Table Of ContentHYPOTHESISANDTHEORYARTICLE
INTEGRATIVE NEUROSCIENCE published:04January2013
doi:10.3389/fnint.2012.00126
Transsynaptic modality codes in the brain: possible
involvement of synchronized spike timing, microRNAs,
exosomes and epigenetic processes
JohnSmythies1*andLawrenceEdelstein2
1CenterforBrainandCognition,UniversityofCaliforniaSanDiego,LaJolla,CA,USA
2MedimarkCorporation,DelMar,CA,USA
Editedby: This paper surveys two different mechanisms by which a presynaptic cell can modulate
SidartaRibeiro,FederalUniversityof thestructureandfunctionofthepostsynapticcell.Wefirstpresenttheevidencethatthis
RioGrandedoNorte,Brazil occurs,andthendiscusstwomechanismsthatcouldbringthisabout.Thefirsthypothesis
Reviewedby: relates to the long lastingeffects that the spikepatterns of presynaptic axons may exert
ClaudioV.Mello,OregonHealthand
by modulating activity–inducible programs in postsynaptic cells. The second hypothesis
ScienceUniversity,USA
MarcosR.Costa,FederalUniversity is based on recently obtained evidence that, the afferent neuron at the neuromuscular
ofRioGrandedoNorte,Brazil junction buds off exosomes at its synapse and carries a cargo of Wg and Evi, which
*Correspondence: are large molecular transsynaptic signalingagents (LMTSAs). Further evidence indicates
JohnSmythies,CenterforBrainand that many types of neurons bud off exosomes containing payloads of various lipids,
Cognition,UniversityofCalifornia
proteins, and types of RNA. The evidence suggests that they are transmitted across
SanDiego,9500GilmanDrive,
LaJolla,CA92039,USA. the synapse and are taken up by the postsynaptic structure either by perisynaptic or
e-mail:[email protected] exosynapticmechanisms,thusmediatingthetransferofinformationbetweenneurons.To
date, the molecular hypothesis has been limited to local interactions within the synapse
ofconcern.Inthispaper,weexplorethepossibilitythatthisrepresentsamechanismfor
information transfer involving the postsynaptic neuron as a whole. This entails a review
of the known functions of these molecules in neuronal physiology, together with an
estimateofthepossibletypesofinformationtheycouldcarryandhow theymightaffect
neurocomputations.
Keywords:molecularcodes,synchronizedoscillations,signalingmolecules,microRNAs,exosomes
INTRODUCTION circumstances during later development into adulthood these
Inthispaperweaddressthegeneralquestionofhowpresynaptic connectionsmaintaintheiroverallpatternalthoughsubjecttoa
cellscanexertlong-lastinginfluencesontheirprojectiontargets dealofsynapticplasticity(Smythies,2002).Thedegreetowhich
and thus affect the modality identity and properties of specific thisismaintainedduringlaterstagesofgrowthbysignalingand
cortical regions. To answer this, we propose two general solu- epigenetic factorstransmitted fromthe presynaptic to the post-
tions.Thefirstisthat,thespikepatternsofthesynapticaxonsmay synapticneuroniscurrentlyunknown.However,underextreme
exert long-lasting effects by modulating activity–inducible gene conditions of massive deafferentation and reafferentation, this
expression programs in postsynaptic cells, with different spik- system can undergo more extensive changes. For example, in
ingpatternspossiblyelicitingtheexpressionofdifferentcohorts experimentsonblindsubjectsskilledinBraille.Ptitoetal.(2008)
ofepigenetic signalingmolecules. The second isthat, presynap- showed that magnetic transcranial stimulation of neurons of
ticcellsexerteffectsthroughthetranssynaptictransmissionofa loci in the optic cortex results in a somatosensory, and not a
varietyofsignalingmolecules,possiblythroughexosome-related visual, experience in these subjects (Ptito et al., 2008). In these
mechanisms.Wewillstartbyconsideringtheevidencethatpresy- cases, some differentiated “visual” cells are “taken over” by the
napticcellsdoinfluencepostsynapticcellsinthismanner. somatosensorysystem,andstarttoprocesssomaticinformation
instead.Thisactivitygeneratessomatosensorysensationsincon-
THEROLEOFTHEPRESYNAPTICNEURONINDETERMINING sciousness(sensationsthatthefingersarebeingtouched)inplace
MODALITY:THEEFFECT OFDEAFFERENTATIONAND of the normal visual sensations. These authors conclude: “Our
REAFFERENTATION data show that the qualitative character of the subject’s experi-
We can approach this topic by asking, what determines the enceisnotdeterminedbytheareaofcortexthatisactive(cortical
modality of a sensory neuron e.g., whether activation of such dominance),butbythesourceofinputtoit(corticaldeference).”
a neuron leads to a visual or to a somatosensory experience. At a functional level, a deafferented cortex (e.g., the visual
During embryogenesis, the newly differentiated cortical areas cortex in the blind) can take over functions of another sensory
each secrete specific attractant molecules that guidethe incom- modality(e.g.,hearing),butonlyifthefunctionsofthetwoare
ing thalamic relay axons to the correct location. In normal homologous.Thatis,spatialhearingfunctionsareimproved,but
FrontiersinIntegrativeNeuroscience www.frontiersin.org January2013|Volume6|Article126|1
SmythiesandEdelstein Transsynapticspikeandmolecularcodes
not tone discrimination (Lomber et al., 2010). If the auditory not include, for our present purposes, transcription factors nor
cortexisdeafferented asinthedeaf,thenvisualspatialdiscrim- processes likemethylation and acetylation. In this paper,more-
ination is improved, but not color functions. By employing an over,wehavefocused,notonmodalitycodeadaptionstoextreme
experimentaldesignthatallowsindependentcontrolofbothspa- conditions,buttotheirnatureunderordinarybrainactivities.
tial and contextual correspondence, Heronet al.(2012) showed
that observers are able simultaneously to adapt to two oppos- DIFFERENTIATIONOFNEURONALCHANGESFOLLOWING
ingtemporalrelationships,providedtheyaresegregatedinspace. ACUTEANDCHRONICDEAFFERENTATION
Nosuchrecalibrationwasobservedwhenspatialsegregationwas Acute deafferentation induces functional changes in the brain’s
replacedbycontextualstimulusfeatures(inthiscase,pitch,and networks, whereas chronic deafferentation also results in struc-
spatialfrequency).Theauthorssuggestthattheseeffectsprovide tural changes. Rerouting sensory pathwaysinthe braininvolves
supportfordedicatedasynchronymechanismsthatinteractwith sensory deafferentation since one part of the cortex is deprived
spatially selective mechanisms early in visual and auditory sen- ofitsnormalinput.Certaincasesofsensorydeafferentationalso
sory pathways. However, this account of “partial” restructuring involve transmodal rerouting of pathways, as when the deaf-
leavesunansweredthequestionofwhetherthereafferentedcortex ferented neurons get invaded by axons belonging to another
shouldbe regarded ashavinganew modality, orhavingunder- modality.Thishasbothimmediateanddelayedeffectsonthesen-
goneamodulationofitsmodality.Weprefertoleavethismatter sorysysteminthebrainconcerned.Theimmediateeffectsaredue
openatthisstage.Ineitherevent,thedataindicatesthatthepresy- todynamicchangesinbrainnetworks.
