Summary Efficient retrograde access to projection neurons for the delivery of sensors and effectors constitutes an important and enabling capability for neural circuit dissection. Such an approach would also be useful for gene therapy, including the treatment of neurodegenerative disorders characterized by pathological spread through functionally connected and highly distributed networks. Viral vectors, in particular, are powerful gene delivery vehicles for the nervous system, but all available tools suffer from inefficient retrograde transport or limited clinical potential. To address this need, we applied in vivo directed evolution to engineer potent retrograde functionality into the capsid of adeno-associated virus (AAV) — a vector that has shown promise in neuroscience research and the clinic. A newly evolved variant, rAAV2-retro, permits robust retrograde access to projection neurons with efficiency comparable to classical synthetic retrograde tracers, and enables sufficient sensor/effector expression for functional circuit interrogation and in vivo genome editing in targeted neuronal populations.

The thin basal and oblique dendrites of cortical pyramidal neurons receive most of the cells' synaptic input, but their integrative properties remain uncertain. Previous studies have most often reported global linear or sublinear summation. An alternative view, supported by biophysical modeling studies, holds that thin dendrites provide a layer of independent computational 'subunits' that sigmoidally modulate their inputs prior to global summation. To distinguish these possibilities, we combined confocal imaging and dual-site focal synaptic stimulation of identified thin dendrites in rat neocortical pyramidal neurons. We found that nearby inputs on the same branch summed sigmoidally, whereas widely separated inputs or inputs to different branches summed linearly. This strong spatial compartmentalization effect is incompatible with a global summation rule and provides the first experimental support for a two-layer "neural network" [The quotes are left in to refer to a standard architecture in the artificial neural network field] model of pyramidal neuron thin-branch integration. Our findings could have important implications for the computing and memory-related functions of cortical tissue.

Basal dendrites are a major target for synaptic inputs innervating cortical pyramidal neurons. At present little is known about signal processing in these fine dendrites. Here we show that coactivation of clustered neighbouring basal inputs initiated local dendritic spikes, which resulted in a 5.9 +/- 1.5 mV (peak) and 64.4 +/- 19.8 ms (half-width) cable-filtered voltage change at the soma that amplified the somatic voltage response by 226 +/- 46%. These spikes were accompanied by large calcium transients restricted to the activated dendritic segment. In contrast to conventional sodium or calcium spikes, these spikes were mediated mostly by NMDA (N-methyl-D-aspartate) receptor channels, which contributed at least 80% of the total charge. The ionic mechanism of these NMDA spikes may allow 'dynamic spike-initiation zones', set by the spatial distribution of glutamate pre-bound to NMDA receptors, which in turn would depend on recent and ongoing activity in the cortical network. In addition, NMDA spikes may serve as a powerful mechanism for modification of the cortical network by inducing long-term strengthening of co-activated neighbouring inputs.