The functional loop linking the frontal lobe and the basal ganglia plays an important role in the control of motor behaviors. To delineate the principal features of motor information processing in the cortico-basal ganglia loop, the present study aimed at investigating how corticostriatal inputs from the primary motor cortex (MI) and the supplementary motor area (SMA) are transposed onto the pallidal complex and the substantia nigra. In macaque monkeys, stimulating electrodes were chronically implanted into identified forelimb representations of the MI and SMA. Subsequently, the distribution of neurons exhibiting orthodromic responses was examined in the caudal putamen to demarcate striatal zones receiving inputs separately or confluently from the MI and SMA. Finally, anterograde double labeling was performed by paired injections of tracers into two of three identified zones: the MI-recipient zone, SMA-recipient zone, and the convergent zone. Data have revealed that inputs from the MI-recipient and SMA-recipient striatal zones were substantially segregated in the pallidal complex and that those from the convergent zone were distributed to fill in blanks made by terminal bands derived from the MI and SMA. On the other hand, striatonigral inputs from the SMA-recipient and convergent zones of the putamen largely overlapped, while the input from the MI-recipient zone was minimal. The present results clearly indicate that the mode to process corticostriatal motor information through the striatopallidal and striatonigral projections is target-dependent, such that the parallel versus convergent rules govern the arrangement of striatopallidal or striatonigral inputs, respectively.
There is a consensus that the motor loop connecting the frontal lobe and the basal ganglia plays a pivotal role in voluntary movement control (Albin et al. 1989; Alexander and Crutcher 1990;Alexander et al. 1986; DeLong 1990). At the first stage of this loop, multiple areas of the frontal lobe send diverse motor information to the striatum which constitutes the main input station of the basal ganglia. Two opposite views have been proposed as to the topography of the cortico-basal ganglia loop (Alexander and Crutcher 1990; Alexander et al. 1986; Mink 1996; Parent and Hazrati 1995a). One hypothesis favors the existence of distinct parallel pathways from and to each cortical area through the basal ganglia, whereas the other supports a convergence from different cortical areas to create a given pathway through the basal ganglia and back to the cortex. Previous anatomical studies have shown that corticostriatal inputs from individual motor-related areas are distributed in an orderly arrangement according to the parallel versus convergent rules (Inase et al. 1996, 1999;Parthasarathy et al. 1992; Strick et al. 1995a,b; Takada et al. 1998a,b, 2001). The two major origins of corticostriatal motor inputs are the primary motor cortex (MI) and the supplementary motor area (SMA). We have recently revealed that while there is a preservation of somatotopy such that striatal zones receiving inputs from hindlimb, forelimb, and orofacial representations of the MI or SMA are apparently separate, corticostriatal input zones from regions of the MI and SMA representing the same body part are arranged in a partly overlapping, but largely segregated, manner (Takada et al. 1998a,b). With regard to the forelimb representation, for example, dense input zones from the MI and SMA are located primarily in the ventrolateral or dorsomedial portion of the caudal aspect of the putamen, respectively, and a striatal zone receiving convergent inputs from both motor-related areas exists in between. Such an input organization implies that motor information from distinct cortical areas is at least partly integrated within the striatum.
It is generally accepted that the outflow from the striatum reaches the two output stations of the basal ganglia, the internal segment of the globus pallidus (GPi) and the substantia nigra pars reticulata (SNr), directly or indirectly via the external segment of the globus pallidus (GPe) and then the subthalamic nucleus (STN) (Alexander and Crutcher 1990; Mink 1996; Mink and Thach 1993; Parent and Hazrati 1995a,b). To advance our understanding of the mechanisms underlying motor information processing in the cortico-basal ganglia loop, it is necessary to elucidate how corticostriatal inputs from the MI and SMA are conveyed through striatopallidal and striatonigral projections before terminating in the GPe, GPi, and SNr. Three hodological strategies seem possible. First, striatal zones receiving inputs from the MI, SMA, or from both project to individual pallidal and nigral regions (parallel processing). Second, striatal zones receiving segregated inputs from the MI and SMA project to common pallidal and nigral regions where corticostriatal inputs from both the MI and the SMA are transposed (convergent processing). Third, striatal zones receiving convergent inputs from both the MI and the SMA project to independent pallidal and nigral regions where corticostriatal inputs from the MI or SMA alone are transposed (redivergent processing). Therefore the present study was undertaken to clarify which strategy determines the distribution patterns of striatal inputs to the GPe, GPi, and SNr.
