Search

A third of the human genome encodes N-glycosylated proteins. These are co-translationally translocated into the lumen/membrane of the endoplasmic reticulum (ER) where they fold and assemble before they are transported to their final destination. Here, we show that calnexin, a major ER chaperone involved in glycoprotein folding is palmitoylated and that this modification is mediated by the ER palmitoyltransferase DHHC6. This modification leads to the preferential localization of calnexin to the perinuclear rough ER, at the expense of ER tubules. Moreover, palmitoylation mediates the association of calnexin with the ribosome-translocon complex (RTC) leading to the formation of a supercomplex that recruits the actin cytoskeleton, leading to further stabilization of the assembly. When formation of the calnexin-RTC supercomplex was affected by DHHC6 silencing, mutation of calnexin palmitoylation sites or actin depolymerization, folding of glycoproteins was impaired. Our findings thus show that calnexin is a stable component of the RTC in a manner that is exquisitely dependent on its palmitoylation status. This association is essential for the chaperone to capture its client proteins as they emerge from the translocon, acquire their N-linked glycans and initiate folding. The EMBO Journal (2012) 31, 1823-1835. doi: 10.1038/emboj.2012.15; Published online 7 February 2012

Over the past years, molecular modeling and simulation techniques have had a major impact on experimental life sciences. They are capable of providing accurate insight into microscopic mechanisms, which are usually difficult to investigate experimentally. Moreover, the integration of experimental data with molecular modeling appears to be a promising strategy to better understand complex biological processes. In this thesis, molecular dynamics simulation has been used in combination with experimental data to investigate two transmembrane proteins: (i) the bacterial chemoreceptor PhoQ and (ii) the Amyloid Precursor Protein (APP). (i) Bacterial two-component system PhoQ and bacterial membranes. Two-component systems (TCSs) are signaling complexes essential for bacterial survival and virulence. PhoQ is the histidine kinase chemoreceptor of the PhoQ-PhoP tandem machine that detects the concentration of cationic species at the inner membrane of Gram-negative bacteria. A full understanding of the PhoQ signal transduction mechanism is currently hindered by the lack of a complete atomistic structure. In this thesis project, the first structural model of the transmembrane (TM) portion of PhoQ from Escherichia coli was assembled, by using molecular simulations integrated with cross-linking disulfide scanning data. Its structural and dynamic features induce a concerted displacement of the TM helices at the periplasmic side, which modulates a rotation at the cytoplasmic end. This supports the idea that signal transduction is promoted through a combination of scissoring and rotational movements of the TM helices. Knowledge of this complex mechanism is essential in order to understand how the chemical stimuli sensed by the periplasmic sensor domain trigger, via the relay of the HAMP domain, the histidine auto-phosphorylation and kinase/phosphatase activity at the cytoplasmic end. The PhoQ sensor domain lies in close proximity to the membrane. Interaction with anionic lipids, such as phosphophatidylglycerols (PG) and cardiolipins (CL), are thought to play a key role in PhoQ activity. Present in bacterial and mitochondrial membranes, cardiolipins have a unique dimeric structure, which carries up to two charges, i.e. one per phosphate group, and under physiological conditions, can be unprotonated or singly protonated. Exhaustive models and characterization of cardiolipins are to-date scarce; therefore an ab initio parameterization of cardiolipin species for molecular simulation consistent with commonly used force fields is proposed here. Molecular dynamics (MD) simulations based on these models indicate a protonation-dependent lipid packing. A noteworthy interaction with solvating mono- and divalent cations is also observed. The proposed models will contribute to the biophysical and biochemical characterizations of bacterial and mitochondrial membranes and membrane-embedded proteins. (ii) Structural and dynamic properties of the Amyloid Precursor Protein. The Amyloid Precursor Protein (APP) is a type I membrane glycoprotein present at the neuronal synapsis. The proteolytic cleavage of its C-terminal segment produces amyloid-β (Aβ) peptides of different lengths, the deposition of which is an early indicator of Alzheimer"s disease (AD). Recently, the backbone structure of the APP transmembrane (TM) domain in detergent micelles was determined by nuclear magnetic resonance (NMR, independently by two different experimental groups). The TM conformations of these two structures are however markedly different. One is characterized by a highly kinked α-helix, whereas the other is mainly straight. Molecular dynamics simulations showed that the APP TM region is highly flexible and its secondary structure is influenced by the surrounding lipid environment. The size of the embedding detergent micelles strongly affects the conformation of the APP α-helix, with solvation being the main driving force for the development of a helical curvature. Once embedded in a membrane bilayer, APP systematically prefers a straight helical conformation. This is also confirmed when analyzing in silico the atomistic APP population observed in double electron-electron resonance (DEER) spectroscopy. In summary, the APP transmembrane domain is highly flexible due to the presence of glycine residues and can readily respond to the lipid environment, a property that might be critical for proteolytic processing by γ-secretase enzymes. The presented thesis work clearly shows how molecular simulations and their interplay with available experimental input can help advance the understanding of the mechanism of complex biological systems and processes on a molecular scale. These results, in particular, go well beyond the current understanding of the functioning of two transmembrane proteins relevant for human health. Furthermore, the computational approaches and procedures developed in these projects will hopefully promote novel integrated strategies for investigating biological systems.

