Skip to main content

The rules of connectivity of genetically-defined long-range projections

Final Report Summary - LONG_RANGE_CC (The rules of connectivity of genetically-defined long-range projections)

The neocortex endows us with advanced cognitive abilities. Given the complexity of the neocortical circuit, the need for establishing simplifying principles of connectivity is paramount. Local cortical circuits form stereotyped connections across the neocortex. Within a cortical column, connections show laminar specificity and target preferentially particular cell-types while avoiding others. In addition, subnetworks of highly interconnected neurons are embedded in the local circuit. These subnetworks receive common local input and have similar functional properties in vivo, suggesting the might work as functional units. In contrast to our knowledge of the local circuits, the connectivity of long-range cortical circuits remains largely unknown. Activity within any cortical area impacts on many other cortical regions via direct monosynaptic inputs. As a consequence, understanding cortical function will require unraveling the logic of cortico-cortical (CC) interactions. CC connections have been implicated in several neurological diseases, such as schizophrenia and autism. Thus, unraveling the organizational logic of CC connections will lead not only to a better understanding of the neural basis of several cognitive functions but also will shed light on the circuits underlying disease states.
Despite the complexity of the cortical network, CC projections can be divided in two stereotypical classes: Projections either terminate in the middle layers (ascending or feedforward inputs; FF) or innervate the lower and upper layers, avoiding the middle ones (descending or feedback inputs; FB). Areas are reciprocally connected with projections of opposite types: areas receiving FF inputs send in return FB projections. We set up to study the organization of CC connections using a combination of optogenetic, electrophysiological and imaging methods. We focused on two main questions : 1) The principles organizing the connectivity of defined long-range cortico-cortical projections. 2) Determining how local and long-range cortical connectivity relate to each other. We found a series of organizational rules determining the connectivity of afferent CC connections.
Using channelrhosopsin-assisted circuit mapping we measured the specificity of FB inputs from high-order visual areas LM and AM to mouse primary visual cortex (V1). We found that both LM and AM inputs preferentially innervated neurons projecting to the source of FB in L5 but not in L2/3. Using a novel approach combing functional 2-photon imaging in single boutons and intrinsic signal imaging we measured the retinotopic specificity of FB inputs from LM to V1. We found an elegant organization where the tuning of afferents inputs for oriented moving edges is linked to the regions they target in V1’s retinotopic representation. LM axons spread around the retinotopically-matched location in V1 perpendicularly to their preferred orientation. Direction-selective axons targeted visual areas shifted from the retinotopically-matched position along the angle of their antipreferred direction. Our results show that, although they appear as anatomically “diffuse”, FB connections are made with a high degree of retinotopic specificity on L1 of V1, and that the distribution of FB synapses depends on their visual tuning . These anatomical observations shed light on the role of FB processing in cortical function and put constraints on models of FB function. Strikingly, the structure we found predicts that feedback inputs selectively target regions in V1 that are unlikely to be activated under the current stimulus sheding light on the function of FB afferent in cortical function.
Finally, we developed a new optical method to measure local divergence of afferent axons in local circuits. Out method measures shared connectivity by analyzing co-fluctuations of evoked responses in neurons after photostimulation of channelrhodopsin-expressing axons using a laser beam that can be rapidly moved using galvanometer mirrors. Using this novel method, we quantified the amount of shared thalamocortical inputs for connected and non-connected cell pairs in L4 and L2/3 of V1. We found that interconnected L4 and L2/3 pyramidal neurons are more likely to receive common input from dLGN than non-connected neurons. Furthermore, L4→L2/3 connected pairs were also more likely to receive inputs from the same thalamocortical axons than non-connected pairs. Our results provide a circuit mechanism for the observed amplification of lateral geniculate inputs by L4 circuits. They also show that, rather than flowing sequentially from L4 to L2/3 as commonly thought, afferent sensory information relayed from the thalamus to the cortex is concurrently processed in both layers by columnar, multilaminar sub-networks of interconnected neurons contacted by the same thalamocortical axons. The specificity of the monosynaptic dLGN→L2/3 inputs we found in this study suggests that L2/3 neurons integrate input from L4 as well as a copy of the specific thalamic responses that gave rise to the L4 responses. They also show exquisite, multi layer specificity in the connectivity of long-range afferents.
In summary, our recent research identified organizational rules of afferent long-range inputs to sensory cortex. We found that these afferents specifically connect with neurons in their target area depending on their projection pattern and local connectivity. In addition the functional properties of individual afferent axons within a projection also determine the region within V1 that they target. The organizing rules we identified during this research project put constrains theories of cortical function. We anticipate they will also help identify the role played by these same afferents during neurological disease states.