From mundane to surprising nonadditivity: drivers and impact on ML models.

Nonadditivity (NA) in Structure-Activity and Structure-Property Relationship (SAR) data is a rare but very information rich phenomenon. It can indicate conformational flexibility, structural rearrangements, and errors in assay results and structural assignment. While purely ligand-based conformational causes of NA are rather well understood and mundane, other factors are less so and cause surprising NA that has a huge influence on SAR analysis and ML model performance. We here report a systematic analysis across a wide range of properties (20 on-target biological activities and 4 physicochemical ADME-related properties) to understand the frequency of various different phenomena that may lead to NA. A set of novel descriptors were developed to characterize double transformation cycles and identify trends in NA. Double transformation cycles were classified into "surprising" and "mundane" categories, with the majority being classed as mundane. We also examined commonalities among surprising cycles, finding LogP differences to have the most significant impact on NA. A distinct behavior of NA for on-target sets compared to ADME sets was observed. Finally, we show that machine learning models struggle with highly nonadditive data, indicating that a better understanding of NA is an important future research direction.

Novel Indane-Containing NBTIs with Potent Anti-Gram-Negative Activity and Minimal hERG Inhibition.

Novel bacterial topoisomerase inhibitors (NBTIs) make up a promising new class of antibiotics with the potential to combat the growing threat of antimicrobial resistance. Two key challenges in the development of NBTIs have been to obtain broad spectrum activity against multidrug-resistant Gram-negative bacteria and to diminish inhibition of the hERG cardiac ion channel. Here we report the optimization of a series of NBTIs bearing a novel indane DNA intercalating moiety. The addition of a basic, polar side chain connected to the indane by an ether or an N-linked secondary amide linkage together with a lipophilicity-lowering modification of the enzyme binding moiety led to compounds such as 2a and 2g which showed excellent broad spectrum potency and minimal hERG inhibition. Compound 2a demonstrated robust bactericidal in vivo activity in a mouse lung infection model with the strain P. aeruginosa ATCC 27853 which is resistant to several clinically relevant antibiotics. Rodent pharmacokinetic studies with 2a revealed an unusual profile characterized by rapid tissue distribution and a prolonged, flat terminal phase. This profile precluded further development of these compounds as potential new antibiotics.

Discovery of a Series of Indane-Containing NBTIs with Activity against Multidrug-Resistant Gram-Negative Pathogens.

The rise of multidrug-resistant (MDR) Gram-negative bacteria is a major global health problem necessitating the discovery of new classes of antibiotics. Novel bacterial topoisomerase inhibitors (NBTIs) target the clinically validated bacterial type II topoisomerases with a distinct binding site and mechanism of action to fluoroquinolone antibiotics, thus avoiding cross-resistance to this drug class. Here we report the discovery of a series of NBTIs incorporating a novel indane DNA binding moiety. X-ray cocrystal structures of compounds 2 and 17a bound to Staphylococcus aureus DNA gyrase-DNA were determined, revealing specific interactions with the enzyme binding pocket at the GyrA dimer interface and a long-range electrostatic interaction between the basic amine in the linker and the carboxylate of Asp83. Exploration of the structure-activity relationship within the series led to the identification of lead compound 18c, which showed potent broad-spectrum activity against a panel of MDR Gram-negative bacteria.

Discovery and characterization of MAPK-activated protein kinase-2 prevention of activation inhibitors.

Two structurally distinct series of novel, MAPK-activated kinase-2 prevention of activation inhibitors have been discovered by high throughput screening. Preliminary structure-activity relationship (SAR) studies revealed substructural features that influence the selective inhibition of the activation by p38α of the downstream kinase MK2 in preference to an alternative substrate, MSK1. Enzyme kinetics, surface plasmon resonance (SPR), 2D protein NMR, and X-ray crystallography were used to determine the binding mode and the molecular mechanism of action. The compounds bind competitively to the ATP binding site of p38α but unexpectedly with higher affinity in the p38α-MK2 complex compared with p38α alone. This observation is hypothesized to be the origin of the substrate selectivity. The two lead series identified are suitable for further investigation for their potential to treat chronic inflammatory diseases with improved tolerability over previously studied p38α inhibitors.

Beyond the scope of free-Wilson analysis. 2: Can distance encoded R-group fingerprints provide interpretable nonlinear models?

