Synaptic Plasticity: The SubcellularLocation of CaMKII Controls Plasticity

Kevin Fox

In a neuron’s dendritic spine, the location of CaMKIIis controlled by a number of interacting factors,including its ability to bind calcium/calmodulin, itsphosphorylation state and the synthesis of new sub-units in the dendrites. New studies have shown thatthe exact location of CaMKII is crucial for the formand endurance of synaptic plasticity.

Our concept of the synapse and synaptic function hasevolved enormously over the past decade. In the past,the synapse has often been treated as a system ofcomplicated switches turned on and off by endoge-nous transmitters and pharmacological agents. Morerecently, however, an understanding of what liesbeneath the membrane on either side of a synapse hasled to a view of synaptic function as the operation ofan intricate set of molecular mechanisms. The synap-tic machine even has real moving parts: AMPA andNMDA glutamate receptors have been shown to slot inand out of the membrane [1,2]; and calcium/calmodulinkinase II (CaMKII) has been shown to move in and outof the ‘post-synaptic density’ — the specialised regionimmediately beneath the postsynaptic neuron’s mem-brane that faces the synaptic cleft — under the controlof its own phosphorylation state [3]. Understanding therole of various synaptic constituents therefore relies,not only on knowing how they are activated and whatthey act upon, but also where they are located withinthe molecular machinery at any one time.

Two groups [4,5] have now reported on the functionalsignificance of the regulated subcellular location ofCaMKII in dendrites. Elgersma et al. [4] have studied theimportance of CaMKII located in the postsynapticdensity. And Miller et al. [5] have looked at the signifi-cance of CaMKII mRNA translation in the dendrites —as opposed to the soma — by preventing the deliveryof CaMKII mRNA to the dendrites. Using advancedmolecular genetic techniques, both groups have shownthat the location of CaMKII can determine its functionand thereby affect the plasticity of the synapse.

In its basal state, CaMKII is phosphorylated onresidue threonine 305 [6]. Elgersma et al. [4] found thatphosphorylation at this site substantially decreases theamount of CaMKII in the post-synaptic density [4]. Theyreplaced CaMKII’s threonine 305 with an aspartate —mutation T305D — which mimics a phosphorylationgroup. They found that this led to a reduction in thetotal amount of CaMKII in post-synaptic-density-enriched fractions to only 28% of the wild-type level.This is consistent with earlier studies which showedthat preventing phosphorylation at the 305/306 sites

increases CaMKII’s affinity for the post-synaptic density[3]. The explanation for this behaviour is probably thatphosphorylation at the inhibitory 305 site preventsbinding of calcium/calmodulin to CaMKII and, as suchbinding is necessary for translocation of CaMKII to thepost-synaptic density, the lack of it leads to CaMKIIbeing marooned in the cytoplasm (Figure 1).

Loss of CaMKII function — whether by directinhibition of the enzyme or inactivation of the gene —is known to prevent plasticity in the hippocampusand cortex [7,8]. CaMKII is ubiquitously distributed inthe cytoplasm and the post-synaptic density,however, so it could potentially influence plasticity ata number of different sites. The T305D mutant form ofCaMKII is characteristically present at low levels inthe post-synaptic density, making it is possible to testwhether CaMKII has to be present in the post-synap-tic density for plasticity. Elgersma et al. [4] found that

Dispatch

Current Biology, Vol. 13, R143–R145, February 18, 2003, ©2003 Elsevier Science Ltd. All rights reserved. PII S0960-9822(03)00077-0

School of Biosciences, Cardiff University, Museum Avenue,Cardiff CF10 3US, UK. E-mail: foxkd@cf.ac.uk

Figure 1. Movement of CaMKII within the synapse anddendritic spine.

A highly schematic view of the possible transition states ofCaMKII (blue squares) is shown for a single subunit. CaMKII isa 12 member heteromer composed of αα and ββ subunits, butsingle subunits are shown here for clarity. CaMKII is translatedin the polyribosomes under the control of calcium influx to thedendrite [13] and is presumably assembled into multimersnearby. CaMKII can move into the post-synaptic density whenit binds calcium/calmodulin during elevated levels of synapticactivity [3]. It can then either bind to the NMDA receptor (notshown) and/or autophosphorylate on residue threonine 286.Further phosphorylation at threonine 305 results in CaMKIIbeing expelled from the post-synaptic density. Binding to theNMDA receptor subunit NR2B may protect CaMKII frominhibitory phosphorylation and thereby prolong further its dwell-time in the post-synaptic density [15]. Note that a number ofpossible intermediate states and transitions have been omittedfor clarity.

