Résumé La ribonucléase P (RNase P) est une ribonucléoprotéine omniprésente dans tous les règnes du vivant, elle est responsable de la maturation en 5’ des précurseurs des ARNs de transfert (ARNts) et quelques autres petits ARNs. L’enzyme est composée d'une sous unité catalytique d'ARN (ARN-P) et d'une ou de plusieurs protéines selon les espèces. Chez les eucaryotes, l’activité de la RNase P cytoplasmique est distincte de celles des organelles (mitochondrie et chloroplaste). Chez la plupart des espèces, les ARN-P sont constituées de plusieurs éléments structuraux secondaires critiques conservés au cours de l’évolution. En revanche, au niveau de la structure, une réduction forte été observé dans la plupart des mtARN-Ps. Le nombre de protéines composant la RNase P est extrêmement variable : une chez les bactéries, environ quatre chez les archéobactéries, et dix chez la forme cytoplasmique des eucaryotes. Cet aspect est peu connu pour les formes mitochondriales. Dans la plupart des cas, l’identification de la mtRNase P est le résultat de longues procédures de purification comprenant plusieurs étapes dans le but de réduire au minimum le nombre de protéines requises pour l’activité (exemple de la levure et A. nidulans). Cela mène régulièrement à la perte de l’activité et de l’intégrité des complexes ribonucléo-protéiques natifs. Dans ce travail, par l’utilisation de la technique de BN-PAGE, nous avons développé une procédure d’enrichissement de l’activité RNase P mitochondriale native, donnant un rendement raisonnable. Les fractions enrichies capables de cette activité enzymatique ont été analysées par LC/MS/MS et les résultats montrent que l’holoenzyme de la RNase P de chacune des fractions contient un nombre de protéines beaucoup plus grand que ce qui était connue. Nous suggérons une liste de protéines (principalement hypothétiques) qui accompagnent l’activité de la RNase P. IV De plus, la question de la localisation de la mtRNase P de A. nidulans a été étudiée, selon nos résultats, la majorité de la mtRNase P est attachée á la membrane interne de la mitochondrie. Sa solubilisation se fait par l’utilisation de différents types de détergent. Ces derniers permettent l’obtention d’un spectre de complexes de la RNase P de différentes tailles. / Abstract RNase P is a ribonucleo-protein complex (an RNA enzyme or ribozyme) that cleaves 5’ leader sequences of precursor tRNAs and a few other small RNAs. It occurs in all three domains of life, Bacteria, Archaea and Eukarya, with the latter containing distinct nuclear and organellar (mitochondrial or plastid) activities. In most instances, the complex contains a single, well-conserved RNA subunit that carries the active center of the enzyme. Yet, compare to bacterial and nuclear P RNA, most mtP RNAs are structurally highly reduced. The number of P proteins is highly variable: one in Bacteria, about four in Archaea, and ten in the cytoplasmic form of Eukarya. Much less is known in the case of mitochondria. MtRNase P is usually purified by using numerous separation steps that include unphysiological conditions, leading to complexes having a minimum number of subunits (e.g., in yeast and Aspergillus nidulans), that often loose their activity. Here, using BN PAGE, we have developed an enrichment procedure for A. nidulans mtRNase P that avoids some of the most disruptive conditions. The protein composition of active fractions was identified with LC/MS/MS, indicating that the RNase P holoenzyme is much larger than previously thought. Finally, the question of mtRNase P localization within mitochondria was investigated, by tracing its RNA subunit by RT PCR. We found that mtRNase P of A. nidulans is a predominantly membrane-attached enzyme; it is in part solubilized by detergents such as digitonin and Triton.

