Investigating the high affinity and low sequence specificity of calmodulin binding to its targets.

Calmodulin (CaM) is a calcium binding protein that regulates a wide range of enzymes. Recently the structures of a number of complexes between CaM and synthetic target peptides have been determined. The peptides correspond to the CaM-binding domain of skeletal and smooth muscle myosin light-chain kinase (MLCK) and calmodulin-dependent protein kinase II alpha. Comparison of the peptide-free and peptide-bound structures reveals that CaM undergoes a large conformational change when forming a complex, resulting in the formation of a binding surface that provides for an optimal interaction with its target. In this work, the available co-ordinates of the NMR solution structure of CaM-skeletal MLCK peptide are used as a basis upon which several molecular models of binding are built. The detailed features of the protein's peptide binding surface are revealed through two-dimensional topographical projections. Negatively charged margins at the binding surface extremities interact strongly with basic peptide residues separated by nine or ten positions. The binding surface core is hydrophobic and displays a groove with four deep pockets, which can accommodate bulky peptide residues at relative positions 4 and 8 (pocket A), 11 (pocket B), 13 (pocket C), 14 and 17 (pocket D). Therefore, both electrostatic and van der Waals' features contribute to the high affinity binding. A search for alternative peptide placements in the binding tunnel reveals the dominant role of specific electrostatic interactions in the binding energy. Apolar interactions are more permissive, such that the hydrophobic side-chains that line the binding tunnel adapt in order to maintain favourable van der Waals' contacts. The model suggests that the structure can accommodate large peptide translations (up to 5 A) and a reversed peptide binding mode, with a little loss in binding interaction energy. These calculations are compared with available experimental data, providing a structural rationale for the low sequence specificity of the CaM target recognition.

A protein/peptide assay using peptide-resin adduct: application to the calmodulin/RS20 complex.

To obtain equilibrium and kinetic constants of a protein/peptide complex, we have developed a rapid procedure which uses peptides specifically linked to a resin. With this peptide-resin adduct, bound and free 125I-labeled protein could be easily separated by simple centrifugation. The feasibility of the method was demonstrated with the calmodulin/RS20 complex, where RS20 is the putative calmodulin binding peptide of the smooth muscle myosin light chain kinase (smMLCK). In addition to the wild-type calmodulin (SYNCAM) expressed in Escherichia coli, we also examined calmodulin mutants with charge reversals called SYNCAM12A (DEE 118-120-->KKK) and SYNCAM18A (EEE 82-84-->KKK and DEE 118-120-->KKK). The kinetic analysis of the interaction between SYNCAM and RS20 associated with titration experiments allowed us to measure dissociation constants (KD) in the range of 10(-9) M, in good agreement with previously published data. Moreover, the binding assays showed that SYN-CAM18A did not interact with RS20, whereas SYN-CAM12A did with a KD around 10(-8) M. The lack of binding of SYNCAM18A to RS20 provides an explanation for the lack of smMLCK activation by SYNCAM18A. Altogether, these results demonstrate that peptide-resin can be used as a tool for separating bound from free protein, thus enabling a rapid and reliable quantification of the protein/peptide interaction.

Endothelin 1: conformation and aggregation.

The features of the far UV CD spectrum of endothelin 1 (ET 1) in water-containing solutions rules out the presence of any alpha-helical contribution, thus questioning the conclusions made by several authors on the basis of NMR investigations. We propose here a structural model, based on a succession of beta turns, which is consistent with both the NMR and the CD data. Using electron microscopy, we show that ET 1 can form "micelles," and the micelles self-associate into percolation clusters which have a fractal dimension of 1.23 in a 2D space. These data, too, are in agreement with our proposed structural model.