Structure Preparation / Side-Chain Rotamer Exploration / Space Groups / Electron Density-Guided Docking / 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. Electron-density guided docking for generating ligand poses will discussed. 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.
Protein Engineering / Protein Properties / Developability / Hot Spot Analysis / Antibody Modeling / Molecular Surfaces
The course covers approaches for structure-based biologics 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 evaluated by generating and examining the molecular surface topology and visualizing protein-protein interactions in 3D and 2D. The application of protein engineering tools for homology modeling and conducting property optimization of an antibody-antigen complex in the context developability will be studied.All the steps necessary for producing and assessing antibody homology models will be described. Methods for evaluating the structure of protein and antibody models will be discussed.
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 aligning and super-positioning protein multimers. A method for superposing proteins that is strictly based on structural motifs will be described.
The course describes advanced SBDD workflows in drug discovery projects and encompasses a range of topics from phamacophore 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).
Biologics Modeling: Protein Alignments, Advanced Protein Modeling and Docking
Protein Alignments / Protein Super-positioning / Loop Modeling / Linker Modeling / Homology Modeling / Protein- Protein Docking
The course covers methods for aligning protein sequences, super-positioning structures, homology modeling fusion proteins and conducting protein-protein docking. In particular, an approach for aligning and super-positioning multiple structures will be described for determining structural and surface protein variations in relation to protein property behavior. 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.
Organizing Structural Project Data and Protein Family Modeling
Automated Data Organization Protocol / Protein-Ligand Interaction Fingerprints / Search Application for Structure- and Sequence-based Queries / Specialized Protein Family Databases
The course describes the MOE Project automated protocol for organizing, managing and sharing data sets across a drug discovery organization. Topics include an overview of the Project Search application used for generating, searching and combining queries, an XML editor for generating specialized protein family databases, and analysis of receptor-ligand contacts through Protein-Ligand Interaction Fingerprints (PLIF) analysis.
From First Principles to Discovery: Reshaping the Way We Do Modeling
José Duca, Global Head of Computer-Aided Drug Discovery, Novartis
We have created a new paradigm to explore possible solutions to two of the most relevant uncertainties in drug discovery: the structure-energy and the in vitro-in vivo relationships. Our novel approaches rely on first principles and imply the need to re-think and re-learn structure-based drug design and its application to drug discovery. The talk will exemplify several aspects of this new approach and its application to real-world examples covering the use of kinetics, dynamic modeling, solvation and flexibility.
Extracting Knowledge from Large In-vitro Metabolic Stability Data Sets Using Matched Molecular Pair Analysis (MMPA)
Hao Zheng, Scientist, Genentech
Through many years of drug discovery effort, pharmaceutical companies have accumulated a large set of in-vitro ADME data. Useful knowledge can be extracted from the data using matched molecular pair (MMP) and statistical testing.
Byron DeLaBarre, Founder, The Consulting Biochemist
"The Consulting Biochemist", Arlington, MA, USA firstname.lastname@example.org
The interaction of small molecules with large proteins is a mainstay of modern drug discovery. Modern tools routinely enable an atomistic understanding of these interactions. There is usually a focus on the orthosteric binding of these small molecules with the active site of the protein. However – the active site typically constitutes only a small part of a functional protein. The surface areas that become exposed as proteins undulate through their natural range of motions present many more allosteric binding sites with which biomolecules can interact. Moreover, these allosteric interactions most likely outnumber the well-studied orthosteric interactions.
An interesting place to explore allosteric interactions can be found in the metabolic pathways responsible for converting nutrients into energy and/or biological building blocks. There are on the order of 103 proteins in these pathways and a similar order of magnitude of small molecules in the metabolome. Therefore, there exists the possibility of 106 single interactions and an even greater number (it has ~5.5 million digits) of multiple interactions. It is highly unlikely that all of these potential interactions would be focused on the active site region of the protein. This leads us back again to allostery. While we know of many examples of important allosteric interactions, the allosteric language of the cell is largely unexplored. Deeper knowledge of this largely hidden information network within the cell may help to better understand key physiological processes and thereby accelerate the pace of drug discovery.
This talk will discuss computational methods for the prediction of protein allosteric binding sites and how one might test such predictions via a rapid biophysical screening of a chemical library. The computational methods are based on understanding the covariance of protein residues and the small molecule libraries are fragment sized and drug discovery oriented. The coupling of these methods will enable us to better exploit the biomolecules we seek to control via small molecule based therapies.
