-
DOI: 10.1126/science.1209236, 623 (2011);334 Science
, et al.Victoria M. HoThe Cell Biology of Synaptic
Plasticity
This copy is for your personal, non-commercial use only.
clicking here.colleagues, clients, or customers by , you can
order high-quality copies for yourIf you wish to distribute this
article to others
here.following the guidelines can be obtained byPermission to
republish or repurpose articles or portions of articles
): November 3, 2011 www.sciencemag.org (this infomation is
current as ofThe following resources related to this article are
available online at
http://www.sciencemag.org/content/334/6056/623.full.htmlversion
of this article at:
including high-resolution figures, can be found in the
onlineUpdated information and services,
http://www.sciencemag.org/content/334/6056/623.full.html#ref-list-1,
20 of which can be accessed free:cites 63 articlesThis article
http://www.sciencemag.org/cgi/collection/neuroscienceNeuroscience
subject collections:This article appears in the following
registered trademark of AAAS. is aScience2011 by the American
Association for the Advancement of Science; all rights reserved.
The title
CopyrightAmerican Association for the Advancement of Science,
1200 New York Avenue NW, Washington, DC 20005. (print ISSN
0036-8075; online ISSN 1095-9203) is published weekly, except the
last week in December, by theScience
on N
ovem
ber 4
, 201
1w
ww
.sci
ence
mag
.org
Dow
nloa
ded
from
-
50. D. D. Bock et al., Nature 471, 177 (2011).51. V. Jain, H. S.
Seung, S. C. Turaga, Curr. Opin. Neurobiol.
20, 653 (2010).Acknowledgments: We thank K. Briggman, H. Hess,
J. Livet,
M. Helmstaedter, and S. Smith for providing images and
A. Karpova for commenting on the manuscript. Our ownwork was
supported by the NIH, Gatsby CharitableFoundation, and a
Collaborative Innovation Award (no.43667) from Howard Hughes
Medical Institute ( J.W.L.),the Deutsche Forschungsgemeinschaft
(W.D.), and the
Max-Planck Society (W.D.). W.D. has provided a technol-ogy
license for block-face EM to Gatan, Incorporated.
10.1126/science.1209168
The Cell Biology of Synaptic PlasticityVictoria M. Ho,1 Ji-Ann
Lee,2 Kelsey C. Martin2,3,4*
Synaptic plasticity is the experience-dependent change in
connectivity between neurons thatis believed to underlie learning
and memory. Here, we discuss the cellular and molecularprocesses
that are altered when a neuron responds to external stimuli, and
how these alterationslead to an increase or decrease in synaptic
connectivity. Modification of synaptic componentsand changes in
gene expression are necessary for many forms of plasticity. We
focus onexcitatory neurons in the mammalian hippocampus, one of the
best-studied model systemsof learning-related plasticity.
The circuitry of the human brain is com-posed of a trillion
(1012) neurons and aquadrillion (1015) synapses, whose
con-nectivity underlies all human perception, emo-tion, thought,
and behavior. Studies in a rangeof species have revealed that the
overall struc-ture of the nervous system is genetically hard-wired
but that neural circuits undergo extensivesculpting and rewiring in
response to a varietyof stimuli. This process of
experience-dependentchanges in synaptic connectivity is called
synap-tic plasticity.
Studies of synaptic plasticity have begun todetail the molecular
mechanisms that underliethese synaptic changes. This research has
exam-ined a variety of cell biological processes, in-cluding
synaptic vesicle release and recycling,neurotransmitter receptor
trafficking, cell adhe-sion, and stimulus-induced changes in gene
ex-pression within neurons. Taken together, thesestudies have
provided an initial molecular bio-logical understanding of how
nature and nurturecombine to determine our identities. As a
result,research on synaptic plasticity promises to pro-vide insight
into the biological basis of manyneuropsychiatric disorders in
which experience-dependent brain rewiring goes awry.
Here we focus on long-lasting forms of plas-ticity that underlie
learning and memory. Weconsider, in turn, each component of the
synapse:the presynaptic compartment, the postsynapticcompartment,
and the synaptic cleft, and discuss
processes that undergo activity-dependent mod-ifications to
alter synaptic efficacy. Long-lastingchanges in synaptic
connectivity require newRNA and/or protein synthesis, and we
discusshow gene expression is regulated within neu-rons.We
concentrate on studies of learning-relatedplasticity at excitatory
chemical synapses in therodent hippocampus because these provide
ex-tensive evidence for the cell biological mecha-nisms of
plasticity in the vertebrate brain. Space
constraints prevent us from addressing any sin-gle mechanism in
depth; instead, our aim is toprovide a framework for understanding
the cellbiology of synaptic plasticity.
Hippocampal Synaptic PlasticityThe successful study of the cell
biology of syn-aptic plasticity requires a tractable
experimentalmodel system. Ideally, such a model should con-sist of
a defined population of identifiable neu-rons and be amenable to
electrophysiological,genetic, and molecular cell biological
manipu-lations. Awell-studiedmodel system for studyingplasticity in
the adult vertebrate nervous systemis the rodent hippocampus (Fig.
1). Critical formemory formation, the anatomy of the hippo-campus
renders it particularly suitable for electro-physiological
investigation. It consists of threesequential synaptic pathways
(perforant, mossyfiber, and Schaffer collateral pathways), eachwith
discrete cell body layers and axonal anddendritic projections (Fig.
