“CaMKII can act as a structural molecule to mediate synaptic remodeling through protein degradation,” said first author Baris Bingol. He and principal investigator Morgan Sheng, who performed the bulk of these studies at the Massachusetts Institute of Technology, have since relocated to Genentech in South San Francisco, California, where Sheng is vice president for neuroscience. In searching for proteins that interact with synaptic proteasomes, the scientists discovered that CaMKII recruits the proteasome to the tips of dendritic spines, where it is needed for remodeling. CaMKII’s kinase activity was not necessary for this recruitment.

CaMKII is involved in long-term potentiation, the enhanced signaling between neurons that underlies memory formation (reviewed in Lisman et al., 2002). Researchers know that upon neural stimulation, CaMKII arrives at the post-synapse within seconds (Shen and Meyer, 1999) and proteasomal proteins follow minutes later (Bingol and Schuman, 2006).

CaMKIIα and β isoforms can form α homo-multimers or α-β hetero-multimers. CaMKIIβ has already exhibited a structural association in that it bundles actin fibers in dendritic spines (Okamoto et al., 2007). CaMKIIα is also essential for synaptic plasticity in the presynaptic region, where its kinase activity is not necessary (Hojjati et al., 2007).

Some studies suggest that CaMKII activity could be diminished in Alzheimer disease. In AD model mice, cortical neurons contain less CaMKII than control animals to begin with, and Aβ oligomer treatment further reduced the density of CaMKII clusters in cortical synapses (Gu et al., 2009). And in rat hippocampal slices, Aβ blocks CaMKII’s autophosphorylation (Zhao et al., 2004). Therefore, lack of CaMKII function might be involved in the cognitive deficits of Alzheimer’s, researchers surmise.

Bingol and colleagues began with the goal of identifying proteins that interact with synaptosome proteasomes from rat forebrain. They purified proteasomes and associated components, then used mass spectrometry to identify the hangers-on. One was CaMKII, an exciting find because of its abundance in the post-synaptic density. However, it was also questionable, Bingol said. CaMKII might simply be tagging along during the purification because the kinase is so plentiful. “It took us a long time to convince ourselves that the interaction is real,” he said. The researchers performed immunoblotting to confirm CaMKII’s identity and employed fluorescence microscopy to show that it colocalized with the proteasome in cultured hippocampal neurons.

Next, the scientists examined CaMKII’s role in controlling proteasome localization. They transfected neurons with GFP-tagged Rpt1, a proteasome subunit. When they stimulated the neurons, Rpt1 redistributed from the dendritic shafts to the spines, where GFP fluorescence brightened 1.4-fold. The scientists co-transfected plasmids encoding CaMKIIα or β, and found that spine fluorescence intensified 1.6-fold, showing that CAMKII draws proteasomes to the post-synaptic area. Conversely, when they applied RNA interference to attenuate expression of endogenous CAMKIIα, spine fluorescence nudged up only 1.2-fold. CAMKIIβ knockdown did not influence Rpt1 distribution, suggesting the enzyme’s α sibling handles the recruitment.

The researchers reasoned that Rpt1 deployment to spines could be due to any effect of neural stimulation, not just CaMKIIα recruitment, so they devised a stratagem to separate the two. To place CaMKIIα in the post-synaptic density without neural stimulation, they relied on the interaction between the FK506 binding protein 12 (FKBP) and its partner, the FKBP-rapamycin binding domain (FRB) (Banaszynski et al., 2005). The two protein domains unite in the presence of rapamycin. Sheng and colleagues hitched FRB to CaMKIIα, and FKBP to the protein PSD-95, which resides in the post-synaptic density. Upon rapamycin treatment, CaMKIIα moved into post-synaptic spines. Immunostaining then demonstrated that the proteasome followed along.

To examine the interaction between the kinase and the proteasome, the authors transfected HEK293 kidney cells with various CaMKIIα mutants. The CaMKIIα constructs co-immunoprecipitated with endogenous Rpt6, another proteasome component. When the researchers added CaMKIIα-T286D, a mutant that mimics the kinase’s auto-phosphorylated state, the kinase pulled down fivefold more Rpt6. This experiment suggests that CaMKIIα, although it can bind the proteasome in either phosphorylation state, interacts more strongly when phosphorylated at T286. But a kinase-dead mutant, T286D/K42R, showed the same association. That indicates that to bind the proteasome, CaMKIIα’s kinase activity is not necessary.

The data are “pretty convincing,” said Nikola Otmakhov of Brandeis University in Waltham, Massachusetts, who was not involved in the study. As a continuation, he said he would like to see researchers define the amino acid sequences that allow CaMKII and the proteasome to interact, and show how preventing that association affects a cell.

CaMKII may provide structure to more than just the proteasome, Otmakhov suggested. The kinase can partner with at least three proteins at once in the post-synaptic density (Robison et al., 2005). Otmakhov noted that not all of the post-synaptic CaMKII is needed to interact with the proteasome population. “It is going to be many, many more proteins which bind to CaMKII as a scaffold,” he predicted.—Amber Dance.

Reference:
Bingol B, Wang CF, Arnott D, Cheng D, Peng J, Sheng M. Autophosphorylated CaMKIIalpha acts as a scaffold to recruit proteasomes to dendritic spines. Cell. 2010 Feb 19;140(4):567-578. Abstract