The primate thalamostriatal systems: Anatomical organization, functional roles and possible involvement in Parkinson’s disease

Abstract

The striatum receives glutamatergic inputs from two main thalamostriatal systems that originate either from the centre median/parafascicular complex (CM/PF-striatal system) or the rostral intralaminar, midline, associative and relay thalamic nuclei (non-CM/PF-striatal system). These dual thalamostriatal systems display striking differences in their anatomical and, most likely, functional organization. The CM/PF-striatal system is topographically organized, and integrated within functionally segregated basal ganglia-thalamostriatal circuits that process sensorimotor, associative and limbic information. CM/PF neurons are highly responsive to attention-related sensory stimuli, suggesting that the CM/PF-striatal system, through its strong connections with cholinergic interneurons, may play a role in basal ganglia-mediated learning, behavioral switching and reinforcement. In light of evidence for prominent CM/PF neuronal loss in Parkinson’s disease, we propose that the significant CM-striatal system degeneration, combined with the severe nigrostriatal dopamine loss in sensorimotor striatal regions, may alter normal automatic actions, and shift the processing of basal ganglia-thalamocortical motor programs towards goal-directed behaviors.

In traditional models of the basal ganglia circuitry, the cerebral cortex is considered as the prime source of excitatory glutamatergic afferents to the striatum, while the thalamus is recognized as the main target of basal ganglia outflow [1–3]. However, it has long been known that the thalamus is also a predominant source of excitatory inputs to the striatum [4, 5], but due to the limited knowledge about the functional role of this system, its integration into the functional circuitry of the basal ganglia has long been neglected. However, converging data from recent studies have highlighted the potential role of the thalamostriatal system from the caudal intralaminar nuclei in alertness and behavioral switching. These findings, combined with evidence that the caudal intralaminar nuclei undergo significant degeneration in Parkinson’s disease, and may serve as a potential target for deep brain stimulation in movement disorders, have set the stage for significant advances in our understanding of the anatomical and functional organization of the thalamostriatal system. In this review, we will highlight these recent developments, and provide a comprehensive analysis of the anatomical substrate through which the thalamostriatal systems could mediate their effects upon behavioral switching and attention shifts in normal and parkinsonian states.

1. Anatomy of the dual thalamostriatal systems

Although the caudal intralaminar nuclei (centromedian/parafascicular complex, CM/PF), represent the main source of thalamic inputs to the striatum, it is important to recognize that thalamostriatal projections also arise from several other thalamic nuclei, including the rostral intralaminar, midline and specific relay nuclear groups [6–21]. Based on their dual thalamic origin, and distinctive anatomical features, the thalamostriatal networks can be divided into two segregated subsystems: (1) the CM/PF-striatal projection and (2) the non CM/PF-striatal projections.

1.1. CM/PF-striatal projection and basal ganglia-thalamostriatal circuits

The intralaminar nuclei of the thalamus are located lateral to the mediodorsal nucleus within the dense axonal meshwork of the internal medullary lamina. They are divided into a rostral group –the central medial, paracentral and central lateral nuclei– and a caudal group which, in primates, consists of the centromedian (or centre median, CM) and the parafascicular (PF) nuclei, that together form the CM/PF complex [22]. Because the CM/PF complex is the main source of thalamic inputs to the striatum [18, 23, 24], the CM/PF-striatal projection has been the most extensively studied thalamostriatal subsystem.

In primates, the CM/PF projects to all functional regions of the striatum in a topographic fashion: (1)The rostral third of PF innervates predominantly the nucleus accumbens (ventral “limbic” striatum); (2) the caudal two thirds of PF project to the caudate nucleus (“associative” striatum); (3) the dorsolateral PF (PFdl) projects to the anterior putamen (“associative” striatum); (4) the medial two thirds of CM innervate the post-commissural putamen (“sensorimotor” striatum); and (5) the lateral third of CM (CMl) provides inputs to the primary motor cortex [18, 23, 24] (Fig. 1). Through these extensive projections, the CM/PF gains access to the whole striatal complex, thereby making the CM/PF-striatal system a functionally organized network that could have broad influences upon motor and non-motor basal ganglia functions.

Fig. 1.

