Membrane Stress and the Role of GYF Domain Proteins

Full text

(1)Membrane Stress and the Role of GYF Domain Proteins.

(2)

(3) Membrane Stress and the Role of GYF Domain Proteins Alexander Georgiev. Stockholm University.

(4) © Alexander Georgiev, Stockholm 2008 ISBN 978-91-7155-683-7 Printed in Sweden by Universitetsservice AB, Stockholm 2008 Distributor: Department of Biochemistry and Biophysics, Stockholm University.

(5) To my daughter Joanna.

(6)

(7) List of Publications. This thesis is based on the following papers, which are referred to in the text by their Roman numerals. I. II. III. IV. ∗. Georgiev, A., Sjöström, M., Wieslander, Å. (2007). Binding specificities of the GYF domains from two Saccharomyces cerevisiae paralogs. Protein Eng Des Sel, 20(9):443–52 Georgiev, A.∗ , Leipus, A.∗ , Olsson, I., Berrez, J-M., Mutvei, A. (2008). Characterization of MYR1, a dosage suppressor of YPT6 and RIC1 deficient mutants. Current Genetics, 53(4):235–47 Georgiev, A., Ge, C., Wieslander, Å. (2008). Lipid-specific interactions of Myr1 suggest a role in sensing membrane stress. Manuscript Wikström, M., Kelly, A.A., Georgiev, A., Eriksson, H.M., Rosén Klement, M., Bogdanov, M., Dowhan, W., Wieslander, Å (2008). Curvature-engineered Escherichia coli bilayers reveal critical lipid head-group size for membrane protein function in vivo. Manuscript. These authors have contributed equally to the paper. Reprints were made with permission from the publishers..

(8)

(9) Contents. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Budding yeast as a model organism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Classification, life cycle and biology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Genetics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Studying vesicular traffic in yeast . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. Membrane traffic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Endomembrane systems in eukaryotic cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Protein translocation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The life cycle of transport vesicles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Rab GTPases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ER to Golgi . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Intra-Golgi traffic and cisternal maturation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Exocytosis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Vacuolar traffic and endocytosis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Retrograde traffic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. Lipids in Saccharomyces cerevisiae . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Lipids in organelles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Synthesis and distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Regulation of synthesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Phosphoinositides . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Lipids and membrane traffic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. Protein-lipid interactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Membrane binding proteins . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Mechanisms of membrane binding . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Lipid binding and vesicular traffic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. Membrane stress . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Spontaneous curvature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Physiological membrane stress . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Examples in eukaryotes and prokaryotes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Importance of membrane stress for homeostasis in unicellular organisms . . . . . . . .. Translational control, RNA decay and stress . . . . . . . . . . . . . . . . . . . . . . . Translation initiation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . RNA degradation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Processing bodies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Nutrient deprivation and the TOR signaling pathway . . . . . . . . . . . . . . . . . . . . . . .. 1 3 3 4 4 5 5 6 6 8 9 10 10 11 12 13 13 14 15 16 17 19 19 19 22 23 23 24 25 26 27 27 28 29 30.

(10) The GYF domain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33 Structural features . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33 GYF domain protein families . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34 Binding partners . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35 Membrane traffic connections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36 Relation to RNA metabolism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36 Integrating knowledge by computational biology . . . . . . . . . . . . . . . . . . . 37 Gene Ontologies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38 Multivariate analysis of protein sequences . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38 Summary of papers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41 The GYF Domain of Myr1 and Smy2 (Paper I) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41 MYR1 – a suppressor of YPT6 and RIC1 mutants (Paper II) . . . . . . . . . . . . . . . . . . 42 Myr1 as a membrane sensor – interactions with lipids (Paper III) . . . . . . . . . . . . . . 43 Lipid-engineered bacteria (Paper IV) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45 Discussion and perspectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47 Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53.

(11) Abbreviations. COPI. coat protein I. COPII. coat protein II. GAP. GTPase activating protein. GEF. guanine exchange factor. GFP. green fluorescent protein. GO. gene ontology. GYF. glycine-tyrosine-phenylalanine, signature amino acid motif of the GYF domain. eIF. eukaryotic initiation factor. ER. endoplasmic reticulum. LUV. large unilamellar vesicle. MYR1. multi-copy suppressor of YPT6 and RIC1 (≡SYH1). NSF. N-ethylmaleimide-sensitive factor. P-body. processing body, cytoplasmic site of mRNA degradation. PCA. principal components analysis. PLS. partial least squares. PA. phosphatidic acid. PC. phosphatidylcholine. PE. phosphatidylethanolamine. PI. phosphatidylinositol. PS. phosphatidylserine. PIPs. phosphoinositides, phosphorylated derivatives of PI. PITP. PI transfer protein. SMY2. suppressor of myo2 mutant. SNARE. soluble NSF attachment receptor. SRP. signal recognition particle. SYH1. Smy2 homolog, synonym for MYR1.

(12)

(13) Introduction. The divide between what we perceive as living or not is six nanometers thin, and is known as the cellular membrane. It is the semi-permeable barrier that topologically defines the cell and is responsible for maintaining the gradients characteristic of living systems. In the following pages, I present a study focused on the interplay between the two major constituents of biological membranes - lipids and proteins. The main objects of this study are the two poly-proline binding GYF domain proteins Myr1 and Smy2 from bakers’ yeast, both identified through genetic screens in trafficking mutants. Two independent lines of evidence connect these proteins to intracellular membrane traffic and to processing of messenger ribonucleic acid (mRNA). These lines cross at the regulation of cellular membrane homeostasis through membrane-protein interactions, a research area characterized by a precipitous accumulation of data during the last few years. Aiming for a deeper understanding of the nature of GYF domain proteins, and of their role in the cellular response to membrane alterations, I outline the fundamental processes in which they take part. To establish the biological context, I also briefly present the unicellular organisms Saccharomyces cerevisiae and Escherichia coli which, despite their idiosyncrasies, have served as models to study membrane-protein interactions in general. An emphasis is placed upon some evolutionary aspects of the different cellular processes and components. Thus, while most of the background is heavily weighted towards budding yeast, examples from other domains of life and especially those dear to us mammals are given when appropriate. Separate chapters in the thesis are devoted to intracellular membrane traffic, biogenesis and dynamics of lipids and membranes, and lipid-protein interactions. Their connections to GYF domain proteins are further evaluated in regard to transcript stability, regulation of translation initiation, and membrane stress. Computational methods used to establish some of these connections are also described. Finally, my contribution to the advancement of knowledge in this field is summarized in a synopsis of papers. It is inevitable that once we succeed reaching the realms of the known we are left facing the unknown, and therefore this work leaves many questions open. It is my conviction though that the years spent on it are merely a preparation for the quest for answering those and many other questions.. 1.

(14)

(15) Budding yeast as a model organism. The budding yeast Saccharomyces cerevisiae is one of the best studied eukaryotic organisms. It has been known to humans and used for millenia for its ability to ferment sugars. In the past few decades, it was also established as an important research tool for genetics and molecular biology. The advantages of yeast over other model systems combine rapid growth, ease of cultivation, ease of genetic manipulation, low pathogenicity, and the conservation of fundamental processes. Major metabolic and signaling pathways, such as cell cycle control and membrane trafficking, are very similar in fungi and animals, and have often been initially studied in yeast [131, 176].. Classification, life cycle and biology The kingdom Fungi is widespread in all possible environments. Nearly 1000 species of yeasts have colonized the depths of oceans, the soils of earth, and our own bodies. S. cerevisiae belongs to the Ascomycota phylum and was named after the beer malt from which it was isolated in 1837 [217]. In nature yeasts are mostly diploid, and undergo meiosis under certain nutrient limitations, only to mate again when conditions permit. Wild type strains are normally homothallic, meaning that haploids are capable of switching their mating type with a high frequency during budding, which allows for rapid diploidization. To make genetic studies feasible, lab strains are selected that have lost the mating type switching ability. Yeast cells are oval and 5-10 µm in size. Both haploid and diploid cells propagate by budding and optimally complete their cell cycle in 90 minutes. Several cellular structures may be observed by light microscopy and with the aid of different staining techniques. Intracellular membranous compartments in yeast include the endoplasmic reticulum (ER), the Golgi, the vacuole, endosomes, multi-vesicular bodies, and transport vesicles (fig. 1). Those are similar in function to the corresponding organelles in metazoans, with the vacuole playing a comparable role to the lysosome of animal cells. The nuclear envelope, a specialized extension of the ER enclosing the genetic material, does not disassemble during cell division, a phenomenon known as closed mitosis and also typical for some protozoans [31]. 3.

