Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Molecular and genetic analyses of complexes and pathways for retrograde transport to the Golgi Quenneville, Nicole R. 2009

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2009_spring_quenneville_nicole.pdf [ 14.32MB ]
Metadata
JSON: 24-1.0067127.json
JSON-LD: 24-1.0067127-ld.json
RDF/XML (Pretty): 24-1.0067127-rdf.xml
RDF/JSON: 24-1.0067127-rdf.json
Turtle: 24-1.0067127-turtle.txt
N-Triples: 24-1.0067127-rdf-ntriples.txt
Original Record: 24-1.0067127-source.json
Full Text
24-1.0067127-fulltext.txt
Citation
24-1.0067127.ris

Full Text

Molecular and Genetic Analyses of Complexes and Pathways for Retrograde Transport to the Golgi  by  Nicole R. Quenneville  B.Sc. University of British Columbia, 2002  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  DOCTOR OF PHILOSOPHY  in (Biochemistry and Molecular Biology)  THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)  April 2009  © Nicole R. Quenneville, 2009  Abstract Proteins and lipids are selectively transported between the Golgi, plasma membrane and endosomes by a network of vesicle-mediated endosomal transport pathways. Trafficking specificity requires the coordination of multiple protein assemblies and signals of compartment identity. Genetic screens, and molecular and biochemical techniques, have revealed many components for endosomal transport, but questions regarding the mechanisms of specificity and the coordination of trafficking pathways remain. The Golgi Associated Retrograde Protein (GARP) complex is required to tether vesicles derived from multiple types of endosomes with the Golgi. In the absence of GARP, retrograde transport from endosomes to the Golgi is abolished, and numerous cargoes are missorted. Mutation of the GARP subunit Vps54 causes motor neuron disease in the mouse, emphasizing the physiological importance of GARP. Tethering requires recognition of multiple membranes, but how GARP recognizes vesicles derived from multiple upstream compartments is not known. In my first body of work, the function of the GARP subunit Vps54 was addressed. The Nterminal portion of Vps54 was found to be important for GARP complex assembly and stability, while the C-terminal portion localized to a compartment with features of an early endosome. In the absence of the C-terminal domain, retrieval of early endosome cargo became dependent on late-endosome retrograde transport. This body of work supports the model that tethers recognize, and possibly distinguish between, upstream compartments. The machinery involved in retrograde transport from endosomes is not fully understood. In my second body of work, genes involved endosomal transport were systematically identified by screening mutant collections with a reporter of early endosome dysfunction. To evaluate the relationships between genes and pathways discovered in this screen, genetic interaction analyses with two phenotypes, growth and endosomal dysfunction, were performed. An analysis of genetic interactions based on trafficking dysfunction revealed interesting genetic relationships between endosomal coat proteins and their regulators. This body of work provides insight into the relationships between endosomal transport pathways and presents a framework to discover relationships between genes and pathways discovered in a genomic screen. Together, this thesis presents a molecular and pathway perspective of endosomal transport that provides insight into pathway specificity and the relationships between pathway components.  ii  Table of Contents Abstract ......................................................................................................................................... ii  List of Tables................................................................................................................................ vi  List of Figures ............................................................................................................................. vii  List of Abbreviations.................................................................................................................... viii  Glossary ....................................................................................................................................... ix  Acknowledgements ...................................................................................................................... xi  Dedication ................................................................................................................................... xii  Co-Authorship Statement ............................................................................................................ xiii  Chapter 1: Introduction ................................................................................................................. 1  1.1 An Overview of Intracellular Transport and Vesicle Transport Machinery........................... 1  1.1.1 Specificity of intracellular transport pathways and organelle identity ............................ 1  1.1.2 Vesicle formation and cargo selection .......................................................................... 4  1.1.3 Tethering and Fusion: role of vesicle tethering complexes and SNARE proteins......... 6  1.2 Endosomal and Retrograde Transport Pathways ............................................................... 8  1.2.1 Retrograde transport from early and late endosomes .................................................. 8  1.2.2 Transport to the degradative compartment: the MVB pathway .................................. 10  1.2.3 Recycling from endosomes to the cell surface ........................................................... 11  1.3 Trafficking Dysfunction and Neurodegenerative Disease ................................................. 12  1.3.1 Involvement of endosomes in neurodegenerative diseases ....................................... 12  1.3.2 Amyotrophic Lateral Sclerosis and the wobbler mouse .............................................. 13  1.4 Genetic Interaction Analysis to Study Retrograde Transport ............................................ 14  1.4.1 Yeast as a model of mammalian trafficking pathways ................................................ 14  1.4.2 Genetic screens: tools for discovery of pathway components .................................... 14  1.4.3 Synthetic genetic analysis and epistasis mapping in yeast ........................................ 16  1.4.4 Genetic interaction analysis in other systems ............................................................. 18  1.4.5 Definitions of genetic interaction ................................................................................. 18  1.4.6 Networks of genetic interaction ................................................................................... 20  1.5 Thesis Summary ............................................................................................................... 21  1.6 References ........................................................................................................................ 23  Chapter 2: Domains within the GARP subunit Vps54 confer separate functions in complex assembly and early endosome recognition ................................................................................. 36  2.1 Preface .............................................................................................................................. 36  2.2 Introduction........................................................................................................................ 37  2.3 Results .............................................................................................................................. 39  iii  2.3.1 The N-terminal domain of Vps54 is sufficient for assembly and stability of the GARP complex ............................................................................................................................... 39  2.3.2 Truncation of Vps54 causes a specific block in early endosome recycling ................ 40  2.3.3 The C-terminal domain of Vps54 localizes to a polarized intracellular compartment . 43  2.3.4 Vps54-C localizes to an early endocytic compartment ............................................... 46  2.3.5 Point mutations within the C-terminal region of Vps54 prevent localization to the polarized early endosome and block early endosome recycling ......................................... 50  2.3.6 GARP is not required for endocytic recycling of the a-factor receptor Ste3................ 53  2.4 Discussion ......................................................................................................................... 54  2.4.1 Assembly of the GARP complex ................................................................................. 55  2.4.2 Recruitment of GARP to upstream compartments ..................................................... 56  2.4.3 Early endosome recycling pathways .......................................................................... 57  2.5 Revision post-publication .................................................................................................. 58  2.6 Materials and Methods ...................................................................................................... 58  2.7 References ........................................................................................................................ 65  Chapter 3: Genetic interaction analyses of trafficking dysfunction highlights relationships between endosomal transport pathways .................................................................................... 70  3.1 Preface .............................................................................................................................. 70  3.2 Background ....................................................................................................................... 71  3.3 Results & Discussion ......................................................................................................... 73  3.3.1 Discovery of retrograde transport factors ................................................................... 73  3.3.2 Genetic interaction profiling highlights putative alleviating genetic interactions ......... 77  3.3.3 Quantitative analysis of genetic interactions recapitulates known genetic interactions with respect to growth and informs pathway dependencies ................................................ 79  3.3.4 Comparing genetic interactions as defined by growth versus trafficking dysfunction . 81  3.3.5 Sub-classification of alleviating interactions with respect to trafficking dysfunction reveals novel masking relationships .................................................................................... 83  3.3.6 Stimulation of recycling pathways may underlie suppressive interactions.................. 86  3.4 Conclusion......................................................................................................................... 90  3.5 Materials and Methods ...................................................................................................... 91  3.6 References ........................................................................................................................ 99  Chapter 4: Discussion and Conclusions ................................................................................... 104  4.1 Summary of Findings ...................................................................................................... 104  4.2 Implications and Significance .......................................................................................... 105  4.2.1 Recruitment of tethers by vesicle coats - How is GARP recruited to membranes? .. 105  4.2.2 Recruitment of tethers by lipid domains - How is GARP recruited to membranes?.. 108  4.2.3 Relationship between GARP and ARL pathway ....................................................... 109  4.2.4 Comparison of yeast and mammalian GARP ........................................................... 111  iv  4.2.5 Role of Vps54 in Amyotrophic Lateral Sclerosis (ALS) ............................................. 112  4.2.6 Trafficking pathway relationships and recycling to the cell surface .......................... 113  4.2.7 Systematic screens of genetic interaction ................................................................ 115  4.3 Future Directions ............................................................................................................. 119  4.3.1 GARP function and mechanisms of membrane recognition ..................................... 119  4.3.2 Cell surface recycling ................................................................................................ 119  4.3.3 Genetic interaction analyses ..................................................................................... 120  4.4 Concluding Statement ..................................................................................................... 121  4.5 References ...................................................................................................................... 122  Appendix A: Erratum ................................................................................................................. 128  Appendix B: Supplemental Figures to Chapter 3 ...................................................................... 129  Appendix C: Meisler et al. (2008) Evaluation of the Golgi trafficking protein VPS54 (wobbler) as a candidate for ALS .................................................................................................................. 132   v  List of Tables Table 2.1 Yeast strains used in Chapter 2 ................................................................................. 61  Table 3.1 Yeast strains used in Chapter 3 ................................................................................. 91   vi  List of Figures Figure 1.1 Schematic of endosomal transport pathways in yeast ................................................ 1  Figure 1.2 Schematic of endosomal transport pathways in mammalian cells .............................. 2  Figure 1.3 Gga proteins simultaneously recognize Arf1-GTP and PI4P, illustrating an example of coincidence detection at the Golgi ............................................................................................ 5  Figure 1.4 Recycling pathways to the cell surface ..................................................................... 12  Figure 1.5 Quantitative reporter of endosomal trafficking dysfunction ....................................... 16  Figure 2.1 The N-terminus of Vps54 is important for GARP complex assembly and stability.... 40  Figure 2.2 C-terminally truncated Vps54 is functional in late endosome to TGN transport, but defective in sorting from early endosomes ................................................................................. 42  Figure 2.3 Differential localization of N- and C-terminal domains of Vps54 ............................... 44  Figure 2.4 The C-terminal domain of Vps54 localizes to sites of polarized growth throughout the cell cycle ..................................................................................................................................... 45  Figure 2.5 Vps54 C-terminus co-localizes with early endocytic markers but not markers of late Golgi or secretory vesicles .......................................................................................................... 47  Figure 2.6 Morphology of Vps54-C expressing cells by electron microscopy ............................ 49  Figure 2.7 Conserved residues in the C-terminus of Vps54 are required for localization to the polarized, early endocytic structure ............................................................................................ 50  Figure 2.8 Point mutations in the C-terminal Vps54 domain abolish sorting from early endosomes without affecting late endosome to TGN transport .................................................. 52  Figure 2.9 Vps54 is not required for recycling of Ste3∆365-GFP .............................................. 54  Figure 2.10 Model of Vps54 function in retrograde transport from early endosomes ................ 55  Figure 3.1 The GFP-SNC1-SUC2 GSS reporter is a model protein for quantitative phenotypic analysis of endosomal transport ................................................................................................. 74  Figure 3.2 Non-essential retrograde transport factors are identified by screens of GSS localization .................................................................................................................................. 76  Figure 3.3 Genetic interaction profiling reveals putative alleviating interactions with retrograde transport queries ......................................................................................................................... 78  Figure 3.4 Genetic interaction analysis as defined by growth discovers known and novel interactions, highlighting pathway interdependencies ................................................................. 81  Figure 3.5 Alleviating GSS genetic interactions can be mapped into multiple modes of interaction ................................................................................................................................... 82  Figure 3.6 Common masking, partial masking and suppressive genes are identified by network analysis ....................................................................................................................................... 84  Figure 3.7 Deletion of NHX1, VPS30 or VPS38 results in efficient recycling to the plasma membrane ................................................................................................................................... 88  Figure 4.1 Schematic model of endosomal transport and cell surface recycling in yeast ........ 110  Figure A.1 Snc1 sorting requires late endosome recycling in cells expressing Vps54-N as the sole copy of Vps54 .................................................................................................................... 128  Figure B.1 Network of genetic interactions based on growth ................................................... 129  Figure B.2 Double mutants are hyper-efficient for recycling Fur4-GFP ................................... 130  Figure B.3 Double mutants are hypersensitive to the toxic uracil analogue 5-FU ................... 131   vii  List of Abbreviations ALS ARF ARL COG CPY EE E-MAP ER ESCRT GAE GAT GARP GDP GTP LE MPR MVB RE SE SGA SGD SNARE TfnR TGN VHS VPS  Amyotrophic Lateral Sclerosis ADP ribosylation factor Arf-like Conserved oligomeric Golgi complex Carboxy-peptidase Y Early endosome Epistasis miniarray profile Endoplasmic reticulum Endosomal sorting complex required for transport γ-adaptin ear domain GGA and TOM (target of myb) domain. Golgi associated retrograde protein Guanosine diphosphate Guanosine triphosphate Late endosome Mannose-6-phosphate receptor (CI-MPR: cation independent) Multi vesicular body Recycling endosome Sorting endosome Synthetic genetic analysis Saccharomyces genome database (www.yeastgenome.org/) Soluble NSF(N-ethylmaleimide sensitive factor) associated receptor Transferrin receptor Trans Golgi network Vps27 (vacuolar protein sorting 27), Hrs (hepatocyte-growth-factorreceptor substrate) and Stam (signal-transducing adaptor molecule) domain. Vacuolar protein sorting  viii  Glossary Adaptor Proteins  Proteins that interact with cargo, membrane features and coat proteins. This term usually refers to the multisubunit clathrin adaptor proteins AP-1, AP-2, and AP-3.  Aggravating Interaction  A type of genetic interaction wherein a double mutant phenotype is worse off than expected. For example synthetic sick or lethal.  Alleviating Interaction  A broad category of genetic interaction wherein a double mutant phenotype is better off than expected. An alleviating interaction can reflect different types of genetic interaction, for example coequal, masking and suppressive interactions are all types of alleviating interactions.  Anterograde  The direction of traffic for secretion (ER to Golgi to plasma membrane), or for degradation after endocytosis (plasma membrane to endosomes to vacuole/lysosome).  Arf  GTPases of the Ras superfamily that are distinguished by an N-terminal extension that includes an amphipathic helix and site for N-myristoylation. Membrane association is stabilized when in the active GTP-bound form.  Arl  Family of Arf-like GTPases. Some members of the Arl family are N-terminally myristoylated and regulated like Arf, while others are N-acetylated to facilitate membrane association.  Coat Proteins  Proteins that interact to deform a membrane and form a transport carrier.  Coequal  Two single mutant phenotypes are very similar to each other, and to the phenotype of their double mutant progeny.  Dubious ORF  A questionable, unconfirmed open reading frame (ORF).  GAP  Guanine Activation Protein - facilitates the hydrolysis of GTP to form GDP for members of Rab, Arf and Arl families.  GEF  Guanine Exchange Factor - regulates the switch from GDP to GTP for members of Rab, Arf and Arl families.  Genetic Interaction  When the combination of two mutant alleles results in a double mutant phenotype that significantly differs from what might be expected if there is no interaction between the two genes.  Gga  A type of conserved monomeric clathrin adaptor that has domains to interact with clathrin, cargo, membranes and accessory proteins.  Masking  A genetic interaction wherein a double mutant phenotype is more similar to one parental single mutant phenotype than another.  ix  Quatrefoil  Refers to the fourfold nature shared by a group of related tethering complexes that includes GARP, COG and EXOCYST.  Query  A yeast strain of a genetic background that permits and eases construction of further strains by synthetic genetic analysis (SGA).  Rab/Ypt  Ras-like GTPases that are peripherally associated with membranes via Cterminal prenylation. Rab proteins are inactive in their GDP bound state and active when bound to GTP.  Retrograde  Trafficking pathways that oppose anterograde transport (Ex. endosomes to Golgi).  Retromer  Conserved sorting nexin complex involved in endosome to Golgi transport of mannose-6-phosphate receptors (homolog Vps10) and other cargoes from the late endosome.  Snx4  Conserved sorting nexin involved in endosome to Golgi transport. In yeast Snx4 forms a complex with Snx41 and Snx42 and is required for the retrieval of Snc1 from endosomes.  Suppressive  In this work, a suppressive interaction refers to a double mutant phenotype that is unlike either parental single mutant phenotype, and deviates in the direction of the wild-type phenotype.  Synthetic Genetic Analysis (SGA)  In SGA analysis, an array of single deletion mutants is crossed with a query strain to produce progeny that will contain deletions or mutations in two alleles: one is specified by the query mutation, and one by the array mutation.  x  Acknowledgements I am grateful to my colleagues and friends who have lent their support, guidance, and expertise during the course of this work. I would like to thank Dr. Liz Conibear for providing her mentorship and expertise. Her attentiveness to my studies and enthusiasm to discuss new findings and ideas were encouraging and instrumental. I am also thankful to Liz for her continued support as I pursued my numerous scientific interests. I’d like to thank members of the Conibear lab, Helen Burston, Mike Davey, Karen Lam, Cayetana Schluter, Chris Tam, Les Grad, Cindy Chao and Ben Montpetit for their help and camaraderie. Mike Davey is also acknowledged for his assistance with robotics and many invertase assays. I thank T.Y Chao for her initial observation of “Cindy Blobs”, and work with the GARP complex. I thank Chris Tam and Ben Montpetit for sharing their insights and observations. I am very thankful to Helen Burston for sharing countless thoughtful discussions and laughs in the lab. I’d like to acknowledge John Aitchison and colleagues at the Institute for Systems Biology for the opportunity to collaborate. Greg Carter is acknowledged for his help, expertise and advice in analyses of genetic interaction. I’d like to thank Ramsey Saleem and Jennifer Smith for their assistance and advice with protein purification and gradient fractionation. I thank Jenny Bryan and Rick White for their expertise, advice and contribution to our statistical analyses, and Beverly Wendland, for assistance with electron microscopy. I am grateful to Peter Novick, Ben Glick, Jeffery Gerst, Daniel Urban-Grimal and Charlie Boone for providing reagents. I also thank my supervisory committee, Lawrence McIntosh, Phil Hieter and John Aitchison for their guidance and support throughout my PhD studies, and their careful reading of this thesis. I am indebted to my family for their continual support and interest throughout my academic studies. Finally, I am grateful to the Department of Biochemistry and Molecular Biology, the Child and Family Research Institute, and the Michael Smith Foundation for Health Research for Graduate scholarship funding.  