P

Ca/CaM

T286T305

P

P

PP1

T286T305

Autophos

T286T305

T286T305 PP1/PP2A

Polyribosomes

CaMKII

Autophos

PSD

Dendritic spine

Current Biology

hippocampal LTP — induced in wild-type animals bytheta-burst stimulation or a 100 Hz tetanus of 1second duration — was abolished in T305D mutantmice. This implies that CaMKII is required in the post-synaptic density for LTP induction [4]. The samemutants also exhibited severely impaired memory inthe water maze task, suggesting that in vivo hip-pocampal plasticity also depends on the presence ofCaMKII in the post-synaptic density [4].

So why does CaMKII need to be located in the post-synaptic density for plasticity? Many substrates ofCaMKII are located in the post-synaptic density andthe molecules almost certainly need to come intoclose proximity in the appropriate combinations tointeract. Once in the post-synaptic density, CaMKIIcan phosphorylate substrates including — but notlimited to — the GluR1 subunit of the AMPA channel,SynGap and the NMDA channel. CaMKII may alsocontrol delivery of newly formed AMPA channels intothe membrane from the post-synaptic density, orcreate anchoring assemblies for delivery of AMPAchannels into the membrane from this location [9]. Theaction of CaMKII in this regard is an example of amore general principle of cellular organization. Complexsignal transduction pathways are often organized incells by a system of scaffold, anchoring and adaptorproteins which bring enzymes and substrates intoclose proximity to increase the efficiency of trans-duction and restrict the possible molecular interac-tions to just those required [10].

Not only is the localization of CaMKII carefullyregulated in the dendritic spine, but so too is thesupply of available CaMKII to the dendritic spine. Mostproteins are synthesized in the soma and transportedto their final location within the cell, but like severalother neuronal proteins, CaMKII is translated in thedendrites by polyribosomes proximal to dendriticspines [11,12]. Synthesis of CaMKII is calcium-depen-dent and can be induced by sensory activity capableof inducing plasticity [13,14]. The functional signifi-cance of this mechanism has been studied by Miller etal. [5]. By altering the sequence of the 3′ untranslatedregion that addresses CaMKII mRNA for delivery to thedendrites, this group was able to produce an animalthat contained CaMKII at normal levels in the soma butat greatly reduced levels in the dendrites. The CaMKIIin the post-synaptic density was reduced to 17% ofcontrol levels in this mutant, indicating that most, butnot all, CaMKII in the post-synaptic density is normallygenerated in the dendrites [5].

Curiously, the early phase of LTP was unaffected bythe low level of CaMKII in the post-synaptic density.But LTP was impaired from 2 hours after the tetanusonward. Similarly, short-term memory over a period of30 minutes was unaffected in the mutants, but longer-term memory was impaired. One interpretation of thisresult is that a supply of newly synthesized CaMKIIsubunits is required for maintenance of the laterstages of LTP and memory. Why this might be thecase is not clear at present.

The mutations discussed here lead to a static orend-state distribution of CaMKII and do not necessar-ily tell us about the dynamic behaviour of wild-type

CaMKII under various phosphorylation conditions. Butsome insight can be gained by comparison with cellculture studies, where dynamics can be studied usingGFP-tagged CaMKII [3]. One possible sequence ofevents is as follows: synaptic stimulation capable ofproducing late phase LTP increases the enzymaticactivity of CaMKII located in the post-synapticdensity, because the kinase becomes phosphorylatedat threonine 286, increasing its affinity for calcium/calmodulin and thereby lengthening its dwell-time inthe post-synaptic density (Figure 1). Very high levelsof CaMKII kinase activity then lead to phosphorylationat the threonine 305 site and consequent expulsion ofCaMKII from the post-synaptic density. According tothis argument, maintaining the level of active CaMKIIthen requires replenishment of the post-synapticdensity from the pool of newly synthesized CaMKII.Without the replenishment, late phase LTP does notdevelop fully [5].