The diplomonads are a diverse group of eukaryotic microbes found in oxygen limited environments such as the intestine of animals were they may cause severe disease. Among them, the prominent human parasite Giardia intestinalis non-invasively colonizes the small intestine of humans and animals where it induces the gastrointestinal disease giardiasis. Two of the eight genetic groups of G. intestinalis, assemblage A and B, are known to infect humans and have zoonotic potential. At the start of project, genome scale data from assemblage B-H was either sparse or entirely missing. In this thesis, genome sequencing was performed on the assemblage B isolate GS (Paper I) and the P15 isolate (Paper III) of the hoofed-animals specific assemblage E to investigate the underlying components of phenotypic diversity in Giardia. Comparisons to assemblage A isolate WB revealed large genomic differences; entirely different repertoires of surface antigens, genome rearrangements and isolate specific coding sequences of potential bacterial origin. We established that genomic differences are also manifested at the transcriptome level (Paper VIII). In a follow up analysis (Paper IV) we concluded that the Giardia assemblages are largely reproductively isolated. The large genomic differences observed between Giardia isolates can explain the phenotypic diversity of giardiasis. The adaptation of diplomonads was further studied in Spironucleus barkhanus (Paper II), a fish commensal of grayling, that is closely related to the fish pathogen Spironucleus salmonicida, causative agent of systemic spironucleosis in salmonid fish. We identified substantial genomic differences in the form of divergent genome size, primary sequence divergence and evidence of allelic sequence heterozygosity, a feature not seen in S. salmonicida. We devised a transfection system for S. salmonicida (Paper VI) and applied it to the study of the mitochondrial remnant organelle (Paper VII). Our analyses showed that S. salmonicida harbor a hydrogenosome, an organelle with more metabolic capabilities than the mitosome of Giardia. Phylogenetic reconstructions of key hydrogenosomal enzymes showed an ancient origin, indicating a common origin to the hydrogenosome in parabasilids and diplomonads. In conclusion, the thesis has provided important insights into the adaptation of diplomonads in the present and the distant past, revealing hidden diversity.

In mammalian cells, homeostasis and fate during development relies on the proper transport of membrane-bound cargoes to their designated cellular locations. The hetero-tetrameric adaptor protein complexes (APs) are required for sorting and concentration of cargo at donor membranes, a crucial step during targeted transport. AP2, which functions at the plasma membrane during clathrin-mediated endocytosis, is well characterized. In contrast, AP1 a clathrin adaptor mediating the delivery of lysosomal hydrolases via mannose 6-phosphate receptors (MPRs) and AP3 an adaptor ensuring the proper targeting of lysosomal membrane protein are difficult to study by classic cell biology tools. To gain new insights on these APs, our lab has previously designed an in vitro system. Reconstituted liposomes were modified with small peptides mimicking the cytosolic domains of bona fide cargoes for AP1 and AP3 respectively and thereby enabling the selective recruitment of these APs and the identification of the interacting protein network. In the study at hand we utilize above-described liposomes to generate supported lipid bilayers and Giant Unilamellar Vesicles (GUVs), large-scale membrane systems suited for analysis by fluorescence microscopy. By using cytosol containing fluorescently-tagged subunits, we visualized clathrin coats on artificial membranes under near physiological conditions for the first time. Moreover, we demonstrated clathrin-independent recruitment of AP3 coats on respective GUVs. Presence of active ARF1 was sufficient for the selective assembly of AP1-dependent clathrin coats and AP3 coats on GUVs. By using dye-conjugated ARF1, we show that ARF1 colocalized with AP3 coats on GUVs and that increased association of ARF1 with GUVs coincided with AP1-dependent clathrin coats. Our previous study identified members of the septin family together with AP3 coats on liposomes. Here we show on GUVs, that active ARF1 stimulated the assembly of septin7 filaments, which may constrain the size and mobility of AP3 coats on the surface. Subsequent cell biology studies in HeLa cells linked septins to actin fibers on which they may control mobility of AP3-coated endosomes and thus