Neurons in the basal ganglia output nuclei display rhythmic burst firing after chronic nigrostriatal lesions. The thalamocortical network is a strong endogenous generator of oscillatory activity, and the striatum receives a massive projection from the cerebral cortex. Actually, the membrane potential of striatal projection neurons displays periodic shifts between a very negative resting potential (down state) and depolarizing plateaus (up states) during which they can fire action potentials. We hypothesized that an increased excitability of striatal neurons may allow transmission of cortical slow rhythms through the striatum to the remaining basal ganglia in experimental parkinsonism. In vivo intracellular recordings revealed that striatal projection neurons from rats with chronic nigrostriatal lesions had a more depolarized membrane potential during both the down and up states and an increased firing probability during the up events. Furthermore, lesioned rats had significantly fewer silent neurons than control rats. Simultaneous recordings of the frontal electrocorticogram and membrane potential of striatal projection neurons revealed that the signals were oscillating synchronously in the frequency range 0.4–2 Hz, both in control rats and rats with chronic nigrostriatal lesions. Spreading of the slow cortical rhythm is limited by the very low firing probability of control rat neurons, but a slow oscillation is well reflected in spike trains of ∼60% of lesioned rat neurons. These findings providein vivo evidence for a role of dopamine in controlling the flow of cortical activity through the striatum and may be of outstanding relevance for understanding the pathophysiology of Parkinson's disease.
- cerebral cortex
- in vivo intracellular recording
- Parkinson's disease
- neuronal firing patterns
It is conventionally accepted that a substantial part of the influence of dopamine (DA) on sensorimotor and cognitive functions is related to its modulatory effects on striatal processing of cortical input (Albin et al., 1989;Hollerman et al., 2000). Dopamine inputs from the substantia nigra converge with glutamatergic inputs from the cerebral cortex at the dendrites of striatal projection neurons (Smith et al., 1998), which are called “medium spiny neurons” (MSNs). In vivointracellular recordings from MSNs revealed that their membrane potential alternates between two steady-state values. A very polarized resting potential (“down state”) is interrupted by periods of sustained depolarization (“up states”) lasting a few hundred milliseconds (Wilson and Groves, 1981; Calabresi et al., 1990; Wilson, 1993). There is evidence supporting the theory that the up events are driven by excitatory inputs from the cerebral cortex and thalamus (Wilson et al., 1983; Wilson 1993; Plenz and Aertsen, 1996).In vitro studies indicate that the ionic conductances shaping the fluctuating membrane potential of MSNs are modulated by DA (Nicola et al., 2000). The up events are currently perceived as “enabling states,” during which synchronously depolarized MSNs translate afferent activity into sequences of action potentials, allowing transmission of information to structures receiving striatal projections (Wilson, 1993; O'Donnell and Grace, 1995).
Extracellular single unit recordings established that after nigrostriatal system lesions, the firing pattern of neurons in nuclei receiving strong striatal innervation, like the globus pallidus (GP) and substantia nigra pars reticulata (SNpr), shifts from a regular pattern to a rhythmic burst firing mode (Pan and Walters, 1988; MacLeod et al., 1990; Murer et al., 1997; Ni et al., 2000). Recent studies from our laboratory indicate that as much as 40% of SNpr units recorded from 6-hydroxydopamine (6-OHDA)-lesioned rats display rhythmic ∼1 Hz firing rate modulations (Tseng et al., 2001). Our findings showing a strong modulatory effect of intrastriatally administered DA receptor agonists on the firing pattern of SNpr units in 6-OHDA-lesioned rats support a role of the striatum in the genesis of this phenomenon (Tseng et al., 2000). Interestingly, it has been reported that the membrane potential fluctuations of MSNs display a weak periodicity, also with a frequency of ∼1 Hz (Stern et al., 1997). In slow wave sleep and anesthesia, rhythmic modulations of firing rate occur in neurons throughout the cerebral cortex and thalamus. Rhythmic activity of the thalamocortical network is reflected in the electroencephalogram (EEG) of animals and humans as oscillations of different frequencies (including a slow ∼1 Hz oscillation) that are synchronized over widespread areas of the cerebral cortex (Steriade, 1999). Therefore, we hypothesized that an altered functional state of striatal projection neurons may facilitate the transmission of slow cortical rhythms to the basal ganglia output nuclei in experimental parkinsonism. In this manuscript, we report the existence of a strong correlation between cortical slow rhythmic activity and membrane potential fluctuations of striatal neurons. Furthermore, we describe modifications in the membrane potential of striatal neurons from 6-OHDA-lesioned rats that may allow transmission of slow cortical rhythms toward basal ganglia output nuclei in experimental parkinsonism.