One aspect of the above theory which is not clear,is how the newly synthesized CaMKII gets into thepost-synaptic density without further calcium influxand hence calcium/calmodulin binding (Figure 1). It isconceivable that a sufficient quantity of newly synthe-sized CaMKII is able to drive the kinase molecules intothe post-synaptic density by diffusion, or alternativelythe NMDA channel may be sensitized to admit higherlevels of calcium during normal synaptic transmissionfor some time after the initial plasticity inducing event.Another aspect that is not clear is why newly synthe-sized CaMKII would be required rather than recycledCaMKII. Clearly, many questions remain to beanswered, but by linking the location of CaMKII to itsfunction, the studies discussed above give an impor-tant rationale for further investigation in this field.

References1. Shi, S., Hayashi, Y., Esteban, J.A. and Malinow, R. (2001). Subunit-

specific rules governing AMPA receptor trafficking to synapses inhippocampal pyramidal neurons. Cell 105, 331–343.

2. Barria, A. and Malinow, R. (2002). Subunit-specific NMDA receptortrafficking to synapses. Neuron 35, 345–353.

3. Shen, K. and Meyer, T. (1999). Dynamic control of CaMKII translo-cation and localization in hippocampal neurons by NMDA receptorstimulation. Science 284, 162–166.

4. Elgersma, Y., Fedorov, N.B., Ikonen, S., Choi, E.S., Elgersma, M.,Carvalho, O.M., Giese, K.P. and Silva, A.J. (2002). Inhibitoryautophosphorylation of CaMKII controls PSD association, plastic-ity, and learning. Neuron 36, 493–505.

5. Miller, S., Yasuda, M., Coats, J.K., Jones, Y., Martone, M.E. andMayford, M. (2002). Disruption of dendritic translation of CaMKIIal-pha impairs stabilization of synaptic plasticity and memory consol-idation. Neuron 36, 507–519.

6. Colbran, R.J. (1993). Inactivation of Ca2+/calmodulin-dependentprotein kinase II by basal autophosphorylation. J. Biol. Chem. 268,7163–7170.

7. Silva, A.J., Stevens, C.F., Tonegawa, S. and Wang, Y. (1992). Defi-cient hippocampal long-term potentiation in alpha-calcium-calmod-ulin kinase II mutant mice. Science 257, 201–206.

8. Glazewski, S., Giese, K.P., Silva, A. and Fox, K. (2000). The role ofalpha-CaMKII autophosphorylation in neocortical experience-dependent plasticity. Nat. Neurosci. 3, 911–918.

9. Lisman, J., Schulman, H. and Cline, H. (2002). The molecular basisof CaMKII function in synaptic and behavioural memory. Nat. Rev.Neurosci. 3, 175–190.

10. Pawson, T. and Scott, J.D. (1997). Signaling through scaffold,anchoring, and adaptor proteins. Science 278, 2075–2080.

11. Steward, O. and Levy, W.B. (1982). Preferential localization ofpolyribosomes under the base of dendritic spines in granule cells ofthe dentate gyrus. J. Neurosci. 2, 284–291.

DispatchR144

12. Bagni, C., Mannucci, L., Dotti, C.G. and Amaldi, F. (2000). Chemicalstimulation of synaptosomes modulates alpha -Ca2+/calmodulin-dependent protein kinase II mRNA association to polysomes. J.Neurosci. 20, RC76.

13. Wells, D.G., Dong, X., Quinlan, E.M., Huang, Y.S., Bear, M.F.,Richter, J.D. and Fallon, J.R. (2001). A role for the cytoplasmicpolyadenylation element in NMDA receptor-regulated mRNA trans-lation in neurons. J. Neurosci. 21, 9541–9548.

14. Hendry, S.H. and Kennedy, M.B. (1986). Immunoreactivity for acalmodulin-dependent protein kinase is selectively increased inmacaque striate cortex after monocular deprivation. Proc. Natl.Acad. Sci. U.S.A. 83, 1536–1541.

15. Bayer, K.U., De Koninck, P., Leonard, A.S., Hell, J.W. and Schul-man, H. (2001). Interaction with the NMDA receptor locks CaMKII inan active conformation. Nature 411, 801–805.

Current BiologyR145