Karl-Heinz Baringhaus, Gerhard Hessler, Hans Matter and Christian Buning,
Sanofi-Aventis Deutschland GmbH, 65926 Frankfurt am Main, Germany

Molecular Design plays an important role in modern Drug Discovery. It is applied in lead generation for turning hits into leads and in lead optimization for fine-tuning of leads.

This presentation outlines Molecular Design principles in particular in small molecule lead optimization. For example, application of our SALI Explorer in structure-activity relationships (SAR) fosters the early identification of SAR hot spots within lead series. Subsequent quantitative structure-activity relationship is then applied to optimize compounds against the target of interest.

Most often lead series suffer from not acceptable ADMET profiles and/or anti-target activities. We have implemented several predictive data mining tools in the field of ADMET- and anti-target (e.g. hERG, CYPs) modeling. A few examples outline the scope and applicability of these tools for Medicinal Chemists and Drug Designers. In addition, in silico profiling of molecules against off-targets allows fast and early identification of liabilities of putative lead series.

The increasing interest in peptides and biologicals (e.g. proteins, antibodies) as drugs opens new opportunities for Molecular Modeling. For example, peptide design becomes more and more important in order to synthesize only the most active peptides while taking into account their physicochemical properties.

It is now twenty years since the first publication on fragment based drug discovery and the methods have matured in the past ten years to provide a powerful addition to the armory of methods used in ligand discovery¹. The central tenet is that a small library of low MW compounds can effectively sample chemical space to identify compounds that bind to a range of protein targets; these small fragments can then be grown into potent ligands (usually by structure-guided design)². Some of these compounds have been developed into effective inhibitors to probe the mechanism of action of certain protein targets. Others have been optimized to provide many clinical candidates³ and there are now two fragment-derived compounds on the market.

In this presentation, I will briefly summarize the methods and then focus on the various areas where computational and cheminformatics methods can have impact. These include the design of libraries (including novel libraries of so-called 3D fragments⁴), computational screening of fragments^5,6, using fragments to identify novel mechanisms of enzyme action⁷ and the difficult challenge of growing and merging fragments to give potent ligands.

Ben Allen, Colin Stubberfield, Sree Vadlamudi, Marie Weston, Jonny Wray

Network pharmacology models cells as networks of interacting proteins. Within this paradigm, a disease state is identified as a disorder of the entire network, and the target for pharmacological intervention becomes a set of proteins. Network modelling can be used to identify a set of key proteins and their interactions which can be targeted to correct the network disorder in a disease setting.

The chemoinformatic problem is then to identify compounds with desirable interactions with the set of proteins. The e-Therapeutics method starts by generating databases of protein-protein interactions (PPI) and compound-protein interactions (CPI), using a combination of public and proprietary collections of experimental results and computational modelling.

Biological and medical expertise is used to identify key proteins that participate in the networks that mediate disease targets of interest, and PPI data is used to construct networks based on the targets. This complex network of proteins is then analysed to identify multiple intervention points that, if impacted simultaneously, will disrupt the disease-related network. The next step is to identify potential small molecule modulators with the best overall impact in the disease-related network using proprietary computational methods. This search process takes account of the chemical biology profile of the molecules in proprietary CPI database. The lists of predicted active compounds are then triaged based on phys.chem profile, synthetic tractability and the shortlists of selected compounds are screened in cell based disease specific assays.

We believe that, by accounting more realistically for both the complexity of disease at the outset and for the many effects of the presence of a drug molecule in the body, our approach has the potential to discover more effective treatments. The e-Therapeutics network pharmacology approach has enabled us to successfully identify multiple new chemical entities for targeting Telomerase and the Hedgehog pathway in cancer and TNF alpha in anti-inflammatory disease projects. Lead optimisation is ongoing for the selected small molecule modulators. Details of the network analyses and cell-based screening results will be discussed.

The major trend in drug discovery research over the last 30 years has been to move resources from in vivo based pharmacology towards more reductionist in vitro based approaches emphasising high through-put, genomic, molecular, mechanistic and target-based aspects as key factors for predicted success. Computational chemistry is easily applied and serves, in many respects, as a cornerstone in this new strategy.

An important driver for this change was the assumption that the new approaches would be more rational. The outcome did not turn out to be an obvious success and a partial rebound has been observed in the field with “new” approaches emerging, most of them still rather reductionist in nature, nevertheless emphasising instead the importance of phenotypic and systemic aspects. Computational biology is becoming increasingly important here, since the biology data generated is much more complex.

