Macroscale cortical organization and a default-like apex transmodal network in the marmoset monkey

Abstract

Networks of widely distributed regions populate human association cortex. One network, often called the default network, is positioned at the apex of a gradient of sequential networks that radiate outward from primary cortex. Here, extensive anatomical data made available through the Marmoset Brain Architecture Project are explored to show a homologue exists in marmoset. Results reveal that a gradient of networks extend outward from primary cortex to progressively higher-order transmodal association cortex in both frontal and temporal cortex. The apex transmodal network comprises frontopolar and rostral temporal association cortex, parahippocampal areas TH / TF, the ventral posterior midline, and lateral parietal association cortex. The positioning of this network in the gradient and its composition of areas make it a candidate homologue to the human default network. That the marmoset, a physiologically- and genetically-accessible primate, might possess a default-network-like candidate creates opportunities for study of higher cognitive and social functions.

Fig. 1

Flat map format and candidate zones of interest. All cortical areas are displayed on a flat map that minimizes distortion. The lateral (top left) and midline (top right) show the volume surface models of the marmoset cortex color-coded corresponding to the flat map representation below. Relevant areas are labeled for orientation as well as MOT (primary motor cortex A4ab) and AUD (auditory cortex involving multiple primary auditory areas). The major zones of interest in this paper are highlighted by blue rectangles: (I) frontopolar A10, (II) posterior midline A29a–c, A23, caudal A30, (III) rostral temporal association cortex TE3/TPO/PGa/IPa, (IV) posterior parietal cortex Opt/PG, and (IV) parahippocampal cortex TH/TF. These are not the only zones implicated in the default network but represent five zones that are candidate homologs to several of the well-studied regions implicated in default network-A in the human. Area labels use nomenclature of the Paxinos et al. atlas

Fig. 2

Aggregate analysis of anatomical connectivity reveals a macroscale organization of networks. Each row displays a candidate network that is based on a frontal injection (left column), replication of the frontal injection (middle column), and corroboration of the network using a posterior injection (right column). Flat map format from Fig. 1. These displayed networks represent a partial subset of possible networks that could be plotted and do not reflect the full complexity of the projection patterns. Nonetheless, they reveal a sequence by which progressively more distributed networks populate the cortex as one goes from primary motor cortex (A4ab) through to frontopolar A10. In each map, the tracer injection is shown with a red dot (all retrograde). Injection cases are labeled in the bottom right as annotated in the Marmoset Brain Architecture Project archive

Fig. 3

Anatomical evidence for an apex transmodal network in the marmoset. a–f Multiple injections from frontopolar cortex A10 and adjacent areas are illustrated. The blue bounding boxes in a identify five target zones hypothesized to be part of the default network-A, labeled I–V. The area labels in b are displayed for key areas useful for orienting to components of default network-A. The blue arrow in a notes projections originating in A8ab, which is examined in more detail in Fig. 5. The gray arrow in a notes projections within and along the border of A47O/A13L/A13M that are discussed in the text. Arrows also illustrate similar patterns from temporal injections in Fig. 4e

Fig. 4

Evidence for involvement of temporal association cortex in the apex transmodal network. Injections in estimated TPO (a, b) are distinct from injections to nearby auditory areas AuCPB and AuML (c, d). TPO injections receive projections from the full constellation of regions predicted as components of the default network-A candidate as well as nearby auditory areas, whereas the auditory areas receive projections predominantly from adjacent auditory areas. e Injection of TE3 also yields a default-network-like pattern, while largely sparing auditory cortex. f The gradient of temporal lobe projections to frontal regions is illustrated by combining the anterior projection patterns of Fig. 2 into a single multicolored image: A4ab, purple; A6DC, blue; A6DR, green; A8aV, yellow; A47L, orange; A10, red. A clear progression into rostral temporal association cortex is observed. The three blue circles mark the locations of the TPO and TE3 injections and the two tan circles mark the auditory cortex injections (AuCPB and AuML). In addition to revealing the network gradient, this composite image illustrates a region of temporal cortex where transmodal association areas that are part of the default-network-like candidate are near to auditory sensory areas (see also Supplementary Fig. 1)

Fig. 5

Additional areas may be components or subdivisions of the apex transmodal network. The human default network contains regions outside of the five-targeted zones, including additional prefrontal regions and potentially more extensive areas along the posterior midline. Injections to two relevant frontal areas are shown, A8b (a, b) and A8aD (e, f). The two prefrontal areas are displayed alongside posterior PGM (c, d). A8b receives projections from within or near several zones of the candidate default network, including TPO and regions along the posterior cingulate (A30, A29a-c, A23V, and ProsSt). A8b also receives projections from parahippocampal area TH. A8aD displays overlap. PGM possesses some overlap, in particular for the injection in c that is in the ventral portion of PGM bordering A32V and A30. The position of the Ventral PGM injection site is shown by a red diamond in panel a to fully appreciate its location in relation to the borders of areas and in relation to the A8b injection pattern. The PGM injection falling fully within PGM reveals few (if any) projections from ventral posterior cingulate and generally spares temporal association and parahippocampal cortex

Fig. 6

The apex transmodal network is spatially distinct from a canonical distributed sensory-motor network. Images illustrate the contrast between injections within the apex transmodal network and the sensory-motor network linked to A8aV-MT. In each image, averages of multiple injections are contrasted. Displayed to the left are average projection maps for injections to frontopolar cortex (red) contrasted with A8aV injections (yellow). The distinct patterns of projections in posterior cortex are striking including that the apex transmodal network involves extensive regions of rostral temporal association cortex that are juxtaposed to caudal sensory-aligned regions (dashed rectangle) as well as a punctate region of parietal association cortex near PG/Opt that is surrounded by regions more aligned to the sensory-motor network (blue arrow). Displayed to the right are average projections for rostral temporal injections (TPO/TE3) contrasted to MT injections. This map reveals the anterior separation of the networks including the region near to A8aV (a putative homolog of FEF) that is preferentially distinct and caudal to the apex transmodal network