An amber obligate active site-directed ligand evolution technique for phage display

Abstract

Although noncanonical amino acids (ncAAs) were first incorporated into phage libraries through amber suppression nearly two decades ago, their application for use in drug discovery has been limited due to inherent library bias towards sense-containing phages. Here, we report a technique based on superinfection immunity of phages to enrich amber-containing clones, thus avoiding the observed bias that has hindered incorporation of ncAAs into phage libraries. We then take advantage of this technique for development of active site-directed ligand evolution of peptides, where the ncAA serves as an anchor to direct the binding of its peptides to the target's active site. To demonstrate this, phage-displayed peptide libraries are developed that contain a genetically encoded butyryl lysine and are subsequently used to select for ligands that bind SIRT2. These ligands are then modified to develop low nanomolar inhibitors of SIRT2.

Fig. 1. Overview of phage-assisted active site-directed ligand evolution.

a A diagram that illustrates the interaction between a protein target and a ligand. b The genetic incorporation of a ligand-fused ncAA into phage-displayed peptides for active site-directed binding to a protein target that is followed by stringent wash and elution to select high-affinity and selective phages. c ncAAs that have been genetically incorporated and can potentially serve as ligands for epigenetic regulators (HAT histone acetyltransferase, KDM protein lysine demethylase, KMT protein lysine methyltransferase, Bromo, YEATS, chromo, Tudor, and PhD are epigenetic reader domains).

Fig. 2. A schematic representation of a method for constructing an amber-obligate phage display library by superinfection-immunity-based selection.

a, b Following cloning, the naive phagemid library is used to transform non-amber-suppressing E. coli bearing an F sex pilus. The expression of pIII is induced with IPTG, and shortly after, cells are superinfected with the CM13 helper phage (a). The expression of pIII in cells harboring a copy of the library that contains only sense codons renders these cells immune to superinfection (b). c Changing the media to one containing kanamycin allows for the selective growth of cells harboring a copy of the library that contains in-frame amber codons. Cells harboring a copy of the library with deleterious mutations also pass this selection. d, e The phagemid library is purified (d) and used to transform DH5α, an amber-suppressing strain of E. coli (e). Complementation of the phagemid with a pIII-knockout helper phage in E. coli DH5α allows for the production of phagemid particles only from cells harboring a copy of the peptide-pIII library containing in-frame amber codons.

Fig. 3. Characterization of the amber-obligate phage library generated from superinfection-immunity-based selection.

a During two rounds of superinfection-immunity-based selection, the percentage of the population that was susceptible to superinfection by the CM13 helper phage (e.g., clones that contained amber codons) was monitored by growth on kanamycin. The increase in the number of cells that were susceptible to superinfection between rounds 1 and 2 is indicative of the successful selection of library clones containing amber codons. b The occurrence percentages of TAG codons across the seven randomized positions in the enriched amber- obligate phage library. c A heat map showing the occurrence abundance (amount at position/total sequences) of each amino acid or the TAG codon across all library positions. d A heat map showing deviation from theoretical randomization for each amino acid or the TAG codon at all library positions. Deviation is calculated by (observed abundance-expected abundance)/expected abundance.

Fig. 4. The genetic incorporation of ncAAs into a phage-displayed peptide library.

a A schematic representation of the three-plasmid system used to incorporate ncAAs into the peptide library. b The structures of phenylalanine derivatives 1–4. c Phage yield in the presence and absence (NA) of 4 mM ncAAs 1–4. The yield is displayed in millions of colony-forming units per milliliter of culture media. Error bars represent one standard deviation of the mean of three independent experiments (n = 3). d The structures of two post-translationally modified lysines, 5 and 6. e Phage yield in the presence and absence (NA) of 5 and 6. The yield is displayed in millions or hundreds of thousands of colony-forming units for compounds 5 and 6, respectively. Error bars represent one standard deviation of the mean of four independent experiments (n = 4). Data for the incorporation of ncAAs are available in the Source Data File.

Fig. 5. Phage-assisted identification of peptide inhibitors for SIRT2.

a Amino acid sequences identified through phage selection of peptides that bind to SIRT2. X = N^Ɛ-butyryl-l-lysine. b Binding and inhibition parameters of the selected peptides as N^Ɛ-butyryl-l-lysine (BuK), N^Ɛ-thiobutyryl-l-lysine (tBuK), or N^Ɛ-thiomyristoyl-l-lysine (tMyK) derivatives, along with standard compounds TB and TM. IC₅₀ values of tMyK derivatives are highlighted in red. K_d values are given as the mean ± standard deviation of three individual experiments (n = 3). IC₅₀ values are given as the mean ± standard deviation of two individual experiments (n = 2). Source data are available in the Source Data file.

Fig. 6. Binding and inhibition of human sirtuins by synthetic ligands derived from selected peptides.

a Fluorescent polarization assay of SIRT2 binding to FITC-conjugated S2P03 and its deacylated counterpart S2P03-K. Values are reported as the mean of three independent experiments (n = 3). b Inhibition of human SIRT1–3 by S2P04-tBuK (left) and TB (right). Values are reported as the mean of two independent experiments (n = 2). c Inhibition of SIRT2 by four tMyK-containing peptides. Values are reported as the mean of two independent experiments (n = 2). d Interactions between S2P03-tBuK and SIRT2 predicted by molecular dynamic simulation. Data associated with all inhibition and binding curves can be found in the Source Data File.