1). Synaptic plasticityhas been studied in all three hippocampal
path-ways. Distinct stimuli elicit changes in synaptic
1Interdepartmental Program in Neurosciences, University
ofCaliforniaLos Angeles (UCLA), BSRB 390B, 615 Charles E.Young
Drive South, Los Angeles, CA 900951737, USA. 2De-partment of
Biological Chemistry, UCLA, BSRB 310, 615Charles E. Young Drive
South, Los Angeles, CA 900951737,USA. 3Department of Psychiatry and
Biobehavioral Sciences,UCLA, BSRB 390B, 615 Charles E. Young Drive
South, LosAngeles, CA 900951737, USA. 4Brain Research
Institute,UCLA, BSRB 390B, 615 Charles E. Young Drive South,
LosAngeles, CA 900951737, USA.
*To whom correspondence should be addressed.
E-mail:[email protected]
Dentate gyrus
Record
Stimulate
CA3
CA1
Perforant (axons from entorhinal cortex)Mossy fiber (axons from
dentate granule cells)Schaffer collateral (axons from CA3 pyramidal
cells)
Synaptic pathways
Fig. 1. Hippocampal synaptic plasticity. The rodent hippocampus
can be dissected and cut into transverseslices that preserve all
three synaptic pathways. In the perforant pathway (purple), axons
from theentorhinal cortex project to form synapses (yellow circles)
on dendrites of dentate granule cells; in themossy fiber pathway
(green), dentate granule axons synapse on CA3 pyramidal neuron
dendrites; and inthe Schaffer collateral pathway (brown), CA3 axons
synapse on CA1 dendrites. The dentate, CA3, and CA1cell bodies form
discrete somatic layers (dark blue lines), projecting axons and
dendrites into definedregions. Electrodes can be used to stimulate
axonal afferents and record from postsynaptic follower cells,as
illustrated for the Schaffer collateral (CA3-CA1) pathway.
www.sciencemag.org SCIENCE VOL 334 4 NOVEMBER 2011 623
REVIEWS
on N
ovem
ber 4
, 201
1w
ww
.sci
ence
mag
.org
Dow
nloa
ded
from
-
efficacy; high-frequency stimuli pro-duce synaptic strengthening
calledlong-term potentiation (LTP), andlow-frequency stimulation
producessynaptic weakening, called long-term depression (LTD). LTP
andLTD can also be produced by spiketimingdependent plasticity,
inwhichthe relative timing of pre- and post-synaptic spikes leads
to changes insynaptic strength (1). Different pat-terns of
stimulation elicit changes insynaptic strength that persist over
var-ious time domains, with long-lastingforms, but not short-term
forms, re-quiring new RNA and protein syn-thesis (2).
Hippocampal plasticity is studiedin in vivo and in vitro
preparations.Implanted electrodes can be used tostimulate and
record from hippocam-pal pathways in living animals. Thehippocampus
can be dissected outof the brain and cut into 300- to 500-mm-thick
transverse slices that canbe maintained and recorded fromfor hours
(Fig. 1). Slices can alsobe kept as organotypic slice culturesfor
weeks, preserving many aspectsof their architecture. Finally,
hippo-campal neurons can be studied indissociated cultures, which
are par-ticularly amenable to manipulationand dynamic imaging of
individ-ual neurons and synapses. The de-velopment of genetically
modifiedmice and vectors for acute manipu-lation of gene expression
completea rich tool-kit for studies of the celland molecular
biology of hippocampal synapticplasticity.
Presynaptic Mechanisms of PlasticityCommunication at chemical
synapses involvesthe release of neurotransmitter from the
presyn-aptic terminal, diffusion across the cleft, andbinding to
postsynaptic receptors (Figs. 2 and3). Chemical neurotransmission
is rapid (occur-ring in milliseconds) and highly regulated.
Thepresynaptic terminal contains synaptic vesiclesfilled with
neurotransmitter and a dense matrixof cytoskeleton and scaffolding
proteins at thesite of release, the active zone. Varying the
prob-ability of neurotransmitter release provides onemechanism for
altering synaptic strength duringneuronal plasticity.
Synaptic vesicle release can be subdividedinto distinct steps,
including vesicle mobilization,docking, priming, fusion, and
recycling. Althougheach of these steps may be regulated by
activity,we will highlight three: vesicle mobilization,docking, and
priming.
Synapsins and synaptic vesicle mobilization.The population of
synaptic vesicles within a pre-synaptic terminal exist in three
states: the readily
releasable pool docked at the active zone; therecycling pool,
which can be released with mod-erate stimulation; and the reserve
pool, which isonly released in response to strong stimuli. A
fam-ily of phosphoproteins called synapsins tethersynaptic vesicles
to the actin cytoskeleton andto one another. Neuronal stimulation
activateskinases that phosphorylate synapsins to modulatesynaptic
vesicle tethering and thereby alter thenumber of synaptic vesicles
available for release(3). Synapsin knockout mice have reduced
re-serve pools of synaptic vesicles and demon-strate deficits in
learning and memory as wellas various forms of plasticity (4),
indicating thatactivity-dependent modulation of synaptic ves-icle
mobilization is critical to neuronal and be-havioral
plasticity.
RIM proteins and synaptic vesicle dockingand priming. For
synaptic vesicles to becomefusion competent, they must undergo
dockingand priming, in which vesicle and plasma mem-brane soluble
NSF-attachment protein receptor(SNARE) proteins are brought into
close contactto allow rapid fusion following calcium influx.The
Rab3-interacting molecule (RIM) familyof proteins is critical for
this process (5). As
large, multidomain proteins, RIMs act as scaf-folding proteins
to cluster calcium channels in theactive zone (6) and interact with
Munc-13 (7),a priming factor required for efficient SNAREcomplex
formation and membrane fusion. RIMis a substrate for
phosphorylation by protein ki-nase A (PKA) and is required for
mossy fiberLTP (8).
