–8571. Selesnick, I.W., Lang, M., Burrus, C.S., 1996. Constrained least square design of FIR filters without specified transition bands. IEEE Trans. Signal Process. 44, 1879–1892. Stanford, I.M., 2003. Independent neuronal oscillators of the rat globus pallidus. J. Neurophysiol. 89, 1713–1717. Stoney, S.D., Thompson, W.D., Asanuma, H., 1968. Excitation of pyramidal tract cells by intracortical microstimulation: effective extent of stimulating current. J. Neurophysiol. 31, 659–669. Tass, P., Rosenblum, M.G., Weule, J., Kurths, J., Pikovsky, A., Volkmann, J., Schnitzler, A., Freund, H.J., 1998. Detection of n:m phase locking from noisy data: application to magnetoencephalography. Phys. Rev. Lett. 81, 3291–3294. Tehovnik, E.J., Slocum, W.M., 2003. Using ocular dominance to infer the depth of the visual input layers of V1 in behaving macaque monkey. J. Neurosci. Meth. 125, 121–128.
Change in phase synchronization of local field potentials in anesthetized rats after chronic dopamine depletion Kyoung-Min Lee a,∗ , Tae-Beom Ahn a , Beom S. Jeon a , Dong Gyu Kim b a
Department of Neurology, Seoul National University, 28 Yongon-Dong Chongno-Gu, Seoul 110-744, South Korea b Department of Neurosurgery, Seoul National University, Seoul, South Korea Received 29 October 2003; accepted 16 February 2004 Available online 9 April 2004
Abstract In order to investigate temporal characteristics of oscillatory neural activity and the effect of chronic dopamine depletion on it, local field potentials were measured in anesthetized rats without or with a 6-OHDA lesion at either the ventral tegmental area or the substantia nigra compacta, using a pair of electrodes that were separated by 120 m. Coupling of neural activity in this mesoscopic scale was measured by a synchronization index that quantified the distribution of differences in instantaneous phase between the two potentials recorded. Phase synchrony was significantly stronger at deep basal ganglia sites, more so than at cortical sites, over a ␥ range (30–120 Hz) in normal rats. After chronic dopamine depletion, this synchrony was no longer observable, especially after a substantia nigra lesion, while there was an increase in phase synchrony in the cortex at a lower frequency band (10 Hz). These findings are consistent with previously reported findings that the effect of dopaminergic innervation on oscillatory neural activity varies from synchronizing to desynchronizing, depending on the structure innervated and the frequency of the oscillation. © 2004 Elsevier Ireland Ltd and The Japan Neuroscience Society. All rights reserved. Keywords: Phase synchrony; Oscillatory neural activity; Dopamine depletion; Parkinson’s disease; Rat brain
1. Introduction Alteration in oscillatory activity of neurons or neuronal assembles has been implicated in the pathogenesis of Parkinson’s disease (PD). It has been demonstrated that periodic, oscillatory bursting in neurons at the internal segment of the globus pallidus (GPi) of MPTP-treated monkeys and patients with PD is synchronized (Nini et al., 1995; Bergman et al., 1998; Hurtado et al., 1999; Raz et al., 2000). Similarly, high-frequency oscillations were reported in the subthalamic nucleus (STN) of PD patients with tremor (Levy et al., 2000), and enhanced synchronized population bursting was observed in the primary motor cortex of the MPTP primate model (Goldberg et al., 2002). The origin of such burst firing, oscillations, or neuronal synchronization is not entirely clear. They can arise through the properties of individual cells and/or network properties of neuronal ensembles. On the one hand, synchronized ac∗ Corresponding author. Tel.: +1-82-2-760-2985; fax: +1-82-2-3672-7553. E-mail addresses: [email protected], [email protected] (K.-M. Lee).
