|Home | About | Journals | Submit | Contact Us | Français|
Memories are remarkably persistent, but rely on transient signaling. The prion-like properties of CPEB suggested a solution to this problem. The paper by Krüttner et al demonstrates that the prion-like domain of Drosophila CPEB functions independently of its RNA binding domain for memory.
A fundamental property of the brain is that perceptual experiences drive modifications in number and strength of synaptic connections among neurons. These modifications of synaptic strength and connectivity are thought to be the neural correlates of cognition, which is constantly shaped by experience. Because synapse specificity is fundamental to neuronal plasticity, local protein synthesis at activated synapses plays a key role in establishing this spatial specificity. Although the mechanisms governing synapse specific protein translation are not fully understood, a “synaptic tagging” mechanism that restricts new protein synthesis to activated synapses has been proposed (Redondo and Morris, 2011).
Although the “synaptic tagging” model can explain the induction of protein synthesis to support synapse specific modifications, the remarkable persistence of memory presents a conceptual challenge because cellular signaling cascades typically provide only transient activation. How then can spatially distinct and mRNA specific translational activation be maintained? A potential mechanism was suggested with the discovery of prion-like properties of RNA binding protein CPEB (cytoplasmic polyadenylation element binding protein), one of the key molecules involved in activity dependent local translation (Si et al., 2003a; Si et al., 2003b).
Prions are proteins that exhibit two remarkable properties (Shorter and Lindquist, 2005). First is the ability to adopt alternate physical conformations that have distinct functional impacts. Second is that at least one of the isoforms has the ability to promote conformational changes in trans. This provides a self-perpetuating property to the functional impact of the conformational switch. Prions were first characterized in the context of transmissible neurodegenerative disorders, but prion like proteins have also been discovered in non-pathological contexts, where it is thought that prion-like protein conformation is altered by cellular signaling events. For example in yeast, several well characterized prion-like proteins have been described that undergo a self-perpetuating switch from soluble to aggregating oligomers in response to stressful conditions (Shorter and Lindquist, 2005). The idea that this mechanism might be at play in long-term synaptic plasticity came from the seminal discovery that CPEB has prion-like properties (Si et al., 2010; Si et al., 2003a; Si et al., 2003b).
CPEB was originally described in the context of early metazoan development, during which the translation of maternally deposited mRNAs are also spatiotemporally regulated. In Xenopus oocytes, CPEB was shown to activate the translation of dormant mRNAs by binding to the cytoplasmic polyadenylation elements (CPEs) within the 3′UTR of specific mRNAs. It was subsequently discovered that CPEB was also present in the dendritic layer of the hippocampus, at synapses in cultured hippocampal neurons, and in the postsynaptic density of biochemically fractionated synapses. Indeed, local protein synthesis underlying long-term synaptic plasticity was shown to rely on CPEB. A number of activity dependent and synaptically localized mRNAs regulated by CPEB have been identified, including CaMKII, and a number of other signaling molecules and translational regulatory factors have also been linked to CPEB, including 4E-BPs (eIF4E-binding proteins) (Darnell and Richter, 2012).
The surprising hypothesis that CPEB-mediated translational regulation at the synapse reflected a prion-like self-perpetuating mechanism came from the discovery that Aplysia CPEB possess a glutamine-rich domain and when fused to a yeast reporter gene, this glutamine-rich domain supports the prion-like switch (Si et al., 2003b). CPEB in Aplysia also was demonstrated to play a critical role in branch-specific local translation underlying long-term facilitation (LTF), a cellular correlate of memory. In this classic paradigm, the synapse between a specific sensory neuron branch and motor neuron is stimulated with spaced pulses of serotonin resulting in long-lasting plasticity. However, the LTF was abolished when antisense oligonucleotides were delivered to the specific synapse to knockout local CPEB. Importantly, the Aplysia LTF model also revealed that CPEB expression is itself up-regulated locally at activated synapses (Si et al., 2003a). An SDS-resistant CPEB oligomer can be immunopurified from Aplysia neurons. The formation of such oligomer was enhanced by serotonin treatment and promoted LTF, suggesting that the activity dependent induction of CPEB plays a role in plasticity. These findings raised the provocative hypothesis that neural activity induces CPEB to undergo a self-sustaining conformational change that then helps to maintain a translationally active state for some mRNAs at the synapse. The roles of the RNA-binding and prion-like functions of CPEB were not easily deciphered, in part because of the potential contributions of the various known isoforms of CPEB. Studies in flies, including the new one from Krüttner et al., leverage the advantages of the fly model system for precise genetic manipulations. The findings nicely complement the results from Aplysia and mammalian systems, where cell biology was more tractable.