Postsynaptic Mechanisms of PlasticityMost principle neurons in
the brain are studdedwith membrane protuberances called dendrit-ic
spines, which are the postsynaptic compart-ments. Spines are
heterogeneous in shape (Fig.2), but consist of a bulbous head and a
thinnerneck that connects the spine to the dendriticshaft; the size
of the spine head and the vol-ume of the spine correlate with
synaptic strength(9, 10), with large spine heads containing
moreneurotransmitter receptors, reflecting greater syn-aptic
strength. Spines serve as compartmental-ized signaling units, and
the number and shapeof spines change during synaptic plasticity
(11).At the ultrastructural level, the postsynaptic com-partment is
characterized by an electron-densepostsynaptic density (PSD), which
consists ofneurotransmitter receptors and an extensive net-work of
scaffolding proteins.
Activation of postsynaptic kinases in thespine: CaMKII and PKMz.
LTP and LTD induc-tion are both dependent on postsynaptic
increasesin intracellular calcium, with LTP requiring
largeincreases in calcium concentrations and LTDbeing dependent on
smaller calcium increases.The increase in calcium activates
multiple down-stream signaling enzymes, including the
kinasescalcium/calmodulin-dependent protein kinase II(CaMKII) and
protein kinase C (PKC).
LTP induction in the CA1 region of the hip-pocampus requires
CaMKII activity (12, 13),and transgenic mice lacking the a isoform
havedefective LTP and spatial learning (14, 15).CaMKII undergoes
autophosphorylation in re-sponse to increases in Ca2+-bound
calmodulin,which renders the kinase autonomously active.Neuronal
activity also translocates CaMKII tothe PSD, where it can
phosphorylate many PSDproteins, including glutamate receptors. The
auto-phosphorylation of CaMKII is essential for LTPinduction and,
perhaps, its maintenance (16) [butsee (17)].
The brain-restricted atypical PKC isoform,protein kinase M zeta
(PKMz), is constitutivelyactive and thus phosphorylates targets in
the ab-sence of extracellular stimulation. PKMzmRNAis targeted to
dendrites where activity-dependentsignaling cascades regulate its
local translationduring LTP and LTD (18). PKMz is sufficientand
necessary for LTP maintenance and for themaintenance of long-term
memories, and PKMzactivation may perpetuate synaptic plasticity
andmemory (18, 19).
Activity-dependent modulation of post-synaptic glutamate
receptors. The main excitatoryneurotransmitter in the brain is
glutamate, which
Fig. 2. The ultrastructure of the synapse. Neurons
communicatewith one another at chemical synapses. (A) Electron
micrographfrom area CA1 in adult rat hippocampus. The CA1 dendritic
shaftis colorized in yellow, the spine neck and head in green,
thepresynaptic terminal in orange, and astroglial processes in
blue.Scale bar, 0.5 mm. (B) Three-dimensional reconstruction of
an8.5-mm-long dendrite (yellow) with the PSDs labeled in red.
Notethe variation in spine and PSD size and shape. Scale cube, 0.5
mm3.Reproduced with permission from Elsevier (63).
4 NOVEMBER 2011 VOL 334 SCIENCE www.sciencemag.org624
REVIEWS
on N
ovem
ber 4
, 201
1w
ww
.sci
ence
mag
.org
Dow
nloa
ded
from
-
activates several postsynaptic receptors. Two typesof ionotropic
glutamate receptorsa-amino-3-hydroxy-5-methyl-4-isoxazolepropionic
acid(AMPA) and N-methyl D-aspartate (NMDA)have central roles in
hippocampal synaptic plas-ticity. Both are ligand-gated ion
channels andhave unique properties that subserve differentphases of
synaptic plasticity. NMDA-type glu-tamate receptors (NMDARs) are
calcium per-meable and, when activated, allow an influx ofcalcium
needed for the induction of LTP. How-ever, NMDARs do not conduct
current at restingpotentials because their channel pores are
blockedby magnesium cations. Consequently, NMDARshave been called
coincidence detectors because,to conduct current, they require both
presynaptictransmitter release as well as postsynaptic
de-polarization to relieve the magnesium block.AMPA-type glutamate
receptors (AMPARs) areimportant for the expression and
maintenanceof LTP. Unlike NMDARs, AMPARs can be ac-
tivated by ligand binding at resting potentialsto allow current
flow. Increased conductancethrough AMPARs is responsible for the
increasein synaptic strength during NMDAR-dependentLTP at CA1
synapses.
Given the importance of AMPARs in de-termining synaptic
strength, much effort hasfocused on delineating the mechanisms
thatregulate their function. Regulated phosphoryl-ation can change
AMPAR function by changingthe open probabilities and conductances
of thereceptors. However, changes in channel proper-ties are
unlikely to account for the drastic changesin AMPAR function seen
with LTP (20). Instead,changes inAMPAR function during synaptic
plas-ticity are mostly due to phosphorylation-inducedchanges in its
abundance at the synapse.
