Extending a stapled MDM2 peptide to 4.1 hours in whole blood

The p53 transactivation domain binds MDM2 through three hydrophobic residues on one face of a short alpha helix. Hydrocarbon stapling holds that helix in its bound conformation, which raises affinity and shields the backbone from proteolysis, and the same hydrophobic face that drives binding also drives membrane interaction.

The partner's series had been optimised against MDM2 affinity and human plasma stability. Both endpoints looked finished: single-digit nanomolar Ki, and plasma half-lives long enough that stability had stopped being discussed in programme reviews.

Cellular activity did not track affinity, and in vivo exposure did not track plasma stability. The series had two unexplained rank inversions that had persisted through four design cycles.

Plasma is a poor proxy for a peptide's proteolytic environment. It is depleted of the cell-surface and cell-associated peptidases that act on circulating peptides, so a sequence cleaved rapidly at a membrane-proximal site can present a long plasma half-life and a short one everywhere that matters.

Opsin re-measured the series across four matrices: human plasma, serum, whole blood, and liver S9. Parent disappearance and metabolite formation were tracked in each, so every half-life is reported with the matrix it came from and the cleavage sites it implies.

Whole blood separated the series where plasma had not. Sequences clustered by the position of their primary cleavage site, and the cluster with the shortest whole blood half-life shared a single exposed dibasic motif that plasma had left largely intact.

Haemolysis was measured in the same campaign, since the amphipathic face responsible for membrane disruption is the face the staple stabilises. Eleven sequences were made to move the cleavage site and reduce the exposed hydrophobic surface without breaking the helix.

Whole blood half-life varied more between donors than the assay tolerance allowed. Across three donors the value for OPS-PEP-22 spanned 3.4 to 5.1 hours, so the ranking is reliable at the level of clusters and not between sequences that sit within an hour of each other.

Passive permeability did not move. The sequence remains dependent on an active uptake route that has not been identified, which leaves tissue distribution as an open question that the stability work did not answer.

Two rank inversions that had been treated as assay noise were the signal. Once the whole blood half-life was in the ranking, cellular activity and exposure both tracked it, and the sequences that had failed unexpectedly in vivo were the sequences with the exposed dibasic motif.

The haemolysis result matters for the same reason. Removing exposed hydrophobic surface improved membrane tolerance and cost nothing at the target interface, because the residues that contact MDM2 and the residues that lyse erythrocytes were not the same residues.

Whole blood stability used fresh heparinised human blood from three donors, incubated at 37 degrees Celsius with shaking, quenched into acetonitrile at seven time points to 8 hours. Metabolites were assigned by high-resolution mass spectrometry against the predicted cleavage set.

Haemolysis was measured on washed human erythrocytes at 2 percent haematocrit over 1 hour, with Triton X-100 as the full-lysis control. Sequences reaching only partial lysis at the top concentration are reported as greater-than values, with no extrapolation.