X-Ray Crystallography: Structure Preparation, Electron Density and Solvent Analysis
Structure Preparation / Side-Chain Rotamer Exploration / Electron Density Maps / Solvent Analysis with 3D-RISM
The course will cover methods for evaluating, analyzing and refining protein models derived from X-ray
crystallographic data. Topics related to protein structure preparation and side-chain conformational analysis and
placement will be discussed. Visualization and interpretation of electron density maps for assessing protein models
will be described. The 3D-RISM method for predicting and refining the placement of solvent molecules in protein
models will also be presented.
Structure-Based Drug Design and Ligand Modification
Molecular Surfaces and Maps / Ligand Interactions / Conformational Searching / Ligand Optimization / Ligand Selectivity / Protein Alignments and Superposition
The course covers MOE applications for interactive structure based design. Examples include active site visualization, protein-ligand contact analysis and ligand modification/optimization in the receptor pocket. Conformational searching and analysis of the ligand to assess ligand flexibility will be discussed. A protocol for aligning and superposing protein complexes in the context of protein selectivity will be studied.
Fragment-Based Drug Design Scaffold Replacement / Medicinal Chemistry Transformations / Fragment Linking / R-Group Screening / Ligand Growing / Pharmacophores / Fragment Databases
Fragment-Based Drug Design
Scaffold Replacement / Medicinal Chemistry Transformations / Fragment Linking / R-Group Screening / Ligand Growing / Pharmacophores / Fragment Databases
The course will focus on fragment-based drug design tools in MOE. Combinatorial fragment design and scaffold replacement in the receptor active site will be covered in detail, along with approaches for fragment linking and growing. A method for generating a series of closely related derivatives through medicinal chemistry transformations and the reaction based combinatorial builder will be presented. The use of pharmacophores and 2D/3D descriptors to guide drug design processes will also be discussed.
The course covers methods for analyzing and optimizing peptide-protein interactions in the active site. Topics related to peptide-protein structure preparation, peptide sequence optimization using natural and non-natural acids and conformational analysis will be discussed. Peptide-protein docking and protein-ligand interactions to analyze contact points will be described. The course will also cover advanced conformational searching using distance restraints.
The course describes the application of pharmacophores in drug discovery projects and encompasses a range of topics from pharmacophore query generation to pharmacophore refinement and searching of structural databases. A new approach based on Extended Hückel Theory (EHT) for producing pharmacophore models with encoded interaction energies in the context of ligand-based and structure-based projects will be described. The generation and analysis of protein-ligand interaction fingerprints (PLIF) will be presented along with the application of PLIF for producing pharmacophore queries. A method for combining pharmacophore models and PLIF with linear and binary QSAR models for consensus modeling will also be described.
Protein Engineering / Protein Properties / Developability / Hot Spot Analysis / Antibody Modeling / Humanization / Molecular Surfaces
The course covers approaches for structure-based antibody design and includes protein-protein interactions analysis, in silico protein engineering, affinity modeling and antibody homology modeling. The interaction of a co-crystallized antibody-antigen complex will be studied by generating and examining the molecular surfaces and visualizing protein-protein interactions in 3D and 2D. Antibody properties will be evaluated using specialized calculated protein property descriptors and analyzing protein patches. The application of protein engineering tools for homology modeling and conducting property optimization of antibodies in the context of developability will be studied. Antibody optimization examples will include identification of glycosylation sites and analysis of correlated pairs using a specialized antibody database. An approach for humanizing antibody homology models will be discussed. All the steps necessary for producing and assessing antibody homology models will be described.
The course describes advanced SBDD workflows in drug discovery projects and encompasses a range of topics from pharmacophore query generation to protein-ligand interaction fingerprints. More specifically, the course will cover the application of pharmacophores in the context of protein-ligand docking, scaffold replacement and R-group screening. A method for querying a 3D project database will also be presented along with the generation and analysis of protein- ligand interaction fingerprints (PLIF).
The course covers protein alignments and homology modeling in MOE. The homology modeling section covers the complete sequence to structure workflow, including template searching, query alignment, adjustment of the alignment between the sequence and the template, homology model building and refinement, along with applications for assessing model quality. Refinement of protein loops and loop conformations using the Loop Modeler and LowModeMD conformational searching will be discussed. Protein contacts will be evaluated after alignment and superposition of protein multimers. A method for superposing proteins that is strictly based on structural motifs will be described.
