De novo development of small cyclic peptides that are orally bioavailable

Abstract

Cyclic peptides can bind challenging disease targets with high affinity and specificity, offering enormous opportunities for addressing unmet medical needs. However, as with biological drugs, most cyclic peptides cannot be applied orally because they are rapidly digested and/or display low absorption in the gastrointestinal tract, hampering their development as therapeutics. In this study, we developed a combinatorial synthesis and screening approach based on sequential cyclization and one-pot peptide acylation and screening, with the possibility of simultaneously interrogating activity and permeability. In a proof of concept, we synthesized a library of 8,448 cyclic peptides and screened them against the disease target thrombin. Our workflow allowed multiple iterative cycles of library synthesis and yielded cyclic peptides with nanomolar affinities, high stabilities and an oral bioavailability (%F) as high as 18% in rats. This method for generating orally available peptides is general and provides a promising push toward unlocking the full potential of peptides as therapeutics.

Fig. 1. Synthesis and metabolic stability of dithioether cyclic peptides.

a, General depiction of the library synthesis strategy. Linear dithiol peptides are cyclized with bis-electrophiles and are N-acylated with carboxylic acids to generate large cyclic peptide libraries. b, Chemical structures of random cyclic peptides 1–10 and half-lives in RLMs. Mean and s.d. values of three independent measurements are indicated. c, Comparison of metabolic stability in RLMs for a peptide that degrades rapidly (peptide 1, left) versus slowly (peptide 2, right). The concentration of intact peptide is shown by the black curve. The concentration of metabolized peptide is shown in blue for the addition of a single oxygen and in red for the addition of two oxygens. Mean and s.d. values of three independent measurements are indicated. d, CL for the 10 peptides and two oral drugs in RLMs. Mean and s.d. values of three independent measurements are shown. Source data

Fig. 2. Cyclization and N-acylation of peptides for library generation.

a, Cyclization and acylation reactions shown for an example peptide. The peptide was analyzed by HPLC before and after the reactions measuring absorbance at 220 nm. Peaks for the peptide and the desired products are shown in red. The asterisk marks the peak corresponding to the quenched linker that became smaller after the acylation reaction. b, Cyclization of peptides with bis-electrophilic linkers L1–L4. Cyclization yield determined by LC–MS is shown. c, Efficiency of N-acylation reactions. Cyclic peptides and carboxylic acids activated as NHS esters were transferred by acoustic dispensing. Reaction yield is indicated as percent of cyclic peptide acylated with carboxylic acid and was determined by LC–MS. Source data

Fig. 3. Synthesis of library of 8,448 cyclic peptides and screen against thrombin.

a, Structures of amino acids, bis-electrophilic cyclization linkers and carboxylic acids used to synthesize the library. b, MW, PSA, number of HBDs, cLogP, number of HBAs and nRotBs of the cyclic peptides in the library. Peptides outside the rule-of-five (Ro5) space defined by Lipinski et al. are shown in gray. Asterisk indicates the physicochemical properties defined by Veber et al.. c, Chemical structures and activities of purified cyclic peptides of group 1 indicated in Extended Data Fig. 2. The common core of the structure is colored in black. Mean and s.d. values of three independent measurements are shown. Source data

Fig. 4. Thrombin inhibitor characterization and sub-library screen.

a, In vitro characterization of peptides 11–30. K_i and s.d. values of three independent measurements are shown. Proteolytic stability was measured by incubating peptide in SGF and SIF for 8 h at 37 °C. Mean and s.d. values of three independent measurements are shown. Passive membrane permeability was measured by PAMPA wherein cyclic peptides (50 μM) were added to the donor wells and incubated for 18 h. The permeability is indicated as the concentration in the acceptor well relative to the concentration in the wells at equilibrium (in %). Mean and s.d. values for three independent measurements are shown. logP_app is shown in Supplementary Table 2. Metabolic stability was measured by incubating the peptides with RLMs at 37 °C, and the CL was calculated based on the half-life for the disappearance of the peptide. Mean and s.d. values for three independent measurements are shown. b, Amino acids and cyclization linkers used for the synthesis of a sub-library based on cyclic peptide 11. c, In vitro characterization of peptides 31–37 as in a. For all four parameters, mean and s.d. values for three independent measurements are shown. Permeabilities were measured by incubating the PAMPA plates for 12 h. Source data