napticneuronexertsapowerfulinfluenceoverthestructureand Delayedeffects,however,involvemuchrewiring.Recentstud-
functionofthepostsynapticneuronthatmustbetransmittedby ies show that, in the long term, substantial reorganization in
someformofcodeoragent. subcortical structures, including the brainstem and thalamus,
Anothersourceofinformationaboutsensorymodalitydeter- occurs that maybe of sufficient extent to account for, orplaya
mination comes from sensory input rerouting [diverting the largepartin,representationalplasticityinsomatosensorycortex
sensoryinflowofonesystemtothecortexofanothersystemby (Jones, 2000). Extensive interhemispheric corticocortical reor-
neonatal diversion of e.g., retinal axons to the auditory thala- ganization can occur in the rodent brain following peripheral
mus(cross-modal rewiring)]. This operation leadsto profound nervedeafferentation(Pelledetal.,2007).FMRIdatashowsthat
changesondiversecomponentsofcorticalcircuitry, bothatthe long-term reorganization of the somatosensory cortex, follow-
anatomicalandfunctional levels(GaoandPallas, 1999;Sharma ing spinal cord injury in humans, is associated with changes in
etal.,2000).Sensoryinputreroutingcanalsoleadtochangesin local cortical anatomy and provide “compelling evidence” that
intracolumnar information processing in the postsynaptic neu- such reorganization in humans results from the growth of new
ron. For example, if the input to an auditory neuron in the lateral connections from adjacent cortex into the deafferentated
cortexisreplacedbyavisualinput,thentheresponsecharacter- portion,andnotsimplyfromtheunmaskingofalreadyexisting
istics of information processing in its cortical columns changes lateralconnections(Hendersonetal.,2011).
so asto mimic the systems employed in a visual neuron. These Olfactory deafferentation induces whisker tactile hypersensi-
auditory cortical cells develop visual response properties such tivity. In studies in mice 1 week after olfactory deafferentation,
as direction selectivity, orientation tuning, and simple/complex Nietal.(2010)showedthat,thereresultsarecruitmentofmore
receptive-field structure (Roe et al., 1992). The auditory cortex GABAergic neurons and their fine processes in the barrel cor-
alsodevelopsretinotopicmaps[Roeetal.,1990;andseeLinden tex,aswellasanup-regulationoftheircapacitytoencodeaction
and Schreiner (2003) for a comprehensive review]. Rerouting potentials. The hyperpolarization driven by inhibitory inputs
visual inputs to the auditory thalamus can also reorganize cal- strengthenstheencodingabilityoftheirtargetcells.
losalconnectionsintheauditorycortex,causingbothareduction Churchill et al. (2004) conducted Golgi studies in the
intheir extent anda reorganizationof the pattern(Pallaset al., somatosensorycortexinprimatesfollowingdeafferentation.They
1999). Chowdhury and De Angelis (2008) have extended the showedthat,afterdenervation,thereisasystematicchangeinthe
range of cortical plasticity by showing, in depths perception dendriticarborizationpatternofbothlayerII/IIIpyramidaland
discrimination experiments, that the contribution of particular layerIVspinystellatecellsinthecontralateralhandregionofarea
brain areas to task performance can change dramatically as a 3b, compared to unaffected cortical areas. This was marked by
resultoflearningnewtasks. a progressive expansion of distal regions of the dendritic arbor,
These changeswould appearto be the resultof somesignal- bothbasilarandapical,withnoappreciablechangesproximally.
ing system in the afferent axons. This implies that these axons Recently considerable attention has been paid to the role of
are carrying a code that is most probably contained either in single neurons, particularly their dendritic arbors, in informa-
thespatiotemporal patternsofitsspiketrains, orbysomeform tion processing (London and Häusser, 2005; Gollo et al., 2009;
of molecular or epigenetic signal, or both. The details of how Klausberger, 2009; Branco et al., 2010). In particular a paper
thesemodality-relatedcodessystemshavesuchremarkableeffects by Legenstein and Maass (2011) is relevant to the subject of
are not currently known. The aim of this present paper is to this review. They propose that non-linear processing in den-
enquirewhatformthiscodemighttakeandhowitcouldbring dritic branches endows individual neurons with the ability to
abouttheextensiveeffectsonthepostsynapticneuronreported. perform complex computational operations necessary to solve
We should make it clear by “epigenetic factor” here we mean forexamplethebindingproblem.Theyinvestigated howexper-
firstordermolecules,suchasspecificproteinsandRNAs.Wedo imentally observed plasticity mechanisms in dendritic arbors,
FrontiersinIntegrativeNeuroscience www.frontiersin.org January2013|Volume6|Article126|2
SmythiesandEdelstein Transsynapticspikeandmolecularcodes
such asdepolarization-dependentspike-timing-dependent plas- in barrel cortex to 15ms in visual cortex. They concluded that,
ticity and branch-strength potentiation, could be integrated to differentcorticalareasareadaptedtothespecificstructureofthe
conductself-organizednonlinearneuralcomputationswithden- inputsignalstheyprocess,andthatprecisespiketimingmayplay
driticspikes.Itseemspossible,therefore,thatchangesindendritic amoreimportantroleforsomecorticalareasthanforothers.
arborneurocomputationsmaybeinvolved.Changesinthespa- Thereisalsoaneedtofindouthowthesespikecodesproduce
tial arrangementofinterlaminar glia maybeanintegral partof theirtranssynapticeffects.Wecouldfindnopaperspublishedthat
thelong-termprocessofstructuralreorganizationofthecerebral tacklethissubject.
cortex following cortical deafferentation (Reisin and Colombo,
2004). EVIDENCEFORTHEACTIVITYOFLARGEMOLECULAR
TRANSSYNAPTICSIGNALINGAGENTS(LMTSAs)IN
SPIKETRAINMODALITYCODES NEURALCOMMUNICATION
Spikes,andburstsofdifferentdurations,codefordifferentstim- Many types of cells, including neurons, bud off exosomes from
ulus features. The biophysical mechanism of spike generation their plasma membranes into the extracellular environment.
enablesindividualneuronstoencodedifferentstimulusfeatures Exosomesaresmalllipoproteinvesiclesderivedfromtheintralu-
into distinct spikepatterns(KepecsandLisman,2003). Cortical minal membranesofmultivesicularbodies (MVB)ofthe endo-
regular-spikingneuronscanpropagatefilteredtemporalinforma- cytotic pathway. They are expelled into the extracellular space
tion in a reliablewaythroughthe network, and with high tem- uponfusionoftheMVBwiththeplasmamembrane(Fauréetal.,
poralaccuracy(Asaiet al., 2008). Sincethe changes induced by 2006).Endocytosisisasimilarprocess,actingintheotherdirec-
deafferentationcantransferacrosssynapses(aswhenanewinput tion.Thistransportsmanyactivatedmembranereceptorsintothe
to thethalamusinducessuchchangesinthecortex)themecha- cell (Smythies, 2002). However, for long the only form of exo-
nismthatdoesthismustinvolvetranssynaptictransfer.Thebrain cytosis recognized in neurons was the familiar synaptic vesicle
isahighlymodularstructure.Therefore,spikingactivitymustbe neurotransmitter and neuromodulator system. Exosomes carry
abletopropagatefromonemoduletoanotherwhilepreserving molecules between cells. In a recent review, Tetta et al. (2012)
theinformationitcarries(Kumaretal.,2010).Ifthebrainuses state “Extracellular vesicles, including exosomes and microvesi-
spike timing as a means of information processing, other neu- cles,maydeliverlipidsandvariousfunctionaltranscripts,released
ronsreceivingspatiotemporalspikesfromsuchsensoryneurons fromthecelloforigin,totargetcells.Sinceextracellularvesicles
must also be able to treat information included in the inter- containdefinedpatternsofmRNA,microRNA,longnon-coding
spikeintervals(Masudaand Aihara,2002). Significantamounts RNA,andoccasionallygenomicDNA,theymaytransfergenetic
of visual information are represented with high precision by information which induces transient or persistent phenotypic
detailsofthespiketrainatmillisecondandsub-millisecondpre- changes in recipient cells”. In another recent paper, O’Loughlin
cision(Nemenmanetal.,2008).Diesmannetal.(1999)statethat etal.(2012)state“Exosomesplayanimportantroleinendoge-
precisely synchronized action potentials can propagate within a nouscell-to-cell communication... [andhavebeen]... shown
modelofcorticalnetworkactivitythatrecapitulatesmanyofthe to be capable of traversing biological barriers and to natu-
featuresofbiologicalsystems.Axonscancarrymultiplecodesin rallytransportfunctionalnucleicacidsbetweencells.”Kolesand
the spatiotemporal patterns of their spike trains (Kayser et al., Budnik (2012) conclude, “Exosomes, small secreted microvesi-
2009). Singer (2009) proposes that axons can carry two mes- cles,areimplicatedinintercellularcommunicationindiversecell
sagesinparallel.Thefirstindicatesthepresenceofthefeatureto types, transporting protein, lipid, and nucleic acid cargo that
whichtheirneuronsaretuned.Thesecondconveystheinforma- impact the physiology of recipient cells.” Although the transsy-
tionwithwhich otherneurons(specifictarget cellsormembers naptictransportofsuchmoleculesinthecaseofneuronshasbeen
ofacoherentassembly)theyarecommunicating.Thefirstmes- experimentally limited so far as to signaling proteins (Wg and
sage is carried by a rate code. Singer proposes that the second Evi), it seems highly unlikely that neurons would be exceptions
code is a function of the precise timing relationships between to the general rule that exosomes transport a variety of nucleic
individualspikesofdistributed neurons(temporal code). These acids as well. We will use the term large molecular transsynap-
relationsareestablished,hesuggests,eitherbythetimingofexter- ticsignalingagents(LMTSAs)todefinethesecargoescarriedby
nalevents(stimuluslocking),orbyinternaltimingmechanisms exosomes. Potential LMTSAsincludelipids,trophins, and mor-
basedonanoscillatorymodulationofneuronalresponsesindif- phogeneticproteins,mRNAandmicroRNAs,butnotperinuclear
ferentfrequencybands.Therefore,weneedtodiscoverwhatthe transcriptionfactors.Thefirstlineofenquirythatwewillfollow
individual spike codes are that carry the modality information. istoask—whatfunctiondotheseagentshaveinneurons?
