05 Jun A MS approach to understand the most metabolically liable amide bonds in peptides

Ismael Zamora(1,2), Tatiana Radchennko(1,3), Elisabeth Ortega-Carrasco(2)
1. Lead Molecular design, S.L.,Spain; 2. Molecular Discovery, Ltd, UK; 3. Pompeu Fabra University, Barcelona, Spain

Introduction Protein/peptide derived drugs are becoming very relevant as a new therapeutic group for medical care. In all cases, these agents must have the following properties: “specificity, affinity, solubility, stability and safety” and the knowledge of these parameters can help to understand aspects such as bioavailability, main effects and side-effects of the drug and its metabolites. Moreover, there are several in-silico methodologies trat try to predict peptide cleavage sites for different proteases and thus predict metabolic liability of different amide bonds depending on incubation media. The aim of this poster is to present a new methodology to analyse the MS driven experimental determined metabolite structures to udnerstand which is the frequancy of the metabolic liability of the different amide bonds Methods The poster will show a doftware application which could use MSMS information from experiments with incubations in different media (human hepatocite, gastrointestinal fluid etc) to do the prediction of the metabolites structure for the peptide-parent, store information about parents and metabolites structure and its properties in a searchable format in the database. Moreover a frequency analysis of the broken bonds depending on the media is done in order to derive a set of rules to analyse and predict metabolic liability of different amide bonds depending on incubation media. Preliminary Data In order to show concept we collected experimental information about several incubations of three peptides (linear, cyclic and non-natural aminoacids) incubated in human hepatocyte media. The main algorithm at the first stage uses MassMetaSite and WebMetabase to do the metabolites structure prediction. At the next step we stored peptides-parent and metabolites structure in the WebMetabase annotated with monomers (blocks of amino acids between amide bonds) and its connectivity. It was seen that we were able to do the chemical aware search of peptide´s part of any size. Moreover peptides were annotated with molecular descriptors to describe its pharmacophoric and physicochemical properties. It also let us to do similarity search based on these properties. Developed search algorithm was not based on the theoretical mass spectrum or sequence alignment, but rather in performing an exact search of peptides and its metabolites monomers sequences. Moreover a similarity search of similar monomers was performed based on the physicochemical properties and/or pharmacophoric interaction. Using stored information we performed a frequency analysis of the chemical moieties (e.i. amide bonds between two monomers) involved in the metabolic reactions within this database. It was seen that there was 36 moieties, 13 bonds were met 2 times and 2 were met in te database 4 times. Amide bonds between Lysine and Leucine were met 4 time and were broken 4 times. Amide bond between Leucine and Serine was met 4 times and was broken 2 times. Novel Aspect Development of the rule se for structure predictions of the cleavage sites in the peptides based on HRMS data.