their maturation. An actin nucleation complex, based on CYFIP1 was identified together with AP1 on liposomes before. Here we show on GUVs, that CYFIP1 is recruited on the surface surrounding clathrin coats. Upon supply of ATP, sustained actin polymerization generated a thick shell of actin on the GUV surface. The force generated by actin assembly lead to formation of long tubular protrusions, which projected from the GUV surface and were decorated with clathrin coats. Thereby the GUV model illustrated a possible mechanism for tubular carriers formation. The importance of CYFIP1-reliant actin polymerization for the generation of MPR-positive tubules at the trans-Golgi network (TGN) of HeLa cells was subsequently demonstrated in our lab. The notion that tubulation of artificial membranes could be triggered by actin polymerization allowed us to perform a comparative mass spectrometry screen. By comparing the abundance of proteins on liposomes under conditions promoting or inhibiting actin polymerization, candidates possibly involved in stabilization, elongation or fission of membrane tubules could be identified. Among the proteins enriched under conditions promoting tubulation, we identified type I phosphatidylinositol-4-phosphate 5-kinases. Their presence suggested an involvement of phosphatidylinositol 4,5-bisphosphate (PI(4,5)P2) in tubule formation. By cell biology studies in HeLa we show, that down regulation of these enzymes altered the dynamics of fluorescently-tagged MPRs, illustrating the importance of locally confined PI(4,5)P2 synthesis during formation of coated carriers at the TGN. Bin–Amphiphysin–Rvs (BAR) domains are known to sense membrane curvature and induce membrane tubulation. Among various BAR domain proteins, Arfaptin2 was enriched under conditions allowing tubulation of liposomes. By microscopy studies on HeLa cells we show, that Arfaptin2 as well as its close paralog Arfaptin1 were present on AP1-coated MPR tubules emerging from the TGN. We further show, that tubule fission occurred at regions were Arfaptin1 is concentrated and that simultaneous down regulation of both Arfaptins lead to increased number and length of MPR tubules. Since fission of coated transport intermediates at the TGN is poorly understood, our findings contribute a valuable component towards a model describing the entire biogenesis of coated post-Golgi carriers. In conclusion, combining artificial membrane systems and cell biology studies allowed us to propose new models for formation as wall as for fission of AP1-coated transport intermediates at the TGN. Further we gained new insights on AP3 coats and the possible involvement of septin filaments in AP3-dependent endosomal maturation.

Résumé La ribonucléase P (RNase P) est une ribonucléoprotéine omniprésente dans tous les règnes du vivant, elle est responsable de la maturation en 5’ des précurseurs des ARNs de transfert (ARNts) et quelques autres petits ARNs. L’enzyme est composée d'une sous unité catalytique d'ARN (ARN-P) et d'une ou de plusieurs protéines selon les espèces. Chez les eucaryotes, l’activité de la RNase P cytoplasmique est distincte de celles des organelles (mitochondrie et chloroplaste). Chez la plupart des espèces, les ARN-P sont constituées de plusieurs éléments structuraux secondaires critiques conservés au cours de l’évolution. En revanche, au niveau de la structure, une réduction forte été observé dans la plupart des mtARN-Ps. Le nombre de protéines composant la RNase P est extrêmement variable : une chez les bactéries, environ quatre chez les archéobactéries, et dix chez la forme cytoplasmique des eucaryotes. Cet aspect est peu connu pour les formes mitochondriales. Dans la plupart des cas, l’identification de la mtRNase P est le résultat de longues procédures de purification comprenant plusieurs étapes dans le but de réduire au minimum le nombre de protéines requises pour l’activité (exemple de la levure et A. nidulans). Cela mène régulièrement à la perte de l’activité et de l’intégrité des complexes ribonucléo-protéiques natifs. Dans ce travail, par l’utilisation de la technique de BN-PAGE, nous avons développé une procédure d’enrichissement de l’activité RNase P mitochondriale native, donnant un rendement raisonnable. Les fractions enrichies capables de cette activité enzymatique ont été analysées par LC/MS/MS et les résultats montrent que l’holoenzyme de la RNase P de chacune des fractions contient un nombre de protéines beaucoup plus grand que ce qui était connue. Nous suggérons une liste de protéines (principalement hypothétiques) qui accompagnent l’activité de la RNase P. IV De