Biologics: Protein Alignments, Modeling and Docking
Protein Alignments and Superposition / Loop and Linker Modeling / Homology Modeling / Protein- Protein Docking / Protein Solubility Analysis / 2D Hot Spot Mapping / PLIF / Biologics QSAR Modeling
The course covers methods for aligning protein sequences, superposing structures, homology modeling fusion proteins and conducting protein-protein docking. In particular, an approach for aligning and superposing multiple structures will be described for determining structural and surface protein variations in relation to protein property modulation. A method for grafting and refining antibody CDR loops as well as using a knowledge-based approach to scFv fusion protein modeling using the MOE linker application will be described. An approach to generate homology models of a murine antigen structure from a human template as well as protein-protein docking of an antibody to an antigen will be discussed. A QSAR model for predicting and analyzing protein/biologics solubility will be described.
Ligand-Based Drug Design and SAR Analysis R-Group Profiles and Analysis / MOEsaic / MMP Analysis / Descriptor Calculations / Conformational Searching / Molecular Alignments / Pharmacophore Modeling and Searching / Diversity Analysis
The course covers essential in silico methods needed for guiding drug discovery projects in the absence of a protein structure. Analysis of SAR through R-group profiling and matched molecular pairs (MMP) analysis using MOEsaic to determine relationships among a chemical series are examined. Molecular descriptor calculations and their application for determining property correlations along with diversity analysis are described. Molecular alignments and conformational analyses of a congeneric series are explored to assess the impact of ligand substituents. An approach for developing pharmacophore queries is discussed. Management and manipulation of MOE databases are also covered.
Exploring Molecular Dynamics Simulations of Membranes in Drug Discovery
Jose Duca, Global Head of Computer-Aided Drug Discovery, Novartis
An important mission of our group is to advance the science of computational chemistry. We have embarked on new research to try to shed light on problems that have been elusive to the drug discovery scientific community. We have published on kinetics, solvation and non-equilibrium conditions, to mention a few. Our most recent efforts are centered on membrane simulations with the overarching goal of understanding membranes at an atomic level. Membranes are key to many biological processes including absorption, distribution, metabolism, and excretion (ADME) profiles of drug candidates.
This presentation will focus on the use of massive molecular dynamics simulations to understand ligand binding to membranes (or membrane partition), the effect compounds partition on membrane biophysics and structure-kinetics relationships of passive membrane permeation from multiscale modeling.
Tactics for amplifying the power of MOE in drug design
Hongmao Sun, Scientist, National Institutes of Health
Structural information of drug targets has proven a valuable asset to drug discovery projects. MOE is a prestigious software package for handling structural information of both proteins and ligands. One unique feature of MOE is that MOE provides a broad platform allowing users to extend and enhance its functionalities. I will discuss a case study, where fragmentation function provided by MOE is combined with data mining of the Cambridge Structure Database (CSD), and leading to discovery of novel XIAP inhibitors. In another case study, pharmacophore models generated with MOE are subjected to a genetic algorithm guided optimization, which leads to a highly selective pharmacophore model.
Bringing Computational Chemistry Capabilities into the Hands of Medicinal Chemists
Jibo Wang, Senior Research Advisor, Eli Lilly
During the optimization of hit/lead compounds, medicinal chemistry teams operate on a "design-synthesis-test" cycle where novel structural hypotheses are constantly generated and evaluated. During this iterative process good design ideas are identified and incrementally evolved into better ones. Many state of the art computational chemistry techniques, from predictive methodologies to 3D modeling to synthesis evaluation can be applied during the design stage to increase the possibility of success. Traditionally, these computational evaluations are done by computational chemistry specialists supporting these projects. Over the last few years, an increasing number of computational chemistry technologies and tools have improved both in performance and ease of use. This improvement creates opportunities for wider deployment and use, for example among non-computational chemistry experts. In this presentation we present our efforts to transform computer-aided drug design at Eli Lilly by engaging medicinal chemists as end users of these tools with computational chemists assuming an advisory/supporting role. In our experience this setting can create efficiencies during design and free up computational chemistry specialists to solve more challenging problems. Over the last few years, we have focused on building infrastructure/interfaces to facilitate this transition. In this talk, we are sharing our strategy, present the progress we have made through examples of our work and point out our future directions.
