Fabien Fontaine

Enhanced metabolite identification with MSE and a semi-automated software for structural elucidation

The identification of metabolites is almost exclusively done with liquid chromatography/tandem mass spectrometry (LC/MSMS) and despite the enormous progress in the development of these techniques and software for handling of data this is a time-consuming task. In this study the use of quadrupole time-of-flight (QTOF)-generated MS(E) and MS/MS data were compared with respect to rationalization of metabolites. In addition Mass-MetaSite, a semi-automated software for metabolite identification, was evaluated. The program combines the information from MS raw data, in the form of collision-induced dissociation spectra, with a prediction of the site of metabolism in order to assign the structure of a metabolite. The aim of the software is to mimic the rationalization of fragment ions performed by a biotransformation scientist in the process of structural elucidation. For this evaluation, metabolite identification in human liver microsomes was accomplished for 19 commercially available compounds and 15 in-house compounds. The results were very encouraging and for 96% of the metabolites the same structures were assigned using MS(E) compared with MSMS acquired data. The possibility of using MS(E) could considerably reduce the analysis time. Moreover, Mass-MetaSite performed well and the correct assigned structure, compared to manual inspection of the data, was picked in the first rank in ∼80% of the cases. In conclusion MS(E) could be successfully used for metabolite identification in order to reduce time of analysis and Mass-MetaSite could alleviate the work of a biotransformation scientist and decrease the workload by assigning the structure for a majority of the metabolites.

High-throughput, computer assisted, specific MetID. A revolution for drug discovery

One of the key factors in drug discovery is related to the metabolic properties of the lead compound, which may influence the bioavailability of the drug, its therapeutic window, and unwanted side-effects of its metabolites. Therefore, it is of critical importance to enable the fast translation of the experimentally determined metabolic information into design knowledge. The elucidation of the metabolite structure is the most structurally rich and informative end-point in the available range of metabolic assays. A methodology is presented to partially automate the analysis of this experimental information, making the process more efficient. The computer assisted method helps in the chromatographic peak selection and the metabolite structure assignment, enabling automatic data comparison for qualitative applications (kinetic analysis, cross species comparison).

Post-acquisition analysis of untargeted accurate mass quadrupole time-of-flight MSE data for multiple collision-induced neutral losses and fragment ions of glutathione conjugates

RATIONALE Analytical methods to assess glutathione (GSH) conjugate formation based on mass spectrometry usually take advantage of the specific fragmentation behavior of the glutathione moiety. However, most methods used for GSH adduct screening monitor only one specific neutral loss or one fragment ion, even though the peptide moiety of GSH adducts shows a number of other specific neutral fragments and fragment ions which can be used for identification. METHODS Nine reference drugs well known to form GSH adducts were incubated with human liver microsomes. Mass spectrometric analysis was performed with a quadrupole time-of-flight mass spectrometer in untargeted accurate mass MS(E) mode. The data analysis and evaluation was achieved in an automated approach with software to extract and identify GSH conjugates based on the presence of multiple collision-induced neutral losses and fragment ions specific for glutathione conjugates in the high-energy MS spectra. RESULTS In total 42 GSH adducts were identified. Eight (18%) adducts did not show the neutral loss of 129 but were identified based on the appearance of other GSH-specific neutral losses or fragment ions. In high-energy MS(E) spectra the GSH-specific fragment ions of m/z 308 and 179 as well as the neutral loss of 275 Da were complementary to the commonly used neutral loss of 129 Da. Further, one abundant (yet unpublished) GSH conjugate of troglitazone formed in human liver microsomes was found. CONCLUSIONS A software-aided approach was developed to reliably retrieve GSH adduct formation data out of untargeted complex full scan QTOFMS(E) data in a fast and efficient way. The present approach to detect and analyze multiple collision-induced neutral losses and fragment ions of glutathione conjugates in untargeted MS(E) data might be applicable to higher throughput to assess reactive metabolite formation in drug discovery.

