Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

Recent electrophysiological studies of single-unit spiking activity in several basal ganglia nuclei of awake rats have revealed periodic variations in firing rate at s-to-min time scales (Twery et al., 1996; Allers et al., 1998; Ruskin et al., 1999). These multisecond oscillations were found in a majority of neurons in the globus pallidus (GP), substantia nigra pars reticulata, entopeduncular nucleus, and subthalamic nucleus. Activation of dopamine (DA) receptors with the directly-acting mixed D1/D2 agonist apomorphine caused striking changes in these slow patterns. Specifically, apomorphine increased the frequency of oscillations: periods of oscillatory cycles decreased from a mean of ~30 to ~15 s (Ruskin et al., 1999). A combination of selective D1 and D2 agonists also decreased oscillatory period (Ruskin et al., 1999). These data reveal that modulation of multisecond oscillations is a newly described means by which DA receptors can influence basal ganglia physiology.

Activation of DA receptors also causes striking behavioral changes in animals and humans. DA agonists induce hyperactivity as well as stereotypic behavior (Ernst, 1967; Lyon and Robbins, 1975). In normal rodents, DA agonist-induced stereotypic behavior takes the form of strong, repetitive sniffing or chewing. These behavioral effects of DA agonists involve both the D1 and D2 subfamilies of DA receptors. For instance, the stereotypic behavior induced by apomorphine or the DA-releasing agent d-amphetamine is blocked by either D1 or D2 subfamily antagonists (Fog et al., 1971; Honda et al., 1977; Iorio et al., 1983; Molloy and Waddington, 1984). Also, injection of selective D1 and D2 agonists together produces a stronger behavioral response than does injection of either alone (Braun and Chase, 1986; Arnt and Perregaard, 1987; Walters et al., 1987; White et al., 1988). These behavioral data illustrate that D1 and D2 receptors interact positively, or synergistically, in controlling motor function.

The effects of DA agonists on behavior are typically ascribed to DA receptor-mediated changes in the mean firing rate of basal ganglia neurons, and, as for behavior, many of these net shifts in tonic firing rate are controlled synergistically by D1 and D2 receptors in normal animals. For instance, studies using either systemic drug injection or local drug iontophoresis have shown that D1 and D2 receptors synergize to change firing rates in striatal nuclei (White, 1987; Hu and Wang, 1988; Rosa-Kenig et al., 1993). Cooperative D1/D2 effects are also found in electrophysiological studies of other basal ganglia nuclei, and are particularly evident in the DAergic control of GP firing rates. Apomorphine and d-amphetamine increase mean firing rates of pallidal neurons roughly 2-fold (Bergstrom and Walters, 1981; Bergstrom et al., 1982), and this effect is blocked by either D1 or D2 antagonists (Carlson et al., 1986). Although selective D1 agonists have variable effects on pallidal neuron firing rate, and D2 agonists alone cause mild rate increases, their combination produces large excitatory effects (Carlson et al., 1987; Walters et al., 1987; Ruskin et al., 1998). Therefore, D1 and D2 receptors interact positively to shift mean firing rates in the GP (and other basal ganglia nuclei).

As already noted, however, neuronal firing activity in the basal ganglia is characterized not only by firing rate but also by slow oscillatory patterns in firing rate. The prominent effects of apomorphine on multisecond oscillations in the GP and entopeduncular nucleus described above were reversed by either SCH 23390 or eticlopride (D1 and D2 subfamily antagonists, respectively), suggesting that both D1 and D2 receptor subfamilies were involved (Ruskin et al., 1999). To more fully characterize the D1/D2 receptor control of this phenomenon, and to extend the previous results, which only used a single combined dose of D1 and D2 agonists, the present study examines the effects of selective D1 and D2 agonists, administered alone or sequentially at several doses.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Electrophysiology. Sprague-Dawley rats (Taconic Farms, Germantown, NY) weighing 250 to 400 g were used. Extracellular recordings of tonically active single-units (with biphasic Type II waveforms; Kelland et al., 1995) in the GP were performed in artificially respired, locally anesthetized rats as described previously (Ruskin et al., 1998). All surgical procedures have been described previously (Ruskin et al., 1998), and were conducted in accordance with National Institutes of Health guidelines (Cohen et al., 1985). Rats were tracheotomized under anesthesia with halothane (Halocarbon Laboratories, River Edge, NJ) and the trachea was intubated with a 13- or 14-gauge cannula. To prevent discomfort, incision and pressure sites were thoroughly infiltrated with the long-acting local anesthetic mepivacaine HCl (Sanofi Winthrop, New York, NY), 2% lidocaine anesthetic gel (Astra USA, Westborough, MA) was applied to the outside of the tracheal cannula and the tips of the stereotaxic ear bars, and corneal drying was prevented with Lacri-Lube (Allergan Pharmaceuticals, Irvine, CA). After placement in a stereotaxic frame and the completion of all surgical procedures, halothane anesthesia was discontinued, and rats were paralyzed with the injection of gallamine triethiodide (16 mg/kg) through a tail vein. Rats were then artificially ventilated at a rate adjusted to maintain expired CO₂ levels between 3.4 and 4.5%. Supplements of gallamine were given as needed. Body temperature was maintained with a heating pad.