(16) Figure 1: Yeast cell - schematic representation. 1. ER, 2. Golgi cisternae at different levels of maturation, 3. Vacuole, 4. Nucleus, 5. Early endosome, 6. Pre-vacuolar compartment, 7. Lipid droplets, 8. Bud, 9. Cell wall, 10. Mitochondrion. 11. Peroxisomes.. Genetics Yeast is a widely used object of classical genetics, aided by the existence of both haploid and diploid vegetative forms. Simple techniques for tetrad dissection allow for the recovery of the products of meiosis, and the ability to grow on synthetic media with completely defined composition makes selection of diverse markers possible. S. cerevisiae was the first eukaryote to have its entire genome sequenced, in 1996 [73].The yeast genome is organized in 16 chromosomes, 12 million nucleotides, and 5795 protein coding genes. While 1/4 of the yeast genes are similar to human ones, a comparison between 13 Hemiascomicetous species determined that another 1/3 are specific for Ascomycetes [144]. The number of characterized yeast genes is steadily growing but despite earlier predictions that the functions of all genes will be known already by 2007, there are still more than 1000 uncharacterized genes, many of them yeastspecific [102, 187].. Studying vesicular traffic in yeast Budding yeast has traditionally served as a model organism for the dissection of membrane traffic. One of the first studies dedicated to vesicular traffic and secretion is the work by Peter Novick and Randy Schekman [173, 176]. The conservation of the mechanism of membrane traffic and the lack of tissue specialization in yeast have allowed determining the function of many protein regulators with homologs in metazoan species.. 4.

(17) Membrane traffic. Living cells are highly organized structures in a fragile dynamic equilibrium. To maintain their organization during normal biosynthetic activity and growth, they have developed pathways for regulated delivery of macromolecules. Membrane traffic is of special interest here and is presented in this chapter. Other types of traffic, such as nucleo-cytoplasmic, import of proteins into mitochondria and chloroplasts, or transport of molecules and ions across membranes, are outside the scope of this work.. Endomembrane systems in eukaryotic cells Nominally, the difference between prokaryotes and eukaryotes is the presence of a nucleus (karyon). It is in fact the existence of an evolved membrane system in eukaryotes that allows for the increased cell complexity by compartmentalization; indeed, even the eponymous organelle organizing the genetic material in eukaryotes may be regarded as a result of this compartmentalization [21]. The endomembrane compartments of animals and plants are similar to those present in yeast as described above. Plant cells have vacuoles while animal cells have lysosomes; due to the functional similarities and homologous proteins in the two types of organelles we regard them as analogous. Multicellular organisms with specialized tissues have evolved types of membrane compartments specific for new functions, such as melanosomes or secretory granules of neurons containing respectively melanin or neurotransmitters. Membrane compartments differ by the lipid and protein composition of their membranes. Their identity is a result of their sequential biogenesis and the directional routes of exchange of membrane material between them. A typical biosynthetic pathway followed by a secretory protein would begin with synthesis at membrane-bound polysomal complexes at the ER, followed by translocation through the membrane leading to either membrane insertion (for integral proteins) or to lumenal localization. Membrane material travels further via vesicle and tubular intermediates, to an intermediate compartment between ER and Golgi, then to early, mid and late Golgi, and eventually by secretory vesicles to the plasma membrane [181]. Outgoing traffic in this direction is referred 5.

(18) to as anterograde, in contrast to retrograde traffic. The latter includes internalization of plasma membrane material by endocytosis, transport of endocytic vesicles to a pre-endosomal compartment, with a final destination either in the vacuole (for degradation) or backwards via Golgi to the ER.. Protein translocation Protein translocation at the ER, currently understood at the structural level [234], is achieved through the translocon composed of Sec61, Sec62, Sec63, Sec71 and Sec72 [191, 192]. The precursors of secretory proteins pass from the cytosol into the lumen through a membrane channel formed by Sec61. This process requires ATP and molecular chaperons of the hsp70 family, exemplified by yeast Ydj1 and E. coli DnaJ. Depending on whether the polypeptide synthesis is complete before or after the onset of translocation, two major routes exist. For yeast, post-translational translocation is typical. It requires that protein precursors are completely synthesised in the cytosol and maintained there in semi-folded state by chaperons homologous to E. coli DnaJ. The co-translational translocation is used by other secretory precursors, especially in mammalian cells. Upon such translocation the need for chaperons is relieved, however another set of cytosolic factors known as the signal recognition particle is needed [87, 123]. A ribosome–nascent-chain–SRP complex binds to the translocon channel, after which the SRP is released and the translating ribosome remains attached to the membrane via interactions with the channel [179, 192]. In the ER lumen another hsp70 chaperon exists known as BiP, encoded by the KAR2 gene in S. cerevisiae. It binds to a peptide loop of Sec63 containing a 70 residue region homologous to the ’J box’ of the DnaJ family of proteins. The BiP protein serves as a molecular ratchet pulling the polypeptide chain inside the lumen.. The life cycle of transport vesicles For prokaryotes, the process of secretion is complete with protein translocation through the membrane. In contrast, this is the just the beginning for eukaryotes. Once in the ER, proteins continue to traverse the endomembrane system in a regulated fashion, being carried inside transport vesicles. Before going into more detail about the specific steps of the transport, one needs to understand the general mechanism of formation of the carriers, their molecular anatomy, and the way they are guided towards and fuse with their targets. 6.

(19) Membrane vesicles emerge through invaginations of a donor membrane that pinch off to form a spherical vesicle, typically 50-100 nm in diameter. The initial bending of the membrane is caused by successive attachment of protein molecules to its cytoplasmic side. These proteins include small GTPases, coat proteins, and cargo adapters. Once the vesicle is formed the membrane undergoes fission, thus releasing the completed carrier in the cytosol. At this point, many vesicles “shed off” their protein coats to facilitate later fusion.. Figure 2: Vesicle cycle, adapted from Behnia & Munro [15].. Depending of the type of vesicle, it travels through the cell either by diffusion or through a vectorial transport along cytoskeletal elements. Upon reaching its final destination, the target membrane, vesicles are tethered by long filamentous proteins or multi-protein complexes. Specific attachment to the correct membrane is ensured by the tethering mechanism, but also enhanced by the next step, docking of the vesicle to the membrane with the participation of both vesicle and target SNAREs. Finally, the membrane of the vesicle fuses with the target membrane and the soluble cargo is released inside the recipient compartment (fig. 2). It is very important, from an evolutionary perspective, to realize that despite the strict specificity existing when it comes to the individual protein regulators of the vesicle life cycle, the overall scheme of the process is the same at each stage and indeed involves homologous proteins. Sometimes an intermediate stage in vesicle formation is stabilized leading to formation of tubules instead. This is commonly observed in some organelles, but can also be the result of aberrant vesicle formation due to mutations in the protein regulator genes [51, 130]. 7.

