ACC-striatal Projections are Uniquely Positioned to Interact With an Unusually Functionally Diverse Set of Cortical Areas

Background: Cortico-striatal projections play a central role in providing the basal ganglia with integrated information relevant to action selection and execution, making them a key candidate for the treatment of goal-directed and habitual disorders such as addiction and OCD. Probabilistic models have successfully summarized the spatial topography of this circuitry (Averbeck et al., 2014). For example, projections originating caudally from motor control areas terminate in the dorsolateral striatum, whereas rostral projections from ventromedial prefrontal cortex (PFC) terminate in ventromedial striatum. However, unlike the dorsolateral and ventromedial striatal sectors, the rostral and central striatum receives converging projections from multiple areas. This region is particularly complex and serves as an interface between functionally diverse cortical regions, the orbitofrontal cortex (OFC), ventrolateral PFC (vlPFC), and anterior cingulate cortex (ACC). Although there is an underlying topography to these connections, this alone may not be sufficient to understand the complexity of their interactions. The ACC is a particularly good example with its functional diversity and distributed striatal projections. The goal of this study was to develop a data-driven methodology that works in tandem with the more traditional topographic approach, to uncover additional key organizational principles of cortico-striatal circuitry. We focus specifically on the relationship between ACC-striatal projections to those originating in other areas of PFC. Methods: We analyzed 30 anterograde injections sites evenly distributed throughout macaque PFC (Averbeck et al., 2014). For each injection, we tabulated the loci of its terminal projections according to a striatal grid composed of 600 μm isotropic voxels. We then computed the similarity of the projections among every pair of cortical injections using a metric that corrects for chance agreement (adjusted rand index), and clustered cortical regions using spectral methods on the resulting similarity matrix. This identified two dominant gradients of cortical organization. We then visually examined the pattern of projections of every cortical cluster. Finally, based on previous qualitative observations that PFC projections to the striatum are mirrored along a dorso-ventral axis in ACC (Tang et al., 2019), we asked whether PFC regions could be similarly organized based on their connectivity to this region. We replicated the clustering analysis using 10 cortico-cortical retrograde injections distributed along the ACC. For each injection site, we computed the proportion of cells that projected to it from each area in PFC, and clustered cortical areas by the similarity of their projection pattern across injection sites. Results: The primary gradient organized cortical areas based on the location of their terminal projections along a ventromedial to dorsolateral axis in striatum. The secondary gradient captured the size of these projections. In line with classic topographic findings, the combined gradients positioned areas in ventromedial PFC as focally projecting to ventromedial striatum, whereas motor and dorsal PFC areas projected focally to dorsolateral striatum. In contrast, regions in ACC and portions of OFC had broad projections in central striatum, displaying partial overlap with the two more focal clusters. We found no significant relationship among injection volume, size of terminal projections, or the mean similarity of a given area (all pairwise correlations p > 0.1). Clustering performed on retrograde ACC injections resulted in a comparable primary gradient in PFC; motor and limbic areas primarily projected to dorsal and subgenual ACC, respectively, with remaining areas mostly projecting to rostral ACC. Conclusions: In this project, we used a data-driven approach to further characterize the underlying organizational principles of cortico-striatal circuitry. We used this approach and identify that ACC stands out in its connectivity within this network. We show that projections to striatum are organized by the size of their terminal fields in addition to their rostro-caudal location. The combination of these axes of organization distinguished ACC as an area with particularly widespread connections to the striatum. These results were not explained by methodological issues, such as injection volume. Further, we note commonalities between the organization of connections between prefrontal areas and both the ACC and the striatum. Based on these findings, we conclude that the ACC integrates information from diverse functional areas and relays this information to other functional groups via its relatively large projections to striatum. The mirrored cortico-cortical organization observed here raises the possibility that an individual’s topography of cortico-striatal projections, which can be difficult to precisely map using neuroimaging methods, could be approximated using cortico-cortical connectivity with ACC. This possibility could facilitate the identification of cortical targets for transcranial stimulation. Future work will test these findings using networks derived from diffusion and functional MRI in human and non-human primates