MOLECULAR AND CELLULAR BIOLOGY, Dec. 2010, p. 5658–5671
Copyright © 2010, American Society for Microbiology. All Rights Reserved.
Vol. 30, No. 24
CPEB4 Is a Cell Survival Protein Retained in the Nucleus upon
Ischemia or Endoplasmic Reticulum Calcium Depletion?†
Ming-Chung Kan,1‡ Aparna Oruganty-Das,1‡ Amalene Cooper-Morgan,1Guang Jin,2
Sharon A. Swanger,3Gary J. Bassell,3Harvey Florman,4Klaus van Leyen,2
and Joel D. Richter1*
Program in Molecular Medicine1and Department of Cell Biology,4University of Massachusetts Medical School, Worcester,
Massachusetts 01605; Neuroprotection Research Laboratory, Massachusetts General Hospital, Charlestown, Massachusetts 021292;
and Departments of Cell Biology and Neurology, Emory University School of Medicine, Atlanta, Georgia 303223
Received 21 June 2010/Returned for modification 6 August 2010/Accepted 1 October 2010
The RNA binding protein CPEB (cytoplasmic polyadenylation element binding) regulates cytoplasmic polyade-
nylation and translation in germ cells and the brain. In neurons, CPEB is detected at postsynaptic sites, as well as
in the cell body. The related CPEB3 protein also regulates translation in neurons, albeit probably not through
polyadenylation; it, as well as CPEB4, is present in dendrites and the cell body. Here, we show that treatment of
neurons with ionotropic glutamate receptor agonists causes CPEB4 to accumulate in the nucleus. All CPEB
proteins are nucleus-cytoplasm shuttling proteins that are retained in the nucleus in response to calcium-mediated
signaling and alpha-calcium/calmodulin-dependent kinase protein II (CaMKII) activity. CPEB2, -3, and -4 have
conserved nuclear export signals that are not present in CPEB. CPEB4 is necessary for cell survival and becomes
analysis indicates that nuclear accumulation of CPEB4 is controlled by the depletion of calcium from the ER,
specifically, through the inositol-1,4,5-triphosphate (IP3) receptor, indicating a communication between these
organelles in redistributing proteins between subcellular compartments.
The cytoplasmic polyadenylation element binding (CPEB)
protein, a sequence-specific RNA binding protein, is found in
the cell body and at synapses of hippocampal and other neu-
rons; in response to synaptic experience, CPEB promotes poly-
adenylation and translation (37, 49, 50). CPEB knockout mice
(44) have deficiencies in synaptic plasticity, particularly, theta
burst-induced long-term potentiation (LTP) (1, 51), as well as
in particular forms of hippocampal-dependent memories (3).
Some of these CPEB-related functions may be related at least
in part to its ability to direct CPE-containing RNA transport in
dendrites (16), in addition to its regulation of translation (51).
In neurons, most of the CPEB-related proteins CPEB3 and
CPEB4 are found in the cell soma. However, a relatively small
amount of these proteins is localized to synaptic regions and
cofractionates with postsynaptic density (PSD) proteins, suggest-
ing possible roles in RNA translation and/or localization (17).
While investigating the possible translocation of CPEB3 and -4 to
dendritic spines in response to N-methyl-D-aspartate receptor
(NMDAR) activation, as is the case with the fragile X mental
retardation protein (FMRP) (11), we noticed a surprising relo-
calization of CPEB4 from the cell soma to the nucleus. While the
treatment of neurons with ionotropic glutamate receptor-activat-
ing agents, such as glutamate and ?-amino-3-hydroxy-5-methyl-
4-isoxazolepropionic acid (AMPA), as well as NMDA, induced
not. CPEB4 remained cytoplasmic when calcium/calmodulin-de-
pendent protein kinase II (CaMKII) activity was inhibited, sug-
gesting a link between calcium levels and nuclear import and/or
retention. When fused to epitope tags and expressed in neurons,
CPEB, CPEB3, and CPEB4 also became concentrated in the
nucleus in response to NMDAR activation. All three proteins,
however, were also nuclear when neurons were treated with lep-
tomycin B (LMB), an agent that inactivates the nuclear export
factor Crm1, indicating that CPEB family proteins are continu-
ously shuttling between nucleus and cytoplasm. Moreover, these
data suggest that NMDAR activation does not necessarily induce
nuclear translocation but, rather, induces nuclear retention of the
One physiological event that causes widespread glutamate
overload is ischemia, which occurs when the brain’s supply of
glucose and oxygen is disrupted by stroke or cardiac arrest
(35). ATP production is reduced when the blood supply is
insufficient, causing neuron polarization and the accumulation
of glutamate in the extracellular space due to reversed uptake
(40). This excess glutamate is responsible for neuron death in
transient ischemia (5), and its stimulation of massive calcium
influx through the NMDA receptors plays a major role in
excitotoxicity (45, 46). A mouse model for transient focal is-
chemia caused nuclear accumulation of CPEB4, as did an in
vitro model for oxygen and glucose deprivation. Lentivirus-
mediated small hairpin RNA (shRNA) knockdown of CPEB4
demonstrates that this protein is essential for neuron survival
and might suggest that its nuclear accumulation is a physiolog-
ical response to high levels of intracellular calcium. However,
additional experiments with cultured neurons show that the
localization of CPEB4, 3,5-dihydroxyphenylglycine
* Corresponding author. Mailing address: Program in Molecular
Medicine, 373 Plantation St., Suite 204, University of Massachusetts
Medical School, Worcester, MA 01605-2377. Phone: (508) 856-8615.
Fax: (508) 856-4289. E-mail: firstname.lastname@example.org.
‡ These authors were equal contributors.
† Supplemental material for this article may be found at http://mcb
?Published ahead of print on 11 October 2010.
depletion of calcium from the endoplasmic reticulum (ER)
rather than elevated levels of cytosolic calcium per se is respon-
sible for CPEB4 nuclear accumulation. We have also demon-
strated the involvement of IP3 receptors in this ER calcium
depletion-mediated nuclear accumulation of CPEB4. We pro-
pose that the dispatch of a signal that is dependent upon the
expulsion of ER calcium signals CPEB4 to accumulate in the
nucleus, where it may offer some protection against cell death.
MATERIALS AND METHODS
Hippocampal neuron culture. The culture of primary rat hippocampal neurons
was performed as described previously (2), with a typical plating density of 1.8 ?
104cells/cm2cultured in Neurobasal medium (Invitrogen) containing B27 sup-
plement (B27 medium) and glutamine (1 ?g/ml). Cytosine arabinoside (Ara-C)
(1 ?M) was added at day 3 after plating in vitro (DIV3) to prevent glial cell
Lentiviral vector construction and virus production. Lentiviruses expressing
CPEB3 and CPEB4 were constructed by inserting myc-CPEB3 and myc-CPEB4
into the BamHI and XhoI sites of pFugw vector (Addgene). For virus produc-
tion, 10 ?g of virus transfer vector that expressed various CPEBs, 7.5 ?g of
gag-pol-expressing vector psPAX2 (from Addgene), and 5 ?g of vesicular sto-
matitis virus G (VSV-G)-expressing vector pMD2.G (from Addgene) were co-
transfected into 1 ? 107293T cells plated in 10-cm culture dishes using Lipo-
fectamine 2000 (Invitrogen). Three hours after transfection, the medium was
replaced with Neurobasal medium containing B27 supplement (B27 medium).
Forty-eight hours after transfection, the medium was collected and passed
through a 0.45-mm filter; the virus titer in the filtrate was calculated by serial
dilution to determine the minimum amount of virus that could infect 90% of
neurons plated at 1.8 ? 104cells/cm as assayed by immunocytochemistry for the
myc-tagged fusion proteins.
CPEB4 knockdown. Lentiviruses expressing shRNAs against CPEB4 (pLL3.7-
syn-KD2 and pLL3.7-syn-KD3) followed the procedure reported in reference 17.