Five female Japanese monkeys (Macaca fuscata) weighing 3.8–6.3 kg were used for this study. In a series of our anterograde double-labeling experiments, paired injections of tracers were made into two of the three electrophysiologically identified zones of the putamen—an “MI-recipient zone,” an “SMA-recipient zone,” and a “convergent zone”—that had been demarcated by testing whether a given zone receives inputs from forelimb representations of the MI and SMA separately (from the MI or SMA alone) or confluently (from both the MI and the SMA). Each experiment was designed to examine the distribution patterns of striatopallidal and striatonigral inputs from the MI-recipient and SMA-recipient zones (Monkeys 1 and 2), from the SMA-recipient and convergent zones (Monkeys 3 and 4), and from the MI-recipient and convergent zones (Monkey 5) (see Table1). The experimental protocol was approved by the Animal Care and Use Committee of the Tokyo Metropolitan Institute for Neuroscience, and all experiments were conducted in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals.
Under general anesthesia with ketamine hydrochloride (10 mg/kg body wt, im) and pentobarbital sodium (25 mg/kg body wt, iv), each monkey was positioned in a stereotaxic apparatus and prepared surgically for electrophysiological mapping and subsequent tracer injections. Briefly, small stainless steel screws as anchors were affixed to the exposed skull and then covered with transparent acrylic resin, and two stainless steel tubes were mounted in parallel over the frontal and occipital lobes for head fixation (see Hatanaka et al. 2001; Inase et al. 1999; Nambu et al. 1996, 2000; Takada et al. 1998a, 2001;Tokuno et al. 1997).
Mapping of MI and SMA for implanting stimulating electrodes
Several days after the surgery, stimulating electrodes were chronically implanted into forelimb representations of the MI and SMA that had been defined by means of intracortical microstimulation (ICMS). The monkeys were anesthetized with ketamine hydrochloride (10 mg/kg body wt, im) and xylazine hydrochloride (1–2 mg/kg body wt, im), and, then, seated quietly in a monkey chair with their head fixed in a stereotaxic frame attached to the chair. After partial removal of the skull over the frontal lobe, a glass-coated Elgiloy-alloy microelectrode (0.5–1.5 MΩ at 1 kHz) was introduced perpendicular to the cortical surface at 1,000- to 1,500-μm intervals using a hydraulic microdrive. Extracellular unit activity was recorded while neuronal responses to somatosensory stimuli (skin touch or passive joint movement) were examined. Following these unit recordings, ICMS was attempted. Currents of <50 μA were delivered with a train of 12 (for MI mapping) or 22 (for SMA mapping) cathodal pulses (200-μs duration at 333 Hz), and evoked movements of various body parts were observed. According to the somatotopical map, pairs of bipolar stimulating electrodes (prepared with 200-μm-diameter enamel-coated stainless steel wires; intertip distance, 2 mm) were implanted into the forelimb regions of the MI and SMA, respectively, at a depth of 4 or 5 mm from the dural surface (Nambu et al. 2000). The electrodes were then covered with transparent acrylic resin.
Identification of MI-recipient, SMA-recipient, and convergent striatal zones
After combined anesthesia with ketamine hydrochloride (10 mg/kg body wt, im) and xylazine hydrochloride (1–2 mg/kg body wt, im), the monkeys were seated in the same monkey chair with their head restrained, and a skull portion over the orofacial region of the MI was removed. To preserve the removed portion, a rectangular chamber was attached to the skull with transparent acrylic resin. An Elgiloy-alloy microelectrode was inserted obliquely (45° from vertical in the frontal plane) through the dura mater into the putamen for recording neuronal activity. The unitary activity of putamen neurons was amplified and displayed on an oscilloscope. Then, orthodromic spike discharges evoked by electrical stimulation in the MI and SMA (0.3-ms-duration single pulses; usually 0.3–0.5 mA and, sometimes, ≤0.8 mA at 0.4–0.8 Hz) were observed. Based on the pattern of orthodromic responses to cortical stimulation and the distribution of putamen neurons exhibiting such responses, we determined three striatal zones receiving inputs from the MI and SMA: the MI-recipient, SMA-recipient, and convergent zones.
