Associative learning rests on the strengthening of synaptic inputs that show coincident activity over extended periods of time. A notable exception is provided by supervised associative learning in the cerebellum. Parallel fiber (PF) - Purkinje cell synapses, whose activity predicts a climbing fiber (CF)-mediated error signal, undergo long-term depression (LTD). Since Purkinje cells are inhibitory neurons, classic Marr-Albus-Ito theories of cerebellar function state that LTD at glutamatergic PF inputs causes disinhibition of target cells in the cerebellar nuclei, thus enabling motor learning. However, more recent evidence challenges the notion of LTD as the only, or the predominant, cellular mechanism underlying associative motor learning. For example, findings from our laboratory show that in mice Purkinje cell excitability is enhanced after eyeblink conditioning (delay EBC), and that mice with a Purkinje cell-specific knockout of SK2-type K+ channels show reduced EBC. SK2 channels are small conductance, calcium-dependent K+ channels that are downregulated in a form of non-synaptic (?intrinsic?) plasticity, which enhances Purkinje cell excitability. Intrinsic plasticity is co-induced with long-term potentiation (LTP) at PF synapses. A scenario emerges, in which an intrinsic plasticity-assisted potentiation of those PF inputs that warn of an upcoming error signal (without contributing to it) enables EBC learning, possibly in parallel with depression at other PF synapses, whose activity continues to predict the error signal throughout learning. This scenario is in line with an adaptive filter model of cerebellar learning, in which bidirectional synaptic weight adjustment under supervision of a teacher signal is crucial for the fine-tuning of motor output. Here, we plan to use two-photon measurements of GCaMP6f-encoded, dendritic calcium signals in Purkinje cells of awake mice to test the hypothesis that during EBC the dendritic input map is restructuring. We predict that this map plasticity does not only consist of depression of response amplitudes at some PF synapses, but also the emergence of responses at other PF inputs, whose activity shifts from predicting the unconditioned stimulus (US; periorbital airpuff) to predicting the occurrence of the developing eyelid closure during EBC. We will examine how SK2-dependent intrinsic plasticity contributes to response strengthening, with a focus on possible roles of dendritic calcium spikes in synapse stabilization and clustering, motifs that have been identified as important cellular mechanisms in hippocampal place field formation. Using genetically modified mice with blockade of intrinsic plasticity (L7-SK2 knockout), LTP (L7-PP2B) and LTD (CaMKII T305D), respectively, we will further delineate the specific roles of these plasticity mechanisms in map re-organization and motor learning. Finally, using double-patch recordings from Purkinje cell dendrites and somata in vitro, we will examine the mechanisms of interaction between LTP and intrinsic plasticity that both seem to co-exist and complement each other in EBC. We will test the hypothesis that LTP stabilizes synaptic inputs, while intrinsic plasticity regulates synaptic penetrance, i.e. the predictive control of the EPSP amplitude over the spike output.
The cellular mechanisms underlying learning and memory have been intensely studied over decades, resulting in the widely accepted conclusion that experience-dependent changes in synaptic input weight (as in long-term potentiation; LTP) allow for the storage of information. Here, we hypothesize that this view is too narrowly focused on synaptic memory components, and that instead plasticity of membrane excitability (?intrinsic plasticity?) is required to promote LTP and synapse clustering (induction) as well as to permit potentiated synapses to trigger spike and spike burst firing (functional expression). We will test this hypothesis using two- photon recordings of dendritic calcium responses in cerebellar Purkinje cells in awake mice during a motor learning task, as well as by performing somato-dendritic double-patch recordings to examine the impact of synaptic weight and membrane excitability, respectively, on the generation of appropriate spike output patterns.
Showing the most recent 10 out of 16 publications