(20) Rab GTPases The Rab family of small GTPases (as opposed to large GTPases, such as various structural proteins or G proteins) is the largest subfamily of RASrelated GTPases. More than 60 homologs exist in humans, and 11 in yeast [128, 188]. As their name suggests, they catalyse GTP hydrolysis to GDP and phosphate, a reaction coupled to structural rearrangements within the GTPase molecule and hence to its activity. The functional cycle of a Rab GTPase (fig. 3) begins with synthesis in the cytoplasm, followed by post-translational prenylation necessary for membrane association. Initially Rabs bind to GDP which is more abundant than GTP, and in complex with a Rab escort protein they travel to their site of action. The latter is marked by the specific guanine exchange factors (GEF) which ensure that Rabs will only be activated at the correct site. Activation itself is catalysed by the GEF and consists of ejecting the GDP molecule and replacing it for a GTP [75]. Once activated, Rabs remain attached to the membrane and may travel in a bound state with the transport vesicle. When specific effectors become available, Rabs complete their cycle by interacting with the effectors and hydrolyzing GTP, a process stimulated by GTPase activator proteins, GAPs. At this point, Rab-GDP is extracted from the membrane to be reused for another cycle (see also Goody et al. [75] for a recent review).. Figure 3: Functional cycle of Rabs.. Rab/Ypt proteins participate in four major steps of membrane traffic (fig. 2), cargo sorting, myosin dependent transport, vesicle tethering, and membrane fusion, recently reviewed by Grosshans et al. [78]. In contrast to the evolutionary conservation of nucleoporins and SNAREs [117], the Rab GTPase family has diversified in parallel with the specialization of endomembrane systems in multicellular organisms [78, 80, 205]. Yeast contains a minimal set of Rabs [128] which can be regarded as a reference to the more diverse Rabs in animals and plants. The site of action and the set of activators and effectors known for the yeast Rabs are listed in Table 1. 8.

(21) Table 1: Rab GTPases in S. cerevisiae and their known interaction partners.. GTPase. localization GEF. effectors. GAP. Sec4. PM. Sec2. Sec15 (exocyst), Sro7. Msb3, Msb4. Ypt1. Golgi. TRAPP. Sec34/35 (COG). Gyp1. Ypt31/32. endosome, Golgi. ?. Sec2, Rcy1. ?. Ypt51/52/53 endocytic carriers. ?. Vac1, CORVET. ?. Ypt6. endosome, Golgi. Ric1/Rgp1 complex. GARP/VFT. Gyp6. Ypt7. vacuole. Vps39. VPS/HOPS. Gyp7. ER to Golgi Inside the ER, protein precursors are exposed to a changed environment of higher oxidation and to a number of modification and maturation enzymes. First, the signal peptide is cleaved off by the Sec11/Spc1-3 complex [23, 55, 160, 176]. Enzymes of the protein disulfide isomerase family are responsible for rearranging the disulfide bridges upon refolding of the protein after translocation [57]. The ER is also site of initial post-translational modification by glycosylation [220]. From the ER, proteins may continue further by being exported in membrane vesicles. During the formation of vesicles, cargo selection is important since certain proteins, such as the processing enzymes, need to be retained in the ER. A typical retention signal for ER is KDEL (HDEL in yeast) [186], while motifs such as FF or ExD serve as signals for packing into vesicles for transport. Vesicles bud from the ER with the participation of the COPII complex, consisting of the small GTPase Sar1 and the dimers Sec13-Sec31 and Sec23Sec24 [7, 130]. Sar1 is activated by Sec12 through binding to GTP [161, 162], via a mechanism common to small GTPases. As a results, the N-terminal amphipathic α-helix of Sar1 inserts into the membrane, causing a local deformation of the lipid bilayer and subsequently tubulation [130]. The presence of bulky hydrophobic residues in the α-helix is essential for its ability to induce membrane curvature. After attachment to the membrane, Sar1 recruits the Sec23/Sec24 dimer [13]. Sec23 and Sec24 are structurally similar [129], however Sec23 is responsible for Sar1 binding, while Sec24 fulfills the task of recruiting cargo, which additionally stabilizes the association with the membrane [7]. Fi9.

(22) nally, the tetramer composed of two Sec13 and two Sec31 subunits binds, and subsequent polymerization forms the COPII cage [212]. Vesicles leaving the ER are exclusively targeted towards the Golgi. While in mammals this proceeds through an intermediate compartment, ERGIC, S. cerevisiae apparently delivers the vesicles directly to the cis-Golgi. Vesicles are tethered to the Golgi by the transport complex TRAPP [106] and by Uso1 [32] and fuse under the regulation of Ypt1 [204]. This process is stimulated by the Ypt1 GEF, TRAPP [108], and inhibited by Gyp1, a GAP protein for Ypt1 [52]. It also involves the v-SNAREs Sec22, Bet1, Bos1 and Ykt6, and the t-SNARE Sed5 [43, 89, 168].. Intra-Golgi traffic and cisternal maturation The Golgi is by far the most intriguing organelle in membrane traffic, due both to its function as a central sorting station, and to its unique morphologies in different cell types and its diverse biosynthetic activities. Indeed, until it reaches the Golgi, the vesicular pathway is practically unbranched, since all vesicles generated at the ER target Golgi [24]. Yeast Golgi is functionally equivalent to its metazoan counterpart, however it is morphologically different. The Golgi apparatus cisternae are typically closely stacked in mammals, but in S. cerevisiae they are represented by dispersed organelles. Those individual cisternae correspond to the functionally defined early (cis), medial, and late (trans) Golgi, and can be labeled by markers specific for these compartments. They are highly dynamic structures, which are converted directionally from early to later forms, in agreement with the cisternal maturation model [142, 148]. Golgi is the site of formation of COPI coated vesicles, the second coated vesicle type to be described after clathrin [178]. Vesicle formation is initiated by Arf1, a “cousin” of Sar1 GTPase, by a similar mechanism. Membrane insertion of the N-terminal helix of Arf1 follows activation by nucleotide exchange, catalysed by the Sec7 GEF. The COPI complex formed from seven protein in the cytosol binds en block to the membrane [88]. COPI coated vesicles mediate retrograde intra Golgi and Golgi to ER traffic. Other routes leading from the Golgi to the plasma membrane or vacuole are explained in more detail below.. Exocytosis Secretion as the ultimate fate of most of the material participating in membrane traffic has several functions, depending on the type of cell. Non-specialized unicellular eukaryotes such as yeast use secretion in order to expand their surface, therefore this is an essential process 10.

(23) needed for cell growth. Cells that possess any type of extracellular wall or matrix produce and modify it by secreting building materials such as polysaccharides and by exporting the modifying enzymes, which often are retained in the periplasmic space. Finally, highly specialized cells as the epithelial cells or neurons secrete unique enzymatic products or neurotransmitters according to their function. All these processes occur through the regulated formation of secretory vesicles at the Golgi and their targeted transport and fusion with the plasma membrane. The exocyst is a special landmark composed of eight proteins and responsible for the directing exocytosis [79, 224]. It interacts with the Sec4 GTPase, which is present on post-Golgi secretory vesicles after being activated by its GEF Sec2. Correct targeting of secretory vesicles involves the cytoskeleton, and fusion with the plasma membrane requires a specific set of SNAREs [174].. Vacuolar traffic and endocytosis The vacuole is the important “back yard” of the yeast cell where various compounds such as salts and amino acids are stored. The presence of hydrolytic enzymes is essential for the function of the vacuole and their activity is ensured by the environment characterized by lower pH. Aside from the resident vacuolar enzymes, proteins are targeted there for degradation. Membrane material reaches the vacuole via a pre-vacuolar compartment, known as the late endosome, which on its part receives input from the late Golgi or the early endosome. Through a special budding mechanism directed inwards, late endosome compartments are transformed into multivesicular bodies, thus allowing for the degradation of the enclosing membrane material as well as the cargo. Cargo sorted for vacuolar delivery at the late Golgi includes several hydrolases such as carboxypeptidase Y, proteinase A and B [30, 77]. A peptide signal is required for correct sorting to the vacuole, and in its absence the proteins end up at the cell surface [30, 145]. Importantly, exit from the Golgi towards a pre-vacuolar compartment is dependent of Vps34, a PI3 kinase. Subsequent steps have been dissected through the analysis of vacuolar protein sorting (VPS) mutants [12, 193]. Another source of traffic to the vacuole is endocytosis. It is a means to internalize plasma membrane material in order to regulate signaling or permeability by retrieving receptors and transporters from the cellular exterior. An evolutionary conserved process, endocytosis is largely associated with clathrin in yeast as in mammals, even though it is possible in the absence of clathrin in yeast [109]. Clathrin and Ede1 are the first two proteins to engage in endocytosis. The presence of PI(4,5)P2 is necessary at endocytic cites. A specific structure as11.