AMPARs traffic constitutively to and fromthe plasma membrane via
recycling endosomes(21) (Fig. 3). Delivery of AMPARs to synapses
isbelieved to occur first by exocytosis at extra-
synaptic sites followed by lateral diffusion withinthe plasma
membrane to PSDs, where the mo-bility of the receptors is greatly
reduced. Duringremoval of synaptic AMPARs, receptors diffuseaway
from the PSD and then undergo clathrin-mediated, dynamin-dependent
endocytosis. Afterendocytosis, small GTP-binding proteins of theRab
family and effector proteins direct AMPARseither to early (sorting)
endosomes or back tothe plasma membrane (22).
AMPAR trafficking occurs constitutively un-der basal conditions
and is modulated by activitythrough changes in actin and myosin
dynamics(23), as well as AMPAR interactions with scaf-folding
proteins and accessory subunits. One ofthese accessory subunits,
Stargazin, mediates theinteraction between AMPARs and the PSD
pro-tein PSD-95, and this interaction is importantfor synaptic
localization of AMPARs (24). Ac-tivity alters the phosphorylation
of Stargazin,with phosphorylated Stargazin decreasing the
5 P
Internal sources of AMPARs
DegradationEndocytosis
Presynapticaxon
terminal
Postsynapticdendritic
spine
Synapticcleft
Synapticcell adhesion
molecules
VSCCFusion
and exocytosis
EphrinB
EphB
Endocytosis
Synapticvesicle
Neurotransmitteruptake
Neurotransmittermolecules
AMPAR
AMPAR
NMDAR
PSD
1
Exocytosis3
b
cNeurexin
Neuroligin
NCAM
NCAM
PSA-NCAM
PSA-NCAM
2
Lateraldiffusion
4
Vesiclemobilization
Active zone 6Docking
and priming
a
Fig. 3. Activity-dependent modulation of pre-, post-, and
trans-synapticcomponents. Presynaptic: Neurotransmitter vesicle
cycling. Neurotransmitterrelease starts with the filling of
synaptic vesicles, which then dock andundergo priming at the active
zone. Arrival of an action potential inducescalcium influx through
voltage-sensitive calcium channels (VSCCs), whichtriggers membrane
fusion and exocytosis. The synaptic vesicles are thenrecycled via
local reuse (a; kiss and stay), fast recycling (b; kiss andrun), or
clathrin-mediated endocytosis (c). Neurotransmitter release can
beregulated during plasticity as exemplified by the regulation of
synapsinphosphorylation (1) and the regulation of RIM protein
phosphorylation(2). Postsynaptic: AMPA receptor trafficking.
Locally and somatically syn-thesized AMPARs enter a pool of
endosomes that undergo constitutive
and regulated membrane trafficking. During potentiation, greater
receptorinsertion (3) increases the concentration of AMPARs at the
synapse, wherethey are anchored by interactions at the PSD. During
synaptic depression,AMPARs are endocytosed (3). The preferential
location of endocytosis andexocytosis is probably extrasynaptic.
Within the plasma membrane, traf-ficking of AMPARs between the
synapse and the point of insertion or removaloccurs by lateral
diffusion. Extrasynaptic movement of AMPARs increaseswith neuronal
activity (4). Receptor trafficking is modulated by phosphoryl-ation
of AMPAR subunits (5), which influences interactions with
scaffoldingproteins. Trans-synaptic: Synaptic cell adhesion
molecules. PSA-NCAM isincreased following neuronal activity (6).
Lightning bolts indicate activity-dependent processes.
www.sciencemag.org SCIENCE VOL 334 4 NOVEMBER 2011 625
REVIEWS
on N
ovem
ber 4
, 201
1w
ww
.sci
ence
mag
.org
Dow
nloa
ded
from
-
mobility of AMPARs and enhancing AMPARfunction. Blocking
Stargazin phosphorylationor dephosphorylation blocks LTP and LTD,
re-spectively (25).
AMPARs exist as tetramers made up of dif-ferent combinations of
the four subunits, GluA1through 4. The cytoplasmic tails of each
subunitcontain multiple phosphorylation sites that reg-ulate the
trafficking of AMPARs. For example,PKA phosphorylation of S845 in
the long cyto-plasmic tail of GluA1 increases GluA1
surfaceexpression due to both enhanced insertion andattenuated
internalization (26). Conversely, LTDof dissociated cultures and
brain slices results indephosphorylation of S845 and is correlated
withan increased rate of AMPAR endocytosis (27).Knock-in mice with
phosphorylation-deficientmutations at both S831A and S845A display
aloss of NMDA-induced AMPAR internalization,deficits in LTP and
LTD, and have impaired spa-tial memory (28).
Although studies of posttranslational mod-ifications at
individual sites have established arole for regulating GluA1
trafficking and chan-nel properties, they do not fully account for
thechanges in GluA1 function observed with syn-aptic plasticity
(29). Activity-modified residuescontinue to be discovered,
including, for exam-ple, the highly conserved T840
phosphorylationsite, the phosphorylation of which correlates
re-markably well with synaptic strength (30). It islikely that
complex patterns of phosphorylationand of other post-translational
modifications (e.g.,palmitoylation or ubiquitination) combine to
reg-ulate AMPAR localization.
Trans-Synaptic Signaling; the Synaptic CleftThe synaptic cleft
is a ~20-nm junction betweenthe pre- and postsynaptic compartments,
consist-ing of a space through which neurotransmittersdiffuse to
bind postsynaptic receptors, as well asa network of cell adhesion
molecules (CAMs)
that keeps the synapse together. These adhesiveinteractions are
so strong that it is impossible toseparate intact pre- from
postsynaptic compart-ments biochemically.
