To interrogate large-scale neural circuitry at the cellular and synaptic levels, we monitor neural activity during normal brain function and during various manipulations. We use a range of different techniques to carry out this work, including imaging, electrophysiology, optogenetics, and behavior. Currently, the principal model system for our studies is the mouse visual cortex. Here are some examples of the work we do.
Two photon population calcium imaging in vivoIn this figure, neurons in visual cortex expressing GCaMP3 (a genetically-encoded calcium indicator, transfected using an AAV viral vector) are imaged with two-photon microscopy. Visual stimuli (drifting square wave gratings) presented to the mouse evoke orientation tuned, action-potential-associated calcium signals in individual, genetically-targetted neurons. We have used population calcium imaging to reveal principles of wiring in local neuronal populations (next section below; Smith and Hausser, 2010). (figure credit: Smith, Cottam, Keller, and Hausser, unpublished)
Shared receptive field subunitsUsing a type of two-photon population calcium imaging, our data revealed that individual neurons in mouse visual cortex recieve shared input. On the left, the receptive field subregions (in visual space) of 2 to 3 neurons are shown in outline next to diagrams of where the neurons were located in visual cortex (dorsal view). These shared receptive field subregions indicate that the neurons are primarily driven by a small number (1-3) of unique inputs. We know this, because if each neuron recieved strong drive from a large number of inputs, shared receptive field subregions would occur much less commonly (this is illustrated by the diagram on the left, above). Although single receptive field subregions were often shared among a set of neurons, it was rare for neurons to share more than one receptive field subregion overall, and thus the population exhibited diverse firing patterns (Smith and Hausser, 2010). This observation suggests a mechansim for generating diverse receptive fields in a local population, despite a very limited diversity of inputs. We are currently exploring the computational implications of this observation (Smith and Hausser, in preparation).
Nonlinear synaptic integration in vivoNeuronal dendrites have nonlinear, voltage-dependent properties, but it has remained unknown if, and how much, these nonlinear properties contribute to synaptic integration in vivo. To explore this question, we carried out a series of somatic and dendritic patch clamp experiments in visual cortex with evoked sensory responses. We found clear evidence of orientation-tuned, nonlinear electrical events on dendrites and nonlinear contributions to orientation tuning. Here, a dendritic patch clamp recording ~100 microns from the soma reveal orientation-tuned, burst-like responses indicitive of dendritic electrogensis. (Smith and Hausser, in preparation)
Development and plasticity of cortical mapsWe used intrinsic signal optical imaging (left panel) to explore the development of the retinotopic map in mouse visual cortex from just after eye opening at day 13 (right panel shows the development of the ipisilateral eye's retinotopic map). We found that the mechanisms for plasticity in this system are in place much earlier than previously believed. (Smith and Trachtenberg, 2007) Furthermore, the ipsilateral eye representation in visual cortex is highly sensitive to noise, which retards circuit development. Our findings are relevant for a type of amblyopia in which the two eyes are aligned, but have vastly different resolving powers (e.g., one eye is near-sighted and the other is far-sighted). Our data suggested a neural mechanism underlying the efficacy of vision correction over the more standard therapy of patching the weaker eye. Multicenter studies have shown the efficacy of vision correction alone.