Spectral Analysis. Data segments (180-s) were selected from baseline and postdrug times for analysis of firing rate periodicities with the Lomb periodogram (Kaneoke and Vitek, 1996). One segment each was selected from baseline, postagonist and (when present) postantagonist epochs. Data segments after drug treatments were generally taken from the 5- to 10-min range postinjection. Preliminary analyses indicated that oscillatory firing rate variations with periods of several s or several tens of s were best revealed with bin widths of less than 1 s, and so spike trains were smoothed with square, nonoverlapping bins of 200 ms width for Lomb algorithm analysis (for illustrative purposes, example spike trains in Figs. 1 and 2 are shown smoothed with 500-ms bins). The Lomb periodogram (Kaneoke and Vitek, 1996) was used to characterize periodicities in the range of 2 to 60 s (0.5-0.017 Hz). This method was selected because of the ease with which the Lomb algorithm assesses the statistical significance of features in the power spectrum (Scargle, 1982). In the present study, the method (Kaneoke and Vitek, 1996) was modified so that power spectra were taken from smoothed spike trains instead of smoothed autocorrelograms. Oscillations with periods shorter than 2 s are often directly related to ventilation (Ruskin et al., 1999) and were not presently considered. An upper period limit of 60 s was utilized to widely bracket the typical periods of strong postDA agonist oscillations (most often in the 5- to 25-s range); 180-s segments of data were analyzed to provide sufficient time for even the longest period presently examined (60 s) to exhibit three cycles. Spectral peaks in the Lomb periodogram power spectra were considered to be significant at p < .01 in comparison with independent Gaussian random values. Only 1 of 100 spectra of segments of random Gaussian noise would be expected to have peaks above the power level represented by the p = .01 line, hence peaks above this line are highly likely to represent true periodic activity (Horne and Baliunas, 1986; Press et al., 1992). Spectral peak frequency (period) was measured only for spectral peaks with significant height.

View larger version (18K): [in this window] [in a new window]   Fig. 1.   A complete firing rate record of a GP neuron from an awake rat. The baseline activity of this neuron has only small and irregular variations in rate. Little change is seen after injection of 0.26 mg/kg quinpirole, but 1 to 2 min after subsequent injection of 0.39 mg/kg SKF 81297 there is an emergence of high-amplitude, regular oscillations in firing rate, as well as an large overall increase in firing rate. The oscillations persist until the injection of the D2 antagonist eticlopride (0.2 mg/kg). Oscillatory activity is blocked and the overall firing rate increase is reversed by ~2 min after eticlopride injection. All drug injections are i.v. The smoothed spike train is scaled to its highest point (120 Hz).