xi  Dedication  For my parents, thank you for all your support.   xii  Co-Authorship Statement  Chapters 2, 3 and Appendix C are co-authored work. In Chapter 2, I performed and analyzed research shown in all Figures with the following exceptions. Immuno-precipitations, Western blots and FACs analysis shown in Figures 2.1 and 2.4 were preformed by T.Y. Chao. Electron microscopy shown in Figure 2.6 was performed by M.J. McCaffrey. Identification and design of the research, manuscript preparation and data analyses were performed in collaboration with Dr. Elizabeth Conibear. For Chapter 3, my contributions included experimental identification and design, data collection and analyses and manuscript preparation. Data from the initial genetic screen and genetic interaction profiling (Figures 3.2 and 3.3) was collected by Mike Davey. Data normalization, scores and modes of genetic interaction were computed by Rick White. Data analyses and manuscript preparation were performed in collaboration with Drs. Jenny Bryan and Elizabeth Conibear.  Appendix C includes a re-print of “Meisler et al. (2008) Evaluation of the Golgi trafficking protein VPS54 (wobbler) as a candidate for ALS. Amyotrophic Lateral Sclerosis 9(3): 141” [1]. I tested the functionality of the rare VPS54 variant in yeast (Appendix C, Figure 3)  xiii  Chapter 1: Introduction 1.1 An Overview of Intracellular Transport and Vesicle Transport Machinery 1.1.1 Specificity of intracellular transport pathways and organelle identity  Vesicle transport mediates cargo delivery while maintaining the identity and structure of each organelle. Trafficking pathways are highly conserved from yeast, Saccharomyces cerevisiae, to mammalian cells enabling studies of analogous pathways and organelles across systems. Newly synthesized proteins targeted for secretion or for delivery to the lysosome (vacuole in yeast), are first transported from the endoplasmic reticulum (ER) to the cis-Golgi (yeast early Golgi). Upon delivery to the trans-Golgi network (TGN; yeast late Golgi) cargo proteins are selectively incorporated into one of many different types of vesicles targeted for delivery to either the plasma membrane, endosomes or directly to the vacuole (Figures 1.1, 1.2). Opposing these anterograde transport pathways, are retrograde transport pathways that can return proteins to their intended residence, helping to maintain organelle function and identity [2, 3]. Retrograde transport is also used by some endocytosed proteins; rather than being targeted for degradation, these proteins are selectively retrieved from endosomes and are delivered to the late-Golgi.  Figure 1.1 Schematic of endosomal transport pathways in yeast. Anterograde and retrograde transport pathways connecting the Golgi, plasma membrane, endosomes and vacuole are shown with arrows. Trafficking pathways used by the receptor for the vacuolar hydrolase CPY (Vps10) and by the v-SNARE Snc1 are coloured blue and green respectively. Retrograde 1  transport to the Golgi occurs from early endosomes (EE) or late endosomes (LE). Vps10 recycles between the Golgi and late endosome (blue arrows). Anterograde transport to the lateendosome is mediated by Gga coat proteins, and retrograde transport is mediated by the retromer complex. The secretory vesicle SNARE Snc1 recycles between the Golgi, plasma membrane and early endosomes (green arrows). Retrograde transport of Snc1 from early endosomes is mediated by the Snx4/41/42 complex. The GARP complex tethers vesicles derived from either early or late endosomes with the late Golgi.  Figure 1.2 Schematic of endosomal transport pathways in mammalian cells. Trafficking pathways connecting the trans-Golgi network, endosomes, plasma membrane and lysosome are depicted with arrows and compartments that are related by maturation are depicted with dotted lines. Endocytosed proteins are first delivered to the sorting endosome (SE) which sorts proteins to be recycled into tubules or directly back to the surface, and retains proteins for lysosomal degradation in the body of the endosome. Some tubules become a tubular vesicular compartment referred to as the recycling endosome (RE). From the recycling endosome, proteins can be recycled to the plasma membrane (eg. transferrin receptor (TfnR)) or retrieved via retrograde transport to the TGN (eg. TGN46 or Shiga Toxin). As the body of the SE accumulates inward budding vesicles (mediated by activity of ESCRT) it matures into the lateendosome (LE). Retromer retrieves cargo (such as MPRs) from the SE or LE for retrograde transport to the TGN. Like Vps10 in yeast, MPRs recycle between the TGN and endosomes (SE or LE) by Gga and retromer dependant pathways (blue arrows). The GARP complex is important for retrograde transport of multiple cargoes, suggesting it tethers multiple types of vesicles with the TGN in mammalian cells. In each trafficking pathway, selective transport is achieved by the combination of sorting signals exhibited by cargoes and recognition of these signals by coat and adaptor proteins (reviewed in [2, 4, 5]). Sorting signals are often short, linear motifs present in the cytosolic portion of transmembrane cargoes, but can also include post-translational modifications such as 2  ubiquitination or phosphorylation (reviewed in [6-8]). Coat proteins are recruited from the cytosol, and their assembly into a supramolecular complex helps to drive vesicle formation. Adaptor proteins provide a link between the cargo, membrane and vesicle coat, and together the assembly of coat and adaptor complexes deforms the membrane to form a vesicle (reviewed in [2, 4, 9]). In addition to recognition of cargo, coat and adaptor complexes also recognize features of the organelle membrane, such as regulatory GTPases and phosphatidylinositol species (reviewed in [10]). GTPases of the Ras superfamily, including Arf and Arf-like (Arl) and Rab (yeast Ypt) families are involved in various anterograde and retrograde trafficking pathways (reviewed in [11, 12]). In each case, the GTPase cycle is regulated by the activity of a GTP exchange factor (GEF) and a GTPase activation protein (GAP). GEFs promote exchange of the GTPase from an inactive GDP to the active GTP bound state, and GAPs promote subsequent GTP hydrolysis. The minor lipid phosphatidylinositol (PI) is phosphorylated and dephosphorylated in a spatially regulated manner at positions 3, 4, and 5 of the inositol ring to produce specific PI species on organelle membranes (reviewed in [10, 13]). For instance, PI(4,5)P2 is enriched on the inner leaflet of the plasma membrane, PI3P on the cytosolic face of endosomes, PI(3,5)P2 on the vacuole, and PI4P on the Golgi. Many proteins involved in sorting, including adaptors, contain domains that recognize these PI species with varying levels of specificity: pleckstrin homology (PH); phox homology (PX); Fab1, YotB, Vac1, EEA1 (FYVE); epsin amino terminal homology (ENTH) and AP180 amino-terminal holomology (ANTH) domains. The restricted localizations of PI species however, do not fully explain the restricted localizations of proteins containing PI binding domains. For instance, sorting nexins such as Snx4 and subunits of the retromer complex both contain PX domains that enable binding to PI3P at endosomes, but each binds and functions at distinct endosomal compartments [14, 15]. The coincidence detection hypothesis suggests that sorting proteins recognize a combination of determinants simultaneously including PI species, activated GTPases, cargo signals, other lipids and even membrane curvature [13]. Hence, the sorting of cargo involves the coordinated action and regulation of multiple determinants of specificity. While evidence suggests specificity is achieved by combinatorial, multivalent interactions, the precise composition of determinants and machinery for each intracellular trafficking pathway is not fully defined. Despite intricate mechanisms of specificity, in the absence of sorting machinery, cargoes can sometimes reach their intended destinations by alternative or bypass pathways. In the yeast 3  secretory system, two types of secretory vesicles of differing densities are found to exit the Golgi [16]. A dense vesicle type contains a specific subset of cargoes and requires clathrin for its formation. In clathrin mutants, cargoes that are normally transported in dense vesicles are instead present in light secretory vesicles [16, 17]. In another example, Pep12, a soluble Nethylmaleimide-sensitive attachment factor receptor (SNARE) protein, is normally sorted from the Golgi directly to the late endosome due to its linear sorting signal (FSD) and the sorting proteins Gga1 and Gga2 [18]. In gga1∆gga2∆ double mutants, Pep12 is instead delivered to the early endosome, from where it can ultimately reach the late endosome [18]. Intracellular trafficking pathways are indeed selective but the interconnected nature of both pathways and organelles can bring about surprising consequences in the face of mutation. The relationships between intracellular trafficking pathways and mechanisms that enable such bypass routes are not well understood. 1.1.2 Vesicle formation and cargo selection  A vesicle transport pathway begins when cargo is sequestered into regions of a membrane by the action of coats. Coats can take a number of different forms and though specific for a particular pathway, components of coats can be shared. Three better understood coat assemblies include COPII and COPI, which function in vesicle transport between the ER and Golgi respectively, and clathrin coats (reviewed in [2, 4]). The well known clathrin coat is formed by triskelions of clathrin heavy and light chains and, as an example of a shared coat component, is involved in vesicle budding at multiple compartments including the TGN, plasma membrane and endosomes [4]. Clathrin itself does not contain domains for association with membranes. Instead, clathrin associates with adaptor protein complexes which, as outlined above, recognize cargo, clathrin coat proteins, and features of the organelle membrane [4]. The conserved adaptor protein (AP) complexes AP-1, AP-2 and AP-3 function in unique trafficking pathways but share a common organization consisting of a core domain and two ear or appendage domains connected to the core by long unstructured hinge domains. Each of these domains mediates a subset of the protein or lipid interactions that the adaptor protein coordinates. Monomeric clathrin adaptors, such as the gamma-ear-containing Arf-binding (Gga) proteins, also exist in both yeast and mammalian cells [7, 19-22]. Similar to the multi-protein adaptor assemblies, Gga proteins have a modular organization and participate in similar types of interactions to transport cargo between the TGN and endosomes.  4  There are two forms of Gga in yeast (Gga1 and Gga2) which are thought to be largely redundant, and three forms of Gga in mammalian cells which may have unique sorting roles [7, 19-22]. Gga proteins consist of four domains: VHS, GAT, a hinge and GAE (reviewed in [7, 23]). The interactions observed for these domains serves as an example to illustrate the mechanisms by which adaptors modulate interactions with coats, cargo, regulatory proteins and the membrane. Many of the specific interactions mediated by these domains are conserved in both yeast and human forms, for instance interaction with clathrin is mediated by a variant of the clathrin box motif that is present in the hinge region of Gga [7]. Similarly, the GAT domain is involved in cargo recognition as it binds to ubiquitin moieties post-translationally added to cargo, and in membrane recognition as it binds to the Golgi localized GTPase Arf1 [20, 24-28]. In both yeast and mammalian cells Gga proteins also bind to PI4P, which is enriched in Golgi membranes, but this interaction occurs through the GAT domain in mammalian cells and the VHS domain in yeast [29, 30]. Binding of Arf1-GTP and PI4P can occur simultaneously, and in mammalian cells, binding of PI4P promotes the interaction with ubiquitin, illustrating an example of coincidence detection at the Golgi ([13, 29], Figure 1.3).  Figure 1.3 Gga proteins simultaneously recognize Arf1-GTP and PI4P, illustrating an example of coincidence detection at the Golgi. A. Gga protein domain organization and selected binding partners are indicated (adapted from [29]). In mammalian cells PI4P interacts with the GAT domain, and in yeast, interaction occurs with the VHS domain. B. Schematic depicting simultaneous interaction of Gga proteins with Arf1 and PI4P which are enriched at the Golgi. Arf1 is involved in many trafficking pathways, simultaneous interaction with PI4P and Arf1-GTP helps to confer the specificity of Gga interaction with the TGN or late-Golgi. In mammalian cells, 5  interaction with PI4P promotes interaction with ubiquitin present in cargo. Gga proteins also interact with ubiquitin in yeast. Gga also interacts with accessory proteins, such as Ent3, through its GAE domain, which are also involved in sorting cargo. Combinatorial physical interactions help to confer the specificity of trafficking pathways. The VHS and GAE domains are involved in cargo recognition. In mammalian cells, the VHS domain recognizes an acidic di-leucine (DXXLL) sorting signal that is present in the cytoplasmic tail of mannose 6-phosphate receptors (MPR) and other transmembrane receptors [31-34]. Yeast Gga proteins do not recognize DXXLL motifs, but like human Ggas they are also involved in sorting receptors (Vps10) [20, 21]. Finally, the GAE domain interacts with accessory proteins that are involved in cargo selection and membrane binding [7]. In yeast, two epsin related accessory proteins, Ent3 and Ent5, bind to Gga [35]. While Ent3 and Ent5 may partially substitute for each other’s function, each accessory protein shows a preferential role in vivo with Gga or AP-1 adaptor complexes respectively [36, 37]. Ent3 and Ent5 each contain a PI3P binding domain, ENTH and ANTH respectively [35, 38]. As mentioned in the previous section, Gga proteins are required for sorting the SNARE protein Pep12 from the late-Golgi to the late endosome [18]. This function is mediated by a linear FSD motif in the cytosolic N-terminus of Pep12 but an interaction between Gga and Pep12 could not be observed [18]. Interestingly, Ent3 was recently shown to bind the FSD motif of Pep12 and is responsible for the Gga mediated transport of Pep12 [39]. The cooperative action of Gga and Ent3 in sorting the cargo Pep12 from the Golgi again illustrates the combinatorial process of protein sorting. Although a role for Gga in anterograde transport from the Golgi to late endosomes is established, there is also evidence that Gga proteins may have a role at endosomes [7, 40, 41]. In both yeast and mammalian cells, Gga proteins have been localized to endosomes, and may play a role in the retrograde transport of cargo to the Golgi [40, 41]. While there is evidence of a putative role for Gga at endosomes, the mechanisms and significance of this role are unknown [7]. Mechanisms of protein sorting and vesicle formation clearly involve the action of multiple proteins, signals and regulatory factors. Ultimately a successful vesicle budding event occurs when the appropriate cargo is packaged with the appropriate machinery and signals, to ensure fusion with the correct downstream compartment. 1.1.3 Tethering and Fusion: role of vesicle tethering complexes and SNARE proteins  Subsequent to budding, the vesicle must recognize and fuse with the correct downstream compartment. Like cargo sorting and vesicle budding, tethering is a specific, 6  regulated event. The specificity of vesicle fusion is thought to arise by the combined action of tethering complexes, SNARE proteins and GTPases, predominantly of the Rab (Ypt) family. SNARE proteins mediate the fusion of opposed lipid bilayers when a vesicle SNARE (v-SNARE) forms a trans-SNARE complex with three cognate target SNAREs (t-SNAREs) in the appropriate downstream membrane ([42, 43] reviewed in [44]). SNARE proteins are typically small type II transmembrane proteins with large cytosolic N-terminal domains that interact with various regulatory proteins. SNARE proteins can confer some specificity in complex formation, but there is evidence to suggest that the specificity observed in vivo cannot be fully explained by the specificity of pairings observed in vitro [45-47]. Tethering complexes help to confer the specificity of membrane fusion, interacting with SNARE proteins and activated Rabs. Different types of tethers exist, including long coiled-coil tethers and multisubunit assemblies (reviewed in [46]). Coiled-coil type tethers mainly function at the Golgi, and in mammalian cells, at endosomes. Examples of these tethers include the essential protein Uso1 at the yeast early Golgi, and EEA1 at mammalian early endosomes; each is capable of forming long coiled-coil assemblies [48, 49]. Several conserved multisubunit tethering complexes also exist and these can be further grouped into those that are related on the basis of a conserved domain and are termed “quatrefoil”, and those that are not [46]. The quatrefoil complexes include the EXOCYST, a well studied essential tethering complex required for secretion, the conserved oligomeric Golgi (COG) complex, which acts in Golgi transport, and the Golgi associated retrograde protein (GARP) complex which acts at the late-Golgi to tether vesicles derived from endosomes [50-54]. Non-quatrefoil multi-subunit tethering complexes and their sites of activity include HOPS (vacuole), Dsl1 (ER), and TRAPP (Golgi) [46]. To enable tethering, a tethering complex must recognize features of both the originating upstream compartment and the downstream target compartment. At least one of these recognition events is mediated by interactions with Rab proteins, and in some cases the tether provides GEF activity, thereby activating the Rab [55]. Tethering complexes can also interact with SNARE proteins, but in the case of GARP, loss of this interaction does not affect membrane localization [53]. There is evidence that tethering complexes can interact with upstream compartments. In mammalian cells, subunits of EXOCYST, which exerts its tethering function at polarized sites of secretion, are localized to the TGN and recycling endosomes which both form secretory vesicles [56, 57]. The COG complex is, like the GARP complex, responsible for tethering vesicles derived from multiple upstream compartments. Subunits of the COG 7  complex have been shown to interact with coat components, and are also involved in protein sorting at the ER [58, 59]. Other tethering complexes including TRAPP and Dsl1 interact with coat components [60, 61]. Observations of tether-coat interactions prompt the question of timing; do tethers associate with coat proteins while the vesicle forms, or does a budded vesicle maintain its coat long enough to be recognized by a tether [9]? In the latter model, binding of tethering complexes could initiate coat disassembly [9]. Further investigations into the mechanisms by which tethering complexes are able to associate with specific subsets of membranes is an important question for understanding the specificity of trafficking pathways.  1.2 Endosomal and Retrograde Transport Pathways 1.2.1 Retrograde transport from early and late endosomes  The late-Golgi and endosomes are connected by anterograde and retrograde vesicle transport pathways (Figures 1.1, 1.2). Retrograde transport pathways can originate from early or late endosomes and are used to retrieve Golgi resident proteins, recycle vesicle transport machinery such as SNAREs, and to retrieve vacuolar hydrolase receptors after they have delivered their cargo (Figure 1.1). At the late endosome a conserved sorting complex called retromer acts to retrieve cargo such as the yeast carboxy-peptidase Y (CPY) receptor, Vps10 [62], or the mannose-6-phosphate receptors (MPR) in mammalian cells [63]. The retromer complex is a heteropentameric complex consisting of a dimer of sorting nexins (Vps5 and Vps17) and trimer of Vps26, Vps29 and Vps35 [62]. Sorting nexins contain the PI3P binding domain, phox-homology (PX). Localization of retromer requires the function of the PI3-kinase Vps34, as well as proteins that recruit Vps34 to endosomal membranes, Vps30 and Vps38 (complex II) [15, 64]. Recent studies of the structure and function of retromer support the model that the retromer complex sorts cargo into tubulo-vesicular structures that may lack a conventional outer coat [65, 66]. Retromer is responsible for sorting a diverse set of cargo, and while residues important for retrieval of Vps10 and MPRs have been discovered, the apparent lack of a consensus sequence may also suggest a unique type of cargo recognition [66]. Cargo proteins can also be retrieved from early endosomes, but the full composition of complexes involved is not as well defined. The secretory vesicle SNARE Snc1 is retrieved from early endosomes in a process that does not require retromer, but does require the sorting nexins Snx4, Snx41 and Snx42 ([14, 67], Figure 1.1). Like retromer, these sorting nexins also bind PI3P, which is enriched at endosomal membranes [14]. Other complexes involved in Snc1 8  retrieval from early endosomes include the Arf-GAP Gcs1, which physically interacts with Snx4, and components of the COPI coat [68]. Snx4, and a protein involved in recycling from endosomes Rcy1, also physically interact with Snc1, suggesting that all of these components may function together in the retrieval of Snc1 from early endosomes [14, 68-70]. Rcy1, contains a CAAX box for membrane localization and an F-box that mediates association with a ubiquitin ligase component, Skp1, and requires each of these domains for its role in recycling Snc1 [70, 71]. Also involved in early endosome to Golgi recycling is the amino phospholipid translocase complex Cdc50/Drs2 which maintains lipid asymmetry [72], and the GTPases Ypt31/32 [69, 73]. The mechanisms that coordinate activity of all of these factors have yet to be clearly defined. While cargo is selectively retrieved from early and late endosomes, there is some degree of shared machinery between the two compartments. For example, Snx4 is required for retrieval of Snc1 at early endosomes, but also forms a complex with a sorting protein Btn2 and components of retromer [74]. The Btn2 complex is required to retrieve the cargo Yif1 from late endosomes, but is not involved in the retrograde transport of many other tested cargo, including Snc1 [74]. Another sorting nexin, Grd19 (or Snx3), associates with retromer subunits to recycle the iron transporter Fet3-Ftr1 [75]. The capacity for retromer subunits to associate with other sorting proteins may provide a mechanism to expand the repertoire of cargo sorted by retromer from endosomes [5, 75]. The occurrence of shared components between unique trafficking pathways is a common theme. The GARP complex tethers vesicles that are derived from both early and late endosomes, and is therefore required for retrograde transport of Snc1, Vps10 and other retrograde transport cargo [52-54, 76]. The GARP complex (comprised of four subunits Vps52, Vps53, Vps54 and Vps51) interacts with the Golgi localized Rab protein Ypt6 which is required for GARPs steady state Golgi localization, but does not enable GARPs association with all membranes [53, 54]. The smallest GARP subunit, Vps51, interacts with the N-terminal domain of the endosomal and Golgi t-SNARE Tlg1 [53, 77]. In early models, this interaction was thought to play a regulatory role in SNARE complex formation, but more recently it was found that loss of the interaction does not affect trafficking function in vivo [53, 78]. GARP is also an effector of the Arf-like GTPase, Arl1, but the functional significance of this interaction has yet to be established [79]. Interestingly, it was recently shown that another Golgi tether in mammalian cells, Golgin GCC185, interacts with Rab6 and that this interaction promotes association of GCC185 with Arl1, providing a mutli-layered mechanism for Golgi localization [80]. The GARP 9  complex was recently characterized in mammalian cells and is found to have similar properties to the yeast GARP complex, although a homolog for Vps51 has not been discovered [81, 82]. Human GARP interacts with a homolog of Ypt6, Rab6a, is localized to the TGN and also interacts with a SNARE involved in retrograde transport, Syntaxin10 [81, 82]. Depletion of GARP subunits by RNA interference results in missorting of the hydrolase cathepsinD due to a lack of cation independent MPR (CI-MPR) recycling, and a lack of retrograde transport of other cargoes, establishing that the function of GARP is conserved [82]. Retrograde transport to the Golgi is critical for recycling cargo proteins such as vesicle transport machinery, maintaining the localization of Golgi resident proteins, and correct delivery of soluble vacuolar hydrolases to the vacuole. Many proteins involved in retrograde transport have been discovered, but the relationships between protein complexes and pathways are only partially understood [3]. There is much to be learned about retrograde transport in both yeast and mammalian systems. 