Single injections of two different anterograde tracers, wheat germ agglutinin-conjugated horseradish peroxidase (WGA-HRP; Toyobo, Osaka, Japan) and biotinylated dextran amine (BDA; Molecular Probes, Eugene, OR; 3000 M r) were made into two of the three identified zones of the putamen once the patterns of cortically induced orthodromic responses had been confirmed. Using a 10-μl Hamilton microsyringe modified for extracellular recording (Tokuno et al. 1998), we slowly deposited a volume of 0.5–0.9 μl of a 0.4–1% aqueous solution of WGA-HRP and a volume of 0.85–0.9 μl of a 20% aqueous solution of BDA (see Table 1). The BDA injection was carried out 18–19 days prior to the WGA-HRP injection.
After a survival period of 3 days for WGA-HRP and 21–22 days for BDA, the monkeys were anesthetized deeply with an overdose of pentobarbital sodium (50 mg/kg body wt, iv) and perfused transcardially with 0.1 M phosphate-buffered saline (PBS), pH 7.3, followed by 8% formalin dissolved in 0.1 M PBS, pH 7.3, and, finally, the same fresh PBS containing 10% and then 30% sucrose. The removed brains were saturated with 30% sucrose in 0.1 M PBS, pH 7.3, at 4°C and cut serially into 60-μm-thick coronal sections on a freezing microtome. Every sixth section was histochemically stained for WGA-HRP, and the adjacent sections were stained for BDA. For visualizing injected and transported WGA-HRP, the sections were reacted with tetramethylbenzidine. For visualizing injected and transported BDA, the sections were treated with avidin-biotin-peroxidase complex (ABC Elite, Vector Laboratories, Burlingame, CA) and then reacted with diaminobenzidine. All sections were mounted onto gelatin-coated glass slides and counterstained with 1% Neutral Red. Other technical details were as previously described (Inase et al. 1999;Nambu et al. 1996; Takada et al. 1998a,2001).
The injection sites of WGA-HRP and BDA in the putamen were charted in tracings of equidistant coronal sections by the aid of a profile projector. The patterns of anterograde labeling in the GPe, GPi, and SNr were plotted in projection drawings of representative coronal sections by the aid of both a profile projector and a light microscope. In cases where tetramethylbenzidine-reacted (WGA-HRP-stained) sections were subjected to some shrinkage, their projection drawings were expanded as large as BDA-stained sections. Then, the projection drawings of the WGA-HRP-stained sections were superimposed on those of the adjacent BDA-stained sections (60 μm apart) according to local vascular landmarks. In a similar way, the distribution of retrogradely labeled neurons in the cerebral cortex was depicted in a series of coronal sections. The pattern of retrograde labeling over the frontal and parietal lobes was reconstructed on the unfolded cortical map prepared as described elsewhere (Hatanaka et al. 2001;Tokuno et al. 1997).
Implantation of stimulating electrodes into identified MI and SMA
Initially, ICMS mapping was performed to define regions representing the forelimb in the MI and SMA as a guide for implantation of the stimulating electrodes. The forelimb regions of the MI and SMA were successfully identified on the anterior bank of the central sulcus and the medial wall of the hemisphere, respectively (Fig.1 A). The central parts of these forelimb regions were selected as sites for implantation of pairs of stimulating electrodes (Fig. 1 A, shaded circles connected with arrows). In all five monkeys, two pairs of electrodes were implanted into the MI, and one pair into the SMA (Fig. 1 A). During the daily experimental sessions, movement of no body part other than the forelimb was elicited by MI or SMA stimulation (at the strength of ≤0.8 mA), suggesting that current spread to neighboring regions representing the hindlimb and orofacial area was negligible. Histological examination revealed that each tip of the six electrodes was located within the gray matter of the precentral gyrus or the medial superior frontal gyrus (data not shown).