Role of CAMs in synaptic plasticity.TheCAMsthat localize to the
synaptic cleft include membersof the cadherin, integrin, and
immunoglobulin-containing CAMs, as well as neurexins and
neu-roligins. Much research has focused on trying tounderstand
whether and how CAMs mediatesynapse specificity during neural
circuit forma-tion. Here we focus on the regulation of synapticCAMs
during experience-dependent synapticplasticity, limiting our
discussion to just two ofmany examples.
One such example involves the addition oflarge sialic acid
homopolymers to the neuralcell adhesion molecule (NCAM) to form
poly-sialylated NCAM (PSA-NCAM), which decreaseshomophilic adhesion
to allow new synaptic re-modeling and growth. The ratio of
PSA-NCAM
1
2
3
5
Transcription
Export
RNAgranule
RNA trafficking Synapse to nucleus signalsSplicing and
processing
RBPsRBPs
AAAAAAAAAAPABP
elF4E4EBP
m7G
40S
Localtranslation
Proteasome
Localdegradation
Nascentpolypeptide
AAAAAAAAAAPABP
elF4E
40SelF4Gm7G
elF2
GTP
Met-i
4
AAAAAAAAAA
m7G
60SeEF240S
Folded proteinUbiquitin
Degradedprotein
Ub Ub Ub Ub
Fig. 4. Local regulation of the synaptic proteome. Synaptic
plasticity modifiesgene expression at many levels. Strong
stimulation of synapses triggers signalsthat are sent to the
nucleus to modify RNA synthesis. Synaptic activity alsomodifies
protein synthesis, and has been found to act at several key
stepsduring translation: (1) Relief of repression, e.g.,
RISC-mediated repression; (2)modification of translational
initiation to allow 4E-4G interaction and re-cruitment of 40S; (3)
formation of the preinitiation complex; and (4) de-phosphorylation
of eEF2 to allow for catalysis of ribosome translocation
during translational elongation. To counterbalance local protein
synthesis,local protein degradation also occurs at synapses (5).
Together, these regu-lated steps in protein addition and removal
allow for rapid, spatially restrictedcontrol of the synaptic
proteome. Lightning bolts indicate activity-dependentprocesses.
RBP, RNA binding proteins such as exon junction complexes,
RISCmachinery, Staufen, CPEB, etc. (Note: Although local
translation in dendrites isa well-accepted phenomenon, it has not
been demonstrated to occur inspines.)
4 NOVEMBER 2011 VOL 334 SCIENCE www.sciencemag.org626
REVIEWS
on N
ovem
ber 4
, 201
1w
ww
.sci
ence
mag
.org
Dow
nloa
ded
from
-
to NCAM increases following hippocampallearning tasks, and
inactivation of the enzymethat adds the polysialic moieties blocks
hippo-campal learning and plasticity (31). The in-crease in
PSA-NCAM is thought to promotesynaptic remodeling during persistent
forms ofplasticity.
Another family of CAMs that play a role inhippocampal plasticity
includes the synapticallylocalized receptor tyrosine kinase ephrins
andephrin receptors (Eph receptors). Initially studiedin the
context of neural development, ephrins andEph receptors have also
been found to be essen-tial for hippocampal LTP and LTD in the
adultbrain (32). Specific ephrins and Eph receptorsregulate the
localization and function of NMDAreceptors, and can thereby
modulate synapticstrength in response to activity. Experimentsusing
inhibitory ephrin and Eph receptor peptideshave revealed that both
molecules are required,in a kinase-independent manner, for mossy
fiberhippocampal LTP (33).
Trans-synaptic signaling by retrogrademessen-gers. Another means
of trans-synaptic signalinginvolves diffusible, membrane-soluble
messen-gers. The CB1 and CB2 cannabinoid receptorswere initially
identified as receptors for canna-binoid, the active ingredient of
THC/marijuana.This led to the identification of endogenous CB1and
CB2 ligands, called endocannabinoids. Endo-cannabinoids have
emerged as important modu-lators of plasticity, initially at
inhibitory synapses,and more recently at excitatory synapses (34).
De-polarization and activation of a variety of receptorshave been
shown to activate release of endo-cannabinoids from the
postsynaptic compartmentand binding to presynaptic CB receptors,
result-ing in a suppression of neurotransmitter release(and thus
regulating presynaptic plasticity). Thisform of plasticity is
called endocannabinoid-LTD,or eCB-LTD. Endocannabinoid signaling is
re-quired for extinction but not acquisition of spatialmemories
(35).
The Tripartite Synapse: Glia andSynaptic PlasticityOnce thought
of as the support cells of thenervous systems, glial cells are now
consideredessential partners in synapse formation,
synaptictransmission, and plasticity (36). Astrocytes sur-round the
synapse (Fig. 2), forming a tripartitesynapse, composed of neuronal
pre- and post-synaptic compartments as well as
surroundingastrocytes. Synaptically localized glia release
neu-roactive molecules that influence neuronal com-munication. For
example, release of D-serine (acoactivator of the NMDA receptor)
from glia isrequired for LTP of hippocampal Schaffer col-lateral
synapses (37) [although see also (38)].Ephrin and Eph receptor
signaling between neu-rons and glia regulates the uptake of
glutamatethrough glial glutamate transporters and there-by affects
neurotransmission and synaptic plas-ticity (39). The release of
lactate from astrocytesand uptake by neurons has also been
reported
to be required for long-term hippocampal mem-ory and plasticity
(40).
