SUMMARY THESIS YVETTE C. VAN DONGEN

The main aim of the present study was to determine how integration of information may take place within and between cortical-basal ganglia-thalamocortical systems in order to better understand how the complex neuronal forebrain circuits contribute to the production of goal-directed and complex behavior. As introduced in the first Chapter of this thesis, the basic principle of the cortical-basal ganglia relationships is essentially based on the parallel processing of information in functionally distinct basal ganglia-thalamocortical circuits (Alexander et al., 1986, 1990; Groenewegen et al., 1987; Wiesendanger et al., 2004). Theoretical and experimental approaches to explain and predict how these basal ganglia circuits function and interact, have been taken at different levels of analysis. In very general terms, it has been suggested that the basal ganglia may play an important role, in close association with the (pre)frontal cortex, in selecting an appropriate motor or behavioral output in a particular context (Pennartz et al., 1994; Mink, 1996; Redgrave et al., 1999). For instance, Pennartz et al. (1994) focused on the microcircuit level in proposing in the ‘ensemble hypothesis' of striatal function a prominent role for medium-sized spiny projection neurons (MSN) of the rat nucleus accumbens (Acb) with a relatively ‘long-range' complex of axonal ramifications from MSN for cross-talk and cross-regulation of activity within, as well as between, compartments or functionally distinct neuronal ensembles. At the system level, Haber et al. (2000) proposed that the striato-nigrostriatal pathways in primates form an ascending spiral from the Acb shell to the dorsolateral striatum, which might argue for a serial, hierarchical organization of behavior involving successively more dorsal parts of the striatum as proposed by Redgrave et al. (1999). These striato-nigrostriatal relationships comprise both reciprocal and non-reciprocal components suggesting that the DA system is involved in both feedback and feedforward loops with the striatum. We sought to determine to what extent the functional-anatomical organization of the microcircuitry of the Acb would lend support for the ‘ensemble hypothesis' ( Chapter 2 and Chapter 3 ) and, in addition, to what extent the Acb shell would exert a synaptic influence on the two subpopulations of nigrostriatal neurons that innervate the sensorimotor dorsolateral striatum in rodents ( Chapter 4 ).

SUMMARY OF THE FINDINGS BY CHAPTER

In Chapter 2, the anatomical organization of intrastriatal communication between the two major subregions of the rat Acb, i.e., shell and core, was investigated. As highlighted in the first Chapter of the present thesis, the Acb is thought to subserve different aspects of adaptive and emotional behaviors. The anatomical substrates for such actions are multiple, parallel ventral striatopallidal output circuits originating in the Acb shell and core subregions. Several indirect ways of interaction between these two subregions and their associated circuitry have been proposed, in particular through striato-pallido-thalamic (Zahm and Brog, 1992; Joel and Weiner, 1994; Groenewegen et al., 1994; O'Donnell et al., 1997) and dopaminergic pathways (Otake and Nakamura, 2000; Haber et al., 2000). In addition to these indirect, multisynaptic pathways between the Acb shell and core, Heimer et al. (1991) described intrastriatal ‘associational' projections in the ventral striatum. According to Heimer and colleagues (1991), these intrastriatal Acb fibers represent a fine and delicate system that are rather restricted in its distribution. These authors further noted that the ‘intrastriatal association fibers' could cross the shell-core boundary. In order to investigate to which extent the Acb shell and core are directly interconnected and whether a specific organization underlies the intra-accumbens distribution of axon collaterals, we placed small injections of the anterograde neuroanatomical tracers Phaseolus vulgaris- leucoagglutinin (PHA-L) and biotinylated dextran amine (BDA) in different parts of the Acb. These experiments were supplemented by single-cell juxtacellular injections with the tracer neurobiotin, in order to characterize the neurons that give rise to intrastriatal projections.

Our results confirm the observations by Heimer et al. (1991) of intrastriatal ‘associational' projections and show, in addition, the specific organization of these projections using much smaller injections in different subareas of the Acb. For instance, we demonstrate for the first time widespread intra-accumbens projection patterns, including reciprocal projections between specific parts of the shell and core. However, fibers originating in the core reach more distant areas of the shell, including the rostral pole (i.e., the predominant calbindin-poor part of the shell anterior to the core) and striatal parts of the olfactory tubercle, than those arising in the shell and projecting to the core. The latter projections are more restricted to the border region between the shell and core. The density of the fiber labeling within both the shell and core was quite similar. Moreover, specific intrinsic projections within shell and core were identified, including a relatively strong projection from the rostral pole to the rostral shell, reciprocal projections between the rostral and caudal shell, as well as projections within the core that have a caudal-to-rostral predominance. The results of the single-cell juxtacellular tracing experiments show that MSN and medium-sized aspiny neurons (most likely fast- spiking) contribute to these intra-accumbens projections. As to the dorsal striatum, several studies using intracellular filling have shown that MSN (Bishop et al., 1982; Kawaguchi et al., 1989, 1990), as well as interneurons (Wilson et al., 1990; Kawaguchi, 1993), have axons that project up to 1 mm away from the parent cell body. While such neurons are GABAergic, the intrastriatal projection patterns indicate the existence of lateral inhibitory interactions within, as well as between, shell and core subregions of the Acb. Although not much is known about the functional aspects of shell-to-core and core-to-shell projections, the available evidence supports the hypothesis that the core may convey information to the shell to overrule its modulation of primary, non-learned behavior (e.g. the sight of an apple may counteract the shell-mediated inhibition of feeding)(Parkinson et al., 1999). Finally, our present findings of reciprocal connections between the rostromedial and caudomedial shell suggest a possible involvement in funneling the motivational valence of positive (appetitive) vs. negative (aversive) states (Kelley, 1999, 2004).

