The aim of this study was to design and optimize new PCR primers for detection of potential cyanobacterial producers of cytotoxic lipopeptides puwainaphycins and minutissamides in environmental samples. Samples from two distinct localities were tested, as suggested based on preliminary data. The first set of samples consisted of cyanobacterial soil biofilms from sheep pastures affected by Alveld illness in Norway. The other one contained samples of planktic cyanobacaterial blooms from Protected Landscape Area Třeboň and its vicinity. Three different approaches were used for evaluation of the presence of cyanobacterial lipopeptide producers: microscopy, PCR with the designed primeres, and liquid chromatography-mass spectrometry analysis. Results of this study confirmed the specificity of the newly designed PCR primers. The presence of producers of puwainaphycins/minutissamides was proven at both tested localities.

Les peptides vecteurs ou CPP sont de petits peptides, en général de taille inférieure à 30 acides aminés. Parmi les nombreux CPP décrits dans la littérature, les peptides riches en arginine ont fait l'objet d'une attention particulière. Plusieurs modifications chimiques du squelette peptidique conduisant à une distribution spatiale différente des groupes fonctionnels, ou encore l'introduction de chaînes aliphatiques ont été effectuées pour accroitre la capacité du peptide à traverser la membrane de la cellule. L'objectif de ce travail a été le développement de nouveaux peptides vecteurs cycliques contenant un domaine cationique minimal et pouvant être acylés par une chaîne aliphatique. Quinze nouveaux transporteurs cycliques, dont les peptides vecteurs classiques Pénétratine et R6W3ont été synthétisés. La cyclisation tête-queue par ligation chimique native a été rendue possible par l'introduction d'un résidu cystéine et d'une fonction thioester (ou précurseur) respectivement aux extrémités N et C-terminales des différentes séquences de CPP. Leur aptitude à transporter le peptide bioactif PKCi dans des cellules CHO a été évaluée par quantification de la cargaison internalisée en utilisant la spectrométrie de masse MALDI-TOF. Les résultats indiquent une meilleure internalisation essentiellement par voie d'endocytose dépendante des glycosaminoglycanes, suite à la cyclisation des CPP comparés à leur version linéaire. De toute la série des lipopeptides testés dans ce projet, deux séquences se distinguent par leur capacité remarquable à franchir les membranes cellulaire : les CPP [C12-R4] et [C12-R7]. / Cell-penetrating peptides (CPPs) are short, cationic or amphipatic peptides, usually containing less than 30 amino acids, which are able to deliver various bioactive cargoes inside cells. Among the many CPPs described in the literature, the arginine-rich peptides have attracted particular attention. Several chemical modifications of the peptide backbone leading to different spatial distributions of the CPP functional groups, or the introduction of aliphatic chains have been made to enhance their internalization efficiency. The aim of this work was the synthesis of new cyclic CPPs containing a minimal cationic domain and their functionalisation with an aliphatic chain. We have synthesised a small library of fifteen new cyclic carriers including the classical CPPs Penetratin and R6W3 using native chemical ligation (NCL) in solution. The introduction of an N-terminal Cys residue and of a C-terminal thioester (or precursor) in the initial linear peptide sequence allowed the head-to-tail cyclisation. The efficiency of cargo delivery in CHO cells was measured by MALDI-TOF mass spectrometry. We found that cyclisation of CPPs improved their internalisation efficiency mostly by glycosaminoglycan-dependent endocytosis. Among the whole series of lipopeptides tested in this project, two sequences are distinguished by their remarkable ability to cross cellular membranes: the peptides [C12-R4] and [C12-R7].

This thesis deals with the analysis of natural peptide libraries using mass spectrometry. In the course of the study, both ribosomal and non-ribosomal classes of peptides have been investigated. Microheterogeneity, post-translational modifications (PTM), isobaric amino acids and disulfide crosslinks present critical challenges in routine mass spectral structure determination of natural peptides. These problems form the core of this thesis. Chapter 2 describes an approach where chemical derivatization, in unison with high resolution LC-MSn experiments, resulted in deconvolution of a microheterogenous peptide library of B. subtilis K1. Chapter 3 describes an approach for distinction between isobaric amino acids (Leu/Ile/Hyp), by the use of combined ETD-CID fragmentation, through characteristic side chain losses. Chapters 4-6 address a long standing problem in structure elucidation of peptide toxins; the determination of disulfide connectivity. Through the use of direct mass spectral CID fragmentation, a methodology has been proposed for determination of the S-S pairing schemes in polypeptides. Further, an algorithm DisConnect has been developed for a rapid and robust solution to the problem. This general approach is applicable to both peptides and proteins, irrespective of the size and the number of disulfide bonds present. The method has been successfully applied to a large number of peptide toxins from marine cone snails, conotoxins, synthetic foldamers and proteins. Chapter 7 describes an attempt to integrate next generation sequencing (NGS) data with mass spectrometric analysis of the crude venom. This approach couples rapidly generated cDNA sequences, with high-throughput LC-ESI-MS/MS analysis, which provides mass spectral fragmentation information. An algorithm has been developed that allows the construction of a putative conus peptide database from the NGS data, followed by a protocol that permits rapid annotation of tandem MS data. The approach is exemplified by an analysis of the peptide components present in the venom of Conus amadis, yielding 225 chemically unique sequences, with identification of more than 150 sites of PTMs. In summary, this thesis presents different methodologies that address the existing limitations of de novo mass spectral structure determination of natural peptides and presents new methodologies that permit for rapid and efficient analysis of complex mixtures.