Protein-protein interactions regulate most aspects of life cycle thereof being the most attractive and perspective target for contemporary drug development. This field is still not well explored and there are no common rules for proteins involved in this type of interaction. Such intricate biological systems cannot be cost-efficiently tackled using conventional high-throughput screening methods and algorithms. However, fragment-based approach showed fruitful results in search of new PPI inhibitors. We created new library of PPI fragments with dedicated design that consists of systematical knowledge of common PPI inhibitors and selection by privileged structural motifs.

Using our PPI Fragments for hit finding you receive multiple benefits allowing you to save on time and costs in lead generation. Each fragment from the library can be easily followed with available in stock analogues or synthesis of new derivatives through our REAL Database technology.

Key features

Recent researches in the field of PPI fragments revealed the requirement of higher molecular weight molecules compared to common fragments due to a larger contact surface area. Therefore “Rule of four” restriction were used to extract initial PPI fragments subset, as they are typically larger and more lipophilic. Additionally, machine-learning method (decision tree) with number of validated descriptors was used to refine the library.

The “hot-spots” concept has been applied, implying the use of “key” amino acid residues involved in PPIs. A significant number of the selected fragments contains groups/moieties which correspond to these hot-spots. Additionally, library was enriched with the molecules having alpha helix-like structure, able to mimic protein motifs. Also since hydrogen bonds often play a crucial role in PPIs the preference was set for the molecules bearing at least one H-bond donor (>70%) and one H-bond acceptor (100%).

Examples of the molecules

Z2010009926

Z1728138994

Z1446845287

Z1436203819

Z1892408066

Z1634931513

Z991146812

Z1868463771

Indoleamine 2,3-dioxygenase (IDO) is a heme-containing protein that catalyzes the oxidative cleavage of the C₂−C₃ double bond of the indole in tryptophan to provide N-formylkynurenine.

Inhibition of this pathway is perspective and very attractive for the treatment of many neurological disorders, such as Alzheimer’s disease, Parkinson’s disease and cerebral ischemia.

In addition overexpression of IDO1 has been reported in many tumor cells that lead to consideration of IDO1 as one of the key factors that contributes to cancer immunosuppression in tumor microenvironment.

Library design

According to the literature data, the key interaction of IDO inhibitors with the biological target is coordination to Fe ion of the heme in the active site. It is this interaction which prevents participation of the Fe ion in indole amine oxidation.

Significant number of the known IDO inhibitors is hydrophilic compounds that have relatively low molecular weight. Therefore, strict PhysChem parameter filters (lead-likeness, Veber rules) were applied for initial compound set selection. We have developed a holistic approach to the library design that includes ligand-/receptor-based virtual screening and enrichment with heme-binding fragments along with new chemotypes selection.

• Ligand-based approach: Compounds with similar scaffolds, chemotypes and key moieties were picked out from our stock collection.
• Docking: Virtual screening was carried out with heme coordination constraint and other important structural features that are specific for vast majority of potent ligands. The ability of ligands to fill the ambient sub-pockets was set as preferable.
• Selection by Substructure Search: Critical search of molecules bearing fragments reported to form coordination bonds with heme. Compounds with unique chemotypes and appropriate shapes were included to the library.
• Molecular properties: 100% Ro5 compliant, and 99.5 % lead-like.

Examples of the molecules in the library

Chelating fragments included in the library and examples of utilized reference compounds

Protein folding is very important as attractive field in new drug development paradigm. Control of these processes in the cell achieves through action of assembly of enzymes known as molecular chaperones. This set of diverse protein families assists a large variety of processes involving folding, translocation, unfolding, disaggregation and homeostasis of proteins within the cellular environment. In spite of intensive studies, the spectrum of cellular substrates and functions mediated by these different chaperones remains largely undefined. Meanwhile, targeting molecular chaperones is proven to be crucial for the prevention of the many deleterious effects of protein misfolding and aggregation, which might lead in the end in cell death, in neurodegeneration and in other protein misfolding diseases.

Enamine has been following investigation on molecular chaperones for years and developed a library of 2 468 synthetic compounds potentially targeted molecular chaperones. The emphasis has been made on most promising and studied targets: Heat shock proteins (Hsp90, Hsp82, Hsp27), Chaperone activity of bc1 complex-like and histone-chaperone ASF1a complex (ASF1-histone interaction). In addition, the library was enriched with bioisosteric replacement and compounds bearing new chemotypes predominantly with polar scaffolds and cores to increase novelty of the targeted sets.

The library is available in convenient and ready to ship pre-plated formats. Also, the library can be delivered in any other custom formats within a week only.

Library design

• Ligand based approach: pharmacophore searches, bioisosteric replacement, 3D shape-based screening.
• Protein/target-based: Docking, in silico screening with protein-ligand key interaction feature constraints.
• Strict MedChem filters including PAINS and other industry affiliated structural filters and rules.

The sets of the reported active molecules, collected for the targets mentioned above, were carefully analyzed. Series of 3D pharmacophore models within volume restriction constraints were created and further validated with the set of reference actives and non-active ligands (Figure 1). Enamine’s MedChem stock compound collection (Ro5 compliant and filtered through series of MedChem filters) was then screened against these models. The results were inspected visually, and molecules derived from trivial chemotypes, as well as those poorly matching the pharmacophore models, were removed. The bioisosteric replacement of the core structures and selection of compounds by privileged motifs were also used to enrich the library with new valuable structures and prospective drug/lead-like compounds.

Figure 1. Examples of 3D pharmacophore models. In the cases when protein 3D structure was known superposition of ligand and protein key features was used to create pharmacophore model.

Examples of the molecules in the library

Z1170024658

Z1182663584

Z1373679952

Z94608343

Z2193819144

Z1661486482

Z1883306029

Z2043588608

Bioisosteric replacements of the adenosine core of known highly active molecular chaperones ligands

Z319035652

Z1262404427

Z1083983216

Z1037335650

Z1037335452

Z1070385388

Z108193020

Z1051102912

Tubulin targeted library was designed using a combination of different approaches, including molecular docking, substructure and similarity, topological analogs search, molecular parameters restrictions, and specially developed structural filters.

Tubulin library is available in pre-plated format and can be quickly delivered in any customized format.

Library design

Protein structures recorded recently in PDB were considered and analyzed to be most suitable for in silico screening: 4YJ2, 5C8Y, 5CA1. The main features of protein-ligand interaction in the binding site of all three tubulin structures are very similar and could be superimposed. Hence, the protein docking model was built based on the features of critical amino acid residues in the binding pocket and ligand interactions observed in analyzed protein structures.

The docking models have been validated with a reference set of known actives (110 ligands) and nonactive molecules. Molecular queries and docking constraints were corrected according to the activity of the reference compound set.

Fig. 1. Example of hit binding confirmation after docking calculation

Fig. 2. 2D ligand interaction diagram of hit compound Z666232790 with a high docking score

2D Similarity search & topological analogues search

• The reference compound set was carefully compiled from available literature sources and databases: ChEMBL, BindingDB, and PudChem.
• A Tanimoto similarity range of 95–80% was used for compound selection.
• Topological fields and bioisosteric group replacement were used to search for analogs of the most potent tubulin inhibitors.

Examples of active molecules

Approximately 900 compounds were selected using this approach.

Examples of molecules in Tubulin Targeted Library

Filters applied to the Library

• In-house developed MedChem filters of unwanted structural fragments were applied to Enamine stock compound Collection as a pre-selection procedure.
• PAINS, Eli Lilly, REOS, and trivial functionalities filters included.
• Full Rule of Five compliance.