1.2.2 Transport to the degradative compartment: the MVB pathway  At the plasma membrane, endocytosed proteins are targeted for degradation by the addition of ubiquitin, and are transported through endosomes enroute to the vacuole. The late endosome also receives biosynthetic cargo targeted for vacuolar delivery and thus acts as a point of intersection. At the late endosome, a unique type of vesicle with a topology opposite to that of other budding events is formed. As the late endosome matures, it accumulates these vesicles and is called the multivesicular body (MVB). In the last several years much has been learned about the machinery and mechanisms responsible for the inward budding of vesicles, which is mediated by the endosomal sorting complex required for transport (ESCRT) (reviewed in [83]). Initial identification of much of the ESCRT machinery was accomplished by phenotypic screens in yeast for mutants defective in vacuolar delivery (reviewed in [84]). Morphological classification of the vacuoles of these vacuolar protein sorting (vps) mutants discovered a unique class of mutants termed Class E, which accumulate a proteolytically active enlarged endosomal compartment [85]. In recent years, many structural, genetic and molecular analyses have revealed a mechanism whereby ubiquitinated cargo is recognized and sorted into inward budding vesicles by the sequential and/or cooperative action of three multisubunit ESCRT complexes I, II and III and an initial ESCRT complex, ESCRT-0 [83, 86]. Defects in any component of the ESCRT machinery prevents formation of the MVB, resulting in accumulation 10  of cargo in the Class E compartment. Under normal conditions however, the limiting membrane of the MVB fuses with the vacuole (or lysosome) enabling degradation of the intralumenal vesicles in the vacuole. 1.2.3 Recycling from endosomes to the cell surface  Some cargo recycles to the plasma membrane directly from endosomes, as opposed to using retrograde transport pathways that involve the Golgi (Figure 1.4). Harsay, Bretcher and Sheckman first described a possible route for such recycling in yeast with their discovery that subsets of biosynthetic cargo are sorted to either light or dense secretory vesicle populations, proposing that the dense population passes through endosomes [16, 17]. While it has not been shown that these are the same vesicle population, a subset of cargo in yeast does recycle to the plasma membrane directly from endosomes, including the receptor for the mating pheromone afactor, Ste3 and the uracil transporter Fur4 ([87, 88], Figure 1.4). In mammalian cells, recycling from endosomes to the cell surface is more commonly described, and can occur for proteins such as transferrin receptor from the recycling endosome [89]. The machinery involved in recycling from endosomes to the cell surface in yeast is not known. However, a truncation mutant of Sec5, a component of the secretory vesicle tethering complex EXOCYST, was found to be defective for recycling Ste3 [90]. Hyper-efficient recycling of Fur4 from endosomes to the cell surface occurs when transport through endosomes is blocked by deletion of the late endosome SNARE, Pep12, or by deletion of ESCRT machinery [88]. Fur4 recycling requires members of a late endosome and vacuole tethering complex, Vps Class C, but the significance of this is not understood [88]. Polarized plasma membrane localization of Snc1 was also found to be enhanced in ESCRT mutants and in a subset of COPI coat mutants proposed to be involved in numerous endosomal transport pathways, suggesting Snc1 may also be capable of recycling from late endosomes directly to the cell surface [68, 91].  11  Figure 1.4 Recycling pathways to the cell surface. Proteins can be recycled back to the plasma membrane by multiple routes in both yeast and mammalian cells. A. Cargo such as Snc1 in yeast, or TGN46 in mammalian cells, are retrieved from endosomes for delivery to the Golgi. Upon reaching the late Golgi, Snc1 is incorporated into newly forming secretory vesicles, and is thus recycled for another round of secretion. B. Other cargoes, such as Fur4 and Ste3 in yeast, or transferrin receptor in mammalian cells, are recycled from endosomes directly to the surface. Multiple types of endosomes (sorting and recycling endosomes) are capable of recycling proteins directly to the plasma membrane without transiting the Golgi. In this simplified depiction both types of endosomes are grouped together.  1.3 Trafficking Dysfunction and Neurodegenerative Disease 1.3.1 Involvement of endosomes in neurodegenerative diseases  Endosomal dysfunction is recognized as a common feature of several neurological diseases, including Alzheimer’s Disease, Downs syndrome, Neimann-Pick Type C and Huntington’s Disease [92, 93]. Alterations in endosome morphology and function have been noted as early events in the progression of Alzheimer’s Disease [92]. These endosomal defects may lead to elevated toxic production of amyloid-beta peptide (Aβ), or lead to Aβ independent mechanisms of toxicity, such as alterations in signalling pathways that involve endosomes [92, 94]. Aβ is produced by transmembrane protease cleavage of amyloid precursor protein (APP) by β-and γ-secretase at least in part at endosomes [66, 92, 94]. Defects in the retromer complex, which is important for retrieving cargo from endosomes to the TGN, have recently been implicated in Alzheimer’s disease. The retromer complex may play a role in retrieving the sortilin receptor (sorlA) and its cargo APP away from endosomes, reducing the production of Aβ [66, 94]. Knock-down of retromer components in cell culture accelerates the production of Aβ, and mutations in the receptor sorLA are associated with late onset Alzheimer’s disease [94]. 12  Endosomal dysfunction is increasingly implicated in inherited and sporadic forms of various neurological diseases, and therefore a deep understanding of the machinery and regulation of endosomal transport has important implications [92-95]. 1.3.2 Amyotrophic Lateral Sclerosis and the wobbler mouse  Mutations in the GARP tethering complex were implicated in Amyotrophic Lateral Sclerosis (ALS) when the cause of motor neuron disease in the wobbler (wr) mouse, a fifty-year old model of ALS resulting from a spontaneous autosomal recessive mutation, was mapped to VPS54 [96]. ALS, or Lou Gehrig’s disease, is a fatal disease characterized by progressive and selective loss of motor neurons in the brain and spinal cord. The wobbler mouse carries a missense mutation encoded in the C-terminal portion of Vps54 (L967Q). Transgenic expression of wildtype VPS54 completely rescued motor neuron disease of the wobbler mutant (wr/wr), establishing that the L976Q missense mutation is causative of disease in the mouse [96]. This finding prompted efforts to sequence human VPS54 in sporadic and familial ALS patients (Appendix C), and future work to sequence other GARP subunits. The vast majority of ALS cases are sporadic (90-95%), with no known Mendelian genetic component, emphasizing the complex genetics of this disease [97]. Insight into the possible pathways involved in sporadic ALS may be revealed by studying the molecular defects caused by rare familial mutations. Approximately 20% of familial cases of ALS are attributed to mutations in superoxide dismutase (SOD1) (reviewed in [97]). Four other Mendelian mutations have been discovered that can cause ALS, and three of these are involved in vesicle transport: alsin (ALS2), synaptobrevin/VAMP (vesicle associated membrane protein)-associated protein B (VAPB), and dynactin [97-100]. Alsin contains three guanine nuculeotide exchange factor (GEF) domains, and is an exchange factor for the endosomal Rab protein Rab5 [101]. VAPB is named as such because of a yeast-two-hybrid interaction with the vesicle SNARE synaptobrevin [97]. Recent characterization of VAPB suggests involvement in signalling and transcription [102, 103]. The dynein/dynactin complex is important for the retrograde movement of proteins, neurofilaments, vesicles and organelles along microtubules of the axon (reviewed in [95]). Defects in retrograde transport are advanced as one of the many possible mechanisms that can lead to selective loss of motor neurons [95, 97]. In support of this, reduced retrograde axonal transport is apparent in both ALS patients and mouse models, and can be observed early in disease progression. 13  Many pathways have been implicated in the progression of ALS, including mitochondrial dysfunction, protein aggregation, neuronal excitotoxicity as well as effects due to non-neuronal cell-types [97]. While selective loss of motor neurons in ALS can likely occur due to multiple mechanisms, aberrations in axonal transport can affect multiple pathways, including mitochondrial function, protein aggregation/degradation, and delivery of neurotrophic factors [95], suggesting a deeper understanding of factors involved in retrograde transport is an important facet of understanding and developing therapies for this disease.  1.4 Genetic Interaction Analysis to Study Retrograde Transport 1.4.1 Yeast as a model of mammalian trafficking pathways  In mammalian cells, endosomal transport involves multiple types of endosomes, with characteristic morphologies or intracellular locations, that are related to each other through compartment maturation ([89], Figure 1.2). The conservation of trafficking machinery at every stage of vesicular transport is very high [104], and models of endosome organization in yeast are thought to encompass much of the complexities of mammalian cells, including compartment maturation, specialized endosomal compartments and recycling pathways from endosomes to the cell surface [3, 5]. Our understanding of the sites of action and functional properties of many types of endosomal trafficking machinery is only partially complete in either yeast or mammalian cells. As endosomal transport in yeast reflects much of the complexity of mammalian endosomal transport, it presents an excellent model for genetic and biochemical analyses and has been used in trafficking studies for decades. Studies of conserved protein machineries in yeast are often directly applicable to mammalian cells, wherein whole protein complexes and their interaction partners are often conserved. Discovery of trafficking machinery and their pathway relationships in yeast can thus inform relationships between gene products in mammalian systems. 1.4.2 Genetic screens: tools for discovery of pathway components  Genetic screens are a well-established, commonly used approach to evaluate gene function for a wide variety of cellular pathways. In yeast, genetic screens have been immensely successful for discovering genes involved in the secretory, Golgi, endosomal and vacuolar 14  transport pathways [85, 105-110]. Characterization of mutants discovered in these screens have revealed roles in cargo sorting, vesicle budding, tethering and fusion, emphasizing the power of screens to discover genes with direct roles in trafficking [84]. Screens can be designed to focus on a particular trafficking pathway by measuring mislocalization of unique cargo. For example, genes involved in retrograde transport from early or late endosomes can be distinguished based on localization of the v-SNARE Snc1, and secretion of the vacuolar hydrolase carboxypeptidase Y (CPY), respectively (Figure 1.1). The receptor for CPY, Vps10, constantly recycles between the Golgi and late endosome [111]. While in the Golgi, the luminal C-terminal portion of Vps10 binds newly synthesized precursor or pro-CPY and together, CPY and its receptor are packaged into vesicles targeted for the late endosome. In the acidic environment of the late endosome, CPY dissociates from its receptor and is delivered to the vacuole where it matures and functions, while Vps10 is retrieved by retrograde transport from the late endosome to the Golgi. In the absence of its receptor, proCPY is aberrantly packaged into secretory vesicles and is secreted into the extra-cellular space. In one simple method to detect secreted pro-CPY, yeast colonies are grown on top of a nitrocellulose membrane, washed away, and secreted CPY is detected by Western blot. Mutants that missort Vps10 also secrete CPY to varying extents as is seen for vps10∆ strains. Measuring the degree of CPY missorting has been a well used screening approach identifying many genes involved in endosomal and vacuolar transport [106-108, 110, 112, 113]. Localization of the v-SNARE Snc1 can be used to detect defects in retrograde transport from early endosomes to the Golgi. Snc1 is cyclically transported between the Golgi, plasma membrane and endosomes and facilitates fusion at both the plasma membrane and Golgi. In wild-type cells, Snc1 is localized to the plasma membrane in a polarized distribution. Defects in retrograde transport from early endosomes, but not late endosomes, result in the intracellular accumulation of Snc1-GFP [14, 67]. In our work, we created a quantitative, high-throughput reporter of endosomal transport based on Snc1 (Chapter 3, Figure 1.5). Chimeric reporters are a common tool in intracellular trafficking that can help to dissect sorting signals and identify the machinery that recognizes them [18, 108, 114, 115]. The chimeric reporter we employed involved a fusion of the sucrose converting enzyme invertase, encoded by the gene SUC2, to the C-terminus of GFP-Snc1 (Figure 1.5). This topology ensures invertase is exposed to the extracellular space, and allowed detection of plasma membrane localization by whole colony assays of invertase activity. This ease of quantitative measure compared to imaging GFP-Snc1 15  localization, enabled sensitive, high-throughput studies of retrograde transport. Use of this reporter enabled identification and analysis of genes and pathways involved in endosomal and Golgi transport (Chapter 3).  Figure 1.5 Quantitative reporter of endosomal trafficking dysfunction. Chimeric fusion of GFP, the secretory vesicle v-SNARE, Snc1, and the enzyme invertase, Suc2, comprises the GSS reporter. Sorting signals in Snc1 cause GSS to follow a similar recycling itinerary to GFP-Snc1. In wild-type cells, GSS is present on the plasma membrane (PM) as well as internal membranes. The topology of the GSS reporter exposes invertase to the extracellular space. Membrane impermeant reagents are used to quantify the activity of invertase, and thus the extent of GSS plasma membrane localization. Mutants that alter GSS localization to the plasma membrane are detected by relative changes in GSS activity. 1.4.3 Synthetic genetic analysis and epistasis mapping in yeast  Genetic screens effectively identify a list of genes involved in a common process, but with this information alone, the significance of each hit is difficult to assess. Combining mutant alleles in the same organism enables genetic interaction to be measured. Since genes that genetically interact tend to belong to the same broad functional category, or in parallel pathways required for a common process, genetic interactions can relate genes on the basis of pathways [116, 117]. High-throughput genetic interaction analysis became possible in yeast with the systematic creation of genome wide collections of allele disruptions, and the development of synthetic genetic array (SGA), and diploid-based synthetic lethal analysis by microarray (dSLAM) technologies [118-121]. Yeast exists in two haploid mating types termed MATa and MATα, and also exists in the diploid form. Each of the ~6000 genes in the yeast genome has been disrupted by insertion of a drug resistance gene flanked by a unique 20-bp bar code identifier and common PCR 16  amplification sites [118, 122]. This effort created four sets of yeast knock out collections: ~4800 MATa and MATα haploid deletion mutants, ~4800 homozygous diploid mutants and ~6000 heterozygous diploid mutants [122]. Collections of essential gene disruptions have also been constructed and systematically expanded [123-125]. These systematic collections of allele disruptions have enabled many genetic screens for diverse cellular processes, often making unexpected connections with trafficking genes (reviewed in [126]). SGA analysis is an array based method for the systematic construction and analysis of double mutants [121]. Briefly, ordered arrays of a haploid yeast deletion collection are mated with a query strain of the opposite mating type that carries an allele disruption marked with a unique selectable marker, producing diploid cells. Diploid cells are then induced to undergo meiosis and through a series of array replication steps that select knock-outs based on mating type and drug resistance, haploid double mutants are created [121]. In array based methods, the identity of the double mutant is inferred from its position on an ordered array and can be confirmed using PCR strategies. dSLAM involves somewhat similar techniques to produce double mutants, however, strains are grown together in batch culture and double mutants are identified by hybridization of their unique tags to oligonucleotide arrays [119]. A recent approach for construction of double mutants called Genetic Interaction Mapping (GIM) involves a combination of techniques [127]. Early systematic screens of genetic interaction focussed on detection of synthetic sickness and lethality (SSL), aiming to identify all SSL interaction partners for each gene in the yeast genome and to relate genes in the context of pathways [116]. Comprehensively defining all genetic interactions in the yeast genome requires quantitative measure of approximately 18 million double mutant phenotypes. Recent work has shown that a more focussed analysis of genetic interactions that involves logically selected subsets of the genome can provide much information regarding gene function and pathway relationships [128, 129]. This approach, termed epistasis miniarray profiling (E-MAP), enables detection of synthetic sick/lethal interactions, but also allows detection of other types of genetic interaction [128]. E-MAP analysis of genetic interaction produces a continuous score for the degree by which each pair of genes interacts [128, 130]. This score can then be used to detect similar genetic interaction profiles for each gene using cluster analysis, and the combination of both properties (correlation of profiles and interaction score) enables prediction of putative protein 17  complexes or members of the same pathway [128, 130]. The capacity for E-MAP analysis to make predictive statements of gene function was thought to rely on the large and comprehensive number of measurements [130]. However, a recent re-analysis of E-MAP data suggests that much of the information gained could be achieved with approximately half the queries tested [131], suggesting that a comprehensive analysis of genetic interactions is not required to gain much of the valuable information regarding gene function and pathway relationships. Recently, a strategy to measure both aggravating and alleviating genetic interactions with a continuous score on a genome wide scale was developed, which may prove to be valuable for understanding properties of genetic interaction networks [127]. Also, another important element of genetic interaction analysis is a consideration of the phenotype analyzed. The overwhelming majority of systematic genetic interaction analyses have used growth as a phenotype; however, alternative phenotypes such as protein localization or expression of a reporter construct could provide unique information about the ways that genes interact. 1.4.4 Genetic interaction analysis in other systems  With advancements in RNA interference and high-throughput imaging platforms, genetic screens in mammalian cell lines and other higher organisms are possible and prevalent (reviewed in [132-134]). Recent work has accomplished combinatorial knock-downs allowing analysis of genetic interaction, including work with non-growth phenotypes. For example, in Drosophilia, knock-down of kinases and phosphatases were assessed for their effect on JNK signalling as determined by activity of a FRET-based reporter. From this initial screen, additional targets were selected to test in combination, resulting in identification of additional JNK suppressors and enhancers [135]. Screens for synthetic lethality and sickness by combinatorial RNA interference have also been accomplished in C. elegans [136, 137]. Recently, tools for high-throughput genetic interaction analyses in Schizosaccharomyces pombe, a yeast strain that is distantly related to S. cerevisiae, as well as tools for genetic interaction analysis in E. coli were developed [138, 139]. Analyses of divergent genetic interaction networks should help to inform the extent and context in which networks of genetic interaction are conserved [140-142] 1.4.5 Definitions of genetic interaction  Use of the term genetic interaction has taken on different meanings [143]. One commonly applied definition, originally coined by Fisher in statistical genetics, is significant  18  deviation of a double mutant phenotype from a neutral expectation [144]. A neutral expectation can also be defined in multiple ways, and these can be generally categorized into empirical and theoretical models. In an empirical definition, it is assumed that genetic interactions are rare and a neutral expectation is defined as the average phenotype observed for a given gene deletion when combined with all other genes. That is, the neutral expectation is the average normalized double mutant phenotype for each gene [130]. This model of neutrality has been successfully used to define genetic interactions in E-MAP analysis as discussed above. A theoretical definition of the neutral expectation computes an expectation based on the phenotype observed for each single gene mutation and a model of neutrality, such as multiplicative, log or additive [144, 145]. In either case, with the Fisher definition of genetic interaction, two genes can combine to produce a phenotype that is either more severe than expected, neutral, or less severe than expected [146]. Interactions that are more severe than expected are also called aggravating, or with respect to growth, synthetic sick or lethal [121, 128, 130]. If an observed double mutant phenotype is less severe than expected, the genetic interaction is called alleviating (also referred to as diminishing returns) [116, 128, 143]. This terminology (aggravating, neutral, and alleviating) can be thought of as a coarse grain perspective to describe the relationship between two genes. Since comparison of the observed double mutant phenotype with its neutral expectation can be accomplished with a t-test (or similar approach), these initial broad terms are useful for categorization. Alleviating interactions are called information rich because this broad term encompasses a number of classic and familiar ways in which two genes can interact. Interpretation of the type of genetic interaction is accomplished by comparing single and double mutant phenotypes [145, 147]. For instance, if a double mutant phenotype is very similar to the phenotype of each of its parental single mutant phenotypes, this is called coequal [145]. The classical molecular or genetic definition of the term epistasis coined by Bateson, wherein the activity of one gene masks the activity of another, is also included in the broad category of alleviating interactions (reviewed in [143, 148]). To avoid confusion between Bateson and Fisher definitions of epistasis, in this thesis the term masking is used to describe Bateson epistasis. Because of the capacity to categorize alleviating interactions, an analysis of alleviating genetic interactions can provide an understanding of the functional relationships between gene products. Moreover, these analyses can uncover some of the complex relationships between pathways.  