Résumé
The goal of my thesis was to discover novel peptide hormones (PH) in the human genome. Peptide hormones (PH) are an important class of molecules that are involved in a broad range of normal and pathological physiological processes. PH are small secreted proteins that are processed from larger proteins, called precursors. In the first part of the thesis I introduce hidden Markov models (HMM)-based algorithms that allow for the modelling of those PH precursor sequences. This leads me to derive models of some of their well-known characteristic features, including signal peptides and prohormone convertase cleavage sites. Next, I make use of orthology information and modify the HMM algorithms so that the HMM can incorporate a conservation score that is calculated along protein sequence alignments. I show that this idea brings significant improvements in the quality of predictions and I conclude the chapter by drawing a list of potential candidate peptide hormones that was obtained by running the HMM algorithms on genome-wide protein sequence data. In the second and third part of the thesis I present data on the most interesting candidates that we named "spexin" and "augurin". I show in vitro subcellular localization, secretion and processing of those proteins after transfection of their coding DNA into endocrine cells. I then show mRNA and protein in vivo expression data in the mouse for those two proteins, and present functional data on spexin. I conclude both chapters by making some speculations on the functional significance of those two proteins. In the fourth part of the thesis I present data on the expression, secretion and processing of four extra candidate peptide hormones.