SHOP: A Method For Structure-Based Fragment and Scaffold Hopping

We present a method for fragment/scaffold substitution based on protein–ligand interactions. This concept goes beyond bioisosteric replacement, which only uses the structure of the fragment to replace as query. The methodology is validated with more than 10 biological targets relevant for drug discovery.

Comparison of biomolecules on the basis of Molecular Interaction Potentials

Molecular Interaction Potentials (MIP) are frequently used for the comparison of series of compounds displaying related biological behaviors. These potentials are interaction energies between the considered compounds and relevant probes. The interaction energies are computed in the nodes of grids defined around the compounds. There is a need of detailed and objective comparative analyses of MIP distributions in the framework of structure-activity studies. On the other hand, MIP-based studies do not have to be restricted to series of small ligands, since such studies present also interesting possibilities for the analysis and comparison of biological macromolecules. Such analyses can benefit from the application of new methods and computational approaches. The new software MIPSim (Molecular Interaction Potentials Similarity analysis) has recently been introduced with the purpose of analyzing and comparing MIP distributions of series of biomolecules. This program is transparently integrated with other programs, like GAMESS or GRID, which can be used for the computation of the potentials to be analyzed or compared. MIPSim incorporates several definitions of similarity coefficients, and is capable of combining several similarity measures into a single one. On the other hand, MIPSim can perform automatic explorations of the maximum similarity alignments between pairs of molecules.

Software-aided cytochrome P450 reaction phenotyping and kinetic analysis in early drug discovery.

RATIONALE Cytochrome P450 (CYP450) reaction phenotyping (CRP) and kinetic studies are essential in early drug discovery to determine which metabolic enzymes react with new drug entities. A new semi-automated computer-assisted workflow for CRP is introduced in this work. This workflow provides not only information regarding parent disappearance, but also metabolite identification and relative metabolite formation rates for kinetic analysis. METHODS Time-course experiments based on incubating six probe substrates (dextromethorphan, imipramine, buspirone, midazolam, ethoxyresorufin and diclofenac) with recombinant human enzymes (CYP1A2, CYP2C9, CYP2C19, CYP2D6 and CYP3A4) and human liver microsomes (HLM) were performed. Liquid chromatography/high-resolution mass spectrometry (LC/HRMS) analysis was conducted with an internal standard to obtain high-resolution full-scan and MS/MS data. Data were analyzed using Mass-MetaSite software. A server application (WebMetabase) was used for data visualization and review. RESULTS CRP experiments were performed, and the data were analyzed using a software-aided approach. This automated-evaluation approach led to (1) the detection of the CYP450 enzymes responsible for both substrate depletion and metabolite formation, (2) the identification of specific biotransformations, (3) the elucidation of metabolite structures based on MS/MS fragment analysis, and (4) the determination of the initial relative formation rates of major metabolites by CYP450 enzymes. CONCLUSIONS This largely automated workflow enabled the efficient analysis of HRMS data, allowing rapid evaluation of the involvement of the main CYP450 enzymes in the metabolism of new molecules during drug discovery.

Conformationally constrained butyrophenones as new pharmacological tools to study 5-HT2A and 5-HT2C receptor behaviours

