There is abundant evidence that the calcium/calmodulin-dependent protein kinase αCaMKII plays a critical role in the genesis of LTP in the Schaffer-collateral projection to hippocampal CA1 pyramidal cells. Specific inhibitors of the enzyme prevent the development of LTP (Malenka et al. 1989), and genetically engineered mice with a targeted deletion of the gene encoding αCaMKII fail to exhibit LTP in area CA1 (Silva et al. 1992). In a widely discussed model of the mechanism by which the enzyme affects the processes underlying LTP, a rise in intracellular Ca²⁺ caused by the LTP-inducing stimulation leads to activation of the enzyme and autophosphorylation on threonine 286 (Lisman et al. 2002). Phosphorylation on this residue in the multimeric enzyme results in autonomous activation, a state in which the enzyme no longer requires Ca²⁺-calmodulin for activity. The persistence of autonomous activity, and resulting phosphorylation of substrates, including AMPA receptor subunits, has been suggested to contribute to the maintenance of LTP (Lisman et al. 2002).

We have examined levels of autonomous activity in hippocampal slices in which LTP was induced either by titanic stimulation, or by exposure to the K⁺ channel blocker tetraethylammonium (TEA). In contrast to an earlier report of a persistent increase in autonomous activity following tetanus-induced LTP (Fukunaga et al. 1993), we did not observe a sustained increase in autonomous activity in either case. However, TEA-induced LTP led to a persistent increase in Thr286 phosphorylation, as observed by others for tetanus-induced LTP. Moreover, in animals with a threonine to alanine point mutation at residue 286, TEA failed to induce LTP, as previously demonstrated for tetanus-induced LTP (Giese et al. 1998). Thus, phosphorylation at Thr286 rather than persistent autonomous activation of the enzyme is the critical factor determining whether or not LTP is produced. In a further series of experiments we monitored LTP in vivo in the T286A mutant, and find that while LTP in area CA1 is absent in the mutant, it is relatively unaffected in the dentate gyrus. Thus, the current model for the involvement of αCaMKII in LTP requires modification in area CA1, and cannot account for LTP in the dentate gyrus.