Protein-protein interactions (PPIs) and protein surfaces are considered challenging targets for drug discovery. In this field, conventional medicinal chemistry (i.e. small molecules) has largely failed to provide effective hits. On the other hand, peptides are endowed with a higher degree of structural flexibility (which allows them to better adapt to irregular targets) and are able to display a variety of tailor-made topologies, emerging as an alternative to target proteins that were considered undruggable. In this thesis, we have explored the potential of designed peptides to modulate the function of two therapeutically relevant protein targets involved in cancer (epidermal growth factor, EGF) and cognitive disorders (prolyloligopeptidase, POP).
Regarding the discovery of new peptide ligands against EGF, a rational design strategy has been applied, based on the wealth of structural data reported on the EGF-EGFR system. Peptide docking tools have allowed de novo design of a family of tripeptide ligands, which have shown a reproducible, albeit weak, binding to EGF. In order to obtain more active candidates, the relevant interacting regions of EGFR have been identified and mimicked with a diversity of cycle-constrained peptides. These are larger and structurally richer scaffolds that have proved more efficient in targeting a small and featureless protein, such as EGF. The best peptide hit, a 28-mer cyclic miniprotein (cp28), has served as the starting point in a computer-guided optimization process that strived for more active and structurally constrained analogues. Several rounds of computational design and biophysical screening have resulted in the generation of a series of bicyclic peptides that mimic the mode of binding of cp28 to EGF, albeit with reduced size, increased hydrophilicity and a more restrained topology. The chemical synthesis of these complex bicyclic molecules was enabled by state-of-the-art native chemical ligation techniques.
In order to assess the binding of our peptides with EGF, an array of suitable biophysical techniques was explored, and the most suitable ones were implemented to our discovery process. In particular, NMR spectroscopy (combined with expression of recombinant 15N-EGF) has allowed the monitoring of ligand-induced changes on the protein NMR spectra, an experiment which provided invaluable information on the binding mode of our peptides. In parallel, a recently developed acoustic biosensor (SAW) was set up as a low-cost, label-free, and highly sensitive technique to quantify the interactions with EGF.
In addition, our best peptide candidates (cp28 and cp23G) were able to disrupt the EGF-EGFR interaction, an effect that has been tested in several cell-like and living cell assays. Indeed, these peptides were able to halt the proliferation of EGFR(+) human carcinoma cells, an effect that underlines their biological efficacy. Moreover, the bicycle-constrained analogues display an exceptional level of biological stability, with most of the peptide still intact after serum incubation at 37ºC for 24 hours.
In the last part of this thesis, a series of bioactive peptides has been designed with a fundamentally different mechanism of action. Peptides typically possess fast dissociation rates from the protein target, a feature that represents an obstacle when competing with endogenous ligands for the binding to cavities, such as catalytic sites in enzymes. As a proof of concept, a novel class of covalent-acting peptidomimetics were developed to supress the activity of POP, a protease involved in neurodegenerative disorders. In these bifunctional molecules, the peptide backbone and side chains formed a template that selectively binds to the POP active site, whereas a novel sulfonyl fluoride electrophile was optimally positioned to react with the catalytic Ser residue. These compounds showed a high potency in vitro, being able to inactivate POP at low nM concentrations, and their mechanism of action as irreversible inhibitors was confirmed by kinetic assays. Moreover, they displayed a remarkable selectivity against closely related proteases, and they were able to permeate through a lipid bilayer that mimics the composition of the blood-brain barrier.
In summary, our findings show how two completely different classes of peptides, bicycle-constrained miniproteins and covalent-acting peptidomimetics, with binding affinities several orders of magnitude apart, can be efficiently designed to target specific protein surfaces. With PPIs and challenging binding sites becoming the focus of current drug discovery projects, these type of ligands are ideally positioned to deliver new drugs for the treatment of disease.
© 2001-2024 Fundación Dialnet · Todos los derechos reservados