Nuno Miguel Macarico A. da Costa

Email:	ndacosta@ini.phys.ethz.ch
Work phone:	+41 44 635 3044
Home page:	http://www.ini.uzh.ch/~ndacosta

“Our ignorance is such that, even today, to a question so important and yet so simple as the general connections of nerve cells, science does not have an answer for the more or less probable hypotheses” Ramón y Cajal (1891)
Mapping the circuit diagram of the cortex has been a long term goal in neuroscience, and in recent years there has been an intensification of this effort, to completely reconstruct, at the ultrastructural level, a volume of neocortex. We offer an alternative methodology to achieve the same goal. Instead of mapping every synapse onto every neuron of the cortical microcircuit, we describe the statistics of connectivity. We combine light and electron microscopy to precisely map the synapses of an afferent pathway onto their target neurons. By doing this we are able determine the number of synapses a pathway forms with their target neurons, the size of the synapses, their target preference (dendritic shafts or spines), the dendrites onto which the synapses are formed, and the statistics of the locations of these synapses along the dendritic tree. Such complete structural information is crucial to any biologically-realistic model of cortical computation. We use as a model system the primary visual cortex of the cat, which allows us to put our structural results together with more than 50 years of physiology collected at the circuit level, the cell level and the synaptic level.
Previously we have precisely mapped and described the thalamocortical input to spiny stellates and corticothalamic neurons and show that it provides a minute percentage of all the excitatory synapses on these neurons. Most synapses in the cortex are provided by local neurons, and now we intend to extend our work to map the distribution of intracortical synapses into their target neurons. We intend to combine light and electron microscopy to precisely map synapses of an afferent pathway on to their target neurons. Our goal is to develop methodologies that will allow us to morphologically identify different presynaptic terminals at the ultrastructural level, and then map the locations of these terminals on the dendrites of cortical neurons which have been previously reconstructed at the light microscopy level. Such a fingerprint of a bouton (or synaptic connection), if it exists, is a powerful tool, and much more efficient than labeling all the different inputs to a neuron. This implies reconstructing the dendrites of major neuron types. This was done before by reconstructing the entire dendrite, however this is not the only option: sampling smaller portions of dendrite, but from more cells and more dendrites, will with the same likely effort give statically more significant results. Parts of the process could also more easily be automated, because the individual dataset of EM serial sections would be smaller.
The connectivity maps of Binzegger et al. (2004) give us an idea not only of the size of the problem that we are faced with, but also of the different possible inputs to each neuron and the probability of finding them. The results of these project might provide us with an impressive tool for connectivity, but the challenge lying ahead is not only to obtain the “general connections of nerve cells” but also to articulate it with the knowledge that we have of its physiology.