View larger version (39K): [in this window] [in a new window]   Fig. 2.   Lomb periodogram power spectra and the spike trains from which they were derived, showing examples of basal and postdrug smoothed GP spike trains (A-D). Drugs are injected sequentially at 10-min intervals during each recording. Spike trains are shown scaled to their highest point (in Hz). Mean firing rate is also given in the lower left corner of each spike train plot. Segments (90-s) of the 180-s trains used for analysis are shown. Beneath spike trains are the corresponding power spectra, which are shown with a line signifying the p = .01 significance level; spectral peaks above this line indicate significant periodicities (Horne and Baliunas, 1986; Kaneoke and Vitek, 1996). Several cases have multiple significant peaks. Spectra for a neuron are scaled to the highest spectral peak for that neuron. A, analysis of the baseline of this neuron reveals a spectral peak (at 29 s, or 0.034 Hz) just crossing the significance line. There are no significant spectral peaks after injection of 1.0 mg/kg quinpirole; however, subsequent injection of 1.0 mg/kg SKF 81297 causes a striking oscillation in firing rate and the appearance of powerful spectral peaks. The tallest spectral peak has a period of 7.0 s; another strong adjacent peak has period of 7.9 s. These periods correspond to frequencies of 0.13 to 0.14 Hz. Significant periodic activity is abolished by injection of 0.2 mg/kg eticlopride. B, there are no significant periodicities in baseline or postSKF 81297 (0.39 mg/kg) epochs from this neuron. Subsequent injection of 0.1 mg/kg quinpirole causes notable (although somewhat irregular) firing rate oscillations and several strong spectral peaks. The tallest of these peaks has a 15-s period (corresponding to 0.067 Hz). Injection of eticlopride (0.2 mg/kg) eliminates the strong oscillatory activity; the power spectrum has one significant but relatively low-power peak, at 57 s. C, baseline fluctuations in firing rate of this neuron are revealed as multiple significant spectral peaks, the tallest of which has a 25-s period. After injection of 5.0 mg/kg SKF 81297, stronger, more regular firing rate oscillations appear; the tallest spectral peak here is at 11 s (0.091 Hz). Subsequent injection of SCH 23390 (2.0 mg/kg) greatly reduces the oscillatory activity. D, in this neuron, SKF 82958 (1.0 mg/kg) induces isolated, periodic increases in firing rate, leading to a strong spectral peak at 12 s. After subsequent quinpirole (1.0 mg/kg), the rate of the oscillations increases; the main spectral peak is at 6.6 s.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Slow Oscillations in Baseline Activity. Ninety GP neurons were recorded from awake, artificially respired rats. Although some of these neurons appeared to have fairly stationary basal firing rates (Figs. 1 and 2B), in most neurons basal firing rates showed slow increases and decreases on time scales of many s or tens of s (Fig. 2, A and C). Spectral analysis revealed oscillations in basal firing rate to be significantly periodic in many cases: significant spectral peaks (in the 2- to 60-s range of periods) were found in spectra from baseline spike trains of 64 of 90 neurons (71%). Power in these spectra was distributed widely over the analyzed range, as can be seen in the distribution of periods of the `main' (most powerful) spectral peaks from baseline epochs (illustrated for many different treatment groups in Figs. 3 and 4). The mean ± S.E.M. for main peak period of all basal spike trains was 32.5 ± 2.0 s (Table 1). The wide distribution of oscillatory frequencies was also evident when considering all significant spectral peaks (instead of only the main spectral peak) in the baseline power spectra of several single neurons having multiple significant spectral peaks, e.g., Fig 2C.

View larger version (28K): [in this window] [in a new window]   Fig. 3.   Effects of D1 and D1/D2 receptor activation on oscillations in firing rate. Data points indicate the main, or most powerful, spectral peak of a neuron within a given condition. Mean ± S.E.M. is also shown. Cases without significant periodicity (i.e., no spectral peaks cross the p = .01 significance level) are indicated at the top of each graph ("no osc".). Drugs are injected i.v. in the order shown in each graph, from left to right. Drug doses (in parentheses) are in mg/kg. Number of neurons is shown at the bottom of each bar. ANOVA results (F and p values) for each treatment group are given above each graph. A, there is no overall drug effect of a low dose of SKF 81297 (0.39 mg/kg) followed by a low dose of quinpirole (0.1 mg/kg). B and C, in each of these groups, 1.0 mg/kg SKF 81297 significantly shortens periods. Subsequent injection of quinpirole (0.26 or 1.0 mg/kg) causes further shortening. The effects are reversed by SCH 23390. D, SKF 81297 (5 mg/kg) significantly reduces main spectral peak periods, and this effect is reversed by SCH 23390. SKF: SKF 81297, quin: quinpirole, etic: eticlopride, SCH: SCH 23390, ns: nonsignificant. *p < .05, **p < .01, ***p < .001 Dunnett's t tests.