The fruit fly made a debut in the CPEB literature with the discovery that Orb2, the Drosophila ortholog of CPEB2, plays a role in courtship memory (Keleman et al., 2007). Flies offer the combination of a tremendous genetic toolbox and a rich array of well-studied memory paradigms including visual memory, both appetitive and aversive forms olfactory memory, and memory of courtship experience. In each case, many of the genetic pathways and neural circuits have been elucidated, which provides a considerable leg up for mechanistic investigations. Many of the key regulators of local translation in Aplysia and mammals are conserved and at play in the fly brain (Barbee et al., 2006; Dubnau et al., 2003). In the courtship learning paradigm (Keleman et al., 2007; Keleman et al., 2012), male flies can learn to discriminate between virgin and mated females if their courtship attempts have previously been rejected by a mated female. Such courtship memory can be short-term or long-lasting, depending on the training protocol used. Keleman et al. 2007 first linked the Drosophila CPEB protein Orb2 with this particular long-term courtship memory paradigm. Drosophila Orb2, together with vertebrate CPEB2-4 belongs to the CPEB2 subfamily, while Drosophila Orb, Xenopus CPEB, vertebrate CPEB1 and Aplysia CPEB belongs to the CPEB1 subfamily. However, Drosophila Orb2, similar to Aplysia CPEB, does contain an N-terminal glutamine-rich prion-like domain. The two major protein isoforms (Orb2A and Orb2B) produced from the orb2 locus share not only this glutamine-rich domain, but also a C-terminal RNA binding domain. Null mutant of orb2 or mutants in which the glutamine-rich domain is deleted from both isoforms each exhibit defective courtship suppression memory 24 hours after training, though they exhibit memory 30 minutes after training. Keleman et al. 2007 also showed that Orb2 is required for long-term memory only shortly after training and mapped the requirement of Orb2 to a specific subset of neurons (γ neurons) in the Drosophila mushroom bodies (MB), a known site of associative learning for several different tasks in fruit flies (Keleman et al., 2007; Qin et al., 2012).
These studies establish the fly courtship assay as a platform for investigation of the mechanism by which CPEB function impacts memory. Because the orb2 gene produces multiple isoforms and contains several functional domains, including the prion-like interaction domain, traditional gain-of-function or loss-of-function studies are not sufficient to dissect the role of RNA-binding versus prion-like action for long-term memory. In this issue of Neuron, Krüttner et al. 2012 made elegant use of the fly genetics toolset to generate isoform-specific manipulations of RNA-binding and prion-like domains within the context of the endogenous locus. They created a deletion of the endogenous orb2 locus and replaced it with engineered variants that give expression of only Orb2A (orb2 B) or Orb2B (orb2 A). In each case, they tagged the protein that was expressed with GFP as a reporter and tested the engineered allele for rescue of the behavioral phenotype. They further engineered isoform-specific deletions of the glutamine-rich domain (to make orb2 Q B and orb2 A Q) or replaced the glutamine-rich domain with similar domains from orthologous CPEBs. Similar modifications also were systematically generated for the RNA binding domain (RBD) to make orb2RBD* B and orb2RBD* A (RBD* denotes the mutated RBD), replace the RBD with other RBDs from orthologous CPEBs, and swap the RBDs of the two isoforms. Because all of the above modifications were placed back into the original genomic context, proper expression levels and distribution were ensured. Together, these reagents permit the independent manipulation of orb2A and orb2B as if they are independent loci and provide the means to test the roles of each of the two major domains within each of the two isoforms.
