Fig 1.
DCP phage library design and production.
(A) Structures of wild-type DCPs utilized for constructing the DCP phage libraries. The residues and loops (L) highlighted in pink were subjected to hard randomization or extended to create libraries. Structures were derived from the following PDB files: Mch1 (2M2Q), gurmarin (1C4E), Asteropsin (2LQA), AMP-1 (1MMC), and CPI (4CPA). (B) Expression analysis of four wild-type DCPs with SUMO tag (kB1: kalata B1, MW: 15908.7 kDa; kB1 thr: kalata B1 thrombin binder, MW: 15913.8 kDa; AVR9, MW: 16629.5 kDa; circulin A, MW: 16225.2 kDa). Bacterial expression vectors were generated containing open reading frame with His(6) tag, SUMO, and a TEV protease cutting site at the N-termini of DCPs. The expression profile of the fusion proteins was analyzed with SDS-PAGE. T: total lysate; S: soluble fraction; E: elute. (C) Folding analysis of recombinant wild-type cellulose binding domain (CBD). After TEV protease digestion, the folding profile of CBD was analyzed with LC-MS to show success removal of the SUMO-His(6) tag and confirms the identity of fully oxidized CBD. (D) DCP production using chemical synthesis and thermodynamic oxidation. After linear peptide synthesis and purification, small scale folding analysis was performed using various folding conditions to determine the optimal folding condition yielding a major product containing the three disulfide bonds. The folding product then was desalted using a C18 column, lyophilized and purified through HPLC. The final product was lyophilized and analyzed with LC-MS. (E) Workflow of DCP phage panning. DCP phage libraries were used to select against specific protein targets, followed by washing and elution of the phage displaying DCPs binding to the target. The selection process was repeated four times to improve binder affinity as well as specificity. Individual phage containing DCP binders was isolated and ranked using phage spot ELISA, Sanger sequencing and NGS. The cartoon was created with BioRender.com.
Fig 2.
Identification of DCP binders to human IgG Fc or human serum albumin.
(A) Representative sequence consensus analysis of top binders from phage spot ELISA. From left to right: Ig-CON DCPs (loop 2, 11 aa variants, n = 21); Ig-AMP DCPs (loop 2, 4 aa variants, n = 9); Hs-CHR DCPs (loop 3, 4, 5 and N-term, n = 9). (B) Recombinant production of Ig-DCPs (MW: 16–17 kD). Representative SDS-PAGE analysis of purified His(6)-SUMO-TEV-DCPs. Left panel: Ig-CON-1, MW: 16276.2 kDa; Ig-CON-2, 16625.6 kDa; Ig-CON-3, 16451.3 kDa; Ig-CON-4, 16283.2 kDa; Ig-CON-5, 15994.9 kDa; Ig-CON-6, 16438.4 kDa; Ig-CON-7, 16629.5 kDa; middle panel: Ig-AMP-1, MW: 16114.1 kDa; Ig-AMP-2, MW: 16057 kDa; Ig-AMP-3, MW: 16114 kDa; Ig-AMP-4, MW: 16073 kDa; right panel: Ig-EET-1, MW: 16555.5 kDa; Ig-EET-2, MW: 16306.2 kDa; Ig-EET-3, MW: 16171.1 kDa; Ig-EET-4, MW: 16415.4 kDa.
Fig 3.
Synthetic DCPs bind to human IgG Fc or human serum albumin with sub-micromolar affinity.
(A) Sequences of top DCP binders to human IgG Fc or human serum albumin. Loops that have been changed from the wild-type DCP sequences are highlighted in color. DCP variants with signal (binding against the target) to noise (binding against the control) (s/n) ratios (from phage spot ELISA) higher than 4 were selected and tested according to the procedure illustrated in Fig 1D, and the lead DCPs were chemically synthesized and analyzed with phage ELISA and SPR to identify the DCPs with the highest binding affinity from each scaffold. (B) Representative SPR sensorgrams showing dose response binding of top DCP binders to immobilized human IgG Fc (measured by single-cycle kinetics). Black: raw data; red: fitted data. (C) Phage competition ELISA shows synthetic DCPs compete with DCPs displayed on phage surface for binding against human IgG Fc or serum albumin. (D) Comparison of IC50 and KD values of the top DCP binders against human IgG Fc or serum albumin. Open bar, IC50 values from phage ELISA; filled bar, KD values from SPR. All values were averaged from three independent runs. Error bar: standard deviation from at least three independent experiments. (E) Representative SPR sensorgram showing dose response binding of Hs-CHR-3 to immobilized human serum albumin (measured by multi-cycle kinetics with a 3-fold serial dilution starting from 5 μM).
