Haddock structure determination of CaM/pT19 PSD‐95 (1–21)

The molecular docking of Ca²⁺/CaM and pT19 PSD‐95 (1–21) was performed using the Haddock d‐level 2.2 web server as described (van Zundert & Bonvin, 2014). Residual dipolar couplings, chemical shift perturbation, and mutagenesis data were used as structural restraints. For active restraints or ambiguous interaction restraints (AIR), chemical shift perturbation was used, selecting residues whose chemical shift perturbation falls above the average perturbation. Charge reversal mutagenesis data (Table 1) that defined the intermolecular salt bridge between E17(PSD‐95)/R126 (CaM) and pT19(PSD‐95)/K115 (CaM) were used as unambiguous restraints.

The initial docking calculation used the structure of CaM (2MES) and helical structure of PSD‐95 N‐terminal peptide (residues 1–14; Zhang et al, 2014) as input structures for Haddock. PSD‐95 residues 14–21 were defined to be fully flexible for docking. UCSF Chimera (Pettersen et al, 2004) was used to introduce a phosphate group at the T19 position. Fifty‐nine AIR restraints were used for CaM based on chemical shift perturbation data and 14 active restraints were used for PSD‐95 from chemical shift perturbation of PSD‐95 previously defined by (Zhang et al, 2014). Unambiguous restraints were introduced to define key intermolecular interactions (Q15 (PSD‐95)/A147 (CaM); E17 (PSD‐95)/R126 (CaM); and pT19 (PSD‐95)/K115 (CaM)); the E17 (PSD‐95)/R126 (CaM) interaction was verified by mutagenesis (Table 1). In addition, an unambiguous restraint between C3 (PSD‐95) and I63 (CaM) was added based on visual inspection of the structure refined by RDC restraints. This interaction is supported by earlier data indicating that C3E substitution in the N‐terminal PSD‐95 peptide 1–13 decreased binding to Ca²⁺/CaM (Zhang et al, 2014).

Initial docking calculations used AIRs based on chemical shift perturbation, grouping the top 200 structures for simulated annealing and water refinement. The lowest energy structures were then run again, adding unambiguous restraints based on mutagenesis data. Rigid‐body docking, simulated annealing, and water refinement were run using the top 200 structures. RDC restraints assigned to CaM were then added using the Sani statement, with tensor values Dr and Da calculated using the program PALES (Zweckstetter, 2008). A total of 74 RDC values were used from residues found in regions of regular secondary structure and as deemed reliable by the PALES calculation.