Arrangement of MI-recipient, SMA-recipient, and convergent striatal zones
Next, cortically induced orthodromic spike discharges were recorded from the putamen at its caudal levels to map striatal zones receiving inputs independently or convergently from forelimb representations of the MI and SMA. In Monkey 1, striatal neurons displaying orthodromic responses to MI stimulation only were localized in the ventrolateral aspect of the caudal putamen (i.e., the MI-recipient zone), while those exhibiting orthodromic responses to SMA stimulation only were localized in the dorsomedial aspect (i.e., the SMA-recipient zone) (Fig. 1 B). The majority of putamen neurons responding orthodromically to stimulation in both the MI and the SMA were distributed in a zone (i.e., the convergent zone) between the MI-recipient and the SMA-recipient zones (Fig. 1 B). Essentially the same results were obtained in the other four monkeys (see Monkeys 3 and 5 in Figs. 6 A and7 A). Thus the present intrastriatal recordings revealed that, at the caudal levels of the putamen, zones receiving inputs from forelimb representations of the MI and SMA were arranged from dorsomedial to ventrolateral in the order of the SMA-recipient, convergent, and MI-recipient zone. This was largely in accordance with the data of our previous work and those of other anatomical reports (Strick et al. 1995a,b; Takada et al. 1998a,b), showing that major corticostriatal inputs from the forelimb regions of the MI and SMA terminate ventrolaterally or dorsomedially within the caudal putamen, respectively, and that a spatial overlap of both inputs occurs in an intermediate region between the sole terminal fields from the MI and SMA. Under the guidance of intrastriatal mapping, paired injections of WGA-HRP and BDA were made, respectively, into the MI-recipient or SMA-recipient zone inMonkey 1 (Figs. 1 B and2), into the SMA-recipient or MI-recipient zone in Monkey 2, into the SMA-recipient or convergent zone in Monkeys 3 and 4 (Fig.6 A), and into the MI-recipient or convergent zone inMonkey 5 (Fig. 7 A) (see also Table 1).
Distribution of corticostriatal neurons
In each of the five monkeys, the validity of the injection sites in the putamen was confirmed by analyzing the distribution pattern of retrogradely labeled neurons in the cerebral cortex. Referring to previous anatomical studies, special attention was paid to the origin in the frontal and parietal lobes of sensorimotor corticostriatal projections to the caudal putamen. After tracer injection into the ventrolaterally situated MI-recipient zone of the putamen, a multitude of labeled neurons were seen around the traces of the stimulating electrodes in the precentral gyrus, and a particular abundance was seen on the anterior bank of the central sulcus (Monkeys 1, 2, and 5; see Fig.3, blue circles). This finding was compatible with the results of our anterograde tract-tracing study (Tokuno et al. 1999) showing that projection fibers from the “bank region” of forelimb representation of the MI terminate ventrolaterally rather than dorsomedially within the caudal putamen. Likewise, tracer injection into the SMA-recipient striatal zone yielded massive neuronal labeling around the traces of the stimulating electrodes in the medial superior frontal gyrus (Monkeys 1–4; see Fig. 3, red circles). After BDA injection into the convergent zone of the putamen in Monkeys 3–5, many cortical neurons were labeled around the traces of the stimulating electrodes in both the MI and the SMA.
