Protease cleavage site recognition motifs can be identified using protease substrate discovery methodologies, but typically exhibit non-optimal specificity and activity. To enable evolutionary optimization of substrate cleavage kinetics, a two-color cellular library of peptide substrates (CLiPS) methodology was developed. Two-color CLiPS was applied to identify peptide substrates for the tobacco etch virus (TEV) protease from a random pentapeptide library, which were then optimized by screening of a focused, extended substrate library. Quantitative library screening yielded seven amino acid substrates exhibiting rapid hydrolysis by TEV protease and high sequence similarity to the native seven-amino-acid substrate, with a strong consensus of EXLY phi QG. Comparison of hydrolysis rates for a family of closely related substrates indicates that the native seven-residue TEV substrate co-evolved with TEV protease to facilitate highly efficient hydrolysis. Consensus motifs revealed by screening enabled database identification of a family of related, putative viral protease substrates. More generally, our results suggest that substrate evolution using CLiPS may be useful for optimizing substrate selectivity and activity to enable the design of more effective protease activity probes, molecular imaging agents, and prodrugs. Biotechnol. Bioeng. 2010;106: 339-346. (C) 2010 Wiley Periodicals, Inc.