Background: Amyloid- neurotoxicity depends on the specificity of the proteolytic cleavage of the amyloid precursor protein (APP) transmembrane domain. Results: The APP transmembrane -helix is straight in a biological membrane bilayer. Conclusion: The flexibility of APP is key for adapting to the lipid environment and modulating proteolytic processing by -secretase. Significance: The dynamic characterization of APP is expected to rationalize the design of -secretase modulators. The amyloid precursor protein (APP) is a widely expressed type I transmembrane (TM) glycoprotein present at the neuronal synapse. The proteolytic cleavage by -secretase of its C-terminal fragment produces amyloid- (A) peptides of different lengths, the deposition of which is an early indicator of Alzheimer disease. At present, there is no consensus on the conformation of the APP-TM domain at the biological membrane. Although structures have been determined by NMR in detergent micelles, their conformation is markedly different. Here we show by using molecular simulations that the APP-TM region systematically prefers a straight -helical conformation once embedded in a membrane bilayer. However, APP-TM is highly flexible, and its secondary structure is strongly influenced by the surrounding lipid environment, as when enclosed in detergent micelles. This behavior is confirmed when analyzing in silico the atomistic APP-TM population observed by residual dipolar couplings and double electron-electron resonance spectroscopy. These structural and dynamic features are critical in the proteolytic processing of APP by the -secretase enzyme, as suggested by a series of Gly(700) mutants. Affecting the hydration and flexibility of APP-TM, these mutants invariantly show an increase in the production of A38 compared with A40 peptides, which is reminiscent of the effect of -secretase modulators inhibitors.

The worldwide increase and proliferation of drug resistant microbes, coupled with the lag in new drug development, represents a major threat to human health. In order to reduce the time and cost for exploring the chemical search space, drug discovery increasingly relies on computational biology approaches. One key step in these approaches is the need for the rapid and accurate prediction of the binding affinity for potential leads. Here, we present RosENet (Rosetta Energy Neural Networks), an ensemble of three-dimensional (3D) Convolutional Neural Networks (CNNs), which combines voxelized molecular mechanics energies and molecular descriptors for predicting the absolute binding affinity of protein–ligand complexes. By leveraging the physicochemical properties captured by the molecular force field, our ensemble model achieved a Root Mean Square Error (RMSE) of 1.24 on the PDBBind v2016 core set. We also explored some limitations and the robustness of the PDBBind data set and our approach on nearly 500 structures, including structures determined by Nuclear Magnetic Resonance and virtual screening experiments. Our study demonstrated that molecular mechanics energies can be voxelized and used to help improve the predictive power of the CNNs. In the future, our framework can be extended to features extracted from other biophysical and biochemical models, such as molecular dynamics simulations., Journal of Chemical Information and Modeling, 60 (6), ISSN:1549-9596, ISSN:0095-2338, ISSN:1520-5142

Background Carbohydrates are a class of large and diverse biomolecules, ranging from a simple monosaccharide to large multi-branching glycan structures. The covalent linkage of a carbohydrate to the nitrogen atom of an asparagine, a process referred to as N-linked glycosylation, plays an important role in the physiology of many living organisms. Most software for glycan modeling on a personal desktop computer requires knowledge of molecular dynamics to interface with specialized programs such as CHARMM or AMBER. There are a number of popular web-based tools that are available for modeling glycans (e.g., GLYCAM-WEB (http://https://dev.glycam.org/gp/) or Glycosciences.db (http://www.glycosciences.de/)). However, these web-based tools are generally limited to a few canonical glycan conformations and do not allow the user to incorporate glycan modeling into their protein structure modeling workflow. Results Here, we present Glycosylator, a Python framework for the identification, modeling and modification of glycans in protein structure that can be used directly in a Python script through its application programming interface (API) or through its graphical user interface (GUI). The GUI provides a straightforward two-dimensional (2D) rendering of a glycoprotein that allows for a quick visual inspection of the glycosylation state of all the sequons on a protein structure. Modeled glycans can be further refined by a genetic algorithm for removing clashes and sampling alternative conformations. Glycosylator can also identify specific three-dimensional (3D) glycans on a protein structure using a library of predefined templates. Conclusions Glycosylator was used to generate models of glycosylated protein without steric clashes. Since the molecular topology is based on the CHARMM force field, new complex sugar moieties can be generated without modifying the internals of the code. Glycosylator provides more functionality for analyzing and modeling glycans than any other available software or webserver at present. Glycosylator will be a valuable tool for the glycoinformatics and biomolecular modeling communities., BMC Bioinformatics, 20 (1), ISSN:1471-2105