What are the consequences of CaMKII translocation? How does the activation of this kinase contribute to the establishment of LTP and in which way is this contribution related to the formation of new memories? CaMKII phosphorylates several PSD-associated proteins, including PSD-95, a and P tubulin, the GTPase dynamin, the type IV intermediate filament protein a-internexin and cAMP phosphodiesterase,120 although the functional consequences of most of these phosphorylations are not known. Despite this fact, there exists a considerable amount of evidence indicating that one of the principal events mediated by CaMKII during the early phase of LTP is related with the upregulation of AMPA receptors. As mentioned above, CaMKII phosphorylates the GluR1 subunit of the rAMPA at Ser 831 and it has been shown that this phosphorylation potentiates AMPA-mediated currents,4,58,69,104 presumably through a mechanism that involves the stabilisation of the receptor into a high conductance state.21 rNMDA stimulation promotes the Ca2+ and CaMKII-dependent phosphorylation of the rAMPA104 and, moreover, induction of LTP is associated with a delayed enhancement in AMPA-mediated responses which is accompanied by the CaMKII-dependent phosphoryla-tion of GluR1 at Ser 831.5 Interestingly, inhibitory avoidance learning in the rat is also associ ated with an increase in hippocampal GluRl phosphorylation, possibly through a mechanism involving CaMKII activation.14
Besides the direct effect that phosphorylation of the GluRl subunit might have on the electrophysiological properties of the rAMPA, CaMKII could mediated some other responses that are known to involve plastic changes of this receptor. A high proportion of synapses in the CA1 area of the hippocampus transmit with rNMDA but not with rAMPA, indicating that they are nonfunctional at normal resting potentials. Surprisingly, these same synapses acquire AMPA-type responses following LTP induction.54 These findings, together with earlier results showing that both glycine-induced LTP in hippocampal slices72,84 and in vivo induced LTP in the rat hippocampus63,107 produce an increase in the number of 3H-AMPA binding sites, fuelled the appearance of the "silent synapse" hypothesis (for a recent review see ref. 61). This hypothesis postulates that LTP produces the conversion of synapses lacking rAMPA responses into fully functional AMPAergic terminals, maybe by means of the insertion of previously extrasynaptically localised rAMPA. This idea received further support when it was shown that induction of LTP causes the rNMDA-dependent redistribution of green fluorescent protein-tagged GluR1 from intracellular pools into dendritic spines.87 Although this process has been reported to be independent of GluR1 Ser 831 phosphorylation, it can be mimicked by increasing the activation state of CaMKII35 and quite recently it has been shown that spontaneous activation of rNMDA in hippocampal neuronal cultures provokes the rapid recruitment of rAMPA into morphological silent synapses (i.e., synapses that contains rNMDA but not rAMPA), an event accompanied by the translocation of CaMKII into those synapses and the phosphorylation of GluR1 at Ser 831.55 These results are in agreement with both, recent findings showing that KN-62 blocks the NMDA-induced increase in GluR1 and GluR2/3 associated with hippocampal synaptic plasma membranes (SPM)12 and those indicating that incubation of hippocampal SPM under conditions suitable for CaMKII activation and autophosphorylation promotes a CaMKII-dependent increase in PSD-associated 3H-AMPA binding sites,14 suggesting the participation of CaMKII in this insertion mechanism and providing a plausible explanation to the observed increase in hippocampal 3H-AMPA binding sites that follows inhibitory avoidance learning in the rat.15
A role for CaMKII in Ca2+ dendrite membrane trafficking has been suggested.59 There is experimental evidence for the formation of new synapses (or the remodelling of existing ones) after the induction of LTP1,71 as well as during the consolidation phase of both an avoidance training and the water maze learning task (see ref. 74,75 and Geinisman et al, this book). Tetanic stimulation promotes a rNMDA-dependent increase in the number of dendritic spines surrounding the area of stimulation, which is blocked by KN-93, a specific CaMKII inhibitor.22,60,108 Moreover, in Drosophila, expression of a constitutively active form of CaMKII results in the phosphorylation of the disc large protein (DLG), a homologue of the mammalian PSD family of proteins, causing rearrangement of the synaptic structure.53 These findings suggest that long-lasting CaMKII up-regulation, as it occurs during activity-dependent synaptic plasticity, could help in the synaptic remodelling of the PSD scaffold.