MATERIALS AND METHODS
Male adult Sprague Dawley rats, weighing 190–220 gm, were randomly assigned to receive a 6-OHDA lesion or a sham lesion. The animals were maintained on a 12 hr light/dark cycle with food and tap water available ad libitum until the time of the experiment.In vivo intracellular recordings of striatal neurons were performed 6–10 weeks after the lesion.
Lesions. A severe unilateral lesion of mesencephalic dopaminergic neurons was obtained with the neurotoxin 6-OHDA hydrobromide (Sigma, St. Louis, MO), as reported previously (Murer et al., 1997). Briefly, the rats were anesthetized with pentobarbital (50 mg/kg, i.p.), and with the aid of a stereotaxic instrument (David Kopf Instruments, Tujunga, CA), they were injected with the neurotoxin (8 μg free base in 4 μl of 0.1% ascorbic acid) at the medial forebrain bundle (stereotaxic coordinates: anterior (A), −4.3 from bregma; lateral (L), 1.6; H, 8.3 below the cortical surface) (Paxinos and Watson, 1997). The control group received only the ascorbic acid solution. The effectiveness of the lesion was evaluated with the stepping test (Olsson et al., 1995; Chang et al., 1999) and further confirmed postmortem by means of immunohistochemistry (Murer et al., 1997). The stepping test was repeated three times between days 15 and 20 after surgery in all the rats. Only those 6-OHDA-lesioned rats that showed less than two adjusting steps with the forelimb contralateral to the lesion during each trial were selected for the experiments (see Fig. 1).
Electrophysiological recordings. The rats were anesthetized with urethane (1.2 gm/kg, i.p.), treated with a local anesthetic on the scalp and pressure points, and secured to the stereotaxic frame. Temperature was maintained between 36 and 37°C with a heating pad. Additional urethane was administered throughout the experiment as necessary to maintain a constant level of anesthesia, as determined by testing the hindlimb compression reflex, and electrocorticographic recordings when available (customarily, supplements of 0.4 gm/kg, s.c., every 3–4 hr).
A concentric bipolar stimulating electrode (SNE-100; Better Hospital Equipment, Rockville Centre, NY) was placed in the frontal cortex (A, 3 mm anterior to the bregma; L, 1.5 mm; and H, 2.5 mm below the cortical surface) (Paxinos and Watson, 1997) to allow delivery of 300 μsec square wave pulses, 300–500 μA in amplitude (10–15 pulses at 0.5 Hz). In many experiments, an additional concentric bipolar electrode was located 1 mm lateral to the stimulating electrode, with the tip inserted ∼1.5 mm below the cortical surface, to obtain a differential recording of the electrocorticogram (ECoG). The cortical signal was filtered (0.1–300 Hz), amplified (ER-98; NeuroData, Delaware Water Gap, PA), and sent to an A/D converter (DigiData 1200; Axon Instruments, Foster City, CA).
Intracellular recordings were obtained from a striatal region located 1.0 mm anterior to bregma, 2.5–3.0 mm lateral to the midline, and 3.0–5.0 mm below the cortical surface (Paxinos and Watson, 1997). The microelectrodes were pulled (Rhema-Labortechnik microelectrode puller; Campden Instruments, Loughborough, UK) from 1.2 mm outer diameter borosilicate glass capillaries (WPI, Sarasota, FL) and filled with 2 m K-acetate and 2% Neurobiotin (Research Biochemicals, Natick, MA); they had resistances ranging from 60 to 130 MΩ. The recorded signal was sent to a conventional bridge amplifier (IR-283; NeuroData) and to the A/D converter. The microelectrode was advanced with a hydraulic micromanipulator through the cortex up to the striatum (2.5–3 mm below the cortical surface), while continuously measuring electrode resistance by passing 0.5 nA, 100 msec current pulses. The microelectrode was left in place for 20–30 min and then slowly advanced through the striatum until a neuron was impaled. Recordings had to fulfill all of the following criteria to be included in the study: (1) resting membrane potential of at least −55 mV; (2) action potential amplitude >45 mV measured from threshold and duration inferior to 1 msec measured at half maximal amplitude; and (3) input resistance >20 MΩ. All of these parameters had to remain stable for at least 10 min (without any hyperpolarizing current). Input resistance was estimated from the membrane potential measured 80–90 msec after the onset of small amplitude hyperpolarizing or depolarizing (subthreshold) current pulses. Ordinarily, several epochs of spontaneous activity lasting 1–2 min could be recorded, and the activity evoked by cortical stimulation could be evaluated at least twice at intervals of 5–10 min. All signals were acquired with Axoscope 1.1 (Axon Instruments) at a sampling rate of 10 kHz.