We could find only one paper in the literature on this topic.
Adirectinvestigationofmodalcodesinthecortexhasshownthat PROTEINS
neuronalspikingpatternsareregularinmotorareas,randomin Afferentaxonscouldinitiatetranssynapticmodulationbysecret-
thevisualareas,and“bursty”intheprefrontalarea(Shinomoto ingagentssimilartoneuroserpinanddoublecortin.Neuroserpin
etal.,2009).Supportingevidenceinpartforourhypothesishas isanaxonally-secretedproteinmemberoftheserpinsuperfamily
beenobtainedbyYangandZador(2012),whocarriedoutexper- ofserineproteaseinhibitors,andiswidelyexpressedthroughout
imentsonratsmeasuringtheirabilitytodiscriminateminimum the cerebral cortex, hippocampus and amygdala (Berger et al.,
timingdifferencesofelectrical stimulideliveredtodifferentcor- 1999; Lee et al., 2012). These authors report that neuroser-
ticalmodalities.Theyfoundwidedifferencesrangingfrom1ms pin mRNA is increased in cultured hippocampal neuronsupon
FrontiersinIntegrativeNeuroscience www.frontiersin.org January2013|Volume6|Article126|3
SmythiesandEdelstein Transsynapticspikeandmolecularcodes
depolarizationbymeansofelevatedextracellularKCl.Theyfur- Fauré et al. (2006) found that, exosomes released by cortical
theroutlineaneurochemicallinkbetweenanartificiallyinduced neuronscontaintheL1celladhesionmolecule,theGPI-anchored
depolarizationandneuralplasticity,whichcouldbefollowed,in prion protein, and the GluR2/3 but not the NR1 subunits of
the case of a natural synapse, if the afferent axon secreted neu- glutamate receptors. They also found that exosomal release is
roserpinorasimilaragent.Neuroserpinhasalsobeenfoundto regulated by K+-induced depolarization. They concluded that
modulate the growth and shape of axons and dendrites in the exosomes might have a regulatory function at synapses. In a
hippocampus (Borges et al., 2010). The doublecortin family of later paper using mature cortical neurons in culture, Lachenal
proteinsmodulatemicrotubularfunctionindevelopingneurons et al. (2011) observed exosomes being released from the soma-
(Dijkmansetal.,2010). todendritic compartment of neurons. They also found that this
Additionalproteinsofinterest(e.g.,Licelladhesionmolecule, process was modulated by glutamatergic synaptic activity, and
GPI-anchored prion protein, Glu 2/3 receptor subunit, brain- similarly concluded that exosomes might take part in normal
derived neurotrophic factor, neurotrophins) will be discussed synaptic activity. In the most recent paper from this group at
below. INSERM,Chivetetal.(2012)concluded:“Exosomescouldthus
representanidealmechanismforinterneuronaltransferofinfor-
THEROLEOFmicroRNAsANDEXOSOMES mation.” This is a conclusion with which we agree. Recently,
Wefeelthatpromisingcandidatesfortheroleofmolecularcarri- Turola et al. (2012) state: “Microvesicles (MVs) [a.k.a. exo-
ersofinformationcodessofarliesintherecentlydevelopedfields somes] are released from almost all cell brain types into the
of microRNAs. There is now considerable data to indicate that microenvironmentandareemergingasanovelwayofcell-to-cell
microRNAs modulate a number of neuronal functions. To give communication.”
someexamples: The L1 cell adhesion molecule, found in synaptic exosomes
by Fauré et al. (2006), has many functions that might link its
− Cohen et al. (2011) have identified a developmentally and appearance in the postsynaptic neuron with structural modal-
activity-regulated microRNA (miR-485) that controls den- ity modulation. Forexample, it is involved in axonguidance; it
dritic spine number and synapse formation in an activity- alters the expression of transcription factors in murine neocor-
dependenthomeostaticmanner. tex (Kishimoto et al., 2012); it facilitates dendritic and axonal
− Transcription ofthe microRNAmiP335 ispromoted by nat- compartmentalization (Winther et al., 2012); it regulates the
urallyevokedsynapticactivity attheclimbingfiber-Purkinje developmentofseptalcholinergicneurons(Cuietal.,2011);peri-
cellsynapseinthemousecerebellarflocculus. somaticGABAergicinnervationinprefrontalcortexisregulated
ThetargetmRNAsforthismicroRNAhavebeenidentifiedas byankyrininteractionwiththeL1celladhesionmolecule(Guan
calbindinand14-3-3-theta(Barmacketal.,2010). and Maness, 2010); it acts transcellularly to promote synaptic
− Impey et al. (2010) report that, neuronal activity regulates maturationontheneuronsinculture(Triana-Baltzeretal.,2008);
spine formation, in part, by increasing miR132 transcrip- andacloserelative,neuralrecognitionmoleculeclosehomologof
tion, which in turn activates a Rac1-Pak actin remodeling L1(CHL1),hasbeenshowntoregulatetheorientationofapical
pathway. dendritesinthemousecortex(Yeetal.,2008).