Regulating Gene Expression WithinNeurons During
PlasticitySignaling from synapse to nucleus to
regulatetranscription. Long-lasting forms of synaptic plas-ticity,
such as those underlying long-term memo-ry, require new RNA
synthesis (2). This indicatesthat synaptic signals must be relayed
to the nu-cleus to regulate transcription.
Synapse-to-nucleussignaling poses a unique set of challenges
inneurons, where the distance between the syn-apse and nucleus can
be appreciable. Neuronsare specialized for rapid communication
betweencompartments via electrochemical signaling,
withdepolarization at the synaptic terminal leadingto
depolarization at the soma in less than amillisecond. Calcium
influx can occur throughvoltage- and ligand-gated ion channels.
Cytosol-ic calcium can also be released from intracellularpools
following activation of Gq-coupled recep-tors such as metabotropic
GluRs (mGluRs). Eachroute of calcium influx induces different
programsof gene induction (41).
Soluble signals can also be transported fromthe synapse to the
nucleus by slower, microtubule-andmotor proteindependent pathways
(42). Thisclass of signals includes kinases and transcrip-tional
regulators that function to alter transcrip-tion. These slower
pathways of signaling to thenucleus may sustain changes in gene
expressionfor time periods extending beyond the
initialstimulus.
To obtain a global view of how transcrip-tion is altered during
activity-dependent plastic-ity, expression profiling has been used
to identifychanges in transcription following depolariza-tion of
cultured mouse neurons. Such studieshave identified several hundred
activity-regulatedgenes (41). Genome-wide analyses of
transcrip-tion factor binding sites of the activated geneshave
revealed that the transcription factors CREB,MEF2, and Npas4
control the activity-dependenttranscription of a large number of
downstreamactivity-regulated genes (41). These
downstreamtranscription factors regulate the expression
ofoverlapping but distinct subsets of activity-regulatedgenes,
suggesting that the precise temporal, spa-tial, and
stimulus-specific cellular response isachieved by the combinatorial
control by differ-ent transcription factors.
Local protein synthesis. Despite requiringnew transcription, LTP
and LTD can occur in aspatially restricted manner, raising the
questionof how gene expression in neurons can be lim-ited to
subsets of synapses and not generalizedto the entire cell. One way
of locally changingthe proteome in neurons is through
regulatedtranslation of localized mRNAs (Fig. 4).
The existence of local translation in dendritesof mature neurons
was first suggested by elec-tron micrographic identification of
polyribosomesin hippocampal dendrites (43). Studies in hip-pocampal
slices in which dendrites had been se-
vered from cell bodies found that such dendritesretain the
ability to express long-lasting LTP andLTD, indicating that local
translation can medi-ate long-term modification of synaptic
strength(44, 45).
Studies of mRNA localization have led tothe identification of
cis-acting RNA elementsthat bind to RNA-binding proteins to
undergoexport from the soma into the dendrite (46). Al-though
several dendritic localization elementshave been identified, there
is to date no consensuson their sequence or structure. Among the
best-studied RNA binding proteins involved in den-dritic mRNA
localization are Staufen, Zipcodebinding protein 1 (ZBP1), and
hnRNPA2 (46).These proteins bind cis-acting elements and as-semble
transcripts into larger RNA transportgranules, which travel in a
kinesin-dependentmanner along microtubules to their final
destina-tion. Whether localized RNAs undergo directedtargeting,
anchoring, or stabilization at specificsites remains an open
question.
In terms of translational regulation, studieshave revealed
activity-dependent regulation oftranslation initiation and
elongation. A mech-anism of translational regulation known to
occurat synapses involves the cytoplasmic polyade-nylation element
binding protein (CPEB). CPEBbinding to 3 untranslated regions
(3UTRs) re-presses translation. However, CPEB
undergoesphosphorylation in an activity-dependent man-ner to
recruit other proteins that increase thepolyadenylate [poly(A)]
tails of mRNAs. Sub-sequently, poly(A) binding protein (PABP)
isrecruited to the elongated poly(A) tail, which inturn recruits
eukaryotic translation initiation fac-tor 4g (eIF4G) to interact
with eukaryotic trans-lation initiation factor 4E (eIF4E) to
promotetranslation initiation (47). CPEB localizes tosynapses,
where it regulates translation of den-dritically localized CamKIIa
mRNA (48, 49).
Another activity-dependent means of regu-lating translation
initiation involves phospho-rylation of eIF4E-binding proteins
(4E-BPs).Hypophosphorylated 4E-BPs bind eIF4E andprevent
translation initiation; phosphorylated4E-BP dissociates from eIF4E
and relieves trans-lational inhibition. In neurons, activity
increases4E-BP phosphorylation and stimulates translation(50).
Studies in 4E-BP2 knockout mice foundthat E-LTP stimulation
protocols could induceL-LTP in brain slices. Recently, two
additional4E-BPs have been identified in neurons: neuro-guidin and
the cytoplasmic FMRP interactingprotein (CYFIP). Whereas 4E-BP1 and
2 are be-lieved to affect general translation, these new4E-BPs may
preferentially affect subgroups oftranscripts within dendrites (51,
52).
Activity can also regulate translational elon-gation during
synaptic plasticity. For example, theelongation factor eukaryotic
translation elonga-tion factor 2 (eEF2) undergoes
activity-dependentchanges in phosphorylation. Phosphorylation
ofeEF2 decreases the rate of translation. Whereasaction potentials
decrease eEF2 phosphorylation
www.sciencemag.org SCIENCE VOL 334 4 NOVEMBER 2011 627
REVIEWS
on N
ovem
ber 4
, 201
1w
ww
.sci
ence
mag
.org
Dow
nloa
ded
from
-
(thereby increasing translation), spontaneous re-lease of
neurotransmitter increases eEF2 phos-phorylation and decreases
translation (53). Theseeffects occur locally at synapses,
indicating thatone function of spontaneous release may be
tosuppress local translation and thereby stabilizesynapses.