Segregated basal ganglia-thalamostriatal circuits. On the left, the illustration shows the pattern of distribution of color-coded thalamic inputs from the CM/PF complex to three rostrocaudal levels (stereotaxic coordinates at the bottom left) of the striatal complex in monkeys. Apart from the lateral 1/3 of the CM (CMl) which projects mainly to the motor cortex, the rest of the complex is tightly linked in a topographical fashion with the dorsal and ventral striatum. On the right, the functional circuits are indicated. The sensorimotor GPi (ventrolateral 2/3) projects to the CM. The limbic GPi (rostromedial and ventral pallidum) innervate the rostral PF, and the associative GPi (dorsal 1/3) provides inputs to the dorsolateral PF (PFdl). In turn, CM/PF neurons project back to the corresponding functional territories of the striatum (black arrows). The substantia nigra reticulata (SNr) innervates PF neurons that project to the caudate nucleus. Additional abbreviations: A, anterior; AC, anterior commissure; ACC, Accumbens; CD, caudate; GPe, globus pallidus, external segment; GPi, globus pallidus internal segment; IC, internal capsule; PF, parafascicular nucleus; PFdl, dorslateral parafasccular nucleus; Pre-comm., Pre-commissural; Put, Putamen; Th, thalamus.

In addition to these massive and tight connections with the striatum, the CM/PF complex is integrated within the basal ganglia circuitry via direct inputs from the two main output nuclei of the basal ganglia, the internal globus pallidus (GPi) and the substantia nigra pars reticulata (SNr). These topographic and functionally organized projections originate from axon collaterals of GPi and SNr afferents to the ventral anterior/ventral lateral (VA/VL) nuclear complex [24–27]. Through these connections, the CM/PF is part of functionally segregated basal ganglia-thalamostriatal loops that process sensorimotor, associative and limbic information in primates (Fig. 1).

Although the overall organization of these projections has also been described in rodents, it is important to recognize that the caudal intralaminar nuclear complex in these species is solely made up of a single nuclear mass called PF of which the lateral sector is considered as the homologue of the primate CM that projects to the sensorimotor striatum, while the medial sector corresponds to the primate PF proper, connected with associative and limbic striatal regions [7, 28].

1.2. Non CM/PF-striatal system

The existence of thalamostriatal projections from thalamic nuclei other than the CM/PF has long been recognized [5, 29]. Studies using retrograde and anterograde labeling have identified thalamic inputs to the striatum from the midline nuclei, rostral intralaminar nuclei, VA/VL, mediodorsal nucleus, and the pulvinar [6, 7, 10, 13, 15, 17, 22, 28, 30–39].

All areas of the striatum receive modest thalamic glutamatergic projections from non CM/PF nuclei, with a certain degree of topographical organization and specificity. In primates and non-primates, the nucleus accumbens (ventral striatum) receives inputs from dorsal midline thalamic nuclei [7, 17, 22, 31], while the rostral intralaminar group provides afferents to the dorsal striatum [22, 30, 33, 34, 38].

In primates, regions of the VL interconnected with specific areas of the motor cortex project to sectors of the sensorimotor striatum that also receive inputs from the same cortical regions, suggesting convergence and interactions of functionally related corticostriatal and thalamostriatal systems [14]. A similar pattern of connectivity has also been suggested for relationships between VA, associative frontal cortices and related target sites in the caudate nucleus and anterior putamen [14, 15].

1.3. Anatomical Differences between the CM/PF-striatal and other thalamostriatal systems

In addition to their origin, the CM/PF- and non CM/PF-striatal systems display other striking anatomical differences, as illustrated in Figure 2.

Fig. 2.

Dual thalamostriatal systems from CM/PF and non CM/PF nuclei. The thalamus provides striatal cells with two independent and anatomically different systems. The main anatomical differences between the two systems are summarized.

For instance, in contrast to most thalamic nuclei that provide major inputs to specific cortical regions, and a more modest diffuse projection to the striatum [11, 12, 15, 28], CM/PF neurons provide massive topographically organized striatal projections with only sparse and diffuse collateral projections to the cerebral cortex [6, 20, 40].

The pattern of striatal innervation is also quite different between the two systems. While the CM/PF complex provides dense, focal and highly convergent inputs to the striatum, striatal projections from other thalamic nuclei are more diffuse and sparsely distributed, thereby implying that terminals from individual CM/PF axonal projections may provide a more massive focused innervation of a restricted pool of striatal neurons than inputs from non CM/PF thalamic nuclei [10, 20, 30, 41]. Such a different pattern of termination may impact the synaptic strength of these two systems on striatal neuronal activity.

