Machine learning approaches to evaluate correlation patterns in allosteric signaling: A case study of the PDZ2 domain.

Many proteins are regulated by dynamic allostery wherein regulator-induced changes in structure are comparable with thermal fluctuations. Consequently, understanding their mechanisms requires assessment of relationships between and within conformational ensembles of different states. Here we show how machine learning based approaches can be used to simplify this high-dimensional data mining task and also obtain mechanistic insight. In particular, we use these approaches to investigate two fundamental questions in dynamic allostery. First, how do regulators modify inter-site correlations in conformational fluctuations (Cij)? Second, how are regulator-induced shifts in conformational ensembles at two different sites in a protein related to each other? We address these questions in the context of the human protein tyrosine phosphatase 1E's PDZ2 domain, which is a model protein for studying dynamic allostery. We use molecular dynamics to generate conformational ensembles of the PDZ2 domain in both the regulator-bound and regulator-free states. The employed protocol reproduces methyl deuterium order parameters from NMR. Results from unsupervised clustering of Cij combined with flow analyses of weighted graphs of Cij show that regulator binding significantly alters the global signaling network in the protein; however, not by altering the spatial arrangement of strongly interacting amino acid clusters but by modifying the connectivity between clusters. Additionally, we find that regulator-induced shifts in conformational ensembles, which we evaluate by repartitioning ensembles using supervised learning, are, in fact, correlated. This correlation Δij is less extensive compared to Cij, but in contrast to Cij, Δij depends inversely on the distance from the regulator binding site. Assuming that Δij is an indicator of the transduction of the regulatory signal leads to the conclusion that the regulatory signal weakens with distance from the regulatory site. Overall, this work provides new approaches to analyze high-dimensional molecular simulation data and also presents applications that yield new insight into dynamic allostery.

Stimulation of Nipah Fusion: Small Intradomain Changes Trigger Extensive Interdomain Rearrangements.

Nipah is an emerging paramyxovirus that is of serious concern to human health. It invades host cells using two of its membrane proteins-G and F. G binds to host ephrins and this stimulates G to activate F. Upon activation, F mediates virus-host membrane fusion. Here we focus on mechanisms that underlie the stimulation of G by ephrins. Experiments show that G interacts with ephrin and F through separate sites located on two different domains, the receptor binding domain (RBD) and the F activation domain (FAD). No models explain this allosteric coupling. In fact, the analogous mechanisms in other paramyxoviruses also remain undetermined. The structural organization of G is such that allosteric coupling must involve at least one of the two interfaces-the RBD-FAD interface and/or the RBD-RBD interface. Here we examine using molecular dynamics the effect of ephrin binding on the RBD-RBD interface. We find that despite inducing small changes in individual RBDs, ephrin reorients the RBD-RBD interface extensively, and in a manner that will enhance solvent exposure of the FAD. While this finding supports a proposed model of G stimulation, we also find from additional simulations that ephrin induces a similar RBD-RBD reorientation in a stimulation-deficient G mutant, V209 VG → AAA. Together, our simulations suggest that while inter-RBD reorientation may be important, it is not, by itself, a sufficient condition for G stimulation. Additionally, we find that the mutation affects the conformational ensemble of RBD globally, including the RBD-FAD interface, suggesting the latter's role in G stimulation. Because ephrin induces small changes in individual RBDs, a proper analysis of conformational ensembles required that they are compared directly-we employ a method we developed recently, which we now release at SimTK, and show that it also performs excellently for non-Gaussian distributions.

The disruption of prenylation leads to pleiotropic rearrangements in cellular behavior in Staphylococcus aureus.

Prenylation is the addition of prenyl groups to peptide chains or metabolites via the condensation of geranyl- or isopentenyl-diphosphate moieties by geranyltranstransferases. Although this process is extensively studied in eukaryotes, little is known about the influence of prenylation in prokaryotic species. To explore the role of this modification in bacteria, we generated a mutation in the geranyltranstransferase (IspA) of Staphylococcus aureus. Quite strikingly, the ispA mutant completely lacked pigment and exhibited a previously undescribed small colony variant-like phenotype. Further pleiotropic defects in cellular behavior were noted, including impaired growth, decreased ATP production, increased sensitivity to oxidative stress, increased resistance to aminoglycosides and cationic antimicrobial peptides, and decreased resistance to cell wall-targeting antibiotics. These latter effects appear to result from differences in envelope composition as ispA mutants have highly diffuse cell walls (particularly at the septum), marked alterations in fatty acid composition and increased membrane fluidity. Taken together, these data present an important characterization of prokaryotic prenylation and demonstrate that this process is central to a wealth of pathways involved in mediating cellular homeostasis in S. aureus.

