04 Jun Mass-MetaSite and WebMetabase: Tools for the identification and prediction of protease cleavage sites in peptide drugs

Tatiana Radchenko(1,2), Christopher Kochansky(3), Alison Bateman(3),)Andreas Brink(4) Fabien Fontaine(1), Luca Morettoni,(5) Ismael Zamora(1,2)
1. Pompeu Fabra University, Dr, Aiguader 80, Barcelona Spain; 2.Lead Molecular Design, S.L, Vallés 96-102 (L27), Sant Cugat del Vallés, Spain; 3.Merck & Co, 770 Sumneytown Pike, West Point, PA, 19486-0004, USA; 4. Pharmaceutical Sciences, Pharma Research and Early Development, Roche Innovation Center Basel F. Hoffmann-La Roche Ltd., Basel, Switzerland; 5.Molecular Discovery Ltd, London, UK

Introduction The identification of potential cleavage sites in peptide drugs, of metabolite structures and of the proteases involved is important to optimize pharmacokinetic properties of peptide drugs. Several sequence-based bioinformatics tools such as PeptideCutter can be used to identify the sites of cleavage. But their predictions are limited to the 20 natural amino acids and have difficulties in the case of cyclic peptides. Here we present a new approach to analyze experimental high resolution mass spectrometry data to identify in vitro peptide metabolites and determine their structures. Based on this data an in silico analysis tool was developed to compute susceptibility of the peptide bond for cleavage by using a frequency analysis to finally predict potential cleavage sites of peptides. Methods Experimental data: Incubations were performed using selected proteases (serine, aspartic and matrix-metalloproteases). Samples were analyzed using a Q-Exactive instrument in data dependent mode. All data acquired from the LC/MS system were processed using Mass-Metasite and WebMetabase to detect and display the peptide related chromatographic peaks together with the structural elucidation. Workflow: Uploaded information about peptide cleavage data from MEROPS database for the seven human proteases in WebMetabase and calculated frequency for each case. We predicted cleavage sites for 10 investigated commercial peptides (goserelin, leuprolide, deslorelin, gonadorelin, buserelin, histrelin, GLP1, taspoglutide, exenatide, liraglutide). We processed experimental data of incubations into WebMetabase and checked “real” cleavage sites. Preliminary Data HRMS data were collected for two sets of peptides. First, synthetic analogues for the luteinizing-hormone releasing hormone and gonadorelin itself (LHRH) were incubated with trypsin, chymotrypsin, elastase and pepsin. Second Glucagon-Like Peptide-1 and three of its analogs (taspoglutide, liraglutide, exenatide) were incubated with NEP, DPP4 and MMP9. For all LHRH compounds the cleavage site was identified in the site where Trp was positioned in P1 for chymotrypsin and trypsin. Selectivity was identified for pancreatic elastase where Ser was positioned in P1 and trypsin where Leu was positioned in P1 and Ser in prime P1. For the GLP1 cleavage site was identified where Ala was positioned in P1 and Glu in P1 prime. Using the MEROPS public data the frequency of the broken bonds for each of the protease analyzed was obtained. The frequency analysis obtained is in agreement with the reported one in all the cases analyzed (trypsin, chymotrypsin, elastase, pepsin, NEP, DPP4 and MMP9). For example in the case of chymotrypsin where Trp amino acid is involved in most of the cleavages for the proteases investigated here. This result is in agreement with the findings that the Trp-Ser bond is the most likely to be broken by trypsin, chymotrypsin, elastase and pepsin. These results indicate that the shown approach could be useful to elucidate susceptible cleavage sites in peptide drugs with the big advantage not being limited to natural amino acids. Novel Aspect A new approach to investigate peptide metabolism and to in silico predict cleavage sites based on the predicted metabolites structures.