A peptide that binds its target beautifully in a test tube and does nothing in a cell is a familiar disappointment. The usual diagnosis — "it did not get in" — is correct but unhelpful, because there are two quite separate ways a peptide can get in and they call for opposite design decisions. This page is about the second one: passive diffusion through the lipid bilayer, the route taken by cyclosporine A and the small number of macrocycles that behave like it.
Many cell-penetrating peptides rely on high cationic charge: arginine-rich sequences engage the anionic cell surface and enter through direct translocation and endocytic routes. Mechanisms vary with sequence, cargo and cell type, and uptake into a cell does not by itself mean delivery to the cytosol. Passive diffusion is the reverse problem. Here charge is the central obstacle, because moving a charged species into the hydrocarbon interior of a bilayer carries a large desolvation penalty, and the design effort goes into hiding polarity rather than displaying it. How much any particular charge costs depends on the rest of the molecule and on the conditions of the assay.
Confusing the two is a real and common error. A design that adds arginines to "improve uptake" of a macrocycle intended to diffuse passively will usually make it worse. If cationic uptake is the route you want, see our page on cell-penetrating peptides and on cell permeable peptides; everything below concerns the passive route.
To cross a membrane passively a molecule has to leave water behind. Every polar group that was hydrogen bonded to solvent must give those bonds up, and the energy cost of doing so is what blocks most peptides. Backbone amide N–H donors are the worst offenders: a linear peptide carries one per residue, each firmly solvated, and the cumulative penalty is prohibitive. This is why molecular weight alone is a poor predictor. The relevant quantity is how much polar surface the molecule actually exposes at the moment it tries to enter the bilayer.
That phrasing — at the moment it tries to enter — is the whole insight. Exposed polarity is not a fixed number.
Cyclosporine A is the canonical case and remains the clearest teaching example. Its crystal structure shows a closed, twisted β-pleated-sheet conformation held by four intramolecular hydrogen bonds, and the same closed form is seen by NMR in chloroform. Bound to its target cyclophilin, by contrast, it adopts an open conformation with none of those intramolecular hydrogen bonds1,3,6.
The explanation for its unexpected permeability, first proposed two decades ago and now the standard account, is that it switches: extended in water, where the backbone amides hydrogen bond to solvent, and folded in the membrane interior, where those same donors satisfy each other instead2. Intramolecular hydrogen bonds shield polar functionality and reduce the energetic penalty of desolvation2. A molecule that does this is called a chameleon, and chameleonicity — solvent-dependent conformational change — has become a design parameter in its own right.
The structural features associated with passive permeability in macrocycles share one property: they all remove or hide backbone N–H donors3.
The counterintuitive result worth knowing. Cyclosporine E illustrates the point. It differs from cyclosporine A by a single backbone methylation, yet it is about an order of magnitude less permeable. The clearest difference between the two is not which conformations are available but how quickly the molecule moves between them: the interconversion rates between the states favoured in water and in a non-polar environment are up to ten times slower for cyclosporine E5. Over-stabilising the folded form does not help, because a conformation locked in the closed form cannot reopen to dissolve in water on one side or to bind its target on the other. What is wanted is a switch, not a preference.
The same logic applies to lipophilicity. It is tempting to treat the problem as one-dimensional — more lipophilic, more permeable — but permeability and solubility pull against each other, and a compound that partitions enthusiastically into membranes may simply stay there. Permeability optimisation in this space therefore means tuning overall lipophilicity rather than maximising it, and series work shows that useful permeability is reachable by more than one structural strategy4. A series of 56 semipeptidic macrocycles varying in stereochemistry, N-methylation and lipophilicity showed adequate passive permeability even with properties outside Lipinski's rule of five4, which is the encouraging half of the picture; the discouraging half is that the successes are not predicted by two-dimensional descriptors.
Because the governing property is conformational behaviour rather than composition, two macrocycles with identical molecular formulae and identical calculated properties can differ enormously in permeability. That has three practical consequences.
For the synthetic routes used to build and close these scaffolds, see cyclic and constrained peptides; for what macrocyclic binders are increasingly being used for, see induced proximity and peptide degraders.
| Barrier | Desolvating backbone N–H donors, not molecular weight as such |
| Mechanism | Conformational switching between open in water and closed in membrane |
| Main levers | N-methylation, internal H-bonds, ester or heterocycle backbone, side chain shielding |
| Trap | Over-stabilising the closed form slows permeation |
| Not this route | Cationic cell-penetrating sequences, which work by a different mechanism entirely |
LifeTein synthesises macrocyclic and N-methylated peptides, including head-to-tail and side-chain cyclisations, N-methyl amino acid incorporation at specified positions, and D-amino acid and stereochemical variants. If you are scanning N-methylation positions across a scaffold, tell us the parent sequence and the positions to vary and we will quote the set.