19F NMR chemical shift prediction with fluorine fingerprint descriptor

QM assisted ML for 19F NMR chemical shift prediction

BACKGROUND

Ligand-observed 19F NMR detection is an efficient method for screening libraries of fluorinated molecules in fragment-based drug design campaigns. Screening fluorinated molecules in large mixtures makes 19F NMR a high-throughput method. Typically, these mixtures are generated from pools of well-characterized fragments. By predicting 19F NMR chemical shift, mixtures could be generated for arbitrary fluorinated molecules facilitating for example focused screens.

METHODS

In a previous publication, we introduced a method to predict 19F NMR chemical shift using rooted fluorine fingerprints and machine learning (ML) methods. Having observed that the quality of the prediction depends on similarity to the training set, we here propose to assist the prediction with quantum mechanics (QM) based methods in cases where compounds are not well covered by a training set.

RESULTS

Beyond similarity, the performance of ML methods could be associated with individual features in compounds. A combination of both could be used as a procedure to split input data sets into those that could be predicted by ML and those that required QM processing. We could show on a proprietary fluorinated fragment library, known as LEF (Local Environment of Fluorine), and a public Enamine data set of 19F NMR chemical shifts that ML and QM methods could synergize to outperform either method individually. Models built on Enamine data, as well as model building and QM workflow tools, can be found at https://github.com/PatrickPenner/lefshift and https://github.com/PatrickPenner/lefqm .

Prediction of chemical shift in NMR: A review

Calculation of solution-state NMR parameters, including chemical shift values and scalar coupling constants, is often a crucial step for unambiguous structure assignment. Data-driven (sometimes called empirical) methods leverage databases of known parameter values to estimate parameters for unknown or novel molecules. This is in contrast to popular ab initio techniques that use detailed quantum computational chemistry calculations to arrive at parameter estimates. Data-driven methods have the potential to be considerably faster than ab inito techniques and have been the subject of renewed interest over the past decade with the rise of high-quality databases of NMR parameters and novel machine learning methods. Here, we review these methods, their strengths and pitfalls, and the databases they are built on.

Efficient Screening of Target-Specific Selected Compounds in Mixtures by 19F NMR Binding Assay with Predicted 19F NMR Chemical Shifts

Ligand-based 19 F NMR screening is a highly effective and well-established hit-finding approach. The high sensitivity to protein binding makes it particularly suitable for fragment screening. Different criteria can be considered for generating fluorinated fragment libraries. One common strategy is to assemble a large, diverse, well-designed and characterized fragment library which is screened in mixtures, generated based on experimental 19 F NMR chemical shifts. Here, we introduce a complementary knowledge-based 19 F NMR screening approach, named 19 Focused screening, enabling the efficient screening of putative active molecules selected by computational hit finding methodologies, in mixtures assembled and on-the-fly deconvoluted based on predicted 19 F NMR chemical shifts. In this study, we developed a novel approach, named LEFshift , for 19 F NMR chemical shift prediction using rooted topological fluorine torsion fingerprints in combination with a random forest machine learning method. A demonstration of this approach to a real test case is reported.

Advances in the study of GPCRs by 19F NMR

Crystallography and cryo-electron microscopy have advanced atomic resolution perspectives of inactive and active states of G protein-coupled receptors (GPCRs), alone and in complex with G proteins or arrestin. ¹⁹F NMR can play a role in ascertaining activation mechanisms and understanding the complete energy landscape associated with signal transduction. Fluorinated reporters are introduced biosynthetically via fluorinated amino acid analogs or chemically, via thiol-specific fluorinated reporters. The chemical shift sensitivity of these reporters makes it possible to discern details of conformational ensembles. In addition to spectroscopic details, paramagnetic species can be incorporated through orthogonal techniques to obtain distance information on fluorinated reporters, while T₂-and T₁-based relaxation experiments provide details on exchange kinetics in addition to fluctuations within a given state.

Fluorine NMR functional screening: from purified enzymes to human intact living cells

The substrate- or cofactor-based fluorine NMR screening, also known as n-FABS (n fluorine atoms for biochemical screening), represents a powerful method for performing a direct functional assay in the search of inhibitors or enhancers of an enzymatic reaction. Although it suffers from the intrinsic low sensitivity compared to other biophysical techniques usually applied in functional assays, it has some distinctive features that makes it appealing for tackling complex chemical and biological systems. Its strengths are represented by the easy set-up, robustness, flexibility, lack of signal interference and rich information content resulting in the identification of bona fide inhibitors and reliable determination of their inhibitory strength. The versatility of the n-FABS allows its application to either purified enzymes, cell lysates or intact living cells. The principles, along with theoretical, technical and practical aspects, of the methodology are discussed. Furthermore, several applications of the technique to pharmaceutical projects are presented.

Escherichia coli Topoisomerase IV E Subunit and an Inhibitor Binding Mode Revealed by NMR Spectroscopy

Bacterial topoisomerases are attractive antibacterial drug targets because of their importance in bacterial growth and low homology with other human topoisomerases. Structure-based drug design has been a proven approach of efficiently developing new antibiotics against these targets. Past studies have focused on developing lead compounds against the ATP binding pockets of both DNA gyrase and topoisomerase IV. A detailed understanding of the interactions between ligand and target in a solution state will provide valuable information for further developing drugs against topoisomerase IV targets. Here we describe a detailed characterization of a known potent inhibitor containing a 9H-pyrimido[4,5-b]indole scaffold against the N-terminal domain of the topoisomerase IV E subunit from Escherichia coli (eParE). Using a series of biophysical and biochemical experiments, it has been demonstrated that this inhibitor forms a tight complex with eParE. NMR studies revealed the exact protein residues responsible for inhibitor binding. Through comparative studies of two inhibitors of markedly varied potencies, it is hypothesized that gaining molecular interactions with residues in the α4 and residues close to the loop of β1-α2 and residues in the loop of β3-β4 might improve the inhibitor potency.