CM/PF terminals target both MSNs and striatal interneurons. In fact, with the exception of calretinin-positive cells, all types of striatal interneurons are contacted by CM/PF terminals [42, 43]. The cholinergic interneurons, in particular, receive a dense innervation from CM/PF terminals in rats and monkeys [42, 44, 45]. The functional significance of this tight relationship between the CM/PF-striatal system and cholinergic interneurons will be discussed in more detail in section 2.3. Although both direct and indirect striatal projection neurons receive CM/PF inputs, there is a slight preference for CM/PF-striatal afferents towards direct pathway neurons that project to GPi in monkeys [41]. However, non CM/PF-striatal terminals contact almost exclusively MSNs [30, 46], without any specific discrimination between direct or indirect pathway neurons in rats [47].

At the ultrastructural level, the majority of CM and PF terminals form asymmetric synapses with dendrites of striatal projection neurons and interneurons [41, 42, 46, 48, 49]. In contrast, most thalamic terminals from non-CM/PF sources display a pattern of synaptic connectivity similar to that of glutamatergic corticostriatal afferents, i. e., they contact almost exclusively dendritic spines of projection neurons [21, 46].

Dopaminergic terminals from the substantia nigra compacta (SNc), known as key regulators of glutamatergic transmission in the striatum, exhibit different spatial relationships with thalamic terminals from CM/PF versus other thalamic nuclei on the surface of striatal projection neurons. Non-CM/PF thalamic terminals contact dendritic spines in close proximity to dopaminergic boutons, while CM/PF terminals show no spatial closeness to dopaminergic afferents [46, 50, 51]. Nevertheless, despite their distant location, it is possible that dopaminergic and CM/PF inputs terminate on the same striatal neurons, and that the non-synaptic volume transmission of dopamine mediates functional interactions between these neural systems [51, 52].

The striatum is a non-homogenous structure made up of at least two different compartments referred to as “extrastriosomal matrix” (or matrix) and “patches” (or striosomes [53, 54]) that are recognized by their differential anatomical, neurochemical and, most likely, functional characteristics. These compartments also differ in their main sources of thalamic innervation. Afferents from the CM/PF tend to terminate in the matrix, [49, 55], while inputs from other thalamic nuclei are less selective and more widely distributed across both compartments, although terminals from some midline nuclei target selectively the patches [7, 39, 46, 56].

Thus, in light of these striking anatomical differences between striatal inputs from CM/PF versus non CM/PF thalamic nuclei, the thalamostriatal network must be seen as a composite system made up of two major sets of axonal projections that display a unique anatomical organization and a differential pattern of synaptic connectivity within the striatal microcircuitry depending on their thalamic origin. The anatomical and functional organization of this dual thalamostriatal system will now be examined and compared with the massive corticostriatal network, which provides the bulk of glutamatergic excitatory drive to the striatum.

1.4. Thalamostriatal versus corticostriatal projections: Their differences and commonalities

While both cortical and thalamic projections provide glutamatergic inputs to the striatum, these neural systems differ significantly in various anatomical, neurochemical and functional grounds. In this section, the specific features that characterize thalamic versus cortical projections to the striatum will be highlighted in light of the functional roles these neural networks may play in the physiology and pathophysiology of the basal ganglia in normal and diseased conditions.

First, these striatal afferent projections can be differentiated by the segregated expression of vesicular glutamate transporter type 1 (vGlut1) in cortical terminals and vesicular glutamate transporter type 2 (vGlut2) in thalamic terminals [57, 58]. In both rodent and primate striata, double electron microscopy immunocytochemistry demonstrated that vGlut1 is specifically associated with cortical glutamatergic terminals, while vGlut2 is confined to thalamic afferents; with less than 5% of total glutamatergic striatal terminals co-expressing both vGluTs [46, 59]. These studies also allowed to quantify the relative prevalence of cortical over thalamic glutamatergic terminals in the rat and monkey striatum, revealing that vGlut1- and vGlut2-positive terminals represent about 50% and 20% respectively of total glutamatergic terminals in the monkey striatum [59], while these percentages are about 35% and 25% for vGlut1 and vGlut2 respectively in the rat striatum [60]. A considerable proportion of putative glutamatergic terminals do not express either vGlut1 or vGlut2 in rat and monkey, suggesting the presence of another vesicular glutamate transporter, yet to be identified [59, 60].

