A deep encoder-decoder framework for identifying distinct ligand binding pathways.

The pathway(s) that a ligand would adopt en route to its trajectory to the native pocket of the receptor protein act as a key determinant of its biological activity. While Molecular Dynamics (MD) simulations have emerged as the method of choice for modeling protein-ligand binding events, the high dimensional nature of the MD-derived trajectories often remains a barrier in the statistical elucidation of distinct ligand binding pathways due to the stochasticity inherent in the ligand's fluctuation in the solution and around the receptor. Here, we demonstrate that an autoencoder based deep neural network, trained using an objective input feature of a large matrix of residue-ligand distances, can efficiently produce an optimal low-dimensional latent space that stores necessary information on the ligand-binding event. In particular, for a system of L99A mutant of T4 lysozyme interacting with its native ligand, benzene, this deep encoder-decoder framework automatically identifies multiple distinct recognition pathways, without requiring user intervention. The intermediates involve the spatially discrete location of the ligand in different helices of the protein before its eventual recognition of native pose. The compressed subspace derived from the autoencoder provides a quantitatively accurate measure of the free energy and kinetics of ligand binding to the native pocket. The investigation also recommends that while a linear dimensional reduction technique, such as time-structured independent component analysis, can do a decent job of state-space decomposition in cases where the intermediates are long-lived, autoencoder is the method of choice in systems where transient, low-populated intermediates can lead to multiple ligand-binding pathways.

Solvent's Role in Cavity-Ligand Recognition Would Depend on the Mode of Ligand Diffusion.

Solvent is known to play crucial roles in dictating the thermodynamics and kinetics of the biomolecular recognition process. Here, we show that the extent of significance of water in modulating the ligand recognition process is critically contingent on the ligand diffusion and on the constraints introduced on it. Toward the end, we use a well-known prototypical system of spherical ligand diffusing freely toward a hydrophobic concave cavity in explicit water. We analyze a large series of adaptively sampled unbiased molecular dynamics simulation trajectories within the framework of time-structured independent component analysis (TICA). Our quantitative investigations reveal that water would play a significant role in the ligand recognition process, provided that the ligand is constricted to diffuse along a centro-symmetric fashion. On the contrary, water's contribution in the ligand recognition process would diminish to a negligible value if the ligand freely diffuses toward the pocket. A Markov state model (MSM) constructed using the simulated trajectories identifies a set of transiently populated metastable states comprising partially ligand-unbound macro states, alongside ligand-bound and ligand-unbound pose and gives rise to multiple transition paths of ligand in its way to the hydrophobic cavity. Lifting the restriction on ligand movement changes its binding pathway, time scales, and the extent of the role of solvent in modulating the recognition process.

A deep autoencoder framework for discovery of metastable ensembles in biomacromolecules.

Biomacromolecules manifest dynamic conformational fluctuation and involve mutual interconversion among metastable states. A robust mapping of their conformational landscape often requires the low-dimensional projection of the conformational ensemble along optimized collective variables (CVs). However, the traditional choice for the CV is often limited by user-intuition and prior knowledge about the system, and this lacks a rigorous assessment of their optimality over other candidate CVs. To address this issue, we propose an approach in which we first choose the possible combinations of inter-residue Cα-distances within a given macromolecule as a set of input CVs. Subsequently, we derive a non-linear combination of latent space embedded CVs via auto-encoding the unbiased molecular dynamics simulation trajectories within the framework of the feed-forward neural network. We demonstrate the ability of the derived latent space variables in elucidating the conformational landscape in four hierarchically complex systems. The latent space CVs identify key metastable states of a bead-in-a-spring polymer. The combination of the adopted dimensional reduction technique with a Markov state model, built on the derived latent space, reveals multiple spatially and kinetically well-resolved metastable conformations for GB1 ß-hairpin. A quantitative comparison based on the variational approach-based scoring of the auto-encoder-derived latent space CVs with the ones obtained via independent component analysis (principal component analysis or time-structured independent component analysis) confirms the optimality of the former. As a practical application, the auto-encoder-derived CVs were found to predict the reinforced folding of a Trp-cage mini-protein in aqueous osmolyte solution. Finally, the protocol was able to decipher the conformational heterogeneities involved in a complex metalloenzyme, namely, cytochrome P450.

