Rescue of secretion of rare-disease-associated misfolded mutant glycoproteins in UGGT1 knock-out mammalian cells.

Endoplasmic reticulum (ER) retention of misfolded glycoproteins is mediated by the ER-localized eukaryotic glycoprotein secretion checkpoint, UDP-glucose glycoprotein glucosyl-transferase (UGGT). The enzyme recognizes a misfolded glycoprotein and flags it for ER retention by re-glucosylating one of its N-linked glycans. In the background of a congenital mutation in a secreted glycoprotein gene, UGGT-mediated ER retention can cause rare disease, even if the mutant glycoprotein retains activity ("responsive mutant"). Using confocal laser scanning microscopy, we investigated here the subcellular localization of the human Trop-2-Q118E, E227K and L186P mutants, which cause gelatinous drop-like corneal dystrophy (GDLD). Compared with the wild-type Trop-2, which is correctly localized at the plasma membrane, these Trop-2 mutants are retained in the ER. We studied fluorescent chimeras of the Trop-2 Q118E, E227K and L186P mutants in mammalian cells harboring CRISPR/Cas9-mediated inhibition of the UGGT1 and/or UGGT2 genes. The membrane localization of the Trop-2 Q118E, E227K and L186P mutants was successfully rescued in UGGT1-/- cells. UGGT1 also efficiently reglucosylated Trop-2-Q118E-EYFP in cellula. The study supports the hypothesis that UGGT1 modulation would constitute a novel therapeutic strategy for the treatment of pathological conditions associated to misfolded membrane glycoproteins (whenever the mutation impairs but does not abrogate function), and it encourages the testing of modulators of ER glycoprotein folding quality control as broad-spectrum rescue-of-secretion drugs in rare diseases caused by responsive secreted glycoprotein mutants.

AutoDock Bias: improving binding mode prediction and virtual screening using known protein-ligand interactions.

SUMMARY: The performance of docking calculations can be improved by tuning parameters for the system of interest, e.g. biasing the results towards the formation of relevant protein-ligand interactions, such as known ligand pharmacophore or interaction sites derived from cosolvent molecular dynamics. AutoDock Bias is a straightforward and easy to use script-based method that allows the introduction of different types of user-defined biases for fine-tuning AutoDock4 docking calculations. AVAILABILITY AND IMPLEMENTATION: AutoDock Bias is distributed with MGLTools (since version 1.5.7), and freely available on the web at http://ccsb.scripps.edu/mgltools/ or http://autodockbias.wordpress.com. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

An efficient use of X-ray information, homology modeling, molecular dynamics and knowledge-based docking techniques to predict protein-monosaccharide complexes.

Unraveling the structure of lectin-carbohydrate complexes is vital for understanding key biological recognition processes and development of glycomimetic drugs. Molecular Docking application to predict them is challenging due to their low affinity, hydrophilic nature and ligand conformational diversity. In the last decade several strategies, such as the inclusion of glycan conformation specific scoring functions or our developed solvent-site biased method, have improved carbohydrate docking performance but significant challenges remain, in particular, those related to receptor conformational diversity. In the present work we have analyzed conventional and solvent-site biased autodock4 performance concerning receptor conformational diversity as derived from different crystal structures (apo and holo), Molecular Dynamics snapshots and Homology-based models, for 14 different lectin-monosaccharide complexes. Our results show that both conventional and biased docking yield accurate lectin-monosaccharide complexes, starting from either apo or homology-based structures, even when only moderate (45%) sequence identity templates are available. An essential element for success is a proper combination of a middle-sized (10-100 structures) conformational ensemble, derived either from Molecular dynamics or multiple homology model building. Consistent with our previous works, results show that solvent-site biased methods improve overall performance, but that results are still highly system dependent. Finally, our results also show that docking can select the correct receptor structure within the ensemble, underscoring the relevance of joint evaluation of both ligand pose and receptor conformation.

Cosolvent-Based Protein Pharmacophore for Ligand Enrichment in Virtual Screening.