Besides their useful application as markers of cortico- or thalamostriatal boutons, the selective expression of vGlut1 or vGlut2 may confer unique functional properties to the cortical and thalamic striatal afferent systems. In other brain regions, the presence of vGlut1 or vGlut2 in axon terminals is associated with low and high probability of transmitter release respectively and a different degree of synaptic plasticity [58, 61]. Consistent with these descriptions, recent in vitro electrophysiological data have shown that thalamostriatal synapses exhibit higher probability of glutamate release than corticostriatal synapses, and differ in their short-term synaptic plasticity [62]. These studies also revealed that thalamic and cortical inputs to the striatum differ in ratio and composition of NMDA and AMPA glutamate receptors [62, 63].

In regard to their synaptology, cortical inputs to the striatum target dendritic spines of medium spiny projection neurons, with very rare incidence of axo-dendritic synapses, a pattern reminiscent of the non CM/PF-striatal system, but strikingly different from the CM/PF-striatal projection (see previous section) [46, 48–50, 64, 65]. There is evidence that thalamic and cortical projections differ in the degree of innervation of specific populations of striatal interneurons. For instance, although CM/PF terminals provide a strong input to the proximal dendrites and cell bodies of cholinergic interneurons, cortical terminals only contribute scarce inputs to the distal dendrites of these neurons in primates and non-primates [42, 44, 45, 66]. These data are supported by slice electrophysiology data showing that thalamic, but not cortical, stimulation evokes patterned responses in cholinergic interneurons [67]. The functional importance of the CM/PF-striatal system in regulating cholinergic interneurons activity in learning and behavioral switching/reinforcement is discussed in section 2.3. In contrast to cholinergic cells, striatal GABAergic parvalbumin-positive interneurons (putatively fast-spiking interneurons [68]) appear to receive a significantly stronger cortical than thalamic innervation in rats [43], though such may not be the case in primates [42].

A recent study has revealed a neurochemical feature that appears to be specific for the CM/PF-striatal system. In rats, PF-striatal terminals express immunoreactivity for a protein called cerebellin 1, which was found to play an important role in shaping dendritic structure of striatal MSNs [69]. So far, there is no evidence that cortical terminals contain cerebellin 1 or related proteins.

2. Role of the CM/PF-striatal system in attention and behavioral switching

As described above, the CM/PF is, by far, the main source of thalamic inputs to the striatum. Although our understanding of the functional significance of the thalamostriatal systems remains limited, the recent interest towards the possible role of CM/PF neurons in attention and its importance in regulating cholinergic interneurons activity, combined with evidence for significant CM/PF degeneration in PD patients, have set the stage for a deeper understanding of the importance of the CM/PF-striatal system in mediating basal ganglia responses to unpredicted stimuli, and the potential consequences of the degeneration of this system towards behavioral switching deficits in PD. Despite significant anatomical evidence for thalamostriatal projections that originate outside the CM/PF (see above), the paucity of information on the functional role of these non-CM/PF-striatal projections limits considerably our interpretation of the importance of these systems in the functional circuitry of the basal ganglia. Thus, the following discussion will be entirely devoted to the CM/PF-striatal system, and its possible implication in attention and behavioral switching in normal and parkinsonian conditions. We will conclude with a brief overview of future studies one should consider to move this field forward.

2.1. CM/PF neurons: afferents and functions

In order to critically examine the mechanisms by which the CM/PF-striatal system may contribute to the regulation of behavioral switching in response to attention-related stimuli, it is important to recognize the main sources of inputs that contribute to the regulation of CM/PF neuronal activity. In addition to the prominent basal ganglia GABAergic projections from the GPi and SNr (see section 1.1.), the CM receives inputs from motor, premotor and somatosensory cortices [70–76], while cortical inputs to PF originate preferentially from the frontal and supplementary eye fields [77, 78], and associative areas of the parietal cortex [79, 80].The CM/PF complex also receives strong inputs from various subcortical sources, including the pedunculopontine tegmental nucleus [81–83], the superior colliculus [84–88], the cerebellum [89, 90], the raphe nuclei and locus coeruleus [89, 91, 92], and from brainstem regions of the mesencephalic, pontine and medullary reticular formation [89, 93–99]. Because of these significant ascending connections from the reticular formation and various brainstem regions, combined with the traditional view that the CM/PF and other intralaminar thalamic nuclei are the sources of widely distributed “nonspecific” thalamocortical projections, the intralaminar nuclei were considered part of the ascending “reticular activating system” that regulates arousal and attention (as reviewed in [22]). In line with this concept, functional imaging studies in humans have demonstrated a significant increase of activity in CM/PF during processing of attention-related stimuli [100–102]. Studies in behaving monkeys have provided direct evidence that one of the main roles of CM/PF neurons is, indeed, to process sensory stimuli related to shifts in attention and action bias [103–105].

