Here we describe CoESPRIT, a library scale construct screening methodology that incorporates co-expression of bait proteins into our previously reported protocol for identifying soluble sub-constructs of a target protein. It allows identification of protein domains or fragments that interact with a fixed bait protein and is therefore similar in concept to deletion analyses by yeast-2-hybrid screening, but with the important distinction that the yields of purified protein are compatible with downstream studies such as crystallisation, NMR, biophysical methods, as well as recombinant vaccine testing, that require multi-milligram quantities of material. Additionally, the E. coli host provides the option of labelling proteins with isotopes or heavy atoms as needed that is less straightforward with other organisms. Compatible plasmids were designed that permitted screening of the library of constructs in a strain prepared as competent cells coexpressing the bait. The first colony screening step identifies putatively soluble forms of the target that may or may not complex with the bait. As recently described, simultaneous analysis of fused N- and C-terminal peptide tags from colony signals of a truncated target provides a means to eliminate the out-of-frame constructs that are a majority species in DNA truncation libraries, together with in-frame gene products that are post-translationally degraded. A second step of purification screening of initial positives confirms the soluble phenotype of the target, and identifies library members that co-purify with the bait. The main advantages of this approach are that expression of unsuspected or difficult-to-predict domains can be achieved without prior knowledge of the domain content of the target, and that folding-upon-binding effects can stabilise fragile or otherwise insoluble constructs where complex formation is required to form a hydrophobic core and protect the target from aggregation or proteolysis. A standard and convenient screening capacity is approximately 27,648 clones that occupies seventy-two 384 well plates; this can be picked in two days and then arrayed onto a single membrane for expression screening. Such a capacity can be used to screen a library of a single target with high oversampling for unidirectionally truncated genes, or with about 5% coverage of total diversity for inserts truncated at both ends simultaneously. Here we constructed two unidirectional truncation libraries and screened them against 4 different baits, in parallel on the same membrane. At this level of sampling it would not be feasible to screen a target library against a bait library since the diversity of clones in such an experiment would greatly exceed the screening capacity of the automation. However, a sparser sampling of both constructs randomised simultaneously might in some cases allow Everolimus mTOR inhibitor direct identification of complexes that could be further refined by subsequent steps of rescreening individually, or by limited proteolysis and mass spectrometry depending on the requirements of the downstream application. In a first step in this direction, we have demonstrated how a library-based screening strategy can permit identification of an optimal form of the bait protein amongst several similar candidates as shown by screening one PB2 library against three different PB1 fragments where only one of the Tasocitinib latter proved competent for complex formation. Such coverage of construct diversity would be practically impossible using classic PCR cloning strategies due to issues of clone handling and cost of reagents. Here it is a relatively simple procedure because all constructs of a target are made in a single reaction tube and plasmid molecules clonally separated by bacterial transformation and robotic colony picking.