The primers for constructing pLL3.7-syn-KD2 were C4-KD2-F (TGGCTGCAGC
TTTTC) and C4-KD2-R (TCGAGAAAAAAGGCTGCAGCATGGAGAGATA
GATTCTCTTGAAATCTATCTCTCCATGCTGCAGCCA). The primers for
constructing pLL3.7-syn-KD3 were C4-KD3-F (TGGCTGCCTCATTTGGCGAA
TAATTTCAAGAGAATTATTCGCCAAATGAGGCAGCCTTTTTTC) and C4-
AATTATTCGCCAAATGAGGCAGCCA). Lentivirus expressing shRNA was
produced by transfecting 293T cells (1 ? 106cells/ml) with transfer vector together
with packaging vectors pSPAX2 and pMD2.G using Lipofectamine 2000. Three
hours after transfection, the cells were washed and then cultured for 48 h after
transfection, when the culture medium was collected and filtered through a 0.2-mm
syringe filter. The virus-containing medium was used directly without concentration;
the titer was determined as the minimal amount needed to infect ?90% of cultured
Image acquisition and processing. A Nikon Eclipse E600 microscope was used
to take fluorescence images; confocal images were acquired using a spinning-disk
confocal microscope (CSU10B; Solamere Technology Group) controlled by Meta-
morph software. Most of the images were taken using a Plan Fluor objective lens
with ?20 magnification and a numerical aperture (NA) of 0.5. The cooled charge-
coupled device (CCD) RT (real time) color camera is made by Diagnostic Instru-
ments, Inc. Images were taken and further processed using SPOT version 3.5.8
software. Where indicated, fluorescence intensity was quantified using Image J
Antibodies and immunohistochemistry. CPEB4 antibody was produced as
described previously (17); hemagglutinin (HA) (16B12) and myc (9E10) mono-
clonal antibodies were produced as ascites fluid (Covance); and C/EBP homol-
ogy protein (CHOP) antibody was purchased from Santa Cruz Biotechnology. A
terminal deoxynucleotidyltransferase-mediated dUTP-biotin nick end labeling
(TUNEL) assay kit was purchased from MBL International. For CPEB4 immu-
nostaining, cells were fixed in 2% paraformaldehyde–phosphate-buffered saline
(PBS)–4% sucrose for 20 min and then blocked in 10% bovine serum albumin
(BSA) for 20 min before overnight incubation with affinity-purified CPEB4
antibody at 4°C. Secondary antibody (Alexa 595-conjugated goat anti-rabbit and
Alexa 488-conjugated goat anti-mouse antibodies) application and washing were
performed as recommended by the manufacturer (Molecular Probes). For some
experiments, neurons treated with tetrodotoxin (TTX) or TTX plus NMDA were
collected and probed for CPEB4 and actin on Western blots.
MCAO and OGD. Middle cerebral artery occlusion (MCAO) was performed
as described previously (47), except that MCAO was extended to 90 min and
reperfusion to 24 h before sacrifice. For oxygen and glucose deprivation (OGD),
minimal essential medium (MEM) is purged with nitrogen for 20 min to remove
oxygen from solution. The gas-purged medium was placed in an anaerobic
chamber for 30 min to equilibrate pH. DIV14 hippocampal neurons were then
cultured in oxygen-purged, glucose-deficient MEM in an anaerobic chamber with
an air mixture of 0.5% oxygen, 10% carbon dioxide, and 89.5% nitrogen for 1 h.
The cells were then moved to normal Neurobasal medium with B27 and cultured
under standard conditions for various times before fixation.
In vitro nuclear import assay. HeLa cells grown on Lab-Tek chamber slides
were permeabilized with digitonin (40 mg/ml) in TB buffer [20 mM HEPES, pH
7.4, 110 mM potassium acetate (KOAc), 2 mM Mg(OAc)2, 2 mM dithiothreitol,
1 mM EGTA, and protease inhibitor] for 5 min on ice. The cells were then
washed twice with TB buffer plus BSA (10 mg/ml). After the second wash, the
import reaction mixture was added (2 ?l 100 mg/ml BSA, 8 ml HeLa cytosol with
ATP regeneration system, 2 ml glutathione S-transferase [GST]-CPEB4 RNA
binding domain [RBD], and 8 ml TB buffer). The ATP regeneration system
contained 1 mM ATP, 5 mM phosphocreatine, and 20 units/ml creatine phos-
phokinase. Permeabilized HeLa cells were incubated in nuclear import reaction
mixture for 20 min at 25°C; the cells were then washed with cold TB buffer before
fixation with 4% formaldehyde–PBS for 10 min. The fixed cells were stained with
anti-GST antibody to detect nuclear import substrate and counterstained with
Pharmacological treatment of primary neuron culture. Freshly prepared glu-
tamate (100 ?M), NMDA (100 ?M), AMPA (300 ?M), DHPG (100 ?M), or
ionomycin (5 ?M) was applied to DIV16 hippocampal neurons for 1 h before
fixation and immunostaining. In some experiments, APV (2-amino-5-phospho-
novaleric acid, 20 ?M), Ant-AIP-II (10 ?M; Calbiochem), or EGTA (2 mM) was
added to the cells 20 min before NMDA. BAPTA/AM [1,2-bis(o-aminophe-
noxy)ethane-N,N,N?,N?-tetraacetic acid tetra(acetoxymethyl) ester, 50 ?M; Cal-
biochem] was added to cultures for 20 min and then replaced with culture
medium for 40 min. Thapsigargin (TG, 2 mM stock; Calbiochem) and tunica-
mycin (TM, 5 mg/ml of 1,000? stock; Calbiochem) were suspended in dimethyl
sulfoxide (DMSO) and added to cultures. 2-APB (2-aminoethoxydiphenyl bo-
rate, 20 ?M) was added to neurons 45 min before NMDA.
Sucrose gradients. Brain tissue from 2-month-old mice was washed once in 1?
PBS, homogenized in 2 ml of 0.8 M sucrose, and centrifuged at 10,000 rpm for
10 min. An amount of 1.5 ml of the supernatant was subsequently removed and
layered on 2.0 M sucrose in an SW41 centrifuge tube. Amounts of 2.25 ml of 1.3
M, 1.95 M, and 2.5 M sucrose were layered on top of the brain lysates, which was
followed by centrifugation at 40,000 ? g for 5 h. An 18-gauge needle was used to
pierce the bottom of the tube, and 1-ml fractions were collected; 200 ?l of each
fraction was removed to a clean microcentrifuge, 2.5 volumes of 100% ethanol
were added, and the tubes were stored overnight at ?20°C. The precipitates were
collected by centrifugation and with 70% ethanol. The pellets were suspended in
100 ?l of 1? SDS sample buffer and analyzed by Western blotting for CPEB4
(antibody dilution of 1:1,000) or protein disulfide isomerase (PDI, antibody
dilution of 1:750; Santa Cruz).
Immunoelectron microscopy. CPEB4-enriched fractions were pooled and di-
luted 5-fold in homogenization buffer (20 mM Tris-HCl, pH 7.5, 2 mM MgCl2,
50 mM KCl, 250 mM sucrose, 1 mM phenylmethylsulfonyl fluoride). Pooled
fractions were then centrifuged for 1 h at 36,000 rpm and the pellet washed once
with 1? PBS. The pellets were then incubated with CPEB4 antibody (1:500) for
1 h at room temperature. The pellets were then washed twice with 1? PBS and
blocked for 15 min with 5% BSA–PBS. Subsequently, the pellets were incubated
with 10 nm goat anti-rabbit gold-conjugated antibody (1:25; Ted Pella, Inc.) for
1 h at room temperature. The pellets were then washed three times for 5 min
each time with 1? PBS and fixed with 2.5% glutaraldehyde in 0.1 M phosphate
buffer, pH 8.0.
Measurement of intracellular Ca2?. Neurons were plated at a density of 0.5 ?
105cells/well in a 96-well dish and maintained in Neurobasal medium containing
B27. DIV16 hippocampal neurons were loaded with 1 ?M Fluo-3/AM, a Ca2?-
sensitive fluorescent probe (Invitrogen) containing 0.02% Pluronic F-127 (Mo-
lecular Probes) in Hanks’ buffered salt solution for 30 min at 37°C. The plates
were gently washed with buffer and incubated for another 30 min at 37°C to
permit de-esterification of intracellular Fluo-3/AM. The fluorescence of Fluo-3 is
a measure of the Ca2?concentration and was determined using a POLARstar
Optima fluorescence plate reader (BMG Labtech) equipped with an excitation
filter set to 485 ? 10 nm and an emission filter set to 520 ? 10 nm. After dye
loading, neurons were treated with various reagents (see Fig. 9 for a complete
list), and the relative fluorescence intensity was determined.
VOL. 30, 2010 NUCLEAR RETENTION OF CPEB45659
Measurement of ER calcium. ER calcium measurements were carried out using
the ER-targeted cameleon construct D1ER (30). Primary hippocampal neurons
(DIV12) were transfected with 1.6 mg of pD1ER using Lipofectamine 2000 and
maintained in Neurobasal medium. Four days after transfection, the appropriate
pharmacological agent (see Fig. 9) was added to the neurons and the fluorescence
resonance energy transfer (FRET) from the donor fluorophore cyan fluorescent
protein (CFP) (441/485 nm excitation/emission) to the acceptor fluorophore yellow
fluorescent protein (YFP) (441/550 nm excitation/emission) was measured. The
magnitude of the FRET signal is proportional to the amount of calcium in the ER.
Cells were imaged on an Olympus IX 70 inverted light microscope. To calibrate the
relative FRET signals, Rminwas obtained by treating the cells with 3 mM EGTA and
2 ?M ionomycin and the FRET signal from untreated cells was set as Rmax. Fluo-
rescence images were background corrected. The emission ratio (FRET/CFP) was
quantified before and after treatment of the neurons, and the percent calcium
Fig. 9 for a schematic of the FRET assay.