− MicroR-181a activity in primary neurons, induced by Smalheiser(2007,2009)suggestedthat,exosomesareinvolved
dopamine signaling, is a negative post-transcriptional regu- in much transsynaptic activity. He based his hypothesis on the
lator of GluA2 expression (Saba et al., 2012). Additionally observation that exosomes contain a mixture of proteins and
these authors report that, miR-181a overexpression reduces RNAsincludingmRNAsandmicroRNAs(Ratajczaketal.,2007;
GluA2 surface expression, spine formation, and miniature Valadi et al., 2007). Furthermore, exosomes express cell recog-
excitatory postsynaptic current (mEPSC) frequency in hip- nition molecules ontheir surface, which facilitates selective tar-
pocampalneurons.ThusmicroR-181acouldregulatesynaptic geting and their uptake into recipient cells. This led him to
function. suggest thatexosomal secretion ofproteinsand RNAsmaybea
− Utilizing a mouse line with a conditional neuronal deletion fundamental mode of communication within the nervous sys-
of Dgcr8, a microRNA biogenesis protein predicted to tem, supplementingtheknownmechanismsofanterogradeand
process microRNAs exclusively, Hsu et al. (2012) produced retrogradesignalingacrosssynapses.
evidence that some microRNAs govern essential aspects of Smalheisergoesontosay:“Inonespecificscenario,exosomes
inhibitory transmission and interneuron developmentin the areproposedtobudfromthelipidraftregionofthepostsynaptic
mammaliannervoussystem. membraneadjacenttothepostsynapticdensity,inamannerthat
is stimulated by stimuli that elicit long-term potentiation. The
MoststudiesofmicroRNAsinneuronshaveconcentratedon exosomes would then transfer newly synthesized synaptic pro-
the effects of these agents on process in the same neuron that teins (such as CAM kinase II alpha) and synaptic RNAs to the
producedthem. However,itispossiblethatmicroRNAs,aswell presynaptic terminal, where they would contribute to synaptic
asother factors, maybe exported from one neuron by synaptic plasticity.”
transfer to other neurons. One mechanism may be by syncy- Note that here he is suggesting that the flow oninformation
tia composed of gap junctions. Another mechanism has been isfromthepostsynapticneuronbacktothepresynapticneuron.
suggested in trail breaking papers by Fauré et al. (2006) and Thismaywellbethecaseforcertainfunctions.However,inaddi-
Smalheiser(2007,2009)relatingtoexosomes. tion,wesuggestthat,toexplainfullytheactionofLMTSAcodes,
FrontiersinIntegrativeNeuroscience www.frontiersin.org January2013|Volume6|Article126|4
SmythiesandEdelstein Transsynapticspikeandmolecularcodes
theflowmayalsobeintheotherdirection.Weproposethatthe glia. Moreover, as far as we are aware there have been no elec-
presynapticneuronmaygenerateanumberofLMTSAs,possibly tron microscope detections of structures resembling exosomes
including microRNAs, which may be transported by the exoso- in between neurons. This suggests that the most likely path-
malsystemintothesynapticcleft,andthencemaybetakenupby wayfortheexosomestraffickingbetweenthepresynapticneuron
endocytotic mechanisms into the postsynaptic neuron. In that, andthepostsynapticneuronmightbeviaglia.Thereareseveral
theymayexerttheepigeneticactionsassociatedwithanumberof reportsofexosomesreactingwithglia(Kramer-Albersetal.,2007;
different types of information processing. In the normal course Guescini et al., 2010; Frühbels et al., 2012). Oligodendrocytes
ofeventsthesefactorswouldrefinethemodeofactionofexisting activated by glutamate secrete exosomes that contain proteins,
circuits. mRNAsandmicroRNAs. Theseexosomesaretakenupbyadja-
centneuronsbyaclathrin-dependentmechanism(Frölichetal.,
REVIEWOFTHEEVIDENCEFORTHETRANSSYNAPTIC 2013). Astrocytes wrap round synapses and interact with them
TRANSPORTOFLMSAsINNEURONS to form what have been called “tripartite synapses” (Haydon,
The evidence that exosomes transport lipids, proteins, a vari- 2001; Gordleeval et al., 2012). To this has been added possi-
ety of RNAs, and even DNA between cells already exists (Koles ble involvement of the intercellular matrix to form tetrapartite
and Budnik, 2012; O’Loughlin et al., 2012; Tetta et al., 2012). synapses (Dityatev and Rusakov, 2011). These offer avenuesfor
Directevidencefortheexistence ofLMSTAsspecificallyinneu- theinterneuronaltransportofLMTAsandmicroRNAs.
rons is presented in the following reports. Menna et al. (2003) Interneuronaltransportofexosomescouldalsobeconducted
haveshowninneonatalratsthatbrain-derivedneurotrophicfac- via gap junctions. There are extensive similarities between neu-
tor (BDNF) is produced by retinal ganglion cells (RGCs) and ronsandelongatedfibercellsthatmakeupintheinteriorofthe
travels in an anterograde direction along the optic nerve. This ocularlens(Frederikseetal.,2012).Electronmicrographsshow
results in modulationof the microanatomy of their synapses in similaritiesbetweentheorganizationoftheirintracellularvesicle
thelateralgeniculatenucleus.Instudiesofthedevelopingchick transport machinery and between lens fiber cell lateral protru-
brain, VonBartheld et al.(1996) report that, neurotrophins are sionsanddendriticspines.Gruijters(2003)reviewsevidencethat
transportedintheanterogradedirection,fromcellbodiestothe intercellularvesicletransportinthelens,possiblycarryinglarge
axonterminals,andthattheintactneurotrophinisreleasedafter molecules, is mediated via gap junctions. There isalso the pos-
anterogradetransport,takenupandutilizedbythepostsynaptic sibility that membrane-bound proteins on the exosome could
neuronviaaxo-dendriticcontacts.Theyconclude“Theseresults bind to complementary receptors on the postsynaptic neuron
suggestthatanterogradelytransportedneurotrophinsmayplaya thatwouldtransferthesignaltotheinteriorbyaconformational
roleinsynapticplasticityandmayhaveeffectsatmorethanone change.
synapsebeyondtheinitialreleasesite.”Thesetworeportsmention
anterograde transport of neurotrophins leading to transsynap- MEMBRANEUTILIZATIONDYNAMICS
tic communication, but do not specifically involve exosomes. Our hypothesis places particular significance on the interneu-
However, exosomes would seem to be the most likely candi- ronaltransportofexosomes,constructedoflipoproteincellmem-
dates to transport large molecules across the synaptic cleft. The branes,ineitherananterograde,orretrogradedirection,orboth,
third report provides evidence that this process actually takes carried by exosomes, which consist of membrane. This process
place. At the Drosophila neuromuscular junction the signaling is subject to membrane utilization dynamics. In the postsynap-
morphogenic protein Wg is transferred inside exosomes (bud- tic neuron the process of endocytosis of receptors is a balanced
dedfrommultivesicularbodies)throughbindingtotheexosomal dynamic process (Smythies, 2002). Upon binding most neuro-
proteinEvi.Theexosome,withitsWgload,isthenreleasedfrom transmitter, orneuromodulatormolecules,thereceptorispack-
thepresynapticterminalto betakenupbythepostsynaptictar- aged into a fold of the surrounding membrane. This pinches
get, in this case muscle cells (Koles and Budnik, 2012: see their off to form a sack that is then trafficked to the endosome sys-
Figure1). tem inside the cell. Here the load molecule is extruded into the
Theauthorssuggestthatthesiteofreleaseisprobablynotthe endosome cavity and the sack fuses with the endosome mem-
activesiteofthesynapse,whichisspecializedforneurotransmit- brane. The load molecules are then subject to a triage process.
terrelease,butintheperiactiveregionaround.Thereisalsothe In this, damaged (oxidized) molecules are routed to lysosomes
possibilitythattransfercouldbemediatedbyextrasynapticmech- and are broken down, and their components recycled. The rest
anisms that are widely spread throughout the brain (Smythies, areenclosedinendosomemembranetoformmoresacks,which
2002). There is another consideration. This release at the neu- are recycled to the surface. Here the load molecule is inserted
romuscular junction is into the subsynaptic reticulum, which into the plasma membrane, and the sack fuses with the plasma
is a series of cisternae peculiar to the neuromuscular junction. membraneitself.Inthiswaymembranesarecontinuallyrecycled
However,theactivesiteattheinterneuronalsynapse,whichlacks so cuttingthe expensive processofsynthesizing newmembrane
anysubsynapticreticulum,isonly20nmwide,whereasthediam- to a minimum. In contrast, at the presynaptic axon terminal,
eterofanexosomeis50–90nm.Sothereisnoroomattheactual theprocessofneurotransmitterorneuromodulatorreleaseisnot
synapse for any such transfer. Therefore, if such transfer takes affected by the extrusion of membranous vesicles. The vesicle
placeatthe interneuronalsynapse, it mustdoso atextrasynap- extrudes its payload at the cell surface and the vesicle is inter-
ticregions.Theproblemisthatthereisnorealextracellularspace nally recycled. Again synthesis of new membrane is kept to a
in the brain. The space between neurons is effectively filled by minimum.