View larger version (23K): [in this window] [in a new window]   Fig. 4.   Effects of D2 and D2/D1 receptor activation, and D2 receptor blockade, on oscillations in firing rate. Data points indicate the main, or most powerful, spectral peak of a neuron within a given condition. Conventions as in Fig. 3. A, although 0.26 mg/kg quinpirole has no effect, subsequent administration of 0.39 mg/kg SKF 81297 strongly shortens oscillation periods. The effect is reversed by a final injection of eticlopride. B, 1.0 mg/kg quinpirole has no consistent effect, but subsequent administration of SKF 81297 (1.0 mg/kg) strongly reduces main spectral peak periods. The effect is reversed by eticlopride. C, a minor reduction in mean period is seen after eticlopride (0.2 mg/kg), with no further reduction after subsequent SKF 81297 (1.0 mg/kg): the overall effect of the drug treatments is not significant. *p < .02, **p < .01 Dunnett's t tests.

Effects of D1 Agonist. The selective D1 subtype agonist SKF 81297 modulated the slow periodicities of GP neurons in a dose-related manner. Visual inspection of spike trains suggested little effect of 0.39 mg/kg SKF 81297; this impression was supported by the lack of significant action on the period of main spectral peaks (Fig. 3A; example in Fig. 2B). A higher dose (1.0 mg/kg) caused visually apparent increases in the rate of oscillations in many neurons, and analysis revealed that this dose had significant effects in two separate groups of neurons (Fig. 3, B and C). In both groups, 1.0 mg/kg SKF 81297 caused the mean of main spectral peak periods to decrease to ~16 s. A higher dose (5.0 mg/kg) also significantly shifted oscillatory activity into a faster range (Fig. 3D; example in Fig. 2C), but did not have a much greater effect than that of 1.0 mg/kg, as the mean period was shifted to 14.1 ± 2.9 s. The effects of 5.0 mg/kg SKF 81297 were reversed by the D1 antagonist SCH 23390 (Figs. 2C and 3D). Another D1 agonist, SKF 82958, was also tested (1.0 mg/kg, n = 9), and had effects similar to those of SKF 81297, namely, significantly decreasing the mean of main spectral peak periods to 14.4 ± 2.2 s (p < .05, Fig. 2D).

Effects of D2 Agonist. The D2 subtype agonist quinpirole modestly increased overall firing rates in most GP neurons at both 0.26 and 1.0 mg/kg, yet had variable effects on slow oscillations. Visual inspection of spike trains showed that strong, regular oscillations were found after quinpirole (particularly after 1.0 mg/kg) in some cases, but not in the majority of neurons (e.g., Figs. 1 and 2A). Spectral analysis revealed that neither 1.0 nor 0.26 mg/kg quinpirole had significant effects on main spectral peak period compared with basal data (Fig. 4, A and B).

Effects of Combined D1 and D2 Agonists. Some of the most striking firing rate oscillations occurred in epochs after the administration of both SKF 81297 and quinpirole. For example, 1.0 mg/kg quinpirole had little effect on slow pattern for the neuron illustrated in Fig. 2A, but the subsequent administration of 1.0 mg/kg SKF 81297 induced high amplitude, regular oscillations with periods of 7 to 8 s. Group data for this regimen are shown in Fig. 4B. In cases in which SKF 81297 was administered first, and had already significantly shortened main spectral peak periods, subsequent injection of quinpirole moved periods into an even faster time range (<10 s in most cases; Fig. 3, B and C). These effects on main spectral peak period of combined SKF 81297 and quinpirole (at 1.0 mg/kg each) occurred regardless of the order of administration (compare Figs. 3C and 4B). There was also no apparent "order effect" for drug-induced increases in overall firing rate (data not shown). Doses of SKF 81297 and quinpirole, which by themselves had no significant effect on oscillations (0.39 and 0.26 mg/kg, respectively), strongly reduced the periods of the oscillations when combined (Figs. 1 and 4A). The combination of SKF 81297 and quinpirole increased oscillatory frequency in a manner that was clearly dose-related, as illustrated by the progressive decrease in main spectral peak period with increasingly greater combined doses of these drugs (Table 1). Even the lowest dose combination (0.39 mg/kg SKF 81297 and 0.10 mg/kg quinpirole) caused some neurons to oscillate in the 5- to 20-s range (Fig. 2B), although the effect was not statistically significant for the population. This latter dose combination was at the threshold for causing significant increases in overall firing rate (a modest 33% increase, p = .05; Table 1). The progressive decrease in oscillatory periods was also evident when considering the integrated area beneath the power spectrum line in the 4- to 10-s band (as a fraction of total integrated area). Increasing combined doses of SKF 81297 and quinpirole caused clear dose-related increases in the amount of power in the 4- to 10-s band, with an increase of 234% compared with baseline at the highest combined dose (Table 1).