Structure-Based Hit Identification of Allosteric HIV Integrase Inhibitors
John Sanders, Principal Scientist, Merck & Co., Inc.
HIV Integrase catalyzes the incorporation of viral DNA into the host genome and is a critical part of the viral life cycle. As a result, integrase has been a protein of therapeutic interest for several decades and, to date, three integrase inhibitors have been approved for clinical use (raltegravir, dolutegravir, and elvitegravir). Clinically significant resistance mutations to drugs targeting HIV integrase have emerged with integrase inhibitor-containing treatment regimens and efforts to develop novel inhibitors of HIV integrase that overcome viral resistance have resulted in the identification of compounds with allosteric binding sites and non-canonical mechanisms of inhibition. An example of these efforts is the broad class of compounds which bind in the LEDGF/p75 binding site on HIV integrase and disrupt the association of these two proteins. Integrase inhibitors displaying this mechanism of action have been shown to possess significant antiviral potency, motivating additional discovery efforts in this area. We describe virtual screening, crystallographic, and additional computational efforts directed towards the identification of novel, allosteric integrase inhibitors.
Computational Modeling of human β-secretase 1 (BACE-1) Inhibitors using Ligand Based Approaches
Govindan Subramanian, Senior Principal Research Scientist, Zoetis
Multiple ligand based in silico quantitative structure activity relationship (QSAR) statistical modeling approaches and tools were used to comprehend the in vitro binding affinities (IC50)
of diverse small molecule human β-secretase 1 (hBACE-1) inhibitors reported in scientific literature. Departing from the standard tradition, only 230 (~13%) small molecules were used for training the system and the prediction performance evaluated using an external validation set of 1476 (~87%) inhibitors. Overall, tangible and systematic improvements were observed as the descriptive information content and complexity of the modeling technique increased. The current results demonstrate that useful and productive models is within reach by choosing appropriate modeling techniques in spite of small datasets and diverse chemical classes, a scenario typical in HTS triaging or patent busting activities.
MOEsaic: Application of Matched Molecular Pairs to Interactive SAR Exploration
Al Ajamian, Director of Business Development, Chemical Computing Group
Managing and analyzing structure activity/property relationship data in medicinal chemistry projects is becoming ever more challenging, with larger data sets and parallel development of different structural series. Tools and methods for the efficient visualization, analysis and profiling of structures therefore remain of deep interest. Here, we will describe a new application, MOEsaic,
which enhances typical medicinal chemistry workflows aimed at interrogating the SAR data through the use of interactive MMP analysis and R-group profiling, for guiding a campaign in its development.
Structural informatics of complex drug targets: Modeling flexible pharmacophores in genome space – the case of NS5A and NS5B directed inhibitors of hepatitis C
James Nettles, Adjunct Faculty-Dept. of Biomedical Informatics, Emory University School of Medicine
Even with advances in structural biology, many important protein targets cannot be fully characterized by structural experiments alone. The new classes of selective combination drugs for treating Hepatitis C represent some of the most potent and effective antivirals ever developed. This talk will summarize our development and use of multi-scale informatics workflows to build multiple receptor/drug models and test for ones that best correlate with our collaborators internal chemogenomic screening data. The resulting methods/models are robust and explain SAR for diverse chemotypes and genotypes outside our original training set.
Can Machine-Learning post-processing of docking results yield an improvement in activity prediction?
Jeff Warrington, Senior Scientist, Cytokinetics
As part of our ongoing efforts in understanding muscle biology, we investigated the use of docking and scoring to optimize our efforts in a small molecule muscle contractility program. Enabled by multiple structures, and a larger well-controlled dataset, our follow-on program offered a unique opportunity to test the abilities of recent advances in machine-learning assisted docking, and field test their use. Herein are described some of the advantages and pitfalls of such an approach.
Improvements and integration of QM and MM strain energy calculations using recently open-sourced Chemalot tools with MOE
Ben Sellers, Scientist, Genentech
An efficient and accurate method to evaluate whether a ligand design is strained would be highly useful. With the aim of identifying strain calculators for computational and medicinal chemists,
we have evaluated a number of both standard and newly-developed quantum and molecular mechanics methods.
