Chemistry Reference
In-Depth Information
Effectively protein-protein interface flatness makes difficult the design of small
molecules (blocking of interactions). Large number of scoring functions
developed over the years for small ligands may be inappropriate. Protocols
developed over the last decades for protein-ligand binding may not be effective
for the challenges of finding small deep crevice-like binding pockets (enzymes or
receptor ligands). Significant conformational change can be observed for proteins
and their surfaces on binding to a partner. Protein-protein contact surfaces can be
large and dominated by hydrophobic, steric, electrostatic and hydrogen bond
interactions. The interfaces are flat, broad and large, often containing residues.
One advantage is a strong contribution from hotspots.
Validated protein-protein inhibitors, machine learning and chemoinformatic
methods have been applied yielding promising patterns and trends. These include
the information that at the interfaces, many Trp, Phe Tyr and Met residues are
present [782]. Lipinski's rule is not typically observed, since the inhibitors
considered (with nanomolar Kds) have molecular weights of at least 650 KDa.
Typically, we expect broad/flat/large protein-protein interfaces. Noncontinuous
epitopes are also compromised by contact residues. Despite the large interfaces,
most of the free energy of binding is contributed by small hot spots. These sites
can be prioritized as drug design targets. In some cases, fragment based ligand
design has been successful for designing these ligands.
Protein-protein interfaces show adaptability, can change from the unbound to the
bound state and are flexible. Created binding cavities not seen in the free protein
can disrupt the protein-protein interactions. Motions of small loop perturbations
and side chains often define the conformational changes. At interfaces,
crevices/pockets can be accessed by small molecules. The crevices may not be
accessible by the protein partner. For static structures of protein-protein
complexes it is difficult to predict their disruption by small molecules.
For docking of inhibitors, NMR and 10 ns molecular dynamics simulations can be
used to predict dynamics and flexibility in protein-protein interactions. Other
methods include MM-PBSA, component analysis of decomposition of solvation
energy, regression based scoring functions, normal mode analysis, coarse grain