After completion of the experimental manipulations, we attempted to label the neurons with Neurobiotin by passing 1 nA, 300 msec positive current pulses at a frequency of 2 Hz for at least 20 min (Kita and Armstrong, 1991). The rats were transcardially perfused with cold saline, followed by 4% paraformaldehyde in PBS. The brain was removed, stored overnight in the same fixative, and then incubated in PBS containing 15% sucrose for 24 hr.
Tissue processing. Serial 40-μm-thick coronal sections were obtained from the entire forebrain and mesencephalon. The localization of the extracellular recording sites was determined from Nissl-stained sections. Neurobiotin was revealed after a slightly modified version of the protocol of Kita and Armstrong (1991). Briefly, the sections were incubated for 4 hr in 0.1 m PBS containing an avidin-biotin-peroxidase complex (1/100, Vectastain Elite ABC kit; Vector Laboratories, Burlingame, CA) and 0.3% Triton X-100, washed in 0.1 m PBS, and incubated for 5–10 min in 0.05 m sodium acetate containing 0.5 mg/ml 3,3′-diaminobenzidine (Sigma), 0.8 mg/ml ammonium nickel sulfate, and 0.2 μl/ml H2O2. Tyrosine-hydroxylase (TH) immunohistochemistry was performed on free-floating sections following a published protocol (Murer et al., 1997).
Analysis of membrane potential fluctuations and firing pattern of striatal neurons. For all the neurons included in the study, at least two 1 min signal segments of basal activity were analyzed. Sampling was reduced to 1000 Hz by using the substitute average module of Axoscope. The signal was browsed to select epochs showing a clear two-state membrane potential. These epochs were used to construct histograms displaying the amount of time spent at any given membrane potential (all-points histograms, pClamp 6; Axon Instruments). For most recorded neurons, the histogram showed a clear bimodal profile fitting to a dual-Gaussian function (see Fig. 2). Fitting was performed with the Levenberg-Marquart method of nonlinear least squares (pClamp 6;Axon Instruments) (Stern et al., 1997). The voltage corresponding to the minimum reached by the function amid its peaks was used to estimate the time of state transitions in the signal. The time spent in the up and down states was estimated with custom-made software that records a state each time that voltage remains for >100 msec below or above the selected transitional value. This approach was preferred to the alternative of estimating time spent in the lower and upper voltage ranges of the distribution directly from the histogram, because the duration of individual states probably influences distinctively the activity of voltage-sensitive ionic conductances and firing probability (Nicola et al., 2000). The mode inside each of the two resulting series of voltage values was considered to represent the steady state reached during the down and up states.
Spectral analysis was used to characterize the periodic components in the membrane potential of striatal neurons. Spectra were computed from 1 min segments of signal (down-sampled to 1000 Hz), yielding a spectral resolution of 0.017 Hz. For the spontaneously active neurons, smoothing with a 20 points moving average was used to truncate the action potentials before performing the fast Fourier transform (FFT). Spectral densities were obtained using a Hamming window (width, 5). Relative power was calculated by normalization to the total power within the frequency range 0.017–10 Hz (the power of frequencies above 10 Hz was negligible). Peaks exceeding the 95% confidence interval of the mean relative spectral power were considered significant.