(24) sociated with endocytosis is the eisosome, large complexes at the cell periphery containing the proteins Pil1, Lsp1, and Sur7 [110, 227].. Retrograde traffic Following endocytosis, membrane material continues from the early endosome either to the vacuole or potentially all the way to the ER. [24]. Endosome-to-Golgi recycling is of particular interest here, since we have also addressed it by suppressor screening in a ypt6 null mutant (paper II). The retrograde transport toward Golgi is of importance for retrieving membrane material and regulators, to be reused in anterograde traffic, e.x. α-factor maturation enzymes and the Vps10 receptor [39]. Retrieval from the endosome requires as yet unknown components of a multimeric coat complex [99, 171]. In a genetic screen for mutations that cause growth defects in combination with a mutant clathrin heavy chain, Bensen et al. [16] found an inactive allele of Ric1. It was also shown that the SNAREs Ykt6 and Sed5, involved in transport to cis-Golgi (see above), were able to alleviate the effects of both Ric1 and Ypt6 deficiencies [16]. The protein complex Ric1/Rgp1 serves as GEF for Ypt6 at the Golgi [210], while Gyp6 is a specific GAP [250]. It was finally confirmed that Ypt6 is involved in retrograde Golgi transport using also a temperature sensitive Ypt6 allele [143].. 12.

(25) Lipids in Saccharomyces cerevisiae. Yeast lipidomics is a comparatively new research field, offering the advantages of yeast as a genetics model to explore the pathways and regulation of lipid synthesis [68]. Thousands of lipid species exist in mammals [50]; in contrast, the yeast lipidome is simpler and the existing collections of yeast mutants allow addressing directly the effects of lipid metabolism perturbations. Lipids in budding yeast include several derivatives of diacylglycerol, sphingolipids, and ergosterol, summarized in table 2. The major phospholipids are represented by PC, PI, PE and PS. The acyl chains of diacylglycerol containing lipids are most commonly palmitoleic, palmitic, oleic and stearic acid [200]. A few differences between yeast and animals should be noted: yeast sphingolipids contain inositol instead of choline, they lack polyunsaturated fatty acids, and the major sterol in yeast cells is ergosterol [68, 260].. Lipids in organelles Lipid composition of various membranous organelles is different, due to both the site of biosynthesis of the lipid species as well as the their uneven distribution with regard to their signaling qualities (fig. 4). Sub-cellular fractionation has been used in order to determine the amounts and distribution of different lipids [200, 260, 261]. The major lipid in yeast as well as in most eukaryotes is PC, other abundant lipid species are PE and PS (Table 2). A special place is reserved for phosphatidylinositol (PI), one of the most abundant lipid species, since inositol metabolism has a tremendous impact on the entire cell [68]. Phosphorylated variants of PI occur in much lower amounts than other lipids, however they play a disproportionately large role as signaling molecules determining the identity of organelles [15]. Apart from building up cellular membranes, lipids serve a storage function in yeast, as triglycerides accumulate in lipid droplets [190]. Lipid composition of yeast is characterized by stability upon constant conditions and the possibility to change with the environment [200, 241, 259, 261]. 13.

(26) Figure 4: Relative composition and sites of lipid synthesis in eukaryotic organelles. Standard lipid abbreviations are used, in addition: Ch - cholesterol, SM - sphingomyelin, BMP - bis-monoacylglycerol-phosphate, Erg - ergosterol, ISL - yeast inositol sphingolipid. Flags indicate location of signaling PIPs. After van Meer et al. [241].. Synthesis and distribution The biosynthesis of lipids in yeast cells occurs in several specialized compartments [164, 200, 261]. Ergosterol and ceramide synthesis are carried out mainly in the ER, while phospholipid production is more spread out, with PS being decarboxylated either in the mitochondria by Psd1, or in the Golgi by Psd2 [122, 229, 261]. Other organelles with a role in lipid biogenesis are the peroxisomes and lipid droplets. Phosphatidic acid (PA) is the precursor of three different phospholipid types, PI, PC and CL. The pathway of direct generation of PC by PA is known as the Kennedy pathway [114]. Phosphatidylserine is made directly from CDP-DAG and serine by the Cho1 PS-synthase [135]. Upon decarboxylation by Psd1 in the mitochondria or by Psd2 in the Golgi, it forms PE, which in turn can be methylated in the Golgi to yield PC via an alternative route to the Kennedy pathway. The distribution of lipids from their site of synthesis to various organelles in achieved by diffusion through the cytosol, through transport 14.

(27) Table 2: Phospholipid contents of budding yeast sub-cellular fractions, adapted from [259, 261].. % of total phospholipid Subcellular fraction. PC. PE. PI. PS. CL. PA. Others. Plasma membrane. 16.8. 20.3. 17.7. 33.6. 0.2. 3.9. 6.9. Secretory vesicles. 35.0. 22.3. 19.1. 12.9. 0.7. 1.2. 8.8. Golgi. 25.5. 24.5. 26.5. 23.5. –. –. –. Vacuoles. 46.5. 19.4. 18.3. 4.4. 1.6. 2.1. 7.7. Nucleus. 44.6. 26.9. 15.1. 5.9. <1.0. 2.2. 4.3. Peroxisomes. 48.2. 22.9. 15.8. 4.5. 7.0. 1.6. –. Light microsomes. 51.3. 33.4. 7.5. 6.6. 0.4. 0.2. 0.5. Mitochondria. 40.2. 26.5. 14.6. 3.0. 13.3. 2.4. –. of membrane material along the secretory pathways, or by specific proteins [211]. Cholesterol is capable of flipping between the leaflets of one membrane and of diffusing between membranes as a single molecule [14]. PITPs such as Sec14 transfer PI and PC molecules inside an internal cavity [158]. Ceramide travels with the cytoplasmic protein CERT [85]. Sphingolipids and sterols leave the Golgi after segregating from the phospholipids in detergent-resistant microdomains [239]. Lipids segregate not only among organelles, but also between the two leaflets of the same membrane. The most well known case is PS, greatly enriched in the inner leaflet of the plasma membrane. This asymmetry is energy requiring and is maintained by the plasma membrane aminophospholipid translocase at the expense of ATP hydrolysis [206]. Finally, lateral segregation of lipids into lipid rafts is yet another means of increasing membrane complexity with major functional implications [9].. Regulation of synthesis On the level of metabolites, lipid biogenesis is strongly influenced by the presence of inositol and choline in the media. When supplied with inositol, cells produce significantly more PI and less PS [113]. This response is related to the increase in transcription level of INO1, the gene for inositol-3-phosphate synthase [49]. INO1 is co regulated together with a set of genes with a common UASI NO element in their promoter regions. Loss-of-function ino1 mutants are auxotrophic for inositol [41]. In contrast, cells overexpressing INO1 excrete inositol in the growth medium. 15.

(28) It was recently shown that Opi1, a PA binding protein, is a repressor of INO1 [140].. Phosphoinositides The hydroxyl groups of the inositol ring (fig. 5) can be modified by covalent addition of a phosphate group, catalysed by various kinases [33]. The phosphorylation derivatives of PI are collectively known as phosphoinositides and have emerged in the last decades as major cellular regulators in eukaryotes. Through specific interaction with a number of protein domains, recently review by Lemmon [133], phosphoinositides have been established as the most important lipid species in membrane-related processes.. Figure 5: Phosphatidylinositol. The hydroxyl positions on the ring are numbered D1 to D6. Common modifications are phosphorylation at positions D3, D4 and D5.. Phosphatidylinositol-3-phosphate, PI(3)P, is generated from PI by the Vps34 kinase [201] and is necessary for protein sorting from late Golgi to the vacuole and for various forms of autophagy [116]. Phosphatidylinositol-4-phosphate, PI(4)P, is synthesized at the Golgi and plasma membrane by the PI 4-kinases Stt4 and Pik1, and to a lesser extent by Lsb6 [61, 84, 255]. Diphosphorylated phosphoinositides are represented by PI(4,5)P2 , generated by MSS4 at the plasma membrane [256], and PI(3,5)P2 , synthesized by Fab1 at the vacuole [252]. PI phosphorylation is reversible and therefore the presence of phosphoinositides is dependent also on the activity of phosphatases. S. cerevisiae has seven distinct phosphoinositide phosphatases, recently reviewed by Strahl & Thorner [216]. There are three types of activities associated with phosphoinositide phosphatases, carried out by the Sac1-like domain [175], by enzymes removing the D5 phosphate [247], or by YMR1 [221]. The three synaptojanin-like phosphatases Inp51, Inp52 and Inp53 have both a Sac1-like domain and an inositol-5-phosphatase 16.