Plasmid construction. Plasmid construction is detailed in the supplemental
NMDA induces nuclear localization of CPEB4. In cultured
hippocampal neurons, CPEB4 is detected mainly in the cyto-
plasm and in dendrites; it is also enriched in postsynaptic
density (PSD) fractions from adult rat brain and hippocampal
neurons. To assess whether CPEB4 changes location in re-
sponse to synaptic activity, neurons were treated with 0.1 mM
NMDA for 40 min in the presence of tetrodotoxin (TTX).
Compared to the results for a control, NMDA treatment un-
expectedly caused strong and permanent nuclear CPEB4 im-
munostaining, which was prevented when APV, an NMDA
antagonist, was added to the cells (Fig. 1A and B; also see Fig.
S1 in the supplemental material for specificity of CPEB4 an-
tibody). Although TTX was used here to silence spontaneous
neural activity, we have observed similar nuclear localization of
CPEB4 even in the absence of TTX (data not shown). The
numbers of neurons in which CPEB4 was predominantly nu-
clear or cytoplasmic or was distributed in both compartments
(see Fig. S2 in the supplemental material for representative
images) as a function of activity show that NMDA caused a
dramatic nuclear localization of CPEB4 that was nearly com-
pletely prevented by APV (Fig. 1A and B, right, comparison of
FIG. 1. Stimulation of NMDA receptor causes CPEB4 nuclear localization. (A) DIV16 hippocampal neurons incubated in TTX for 24 h were
treated with buffer alone (TTX) or NMDA (TTX NMDA) and then fixed and immunostained with affinity-purified CPEB4 antibody. DAPI shows
nuclear DNA staining. The cells were examined by confocal microscopy. To quantify the localization of CPEB4 in the nucleus or cytoplasm, other
cells were stained for CPEB4 where the protein was predominantly cytoplasmic (cyto), nuclear (nuc), or evenly distributed in cytoplasm and
nucleus (cyto/nuc) (see Fig. S2 in the supplemental material for representative images). Using this protein distribution as a standard, the
percentage of cells in which CPEB4 was in these compartments following TTX or NMDA treatment was determined (histogram). The asterisks
refer to a statistically significant difference (P ? 0.01, Student’s t test) between the results for neuron samples treated with TTX versus those treated
with TTX plus NMDA. (B) DIV16 hippocampal neurons incubated in TTX for 24 h were treated with APV (TTX NMDA APV) for 5 min prior
to application of NMDA for 40 min and then fixed and immunostained with affinity-purified CPEB4 antibody. DAPI shows nuclear DNA staining.
The cells were examined by fluorescence microscopy. The asterisks refer to statistical significance (P ? 0.01, Student’s t test). Size bar ? 10 ?m.
5660KAN ET AL.MOL. CELL. BIOL.
TTX versus TTX-plus-NMDA treatment, P ? 0.01, Student’s
t test). Also, upon treatment with lower concentrations of
NMDA (0.01 mM and 0.001 mM), fewer neurons exhibited
nuclear accumulation of CPEB4 (data not shown), indicating a
dose-dependent response of CPEB4 localization to this neu-
NMDA also caused a substantial loss of CPEB4 from den-
drites. While NMDA did not significantly alter the overall
amount of CPEB4 in neurons, it did lead to an increase in the
fluorescence intensity of CPEB4 in the nucleus relative to that
in the cell body without stimulation (see Fig. S3 in the supple-
mental material). These data suggest that dendritic (and cell
body) CPEB4 was not destroyed in an activity-dependent man-
ner but instead was probably transported retrogradely to the
Ligand binding to ionotropic glutamate receptors causes
calcium influx and induces downstream signaling events
through CaMKII (18). To examine whether extracellular cal-
cium is important for nuclear accumulation of CPEB4, EGTA
was applied to neurons before NMDA; this caused CPEB4 to
remain predominantly cytoplasmic (Fig. 2A). To assess
whether NMDA/calcium-induced CPEB4 nuclear accumula-
tion occurs via CaMKII, a membrane-permeable CaMKII in-
hibitory peptide, AIP-II (19, 48), was applied to neurons 20
min before NMDA treatment. This peptide reduced NMDA-
induced CPEB4 nuclear translocation (Fig. 2B), indicating that
extracellular calcium and CaMKII are part of an NMDA-
induced signaling pathway that causes CPEB4 nuclear accu-
mulation. In addition, treatment of neurons with glutamate,
AMPA, or NMDA induced CPEB4 nuclear staining; DHPG, a
metabotropic glutamate receptor agonist, had no effect, how-
ever (Fig. 2B), nor did acetylcholine (calcium opening neuro-
transmitter) or ?-aminobutyrate (GABA) (chloride channel
neurotransmitter) (data not shown). These data, together with
the determination of the number of cells that respond to these
treatments (Fig. 2C), show that ionotropic glutamate receptor
activation induces CPEB4 nuclear accumulation.
CPEB family proteins are nucleus-cytoplasm shuttling pro-
teins. Nuclear accumulation of CPEB4 could be due to active
cytoplasm-to-nucleus import or inhibited nuclear export if this
protein continuously shuttles between the two compartments.
To distinguish between these possibilities, neurons were
treated with leptomycin B (LMB), an inhibitor of the nuclear
export factor Crm1 (21, 29, 33). The application of LMB to
cultured neurons resulted in nuclear accumulation of CPEB4,
suggesting that CPEB4 is a nucleus-cytoplasm shuttling protein
To investigate whether other CPEB family proteins are nu-
cleus-cytoplasm shuttling proteins and accumulate in the nu-
cleus in response to NMDA, DIV14 hippocampal neurons
were infected with lentiviruses expressing epitope-tagged
CPEB1, CPEB3, and CPEB4, which were then subjected to
either LMB or NMDA treatment. These proteins all accumu-
lated in the nucleus when nuclear export was blocked by LMB,
suggesting that they are all shuttling proteins (Fig. 3B). NMDA
treatment also caused these proteins to accumulate in nuclei
(Fig. 3B; quantification of responsive cells is presented in
Fig. 3C), supporting the notion that nuclear accumulation is
a common feature among CPEB proteins in neurons follow-
ing synaptic stimulation. To determine whether nuclear ac-
cumulation of CPEB4 is neuron specific, we treated HeLa
cells and 293T cells with LMB; in both cases, CPEB4 was
retained in the nucleus (unpublished data). We also treated
HeLa cells with digitonin, which causes cytoplasmic leakage
through pores in the plasma membrane. When a fusion of
the CPEB4 RNA binding domain (RBD, i.e., both RNA
recognition motifs [RRMs] and both zinc fingers) to GST
FIG. 2. Effectors of CPEB4 nuclear localization. (A) DIV16 neu-
rons were incubated with EGTA for 20 min prior to application of
NMDA for 40 min; the cells were then fixed and stained for CPEB4.
(B) TTX-treated DIV16 hippocampal neurons were treated with
DHPG, AMPA, glutamate, or NMDA for 40 min before being immu-
nostained with CPEB4 antibody. Other cells were treated with AIP-II
for 20 min and then subjected to NMDA for 40 min before fixation.
(C) Quantification of CPEB4 in the cytoplasm, nucleus, or both nu-
cleus and cytoplasm was determined as described in the legend to Fig.
1. The asterisks refer to statistically significant differences (P ? 0.01,
Student’s t test) between the results for the indicated samples.
VOL. 30, 2010NUCLEAR RETENTION OF CPEB4 5661
was added to the treated cells after they were washed to
remove the leaked cytoplasm, there was no evidence of
nuclear accumulation. However, when the permeabilized
cells were supplemented with fresh HeLa cytosol, nuclear
GST-CPEB4 was readily apparent (Fig. 3D). These data
show that CPEB4 nuclear transport is not neuron specific
and that the RBD contains the nuclear localization signal
(NLS) of CPEB4.
Identification of CPEB4 nuclear import/export cis elements.
To identify the CPEB4 NLS and nuclear export signal (NES),
plasmids encoding serial deletions of CPEB4 (mutants D1 to
D7) were generated (Fig. 4A) and transfected into NIH 3T3
cells, which because of their flattened morphology, are partic-
ularly amenable for use in immunocytochemistry to distinguish
between cytoplasmic and nuclear proteins. A CPEB4 trunca-
tion mutant that lacks the NLS would be expected to be con-
stitutively cytoplasmic irrespective of LMB treatment; protein
with an NES deletion should reside in the nucleus irrespective
of LMB treatment.
