Molecular Dynamics and Machine Learning Give Insights on the Flexibility-Activity Relationships in Tyrosine Kinome.

Tyrosine kinases are a subfamily of kinases with critical roles in cellular machinery. Dysregulation of their active or inactive forms is associated with diseases like cancer. This study aimed to holistically understand their flexibility-activity relationships, focusing on pockets and fluctuations. We studied 43 different tyrosine kinases by collecting 120 µs of molecular dynamics simulations, pocket and residue fluctuation analysis, and a complementary machine learning approach. We found that the inactive forms often have increased flexibility, particularly at the DFG motif level. Noteworthy, thanks to these long simulations combined with a decision tree, we identified a semiquantitative fluctuation threshold of the DGF+3 residue over which the kinase has a higher probability to be in the inactive form.

Mechanisms of influence of the microtubule over-stabilizing ligands on the structure and intrinsic dynamics of α,ß-Tubulin.

The intervention into the cell cycle progression by administering microtubule over-stabilizing ligands that arrest the mitotic cell division by preventing spindle dissociation, is a promising strategy to fight against cancers. The building blocks of the microtubules and the spindles, i.e. the α,ß-tubulin dimer, upon binding of such ligands, stay more comfortably in the microtubular multimeric form; the phenomenon of which is the key to the said over-stabilization. Using two such over-stabilizing ligands, Taxol and Taxotere, the present work reports the collective changes that these ligands induce on the structure and dynamics of the α,ß-tubulin dimer which could be reconciled as the molecular basis of the over-stabilization of the microtubules; the trends have been found to be statistically significant across all independent calculations on them. The ligand binding increases the coherence between the residue communities of the two opposite faces of the ß-subunit, which in a periodic arrangement in microtubule are knwon to form intermolecular contact with each other. This is likely to create an indirect cooperativity between those structural regions and this is a consequence of the reshuffling of the internal network of interactions upon ligand binding. Such reorganizations are also complemented by the increased contributions of the softer modes of the intrinsic dynamics more, which is likely to increase the plasticity of the system favourable for making structural adjustments in a multimer. Further, the ligands are able to compensate the drawback of lacking one phosphate group in protein-GDP interactions compared to the same for protein-GTP and this is in agreement with the hints form the earlier reports. The findings form a mechanistic basis of the enhanced capacity of the α,ß-tubulin dimer to get more favourably accommodated into the microtubule superstructure upon binding either of Taxol and Taxotere.

Conformational States of E7010 Is Complemented by Microclusters of Water Inside the α,ß-Tubulin Core.

The α,ß-tubulin is the building block of microtubules, which is associated with and dissociated from the microtubular architecture complying with the dynamic instability of the microtubules. This dynamic instability has a direct relation with the spindle formation by the microtubules and cell division kinetics. E7010 is one of the promising ligands of an α,ß-tubulin protein that binds at the core of this protein and can diminish the protein's ability to fit to a growing microtubule, thus frustrating cell division. Although X-ray crystallography has reported a specific binding conformation of E7010 in PDB, molecular dynamics (MD) simulations have revealed two other conformational states of the ligand capable of binding to tubulin with stabilities close to that state reported in PDB. To rationalize this quasidegeneracy of ligand binding modes, MD simulations have further revealed that the understanding of the mechanism of E7010-tubulin binding remains incomplete unless the role of water molecules to bridge this interaction is taken into consideration, a very critical insight that was not visible from the PDB structure. Further, these water molecules differ from the standard examples of "bridging" waters which generally exist as isolated water molecules between the receptor and the ligand. In the present case, the water molecules sandwiched between ligand and protein, sequestered from the bulk solvent, integrate with each other by an H-bonds network forming a group, which appear as microclusters of water. The structural packing with the ligand binding pocket and the bridging interactions between protein and ligand take place through such clusters. The presence of this microcluster of water is not just cosmetic, instead they have a crucial impact on the ligand binding thermodynamics. Only with the explicit consideration of these water clusters in the binding energy calculations (MMGBSA) is the stability of the native mode of ligand binding reported in PDB rationalized. At the same time, two other binding modes are elucidated to be quasi-degenerate with the native state and that indicates the further possibility in gaining more entropic stabilization of the complex. The role of such "bridging" water clusters to enhance the protein-ligand interaction will be insightful for designing the next generation prospective compounds in the field of cancer therapeutics.

Flexibility enables to discriminate between ligands: Lessons from structural ensembles of Bcl-xl and Mcl-1.

The proteins of Bcl-2 family, which are promising anti-cancer-drug targets, have substantial similarity in primary sequence and share homologous domains as well as similar structural folds. In spite of similarities in sequence and structures, the members of its pro- and anti- apoptotic subgroups form complexes with different type of partners with discriminating binding affinities. Understanding the origin of this discrimination is very important for designing ligands that can either selectively target a protein or could be made broad ranged as necessary. Using principal component analysis (PCA) of the available structures and from the analysis of the evolution of the binding pocket residues, the correlation has been investigated considering two important anti-apoptotic protein Bcl-xl and Mcl-1, which serve as two ideal representatives of this family. The flexibility of the receptor enables them to discriminate between the ligands or the binding partners. It has been observed that although Bcl-xl and Mcl-1 are classified as homologous proteins, through the course of evolution the binding pocket residues are highly conserved for Bcl-xl; whereas they have been substituted frequently in Mcl-1. The investigation has revealed that the Bcl-xl can adjust the backbone conformation of the binding pocket residues to a larger extent to complement with the shape of different binding partners whereas the Mcl-1 shows more variation in the side chain conformation of binding pocket residues for the same purpose.