In addition, the injection into the MI-recipient striatal zone produced numerous retrogradely labeled neurons in the postcentral gyrus (Fig. 3, blue circles), corresponding to the primary somatosensory cortex from which corticostriatal terminal fields have been reported to overlap those from the MI in the ventrolateral aspect of the caudal putamen (Flaherty and Graybiel 1991, 1993a). Some labeled neurons were further observed on the dorsal and ventral banks of the cingulate gyrus, corresponding probably to the dorsal and ventral cingulate motor areas, and on the posterior bank of the inferior limb of the arcuate sulcus, corresponding probably to the ventral division of the premotor cortex (Fig. 3, blue circles) (seeStrick et al. 1995a,b). After the injection into the dorsomedially situated SMA-recipient zone of the putamen, a number of labeled neurons were found extensively over the frontal lobe, particularly in the dorsal and ventral cingulate motor areas and the dorsal and ventral divisions of the premotor cortex (Fig. 3, red circles) (see McFarland and Haber 2000; Takada et al. 1998a). Such an injection also resulted in retrograde labeling in presumed hindlimb representations of the MI and SMA that are located on the medial wall of the hemisphere as well as in the precentral gyrus (Fig. 3, red circles) (see Hatanaka et al. 2001). This was probably due to partial diffusion of the injection site into the striatal hindlimb sector which has been reported to lie laterally or dorsolaterally within the caudal putamen (McFarland and Haber 2000; Takada et al. 1998a).
Throughout the experiments, the labeled cortical neurons were of the small or medium-sized pyramidal type and were located mainly in layers III and V. Thus we concluded that, for the purpose of the present study, paired injections of WGA-HRP and BDA in each monkey were properly made based on the intrastriatal map prepared. Since similar patterns of striatopallidal and striatonigral terminal distribution were obtained in Monkeys 1 and 2 or Monkeys 3 and 4 (see Table 1), we hereafter describe results inMonkeys 1, 3, and 5 as representative in the following two sections.
Distribution of striatopallidal inputs from MI-recipient, SMA-recipient, and convergent zones
We examined the distribution pattern of anterograde labeling in the pallidal complex from the MI-recipient, SMA-recipient, and convergent zones of the putamen. In Monkey 1, who received paired injections of WGA-HRP and BDA, respectively, into the MI-recipient or SMA-recipient zone (Figs. 1 B, 2, and4), labeled fibers and terminals were profusely observed in both the GPe and the GPi at their caudal two-thirds levels. Dense accumulations of labeled terminals were distributed into multiple bands lying dorsoventrally along the external or internal medullary lamina (Figs. 4 and5, A and B). These terminal bands were often spaced mediolaterally, as described byHazrati and Parent (1992). The WGA-HRP labeling was characterized by punctate filling of fibers and terminals (Fig.5 C), while the BDA labeling was indicated as a cluster of randomly oriented, fine varicose axons (Fig. 5 D). The terminal fields derived from the SMA-recipient striatal zone were largely confined to the dorsal aspect of the pallidal complex, whereas those from the MI-recipient zone were located more ventrally (Fig. 4). Between these two striatopallidal terminal fields, there existed distinct gaps that were virtually devoid of terminal labeling.
In Monkeys 3 and 5, anterogradely labeled striatopallidal terminals emanating from the SMA-recipient zone or MI-recipient zone of the putamen were distributed in a manner essentially the same as in Monkey 1 (Figs.6 B and7 B), although their extent was changed depending on the size of the injection site (Figs.6 A and 7 A). At the caudal two-thirds levels of the GPe and GPi, accumulations of labeled terminals from the convergent zone of the putamen appeared to be located laterally or ventrolaterally to terminal labeling from the SMA-recipient zone (Fig. 6 B), and medially or dorsomedially to that from the MI-recipient zone (Fig.7 B). The terminal fields from the convergent zone occurred, with slight overlap, in pallidal regions immediately adjacent to those from the SMA-recipient and MI-recipient zones (Figs. 5, Eand F, 6 B, and 7 B). They were frequently seen to fill in blanks between the dorsoventrally segregated terminal bands from the MI-recipient and SMA-recipient zones, or between the mediolaterally spaced terminal bands from the MI-recipient or SMA-recipient zone. Thus the overall data revealed that striatopallidal inputs from the MI-recipient, SMA-recipient, and convergent zones of the putamen were spatially segregated in both the GPe and the GPi with no substantial overlap.
