Fig 1: Library sorting by FACS.Sorting of the naive library by FACS and quality assessment of each enriched group by flow cytometry. A Sorting of the naive Atezolizumab scFv library on the yeast surface compared to the WT scFv (after 29 h of off-rate competition). The right (highest) gate was set to collect mutants with slower off-rates while the middle (medium (blue cells)) gate was set for WT-level off-rates and the left (lowest (purple)) gate set for faster off-rates. The y-axis represents scFv expression (measured by the level of V5 epitope/AF647) while the x-axis is PD-L1 binding/PE intensity. % values represent each group’s percent of the total number of sorted cells. B Testing each sorted population for binding intensity relative to the WT after 8 h of off-rate competition. LB are low-binders (faster off-rates), MB are medium binders (WT off-rates), HB are high binders (slower off-rates). Analyzed 105 cells per group, mean fluorescent intensity (MFI) on the x-axis is given per plot, V5 expression (y-axis) from AF647 intensity while binding (x-axis) in PE intensity.
Fig 2: Computational and experimental pipeline.Schematic illustration of the RESP computational and experimental pipeline, U-PD-L1 and B-PD-L1 are unlabeled and biotin-labeled PD-L1, respectively.
Fig 3: AF2 models of designed interface contacts are different than those at the HA-PD1:PD-L1 interface.(A) Despite sharing a similar surface patch on HA-PD1, the designs make unique and different contacts than that of the natural ligand, PD-L1 (PDB 5IUS). One common feature, however, is the insertion of a hydrophobic residue (marked with a black star) into a hydrophobic pocket in HA-PD1. The AF2 models were energy minimized with Rosetta before generating the figures. Figure made in PyMOL where some residues were hidden for clarity and H68HA-PD1 is shown in every panel. (B) The table compares the designed sequences with the corresponding starting scaffold sequence provided as input to EvoPro. Scoring metrics for the design models are also provided for the Rosetta dG/dSASA and buried unsatisfied hydrogen bonds (bUN H-bonds) parameters.
Fig 4: The locations of mutations in the PD-L1-Atezolizumab complex.Location of mutations in the 21 mutants in the structure of the Atezolizumab heavy chain. Mutated residues are labeled and colored in yellow in the heavy chain (green). PD-L1 is colored blue while the light chain is orange. Structure from RCSB Protein Data Bank (5XXY) in Mol* Viewer45, 71.
Fig 5: Determination of the off-rate and KD of mutant 4.Experimental validation of the Koff and KD on the yeast surface. A WT vs. Mutant 4 scFv dissociation after 92 h at RT on the yeast surface, T1/2 is half-life. B Comparison of WT Atezolizumab, Mutant 4, Durvalumab, and Avelumab scFv dissociation over 28 h at RT on the yeast surface. C Binding affinity (KD) measurements determined on the yeast surface between scFv and PD-L1 (3 independent measurements, also see Supplementary Fig. 7)). Source data are provided as a source data file for this figure.
Supplier Page from Sino Biological, Inc. for Human PD-L1 / B7-H1 / CD274 Protein (His Tag), Biotinylated