plus, la question de la localisation de la mtRNase P de A. nidulans a été étudiée, selon nos résultats, la majorité de la mtRNase P est attachée á la membrane interne de la mitochondrie. Sa solubilisation se fait par l’utilisation de différents types de détergent. Ces derniers permettent l’obtention d’un spectre de complexes de la RNase P de différentes tailles. / Abstract RNase P is a ribonucleo-protein complex (an RNA enzyme or ribozyme) that cleaves 5’ leader sequences of precursor tRNAs and a few other small RNAs. It occurs in all three domains of life, Bacteria, Archaea and Eukarya, with the latter containing distinct nuclear and organellar (mitochondrial or plastid) activities. In most instances, the complex contains a single, well-conserved RNA subunit that carries the active center of the enzyme. Yet, compare to bacterial and nuclear P RNA, most mtP RNAs are structurally highly reduced. The number of P proteins is highly variable: one in Bacteria, about four in Archaea, and ten in the cytoplasmic form of Eukarya. Much less is known in the case of mitochondria. MtRNase P is usually purified by using numerous separation steps that include unphysiological conditions, leading to complexes having a minimum number of subunits (e.g., in yeast and Aspergillus nidulans), that often loose their activity. Here, using BN PAGE, we have developed an enrichment procedure for A. nidulans mtRNase P that avoids some of the most disruptive conditions. The protein composition of active fractions was identified with LC/MS/MS, indicating that the RNase P holoenzyme is much larger than previously thought. Finally, the question of mtRNase P localization within mitochondria was investigated, by tracing its RNA subunit by RT PCR. We found that mtRNase P of A. nidulans is a predominantly membrane-attached enzyme; it is in part solubilized by detergents such as digitonin and Triton.

Diss. (sammanfattning) Stockholm : Karolinska institutet, 2007. / Härtill 4 uppsatser.

Les complexes multi-protéiques jouent un rôle crucial dans les cellules vivantes en catalysant et servant d'intermédiaires entre pratiquement toutes les activités cellulaires essentielles. Cependant, un grand nombre de ces machines se trouvent en très faibles quantités dans les cellules en particulier en ce qui concernent les complexes eucaryotes. Ceci est réfractaire à leur extraction à grande échelle et empêche sévèrement l'élucidation de leur structure et fonction. Dans le but de rendre les complexes multi protéiques accessibles par la voie de production recombinante, le groupe Berger a mis au point un ensemble de systèmes d'expression sur mesure pour la surproduction de complexes multi protéiques dans différents organismes hôtes incluant E. coli, les cellules d'insectes et les cellules de mammifères. Ces systèmes et en particulier le système MultiBac baculovirus/cellules d'insecte ont d'ors et déjà grandement contribués à l'étude de l'assemblage structural et fonctionnel à l'échelle moléculaire et atomique de nombreux complexes multi protéiques importants. Cela inclut en particulier le facteur général humain de transcription TFIID, un complexe de ~1.5 MDa qui constitue le sujet de recherche du laboratoire Berger. Mes contributions dans le développement de la technologie pour la production et dans l'élucidation des complexes TFIID humains sont discutées en détails dans cette thèse. / Multiprotein complexes play a crucial role in living cells by catalyzing and mediating virtually all essential cellular activities. However, many of these essential machines exist in very low endogenous amount in cells, in particular for eukaryotic complexes. This is refractory to large-scale extraction from native source material, severely impeding the elucidation of their structure and function. In order to make multiprotein complexes accessible by means of recombinant production, the Berger laboratory has developed an array of advanced expression systems tailor-made for overproducing multiprotein complexes in various host organisms including E. coli, insect cells and mammalian cells. Those systems, in particular the MultiBac baculovirus/insect cell system have already greatly contributed to studying the structural and functional assemblies of numerous important multiprotein complexes in molecular and atomic detail. Notably, this includes also the human general transcription factor TFIID, a ~1.5 MDa complex, which is the research focus of the Berger laboratory. My contributions to the expression technology development and to the structural elucidation of human TFIID complexes are discussed in details in this thesis.