CM/PF neurons which, at rest, have low firing rates and burst-like discharge pattern, increase their activity in response to a wide variety of sensory stimuli (visual, auditory or tactile), and habituate rapidly after repeated presentation of the stimulus [103]. On the basis of the response latency to sensory stimuli, Kimura and colleagues have classified CM/PF neurons in short-latency or long latency facilitation (SLF or LLF respectively, [103–105]). SLF neurons are found more frequently in PF, whereas LLF cells lay preferentially in CM. Responses of both types of neurons are independent of the rewarding attributes of the stimuli (Fig. 3A, B).

Fig. 3.

Sensory responses of two types of CM/PF neurons, and a striatal TAN recorded during the presentation of a stimulus with reward (WR) and stimulus without reward (WOR). Spike raster and histograms aligned to the time of presentation of the stimulus. A: representative activity of a CM neuron with long-latency facilitation following stimulus presentation (LLF). B: activity of a PF neuron showing short-latency facilitation after stimulus (SLF). C: activity of a TAN. Note that thalamic responses occur in both WR and WOR tasks, whereas TAN responses occur only in the WR task (from Matsumoto et al, 2001)

CM/PF inactivation impairs performance of monkeys trained in attention-related tasks [104], and most CM neurons are activated when task conditions demand a change in response type after unpredicted events. In rats, performance in a reversal learning task, which requires shifting choice patterns and behavioral flexibility, is impaired after pharmacological inactivation of PF [106]. Based on these evidences, Kimura and colleagues have proposed that the CM/PF complex is particularly relevant for re-directing attention or behavior from biased actions [105, 107]. Thus, CM/PF neurons play an important role in attention re-directing and shifting behavioral choices under unexpected conditions.

2.2. Control of striatal activity by CM/PF thalamic inputs

Although CM/PF projections to the striatum are massive, their impact on the activity of striatal neurons remains poorly understood. However, significant effort using in vitro and in vivo preparations in rats and primates has been devoted to characterize the physiological effects of thalamic inputs upon striatal neurons activity. Early studies in anesthetized cats and rats, using in vivo intracellular recording methods, described short latency (likely monosynaptic) excitatory postsynaptic potentials in striatal neurons following electrical stimulation of the CM/PF complex [108–110]. Wilson and colleagues further characterized these effects, and demonstrated that both striatal projection neurons and tonically active neurons (TANs; putative cholinergic interneurons) often display early excitatory responses to electrical stimulation of intralaminar nuclei. However, they also described prolonged inhibitory and long latency excitatory responses following these stimulations [111, 112], suggesting polysynaptic responses to thalamic stimulation that might induce complex patterns of striatal activity in response to intralaminar nuclei activation.

Recent data from rhesus monkeys, using in vivo extracellular single unit recording methods, further demonstrated the intricate nature of the physiological responses CM/PF stimulation elicits in striatal neurons [113]. Following CM stimulation, a large proportion of phasically active neurons (PANs, putative MSNs) increase their firing rate, while TANs (putative cholinergic interneurons) display complex responses that include short- and long-latency increases and decreases in activity (Fig. 4) [113]. These complex events are correlated with evidence that CM or PF stimulation results either in a glutamate-mediated increase in striatal acetylcholine (ACh) [114], or a reduction in ACh levels that is abolished by intrastriatal pharmacological blockade of GABA-A receptors [113, 115]. Therefore, activation of CM/PF connections to the striatum can induce either increase or decrease in ACh release; the former being most likely mediated by direct, monosynaptic, glutamatergic afferents from the CM/PF onto cholinergic interneurons (see section 1.3.), while the latter probably results from CM/PF-induced activation of MSNs or GABAergic interneurons that, in turn, inhibit cholinergic interneurons. Based on these findings, we propose that the CM exerts strong modulation of both striatal projection neurons and interneurons, mediated by a complex interplay between direct monosynaptic CM-striatal glutamatergic inputs and multisynaptic influences through CM-mediated activation of GABAergic MSNs and interneurons which, via intrinsic microcircuits, could indirectly reduce striatal activity.

Fig. 4.

Responses of PANs (putatively MSNs) and TANs (cholinergic interneurons) to electrical stimulation of CM, in rhesus monkeys. The stimulation (100 Hz, 100 pulses) is indicated by the shaded area. Right: Example of a PAN responding with increased firing to CM stimulation. Left: Example of a TAN responding with a brief decrease followed by an increase in firing. The histograms and rasters are aligned to the start of stimulation trains. Bottom: Summary of responses. While the majority of PANs presented increases in firing rate, most TANs presented combinatory (increases and decreases) responses (from Nanda et al, 2009).