Although the entire CPEB4 protein was sequentially deleted
(Fig. 4A), none of the mutants remained cytoplasmic when
cells were treated with LMB. Thus, CPEB4 probably has two
or more independently acting NLSs. However, CPEB4 trunca-
tion mutant D4, lacking residues 351 to 463, was nuclear in the
absence or presence of LMB, indicating that this truncated
region contains the NES (Fig. 4B; quantification of responsive
cells is presented in Fig. 4C). The expression levels of the
CPEB4 mutant proteins are shown on the lower right in
Using the Multalin program (6) that compares protein se-
quence similarity, a sequence from the peptide deleted in mu-
tant D4 was aligned with corresponding regions from CPEB2,
-3, and -4 from rat, human, Xenopus laevis, zebrafish, and
Drosophila melanogaster (in this case, a single CPEB4-like pro-
tein called Orb2). As shown in Fig. 5A, CPEB4 residues 383 to
397 within the deleted peptide are highly conserved among all
the protein sequences examined. Leucine residues are often
found in NESs; mutation of two of them at positions 385 and
390 (arrows and denoted in boldface) to alanine (CPEB4 LL/
AA) caused the accumulation of CPEB4 in the nucleus of
transfected 3T3 cells irrespective of whether they were treated
with LMB (Fig. 5A, bottom left; quantification of cells is at the
bottom right). Thus, these leucine residues are essential for
CPEB4 nuclear export. To assess sufficiency of nuclear export,
various segments of CPEB4 containing the NES were fused to
bacteriophage MS2 expressing enhanced green fluorescent
protein (EGFP-MS2), which also contained a simian virus 40
(SV40) T NLS (Fig. 5B, N-EGFP). As expected, the control
N-EGFP lacking a CPEB4 NES was nuclear in transfected 3T3
cells (Fig. 5B, middle, left). Upon fusion with the CPEB4
NES-containing fragment, three of the fusion proteins,
N-EGFP-268, N-EGFP-145, and N-EGFP-83, became local-
ized to the cytoplasm in the absence but to the nucleus in the
presence of LMB. However, two proteins lacking CPEB4
residues 341 to 376 (N-EGFP-110 and N-EGFP-48) were
evenly distributed in both the nuclear and cytoplasmic com-
partment, suggesting a lack of NES function (Fig. 5B, mid-
dle; quantification of cells is at the bottom). From these
FIG. 3. CPEB family proteins are nucleus-cytoplasm shuttling proteins. (A) DIV16 hippocampal neurons were treated with 0.1% ethanol alone
or with 10 nM LMB for 40 min before fixation and then immunostained for endogenous CPEB4. (B) DIV16 neurons infected with lentivirus
expressing HA-CPEB1, myc-CPEB3, or myc-CPEB4 for 2 days were treated with 10 nM LMB for 1 h and then immunostained with HA or myc
antibodies. Other infected cells were stimulated with NMDA for 1 h prior to immunostaining. (C) Quantification of CPEB1, CPEB3, or CPEB4
in the nucleus or cytoplasm following the treatments described for panel B. The asterisks refer to statistically significant differences (P ? 0.01,
Student’s t test) between the results for the indicated samples. C4RBD, CPEB4 RNA binding domain. (D) HeLa cells were treated with digitonin,
and the permeabilized cells were incubated with an ATP-regenerating system, purified recombinant GST-CPEB4 RBD fusion protein, with or
without HeLa cell cytosol. After 40 min, the cells were fixed and stained with GST antibody. Size bar ? 10 ?m.
5662KAN ET AL.MOL. CELL. BIOL.
results, we conclude that CPEB4 residues 341 to 424 con-
stitute a minimal NES.
Brain ischemia causes nuclear accumulation of CPEB4.
Glutamate plays dual roles in the brain: at physiological levels,
it induces excitatory synapse activation and plasticity, but un-
der pathological conditions, such as ischemia and epilepsy,
elevated levels of extracellular glutamate cause neuron degen-
eration through excitotoxicity. To determine whether CPEB4
FIG. 4. Identification of the CPEB4 shuttling elements. (A) The CPEB4 internal deletion constructs are depicted. The boxes indicate the parts
of the protein that were deleted, and the gray boxes indicate the known functional domains (RNA recognition motifs RRM1 and RRM2 and two
zinc fingers [ZF]) of CPEB4. The bar in the N terminus is a myc epitope tag. (B) NIH 3T3 cells were transfected with plasmid DNA encoding the
proteins shown in panel A; 12 h later, the cells were treated with LMB for 1 h prior to fixation. Antibody against the myc epitope was used for
immunostaining. Exogenous proteins and ?-tubulin were monitored by immunoblotting. Size bar ? 10 ?m. (C) Quantification of the CPEB4
proteins in the nucleus and cytoplasm as described in the legend to Fig. 1. The asterisks refer to statistically significant differences (P ? 0.01,
Student’s t test) between the results for the indicated samples.
VOL. 30, 2010NUCLEAR RETENTION OF CPEB4 5663
FIG. 5. Identification of CPEB4 nuclear export signal. (A) Alignment of the human CPEB4 protein sequence from residues 383 to 397 with
homologous regions from rat and zebrafish CPEB4 and human and Xenopus CPEB2 and CPEB3. Two arrows point to conserved leucine residues that
have been mutated to alanine in the LL-AA mutant. NIH 3T3 cells were transfected with DNA encoding myc-tagged wild-type or LL-AA mutant CPEB4
proteins for 12 h, treated with LMB for 1 h and then fixed and immunostained for the myc-tagged protein. Quantification of the CPEB4 proteins in the
nucleus and cytoplasm are as described in the legend to Fig. 1. The asterisks refer to statistically significant differences (P ? 0.01, Student’s t test) between
the results for the indicated samples. (B) Fusion proteins used to identify the minimal CPEB4 NES domain. Various regions of the CPEB4 coding region
(the numbers refer to amino acid residues) that contain the NES were fused to the SV40 T NLS, the bacteriophage MS2 coat protein, and EGFP. The
dark bar indicates the leucine residues indicated in panel A (top). Middle, NIH 3T3 cells were transfected with DNA encoding the fusion proteins noted
above. The image in the left panel shows that the NLS-EGFP-MS2 protein, without any CPEB4 sequence, was nuclear. The other panels show the
nuclear-cytoplasmic distribution of the fusion proteins in neurons, some of which were treated with LMB. Size bar ? 10 ?m. Bottom, quantification of
the GFP-CPEB4 fusion proteins in the nucleus or cytoplasm as described in the legend to Fig. 1. The asterisks refer to statistically significant differences
(P ? 0.01, Student’s t test) between the results for the indicated samples.
nuclear accumulation is induced by glutamate release in vivo,
brain sections from rats subjected to high-frequency stimulation
(HFS) for 90 min were probed with CPEB4 antibody. While this
HFS paradigm induced the expression of the immediate early
gene that encodes activity-regulated cytoskeleton-association pro-
tein (Arc) in granular cells of the dentate gyrus (41), it did not
cause CPEB4 nuclear accumulation (unpublished data).
To investigate whether pathological levels of glutamate or
the stress associated with it might cause CPEB4 nuclear accu-
mulation, we turned to a mouse model for transient focal
ischemia. In this paradigm, the middle cerebral artery is oc-
cluded (MCAO) for 90 min, which causes a focal deprivation
of blood flow, followed by a reperfusion of blood for 24 h
before sacrifice and histological preparation (47). A clear in-
farction was evident in the ipsilateral portion of the brain,
which not only caused neuron death (as determined by
TUNEL staining) but also dramatically reduced CPEB4 stain-
ing (Fig. 6A). However, in the motor cortex and insular cortex,
which are in the penumbra of the severely affected area,
CPEB4 staining was enriched in the nucleus. In contrast,
CPEB4 staining was cytoplasmic in the corresponding con-
tralateral regions (Fig. 6B).
Ischemia causes not only hypoxia but also hypoglycemia in
the affected part of the brain. One cell culture model that
mimics these two ischemia-induced deficits is oxygen-glucose
deprivation (OGD); here, neurons are cultured under condi-
tions of reduced atmospheric oxygen and glucose. Hippocam-
pal neurons (DIV14) were subjected to OGD treatment for 1 h
and then returned to normal culture conditions for 3 h before
fixation and staining for CPEB4 and apoptosis. In control cells
grown under normal conditions, CPEB4 was cytoplasmic and
TUNEL staining was absent (Fig. 7A). After 1 h of OGD
treatment and 3 h of recovery, CPEB4 protein was undetect-
able in most neurons, but in others, it was concentrated in the
nucleus. Interestingly, CPEB4 nuclear staining was inversely
correlated with TUNEL staining (Fig. 7A, right), suggesting
that nuclear CPEB4 may offer some protection against apop-
tosis. The results of the above-mentioned MCAO and OGD
experiments suggest that excessive glutamate causes CPEB4
nuclear accumulation. However, this nuclear accumulation
could be due to stress caused by the excessive glutamate.