Translation may also be regulated throughthe microRNA (miRNA)
pathway, where eachmiRNA can potentially regulate hundreds
oftranscripts and hence coordinate the expressionof many genes.
Many miRNAs are relativelymore abundant in, or restricted to, the
brain.While miRNAs can regulate cell-wide levels oftranslation,
their posttranscriptional mode ofaction makes them especially well
suited to reg-ulating distally localized transcripts.
SpecificmiRNAs have been found in dendrites and syn-apses, and
components of the RNA-induced si-lencing complex (RISC) machinery
itself havebeen found to be altered by activity (54).
Consistent with the importance of regulatingsynaptic AMPAR
concentrations during plasticity,the mRNAs encoding GluA1 and GluA2
haveboth been detected in hippocampal dendritesand found to undergo
activity-dependent changesin localization and translation (55, 56).
Furtherlinking local translation with synaptic AMPARabundance,
local eEF2-dependent translationof Arc mRNA has been shown to
trigger endo-cytosis of AMPARs during mGluR-mediatedhippocampal LTD
(57, 58).
Local protein degradation. The local pro-teome is regulated not
only by local translationbut also by protein degradation through
theubiquitin proteasome system (Fig. 4). Bothprotein synthesis and
degradation are requiredfor the maintenance of late-phase LTP,
suggestingthat protein degradation is needed to counter-balance
protein synthesis during plasticity (59).Like local translation,
protein degradation canbe regulated within dendrites. Ubiquitin and
pro-teasomal subunits have been found in dendritesand at synapses,
and stimulation of hippocampalneurons triggers proteasome-dependent
changesin the composition of PSD proteins (60). Activity-dependent
degradation involves redistributionof proteasomes from dendritic
shafts to spines(61). Notably, the ubiquitin proteasome
pathwayalters AMPAR trafficking and degradation atsynapses during
plasticity (62).
PerspectivesAs the above examples illustrate, cell
biologicalapproaches have provided a detailed under-standing of
many aspects of activity-dependentplasticity. By focusing on
molecular processesoccurring within individual neurons and
sub-cellular compartments, we now understand spe-cific processes
that are modulated by experienceto change synaptic strength. These
involve alter-ations in neurotransmitter release,
trans-synapticsignaling, postsynaptic receptor dynamics, andgene
expression within neurons. Distinct plastic-ity mechanisms are used
at different types of syn-
apses. For instance, LTP at mossy fiber synapsesoccurs primarily
through presynaptic changes,whereas LTP at Schaffer collateral
synapses oc-curs mostly through postsynaptic mechanisms.The end
result of many of the processes we havedescribed is to regulate the
concentration of glu-tamate receptors, indicating that this is a
majorpostsynaptic determinant of synaptic strength dur-ing
plasticity.
Together, each of these cell biological mech-anisms provides
potential therapeutic targetsfor diseases in which brain plasticity
is dysfunc-tional. However, they fall short of elucidatinghow
complex circuits are altered by experienceto store information and
alter behavior. This willrequire the development of tools for
investigatingboth the dynamic nano-architecture of the syn-apse and
the neural circuit as a whole. A par-ticular challenge is to study
plasticity in neuralcircuits in living animals, and to
developmethodsto examine, and computational frameworks
tounderstand, how all components of a circuit areregulated to alter
circuit function dynamically.The development of methodologies for
super-resolution time-lapse imaging of synapses, neu-rons, and
circuits in live animals promises tomove the field forward toward a
more nuancedand complete understanding of the experience-dependent
plastic changes in the brain that me-diate learning and memory.
References and Notes1. Y. Dan, M. M. Poo, Physiol. Rev. 86, 1033
(2006).2. E. R. Kandel, Science 294, 1030 (2001).3. F. Cesca, P.
Baldelli, F. Valtorta, F. Benfenati,
Prog. Neurobiol. 91, 313 (2010).4. T. W. Rosahl et al., Nature
375, 488 (1995).5. P. S. Kaeser et al., Cell 144, 282 (2011).6. Y.
Han, P. S. Kaeser, T. C. Sdhof, R. Schneggenburger,
Neuron 69, 304 (2011).7. L. Deng, P. S. Kaeser, W. Xu, T. C.
Sdhof, Neuron 69,
317 (2011).8. P. E. Castillo, S. Schoch, F. Schmitz, T. C.
Sdhof,
R. C. Malenka, Nature 415, 327 (2002).9. S. A. Kirov, K. M.
Harris, Nat. Neurosci. 2, 878
(1999).10. A. Holtmaat, K. Svoboda, Nat. Rev. Neurosci. 10,
647 (2009).11. F. Engert, T. Bonhoeffer, Nature 399, 66
(1999).12. R. C. Malenka et al., Nature 340, 554 (1989).13. R.
Malinow, H. Schulman, R. W. Tsien, Science 245, 862
(1989).14. A. J. Silva, R. Paylor, J. M. Wehner, S. Tonegawa,
Science
257, 206 (1992).15. A. J. Silva, C. F. Stevens, S. Tonegawa, Y.
Wang, Science
257, 201 (1992).16. J. Lisman, H. Schulman, H. Cline, Nat. Rev.
