The most frequently used binding proteins used in research are monoclonal antibodies, made by the >40-year-old hybridoma technology, some with questionable performance [1,2]. More recently, recombinant antibodies and non-antibody scaffolds, selected from synthetic libraries, have started to provide access to molecularly defined molecules [3,4]. Nonetheless, all of these approaches require one to treat every target as a completely new project. This is currently unavoidable for folded proteins. However, we hypothesized that for unfolded proteins or unfolded stretches (tags, posttranslationally modified tails, denatured proteins on western blots), termed "peptides" for simplicity, the regularity of the peptide main chain can be exploited. If true, a modular detection system can be devised, which would ultimately allow one to generate a sequence-specific binding protein without experimentation.
The basis of our approach are Armadillo Repeat Proteins [5-15], which bind peptides in a completely extended way, providing a pocket for each side chain, and thus access to a modular approach. Combining evolutionary engineering, NMR, X-ray crystallography and structure-based computation, we have now achieved well crystallizing ArmRPs with bound peptides, picomolar affinities, and a well functioning selection and evolution technology, as well as a portfolio of biochemical and biophysical analysis technologies for the engineered ArmRPs. Progress in the various aspects will be summarized.