19  1.4.6 Networks of genetic interaction  Like other large scale data-types, systematic analyses of genetic interaction are often interpreted using network analysis. Network diagrams generally involve depictions of genes as nodes (circles) and the data connecting two genes is represented by an edge (a line). Large scale genetic interaction networks produced from publicly available data repositories such as Saccharomyces Genome Database (SGD), predominantly describe aggravating, synthetic sick and lethal relationships between genes. Integration of this data with other high-throughput datatypes such as protein interaction data improves interpretation, and many strategies to accomplish network integration and analysis have been developed (reviewed in [149-151]). Integrated analyses have assisted with the biological interpretation of aggravating interactions, establishing that for non-essential genes, synthetic sick or lethal interactions tend to occur between parallel pathways [117]. But, analysis of genetic interaction networks alone can provide information about the pathway to which a gene belongs and can be used to predict gene function [146, 152, 153]. A consideration of alleviating interactions increases the amount of information present in a genetic interaction network because these interactions can be further categorized [145, 147]. St. Onge et al. categorized alleviating interactions into one of five interaction types based on the relative growth sensitivity of single and double mutants for each interacting pair. These classifications involved Bateson definitions of masking epistasis, and other categories such as coequal and suppression. Analysis of these relationships is assisted with network diagrams that allow common interactions and trends to be visualized [145]. A more complex categorization of genetic interactions does not first segregate interactions into alleviating and aggravating, but rather considers all possible relationships between single, double and wild-type phenotypes to construct an inequality, and then uses this inequality to classify gene pairs into one of nine categories of interaction [147]. This dense network of interactions allows complex analyses of the relationships between genes and pathways, illustrating that a network of genetic interaction can reflect the complex relationships, and flow of information between pathways [147]. In summary, the capacity to map alleviating interactions into more complex types of genetic interaction allows an in depth analysis of the genetic relationships between pathways, which may reflect, or provide novel insight into, the biochemical relationships among their gene products.  20  1.5 Thesis Summary Intracellular trafficking pathways are governed by the dynamic interplay of many determinants of specificity: cargo sorting signals, lipids, sorting complexes, coat proteins, GTPases, tethering complexes, SNARE proteins and numerous regulators. The complex relationships between determinants of specificity are mirrored by the complexity of the pathways themselves. Trafficking pathways are not isolated units: organelles mature, machinery can be shared between pathways and cargo can take alternative or bypass routes. The importance of understanding intracellular transport is highlighted by the fact that mutations in trafficking machinery and regulators can result in numerous diseases, including neurodegeneration. Deciphering both the components and pathways that comprise the endosomal transport system has benefited from the combined efforts of biochemical, molecular and imaging studies as well as phenotypic screens, genetic interactions and the construction of model proteins designed to follow pathways connecting the Golgi, endosomes and vacuole. In this thesis a combination of molecular analyses (Chapter 2) and larger scale genetic approaches (Chapter 3) have been applied to study the machinery and pathways involved in retrograde transport from endosomes to the Golgi. The GARP complex is one of many examples of shared machinery as it is responsible for tethering vesicles derived from both early and late endosomes with the late Golgi. Tethering complexes must recognize features of both the upstream and downstream compartment to mediate their activity. While tethers are known to interact with both SNAREs and regulatory GTPases, the mechanisms and timing of membrane recognition are not fully understood. GARP interacts with the t-SNARE Tlg1, but this interaction is not required for Golgi localization, and loss of the interaction does not impact GARP function in vivo [78]. Interaction with the activated form of the Rab GTPase Ypt6 is required for GARP’s Golgi localization, but loss of YPT6 does not abolish GARP’s association with membranes, suggesting another determinant is responsible for upstream compartment recognition. In my first body of work (Chapter 2), I evaluated the hypothesis that different domains within the Vps54 protein, and thus the GARP complex, are involved in unique membrane recognition events. By evaluating truncation mutants of Vps54, I found that the N-terminal portion is important for GARP complex assembly and stability, while the C-terminal portion localized to a compartment with features of an endosome. In the absence of the C-terminal domain, retrieval of Snc1 from early endosomes to the Golgi 21  became dependant on the late endosome to Golgi retrieval pathway. These data suggest the Cterminal domain is important for early endosome to Golgi pathway function. These studies provide insight into the mechanisms of tethering complex function, and illustrate the capacity for retrograde transport from late endosomes to compensate for defects in early endosome transport. The complexity of membrane recognition events, capacity for cargo to be transported on bypass pathways and involvement of retrograde transport in neurodegenerative disease motivates an analysis of retrograde transport from a pathway perspective. Genetic interactions can reveal complex relationships between genes and by extension between pathways. Much of the power of genetic interaction analyses comes from the discovery of alleviating genetic interactions, as these encompass a number of classical forms of interaction such as Bateson masking epistasis and suppression. Genetic interactions in the genome are rare, but can be enriched by studying a set of functionally related genes. As genetic interactions are expected to be context dependant, it is possible that genetic interactions could also be enriched by measuring a phenotype that closely relates to gene function. In my second body of work (Chapter 3), I evaluated the hypothesis that alleviating genetic interactions can be enriched based on gene selection and phenotypic measure, by using the combination of a genetic screen and analyses of genetic interaction that are targeted for genes involved in endosomal transport. The hypothesis that alleviating interactions can reveal relationships between pathways was explored by mapping alleviating interactions into categories, and visualizing pathway trends with network analyses. These analyses ultimately reveal interdependencies and more complex relationships between endosomal, Golgi and vacuolar trafficking pathways, uncovering a potentially common way to enhance trafficking to the cell surface. Taken together, this thesis will investigate possible mechanisms by which trafficking machinery confers specificity and how this translates to the complex relationships between endosomal and Golgi trafficking pathways.  22  1.6 References 1.  Meisler MH, Russ C, Montgomery KT, Greenway M, Ennis S, Hardiman O, Figlewicz DA, Quenneville NR, Conibear E, Brown RH, Jr.: Evaluation of the Golgi trafficking protein VPS54 (wobbler) as a candidate for ALS. Amyotroph Lateral Scler 2008, 9(3):141-148.  2.  Bonifacino JS, Glick BS: The mechanisms of vesicle budding and fusion. Cell 2004, 116(2):153-166.  3.  Bonifacino JS, Rojas R: Retrograde transport from endosomes to the trans-Golgi network. Nat Rev Mol Cell Biol 2006, 7(8):568-579.  4.  Kirchhausen T: Three ways to make a vesicle. Nat Rev Mol Cell Biol 2000, 1(3):187198.  5.  Seaman MN: Endosome protein sorting: motifs and machinery. Cell Mol Life Sci 2008, 65(18):2842-2858.  6.  Bonifacino JS, Traub LM: Signals for sorting of transmembrane proteins to endosomes and lysosomes. Annu Rev of Bioch 2003, 72(1):395-447.  7.  Bonifacino JS: The GGA proteins: adaptors on the move. Nat Rev Mol Cell Biol 2004, 5(1):23-32.  8.  Mukhopadhyay D, Riezman H: Proteasome-independent functions of ubiquitin in endocytosis and signaling. Science 2007, 315(5809):201-205.  9.  Spang A: The life cycle of a transport vesicle. Cell Mol Life Sci 2008, 65(18):27812789.  10.  Behnia R, Munro S: Organelle identity and the signposts for membrane traffic. Nature 2005, 438(7068):597-604.  11.  Zerial M, McBride H: Rab proteins as membrane organizers. Nat Rev Mol Cell Biol 2001, 2(2):107-117.  12.  Munro S: The Arf-like GTPase Arl1 and its role in membrane traffic. Biochem Soc Trans 2005, 33(Pt 4):601-605.  13.  Carlton JG, Cullen PJ: Coincidence detection in phosphoinositide signaling. Trends Cell Biol 2005, 15(10):540-547.  23  14.  Hettema EH, Lewis MJ, Black MW, Pelham HR: Retromer and the sorting nexins Snx4/41/42 mediate distinct retrieval pathways from yeast endosomes. EMBO J 2003, 22(3):548-557.  15.  Burda P, Padilla SM, Sarkar S, Emr SD: Retromer function in endosome-to-Golgi retrograde transport is regulated by the yeast Vps34 PtdIns 3-kinase. J Cell Sci 2002, 115(20):3889-3900.  16.  Harsay E, Bretscher A: Parallel secretory pathways to the cell surface in yeast. J Cell Biol 1995, 131(2):297-310.  17.  Harsay E, Schekman R: A subset of yeast vacuolar protein sorting mutants is blocked in one branch of the exocytic pathway. J Cell Biol 2002, 156(2):271-286.  18.  Black MW, Pelham HRB: A Selective Transport Route from Golgi to Late Endosomes that Requires the Yeast GGA Proteins. J Cell Biol 2000, 151(3):587-600.  19.  Boman AL, Zhang C, Zhu X, Kahn RA: A family of ADP-ribosylation factor effectors that can alter membrane transport through the trans-Golgi. Mol Biol Cell 2000, 11(4):1241-1255.  20.  Dell'Angelica EC, Puertollano R, Mullins C, Aguilar RC, Vargas JD, Hartnell LM, Bonifacino JS: GGAs: A Family of ADP Ribosylation Factor-binding Proteins Related to Adaptors and Associated with the Golgi Complex. J Cell Biol 2000, 149(1):81-94.  21.  Hirst J, Lui WWY, Bright NA, Totty N, Seaman MNJ, Robinson MS: A Family of Proteins with {gamma}-Adaptin and VHS Domains that Facilitate Trafficking between the Trans-Golgi Network and the Vacuole/Lysosome. J Cell Biol 2000, 149(1):67-80.  22.  Poussu A, Lohi O, Lehto V-P: Vear, a Novel Golgi-associated Protein with VHS and gamma -Adaptin "Ear" Domains. J Biol Chem 2000, 275(10):7176-7183.  23.  Boman AL: GGA proteins: new players in the sorting game. J Cell Sci 2001, 114(Pt 19):3413-3418.  24.  Scott PM, Bilodeau PS, Zhdankina O, Winistorfer SC, Hauglund MJ, Allaman MM, Kearney WR, Robertson AD, Boman AL, Piper RC: GGA proteins bind ubiquitin to facilitate sorting at the trans-Golgi network. Nat Cell Biol 2004, 6(3):252-259.  25.  Puertollano R, Bonifacino JS: Interactions of GGA3 with the ubiquitin sorting machinery. Nat Cell Biol 2004, 6(3):244-251.  24  26.  Shiba Y, Katoh Y, Shiba T, Yoshino K, Takatsu H, Kobayashi H, Shin H-W, Wakatsuki S, Nakayama K: GAT (GGA and Tom1) Domain Responsible for Ubiquitin Binding and Ubiquitination. J Biol Chem 2004, 279(8):7105-7111.  27.  Boman AL, Salo PD, Hauglund MJ, Strand NL, Rensink SJ, Zhdankina O: ADPRibosylation Factor (ARF) Interaction Is Not Sufficient for Yeast GGA Protein Function or Localization. Mol Biol Cell 2002, 13(9):3078-3095.  28.  Puertollano R, Randazzo PA, Presley JF, Hartnell LM, Bonifacino JS: The GGAs Promote ARF-Dependent Recruitment of Clathrin to the TGN. Cell 2001, 105(1):93102.  29.  Wang J, Sun H-Q, Macia E, Kirchhausen T, Watson H, Bonifacino JS, Yin HL: PI4P Promotes the Recruitment of the GGA Adaptor Proteins to the Trans-Golgi Network and Regulates Their Recognition of the Ubiquitin Sorting Signal. Mol Biol Cell 2007, 18(7):2646-2655.  30.  Demmel L, Gravert M, Ercan E, Habermann B, Muller-Reichert T, Kukhtina V, Haucke V, Baust T, Sohrmann M, Kalaidzidis Y et al: The Clathrin Adaptor Gga2p Is a Phosphatidylinositol 4-phosphate Effector at the Golgi Exit. Mol Biol Cell 2008, 19(5):1991-2002.  31.  Nielsen MS, Madsen P, Christensen EI, Nykjaer A, Gliemann J, Kasper D, Pohlmann R, Petersen CM: The sortilin cytoplasmic tail conveys Golgi-endosome transport and binds the VHS domain of the GGA2 sorting protein. EMBO J 2001, 20(9):2180-2190.  32.  Puertollano R, Aguilar RC, Gorshkova I, Crouch RJ, Bonifacino JS: Sorting of mannose 6-phosphate receptors mediated by the GGAs. Science 2001, 292(5522):1712-1716.  33.  Zhu Y, Doray B, Poussu A, Lehto V-P, Kornfeld S: Binding of GGA2 to the Lysosomal Enzyme Sorting Motif of the Mannose 6-Phosphate Receptor. Science 2001, 292(5522):1716-1718.  34.  Takatsu H, Katoh Y, Shiba Y, Nakayama K: Golgi-localizing, gamma-adaptin ear homology domain, ADP-ribosylation factor-binding (GGA) proteins interact with acidic dileucine sequences within the cytoplasmic domains of sorting receptors through their Vps27p/Hrs/STAM (VHS) domains. J Biol Chem 2001, 276(30):2854128545.  35.  Duncan MC, Costaguta G, Payne GS: Yeast epsin-related proteins required for Golgi-endosome traffic define a [gamma]-adaptin ear-binding motif. Nat Cell Biol 2003, 5(1):77-81.  25  36.  Costaguta G, Duncan MC, Fernandez GE, Huang GH, Payne GS: Distinct roles for TGN/endosome epsin-like adaptors Ent3p and Ent5p. Mol Biol Cell 2006, 17(9):3907-3920.  37.  Copic A, Starr TL, Schekman R: Ent3p and Ent5p exhibit cargo-specific functions in trafficking proteins between the trans-Golgi network and the endosomes in yeast. Mol Biol Cell 2007, 18(5):1803-1815.  38.  Duncan MC, Payne GS: ENTH/ANTH domains expand to the Golgi. Trends Cell Biol 2003, 13(5):211-215.  39.  Chidambaram S, Zimmermann J, von Mollard GF: ENTH domain proteins are cargo adaptors for multiple SNARE proteins at the TGN endosome. J Cell Sci 2008, 121(3):329-338.  40.  Wahle T, Prager K, Raffler N, Haass C, Famulok M, Walter J: GGA proteins regulate retrograde transport of BACE1 from endosomes to the trans-Golgi network. Molecular and Cellular Neuroscience 2005, 29(3):453-461.  41.  Singer-Kruger B, Lasic M, Burger AM, Hausser A, Pipkorn R, Wang Y: Yeast and human Ysl2p/hMon2 interact with Gga adaptors and mediate their subcellular distribution. EMBO J 2008, 27(10):1423-1435.  42.  Sollner T, Whiteheart SW, Brunner M, Erdjument-Bromage H, Geromanos S, Tempst P, Rothman JE: SNAP receptors implicated in vesicle targeting and fusion. Nature 1993, 362(6418):318-324.  43.  Weber T, Zemelman BV, McNew JA, Westermann B, Gmachl M, Parlati F, Sollner TH, Rothman JE: SNAREpins: minimal machinery for membrane fusion. Cell 1998, 92(6):759-772.  44.  Malsam J, Kreye S, Sollner TH: Membrane fusion: SNAREs and regulation. Cell Mol Life Sci 2008, 65(18):2814-2832.  45.  Parlati F, McNew JA, Fukuda R, Miller R, Sollner TH, Rothman JE: Topological restriction of SNARE-dependent membrane fusion. Nature 2000, 407(6801):194198.  46.  Whyte JRC, Munro S: Vesicle tethering complexes in membrane traffic. J Cell Sci 2002, 115(13):2627-2637.  47.  Pfeffer SR: Transport Vesicle Docking: SNAREs and Associates. Annual Review of Cell and Developmental Biology 1996, 12(1):441-461.  26  48.  Yamakawa H, Seog DH, Yoda K, Yamasaki M, Wakabayashi T: Uso1 protein is a dimer with two globular heads and a long coiled-coil tail. J Struct Biol 1996, 116(3):356-365.  49.  Dumas JJ, Merithew E, Sudharshan E, Rajamani D, Hayes S, Lawe D, Corvera S, Lambright DG: Multivalent endosome targeting by homodimeric EEA1. Mol Cell 2001, 8(5):947-958.  50.  TerBush DR, Novick P: Sec6, Sec8, and Sec15 are components of a multisubunit complex which localizes to small bud tips in Saccharomyces cerevisiae. J Cell Biol 1995, 130(2):299-312.  51.  TerBush DR, Maurice T, Roth D, Novick P: The Exocyst is a multiprotein complex required for exocytosis in Saccharomyces cerevisiae. EMBO J 1996, 15(23):64836494.  52.  Conibear E, Stevens TH: Vps52p, Vps53p, and Vps54p form a novel multisubunit complex required for protein sorting at the yeast late Golgi. Mol Biol Cell 2000, 11(1):305-323.  53.  Conibear E, Cleck JN, Stevens TH: Vps51p mediates the association of the GARP (Vps52/53/54) complex with the late Golgi t-SNARE Tlg1p. Mol Biol Cell 2003, 14(4):1610-1623.  54.  Siniossoglou S, Pelham HR: An effector of Ypt6p binds the SNARE Tlg1p and mediates selective fusion of vesicles with late Golgi membranes. EMBO J 2001, 20(21):5991-5998.  55.  Wurmser AE, Sato TK, Emr SD: New Component of the Vacuolar Class C-Vps Complex Couples Nucleotide Exchange on the Ypt7 GTPase to SNARE-dependent Docking and Fusion. J Cell Biol 2000, 151(3):551-562.  56.  Prigent M, Dubois T, Raposo G, Derrien V, Tenza D, Rosse C, Camonis J, Chavrier P: ARF6 controls post-endocytic recycling through its downstream exocyst complex effector. J Cell Biol 2003, 163(5):1111-1121.  57.  Zhang XM, Ellis S, Sriratana A, Mitchell CA, Rowe T: Sec15 is an effector for the Rab11 GTPase in mammalian cells. J Biol Chem 2004, 279(41):43027-43034.  58.  Morsomme P, Riezman H: The Rab GTPase Ypt1p and tethering factors couple protein sorting at the ER to vesicle targeting to the Golgi apparatus. Dev Cell 2002, 2(3):307-317.  59.  Suvorova ES, Duden R, Lupashin VV: The Sec34/Sec35p complex, a Ypt1p effector required for retrograde intra-Golgi trafficking, interacts with Golgi SNAREs and COPI vesicle coat proteins. J Cell Biol 2002, 157(4):631-643. 27  60.  Cai H, Yu S, Menon S, Cai Y, Lazarova D, Fu C, Reinisch K, Hay JC, Ferro-Novick S: TRAPPI tethers COPII vesicles by binding the coat subunit Sec23. Nature 2007, 445(7130):941-944.  61.  Andag U, Neumann T, Schmitt HD: The coatomer-interacting protein Dsl1p is required for Golgi-to-endoplasmic reticulum retrieval in yeast. J Biol Chem 2001, 276(42):39150-39160.  62.  Seaman MN, McCaffery JM, Emr SD: A membrane coat complex essential for endosome-to-Golgi retrograde transport in yeast. J Cell Biol 1998, 142(3):665-681.  63.  Ghosh P, Dahms NM, Kornfeld S: Mannose 6-phosphate receptors: new twists in the tale. Nat Rev Mol Cell Biol 2003, 4(3):202-212.  64.  Kihara A, Noda T, Ishihara N, Ohsumi Y: Two Distinct Vps34 Phosphatidylinositol 3Kinase Complexes Function in Autophagy and Carboxypeptidase Y Sorting in Saccharomyces cerevisiae. J Cell Biol 2001, 152(3):519-530.  65.  Hierro A, Rojas AL, Rojas R, Murthy N, Effantin G, Kajava AV, Steven AC, Bonifacino JS, Hurley JH: Functional architecture of the retromer cargo-recognition complex. Nature 2007, 449(7165):1063-1067.  66.  Bonifacino JS, Hurley JH: Retromer. Curr Opin in Cell Biol 2008, 20(4):427-436.  67.  Lewis MJ, Nichols BJ, Prescianotto-Baschong C, Riezman H, Pelham HRB: Specific Retrieval of the Exocytic SNARE Snc1p from Early Yeast Endosomes. Mol Biol Cell 2000, 11(1):23-38.  68.  Robinson M, Poon PP, Schindler C, Murray LE, Kama R, Gabriely G, Singer RA, Spang A, Johnston GC, Gerst JE: The Gcs1 Arf-GAP Mediates Snc1,2 v-SNARE Retrieval to the Golgi in Yeast. Mol Biol Cell 2006, 17(4):1845-1858.  69.  Chen SH, Chen S, Tokarev AA, Liu F, Jedd G, Segev N: Ypt31/32 GTPases and their novel F-box effector protein Rcy1 regulate protein recycling. Mol Biol Cell 2005, 16(1):178-192.  70.  Galan J-M, Wiederkehr A, Seol JH, Haguenauer-Tsapis R, Deshaies RJ, Riezman H, Peter M: Skp1p and the F-Box Protein Rcy1p Form a Non-SCF Complex Involved in Recycling of the SNARE Snc1p in Yeast. Mol Cell Biol 2001, 21(9):3105-3117.  71.  Wiederkehr A, Avaro S, Prescianotto-Baschong C, Haguenauer-Tsapis R, Riezman H: The F-box protein Rcy1p is involved in endocytic membrane traffic and recycling out of an early endosome in Saccharomyces cerevisiae. J Cell Biol 2000, 149(2):397-410.  28  72.  Sakane H, Yamamoto T, Tanaka K: The functional relationship between the Cdc50pDrs2p putative aminophospholipid translocase and the Arf GAP Gcs1p in vesicle formation in the retrieval pathway from yeast early endosomes to the TGN. Cell Struct Funct 2006, 31(2):87-108.  73.  Furuta N, Fujimura-Kamada K, Saito K, Yamamoto T, Tanaka K: Endocytic recycling in yeast is regulated by putative phospholipid translocases and the Ypt31p/32pRcy1p pathway. Mol Biol Cell 2007, 18(1):295-312.  74.  Kama R, Robinson M, Gerst JE: Btn2, a Hook1 Ortholog and Potential Batten Disease-Related Protein, Mediates Late Endosome-Golgi Protein Sorting in Yeast. Mol Cell Biol 2007, 27(2):605-621.  75.  Strochlic TI, Setty TG, Sitaram A, Burd CG: Grd19/Snx3p functions as a cargospecific adapter for retromer-dependent endocytic recycling. J Cell Biol 2007, 177(1):115-125.  76.  Conboy MJ, Cyert MS: Luv1p/Rki1p/Tcs3p/Vps54p, a yeast protein that localizes to the late Golgi and early endosome, is required for normal vacuolar morphology. Mol Biol Cell 2000, 11(7):2429-2443.  77.  Siniossoglou S, Pelham HR: Vps51p links the VFT complex to the SNARE Tlg1p. J Biol Chem 2002, 277(50):48318-48324.  78.  Fridmann-Sirkis Y, Kent HM, Lewis MJ, Evans PR, Pelham HRB: Structural Analysis of the Interaction Between the SNARE Tlg1 and Vps51. Traffic 2006, 7(2):182-190.  79.  Panic B, Whyte JR, Munro S: The ARF-like GTPases Arl1p and Arl3p act in a pathway that interacts with vesicle-tethering factors at the Golgi apparatus. Curr Biol 2003, 13(5):405-410.  80.  Burguete AS, Fenn TD, Brunger AT, Pfeffer SR: Rab and Arl GTPase family members cooperate in the localization of the golgin GCC185. Cell 2008, 132(2):286-298.  81.  Liewen H, Meinhold-Heerlein I, Oliveira V, Schwarzenbacher R, Luo G, Wadle A, Jung M, Pfreundschuh M, Stenner-Liewen F: Characterization of the human GARP (Golgi associated retrograde protein) complex. Exp Cell Res 2005, 306(1):24-34.  82.  Perez-Victoria FJ, Mardones GA, Bonifacino JS: Requirement of the human GARP Complex for mannose 6-phosphate-receptor-dependent sorting of cathepsin D to lysosomes. Mol Biol Cell 2008, 19(6):2350-2362.  83.  Hurley JH: ESCRT complexes and the biogenesis of multivesicular bodies. Curr Opin Cell Biol 2008, 20(1):4-11.  29  84.  Bowers K, Stevens TH: Protein transport from the late Golgi to the vacuole in the yeast Saccharomyces cerevisiae. Biochim Biophys Acta Mol Cell Res 2005, 1744(3):438-454.  85.  Raymond CK, Howald-Stevenson I, Vater CA, Stevens TH: Morphological classification of the yeast vacuolar protein sorting mutants: evidence for a prevacuolar compartment in class E vps mutants. Mol Biol Cell 1992, 3(12):13891402.  86.  Nickerson DP, Russell MR, Odorizzi G: A concentric circle model of multivesicular body cargo sorting. EMBO Rep 2007, 8(7):644-650.  87.  Chen L, Davis NG: Recycling of the yeast a-factor receptor. J Cell Biol 2000, 151(3):731-738.  88.  Bugnicourt A, Froissard M, Sereti K, Ulrich HD, Haguenauer-Tsapis R, Galan J-M: Antagonistic Roles of ESCRT and Vps Class C/HOPS Complexes in the Recycling of Yeast Membrane Proteins. Mol Biol Cell 2004, 15(9):4203-4214.  89.  Maxfield FR, McGraw TE: Endocytic recycling. Nat Rev Mol Cell Biol 2004, 5(2):121132.  90.  Sommer B, Oprins A, Rabouille C, Munro S: The exocyst component Sec5 is present on endocytic vesicles in the oocyte of Drosophila melanogaster. J Cell Biol 2005, 169(6):953-963.  91.  Gabriely G, Kama R, Gerst JE: Involvement of Specific COPI Subunits in Protein Sorting from the Late Endosome to the Vacuole in Yeast. Mol Cell Biol 2007, 27(2):526-540.  92.  Nixon RA: Endosome function and dysfunction in Alzheimer's disease and other neurodegenerative diseases. Neurobiol Aging 2005, 26(3):373-382.  93.  Bronfman FC, Escudero CA, Weis J, Kruttgen A: Endosomal transport of neurotrophins: roles in signaling and neurodegenerative diseases. Dev Neurobiol 2007, 67(9):1183-1203.  94.  Small SA: Retromer sorting: a pathogenic pathway in late-onset Alzheimer disease. Arch Neurol 2008, 65(3):323-328.  95.  Ström A-L, Gal J, Shi P, Kasarskis EJ, Hayward LJ, Zhu H: Retrograde axonal transport and motor neuron disease. J of Neurochem 2008, 106(2):495-505.  96.  Schmitt-John T, Drepper C, Muszmann A, Hahn P, Kuhlmann M, Thiel C, Hafner M, Lengeling A, Heimann P, Jones JM et al: Mutation of Vps54 causes motor neuron 30  disease and defective spermiogenesis in the wobbler mouse. Nat Genet 2005, 37(11):1213-1215. 97.  Pasinelli P, Brown RH: Molecular biology of amyotrophic lateral sclerosis: insights from genetics. Nat Rev Neurosci 2006, 7(9):710-723.  98.  Yang Y, Hentati A, Deng HX, Dabbagh O, Sasaki T, Hirano M, Hung WY, Ouahchi K, Yan J, Azim AC et al: The gene encoding alsin, a protein with three guaninenucleotide exchange factor domains, is mutated in a form of recessive amyotrophic lateral sclerosis. Nat Genet 2001, 29(2):160-165.  99.  Hadano S, Hand CK, Osuga H, Yanagisawa Y, Otomo A, Devon RS, Miyamoto N, Showguchi-Miyata J, Okada Y, Singaraja R et al: A gene encoding a putative GTPase regulator is mutated in familial amyotrophic lateral sclerosis 2. Nat Genet 2001, 29(2):166-173.  100.  Nishimura AL, Mitne-Neto M, Silva HC, Richieri-Costa A, Middleton S, Cascio D, Kok F, Oliveira JR, Gillingwater T, Webb J et al: A mutation in the vesicle-trafficking protein VAPB causes late-onset spinal muscular atrophy and amyotrophic lateral sclerosis. Am J Hum Genet 2004, 75(5):822-831.  101.  Topp JD, Gray NW, Gerard RD, Horazdovsky BF: Alsin is a Rab5 and Rac1 guanine nucleotide exchange factor. J Biol Chem 2004, 279(23):24612-24623.  102.  Tsuda H, Han SM, Yang Y, Tong C, Lin YQ, Mohan K, Haueter C, Zoghbi A, Harati Y, Kwan J et al: The amyotrophic lateral sclerosis 8 protein VAPB is cleaved, secreted, and acts as a ligand for Eph receptors. Cell 2008, 133(6):963-977.  103.  Gkogkas C, Middleton S, Kremer AM, Wardrope C, Hannah M, Gillingwater TH, Skehel P: VAPB interacts with and modulates the activity of ATF6. Hum Mol Genet 2008, 17(11):1517-1526.  104.  Bock JB, Matern HT, Peden AA, Scheller RH: A genomic perspective on membrane compartment organization. Nature 2001, 409(6822):839-841.  105.  Novick P, Field C, Schekman R: Identification of 23 complementation groups required for post-translational events in the yeast secretory pathway. Cell 1980, 21(1):205-215.  106.  Rothman JH, Stevens TH: Protein sorting in yeast: mutants defective in vacuole biogenesis mislocalize vacuolar proteins into the late secretory pathway. Cell 1986, 47(6):1041-1051.  107.  Bankaitis VA, Johnson LM, Emr SD: Isolation of yeast mutants defective in protein targeting to the vacuole. Proc Natl Acad Sci U S A 1986, 83(23):9075-9079.  