To further clarify the anatomical organization of intrastriatal communication within the two major subregions of the rat Acb, i.e., shell and core, in the studies described in Chapter 3 we investigated the three-dimensional (3D) organization of dendrites and axons of MSN of the Acb in relation to subregional (shell-core) and compartmental (patch-matrix) boundaries. Previous studies in the primate caudate nucleus have demonstrated that MSN have preferred dendritic orientations that tend to parallel the orientations of the striosomes (striatal compartments; Walker et al., 1993). Moreover, recurrent axon collaterals of MSN in the rat dorsal striatum have been categorized into two types, i.e., restricted and widespread (Kawaguchi et al., 1990), which tend to respect the boundaries between patch and matrix (Kawaguchi et al., 1989). As highlighted in the general introduction of this thesis, the Acb has a highly complex compartmental organization. Arts and Groenewegen (1992) described the relationships of the dendrites of MSN with the compartmental structure of the Acb using in vitro

Our results show for the first time that dendritic arbors of MSN in both the Acb shell and core subregions are preferably oriented, i.e., that they are flattened in at least one of the 3D-planes. The preferred orientations are influenced by the shell-core and patch-matrix boundaries, they conform to the boundaries, suggesting a particular interaction between the MSN and their afferents. This supports the idea of parallel and independent processing of information in both the shell and core to be transmitted to their respective segregated target areas in the basal forebrain, diencephalon and mesencephalon. Dendritic orientations of MSN of the Acb core are more heterogeneous, than those of the Acb shell and dorsal striatum, suggesting a more complex distribution of striatal inputs within the Acb core. While dendrites respect the shell-core and patch-matrix boundaries in shell and core, we demonstrate for the first time that local axon collaterals may cross these boundaries. Finally, different degrees of overlap between dendritic and axonal arborizations of individual MSN were identified, suggesting various possibilities of lateral inhibitory interactions within, as well as between, functionally distinct territories of the Acb. In this way, lateral inhibition between MSN would support a ‘Winner-take-all' mechanism in the selection of striatal outputs.

In Chapter 4, we investigated the indirect dopaminergic pathway of interaction between the rat Acb shell and the nigrostriatal neurons innervating the sensorimotor region of the dorsal striatum. A recent analysis of the pattern of striatonigral and nigrostriatal projections in both rats and primates revealed that each functional territory of the dorsal striatum is innervated by two main subpopulations of nigrostriatal dopaminergic neurons (Maurin et al., 1999; Haber et al., 2000). The first population, denominated “proximal”, occupies a position in register with the striatonigral projections, and may be involved in reciprocal striato-nigrostriatal connections. The second population, denominated “distal”, is likely involved in non-reciprocal connections with the dorsal striatum. However, both the anatomical organization of Acb shell projections to the nigrostriatal dopaminergic neurons innervating the sensorimotor striatum and the synaptic influence exerted by the Acb shell on these neurons, remain to be determined. These issues were addressed in the rat using neuroanatomical and electrophysiological approaches.

Combined anterograde tracing from the Acb shell with retrograde tracing from the sensorimotor region of the dorsal striatum revealed that labeled fibers from the Acb shell overlap retrogradely labeled nigrostriatal neurons located in the medial (substantia nigra pars compacta (SNC) and the lateral ventral tegmental area (VTA) but avoid the nigrostriatal neurons located laterally. In addition, stimulation of the Acb-shell induced an inhibition, as noted by a decrease of firing rate, of dopaminergic nigrostriatal neurons projecting to the sensorimotor striatal territory. In agreement with the anatomical observations, these responses were observed in nigrostriatal neurons located in the medial SNC and the lateral VTA but not in nigrostriatal neurons located laterally. These data further establish the existence of a functional-anatomical link between the Acb shell and the sensorimotor striatum via dopaminergic nigrostriatal neurons. The present study also reveals that among the dopaminergic nigrostriatal neurons innervating the sensorimotor striatal territory, only the subpopulation located in the medial SNC and lateral VTA receives an inhibitory input from the Acb shell. This indicates a functional heterogeneity within the population of dopaminergic neurons innervating a given striatal territory. This suggests that not all dopaminergic neurons act in the same way, i.e., in reacting to unexpected rewards (Schultz, 1997). Rather, some dopaminergic neurons may get different information than others, which should enable them to react differently to different stimuli or context.

To summarize, our present studies in rats ( Chapter 2-4 ) outline several nodal points at which information from separate cortical-basal ganglia-thalamocortical loops can influence each other. Chapter 2 and Chapter 3 describe the interacting neural networks of shell and core at the level of the Acb. The directionality in these intranuclear projections indicates that in addition to a ‘limbic-to-motor' flow, ‘motor-to-limbic' transfer of information may also occur. These data argue against a strict spiralling hierarchy, in which the shell would primarily influence the core. Rather, the core may also influence the shell directly. Chapter 4 describes an open component of the striato-nigrostriatal pathway in the rat that allows direct interactions between segregated corticostriatal channels through non-reciprocal connections to the substantia nigra. The directionality in these projections indicate a ‘ventral-to-dorsal' transfer of information.