Together with CaMKII, PKA and PKC, the intracellular signalling pathway mediated through the activation of the mitogen-activated protein kinases ERK1/2 is one the best characterised in relation to plastic mechanisms (for recent reviews see ref. 99 and Selcher et al in this book). Up-regulation of ERK1/2 is classically attained through the sequential activation of the GTPase Ras, Raf-1 and the ERK kinase MEK and this pathway can modulate the phosphorylation state of the transcription factors ElK-1 and CREB.19 A couple of years ago, and almost simultaneously, Huganir's and Kennedy's groups at the John Hopkins University and California Institute of Technology, respectively, reported the existence of a novel, synaptically-localized Ras-GTPase activating protein named p135 SynGAP p135 SynGAP colocalizes with PSD-95, the rNMDA and the synapse associated protein SAP-102 and it is almost exclusively present in hippocampal neurons. p135 SynGAP stimulates Ras GTPase activity, suggesting that it is negatively coupled to the activation of the ERK pathway.18,51 Interestingly, p135 SynGAP is a substrate of CaMKII which inhibits its Ras-GTPase activating activity. Moreover, it has been shown that in cortical neurons, inhibition of CaMKII partially blocks the NMDA-induced phosphorylation of ERK1/2.45 These findings encourage the tempting hypothesis that, by means of blocking the inactivation of GTP-bound Ras, CaMKII could directly couple rNMDA stimulation with the up-regulation of the ERK pathway and hence, indirectly contribute to the activation of the inducible transcription factors that is needed for the establishment of late LTP and the formation of long-lasting memories (see ref. 99 and also Selcher et al and Frankland and Josselyn in this book]. In addition, there exists evidence suggesting that, in turn, ERK1/2 could also control CaMKII up-regulation during LTP. When LTP is induced into the CA1 area of the rat hippocampus using a protocol that involves theta-pulse stimulation and activation of P adrenergic receptors, it produces an ERK-dependent potentiation which is accompanied by a transient, colocalized and rNMDA-dependent increase in ERK and CaMKII phosphoryla-tion. These early increases are followed by a delayed, actinomycin-D/anisomycin-dependent augmentation of CaMKII protein levels. Both, CaMKII phosphorylation and expression are blocked by preincubation of the hippocampal slices in the presence of the MEK inhibitor PD98059, indicating that ERK1/2 are likely to participate as an up-stream factor in the events that regulate CaMKII function during neuronal plasticity.29
As mentioned above CaMKII subunits are encoded by a family of 4 related genes: a,P,y and 6 and it has been described that alternative splicing can generate isoforms containing a nuclear localization signal (NLS) homologous to that present in the SV40 large T antigen.11,94 As demonstrated by immunostaining of brain sections11 and despite the large size of the holoenzyme (400-600 kDa), it seems that CaMKII does have access to the neuronal nucleus where it could directly regulate the activity of different transcription factors and hence gene expression. Overexpression of a nuclear-localised isoform of the 6 CaMKII subunit promotes BDNF transcription in NG108-15 cells101 and it has been shown that, in hippocampal neurons, Ca2+/ calmodulin-dependent kinases intensify the activity and expression of the CCAAT enhancer element-binding protein P (C/EBPP),121 a transcription factor involved in the switch from short to long-term facilitation in Aplysia2 and in the consolidation of hippocampal dependent memories.1 5
As well as PKA and the ERK-activated kinase p90RSK, CaMKII can phosphorylate CREB at Ser 133  but it is not able to promote the activity of this transcription factor, maybe due to the phosphorylation of a secondary site at Ser 142 that prevents CREB dimerization and binding to the CREB-binding protein (CBP).117 Despite this fact, there are several lines of evidence suggesting the involvement of CaMKII in the activation of CREcontaining genes. At this respect, early studies have shown that CaMKII is able to directly phosphorylate the activating transcription factor 1 (ATF-1; a member of the ATF/CREB family of transcription factors) on Ser 63, suggesting that in that way it could mediate transactivation respon sive genes.88 (See Fig. 2 for a schematic diagram of the postsynaptic effects of CaMKII activation during plastic events).
Was this article helpful?