Structure-Based Drug Design of the Eg5 Inhibitor NVP-BQS481 Dirksen Bussiere, Director, Structural and Biophysical Chemistry, Novartis
Integrated X-ray Crystal Water Site Determination using MOE, 3D-RISM, and Phenix/DivCon
Lance Westerhoff, President and General Manager, QuantumBio Inc.
Structure Based Drug Discovery (SBDD) is employed by virtually all pharmaceutical R&D organizations, and understanding the protein:ligand complex structure along with explicit solvent effects is necessary to obtain meaningful results from docking, thermodynamic calculations, and active site exploration. Phenix/DivCon is able to accurately elucidate the protein:ligand complex structure through in situ treatment of the structure using quantum mechanics; however, at standard SBDD resolutions, the crystallographic data unambiguously reveals only a small fraction of water molecules in protein crystals - even within the first hydrogen shell of the protein molecule. Further, the implicit solvent correction in conventional methods does not take into account non-linear effects of hydrogen bonding and dispersion interactions introduced by the nearest hydration shells. To address this deficiency, we have used the 3D Reference Interaction Site Model (3D-RISM) method as implemented in MOE to filter crystallographic map data and create a more complete first solvation shell of the biomolecular complex. The combination, implemented within the Phenix/DivCon refinement workflow, allows us to capture weaker difference density peaks and thus "rescue" water sites that are normally undetectable using conventional crystallographic protocols.
Correlating Protein-ligand Activity to Quantum-mechanics/molecular-mechanics Binding Energies
Alejandro Crespo, Associate Principal Scientist, Merck
Victoria Lim and Alejandro Crespo
Understanding and quantifying protein-ligand interactions is a key step of the drug discovery process. However, obtaining a reliable correlation between binding and activity that can be used with predictive power is still an active area of research. This is because most ranking and scoring methods are based on classical-mechanics approximations and hence, the underlying level of theory may not be adequate for describing the intrinsic nature of protein-ligand interactions or desolvation effects. In this work, we propose applying a high-level ab initio method to predict relative binding energies, based on a quantum-mechanical description of the ligand and the surrounding residues in combination with a classical-mechanics treatment of the remaining environment. Our results suggest QM/MM scoring is better suited for ranking congeneric ligand series where induced fit effects are smaller. On average, QM/MM scores are better than those based on classical mechanics alone and are comparable to the state-of-the-art FEP scoring implementation.
Ligand efficiency is a simple, perhaps overly simple, concept that has been widely adopted in the industry. In spite of this, the concept has numerous sharp critics that argue the application of LE has actually had a negative impact on drug discovery. I will discuss some of the problems with LE and describe an alternative approach to answering the fundamental question of how effectively a given ligand binds to its target.
Quantifying long standing organic chemistry principles for small molecules potential energy surface prediction
Nicolas Moitessier, Professor, McGill University
Qualitative chemical principles such as the effect of electronegativity on conformational preferences (through hyperconjugation), the impact of steric clashes on stereochemical outcome of reactions and the impact of resonance and inductive effects on the shape and reactivity of molecules have been used to rationalize experimental observations. In parallel, computational chemistry has evolved to a point where these observations can often be measured quantitatively. While computational chemists speak about electron densities and molecular orbitals, organic chemists speak about partial charges and localized molecular orbitals. We proposed to bridge these two schools by encoding and quantifying chemistry knowledge and qualitative principles aiming at developing predictive methods. The broad applicability of these principles ensures transferability of this approach to any complex molecule. As an application, we thought to understand the conformational behaviors of molecules and to encode this knowledge back into a tool computing conformational potential energy. We will present our hypothesis, the modeled principles together with applications/validations of this novel approach to sets of molecules.
Application of Hückel Theory Descriptors to QSPR Models and pKa
Paul Labute, President and CEO, Chemical Computing Group
All alternative to the method of Group Contributions (atom types) in QSPR modeling is presented. A small number of descriptors derived from a modified Hückel Theory (2D) calculation are used in place of tens or hundreds of atom type descriptors for creating QSPR models of Molar Refractivity, logP, logS, Boiling Point and Free Energy of Hydration. The method is extended to the calculation of pKa and logD. The results of computational experiments demonstrate the validity of the approach.