Fig 4.
DCPs bind at distinct epitopes on IgG and human serum albumin.
(A) Ig-DCPs bind less potently to rabbit and very weakly to mouse IgG Fc. They specifically bind to human IgG Fc over human IgA or IgM. Protein A was used as a control. Red: estimated values due to poor fitting. *: estimated value due to possible non-specific binding. (B) Hs-Chr-3 binds to human and rat (weakly), but not to rabbit or mouse albumin (N.A., not applicable). All KD values were obtained using SPR. SA21 was used as a control. (C) Select Ig- and Hs-DCPs are stable in 100% human serum for at least 24 h at 37°C. DCP samples were analyzed using LC-MS and percentage of remaining peptide was calculated. Open bar, 0 h; solid bar: 24 h. Error bar: standard deviation from three independent experiments. (D) Ig-AMP-1 and Ig-CON-5 bind to a similar epitope. In the phage competition ELISA assay, Ig-AMP-1 and Ig-CON-5, but not Ig-EET-1, potently inhibited the interaction between biotinylated hIgG Fc and phage displaying Ig-AMP-1. (E) In the protein A-Fc interaction ELISA, the hIgG Fc binding peptide, Fc-III, but not Ig-CON-5 or Ig-AMP-1, disrupted the protein A (immobilized on the plate)-hIgG Fc binding. Ig-EET-1 also partially disrupted the binding between protein A and hIgG Fc. (F) Ig-EET-1 binds to a different epitope from Fc-III. In the phage competition ELISA assay, Ig-EET-1, but not Fc-III, potently inhibited the interaction between biotinylated hIgG Fc and phage displaying Ig-EET-1. Error bar: standard deviation from duplicate samples. Representative data from at least three independent experiments are shown. (G) Model of the mode of action of Ig-EET-1, Ig-AMP-1, Ig-CON-5 and Fc-III. Fc-III binds to the protein A binding pocket (the hinge between CH2 and CH3 domains) on hIgG Fc, whereas Ig-EET-1 binds to an epitope partially overlapping with the protein A binding site. Fc-III and Ig-EET-1 do not share binding epitopes. Ig-CON-5 and Ig-AMP-1 largely share the same binding epitope on hIgG Fc, which is away from the protein A binding pocket. They both partially share binding epitopes with Ig-EET-1. (H) Hs-CHR DCPs share partial binding epitope with SA21. In the phage competition ELISA assay, all synthetic Hs-CHR DCPs potently inhibited the interaction between biotinylated human serum albumin (immobilized on the plate) and phage displaying Hs-CHR-3. SA21 partially disrupted the interaction. Error bar: standard deviation from duplicate samples. Representative data from at least three independent experiments are shown. (I) Model of the mode of action of Hs-CHR DCPs. Hs-DCPs bind to the similar epitope on human serum albumin, which partially overlap with the SA21 binding site.
Fig 5.
Discovery of DCP binders to human VEGF-A (V-DCP), human PDGF (P-DCP) or Ly6E (L-DCP).
(A) Sequences of V-DCP, P-DCP or L-DCP. Loops that have been changed from the wild-type DCP sequences are highlighted in color. (B, C) Representative SPR sensorgrams showing dose response binding of top DCP binders to immobilized protein targets (V-EET-3 and V-CON-1 were measured by single-cycle kinetics; V-AMP-1 and P-EET-6 were measured by multi-cycle kinetics with a 3-fold serial dilution starting from 50 μM). Black: raw data; red: fitted data. (D) Representative SPR sensorgrams showing dose response binding of L-EET-7 to immobilized Ly6E-rat IgG Fc (left panel; black: raw data; red: fitted data), or rat/human IgG Fc (right panel; black: raw data against rat IgG Fc; red: fitted data against rat IgG Fc; blue: raw data against human IgG Fc). All were measured by single-cycle kinetics.