(29) domain, and in general most phosphatases have partially overlapping functions in the cell [216]. Interestingly, all yeast synaptojanin-like phosphatases possess a poly-proline motif known to be the target of the GYF domain, as explained below. However, no potential interactions in this regard have been investigated so far, except for an in vitro binding assay with synthetic peptides [119]. Therefore, a putative link between synaptojanin-like phosphatases and GYF domain proteins remains to be elucidated.. Lipids and membrane traffic The first strong link of lipid biogenesis and vesicular traffic was provided by the PITP protein Sec14, necessary for vesicle export from the Golgi [1, 11, 158, 215]. It was shown that the presence of PI(4)P is necessary for secretion, and that the defect associated with lack of Sec14 could be relieved by either deleting the Sac1 phosphatase or overexpressing the Pik1 kinase [83], essentially two alternative means of up-regulating the amount of PI(4)P at the Golgi membrane (see above). Another example of the need for specific lipids for traffic is the requirement for PI(4,5)P2 at the plasma membrane for endocytic internalization of ion channels and clathrin dynamics [218, 243, 262]. Similarly, depletion of PI(3)P from endosomal compartments caused block of endosomal trafficking [59]. The likeliest mechanism by which PIP depletion affects traffic is by removing the “landmarks” specific for different organelles, thus preventing effector protein binding [15, 240]. In addition to the signaling role played by the minor lipid species, the more abundant lipids are also important for various steps in traffic. For instance, PA, produced by the action of phospholipase D upon PC, has been shown to promote fusion of secretory vesicles with the plasma membrane [101]. PE was recently found to be required for vacuolar protein targeting [165]. Another recent study followed the intracellular distribution of PS with a fluorescent biosensor, to find it enriched in endosomal and lysosomal compartments, with the potential to attract there fusogenic effectors [253]. Ceramide, produced by hydrolysis of sphingomyelin, was required for formation of exosomes, likely for its conical shape, inducing spontaneous negative curvature at the endosome membrane [228]. Several lines of evidence implicate sterols in targeting to the vacuole or the plasma membrane [9, 189, 233]. In a more indirect fashion, cholesterol was also suggested to affect cellular polarization through the formation of PI(4,5)P2 enriched lipid domains [74]. In the majority of the above examples, it is the biophysical properties of the lipid species involved that cause the observed effects, rather than effectors recognizing the lipids.. 17.

(30)

(31) Protein-lipid interactions. Membrane binding proteins The current view of the lipid bilayer membrane is formed after decades of studies, from the discovery that the lipids are arranged in two monolayers [76], to the detailed molecular models in modern simulations [156]. The fluid mosaic model of Singer and Nickolson [209] is an important milestone, since it grouped for the first time the membrane associated proteins into integral and peripheral [157]. Unlike proteins that span the lipid bilayer and therefore are an integral part of it, many others are capable of peripheral association with membranes (fig 6). This makes possible transient interactions subject to regulation by modification of both lipid bilayer and protein.. Figure 6: The lipid bilayer with associated proteins. Typical strategies for membrane attachment are illustrated. A, lipid anchors; B, insertion of an amphiphillic helix; C, electrostatic interactions with anionic lipids; D, specific lipid head group recognition.. Mechanisms of membrane binding Covalent lipid modifications Lipidation is common for the Ras-related small GTPases [86]. Members of the Rab family are post-translationally prenylated at the conserved CxC or CC motif at their C-terminus [115]. In complex with an escort protein, the 19.

(32) Rabs are modified by a type II geranylgeranyltransferase by addition of two geranylgeranyl group to two C-terminal cysteines in the CC or CXC motif [56], a modification that is essential for their function. Another trafficking regulator modified by lipidation is the SNARE Ykt6 [150].. Amphipathic helices Several groups of proteins employ amphipathic helices for membrane association. The crystal structure of MurG, member of the large family of glycosyltranferases, suggested an amphipathic helix as a likely membrane association site [82]. A combined structural/biophysical study of analogous helix from the Acholeplasma laidlawii MGS verified its membrane-binding capacity [139]. Another example for controlled membrane association by similar structure are the GTPases of the ARF family (as described previously, yeast Sar1 and Arf1 are involved in control of vesicle budding). The mechanism of membrane insertion of the helix upon activation by GTP binding has been elucidated with the help of existing crystal structures of several ARF proteins (fig. 7).. Figure 7: Interswitch toggle mechanism for Arf activation (after Pasqualato et al. [184]). Upon binding, the γ−phosphate of GTP displaces a conserved Asp residue, causing a shift in the interswitch and making the amphipathic N-terminal helix available for membrane association. Arf6-GDP (left) and Arf6-GTPγS (right) are shown in the same orientation.. Electrostatic interactions Positive charge clusters in proteins cause commonly membrane association via electrostatic attraction to the negatively charged anionic lipid heads. Well studied examples are the BAR-domain proteins and protein kinase C (PKC). BAR was named after the proteins where it 20.

(33) originally was found (bin/Amphiphysin/Rvs161-167), which are needed for the final stage of vesicle formation, scission of the membrane [66]. They are characterized by a concave shape stabilizing membrane curvature upon binding; the latter is dominated by electrostatic interactions with negatively charged lipids such as PS and PI. The C2 domain of PKC-α, also found in phospholipases, binds to PS containing membranes in the presence of Ca2+ [20]. Interestingly, a negatively charged amino acid residue (Asp55) and the bridging role of Ca2+ are required for this binding.. Specific lipid head recognition The first mechanism of binding described above requires no special lipids in the bilayer, while the next two generally depend on any negative charge in the membrane. However, a large number of protein possess specialized structural motifs targeting specific lipid head groups, as recently reviewed by Lemmon [133]. The first example is the C1 domain from PKC (mentioned earlier regarding its C2 domain), identified as binding phorbol esters and DAG (reviewed by Colon-Gonzalez & Kazanietz [38]). It is a zinc-finger domain of about 50 amino acids. For efficient binding, it requires not only the specific head group but also an acidic membrane environment [112]. Another example are PH domains, named after a region of homology to pleckstrin, a substrate of PKC in platelets [90]. They bind to the PI(4,5)P2 lipid, but also to its free head-group, inositol-1,4,5-triphosphate [67, 134]. Following the solution of a crystal structure of substrate-bound PH domain [58], more than 150 articles in the past 20 years describe different aspects and occurrences of PH domains, thus underlining their importance as effectors of lipid second-messengers. The phosphoinositide binding site of PH domains is characterized by the KXn(K/R)XR motif in which the basic side chains form the interactions with phosphate groups [104]. FYVE domains (an acronym for Fab1, YOTB/ZK632.12, Vac1, and EEA1) recognize PI(3)P through another type of zinc-finger and the RR/KHHCR basic motif [213]. Unlike PH domains, FYVE prefer membrane inserted PI(3)P to the free I(1,3)P2 head-group for binding [124]. Importantly, FYVE domain binding efficiency is increased by dimerization, allowing for binding of multiple PI(3)P targets in the membrane. In the case of the EEA1 protein, the dimerization is achieved by coiled-coil domains [53]. Sorting nexins, important in membrane trafficking [99, 172], have a type of PI(3)P selective domain known as PX domain, reviewed by Seet & Hong [203]. PROPPINs are β−propeller proteins that bind PI(3,5)P2 [152]. A final example are the ENTH (and related ANTH) domains, found in the epsin N-terminal homology region of clathrin adapter proteins [63]. They bind PI(4,5)P2 , however with low affinity. 21.