To investigate whether CPEB4 is important for neuron sur-
vival, cultured hippocampal neurons were infected with lenti-
viruses expressing GFP, as well as two different shRNAs for
CPEB4. The results in Fig. 7B show that, while the KD2
shRNA (see “CPEB4 knockdown” in Materials and Methods)
effectively reduced CPEB4 levels, KD3 did not. Neither
shRNA affected CPEB3 or tubulin levels. Neurons were also
doubly infected with viruses expressing KD2 shRNA and
CPEB4 or CPEB4 ?NES, which lacks the nuclear export signal
(Fig. 7C). CPEB4 and CPEB4 ?NES, which resides solely in
the nucleus, were mutated so as to maintain the proper amino
acid sequence but not to anneal with the shRNA. Most neu-
rons died when CPEB4 was reduced (KD2 shRNA), but neu-
rons survived up to 3 days when CPEB4 levels were restored
with either full-length CPEB4 or CPEB4 ?NES (Fig. 7C).
Thus, nuclear CPEB4 is necessary for neuron survival.
CPEB4 is present on the ER, and its nuclear localization is
induced by ER calcium depletion. Transient ischemia induces
protein aggregation in the ER, possibly due to inhibited fold-
ing capacity when luminal calcium levels are reduced (15). This
possibility suggests that CPEB4 nuclear accumulation might
also be mediated by reduced ER calcium; consequently, a
membrane-permeable calcium chelator, BAPTA/AM, was
used to immobilize free calcium inside the ER. Twenty min-
utes after the addition of BAPTA/AM, neurons were placed in
fresh medium for an additional 40 min. After the membrane-
permeable moiety of BAPTA/AM is cleaved by cytosolic es-
terases, the remaining BAPTA becomes trapped intracellu-
larly. Because the dissociation constant of BAPTA for calcium
is close to the cytosolic calcium level, BAPTA does not sub-
stantially affect cytosolic calcium levels (31). On the other
hand, BAPTA targets free calcium in the ER because of its
FIG. 6. Ischemia causes CPEB4 protein to become concentrated in
the nucleus. (A) A frozen section of brain taken from a mouse that had
a middle cerebral artery occlusion (MCAO) performed was fixed and
stained for CPEB4. A consecutive section from the same animal was
labeled by TUNEL staining. The white boxes refer to regions of the
motor cortex (MC) and insular cortex (IC) that were examined under
higher magnification, as shown in panel B. Size bar ? 1 mm. (B) A
section from the ischemic brain was immunostained with anti-CPEB4
antibody. DAPI staining shows nuclei. The images were taken from the
motor cortex (MC) or insular cortex (IC); ipsilateral (Ipsi) and con-
tralateral sides (Con) of these regions are shown. Size bar ? 20 ?m.
VOL. 30, 2010NUCLEAR RETENTION OF CPEB45665
high calcium level (?700 ?M) (8). This chelation of ER cal-
cium induced the nuclear accumulation of CPEB4 (Fig. 8A).
A reduction in ER calcium diminishes protein chaperone ac-
tivity in the lumen, causing the accumulation of unfolded protein
that will induce the ER stress response (31). To determine
whether CPEB4 nuclear accumulation is a response to ER cal-
cium depletion or ER stress, cells were incubated with thapsigar-
gin (TG) and tunicamycin (TM). While both agents activate the
unfolded protein response (UPR), TG does so by causing ER
calcium efflux, while TM disrupts protein glycosylation. In neu-
rons treated with 16 ?M TG, CPEB4 protein began to accumu-
late in the nucleus 30 min after application of the drug (Fig. 8B).
However, CPEB4 remained cytoplasmic when cells were treated
the expression of C/EBP homology protein (CHOP) (Fig. 8C).
These data suggest that CPEB4 nuclear accumulation is induced
by ER calcium depletion. The results of a dose-response experi-
ment demonstrate that while a 1-h treatment with 4 ?M TG had
no effect on CPEB4 localization, 8 ?M caused an even distribu-
tion between nucleus and cytoplasm and 16 ?M TG caused
strong nuclear CPEB4 staining (Fig. 8D). These data are consis-
tent with the notion that the retention of CPEB4 in the nucleus is
triggered by ER calcium depletion.
To investigate how ER calcium depletion might stimulate
CPEB4 nuclear localization, mouse brain lysate was underlain
on a discontinuous sucrose gradient and then centrifuged; mol-
ecules associated with membranes in such a flotation assay
should band at a sucrose concentration with a similar density.
FIG. 7. Oxygen and glucose deprivation and CPEB4-mediated neuron survival. (A) CPEB4 nuclear localization in DIV14 hippocampal neurons
after OGD treatment. Hippocampal neurons were incubated in medium without glucose in an atmosphere deprived of oxygen for 1 h (OGD 1 h),
which was followed by recovery in normal culture medium in atmosphere containing oxygen for 3 h (rec 3 h). Control (Con) refers to cells without
OGD treatment. The images show CPEB4 staining, TUNEL staining, and CPEB4/TUNEL/DAPI staining to show the location of nuclei. Right,
quantification of the CPEB4 proteins in the nucleus and cytoplasm as described in the legend to Fig. 1. The asterisks refer to a statistically
significant difference (P ? 0.01, Student’s t test) between the results for the indicated samples. (B) Hippocampal neurons were cultured with
lentivirus expressing GFP only (V) or GFP and two different CPEB4 shRNAs (KD2 and KD3). Some neurons were also cultured with two
lentiviruses expressing KD2 and CPEB4 (C4), containing mutations to prevent knockdown but still encoding the correct protein. Extracts from the
cells were probed for CPEB4, CPEB3, tubulin, and GFP. (C) Survival of the neurons infected with some of the viruses noted above was determined
(n ? 200). Error bars represent standard errors of the means.
5666KAN ET AL.MOL. CELL. BIOL.
Fig. 8E shows that protein disulfide isomerase (PDI), an ER-
specific marker, banded in fractions 8 to 10; CPEB4 was most
prevalent in fractions 8 and 9. The material in these fractions
was then pelleted and analyzed by immunoelectron microscopy
using CPEB4 antibody. Gold particles marking CPEB4 were
found to colocalize with ER structures, thus confirming the pres-
ence of CPEB4 on the ER (Fig. 8E). Although we do not know
what other molecules might be involved in tethering CPEB4 to
the ER, we surmise that calcium depletion from this structure
releases CPEB4 and facilitates its nuclear localization.
We have conducted several additional experiments to delve
further into the notion that CPEB4 nuclear accumulation is
driven substantially by ER calcium depletion as opposed to an
increase in cytosolic calcium. Because the various agents used
in this study to alter calcium levels would likely do so to varying
extents, we measured relative intracellular calcium levels with
Fluo-3, a calcium-sensitive fluorescent indicator. DIV16 neu-
rons were incubated with Fluo-3 followed by treatment with
each agent (NMDA, ionomycin, BAPTA/AM, thapsigargin,
etc.); fluorescence intensity was then determined using a fluo-
rescence plate reader. The results in Fig. 9A show that a high
intracellular calcium concentration per se did not induce
CPEB4 to become predominantly nuclear; indeed, BAPTA/
AM, which reduced cytosolic calcium, resulted in nuclear
CPEB4. In this case, chelation of ER calcium was most likely
responsible for the nuclear accumulation of CPEB4. Taken
FIG. 8. Relationship between ER calcium levels and CPEB4 nuclear localization in cultured hippocampal neurons. (A) DIV16 hippocampal
neurons that had been treated with TTX for 24 h were incubated with BAPTA/AM for 20 min. The neurons were then washed, cultured in fresh
medium, and then treated with either DMSO or NMDA for 40 min before fixation and immunostaining for CPEB4. (B) DIV16 hippocampal
neurons were treated with DMSO as a control or 16 ?M thapsigargin (TG) for 30 min or 1 h. At the end of treatment, neurons were fixed and
stained with CPEB4 antibody and DAPI. (C) DIV16 hippocampal neurons were treated with tunicamycin for 1 h, 4 h, or 6 h before fixation and
immunostained with CPEB4 or CHOP antibodies. (D) DIV16 hippocampal neurons were treated with 4, 8, or 16 ?M thapsigargin for 1 h before
fixation and immunostaining for CPEB4. DAPI staining shows location of the nucleus. Size bar ? 10 ?m. (E) Sucrose gradient fractionation was
performed on postnuclear supernatants of brain lysates; the fractions were immunoblotted for CPEB4 and the ER marker PDI. The CPEB4-
enriched fractions (fractions 8 to 10) were then analyzed by immunoelectron microscopy using CPEB4 antibody. Size bar ? 200 nm.
VOL. 30, 2010NUCLEAR RETENTION OF CPEB4 5667
together, these data indicate that the reduction of ER calcium
stores and probably not an increase in cytosolic calcium per se
is responsible for CPEB4 nuclear accumulation.