tivity was not observed in the GPi, the external segment of the globus pallidus (GPe), or the substantia nigra pars reticulata (SNr) of non-tremulous PD patients (Levy et al., 2002), suggesting that overt neuronal synchronization is not intrinsic to the parkinsonian basal ganglia. It has been postulated that the coupled STN–GP network constitutes a central pattern generator that is able to maintain synchronized burst discharges (Plenz and Kital, 1999), upon which the cortex provides an oscillatory drive (Magill et al., 2000). After dopamine depletion, oscillatory cortical activity is expressed more powerfully by the STN–GP network, and the manner in which cortical information is processed by the network is altered profoundly (Magill et al., 2001). On the other hand, a small proportion of STN and GP cells show oscillation in the absence of cortical input, suggesting that either a local network of STN–GP neurons confined within the basal ganglia or intrinsic neuronal properties are able to sustain such activity (Magill et al., 2001). In a recent experiment where whole-cell patch-clamp techniques were used to record from single and pairs of GP cells, it was observed that type A neurons of the GP independently fire at a frequency determined by intrinsic subthreshold membrane
0168-0102/$ – see front matter © 2004 Elsevier Ireland Ltd and The Japan Neuroscience Society. All rights reserved. doi:10.1016/j.neures.2004.02.008
180
K.-M. Lee et al. / Neuroscience Research 49 (2004) 179–184
oscillations, while GABA inhibitory postsynaptic potentials can produce rebound spiking that promotes coherent firing in multiple GP neurons through a phase realignment of this intrinsic oscillatory activity (Stanford, 2003). In view of these issues on the synchronization of neuronal activity in Parkinson’s disease, the main objectives of this study were: (1) To examine the temporal dynamics of phase synchronization at various structures in the brain, especially focusing on the synaptic input activity of each region. While spikes mainly represent the output activity of neural elements (Legatt et al., 1980; Freeman, 1975), local field potentials (LFPs) reflect more of input synaptic activity (Mitzdorf, 1987; Juergens et al., 1999). Therefore, we chose to examine LFPs and measure their phase synchronization. For computing the degree of phase synchrony from bivariate data obtained using a pair of electrodes, we employed an approach developed by Tass et al. (1998). (2) To assess the role of dopaminergic innervation on phase synchronization comparing normal brains and those with chronic dopamine depletion, we recorded also from a rat PD model where a 6-OHDA lesion was made either at the ventral tegmental area (VTA) or at the substantia nigra pars compacta (SNc).
2. Methods Experiments were carried out on 18 adult male Sprague– Dawley rats (weight ranging from 180 to 200 g; seven in the normal control group, five in the VTA-lesion group, and six in the SNc-lesion group) and were conducted in accordance with the institutional guidelines on animal research. 2.1. Unilateral lesion of dopaminergic neurons Anesthesia was induced with chloral hydrate (400 mg/ 100 g wt.). Twenty-five minutes before the injection of 6-OHDA, all animals received a pre-treatment with desipramine (12.5 mg/100 g wt.) to minimize the uptake of 6-OHDA by noradrenergic neurons, and five minutes before the OHDA injection, pargyline (25 mg/100 g wt.) was administered to maximize the toxic effects on dopaminergic neurons. Using a stereotaxic frame (David Kopf Instruments, Tujunga, CA, USA), 6-OHDA (8 g/4 l in 0.2% ascorbic acid solution) was injected into the VTA (at A-P 3.4 mm, left 0.8 mm, and D-V 2.2 mm from the interaural line, aimed at 10◦ angle from the vertical) or the SNc (at AP 4.2 mm, left 1.8 mm, and D-V 2.0 mm from the interaural line). The lesions were verified by tyrosine hydroxylase immunohistochemistry after the animals were sacrificed at the end of a recording session. 2.2. Recording the local field potentials Electrophysiological recordings were performed using tungsten stereotrodes (WPI, Sarasota, FL, USA, Model
