Divergence-convergence ratios govern functional circuitry in the early visual pathway

Resumo

One of the main goals of visual neuroscience is to understand how the early visual pathway (EVP) works, that is, what are the operations that it performs and why. For this, structure, function and behavior must be related. However, relating structure to function in a coherent way is not so easy. Even though connectomics is a growing and promising field, in many cases, it can give an excess of information that we do not know yet how to interpret or relate to cortical function. To have a complete map of the synapsis of a cortical structure does not necessary mean that it is possible to understand how the weights are distributed across the circuit and therefore how the functional architecture is built. In this work I will show how a comparative neurobiology approach (Pettigrew 2004) is really useful to build models of cortical function. It enables to identify the basic parameters that may constrain the development of cortical circuits, which gives the basis to generate hypothesis of the operations the visual system is performing, and to proof them using computational models. In the first place, I present an extension of our previous statistical wiring model (Martinez et al 2014) that, using a single coverage optimization principle, maximizes the transfer of visual information from the retina to the primary visual cortex (V1), and accounts for the functional and topological differences found at the level of V1 across the mammalian phylogenetic tree. In particular, the model reproduces the experimentally derived differences in cortical magnification factor, which we express in terms of relative retina-to-V1 cortical area; it describes the existence of a continuum of different divergence-convergence ratios (DCr) between the retina and V1 through the lateral geniculate nucleus of the thalamus (LGN); and finds a developmental threshold, based on the extent of the emergent local correlations, that explains the transition between the decorrelated V1 salt-and-pepper cortical structures typical of rodents, and the topologically organized cortical orientation-preference maps (OPMs) characteristic of carnivores and primates. The model correctly predicts the cortical structure of all mammalian species that have been experimentally explored, and makes clear testable predictions about how the cortical topology of other transition species, those that are right at the threshold of continuous distribution, should be. In the second place, since retinotopy is the most present and relevant functional constraint across mammalian species. A mathematical model of the development of correctly oriented topographic maps in visual cortex is presented that takes into account the synchronization between the two retinas mediated by retino-retinal (R-R) connections, to explain the role of molecular and activity dependent mechanisms in the establishment of retinotopic maps across phylogeny. Demonstrating that the presence or absence of R-R connections across different species is completely related to their different DCr values through phylogeny. In third place, we present a set of experimental results where V1 of a strain of mutant mice (Brn3b-Zic2) that have a larger proportion of ipsilateral fibres, was characterized using optical imaging of intrinsic signals. This manipulation alters the input that reaches V1 and, therefore, increases the DCr of the ipsilateral fibres through the EVP. As a result, Brn3b-Zic2 mice develop Ocular Dominance (OD) columns that resemble those presented in Ocular Dominance maps (ODM) of carnivores and primates, which emphasizes the role that the DCr plays in the development of V1 functional structure.