(34) Lipid binding and vesicular traffic All protein regulators of membrane trafficking require some mechanism of interaction, or association, with the membrane. A few are integral membrane proteins, most notably the SNAREs, which usually have a single C-terminal membrane spanning domain (19 out of 22 in yeast, plus one lipidated). The majority of the trafficking regulators though are peripherally attached to the membrane, often transiently and in equilibrium with a cytosolic pool. Some examples given previously include the two GTPase families, ARF and RAB; their GEFs; their effectors, multimolecular tethering complexes such as TRAPP, HOPS and CORVET; sorting nexins and amphiphysin. It should already have become clear that the lipids in the organellar membranes play a substantial role in the peripheral attachment of the above listed proteins. Thus, a combination of protein-lipid and protein-protein interactions drives the correct assembly of trafficking regulators at specific membrane sites. While this idea may seem transparent nowadays, it is in fact relatively young, with one of the first hints been published only a few years ago by Levine & Munro [136].. 22.

(35) Membrane stress. The biophysical notion of membrane stress is associated with the accumulated curvature stress as the formation of a bilayer constrains the non-bilayer prone lipids within a plane. This is separate from the physiological membrane stress described later.. Spontaneous curvature The planar arrangement of amphipathic lipid molecules minimizes the surface of the polar/non-polar interface, thus maximizing the internal contacts in the aqueous phase, and is therefore energetically profitable. This does however come at a price, since the shape that the lipid molecules assume within a bilayer does not always guarantee a tight arrangement. The most abundant phospholipid in eukaryotes, PC, has a nearly cylindrical shape, however other common lipids such as PE, CL, and some glycolipids, resemble a cone1 and are therefore non-bilayer prone (fig. 8). This results in a spontaneous tendency of the monolayers to curve, suppressed inside the bilayer formed of two monolayers with opposite topology [93]. Thus, the presence of non-bilayer lipids in the membrane causes accumulation of curvature stress.. Figure 8: Lipid shapes and the corresponding spontaneous monolayer curvature. Membrane curvature is also related to the lateral pressure along the normal of the bilayer [93, 146, 223]. The term lateral pressure is used to describe the forces within the plane of the bilayer. These forces are different at 1 The term “inverted” cone has been used to describe the shape of lysolipids. Since the math-. ematical definition of a cone does not imply an orientation in space, this term is regarded here as confusing and is therefore avoided.. 23.

(36) different depths, leading to the characteristic pressure profiles of bilayers [146, 223]. The intrinsic curvature and lateral pressure are being actively modified by membrane proteins (recently reviewed by Ces & Mulet [34], McMahon & Gallop [149], Zimmerberg & Kozlov [258]). One line of research has elucidated the control of membrane homeostasis in Acholeplasma laidlawii by maintaining constant spontaneous curvature [180]. This regulation can be achieved by controlling the activity of lipid synthesizing enzymes dependent on curvature [42, 244]. Recently, effects on membrane curvature by lipid modifications have been shown to be relevant to eukaryotes as well, in a study on yeast depleted from PC [25]. A certain amount of stored curvature stress is essential for membrane function. This stress becomes effective in events of membrane fusion or fission, such as those occurring during cell division or membrane traffic. One example is the membrane intercalation of an amphipathic helix by the Sar1 GTPase during vesicle formation at the ER, described previously [184]. This process introduces locally positive curvature, which is essential for membrane budding [18, 130]. Reciprocally, after GTP hydrolysis and dissociation of Sar1, the membrane maintains a certain amount of curvature stress, which is relieved upon fusion with the target membrane [6, 130]. The energetics of membrane binding and bending by proteins determine their function as either modulators of membrane curvature of merely sensors [258]. When the energy of attachment of a protein (such as one with a BAR domain) is high, it is sufficient to modify the shape of the membrane to follow that of the protein surface. On the other hand, when the free energy released by the interaction is low, the protein will only bind if the membrane already has a suitable curvature, thereby the association event will be equivalent to sensing the membrane curvature. Thus, the role of such peripheral proteins is determined by their affinity to specific membranes relative to the rigidity of the membranes. Other mechanisms by which proteins modify membrane curvature include pulling by cytoskeletal elements or motor proteins and oligomerization of asymmetric transmembrane proteins [149]. As proteins can shape membranes, the opposite is also true – membrane curvature and lateral pressure profiles affect the binding, conformation and activity of some proteins, reviewed by Marsh [146], van den Brink-van der Laan et al. [237].. Physiological membrane stress In the context of this work, membrane stress is broadly defined as any nonphysiological change in the biophysical properties of biological membranes affecting the function of membrane-associated proteins. Possible causes 24.

(37) may include altered bilayer composition, the effects of small, often amphipathic molecules, or deviation from a normal protein-to-lipid ratio.. Examples in eukaryotes and prokaryotes Examples of membrane stress caused by changed composition are found in bacterial strains in which the lipid biosynthetic pathways have been disrupted [45, 91, 155]. Escherichia coli defective in PG synthesis grows slowly and requires sucrose and Mg++ supplemented medium [155]. Strains of E. coli with inactive pss gene lack the major membrane phospholipid PE and are characterized by divalent cation dependence and slow growth [45]. In addition, electron transfer and transmembrane transport functions are affected in these strains [22, 153]. Interestingly, similar defects were not observed in Bacillus subtilis lacking PE due to disruption of PS decarboxylase, however glucosyldiacylglycerol levels increased two- to fourfold in this organism [147]. Engineering of lipid composition in eukaryotes is more challenging, since multiple biosynthesis pathways often exist (see above). PE was found to be essential for growth of yeast [19]. Strains substantially depleted for PE demonstrated strong negative effects on amino acid transporters [194]. These effects have been attributed to defective protein secretion in the complete absence of PE [177]. Phospholipid composition has direct implication for membrane protein structure and topology. In agreement with the "positive-inside" rule [245], the anionic phospholipids PG and CL were shown as determinants of transmembrane protein orientation in E. coli [238]. A zwitterionic lipid, PE, was also necessary for the correct organization of the γ−aminobutyric acid permease [257]. Some small molecules also have the capacity to cause membrane stress. Alcohols and detergents have an effect on lipid packing and the gel – liquid crystalline phase transition temperature [248]. Trifluoroethanol and other small alcohols destabilize the KcsA tetramer by inserting in the headgroup region of the lipid bilayer, resulting in a decreased lateral pressure in the acyl chain region [235, 236]. The antipsychotic drug chlorpromazine has been used to cause effects similar to blocking of membrane traffic after inserting in the membrane interface and subsequent "stretching" of the lipid bilayer [46]. The mechanism of action of chlorpromazine relied on binding to negatively charged phospholipids and disrupting association of traffic-regulator proteins with the membrane [44].. 25.

(38) Importance of membrane stress for homeostasis in unicellular organisms As illustrated by the above examples, physiological membrane stress is common to unicellular organisms. They are more likely to be exposed to effects of small molecules or to rapid changes in their environment. Therefore, it can be expected that different mechanisms may have evolved to deal with this stress, similar to the those controlling channel opening by mechanosensation [5]. One such possible mechanism is described in more detail in the next chapter.. 26.