#TST33A20KT, outer diameter 0.365 mm) which have two electrode tips, separated by 120 m, each with an impedance of 2.0 M. In the case of lesioned animals, the recording was performed 3–4 weeks after a lesion was made. After anesthesia was established using ketamine (bolus injection of 100 mg/kg i.p. and supplementary injection of 50 mg/kg i.m.), an animal was positioned in the same stereotaxic frame as used for lesion-making (David Kopf Instruments, Tujunga, CA, USA). A craniotomy was done and the dura was opened above the aimed recording site (at A-P −0.6 mm, lateral 3.0 mm from the bregma). After an electrode was approximately positioned above the exposed surface of the brain, it was slowly advanced downward using a hydraulic microdrive (Stoelting Co., Illinois, USA). A contact to the brain tissue was noted visually and at the same time by monitoring a sudden change in the noise pattern. The contact point was marked as a reference for the depth of subsequent recordings. The electrode was then slowly advanced downward and stopped at each millimeter up to 6 mm in depth. At least 3 min of rest was allowed for the brain tissue to settle before LFPs were recorded for half a minute. According to a rat brain atlas (Paxinos and Watson, 1986), our recorded sites were located at the frontoparietal motor cortex (CTX, for sites 1 and 2 mm deep from the surface), the caudate-putamen (CP, for sites 3, 4, and probably 5 mm deep from the surface), and the globus pallidus (GP, for sites possibly 5 and 6 mm deep from the surface) (Fig. 1). However, these localizations were approximate at best, considering expected anatomical variability across individual animals. LFP was sampled and stored at 100 kHz using TDT System 2 (Tucker-Davis Technologies, FL, USA), and then sub-sampled to 1 kHz for further offline analysis by averaging over 100 data points. The dual recordings were both effectively monopolar since a common reference lead from the pre-amplifier was connected to the subcutaneous tissue around the craniotomy site with an impedance negligible compared to that of the recording electrodes. 2.3. Computation of phase synchrony Quantification of the degree of phase synchrony between signals that were recorded simultaneously from the two electrodes of a stereotrode was performed using the following steps (Fig. 2): (1) LFPs were bandpassfiltered at 10 Hz steps from 10 to 150 Hz with a bandwidth of 5 Hz using a linear-phase FIR filter designed by constrained least-squares (Selesnick et al., 1996) implemented in MATLAB (The Mathworks Inc., Natick, MA, USA). (2) Instantaneous phases of each filtered signal were computed by the analytic signal approach using the Hilbert transform. (3) The distribution of the relative phase, Ψ = (Φ modulo 2π), between the two signals was obtained. Since our data was not necessarily stationary, we adopted a sliding window approach in which the distribu-
K.-M. Lee et al. / Neuroscience Research 49 (2004) 179–184
181
CTX CP GP
(A) 1mm Normal VTA-lesion SNc-lesion Recording Depth Channel1 Channel2 Channel1 Channel2 Channel1 Channel2 1mm
2mm
3mm
4mm
5mm
6mm 0.2 mV
(B)
0.5 sec
Fig. 1. (A) Presumed recording sites schematically depicted over a coronal plane of the rat brain at −0.6 mm from the bregma. The background was hand-drawn based on the dimensions provided in Paxinos and Watson (1986). (B) Examples of data from one animal in each of the normal, VTA-lesion, and SNc-lesion groups at each recording depth. Shown is a pair of half-second segments of data simultaneously obtained from the two electrodes of a stereotrode.
tion function of Ψ (t) for each time point t was computed from the time window of (t − 100 ms, t + 100 ms). (4) We then computed the synchronization indexfor each t, SI(t) = (Hmax –H(t))/Hmax , where H(t) = − pk ln(pk ), Shannon’s entropy of the distribution of Ψ with pk denoting the relative frequency of finding Ψ within the kth bin, k = 1, . . . , 20. Hmax was the maximum entropy obtainable and equaled to ln(P), where P was the number of partitions (n = 20), in the case of a uniform distribution of Ψ . The degree of phase synchrony reflected by SI increases monotonically in a range from 0 (no synchronization) to 1 (perfect synchrony). (5) Since we do not know the distribution of SI, we used a bootstrap method to assess the statistical significance of the result (Efron and Tibshrani, 1993). Five bootstrap samples were generated by pairing two recording data randomly chosen from the pool of all animals (n = 18). Each bootstrap sample was subjected to the same steps of data analysis as described above for the calculation of Ψ boot , SIboot and the mean of five SIboot values. This bootstrap sampling procedure was replicated 200 times to yield a bootstrap distribution of mean SIboot . We took the 99th percentile of the distribution (SIboot 99%) as a confidence limit. (6) For each animal group, a group mean SI at each recording depth was compared with SIboot 99%
Fig. 2. (A) Steps in the calculation of phase synchrony are shown; see Section 2 for details; (B) steps of data analysis are shown with an example of raw data. The data are the same as those shown for VTA-lesion group at 6 mm recording depth in Fig. 1B. Only 30 and 60 Hz filtered data are shown for clarity.
