The pairing of heavy and light chain V-genes from each family occurs in proportion to their abundance in the library (data not shown), indicating random pairing as expected with the library construction AG-014699 cost method that was employed.

Previous data suggests random pairing also occurs in the human repertoire (de Wildt et al., 1999). Each library was assessed by selection against seven targets: gastrin (a 14 amino acid peptide), β-galactosidase (a bacterial protein, β-gal), human proteins insulin receptor in complex with insulin (InsR + Ins), TIE1, TIE2, TIE2 in complex with angiopoeitin 1 (ANG1), and TIE2 in complex with angiopoeitin 2 (ANG2). Three rounds of panning were performed for each target using previously described panning methods (Hoet et al., 2005 and Bhaskar et al., 2012). For each target, five to ten 96-well HSP inhibitor plates of clones were screened either by ELISA (gastrin, β-gal, TIE1, TIE2, and TIE2 complexes) or flow cytometry (InsR + Ins (Bhaskar et al., 2012), TIE2, and TIE2 complexes) for binding to the target. The clones that bound to their target were sequenced to identify unique clones. The unique clones were then analyzed for VH and VL family representation (Fig. 4), for CDR3 length of the VH and VL, and to assess the germline representation in FR1–FR3 of the selected clones (Table 2). Once unique clones were identified for each

target, further Florfenicol characterization of those clones was performed. For both libraries, the unique clones that bound to β-gal and TIE1 were prepared

as soluble antibody fragments in periplasmic extracts (PPE) and the KD (equilibrium dissociation constant) was determined using Biacore. For both targets, multiple antibody fragments with high affinities (single-digit nM to triple-digit pM) were identified. Table 2 lists the best KD identified for each target per library (Fig. S3). When screening the panning campaigns of TIE2 in complex with either of its ligands (ANG1 or ANG2), antibody fragments in PPE were screened by flow cytometry for binding to TIE2 or TIE2–ligand complex, and screened by ELISA for binding to ANG1 or ANG2. Binders in three categories emerged: single-protein binders (TIE2, ANG1 or ANG2), TIE2/ANG1-complex binders, or TIE2/ANG2-complex binders (Table 3). A subset of 10 Fab clones and 8 scFv clones that bound TIE2 was reformatted as IgG and the KD for each clone was determined with Biacore (Table 2 and Fig. S3). For clones from XscFv2, 6 out of 8 clones have KDs > 8 nM. For clones from XFab1, 9 out of 10 have KDs > 11 nM with two of these clones having KDs in the pM range (Fab09 = 800 pM and Fab10 = 500 pM). The sequences of the 591 unique selected clones for both libraries were compared to each other and aligned to the closest germline sequence.