Distribution of striatonigral inputs from MI-recipient, SMA-recipient, and convergent zones
Finally, we investigated the distribution pattern of anterograde labeling in the substantia nigra from the MI-recipient, SMA-recipient, and convergent zones of the putamen. In Monkey 1, densely labeled fiber bundles from the SMA-recipient zone were seen in the lateral aspect of the SNr at its rostral levels. More caudally, accumulations of labeled terminals were widely observed in the mediolaterally central aspect of the SNr to be embedded within plexuses of the labeled fibers (Fig. 8;Monkey 1). In contrast, only a small number of fibers and terminals were labeled from the MI-recipient striatal zone. These were distributed laterally to overlap, at least partly, the terminal accumulations from the SMA-recipient zone (Fig. 8; Monkey 1). In Monkey 3, on the other hand, anterogradely labeled terminal fields from the SMA-recipient and convergent zones of the putamen displayed an extensive overlap in the mediolaterally central aspect of the SNr (Fig. 8; Monkey 3). The distribution pattern of labeled striatonigral terminals in Monkey 5 was compatible with the patterns in Monkeys 1 and3. Terminal labeling from the convergent zone occurred in the mediolaterally central aspect of the SNr, and only small terminal fields from the MI-recipient zone were located laterally with a partial overlap (Fig. 8; Monkey 5). Unlike the laminar distribution of striatopallidal terminals, striatonigral terminals from each of the MI-recipient, SMA-recipient, and convergent zones gathered in a single large mass. The overall results indicate that striatonigral inputs from the SMA-recipient and convergent zones of the putamen spatially overlap in the SNr, while input from the MI-recipient zone is minimal.
In each of the three monkeys, anterograde labeling was also found in the lateral or ventrolateral portions of the substantia nigra pars compacta (SNc). A number of retrogradely labeled neurons were seen mainly in the SNc and, additionally, in the SNr. However, no clear topography was detected in their distribution (Fig. 8).
The present study focused on investigating the transposition of corticostriatal motor inputs from the MI and SMA onto the GPe/GPi and SNr. To analyze how striatopallidal and striatonigral inputs from three distinct regions of the caudal putamen are arranged, each of which receives projection fibers from the MI alone (MI-recipient zone), the SMA alone (SMA-recipient zone), or both the MI and the SMA (convergent zone), we employed an experimental paradigm consisting of electrophysiological mapping combined with an anterograde double-labeling technique. Our results clearly demonstrate that corticostriatal motor information is differentially processed in the GPe/GPi and SNr.
A thorough survey of previous anatomical data in primates reveals that the organization of striatopallidal and striatonigral projections remains controversial. Together with a remarkable decrease in the number of neurons from the source to the targets (Percheron and Filion 1991), three lines of evidence suggest that striatal inputs to the GPe/GPi and SNr are convergent. First, the dendrites of individual pallidal and nigral neurons, oriented perpendicular to the incoming striatal axons, span as far as 1.5 mm in diameter (François et al. 1987; Percheron et al. 1984; Yelnik et al. 1984). Such a wide range of dendritic arborization would maximize the potential for convergence of striatopallidal and striatonigral inputs. Second, terminal fields derived from a single, small region of the striatum are so large in the GPe/GPi and SNr as to encompass a great proportion of each structure (Hazrati and Parent 1992; Hedreen and DeLong 1991; Parent and Hazrati 1994, 1995a), thereby allowing a high degree of overlap of pallidal and nigral input zones from at least neighboring striatal regions. Third, a set of experiments to study the input-output relationship of the striatum has demonstrated that multiple zones of the putamen receive input from the sensorimotor cortex and, in turn, send output to a single, restricted area of the GPe/GPi (Flaherty and Graybiel 1994; Graybiel et al. 1994). This favors a pattern of information processing in the basal ganglia that involves corticostriatal divergence followed by striatopallidal reconvergence.