of the same frequency band, and taken as significant (P < 0.01) if the ratio of SI over SIboot 99% was greater than 1. (7) For testing the significance of a change in lesion groups compared with the normal, a distribution of SI differences was obtained from the 200 bootstrap replications described above. If the absolute value of an SI difference between a lesion group and the normal was greater than that of 99% of the distribution, then it was taken as significant (P < 0.01). With these steps we were essentially calculating a synchrony between instantaneous phases of two signals. The advantage of such a procedure over more conventional measures, such as frequency coherence or magnitude squared coherence, has previously been pointed out (Lachaux et al., 1999). First, coherence is a measure of the linear covariance between two spectra. As such, it requires that data are stationary in time in continuous measurements, or across trials in repeated measurements. Since this assumption can rarely be validated, methods using instantaneous phase calculation and thus not requiring stationarity are preferred. Second, coherence is influenced by amplitude covariance as well as phase covariance. The relative contribution of amplitude or phase to covariance is not cleanly separable. In situations
182
K.-M. Lee et al. / Neuroscience Research 49 (2004) 179–184
Fig. 4. The differences in mean SI between VTA-lesion group and normal control (A), and between SNC-lesion group and normal control (B) are shown. The SI difference ratio, or the ratio of the observed SI difference with respect to the 99th percentile level of absolute SI difference in bootstrap samples are coded as indicated by the gray scale at the bottom of the figure. Only cells with an absolute SI difference ratio greater than one (i.e., P < 0.01) are shown for clarity.
Fig. 3. Synchronization index (SI) ratio, or the ratio of SI relative to the 99th percentile of SI in bootstrap samples, is shown for each recording depth and at each frequency analyzed for phase synchronization. Only the cells with SI ratio higher than 1 (i.e., P < 0.01) are coded as indicated by the gray-to-black scale bar at the bottom of the figure, for normal controls (A), VTA-lesion group (B), and SNC-lesion group (C).
where one wants to investigate the phase relationship between two signals independent of their relative amplitudes, comparison of instantaneous phase is better.
3. Results Histological examination showed that no more than 1% of dopaminergic cells survived to be stained in the SNc of the injected side compared to the opposite side in all animals of SNc-lesion group. Removing dopaminergic neurons in the target area was similarly successful in all animals of VTAlesion group, but we noted additional loss of dopaminergic neurons in SNc of the same side in varying degrees (ranging from 30 to 80%). Fig. 3A–C depicts the average SI at each recording site and frequency band for each experimental group. For consistency across cells in the plot, the mean SI was grayscalecoded after normalizing to the 99th percentile value from surrogate samples, i.e., SIboot 99% of each cell. Brighter color denotes higher phase synchronization, while darker color indicates a level of phase synchrony not significantly different from that expected from bootstrap samples (see Section 2). As shown in Fig. 3A, normal animals showed higher phase synchrony at deep basal ganglia sites than at cor-
tical sites over a ␥ range (20–100 Hz). More specifically, strong phase synchronization was observed at 60 Hz at 3 mm depth, and around 20–30, 50, and 80 Hz at depths of 5 and 6 mm. Similar plots of the mean SI of the VTA-lesion and the SNc-lesion groups are shown in Fig. 3B and C, respectively. The most notable differences observed between the VTAlesion group and the normal were that the former showed higher SI in the cortex at 10 and 100 Hz and, in contrast, lower SI in the deep GP at 30 Hz (Fig. 4A). In addition, lower SI for the VTA group was observed at 60 Hz at 3 mm depth. In the case of the SNc-lesion group (Fig. 4B), SI was overall lower than the normal in the basal ganglia in broad bands of frequency ranging from 20 to 110 Hz. This decrease of SI was more pronounced than that observed in VTA group, and was most conspicuous at a high ␥ range (60–80 Hz at the depths of 3 and 6 mm). In contrast, there was an increase in SI observed at 3 mm depth at the low 10 Hz band. The differences noted between SNc-to-normal and VTA-to-normal comparisons were not statistically significant in a direct comparison between SNc-lesion and VTA-lesion groups. In an effort to examine the possibility that the observed changes in SI were related to other characteristics of the recorded data, a power spectral density (PSD) was estimated using the Thompson’s multitaper method (Perceval and Walden, 1993). The PSDs obtained from lesioned animals were within the 95% confidence intervals of the PSDs of control animals at all frequency bands and recording sites. Nor was there a systematic difference in PSD across recording sites that could be correlated with the difference of observed SI changes between cortical and subcortical sites.
