Designed armadillo repeat proteins (dArmRPs) bind extended peptides in a modular way. The consensus version recognises alternating arginines and lysines, with one dipeptide per repeat. For generating new binding specificities, the rapid and robust analysis by crystallography is key. Yet, we have previously found that crystal contacts can strongly influence this analysis, by displacing the peptide and potentially distorting the overall geometry of the scaffold. Therefore, we now used protein

more »

... gn to minimise these effects and expand the previously described concept of shared helices to rigidly connect dArmRPs and designed ankyrin repeat proteins (DARPins), which serve as a crystallisation chaperone. To shield the peptide-binding surface from crystal contacts, we rigidly fused two DARPins to the N-and C-terminal repeat of the dArmRP and linked the two DARPins by a disulfide bond. In this ring-like structure, peptide binding, on the inside of the ring, is very regular and undistorted, highlighting the truly modular binding mode. Thus, protein design was utilised to construct a well crystallising scaffold that prevents interference from crystal contacts with peptide binding and maintains the equilibrium structure of the dArmRP. Rigid DARPin-dArmRPs fusions will also be useful when chimeric binding proteins with predefined geometries are required. Designed armadillo repeat proteins (dArmRPs) bind extended peptides in a modular way. The consensus version recognises alternating arginines and lysines, with one dipeptide per repeat. For generating new binding specificities, the rapid and robust analysis by crystallography is key. Yet, we have previously found that crystal contacts can strongly influence this analysis, by displacing the peptide and potentially distorting the overall geometry of the scaffold. Therefore, we now used protein design to minimise these effects and expand the previously described concept of shared helices to rigidly connect dArmRPs and designed ankyrin repeat proteins (DARPins), which serve as a crystallisation chaperone. To shield the peptide-binding surface from crystal contacts, we rigidly fused two DARPins to the Nand C-terminal repeat of the dArmRP and linked the two DARPins by a disulfide bond. In this ring-like structure, peptide binding, on the inside of the ring, is very regular and undistorted, highlighting the truly modular binding mode. Thus, protein design was utilised to construct a well crystallising scaffold that prevents interference from crystal contacts with peptide binding and maintains the equilibrium structure of the dArmRP. Rigid DARPin-dArmRPs fusions will also be useful when chimeric binding proteins with predefined geometries are required. Designed Armadillo repeat proteins (dArmRPs) bind to elongated peptide sequences and may eventually complement classical detection antibodies (a review on alternative binding scaffolds can be found in ref. 1 , their use in therapeutics is reviewed in ref. 2 ). dArmRPs are based on the helical natural armadillo repeat proteins (nArm-RPs), an α-solenoid repeat protein family that binds to stretches of unfolded regions of proteins 3-5 . Over the last years, a monomeric and well expressing scaffold has been derived from ArmRPs by protein engineering 6-10 (further reviewed in refs 11,12 ). Stable and regularised repeats have been derived from the more irregular repeats of nArmRPs. In these proteins, a varying number of internal repeats, binding the peptide, and capping repeats, which shield the hydrophobic core from the aqueous environment, are stacked together to form a superhelical repeat protein (Fig. 1a) . By stacking the internal repeats, or binding modules, a large concave and solvent-exposed binding surface is formed (Fig. 1a) . Ideally, different modules, each recognising two side-chains of the peptide target, would be recombined to bind to arbitrary peptide sequences. These binding modules are obtained by preselection from libraries and/or protein design. In principle, these modules could be reassembled in any desired arrangement to bind new targets, thereby making costly individual selections against new targets unnecessary. So far, modules binding to arginine-and lysine-rich peptides have been designed and the development of new binding pockets and modules is under way, both by selection techniques and by computational design methods (unpublished data). A key analysis technique for the design process is macromolecular crystallography, and numerous crystal structures have confirmed the sequence-specific binding of dArmRP to different target peptides 10,13-16 . However, it soon became clear that crystal forces can have a large impact on the outcome and interpretation of experimental crystal structures 13, 14, 16 . By analysing 27 crystal structures of dArmRPs without peptide or in complex with their cognate target peptides, we found that, depending on the crystallisation conditions and