Strain energy workflows were constructed using our recently open-sourced modeling package, "Chemalot", which not only enabled us to quickly evaluate methods but also integrate the chosen workflows within MOE. Finally, we will share some applications of using MOE and our strain calculations on small-molecule drug-design teams.
Identification and Optimization of Potent and Selective Inhibitors of PAK1
Edward J. Hennessy, Associate Principal Scientist, AstraZeneca
The p21-activated kinases (PAKs) comprise a family of six serine/threonine protein kinases, divided into two subgroups based on sequence and structural homology: group I (PAKs 1-3) and group II (PAKs 4-6). Multiple PAK family members have been shown to interact directly or indirectly with a wide variety of other proteins, and as a result have been demonstrated to regulate a variety of cellular activities, such as cytoskeletal reorganization, cellular motility, and survival.
In recent years, emerging data has pointed to the role of PAK1 signaling in certain human cancers, suggesting that inhibitors of PAK1 kinase activity may have utility as antitumor agents. However, due to the relatively open and flexible ATP-binding site of this kinase, the identification of potent PAK1 inhibitors with high kinase selectivity and drug-like physical properties has remained a challenge. In this presentation, we will discuss our efforts towards this goal, highlighting the discovery and optimization of two structurally distinct chemical series of PAK1 inhibitors.
Discovery and profiling of novel, intestinally-restricted oral pan-JAK inhibitors for the treatment of inflammatory bowel diseases
Jennifer Kozak, Research Scientist, Theravance
There remains a significant need for improved therapies to treat inflammatory bowel diseases, including ulcerative colitis (UC). The oral JAK inhibitor tofacitinib has demonstrated clinical effectiveness in treating UC patients in Phase 3 trials, although its usage may be limited by adverse events resulting from systemic drug levels. Here we report the structure-based design and profiling of a series of novel, pan-JAK inhibitors designed to be intestinally-restricted thereby minimizing systemic side effects.
Structure-Based Optimization of a Potent, Selective and CNS penetrable p70S6K/AKT Inhibitor M2698 for the Treatment of Tumors with PAM Pathway Genomic Alterations
Igor Mochalkin, Associate Director, Medicinal Chemistry and Lead Optimization, EMD Serono
The PI3K/Akt/mTOR (PAM) pathway is an essential signaling regulator of cell growth, proliferation, and metabolism. Aberrant hyperactivation of the pathway through multiple mechanisms, including loss of tumor suppressor PTEN function, amplification or mutation of Akt and exposure to carcinogens leads to dysregulated cell growth and survival. Clinical evidence suggest that the PAM pathway is an attractive therapeutic target because it serves as a convergence point to cellular processes that contribute to the initiation and maintenance of cancer.
Herein, we present the successful optimization of the quinazoline-8-carboxamide (QCA) series of dual p70S6K/Akt inhibitors. The initial lead MSC2120352 was identified in a focused, kinase library screen. MSC2120352 binds in the ATP-binding pocket of the kinase: the QCA amide group forms bidentate interactions with the hinge region (Glu173-Leu175); the electron-rich π system of the benzyl binds in the G-loop, utilizing noncovalent cation-π interactions with the catalytic lysine-123. Using rational structure-based design, MSC2120352 was subsequently optimized to a highly potent, kinase-selective, CNS penetrant p70S6K/Akt inhibitor M2698.
M2698 delivers a novel mode of PAM pathway inhibition with a different mechanism of action, compared to single-node Akt or p70S6K inhibitors or mTOR rapalogs. In a cellular context, inhibition of p70S6K by M2698 leads to potent inhibition of ribosomal protein S6 phosphorylation, while inhibition of Akt blocks the negative effects of a compensatory feedback loop. In addition, M2698 potently inhibits proliferation of multiple solid tumors, including those with PAM pathway genomic alterations. In vivo pharmacokinetic and efficacy studies of M2698 indicated a dose dependent tumor growth inhibition in models with PAM pathway genomic alterations, including, triple negative breast cancer, Her2+ breast cancer, and glioblastoma. On the basis of its superior in vitro and in vivo profile, M2698 was selected for further development and is currently being evaluated in Phase I clinical studies.