Chez les eucaryotes pluricellulaires, la différenciation des cellules repose en partie sur l’activation oula répression des gènes. Les profils d’expression génique mis en place vont perdurer d’une générationcellulaire à l’autre. Ce phénomène met en jeu des mécanismes épigénétiques qui remodèlentlocalement la structure de la chromatine. Chez Drosophila melanogaster, les protéines des groupesPolycomb (PcG) et Trithorax (TrxG) participent au maintien du profil d’expression des gènes au coursdu développement. Les protéines PcG maintiennent les gènes réprimés tandis que les protéines TrxGmaintiennent les gènes activés. Une troisième classe de protéines nommée Enhancers of Trithoraxand Polycomb (ETP) module l’activité des PcG et TrxG. Dorsal Switch Protein 1 (DSP1) est uneprotéine HMGB (High Mobility Group B) classée comme une ETP. Par tamisage moléculaire, nousavions montré que la protéine DSP1 était présente au sein de complexes de poids moléculaire de 100kDa à 1 MDa. Le travail de thèse présenté ici a pour but d’identifier les partenaires de la protéineDSP1 dans l’embryon et de mieux connaître les propriétés biochimiques de DSP1. Premièrement, j’aimis en place puis effectué l’immunopurification des complexes contenant DSP1 dans des extraitsprotéiques embryonnaires. Cette approche nous a permis d’identifier 23 partenaires putatifs de laprotéine DSP1. Parmi ces protéines, nous avons identifié la protéine Rm62 qui est une ARN hélicaseà boîte DEAD. Les relations biologiques entre DSP1 et Rm62 ont été précisées. Deuxièmement, j’aidéterminé, par une approche biochimique, de nouvelles caractéristiques physico-chimiques de laprotéine DSP1. / In multicellular organism, the identity of cell is determined by several factors playing on genesexpression. Once established, the gene expression pattern is transmitted to daughter cells through aprocess involving epigenetic mechanisms that locally reshape the structure of chromatin. In Drosophilamelanogaster, the Polycomb (PcG) and trithorax (trxG) group genes are involved in the maintenanceof gene expression profile during development. Inside multimeric complexes, PcG proteins maintaingenes in repressed state whereas TrxG maintain genes active. A third class of proteins, calledEnhancers of Trithorax and Polycomb, regulate PcG and TrxG activities. Dorsal Switch Protein 1(DSP1) is a High Mobility Group B protein acting as an ETP. But DSP1 has not yet been identified inPcG or TrxG complexes. On the basis of gel filtration analysis of protein complexes in embryo nuclearextracts, it appears that the majority of DSP1 is present in complex(es) from 100 kDa to 1MDa. Aimsof present work are the identification of DSP1 protein partners in drosophila embryo and thecharacterization of biochemical properties of DSP1. Firstly, I used immunopurification from drosophilaembryonic nuclear extracts. The proteins purified with DSP1 were characterized through sequencingof peptides from individual protein bands by mass spectrometry. Among identified proteins, wefocused on the DEAD Box RNA helicase, Rm62. The role of interaction between DSP1 and Rm62 hasbeen characterized. Secondly, I have identified a new physicochemical aspect of DSP1 protein.

Le système de sécrétion de type II (SST2) permet la sécrétion de protéines repliées à travers la membrane externe chez les bactéries à Gram-négatif. Le SST2 est une nano-machine enchâssée dans l’enveloppe bactérienne, proche par sa composition et structure aux systèmes d’assemblage des pili de type IV (PT4) impliqués, entre autres, dans d’adhésion et motilité. Chez Klebsiella oxytoca, la surexpression des gènes pul codant le SST2 permet l’assemblage de pili composées des sous-unités PulG. Ceci suggère qu’en conditions physiologiques l’assemblage d’un pseudopilus périplasmique permet la sécrétion du substrat spécifique du SST2, la pullulanase. Dans ce projet nous avons exploré le mécanisme moléculaire de l’assemblage du pseudopilus en se focalisant sur les interactions de PulG avec les composants du SST2 dans la membrane interne. En utilisant l’approche de double-hybride bactérien, nous avons établi le réseau d’interactions de PulG avec les pseudopilins mineures PulH, I, J et K et avec la plateforme d’assemblage (PA). Pour valider ces interactions, nous avons combiné des techniques de biochimie (co-purification par affinité, pontage cystéine et chimique) avec des analyses fonctionnelles de sécrétion et de formation du pseudopilus. Nous avons mis en évidence des