View larger version (95K): [in this window] [in a new window]   Fig. 5.   Spectrograms illustrating the time-dependent effects of DAergic drugs on multisecond periodicities in firing rate. y-axes give both frequency and period (left and right axes, respectively). Doses of injected drugs are shown (in mg/kg). A, spectrogram of the recording shown in Fig. 1. Quinpirole (`quin') at 0.26 mg/kg has little effect, apart from a temporary postinjection emergence of power at low (<0.05 Hz) frequencies. SKF 81297 (`SKF'; 0.39 mg/kg) given 10 min later causes a prompt emergence of a discrete band of power at ~0.15 Hz. This band decreases slightly in frequency over the next few min, stabilizing at ~0.10 Hz. The band remains until the subsequent injection of eticlopride (`etic'). B, this spectrogram is from the same neuron analyzed with the Lomb method in Fig. 2A. The four 180-s segments analyzed in Fig. 2A are indicated by dashed lines under the spectrogram. Quinpirole (1.0 mg/kg) has little effect, although some increase in power is noted 8 to 9 min after injection. A subsequent injection of SKF 81297 (1.0 mg/kg) causes the prompt emergence of a discrete band of power in the 0.15- to 0.20-Hz range. The band slowly drifts in frequency; within the 10 min until the injection of the next drug there is an overall slight decrease in frequency to 0.10 to 0.15 Hz. Injection of eticlopride disrupts the band.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

This study illustrates that the spiking activity of most tonically-active pallidal neurons in awake, immobilized rats is not stationary, but rather oscillates periodically at time scales of many seconds. These slow oscillations in firing rate were also observed in another experimental group of GP neurons, as well as in entopeduncular and substantia nigra neurons (Ruskin et al., 1999). Spectral analysis revealed that these oscillations are periodic in a majority of neurons and generate statistically-significant spectral peaks. The spectral peaks from baseline pallidal spike trains are distributed widely across the presently examined range of periods. Although some GP neurons have periodicities in a slightly faster range (0.8-2 s) that are related to artificial ventilation, longer period oscillations are presumably generated by endogenous processes and are sensitive to general anesthesia (Ruskin et al., 1999). Periodic activity in the 2- to 60-s range was strongly modulated by DAergic agonists, and in particular, oscillatory frequency was increased in a dose-related manner by combined treatment with selective D1 and D2 agonists.

Although the present work focuses on data from locally-anesthetized rats that have been immobilized with a paralytic agent and held with a stereotaxic frame, slow periodicities in electrical activity are not peculiar to this preparation. Oscillations in the seconds-to-minutes range have been reported in studies of cerebral cortical activity in awake subjects utilizing electroencephalography or measurement of direct cortical potentials (Aladjalova, 1957; Norton and Jewett, 1965; Ehlers and Foote, 1984; Keidel et al., 1990; Novak and Lepicovska, 1992). In these studies, the subjects (including both humans and animals) were either unrestrained but resting or lightly restrained. Single-unit recordings have shown that slow, regular oscillations in the spiking activity of lateral geniculate neurons are present in freely-moving rats (Albrecht et al., 1998). More specifically with respect to the basal ganglia, Rebec and colleagues (Gulley et al., 1998) have found slow periodicities in substantia nigra pars reticulata spike trains from freely-moving rats (termed "irregular" in their report, referring to the variation between the local rate and the overall average rate).