31  108.  Robinson JS, Klionsky DJ, Banta LM, Emr SD: Protein sorting in Saccharomyces cerevisiae: isolation of mutants defective in the delivery and processing of multiple vacuolar hydrolases. Mol Cell Biol 1988, 8(11):4936-4948.  109.  Wada Y, Ohsumi Y, Anraku Y: Genes for directing vacuolar morphogenesis in Saccharomyces cerevisiae. I. Isolation and characterization of two classes of vam mutants. J Biol Chem 1992, 267(26):18665-18670.  110.  Bonangelino CJ, Chavez EM, Bonifacino JS: Genomic Screen for Vacuolar Protein Sorting Genes in Saccharomyces cerevisiae. Mol Biol Cell 2002, 13(7):2486-2501.  111.  Cooper AA, Stevens TH: Vps10p cycles between the late-Golgi and prevacuolar compartments in its function as the sorting receptor for multiple yeast vacuolar hydrolases. J Cell Biol 1996, 133(3):529-541.  112.  Rothman JH, Howald I, Stevens TH: Characterization of genes required for protein sorting and vacuolar function in the yeast Saccharomyces cerevisiae. Embo J 1989, 8(7):2057-2065.  113.  Schluter C, Lam KKY, Brumm J, Wu BW, Saunders M, Stevens TH, Bryan J, Conibear E: Global Analysis of Yeast Endosomal Transport Identifies the Vps55/68 Sorting Complex. Mol Biol Cell 2008, 19(4):1282-1294.  114.  Nothwehr SF, Roberts CJ, Stevens TH: Membrane protein retention in the yeast Golgi apparatus: dipeptidyl aminopeptidase A is retained by a cytoplasmic signal containing aromatic residues. J Cell Biol 1993, 121(6):1197-1209.  115.  Valdez-Taubas J, Pelham HR: Slow diffusion of proteins in the yeast plasma membrane allows polarity to be maintained by endocytic cycling. Curr Biol 2003, 13(18):1636-1640.  116.  Tong AHY, Lesage G, Bader GD, Ding H, Xu H, Xin X, Young J, Berriz GF, Brost RL, Chang M et al: Global Mapping of the Yeast Genetic Interaction Network. Science 2004, 303(5659):808-813.  117.  Kelley R, Ideker T: Systematic interpretation of genetic interactions using protein networks. Nature Biotechnol 2005, 23:561-566.  118.  Winzeler EA, Shoemaker DD, Astromoff A, Liang H, Anderson K, Andre B, Bangham R, Benito R, Boeke JD, Bussey H et al: Functional characterization of the S. cerevisiae genome by gene deletion and parallel analysis. Science 1999, 285(5429):901-906.  119.  Pan X, Yuan DS, Xiang D, Wang X, Sookhai-Mahadeo S, Bader JS, Hieter P, Spencer F, Boeke JD: A robust toolkit for functional profiling of the yeast genome. Mol Cell 2004, 16(3):487-496.  32  120.  Suter B, Auerbach D, Stagljar I: Yeast-based functional genomics and proteomics technologies: the first 15 years and beyond. Biotechniques 2006, 40(5):625-644.  121.  Tong AHY, Evangelista M, Parsons AB, Xu H, Bader GD, Page N, Robinson M, Raghibizadeh S, Hogue CWV, Bussey H et al: Systematic Genetic Analysis with Ordered Arrays of Yeast Deletion Mutants. Science 2001, 294(5550):2364-2368.  122.  Giaever G, Chu AM, Ni L, Connelly C, Riles L, Veronneau S, Dow S, Lucau-Danila A, Anderson K, Andre B et al: Functional profiling of the Saccharomyces cerevisiae genome. Nature 2002, 418(6896):387-391.  123.  Mnaimneh S, Davierwala AP, Haynes J, Moffat J, Peng WT, Zhang W, Yang X, Pootoolal J, Chua G, Lopez A et al: Exploration of essential gene functions via titratable promoter alleles. Cell 2004, 118(1):31-44.  124.  Dohmen RJ, Varshavsky A: Heat-inducible degron and the making of conditional mutants. Methods Enzymol 2005, 399:799-822.  125.  Ben-Aroya S, Coombes C, Kwok T, O'Donnell KA, Boeke JD, Hieter P: Toward a comprehensive temperature-sensitive mutant repository of the essential genes of Saccharomyces cerevisiae. Mol Cell 2008, 30(2):248-258.  126.  Quenneville N, Conibear E: Toward the Systems Biology of Vesicle Transport. Traffic 2006, 7(7):761-768.  127.  Decourty L, Saveanu C, Zemam K, Hantraye F, Frachon E, Rousselle J-C, FromontRacine M, Jacquier A: Linking functionally related genes by sensitive and quantitative characterization of genetic interaction profiles. Proc Natl Acad Sci 2008, 105(15):5821-5826.  128.  Schuldiner M, Collins SR, Thompson NJ, Denic V, Bhamidipati A, Punna T, Ihmels J, Andrews B, Boone C, Greenblatt JF et al: Exploration of the Function and Organization of the Yeast Early Secretory Pathway through an Epistatic Miniarray Profile. Cell 2005, 123(3):507-519.  129.  Conibear E: An E-MAP of the ER. Cell 2005, 123(3):366-368.  130.  Collins S, Schuldiner M, Krogan N, Weissman J: A strategy for extracting and analyzing large-scale quantitative epistatic interaction data. Genome Biology 2006, 7(7):R63.  131.  Casey FP, Cagney G, Krogan NJ, Shields DC: Optimal stepwise experimental design for pairwise functional interaction studies. Bioinformatics 2008:btn472.  132.  Wheeler DB, Carpenter AE, Sabatini DM: Cell microarrays and RNA interference chip away at gene function. Nat Genet 2005, 37 Suppl:S25-30. 33  133.  Carpenter AE, Sabatini DM: Systematic genome-wide screens of gene function. Nat Rev Genet 2004, 5(1):11-22.  134.  Moffat J, Sabatini DM: Building mammalian signalling pathways with RNAi screens. Nat Rev Mol Cell Biol 2006, 7(3):177-187.  135.  Bakal C, Linding R, Llense F, Heffern E, Martin-Blanco E, Pawson T, Perrimon N: Phosphorylation Networks Regulating JNK Activity in Diverse Genetic Backgrounds. Science 2008, 322(5900):453-456.  136.  Byrne A, Weirauch M, Wong V, Koeva M, Dixon S, Stuart J, Roy P: A global analysis of genetic interactions in Caenorhabditis elegans. J Biol 2007, 6(3):8.  137.  van Haaften G, Vastenhouw NL, Nollen EAA, Plasterk RHA, Tijsterman M: Gene interactions in the DNA damage-response pathway identified by genome-wide RNA-interference analysis of synthetic lethality. Proc Natl Acad Sci U S A 2004, 101(35):12992-12996.  138.  Roguev A, Wiren M, Weissman JS, Krogan NJ: High-throughput genetic interaction mapping in the fission yeast Schizosaccharomyces pombe. Nat Meth 2007, 4(10):861-866.  139.  Butland G, Babu M, Diaz-Mejia JJ, Bohdana F, Phanse S, Gold B, Yang W, Li J, Gagarinova AG, Pogoutse O et al: eSGA: E. coli synthetic genetic array analysis. Nat Meth 2008, 5(9):789-795.  140.  Tischler J, Lehner B, Fraser AG: Evolutionary plasticity of genetic interaction networks. Nat Genet 2008, 40(4):390-391.  141.  Dixon SJ, Fedyshyn Y, Koh JLY, Prasad TSK, Chahwan C, Chua G, Toufighi K, Baryshnikova A, Hayles J, Hoe K-L et al: Significant conservation of synthetic lethal genetic interaction networks between distantly related eukaryotes. Proc Natl Acad Sci 2008, 105(43):16653-16658.  142.  Roguev A, Bandyopadhyay S, Zofall M, Zhang K, Fischer T, Collins SR, Qu H, Shales M, Park H-O, Hayles J et al: Conservation and Rewiring of Functional Modules Revealed by an Epistasis Map in Fission Yeast. Science 2008, 322(5900):405-410.  143.  Boone C, Bussey H, Andrews BJ: Exploring genetic interactions and networks with yeast. Nat Rev Genet 2007, 8(6):437-449.  144.  Mani R, St Onge RP, Hartman JLt, Giaever G, Roth FP: Defining genetic interaction. Proc Natl Acad Sci U S A 2008, 105(9):3461-3466.  34  145.  St. Onge RP, Mani R, Oh J, Proctor M, Fung E, Davis RW, Nislow C, Roth FP, Giaever G: Systematic pathway analysis using high-resolution fitness profiling of combinatorial gene deletions. Nat Genet 2007, 39(2):199-206.  146.  Segre D, DeLuna A, Church GM, Kishony R: Modular epistasis in yeast metabolism. Nat Genet 2005, 37(1):77-83.  147.  Drees B, Thorsson V, Carter G, Rives A, Raymond M, Avila-Campillo I, Shannon P, Galitski T: Derivation of genetic interaction networks from quantitative phenotype data. Genome Biol 2005, 6(4):R38.  148.  Phillips PC: The Language of Gene Interaction. Genetics 1998, 149(3):1167-1171.  149.  Beyer A, Bandyopadhyay S, Ideker T: Integrating physical and genetic maps: from genomes to interaction networks. Nat Rev Genet 2007, 8(9):699-710.  150.  Ulitsky I, Shlomi T, Kupiec M, Shamir R: From E-MAPs to module maps: dissecting quantitative genetic interactions using physical interactions. Mol Syst Biol 2008, 4.  151.  Bandyopadhyay S, Kelley R, Krogan NJ, Ideker T: Functional Maps of Protein Complexes from Quantitative Genetic Interaction Data. PLoS Computational Biology 2008, 4(4):e1000065.  152.  Ma X, Tarone AM, Li W: Mapping Genetically Compensatory Pathways from Synthetic Lethal Interactions in Yeast. PLoS ONE 2008, 3(4):e1922.  153.  Ye P, Peyser BD, Pan X, Boeke JD, Spencer FA, Bader JS: Gene function prediction from congruent synthetic lethal interactions in yeast. Mol Syst Biol 2005, 1.  35  Chapter 2 Domains within the GARP subunit Vps54 confer separate functions in complex assembly and early endosome recognition 2.1 Preface Vesicle transport mediates the delivery of cargo while maintaining the identity of each organelle. Selective delivery requires mechanisms to distinguish and recognize organelles. Tethering complexes are thought to function at the interface of vesicles and the correct downstream compartment, helping to mediate the specificity of transport. How tethering complexes recognize and distinguish membranes is poorly understood. Moreover, the timing of membrane recognition and tethering is also unknown. In this Chapter, the mechanisms by which the conserved Golgi Associated Retrograde Protein (GARP) complex recognizes multiple endosome-derived membranes are explored.  A version of this chapter has been published. Quenneville NR, Chao TY, McCaffrey JM, Conibear E. (2006) Domains within the GARP subunit Vps54 Confer Separate Functions in Complex Assembly and Membrane Recognition. Molecular Biology of the Cell 17(4): 1859-70.  36  2.2 Introduction Selective transport between organelles requires that transport intermediates be targeted to appropriate downstream compartments. While the pairing of cognate SNARE proteins on donor and acceptor membranes is critical for membrane fusion, multi-subunit tethering complexes contribute to the specificity of vesicle transport in vivo by linking transport vesicles to their target membranes in a long-range interaction that may promote SNARE complex formation (reviewed in [1, 2]). Tethering complexes must therefore recognize membrane-associated factors on both the donor and recipient compartment. Interaction of tethering complexes with activated Rab GTPases is considered important for the initial recruitment of tethers to membranes as well as for subsequent fusion events. While Rab proteins are important determinants of membrane recognition, a single specificity factor can only be sufficient to explain recruitment to a single compartment. Additional factors implicated in determining organelle identity and membrane recruitment include inositol phospholipids, small GTPases of the Rho, Arf, and Arl families, and vesicle coat components [3]. Interactions with such determinants have been demonstrated for both coiled-coil and multisubunit tethering complexes (reviewed in [2, 3]). The exocyst, a well-studied tethering complex required for fusion at the cell surface, interacts with active GTPases of the Rab, Rho and Ral families on secretory vesicles and at the plasma membrane [4-6]. Although binding of the exocyst subunit Sec15 to the secretory vesicle Rab protein Sec4 promotes formation of a subcomplex on the vesicle containing six of the eight exocyst subunits, assembly of the full complex occurs only when the vesicle reaches the plasma membrane, where the two remaining subunits are localized by interaction with the Rho-like GTPase Cdc42 [7]. However, recent work indicating that Cdc42 restricts the exocyst to sites of polarized growth but does not mediate its association with the plasma membrane per se [8], suggests there is much to be learned about how tethering complexes are recruited to membranes. Studies in diverse cell types have demonstrated that tethering complexes may first become membrane associated at compartments upstream of the organelle to which they tether vesicles. In mammalian cells, the exocyst components Sec10 and Sec15 have been localized to both the TGN and recycling endosomes [9, 10], and in Drosophila, the exocyst component Sec5 37  is found at endocytic coated pits [11]. Similarly, the COG complex, which is implicated in tethering vesicles at the cis-Golgi that are derived from the ER, later Golgi cisternae and endosomes, interacts with COPI coats in both yeast and mammalian cells and is involved in sorting cargo at the ER [12, 13]. Association with upstream compartments prior to vesicle budding may provide part of the mechanism by which multi-subunit tethers become incorporated into specific classes of transport vesicle. The problem of membrane recognition is compounded by the fact that many tethering complexes are responsible for more than one trafficking pathway and must therefore recognize multiple upstream compartments. Mutations within subunits of the exocyst and COG complexes have been identified that are defective for a subset of pathways controlled by these tethers, supporting the idea that tethers participate in different sets of interactions at each organelle [11, 14, 15]. For example, mutations in Vps41, a subunit of the vacuolar tether HOPS, disrupt Golgi to vacuole transport while vacuolar transport from other compartments remains unaffected [16]. The GARP (UGUolgi UAUssociated URUetrograde UPUrotein) complex, (also referred to as the VFT complex) is a member of the “quatrefoil” family of multi-subunit tethering complexes that includes the exocyst and COG complexes [1]. The GARP complex is responsible for tethering vesicles derived from both early and late endosomes at the TGN [17-20]. Mutation of any one of the four GARP subunits - Vps51, Vps52, Vps53 or Vps54 - impairs the retrieval of the secretory vesicle v-SNARE Snc1 from early endosomes, and the recycling of the carboxypeptidase Y (CPY) receptor, Vps10, from late endosomes [20, 21]. Both of these retrograde pathways also require the Rab protein Ypt6, which interacts with Vps52 when in its GTP-bound form [19, 22]. Deletion of YPT6 prevents the localization of GARP to the TGN, causing it to become diffusely localized. However, loss of Ypt6 does not alter the amount of GARP associated with membranes in vivo [20]. These observations imply that GARP interacts with vesicles or other small, dispersed organelles by Ypt6-independent mechanisms. Because GARP acts in retrograde traffic from two classes of endosomes, distinct components of the complex may determine its interactions with different classes of vesicles. Here we present evidence that the recognition of upstream compartments is a conserved feature of tethering complexes. We identify a conserved C-terminal domain within the GARP subunit Vps54 that is specifically required for recycling from early, but not late, endosomes. Furthermore, this C-terminal domain is necessary and sufficient for recruitment to a 38  polarized early endocytic compartment. These results suggest that the GARP tethering complex recognizes distinct features of early and late endosomes and is recruited to upstream compartments prior to or during the budding of transport vesicles. This may ensure that forming vesicles have the full complement of factors to direct subsequent docking with the correct target organelle.  2.3 Results 2.3.1 The N-terminal domain of Vps54 is sufficient for assembly and stability of the GARP complex  Vps54 (Luv1/Tcs3/Cgp1), the largest subunit of the GARP tethering complex, contains several domains that are conserved from yeast to humans, including an N-terminal region with a predicted coiled-coil motif proposed to act in complex assembly [1, 22, 23]. To determine if one portion of the protein is sufficient for assembly and stability of the GARP complex, N-terminal (Vps54-N; residues 1-430) and C-terminal (Vps54-C; residues 593-889) fragments were HAtagged and co-expressed in strains that lacked endogenous Vps54 but contained functional myc-tagged forms of Vps51, Vps52 and Vps53 [18]. By co-immunoprecipitation, we determined the N-terminal domain, but not the C-terminal portion of Vps54, was present in a complex that contained each of the other GARP complex subunits (Fig. 2.1A). These results suggest that the N-terminal region of Vps54 is largely sufficient for complex assembly. Because Vps54-N is less stable than the full-length protein, experiments were carried out with domains expressed from high copy plasmids. Nevertheless, co-purification of Vps52 and Vps53 with Vps54-N was slightly reduced compared to full length Vps54 (Fig. 2.1A, lanes 7 and 11), indicating that the Cterminal domain of Vps54 is not required for assembly of the GARP complex but may contribute to its stability. With the exception of Vps51, loss of one GARP subunit causes the remaining subunits to be rapidly degraded [18]. We used this characteristic to determine if the N-terminal domain of Vps54 was sufficient to stabilize the other components of the GARP complex. By western blot, steady state levels of Vps52 and Vps53 were found to be approximately 80% of wild-type levels in strains expressing Vps54-N (Fig. 2.1B). Although Vps52 and Vps53 were still present in cells expressing Vps54-C, steady state levels of each protein was reduced to approximately 30% of wild type, similar to those seen when Vps54 is absent (data not shown). The ability of the N-  39  terminal region to stabilize other GARP complex subunits is consistent with the idea that the Cterminal 459 residues of Vps54 are not required for the formation of a stable core complex.  Figure 2.1.The N-terminus of Vps54 is important for GARP complex assembly and stability. Plasmids encoding HA-tagged full length Vps54 (FL; pCC9), its N-terminal region (N; pCC10), or C-terminal region (C; pCC6), were expressed in vps54∆ strains containing integrated myctagged Vps51 (CCY7, lanes 1-4), Vps52 (CCY5, lanes 5-8) or Vps53 (CCY6, lanes 9-12). A) Whole cell lysates were subjected to immunoprecipitation under non-denaturing conditions using HA antibodies and co-purifying HA- and myc-tagged GARP subunits were detected by Western blot. B) Steady-state levels of GARP subunits in the strains described above determined by anti-myc Western blotting of whole cell lysates. Protein abundance was determined by densitometry and expressed as a percentage of levels in cells expressing full length Vps54. Note that the N-terminal region of Vps54 largely stabilizes the other GARP subunits. 2.3.2 Truncation of Vps54 causes a specific block in early endosome recycling  To evaluate whether a GARP core complex lacking the C-terminal domain of Vps54 is functional for retrograde transport to the TGN, we analyzed the distribution of the marker proteins GFP-Snc1 and carboxypeptidase Y (CPY). Wild-type localization of these proteins depends on retrograde transport from early or late endosomes respectively [24, 25]; reviewed in [26] (Fig. 2.2A), and each is mislocalized in GARP deletion mutants [18, 20]. CPY, a vacuolar 40  hydrolase, is secreted from the cell if its receptor, Vps10, fails to recycle from late endosomes back to the Golgi. Overexpression of Vps54-N reduced the amount of secreted CPY in strains lacking endogenous Vps54, establishing that the GARP core complex is partially functional for late endosome recycling (Fig. 2.2B). Consistent with evidence that the Vps54 C-terminal domain is not sufficient for GARP complex assembly, overexpression of Vps54-C did not reduce the amount of CPY secreted from vps54∆ mutants. The v-SNARE Snc1 maintains its steady state plasma membrane localization by continually recycling through early endosomes and is mislocalized to the vacuole in cells lacking a functional GARP complex [20, 24]. Consistent with the partial rescue of CPY sorting defects, expression of Vps54-N partially restored the cell surface localization of Snc1 in vps54∆ mutants. In these cells, GFP-Snc1 was clearly observed at the plasma membrane at sites of polarized growth and was also detected in the vacuole (Fig. 2.2C). Defects in the early endosome recycling of Chs3 do not completely block its cell surface localization, due to an alternate retrieval pathway from late endosomes [27]. In wild type cells, Snc1 does not transit late endosomes and consequently its localization is not affected by loss of VPS5, a subunit of the late endosome-specific retromer complex [24, 28]. However, Snc1 may also be diverted into a late endosome recycling pathway when early endosome retrieval to the TGN is blocked (see Fig. 2.2A). To test this hypothesis, we evaluated the localization of GFPSnc1 in vps54∆ vps5∆ double mutants expressing the N-terminal domain of Vps54 (Fig. 2.2D). Consistent with previous results, Snc1 localization was unaffected by the vps5 mutation when full length Vps54 was present. However, the appearance of Snc1 at the plasma membrane became completely dependent on VPS5 function in cells expressing the C-terminally truncated Vps54 allele. In these cells, Snc1 was localized to structures that may correspond to highly fragmented vacuoles. This suggests that GFP-Snc1 maintains its plasma membrane localization by recycling from late endosomes when it cannot follow a retrograde pathway from early endosomes to the TGN.  41  Figure 2.2. C-terminally truncated Vps54 is functional in late endosome to TGN transport, but defective in sorting from early endosomes. A) GARP is required for retrograde traffic from both early and late endosomes. Vacuolar localization of CPY requires retrograde traffic from late endosomes (LE) as represented by the dotted line, while plasma membrane localization of Snc1 requires retrograde traffic from early endosomes (EE) as represented by a dashed line. Vps5, a subunit of the retromer, is specifically required for LE to TGN transport. B) Vps54-N partially 42  rescues the CPY sorting defect of vps54∆. Secreted CPY was detected by colony overlay assay from 1/1000 ODB600B units of vps54∆ cells (BY54∆) transformed with plasmids to express wildtype Vps54 (pCC9), Vps54-N (pCC10), Vps54-C (pCC6), or vector alone (pVT102L). C) Overexpressing the N-terminal region of Vps54 partially restores localization of Snc1 to the plasma membrane. Snc1-GFP expressed from a CEN plasmid (pCS07) in the strains described in (B) was visualized by fluorescence microscopy. D) The ability of Vps54-N to localize Snc1GFP to the plasma membrane requires late endosome-to-TGN transport. Snc1 was coexpressed with truncated forms of Vps54 in vps5∆ vps54∆ double mutants (NQY116) using plasmids described above. Overexpression of Vps54-C was not able to rescue the plasma membrane localization of GFP-Snc1, which was largely missorted to the vacuole. We also observed an unusual intracellular structure localized to sites of polarized growth in cells expressing Vps54-C that was unaffected by the presence or absence of VPS5 (Fig. 2.2C,D, discussed below). Taken together, our results indicate that the conserved C-terminal domain of Vps54 is largely dispensable for GARP complex assembly and for late endosome recycling, but is required for the retrieval of Snc1 from early endosomes. 2.3.3 The C-terminal domain of Vps54 localizes to a polarized intracellular compartment  The GARP complex is localized to the yeast trans-Golgi network (TGN) at steady state and appears as 3-5 spots by immunofluorescence microscopy [18, 19]. Consistent with our finding that the GARP core complex is partially functional for retrograde transport to the Golgi, expression of Vps54-N completely restored the TGN localization of GFP-Vps52 in vps54∆ cells (Fig. 2.3A). We have previously observed that HA-tagged Vps54 is much more difficult to visualize than Vps52-HA by fluorescence microscopy despite being present in the GARP complex in 1:1 stoichiometry, suggesting the HA epitope is masked when Vps54 is assembled in a complex. Accordingly, both full length and truncated forms of Vps54 appeared hazy and cytosolic when overexpressed, even though a substantial proportion of each protein was associated with high speed membranes by subcellular fractionation (data not shown). Interestingly, the C-terminal domain of Vps54 localized to 1-2 punctate structures with a polarized distribution in both vps54∆ and wild-type strain backgrounds (Fig. 2.3B). This localization pattern, which is atypical of the TGN, was present in approximately 20% of the cell population in vps54∆ but was less frequently observed in wild-type backgrounds. Overexpressed full length Vps54 also localized to a similar structure in approximately 5% of the cell population.  43  Figure 2.3. Differential localization of N- and C-terminal domains of Vps54 A) Vps54-N, but not Vps54-C, directs the localization of Vps52-GFP to the TGN. The vps54∆ strain (CSY51) which contains an integrated copy of Vps52-GFP, was transformed with plasmids for the expression of HA-tagged full-length (pCC9), N-terminal (pCC10) or C-terminal (pCC6) forms of Vps54 and cells were viewed by fluorescence microscopy. B) Vps54-C localized to a polarized structure. vps54∆ cells (BY54∆) expressing the Vps54 plasmids described in (A), as well as wild-type cells (BY4742) expressing Vps54-C (pCC6), were fixed, labeled with a monoclonal anti-HA antibody followed by an FITC-conjugated secondary antibody, and observed by fluorescence microscopy. The localization of Vps54-C varied throughout the cell cycle, roughly corresponding to sites of polarized growth (Fig. 2.4A). This polarized pattern is reminiscent of the early endosomal marker protein Rcy1 [29]. As seen for Rcy1, treatment with the actin inhibitor Latrunculin B resulted in the mislocalization of Vps54-C-labeled structures, which became more numerous and dispersed (Fig. 2.4B). Treatment of MATa yeast cells with the alpha factor pheromone causes G1 cell cycle arrest and the formation of mating projections referred to as shmoos. Like Rcy1, Vps54-C re-polarized to shmoo tips after alpha factor treatment (Fig. 2.4C). Following treatment with the tubulin inhibitor nocodazole, which disrupts microtubules and 44  results in G2/M cell cycle arrest, single Vps54-C labeled structures were seen at the bud neck of large-budded cells. This indicates that the polarized localization of Vps54-C requires an intact actin cytoskeleton, but not microtubules.  