Using the above genetic “parts list”, it was possible to create conditions where each of the two orb2A alleles provided (1) no Orb2A, (2) Orb2A with RBD mutation, (3) Orb2A with glutamine-rich domain deletion or (4) intact Orb2A. By mixing and matching combinations of each of these engineered alleles and/or a wild allele for both orb2A and orb2B, every possible combination could be created. The replacement constructs with corresponding domains from homologous CPEBs further expanded the possible combinations. With this beautiful genetic resource, Krüttner et al. examined the long-term memory phenotype with approximately thirty relevant viable allele combinations. They were able to demonstrate that the glutamine-rich domain in Orb2A is both required and sufficient for long-term memory while the RBD is essential for function of Orb2B but not of Orb2A. The former conclusion is consistent with recent evidence published earlier this year (Majumdar et al., 2012) that a single point mutation in the unique N-terminal extension region of Orb2A impairs long-term memory retention for both courtship memory and appetitive olfactory memory. The essential role of the glutamine-rich domain from Orb2A leads both groups to propose that Orb2A aggregates upon neuronal activity. Indeed, Orb2 oligomer were immunoprecipitated from brain extracts when MB, dopaminergic or serotonergic neurons were acutely activated by a temperature-sensitive dTrpA1 channel that depolarizes neurons. Orb2 oligomer formation was also induced upon tyramine, dopamine, octopamine and serotonin stimulation.
The Orb2 oligomer is resistant to many treatments including RNase, high salt, detergents, denaturants and even boiling and its formation is independent of phosphorylation, N-glycosylation, ubiquitination, sumoylation or acetylation. Importantly, Orb2 oligomer formation requires the expression of the Orb2A isoform with an intact glutamine-rich domain despite the fact that this isoform seems to express at a much lower levels than Orb2B. Orb2 oligomer formation is abolished when Orb2A glutamine-rich domain is deleted or replaced with the point mutation isoform (Orb2AF5Y), which aggregates with much lower affinity (Krüttner et al., 2012; Majumdar et al., 2012). In contrast, glutamine-rich domain of Orb2B is not required for oligomer formation. Instead, the Orb2B isoform appears to require RNA binding function. One of the most striking observations is that an inter-allelic combination that can express orb2RBD* B (Orb2A with glutamine-rich domain and mutated RBD) and orb2 A Q (Orb2B with no glutamine-rich domain but intact RBD) produces perfectly normal long-term courtship memory (Krüttner et al., 2012).
Together, Krüttner et al., 2012 and Majumdar et al 2012 provide a compelling dissection of the roles of Orb2A and Orb2B in memory formation. The studies support a model in which the glutamine-rich domain of Orb2A, though expressed at low levels in neurons, is the primary effector in initiating oligomer formation in response to neural activity and that this feature is important for local translation under the control of Orb2B via its RNA binding domain. This model also suggests the possibility that this conformational change is self-perpetuating, thereby sustaining the translational activation state of specific targets at the synapse.
What more can the fly system add to this question? Three areas come to mind. First, it will be important to determine the relevant mRNA targets whose activity-dependent regulation persists over time. The binding specificity offered by the RBD domain of Orb2B, which is shown by Krüttner et al., 2012 to be indispensable for courtship memory formation, may provide an entry point to this question. Second, investigation of the neural circuits in which such activity dependent translation underlies memory will also be important. Such studies will provide a genes-to-circuit-to-behavior integration, and also a place in the brain to look for behaviorally relevant regulatory effects. Although the initial acquisition of courtship memory, like olfactory memory, appears to occur in MB, through the activation of Dopamine receptors in the MB γ neurons (Keleman et al., 2012; Qin et al., 2012), the site of de novo gene expression underlying olfactory memory has recently been localized outside of MB (Chen et al., 2012). With courtship memory, GAL4-mediated over-expression of either Orb2A or Orb2B in MB neurons is sufficient to rescue the memory defect in orb2 mutants that lack the glutamine-rich domain (Keleman et al., 2007). Therefore, to formally demonstrate that Orb2A-mediated oligomer formation and subsequent CPEB-dependent local translational regulation are induced selectively in MB γ neurons, it will be important to rescue the mutant alleles with Orb2A glutamine-rich domain and Orb2B RNA binding domain each restricted to γ neurons. Finally, the mechanistic details of local translation will likely involve other regulatory molecules, some of which have already been implicated in memory and plasticity in Drosophila (Barbee et al., 2006; Dubnau et al., 2003). A protein of particular interest is Pumilio, another RNA binding protein whose function is required for long-term olfactory aversive memory (Dubnau et al., 2003) and which also contains an aggregation-prone prion-like domain (Salazar et al., 2010). An understanding of the function of prion-like proteins in normal neuronal physiology will provide context to decipher the impact of pathological effects of aggregation prone prion-like proteins in neurodegenerative disorders.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.