Toward a deeper understanding of allostery and allotargeting by computational
approaches
Understanding allosteric mechanisms of action and their modulation by ligand binding (allo-
targeting) gained importance in recent years, as allosteric modulators allow for selective
interference with specific protein-protein interactions (PPI) or cellular pathways. Yet, despite the
growth of data and methodologies, we still lack a solid understanding of allosteric mechanisms
that underlie biological function. We propose that a completely new framework, with focus on the
change in structural dynamics rather than changes in the states only, is needed. Furthermore,
rather than limiting our attention to transitions between two end-states (e.g. open/closed forms of
a protein), one needs to consider the complete ensemble of conformers, and evaluate the effect
of intermolecular interactions or mutations vis-à-vis the changes elicited in the conformational
landscape. Toward this goal, we propose to develop, implement, and apply innovative
computational models and methods that will focus on the essential dynamics of biomolecular
systems. Essential dynamics refers to the global modes of motions intrinsically accessible to the
overall structure, i.e. they cooperatively engage most, if not all, structural elements of the biological
assembly. We propose to: (1) develop, test, and validate an essential site scanning analysis
(ESSA) methodology for predicting ‘essential’ sites that dominate the essential dynamics, and
discriminating allosteric sites among them (Aim 1), (2) enhance the capability and accuracy of our
pathogenicity predictor, RHAPSODY, for evaluating the impact of mutations (single amino acid
variants) on biological function, by including in our machine learning algorithm the features derived
from global motions of biomolecular systems, the signature dynamics of protein families, and the
experimentally resolved PPIs (Aim 2), and (3) develop a hybrid methodology for efficient
assessment of conformational landscapes applicable to proteins containing cryptic sites and cryo-
EM structures (Aim 3), and finally extend and integrate these new methodologies to enable their
efficient translation to biomedical and pharmacological applications. Method development, testing,
validation, and further extensions will entail rigorous benchmarking against other methods and/or
relevant databases where applicable, in addition to detailed case studies in collaboration with
other labs (see support letters from six experimental and one computational collaborator).
Integration of the methodologies within our well-established application programming interface
ProDy will enable efficient dissemination and wide usage of the new technologies by the broader
community.