Neurosci. 3,
175 (2002).17. I. Buard et al., J. Neurosci. 30, 8214 (2010).18.
T. C. Sacktor, Nat. Rev. Neurosci. 12, 9 (2011).19. R. Shema, T. C.
Sacktor, Y. Dudai, Science 317, 951
(2007).20. B. K. Andrsfalvy, J. C. Magee, J. Physiol. 559,
543
(2004).21. M. Park, E. C. Penick, J. G. Edwards, J. A.
Kauer,
M. D. Ehlers, Science 305, 1972 (2004).22. P. van der Sluijs, C.
C. Hoogenraad, Semin. Cell Dev. Biol.
22, 499 (2011).23. Z. Wang et al., Cell 135, 535 (2008).24. E.
Schnell et al., Proc. Natl. Acad. Sci. U.S.A. 99, 13902
(2002).
25. S. Tomita, V. Stein, T. J. Stocker, R. A. Nicoll,D. S.
Bredt, Neuron 45, 269 (2005).
26. H. Y. Man, Y. Sekine-Aizawa, R. L. Huganir, Proc. Natl.Acad.
Sci. U.S.A. 104, 3579 (2007).
27. M. D. Ehlers, Neuron 28, 511 (2000).28. H. K. Lee et al.,
Cell 112, 631 (2003).29. J. Boehm et al., Neuron 51, 213 (2006).30.
J. Y. Delgado et al., J. Neurosci. 27, 13210 (2007).31. L.
Bonfanti, Prog. Neurobiol. 80, 129 (2006).32. K. O. Lai, N. Y. Ip,
Curr. Opin. Neurobiol. 19, 275
(2009).33. A. Contractor et al., Science 296, 1864 (2002).34. B.
D. Heifets, P. E. Castillo, Annu. Rev. Physiol. 71,
283 (2009).35. G. Marsicano, P. Lafentre, Curr. Top. Behav.
Neurosci. 1,
201 (2009).36. C. Eroglu, B. A. Barres, Nature 468, 223
(2010).37. C. Henneberger, T. Papouin, S. H. Oliet, D. A.
Rusakov,
Nature 463, 232 (2010).38. C. Agulhon, T. A. Fiacco, K. D.
McCarthy, Science 327,
1250 (2010).39. A. Filosa et al., Nat. Neurosci. 12, 1285
(2009).40. A. Suzuki et al., Cell 144, 810 (2011).41. P. L. Greer,
M. E. Greenberg, Neuron 59, 846
(2008).42. T. H. Chng, K. C. Martin, Curr. Opin. Neurobiol.
21,
345 (2011).43. O. Steward, W. B. Levy, J. Neurosci. 2, 284
(1982).44. H. Kang, E. M. Schuman, Science 273, 1402
(1996).45. K. M. Huber, M. S. Kayser, M. F. Bear, Science
288,
1254 (2000).46. M. A. Kiebler, G. J. Bassell, Neuron 51, 685
(2006).47. J. D. Richter, Trends Biochem. Sci. 32, 279
(2007).48. L. Wu et al., Neuron 21, 1129 (1998).49. D. G. Wells
et al., J. Neurosci. 21, 9541 (2001).50. M. Costa-Mattioli, W. S.
Sossin, E. Klann, N. Sonenberg,
Neuron 61, 10 (2009).51. M. Y. Jung, L. Lorenz, J. D. Richter,
Mol. Cell. Biol. 26,
4277 (2006).52. I. Napoli et al., Cell 134, 1042 (2008).53. M.
A. Sutton et al., Cell 125, 785 (2006).54. S. Banerjee, P. Neveu,
K. S. Kosik, Neuron 64, 871
(2009).55. W. Ju et al., Nat. Neurosci. 7, 244 (2004).56. S. Y.
Grooms et al., J. Neurosci. 26, 8339 (2006).57. M. W. Waung, B. E.
Pfeiffer, E. D. Nosyreva, J. A. Ronesi,
K. M. Huber, Neuron 59, 84 (2008).58. S. Park et al., Neuron 59,
70 (2008).59. R. Fonseca, R. M. Vabulas, F. U. Hartl, T.
Bonhoeffer,
U. V. Ngerl, Neuron 52, 239 (2006).60. M. D. Ehlers, Nat.
Neurosci. 6, 231 (2003).61. B. Bingol, E. M. Schuman, Nature 441,
1144
(2006).62. A. M. Mabb, M. D. Ehlers, Annu. Rev. Cell Dev. Biol.
26,
179 (2010).63. J. Bourne, K. M. Harris, Curr. Opin. Neurobiol.
17, 381
(2007).Acknowledgments: We apologize to those whose
primary work could not be cited because of spaceconstraints. We
thank J.T. Braslow, T. J. ODell,and F. E. Schweizer for comments on
the manuscript;J. Bourne and K. M. Harris for hippocampalelectron
microscopy images; and all membersof the Martin lab for helpful
discussions. Supportcomes from NIH grants R01 MH077022 and
R01NS045324 (to K.C.M.), the Medical ScientistTraining Program (NIH
T32 GM008042), and theNeurobehavioral Genetics Training Program
(NIHgrant T32 MH073526) (to V.M.H.). A glossary of termsused is
included in the online (HTML) version of thisarticle (Box 1).
10.1126/science.1209236
4 NOVEMBER 2011 VOL 334 SCIENCE www.sciencemag.org628
REVIEWS
on N
ovem
ber 4
, 201
1w
ww
.sci
ence
mag
.org
Dow
nloa
ded
from