Two-photon imaging of structural plasticity underlying classical eyeblink conditioning in mouse barrel cortex

Abstract

Sensory maps of the neocortex are constantly updated to adapt the individual to changes in the outside world that require the association of new sets of stimuli to adequate behavior. It is long known that such adaptation involves macroscopic changes of body representations in sensory maps. The emerging technology of two-photon microscopy together with the availability of transgenic mice that express fluorescent proteins in cortical neurons made it possible to monitor the postsynaptic cell compartments in vivo which are affected by experience dependent structural plasticity: dendritic spines. In the present work I combined classical trace eyeblink conditioning in awake head-fixed mice with two-photon imaging of dendritic spines. Classical conditioning that involves mnemonic processing, i.e. a ‘trace’ period between conditioned and unconditioned stimulus, requires awareness of the association to be formed, and is considered a simple model paradigm for declarative learning. The whisker representation of primary somatosensory cortex, named barrel cortex, is required for the acquisition of the tactile variant of trace eyeblink conditioning. To obtain insight into the cellular mechanisms underlying memory storage I monitored daily performance levels and plastic spine turn processes in test animals which underwent conditioning and in control animals which underwent pseudo conditioning. I showed that one cellular expression of barrel cortex plasticity during learning is substantial spine elimination on layer V neurons’ apical dendrites in layer I. The number of eliminated spines and their time of elimination were tightly related to the observed learning success. Pseudo conditioned animals on the other hand showed low baseline spine turnover rates. Moreover, I found that spine plasticity induced by learning was highly specific for the barrel column receiving signals from the stimulated vibrissa - spines located in an adjacent column were unaffected. The fact that layer I spines receive neuronal signals from associative thalamo-cortical and cortico-cortical circuits, together with the finding of column specific spine elimination observed in this study suggests that spine plasticity may arise via an interaction of ascending sensory (therefore spatially precise) and top-down associative signals.