In an effort to improve sustainability, quality, and productivity of small molecule manufacturing while reducing costs, many pharmaceutical companies are expanding the use of biocatalysts to improve the synthetic route of their small molecules. Since establishing a dedicated enzyme engineering group in 2014, we have evolved enzymes from seven different enzyme classes that have improved the desired catalytic activity from ~120 fold to
over 50,000 fold. Improvements in enzyme stability is critical to the development of industrial biocatalysts that often need to operate under a stressing manufacturing environment with high temperatures and organic solvent concentrations. The application of protein stability calculations to predict stabilizing mutations relevant for industrial biocatalyst evolution will be discussed.
Central Nervous System Multi-Parameter Optimization (CNS MPO) Desirability: A Holistic Assessment of Drug Property and its Application in Discovery Projects
Xinjun Hou, Head of Neuroscience Computational Chemistry, Pfizer
Xinjun Hou, Travis T. Wager, Patrick R. Verhoest and Anabella Villalobos
Using six basic physicochemical properties commonly used in drug design, we developed CNS MPO Desirability score to holistically assess of druglikeness and identify compounds with optimal ADME attributes in one molecule (favorable permeability, minimizing P-glycoprotein efflux, increasing metabolic stability). Using retrospective data analyses and prospective application in real projects in last few years, we will show that he CNS MPO Desirability creates flexibility in design and expands design space, offering advantages over the use of single parameters or hard cutoffs for single or multiple parameters. The use of this tool has played a role in reducing the number of compounds submitted to exploratory toxicity studies and increasing the survival of our drug candidates through regulatory toxicology into First in Human (FIH) studies. Overall, the CNS MPO algorithm has helped to improve the prioritization of design ideas and the quality of the compounds nominated for clinical development.
Computational Modeling and Biophysical Analysis of Novel Biologics
Sandra Rios, Principal Scientist, Merck
Sandra Ríos*, Jennifer M. Johnson**, Brad Sherborne**, Francis Insaidoo* and David Roush*
Merck Research Labs Process Development and Engineering Dept.* and Global Chemistry** Kenilworth, NJ
Bioprocess development modalities (e.g. affinity, ion exchange, hydrophobic interaction chromatography) have been optimized and used extensively in monoclonal antibody purification from where several template purification schemes have emerged. Because of the inherent and engineered variations in therapeutic antibody structures, there is no “one-size-fits-all” when it comes to techniques for purification of monoclonal antibodies (mAbs).
Conditions optimal for antibody purification like elution pH, ionic strength, and temperature may induce aggregation or susceptibility to protease degradation for other biologics. This may pose a challenge for hold times, formulation and overall protein stability. Understanding structure and biophysical properties for novel molecules (e.g. patch analysis for charge and hydrophobicity) may give us a preliminary assessment of purification and formulation challenges and may be exploited to enhance process development to fit novel biologics to current platform purification modalities. To address the aforementioned limitations, computational algorithms were used to build homology models of multiple novel molecules and correlate structural properties with experimental data to assess the impact of these biophysical properties on purification options, including platform fit.
In this study, we employed molecules, which possess antibody-like structural features and yet different in structure. While the MOE antibody modeler tool is useful for optimal CDR (Complement Determining Region) determination, functionality is limited by the single sequence of the novel biologic. Further, homology-modeling tools might not completely capture the complexity of identifying a representative CDR for these molecules. Thus, a hybrid approach may be required such as leveraging homology models from other proteins.
Once a final homology structure is achieved, in-silico docking and biophysical characterization could be explored to understand the impact of the molecular structure in purification, stability and formulation.
Between Large and Small Molecules: Modeling Therapeutic Peptides
Kristin Andrews, Senior Scientist, Amgen
Peptide therapeutics represent a unique niche within drug discovery. Computational methods and strategies developed for either small molecules or large molecules are not always applicable to this class of molecules. Opportunities to impact peptide discovery projects using computational methods will be presented. Examples discussed will include visualization of mutational SAR data, application of protein-protein docking, and peptide stability engineering.