Discovery of New Inhibitors of Baterial Thymidylate Kinase (TMK) Sameer Kawatkar, Senior Scientist, AstraZeneca
Homology Modeling and Electrophysiology Studies to Increase the Selectivity of NaV1.7 Inhibitors
Deping Wang, Associate Principal Scientist, Merck
NaV1.7 is a voltage-gated sodium channel genetically validated to play a critical role in nociception and pain. Local anesthetics (LAs) are effective at reducing pain but lack subtype selectivity, resulting in safety risks. Our team has identified small molecule NaV blockers with selectivity between NaV1.7 and NaV1.5 (cardiac isoform) and identified their binding site by manual patch clamp using an innovative chimeric channel approach. This approach, using a library of NaV1.7/1.5 chimeric constructs, led to the identification of a novel binding pocket in the less-conserved DIV, S2S3 voltage sensor region of the sodium channel alpha-subunit. This site is distinct from both the highly conserved LA binding site (DIV, S6 pore region) and other known toxin inhibition sites. From these efforts, a homology model was constructed to inspire design elements that led to the development of novel and potent aryl-sulfonamide compounds, offering the potential for highly selective NaV1.7 inhibitors. Furthermore, these mutagenesis and homology studies identified key point residues, Y1537 (DIV, S2), W1538 (DIV, S2), and A1585 (DIV, S3), responsible for aryl-sulfonamide binding and selectivity over NaV1.5. This approach continues to inspire target design and aids in rationalization of selectivity over NaV isoforms.
GPCR Drugs, Where We've Been, Where We Are and Where We're Going
Andrew Tebben, Senior Principal Scientist, Bristol-Myers Squibb
GPCRs have long been a rich source for drug discovery targets, currently compromising ~30% of marketed drugs with 17 of the 50 most prescribed drugs working through GPCRs. In contrast to the large set of individual drug substances, the number of distinct targets is significantly smaller. For example, the 17 most prescribed drugs hit only 7 targets. The beta receptor alone accounts for 6 of these 17. This talk will explore the history GPCR drug approvals within the pharmaceutical industry, focusing on coverage and redundancy across therapeutic areas and GPCR families. The talk will also address areas within the GPCR genome that remain undrugged and potential future opportunities.
Improving Covalent Target Modification Using Computational Approaches
Ye Che, Senior Principal Scientist, Pfizer
Abstract: In recent years, there is an increasing interest in the pursuit of targeted covalent drugs as well as designing chemical biology and structural biology probes using covalent target modification. Covalent inhibitors often exhibit slow off-rates, increased binding potency and biochemical efficiency, enhanced selectivity, reduced propensity for target-based drug resistance and prolonged pharmacodynamic effects, whereas covalent chemical biology probes enable target fishing via selective modification of specific residues and enable confidence in rationale studies via target engagement or increasing base-line activation in assay development while covalent structural biology probes enable thermal stabilization of target proteins for crystallography studies. However, covalent inhibitors present unique pharmacokinetic and safety challenges when developed as therapeutic agents. Central to the rational design of covalent inhibitors with improved benefit-risk balance is elevating our understanding of the reaction between electrophilic inhibitors and biological nucleophiles based on physical organic chemistry principles.
This talk will describe ab initio calculations that were used to calculate theoretical reactivities of diverse electrophilic warheads to different biological nucleophiles such as cysteine and serine nucleophiles and comparative studies with simple, electrophilicity descriptor calculations based on ligands alone and the need for higher level theoretical methods. High level ab initio calculations were used to characterize reaction energy surfaces for a broad set of irreversible and slowly reversible warheads targeting nucleophiles in aqueous solutions and in protein binding sites. Results indicate that transition states and reaction barriers are impacted by various factors, such as frontier orbitals, solvation, steric effects and H-bonding activation. In addition, examples of structure-based design of fine-tuning reactivity of covalent war-heads by introducing specific functionalities will also be discussed with illustrative examples that lead to clinical compounds for the treatment of chronic diseases.
Bcrp and P-glycoprotein: Two Peas in a Pod, or Are They?
Elena Dolgikh, Research Scientist, Eli Lilly
Breast Cancer Resistance Protein (BCRP) is a member of the ATP-binding cassette family of transporters collectively known for ATP-dependent transport of a wide range of endo- and xenobiotics. Highly expressed in humans at the gut, bile canaliculi and the blood-brain barrier among other barrier tissues, BCRP has been shown to impact absorption and excretion of a variety of chemotherapeutic drugs while its overexpression in tumors has been associated with increased multidrug resistance.1
BCRP is a half-transporter that is believed to form a homodimer and assume similar topology to that of P-glycoprotein (P-gp), another ABC efflux transporter known to significantly affect pharmacokinetic profiles of many drugs. The two transporters have an overlapping but distinct substrate specificity and while much is known about P-gp’s substrate preference2, that of BCRP’s is much less understood.
