Network structure of the mouse brain connectome with voxel resolution

Abstract

Fine-grained descriptions of brain connectivity are fundamental for understanding how neural information is processed and relayed across spatial scales. Prior investigations of the mouse brain connectome have employed discrete anatomical parcellations, limiting spatial resolution and potentially concealing network attributes critical to the organization of the mammalian connectome.Here we provide a voxel-level description of the network and hierarchical structure of the directed mouse connectome, unconstrained by regional partitioning. We show that integrative hub regions can be directionally segregated into neural sinks and sources, defining a hierarchical axis. We describe a set of structural communities that spatially reconstitute previously described fMRI networks of the mouse brain, and document that neuromodulatory nuclei are strategically wired as critical orchestrators of inter-modular and network communicability. Notably, like in primates, the directed mouse connectome is organized along two superimposed cortical gradients reflecting unimodaltransmodal functional processing and a modality-specific sensorimotor axis. These structural features can be related to patterns of intralaminar connectivity and to the spatial topography of dynamic fMRI brain states, respectively. Together, our results reveal a high-resolution structural scaffold linking mesoscale connectome topography to its macroscale functional organization, and create opportunities for identifying targets of interventions to modulate brain function in a physiologicallyaccessible species.

Cortical silencing results in paradoxical fMRI overconnectivity

“…This interpretation is supported by empirical and computational evidence suggesting that structural and rsfMRI-based connectivity are robustly related. For example, structural and functional connection strengths are correlated both at the whole-brain and mesoscopic scale (Hagmann et al, 2008;Wang et al, 2013;Coletta et al, 2020), and rsfMRI network topography closely recapitulates patterns of anatomical connectivity (Honey et al, 2009;Goni et al, 2014). In keeping with this, experimental resection of callosal connections (O'Reilly et al, 2013) and chemogenetic inactivation of the amygdala have been shown to reduce functional connectivity with regions anatomically linked to the manipulated areas (Grayson et al, 2016).…”

“…Importantly, all the thalamo-cortical regions showing increased rsfMRI connectivity in Kir2.1 mice are characterized by high projections density from the PFC, as seen by comparing the magnitude of intergroup rsfMRI connectivity differences with incoming axonal connectivity strength inferred from a voxel-model of the mouse brain connectome (Coletta et al, 2020) (Fig. 1F, Wilcoxon rank sum test, p<0.0001).…”

“…First, the observation of increased rsfMRI coupling in the face of robust neural silencing is sufficient to disprove a general monotonic relationship between interareal coupling and rsfMRI connectivity and serves as a demonstration for a non-dyadic relationship between these two entities, with implications both for the mechanistic modelling of rsfMRI coupling and the reverse engineering of rsfMRI signals recorded in brain disorders. Second, it should be noted that the light sedation regimen employed in this study is physiologically stable, preserves sensorial responsivity (Orth et al, 2006) and does not induce burst-suppression or cortical hyper-synchronization (Liu et al, 2011), resulting in preserved rsfMRI connectivity topography (Coletta et al, 2020;Whitesell et al, 2020) and state dynamics (Gutierrez-Barragan et al, 2019). These properties make the brain state under which our recordings were carried out not dissimilar from the quiet wakefulness conditions, often bordering into sleep (Tagliazucchi and van Someren, 2017), that characterize most rsfMRI scanning in humans (Reimann and Niendorf, 2020).…”

“…Our thalamic rsfMRI probing suggests that this effect may be possibly gated and propagated by polymodal thalamic nuclei to produce wider cortical over-connectivity. The observation that structural and functional connectivity strength are correlated both at the whole-brain and mesoscopic scale (Hagmann et al, 2008;Wang et al, 2013;Coletta et al, 2020) lies at the basis of structurally-based models of rsfMRI connectivity, according to which spatiotemporal correlations in spontaneous rsfMRI activity are critically shaped and constrained by underlying axonal connectivity (Friston and Büchel, 2006;Alstott et al, 2009;Grayson et al, 2016). A key prediction of these models is that regional inactivation of a neural node would result in reduced functional synchronization in distributed target areas, the spatial extent of which is dictated by the structural topography of the affected region (Alstott et al, 2009).…”

“…To relate the strength of underlying anatomical connectivity to the regions exhibiting increased rsfMRI connectivity with voxel resolution we extracted outgoing projections from the affected PFC regions using a spatially-resampled (0.027 mm 3 ) version of a voxel scale model of the Allen Brain Institute structural connectome (Coletta et al, 2020). We then plotted the strength of PFC-departing structural projections against the corresponding between-group difference in rsfMRI connectivity using the cluster-corrected difference map, and assessed differences in the distribution of overconnected areas with respect to all the brain voxels using a Wilcoxon rank-sum test.…”