K.-M. Lee et al. / Neuroscience Research 49 (2004) 179–184
4. Discussion 4.1. Sources of observed phase synchrony Phase synchrony between two electrode recordings arises when there is synchronized activity in two neuronal populations recorded independently by the electrodes, or due to volume conduction of the same activity reaching the electrodes simultaneously. Although we cannot completely rule out the latter, we believe that the former situation is more likely the case for our recordings for the following reasons. A metal electrode of greater than 1 M impedance can detect signals from large pyramidal cells up to 300 m from the electrode tip (Humphrey and Corrie, 1978), but most recordings are confined to within 200 m (Humphrey and Corrie, 1978; Stoney et al., 1968). Since the radius of the recording field is negatively related to electrode impedance (Legatt et al., 1980), the receptive volume of our 2 M electrode is further restricted. A simple geometric calculation shows that the overlap volume is at maximum 20.8% of the total record volume if 100 m is taken as the radius of signal reception for a 2 M electrode. (The effective volume of overlap where the neural activity can be recorded above a noise level by two electrodes could be much less than the calculated value, because an electrical signal gets weaker as a function of the radius squared. The exact function relating the impedance and the distance of effective reception by an electrode is not known. We assumed a linear inverse relationship, which was inferred from a finding by Tehovnik and Slocum (2003, their Fig. 4F)). Therefore it is likely that each electrode received a signal predominantly from its own neighborhood, and a much smaller contribution from neural elements in an overlapping volume. Furthermore, the observed specificity of SI values may be taken as an indication that the phase synchrony we observed was biological, rather than physical, in nature. If we had been recording from the same neural elements by volume conduction, all the SI would have been significantly much higher than that obtained from our surrogate data analysis, regardless of recording depths, frequency bands, and experimental groups. This is contrary to what we found, namely a systematic variation of SIs across recording depths and frequencies in the normal group, which was then no longer observable after dopamine depletion. We believe these findings together provide further evidence against the volume conduction hypothesis. Neuronal populations are believed to work as ensembles on a mesoscopic scale (Freeman, 2000) of around a few hundred microns, comparable to the separation distance between our two electrodes. Phase synchrony on this scale may arise as a result of a common input to neural elements of the ensemble from outside, or from synchronized oscillation of the networked elements within the ensemble. A feature of our data is compatible with the former view: The observed pattern of phase synchrony was very focal in space and frequency. For example, in normal animals, there was a strong synchronization at 60 Hz which was observed only
183
at sites within the CP and similarly at 80 Hz only in deep GP sites. This is more compatible with the within-ensemble synchrony view, at least in an anesthetized preparation because otherwise one would have expected to find synchronization of the same frequency over many structures. On the other hand, one may wonder if the 80 Hz phase synchrony observed at the GP is driven by the oscillatory activity of similar frequency that was previously recorded from the subthalamic nucleus of the alert healthy rat (Brown et al., 2002). In this case, the phase synchronization may have come from a synchronized input to the recorded ensemble. 