Cyclic peptides: the membrane permeability challenge

A typical linear peptide of 6 to 12 residues faces a cascade of barriers before reaching its systemic target. In the stomach and small intestine, pepsin, trypsin, chymotrypsin, and aminopeptidases hydrolyze amide bonds with evolutionary efficiency. Even when intact fragments reach the enterocyte, their polar profile — multiple hydrogen bond donors and acceptors exposed to solvent — prevents partitioning into the apical membrane. The practical result is oral bioavailability that rarely exceeds 1 to 2 percent.

The second barrier is pharmacokinetic. Small linear peptides are freely filtered by the glomerulus and partially reabsorbed in the proximal tubule by transporters like PEPT1 and PEPT2, but their plasma half-life is measured in minutes. Without modification, therapeutic concentration is sustained only by continuous infusion or frequent subcutaneous dosing, a model incompatible with chronic therapy and adherence.

The third barrier is conformational. A linear chain in aqueous solution samples a vast conformer ensemble; that entropic cost is paid when binding the target and crossing membranes. Restricting flexibility is, by itself, a route to improve affinity and permeability simultaneously.

Head-to-tail cyclization links the N-terminus to the C-terminus through an amide bond, producing a continuous macrocycle. It is the most common strategy in natural peptides like gramicidin S and in synthetic derivatives. It reduces degrees of freedom, protects the ends from exoproteases, and favors conformational states with internalized hydrogen bonds, which lowers the effective polar surface area.

Side-chain-to-tail and side-chain-to-side-chain cyclizations exploit residues like Lys, Asp, Glu, or Cys to close the ring. Disulfide bridges between cysteines are characteristic of marine toxins and conotoxins, enabling highly rigid bicyclic architectures. Lactam bridges (amide between the Lys side chain and Asp or Glu) are orthogonal to disulfides and stable in reducing environments such as the cytosol.

N-methylation of selected peptide bonds is complementary to cyclization. Replacing the amide hydrogen with methyl removes a hydrogen-bond donor, lowers desolvation energy, and, at well-chosen positions, locks the closed conformer. Cyclosporin A combines both strategies: it is a cyclic undecapeptide with seven N-methylations, and its oral bioavailability is reported near 30 percent, an extraordinary value for a molecule above 1,200 Da.

Cyclosporin A remains the paradigmatic case: a natural macrocycle produced by Tolypocladium inflatum, it crosses membranes by passive diffusion despite grossly violating Lipinski's rule of five. Its clinical success in transplantation since the 1980s validated the hypothesis that the chemical space beyond-rule-of-five is habitable.

Octreotide and lanreotide are synthetic cyclic analogues of native somatostatin, stabilized with a disulfide bridge and D-amino acids to resist proteases. They are administered parenterally because their oral bioavailability remains low, but their extended plasma half-lives — hours instead of minutes — made chronic treatment of neuroendocrine tumors and acromegaly feasible.

Other relevant examples include vancomycin (a cyclic glycopeptide for intravenous use), daptomycin (a cyclic lipopeptide), and, more recently, macrocyclic drugs in oncology and antivirals that engage targets considered undruggable by classical small molecules, such as protein-protein interfaces.

The Caco-2 assay uses a polarized monolayer of human colon adenocarcinoma cells cultured on a porous support. It measures apical-to-basolateral and basolateral-to-apical flux of a candidate molecule, yielding an apparent permeability coefficient (Papp) and detecting transporter and efflux effects (P-glycoprotein, BCRP). It is the regulatory standard for predicting oral absorption.

The PAMPA assay (Parallel Artificial Membrane Permeability Assay) uses a phospholipid artificial membrane on a filter, with no cells or transporters. It is faster and cheaper than Caco-2, ideal for early library screening, and isolates the passive diffusion component. For cyclic peptides this matters because most successful macrocycles cross by passive mechanism rather than active transport.

Spectroscopic methods complement these assays, including NMR in solvents of variable polarity, which maps the fraction of internalized hydrogen bonds, and molecular dynamics simulations that estimate free energy of passage through model bilayers. No single method predicts human bioavailability accurately, but the combination yields an actionable profile.

Macrocycle library generation by mRNA display allows screening of more than 10^12 variants against a protein target in a single round. Platforms like RaPID and derivatives incorporate non-canonical amino acids, N-methylations, and non-natural chemical closures, expanding the chemical space far beyond the 20 proteinogenic amino acids.

The beyond-rule-of-five (bRo5) concept recognizes that molecules with molecular weight between 500 and 1,500 Da, high polar surface area, and many rotors can, under appropriate conformational restrictions, behave as oral drugs. Cyclic peptides are the natural inhabitants of this space. Work published in Nature Reviews Drug Discovery and Journal of Medicinal Chemistry maps which subregion of bRo5 space is actually penetrable and which is not.

In parallel, the field invests in formulations that protect the peptide in the digestive tract — enteric capsules, permeation enhancers like SNAC, device-assisted systems — and in conjugates with transport domains that cross epithelia by transcytosis. The combination of macrocyclic chemistry, formulation, and synthetic biology is turning peptides into a mature therapeutic modality.