¹⁹F NMR fingerprints: identification of neutral organic compounds in a molecular container

Improved methods for quickly identifying neutral organic compounds and differentiation of analytes with similar chemical structures are widely needed. We report a new approach to effectively "fingerprint" neutral organic molecules by using (19)F NMR and molecular containers. The encapsulation of analytes induces characteristic up- or downfield shifts of (19)F resonances that can be used as multidimensional parameters to fingerprint each analyte. The strategy can be achieved either with an array of fluorinated receptors or by incorporating multiple nonequivalent fluorine atoms in a single receptor. Spatial proximity of the analyte to the (19)F is important to induce the most pronounced NMR shifts and is crucial in the differentiation of analytes with similar structures. This new scheme allows for the precise and simultaneous identification of multiple analytes in a complex mixture.

Design and generation of highly diverse fluorinated fragment libraries and their efficient screening with improved (19) F NMR methodology

Fragment screening performed with (19) F NMR spectroscopy is becoming increasingly popular in drug discovery projects. With this approach, libraries of fluorinated fragments are first screened using the direct-mode format of the assay. The choice of fluorinated motifs present in the library is fundamental in order to ensure a large coverage of chemical space and local environment of fluorine (LEF). Mono- and poly-fluorinated fragments to be included in the libraries for screening are selected from both in-house and commercial collections, and those that are ad hoc designed and synthesized. Additional fluorinated motifs to be included in the libraries derive from the fragmentation of compounds in development and launched on the market, and compounds contained in other databases (such as Integrity, PDB and ChEMBL). Complex mixtures of highly diverse fluorine motifs can be rapidly screened and deconvoluted in the same NMR tube with a novel on the fly combined procedure for the identification of the active molecule(s). Issues and problems encountered in the design, generation and screening of diverse fragment libraries of fluorinated compounds with (19) F NMR spectroscopy are analyzed and technical solutions are provided to overcome them. The versatile screening methodology described here can be efficiently applied in laboratories with limited NMR setup and could potentially lead to the increasing role of (19) F NMR in the hit identification and lead optimization phases of drug discovery projects.

Technical and practical aspects of (19) F NMR-based screening: toward sensitive high-throughput screening with rapid deconvolution

The technical and practical aspects of (19) F NMR-based screening against a macromolecular target are analyzed in detail. A novel method utilizing the relaxation of (19) F homonuclear double quantum coherence is proposed for performing NMR-based binding assays in a direct- or competition-mode format. A combined strategy based on (19) F NMR chemical shift prediction, 2D (19) F NMR DOSY, and 2D (19) F-(1) H NMR long-range COSY experiments is presented for the deconvolution of complex mixtures of fluorinated molecules generated by either addition of single compounds or by chemical synthesis. The approaches presented here allow the screening of complex mixtures, even in the case where the exact composition is not known, and the rapid identification of the binders contained in the mixtures.

Intermolecular and intramolecular hydrogen bonds involving fluorine atoms: implications for recognition, selectivity, and chemical properties

A correlation between 19F NMR isotropic chemical shift and close intermolecular F⋅⋅⋅H-X contacts (with X=N or O) has been identified upon analysis of the X-ray crystal structures of fluorinated molecules listed in the Cambridge Structural Database (CSD). An optimal F⋅⋅⋅X distance involving primary and shielded secondary fluorine atoms in hydrogen-bond formation along with a correlation between F⋅⋅⋅H distance and F⋅⋅⋅H-X angle were also derived from the analysis. The hydrogen bonds involving fluorine are relevant, not only for the recognition mechanism and stabilization of a preferred conformation, but also for improvement in the permeability of the molecules, as shown with examples taken from a proprietary database. Results of an analysis of the small number of fluorine-containing natural products listed in the Protein Data Bank (PDB) appear to strengthen the derived correlation between 19F NMR isotropic chemical shift and interactions involving fluorine (also known as the "rule of shielding") and provides a hypothesis for the recognition mechanism and catalytic activity of specific enzymes. Novel chemical scaffolds, based on the rule of shielding, have been designed for recognizing distinct structural motifs present in proteins. It is envisaged that this approach could find useful applications in drug design for the efficient optimization of chemical fragments or promising compounds by increasing potency and selectivity against the desired biomolecular target.

Fluorine-protein interactions and ¹⁹F NMR isotropic chemical shifts: An empirical correlation with implications for drug design

An empirical correlation between the fluorine isotropic chemical shifts, measured by ¹⁹F NMR spectroscopy, and the type of fluorine-protein interactions observed in crystal structures is presented. The CF, CF₂, and CF₃ groups present in fluorinated ligands found in the Protein Data Bank were classified according to their ¹⁹F NMR chemical shifts and their close intermolecular contacts with the protein atoms. Shielded fluorine atoms, i.e., those with increased electron density, are observed primarily in close contact to hydrogen bond donors within the protein structure, suggesting the possibility of intermolecular hydrogen bond formation. Deshielded fluorines are predominantly found in close contact with hydrophobic side chains and with the carbon of carbonyl groups of the protein backbone. Correlation between the ¹⁹F NMR chemical shift and hydrogen bond distance, both derived experimentally and computed through quantum chemical methods, is also presented. The proposed "rule of shielding" provides some insight into and guidelines for the judicious selection of appropriate fluorinated moieties to be inserted into a molecule for making the most favorable interactions with the receptor.