(39) Translational control, RNA decay and stress. Cells have evolved numerous coordinated reactions to changes in the environment. Some of these reactions are on the level of single molecules, where the activity of an enzyme or a transporter changes together with alterations of metabolite concentrations or transmembrane potentials. Other reactions involve signaling cascades of multiple molecules, and result in orchestrated responses on a large scale. Here, I will briefly outline the set of events that connect sensing of stimuli including starvation and membrane stress, to a rapid stall in protein translation and potentially degradation of mRNAs.. Translation initiation Upon export from the cell nucleus, mRNA has already undergone extensive processing, most notably it has been spliced to remove non-coding regions (introns), a polyadenine tail has been added to its 3´terminus, and a cap structure has been added at its 5’ end [132]. Messenger RNAs are complexed with proteins that maintain their transcriptional competence, such as the poly(A) binding protein Pab1 [197].. Figure 9: Translation initiation in eukaryotes - 48S complex (after Lehninger et al. [132]). See text for details.. Translation initiation in eukaryotes begins by association of the cap-binding protein complex eIF4F with the cap structure of mRNA. eIF4F consists of the cap-binding protein eIF4E, the RNA helicase eIF4A, and the scaffolding protein eIF4G [98]. The latter mediates the binding of mRNA to the 43S pre-initiation complex by interacting with eIF3 (fig. 9). After binding, the 40S ribosomal subunit scans the mRNA to find the first 27.

(40) AUG codon, followed by dissociation of the initiation factors, binding of the 60S ribosomal subunit, and the beginning of the elongation phase of translation [71, 98, 132]. One regulatory step of translation initiation involves the inhibition of eIF4E-eIF4F complex formation by the 4E binding proteins (4E-BPs) [40, 71]. These proteins compete with eIF4G for binding to eIF4E which necessarily precedes eIF4E-eIF4F binding. Phosphorylation of 4E-BPs in response to growth factors and as a result of the TOR PI-kinases (see below) decreases their affinity for eIF4E and thus promotes translation initiation [71]. Two 4E-BP proteins in yeast are Caf20 and Eap1 [2, 40]. Eap1 in particular was identified by a far-Western blotting assay for eIF4E interactors and was shown to be a second 4E-BP in S. cerevisiae; its disruption leads to temperature sensitive growth and partial resistance to the macrolide antibiotic rapamycin [40].. RNA degradation The three types of ribonucleic acid (RNA) in cells have the same overall primary structure, but generally different sizes and functions. Transfer RNAs are small molecules with specific secondary structures that serve as adapters during protein synthesis in the ribosome. Ribosomal RNAs are larger, and together with the ribosomal proteins fulfill structural and catalytic roles. The most variable type of RNA are the messenger RNAs, transcripts of protein-coding genes with sizes roughly corresponding to the size of the proteins they encode [132]. As all molecules in the cell, RNAs are turned over via constant production of new species and destruction of old ones. Messenger RNAs are subject to special regulation, since they are directly responsible for the protein repertoire that the cell can synthesize. It has been of particular interest to find out what determines mRNA stability, and different factors have been considered in the past [96, 198, 199]. Currently, there is good knowledge on several pathways of mRNA degradation [151, 208]. The degradation of RNAs carrying premature termination codons by the nonsense-mediated decay, as well as the miRNA mediated RNA decay (both recently reviewed by Shyu et al. [208]) are more specialized mechanisms outside the scope of this work. Instead, I will focus below on the more general aspects of RNA degradation. The main exonuclease in yeast acting in the 5’→3’ direction is Xrn1 (Kem1), a conserved enzyme homologous to pacman in Drosophila [107, 159, 166]. In the opposite direction, 3’→5’, degradation is catalysed by the exosome, a large multiprotein complex [154]. However, since none of these enzymes has endonuclease activity, to digest RNA both Xrn1 and the exosome need access to decapped or de-adenylated mRNA respectively. 28.

(41) Figure 10: Pathways of mRNA degradation, adapted from Newbury [167], Parker & Song [183].. Depending on the order of events during mRNA degradation, three major pathways exist [151, 167, 183] (fig. 10). One starts with removal of the cap structure from mRNA, catalysed by the decapping enzymes Dcp1 and Dcp2 [54, 126]. Decapping is also controlled by Dhh1 [60] and the cytoplasmic heptameric complex of LSm1-7 proteins [225]. Decapped RNA can be digested 5’→3’ by Xrn1. Another possibility is to initially de-adenylate the mRNA by adenylase complexes such as those containing Ccr4 and Pop2 [231, 232], which would render it a substrate for the exosome in the 3’→5’ direction. Finally, cleavage of mRNA by endonucleases would generate unprotected 5’ and 3’ ends and result in sensitivity of the fragments to at least one of the above degradation pathways [167].. Processing bodies The site of mRNA degradation in the cytoplasm was shown to be organized in foci containing the respective enzymes and named processing bodies, or P-bodies [207]. In addition to mRNAs, they contain the proteins Dcp1, Dcp2, LSm1-7, Edc3, Scd6, Pat1, Dhh1, Ccr4, Pop2, Not1-5, Ski7, Puf3 and Xrn1 [182, 207]. Importantly, P-bodies are dynamic structures that rapidly change their composition and appearance upon conditions promoting mRNA degradation [28, 207, 222]. In yeast, P-bodies have been shown to form upon stimuli such as glucose limitations and amino acid starvation, osmotic stress and ultraviolet light. It has been demonstrated that defective translation initiation leads to increased de-adenylation and decapping in S. cerevisiae strains with mutant eIF4G, eIF4E and eIF4A [202]. On the other hand, activation of decapping by Dhh1 and Pat1 is required for repression of translation during glucose deprivation [37]. Thus, a functional connection exists between translation and degradation of mRNA. This connection is manifested also in the re29.

(42) Figure 11: mRNA shuttles between polysomes and P-bodies, adapted from Newbury [167].. lation between the corresponding structures (fig. 11), since mRNA moves from the polysomal complexes into P-bodies upon stimulation [28]. However, movement in the opposite direction was also observed, meaning that mRNA can be stored rather than irreversibly degraded in the P-body [28]. Very recently, polyadenylated mRNAs and the translation initiation regulators Pab1, eIF4E, and eIF4G were also detected inside processing bodies [27]. A general view emerging from the interactions between translation initiation and mRNA decay is that the two processes are in competition for engaging with the mRNA [182] (fig. 11). Extrinsic factors, such as different types of stress or growth stimuli, can shift the equilibrium towards either process. Intrinsic factors such as mRNA defects will trigger irreversible degradation and fulfill the quality control function of the mRNA decay machinery.. Nutrient deprivation and the TOR signaling pathway The signals used routinely in the laboratory to trigger P-body assembly are different types of nutrient deprivation [207]. Owing to the interplay between mRNA translation and decay as described above, an interrupted translation initiation step is a likely event preceding P-body formation. Therefore, the question stands how the signal for nutrient limitation is transferred to the protein synthesis machinery in the cell. One well studied route for that signaling goes through the target of rapamycin (TOR) kinases [151, 219]. TOR protein kinases were first identified in S. cerevisiae as the targets of the antifungal agent rapamycin. While addition of rapamycin inhibits protein synthesis and promotes autophagy and glycogen accumulation, the TOR kinases have the opposite effect, i.e. stimulating growth, translation and ribosome biogenesis, as recently reviewed by [195]. 30.

(43) In S. cerevisiae, two complexes exist (TORC1 and TORC2) [141]. They involve some common components and the highly similar Tor1 and Tor2 kinases. TOR complexes often localize to cellular membranes, which is important for their signaling function [195]. Signaling through the Tor pathway is branched, with several targets [195, 219]. One outcome is stimulation of the dephosphorylation of eIF2α by the Sit4 phosphatase [36]. Another is through phosphorylation of the 4EBP proteins which competitively inhibit translation initiation (see above). Adding onto the complexity of connections between TOR signaling and intracellular membranes, a recent study showed a requirement for functional vacuolar protein sorting complexes for efficient TOR signaling [263]. It is becoming evident that signals coming from the cell membrane through the TOR pathway could converge upon regulation of mRNA translation and stability. One important example is the work by Deloche et al. [46] on attenuation of translation initiation in membrane traffic mutants and in cells treated with chlorpromazine (see above). The authors observed a rapid response to shifting to non-permissive temperature of several mutants at different steps of membrane transport. Polysomal complexes were disassembled within 5-10 minutes and a peak of 80S ribosomal subunits was observed [46]. The kinetics of the response indicated it was too fast to be a result of the earlier reported transcriptional repression of rRNA, tRNA and ribosomal protein genes in secretion defective cells [137, 169]. Eventually, it was shown that the TOR pathway acts during membrane stress, since (i) a gcn2∆ mutant was resistant to chlorpromazine treatment; (ii) eIF2α was increasingly more phosphorylated starting at 5 minutes after chlorpromazine treatment; and (iii) the deletion of the 4E-BP Eap1 resulted in only minor redistribution of polysomes during chlorpromazine treatment.. 31.