In a recent study, we presented a novel quantitative-structure-activity-relationship (QSAR) approach, combining R-group signatures and nonlinear support-vector-machines (SVM), to build interpretable local models for congeneric compound sets. Here, we outline further refinements in the fingerprint scheme for the purpose of analyzing and visualizing structure-activity relationships (SAR). The concept of distance encoded R-group signature descriptors is introduced, and we explore the influence of different signature encoding schemes on both interpretability and predictive power of the SVM models using ten public data sets. The R-group and atomic gradients provide a way to interpret SVM models and enable detailed analysis of structure-activity relationships within substituent groups. We discuss applications of the method and show how it can be used to analyze nonadditive SAR and provide intuitive and powerful SAR visualizations.

Potential strategies for increasing drug-discovery productivity.

The productivity challenge facing the pharmaceutical industry is well documented. Strategies to improve productivity have mainly focused on enhancing efficiency, such as the application of Lean Six Sigma process improvement methods and the introduction of modeling and simulation in place of 'wet' experiments. While these strategies have their benefits, the real challenge is to improve effectiveness by reducing clinical failure rates. We advocate redesigning the screening cascade to identify and optimize novel compounds with improved efficacy against disease, not just with improved potency against the target. There should be greater use of disease-relevant phenotypic screens in conjunction with target-based assays to drive medicinal chemistry optimization. An opportunistic approach to polypharmacology is recommended. There should also be more emphasis on optimization of the molecular mechanism of action incorporating understanding of binding kinetics, consideration of covalent drug strategies and targeting allosteric modulators.

Chemical predictive modelling to improve compound quality.

The 'quality' of small-molecule drug candidates, encompassing aspects including their potency, selectivity and ADMET (absorption, distribution, metabolism, excretion and toxicity) characteristics, is a key factor influencing the chances of success in clinical trials. Importantly, such characteristics are under the control of chemists during the identification and optimization of lead compounds. Here, we discuss the application of computational methods, particularly quantitative structure-activity relationships (QSARs), in guiding the selection of higher-quality drug candidates, as well as cultural factors that may have affected their use and impact.

The design and synthesis of novel, potent and orally bioavailable N-aryl piperazine-1-carboxamide CCR2 antagonists with very high hERG selectivity.

A novel N-aryl piperazine-1-carboxamide series of human CCR2 chemokine receptor antagonists was discovered. Early analogues were potent at CCR2 but also inhibited the hERG cardiac ion channel. Structural modifications which decreased lipophilicity and basicity resulted in the identification of a sub-series with an improved margin over hERG. The pharmacological and pharmacokinetic properties of the lead compound from this series, N-(3,4-dichlorophenyl)-4-[(2R)-4-isopropylpiperazine-2-carbonyl]piperazine-1-carboxamide, are described.

The discovery of N-cyclopropyl-4-methyl-3-[6-(4-methylpiperazin-1-yl)-4-oxoquinazolin-3(4H)-yl]benzamide (AZD6703), a clinical p38α MAP kinase inhibitor for the treatment of inflammatory diseases.

A novel, potent and selective quinazolinone series of inhibitors of p38α MAP kinase has been identified. Modifications designed to address the issues of poor aqueous solubility and high plasma protein binding as well as embedded aniline functionalities resulted in the identification of a clinical candidate N-cyclopropyl-4-methyl-3-[6-(4-methylpiperazin-1-yl)-4-oxoquinazolin-3(4H)-yl]benzamide (AZD6703). Optimisation was guided by understanding of the binding modes from X-ray crystallographic studies which showed a switch from DFG 'out' to DFG 'in' as the inhibitor size was reduced to improve overall properties.

Better compounds faster: the development and exploitation of a desktop predictive chemistry toolkit.

Today's drug designer has access to vast quantities of data and an impressive array of sophisticated computational methods. At the same time, there is increasing pressure on the pharmaceutical industry to improve its productivity and reduce candidate drug attrition. We set out to develop a highly integrated suite of design and data analysis tools underpinned by the best predictive chemistry methods and models, with the aim of enabling multi-disciplinary compound design teams to make better informed design decisions. In this article we address the challenges of developing a powerful, flexible and user-friendly toolkit, and of maximising its exploitation by the design community. We describe the impact the toolkit has had on drug discovery projects and give our perspective on the future direction of this activity.

Balancing hERG affinity and absorption in the discovery of AZD5672, an orally active CCR5 antagonist for the treatment of rheumatoid arthritis.

Modifications to a series of potent and selective substituted 1-(3,3-diphenylpropyl)-piperidine phenylacetamide CCR5 antagonists were explored with the aim of reducing affinity at the hERG cardiac ion channel. Replacement of one aromatic ring in the diphenylpropyl region with less lipophilic, saturated heterocyclic rings and subsequent optimisation of the other phenyl ring led to the identification of clinical compound AZD5672 which retained excellent potency while reducing hERG affinity. Modulating lipophilicity affected the interplay between potency, hERG affinity and absorption. AZD5672 was found to have an acceptable balance of these properties and was progressed to a phase II clinical trial to test the hypothesis that inhibition of CCR5 will bring benefits in the treatment of rheumatoid arthritis.