Virtual screening of large compound databases, looking for potential ligands of a target protein, is a major tool in computer-aided drug discovery. Throughout the years, different techniques such as similarity searching, pharmacophore matching, or molecular docking have been applied with the aim of finding hit compounds showing appreciable affinity. Molecular dynamics simulations in mixed solvents have been shown to identify hot spots relevant for protein-drug interaction, and implementations based on this knowledge were developed to improve pharmacophore matching of small molecules, binding free-energy estimations, and docking performance in terms of pose prediction. Here, we proved in a retrospective manner that cosolvent-derived pharmacophores from molecular dynamics (solvent sites) improve the performance of docking-based virtual screening campaigns. We applied a biased docking scheme based on solvent sites to nine relevant target proteins that have a set of known ligands or actives and compounds that are, presumably, nonbinders (decoys). Our results show improvement in virtual screening performance compared to traditional docking programs both at a global level, with up to 35% increase in areas under the receiver operating characteristic curve, and in early stages, with up to a 7-fold increase in enrichment factors at 1%. However, the improvement in pose prediction of actives was less profound. The presented application makes use of the AutoDock Bias method and is the only cosolvent-derived pharmacophore technique that employs its knowledge both in the ligand conformational search algorithm and the final affinity scoring for virtual screening purposes.

Exploring lectin-like activity of the S-layer protein of Lactobacillus acidophilus ATCC 4356.

The surface layer (S-layer) protein of Lactobacillus acidophilus is a crystalline array of self-assembling, proteinaceous subunits non-covalently bound to the outmost bacterial cell wall envelope and is involved in the adherence of bacteria to host cells. We have previously described that the S-layer protein of L. acidophilus possesses anti-viral and anti-bacterial properties. In this work, we extracted and purified S-layer proteins from L. acidophilus ATCC 4356 cells to study their interaction with cell wall components from prokaryotic (i.e., peptidoglycan and lipoteichoic acids) and eukaryotic origin (i.e., mucin and chitin), as well as with viruses, bacteria, yeast, and blood cells. Using chimeric S-layer fused to green fluorescent protein (GFP) from different parts of the protein, we analyzed their binding capacity. Our results show that the C-terminal part of the S-layer protein presents lectin-like activity, interacting with different glycoepitopes. We further demonstrate that lipoteichoic acid (LTA) serves as an anchor for the S-layer protein. Finally, a structure for the C-terminal part of S-layer and possible binding sites were predicted by a homology-based model.

Solvents to Fragments to Drugs: MD Applications in Drug Design.

Simulations of molecular dynamics (MD) are playing an increasingly important role in structure-based drug discovery (SBDD). Here we review the use of MD for proteins in aqueous solvation, organic/aqueous mixed solvents (MDmix) and with small ligands, to the classic SBDD problems: Binding mode and binding free energy predictions. The simulation of proteins in their condensed state reveals solvent structures and preferential interaction sites (hot spots) on the protein surface. The information provided by water and its cosolvents can be used very effectively to understand protein ligand recognition and to improve the predictive capability of well-established methods such as molecular docking. The application of MD simulations to the study of the association of proteins with drug-like compounds is currently only possible for specific cases, as it remains computationally very expensive and labor intensive. MDmix simulations on the other hand, can be used systematically to address some of the common tasks in SBDD. With the advent of new tools and faster computers we expect to see an increase in the application of mixed solvent MD simulations to a plethora of protein targets to identify new drug candidates.

Molecular Dynamics in Mixed Solvents Reveals Protein-Ligand Interactions, Improves Docking, and Allows Accurate Binding Free Energy Predictions.