2.3. CM/PF-striatal effects upon TANs

As described in section 1.3, CM/PF neurons provide strong synaptic inputs to striatal cholinergic interneurons [42] considered to be the tonically active neurons (TANs) in functional studies. In the presence of reward, or reward-related sensory stimuli, TANs display a stereotypic response consisting of a short burst followed by a clear pause and a later excitation (Fig. 3C) [103, 116–120]. However, the responses of TANs are diminished when the reward is delivered in a consistent predictable manner [121], while they are enhanced when the timing of rewards is not predictable [122]. Apicella has proposed that the sensitivity of TANs to changes in the sequence of stimuli indicates these cells might be involved in processing temporal sequence, and participate in the formation of automatic actions [123].

Chemical inactivation of CM/PF abolishes the characteristic reward-related responses of TANs, showing the importance of CM-striatal connections in mediating TANs activity changes in response to reward-related stimuli [103]. In rats, reversal learning is associated with an increase in striatal ACh, which is blocked by PF inactivation [106]. In addition, dopaminergic innervation to the striatum is also an essential regulator of TANs responses to reward-predicting stimuli [124, 125]. This has led to the suggestion that CM/PF provide TANs with information regarding salient events to activate conditional responses, and this information is integrated by TANs with dopaminergic signals from the SNc [103].

Based on findings gathered from a recent in vitro study, Surmeier and colleagues proposed a cellular mechanism by which thalamic regulation of cholinergic interneurons could influence corticostriatal signaling to mediate attentional shifts in response to salient environmental stimuli [67]. Although partly speculative, the authors proposed the interesting working hypothesis that, in response to the presentation of a salient stimulus, the thalamostriatal system can activate cholinergic interneurons which, in turn, regulate corticostriatal signaling and MSNs activity to elicit attention-related behavioral responses [67]. Therefore, because of its critical role in the regulation of striatal microcircuitry in response to attention-related sensory events, inactivation of the CM/PF complex results in contralateral sensory neglect and learning deficits in sensory-related attentional tasks in nonhuman primates [103, 104].

In spite of their significant interest, the main limitation of the in vitro slice experiments described above, and other recent studies [62, 63, 67, 126], is their reliance on non-specific electrical stimulation of the thalamus that most likely involves both the PF-striatal and the non PF-striatal systems (see section 1.2.). Future studies using more specific thalamic stimulation methods that could allow selective activation of the CM/PF-striatal projection are warranted to further address these issues.

In summary, CM/PF projections to the striatum are located to subserve an important control over putative cholinergic interneurons, through which they can regulate their role in attentional direction, behavioral flexibility and formation of automatic actions.

3. How does the CM/PF-striatal system degeneration contribute to set shifts and behavioral switching problems in Parkinson’s Disease?

Parkinson’s Disease (PD) is clinically identified by the motor signs of akinesia, rigidity and tremor at rest. In addition to these characteristic motor signs, most PD patients suffer cognitive deficits, such as impairment in attention tasks, working memory, set shifting and cognitive flexibility related to difficulty in planning, organizing and regulating goal-directed behavior [127, 128]. At the same time, PD patients have a decreased capacity to engage in normal automatic (habitual) control of actions, and become increasingly dependent on a goal-directed mode of action, which impede their normal daily activities [129].

Although degeneration of the nigrostriatal dopaminergic system remains the key pathological feature of PD, it is clear that many other neural systems are also affected [130, 131], including the CM/PF-striatal projection. Evidence from postmortem human brain studies demonstrates that the CM/PF complex presents a 30 to 40% cell loss in PD [132–134]. This thalamic degeneration appears to be specific to CM/PF because neighboring thalamic nuclei remain intact [133]. A similar pattern of degeneration was recently found in the CM/PF complex of parkinsonian monkeys chronically treated for many months with low doses of the toxin MPTP [135]. Furthermore, studies of the synaptic organization of vGlut2-positive (thalamostriatal) terminals in the putamen of MPTP-treated monkeys showed a decrease in the relative prevalence of vGlut2-positive axo-dendritic synapses which, for the most part, originate in CM (see section 1.3.) [59]. This observation is consistent with the possibility that CM inputs to the sensorimotor striatum are partly lost in parkinsonism. Loss of PF neurons has also been reported in some rodent models of parkinsonism ([136–138] but see [139]).