Effect of intrinsic and extrinsic factors on the simulated D-band length of type I collagen.

A signature feature of collagen is its axial periodicity visible in TEM as alternating dark and light bands. In mature, type I collagen, this repeating unit, D, is 67 nm long. This periodicity reflects an underlying packing of constituent triple-helix polypeptide monomers wherein the dark bands represent gaps between axially adjacent monomers. This organization is visible distinctly in the microfibrillar model of collagen obtained from fiber diffraction. However, to date, no atomistic simulations of this diffraction model under zero-stress conditions have reported a preservation of this structural feature. Such a demonstration is important as it provides the baseline to infer response functions of physiological stimuli. In contrast, simulations predict a considerable shrinkage of the D-band (11-19%). Here we evaluate systemically the effect of several factors on D-band shrinkage. Using force fields employed in previous studies we find that irrespective of the temperature/pressure coupling algorithms, assumed salt concentration or hydration level, and whether or not the monomers are cross-linked, the D-band shrinks considerably. This shrinkage is associated with the bending and widening of individual monomers, but employing a force field whose backbone dihedral energy landscape matches more closely with our computed CCSD(T) values produces a small D-band shrinkage of < 3%. Since this force field also performs better against other experimental data, it appears that the large shrinkage observed in earlier simulations is a force-field artifact. The residual shrinkage could be due to the absence of certain atomic-level details, such as glycosylation sites, for which we do not yet have suitable data.

Discerning intersecting fusion-activation pathways in the Nipah virus using machine learning.

The fusion of Nipah with host cells is facilitated by two of their glycoproteins, the G and the F proteins. The binding of cellular ephrins to the G head domain causes the G stalk domain to interact differently with F, which activates F to mediate virus-host fusion. To gain insight into how the ephrin-binding signal transduces from the head to the stalk domain of G, we examine quantitatively the differences between the conformational ensembles of the G head domain in its ephrin-bound and unbound states. We consider the human ephrins B2 and B3, and a double mutant of B2, all of which trigger fusion. The ensembles are generated using molecular dynamics, and the differences between them are quantified using a new machine learning method. We find that the portion of the G head domain whose conformational density is altered equivalently by the three ephrins is large, and comprises ∼25% of the residues in the G head domain. This subspace also includes the residues that are known to be important to F activation, which suggests that it contains at least one common signaling pathway. The spatial distribution of the residues constituting this subspace supports the model of signal transduction in which the signal transduces via the G head dimer interface. This study also adds to the growing list of examples where signaling does not depend solely on backbone deviations. In general, this study provides an approach to filter out conserved patterns in protein dynamics.

Water Dynamics at Protein-Protein Interfaces: Molecular Dynamics Study of Virus-Host Receptor Complexes.

The dynamical properties of water at protein-water interfaces are unlike those in the bulk. Here we utilize molecular dynamics simulations to study water dynamics in interstitial regions between two proteins. We consider two natural protein-protein complexes, one in which the Nipah virus G protein binds to cellular ephrin B2 and the other in which the same G protein binds to ephrin B3. While the two complexes are structurally similar, the two ephrins share only a modest sequence identity of ∼50%. X-ray crystallography also suggests that these interfaces are fairly extensive and contain exceptionally large amounts of waters. We find that while the interstitial waters tend to occupy crystallographic sites, almost all waters exhibit residence times of less than hundred picoseconds in the interstitial region. We also find that while the differences in the sequence of the two ephrins result in quantitative differences in the dynamics of interstitial waters, the trends in the shifts with respect to bulk values are similar. Despite the high wetness of the protein-protein interfaces, the dynamics of interstitial waters are considerably slower compared to the bulk-the interstitial waters diffuse an order of magnitude slower and have 2-3 fold longer hydrogen bond lifetimes and 2-1000 fold slower dipole relaxation rates. To understand the role of interstitial waters, we examine how implicit solvent models compare against explicit solvent models in producing ephrin-induced shifts in the G conformational density. Ephrin-induced shifts in the G conformational density are critical to the allosteric activation of another viral protein that mediates fusion. We find that in comparison with the explicit solvent model, the implicit solvent model predicts a more compact G-B2 interface, presumably because of the absence of discrete waters at the G-B2 interface. Simultaneously, we find that the two models yield strikingly different induced changes in the G conformational density, even for those residues whose conformational densities in the apo state are unaffected by the treatment of the bulk solvent. Together, these results show that the explicit treatment of interstitial water molecules is necessary for a proper description of allosteric transitions.