FrontiersinIntegrativeNeuroscience www.frontiersin.org January2013|Volume6|Article126|5
SmythiesandEdelstein Transsynapticspikeandmolecularcodes
Bearingtheseconsiderationsinmind,ifweconsiderthetrans- this does not rule out the possibility that, later in develop-
fer of exosomes across the synaptic cleft, possibly via glia, it ment, sensory modality can be modulated by a change in the
becomes obvious that portage in only one direction, whether sensory afferent input. In this paper, we are dealing only with
it is anterograde (pre to post) or retrograde (post to pre) will postembryonicstagesofcorticalplasticity.
lead to an unbalanced system. The donor cell will be continu- The parcellation of the cortex into functionally and struc-
ously depleted of membrane, whereas the recipient cell will be turally discrete areas involves what Sur and Rubenstein (2005)
continuously swamped. Furthermore, each vesicle coat is only describeas“... aninterwovencascadeofdevelopmentalevents
usedonce,whichisanenormouslywastefulprocess.Abalanced includingbothintrinsicandextrinsicelements.”Thecorticalpro-
system requires approximately equal traffic in each direction. genitor zonecontainsthe informationthatgenetically generates
So what loads could these two sets of vesicles carry between corticalarea,whereas,lateron,thalamicafferentaxons,through
neurons?Intheanterogradedirectionitwouldbelogicaltosug- activity dependent mechanisms, may impose further cortical
gest the load includes epigenetic factors we have suggested for differentiation via, in part, epigenetic mechanisms. Therefore,
the modulation of local and cell wide information processing although the plastic processes occurring during critical devel-
mechanisms. But, what function would the retrogradely deliv- opmental periods may be different from those occurring after
eredmoleculesperform?Whatinformationdoesthepresynaptic deafferentation, further investigations should enable us to dis-
neuronneedfromthepostsynapticneuron?Cluesaresuggested tinguishbetweenthem.SurandRubensteindescribemanytran-
by the finding by Fauré et al. (2006) that neurons grown in scriptionfactors, adhesionmolecules,axonguidancemolecules,
tissue culture (that presumably act mainly as postsynaptic neu- etc., involved in cortical plasticity. They add, “The patterning
rons being composed mainly of soma) export L1 cell adhesion centers operate in part through generating graded expression
molecules, the GPI-anchored prion protein, and the GluR2/3 of the transcription factors that control histogenetic programs
but not the NR1 subunits of glutamate receptors. The first two forproliferation,neurogenesis, migration,connectivity, andcell
can act as adhesions molecules (among many other functions), death/survival.”Theyconclude,“Activityoperatesthroughmod-
which might seem appropriate,butwhy subunits of AMPAbut ulating the expression and function of almost the entire range
not NMDA receptors? Activated AMPA receptors are endocy- of molecules responsible for neuronal and synaptic function.”
tosedintothepostsynapticneuronbutNMDAreceptorsarenot Further reviews of this topic have been published by Krubitzer
(Lissin et al., 1999). But it is difficult to see why that should (2007)andRakic(2009).Beaudetal.(2012)havedescribedhow
be relevant. One explanation is that this pathway simply gets rewiring in the spinal cord can explain plasticity after periph-
rid of unwanted molecules. Another explanation might be that eral injuries in adult monkeys. Spontaneous electrical activity,
exosomes represent a supplementaryretrograde supplyroute of present at the earliest stages of cortical development, can also
key synaptic components to the site of activity. The antero- modulatethedevelopingcorticalstructure(SurandRubenstein,
grade route from the cell soma to its own active synapses via 2005).
its own axons is often long and slow. The transport route for The exosome-based system we are suggesting would have
the same molecules from the soma of the postsynaptic neuron its dynamic aspects. Activity in the system would normally
to the same location via exosomes would be much shorter and be sufficient to meet the needs of replacing oxidized proteins
faster.ThismightexplainwhyAMPA,butnotNMDA,subunits removedbytheendosometriagesystem.Thenitshouldbecome
are trafficked by exosomes. Since the NMDA receptors are not more active during learning processes to oversee the synthe-
endocytoseduponactivation,theirlifespaninthemembraneis sis of modality-specific new proteins involved in the learn-
much longer that the life span of AMPA receptors. Smalheiser ing process. This reactivation may be reflected in the reported
(2007, 2009) suggests a plausible function for retrograde exo- finding that rates of exosome release are increased by depo-
somesintermsofmodulationofsynapticplasticity.Thisrequires larization of the membrane. A higher rate of activation
the retrograde exosome to carry molecules like CAM kinase II could take place after major shocks such as deafferentation
alpha and mRNAs. However, such molecules were not detected andreafferentation.
by Fauré et al. (2006). This suggests further experiments to
lookforthem. THEMOLECULARBASISOFBINDING
We define the binding problem here as defining the mecha-
DEVELOPMENTALFACTORSINTHERELATIONSHIP nism by which the brain combines information generated by
BETWEENEPIGENETICCODESANDNEURALSTRUCTURE single modality systems into multimodal operations that must
Ofcourse,themicrostructureofthebraincannotbeunderstood underlie the unitary percepts we experience. We do not intend
solely in terms of the postnatal activity of LMTSAs of various to cover every aspect of the physiological basis of binding, but
kinds.Duringembryogenesisthedifferentialmodalityofthecor- to focus upon a new and interesting development. We admit
tex is determined by classical morphogens such as FGF8. Thus that this section is highly speculative, because different mecha-
eachareaofthecortexisdifferentiatedinthemannerdescribedby nisms maycontribute to modality codes inunimodal and mul-
RashandGrove(2006).Theseareassecreteattractantmolecules timodal areas. However, there is no a priori reason why they
that attract the appropriate axons from the thalamus. It seems should not be, if not the same, at least similar. It might seem
therefore that the early specification of the modality of cortical simpler to suggest that evolution would have developed two
areas defines the type of sensory afferents (and, therefore, sen- similar mechanisms to perform two similar processes in related
sorymodality)projectingtothemandnotthecontrary.However, neurons.
FrontiersinIntegrativeNeuroscience www.frontiersin.org January2013|Volume6|Article126|6
SmythiesandEdelstein Transsynapticspikeandmolecularcodes
If unimodal afferent axons in the sensory system have such Theexosomesystemseemstooperateinamannersimilartothe
a marked effect on the functional neuroanatomy of the post- synaptic vesicle system in delivering its payload into the synap-
synaptic neuron, the question naturally arises—what happens ticcleft. Theuptakesystem into thepostsynaptic neuronisalso
whenaxonswhichbelongtotwo(ormore)differentmodalsys- likely to be quick process. The target of a given microRNA is
tems,synapseononemultimodalneuronwithinhighersensory its own specific mRNA, which likely resides within the postsy-
cortex? Does this neuron possess [a] one “intermediate” com- naptic area, implying that the entire operation would be quite
putationalsystem,[b]two(ormore)quasi-independentsystems rapid.
thatsomehowinteractor,[c]anentirelydifferentsystem?Wenow Another questionishowthesystem inamultimodalneuron
know that nearly all of sensory cortex is multimodal in charac- mightreacttoquantitativechangesinitsinputs.Forexample,if
ter. In“primary”sensorycortex onemodeisdominantandthe thebimodalinputtosuchaneuronis50%“A”and50%“B,”how
othersoperateatsubliminallevels.Inhigherpolysensorycortex woulditreacttoachangeinthesensoryinputthatraisesthelevel
theneuronsintegrateallthevariousinputsmoreorlessequally. ofoneoftheseinputsrelativetotheother—sayto70%“A”and
Thesequestionsarerelevanttothequestionoftheneuralbasisof 30%“B?”Thiswouldalterthedetailsofthemodalcode“AB”sent
“binding.” bythisneurontohighercenters.Wouldthischangecarryusable
In multisensory cortex the incoming axons may carry dif- information?
