Aplysia CPEB Can Form Prion-like
Multimers in Sensory Neurons that
Contribute to Long-Term Facilitation
Kausik Si,1,2,* Yun-Beom Choi,5,6Erica White-Grindley,1Amitabha Majumdar,1and Eric R. Kandel3,4,5
1Stowers Institute for Medical Research, 1000 East 50th Street, Kansas City, MO 64110, USA
2Department of Molecular & Integrative Physiology, University of Kansas Medical Center, 3901 Rainbow Boulevard, Kansas City,
KS 66160, USA
3Howard Hughes Medical Institute
4Kavli Institute for Brain Science
5Department of Neuroscience
6Department of Neurology
College of Physicians and Surgeons of Columbia University, New York State Psychiatric Institute, 722 West 168thStreet, New York,
NY 10032, USA
Prions are proteins that can assume at least two
distinct conformational states, one of which is domi-
nant and self-perpetuating. Previously we found that
a translation regulator CPEB from Aplysia, ApCPEB,
that stabilizes activity-dependent changes in syn-
aptic efficacy can display prion-like properties in
yeast. Here we find that, when exogenously ex-
pressed in sensory neurons, ApCPEB can form an
amyloidogenic self-sustaining multimer, consistent
with it being a prion-like protein. In addition, we find
that conversion of both the exogenous and the
endogenous ApCPEB to the multimeric state is
enhanced by the neurotransmitter serotonin and
that an antibody that recognizes preferentially the
multimeric ApCPEB blocks persistence of synaptic
facilitation. These results are consistent with the
idea that ApCPEB can act as a self-sustaining
prion-like protein in the nervous system and thereby
synaptic efficacy to persist for long periods of time.
Activity-dependent changes in synaptic efficacy and synapse
sadio et al., 1999; Frey and Morris, 1997; Kang and Schuman,
1996; Martin et al., 1997; Steward and Schuman, 2001). In the
sensory to motor neuron synapses of the gill-withdrawal reflex
in Aplysia, local protein synthesis is required for at least two
distinct functions: (1) the synthesis of one or more retrograde
signals that travel from the synapse to the cell body to activate
transcription and (2) the synthesis of a local synaptic mark that
stabilizes subsequent functional and structural changes at the
activated synapse (Casadio et al., 1999; Martin et al., 1997; Mini-
aci et al., 2008). Interestingly, the protein synthesis inhibitor
rapamycin, which blocks translation of a subset of synaptic
mRNAs related to growth, does not interfere with the retrograde
signal but only blocks the stabilizing component of the synaptic
mark (Casadio et al., 1999).
In a search for the rapamycin-sensitive stabilizing component
of the synaptic mark in Aplysia, we previously identified a novel,
binding protein (CPEB), ApCPEB (Si et al., 2003a). ApCPEB
belongs to a family of RNA-binding proteins that can act as
both a translational repressor and activator of its target mRNAs
(Barkoff et al., 2000; de Moor and Richter, 1999; Hake and
that the persistence of synaptic facilitation requires ApCPEB-
mediated synaptic protein synthesis not transiently, but continu-
ously for at least 72 hr (Miniaci et al., 2008). As the activity of
ApCPEB is altered by synaptic stimulation and should have
a limited duration, this persistence of activity raised the following
question: How can the altered activity state of ApCPEB be main-
tained for long periods of time?
One plausible solution to the problem of how a short-lived
molecule can produce enduring changes came from our earlier
studies in yeast that revealed ApCPEB to have prion-like proper-
ties (Si et al., 2003b). Prion proteins are unique because the
same protein can assume two distinct conformational states,
one of which is dominant and self-perpetuating (Coustou et al.,
1997; Derkatch et al., 2001; Osherovich and Weissman, 2001;
Prusiner, 1998; Sondheimer and Lindquist, 2000; Wickner,
1994; Wickner et al., 1995). We previously found that in yeast,
like other prion proteins, ApCPEB could exist in two distinct
conformational states, one of which is multimeric, active, and
dominantly self-perpetuating (Si et al., 2003b). Based on the
prion-like properties of ApCPEB in yeast, we proposed that
ApCPEB can assume at least two conformational states: mono-
Cell 140, 421–435, February 5, 2010 ª2010 Elsevier Inc. 421
and either inactive or acts as a repressor. Synaptic activation
leads to the conversion of ApCPEB to a dominant, self-
sustaining, active multimeric state. This creates a self-sustaining
synaptic mark limited to the specific set of synapses that have
been activated, resulting in a sustained period of translation at
the activated synapse that allows for the maintenance of the
synaptic changes. A test of this hypothesis requires an experi-
mental analysis that addresses four questions: First, does
ApCPEB have the properties of a self-perpetuating prion-like
protein in neurons? Second, is the prion-like conversion from
a monomeric to a multimeric form modulated by synaptic stimu-
lation? Third, is ApCPEB active in the multimeric, prion-like
state? Fourth, is the prion-like state of ApCPEB required for
the maintenance of long-term facilitation?
