The beguiling simplicity of the patterns formed by the superficial patch system hides a suprising complexity of construction. Understanding how to build a patch system out of the collaboration of single axons is strangely difficult. We found that a fair amount of information must be shared between neurons to enable them to form the patch system.

Although the clustered axonal arbors of single pyramidal cells have been examined in detail, the precise rules by which these neurons collectively merge their arbors remain unknown. To discover these rules, we generated models of clustered axonal arbors following simple geometric patterns. We found that models assuming spatially aligned but independent formation of each axonal arbor do not produce patchy labelling patterns for large simulated injections into populations of generated axonal arbors ("Crystalline connectivity" figure). In contrast, a model that used information distributed across the cortical sheet to generate axonal projections reproduced every observed quality of cortical labelling patterns ("Reconstructing daisies" figure). We conclude that the patch system cannot be built during development using only information intrinsic to single neurons. Information shared across the population of patch-projecting neurons is required for the patch system to reach its adult state.

Several aspects of the labelling patterns that form the primary description of the superficial patch system are not easily explained. The first difficulty is that the size of a labelled patch never exceeds some maximum size in a given cortical area. Although discrete patch size is largely independent of injection size, very large injections in visual cortex nevertheless reveal a qualitatively different pattern in the patch system. In at least tree shrew, primate and quokka, large pressure injections result in a latticework of labelling immediately surrounding the injection site, composed of walls of labelled terminals and somata surrounding lacunae of relatively unlabelled tissue (Rockland et al. 1982 — see the figure at top). With increasing distance from the injection site, this latticework breaks up into separately labelled patches of the characteristic size for the cortical area that contains them. Retrogradely labelled somata are observed within the lattice walls, and not within lacunae.

We proposed a model that reproduced the complexity of clustered labelling patterns in primary visual cortex. To accomplish this, it was necessary to assume that different points in cortex follow different rules for growing horizontal axonal arbors. Neurons inside cytochrome-oxidase blobs and close to orientation pinwheel centres both had different sets of rules to neurons elsewhere across the simulated cortical surface. Whether a model using fewer than our three connectivity rules could reproduce the labelling patterns in cortex remains an open challenge.

Ours is the first model that captures all features of clustered labelling in primary visual cortex for tracer injections of any size, relying only on known anatomical features of visual cortex. It is precisely the regularity and homogeneity of the patch system in primary visual cortex that makes explaining the patch system there difficult. However, our model is consistent with the concept of like-to-like connectivity and could therefore be generalised to areas of cortex without smoothly regular functional maps.

Publications

This work was published in Cerebral Cortex: DR Muir, RJ Douglas. 2010. From Neural Arbors to Daisies. Cerebral Cortex 21 (5), pp 1118–1133. DOI: 10.1093/cercor/bhq184.

Funding

This work was supported by the European Commission [FP6 2005-015803 DAISY to DRM and RJD]; and the John Crampton Travelling Scholarship to DRM.