The process of learning, and the associated neural plasticity that leads to memory formation, requires the ability to detect and rapidly respond to dynamic changes in the environment. At the level of individual neurons, these responses occur on a timescale that is faster than activity-induced transcription via the coordinated, activity-induced switching of internal molecular states and cellular metabolism. In recent years, our understanding of experience-dependent gene regulation and neuronal adaptation has advanced significantly with the recognition that the structure state of RNA can provide the modifiable context in which this can occur. Structurally labile RNA elements are able to react to changes in ion concentration and metabolite flux, which can lead to altered RBP and RNA-RNA interactions within the cell (
Mortimer et al., 2014). For example, a stem-loop structure in the 3′UTR of brain-derived neurotrophic factor (BDNF) mRNA is structurally responsive to calcium influx, thereby stabilizing this transcript in response to neuronal activity (
Fukuchi & Tsuda, 2010,
Vanevski & Xu, 2015). This structure state also promotes the interaction of BDNF mRNA with the RNA binding protein HuD, which has a direct impact on translation of BDNF (
Allen et al., 2013,
Vanevski & Xu, 2015). Further, an important role for the G-quadruplex RNA structures, which are non-canonical RNA structures organized in stacks of tetrads or G-quartets, in which four guanines are assembled in a planar arrangement by Hoogsteen hydrogen bonding. G-quadruplex RNA has been shown to be critically involved in mediating the localization of CamKIIα and PSD-95 to neurites, which are essential for synaptic plasticity (
Subramian et al, 2011).
Importantly, the dynamic switching of RNA structure states in response to changes in the cellular environment can be influenced by RNA modification. An interesting example of structural lability conferred by RNA modification is the brain-enriched lncRNA MALAT1, which influences synaptogenesis (
Bernard et al., 2010) and is found in nuclear paraspeckles within hippocampal neurons, implying a key role in alternative splicing. When MALAT1 accumulates m6A modifications, its interaction with the RBP heterogeneous nuclear ribonucleoprotein C (HNRNPC) is enhanced, which then promotes its accumulation in paraspeckles (
Liu et al., 2015,
Zhou et al., 2016). We found that a significant number of lncRNAs, including MALAT1, are dynamically expressed in the adult brain in response to fear-related learning (
Spadaro et al., 2015) and that the majority of these lncRNAs contain motifs for m6A and Ψ. The purpose of these modifications on neuronal lncRNAs remains to be determined; direct induction of an experience-dependent structure state change could promote the downstream influence of lncRNAs on RNA-directed epigenetic regulation in learning and memory, and this hypothesis warrants further investigation.