In contrast, other evidence indicates that striatal projections to the GPe/GPi and SNr are organized in a segregated, parallel fashion. Although both the caudate nucleus and the putamen give rise to striatopallidal and striatonigral projections, their contributions to such projections are not necessarily equal. According to previous tract-tracing studies (Hazrati and Parent 1992;Parent and Hazrati 1994, 1995a; Parent et al. 1984), projection fibers from most parts of the caudate nucleus and the rostral putamen—the striatal territory that receives input from the association cortex—terminate more prominently in the SNr, whereas those from the bulk of the caudal putamen—the striatal territory that receives input from the sensorimotor cortex—end more markedly in the GPe/GPi. In addition, finer level analysis has shown that terminal fields from two nearby or adjacent striatal regions do not apparently overlap in either the GPe/GPi or the SNr, but rather interdigitate side by side (Hazrati and Parent 1992;Parent and Hazrati 1994, 1995a). Thus the occurrence of topography and the lack of convergence of striatopallidal or striatonigral inputs are well consistent with the currently prevailing concept of a parallel functional scheme of the cortico-basal ganglia loop (Alexander and Crutcher 1990; Alexander et al. 1986). Using retrograde transneuronal transport of herpes simplex virus type 1, Strick and his colleagues have reported the existence of multiple, segregated output channels in the GPi and SNr through which basal ganglia signals are transmitted to motor and nonmotor cortical areas via the thalamus (Hoover and Strick 1993; Middleton and Strick 1994, 2000a,b). It has also been shown, on the other hand, that multiple channels in relation to functionally distinct striatonigral inputs are not strictly segregated from one another, because SNr neurons within individual channels usually extend part of their dendrites into neighboring channels (Mailly et al. 2001).
In the present study, we have demonstrated that striatopallidal inputs from the MI-recipient and SMA-recipient zones of the putamen are essentially separate in both the GPe and the GPi. As described byHazrati and Parent (1992), these inputs are aggregated to form dorsoventrally elongated multiple bands that are spaced mediolaterally. In full agreement with previous data (Hoover and Strick 1993; Middleton and Strick 2000a,b;Yoshida et al. 1993), the terminal bands derived from the MI-recipient striatal zone are located ventrally to those from the SMA-recipient zone with no substantial overlap. Of particular interest is that striatopallidal inputs from the convergent zone are distributed to fill in blanks between the dorsoventrally segregated terminal bands from the MI-recipient and SMA-recipient zones, or between the mediolaterally spaced terminal bands from the MI-recipient or SMA-recipient zone. Thus our results certainly indicate that the striatopallidal projections from the MI-recipient, SMA-recipient, and convergent zones of the putamen are organized in a segregated, parallel manner.
On the other hand, the arrangement of striatonigral inputs from the three identified zones of the putamen is strikingly different from that of striatopallidal inputs. Our data have shown that there is a great overlap in the SNr of terminal fields from the SMA-recipient and convergent striatal zones. Such striatonigral input convergence occurs mainly in the mediolaterally central aspect of the SNr. Despite the richness in striatonigral inputs from the SMA-recipient and convergent zones, however, projection fibers from the MI-recipient striatal zone do not appear to terminate densely within the SNr. It should be noted here that the SNr contains distinct output channels that connect the basal ganglia with frontal association areas, i.e., areas 9, 12, and 46, via the thalamus, thus implying the possible existence of segregated subloops in the “prefrontal” cortico-basal ganglia loop (see Middleton and Strick 2000a,b). Originally, individual striatopallidal and striatonigral neurons have been considered to be separate although spatially intermingled and, therefore, to project to one of the GPe, GPi, and SNr (Flaherty and Graybiel 1993b; Parent et al. 1984, 1989;Selemon and Goldman-Rakic 1990). It has recently been reported, on the other hand, that a large number of single striatal neurons innervate more than one target simultaneously by way of axon collaterals in the rat and monkey (Kawaguchi et al. 1990; Parent et al. 1995, 2000; Wu et al. 2000). According to Parent et al. (1995), virtually all projection neurons in the monkey striatum issue axons to the GPe, and many of these axons also arborize into both the GPi and the SNr with one preferential target. In this regard, the present results suggest that neurons within the MI-recipient zone of the putamen may only poorly be collateralized to the SNr. Moreover, the paucity of striatonigral input from the MI-recipient zone is supported by a retrograde tract-tracing study (Haber et al. 2000), in which tracer injections into various nigral regions did not notably produce neuronal labeling in the ventrolateral portion of the caudal putamen, corresponding probably to the MI-recipient striatal zone.