Presumably, the degeneration of the CM/PF-striatal system significantly contributes to both motor and non-motor deficits in PD. Based on the findings about the anatomy and functions of the CM/PF-striatal system discussed above, we propose that CM/PF degeneration could underlie set shifting deficits and inability to restore habitual behaviors in PD patients.

3.1. Set shift impairments

PD patients have deficits in set shifting, that is they have problems to alter ongoing behavior in response to sudden changes in their environment, an impairment that can be mediated by loss of dopamine in the associative striatum and/or prefrontal cortex [127] [140]. In light of the recent functional data related to the responses of CM/PF neurons to salient sensory stimuli, we suggest that degeneration of the CM/PF-striatal connection may also contribute to the set shifting problem described in PD. As we have discussed (section 2), CM/PF plays, indeed, a particular important role in redirecting attention to salient stimuli, behavioral flexibility and changing behavior in responses to unpredicted stimuli [107] [106]. Degeneration of caudal intralaminar nuclei, with corresponding loss of modulation over the activity of TANs and instrastriatal circuitry, could result in deficiencies to switch attention and reselect a proper action under changing circumstances. In this manner, the loss of CM/PF neurons could be one of the contributing factors to the set shifting inability in PD.

3.2. Habit vs Goal directed behavior

In a recent review, Redgrave et al (2010) have proposed that the basal ganglia have a prominent role in selecting between goal-directed (voluntary) and a habitual (automatic) control of behavior [129]. A large body of evidence obtained from human, rodent and primate studies indicates that the ventromedial (associative) striatum regulates goal-directed behavior, while the dorsolateral (sensorimotor) striatum is in charge of habitual control [reviewed in 129]. It is well recognized that PD patients show impairments in tasks that are normally controlled automatically, and have difficulties when learning new habits or performing automatic components of movement sequences [141–147]. Due to the reduced capacity of selecting habitual actions, PD patients have to rely on the more time consuming goal-directed action control system. The diminished capacity of basal ganglia to select habitual control system over goal-directed behaviors in PD patients, may be related to the heterogeneous loss of striatal dopamine, which is more severe in the sensorimotor striatum, recognized as the striatal control site for habitual behaviors [129].

We propose that, in addition to the prominent loss of dopamine in the sensorimotor striatal sector, the degeneration of CM neurons that project mostly to the sensorimotor region of the striatum could also play an important role in the reduced capacity to use habitual actions in PD. In section 2.3., we have discussed how animal studies suggest an involvement of TANs in habit formation. The available evidence indicates a strong regulation of TANs by CM/PF projections to the striatum. Although the exact mechanisms remain unknown, the CM/PF degeneration in PD could result in altered control of TANs, and therefore of striatal microcircuitry, with a concomitant deficit of habit formations or expression. Figure 5 presents a summary of the ideas proposed in this section of the manuscript. In PD, the sensorimotor striatum would be particularly affected by the combination of a severe loss of SNc dopamine inputs and a significant degeneration of glutamatergic inputs from the CM [133] [135]. In contrast, the associative striatum would retain a relatively higher level of dopaminergic innervation. In the sensorimotor striatum, the reduced dopamine and glutamatergic CM inputs would result in complex changes in the intrinsic GABAergic and cholinergic striatal microcircuitry, which would underlie the development of attention and set-shift deficits along with an inadequate balance of habitual versus goal-directed behaviors.

Fig. 5.

Schematic illustrating the possible roles of CM and SNc degeneration towards the development of deficits in habitual actions in PD. The massive dopaminergic denervation from the ventral tier of the SNc (SNc v) to the sensorimotor striatum combined with extensive CM cell loss, over the less affected dopaminergic innervation of associative striatal regions, may be the source of attention-related deficits PD patients display in performing habit behaviors (see text for details).

4. Open questions and future studies

Although recent years have witnessed important advances in our understanding of the functional anatomy of the thalamostriatal systems, many unresolved issues remain, that will necessitate careful scrutiny in order to dissect out the significance of these systems in mediating normal basal ganglia function, and their relevance towards the development of various behavioral deficits in PD and other basal ganglia circuitry disorders.