Neurons that did not spontaneously fire any action potential during the recording period were considered “silent.” For the spiking neurons, the number of up states yielding at least one action potential was determined. Interspike interval (ISI) streams were obtained from the intracellular recordings by means of amplitude discrimination. Autocorrelograms were computed from 1 min segments of signal using 10 msec bins over 1000 bins (yielding a total lag of 10 sec). The autocorrelograms were smoothed using a moving average method and then subjected to an FFT, yielding power spectra with 0.1 Hz resolution. Because the purpose of the analysis was to determine whether an ∼1 Hz rhythm can be retrieved from the spike trains of individual striatal neurons, this analysis was only performed on cells firing at rates >1 Hz.
Several measures were performed on the responses induced by cortical stimulation, including the time from the beginning of the stimulus artifact to the peak of the fast depolarizing postsynaptic potential (dPSP), the voltage at the peak of the fast dPSP, and the duration and amplitude of the long-lasting hyperpolarizing and depolarizing changes that follow the fast dPSP (Wilson, 1993). The same rationale that was applied to the analysis of naturally occurring fluctuations in membrane potential was used to examine the long-lasting changes evoked by cortical stimulation. The voltage chosen to establish the occurrence of spontaneous state transitions in the signal was used to define the beginning and end of the long-lasting changes that were evoked by cortical stimulation. The segments of signal including the long-lasting responses were cut off and used to construct histograms of the frequencies distribution of membrane potential, which showed bimodal profiles. The mode inside each part of the frequencies distribution of voltage values was considered to represent the steady state that was reached during the long-lasting hyperpolarizing and depolarizing phases of the response.
Analysis of the simultaneously acquired ECoG and striatal recordings. For each ECoG–intracellular recording pair, at least three disjoined 30 sec epochs (typically 4–6) were studied. Sampling was reduced to 1000 Hz as described above, and the signals were standardized. Cross-correlograms were computed for delays of 3 sec with a resolution of 3 msec. Peaks exceeding three correlated white noise SEs were considered significant. Cross-correlation analysis allowed estimation of the phase shift between the signals (time lag of the highest peak) and the frequency of the dominant coincident oscillatory activity (the reciprocal of the time between two successive peaks) (Lopes da Silva et al., 1989). The similarity in the oscillatory frequency content of the signals was further evaluated using spectral analysis and coherence estimation (Lopes da Silva et al., 1989). Power spectra were computed as described above (resolution for 30 sec signal segments = 0.033 Hz). Coherence was calculated from the cross-spectral density between the two signals normalized by the spectral density of each signal. Coherence attains its highest value when the phase shift and ratio between the amplitudes of the two waveforms remains constant. A significant coherence (>0.75 in at least three disjoined 30 sec signal segments) (Rosenberg et al., 1989) at the dominant frequency of the cross-spectrum, was accepted as an indication of a high probability of oscillatory synchronization and allowed estimation of the phase relationship between the signals. Phase lags were calculated from portions of the phase spectrum encompassing the frequencies showing a significant coherence and a linear phase relationship between the signals (Lopes da Silva et al., 1989). Time series analysis was performed using Statistica 4.2 (Statsoft Inc., Tulsa, OK). Because the ECoG was high-pass filtered at 0.1 Hz, the time lags were corrected using a series of coefficients that were derived from measurements of the phase lag introduced by the filter to digitally generated sinusoidal waveforms (with a 0.1 Hz resolution).
Statistical comparisons. The Fisher exact probability test was used to evaluate the relationship between two dichotomous variables. The Student's t test was used for two-group comparisons involving a single continuous variable. If the data were not normally distributed, or had unequal variances, the Mann–WhitneyU test was used instead of the t test. Normality was assessed with the Kolmogorov–Smirnov test, and homogeneity of variances was assessed with Levene's test. To compare sham-lesioned and 6-OHDA-lesioned rats along two or more variables, ANOVA was used. The Kruskal–Wallis ANOVA by ranks was preferred for multiple comparisons involving interrelated proportions.