ferent modality codes and may transport the different LMTSAs We can also ask whether the spike codes and LMTSA codes
associated with these codes. A modality code is defined simply are translatable into each other. Since the LMTSA code is
by its precise content of LMTSAs. Different individual post- modulated by a variety of neural and synaptic events (Fauré
synaptic neurons would receive, via the exosome system, dif- et al., 2006; Barmack et al., 2010; Impey et al., 2010; Cohen
ferent proportions of these signaling agents. MicroRNAs can et al., 2011; Hsu et al., 2012; Saba et al., 2012; as detailed
modulate a large number of functions throughout the neuron above) it is possible that the temporal pattern of incoming
by their action on mRNAs. All this activity translates into the spikes at a synapse that carry the afferent spike code could
dynamic feature of neurons whereby many parts of the cell is modulate the LMTSA/exosome system. Likewise the emerg-
beingconstantlyreplaced(Smythies,2002).Thisentailsthatthere ing pattern of LMTSA-induced modulation at the postsynap-
will be a wide variety in the functional anatomy of the infor- tic site could modulate the membrane events that lead to the
mation processing mechanisms inside the neurons that receive spike formation and timing that generate the efferent spike
these variegated inputs. For example, one neuron may receive code.
30 visual, 40 somatosensory, and 10 auditory inputs axons.
In another these numbers may be 60, 5, and 40—and so on. EXPERIMENTSTOTESTOURHYPOTHESIS
Eachinvolves adifferent mixofsignalingproteins, mRNAsand The most pressing need is to repeat the experiments, reported
microRNAs.Inaddition,eachneuronwillhaveauniquehistory by Koles and Budnik (2012) in the neuromuscular junction, at
of its activities and impacts of a wide range of neuromod- synapses made by axons on postsynaptic neurons. This would
ulators, in addition to the LMTSAs received. Currently, little answer the question whether exosomes do at the interneu-
attention has been paid to the idea that each and every neu- ronal synapse what they have been shown to do at the neu-
ron in the brain may be unique in this way. This opens up romuscular junction. This should be followed by experiments
a wide range of possible computational mechanisms (Molfese, to test if and how exosomes cross such synapses, particularly
2011). to determine if astrocytes are involved. Then, detailed experi-
Inmosttheoriesoftheneuralbasisof“binding,”attentionis ments are needed to trace the passage of the LMTS agents in
focusedonactivityintheactivityofgroupsofneuronsbelonging the postsynaptic neuron. Experiments are indicated that would
to different modalities arranged in nerve nets. It seems proba- look for specific signaling proteins, mRNA, and microRNAs
blethatsuchactivityisindeedinvolved.However,ourhypothesis in axons of different modalities. It would be informative to
adds another parameter. In a multisensory neuron significant determine the mode of action in more detail of the LMTSAs
“binding”informationmayalsobecontainedwitheachandevery listed above that have been found to modulate neuronal func-
individualneuronbyvirtueoftheirspecificuniqueelectrochem- tion. Further experiments are needed to determine the exact
ical makeup, as we havedescribed. That is to say, for example, mechanisms of transport and transfer of this material down
that the auditory system may have one specific pattern “A” of axons.
computationalfunctionalmachineryorganizedinpartbyitsown
particularcollectionofreceivedLMTSAs.Thevisualsystemlike- CONCLUSION
wise mayhave its own specific system “B” organized in part by The impact of the idea coined by Smalheiser (2007, 2009) and
its own, and different, collection ofreceived LMTSAs.In which Fauréetal.(2006),thatLMTSAs(includingmicroRNAs)canact
case,ahigherbimodalneuron,towhichboththesetwoneurons as signals between neurons and play a role in information pro-
project, willhaveitsownpattern“C”thatorganized byitsown cessingoncognitiveneuroscience,hasbeenlimited.Infact,only
specific collection of LMTSAs. In this case we cansuggest, very onegroup(KolesandBudnik,2012)hastakenupthistopicsince.
roughly,that“C”equals½“A”+½“B.” We hope that our paper will help extend further research into
It might be possible to estimate the degree to which the thisimportantsubject. Mostreviewsaimtopresentadvancesin
LMTSA/exosome system that we have suggested is dynamic. scientificknowledgeintheirsubject.Incontrast,thisreviewdis-
In other words, what is the time frame of its operation? closesalamentablestateofignoranceconcerningakeyfunction
FrontiersinIntegrativeNeuroscience www.frontiersin.org January2013|Volume6|Article126|7
SmythiesandEdelstein Transsynapticspikeandmolecularcodes
of the brain—the possible existence of a hitherto unrecognized thisinformationdown-stream.Wehavehardlyanyideahowthis
widespread signaling system. It now seems probable that the is done: nor how far downthe chain of information processing
afferentaxonscarry,notonlyminute-to-minuteoperativeinfor- thisprocessextends.However,thereseemtobesomepromising
mation in the form of spike trains, but also instructions in the candidatesinthefield.Theseincludeaxonaltemporalspiketrains
form of specific large signaling molecules. It is possible that, and synchronized oscillations, and specific patterns of LMTSAs
undernormalcircumstances, these instructions refine theoper- transported between neurons in the cortical hierarchy by the
ations of the sensory, and probably other, cortex that processes exosomesystem.
REFERENCES regulation of homeostatic synap- Comput. Biol. 5:e1000402. doi: primate somatosensory cortex.
Asai, Y., Guha, A., and Villa, A. tic plasticity. Proc. Nat. Acad. Sci. 10.1371/journal.pcbi.1000402 Annu.Rev.Neurosci.23,1–37.
E. (2008). Deterministic neural U.S.A.108,11650–11611. Gordleeval, S. Y., Stasenkol, S. V., Kayser, C., Montemurro, N. K.,
dynamics transmitted through Cui,X.,Weng,Y.Q.,Frappé,I.,Burgess, Semyanov, A. V., Dityate, A. E., Logothetis, N. K., and Panzeri, S.
neuralnetworks. Neural.Netw.21, A.,GirãodaCruz,M.T.,Schachner, and Kazantsev, V. B. (2012). Bi- (2009). Spike-phase coding boosts
799–809. M.,etal.(2011).Thecelladhesion directional astrocytic regulation of and stabilizes information carried
Barmack, N. H., Qian, Z., and molecule L1 regulates the expres- neuronalactivitywithinanetwork. by spatial and temporal spike
Yakhnitsa, V. (2010). Climbing sionofcholineacetyltransferaseand Front. Comp. Neurosci. 6:92. doi: patterns.Neuron61,4597–4608.
fibersinducemicroRNAtranscrip- thedevelopmentofseptalcholiner- 10.3389/fncom.2012.00092 Kepecs, A., and Lisman, J. (2003).
tion in cerebellar Purkinje cells. gicneurons.BrainBehav.1,73–86. Gruijters,W.Y.(2003).Aregapjunc- Information encoding and com-
Neuroscience171,655–665. Diesmann, M., Gewaltig, M. O., and tionmembraneplaquesimplicated putation with spikes and bursts.
Beaud, M. L., Rouiller, E. M., Bloch, Aertsen,A.(1999).Stablepropaga- inintercellularvesicletransfer?Cell. Network14,103–118.