interactions entre PulG et les protéines de la PA, PulF et PulM, et nous avons analysé en détail l’interface PulG-PulM. Les résultats suggèrent la formation d’un complexe PulK-I-J-H-G dans la membrane interne impliqué dans des étapes précoces de la formation du pseudopilus, à travers les interactions de PulG et PulH avec PulM et PulF. Nos données expérimentales suggèrent un rôle majeur de PulM dans la sécrétion, vraisemblablement durant l’assemblage du SST2 et l’élongation du pseudopilus. Nos travaux collaboratifs mettant en jeu l'analyse par spectroscopie de masse et en dynamique moléculaire in silico révèlent le rôle essentiel des résidus conservés Glu5 et Thr2 de PulG, requis pour l’interaction avec PulM. Ces données suggèrent que Glu5 participe à l'extraction de PulG de la membrane, en neutralisant la charge positive de son peptide N-terminal par des interactions intramoleculaires. Ces résultats permettent d'établir un modèle détaillant les étapes initiales de l’assemblage des pseudopili dans la membrane interne, relevant pour de futures études sur le SST2 et nanomachines homologues. sécrétion de protéinespili de type 4 assemblage de fibres complexes protéiques membranairesinteractions protéine-protéinemicroscopie à immuno-fluorescence simulations en dynamique moléculairedouble-hybride bactérien spectrométrie de masse nanomachines bacteriennes / The type II secretion system (T2SS) drives the translocation of folded, periplasmic proteins across the outer membrane in Gram-negative bacteria. Secretion is carried out by an envelope-spanning nanomachine that is similar to the apparatus that builds type IV pili (T4P), bacterial surface filaments involved in adhesion, motility and other functions. In the Pul T2SS of Klebsiella oxytoca, overexpression of pul genes in plate-grown bacteria allows the assembly of T4P-like surface fibres made of PulG subunits, suggesting that a periplasmic pseudopilus fibre plays a role in the secretion of the type II substrate pullulanase under physiological conditions. In this project, we explored the molecular mechanism of pseudopilus assembly by focusing on the interaction between PulG and the T2SS inner membrane and pseudopili components. The network of interactions of PulG with the minor pseudopilins PulH, I, J and K and the assembly platform (AP) components was established using bacterial two-hybrid analysis. To validate these interactions, we combined biochemical approaches (affinity co-purification, chemical or cysteine cross-linking) with functional assays of secretion and pseudopilus formation. We provide evidence of the interaction between PulG and the AP proteins PulF and PulM, and delve into the PulG-PulM interface. Our results point to the formation of a PulK-I-J-H-G complex in the plasma membrane involved in early steps of fibre assembly, with a determinant role for PulG and PulH interaction with PulM and PulF. We obtained experimental evidence supporting a major role for PulM in pseudopilus assembly and protein secretion, probably by intervening in the assembly of the T2SS apparatus and in pseudopilus elongation. The results of experimental and in silico studies in collaboration with experts in mass spectrometry and molecular dynamics support the essential role of the highly conserved PulG residues Glu5 and Thr2, which participate in PulM binding. In addition, Glu5 probably favours PulG membrane extraction by neutralising its N-terminal positive charge through intra-molecular interaction. These findings shed new light on early membrane events during fibre assembly, and open new and exciting avenues in research on T2SSs and related nanomachines.protein secretiontype 4 pilifibre assemblymembrane protein complexprotein-protein interactionsimmunofluorescence microscopymolecular dynamics simulationsbacterial two-hybrid assaymass spectrometrybacterial nanomachines

京都大学 / 0048 / 新制・課程博士 / 博士(医学) / 甲第22832号 / 医博第4671号 / 新制||医||1047(附属図書館) / 京都大学大学院医学研究科医学専攻 / (主査)教授 林 康紀, 教授 安達 泰治, 教授 岩田 想 / 学位規則第4条第1項該当 / Doctor of Medical Science / Kyoto University / DFAM