Effects of 6-OHDA-induced lesions on the membrane potential of striatal neurons
In vivo intracellular recordings were obtained from 27 striatal neurons from 21 sham-lesioned rats and 31 striatal neurons from 24 6-OHDA-lesioned rats. All neurons successfully injected with Neurobiotin were identified as MSNs (Fig.1). Among the seven successful Neurobiotin injections performed in control rats, all led to the labeling of a single spiny neuron. In contrast, 7 among the 11 successful Neurobiotin injections performed in 6-OHDA-lesioned rats revealed two labeled spiny neurons (p = 0.013, Fisher exact test). An increased probability of labeling more than one striatal neuron after a single intracellular Neurobiotin injection in rats with chronic nigrostriatal lesions has been reported by others (Cepeda et al., 1989; Onn and Grace, 1999) and was suggested to reflect increased “coupling” via gap junctions. Neurons recorded from control and 6-OHDA-lesioned rats did not differ in some of their basic electrophysiological parameters, like input resistance, firing threshold, and action potential duration and amplitude (Table1).
Twenty-two neurons in the control group and 23 in the 6-OHDA-lesioned group showed a fluctuating membrane potential (Fig.2). The membrane potential fluctuations differed between groups (Table 1) (Fig.3). The voltage during both the down and up states was significantly more depolarized in 6-OHDA-lesioned rats than control rats. Considering that up events are perceived as enabling states during which MSNs can fire action potentials (Wilson, 1993; O'Donnell and Grace, 1995), the above described alterations probably underlie the increased firing rates reported for striatal single units in 6-OHDA-lesioned rats (Arbuthnott, 1974; Schultz, 1982; Nisenbaum et al., 1986). Actually, for the spontaneously active neurons, the probability of firing at least one action potential during an up state was significantly higher in 6-OHDA-lesioned than control rats (Fig. 4). Furthermore, striatal neurons from 6-OHDA-lesioned rats spent significantly less time in the down state and required more time to shift from one state to the other (Fig. 3). The contrast between the experimental groups is further emphasized by the finding that control rats had a significantly higher proportion of silent neurons than 6-OHDA-lesioned rats (Table 1).
Spectral analysis of the membrane potential of bistable-like neurons revealed that >90% of the spectral power was concentrated in the frequency range comprised between 0.4–2 Hz (Table2; see Figs. 6 and 7). All two-state neurons had peaks within this frequency range that largely exceeded the 95% confidence interval of the mean relative spectral power. Note that for spontaneously active neurons the action potentials have been truncated before performing the FFT. Consequently, spectral profiles reflect true membrane potential frequency components. State transitions probably explain most of the spectral power within this frequency range (Stern et al., 1997). The frequency of state transitions, as estimated from counts performed on the same signal segments, was 0.95 ± 0.05 in control rats and 1.03 ± 0.07 in 6-OHDA-lesioned rats (p = 0.34, two-tailed t test).
For five neurons in the control group and eight in the 6-OHDA-lesion group, it was not possible to ascertain the existence of a two-state membrane potential. The small number of this kind of recordings in our sample prevented the investigation of between groups differences.
Response of striatal neurons to cortical stimulation
As reported previously (for review, see Wilson, 1993), cortical stimulation evoked a short latency dPSP, followed by a sequence consisting of a prolonged hyperpolarization and a plateau depolarization. The latter sequence closely resembled the spontaneous membrane potential shifts of striatal neurons (Fig.5). Although the short latency dPSP is probably produced by activation of a small volume of cortical tissue around the electrode tip, the hyperpolarization–depolarization sequence is probably driven by a hypersynchronous cortical oscillatory cycle that indicates resumption of the endogenously generated rhythm (Wilson, 1993). After a single pulse applied to a cortical area, neighbor cortical neurons display a short latency dPSP, followed by a long-lasting inhibition and a postinhibitory depolarizing plateau (Amzica and Steriade, 1998); the shape and time course of this response closely resembles that of striatal neurons.
The shape of the response of striatal neurons to cortical stimulation, time to the peak of the fast dPSP, and peak amplitude of the fast dPSP were similar in both experimental groups (Table 2). In agreement with the results reported above regarding naturally occurring state transitions in the signal, the long-lasting hyperpolarization was shorter and had a more depolarized potential in 6-OHDA-lesioned rats than in control rats (Fig. 5). The plateau depolarization had a significantly reduced latency and reached a slightly more depolarized potential (the difference between groups was not significant) in 6-OHDA-lesioned rats (Table 2).
