Hydrophobic membrane proteins perform a variety of important functions in the cell, but their structures are notoriously difficult to solve. Thus, new strategies to obtain crystals of membrane proteins for structure determination are critical. We aim to develop a toolbox of peptide specific single-chain antibody fragment chaperones engineered for hyper-crystallizability. These peptide sequences can be introduced into various regions of membrane proteins without interfering with protein function. The resulting protein-chaperone complex is expected to form a crystal lattice mediated by chaperone interactions. We have developed candidate scFv chaperone proteins binding hexa-histidine (His6) and EYMPME (EE) tags with improved biophysical features influencing crystallization propensity, including peptide affinity, stability and solubility. The scFv libraries were generated using a novel ligation-free technique, MegAnneal, allowing us to rapidly generate large libraries based on 3D5 scFv. We identified two candidate chaperones, 3D5/His_683, specific for His6 and 3D5/EE_48, specific for EE tags. Variants exhibit high solubility (up to 16.6 mg/ml) and nanomolar peptide affinities; complexes of 3D5/EE_48 with EE-tagged proteins were isolated by gel filtration. We have developed design rules for EE peptide placement at terminal, inter-domain or internal loop regions of the target protein to balance peptide accessibility for chaperone binding while retaining rigid protein-chaperone complexes suitable for crystallization. The 3D5/ His_683 crystallized in four different conditions, utilizing multiple space groups. The 3D5/EE_48 scFv was crystallized (3.1 Å), revealing a ~52 Å channel in the crystal lattice, which may accommodate a small peptide-tagged target protein. Our evolution experiments altered scFv surface residues, resulting in use of different crystallization contacts. Analysis of these crystal contacts and those used by crystallized 14B7 scFv variants, led us to postulate that lattice formation is driven by strong crystal contacts. To test this hypothesis, we introduced amino acid changes expected to reduce the affinity of the 3D5/EE_48 energetically dominant crystal contacts. This approach to crystal contact engineering may allow semi-rational control over lattice networks preferred by scFv chaperones. Co-crystallization trials with model proteins are on-going. These engineered scFvs represent a new class of chaperones that may eliminate the need for de novo identification of candidate chaperones from large antibody libraries. / text