In this study we present results of our analysis of a large dataset of compounds determined to be either substrates or nonsubstrates of P-gp and BCRP in in vitro monolayer efflux assays. We compare their physicochemical property profiles with respect to preference for the two transporters and examine in vitro-in vivo correlation between BCRP /P-gp monolayer efflux and passive permeability and in vivo-measured brain-to-plasma ratios and Kp,uu values.
New Scaffolds Targeting DNA Gyrase – Fragment Optimization and Structure Based Drug Design
Jason Cross, Institute Senior Research Scientist, MD Anderson Cancer Center
DNA gyrase is a important target for antibacterials, as evidenced by the broad clinical use of the quinolone class of antibiotics. Target-based resistance to these drugs has driven recent efforts to identify novel methods for inhibiting this enzyme complex. This talk will outline fragment-screening strategies that were employed to find novel chemical leads that were progressed by medicinal chemistry into inhibitors with in vivo activity. The use of X-ray crystallography and structure-based design techniques were critical to the advancement of the project and will be emphasized.
Alexey Teplyakov, Galina Obmolova, Thomas J. Malia, Jinquan Luo, Salman,Muzammil, Raymond Sweet, Juan Carlos Almagro† and Gary L. Gilliland. Janssen Research & Development LLC, 1400 McKean Road, Spring House, PA 19477, USA
The crystal structures of a set of 16 germline variants comprised of four different kappa light chains paired with four different heavy chains have been determined to support antibody therapeutic development. All four heavy chains of the Fabs have the same CDR H3 that was reported in an earlier Fab structure. The structure analyses include comparisons of the overall structures, canonical structures of the CDRs and the VH:VL packing interactions. The CDR conformations for the most part are tightly clustered, especially for the ones with shorter lengths. The longer CDRs with tandem glycines or serines have more conformational diversity than the others. CDR H3, despite having the same amino acid sequence, exhibits the largest conformational diversity. About half of the structures have CDR H3 conformations similar to that of the parent; the others diverge significantly. One conclusion is that the CDR H3 conformations are influenced by both their amino acid sequence and their structural environment determined by the heavy and light chain pairing. The stem regions of 14 of the variant pairs are in the ‘kinked’ conformation and only two are in the extended conformation. The packing of the VH and VL domains is consistent with our knowledge of antibody structure, and the tilt angles between these domains cover a range of 11 degrees. Two out of 16 structures showed particularly large variations in the tilt angles when compared with the other pairings. The structures and their analyses provide a rich foundation for future antibody modeling and engineering efforts.
Quantifying Chemical Liability Risks in Protein-based Biologics via Molecular Modeling
Nikolay Plotnikov, Postdoctoral Fellow, Pfizer
Nikolay Plotnikov and Sandeep Kumar email@example.com
Chemical degradation during manufacturing and storage is one of the main challenges in development of protein-based biologics. Deamidation of asparagine (Asn) and isomerization of aspartic acid (Asp) represent major chemical degradation pathways of monoclonal antibodies (mAb). These adverse chemical reactions can decrease potency of mAbs (1-3) or increase their immunogenicity. Presently, chemical liabilities of Asp-X and Asn-X motifs in a drug candidate are quantified experimentally, predominantly, at the formulation stage. Quantitative reliable modeling would facilitate earlier assessment of chemical degradation risk, at the discovery stage, when it is still possible to modify protein sequence. Empirical statistical models have provided some qualitative insight on degradation risks (4). However, quantitative accuracy of this knowledge-based approach is limited by quality and transferability of models’ training set, since they are based on establishing correlations of the existing experimental data with a set of empirically chosen sequence- and structure-based descriptors. In this contribution, we propose a parameterization-free (ab initio) approach to quantify degradation risks. This approach relies on computing relevant free energy barriers for chemical reactions involved in degradation with physics-based models. Thus, it can be used as a standalone tool (with no experimental data available) or in combination with empirical models to screen potential degradation sites in biologics. We demonstrate application of these tools to examine effects of the protein environment in characterized degradation sites of commercially available biologics and to evaluate existing mechanistic proposals for chemical reactions involved in degradation.
