CAVIAR: a method for automatic cavity detection, description and decomposition into subcavities.

The accurate description of protein binding sites is essential to the determination of similarity and the application of machine learning methods to relate the binding sites to observed functions. This work describes CAVIAR, a new open source tool for generating descriptors for binding sites, using protein structures in PDB and mmCIF format as well as trajectory frames from molecular dynamics simulations as input. The applicability of CAVIAR descriptors is showcased by computing machine learning predictions of binding site ligandability. The method can also automatically assign subcavities, even in the absence of a bound ligand. The defined subpockets mimic the empirical definitions used in medicinal chemistry projects. It is shown that the experimental binding affinity scales relatively well with the number of subcavities filled by the ligand, with compounds binding to more than three subcavities having nanomolar or better affinities to the target. The CAVIAR descriptors and methods can be used in any machine learning-based investigations of problems involving binding sites, from protein engineering to hit identification. The full software code is available on GitHub and a conda package is hosted on Anaconda cloud.

An ABSINTH-Based Protocol for Predicting Binding Affinities between Proteins and Small Molecules.

The core task in computational drug discovery is to accurately predict binding free energies in receptor-ligand systems for large libraries of putative binders. Here, the ABSINTH implicit solvent model and force field are extended to describe small, organic molecules and their interactions with proteins. We show that an automatic pipeline based on partitioning arbitrary molecules into substructures corresponding to model compounds with known free energies of solvation can be combined with the CHARMM general force field into a method that is successful at the two important challenges a scoring function faces in virtual screening work flows: it ranks known binders with correlation values rivaling that of comparable state-of-the-art methods and it enriches true binders in a set of decoys. Our protocol introduces innovative modifications to common virtual screening workflows, notably the use of explicit ions as competitors and the integration over multiple protein and ligand species differing in their protonation states. We demonstrate the value of modifications to both the protocol and ABSINTH itself. We conclude by discussing the limitations of high-throughput implicit methods such as the one proposed here.

Virtual screen to NMR (VS2NMR): Discovery of fragment hits for the CBP bromodomain.

Overexpression of the CREB-binding protein (CBP), a bromodomain-containing transcription coactivator involved in a variety of cellular processes, has been observed in several types of cancer with a correlation to aggressiveness. We have screened a library of nearly 1500 fragments by high-throughput docking into the CBP bromodomain followed by binding energy evaluation using a force field with electrostatic solvation. Twenty of the 39 fragments selected by virtual screening are positive in one or more ligand-observed nuclear magnetic resonance (NMR) experiments. Four crystal structures of the CBP bromodomain in complex with in silico screening hits validate the pose predicted by docking. Thus, the success ratio of the high-throughput docking procedure is 50% or 10% if one considers the validation by ligand-observed NMR spectroscopy or X-ray crystallography, respectively. Compounds 1 and 3 show favorable ligand efficiency in two different in vitro binding assays. The structure of the CBP bromodomain in the complex with the brominated pyrrole 1 suggests fragment growing by Suzuki coupling.

Discovery of Inhibitors of Four Bromodomains by Fragment-Anchored Ligand Docking.

The high-throughput docking protocol called ALTA-VS (anchor-based library tailoring approach for virtual screening) was developed in 2005 for the efficient in silico screening of large libraries of compounds by preselection of only those molecules that have optimal fragments (anchors) for the protein target. Here we present an updated version of ALTA-VS with a broader range of potential applications. The evaluation of binding energy makes use of a classical force field with implicit solvent in the continuum dielectric approximation. In about 2 days per protein target on a 96-core compute cluster (equipped with Xeon E3-1280 quad core processors at 2.5 GHz), the screening of a library of nearly 77 000 diverse molecules with the updated ALTA-VS protocol has resulted in the identification of 19, 3, 3, and 2 µM inhibitors of the human bromodomains ATAD2, BAZ2B, BRD4(1), and CREBBP, respectively. The success ratio (i.e., number of actives in a competition binding assay in vitro divided by the number of compounds tested) ranges from 8% to 13% in dose-response measurements. The poses predicted by fragment-based docking for the three ligands of the BAZ2B bromodomain were confirmed by protein X-ray crystallography.