A low-frequency oscillation is encoded in spike trains of striatal neurons
Most striatal neurons show very low mean firing rates (many are silent) in control rats, but in 6-OHDA-lesioned rats their firing rates are significantly increased (Arbuthnott, 1974; Schultz, 1982;Nisenbaum et al., 1986) (Fig. 4). Because striatal neurons fire action potentials during the up states, and the probability of firing at least one spike during an up state is increased in 6-OHDA-lesioned rats, it seemed likely that the slow membrane potential fluctuation would be transmitted more effectively to striatal target nuclei in rats with nigrostriatal lesions. To provide evidence in support of this hypothesis, we sought for low-frequency oscillations in spike trains of striatal neurons recorded from control and 6-OHDA-lesioned rats. Only those cells having firing rates higher than 1 spike per second were analyzed (Fig. 4). A significant ∼1 Hz periodic frequency component could be retrieved from 4 of 22 control rat recordings (14.8% of the sampled neurons) and from 15 of 23 6-OHDA-lesioned rat recordings (67.7% of the sample) (p = 0.0023, Fisher exact test) (see Fig. 7).
Correlation between the ECoG and the membrane potential of striatal neurons
Intracellular recordings of two-state striatal neurons were obtained simultaneously with the frontal cortex ECoG in 14 instances from 12 control rats, and in 16 instances from 13 6-OHDA-lesioned rats. For 13 signal pairs in the control group and 12 in the 6-OHDA-lesioned group, clear signs of synchronized oscillation were found. In both experimental groups, the power spectra of the signals revealed that the ECoG and the membrane potential of the simultaneously recorded striatal neuron had the highest relative power at a closely similar frequency, usually in the 0.4–2 Hz range (Figs. 6,7). The signals had a very strong coherence at the dominant spectral frequency, suggesting that the waveforms were oscillating synchronously. The experimental groups did not differ in the mean coherence at the dominant frequency or in the phase relationship between the ECoG and intracellular recording (Table3). Phase lags, estimated at the coherent frequency, ranged from −155 to +72 msec (negative means that the ECoG precedes the striatal recording), with the ECoG anteceding the intracellular signal in 15 of 25 instances (Table 3). Similarly, the cross-correlograms revealed a strong correlation between the ECoG and membrane potential fluctuations of striatal neurons (Fig. 6). Lags between the signals, estimated from the cross-correlograms, were consistent with those calculated from the phase spectra (Table 3). The close relationship between the signals was consistently retained over time, as indicated by the analysis of several nonoverlapping 30 sec epochs within time windows spanning several minutes (Fig. 6). For the remaining five signal pairs, a close relationship between the ECoG and intracellular recording could not be evidenced, i.e., they correlated weakly and showed low coherence.
Membrane potential fluctuations of striatal neurons are correlated with slow oscillatory activity in the frontal cortex electrocorticogram
In slow wave sleep and anesthesia, the EEG of animals and humans displays synchronized oscillations of different frequencies. A slow oscillation (∼1 Hz), which is generated intracortically, seems to be responsible for the grouping of delta waves and spindles, and is a major determinant of the shape of the EEG waveform (Steriade, 1999).In vivo intracellular recordings revealed that the membrane potential of cortical neurons shows rhythmic depolarizing and hyperpolarizing shifts that are strongly correlated with the EEG slow oscillatory activity (Amzica and Steriade, 1995). During the depolarizing phase of the membrane potential fluctuation, cortical neurons discharge bursts of action potentials that grant spreading of the rhythm through the cortex and thalamus (Steriade 1999). Our simultaneous recordings of striatal neurons and the frontal ECoG indicate that membrane potential fluctuations reflect spreading of cortical slow rhythms to the striatum. This conclusion is supported by previous findings indicating that up states in striatal neurons are driven by inputs from the cerebral cortex: (1) up states are abolished by decortication (Wilson 1993); (2) striatal neurons do not show up events in slices, but display up states in chronic organotypic cortex–striatum cocultures (Plenz and Aertsen, 1996); (3) corticostriatal neurons display membrane potential fluctuations that resemble those of striatal neurons in many aspects, including the presence of common low-frequency components in the power spectra (Stern et al., 1997). Studies on the origin of state transitions in nucleus accumbens spiny neurons are consistent with our results. In the nucleus accumbens, up events are driven by excitatory hippocampal input (O'Donnell and Grace, 1995), and membrane potential fluctuations are correlated with shifts in hippocampal field potential (Goto and O'Donnell, 2001). Finally, a report by Mahon et al. (2001) indicates that slow waves in the EEG are correlated with depolarizing states both in corticostriatal and striatal neurons under a variety of anesthetic conditions. It is likely that spreading of slow cortical rhythms to the striatum occurs via the massive corticostriatal pathway. In the present study, the mean time lag between cortical and striatal signals was similar to the latency of the fast dPSP that was evoked in striatal neurons by cortical stimulation.