(44)

(45) The GYF domain. The GYF domain, a key feature in the proteins subject of this study, was discovered ten years ago in the CD2BP2 protein as responsible for binding the repeated motif PPPGHR from the T cell adhesion molecule CD2 [170]. A conserved consensus sequence, W X Y X 6−11GP F X 4 M X 2W X 3GY F , was found in various eukaryotic proteomes [64]. Two structural variants of the GYF domain, the CD2BP2-type and the SMY2-type, differ by the length of the first loop and by the residue found at position 8 of the CD2BP2-type GYF domain [64, 120].. Figure 12: Peptide ligand bound to GYF domain from Arabidopsis thaliana GYN4 (PDB 1l2z).. Structural features The GYF domain is a small fold with an α−helix packed against a β−sheet [120]. The proline residues of the ligand form a PPII type helix. Conserved aromatic residues from the domain are responsible for contacts with the ligand. The structural difference between the two types of GYF domains, W/D at position 8, determines the major difference in substrate specificity, a preference for a tryptophan after the prolines/glycine in CD2BP2-type domains, as opposed to a preference for an aliphatic residue and against tryptophan in SMY2-type domains [120]. Another minor structural difference 33.

(46) determined a bias against negative charge and for longer proline repeats for ligands of SMY2-GYF as opposed to the same type MYR1-GYF [70].. GYF domain protein families Both structural types of GYF domains are present in numerous eukaryotic proteins, although their expansion is not as wide as that of similar domains such as WW or EVH1 domains [10, 120]. Examples of the CD2BP2-type are CD2BP2, the first described instance of a GYF-domain protein, involved in the immune response [170], GYN4 from A. thaliana (fig. 12), and Lin1 from budding yeast. Importantly, representatives of this type of GYF domain are spliceosome-associated proteins, as shown by mass spectroscopy, and are readily detectable in the nucleus by fluorescence [17, 118, 125]. The SMY2-type GYF domain is found in Myr1 and Smy2, as well as in homologous proteins from other yeasts. Another studied, albeit with an yet unknown function, group of proteins with a SMY2-type GYF includes the mouse GIGYF1/2 proteins and their human homologs PERQ1/2 [72]. GIGYF1 was described as an interactor of the Grb10 protein, involved in insulin-receptor signaling [72]. Interestingly, the corresponding proline-rich sequence in the human Grb10 does not conform to the sequence requirement for recognition by the GYF domain, which casts a shadow of uncertainty over the relevance of the GIGYF-Grb10 interaction in mice [120]. Very recently, human GIGYF2 was found in a study of familial Parkinson Disease (PD) on 249 patients [127]. Seven different GIGYF2 missense mutations resulting in single amino acid substitutions were present in 12 unrelated PD patients (4.8%) and not in controls; the authors concluded that GIGYF2 is likely a gene with a causal role in familial PD [127].. Figure 13: Examples of GYF domain proteins. While the region of high identity between yeast and mammalian SMY2type GYF domain proteins is limited to the GYF domain itself, additional features such as a coiled-coil domain are present in both Myr1 and Smy2, 34.

(47) as well as in GIGYF/PERQ (fig. 13). However, when considering the possibility of a potential distant relation between these proteins, it is worthwhile to mention that the couple MYR1/SMY2 was shortlisted among 17 “core yeast genes” with homologs in other fungal genomes, but without known homologs in other organisms [100].. Binding partners Yeast Lin1 was identified as interacting with the yeast cohesin complex [17]. Thereafter, it was assigned a role linking chromosome segregation and premRNA splicing [17]. An earlier study using mass spectroscopy had already detected Lin1 in a complex with the U4/U6.U5 small nuclear ribonucleoproteins from S. cerevisiae [214]. Subsequently, a splicing related function was suggested also for the homologous CD2BP2 through binding to the SmB/B’ core splicing component [92, 118]. Eventually CD2BP2 was also found as a constituent of the evolutionary conserved U4/U6.U5 complex [125]. It is currently accepted that Lin1 and CD2BP2 are homologs which both serve auxiliary functions in the formation but not in the activity of the U4/U6.U5 complex, through interactions either with the PRP8 protein or with SmB/B’ [17, 92, 118]. A substantial body of data acquired by using the yeast two-hybrid system (Y2H) has associated SMY2 and MYR1 to mRNA splicing as well [65, 242, 246]. Nevertheless, no functional implications for splicing have been related to these genes so far. While mass spectroscopy of purified cellular components is a generally accepted method of dissecting protein complexes, Y2H studies are notoriously prone to giving false positives. The nuclear localization of most splicing factors is also inconsistent with the demonstrated cytosolic localizations of Myr1 and Smy2 [69, 70, 97, 103]. It is likely that binding of SMY2-type GYF domains to various splicing proteins is merely due to a coincidence of their liberal ligand requirements and artificially created opportunities (in vitro, Y2H), rather than a meaningful functional interaction. A different line of evidence connected MYR1 and SMY2 to cytoplasmlocalized degradation of mRNA. GYF domains from Smy2 and Myr1 were shown to bind strongly to Eap1 [119], an inhibitor of translation initiation [40] (see above). In addition, the capacity to bind a number of factors associated with mRNA decay was shown [70, 119]. Finally, colocalization with P-bodies supported possible interactions of Myr1 and Smy2 with the mRNA decay machinery (Paper I). A very recent study [35] describing a connection of Myo2 to processing bodies may turn out to be highly relevant in this context, see below. Apart from interactions with Eap1, Myr1 and Smy2 have been shown to interact with other translation initiation factors and with polysomal com35.

(48) plexes, by tandem affinity purification [62, 121] and Y2H [105]. Considering the tight functional relations existing between control of translation initiation and mRNA decay, it is tempting to speculate that GYF domain proteins may be the long-sought structural link between the two processes.. Membrane traffic connections Functional data on SMY2-type GYF domain proteins have revealed another aspect of their cellular roles. Smy2 was initially identified and named as a suppressor of the myo2-66 mutant [138]. Myo2 and Myo4 are the two yeast type V myosin motors, and Myo2 is required for the polarized delivery of secretory vesicles by actin-based transport [29]. Thus, the genetic interaction with Smy2 suggested for the first time that Smy2 may be related to vesicular traffic as well. This relation was only recently confirmed by the group of Akihiko Nakano at RIKEN, who described Smy2 as a suppressor of a mutant of the Sec24 component of COPII coats [97]. Myr1 was also linked to membrane traffic by the finding that it is synthetic lethal with Ric1, part of Ypt6 GEF (see above) [226]. Indeed, Ric1 is synthetic lethal with another 146 genes, which reduces the statistical importance of this finding for Ric1, however Myr1 has only been reported for a synthetic phenotype with Ric1. Our own discovery of Myr1 among the ypt6∆ suppressors along with the subsequent finding that Myr1 can rescue ric1∆ confirmed the relevance of this interaction (Paper II). In conclusion, the independent findings that both Smy2 and Myr1 are linked to membrane traffic are corroborating evidence that these interactions are physiologically meaningful.. Relation to RNA metabolism A common theme in GYF domain protein binding partners is the relation to RNA metabolism. Both structural types of GYF domains present in yeast, the CD2BP2-type (Lin1), and the SMY2-type (Myr1 and Smy2) seem to mediate interactions with pre-mRNA splicing and mRNA decay respectively. Thus, an ancient cellular role for all GYF domain containing proteins may well be the regulation of mRNA processing. Indeed, one of the major points of this thesis is that Myr1 and Smy2 affect the fate of mRNA in response to membrane stress caused by trafficking defects.. 36.

No results found