Figure 2.4. The C-terminal domain of Vps54 localizes to sites of polarized growth throughout the cell cycle. (A-C) HA-tagged Vps54-C (pCC6) was visualized by immunofluorescence microscopy in fixed vps54∆ cells (BY54∆). A) Representative images of unbudded cells or cells with small, medium or large buds show polarization of Vps54-C through the cell cycle. B) Localization of Vps54-C to sites of polarized growth was abolished by treatment with the actin inhibitor Latrunculin B. C) Treating cells with the mating pheromone alpha-factor induced G1 arrest and the re-localization of Vps54-C to shmoo tips. Treatment with the microtubule inhibitor nocodazole induced G2/M arrest but did not disrupt the cortical localization of Vps54-C. Cell cycle arrest was confirmed by FACS analysis of DNA content (shown on the right). Peaks corresponding to 1n, 2n and 4n DNA content are indicated. Yeast early endosomes are not morphologically well defined. Although some early endosome markers display a polarized localization similar to that seen with Vps54-C, early endosome markers such as Tlg1 localize to a compartment that is virtually indistinguishable 45  from the TGN, while other putative early endosome markers are found in smaller, more numerous, puncta [30]. Furthermore, a number of organelles have been reported to display polarized distributions in S. cerevisiae [31, 32]. Therefore, a more detailed characterization of the Vps54-C compartment is required to determine its identity. 2.3.4 Vps54-C localizes to an early endocytic compartment  The polarized structure that labeled with Vps54-C was strikingly similar to the Snc1containing compartment previously observed in cells expressing high levels of the Vps54 Cterminal domain (Fig. 2.2C,D). When Vps54-C was co-expressed with GFP-Snc1 in a strain lacking endogenous Vps54, 100% of the structures that were positive for Vps54-C also contained Snc1 (Fig. 2.5A). The trafficking pathway followed by GFP-Snc1 has been well described: GFP-Snc1 is internalized from the plasma membrane and first delivered to early endosomes before being recycled to the trans-Golgi network to be incorporated into newly forming secretory vesicles [24]. To determine if the expression of Vps54-C slows the passage of Snc1 through an organelle that is part of its normal recycling itinerary, we examined the co-localization of the Vps54-C structure with established markers of the TGN, secretory vesicles, and endocytic compartments. The late Golgi marker Sec7-GFP localized to numerous punctate structures that showed no overlap with Vps54-C (Fig. 2.5B). The secretory vesicle Rab protein, Sec4, has been shown to have a polarized localization pattern due to the accumulation of secretory vesicles at sites of cell growth [33]. Indeed, Sec4 was present at bud tips in newly budding cells, where it partially co-localized with Vps54-C (Fig. 2.5C). Vps54-C was often found in two puncta, one near the bud neck of the mother and the other near the tip of the new bud. In these cells, Sec4 co-localized only with Vps54-C present in the new bud.  46  Figure 2.5. Vps54 C-terminus co-localizes with early endocytic markers but not markers of late Golgi or secretory vesicles. A) Snc1-GFP localizes to a punctate polarized structure that colocalizes with the HA-tagged C-terminus of Vps54. B) Vps54-C does not colocalize with the late-Golgi marker Sec7-GFP in NQY111 cells C) Vps54-C and Sec4 co-localize at the bud tip but not in the mother cell. Treatment with Latrunculin B disperses polarized staining patterns and abolishes the co-localization of Vps54-C with Sec4 (D) but not that of Snc1-GFP and Vps54-C (E). F) Vps54-C colocalizes with the styryl dye FM4-64 after a short chase. vps54∆ 47  cells (BY54∆) expressing GFP-tagged Vps54-C (pCC12) were allowed to take up FM4-64 for 5min at 30˚C after labeling the plasma membrane for 15min on ice and viewed by fluorescence microscopy of unfixed cells. G) Vps54-C partially but consistently colocalizes with Rcy1-GFP (arrows). vps54∆ cells expressing GFP-tagged Rcy1 from the GAL1 promoter (pJMG95) and co-expressing HA-tagged Vps54-C from a plasmid (pCC6) were visualized by double label indirect immunofluorescence microscopy after 8h of galactose induction. In panels A-C, E and G, epitope tags were visualized by double label indirect immunofluorescence microscopy using anti-HA and anti-GFP antibodies. In D, the localization of Sec4 was visualized using a monoclonal antibody directed against Sec4. Vps54-C may be localized to two intracellular compartments, both of which contain Snc1, but only one of which corresponds to secretory vesicles. It is also possible that Vps54-C, Snc1 and Sec4 are associated with different compartments that are independently targeted to sites of polarized growth. To test this hypothesis, we took advantage of the observation that the polarized localization of these organelles depends on the actin cytoskeleton [34]. Following a 60 min treatment with Lat B, Sec4-containing secretory vesicles dispersed throughout the cell in finely punctate pattern (Fig. 2.5D). In contrast, Vps54-C and Snc1 co-localized in structures that were larger, less numerous and clearly distinct from secretory vesicles (Fig. 2.5E). To visualize early endosomal compartments, GFP-tagged Vps54-C was expressed in the vps54∆ strain and cells were incubated with the lipid dye FM4-64 for 5 minutes at 30˚C (Fig. 2.5F). FM4-64 clearly labeled Vps54-C compartments after this brief incubation, in addition to numerous small, dispersed organelles similar to previously described early endosomes [30]. Full length Vps54 has previously been found to colocalize with FM4-64 [17]. We also performed indirect immunofluorescence microscopy on vps54∆ strains co-expressing Rcy1-GFP under the control of a galactose-inducible promoter [29] together with HA-tagged Vps54-C. Rcy1-GFP and Vps54-C-HA were rarely visible in the same cells. However, when both signals were present, the proteins consistently co-localized (Fig. 2.5G). The overlap was not always precise, consistent with reports that Rcy1 is found in part on Sec7-containing Golgi compartments [32]. Taken together, these data suggest that Vps54-C localizes to an early endocytic compartment that is accessible to Snc1, FM4-64 and a proportion of Rcy1. By conventional electron microscopy, clusters of 100-200 nm irregularly shaped membrane compartments were observed in or near the small buds of cells expressing Vps54-C that were not found in control strains (Fig. 2.6). These compartments, which have lumenal content of varying densities and appear to contain internal membranes, were frequently found in two distinct groupings within the bud and near the bud neck of the mother cell, reminiscent of 48  the pattern seen by fluorescence microscopy. Double indirect immunoelectron microscopy performed on cryo-sections indicated that Vps54-C was associated with clusters of membranes near the bud neck that were often close to, but not coincident with, patches of actin. Vps54-C appeared to be associated with the limiting membrane of vesicular compartments (Fig 2.6E, arrows) but was more frequently found along disordered membrane profiles that may correspond to tubular elements (arrowheads). These observations suggest that Vps54-C localizes to aberrant compartments that may correspond to enlarged, clustered early endosomes.  Figure 2.6. Morphology of Vps54-C expressing cells by electron microscopy (A-C) Clusters of irregular 100-200nm organelles in the bud and in the mother cell near the mother-bud junction seen in representative cells from a vps54∆ strain (LCY200) expressing Vps54-C (pCC6) but not vector alone are indicated with an asterisk. D) Enlarged view of structures that are boxed in (C). E) Immunoelectron microscopy of the same strain double labeled with anti-HA antibodies conjugated to 5nM gold particles, and anti-actin particles  49  conjugated to 10nM gold particles. HA staining is found on clustered membrane profiles (closed arrowheads) and on the limiting membrane of vesicular profiles (arrows). Scale bars = 0.20µm. 2.3.5 Point mutations within the C-terminal region of Vps54 prevent localization to the polarized early endosome and block early endosome recycling  The C-terminal region of Vps54 contains two conserved stretches of amino acids that may form independent subdomains (Fig. 2.7A). When expressed individually, neither region was sufficient for endosome localization (Fig. 2.7B). However, alanine substitution of residues E689 and W691, which lie within the first conserved region, completely abolished localization of Vps54-C to the polarized structure. These residues are invariant in species ranging from yeast to man, suggesting that the C-terminus of Vps54 engages in conserved interactions that are important for recruitment to early endosomes.  Figure 2.7. Conserved residues in the C-terminus of Vps54 are required for localization to the polarized, early endocytic structure. A) Alignment of Vps54 from S. cerevisiae with homologs in S. pombe (O14093), D. melonogaster (Q9VLC0), C. elegans (Q22639) and human (NP_057600) showing evolutionarily conserved regions in the C-terminal domain. Gray shading shows conservation across 0.5 sequences, while black shading indicates more highly conserved or invariant residues. Arrows indicate residues that were mutated to create Vps54PEWP 50  or Vps54PEWNSP. Sequence alignment was generated with T-coffee [55] B) vps54∆ strains (BY54∆) expressing either Vps54-C (pCC5), truncated forms that each contain one of the conserved regions in the C-terminus (Vps54-CP593-735P, pCC13; Vps54-CP724-889P, pCC4), or Vps54-C harboring point mutations (Vps54-CPEWP, pCC19) from a galactose inducible promoter were grown to log phase in galactose, fixed and processed for immunofluorescence microscopy. Point mutations that abolish the localization of Vps54-C might also be expected to disrupt the early endosome trafficking function of full length Vps54. However, we did not observe an appreciable change in GFP-Snc1 plasma membrane localization when the corresponding mutant, Vps54PEWP, was expressed at endogenous levels in either vps54∆ or vps54∆ vps5∆ strains (Fig. 2.8A). Because redundant interactions contribute to the membrane association of other tethering factors, two highly conserved residues in the second subdomain of Vps54-C, N805 and S806, were also changed to alanine. The resulting quadruple mutant form of Vps54 (Vps54PEWNSP), when expressed from its endogenous promoter, was present at the same level as wild type Vps54 (not shown). Furthermore, this mutant fully complemented the CPY and Snc1 sorting defects of a vps54∆ strain (Fig. 2.8B). Strikingly, the Vps54 quadruple mutant caused the localization of Snc1 to become completely dependent on retrograde trafficking from late endosomes. In vps54∆ vps5∆ cells expressing Vps54PEWNSP, Snc1 was no longer observed at the cell surface and instead was missorted to fragmented vacuolar compartments (Fig. 2.8A), a result that is similar to that seen for Vps54-N. However, Vps54-N was only partially functional for late endosome retrieval even when expressed at high levels, whereas Vps54P EWNSP was stable and fully functional for late endosome retrieval when present at endogenous levels. These data demonstrate that Vps54 contains a conserved determinant that is specifically required for retrograde trafficking from the early endosome.  51  Figure 2.8. Point mutations in the C-terminal Vps54 domain abolish sorting from early endosomes without affecting late endosome to TGN transport. A) The early endosome sorting defect of Vps54PEWNSP is apparent in cells deficient in late endosome recycling. Vps54 (pLC104) and the indicated mutant forms (Vps54PEWP, pCC14; Vps54PNSP, pCC15; Vps54PEWNSP, pCC16; Vps54N, pCC10) were co-expressed with Snc1-GFP (pCS07) in vps54∆ vps5∆ (NQY116) double mutant strains or in a vps54∆ single mutant (BY54∆). Snc1-GFP was visualized by fluorescence microscopy of fixed cells. Neither the E689A W691A nor N805A S806A mutations alone were sufficient to destroy the early endosome sorting function of Vps54, but plasma membrane localization of Snc1 was abolished in vps5∆ strains expressing the Vps54 quadruple mutant. B) The Vps54PEWNSP mutant remains functional for late endosome to TGN transport. CPY secretion was determined by a colony overlay assay for vps54∆ strains (BY54∆) expressing wild type and mutant forms of Vps54 as described in (A). Please see post-publication erratum (2.5; Appendix A)  52  2.3.6 GARP is not required for endocytic recycling of the a-factor receptor Ste3  Ste3, the receptor for the mating pheromone a-factor, undergoes both constitutive and ligand-induced endocytosis through separable mechanisms [35]. Truncating the C-terminal 105 residues of Ste3 blocks constitutive endocytosis but not ligand-stimulated uptake [35]. In the presence of ligand, Ste3∆365 enters early/recycling endosomes that may appear as single polarized structures in the mating projections of highly polarized shmoos. It was recently shown that a truncation mutant of the exocyst component Sec5 is defective for endocytic recycling of the pheromone receptor Ste3 but remains competent for Snc1 recycling and for exocytosis [11]. This suggests that the exocyst has an additional role at an early endosome trafficking step that does not affect Snc1. To determine if GARP is required for the recycling of Ste3∆365 from the yeast early/recycling endosome, we expressed either full length Vps54 or its N-terminal domain in a vps54∆ strain in which endogenous Ste3 was truncated by integrating a GFP tag (Fig. 2.9). As previously described, Ste3∆365GFP was present on both the plasma membrane and the vacuole in the absence of pheromone, and redistributed to intracellular structures following afactor treatment to stimulate endocytosis. Removal of a-factor induced the rapid redistribution of Ste3∆365GFP to the plasma membrane. However, the ability of Ste3∆365GFP to recycle back to the cell surface was unaffected by expression of Vps54-N or -C terminal domains, or even by complete loss of Vps54 (Fig. 2.9; data not shown). The finding that endocytic recycling of Ste3 does not require GARP-dependent retrograde transport to the TGN raises the possibility that Snc1 and Ste3 may transit different classes of early endosomes during their transport.  53  Figure 2.9. Vps54 is not required for recycling of Ste3∆365-GFP. A) vps54∆ Ste3∆365-GFP (NQY113) cells expressing full length Vps54 (pCC9), empty vector or Vps54-C (pCC6) were visualized by fluorescence microscopy before treatment with a-factor (t = 0), immediately after a 45 minute incubation with a-factor (t = 45’ a-factor) and 30 minutes after removing a-factor from the media (t = 30’ wash). Ste3∆365-GFP is recycled from endosomes back to the plasma membrane in both wild type (Vps54) and vps54∆ cells after the 30 minute wash at 30˚C. Expression of Vps54-C does not impair Ste3∆365-GFP recycling or affect its intracellular localization.  2.4 Discussion GARP is a multi-subunit tethering complex that directs the Golgi targeting of retrograde vesicles derived from both early and late endosomes. We find that mutations within a conserved C-terminal domain of the GARP subunit Vps54 specifically abolish retrograde trafficking from early endosomes to the Golgi, without affecting transport on the late endosome-to-Golgi recycling pathway. Furthermore, this C-terminal domain of Vps54 is necessary and sufficient for association with an organelle that has the characteristics of an early endosome. These results suggest that separate domains within the GARP tethering complex dictate recruitment to upstream compartments (Fig. 2.10). The incorporation of tethering factors into budding transport 54  vesicles may ensure that newly formed vesicles contain targeting information to specify their ultimate destination.  Figure 2.10. Model of Vps54 function in retrograde transport from early endosomes. A) In wild type cells, distinct domains within the GARP complex target the complex to retrograde vesicles budding from early and late endosomal membranes. GARP subsequently directs the tethering of these vesicles to the TGN through interactions with Rab and SNARE proteins (not shown). B) Loss of the Vps54-C domain (dark grey appendage) specifically blocks GARP complex recruitment to early endosomes, resulting in the rerouting of Snc1 to late endosomes (light grey arrow), where it is retrieved via the retrograde pathway used by Vps10 (grey arrow). According to this model, a separate, as-yet-unidentified domain (light grey appendage) recruits GARP to late endosomes. 2.4.1 Assembly of the GARP complex  Members of the quatrefoil family of tethering complexes, which includes COG, the exocyst and GARP, are related by conserved sequences in their N-termini that include predicted amphipathic helices thought to be important for complex formation [1]. The predicted amphipathic helix within the N-terminal portion of Vps54 was previously shown to be necessary, but not sufficient, for GARP assembly [22]. Here, we find that a larger N-terminal region consisting of 430 amino acids that includes this helical motif is sufficient to direct the formation of a GARP core complex. In contrast, a C-terminal domain containing the last 296 aa of Vps54 is largely dispensable for complex assembly, but directs the recruitment Vps54 to early endosomes. Computer modeling based on Ginzu domain predictions (Robetta; [36, 37]) indicate that the N-terminal and C-terminal domains may be separated by a long central portion with significant homology to myosin heavy chain (unpublished results). This arrangement may 55  position the C-terminal domain at some distance from the central core region where it would be accessible to membrane recruitment factors. 2.4.2 Recruitment of GARP to upstream compartments  At steady state, many yeast tethering complexes are localized to the compartment at which they function in docking and fusion [18, 23, 33]. Such observations led to early models that tethering complexes act as a "molecular fly paper" at the target membrane to trap specific classes of transport vesicles [38]. Our data support an alternate model (Fig. 2.10), that tethering complexes associate with transport vesicles at the time of their formation and may even contribute to cargo sorting. The localization of the C-terminal domain of Vps54 to an early endosomal compartment suggests that GARP itself may become incorporated into forming transport vesicles as they bud from the donor compartment. Furthermore, Snc1 does not appear to accumulate in transport vesicles in a Vps54PEWNS Pmutant but is instead missorted to late endosomes, suggesting that loss of functional GARP complex impairs either the sorting of Snc1 into retrograde vesicles at the early endosome or formation of the vesicles themselves. However, missorting at the early endosome may be a secondary defect resulting from mislocalization of critical sorting factors in GARP mutants. There is evidence that the processes of tether recruitment and vesicle formation are tightly coupled for other tethering factors. Vps41, a subunit of the HOPS tethering complex involved in vacuole fusion, associates with AP3-coated vesicles destined to fuse with the vacuole through an interaction between an N-terminal region of Vps41 and the AP3 subunit Apl5 [16]. Mutations that block the interaction of Vps41 with the AP3 adaptor complex also block the formation of the AP3-coated transport intermediate. Other tethers have also been shown to interact with vesicle coat proteins. The polarized cell-specific adaptor coat protein AP-1B may play a role in the recruitment of at least two subunits of the exocyst complex to transferrin receptor positive endosomes [39]. What recruits Vps54-C to the polarized early endosome? By analogy with other tethering complexes, Vps54 may interact with components of a vesicle coat that is required specifically for the retrograde sorting of Snc1. However, coat proteins specific for early endosome retrieval have not been characterized. The candidate coat protein Snx4 is likely involved in retrieval of Snc1 at both early and late endosomes, because Snc1 is predominantly missorted to the 56  vacuole in its absence [40]. Furthermore, Snx4 and its interacting partners Snx41 and Snx42 are found at the late endosome/MVB where they colocalize with Snf7 [41]. Gene deletions of other candidates such as Ypt6 and Arl1 [19, 42], which are known GARP interactors, and Arl3, which is required for localization of Arl1, do not prevent association of Vps54-C with the polarized endocytic compartment (unpublished results). Therefore, further work will be required to identify the binding partner of the Vps54-C domain. Although mutation of only one of the two conserved regions within the Vps54 C-terminal domain was sufficient to prevent the localization of the Vps54-C domain to early endosomes, mutation of both regions was required to block early endosome retrieval in vivo. Both may contribute to a single functional interface that retains only partial function in Vps54PEWP mutants. Alternatively, Vps54 may have two binding motifs that interact redundantly with determinants at the early endosome membrane. In fact, the ER tethering factor Dsl1 interacts with two different subunits of the COPI coat through independent binding sites, and knockout of both motifs is required for complete loss of COPI binding in vitro and of loss of function in vivo [43]. 2.4.3 Early endosome recycling pathways  Irregular 100–200 nm organelles were observed by electron microscopy only in Vps54-C expressing cells, suggesting that the overexpression of this domain induces the expansion of a normally dynamic early endosomal compartment. However, GFP-tagged forms of other proteins involved in early endosome recycling, including Gyp2 and Rcy1, are observed in similar structures when expressed at endogenous levels [29],[44], implying that a morphologically similar early endosome also exists in wild-type cells. A truncation mutant of the exocyst component Sec5 was previously shown to be impaired in recycling the a-factor receptor, Ste3, from early endosomes but was not defective for Snc1 sorting [11]. We observe the opposite phenotype for a truncated vps54 mutant: a defect in the recycling of Snc1, but not Ste3. Snc1 may recycle from the early endosome to the Golgi, a pathway that requires GARP, while Ste3 may be sorted directly to the plasma membrane. Alternatively, the plasma membrane recycling of Snc1 and Ste3 could involve different endocytic compartments that are not well characterized in yeast but which may be analogous to the sorting and recycling endosomes of mammalian cells. Recently, human homologues of GARP complex subunits were shown to form a complex that localizes to the 57  TGN and to a perinuclear region described as the recycling endosome [45]. Because the Cterminal region of Vps54 is evolutionarily conserved, the interactions that direct the recruitment of GARP to early endosomes are likely to be present in higher cells. Because late endosomes are thought to mature from early endosomes (for review see [46]), it was conceivable that GARP might interact with a targeting determinant common to both. The discovery of a domain with a specific role in retrograde transport from early endosomes indicates that the GARP complex is likely to recognize distinct determinants on each of the two classes of endosomes and implies that another domain mediates retrograde transport from late endosomes. Further work will be directed towards identifying additional functional domains in the remaining GARP subunits. How tethering complexes are able to recognize and distinguish multiple membranes is important to our understanding of how organelle identity is determined. Our observations support a model for tethering complex function whereby distinct regions of the complex recognize features of each membrane with which it interacts.  2.5 Revision post-publication Subsequent to publication, it was discovered that the Vps54 mutant, Vps54EWNS, is fully functional for Snc1 sorting, contrary to what is shown in Figure 2.8. Further investigation showed the discrepancy arose from an inadvertent error in strain construction. We repeated and expanded our findings with Vps54-N to quantify the degree of Snc1 missorting in vps54∆ and vps54∆vps5∆ strain backgrounds in an erratum submitted to Molecular Biology of the Cell (Appendix A). The reliance of Vps54-N on late-endosome-to-Golgi retrograde transport supports the model that the C-terminal region of Vps54 is important for retrieval from early endosomes, and our conclusions remain unchanged. This erratum was submitted and accepted by the editors of MBC in July 2007, but has yet to be published.  