Investigating the allosteric reverse signalling of PARP inhibitors with microsecond molecular dynamic simulations and fluorescence anisotropy.

The inhibition of the poly(ADP-ribose) polymerase (PARP) family members is a strategy pursued for the development of novel therapeutic agents in a range of diseases, including stroke, cardiac ischemia, cancer, inflammation and diabetes. Even though some PARP-1 inhibitors have advanced to clinical setting for cancer therapy, a great deal of attention is being devoted to understand the polypharmacology of current PARP inhibitors. Besides blocking the catalytic activity, recent works have shown that some PARP inhibitors exhibit a poisoning activity, by trapping the enzyme at damaged sites of DNA and forming cytotoxic complexes. In this study we have used microsecond molecular dynamics to study the allosteric reverse signalling that is at the basis of such an effect. We show that Olaparib, but not Veliparib and HYDAMTIQ, is able to induce a specific conformational drift of the WGR domain of PARP-1, which stabilizes PARP-1/DNA complex through the locking of several salt bridge interactions. Fluorescence anisotropy assays support such a mechanism, providing the first experimental evidence that HYDAMTIQ, a potent PARP inhibitor with neuroprotective properties, is less potent than Olaparib to trap PARP-1/DNA complex.

Identification of a BAZ2A-Bromodomain Hit Compound by Fragment Growing.

BAZ2A is an epigenetic regulator affecting transcription of ribosomal RNA. It is overexpressed in aggressive and recurrent prostate cancer, promoting cellular migration. Its bromodomain is characterized by a shallow and difficult-to-drug pocket. Here, we describe a structure-based fragment-growing campaign for the identification of ligands of the BAZ2A bromodomain. By combining docking, competition binding assays, and protein crystallography, we have extensively explored the interactions of the ligands with the rim of the binding pocket, and in particular ionic interactions with the side chain of Glu1820, which is unique to BAZ2A. We present 23 high-resolution crystal structures of the holo BAZ2A bromodomain and analyze common bromodomain/ligand motifs and favorable intraligand interactions. Binding of some of the compounds is enantiospecific, with affinity in the low micromolar range. The most potent ligand has an equilibrium dissociation constant of 7 µM and a good selectivity over the paralog BAZ2B bromodomain.

In silico fragment-based drug design with SEED.

We report on two fragment-based drug design protocols, SEED2XR and ALTA, which start by high-throughput docking. SEED2XR is a two-stage protocol for fragment-based drug design. The first stage is in silico and consists of the automatic docking of 103-104 fragments using SEED, which requires about 1 s per fragment. SEED is a docking software developed specifically for fragment docking and binding energy evaluation by a force field with implicit solvent. In the second stage of SEED2XR, the 10-102 fragments with the most favorable predicted binding energies are validated by protein X-ray crystallography. The recent applications of SEED2XR to bromodomains demonstrate that the whole SEED2XR protocol can be carried out in about a week of working time, with hit rates ranging from 10% to 40%. Information on fragment-target interactions generated by the SEED2XR protocol or directly from SEED docking has been used for the discovery of hundreds of hits. ALTA is a computational protocol for screening which identifies candidate ligands that preserve the interactions between the optimal SEED fragments and the protein target. Medicinal chemistry optimization of ligands predicted by ALTA has resulted in pre-clinical candidates for protein kinases and bromodomains. The high-throughput, very low cost, sustainability, and high hit rate of the SEED-based protocols, unreachable by purely experimental techniques, make them perfectly suitable for both academic and industrial drug discovery research.

Structural Analysis of Small-Molecule Binding to the BAZ2A and BAZ2B Bromodomains.

The bromodomain-containing protein BAZ2A is a validated target in prostate cancer research, whereas the function of its paralogue BAZ2B is still undefined. The bromodomains of BAZ2A and BAZ2B have a similar binding site for their natural ligand, the acetylated lysine side chain. Here, we present an analysis of the binding modes of eight compounds belonging to three distinct chemical classes. For all compounds, the moiety mimicking the natural ligand engages in essentially identical interactions in the BAZ2A and BAZ2B bromodomains. In contrast, the rest of the molecule is partially solvent-exposed and adopts different orientations with different interactions in the two bromodomains. Some of these differences could be exploited for designing inhibitors with selectivity within the BAZ2 bromodomain subfamily.