J.,Mir,A.,Schwab,M.E.,Wannier, tionofsynchronousspikingincor- Biol.Int.27,711–717. Kishimoto, T.,Itoh,K.,Umekage,M.,
T.,etal.(2012).Invasionoflesion tical neural networks. Nature 402, Guan, H., and Maness, P. F. (2010). Tonosaki, M., Yaoi, T., Fukui, K.,
territory by regenerating fibers 529–533. PerisomaticGABAergicinnervation etal.(2012).DownregulationofL1
after spinal cord injury in adult Dijkmans,T.F.,vanHooijdonk,L.W., inprefrontalcortexisregulatedby perturbs neuronal migration and
macaque monkeys. Neuroscience Fitzsimons,C.P.,andVreugdenhil, ankyrininteractionwiththeL1cell alters the expression of transcrip-
27C,271–282. E. (2010). The doublecortin gene adhesion molecule. Cereb. Cortex tion factors in murine neocortex.
Berger, P., Kozlov, S. V., Cinelli, family and disorders of neuronal 20,2684–2693. J.Neurosci.Res.91,42–50.
P., Krüger, S. R., Vogt, L., and structure. Cent. Nerv. Syst. Agents Guescini,M.,Genedani,S.,Stocchi,V., Klausberger, T. (2009). GABAergic
Sonderegger, P. (1999). Neuronal Med.Chem.10,32–46. andAgnati,L.F.(2010).Astrocytes interneuronstargetingdendritesof
depolarization enhances the tran- Dityatev,A.,andRusakov,D.A.(2011). andGlioblastomacellsreleaseexo- pyramidalcellsintheCA1areaof
scription of the neuronal serine Molecular signals of plasticity at somes carrying mtDNA. J. Neural the hippocampus. Eur.J. Neurosci.
protease inhibitor neuroserpin. thetetrapartitesynapse.Curr.Opin. Trans.117,1–4. 30,947–957.
Mol.Cell.Neurosci.14,455–467. Neurobiol.21,353–359. Haydon,P.(2001).GLIA:listeningand Koles, K., and Budnik, V. (2012).
Borges, V. M., Lee, T. W., Christie, Fauré, J., Lachenal, G., Court, M., talking to the synapse. Nat. Rev. Exosomes go with the Wnt. Cell.
D. L., and Birch, N. P. (2010). Hirrlinger,J.,Chatellard-Causse,C., Neurosci.2,185–193. Log.2,1–5.
Neuroserpinregulatesthedensityof Blot, B., et al. (2006). Exosomes Henderson, L. A., Gustion, S. M., Kramer-Albers, E. M., Bretz, N.,
dendriticprotrusionsanddendritic are released by cultured cortical Macey, P. M., and Siddall, P. J. Tenzer,S.,Winterstein,C.,Mobius,
spineshapeinculturedhippocam- neurons. Mol. Cell. Neurosci. 31, (2011). Functional reorganization W., Berger, H., et al. (2007).
pal neurons. J. Neurosci. Res. 88, 642–648. of the brain in humans follow- Oligodendrocytessecrete exosomes
2610–2617. Frederikse,P.H.,Kasinathan, C., and ing spinal cord injury: evidence containingmajormyelinandstress-
Branco,T.,Clark,B.A.,andHäuser,M. Kleiman, N. J. (2012). Parallels for underlying changes in cor- protectiveproteins:trophicsupport
(2010).Dendriticdiscriminationof betweenneuronandlensfibercell tical anatomy. J. Neurosci. 31, foraxons?ProteomicsClin.Appl.1,
temporalinputsequencesincortical structureandmolecularregulatory 2630–2637. 1446–1461.
neurons.Science329,1671–1675. networks.Dev.Biol.368,255–260. Heron,J.,Roach,N.W.,Hanson,J.V., Krubitzer, L. (2007). The magnificent
Chivet, M., Hemming, F., Pernet- Frölich,D.,Frühbeis,C.,Amphornrat, McGraw, P. V., and Whittaker, D. compromise:corticalfieldevolution
Gallay,K.,Fraboulet,S.,andSadoul, J.,Thilemann,S.,Saab,A.,Kirchoff, (2012).Audiovisualtimeperception inmammals.Neuron56,201–208.
R. (2012). Emerging role of neu- F., et al. (2013). Crosstalk Between isspatiallyspecific.Exp.BrainRes. Kumar,A.,Rotter,S.,andAertsen,A.
ronalexosomes inthecentral ner- Neurons and Glia Involving 218,477–485. (2010). Spiking activity propaga-
vous system. Front. Physiol. 3:145. Exosomes as Vesicular Carriers Hsu, R.,Schofield,C. M.,DelaCruz, tion in neuronal networks: recon-
doi:3389/fphys.2012.00145 ofRNAandProteins.Abstract.ISEV C.G.,Jones-Davis,D.M.,Blelloch, cilingdifferentperspectivesonneu-
Chowdhury, S. A., and De Angelis, Conference. Boston, MA. April R., and Ullian,E.M. (2012).Loss ral coding. Nat. Rev. Neurosci. 11,
G. C. (2008). Fine discrimina- 17–21,2003. of microRNAs in pyramidal neu- 615–627.
tiontrainingaltersthecausalcon- Frühbels,C.,Fröhlich,D.,andKrämer- rons leads to specific changes in Lachenal,G.,Pernet-Gallay,K.,Chivet,
tribution of Macaque area MT Albers,E.M.(2012).Emergingroles inhibitorysynaptictransmissionin M., Hemming, F. J., Belly, A.,
to depths perception. Neuron 60, ofexosomesinneuron-gliacommu- the prefrontal cortex. Mol. Cell. Bodon,G.,etal.(2011).Releaseof
367–377. nication.Front.Physiol.3:119.doi: Neurosci.50,283–292. exosomes from differentiated neu-
Churchill, J. D., Tharp, J. A., 10.3389/fphys.2012.00119 Impey, S., Davarra, M., Lesiak, A., rons and its regulation by synap-
Wellman, C. D., Sengelaub, D. Gao, W. J., and Pallas, S. L. (1999). Fortin,D.,Ando,H.,Varlamova,O., ticglutamatergicactivity.Mol.Cell.
R., and Garraghty, P. E. (2004). Cross-modalreorganizationofhor- et al. (2010). An activity-induced Neurosci.46,409–418.
Morphological correlates of izontalconnectivityinauditorycor- microRNAcontrolsdendriticspine Lee, T. W., Montgomery, J. M., and
injury-induced reorganization tex without altering thalamocor- formation by regulating Rac1-PAK Birch, N. P. (2012). The serine
in primate somatosensory cor- tical projections. J. Neurosci. 19, signaling. Mol. Cell. Neurosci. 43, proteaseinhibitorneuroserpinreg-
tex. BMC Neurosci. 5:43. doi: 7940–7950. 146–156. ulates the growth and maturation
10.1186/1471-2202-5-43 Gollo,L.L.,Kinouchi,O.,andCopelli, Jones, E. G. (2000). Cortical and of hippocampal neurons through
Cohen, J. E., Lee, P. R., Chen, S., M.(2009).Activedendritesenhance subcortical contributions to a non-inhibitory mechanism.
andFields,R.D.(2011).MicroRNA neuronal dynamic range. PLoS activity-dependent plasticity in J.Neurochem.121,561–574.
FrontiersinIntegrativeNeuroscience www.frontiersin.org January2013|Volume6|Article126|8
SmythiesandEdelstein Transsynapticspikeandmolecularcodes
Legenstein, R.,andMaass,W.(2011). exosome-based genetherapy.Curr. controls GluA2 surface expression J. O. (2007). Exosome-mediated
Branch-specific plasticity enables Gene.Ther.12,262–274. inhippocampalneurons.Mol.Cell. transferofmRNAsandmicroRNAs
self-organization of nonlinear Pallas, S. L., Littman, T., and Moore, Biol.32,619–632. is a novel mechanism of genetic
computation in single neurons. D. R. (1999). Cross-modal reor- Sharma, J., Angelucci, A., and exchange between cells. Nat. Cell.
J.Neurosci.31,10787–10802. ganization of callosal connectiv- Sur, M. (2000). Induction of Biol.9,654–659.
