26 December 2003. As government officials scramble to contain the current scare over a cow that tested positive for BSE in Washington state earlier this week, researchers in New York and Boston report that experiments with slugs and yeast suggest prion-like behavior of certain proteins might actually be part of normal memory formation. BSE, Kuru, and Creutzfeldt-Jakob are all neurodegenerative diseases caused by an infectious conformation of a prion protein. The biological conclusion from decades of work on prions is that aberrantly folded versions of these proteins are bad—in fact, lethal. But it may not be so simple. Together, two papers in today's Cell suggest that prion-like conformational changes may be essential for long-term memory. The papers, in one fell swoop, radically change the field’s perspective on prion-like activity and suggest a whole new model to explain how our memories are stored.
In the first paper, Kausik Si and Eric Kandel at Columbia University, New York, together with Susan Lindquist at MIT, reveal that a member of the cytoplasmic polyadenylation element binding protein (CPEB) family—a protein quite separate from the prion protein itself—has an N-terminal domain with prion characteristics. These include a preponderance of polar amino acids, a low propensity for any particular secondary structure, and the potential to exist in both soluble and aggregated form.
Si found the CPEB isoform while studying sequences from the mollusk Aplysia californica. To test if the CPEB prion-like domain actually behaves as a prion, he fused it to a glucocorticoid receptor reporter and expressed the chimera in yeast, a model prion system pioneered by Lindquist (see ARF related news story). The chimera was designed to turn on the β-galactosidase gene, giving rise to blue yeast colonies when grown in the presence of the sugar analog X-Gal. For the most part, yeast expressing the chimera did turn blue, but occasionally they turned white, indicating the reporter system was malfunctioning in one in 100,000 cells. Examining them, Si found no mutations in either the chimera or the reporter gene. This suggested an epigenetic explanation for the expression failure, a result that smacked of prion activity.
To test if the chimera existed in different conformational states, as do prions, Si extracted proteins from blue and white yeast, and found that in white cells, the reporter existed mainly in an aggregated state. To test heritability, Si crossed blue and white cells and found that all offspring were white, suggesting that the aggregation-prone protein was capable of inactivating active protein.
The authors further found that full-length CPEB protein can exist in two distinct states, one of which can activate reporter genes through interaction with the yeast polyadenylation machinery. But, perhaps counterintuitively, when Si used a fluorescent CPEB-GFP construct to probe which protein was the active form, he found that it was the aggregated variant. In support of this, the authors found that aggregated CPEB binds to RNA polyadenylation elements with much higher affinity than the soluble protein.
In the second paper, Si, Kandel, and colleagues demonstrate that CPEB from the sea slug Aplysia is involved in long-term facilitation at synapses. Long-term facilitation, which results in enhanced neurotransmission at repeatedly stimulated synapses, is essential for the process of forming memories, but it is still not fully understood (see ARF related news story). It is known to require the synthesis of new proteins in the cell body, which are transported to and incorporated into synapses that have been repeatedly stimulated. Experiments suggest that the new proteins find the correct synapses because the latter have somehow been "marked." This marking has been shown to depend on synthesis of synaptic protein locally below the activated synapse, and it has been postulated to involve translation from RNAs that are dormant before the synaptic activity began. Enter CPEB, which activates RNA by promoting extension of polyadenosine tails (see Richter, 2001).
Si and colleagues found that a neuronal form of CPEB is expressed in Aplysia and also in Drosophila. To test if the protein is involved in facilitation, Si stimulated neurons with the neurotransmitter 5-HT, finding that this increased expression of CPEB. Significantly, protein synthesis inhibitors blocked this effect, whereas transcriptional inhibitors did not, suggesting that CPEB is itself regulated by translation. But is the protein necessary for facilitation?
To test this, Si used antisense oligonucleotides to inhibit synthesis of new CPEB protein. In neurons stimulated with five pulses of 5-HT, the amplitude of postsynaptic potentials increases after 24 hours by about 60-70 percent, and this persists for 72 hours. However, in cells treated with CPEB antisense nucleotides, the increase at 24 hours was much lower, only about 30 percent, and had returned to baseline by 72 hours. The data suggest that CPEB activity is indeed necessary for long-term facilitation and hence, memory.
Putting the two findings together, Si and colleagues suggest that long-term facilitation may involve a prion-like switch. The "mark" that is necessary for directing newly synthesized proteins may result from a conformational change in CPEB that makes it more active and induces local protein synthesis. In support of this hypothesis, it is worth noting that phosphorylation sites that are required for activation of many CPEB family members are conspicuously absent in the neuronal form of CPEB. But the beauty of the prion-like switch is that, once activated, it is "self-perpetuating and no longer requires for maintenance continued signaling either by kinases or phosphatases," write the authors. Moreover, they write, its activity state is much harder to reverse—an ideal property for something like long-term memory.
"If the general form of this model is indeed referable to neuronal synaptic biology, the results will prove fascinating to explore and nothing less than extraordinary," states Robert Darnell of Rockefeller University, New York, in an accompanying preview. For those interested in mammalian memories, it is worth noting that mouse isoforms of CPEB have been shown to be expressed in neurons.—Tom Fagan.
Si K, Lindquist S, Kandel ER. A neuronal isoform of the Aplysia CPEB has prion-like properties. Cell 2003 December 26;115:879-891. Abstract
Si K, Giustetto M, Etkin A, Hsu R, Janisiewicz AM, Miniaci MC, Kim J-H, Zhu H, Kandel ER. A neuronal isoform of CPEB regulates local protein synthesis and stabilizes synapse-specific long-term facilitation in Aplysia. Cell 2003 December 26;115:893-904. Abstract
Darnell RB. Memory, synaptic translation, and...prions? Cell 2003 December 26;115:767-768. Abstract