We next determined the extent to which some of the agents
that cause CPEB4 nuclear accumulation induce ER calcium
depletion. To address this, neurons were transfected with
D1ER, a plasmid devised by Palmer et al. (30) that encodes an
ER localization signal (derived from calreticulin), CFP, cal-
modulin, M13 calmodulin-binding peptide, YFP, and a KDEL
ER retention sequence. In the ER lumen, calcium binding by
calmodulin will cause it to also bind the M13 peptide. This
interaction will induce a conformational change in the fusion
protein such that CFP and YFP will be juxtaposed. When
transfected neurons are excited by 440-nm light, YFP (the
acceptor) will emit a 535-nm FRET signal. The intensity of the
535-nm signal is therefore proportional to the amount of cal-
cium that is bound to calmodulin (30) (Fig. 9B). To first “cal-
ibrate” the FRET signal, the intensity of 535-nm light emission
from untreated cells was set at zero (i.e., Rmax, maximal ER
calcium), and that from cells treated with ionomycin plus
EGTA was set at ?100 (i.e., Rmin, minimum ER calcium). The
data collected from cells treated with various agents were then
plotted as the percent ER calcium change (Fig. 9C). As ex-
pected, NMDA treatment indeed caused a loss of ER calcium,
as did thapsigargin and BAPTA/AM, although to various ex-
tents (Fig. 9C). Thus, agents that cause CPEB4 to concentrate
in the nucleus do indeed induce depletion of ER calcium (see
also Fig. S4 in the supplemental material).
Finally, we addressed whether the IP3 receptor, which re-
sides predominantly in the ER, can mediate ER calcium de-
pletion and transduce a signal to cause CPEB4 nuclear accu-
mulation. Neurons were treated with 2-aminoethoxydiphenyl
borate (2-APB), an inhibitor of IP3 receptor signaling, with or
without NMDA. The results in Fig. 9D show that 2-APB par-
tially blocked NMDA-induced CPEB4 nuclear accumulation.
Thus, these data implicate the IP3 receptor in mediating the
subcellular localization of CPEB4 in response to depletion of
calcium from the endoplasmic reticulum.
FIG. 9. ER calcium depletion causes CPEB4 to be retained in the nucleus. (A) DIV16 neurons were loaded with 1 ?M Fluo-3/AM for 30 min
and then incubated for another 30 min to allow for de-esterification before being treated with the agents noted in the panel. The cells were then
placed in a fluorescence plate reader to quantify Fluo-3 fluorescence, a relative measure of calcium concentration. Iono, ionomycin; Thaps,
thapsigargin; Glu, glutamate. (B) Schematic structure of the D1ER cameleon construct showing an ER-targeting sequence, CFP, calmodulin
(CaM; the calcium binding domains are depicted as balls with a putative flexible coiled region in between), M13 calmodulin binding peptide, YFP,
and a KDEL ER retention sequence. When the calmodulin moiety binds calcium in the ER, it interacts with the M13 peptide, which in turn brings
CFP and YFP into close proximity. Excitation at 440 nm elicits a 535-nm FRET emission from the YFP; this FRET signal is thus proportional
to the amount of calcium in the ER. (C) DIV12 neurons were transfected with pD1ER cameleon construct, and the FRET and CFP signals were
determined. The maximum and minimum FRET/CFP signals were measured when the cells were untreated or treated with ionomycin and EGTA
and were set at 0 and ?100, respectively. The percentage of ER calcium change was plotted (see also Palmer et al. ). ER calcium was
determined when the cells were treated with NMDA, ionomycin, thapsigargin, or BAPTA-AM. Each experiments was performed 3 times (mean
and standard error of the mean are shown). (D) DIV16 neurons were incubated with 2-APB for 45 min prior to the application of NMDA for 40
min; the cells were then fixed and stained for CPEB4. Right, quantification of the results. The asterisks refer to a statistically significant difference
(P ? 0.01, Student’s t test) between the results for the indicated samples. Size bar ? 10 ?m.
5668 KAN ET AL.MOL. CELL. BIOL.
Most investigations of the biological functions of CPEB, the
most studied member of the CPEB family of proteins, have
concentrated on cytoplasmic events, such as cytoplasmic poly-
adenylation, translation, and RNA transport. The finding that
the CPEB proteins are nucleus-cytoplasm shuttling proteins
suggests new functions for these proteins, involving nuclear
RNA metabolism. For example, the CPEB-interacting pro-
teins CPSF and symplekin are involved in nuclear pre-mRNA
polyadenylation, as well as cytoplasmic polyadenylation (14,
25); thus, it is possible that CPEB also modulates nuclear
polyadenylation, as well as alternative splicing (22). Another
possible role for CPEB proteins is RNA nuclear export and
subcellular localization. Unlike Staufen2 and She2, two nu-
cleus-cytoplasm shuttling proteins that accumulate in nucleoli
when RNA binding activity is disrupted (9, 24), CPEB4 trun-
cation mutants that have part of their RNA binding domains
removed retain their shuttling activity, suggesting that CPEB4
is actively transported across the nuclear membrane instead of
being passively exported by way of tethering to RNA.
The failure to identify a nuclear import signal (NLS) in
CPEB4 using serial deletions suggests that there is more than
one NLS. One putative NLS is probably located in an RNA
binding domain, because a recombinant CPEB4 RNA binding
domain alone is sufficient to induce nuclear import in the
presence of HeLa cell cytosol in an in vitro import assay. On
the other hand, the CPEB4 NES clearly requires leucine res-
idues 386 and 390. Because these residues are present in
CPEB2 and -3, they probably mediate the export of these
proteins as well. Moreover, because they are conserved in, for
example, Drosophila Orb2, we infer that they have a similar
function in invertebrates as well. The observation that the two
leucine residues are not present in CPEB underscores the
divergent nature of these two branches of the CPEB family of
proteins. Nonetheless, the fact that CPEB, like CPEB2, -3, and
-4, shuttles between nucleus and cytoplasm in an NMDA-
stimulated manner indicates that the CPEB family branches
have retained and perhaps even share important nuclear func-
CPEB4 is retained in the nucleus following ischemia. The
nuclear staining of CPEB4 in the penumbral region of the
mouse ischemic brain demonstrates that this pathological con-
dition causes a subcellular redistribution of this protein. The
penumbra represents the area of brain that sustains secondary
damage caused by the diffusion of glutamate and potassium
ions from the immediate site of the infarct, as well as hypo-
perfusion; it is also a target of treatment that aims to reduce
brain injury caused by stroke. In addition to CPEB4, other
proteins have also been shown to translocate to the nucleus
upon ischemia. One of them, apoptosis-inducing factor (AIF),
resides in mitochondria and functions as an oxidoreductase in
cells but translocates to the nucleus and induces chromatin
condensation when apoptosis or necrosis is induced (7, 42).
During ischemia and OGD, nuclear translocation of AIF is
considered to be one of the mechanisms that cause neuron
death (4, 34, 53, 54). HGF (hepatocyte growth factor), which
protects neurons from ischemia-induced cell death when per-
fused into the brain, prevents the translocation of AIF to the
Excessive NMDAR activation and ER calcium depletion. To
examine the relationship between CPEB4 nuclear accumula-
tion and cellular calcium, we treated cells with a variety of
agents and then measured the intracellular calcium concentra-
tion with Fluo-3AM, a cytosolic calcium indicator. In general,
agents that induce increased calcium correlated with CPEB4
nuclear accumulation. However, BAPTA/AM, which resulted
in CPEB4 nuclear accumulation, did not lead to any detectable
increase in cellular calcium. On the other hand, the
BAPTA/AM results are consistent with a decrease in ER cal-
cium. Moreover, because the relationship between excessive
NMDAR stimulation and (possible) ER calcium depletion is
not clear, we examined ER calcium levels in neurons directly,
by fluorescence resonance energy transfer (FRET). In this
case, neurons were transfected with a plasmid (cameleon
D1ER) encoding a fluorescent calcium sensor specifically tar-
geted to the ER (30). Upon treatment with NMDA, ionomy-
cin, thapsigargin, and BAPTA/AM, agents that cause CPEB4
to become concentrated in the nucleus, we observed clear
decreases in ER calcium (Fig. 9). While the amount of ER
depletion varied depending on the agent, the correlation be-
tween nuclear CPEB4 immunostaining and ER calcium deple-
tion was very consistent. Thus, based on these direct measure-
ments, we infer that ER calcium depletion is responsible for
CPEB4 nuclear accumulation.
It has been shown that ER stress is induced in ischemic
tissue, as demonstrated by the accumulation of misfolded pro-
teins (15) and induction of the UPR pathway (27). The UPR is
induced by decreased ER folding capacity or increased synthe-
sis of proteins that transit through the ER. Although it has
been suggested that calcium depletion may be responsible for
ER stress after ischemia, no clear evidence has been provided.
In our results, both ER calcium depletion and the application
of NMDA caused CPEB4 nuclear accumulation; we further
demonstrate that NMDAR activation causes ER calcium de-
pletion, indicating a causal relationship. It has been reported
that activation of NMDAR may induce the release of calcium
from the ER through calcium-induced calcium release (10, 39).
