Memory: From Sensory Circuits To Protein Conformations

Abstract

The ability to form, store, and retrieve memories is an essential capacity of many animals. Only in the past half-century, however, have the key neuronal and molecular events that underlie memory been studied. This has revealed a series of molecular cascades that are triggered by neuronal activation, ultimately leading to the stabilization of otherwise transient synaptic modifications. These synaptic modifications either increase or decrease the efficiency of synaptic transmission, thereby leading to altered neuronal communication. The behavioral events that must precede these molecular and neuronal changes, however, begin with the sensory system, where important information from the external world is identified. Generally, in situations involving associative learning, the experience of a reward or punishment is assumed to be remembered because of the relevance it holds for the organism at the moment of learning. Any pattern that deviates from this general idea provides an indication of exactly which features the processes of memory deem valuable. The deviation we explore here—that a more immediately appealing reward is not always better remembered, that D-arabinose is preferred to L-arabinose, but L-arabinose generates more reliable memories—suggests that there are important aspects beyond the momentary appeal of a reward. Further, L-arabinose, because it is remembered despite having no nutritional value, has allowed identification of a subset of 26 sensory neurons in Drosophila that, when activated, are sufficient to form long-term associative memories. In response to this sensory activation, the biochemical cascades of memory will, in further downstream neurons, trigger the oligomerization of the Drosophila cytoplasmic polyadenylation element binding (CPEB) protein, Orb2. Its oligomerization has so far been described as involving a prion-like conversion from a monomeric form to an amyloidogenic oligomer. Since its initial characterization, however, many types of functional protein oligomerization have been described. The question of whether Orb2’s oligomerization is in fact prion-like can best be addressed by substituting its prion domain with a variety of oligomerization domains, from other amyloid domains to more transient ‘liquid droplet’-type domains, or even standard tetramerization domains. Whether formation of less structured and less stable aggregates can still support the regulatory switch of Orb2 required for memory maintenance is the key question. Finally, Orb2 may not be the only functional prion-like protein in Drosophila, though it is the best characterized. Murashka, a RING-domain E3 ubiquitin ligase, is also involved in memory, and possesses both a disordered domain and features that are characteristic of prion-like proteins. If murashka does in fact undergo a prion-like conversion that is relevant for its role in memory, Orb2 will no longer be a curious outlier but instead the first illustration of what may be a widespread biological phenomenon.