This presentation will argue for the need of alternative drug discovery approaches to strengthen and broaden innovation in the pharmaceutical industry. Questions like what is known and what is not known in the field regarding disease progression, disease mechanism, mechanisms of action of drugs, interactions of molecules in condensed phases, and how does the industry and academia deal with these unknowns in their approaches, will be addressed. The systems pharmacology based approach taken in our labs which have resulted in the Integrative Screening Process (ISP) platform will be described in this context. ISP can be described as the engineer’s approach to drug discovery, where empirical observations on complex read-outs from both novel test compounds and compounds with known effects in man are compared in a standardized manner. Furthermore, we will discuss computational chemistry and biology methods, which are central components of ISP, with a choice of methods and their applications that is different compared to the prevailing use of such methods in drug discovery.

Many simple algorithms are described in the literature to guide good practise in drug discovery, but few are widely accessible as practical tools. Inspired by the needs of academic drug research, The Medicinal Chemistry toolkit has been written as a free and easily accessible tool. The talk will cover a number of case studies of recent drug research, highlighting how the app helps good project decision-making.

Biophysical approaches are routinely used to assess biologics activities, stability and quality. Modern drug discovery operations require characterization of biomolecular interactions to be both time- and cost-effective as well as to be highly precise and reproducible. Here we report applications of two novel predictive, precise, low protein consuming and high throughput methods: nano-Differential Scanning Fluorimetry (nanoDSF ) and MicroScale Thermophoresis (MST) that we apply in our biologics discovery and characterization operations. The examples of the demonstrated effectiveness of the nanoDSF and MST will be presented and discussed.

The audience will gain further insight into:

• new approaches in biologics discovery, development and QC
• advantages applying novel MST and nanoDSF methods
• efficiency of the newly developed technologies and applications for biologics

Harper SA, Raman MC, Rizkallah PJ, Simmons R, Donnellan Z, Dukes J, Bossi G, Le Provost GS, Todorov P, Baston E, Hickman E, Mahon T, Hassan N, Vuidepot A, Sami M, Cole DK, Jakobsen BK.

ImmTAC molecules are bispecific therapeutics that enable the immune system to recognise and kill cancer cells. Recognition occurs via a soluble T cell receptor (TCR) that binds to peptides derived from intracellular cancer targets. Killing is mediated by an anti-CD3 specific scFv that re-directs a potent T cell response.

A TCR can recognize its target peptide only in the context of a cell surface receptor. Affinity for the peptide-receptor complex (pHLA) is typically weak, >µM, allowing them to be highly promiscuous. By contrast it is necessary for ImmTAC molecules to be highly specific. Affinity enhancement whilst engineering specificity is primarily achieved through phage display, targeting mutations in the CDR loops.

Here we present an example of structure led engineering in a TCR that in the context of HLA-A*01 recognises a peptide from the cancer testis antigen Mage A3. This TCR was described in a recent clinical report from another company, where fatal cardiac toxicity followed administration of TCR-engineered T cells. Mutants were designed to reduce off-target cross-reactivity to a structurally related peptide derived from titin a protein expressed in cardiac tissue.

Cristina Sensi, Carlo Pergola

A critical quality attribute (CQA) is a physical, chemical, biological, or microbiological property or characteristic that should be within an appropriate limit, range, or distribution to ensure the desired product quality. Potential CQAs (pCQA) identification is a key initial step in the development of biopharmaceuticals, that depends on linking the expected quality profile to attributes potentially impacting Bioactivity, PK, Immunogenicity, and Safety.

In fact, the identification of pCQAs allows to focus analytical, pharmaceutical, and process development efforts on those attributes where detailed understanding or better control is needed.

Here we propose the use of structural in silico approaches to improve pCQAs identification in early development phases. In particular, an integrated use of structural tools, such as protein and antibody Homology Modeling, Molecular Docking, and Molecular Dynamics Simulations, can be used to increase knowledge on Biopharmaceutical candidates. For instance, this computational approach can allow the evaluation of the steric hindrance for different kind of glycosylation (fucosylation, sialylation, high mannose forms), of the impact of potential misincorporations on the structure, and can simulate different experimental conditions helping initial classification of aggregation potential (i.e., by increasing temperature).

In turn, the information obtained by the in silico work, together with data collected from analytical and bioanalytical laboratories, supports the understanding of biopharmaceutical entities under development, thus advancing knowledge on molecule and on molecule behaviour.

The identification of a biopharmaceutical candidate molecule is often the product of a multi-round selection process based on both biological activity and developability. Whilst biological activity can be understood experimentally in a high-throughput manner, the methods to evaluate potential developmental risks tend to be more low-throughput. Molecules are therefore risk profiled using in silico methods, helping to prioritise molecules for selection and aid interpretation of subsequent experimental data. This presentation will give an outline of the in silico methods employed, and the incorporation of CCG MOE in this process.