While in vitro studies have provided important insights into the possible cellular mechanisms by which thalamic afferents might regulate corticostriatal signaling to control MSNs and TANs activity [62, 67], these studies must be expanded, using specific stimulation of the caudal intralaminar thalamic complex, perhaps with the aid of optogenetic techniques [148, 149], to clearly assess the role of the CM/PF stimulation upon striatal activity. Similarly, the substrate underlying the complex pattern of responses recorded from TANs and MSNs in vivo following CM stimulation [113] must be elucidated. A careful analysis of the possible engagement of intrastriatal GABAergic microcircuits in mediating these effects is warranted [150, 151]. The loss of neurons in the CM/PF complex in PD (and other degenerative diseases, such as progressive supranuclear palsy, Huntington’s disease and Lewy body disease [132, 152, 153]) must be considered in our interpretation of the basal ganglia pathophysiology and learning dysfunctions that characterize these disorders. To do so, the line of research led by Kimura and his colleagues aimed at characterizing the physiological responses of CM/PF neurons to attention-related stimuli must be pursued and expanded to nonhuman primate models of PD. Animal studies have, indeed, provided evidence that CM neurons display abnormal physiological activity in parkinsonism [81, 136, 154–156], but the exact nature of these alterations, and their importance in CM/PF-mediated attention task regulation requires further consideration. The suggestion that CM neuronal loss is a compensatory response to the parkinsonian insult is of interest [157], but awaits further evidence that such a process takes place in the complex scheme of PD pathophysiology.

Our suggestion that the CM/PF neuronal loss may be an important contributor to the deficits in behavioral switching and habit behaviors in PD patients, must be further assessed through careful electrophysiological and behavioral studies in the MPTP-treated nonhuman primate model of PD. We have recently demonstrated that rhesus monkeys chronically treated over a period of 20–26 weeks with low doses of MPTP display brain pathological features that extend beyond the nigrostriatal dopaminergic system to involve other monoaminergic systems, and the CM/PF [158–160]. We believe that the use of this model represents a highly valuable asset to determine the potential role of the CM/PF-striatal system in motor, cognitive and limbic dysfunctions associated with PD.

Finally, it is important to recognize that the functional significance of the non-CM/PF thalamostriatal system is another obscure piece of the puzzle that remains to be clarified, if one hopes to fully understand the possible roles of the dual thalamostriatal systems in the functional circuitry of the basal ganglia. The integration of these connections within functional basal ganglia-thalamostriatal loops reminiscent of those proposed in this review for the various components of the CM/PF-striatal system (see Fig. 1) should be considered (see also [129, 161]). The anatomical and functional relationships between the non CM/PF-striatal connections and related corticostriatal afferents must be examined in great detail to elucidate the mechanisms by which these two neural systems may interact to regulate striatal activity and resulting basal ganglia function. The neurochemical, pharmacological and plastic properties of axo-spinous thalamostriatal versus corticostriatal excitatory synapses must be carefully assessed to determine the substrate through which these systems mediate their effects. Modern transgenic approaches combined with optogenetic stimulation methods could in principle be used to activate or silence specific subpopulations of non-CM/PF thalamostriatal neurons and assess their effects upon striatal activity. Such methods are being successfully used to study other brain systems in rodents (i.e. [162, 163]).

In conclusion, despite the long and unfruitful attempts at characterizing the role of the thalamostriatal systems in the functional organization of the basal ganglia, evidence discussed in this review highlights an interesting path that could shed light into this enigma. The development of proper animal models combined with their use in attention-related tasks that rely on the integrity of subcortical basal ganglia-thalamostriatal loops through the CM/PF complex may guide us towards a deeper understanding of the functional importance of the CM/PF-striatal system in altering the functional balance between the selection of habit and goal-directed behaviors in PD.

Abbreviations

ACh

Acetylcholine

CM

Centromedian nucleus of the thalamus

CM/PF

Centromedian/Parafascicular complex of the thalamus

CMl

Lateral seciton of the CM

GPi

Globus Pallidus, internal segment

LLF

Long latency facilitation neurons of the CM/PF

MPTP

1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine

MSN

Medium spiny neuron of the striatum

PD

Parkinson’s Disease

PF

Parafascicular nucleus of the thalamus

Pfdl

Dorsolateral section of the parafascicular nucleus

SLF

Short latency facilitation neurons of the CM/PF

SNc

Substantia Nigra paras compacta

SNr

Substantia Nigra paras reticulata

TAN

Tonically active neuron of the striatum

VA

Ventral anterior nucleus of the thalamus

vGluT1

Vesicular Glutamate Transporter type 1

vGluT2

Vesicular Glutamate Transporter type 2

VL

Ventral lateral nucleus of the thalamus