This study presents new pharmacological and molecular modelling studies on a recently described series of conformationally constrained butyrophenones. Alignment-free three-dimensional quantitative structure-activity relationship models developed on the basis of GRid Independent descriptors and partial least squares regression analysis, allow feasible predictions of activity of new compounds and reveal structural requirements for optimal affinity, particularly in the case of the 5-HT(2A) receptor. The requirements for the 5-HT(2A) affinity consist in a precise distance between hydrogen bond donor (protonated amino group) and hydrogen bond acceptor groups, as well as an optimal distance between the protonated amino group and the farthest extreme of the compounds. Another significant result has been the characterisation of two structurally similar compounds as interesting pharmacological tools (1-[(4-Oxo-4,5,6,7-tetrahydrobenzo[b]furan-5-yl)ethyl]-4-(6-fluorobenzisoxazol-3-yl)piperidine and 1-[(4-Oxo-4,5,6,7-tetrahydrobenzo[b]furan-6-yl)methyl]-4-(6-fluorobenzisoxazol-3-yl)piperidine). In spite of their structural similarity, the first compound shows clearly higher affinity for the 5-HT(2C) receptor (about 100 fold) and higher Meltzer ratio (1.17 vs. 0.99) than the second. Moreover, the first compound inhibits arachidonic acid release in a biphasic concentration-dependent way in functional experiments at the 5-HT(2A) receptor and it acts as inverse agonist at the 5-HT(2C) receptor, behaviours that are not shown by the second compound.

Use of alignment-free molecular descriptors in diversity analysis and optimal sampling of molecular libraries

The selection of a sample of diverse compounds is a common strategy for exploring large molecular libraries. However, the success of such approach depends on the selection of relevant molecular descriptors and the use of appropriate sampling methods. In the context of pharmaceutical research, the molecular descriptors should be based on physicochemical properties related with the pharmacological behaviour of the compounds. In this sense, the alignment-free GRIND and VolSurf molecular descriptors are promising candidates since they have been successfully used in the modelling of both pharmacodynamic and pharmacokinetic properties of drugs. This work describes the use of such descriptors in the diversity sampling of a library of primary amines and compares the results with those obtained in a previous study that used quantum-mechanical descriptors. As in the previous work, principal component (PC) analysis was applied to reduce the dimensionality and remove redundant information of the original descriptors, and the compounds were sampled on the basis of k-means clustering on the space of the selected PCs. The results of the present study show that VolSurf and GRIND provide similar quality sampling regarding global features of the molecules such as hydrophilicity, however the topology of the compounds is considered differently. The similarity between particular compounds strongly depends on the original descriptors used. However all the sample selections done in the PC space after k-means clustering provide the same apparent diversity in comparison to the whole dataset. The results indicate that there is no best set of descriptors on a diversity basis. The selection of descriptors must be based on the drug features to be investigated.

Correction: Software-aided approach to investigate peptide structure and metabolic susceptibility of amide bonds in peptide drugs based on high resolution mass spectrometry

[This corrects the article DOI: 10.1371/journal.pone.0186461.].

Software-aided workflow for predicting protease-specific cleavage sites using physicochemical properties of the natural and unnatural amino acids in peptide-based drug discovery.

Peptide drugs have been used in the treatment of multiple pathologies. During peptide discovery, it is crucially important to be able to map the potential sites of cleavages of the proteases. This knowledge is used to later chemically modify the peptide drug to adapt it for the therapeutic use, making peptide stable against individual proteases or in complex medias. In some other cases it needed to make it specifically unstable for some proteases, as peptides could be used as a system to target delivery drugs on specific tissues or cells. The information about proteases, their sites of cleavages and substrates are widely spread across publications and collected in databases such as MEROPS. Therefore, it is possible to develop models to improve the understanding of the potential peptide drug proteolysis. We propose a new workflow to derive protease specificity rules and predict the potential scissile bonds in peptides for individual proteases. WebMetabase stores the information from experimental or external sources in a chemically aware database where each peptide and site of cleavage is represented as a sequence of structural blocks connected by amide bonds and characterized by its physicochemical properties described by Volsurf descriptors. Thus, this methodology could be applied in the case of non-standard amino acid. A frequency analysis can be performed in WebMetabase to discover the most frequent cleavage sites. These results were used to train several models using logistic regression, support vector machine and ensemble tree classifiers to map cleavage sites for several human proteases from four different families (serine, cysteine, aspartic and matrix metalloproteases). Finally, we compared the predictive performance of the developed models with other available public tools PROSPERous and SitePrediction.