Chronic nigrostriatal lesions alter the membrane potential of striatal neurons, allowing transmission of slow cortical oscillations to the basal ganglia output nuclei
Spreading of slow cortical rhythms to striatal target nuclei seems to be limited in healthy animals. Neurons of the GP and SNpr display tonic regular firing during natural slow wave sleep (Datta et al., 1991; Urbain et al., 2000) and anesthesia (Pan and Walters, 1988;MacLeod et al., 1990; Murer et al., 1997; Ni et al., 2000). After chronic nigrostriatal lesions, SNpr units display rhythmic burst firing (MacLeod et al., 1990; Tseng et al., 2000) that is strongly modulated by intrastriatal administration of DA receptor agonists (Murer et al., 1997; Tseng et al., 2000). The capability of MSNs to transmit slow cortical rhythms to their target nuclei is constrained by their very low firing probability in control animals. Our results suggest that after chronic nigrostriatal lesions, an increased excitability of striatal neurons allows transmission of slow cortical rhythms to striatal target nuclei (Figs. 7, 8).
Changes in the shape of the fluctuating membrane potential seem to underlie the increased firing probability of striatal neurons in 6-OHDA-lesioned rats. Striatal neurons recorded from rats with nigrostriatal lesions displayed a more depolarized membrane potential, during both the down and up states, and spent less time in the down state. The mechanisms leading to these changes remain to be determined. Current knowledge suggests that they may be multiple and complex. Dopamine increases potassium currents that keep the resting potential near the potassium equilibrium potential (Pacheco-Cano et al., 1996;Waszczak et al., 1998). A reduced activity of these currents may partly explain the more depolarized membrane potential and shorter duration of the down state in rats with nigrostriatal lesions. Dopamine receptors also modulate in a complex way currents that are active within the voltage range reached during the up events. For example, D2 receptor stimulation suppresses currents through L-type calcium channels and consequently, reduces striatal neuron excitability (Hernández-López et al., 2000). There is in vivoevidence demonstrating that L-type calcium channel blockers reduce glutamate-induced bursts in nucleus accumbens neurons (Cooper and White, 2000). Increased currents through L-type calcium channels may have yielded to a more depolarized potential during the up states, and an increased firing probability, in rats with chronic nigrostriatal lesions. In addition, in vitro evidence supports that corticostriatal transmission is increased in slices from 6-OHDA-lesioned rats. Striatal neurons do not show spontaneous firing in control or in DA-depleted slices, but the number and amplitude of spontaneous dPSPs is increased in the DA-denervated striatum (Galarraga et al., 1987; Calabresi et al., 1993). Both presynaptic and postsynaptic mechanisms (Brown and Arbuthnott, 1983; Cepeda et al., 1993; Umemiya and Raymond, 1997) may account for the increase in spontaneous corticostriatal transmission after chronic nigrostriatal lesions.
It may be argued that an increased susceptibility to damage induced by the intracellular recording electrode might have produced the differences in striatal neuronal activity between sham- and 6-OHDA-lesioned rats. This possibility seems unlikely, however, because: (1) reports based on extracellular single unit recordings have demonstrated an increased firing rate of striatal units in 6-OHDA-lesioned rats (Arbuthnott, 1974; Schultz, 1982; Nisenbaum et al., 1986); (2) action potential amplitude and duration, input resistance, and the duration of recordings were not different between control and 6-OHDA-lesioned rats (Table 1); (3) preliminary results indicate that the changes in membrane potential induced by 6-OHDA lesions can be reversed by systemic administration of dopamine receptor agonists, but not by vehicle injections (our unpublished observations).