2.6 Materials and Methods Strain and Plasmid Construction Yeast strains used in this study are described in Table 2.1. SpeI-digested pSEC7-EGFPx3 [31] (gift from B. Glick) was used to integrate a C-terminal GFP tag at the SEC7 locus in strain BY54∆, creating NQY111. NQY108 was made by replacing the KanMX module with NatMX in strain BY54∆ using EcoRI-digested p4339 [47]. To create a chromosomally encoded Ste3∆365GFP mutant, a GFP tag was inserted immediately after residue 365 of Ste3 in LCY200 [18] by 58  homologous recombination [48]. Strain NQY116 was created by the SGA method [47] by mating a vps5∆::kanMX4 strain with a vps54∆::natMX query strain (NQY110) that contained the URA3based, VPS54-complementing plasmid pLC104. Double mutants lacking pLC104 were selected on 5-fluoroorotic acid (5-FOA). LCY230 was created by transformation with pLC89 linearized with HpaI [18]. VPS54 was deleted from LCY230, LCY228 and LCY320 [18, 20], creating CCY5, CCY6 and CCY7 respectively, by transformation with a PCR fragment amplified from BY4741 vps54∆::KanMX4 genomic DNA using oligonucleotides homologous to the 5’ and 3’ UTR of VPS54. CSY18 was created by transforming BY5563 [47] with BglII-linearized pVPS52-EGFPx3 (gift from B. Glick). CSY51 was created by mating CSY18 with BY54∆ and selecting for MATa progeny using the SGA method [47]. To express the C-terminus of Vps54 under control of the Gal1 promoter, an EcoRV/SalI fragment from pLC104 [18] was ligated to EcoRV-SalI cut p423-GAL1 [49], creating pCC5. A HpaI-SalI fragment from pLC104 was ligated to EcoRV-SalI digested p423-GAL1 to create pCC4. To create high copy LEU2 (pCC6) and URA3 (pLC166)-based plasmids expressing the last 296 residues of Vps54 (as well as a C-terminal 3HA tag) from the ADH promoter, pLC104 [18] was digested with XbaI and SalI and sub-cloned into XbaI/XhoI-cut pVT102L and pVT102U, respectively [50]. The plasmids described below were created by homologous recombination. Briefly, DNA fragments were co-transformed with linearized vector into yeast, and recombinant plasmids were recovered in E. coli using standard methods. To create pCC9, encoding full length, HAtagged Vps54, VPS54 sequences were amplified from pLC165 using oligos with homology to pVT102L and VPS54, and co-transformed with XbaI-linearized pCC6. A Vps54 truncation mutant lacking the C-terminal 459 residues was created by integration of a 3xHA::KanMX cassette in pLC104, to create pLC165. Sequences encoding the N-terminal region of Vps54 were introduced into pVT102L by co-transformation of a gel-purified 3kb Bsu36I/SalI-digested fragment of pLC165 with PstI-cut pCC9 to create pCC10. To introduce a GFP tag at the 3’ end of the Vps54 C-terminal fragment in pVT102U, GFP sequences were amplified using oligos homologous to 3’ end of the Vps54 open reading frame and co-transformed with XbaI-digested pLC166, creating pCC12. pCC13 was generated by co-transformation of a PCR product encoding residues 597 - 735 of Vps54 followed by an HA-tag encoded in the reverse oligo, with NcoI-digested pCC5 to yield a construct that expresses residues 593 – 735 followed by an HAtag. 59  To create E689A W691A point mutations in Vps54, a forward primer with mismatches at residues E689 and W691 was used to amplify the C-terminal region of Vps54, using pCC9 as a template. This PCR product was co-transformed with HpaI-cut pCC5 to create pCC19, and with HpaI-cut pLC104 to create pCC14. Similarly, an 85bp reverse primer with mismatches at residues N805 and S806 was used to amplify an 800bp fragment of VPS54, which was cotransformed with HpaI-cut pLC104 to create pCC15. The quadruple point mutant of Vps54 (Vps54PEWNSP) was created by amplifying a 500bp fragment of Vps54 using both forward and reverse mutagenic primers described above. This PCR product was co-transformed with HpaIcut pLC104 to create pCC16. Oligonucleotide sequences are available upon request.  60  Table 2.1: Yeast strains used in Chapter 2 Strain Name BY5563 BY54∆ CCY5 CCY6 CCY7 CSY18 CSY51 LCY200 LCY230 NQY108 NQY110 NQY111 NQY116 NQY113 Y7096  Genotype  Source  MATα his3∆, leu2∆, ura3∆, met15∆ LYS2+ can1∆::MFA1pr-HIS3, lyp1∆ MATa his3∆1 leu2∆0 met15∆0 ura3∆0 vps54∆::KanMX4 MATα ura3-52 leu2-3,112 his3-200 trp1-901 lys2801 suc2-9 pho8::ADE2 VPS52-myc::HIS5 vps54∆::KanMX4 MATα ura3-52 leu2-3,112 his3-∆200 trp1-901 lys2-801 suc2-∆9 pho8∆::ADE2 vps53::TRP1::VPS53-myc vps54∆::KanMX4 MATα ura3-52 leu2-3,112 his3-200 trp1-901 lys2801 suc2-9 pho8::ADE2 VPS51-myc::HIS5 vps54∆::KanMX4 MATα his3∆, leu2∆, ura3 ∆, met15∆ LYS2+ can1∆::MFA1pr-HIS3, lyp1∆ VPS52-EGFPx3URA3 MATa his3∆1 leu2∆0 met15∆0 ura3∆0 Vps52GFP::URA3, vps54∆::KanMX4 MATα ura3-52 leu2-3,112 his3-200 trp1-901 lys2801 suc2-9 pho8∆ADE2 vps54∆TRP1 MATα ura3-52 leu2-3,112 his3-200 trp1-901 lys2801 suc2-9 pho8::ADE2 VPS52-myc::HIS5 MATa his3∆1 leu2∆0 met15∆0 ura3∆0 vps54∆::NatMX4 MATα, can1DEL::STE2pr-Sp_his5 lyp1∆ cyh2, his3∆1, leu2∆0, ura3∆0, met15∆0, LYS2+, vps54∆-NatMX4 MATa his3∆1 leu2∆0 met15∆0 ura3∆0 Sec7GFP::URA3, vps54∆::KanMX4 MATa can1∆::STE2pr-Sp_his5 lyp1-∆cyh2, his3∆1, leu2∆0, ura3∆0, met15∆0, LYS2+, Vps54∆NatMX4; vps5∆-KanMX4 MATα ura3-52 leu2-3,112 his3-200 trp1-901 lys2801 suc2-9 pho8∆ADE2 vps54∆TRP1 Ste3DEL365-GFP::HisMX6 Matα, his3∆1 leu2∆0 ura3∆0 met15∆0 LYS2+, can1∆::STE2pr-Sp_his5 lyp1∆ cyh2  [56] Open Biosystems This study This study This study This study This study [18] This study This study This study This study This study This study A. Tong  Immunoprecipitations and Western blotting 10 ODB600B units of log-phase cells were spheroplasted and stored at -80 as described previously [51]. Frozen spheroplasts were lysed in 1mL NP40 buffer (50mM Tris pH 8.0, 50mM NaCl, 2mM EDTA, 1.2M sorbitol, 1% NP40, 1% PMSF). Clarified lysates were adjusted to equivalent protein concentrations as determined by Bradford assay (Bio-Rad), incubated with rabbit anti-HA 61  antibodies for 2 hours at 4°C, and precipitated with Protein G-sepharose (Amersham Biosciences). After washing in NP40 buffer, immunoprecipitated proteins were released with Thorner Buffer (8M Urea, 5% SDS, 50mM Tris pH 6.8) and subjected to SDS-PAGE. Coimmunoprecipitating proteins were detected by Western blot using standard blotting procedures with mouse anti-myc 4A6 (Upstate) or mouse anti-HA (12CA5) antibodies followed by HRPconjugated secondary goat-anti-mouse antibody (Bio-Rad). Detection and densitometry was carried out on a Fluoro-S MultiImager (Bio-Rad) using Quantity One software. Stability of GARP subunits was determined from yeast cell lysates prepared from spheroplasts as described above. 1 ODB600B unit of each lysate was subjected to SDS-PAGE and Western blotting using monoclonal anti-myc 4AC (Upstate). Secretion of carboxypeptidase Y was determined using a colony immunoblot assay [52]. Fluorescence Microscopy Indirect immunofluorescence microscopy was carried out as described [18, 52] using the following antibodies: cross-absorbed rabbit anti-HA (gift of T. Stevens), mouse anti-HA (HA.11 BAbCo), rabbit anti GFP serum (Molecular Probes), anti-Sec4 mAb C1.2.3 (gift of P. Novick). Secondary antibodies (Molecular Probes) included goat anti-mouse Alexa Fluor 488 or 594 and goat anti-rabbit Alexa Fluor 488 or 594. To follow proteins expressed under the GAL1 promoter, strains were grown overnight in selective media supplemented with 2% raffinose (for GAL induction experiments) or 2% galactose (for steady state expression) and allowed to double twice in YEP-Galactose before fixation and immunoflourescence. Fluorescence microscopy of GFP-expressing strains was performed by growing strains to log phase in selective media, fixing with 4% paraformaldehyde for 20 minutes at room temperature and resuspending in 1xPBS. Membranes were stained with the styryl dye FM4-64 (Invitrogen) by incubating cells on ice for 15 min with 40µM FM4-64 in YPD. Cells were resuspended in pre-warmed YPD, incubated at 30°C for 5 minutes, and resuspended in 1% NaN3, 1% NaF, 100mM Tris pH 8.0. Cells were viewed using a Zeiss Axioplan2 fluorescence microscope, and images were captured with a CoolSnap camera using MetaMorph software and adjusted using AutoQuant Deconvolution and Adobe Photoshop. Electron Microscopy Conventional and immunoelectron microscopy were performed essentially as previously described [53].  62  Conventional electron microscopy. Briefly, cells were grown in YPD medium at 26°C to an A600 of 0.5, harvested by centrifugation, and resuspended in fix (3% glutaraldehyde, 0.1 M sodium cacodylate, 5 mM CaCl2, pH 7.4). Cells were fixed for 1 h at 25°C, washed, and resuspended in M citric acid). The fixed cells were 1.2 M sorbitol in phos-citrate buffer (0.1 M KB2HPOB 4/0.033 B B then treated with glucuronidase and zymolyase for 1 h to remove the cell walls. Cells were embedded in 2% ultra-low gelling temperature agarose (Sigma, typeIX), stained with osmium/thiocarbohydrazide/osmium, en bloc stained in Kellenberger's UA (o.n.), and subsequently embedded in low viscosity Spurr resin. Sections were post-stained with lead citrate and uranyl acetate. 80nm sections were cut on a Leica UCT ultramicrotome and examined on a Philips EM 410 at 80kV. Images were captured with a Soft Imaging System Megaview III camera (Lakewood, CO). Panels were assembled in Adobe Photoshop with only linear adjustments in brightness and contrast. Immunoelectron microscopy. Briefly, exponentially growing cells (30 °C) were fixed in suspension for 15 min by adding an equal volume of freshly prepared 8% formaldehyde in 1X PBS, pH 7.4. The cells were pelleted, resuspended in fresh fixative (4% formaldehyde, 1X PBS, pH 7.4) and incubated for an additional 18–24 h at 4 °C. The cells were washed briefly in PBS and resuspended in 1% low-gelling-temperature agarose. The agarose blocks were trimmed into pieces of 1 mmP3P, cryoprotected by infiltration with 2.3 M sucrose/20% polyvinyl pyrrolidone, pH 7.4, for 2 h, mounted on cryo-pins and rapidly frozen in liquid nitrogen. Ultrathin cryosections were cut on a Leica UCT ultramicrotome equipped with an FC-S cryostage and collected onto formvar/carbon-coated nickel grids. The grids were washed through several drops of 2.5% fetal calf serum (FCS), 1XPBS containing 10 mM glycine (pH 7.4), blocked in 10% FCS for 30 min and incubated overnight in 20 mg ml–1 monoclonal anti-HA antibody (Covance Research Products, Richmond, CA), and rabbit anti-actin antibody. After washing, the grids were incubated for 2 h in 5nm Au donkey anti-mouse and 10nm Au donkey anti-rabbit conjugates (Jackson ImmunoResearch). The grids were washed through several drops of PBS followed by several drops of ddH2O. Grids were then embedded in an aqueous solution containing 3.2% polyvinyl alcohol (10 K)/0.2% methyl cellulose (400 centiposes)/0.1% uranyl acetate. Sections were examined on a Philips 410 transmission electron microscope at 100 kV. Images were captured with a Soft Imaging System Megaview III camera (Lakewood, CO). Panels were assembled in Adobe Photoshop with only linear adjustments in brightness and contrast.  63  Ste3 Recycling Assay Yeast strains were grown to log phase in selective media and were resuspended in YPD to a concentration of 0.75 ODB600B/mL and allowed to grow at 30°C for one hour. 0.75 ODB600B units of each strain was removed and 5mM NaNB3B added for the t=0 time point. A O.5 volume of media containing a-factor (prepared as described in [35] was added to the remainder, and cells were incubated at 30°C for 45 minutes to induce the internalization of Ste3∆365-GFP. 0.75 ODB600B units of each strain was removed and 5mM NaNB3B added for the t=45 minutes time point. The remaining cells were washed in YPD and allowed to incubate at 30°C for various time points, at which time 5mM NaN3 was added. Time points were washed in 1xPBS and GFP visualized by fluorescence microscopy. Other techniques For inhibitor experiments, log phase cultures were treated at 30°C for 2h with 5µg/mL alpha factor or 15µg/mL nocodazole, or for 1h with 200µM Latrunculin B (Sigma) before being processed for immunofluorescence. FACS analysis was performed as described [54].  64  2.7 References 1.  Whyte JR, Munro S: Vesicle tethering complexes in membrane traffic. J Cell Sci 2002, 115(Pt 13):2627-2637.  2.  Lupashin V, Sztul E: Golgi tethering factors. Biochim Biophys Acta 2005, 1744(3):325339.  3.  Munro S: Organelle identity and the organization of membrane traffic. Nat Cell Biol 2004, 6(6):469-472.  4.  Guo W, Roth D, Walch-Solimena C, Novick P: The exocyst is an effector for Sec4p, targeting secretory vesicles to sites of exocytosis. EMBO J 1999, 18(4):1071-1080.  5.  Guo W, Tamanoi F, Novick P: Spatial regulation of the exocyst complex by Rho1 GTPase. Nat Cell Biol 2001, 3(4):353-360.  6.  Zhang X, Bi E, Novick P, Du L, Kozminski KG, Lipschutz JH, Guo W: Cdc42 interacts with the exocyst and regulates polarized secretion. J Biol Chem 2001, 276(50):46745-46750.  7.  Boyd C, Hughes T, Pypaert M, Novick P: Vesicles carry most exocyst subunits to exocytic sites marked by the remaining two subunits, Sec3p and Exo70p. J Cell Biol 2004, 167(5):889-901.  8.  Roumanie O, Wu H, Molk JN, Rossi G, Bloom K, Brennwald P: Rho GTPase regulation of exocytosis in yeast is independent of GTP hydrolysis and polarization of the exocyst complex. J Cell Biol 2005, 170(4):583-594.  9.  Prigent M, Dubois T, Raposo G, Derrien V, Tenza D, Rosse C, Camonis J, Chavrier P: ARF6 controls post-endocytic recycling through its downstream exocyst complex effector. J Cell Biol 2003, 163(5):1111-1121.  10.  Zhang XM, Ellis S, Sriratana A, Mitchell CA, Rowe T: Sec15 is an effector for the Rab11 GTPase in mammalian cells. J Biol Chem 2004, 279(41):43027-43034.  11.  Sommer B, Oprins A, Rabouille C, Munro S: The exocyst component Sec5 is present on endocytic vesicles in the oocyte of Drosophila melanogaster. J Cell Biol 2005, 169(6):953-963.  12.  Suvorova ES, Duden R, Lupashin VV: The Sec34/Sec35p complex, a Ypt1p effector required for retrograde intra-Golgi trafficking, interacts with Golgi SNAREs and COPI vesicle coat proteins. J Cell Biol 2002, 157(4):631-643.  65  13.  Morsomme P, Riezman H: The Rab GTPase Ypt1p and tethering factors couple protein sorting at the ER to vesicle targeting to the Golgi apparatus. Dev Cell 2002, 2(3):307-317.  14.  Mehta SQ, Hiesinger PR, Beronja S, Zhai RG, Schulze KL, Verstreken P, Cao Y, Zhou Y, Tepass U, Crair MC et al: Mutations in Drosophila sec15 reveal a function in neuronal targeting for a subset of exocyst components. Neuron 2005, 46(2):219232.  15.  Zolov SN, Lupashin VV: Cog3p depletion blocks vesicle-mediated Golgi retrograde trafficking in HeLa cells. J Cell Biol 2005, 168(5):747-759.  16.  Darsow T, Katzmann DJ, Cowles CR, Emr SD: Vps41p function in the alkaline phosphatase pathway requires homo-oligomerization and interaction with AP-3 through two distinct domains. Mol Biol Cell 2001, 12(1):37-51.  17.  Conboy MJ, Cyert MS: Luv1p/Rki1p/Tcs3p/Vps54p, a yeast protein that localizes to the late Golgi and early endosome, is required for normal vacuolar morphology. Mol Biol Cell 2000, 11(7):2429-2443.  18.  Conibear E, Stevens TH: Vps52p, Vps53p, and Vps54p form a novel multisubunit complex required for protein sorting at the yeast late Golgi. Mol Biol Cell 2000, 11(1):305-323.  19.  Siniossoglou S, Pelham HR: An effector of Ypt6p binds the SNARE Tlg1p and mediates selective fusion of vesicles with late Golgi membranes. EMBO J 2001, 20(21):5991-5998.  20.  Conibear E, Cleck JN, Stevens TH: Vps51p mediates the association of the GARP (Vps52/53/54) complex with the late Golgi t-SNARE Tlg1p. Mol Biol Cell 2003, 14(4):1610-1623.  21.  Reggiori F, Wang CW, Stromhaug PE, Shintani T, Klionsky DJ: Vps51 is part of the yeast Vps fifty-three tethering complex essential for retrograde traffic from the early endosome and Cvt vesicle completion. J Biol Chem 2003, 278(7):5009-5020.  22.  Siniossoglou S, Pelham HR: Vps51p links the VFT complex to the SNARE Tlg1p. J Biol Chem 2002, 277(50):48318-48324.  23.  Whyte JR, Munro S: The Sec34/35 Golgi transport complex is related to the exocyst, defining a family of complexes involved in multiple steps of membrane traffic. Dev Cell 2001, 1(4):527-537.  24.  Lewis MJ, Nichols BJ, Prescianotto-Baschong C, Riezman H, Pelham HRB: Specific Retrieval of the Exocytic SNARE Snc1p from Early Yeast Endosomes. Mol Biol Cell 2000, 11(1):23-38. 66  25.  Fiedler TA, Karpova TS, Fleig U, Young ME, Cooper JA, Hegemann JH: The vesicular transport protein Cgp1p/Vps54p/Tcs3p/Luv1p is required for the integrity of the actin cytoskeleton. Mol Genet Genomics 2002, 268(2):190-205.  26.  Conibear E, Stevens TH: Multiple sorting pathways between the late Golgi and the vacuole in yeast. Biochim Biophys Acta 1998, 1404(1-2):211-230.  27.  Valdivia RH, Baggott D, Chuang JS, Schekman RW: The yeast clathrin adaptor protein complex 1 is required for the efficient retention of a subset of late Golgi membrane proteins. Dev Cell 2002, 2(3):283-294.  28.  Seaman MN, McCaffery JM, Emr SD: A membrane coat complex essential for endosome-to-Golgi retrograde transport in yeast. J Cell Biol 1998, 142(3):665-681.  29.  Galan JM, Wiederkehr A, Seol JH, Haguenauer-Tsapis R, Deshaies RJ, Riezman H, Peter M: Skp1p and the F-box protein Rcy1p form a non-SCF complex involved in recycling of the SNARE Snc1p in yeast. Mol Cell Biol 2001, 21(9):3105-3117.  30.  Vida TA, Emr SD: A new vital stain for visualizing vacuolar membrane dynamics and endocytosis in yeast. J Cell Biol 1995, 128(5):779-792.  31.  Rossanese OW, Reinke CA, Bevis BJ, Hammond AT, Sears IB, O'Connor J, Glick BS: A role for actin, Cdc1p, and Myo2p in the inheritance of late Golgi elements in Saccharomyces cerevisiae. J Cell Biol 2001, 153(1):47-62.  32.  Chen SH, Chen S, Tokarev AA, Liu F, Jedd G, Segev N: Ypt31/32 GTPases and their novel F-box effector protein Rcy1 regulate protein recycling. Mol Biol Cell 2005, 16(1):178-192.  33.  TerBush DR, Novick P: Sec6, Sec8, and Sec15 are components of a multisubunit complex which localizes to small bud tips in Saccharomyces cerevisiae. J Cell Biol 1995, 130(2):299-312.  34.  Ayscough KR, Stryker J, Pokala N, Sanders M, Crews P, Drubin DG: High rates of actin filament turnover in budding yeast and roles for actin in establishment and maintenance of cell polarity revealed using the actin inhibitor latrunculin-A. J Cell Biol 1997, 137(2):399-416.  35.  Chen L, Davis NG: Recycling of the yeast a-factor receptor. J Cell Biol 2000, 151(3):731-738.  36.  Kim DE, Chivian D, Baker D: Protein structure prediction and analysis using the Robetta server. Nucleic Acids Res 2004, 32(Web Server issue):W526-531.  67  37.  Chivian D, Kim DE, Malmstrom L, Bradley P, Robertson T, Murphy P, Strauss CE, Bonneau R, Rohl CA, Baker D: Automated prediction of CASP-5 structures using the Robetta server. Proteins 2003, 53 Suppl 6:524-533.  38.  Pfeffer SR: Transport Vesicle Docking: SNAREs and Associates. Annual Review of Cell and Developmental Biology 1996, 12(1):441-461.  39.  Folsch H, Pypaert M, Maday S, Pelletier L, Mellman I: The AP-1A and AP-1B clathrin adaptor complexes define biochemically and functionally distinct membrane domains. J Cell Biol 2003, 163(2):351-362.  40.  Hettema EH, Lewis MJ, Black MW, Pelham HR: Retromer and the sorting nexins Snx4/41/42 mediate distinct retrieval pathways from yeast endosomes. EMBO J 2003, 22(3):548-557.  41.  Huh WK, Falvo JV, Gerke LC, Carroll AS, Howson RW, Weissman JS, O'Shea EK: Global analysis of protein localization in budding yeast. Nature 2003, 425(6959):686-691.  42.  Panic B, Whyte JR, Munro S: The ARF-like GTPases Arl1p and Arl3p act in a pathway that interacts with vesicle-tethering factors at the Golgi apparatus. Curr Biol 2003, 13(5):405-410.  43.  Andag U, Neumann T, Schmitt HD: The coatomer-interacting protein Dsl1p is required for Golgi-to-endoplasmic reticulum retrieval in yeast. J Biol Chem 2001, 276(42):39150-39160.  44.  Lafourcade C, Galan JM, Peter M: Opposite roles of the F-box protein Rcy1p and the GTPase-activating protein Gyp2p during recycling of internalized proteins in yeast. Genetics 2003, 164(2):469-477.  45.  Liewen H, Meinhold-Heerlein I, Oliveira V, Schwarzenbacher R, Luo G, Wadle A, Jung M, Pfreundschuh M, Stenner-Liewen F: Characterization of the human GARP (Golgi associated retrograde protein) complex. Exp Cell Res 2005, 306(1):24-34.  46.  Maxfield FR, McGraw TE: Endocytic recycling. Nat Rev Mol Cell Biol 2004, 5(2):121132.  47.  Tong AHY, Evangelista M, Parsons AB, Xu H, Bader GD, Page N, Robinson M, Raghibizadeh S, Hogue CWV, Bussey H et al: Systematic Genetic Analysis with Ordered Arrays of Yeast Deletion Mutants. Science 2001, 294(5550):2364-2368.  48.  Longtine MS, McKenzie A, 3rd, Demarini DJ, Shah NG, Wach A, Brachat A, Philippsen P, Pringle JR: Additional modules for versatile and economical PCR-based gene deletion and modification in Saccharomyces cerevisiae. Yeast 1998, 14(10):953961. 68  49.  Sikorski RS, Hieter P: A system of shuttle vectors and yeast host strains designed for efficient manipulation of DNA in Saccharomyces cerevisiae. Genetics 1989, 122(1):19-27.  50.  Vernet T, Dignard D, Thomas DY: A family of yeast expression vectors containing the phage f1 intergenic region. Gene 1987, 52(2-3):225-233.  51.  Graham LA, Hill KJ, Stevens TH: Assembly of the yeast vacuolar H+-ATPase occurs in the endoplasmic reticulum and requires a Vma12p/Vma22p assembly complex. J Cell Biol 1998, 142(1):39-49.  52.  Conibear E, Stevens TH: Studying yeast vacuoles. Methods Enzymol 2002, 351:408432.  53.  Rieder SE, Banta LM, Kohrer K, McCaffery JM, Emr SD: Multilamellar endosome-like compartment accumulates in the yeast vps28 vacuolar protein sorting mutant. Mol Biol Cell 1996, 7(6):985-999.  54.  Haase SB, Lew DJ: Flow cytometric analysis of DNA content in budding yeast. Methods Enzymol 1997, 283:322-332.  55.  Notredame C, Higgins DG, Heringa J: T-coffee: a novel method for fast and accurate multiple sequence alignment. J Mol Biol 2000, 302(1):205-217.  56.  Zhang L, King O, Wong S, Goldberg D, Tong A, Lesage G, Andrews B, Bussey H, Boone C, Roth F: Motifs, themes and thematic maps of an integrated Saccharomyces cerevisiae interaction network. Journal of Biology 2005, 4(2):6.  69  Chapter 3 Genetic interaction analyses of trafficking dysfunction highlights relationships between endosomal transport pathways 3.1 Preface The specificity of vesicle transport is achieved by coordination of components of the trafficking machinery that recognize signals of membrane identity. As seen in Chapter 2, tethering complexes play an important role in conferring pathway specificity, and are involved in multiple stages of vesicle transport. Evidence from Chapter 2 suggests that GARP interacts with an unidentified early endosome identity determinant, suggesting additional factors involved in early endosomal sorting have yet to be identified. In the absence the Vps54 C-terminal domain, cargo, such as Snc1, may transit a late-endosome to Golgi retrograde transport pathway. These data illustrate the interconnected nature of trafficking pathways. In this chapter, a genetic screen that aimed to identify genes whose protein products are involved in endosomal transport is described. To evaluate the relationships between genes and pathways discovered, genetic interactions were assessed based on two phenotypes, growth and trafficking dysfunction. Although growth is commonly applied in genetic interaction analyses, an assessment of interactions based on trafficking dysfunction highlighted contextual interactions between endosomal coat proteins and their regulators. This work identified genes involved in endosomal transport, revealed new insights into the complex relationships between endosomal transport pathways and has implications for future analyses of genetic interaction.  A version of this chapter will be submitted for publication. Quenneville NR, Davey M, White R, Bryan J, Conibear E. Genetic interaction analyses of trafficking dysfunction highlights relationships between endosomal transport pathways.  