One of the most important biological processes at the molecular level is the formation of protein-ligand complexes. Therefore, determining their structure and underlying key interactions is of paramount relevance and has direct applications in drug development. Because of its low cost relative to its experimental sibling, molecular dynamics (MD) simulations in the presence of different solvent probes mimicking specific types of interactions have been increasingly used to analyze protein binding sites and reveal protein-ligand interaction hot spots. However, a systematic comparison of different probes and their real predictive power from a quantitative and thermodynamic point of view is still missing. In the present work, we have performed MD simulations of 18 different proteins in pure water as well as water mixtures of ethanol, acetamide, acetonitrile and methylammonium acetate, leading to a total of 5.4 µs simulation time. For each system, we determined the corresponding solvent sites, defined as space regions adjacent to the protein surface where the probability of finding a probe atom is higher than that in the bulk solvent. Finally, we compared the identified solvent sites with 121 different protein-ligand complexes and used them to perform molecular docking and ligand binding free energy estimates. Our results show that combining solely water and ethanol sites allows sampling over 70% of all possible protein-ligand interactions, especially those that coincide with ligand-based pharmacophoric points. Most important, we also show how the solvent sites can be used to significantly improve ligand docking in terms of both accuracy and precision, and that accurate predictions of ligand binding free energies, along with relative ranking of ligand affinity, can be performed.

WATCLUST: a tool for improving the design of drugs based on protein-water interactions.

MOTIVATION: Water molecules are key players for protein folding and function. On the protein surface, water is not placed randomly, but display instead a particular structure evidenced by the presence of specific water sites (WS). These WS can be derived and characterized using explicit water Molecular Dynamics simulations, providing useful information for ligand binding prediction and design. Here we present WATCLUST, a WS determination and analysis tool running on the VMD platform. The tool also allows direct transfer of the WS information to Autodock program to perform biased docking. AVAILABILITY AND IMPLEMENTATION: The WATCLUST plugin and documentation are freely available at http://sbg.qb.fcen.uba.ar/watclust/. CONTACT: marcelo@qi.fcen.uba.ar, adrian@qi.fcen.uba.ar.

Solvent structure improves docking prediction in lectin-carbohydrate complexes.

Recognition and complex formation between proteins and carbohydrates is a key issue in many important biological processes. Determination of the three-dimensional structure of such complexes is thus most relevant, but particularly challenging because of their usually low binding affinity. In silico docking methods have a long-standing tradition in predicting protein-ligand complexes, and allow a potentially fast exploration of a number of possible protein-carbohydrate complex structures. However, determining which of these predicted complexes represents the correct structure is not always straightforward. In this work, we present a modification of the scoring function provided by AutoDock4, a widely used docking software, on the basis of analysis of the solvent structure adjacent to the protein surface, as derived from molecular dynamics simulations, that allows the definition and characterization of regions with higher water occupancy than the bulk solvent, called water sites. They mimic the interaction held between the carbohydrate -OH groups and the protein. We used this information for an improved docking method in relation to its capacity to correctly predict the protein-carbohydrate complexes for a number of tested proteins, whose ligands range in size from mono- to tetrasaccharide. Our results show that the presented method significantly improves the docking predictions. The resulting solvent-structure-biased docking protocol, therefore, appears as a powerful tool for the design and optimization of development of glycomimetic drugs, while providing new insights into protein-carbohydrate interactions. Moreover, the achieved improvement also underscores the relevance of the solvent structure to the protein carbohydrate recognition process.

Rescue of secretion of a rare-disease associated mis-folded mutant glycoprotein in UGGT1 knock-out mammalian cells.