We have demonstrated that striatopallidal motor information processing is parallel not only in the GPi, but also in the GPe. However, it has not yet been analyzed how motor information from the GPe is conveyed to the GPi and SNr via the STN along the so-called indirect pathway of the basal ganglia (Alexander and Crutcher 1990; Mink 1996; Mink and Thach 1993; Parent and Hazrati 1995a,b). Thus it is of interest to explore whether or not motor information from each of the MI-recipient, SMA-recipient, and convergent zones of the putamen definitively overlaps in the GPi and SNr through the direct (striato-GPi/SNr) and indirect (striato-GPe-STN-GPi/SNr) pathways.
It has long been believed that the GPi and SNr belong to a single entity that is split rostrocaudally by the internal capsule (Parent 1986). In this view, the two structures are likely to play exactly the same role in the processing of information along the cortico-basal ganglia loop. However, in terms of the parallel versus convergent rules of information processing, the present work provides anatomical evidence that the mode of dealing with corticostriatal motor information from the MI and SMA through the striatopallidal and striatonigral projections is target-dependent, such that the parallel rule governs striatopallidal input distribution, whereas the convergent rule determines striatonigral input distribution (Fig. 9). This strongly implies that the arrangement of the striatopallidal system closely reflects the organization of the corticostriatal system, while that of the striatonigral system does not. It has also been reported that the firing pattern of SNr neurons is less affected in parkinsonian monkeys than that of GPi neurons, suggesting their functional differences in motor behavior (Wichmann et al. 1999).
What is the functional implication of the differential processing patterns of corticostriatal motor information in the GPi and SNr? At pallidal levels, there is a spatial separation of striatopallidal inputs originating from the associative and sensorimotor territories of the striatum, the former being largely confined to the dorsomedial one-third and the latter being largely confined to the ventrolateral two-thirds of the GPi (Hazrati and Parent 1992;Hedreen and DeLong 1991; Parent and Hazrati 1995a; Parent et al. 1984; Selemon and Goldman-Rakic 1990). Such territorial subdivisions are unclear at the level of the SNr, where striatonigral inputs from the two territories seem rather intermingled (Parent and Hazrati 1994,1995a). In addition, there is a greater contribution of the sensorimotor than associative striatal territory to the output system from the GPi, and the inverse is true of that from the SNr (Hazrati and Parent 1992; Middleton and Strick 2000a,b; Parent and Hazrati 1994, 1995a;Parent et al. 1984). It should also be noted, however, that a subpopulation of neurons in the SNr as well as in the GPi display movement-related activity (DeLong et al. 1983;Wichmann et al. 2001). The major outflow from the GPi is directed toward frontal motor areas, including the MI and SMA, via the ventrolateral thalamus, while that from the SNr is directed toward frontal association areas, such as areas 9 and 46, via the ventroanterior and mediodorsal thalami (Ilinsky et al. 1985; Middleton and Strick 2000a,b;Rouiller et al. 1994). Thus well-segregated motor information processed in the cortico-striato-pallidal link might be returned predominantly to the motor cortex, thereby forming a “closed loop,” whereas highly integrated motor information processed in the cortico-striato-nigral link might be conveyed preferentially to the association cortex, thereby forming an “open loop” (Fig. 9).
We thank M. Imanishi, Y. Ito, and E. Mine for technical assistance.
This research was supported by Grants-in-Aid for Scientific Research (C) and for Scientific Research on Priority Areas (A) from the Ministry of Education, Culture, Sports, Science, and Technology of Japan.
Address for reprint requests: M. Takada, Dept. of System Neuroscience, Tokyo Metropolitan Institute for Neuroscience, Tokyo Metropolitan Organization for Medical Research, Fuchu, Tokyo 183-8526, Japan (E-mail:).
- Copyright © 2002 The American Physiological Society