A next-generation Fab library platform directly yielding drug-like antibodies with high affinity, diversity, and developability

Abstract

We previously described an in vitro single-chain fragment (scFv) library platform originally designed to generate antibodies with excellent developability properties. The platform design was based on the use of clinical antibodies as scaffolds into which replicated natural complementarity-determining regions purged of sequence liabilities were inserted, and the use of phage and yeast display to carry out antibody selection. In addition to being developable, antibodies generated using our platform were extremely diverse, with most campaigns yielding sub-nanomolar binders. Here, we describe a platform advancement that incorporates Fab phage display followed by single-chain antibody-binding fragment Fab (scFab) yeast display. The scFab single-gene format provides balanced expression of light and heavy chains, with enhanced conversion to IgG, thereby combining the advantages of scFvs and Fabs. A meticulously engineered, quality-controlled Fab phage library was created using design principles similar to those used to create the scFv library. A diverse panel of binding scFabs, with high conversion efficiency to IgG, was isolated against two targets. This study highlights the compatibility of phage and yeast display with a Fab semi-synthetic library design, offering an efficient approach to generate drug-like antibodies directly, facilitating their conversion to potential therapeutic candidates.

Keywords: Fab; Recombinant antibody libraries; developability; phage display; yeast display.

Figure 1.

Characterization of the therapeutic Fab-phage. (a) schematic representation of library design and assembly. (b) titers of monoclonal Fab phage displaying therapeutic scaffold Fabs infection evaluated on XL1 blue cells. (c) Western blot of Fab-phage displaying therapeutic scaffolds with an anti-Myc antibody. (d) Fab phage ELISA testing the specificity of each of the therapeutic antibody scaffolds for binding to their corresponding antigens (when available as recombinant protein). An unrelated antigen was used as negative control and anti-Fab antibody was used to detect the display. (e) binding specificity of each of the therapeutic antibody scaffolds when displayed as scFab on yeast to their corresponding antigens (when available as recombinant protein). Yeast cells were tested only with the secondary reagents, as well as an unrelated antigen (off-target).

Figure 2.

Schematic representation of the 2-step PCR conversion of Fab phage display into scFab yeast display. (a) graphic representation of the assembly strategy to generate the semi-synthetic Fab library. (b) number of transformants obtained after electroporation of each phage display sublibrary into E. coli TG1. (c) inverse-pcr based amplification of the phage display vector with a 5’ primer annealing to the FW1 of the VH and 3’ primer annealing to the end of the constant light chain region (CK), resulting in the linearization of the entire plasmid without the plasmid elements between the two domains constituting the Fab. (d) the primers confer complementary overhangs compatible to the 5’ and 3’ ends of a donor fragment amplified from the yeast DONOR vector. The donor fragment contains the linker for the generation of the scFab. (e) the intermediate vector is obtained by fusing the donor fragment in frame with the two domains of the Fab by NEBuilder® assembly kit. (f) the cassette from the intermediate vector pool is PCR amplified adding sequence homologous to the final yeast display vector at the ends of the fragment. (g) cloning of the cassette into the yeast display vector via homologous recombination.

Figure 3.

Quality control of the assembled library. a) schematic representation of the pre- and post-packaging experiments for evaluation of clone functionality and Sanger sequencing. Post-packaging refers to colonies picked after packaging the phagemid into phage particles and reinfecting cells, prepackaging refers to colonies picked directly from the library transformation. b) number of in-frame clones identified by Sanger sequencing in the naive library pre- and post-packaging. c) summary of Fab display levels of phage clones pre- and post-packaging tested by sandwich ELISA.

Figure 4.

Quality control of the library by NGS. a) presence of sequence liabilities in the naturally replicated CDRs (HCDR1-2; LCDR1-3) used to generate sublibraries based on PacBio sequencing. b) accumulation plot of Fab library unique clones, compared to previous library iterations from NovaSeq sequencing analysis. c) accumulation plot of unique clones in the individual sublibraries.

Figure 5.

Antibody selection. a) enrichment of phage display selection titers for the 5 sub-libraries selected against hIFNα-2b and SARS-CoV-2 RBD. b) ELISA signals of polyclonal phage after 3 rounds of display tested against hIFNα-2b and SARS-CoV-2 RBD, which were also used as cross-reactivity controls for each over. Display levels were also tested (anti-Fab).

Figure 6.

Sorting results of scFab yeast display. Flow cytometry analysis of scFab yeast display against hIFNα-2b and SARS-CoV-2 RBD, a) after directly subcloning the third rounds of phage display selection and b) after the final yeast sorting.

Figure 7.

Fab phage scFab yeast outputs analysis. Binding activity of monoclonal scFab antibodies displayed on yeast against a) hIFNα-2b and b) SARS-CoV-2 RBD. Binding fold change over the off-target is reported. c) unique HCDR3s for each antigen isolated from the Fab-phage and the scFab-yeast selection outputs, with overlaps indicated.

Figure 8.

PacBio sequencing analysis of the selected scFab yeast clones. a) VL and VH scaffold pairing for both antigen Fab phage selection outputs. Values are normalized by VL (rows sum to 100%). b) VL and VH scaffold pairing for both antigen scFab yeast selection outputs. Values are normalized by VL (rows sum to 100%). c) number of unique clones identified from each sublibrary for each antigen for the Fab phage selection outputs. d) number of unique clones identified from each sublibrary for each antigen for the scFab yeast selection outputs. The number of unique HCDR3s for each antigen isolated from the Fab-phage and the scFab-yeast selection outputs identified by PacBio sequencing of the full Fab or scFab, and corresponding overlaps for e) hIFNα-2b and f) SARS-CoV-2 RBD. ScFab clones ranked by frequency for g) hIFNα-2b and h) SARS-CoV-2 RBD: large circles represent clones shared between scFab and Fab outputs and small circles represent clones exclusive to scFab output. Color indicates clones’ respective rank in the Fab population with scFab-only clones taking on values greater than the highest ranked shared clone.

Figure 9.

scFab to IgG conversion. a) IgG expression levels in supernatant of transfected Expi293F cells b) % IgG conversion and binding to corresponding target.

Figure 10.

Developability assessment. All boxplots in each of the panels reflect the median, first quartile (25^th percentile) and third quartile (75^th percentile) of all the antibodies, with the whiskers reflecting 1.5× the interquartile range (IQR). All datapoints are plotted as jittered datapoints across the plot using the varying metrics with dark blue values indicative of acceptable developable readings and dark red indicating poorly developable measurements. a) antibody affinities measured by surface plasmon resonance (SPR). b) aggregation and fractionation by size-exclusion chromatography (SEC) expressed as a percentage of the main peak using the area under the curve (AUC). c) hydrophobicity profiles assessed by hydrophobic interaction chromatography (HIC), with the AUC of the main peak expressed relative to other peaks across the profile. Fab stabilities determined by d) melting temperature (T_m) collected using the BCM of the fluorescence data from the inflection point (maximum of the first derivative) of the thermal profile and e) the aggregation onset temperature (T_onset) in which we observe a significant elevation of the SLS data. Charge variants were assessed by f) capillary gel electrophoresis (cGE) assessing the AUC of the main peak relative to other peaks across the profile. Aggregate or impurities were analyzed by dynamic light scattering (DLS) at 280 nm absorbance to measure g) the peak of interest mean diameter (nm), h) the peak of interest mass percent (%) and i) the Polydispersity Index (PdI), a unitless metric.