In neurons, the ER forms an extended structure that reaches
synaptic spines (43) and contains two types of calcium-releas-
ing channels, the inositol-1,4,5-triphosphate (IP3) receptor and
the ryanodine receptor. Another report also suggests that the
ryanodine receptor may cause ER calcium release because the
application of a ryanodine receptor inhibitor, dantrolene, pro-
tects neurons from NMDA-mediated excitotoxicity (13). Caf-
feine, which activates the ryanodine receptor, did not cause
CPEB4 nuclear accumulation, probably because it did not
cause sufficient ER calcium depletion and/or because of its
well-known pleiotropic effects. On the other hand, inhibition of
the IP3 receptor with 2-aminoethoxydiphenyl borate (2-APB)
blocked NMDA-induced CPEB4 nuclear accumulation, indi-
cating that the IP3 receptor is of particular importance for
causing CPEB4 to remain nuclear.
Mechanism for ER calcium depletion-induced nuclear re-
tention of CPEB4. The treatment of neurons with TG but not
TM excludes ER stress per se as a possible mechanism inducing
CPEB4 nuclear accumulation. The retention of CPEB4 in the
nucleus after BAPTA/AM incubation suggests that ER cal-
cium depletion plays an inhibitory role in CPEB4 nuclear ex-
port. One key extant question is that of how ER calcium
VOL. 30, 2010 NUCLEAR RETENTION OF CPEB45669
depletion induces CPEB4 retention in the nucleus. Recent
investigations of ER calcium homeostasis suggest that store-
operated calcium entry (SOCE) replenishes ER calcium levels
after depletion. SOCE involves two protein families: the stro-
mal-interacting molecule (Stim) family and plasma membrane
calcium channels, Orai. Stim proteins are located in the ER
membrane and serve as ER lumen calcium level sensors (38,
52). Orai channel proteins interact with aggregated Stim pro-
teins and induce calcium influx when ER calcium is depleted
(12, 23, 26, 32, 36). The influxed cytoplasmic calcium is then
transported into the ER by the sarco-/endoplasmic reticulum
Ca2?-ATPase (SERCA) system (20). Whether calcium in-
fluxed through SOCE triggers CPEB4 nuclear retention upon
ER calcium depletion and NMDA stimulation requires further
We thank Lan Xu for advice on the in vitro nuclear import assay,
Oswald Steward for providing rat brain sections, and Melissa Jung-
nickel and Keith Sutton for help with the calcium fluorescence assays.
We also thank the UMass Medical School imaging facility and the
electron microscopy core facility, Robert Singer for the N-EGFP-MS2
plasmid, Roger Tsien for the Cameleon D1ER plasmid, Arthur Mer-
curio for the use of his hypoxia chamber, and Rachel Groppo for
reading the manuscript.
This work was supported by grants from the NIH (GM46779 and
HD37267). A.C.-M. was supported by grant number 3 R01
GM046779-19S1. Core support from the Diabetes and Endocrinology
Research Center Program Project (DK32520) is gratefully acknowl-
1. Alarcon, J. M., R. Hodgman, M. Theis, Y. S. Huang, E. R. Kandel, and J. D.
Richter. 2004. Selective modulation of some forms of Schaffer collateral-
CA1 synaptic plasticity in mice with a disruption of the CPEB-1 gene. Learn.
2. Banker, G., and K. Goslin. 1988. Developments in neuronal cell culture.
3. Berger-Sweeney, J., N. R. Zearfoss, and J. D. Richter. 2006. Reduced ex-
tinction of hippocampal-dependent memories in CPEB knockout mice.
Learn. Mem. 13:4–7.
4. Cao, G., R. S. Clark, W. Pei, W. Yin, F. Zhang, F. Y. Sun, S. H. Graham, and
J. Chen. 2003. Translocation of apoptosis-inducing factor in vulnerable neu-
rons after transient cerebral ischemia and in neuronal cultures after oxygen-
glucose deprivation. J. Cereb. Blood Flow Metab. 23:1137–1150.
5. Choi, D. W., and S. M. Rothman. 1990. The role of glutamate neurotoxicity
in hypoxic-ischemic neuronal death. Annu. Rev. Neurosci. 13:171–182.
6. Corpet, F. 1988. Multiple sequence alignment with hierarchical clustering.
Nucleic Acids Res. 16:10881–10890.
7. Daugas, E., S. A. Susin, N. Zamzami, K. F. Ferri, T. Irinopoulou, N. Laro-
chette, M. C. Prevost, B. Leber, D. Andrews, J. Penninger, and G. Kroemer.
2000. Mitochondrio-nuclear translocation of AIF in apoptosis and necrosis.
FASEB J. 14:729–739.
8. Demaurex, N., and M. Frieden. 2003. Measurements of the free luminal ER
Ca(2?) concentration with targeted “cameleon” fluorescent proteins. Cell
9. Du, L., and J. D. Richter. 2005. Activity-dependent polyadenylation in neu-
rons. RNA 11:1340–1347.
10. Emptage, N., T. V. Bliss, and A. Fine. 1999. Single synaptic events evoke
NMDA receptor-mediated release of calcium from internal stores in hip-
pocampal dendritic spines. Neuron 22:115–124.
11. Ferrari, F., V. Mercaldo, G. Piccoli, C. Sala, S. Cannata, T. Achsel, and C.
Bagni. 2007. The fragile X mental retardation protein-RNP granules show
an mGluR-dependent localization in the post-synaptic spines. Mol. Cell
12. Feske, S., Y. Gwack, M. Prakriya, S. Srikanth, S. H. Puppel, B. Tanasa, P. G.
Hogan, R. S. Lewis, M. Daly, and A. Rao. 2006. A mutation in Orai1 causes
immune deficiency by abrogating CRAC channel function. Nature 441:179–
13. Frandsen, A., and A. Schousboe. 1992. Mobilization of dantrolene-sensitive
intracellular calcium pools is involved in the cytotoxicity induced by quis-
qualate and N-methyl-D-aspartate but not by 2-amino-3-(3-hydroxy-5-meth-
ylisoxazol-4-yl)propionate and kainate in cultured cerebral cortical neurons.
Proc. Natl. Acad. Sci. U. S. A. 89:2590–2594.
14. Hofmann, I., M. Schnolzer, I. Kaufmann, and W. W. Franke. 2002. Symple-
kin, a constitutive protein of karyo- and cytoplasmic particles involved in
mRNA biogenesis in Xenopus laevis oocytes. Mol. Biol. Cell 13:1665–1676.
15. Hu, B. R., M. E. Martone, Y. Z. Jones, and C. L. Liu. 2000. Protein aggre-
gation after transient cerebral ischemia. J. Neurosci. 20:3191–3199.
16. Huang, Y. S., J. H. Carson, E. Barbarese, and J. D. Richter. 2003. Facilita-
tion of dendritic mRNA transport by CPEB. Genes Dev. 17:638–653.
17. Huang, Y. S., M. C. Kan, C. L. Lin, and J. D. Richter. 2006. CPEB3 and
CPEB4 in neurons: analysis of RNA-binding specificity and translational
control of AMPA receptor GluR2 mRNA. EMBO J. 25:4865–4876.
18. Hudmon, A., and H. Schulman. 2002. Neuronal CA2?/calmodulin-depen-
dent protein kinase II: the role of structure and autoregulation in cellular
function. Annu. Rev. Biochem. 71:473–510.
19. Ishida, A., Y. Shigeri, Y. Tatsu, K. Uegaki, I. Kameshita, S. Okuno, T.
Kitani, N. Yumoto, and H. Fujisawa. 1998. Critical amino acid residues of
AIP, a highly specific inhibitory peptide of calmodulin-dependent protein
kinase II. FEBS Lett. 427:115–118.
20. Jousset, H., M. Frieden, and N. Demaurex. 2007. STIM1 knockdown reveals
that store-operated Ca2? channels located close to sarco/endoplasmic Ca2?
ATPases (SERCA) pumps silently refill the endoplasmic reticulum. J. Biol.
21. Kudo, N., N. Matsumori, H. Taoka, D. Fujiwara, E. P. Schreiner, B. Wolff,
M. Yoshida, and S. Horinouchi. 1999. Leptomycin B inactivates CRM1/
exportin 1 by covalent modification at a cysteine residue in the central
conserved region. Proc. Natl. Acad. Sci. U. S. A. 96:9112–9117.
22. Lin, C. L., V. Evans, S. Shen, Y. Xing, and J. D. Richter. 2010. The nuclear
experience of CPEB: implications for RNA processing and translational
control. RNA 16:338–348.
23. Luik, R. M., M. M. Wu, J. Buchanan, and R. S. Lewis. 2006. The elementary
unit of store-operated Ca2? entry: local activation of CRAC channels by
STIM1 at ER-plasma membrane junctions. J. Cell Biol. 174:815–825.