Orally active achiral N-hydroxyformamide inhibitors of ADAM-TS4 (aggrecanase-1) and ADAM-TS5 (aggrecanase-2) for the treatment of osteoarthritis.

A new achiral class of N-hydroxyformamide inhibitor of both ADAM-TS4 and ADAM-TS5, 2 has been discovered through modification of the complex P1 group present in historical inhibitors 1. This structural change improved the DMPK properties and greatly simplified the synthesis whilst maintaining excellent cross-MMP selectivity profiles. Investigation of structure-activity and structure-property relationships in the P1 group resulted in both ADAM-TS4 selective and mixed ADAM-TS4/5 inhibitors. This led to the identification of a pre-clinical candidate with excellent bioavailability across three species and predicting once daily dosing kinetics.

The design and synthesis of novel N-hydroxyformamide inhibitors of ADAM-TS4 for the treatment of osteoarthritis.

Two series of N-hydroxyformamide inhibitors of ADAM-TS4 were identified from screening compounds previously synthesised as inhibitors of matrix metalloproteinase-13 (collagenase-3). Understanding of the binding mode of this class of compound using ADAM-TS1 as a structural surrogate has led to the discovery of potent and very selective inhibitors with favourable DMPK properties. Synthesis, structure-activity relationships, and strategies to improve selectivity and lower in vivo metabolic clearance are described.

Automated QSAR with a Hierarchy of Global and Local Models.

We present an automated QSAR procedure that is used in AstraZeneca's AutoQSAR system. The approach involves automatically selecting the most predictive models from pools of both global and local models. The effectiveness of this QSAR modelling strategy is demonstrated with a retrospective study that uses a diverse selection of 9 early stage AstraZeneca drug discovery projects and 3 physicochemical endpoints: LogD; solubility and human plasma protein binding. We show that the strategy makes a statistically significant improvement to the accuracy of predictions when compared to an updating global strategy, and that the systematic biases inherent in the global model predictions are almost completely removed. This improvement is attributed to the model selection aspect of the strategy.

Modulators of the human CCR5 receptor. Part 3: SAR of substituted 1-[3-(4-methanesulfonylphenyl)-3-phenylpropyl]-piperidinyl phenylacetamides.

SAR and PK studies led to the identification of N-(1-{(3R)-3-(3,5-difluorophenyl)-3-[4-methanesulfonylphenyl] propyl}piperidin-4-yl)-N-ethyl-2-[4-methanesulfonylphenyl]acetamide as a highly potent and selective ligand for the human CCR5 chemokine receptor with good oral pharmacokinetic properties.

Modulators of the human CCR5 receptor. Part 2: SAR of substituted 1-(3,3-diphenylpropyl)-piperidinyl phenylacetamides.

SAR and DMPK studies led to the identification of substituted N-alkyl-N-[1-(3,3-diphenylpropyl)piperidin-4-yl]-2-phenylacetamides as potent and orally bioavailable ligands for the human CCR5 chemokine receptor.

Stereocontrolled assembly of tetrasubstituted tetrahydrofurans: a concise synthesis of virgatusin.

The condensation of substituted allylsiloxanes with aldehydes leads to the highly stereoselective construction of 2,3,4,5-tetrasubstituted tetrahydrofurans. With electron-rich aryl and alpha,beta-unsaturated aldehydes as substrates, the stereochemical outcome at C5 can be dictated by appropriate choice of Lewis acid. The reaction has been applied to a concise (nine step) synthesis of (+)-virgatusin (ent-1). [reaction: see text]

Modulators of the human CCR5 receptor. Part 1: Discovery and initial SAR of 1-(3,3-diphenylpropyl)-piperidinyl amides and ureas.

Investigation of weak screening hits led to the identification of N-alkyl-N-[1-(3,3-diphenylpropyl)piperidin-4-yl]-2-phenylacetamides and N-alkyl-N-[1-(3,3-diphenylpropyl)piperidin-4-yl]-N'-benzylureas as potent, selective ligands for the human CCR5 chemokine receptor.

Novel, potent and selective anilinoquinazoline and anilinopyrimidine inhibitors of p38 MAP kinase.

SAR studies led to the identification of 4-(3-benzoylamino-6-methyl-anilino)quinazolines as potent and selective inhibitors of p38 MAP kinase. Further optimisation led to the identification of a series of 4-(3-benzoylamino-6-methyl-anilino)pyrimidines as potent inhibitors of the p38 MAP kinase signalling pathway in vitro and in vivo.