In mammalian cells, homeostasis and fate during development relies on the proper transport of membrane-bound cargoes to their designated cellular locations. The hetero-tetrameric adaptor protein complexes (APs) are required for sorting and concentration of cargo at donor membranes, a crucial step during targeted transport. AP2, which functions at the plasma membrane during clathrin-mediated endocytosis, is well characterized. In contrast, AP1 a clathrin adaptor mediating the delivery of lysosomal hydrolases via mannose 6-phosphate receptors (MPRs) and AP3 an adaptor ensuring the proper targeting of lysosomal membrane protein are difficult to study by classic cell biology tools. To gain new insights on these APs, our lab has previously designed an in vitro system. Reconstituted liposomes were modified with small peptides mimicking the cytosolic domains of bona fide cargoes for AP1 and AP3 respectively and thereby enabling the selective recruitment of these APs and the identification of the interacting protein network. In the study at hand we utilize above-described liposomes to generate supported lipid bilayers and Giant Unilamellar Vesicles (GUVs), large-scale membrane systems suited for analysis by fluorescence microscopy. By using cytosol containing fluorescently-tagged subunits, we visualized clathrin coats on artificial membranes under near physiological conditions for the first time. Moreover, we demonstrated clathrin-independent recruitment of AP3 coats on respective GUVs. Presence of active ARF1 was sufficient for the selective assembly of AP1-dependent clathrin coats and AP3 coats on GUVs. By using dye-conjugated ARF1, we show that ARF1 colocalized with AP3 coats on GUVs and that increased association of ARF1 with GUVs coincided with AP1-dependent clathrin coats. Our previous study identified members of the septin family together with AP3 coats on liposomes. Here we show on GUVs, that active ARF1 stimulated the assembly of septin7 filaments, which may constrain the size and mobility of AP3 coats on the surface. Subsequent cell biology studies in HeLa cells linked septins to actin fibers on which they may control mobility of AP3-coated endosomes and thus their maturation. An actin nucleation complex, based on CYFIP1 was identified together with AP1 on liposomes before. Here we show on GUVs, that CYFIP1 is recruited on the surface surrounding clathrin coats. Upon supply of ATP, sustained actin polymerization generated a thick shell of actin on the GUV surface. The force generated by actin assembly lead to formation of long tubular protrusions, which projected from the GUV surface and were decorated with clathrin coats. Thereby the GUV model illustrated a possible mechanism for tubular carriers formation. The importance of CYFIP1-reliant actin polymerization for the generation of MPR-positive tubules at the trans-Golgi network (TGN) of HeLa cells was subsequently demonstrated in our lab. The notion that tubulation of artificial membranes could be triggered by actin polymerization allowed us to perform a comparative mass spectrometry screen. By comparing the abundance of proteins on liposomes under conditions promoting or inhibiting actin polymerization, candidates possibly involved in stabilization, elongation or fission of membrane tubules could be identified. Among the proteins enriched under conditions promoting tubulation, we identified type I phosphatidylinositol-4-phosphate 5-kinases. Their presence suggested an involvement of phosphatidylinositol 4,5-bisphosphate (PI(4,5)P2) in tubule formation. By cell biology studies in HeLa we show, that down regulation of these enzymes altered the dynamics of fluorescently-tagged MPRs, illustrating the importance of locally confined PI(4,5)P2 synthesis during formation of coated carriers at the TGN. Bin–Amphiphysin–Rvs (BAR) domains are known to sense membrane curvature and induce membrane tubulation. Among various BAR domain proteins, Arfaptin2 was enriched under conditions allowing tubulation of liposomes. By microscopy studies on HeLa cells we show, that Arfaptin2 as well as its close paralog Arfaptin1 were present on AP1-coated MPR tubules emerging from the TGN. We further show, that tubule fission occurred at regions were Arfaptin1 is concentrated and that simultaneous down regulation of both Arfaptins lead to increased number and length of MPR tubules. Since fission of coated transport intermediates at the TGN is poorly understood, our findings contribute a valuable component towards a model describing the entire biogenesis of coated post-Golgi carriers. In conclusion, combining artificial membrane systems and cell biology studies allowed us to propose new models for formation as wall as for fission of AP1-coated transport intermediates at the TGN. Further we gained new insights on AP3 coats and the possible involvement of septin filaments in AP3-dependent endosomal maturation.