Abstract
Recently we were able to apply the technique of 3D‐Rapid Prototyping (3D‐RP) to the construction of highly accurate three‐dimensional plastic models of biomolecules [Laub, M. et al. (2001), Materialwiss. Werkstofftech. 32, 926]. These models are derived from x‐ray crystallographic data and therefore represent exact replicas of the depicted molecules. Due to their accuracy these models should be suitable for the modelling of protein‐protein‐interactions. In a first study using 3D‐Rapid Prototyping models of bone morphogenetic protein‐2 (BMP‐2) we were able to identify a novel structural motive on the concave side of this protein which we termed anthelix since a left‐handed helix (radius ca. 0.8–1 nm, pitch 8–9 nm) can be fitted into this groove. Based on these structural findings we identified a 15mer polypeptide (KNMTPYRSPPPYVPP) from the Brookhaven database as a potential physiological ligand. Molecular docking studies using a geometric recognition approach confirmed the anthelix as a possible binding site for this peptide. However in affinity chromatography experiments no binding between BMP‐2 and the immobilized peptide was observed.
As the question arose whether 3D‐Rapid Prototyping is in general suitable for modelling protein‐protein interaction we used dimeric BMP‐2 to study exemplary monomer‐dimer interaction. Molecular docking studies using the monomeric BMP‐2 subunits predicted a structure which is nearly identical to that found in dimeric BMP‐2 (root mean square deviation < 1 Å) proving the suitability of geometric docking. 3D‐RP‐BMP‐2‐monomers (size 140 mm × 75 mm × 65; magnification ca. 22 × 10^6^ fold) constructed from dimeric BMP‐2 could be assembled by hand yielding a structure highly homologous to dimeric BMP‐2. Differences between the 3D‐Rapid Prototyping model of dimeric BMP‐2 and the assembled monomers arose in several gaps at the interface between the two monomers which are not visible in the dimeric structure. These gaps can be explained by the way the solvent‐accessible molecular surface is generated. During this process an exterior probe sphere is rolled over the spherical atoms of the molecule. Distances between the monomers smaller than the diameter of this sphere are bridged thus resulting in a coherent surface.
We conclude that 3D‐Rapid Prototyping is in general eligible for the modelling of protein‐protein‐interaction though there are further efforts needed to increase our understanding of this process.