Genetically encoded discovery of perfluoroaryl macrocycles that bind to albumin and exhibit extended circulation in vivo

Abstract

Peptide-based therapeutics have gained attention as promising therapeutic modalities, however, their prevalent drawback is poor circulation half-life in vivo. In this paper, we report the selection of albumin-binding macrocyclic peptides from genetically encoded libraries of peptides modified by perfluoroaryl-cysteine S_NAr chemistry, with decafluoro-diphenylsulfone (DFS). Testing of the binding of the selected peptides to albumin identified SICRFFC as the lead sequence. We replaced DFS with isosteric pentafluorophenyl sulfide (PFS) and the PFS-SICRFFCGG exhibited K_D = 4-6 µM towards human serum albumin. When injected in mice, the concentration of the PFS-SICRFFCGG in plasma was indistinguishable from the reference peptide, SA-21. More importantly, a conjugate of PFS-SICRFFCGG and peptide apelin-17 analogue (N₃-PEG₆-NMe17A2) showed retention in circulation similar to SA-21; in contrast, apelin-17 analogue was cleared from the circulation after 2 min. The PFS-SICRFFC is the smallest known peptide macrocycle with a significant affinity for human albumin and substantial in vivo circulation half-life. It is a productive starting point for future development of compact macrocycles with extended half-life in vivo.

Fig. 1. Albumin binding peptides.

Previous reports of A macrocyclic peptide: DX-236, B macrocyclic peptide: SA-21, C a linear peptide: FITC-EYEYK_palmESE-NH₂. D This report describes a chemically modified phage-displayed library for discovery of a small macrocyclic albumin binder.

Fig. 2. Panning with chemically modified phage libraries.

A Modification of phage-displayed SXCX₃C disulfide library by DFS to yield OFS-SXCX₃C library. B The OFS-SXCX₃C library panned against a mixture of biotinylated HSA and His-tag expressed T4-GP in solution containing unlabelled milk proteins. Targets were captured separately with avidin beads and Ni-NTA beads affinity beads. In the negative control, OFS-SXCX₃C library was panned against biotinylated ConA and captured with avidin beads. C Volcano plots and D Venn diagram visualizing the sequences from the OFS-SXCX₃C phage-displayed library that were significantly enriched in the HSA screen compared to the naive library or selection against T4-GP, ConA. E A heat map display of the top 25 of 85 hits sequences from differential enrichment results. F Dipeptide motif analysis of all 85 hits. G Selected sequences for chemical synthesis of macrocycles for validation.

Fig. 3. ¹⁹F NMR measurement of macrocycle-albumin interactions.

A ¹⁹F NMR binding assay for macrocycles 12c (PFS-STCHANCGKKK), 14c (PFS-SICRFFCGGG), 15c (PFS-SFCPMFCGGG), 16c (PFS-SLCKRECGGG), and 17c (PFS-STCQGECGGG) at 20 µM against varying concentrations of HSA. B ¹⁹F NMR signals and C extrapolated binding curves for 14c (PFS-SICRFFCGGG) and five alanine mutants (21c, 22c, 23c, 24c, 25c) of this macrocycle.

Fig. 4. Binding between selected macrocycles and HSA measured using fluorescent polarization.

A FP assay measured the K_D of macrocycles 18d and 19d against mouse serum, HSA, and fatty acid-free HSA (n = 3). B The FP assay for BODIPY labelled 18d titrated against mouse serum (black), HSA (green), lysozyme (red), and RNAse A (blue) (n = 2). Bars represent mean values ± SD).

Fig. 5. Docking predictions of HSA-14c complex.

A Macrocycle 14c shows the lowest binding energy when binding to pocket 1 near the Hemin-binding site of HSA; three helical domains (DI, DII, DIII), the subdomains (A and B) and binding sites for ibuprofen and diclofenac are mapped on the same HSA structure. B SMD and umbrella sampling trajectories in site 1 revealed that ARG4 of the 14c forms stable salt bridges with ASP183 of the albumin whereas this salt bridge is absent in other binding modes (C, D). E K_D values measured by ¹⁹F NMR. F ΔΔG for alanine point mutations calculated by free energy perturbation (FEP). G Correlation of experimental ΔΔG calculated from K_D and ΔΔG from FEP calculation for the alanine mutants.

Fig. 6. Pharmacokinetic studies of SICRFFCGGG macrocycle in mice serum.

A Lead sequence SICRFFC modified with N- and C-terminal extensions and different linchpins. B A mixture of compounds, including SICRFFC modifications, alanine scans, together with SA-21 injected into mice and monitored for their retention up to 1 h (n = 3 mice for each peptide). C N- and C-terminal extensions on PFS-SICRFFC macrocycle for payload attachments monitored for circulation time. n = 4 mice for each peptide. D Monitoring the effects of alanine mutants and C-terminal propargyl glycine on retention time. n = 4 mice for each peptide. E Effects of different linchpins on circulation time. n = 4 mice for each peptide. F 14c conjugated to therapeutically relevant payload, apelin-17 analogue (N₃-PEG₆-NMe17A2). Bars and boxes represent mean values ± SD.