Linden, J. F., and Schreiner, C. E. itywithoutalteringthalamocortical visual orientation modules in Von Bartheld, C. S., Byers, M. R.,
(2003). Columnar transformations projections. Proc. Natl. Acad. Sci. auditory cortex. Nature 404, Williams, R., and Bothwell, M.
inauditorycortex?Acomparisonto U.S.A.96,8751–8756. 841–847. (1996). Anterograde transport of
visual and somatosensory cortices. Pelled, G., Chuang, K. H., Dodd, Shinomoto,S.,Kim,H.,Shimokawa,T., neurotrophins and axodendritic
Cereb.Cortex13,83–88. S. J., and Koretsky, A. P. (2007). Matsuno,N.,Funahashi,S.,Shima, transfer in the developing visual
Lissin,D.V.,Carroll,R.C.,Nicoll,R.A., FunctionalMRIdetection ofbilat- K.,etal.(2009).Relatingneuronal system.Nature379,830–833.
Malenka,R.C.,andvonZastrow,M. eral cortical reorganization in the firingpatternstofunctionaldiffer- Winther, M., Berezin, V., and
(1999). Rapid, activation-induced rodent brain following peripheral entiation of cerebral cortex. PLoS Walmod, P. S. (2012).
redistribution of ionotropic glu- nerve deafferentation. Neuroimage Biol.5:e1000433.doi:10.1371/jour- NCAM2/OCAM/RNCAM: cell
tamate receptors in cultured hip- 37,262–273. nal.pcbi.1000433 adhesion molecule with a role
pocampal neurons.J.Neurosci. 19, Ptito, M., Fumal, A., de Noordhout, Singer,W.(2009).Distributedprocess- in neuronal compartmentaliza-
1263–1272. A. M., Schoenen, J., Gjedde, A., ing and temporal codes in neu- tion. Int.J. Biochem. Cell Biol. 44,
Lomber, S. G., Meredith, M. A., and and Kupers, R. (2008). TMS of ronalnetworks.Cogn.Neurodyn.3, 441–446.
Kral,A.(2010). Cross-modal plas- the occipital cortex induces tactile 189–196. Yang, Y., and Zador, A. M. (2012).
ticity in specific auditory cortices sensations in the fingers of blind Smalheiser, N. R. (2007). Exosomal Differences in sensitivity to neu-
underlies visual compensations Braillereaders.Exp.BrainRes.184, transfer of proteins and RNAs at ral timing among cortical areas.
in the deaf. Nat. Neurosci. 13, 193–200. synapsesinthenervoussystem.Biol. J.Neurosci.32,15142–15147.
1412–1418. Rakic,L.(2009).Evolutionoftheneo- Direct.2:35doi:10.1186/1745-6150- Ye,H.,Tan,Y.L.,Ponniah,S.,Takeda,
London,M.,andHäusser, M.(2005). cortex:aperspectivefromdevelop- 2–35 Y., Wang, S. Q., Schachner, M.,
Dendriticcomputation.Annu.Rev. mentalbiology.Nat.Rev.Neurosci. Smalheiser, N. R. (2009). Do neural et al. (2008). Neural recognition
Neurosci.28,28503–28532. 10,724–735. cells communicate with endothe- moleculesCHL1andNB-3regulate
Masuda, N., and Aihara, K. (2002). Rash,B.G.,andGrove,E.A.(2006). lial cells via secretory exosomes apical dendrite orientation in the
Spatiotemporalspikeencodingofa Area and layer patterning in the and microvesicles? Cardiovasc. neocortexviaPTPalpha.EMBOJ.
continuous external signal. Neural developing cerebral cortex. Curr. PsychiatryNeurol.2009:383083.doi: 27,188–200.
Comput.14,1599–1628. Opin.Neurobiol.16,25–34. 10.1155/2009/383086
Menna,E.,Cenni,M.C.,Naska,S.,and Ratajczak, J., Wysoczynski, M., Smythies, J. (2002). The Dynamic
Conflict of Interest Statement: The
Maffei,L.(2003).Theanterogradely Hayek, F., Janoswska-Wiecsorek, Neuron. Cambridge, MA: MIT
authors declare that the research
transportedBDNFpromotesretinal A., Ratajczak, M. Z. (2007). Press.
was conducted in the absence of any
axonremodelingduringeyespecific Membrane-derived microvesicles: Sur, M., and Rubenstein, J. L. R.
commercial or financial relationships
segregation within the LGN. Mol. important and underappreci- (2005).Patterningandplasticityof
thatcouldbeconstruedasapotential
Cell.Neurosci.24,972–983. ated mediators of cell-to-cell the cerebral cortex. Science 310,
conflictofinterest.
Molfese, D. L. (2011). Advancing communication. Leukemia 20, 805–810.
neuroscience through epigenetics: 1487–1495. Tetta, C., Ghigo, E., Silengo, L.,
molecular mechanisms of learning Reisin, H. D., and Colombo, J. A. Deregibus,M.C.,andCamussi,G. Received:14October2012;accepted:13
andmemory.Dev.Neuropsychol.36, (2004). Glial changes in primate (2012). Extracellular vesicles as an December 2012; published online: 04
821–827. cerebralcortexfollowinglong-term emergingmechanismofcell-to-cell January2013.
Nemenman, I., Lewen, G. D., Bialek, sensory deprivation. Brain Res. communication. Endocrine. doi: Citation: Smythies J and Edelstein L
W.,anddeRuyter vanSteveninck, 1000,179–182. 10.1007/s12020-012-9839-0. [Epub (2013)Transsynapticmodalitycodesin
R. R. (2008). Neural coding of Roe, A. W., Pallas, S. L., Hahm, J. aheadofprint]. the brain: possible involvementof syn-
natural stimuli: information at O., and Sur, M. (1990). A map Triana-Baltzer,G.B.,Liu,Z.,Gounko, chronizedspiketiming,microRNAs,exo-
sub-millisecond resolution. PLoS ofvisualspaceinducedinthepri- N. V., and Berg, D. K. (2008). somes and epigenetic processes. Front.
Comput. Biol. 4:e1000025. doi: mary auditory cortex. Science250, Multiple cell adhesion molecules Integr. Neurosci. 6:126. doi: 10.3389/
10.1371/journal.pcbi.1000025 818–820. shapingacomplexnicotinicsynapse fnint.2012.00126
Ni, H., Huang, L., Chen, N., Zhang, Roe,A.W.,Pallas,S.L.,Kwon,Y.H., onneurons.Mol.Cell.Neurosci.39, Copyright © 2013 Smythies and
F., Liu, D., Ge, M., et al. (2010). andSur,M.(1992).Visualprojec- 74–82. Edelstein.Thisisanopen-accessarticle
Upregulation of barrel GABAergic tionsroutedtotheauditorypathway Turola, E., Furlan, R., Bianco, F., distributed under the terms of the
neurons is associated with cross- inferrets: receptive fieldsof visual Matteoli, M., and Verderio, C. CreativeCommonsAttributionLicense,
modal plasticity in olfactory neurons in the primary auditory (2012). Microglial microvesicle which permits use, distribution and
deficit. PLoS ONE 5:e13736. doi: cortex.J.Neurosci.12,3651–3664. secretion and intercellular sig- reproduction inotherforums, provided
10.1371/journal.pone.0013736 Saba,R.,Störchel,P.H.,Aksoy-Aksel, naling. Front. Physiol. 3:149. doi: the original authors and source are
O’Loughlin, A. J., Woffindale, C. A., Kepura, F., Lippi, G., Plant, 10.3389/fphys.2012.00149 credited and subject to any copyright
A., and Wood, M. J. (2012). T. D., et al. (2012). Dopamine- Valadi, H., Ekstrom, K., Bossios, A., notices concerning any third-party
Exosomesandtheemergingfieldof regulated microRNA MiR-181a Sjostrand,M.,Lee,J.J.,andLotvall, graphicsetc.
FrontiersinIntegrativeNeuroscience www.frontiersin.org January2013|Volume6|Article126|9