Nonaffine Displacements Encode Collective Conformational Fluctuations in Proteins.

Identifying subtle conformational fluctuations underlying the dynamics of biomacromolecules is crucial for resolving their free energy landscape. We show that a collective variable, originally proposed for crystalline solids, is able to filter out essential macromolecular motions more efficiently than other approaches. While homogeneous or "affine" deformations of the biopolymer are trivial, biopolymer conformations are complicated by the occurrence of inhomogeneous or "nonaffine" displacements of atoms relative to their positions in the native structure. We show that these displacements encode functionally relevant conformations of macromolecules, and in combination with a formalism based upon time-structured independent component analysis, they quantitatively resolve the free energy landscape of a number of macromolecules of hierarchical complexity. The kinetics of conformational transitions among the basins can now be mapped within the framework of a Markov state model. The nonaffine modes, obtained by projecting out homogeneous fluctuations from the local displacements, are found to be responsible for local structural changes required for transitioning between pairs of macrostates.

On the role of solvent in hydrophobic cavity-ligand recognition kinetics.

A solvent often manifests itself as the key determinant of the kinetic aspect of the molecular recognition process. While the solvent is often depicted as a source of barrier in the ligand recognition process by the polar cavity, the nature of solvent's role in the recognition process involving hydrophobic cavity and hydrophobic ligand remains to be addressed. In this work, we quantitatively assess the role of solvent in dictating the kinetic process of recognition in a popular system involving the hydrophobic cavity and ligand. In this prototypical system, the hydrophobic cavity undergoes dewetting transition as the ligand approaches the cavity, which influences the cavity-ligand recognition kinetics. Here, we build a Markov state model (MSM) using adaptively sampled unrestrained molecular dynamics simulation trajectories to map the kinetic recognition process. The MSM-reconstructed free energy surface recovers a broad water distribution at an intermediate cavity-ligand separation, consistent with a previous report of dewetting transition in this system. Time-structured independent component analysis of the simulated trajectories quantitatively shows that cavity-solvent density contributes considerably in an optimized reaction coordinate involving cavity-ligand separation and water occupancy. Our approach quantifies two solvent-mediated macrostates at an intermediate separation of the cavity-ligand recognition pathways, apart from the fully ligand-bound and fully ligand-unbound macrostates. Interestingly, we find that these water-mediated intermediates, while transient in populations, can undergo slow mutual interconversion and create possibilities of multiple pathways of cavity recognition by the ligand. Overall, the work provides a quantitative assessment of the role that the solvent plays in facilitating the recognition process involving the hydrophobic cavity.

Spontaneous vesicle formation by γ-aminobutyric acid derived steroidal surfactant: Curcumin loading, cytotoxicity and cellular uptake studies.

Cholesterol (Chol) is a ubiquitous steroidal component of cell membrane and is known to modulate the packing of phospholipids within the bilayer. Thus, Chol has been frequently used in the formulation and study of artificial "model membranes" like vesicles and liposomes. In this work, we have developed a novel anionic surfactant by conjugating two biomolecules, cholesterol and γ-aminobutyric acid via a urethane linkage. We have studied its physicochemical behavior in aqueous buffer. The surfactant has been shown to spontaneously form small unilamellar vesicles above a very low critical concentration in aqueous neutral buffer at room temperature. The vesicle phase was characterized by use of fluorescence probe, transmission electron microscopy and dynamic light scattering (DLS) techniques. The vesicle bilayer was found to be much less polar as well as more viscous compared to the bulk water. The vesicle stability with respect to change of temperature, pH, and ageing time was investigated by fluorescence probe and DLS techniques. The loading efficiency of the vesicles for the hydrophobic drug, curcumin, was determined and its release under physiological condition was studied. The in vitro cellular uptake of curcumin-loaded vesicles to human breast cancer cell line (MDA-MB-231) also was investigated. The MTT assay showed that the surfactant was non-cytotoxic up to a relatively high concentration.