Protein-Protein Docking with Sequential Coarse-Grained Minimization
John Gunn, Senior Research Scientist, Chemical Computing Group
Computer modeling of protein-protein docking is an important tool for a variety of applications. This work presents a novel algorithm for performing docking calculations implemented using the MOE software package. Protein structures are represented by a coarse-grained bead model which accurately reproduces the Van der Waals, electrostatic, and solvation energies of the corresponding all-atom model at the same resolution. An exhaustive search of translational coordinates using a FFT approach is followed by a series of partial minimization and filtering steps to progressively reduce the number of solutions. Results are presented validating the performance of the method on the protein docking benchmark 5.0, as well as using automatically-generated constraints in the case of antibody receptors and simulated site constraints in the general case to reflect realistic applications. The output poses generated by the program are shown to produce good high-quality structures when further refined using all-atom minimization.
Thinking Ahead: Hit Optimization Using In Silico, Functional and Analytical Tools in Early Discovery for Better Lead Molecules in Late Discovery
Vanita Sood, Associate Director, EMD Serono
PhD, Global Head of Drug Structure, Prediction and Design Discovery Technologies, EMD Serono Research and Development Institute
A critical step in the development of biologic drugs is the development of a process to express, formulate and stably store the antibody under relevant conditions for supplying clinical trials and, eventually, market. In typical antibody discovery work streams, early discovery research teams focus intensively on affinity and potency of candidate hit molecules, and select any sequence that performs well according to these criteria. These sequences may not be optimal for downstream analytics and manufacture, and can potentially contain sequence based liabilities such as deamidation motifs, cryptic glycosylation sites, oxidation sites, and others. A process for rapidly optimizing an antibody sequence to remove these liabilities while maintaining potency will be presented. The process relies on computational sequence analysis, molecular modeling and high throughput cloning, expression and testing of multiple sequence variants using automation. Data to support the improved stability of these optimized variants will be presented in addition to an unexpected improvement in potency.
Engineering & Screening Antibodies for Improved Manufacturability
Neeraj Agrawal, Scientist, Amgen
Therapeutic antibodies generated from immunization or in vitro display approaches quite often have undesirable manufacturability properties and thus are not well-suited for further development. These manufacturability liabilities could arise due to chemical modification of antibody amino acids, antibody aggregation, viscosity or thermal instability, etc. Proactive antibody engineering and in slico manufacturability assessment, based on sequence, structure analysis combined with past experiences, allows remediating these liabilities early on and ensures that only functionally active and manufacturable molecules progress to the development phase. This talk will focus on sequence and structure based analysis of antibodies to identify their potential manufacturability liabilities, to remediate these liabilities via rational protein engineering, and to rank antibody candidates for their aggregation & viscosity liabilities.
Predicting Changes in Antibody-antigen Binding Affinities
Sarah Sirin, Senior Scientist, Structural Bioinformatics, Abbvie
Antibodies (Abs) are a crucial component of the immune system and are often used as diagnostic and therapeutic agents. The need for high-affinity and high-specificity antibodies in research and medicine is driving the development of computational tools for accelerating antibody design and discovery. We report a diverse set of antibody binding data with accompanying structures that can be used to evaluate methods for modeling antibody interactions. Our Antibody-Bind (AB-Bind) database includes 1101 mutants with experimentally determined changes in binding free energies (ΔΔG) across 32 complexes. The database, AB-Bind, was used to benchmark computational scoring potentials for their ability to predict observed changes in binding free energies. Although there was a clear signal in tests discriminating mutations that improved/reduced binding, the prediction performance of all methods was modest, indicating a continued need to improve computational approaches for binding affinity predictions.
Computational Assessment of Pharmaceutical Properties for Protein Therapeutics
Stanley Krystek, Senior Principal Scientist, Bristol-Myers Squibb
Challenges for developing antibody-based therapies include protein variants often differ in their biophysical and biochemical properties it is essential to characterize protein stability along with pharmacokinetic and pharmacodynamic properties. Following the identification of a series of potent antibodies a substantial part of the optimization process focuses on comparing and improving the developability properties of lead molecules. We have developed a series of computational methods that augment experimental methodologies and are part of a strategy used to understand and optimize the structure of clinical candidates increasing protein homogeneity and providing a strategy for development. Using a web-based platform the computational analysis of biologic candidate molecules provides scientists comprehensive visualization of potential liabilities using annotations on both antibody sequence and 3D molecular structure. Finally, an automated pipeline was created to map the calculated properties on to the 3D molecular structure for interactive viewing of the 3D structure directly in the web browser.