Although the D2 agonist quinpirole alone did not consistently modulate firing rate periodicities within the population of units recorded, the D1 agonist SKF 81297 given alone (at 1.0 and 5.0 mg/kg) significantly shifted the average of spectral peak periods into a faster range. It is likely that this effect of SKF 81297 is specifically mediated by D1, and not D2, receptors, because these doses have little effect on substantia nigra DA neuron firing rate, which provides a sensitive functional assay for D2 receptor activation (Ruskin et al., 1998). The present data therefore dissociate DAergic control of firing rate and firing pattern, because SKF 81297 had no consistent effect on overall GP firing rates, but rather caused a range of effects on mean firing rate which differed from unit to unit (Ruskin et al., 1998). Hence, although the most robust modulation of pallidal slow periodicities in this study occurred after combined D1/D2 receptor activation, which also resulted in consistent increases in overall firing rates, DAergic modulation of pallidal slow periodicities is not limited to those neurons that exhibit an overall increase in firing rate due to the DAergic treatment. Additionally, recordings in the entopeduncular nucleus and substantia nigra pars reticulata illustrate apomorphine-induced increases in oscillatory frequency in neurons with overall decreases, overall increases, or little overall change in firing rate due to the apomorphine (Ruskin et al., 1999). The D1 receptor modulation of firing pattern in the GP is not in agreement with basal ganglia models that hypothesize a segregated influence of D1 receptors on the "direct" (striatonigral, striatoentopeduncular) basal ganglia pathway, which bypasses the GP, but are concordant with previous data demonstrating D1 receptor control of pallidal activity (Carlson et al., 1986; Trugman and James, 1993; Ruskin and Marshall, 1995).

Although SKF 81297 injected alone increased the oscillatory frequency, this effect was not completely independent of D2 receptors. SKF 81297 no longer had a significant effect on spectral peak period or integrated power in the 4- to 20-s range when administered after a prior injection of the D2 antagonist eticlopride. This result illustrates that the D1 receptor control of slow oscillations in firing rate is at least partly dependent on endogenous DA activity at D2 receptors. A similar dependence of D1 agonist effects on D2 receptor tone has been found in studies of subthalamic and striatonigral function in normal animals (Kreiss and Walters, 1997; Wang and McGinty, 1997). Regarding the behavioral effects of D1 agonists in normal animals, possibly the best studied is the strong grooming response, which some reports have found to be attenuated by removal of endogenous D2 receptor tone (Braun and Chase, 1986; Murray and Waddington, 1989), although data from other studies have not supported this finding (Molloy and Waddington, 1984; White et al., 1988). Receptor subtype interdependence is not unidirectional, however; the moderate increase in overall firing rate in the GP due to D2 agonist administration is dependent on endogenous D1 receptor tone (Carlson et al., 1986), as is the moderate motor activation due to D2 receptor activation (Gershanik et al., 1983, Johnson et al., 1976; Walters et al., 1987).

Positive interactions of D1 and D2 receptors were not only evident in the dependence of D1 agonist effects on endogenous D2 receptor tone, but were also found with combined administration of D1 and D2 agonists. For instance, low doses of SKF 81297 (0.39 mg/kg) and quinpirole (0.26 mg/kg), which alone have no significant effect on slow periodicities, together significantly decreased main spectral peak periods. Also, although SKF 81297 at higher doses (1.0 or 5.0 mg/kg) had significant effects when given alone, the apparent maximal effect of this drug was to shift spectral peak period means from 30-40 to 14-16 s. The further addition of quinpirole (0.26 or 1.0 mg/kg) shifted spectral peaks to even shorter periods (means of 8-10 s). Hence, the D1 receptor control of slow oscillations in pallidal firing rates is potentiated by D2 receptor activity. In a previous study, the mixed D1/D2 receptor agonist apomorphine (0.32 mg/kg) reduced spectral peak periods to ~15 s in the GP (Ruskin et al., 1999). Because the combination of SKF 81297 and quinpirole was able to reduce spectral peak periods even more, it is likely that 0.32 mg/kg is a submaximal dose of apomorphine for modulating slow firing rate oscillations. A synergistic D1/D2 control of slow firing rate oscillations is also apparent in recordings of subthalamic nucleus neurons in neurologically-intact rats (Allers et al., 1998).