4.2. Change in phase synchrony after chronic dopamine depletion We observed that, after chronic dopamine depletion, phase synchrony increased in the cortex at frequency bands lower than 20 Hz, whereas it decreased in the basal ganglia at higher frequency bands. The decrease of phase synchrony of deep basal ganglia sites suggests that dopaminergic innervation may serve as a phase synchronizing input to this area, consistent with a previous observation that the spectral peak in the upper ␥ band of local field potentials recorded from the STN in rats was increased by systemic injection of a D2 dopamine agonist (Brown et al., 2002). At the same time, dopaminergic inputs seem to have a desynchronizing influence on the cerebral cortex, resulting in an increase of SI in this structure at lower ␥ frequencies when removed. Previous studies similarly have reported that effects of dopamine depletion on temporal dynamics of neural activity vary across brain structures and frequency ranges. The power within the STN and the coherence between the STN and the GPi were shown to be dominated by activity with a frequency of less than 30 Hz in the absence of dopaminergic medication in patients with PD, whereas it was dominated by activity at 70–85 Hz in the presence of exogenous dopaminergic stimulation (Brown et al., 2001). It was also shown that, while the lower frequency coherence was attenuated by movement, the coherence in the higher frequency range was increased with movement, indicating that the dynamic coordination of oscillatory synchronization in the basal ganglia and the subcortical–cortical motor loops varies in the frequency domain (Cassidy et al., 2002). Even though some differences in SI changes were noted between the two lesion models when they were separately compared to the normal, none of these differences were statistically significant when a direct comparison was made between the two lesion groups. Given the fact that VTA and SNc differ in their predominant projection areas, namely the cortex and the basal ganglia, one would have expected to find a systematic difference in the changes of SI. One of the reasons for our failure to observe this difference was that there was some loss of dopaminergic neurons in SNc, in addition to the loss in VTA, in the VTA-lesion group. Therefore, whether the changes in SI would differ after VTA- and SNclesions could not be answered definitely by our investigation.
184
K.-M. Lee et al. / Neuroscience Research 49 (2004) 179–184
The importance of synchronized activity in basal ganglia and its connection to the cortex in the control of motor behavior is emphasized in a hypothesis by Brown and Marsden (1998). They proposed that the basal ganglia (including those dopaminergic source areas such as SNc and VTA) act to release selected cortical elements from idling alpha rhythms so that they may become coherent (synchronized) in the ␥ range, which makes those channels of motor processing necessary for achieving a given movement favored and bound together. The failure of this mechanism is suggested to contribute to the bradykinesia found in Parkinson’s disease. The current study adds to this line of thinking in that the effect of dopaminergic innervation may vary from synchronizing to desynchronizing, depending on the structure innervated and the frequency of oscillation.
Acknowledgements We thank Dr. Joydeep Bhattacharya for his suggestion on data analysis methods. We also thank Hejin Koh, Minseok Kang, Jinsook Roh, Kyongmi Kim, and Dr. Seon-Ha Baek for their assistance in conducting experiments. Supported by Seoul National University Hospital Research Fund (Grant No. 04-2000-018).
References Bergman, H., Raz, A., Feingold, A., Nini, A., Nelken, I., Hansel, D., Ben Pazi, H., Reches, A., 1998. Physiology of MPTP tremor. Mov. Disord. 13 (Suppl. 3), 29–34. Brown, P., Kupsch, A., Magill, P.J., Sharott, A., Harnack, D., Meissner, W., 2002. Oscillatory local field potentials recorded from the subthalamic nucleus of the alert rat. Exp. Neurol. 177, 581–585. Brown, P., Marsden, C.D., 1998. What do the basal ganglia do. Lancet 351, 1801–1804. Brown, P., Oliviero, A., Mazzone, P., Insola, A., Tonali, P., Di, L.V., 2001. Dopamine dependency of oscillations between subthalamic nucleus and pallidum in Parkinson’s disease. J. Neurosci. 21, 1033– 1038. Cassidy, M., Mazzone, P., Oliviero, A., Insola, A., Tonali, P., Di, L.V., Brown, P., 2002. Movement-related changes in synchronization in the human basal ganglia. Brain 125, 1235–1246. Efron, B., Tibshrani, R.J., 1993. An Introduction to the Bootstrap. Chapman & Hall, New York. Freeman, W.J., 1975. Mass Action in the Nervous System. Academic Press, New York. Freeman, W.J., 2000. How Brains Make Up their Minds. Columbia University Press, New York, NY. Goldberg, J.A., Boraud, T., Maraton, S., Haber, S.N., Vaadia, E., Bergman, H., 2002. Enhanced synchrony among primary motor cortex neurons in the 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine primate model of Parkinson’s disease. J. Neurosci. 22, 4639–4653.
