(720f) Calmodulin Stabilizes Camkii Autophosphorylation through Structural Exclusion of Phosphatase

Neurological disorders such as Alzheimer’s disease affect more than 15 million Americans, yet for these disorders, no effective therapeutics exist. In order to advance, the field requires a quantitative understanding of how protein signaling within neurons could become dysregulated in diseased patients. In particular, calcium/calmodulin-dependent kinase II (CaMKII), which accounts for 1-2 of percent all protein in the human hippocampus, is essential to healthy learning and memory formation. Here, we integrate computational modeling with bench-top experimentation to characterize a novel mechanism regulating CaMKII activity.

Experimental studies repeatedly implicate CaMKII as a regulator of synaptic plasticity, the process by which neuronal connections dynamically shift in size and excitability. Yet, in vitro experiments often lack the spatiotemporal resolution required to characterize the mechanisms regulating CaMKII activity. To complement experiments with the native enzyme and provide spatiotemporal resolution, we increasingly use computational models. Problematically, most computational models of CaMKII sacrifice biological accuracy, reducing the dodecameric architecture of native CaMKII enzyme to monomers, because the native enzyme’s structural complexity exceeds modern computing limits. To reduce computational expense yet preserve the complexity of intra-subunit interactions within the dodecameric CaMKII enzyme, we use a novel rule-based method in the open-source software MCell to deliver the most comprehensive model to-date of the native CaMKII enzyme.

Our rule-based, spatial-stochastic model of CaMKII explicitly accounts for each of the twelve subunits on CaMKII holoenzymes. Each subunit is modeled in a simultaneous combination of states: undocked or docked to the protein’s central hub, open or closed, active or inactive, bound or unbound, and phosphorylated or not (at two key amino acid residues). By varying the rules defining subunit state transitions, we can explore biophysical mechanisms regulating CaMKII activity. For example, studies indicate that calmodulin (CaM) and protein phosphatase 1 (PP1) bind CaMKII at overlapping sites, suggesting mutual steric hindrance. To explore the functional relevance of this structural exclusion of ligand binding to CaMKII, we first implement an “exclusive” model in which CaMKII-bound CaM and PP1 mutually exclude each other from binding the same CaMKII subunit. For comparison, we also implement a “non-exclusive” model in which CaM and PP1 agnostically bind CaMKII. Our results indicate that in the exclusive model only, phosphorylation of CaMKII at Thr-286 (pThr-286, a modification necessary for synaptic plasticity) is maintained over time. Thus, we predict that mutual steric hindrance of PP1 and CaM for CaMKII binding is a mechanism by which CaMKII phosphorylation levels are regulated.

Next, we evaluate our proposed mechanism regulating CaMKII pThr-286 in an in vitro phosphorylation assay. Inspired by previous work from Urakubo et al., we maximally phosphorylate a sample of purified CaMKII holoenzymes. After isolating the phosphorylated CaMKII by size-exclusion chromatography, we introduce PP1 along with or without CaM, and monitor phosphatase activity over time. Preliminary results suggest that pThr-286 levels decline more slowly in the presence of CaM, supporting our hypothesis that CaM excludes PP1 binding to CaMKII.

By integrating computational modeling and in vitro experimentation, we have predicted and experimentally validated a previously unidentified mechanism regulating CaMKII activity. In light of this mechanism, it may be necessary to account for steric hindrance by CaM and PP1 in future therapeutic design.