In this paper, we find that in the sensory neurons of Aplysia
both the exogenously expressed and the endogenous Aplysia
CPEB exist in two states, one of which is multimeric. The exog-
enous multimers in the sensory neuron result from a homotypic
interaction of ApCPEB. The exogenous multimers are amyloido-
genic and self-sustaining. These features are consistent with
ApCPEB having prion-like properties in the sensory neuron. In
addition we find that the multimerization of both endogenous
and exogenous ApCPEB is enhanced by repeated pulses of
5-HT, the stimulus that induces sustained change in synaptic
strength. Finally, we find that ApCPEB is active in its multimeric
state. An antibody that preferentially recognizes the multimeric
state selectively inhibits the persistence of long-term facilitation
of sensory-motor neuron synapse. These results are consistent
with the notion that ApCPEB serves a physiological function in
its self-sustaining prion-like multimeric state.
Aplysia CPEB Has Both Low and Stable High Molecular
Weight States in the Sensory Neuron
Prion-like proteins have two physically distinct states in the cell,
one of which is multimeric. The conversion from the monomeric
to the multimeric state is enhanced by increases in protein
concentration. Earlier we overexpressed ApCPEB in yeast and
found that the overexpressed protein formed puncta that had
similar properties to those of yeast prions (Si et al., 2003b). To
determine whether ApCPEB behaves similarly in neurons as it
does in yeast, we overexpressed EGFP-tagged ApCPEB in the
sensory neurons of Aplysia (Figure 1A) and noticed punctate
green fluorescence (>3 pixels) in all expressing neurons
(Figure 1C). The punctate appearance was not due to the
EGFP tag because ApCPEB fused to the fluorescent protein
KiKGR (Figure S1B available online and Figure 5A) or to HA
peptide (Figures 1D and 1D0) formed similar puncta (Tsutsui
et al., 2005). Efficient puncta formation requires the N-terminal
prion-like domain. Similar to what we have observed in yeast
previously, deletion of N-terminal 252 amino acids considerably
reduced the punctate appearance of ApCPEB (number of
puncta/neurite; full-length ApCPEB; 50.37 ± 5.68, N-terminal
deletion; 5.83 ± 1.19, n = 8, p < 0.001, one-way ANOVA)
(Figure 1E, Figure S2, and Table S1) (Si et al., 2003b). Thus,
when overexpressed in sensory neurons, ApCPEB can form
punctate structures consistent with a multimeric state, and the
punctate appearance of ApCPEB requires the prion-like domain.
However, we cannot exclude the possibility that the reduced
puncta formation in N-terminal-deleted ApCPEB in sensory
neurons was due to a reduction in the level of the protein.
As we observed the punctate ApCPEB upon overexpression
in sensory neurons, it raises the question, does endogenous
Aplysia CPEB exist in a similar multimeric state or is multimeriza-
tion an artifact of overexpression? To address this question, we
took advantage of an antibody, Ab464, that we had raised
againstthe aggregated recombinant N-terminal 213aminoacids
and found that it specifically stains the ApCPEBEGFP puncta in
the sensoryneuron (Figures 1Fand 1G). Wenext immunoprecip-
itated endogenous ApCPEB from central nervous system (CNS)
protein extracts (?0.5 mg of protein) with Ab464 and probed the
immunoprecipitate with two other antibodies, one raised against
the soluble N-terminal 213 amino acids (Ab463) and the other
against a 17 amino acids peptide (Ab77) from the C-terminal
end of the protein (Figure 1H). With both antibodies we detected
only a high molecular weight ApCPEB (>181 kDa) in the Ab464
immunoprecipitate but not in control IgG immunoprecipitate
(Figure 1H). The high molecular weight form of ApCPEB is
most likely not due to association with other proteins or mRNA
because it is stable after boiling 5 min in 10% SDS, which should
dissociate most interacting molecules. Moreover, western anal-
any monomeric ApCPEB, whereas both Ab463 and Ab77 recog-
nized the monomeric form (Figure 1I and Figure S1C).
