Connectivity motifs underlying neuronal computations in the adult OB

One of the great challenges in neuroscience is to explain the function of neuronal circuits as a consequence of the connections and interactions between large numbers of neurons. To understand the dynamics of small microcircuits, it is indispensable to study the connectivity between neurons and its influence on the function of the intact system. However, for most neural circuits the detailed wiring diagram is unknown. The zebrafish is an excellent model system to address these questions because it is small in size, facilitating large-scale neuronal reconstructions, and because a variety of established tools can be combined to identify neurons and measure neuronal activity patterns. The olfactory bulb (OB) is the first stage of information processing in the olfactory system and has become a model for studies of neural circuit structure and function in the intact brain. We used serial block face electron microscopy to obtain a large high-resolution anatomical dataset from the adult olfactory bulb. Focused reconstruction from a small subset of neurons revealed the complete structure of mitral cells. Synapse mapping revealed their complete synaptic map. In higher vertebrates an excitatory interneuron, external tufted cell, has been shown to receive direct synaptic input and then relay the information to the mitral cells. We observed that mitral cells in adult zebrafish receive direct synaptic inputs from the olfactory receptor neurons. Both these scenarios of information transfer can have different computational consequences in the OB.
Two recent computational studies focused on reciprocal connectivity in the OB between mitral cells and granule cells. One model predicts that granule cells activity is ultra-sparse and mitral cell activity is sparse, and that this type of reciprocal circuit helps to perform decorrelation in the OB. Another computational model used the same mathematical approach but comes with different predictions. It puts no constrains on the sparseness of MC and GC cells. It further predicts that the reciprocal circuit performs contrast normalization and brings the OB into a variance-balanced state. We tested both these models experimentally and saw indeed that the granule cell activity was not sparse, thereby showing that the first model is not biologically plausible in the OB. The fraction of reciprocal synapses out of all the input synapses on the mitral cells was significant. With this fraction the olfactory circuit still performs strong contrast normalization thereby showing that the second model is biologically plausible.