Endoplasmic reticulum (ER) retention of mis-folded glycoproteins is mediated by the ERlocalised eukaryotic glycoprotein secretion checkpoint, UDP-glucose glycoprotein glucosyl-transferase (UGGT). The enzyme recognises a mis-folded glycoprotein and flags it for ER retention by reglucosylating one of its N-linked glycans. In the background of a congenital mutation in a secreted glycoprotein gene, UGGT-mediated ER retention can cause rare disease even if the mutant glycoprotein retains activity ("responsive mutant"). Here, we investigated the subcellular localisation of the human Trop-2 Q118E variant, which causes gelatinous droplike corneal dystrophy (GDLD). Compared with the wild type Trop-2, which is correctly localised at the plasma membrane, the Trop-2-Q118E variant is found to be heavily retained in the ER. Using Trop-2-Q118E, we tested UGGT modulation as a rescue-of-secretion therapeutic strategy for congenital rare disease caused by responsive mutations in genes encoding secreted glycoproteins. We investigated secretion of a EYFP-fusion of Trop-2-Q118E by confocal laser scanning microscopy. As a limiting case of UGGT inhibition, mammalian cells harbouring CRISPR/Cas9-mediated inhibition of the UGGT1 and/or UGGT2 gene expressions were used. The membrane localisation of the Trop-2-Q118E-EYFP mutant was successfully rescued in UGGT1-/- and UGGT1/2-/- cells. UGGT1 also efficiently reglucosylated Trop-2-Q118E-EYFP in cellula. The study supports the hypothesis that UGGT1 modulation constitutes a novel therapeutic strategy for the treatment of Trop-2-Q118E associated GDLD, and it encourages the testing of modulators of ER glycoprotein folding Quality Control (ERQC) as broad-spectrum rescueof-secretion drugs in rare diseases caused by responsive secreted glycoprotein mutants.

A quinolin-8-ol sub-millimolar inhibitor of UGGT, the ER glycoprotein folding quality control checkpoint.

Misfolded glycoprotein recognition and endoplasmic reticulum (ER) retention are mediated by the ER glycoprotein folding quality control (ERQC) checkpoint enzyme, UDP-glucose glycoprotein glucosyltransferase (UGGT). UGGT modulation is a promising strategy for broad-spectrum antivirals, rescue-of-secretion therapy in rare disease caused by responsive mutations in glycoprotein genes, and many cancers, but to date no selective UGGT inhibitors are known. The small molecule 5-[(morpholin-4-yl)methyl]quinolin-8-ol (5M-8OH-Q) binds a CtUGGTGT24 "WY" conserved surface motif conserved across UGGTs but not present in other GT24 family glycosyltransferases. 5M-8OH-Q has a 47 µM binding affinity for CtUGGTGT24in vitro as measured by ligand-enhanced fluorescence. In cellula, 5M-8OH-Q inhibits both human UGGT isoforms at concentrations higher than 750 µM. 5M-8OH-Q binding to CtUGGTGT24 appears to be mutually exclusive to M5-9 glycan binding in an in vitro competition experiment. A medicinal program based on 5M-8OH-Q will yield the next generation of UGGT inhibitors.

MotSASi: Functional short linear motifs (SLiMs) prediction based on genomic single nucleotide variants and structural data.

Short linear motifs (SLiMs) are key to cell physiology mediating reversible protein-protein interactions. Precise identification of SLiMs remains a challenge, being the main drawback of most bioinformatic prediction tools, their low specificity (high number of false positives). An important, usually overlooked, aspect is the relation between SLiMs mutations and disease. The presence of variants in each residue position can be used to assess the relevance of the corresponding residue(s) for protein function, and its (in)tolerance to change. In the present work, we combined sequence variant information and structural analysis of the energetic impact of single amino acid substitution (SAS) in SLiM-Receptor complex structure, and showed that it improves prediction of true functional SLiMs. Our strategy is based on building a SAS tolerance matrix that shows, for each position, whether one of the possible 19 SAS is tolerated or not. Herein we present the MotSASi strategy and analyze in detail 3 SLiMs involved in intracellular protein trafficking (phospho-independent tyrosine-based motif (NPx[Y/F]), type 1 PDZ-binding motif ([S/T]x[V/I/L]COOH) and tryptophan-acidic motif ([L/M]xW[D/E])). Our results show that inclusion of variant and structure information improves both prediction of true SLiMs and rejection of false positives, while also allowing better classification of variants inside SLiMs, a result with a direct impact in clinical genomics.

Clamping, bending, and twisting inter-domain motions in the misfold-recognizing portion of UDP-glucose: Glycoprotein glucosyltransferase.