These results are consistent with the idea that endogenous
ApCPEB can exist both as a monomer and as a stable multimer,
which migrates as a high molecular weight protein. Since the
composed of ApCPEB the high molecular weight protein should
contain at least three or more ApCPEB molecules. Our attempts
to sequence the high molecular weight protein by mass spec-
trometry have so far been unsuccessful. Therefore, we cannot
rule out the possibility that the high molecular weight protein
contains other proteins in addition to ApCPEB or that some
stable posttranslational modification of ApCPEB shifts its
Aplysia CPEB Puncta Are Distinct from PolyQ
The Aplysia CPEB N-terminal domain is rich in Q residues and
polyQ rich proteins readily form aggregates in neurons. We
therefore asked, do ApCPEB form puncta simply because they
have a certain number of Q residues or does the prion-like
domain contain additional structural information that allows the
protein to form puncta in neurons? To address this question
we fused either 72Q residues or the Q/N-rich prion domain of
protein in sensory neurons (Figures 2B and 2C). We chose to
attach 72Q residues because the N-terminal 160 amino acids
to produce multimers in yeast cells (Krobitsch and Lindquist,
rescence (number of puncta/neurite; 72Q, 0.5 ± 0.5, n = 7;
Sup35NM domain, 0.5 ± 0.32, n = 8) in the sensory neuron neu-
422 Cell 140, 421–435, February 5, 2010 ª2010 Elsevier Inc.
ofQ/Nresidue number (Figures2Band 2Cand FigureS2). These
results led us to wonder whether the prion domain of ApCPEB
(ApPRD), which can form puncta in yeast albeit less efficiently,
would be able to form puncta in neurons. To our surprise we
failed to detect any puncta in sensory neurons expressing the
N-terminal domain of ApCPEB fused to EGFP (n = 6)
(Figure 2D and Figure S2).
These results have led us to the following two conclusions.
First, the N-terminal prion-like domain of ApCPEB functions
differently than simple polyQ repeats. In addition to the Q
residues, the multimerization of ApCPEB in neuron requires
additional structural components. Second, unlike in yeast, the
ApCPEB prion-like domain functions most efficiently in neurons
only in the context of the full-length protein.
Puncta Formation Is a Unique Property of Aplysia CPEB
Is multimerization a unique property of neuronal CPEB variants,
or can any CPEB form multimers if overexpressed in Aplysia
sensory neurons? Mouse CPEB1 has significant sequence
homology with Aplysia CPEB except for the N-terminal prion-
like domain, which is unique for Aplysia CPEB and its homologs
in the other species. So we expressed EGFP-tagged mouse
CPEB1 (mCPEB1) in sensory neurons and observed only diffuse
fluorescence but no detectable puncta (Figure 2E and
Figure S2). To address whether the lack of multimerization is
due to lack of a prion-like domain we fused the N-terminal 252
amino acids of ApCPEB to mouse CPEB1 (ApPRDmCPEB1).
We now observed that the fusion protein formed punctate
structures in the sensory neuron (number of puncta/neurite;
Figure 1. Both Exogenously Expressed and
Endogenous ApCPEB Form Multimers in
ApCPEB tagged with either fluorescence proteins
or HA peptide expressed from a constitutively
active promoter containing the retroviral RSV
promoter and eight binding sites for the transcrip-
tion activator AP1. The constructs were injected
into the sensory neuron cell body in the sensory
(SN)–motor (MN) neuron coculture. After 2 days
the cells were either imaged directly or stained
(B)EGFP alone produced adiffusedfluorescence.
The fluorescence puncta visible in the cell body of
the sensory neuron is autofluorescence of endog-
enous pigment granules. Because of these
pigment granules we have excluded the cell
body region from our analysis and all quantifica-
tion was done from the neurite region.
(C) ApCPEB fused to EGFP formed punctate
(D) ApCPEB tagged to HA, stained with mouse
anti-HA antibodies, and visualized with Cy3-
coupled anti-mouse IgG antibody. (D0) The same
cell in (D) is shown in a higher magnification
(scan zoom 2.4). The punctate ApCPEB staining
is clearly visible.
(E) ApCPEB lacking N-terminal 252 amino acids
in (D0) where the scale bar is 10 mm. All numbers
are mean ± standard error of the mean (SEM).
(F–I) Endogenous ApCPEB form stable multimers.