Fig. 5. Peptides cyclized by hydrocarbon linkers (based on peptide 32).

a, Structures of peptides 38–44 cyclized by RCM. The configuration of the alkene linker was not determined. b, Synthesis of peptides via RCM on solid phase. (i) Reductive amination: amine, NaBH₃CN, DMF, AcOH 1%, 14 h at RT; (ii) SPPS: amino acid, HATU, DIPEA, DMF, 1–14 h at RT; (iii) RCM: 1st Generation Grubbs Catalyst (30 mol%), 1,2-dichloroethane, 2 h at RT under inert atmosphere; (iv) Cleavage: TFA:DCM:TIPS (50:49:1), 1 h at RT. c, Inhibition of α-thrombin. Mean and s.d. values of three independent measurements are shown. d, PAMPA of cyclic peptides (50 μM) incubated for 13 h. Mean and s.d. values for three independent measurements are shown. e, Comparison of the metabolic stability in RLMs of peptide 32 and RCM analogue 43. Mean and s.d. values for three independent measurements are shown. f, Oral availability of peptide 43. Plasma samples were analyzed by LC–MS after p.o. and i.v. administration to rats (n = 3; 10 mg kg⁻¹). s.d. values are indicated. g, In vivo gastrointestinal stability of peptide 43. Organs were harvested 30 min after p.o. administration to a rat (n = 1), and the quantity of peptide was analyzed by LC–MS. Oxidized peptide refers to a +16 mass adduct on the LC–MS. RT, room temperature. Source data

Fig. 6. Cyclic peptides with improved metabolic stability and oral bioavailability.

a, Sub-library based on 43 in which the carboxylic acid chlorothiophene (shown in blue) was substituted with the 17 analogues shown on the right using the acoustic dispensing library synthesis strategy. Heat map of human and rat thrombin inhibition of the sub-library is shown below. The screen was carried out at a peptide concentration of 10 μM. b, Chemical structure of peptide 46. The molecule occurs in a 1:2 mixture of E:Z isomers, as determined by NMR (Supplementary Note 1). c, Inhibition of α-thrombin. Mean and s.d. values of three independent measurements are shown. d, Passive membrane permeability of the cyclic peptides (50 μM) was measured by PAMPA (12-h incubation). Mean and s.d. values of three independent measurements are shown. e, Metabolic stability of 46 in RLMs compared to the precursor compounds 32 (dithioether, chlorothiophene group) and 43 (RCM, chlorothiophene group). Mean and s.d. values for three independent measurements are shown. f, Specificity profiling of 46. APC, activated protein C; PK, plasma kallikrein. Mean values of three independent measurements are shown. s.d. values are indicated for thrombin. g, Oral availability of 46 in rats (n = 3; 10 mg kg⁻¹). s.d. values are indicated. Source data

Extended Data Fig. 1. Synthesis of peptides for library generation.

Structures of random dithiol peptides P1–P8 synthesized in 96-well plates and purified by precipitation in diethyl ether. The purity assessed by LC-MS is indicated. Source data

Extended Data Fig. 2. Screening 8,448 cyclic peptides against thrombin.

a, Quantification of the 384 linear dithiol peptides using Ellman’s reagent. b, Heat map with residual thrombin activity measured for each cyclic peptide, divided into the two linkers used for cyclization and the four different generated peptide formats. Individual peptide sequences are aligned on the vertical axis, and the acids used for N-acylation are on the horizontal axis. The screen was carried out at a peptide concentration of 10 μM. Cyclic peptides forming a structurally related cluster and containing the most active hit are highlighted as Group 1. c, Thrombin inhibition of replicate reactions for the 20 best hits. Source data

Extended Data Fig. 3. Thrombin inhibitor sub-library screen.

Combined screening for permeability (PAMPA, 14 h, 5 μM) and thrombin inhibition. The heat maps show thrombin inhibition in donor (upper) and acceptor wells (lower). The screen was carried out at a theoretical equilibrium concentration of 1 μM. Source data