The effects of DA agonists on slow oscillations in firing rate were reversed by either D1 or D2 antagonists. However, the postantagonist spike trains of many pallidal neurons still retained some periodic activity as shown by the presence of significant peaks in the power spectra. Significant spectral peaks remained even in conditions in which both D1 and D2 receptors were blocked (Fig. 4C, after final injection), and under no postantagonist condition was the distribution of main spectral peaks significantly different from basal. These results suggest that slow periodicities in pallidal firing rate can occur even in the absence of normal DAergic tone, although they are robustly modulated by increases in DA receptor activity. This conclusion is supported by studies in which unilateral lesion of the midbrain DA neurons did not abolish slow firing rate oscillations in the ipsilateral substantia nigra pars reticulata or subthalamic nucleus (Twery et al., 1996; Allers et al., 1998). In contrast, multisecond firing rate periodicities are highly sensitive to the modulation of brain function by anesthetics. Periodicities in the 2- to 60-s range are virtually eliminated by general anesthesia with urethane or chloral hydrate (Ruskin et al., 1999), suggesting that this phenomenon is sensitive to the overall changes in central activity caused by anesthesia. Studies from other laboratories have shown that at least some neurons in the ventral pallidum slowly oscillate in chloral hydrate-anesthetized animals, with periods ranging from slightly slower than 1 Hz (Lavin and Grace, 1998) to slower than once per minute (Maslowski and Napier, 1991).

Periodicities in neuronal activity have been reported across a wide range of frequencies, from 200 Hz in entorhinal cortex (Chrobak and Buzsáki, 1998) to one cycle per day in the suprachiasmatic nucleus (Weaver, 1998). In the present study, the spiking activity of many GP neurons was found to oscillate at rates of one cycle every several seconds to one cycle per minute. The medium spiny neurons of the striatum, which project directly to the GP, are notable for periodically switching between depolarized and hyperpolarized states (Wilson and Groves, 1981). Because this switching typically occurs at rates of ~1 Hz in awake animals, it appears unlikely that it relates directly to the slower phenomenon in the present study. Previous studies have shown that GP neurons in awake subjects can oscillate at frequencies of 4 to 10 Hz (Nini et al., 1995; Hutchison et al., 1997). However, periodicities in this range were reported to occur mostly after DA depletion. The slower periodicities in the present study remain after DA receptor blockade, implying different underlying mechanisms or at least a differential DAergic control of pallidal firing rate periodicities in these two ranges. Although there is some dissociation of D1 and D2 receptor influences on GP activity (in that D1 agonist alone moderately influenced multisecond periodicities, whereas D2 agonist alone moderately increased overall firing rates), the major finding of the present study is that D1 and D2 receptors synergize to increase oscillatory frequency in the GP, much as they synergize to increase overall firing rates in this structure (Carlson et al., 1987; Walters et al., 1987).

Slow oscillations in basal ganglia firing rate could act to coordinate neuronal activity responsible for controlling motor sequences, movement timing, or attentional processes. Multisecond periodicities in vigilance/arousal state and motor output have been noted in resting monkeys and humans, respectively (Ehlers and Foote, 1984; Keidel, 1989), and the efficacy of human sensory perception and short-term memory formation can fluctuate periodically at these time scales (Stebel and Sinz, 1971; Thoss et al., 1997). The DA agonist-induced effects on multisecond oscillations in GP firing rate demonstrate that increased DA receptor activation causes an abnormal temporal patterning of basal ganglia activity, which may be related to the abnormal stereotypic behaviors caused by these DA agonists. In particular, an increase in oscillatory frequency in the basal ganglia could underlie the stimulant-induced increase in the rate of motor sequence expression and in internal "clock speed" associated with time perception (Lyon and Robbins, 1975; Meck, 1996). At a cellular level, stationary firing activity and slowly oscillating firing activity might differently affect intracellular processes. For instance, in the absence of a change in overall average firing rate, a switch from stationary to oscillatory spiking activity could result in phasic and periodic increases in intracellular calcium which have been shown in vitro to have specific effects on gene expression (Dolmetsch et al., 1998; Li et al., 1998). Although the exact relationship between cognitive, motor, and cellular processes and firing rate oscillations in the seconds-to-minute range remains speculative, the D1/D2 receptor modulation of these periodicities represents a novel means by which DA influences basal ganglia physiology and possibly behavior.