Humphrey, D.R., Corrie, W.S., 1978. Properties of pyramidal tract neuron system within a functionally defined subregion of primate motor cortex. J. Neurophysiol. 41, 216–243. Hurtado, J.M., Gray, C.M., Tamas, L.B., Sigvardt, K.A., 1999. Dynamics of tremor-related oscillations in the human globus pallidus: a single case study. Proc. Natl. Acad. Sci. U.S.A. 96, 1674–1679. Juergens, E., Guettler, A., Eckhorn, R., 1999. Visual stimulation elicits locked and induced gamma oscillations in monkey intracortical- and EEG-potentials, but not in human EEG. Exp. Brain Res. 129, 247–259. Lachaux, J.P., Rodriguez, E., Martinerie, J., Varela, F.J., 1999. Measuring phase synchrony in brain signals. Human Brain Map. 8, 194–208. Legatt, A.D., Arezzo, J., Vaughan, J., 1980. Averaged multiple unit activity as an estimate of phasic changes in local neuronal activity: effects of volume-conducted potentials. J. Neurosci. Meth. 2, 203–217. Levy, R., Hutchison, W.D., Lozano, A.M., Dostrovsky, J.O., 2000. Highfrequency synchronization of neuronal activity in the subthalamic nucleus of parkinsonian patients with limb tremor. J. Neurosci. 20, 7766–7775. Levy, R., Hutchison, W.D., Lozano, A.M., Dostrovsky, J.O., 2002. Synchronized neuronal discharge in the basal ganglia of parkinsonian patients is limited to oscillatory activity. J. Neurosci. 22, 2855–2861. Magill, P.J., Bolam, J.P., Bevan, M.D., 2000. Relationship of activity in the subthalamic nucleus–globus pallidus network to cortical electroencephalogram. J. Neurosci. 20, 820–833. Magill, P.J., Bolam, J.P., Bevan, M.D., 2001. Dopamine regulates the impact of the cerebral cortex on the subthalamic nucleus–globus pallidus network. Neuroscience 106, 313–330. Mitzdorf, U., 1987. Properties of the evoked potential generators: current source-density analysis of visually evoked potentials in the cat cortex. Int. J. Neurosci. 33, 33–59. Nini, A., Feingold, A., Slovin, H., Bergman, H., 1995. Neurons in the globus pallidus do not show correlated activity in the normal monkey, but phase-locked oscillations appear in the MPTP model of parkinsonism. J. Neurophysiol. 74, 1800–1805. Paxinos, G., Watson, C., 1986. The Rat Brain in Stereotaxic Coordinates. Academic Press, New York. Perceval, D.B., Walden, A.T., 1993. Spectral Analysis for Physical Applications: Multitaper and Conventional Univariate Techniques. Cambridge University Press, Cambridge, UK. Plenz, D., Kital, S.T., 1999. A basal ganglia pacemaker formed by the subthalamic nucleus and external globus pallidus. Nature 400, 677– 682. Raz, A., Vaadia, E., Bergman, H., 2000. Firing patterns and correlations of spontaneous discharge of pallidal neurons in the normal and the tremulous 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine vervet model of parkinsonism. J. Neurosci. 20, 8559–8571. Selesnick, I.W., Lang, M., Burrus, C.S., 1996. Constrained least square design of FIR filters without specified transition bands. IEEE Trans. Signal Process. 44, 1879–1892. Stanford, I.M., 2003. Independent neuronal oscillators of the rat globus pallidus. J. Neurophysiol. 89, 1713–1717. Stoney, S.D., Thompson, W.D., Asanuma, H., 1968. Excitation of pyramidal tract cells by intracortical microstimulation: effective extent of stimulating current. J. Neurophysiol. 31, 659–669. Tass, P., Rosenblum, M.G., Weule, J., Kurths, J., Pikovsky, A., Volkmann, J., Schnitzler, A., Freund, H.J., 1998. Detection of n:m phase locking from noisy data: application to magnetoencephalography. Phys. Rev. Lett. 81, 3291–3294. Tehovnik, E.J., Slocum, W.M., 2003. Using ocular dominance to infer the depth of the visual input layers of V1 in behaving macaque monkey. J. Neurosci. Meth. 125, 121–128.