Large scale production of the recombinant human cathepsins L and S was optimized. Final purity was nearly 100%, yield 65% and 41%, respectively. Cost-effective expression in Escherichia coli, inclusion body purification and solubilization followed modified standard protocols. Most contribution to the remarkable increase in over-all efficiency came from the subsequent steps: folding by dilution, selective HIC-capturing of the folded proenzymes, and auto-activation. The effort to optimize the process parameters for folding and activation was greatly reduced by central composite fractional factorial experimental design, considering curved responses as well as factor interactions. Theoretical and practical features of this powerful tool for experimental design are given. Yield in procathepsin S folding could be further increased by addition of an excess of its own native propeptide with known intramolecular chaperone function. This corroborates literature data on proenzyme folding and is broadly discussed in the light of the lower conformational stability of the prodomain compared to the catalytic unit. Auto-activation kinetics was largely different between the two related proenzymes; from its time course contribution of uni- and bimolecular events in proregion hydrolysis and rate constants were estimated. (c) 2007 Elsevier Inc. All rights reserved.