G protein-coupled receptors (GPCRs) are the largest class of pharmaceutical protein targets, yet drug development is encumbered by a lack of information about their molecular structure and conformational dynamics. Most mechanistic and structural studies as well as in vitrodrug screeningwith purifiedreceptorsrequire detergent solubilization of the GPCR, but typically these proteins exhibit only low stability in detergent micelles. We have developed the first directed evolution method that allows the directselection of GPCRsstable in a chosen detergent from libraries containing over one hundred million individual variants. The crucial concept was to encapsulate singleEscherichia coli cells of a library, each expressing a different GPCR variant, to form detergent-resistant, semi-permeable nano-containers. Unlike naked cells, these containers arenot dissolved by detergents, allowing us tosolubilize the GPCR proteins in situ while maintaining an association with the protein's genetic information, a prerequisite for directed evolution. The pore size was controlled to permit GPCR ligands to permeate, but the solubilized receptor to remain within the nano-capsules.Fluorescently labeled ligands were used to bind tothose GPCR variants inside the nano-containers that remained active in the detergent tested.By using fluorescence-activated cell sorting, detergent stable mutants derived from two different family A GPCRs could be identified, some with the highest stability reportedin short-chain detergents. In principle, this method, named Cellular High throughput Encapsulation, Solubilization and Screening (CHESS), is not limited to engineering stabilized GPCRs, but could be used to stabilize other proteins for biochemical and structural studies.