24. Macchi, P., A. M. Brownawell, B. Grunewald, L. DesGroseillers, I. G.
Macara, and M. A. Kiebler. 2004. The brain-specific double-stranded RNA-
binding protein Staufen2: nucleolar accumulation and isoform-specific ex-
portin-5-dependent export. J. Biol. Chem. 279:31440–31444.
25. Mandel, C. R., S. Kaneko, H. Zhang, D. Gebauer, V. Vethantham, J. L.
Manley, and L. Tong. 2006. Polyadenylation factor CPSF-73 is the pre-
mRNA 3?-end-processing endonuclease. Nature 444:953–956.
26. Mercer, J. C., W. I. Dehaven, J. T. Smyth, B. Wedel, R. R. Boyles, G. S. Bird,
and J. W. Putney, Jr. 2006. Large store-operated calcium selective currents
due to co-expression of Orai1 or Orai2 with the intracellular calcium sensor,
Stim1. J. Biol. Chem. 281:24979–24990.
27. Morimoto, N., Y. Oida, M. Shimazawa, M. Miura, T. Kudo, K. Imaizumi,
and H. Hara. 2007. Involvement of endoplasmic reticulum stress after mid-
dle cerebral artery occlusion in mice. Neuroscience 147:957–967.
28. Niimura, M., N. Takagi, K. Takagi, R. Mizutani, N. Ishihara, K. Matsumoto,
H. Funakoshi, T. Nakamura, and S. Takeo. 2006. Prevention of apoptosis-
inducing factor translocation is a possible mechanism for protective effects of
hepatocyte growth factor against neuronal cell death in the hippocampus
after transient forebrain ischemia. J. Cereb. Blood Flow Metab. 26:1354–
29. Nishi, K., M. Yoshida, D. Fujiwara, M. Nishikawa, S. Horinouchi, and T.
Beppu. 1994. Leptomycin B targets a regulatory cascade of crm1, a fission
yeast nuclear protein, involved in control of higher order chromosome struc-
ture and gene expression. J. Biol. Chem. 269:6320–6324.
30. Palmer, A. E., C. Jin, J. C. Reed, and R. Y. Tsien. 2004. Bcl-2-mediated
alterations in endoplasmic reticulum Ca2? analyzed with an improved ge-
netically encoded fluorescent sensor. Proc. Natl. Acad. Sci. U. S. A. 101:
31. Paschen, W., S. Hotop, and C. Aufenberg. 2003. Loading neurons with
BAPTA-AM activates xbp1 processing indicative of induction of endoplas-
mic reticulum stress. Cell Calcium 33:83–89.
32. Peinelt, C., M. Vig, D. L. Koomoa, A. Beck, M. J. Nadler, M. Koblan-
Huberson, A. Lis, A. Fleig, R. Penner, and J. P. Kinet. 2006. Amplification
of CRAC current by STIM1 and CRACM1 (Orai1). Nat. Cell Biol. 8:771–
33. Petosa, C., G. Schoehn, P. Askjaer, U. Bauer, M. Moulin, U. Steuerwald, M.
Soler-Lopez, F. Baudin, I. W. Mattaj, and C. W. Muller. 2004. Architecture
of CRM1/Exportin1 suggests how cooperativity is achieved during formation
of a nuclear export complex. Mol. Cell 16:761–775.
34. Plesnila, N., C. Zhu, C. Culmsee, M. Groger, M. A. Moskowitz, and K.
Blomgren. 2004. Nuclear translocation of apoptosis-inducing factor after
focal cerebral ischemia. J. Cereb. Blood Flow Metab. 24:458–466.
35. Plum, F. 1983. What causes infarction in ischemic brain? The Robert
Wartenberg Lecture. Neurology 33:222–233.
36. Prakriya, M., S. Feske, Y. Gwack, S. Srikanth, A. Rao, and P. G. Hogan.
2006. Orai1 is an essential pore subunit of the CRAC channel. Nature
37. Richter, J. D. 2007. CPEB: a life in translation. Trends Biochem. Sci. 32:
38. Roos, J., P. J. DiGregorio, A. V. Yeromin, K. Ohlsen, M. Lioudyno, S. Zhang,
O. Safrina, J. A. Kozak, S. L. Wagner, M. D. Cahalan, G. Velicelebi, and
5670 KAN ET AL.MOL. CELL. BIOL.
K. A. Stauderman. 2005. STIM1, an essential and conserved component of
store-operated Ca2? channel function. J. Cell Biol. 169:435–445.
39. Rose, C. R., and A. Konnerth. 2001. Stores not just for storage. Intracellular
calcium release and synaptic plasticity. Neuron 31:519–522.
40. Rossi, D. J., T. Oshima, and D. Attwell. 2000. Glutamate release in severe
brain ischaemia is mainly by reversed uptake. Nature 403:316–321.
41. Steward, O., C. S. Wallace, G. L. Lyford, and P. F. Worley. 1998. Synaptic
activation causes the mRNA for the IEG arc to localize selectively near
activated postsynaptic sites on dendrites. Neuron 21:741–751.
42. Susin, S. A., H. K. Lorenzo, N. Zamzami, I. Marzo, B. E. Snow, G. M.
Brothers, J. Mangion, E. Jacotot, P. Costantini, M. Loeffler, N. Larochette,
D. R. Goodlett, R. Aebersold, D. P. Siderovski, J. M. Penninger, and G.
Kroemer. 1999. Molecular characterization of mitochondrial apoptosis-in-
ducing factor. Nature 397:441–446.
43. Svoboda, K., and Z. F. Mainen. 1999. Synaptic [Ca2?]: intracellular stores
spill their guts. Neuron 22:427–430.
44. Tay, J., and J. D. Richter. 2001. Germ cell differentiation and synaptonemal
complex formation are disrupted in CPEB knockout mice. Dev. Cell 1:201–
45. Tymianski, M., M. P. Charlton, P. L. Carlen, and C. H. Tator. 1993. Source
specificity of early calcium neurotoxicity in cultured embryonic spinal neu-
rons. J. Neurosci. 13:2085–2104.
46. Tymianski, M., M. C. Wallace, I. Spigelman, M. Uno, P. L. Carlen, C. H.
Tator, and M. P. Charlton. 1993. Cell-permeant Ca2? chelators reduce
early excitotoxic and ischemic neuronal injury in vitro and in vivo. Neuron
47. van Leyen, K., H. Y. Kim, S. R. Lee, G. Jin, K. Arai, and E. H. Lo. 2006.
Baicalein and 12/15-lipoxygenase in the ischemic brain. Stroke 37:3014–3018.
48. Watterson, D. M., S. Mirzoeva, L. Guo, A. Whyte, J. J. Bourguignon, M.
Hibert, J. Haiech, and L. J. Van Eldik. 2001. Ligand modulation of glial
activation: cell permeable, small molecule inhibitors of serine-threonine
protein kinases can block induction of interleukin 1 beta and nitric oxide
synthase II. Neurochem. Int. 39:459–468.
49. Wells, D. G., X. Dong, E. M. Quinlan, Y. S. Huang, M. F. Bear, J. D. Richter,
and J. R. Fallon. 2001. A role for the cytoplasmic polyadenylation element
in NMDA receptor-regulated mRNA translation in neurons. J. Neurosci.
50. Wu, L., D. Wells, J. Tay, D. Mendis, M. A. Abbott, A. Barnitt, E. Quinlan, A.
Heynen, J. R. Fallon, and J. D. Richter. 1998. CPEB-mediated cytoplasmic
polyadenylation and the regulation of experience-dependent translation of
alpha-CaMKII mRNA at synapses. Neuron 21:1129–1139.
51. Zearfoss, N. R., J. M. Alarcon, P. Trifilieff, E. Kandel, and J. D. Richter.
2008. A molecular circuit composed of CPEB-1 and c-Jun controls growth
hormone-mediated synaptic plasticity in the mouse hippocampus. J. Neuro-
52. Zhang, S. L., Y. Yu, J. Roos, J. A. Kozak, T. J. Deerinck, M. H. Ellisman,
K. A. Stauderman, and M. D. Cahalan. 2005. STIM1 is a Ca2? sensor that
activates CRAC channels and migrates from the Ca2? store to the plasma
membrane. Nature 437:902–905.
53. Zhao, H., M. A. Yenari, D. Cheng, O. L. Barreto-Chang, R. M. Sapolsky, and
G. K. Steinberg. 2004. Bcl-2 transfection via herpes simplex virus blocks
apoptosis-inducing factor translocation after focal ischemia in the rat.
J. Cereb. Blood Flow Metab. 24:681–692.
54. Zhu, C., L. Qiu, X. Wang, U. Hallin, C. Cande, G. Kroemer, H. Hagberg, and
K. Blomgren. 2003. Involvement of apoptosis-inducing factor in neuronal
death after hypoxia-ischemia in the neonatal rat brain. J. Neurochem. 86:
VOL. 30, 2010 NUCLEAR RETENTION OF CPEB45671