Developing a platform for in-silico protein design: Applications to thermostability QSAR modeling
Kenneth McGuinness, Postdoctoral Fellow, Merck
Informed protein design enables better biologics and green chemistry. To balance efficiency and accuracy, prudent design methodologies incorporating sufficient and necessary information are required. Descriptors used to predict changes in protein thermostability ranged from simple sequenced based to more costly energy terms. Using three subsets of the proTherm database, classification and regression quantitative structure relationship models were developed. Classification models enhanced predictive power. Although energy terms were deemed most important in a random forest model, MOE protein descriptors alone were sufficient for thermostability predictive enhancement in an industry setting.
Macromolecular Structure-Activity Relationship: Are We Ready to Use It for Protein Therapeutics Design?
Lei Jia, Scientist, Amgen
Quantitative Structure-Activity Relationship (QSAR) models have been maturely applied in small molecule drug design and have significant contributions to advance developments in this field. The machine learning foundation of the QSAR models can also be applied in the macromolecule regime. Due to the increase of protein therapeutics especially monoclonal antibodies developments, there is an increasing trend to apply the similar method, Macromolecular Structure-Activity Relationship (MSAR), in protein design and engineering. MSAR and similar methods have been developed to predict protein stability, biophysical properties, and chemical hotspots etc. The tools have a potential to be integrated as key steps in biologics drug development and make strong impacts in this field. Are we there yet? Some cases will be discussed in the presentation.
Prediction of Protein-Protein Binding Sites and Epitope Mapping
John Gunn, Senior Research Scientist, Chemical Computing Group
Computational modeling of protein-protein interactions is increasingly playing a more important role in the design and optimization of biologics. Computational methods for predicting protein-protein binding sites or epitopes have profound applications in a number of areas in the development of biologics, from understanding the mode of action to the modulation of protein properties. This work presents a novel algorithm for predicting likely antigen epitopes from protein-protein docking results using the MOE software platform. The approach generates an ensemble of poses which represent the most favorable interactions. Protein-protein residue contacts are then used to generate interaction fingerprints which serve to identify Boltzmann-weighted clusters of poses and extract consensus epitope residues. This method produces at least one predicted epitope with significant overlap to the native structure ranked in the top five clusters.
Optimizing Protein Properties in the Cloud with MOE
Essam Metwally, Senior Scientist, Chemical Computing Group
Protein-display systems, such as phage-display, have demonstrated the power of simultaneously examining: stability; the ability to bind to a target; and the impact of site-specific mutations. Unfortunately, traditional approaches to computer aided molecular design are often intractable and unable to prospectively suggest viable mutations due to the sheer size of the available mutations space. Leveraging the power MOE deployed in a cloud environment, we present a Virtual Phage Display approach which is able to produce predictive models for both thermostability and solubility in under an hour. This MOE-based methodology coupled with the power of the cloud allows modeling to be a prospective exercise, informing and influencing projects from the outset, before time and effort are expended at the bench.
Computational modeling of antibody-antigen interactions
Arvind Sivasubramanian, Senior Scientist I Computational Biology, Adimab
Benchmark sets containing the crystal structures of protein complexes and the unbound structures of the binding partners have been critical in the development of computational methods to model protein:protein interactions. Using these benchmarks, several protein:protein docking algorithms have been developed and tested to predict the 3D structure of protein complexes using the conformations of the individual partners. Despite great strides in this field, there remains room for improvement in the modeling of antibody:antigen complexes. In this work, we present an updated structural benchmark for antibody:antigen docking. The benchmark consists of 291 unique complexes curated from the Protein Data Bank (PDB) and features > 150 unique antigens, representing a substantial increase over similar existing resources. The complexes in the benchmark represent a diverse range of interfaces with respect to size, electrostatic character, and conformational change upon binding. Using this updated benchmark, we propose several improvements to improve antibody:antigen docking performance relative to existing methods.
Structural analysis of antibody-antigen complexes by crystallography and modelling
Rafael Depetris, Principal Scientist, Kadmon Pharmaceuticals
X-Ray crystallography is a very accurate tool to determine the structure of small and large proteins and protein complexes at high resolution. However, obtaining good quality crystals can be a bottleneck that undermines even the most determined efforts. Building models of the target proteins and complexes is an extremely helpful step for this purpose. It can not only produce preliminary information but also help in further protein engineering. Nevertheless, a careful approach is needed during the modeling process in order to establish the selection of binding site and set up proper validation. We will present an example and show how generating models helps to know better the proteins that one is dealing with.