Multiple pathways may be involved in the transmission of slow cortical rhythms to the globus pallidus and basal ganglia output nuclei
The striatum is probably not the only structure involved in transmission of slow cortical rhythms to the remaining basal ganglia. Recent work by Magill et al. (2000) demonstrated that slow cortical rhythms are reflected in spike trains of subthalamic nucleus (STN) and GP neurons in ketamine-anesthetized nonlesioned rats and suggested that cortical rhythms are propagated via cortico-STN-GP connections. Most GP neurons show tonic regular firing in natural slow wave sleep (Urbain et al., 2000), as well as under the influence of other anesthetics (Pan and Walters, 1988; Magill et al., 2000; Ni et al., 2000) and in paralyzed animals (Pan and Walters, 1988), however, suggesting that ketamine may propitiate the propagation of slow cortical rhythms through the basal ganglia. After chronic nigrostriatal lesions, GP neurons display rhythmic burst firing in any condition tested (Pan and Walters, 1988; Ni et al., 2000). A role for the STN in spreading cortical rhythms through the basal ganglia in the parkinsonian condition is supported by previous reports showing that the proportion of SNpr units displaying rhythmic bursting is reduced to half (Tseng et al., 2000), and rhythmic bursting in the GP is almost abolished (Ni et al., 2000) after STN lesions in 6-OHDA-lesioned rats.
Implications for Parkinson's disease
In the awake monkey, basal ganglia output nuclei neurons display nonoscillatory firing patterns. However, after methyl-phenyl-tetrahydropyridine-induced lesions, a significant proportion of units in the STN and basal ganglia output nuclei exhibit an oscillatory pattern of action potential firing (Bergman et al., 1994; Nini et al., 1995; Wichmann et al., 1999; Raz et al., 2000). It is becoming increasingly evident that a substantial proportion of units in the STN and GP exhibit oscillatory firing in individuals with Parkinson's disease (Hurtado et al., 1999; Levy et al. 2000; Magnin et al., 2000). The oscillatory basal ganglia output is supposed to give rise to tremor, one of the cardinal signs of the illness. Furthermore, it has been postulated that an oscillatory basal ganglia output may disrupt cortical information processing by promoting the recruitment of thalamocortical circuits in rhythmic firing patterns (Brown and Marsden, 1998). Although the phenomenon resembles what happens in rats after 6-OHDA-induced lesions, the main frequency of oscillatory activity in behaving parkinsonian primates (3–20 Hz) is higher than that found in anesthetized 6-OHDA-lesioned rats (0.4–2 Hz). If abnormal spreading of cortical slow rhythms through the striatum and STN underlies the rhythmic firing pattern of output nuclei neurons in parkinsonism, the different oscillatory periods observed in awake parkinsonian primates and anesthetized 6-OHDA-lesioned rats may merely reflect the distinct dominant EEG frequencies that characterize each behavioral state. It seems possible that a population of more excitable striatal neurons will be prone to transfer high-frequency cortical rhythms (like those observed in the cortex during quiet waking) to striatal targets. The reduced latency of the plateau depolarization that was evoked by cortical stimulation in 6-OHDA-lesioned rats suggests that more excitable striatal neurons can be driven easily to a depolarized (enabling) state. We are not aware of studies aimed at characterizing the oscillatory behavior of basal ganglia neurons in parkinsonian monkeys during slow wave sleep, nor those of awake 6-OHDA-lesioned rats. Undoubtedly, this issue will be clarified in the near future.
This work was supported by the Ministerio de Salud y Acción Social de la Nación (Beca Carrillo-Oñativia), Fundación Antorchas, Consejo Nacional de Investigaciones Cientı́ficas y Técnicas, y Universidad de Buenos Aires (Argentina). We thank Dr. Patricio O'Donnell for helpful comments on an earlier version of this manuscript, Dr. José L. Ferrán for excellent assistance with microphotography and artwork, and Departamento de Ciencias Biológicas, Facultad de Ciencias Exactas y Naturales, Universidad de Buenos Aires, for their help with time series analysis.
Correspondence should be addressed to Kuei Y. Tseng, Departamento de Fisiologı́a, Facultad de Medicina, Universidad de Buenos Aires, Paraguay 2155, Buenos Aires (1121), Argentina. E-mail:.