Membranous tubules are very abundant structures in living cells and form or are part of most intracellular organelles. The Golgi apparatus is mainly formed by tubules, which adopt different geometries and conformations. However, their physiological role has not yet been established and this is mainly due to the almost absolute lack of knowledge about the biochemical mechanisms regulating their formation, maintenance and disruption. The aim of this thesis was to investigate in a systematic way these mechanisms. The first step has been to set up an in vitro morphological assay suitable for the visualisation of Golgi-associated tubules in isolated Golgi stacks. This assay was based on electron microscopy and specifically on negative staining of whole-mount preparations. It allowed both qualitative and quantitative analysis of the morphological changes of Golgiassociated tubules after in vitro incubations. This assay was then used for screening several molecules or experimental conditions for their effect on tubular homeostasis. Among them, the most significant was BARS (BFA-dependent ADP-Ribosylation Substrate), a protein previously implicated in the maintenance of Golgi architecture. BARS has been found to cause the selective breakdown of the tubular part of the Golgi complex promoting fission events which convert the tubular structures into clusters of vesicles. This effect correlated with the enzymatic activity of BARS, which acts as an acyl-CoA dependent lysophosphatidic acid acyl transferase (LPAAT), increasing phosphatidic acid (PA) levels in Golgi membranes. This suggests that local modifications of the composition of the lipid bilayer is a possible mechanism for the fission of membranous tubules.

Dans la cellule hôte, Toxoplasma gondii se développe dans une vacuole parasitophore (VP) caractérisée par un réseau de nanotubes membranaires (RNM) dont la composition, le mécanisme de formation et la fonction sont obscures. Quelques protéines GRA, dont GRA2 et GRA6, sont sécrétées dans la VP à partir des granules denses puis ciblées au RNM. Cette localisation s'accorde avec l'hélice alpha-hydrophobe de GRA6 et les hélices alpha-amphipathiques de GRA2. Avant et après sécrétion dans la VP, les protéines GRA sont partiellement solubles. Le phénotype de parasites délétés de leur(s) gène(s) GRA2 et/ou GRA6 révèle que ces 2 protéines sont indispensables à la formation du RNM. J'ai montré 1) qu'avant leur insertion dans les membranes de la VP, la solubilité des protéines GRA est préservée grâce à des interactions hydrophobes avec peut être, des micelles de l'espace vacuolaire ; 2) que GRA12, une nouvelle protéine du RNM, n'interagit pas avec GRA2 dans ces membranes. 3) que l'adressage spécifique de GRA6 au RNM est déterminé par son domaine N-terminal hydrophile. 4) J'ai montré que GRA2 recombinante a une affinité pour le phosphatidyl inositol (4, 5) diphosphate avec lequel elle interagit via ses hélices alpha-amphipathiques. GRA2 déforme des liposomes de courbure membranaire importante pour générer de courts tubules membranaires. La tubulation est accentuée par GRA6 qui s'associe aux liposomes, quelque soit leur diamètre. Ces résultats valident le rôle direct de GRA2 et GRA6 dans la formation du RNM et laissent envisager un modèle de sa formation, dans lequel GRA6 favoriserait l'assemblage de vésicules lipidiques que GRA2 fusionnerait en tubules membranaires. / Within the host cell, Toxoplasma gondii multiplies in a parasitophorous vacuole (PV) characterized by a membranous nanotubular network (MNN). Its components, the mechanism of its formation and its function remain unknown. A few GRA proteins, including GRA2 and GRA6, are secreted from the dense granules into the PV and are targeted to the MNN. This location is in agreement with the hydrophobic alpha-helix predicted in GRA6 and with the GRA2 amphipatic alpha-helices. However, before and after their secretion in the PV, the GRA proteins are partially soluble. The phenotypic analysis of parasites deleted from their GRA2 and/or GRA6 gene(s) had shown that both these proteins are indispensable for MNN formation. During my thesis, I showed that before their insertion into the PV membranes, the GRA proteins solubility is preserved by establishing hydrophobic interactions, likely with micelles in the PV space. I also showed that GRA12, a novel MNN-associated protein, does not interact with GRA2 within these membranes. Using GRA6 as a model of study, I contributed to demonstrate that the GRA6 specific targeting to the MNN relies on its N-terminal hydrophilic domain. I demonstrated that recombinant GRA2 recognizes inositol (4, 5) biphosphate with which it interacts via its amphipatic alpha-helices. GRA2 deforms liposomes of steep membrane curvature into short membranous tubules. The tubulation is increased by GRA6 which associates with liposomes independently of their diameter. These results validate the direct role of both GRA2 and GRA6 in MNN formation and led us to propose a model in which GRA6 would tether vesicles, the fusion of which would be induced by GRA2.