Derivatives of 3-Amino-2-methylpyridine as BAZ2B Bromodomain Ligands: In Silico Discovery and in Crystallo Validation.

The 3-amino-2-methylpyridine derivative 1 was identified as ligand of the BAZ2B bromodomain by automatic docking of nearly 500 compounds, selected on the basis of previous fragment hits. Hit expansion by two in silico approaches, pharmacophore search followed by docking, and substructure search resulted in five additional ligands. The predicted binding mode of the six 3-amino-2-methylpyridine derivatives was validated by protein crystallography. A small displacement of residues 1894-1899 of the ZA loop is observed for two of the six ligands. In all structures, the pyridine head is involved in a water-mediated hydrogen bond with the side chain of the conserved Tyr1901 while the 3-amino linker acts as hydrogen bond donor for the backbone carbonyl of Pro1888. Heterogeneous orientations are observed for the tail groups (i.e., the 3-amino substituents). The sulfonyl group in the tail of compounds 1 and 2 is involved in a hydrogen bond with the backbone amide of Asn1894.

A Small-Molecule Inhibitor of Lin28.

New discoveries in RNA biology underscore a need for chemical tools to clarify their roles in pathophysiological mechanisms. In certain cancers, synthesis of the let-7 microRNA tumor suppressor is blocked by an RNA binding protein (RBP) Lin28, which docks onto a conserved sequence in let-7 precursor RNA molecules and prevents their maturation. Thus, the Lin28/let-7 interaction might be an attractive drug target, if not for the well-known difficulty in targeting RNA-protein interactions with drugs. Here, we describe a protein/RNA FRET assay using a GFP-Lin28 donor and a black-hole quencher (BHQ)-labeled let-7 acceptor, a fluorescent protein/quencher combination which is rarely used in screening despite favorable spectral properties. We tested 16 000 molecules and identified N-methyl-N-[3-(3-methyl[1,2,4]triazolo[4,3-b]pyridazin-6-yl)phenyl]acetamide, which blocked the Lin28/let-7 interaction, rescued let-7 processing and function in Lin28-expressing cancer cells, induced differentiation of mouse embryonic stem cells, and reduced tumor-sphere formation by 22Rv1 and Huh7 cells. A biotinylated derivative captured Lin28 from cell lysates consistent with an on-target mechanism in cells, though the compound also showed some activity against bromodomains in selectivity assays. The Lin28/let-7 axis is presently of high interest not only for its role as a bistable switch in stem-cell biology but also because of its prominent roles in numerous diseases. We anticipate that much can be learned from the use of this first reported small molecule antagonist of Lin28, including the potential of the Lin28/let-7 interaction as a new drug target for selected cancers. Furthermore, this approach to assay development may be used to identify antagonists of other RBP/RNA interactions suspected to be operative in pathophysiological mechanisms.

Binding Mode of Acetylated Histones to Bromodomains: Variations on a Common Motif.

Bromodomains, epigenetic readers that recognize acetylated lysine residues in histone tails, are potential drug targets in cancer and inflammation. Herein we review the crystal structures of human bromodomains in complex with histone tails and analyze the main interaction motifs. The histone backbone is extended and occupies, in one of the two possible orientations, the bromodomain surface groove lined by the ZA and BC loops. The acetyl-lysine side chain is buried in the cavity between the four helices of the bromodomain, and its oxygen atom accepts hydrogen bonds from a structural water molecule and a conserved asparagine residue in the BC loop. In stark contrast to this common binding motif, a large variety of ancillary interactions emerge from our analysis. In 10 of 26 structures, a basic side chain (up to five residues up- or downstream in sequence with respect to the acetyl-lysine) interacts with the carbonyl groups of the C-terminal turn of helix αB. Furthermore, the complexes reveal many heterogeneous backbone hydrogen bonds (direct or water-bridged). These interactions contribute unselectively to the binding of acetylated histone tails to bromodomains, which provides further evidence that specific recognition is modulated by combinations of multiple histone modifications and multiple modules of the proteins involved in transcription.