UDP-glucose:glycoprotein glucosyltransferase (UGGT) flags misfolded glycoproteins for ER retention. We report crystal structures of full-length Chaetomium thermophilum UGGT (CtUGGT), two CtUGGT double-cysteine mutants, and its TRXL2 domain truncation (CtUGGT-ΔTRXL2). CtUGGT molecular dynamics (MD) simulations capture extended conformations and reveal clamping, bending, and twisting inter-domain movements. We name "Parodi limit" the maximum distance on the same glycoprotein between a site of misfolding and an N-linked glycan that can be reglucosylated by monomeric UGGT in vitro, in response to recognition of misfold at that site. Based on the MD simulations, we estimate the Parodi limit as around 70-80 Å. Frequency distributions of distances between glycoprotein residues and their closest N-linked glycosylation sites in glycoprotein crystal structures suggests relevance of the Parodi limit to UGGT activity in vivo. Our data support a "one-size-fits-all adjustable spanner" UGGT substrate recognition model, with an essential role for the UGGT TRXL2 domain.

Origin and Evolution of Two Independently Duplicated Genes Encoding UDP- Glucose: Glycoprotein Glucosyltransferases in Caenorhabditis and Vertebrates.

UDP- glucose: glycoprotein glucosyltransferase (UGGT) is a protein that operates as the gatekeeper for the endoplasmic reticulum (ER) quality control mechanism of glycoprotein folding. It is known that vertebrates and Caenorhabditis genomes harbor two uggt gene copies that exhibit differences in their properties.Bayesian phylogenetic inference based on 195 UGGT and UGGT-like protein sequences of an ample spectrum of eukaryotic species showed that uggt genes went through independent duplications in Caenorhabditis and vertebrates. In both lineages, the catalytic domain of the duplicated genes was subjected to a strong purifying selective pressure, while the recognition domain was subjected to episodic positive diversifying selection. Selective relaxation in the recognition domain was more pronounced in Caenorhabditis uggt-b than in vertebrates uggt-2 Structural bioinformatics analysis revealed that Caenorhabditis UGGT-b protein lacks essential sequences proposed to be involved in the recognition of unfolded proteins. When we assayed glucosyltrasferase activity of a chimeric protein composed by Caenorhabditis uggt-b recognition domain fused to S. pombe catalytic domain expressed in yeast, no activity was detected.The present results support the conservation of the UGGT activity in the catalytic domain and a putative divergent function of the recognition domain for the UGGT2 protein in vertebrates, which would have gone through a specialization process. In Caenorhabditis, uggt-b evolved under different constraints compared to uggt-a which, by means of a putative neofunctionalization process, resulted in a non-redundant paralog. The non-canonical function of uggt-b in the worm lineage highlights the need to take precautions before generalizing gene functions in model organisms.

The Structural Biology of Galectin-Ligand Recognition: Current Advances in Modeling Tools, Protein Engineering, and Inhibitor Design.

Galectins (formerly known as "S-type lectins") are a subfamily of soluble proteins that typically bind ß-galactoside carbohydrates with high specificity. They are present in many forms of life, from nematodes and fungi to animals, where they perform a wide range of functions. Particularly in humans, different types of galectins have been described differing not only in their tissue expression but also in their cellular location, oligomerization, fold architecture and carbohydrate-binding affinity. This distinct yet sometimes overlapping distributions and physicochemical attributes make them responsible for a wide variety of both intra- and extracellular functions, including tremendous importance in immunity and disease. In this review, we aim to provide a general description of galectins most important structural features, with a special focus on the molecular determinants of their carbohydrate-recognition ability. For that purpose, we structurally compare the human galectins, in light of recent mutagenesis studies and novel X-ray structures. We also offer a detailed description on how to use the solvent structure surrounding the protein as a tool to get better predictions of galectin-carbohydrate complexes, with a potential application to the rational design of glycomimetic inhibitory compounds. Finally, using Gal-1 and Gal-3 as paramount examples, we review a series of recent advances in the development of engineered galectins and galectin inhibitors, aiming to dissect the structure-activity relationship through the description of their interaction at the molecular level.