(F) ApCPEB multimers in sensory neuron were
specifically recognized by Ab464 raised against
the aggregated N-terminal domain of ApCPEB.
showed distinct punctate staining. (G) The stain-
ing was specific because it was absent in rabbit
IgG control. Scale bar, 20 mm. (H) Immunoprecip-
itation(IP) of Aplysia totalCNS extract withcontrol
rabbit IgG, Ab533, which recognizes the ApCPEB
monomer in western analysis but not in immuno-
precipitation, and the multimer-specific Ab464.
The SDS-resistant ApCPEB multimer is indicated with an asterisk (*). (I) Western blotting of total CNS extract of Aplysia with affinity-purified antibodies Ab77,
Ab463, Ab464, and Ab533. Except Ab464 all the other antibodies recognized the monomeric ApCPEB. The monomeric ApCPEB is indicated with an arrowhead.
The blots were overexposed to ensure detection of all immunoreactive polypeptides.
See also Figures S1 and S2 and Table S1.
Cell 140, 421–435, February 5, 2010 ª2010 Elsevier Inc. 423
63.75 ± 13.32, n = 8, p < 0.001, one-way ANOVA) (Figure 2F and
The ability of mouse CPEB1 to form puncta in the sensory
neurons is specific to the N-terminal domain of Aplysia CPEB.
The addition of 72Q residues in front of the mouse CPEB1
(72QmCPEB1) failed to induce any puncta (number of puncta/
neurite; 0.33 ± 0.21, n = 6) (Figure 2G and Figure S2). The punc-
tate appearance was also not due to mouse CPEB1 forming
a large number of RNA-protein particles because CPEB1 was
able to form puncta when its RNA-binding domain was mutated
(ApPRDmCPEB1rbm) (number of puncta/neurite; 90.5 ± 15.95,
n = 6, p < 0.001, one-way ANOVA) (Figure 2H and Figure S2).
We next asked, can the ApCPEB prion-like domain induce
puncta formation of any RNA-binding protein or is it restricted
Figure 2. Multimerization Is a Unique Prop-
erty of ApCPEB and Distinct from PolyQ
The green fluorescent protein EGFP fused to full-
length Aplysia CPEB (A), to 72 glutamine residue
(72Q) (B), to yeast prion Sup35 N-terminal NM
prion domain (Sup35NM) (C), or to ApCPEB
N-terminal 252 amino acids (ApPRD) (D). All
constructs exhibited diffuse staining, except
full-length Aplysia CPEB with its characteristic
(E) Mouse CPEB1 (mCPEB1) fused to EGFP
produced diffuse fluorescence when expressed
in Aplysia sensory cell.
(F) Fusion of ApCPEB N-terminal 252 amino acids
(ApPRD) to the N-terminal end of mouse CPEB1
resulted in a punctate appearance.
(G) 72Q residues fused to the N-terminal end of
mouse CPEB1 were unable to produce punctate
(H) RNA-binding ability is not necessary for the
punctate appearance of mouse CPEB1 fused to
(I) ApCPEB N-terminal 252 amino acids fused to
Aplysia RNA-binding protein Staufen does not
result in similar punctate structures.
All numbers are mean ± SEM. Scale bar, 20 mm.
See also Figure S2 and Table S1.
to the CPEB family? To address this
question we fused the ApCPEB N-
terminal domain to the RNA-binding
protein staufen from Aplysia (Figure 2I).
produced a few puncta and only in some
sensory neurons (number of puncta/neu-
rite; 1.33 ± 1.3, n = 6) (Figure S2). These
mRNA-protein particles as has been
staufen protein (Kohrmann et al., 1999).
These observations with staufen rein-
forced the earlier conclusion that CPEB
proteins have a latent capacity to multi-
merize, and the addition of the prion-like domain accelerates
Aplysia CPEB Puncta Are due to Self-Assembly
of the Protein
The punctate pattern seen with ApCPEB is unlikely to be RNP
particles or P body structures involved in the degradation or
storage of mRNA for the following reasons. First, the ability to
form either RNP particles or P bodies requires RNA-binding
activity (Huang et al., 2003; Wilczynska et al., 2005). We find
that ApCPEB still formed puncta in sensory neurons even
when it lacked mRNA-binding capacity (number of puncta/neu-
rite; 140.83 ± 35.95) (Figure S3B). Second, the ApCPEB puncta
did not colocalize with P body marker Dcp1 (Figure S3C)