Ligand Binding Swaps between Soft Internal Modes of α,ß-Tubulin and Alters Its Accessible Conformational Space.

The dynamic instability of the microtubule originates from the conformational switching of its building block, that is, the α, ß-tubulin dimer. Ligands occupying the interface of the α-ß dimer bias the switch toward the disintegration of the microtubule, which in turn controls the cell division. A little loop of tubulin is structurally encoded as a biophysical "gear" that works by changing its structural packing. The consequence of such change propagates to the quaternary level to alter the global dynamics and is reflected as a swapping between the relative contributions of dominating internal modes. Simulation shows that there is an appreciable separation between the conformational space accessed by the liganded and unliganded systems; the clusters of conformations differ in their intrinsic tendencies to "bend" and "twist". The correlation between the altered breathing modes and conformational space rationally hypothesizes a mechanism of straight-bent interconversion of the system. In this mechanism, a ligand is understood to bias the state of the "gear" that detours the conformational equilibrium away from its native preference. Thus, a fundamental biophysical insight into the mechanism of the conformational switching of tubulin is presented as a multiscale process that also shows promise to yield newer concept of ligand design.

Dynamic and Static Water Molecules Complement the TN16 Conformational Heterogeneity inside the Tubulin Cavity.

TN16 is one of the most promising inhibitors of α, ß dimer of tubulin that occupies the cavity in the ß-subunit located at the dimeric interface, known as the colchicine binding site. The experimentally determined structure of the complex (Protein Data Bank entry 3HKD) presents the conformation and position of the ligand based on the "best fit", keeping the controversy of other significant binding modes open for further investigation. Computation has already revealed that TN16 experiences fluctuations within the binding pocket, but the insight from that previous report was limited by the shorter windows of sampling and by the approximations on the surrounding environment by implicit solvation. This article reports that in most of the cases straightforward MMGBSA calculations of binding energy revealed a gradual loss of stabilization that was inconsistent with the structural observations, and thus, it indicated the lack of consideration of stabilizing factors with appropriate weightage. Consideration of the structurally packed water molecules in the space between the ligand and receptor successfully eliminated such discrepancies between the structure and stability, serving as the "litmus test" of the importance of explicit consideration of such structurally packed water in the calculations. Such consideration has further evidenced a quasi-degenerate character of the different binding modes of TN16 that has rationalized the observed intrinsic fluctuations of TN16 within the pocket, which is likely to be the most critical insight into its entropy-dominated binding. Quantum mechanical calculations have revealed a relay of electron density from TN16 to the protein via a water molecule in a concerted manner.

Interactions between Bcl-xl and its inhibitors: Insights into ligand design from molecular dynamics simulation.

The Bcl-xl protein is a potential drug target for cancer, and it has a relatively flat and flexible binding pocket. ABT263 is one of the most promising molecules that inhibit Bcl-xl, and it was developed from its precursor ABT737 with suitable substitutions. However, the structural and mechanistic implications of those changes have not yet been reported. Molecular dynamics simulation has revealed that the conformational microstates of the complex of Bcl-xl and ABT263 shows heterogeneity at the binding interface with Bcl-xl in contrast to the precise interactions witnessed in case of ABT737. This occurs because not all the functional groups of ABT263 are able to anchor into the binding pocket simultaneously at the time of complexation; leaving at least one group weakly associated every time. The insight into the mechanism shows that, in spite of such mutual exclusivity, the resultant effect becomes beneficial, i.e., becomes more effective than ABT737. Going against the traditional belief, the calculations also confirm that there is no benefit of reshaping the highly flexible binding pocket to allow the ligand to be comfortably accommodated and avoid conflicting orientations of the functional groups, as the destabilization becomes active from other sources. These structural clues and in-silico tests suggest possible avenues for improving the binding affinity of ABT263 through further in-vitro and in-vivo tests.

Adaptability in protein structures: structural dynamics and implications in ligand design.

The basic framework of understanding the mechanisms of protein functions is achieved from the knowledge of their structures which can model the molecular recognition. Recent advancement in the structural biology has revealed that in spite of the availability of the structural data, it is nontrivial to predict the mechanism of the molecular recognition which progresses via situation-dependent structural adaptation. The mutual selectivity of protein-protein and protein-ligand interactions often depends on the modulations of conformations empowered by their inherent flexibility, which in turn regulates the function. The mechanism of a protein's function, which used to be explained by the ideas of 'lock and key' has evolved today as the concept of 'induced fit' as well as the 'population shift' models. It is felt that the 'dynamics' is an essential feature to take into account for understanding the mechanism of protein's function. The design principles of therapeutic molecules suffer from the problems of plasticity of the receptors whose binding conformations are accurately not predictable from the prior knowledge of a template structure. On the other hand, flexibility of the receptors provides the opportunity to improve the binding affinity of a ligand by suitable substitution that will maximize the binding by modulating the receptors surface. In this paper, we discuss with example how the protein's flexibility is correlated with its functions in various systems, revealing the importance of its understanding and for making applications. We also highlight the methodological challenges to investigate it computationally and to account for the flexible nature of the molecules in drug design.