424 Cell 140, 421–435, February 5, 2010 ª2010 Elsevier Inc.
Without Thioflavin S
ApCPEB fibers after 12 hours
ApCPEB fibers after 2 hours
Figure S4. ApCPEB Can Form Amyloids, Related to Figure 4
(A) Time course of recombinant ApCPEB multimerization visualized with Cy3-labeled protein. The large aggregates formed 2 hr after initiation of refolding are
most likelymisfolded protein aggregates. Forfluorescent fiber formationApCPEBwaslabeled withCy3using aCy3-mono maleimide labeling kit (GE Healthcare)
as per manufacturer’s instructions. The labeled protein was mixed with 419 picomoles of the unlabeled protein and dialyzed against the fiber formation buffer
containing 1M Urea, 100 mM KCl, 10 mM Na-HEPES (pH7.6), 1 mM DTT, 0.1 mM CaCl2, 1 mM MgCl2, and 5% Glycerol at room temperature. Dialysis was per-
formed in Slide-A-Lyzer mini Dialysis units of 7000 MWCO (Pierce). Five microliter aliquots were taken at the indicated times and plated on polylysine-coated
glass bottom Mattek dishes. The dishes were washed after 30 min with the fiber formation buffer and the fluorescent fibers were imaged using confocal micros-
copy. Scale bar, 50 mm.
(B) ApCPEB fibrils visualized by EM at the indicated time after initiation of refolding. Left panel: In addition to the fibers misfolded protein aggregates, indicated
with a white arrow, are visible at 2 hr. Scale bar, 200 nm. Right panel: Electron micrograph of purified ApCPEB 12 hr after initiation of refolding. Large fibers are
visible. Scale bar, 1 mm.
(C and D) Controls for intracellular thioflavin S staining. (C) Aplysia sensory neuron expressing HA-tagged ApCPEB incubated with buffer only. (D) Uninjected
sensory neuron stained with thiofalvin S. The punctate thioflavin S staining was visible only in ApCPEB-expressing neuron. Left panel imaged at 534 nm. Right
panel imaged for thioflavin S at 485 nm. Scale bar, 20 mm.
S4 Cell 140, 421–435, February 5, 2010 ª2010 Elsevier Inc.
% Change in number
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
10-40% Sucrose Gradient
Intensity plot of RAHA+5-HT Intensity plot of RBHA+5-HT
Figure S5. Stimulation with Serotonin Induces Multimerization of Only Drosophila Orb2RA but Not of Orb2RB, Related to Figure 6
Drosophila Orb2RA or RB cDNA tagged with HA were injected in Aplysia sensory neurons in sensory-motor neuron coculture and either mock treated or treated
with 5 pulses of 10 mM 5-HT.
(A) Untreated Orb2RAHA-expressing cells did not show any puncta that were >3 pixels.
(B) Orb2RAHA-expressing cells treated with 535-HT formed visible puncta.
(C and D) In contrast Drosophila Orb2RBHA did not form any detectable puncta in untreated (C) or in 5-HT-treated cells (D).
(B0and D0) A Higher-magnification image of indicated regions in (B) and (D), respectively. The puncta are clearly visible only in RAHA, but not in RBHA.
(E and F) The intensity plot of single 1 mm confocal section. The change in fluorescence intensity in case of RA indicates clustering of RAHA molecule, whereas in
RBHA there is no fluctuation in the intensity plot, suggesting that most of the RBHA is either in a diffused monomeric or in a small multimeric state.
other pictures, 20 mm.
(H) The majority of endogenous ApCPEBexists as amonomeric protein.Approximately 1.5 mg of totalprotein extracts from AplysiaCNS was analyzed in a10%–
ular weight of80 kDa. Weobservedafractionof ApCPEBmigrating similarin sizeof adimeric form of theprotein. However, it could also beApCPEBinacomplex
with other proteins. When we blotted with Ab464 we did not see any immunoreactive band.
Data are represented as mean ± SEM.
Cell 140, 421–435, February 5, 2010 ª2010 Elsevier Inc. S5
Aplysia CNS extract
Figure S6. When Overexpressed ApCPEBEGFP Multimers Can Form outside the Synaptic Area, Related to Figure 7
immunoreactive band is indicated.
(B and C) Immunostaining of sensnory-motor neuron culture injected with empty vector (B) or with indicated split-fluorescent protein constructs (C) with anti-
(D) Higher-magnification images of the boxed region in (C). Scale Bar: 20 mm in (C) and 10 mm in (D).
S6 Cell 140, 421–435, February 5, 2010 ª2010 Elsevier Inc.