70  3.2 Background Genetic interactions can reveal the interplay between biological pathways and the functional relationships among genes [1-3]. A genetic interaction occurs when the combination of two genetic mutations produces a phenotype that differs from what is expected [4]. In this way genetic interactions can be described as either aggravating or alleviating [5]. Aggravating interactions refer to the relationship between two genes if the double mutant phenotype is more severe than expected; an extreme case with respect to an analysis of fitness is synthetic lethality [6]. For non-essential genes, aggravating interactions tend to occur between parallel pathways, or within broadly defined functional categories [6-8], revealing inter-dependencies of biological processes. Conversely, alleviating interactions refer to the case when a double mutant phenotype is less severe than expected. This type of interaction, also known as diminishing returns, tends to occur between gene pairs with close functional associations [1-3, 9]. Within the broad classification of alleviating interactions lies a wealth of classical types or modes of genetic interaction such as Bateson masking epistasis and coequality [9]. Further mapping of alleviating genetic interactions into modes can reveal complex relationships between genes and pathways [9, 10]. Thus, measurement of both aggravating and alleviating interactions can group genes into pathways or functional modules and provide insight into their dependencies and relationships. Genetic interactions as measured in the yeast genome are rare, but can be enriched by choosing to study a set of functionally or logically defined genes [1-3, 9]. Genetic screens are an effective approach to identify genes involved in a common biological process and with the availability of genome-wide knock-out collections in yeast and advancements in RNA interference in other systems, the interrogation of genotype-phenotype relationships is very accessible. Since a genetic screen should reveal many of the genes involved in a given biological process, an analysis of genetic interactions among these genes should be highly informative of gene and pathway relationships [11]. Moreover, alleviating genetic interactions may be prevalent since genes discovered with screening are expected to share close functional associations. Because they can be broken down into classical forms of interaction, alleviating interactions can reflect relationships between pathways that may represent the underlying biochemical relationships between gene products [9, 10]. For example, categorization of alleviating interactions measured within a group of homologous recombination genes revealed a sequence of masking and suppressive interactions that reflect the known biochemical order in 71  which their gene products act [9]. Hence alleviating interactions are a particularly informative type of genetic interaction. Alleviating interactions may also be enriched by measuring a phenotype that is closely related to expected gene function. In a quantitative analysis of genetic interactions among a group of genes involved in DNA damage, more alleviating interactions were observed when fitness was measured in the presence of the DNA damage agent, MMS [9]. Although most genetic interaction analyses have focused on growth, any quantitative phenotype can be used to map genetic interactions [10]. In the same way that diverse quantitative phenotypes are used in screens to target specific biological processes, for example microscopy based-localization screens, or expression or activity of a reporter construct, or morphological changes, genetic interactions related to a functionally proximal non-growth phenotype may be particularly informative for the biological question or pathway targeted. Interpretation of genetic interactions resulting from such analyses may require unique considerations that have not been extensively explored. For example, in the context of intracellular trafficking, organelles are often connected by multiple pathways, and machinery and identity determinants can be shared, increasing the difficulty of interpreting a phenotype. The Golgi and endosomes are major sorting centers in the cell, coordinating secretion, degradation, and retrieval pathways for recycling cargoes. A combination of sorting signals, signal recognition by regulated trafficking machinery, and determinants of organelle identity, directs proteins to their appropriate destination. Endocytosed proteins are typically targeted for degradation in the vacuole, but proteins can also be selectively retrieved from either early or late endosomes through trafficking pathways that are collectively referred to as retrograde transport (reviewed in [12]). Retrograde transport can occur from two types of endosomes, early and late (Figure 3.1). Previous phenotypic screens of retrograde transport have focussed on identifying machinery required for retrieval from late endosomes back to the Golgi [13-15]. Retrograde transport from early endosomes is also important for recycling cargo, and is used by the secretory vesicle SNARE, Snc1 [16]. After fusion with the plasma membrane (secretion), Snc1 is endocytosed and transported to early endosomes. Upon retrieval from early endosomes to the Golgi, Snc1 is again incorporated into forming secretory vesicles (Figure 3.1a, green arrows). Mechanistic studies have identified several genes involved in retrograde transport of 72  Snc1 from early endosomes including putative coat proteins (Snx4/41/42, COPB), phospholipid translocases (Cdc50, Drs2), regulatory factors (Rcy1, Ypt31/32, Gcs1) and the Golgi associated retrograde protein (GARP) tethering complex [17-21]. While molecular studies have revealed many of the components required for retrograde transport, a comprehensive screen targeted to early endosomal transport has not been performed, raising the possibility that new factors have yet to be discovered. Moreover, the interplay between pathways involving the Golgi, endosomes and cell surface is not well understood. Genetic interactions hold potential for revealing higher order relationships among endosomal trafficking pathways as well as relationships among trafficking machineries. In this work, we enriched for genes involved in retrograde transport by performing a genetic screen for protein missorting. Targeted genetic interaction analyses that separately define genetic interactions based on growth, or trafficking dysfunction, identified alleviating genetic interaction partners of retrograde transport queries. Further mapping of alleviating genetic interactions revealed unanticipated relationships between retrograde transport machinery and regulators, reflecting the complex relationships between endosomal transport pathways. This work illustrates the power of targeting alleviating genetic interactions for informing relationships between genes and pathways, and highlights the value of expanding genetic interaction analyses to other non-growth phenotypes.  3.3 Results & Discussion 3.3.1 Discovery of retrograde transport factors To enable systematic identification of genes involved in retrograde transport from early endosomes, we designed a reporter of retrograde transport suitable for high-throughput analysis. The reporter was a chimeric fusion of GFP-Snc1 and the sucrose converting enzyme invertase, encoded by the gene SUC2 (GSS) [22]. Fusion of invertase to the C-terminus of Snc1 confered a quantitative characteristic to the GSS reporter, and enabled a measure of cell surface localization. Like Snc1-GFP, GSS was localized to the plasma membrane of growing buds and to internal punctate structures. Furthermore, cell surface localization required retrograde transport from endosomes to the Golgi (Figure 3.1). The extent of cell surface GSS was quantified using techniques that are similar to those used to measure colony growth on agar plates [23, 24]. Images of GSS assay plates were inverted and spot pixel intensity was 73  quantified with microarray spot finding software (Materials and Methods, [22]). Colonies with low levels of GSS activity are assigned high pixel intensity, and received a high GSS missorting score. This proof of principle suggests that whole colony detection of GSS activity acts as a quantitative phenotype to reflect changes in Golgi and endosomal trafficking pathways.  Figure 3.1. The GFP-SNC1-SUC2 GSS reporter is a model protein for quantitative phenotypic analysis of endosomal transport. A. Snc1 is cyclically transported between the Golgi, plasma membrane and endosomes (green arrows). GSS, a Snc1 chimera with GFP and the enzyme invertase at its N- and C-termini respectively, is predicted to follow a similar itinerary. Black arrows indicate other trafficking pathways between the Golgi and endosomes. B. Log-phase growing wild-type or vps51∆ cells expressing Snc1-GFP or GSS were viewed by fluorescence microscopy using FITC and DIC optics. C. Wild-type and vps51∆ strains expressing GSS are grown overnight on agar plates to form colonies, which are then over-layed with assay reagents. An assay for activity of surface localized GSS results in production of a brown colour (Materials 74  and Methods). Images of scanned plates are inverted and pixel intensity is quantified producing a GSS missorting score. To systematically discover retrograde transport factors, we crossed the GSS reporter into the MATa and MATα haploid yeast deletion collections using synthetic genetic array (SGA) techniques [6], and scored GSS localization of all non-essential yeast deletion mutants. While most mutants had relatively wild-type GSS scores, mutants that had either more or less GSS activity were discovered (Figure 3.2, [22]). Mutants with higher levels of GSS activity than wildtype were enriched for functions in endocytosis [22], while mutants with very low levels of GSS at the surface have known roles in the retrieval of Snc1 from early endosomes (Figure 3.2). Genes that share a common function tend to share common phenotypes [25]. To investigate the functions of genes with high GSS missorting scores, we clustered publicly available chemical genomic data for the top-most 384 genes (those above the threshold shown in Figure 3.2a). Cluster analysis, and an examination of the functions of the top-ranked GSS missorting mutants, points to an enrichment of several pathways and complexes including but not exclusive to those involved in trafficking (Figure 3.2). For example genes with GSS missorting phenotypes also have roles in mitochondrial function, chromatin remodelling, transcription and translation. While it is not clear that these pathways are directly related to transport, several mitochondrial poisons did affect the localization of Snc1 (B. Montpetit, personal communication). We also discovered protein complexes and pathways with direct roles in trafficking, such as the conserved oligomeric Golgi (COG) tethering complex, the Arf-like (ARL) regulatory pathway, and the endosomal phosphotidylinositol-3-kinase (PI3-kinase) regulators (Vps30 and Vps38) as well as putative coat and accessory proteins. Many of these genes have known roles in Snc1 sorting. For example Vps30 and Vps38 are required for the production of PI3P at endosomes, an important lipid determinant for endosome identity, and for the localization of phox-homology (PX) domain containing proteins such as the sorting nexin Snx4 [17, 26]. All known components of the ARL GTPase regulatory pathway, as well as the GTPase activating protein (GAP) Gcs1, were also discovered, further implicating this pathway in Snc1 sorting [21, 27]. New components involved in Snc1 sorting were also discovered. For instance, analysis of screen results prompted studies of a previously uncharacterized ORF, enabling discovery of a new protein associated with the TRAPPII Golgi tethering complex [28]. In assessing genes with known trafficking roles that were discovered in this screen, it is curious that we discovered multiple tethering complexes (GARP, COG, TRAPP), multiple coats (GGA, SNX, retromer), and GTPases involved in tether localization (Ypt6, ARL). While it is possible all 75  these components are involved in a common pathway, it is likely we have identified multiple endosomal and Golgi trafficking pathways. Taken together, a screen of the non-essential deletion collections discovered a set of genes that includes retrograde transport factors and components of several other pathways.  Figure 3.2. Non-essential retrograde transport factors are identified by screens of GSS localization. A. The average GSS missorting score for each gene in the MATa and MATα haploid yeast deletion collections are plotted in ascending order; the uppermost 1000 genes are shown. Dotted line illustrates the threshold above which genes were selected for further analysis (384 genes selected). B. Functions of the 50 top-ranking GSS missorting mutants are shown. Many have known trafficking functions or are known to be involved in Snc1 sorting. The 76  identity, function, score and rank of these genes are shown. Mitochondrial genes are also highly prevalent among the top-ranking genes. C. Hierarchical clustering the top-ranking 384 GSS missorting mutants (y-axis) with chemical-genomic data (x-axis, [25]) illustrates several pathways that are enriched in the top 384 genes: trafficking and retrograde transport, chromatin regulation (SWR1, and HIR complexes) and ribosome biogenesis for example. 3.3.2 Genetic interaction profiling highlights putative alleviating genetic interactions  Chemical genomic profiling clearly highlights pathways based on correlated phenotypes (Figure 3.2), but chemical perturbations often affect multiple pathways simultaneously, and because drug targets are often unknown, a deeper interpretation of pathway interaction is limited. By comparison, genetic disruption allows a more precise perturbation of pathways. EMAP analysis, the comprehensive analysis of genetic interactions involving all possible combinations of target genes, can provide an information rich dataset of both aggravating and alleviating interactions [2, 3]. However, recent work suggests the major sub-groups can be recovered using only a fraction of the data [29], suggesting an effective framework for undertaking comprehensive genetic interaction analysis is to focus interrogation, or query design, on the most influential genes. As we are interested in revealing subgroups from our primary screen that are most relevant to endosomal transport, we expect these could be revealed by genetic interaction analysis of a logically selected set of query genes involved in various elements of intracellular transport. To determine which genes from our primary screen genetically interact with genes involved in intracellular transport, we undertook a genetic interaction profiling analysis. Using SGA techniques, a mini-array comprised of 384 genes discovered in our primary screen was crossed with a spectrum of 50 query genes that represent numerous trafficking pathways, including retrograde transport. Resulting double mutant progeny were assayed to determine their GSS sorting phenotype. Because genetic interactions are rare, large deviation from a common double mutant phenotype may represent a propensity for genetic interaction. In an attempt to visualize double mutant phenotypes with respect to genetic interactions, data was expressed relative to the median phenotype of both queries and arrays and analyzed by hierarchical clustering each axis (Figure 3.3). Three significant and large array clusters were apparent from this analysis (correlation greater than 0.5; membership greater than 5). Furthermore, retrograde transport queries both clustered together on the x-axis, and had severe  77  phenotypic alterations when crossed with members of clusters 1 and 3 (Figure 3.3), suggesting this method highlights genetic interaction partners of retrograde transport queries.  Figure 3.3. Genetic interaction profiling reveals putative alleviating interactions with retrograde transport queries. Heatmap resulting from hierarchical clustering double mutant GSS phenotypes as measured after crossing the top 384 genes identified in the initial screen (y-axis) with a panel of 50 trafficking queries (x-axis). Three significant clusters that putatively interact with the retrograde transport queries ATG20, ENT3, SNX4, and GGA2 are shown (labelled 1, 2, 3). Gene names are coloured according to their rank from the primary GSS screen. The retrograde transport query cluster is comprised of some of the highest ranking genes. Genes belonging to clusters 1 and 3 are generally lower ranking, while many genes in cluster 2 are rank equally with retrograde transport queries. 78  The retrograde transport queries ATG20, ENT3, SNX4 and GGA2 exhibit the most severe single mutant phenotypes observed in our primary screen (all four are in the top 20). The apparent alteration of their phenotypes when crossed with members of clusters 1 and 3, many of which have less severe single mutant GSS missorting phenotypes, suggests cluster members may participate in a type of alleviating interaction with retrograde transport genes. That is, the phenotype of the double mutant is less severe than expected given the relative trafficking dysfunction observed for each single mutant. Similarly, members of cluster 2, which is comprised of several genes with severe single mutant GSS phenotypes, including the retrograde transport query genes themselves, appear to have extreme phenotypes when crossed with many different trafficking queries except the retrograde transport cluster. We hypothesized that these genes may also reflect a type of alleviating interaction with retrograde transport that is distinct from clusters 1 and 3. To test this hypothesis, we undertook a comprehensive and quantitative analysis of genetic interactions among 47 genes identified. 3.3.3 Quantitative analysis of genetic interactions recapitulates known genetic interactions with respect to growth and informs pathway dependencies  Quantitative genetic interaction refers to a significant deviation of a double mutant phenotype from its neutral expectation, computed using a model of neutrality and parental single mutant phenotypes [4]. To enable a more precise measure of genetic interaction, we applied estimates of single mutant phenotypes measured in replicate (n=36), and additive or multiplicative models of neutrality, to estimate a neutral expectation (Materials and Methods). The extent of interaction was calculated for each pair of genes (1081 pairs) by computing a tstatistic to compare the observed phenotype with its expectation. The t-statistic magnitude reflects the extent of deviation while the sign reflects the direction. With respect to growth, a negative t-statistic implies an aggravating interaction (double mutants are smaller than expected), while positive implies an alleviating interaction. The selection of genes for quantitative genetic interaction analysis was based on their propensity to putatively interact with retrograde transport genes in the context of their trafficking phenotypes. To evaluate whether these selected genes genetically interact in the context of fitness, we used colony size as a phenotypic measure of fitness and scored genetic interactions. At a threshold of two standard deviations from the mean growth t-statistic (-0.491), 57 genetic interactions are identified: 48 aggravating (t-statistic < -6.16), and 9 alleviating (t-statistic > 5.12), of the 1081 tested pairs. 79  Since aggravating interactions can reveal pathway inter-dependencies, or pathways that buffer one another [1], we reasoned that the meaning of these synthetic sick relationships could be best interpreted by considering relationships between known pathways. Tethering complexes and GTPases involved in tether localization were abundant in our initial screen, and many of these were also present in the growth genetic interaction data. Figure 3.4 illustrates significant (>2 s.d.) interactions involving first neighbours of these complexes and pathways (GARP, COG, ARL). The full network at 2 standard deviations can be viewed in Appendix B (Figure B.1). Aggravating interactions that were both known and reproduced here (dashed line), or first described here (full line) between the COG tethering complex, the Golgi SNARE Gos1, and members of the ARL GTPase pathway, are particularly striking due to their abundance (Figure 3.4). Aggravating interactions between the COG complex components COG5 and COG7, as well as with the Golgi SNARE, GOS1, and the ARL pathway genes ARL1, ARL3 and SYS1, have been described previously [2, 8]. Our findings expand on these known relationships to include the N-terminal acetyltransferase genes MAK3 and at a lesser significance, MAK10, which are required for the membrane localization of Arl3 [30, 31]. The ARL pathway is a regulatory GTPase cascade involved in membrane localization at the Golgi and possibly endosomes [30-34]. The COG complex also exhibits synthetic sick interactions with other endosomal trafficking components including coat proteins (sorting nexins) and the PI3-kinase subunits required for their localization. Deletion of the COG complex or Gos1 disrupts the morphology of the Golgi [35, 36], while disruption of the ARL pathway, coat components or PI3P alone does not affect Golgi structure but does impact Golgi/endosomal localization of specific sorting proteins, disrupting transport for numerous cargoes. Perhaps the reduced fitness observed when both Golgi and endosomal pathways are disrupted is due to the general defect in Golgi traffic or function combined with specific defects in Golgi and endosomal sorting. Synthetic sick interactions were also observed among sorting nexins involved in different retrograde transport pathways. SNX4, which is required for early endosome to Golgi transport, appears synthetic sick with PEP8, which encodes a component of the late-endosome to Golgi transport complex, retromer. SNX4 was previously shown to exhibit synthetic growth defects with a gene encoding another retromer component, VPS17 [17]. While trafficking pathways are cargo specific, cargoes have also been found to follow bypass pathways in the absence of sorting machinery [37, 38]. Perhaps the aggravating interaction between these classes of sorting complexes reflects a loss of alternative pathways to traffic cargo from endosomes to the 80  Golgi. Among the tethers, sorting nexins and ARL pathway genes shown in Figure 3.4, very few known synthetic sick or lethal interactions were missed. The exception to this is that known synthetic sick interactions between Ric1 and many genes involved in endosomal and Golgi transport (eg: COG complex, ARL pathway, Vps30/38) were not detected here. In summary, an analysis of genetic interactions based on growth recapitulated and expanded known relationships and provides insight into the general pathway inter-dependencies of Golgi and endosomal transport.  Figure 3.4. Genetic interaction analysis as defined by growth discovers known and novel interactions, highlighting pathway interdependencies. Aggravating interactions (red edges) between endosomal and Golgi trafficking pathways that were previously described and recapitulated here (dashed line), or previously unknown (solid line) are illustrated. Pathways are shown with coloured circles. Alleviating interactions are green. Interactions shown are two standard deviations from the mean growth t-statistic (-0.491). 3.3.4 Comparing genetic interactions as defined by growth versus trafficking dysfunction  To further define classes of genetic interaction for the set of genes defined by genetic interaction profiling (Figure 3.3), we quantified genetic interactions with respect to GSS using techniques analogous to those used for growth (Materials and Methods). Strains with strong 81  recycling defects have high GSS missorting scores. Therefore, a negative t-statistic reflects an alleviating interaction since a GSS sorting phenotype that is close to wild-type will have a smaller numeric value than that of a strongly defective strain. The extent of genetic interaction was computed for each of the 1081 double mutants and the distribution of t-statistics evaluated (Figure 3.5). Interestingly, the mean GSS t-statistic was less than zero (-4.85), suggesting that overall there is an enrichment of alleviating interactions based on GSS phenotype. This trend of negative t-scores occurs whether a multiplicative or additive model of neutrality is applied to the data (data not shown).  Figure 3.5. Alleviating GSS genetic interactions can be mapped into multiple modes of interaction. A. Scatter plot of growth t-statistics (x-axis) versus GSS t-statistics (y-axis) for each of the 1081 double mutants illustrates a slight positive correlation (0.457). The tendency for alleviating GSS dark double mutants to be aggravating based on growth helped to define 76 stringent alleviating GSS interactions where no growth defect was apparent B. A comparison of each double mutant GSS phenotype with its two parental single mutant phenotypes enables categorization of alleviating interactions into one of 5 categories: coequal, masking, partial masking, intermediate and suppression. One example of each interaction type from the dataset is shown with GSS mis-sorting phenotype on the y-axis. Blue and red bars represent single mutant phenotypes; green bars represent double mutant phenotypes. For visual reference, the black dashed line reflects the mean wild-type phenotype. Although the phenotypic context of a genetic interaction is expected to be a relevant defining characteristic, it is not clear whether genes that genetically interact under one phenotypic context will also interact in another. A scatter plot of growth versus GSS t-statistic for each of the 1081 gene pairs illustrates that while many gene pairs may participate in alleviating interactions based on GSS, most gene pairs do not interact based on growth (Figure 3.5; compare means). However, a slight positive correlation (0.457) is apparent, which reflects our 82  observation that very small colonies tended to exhibit a GSS phenotype similar to wild-type cells (not shown). To identify a set of high-confidence alleviating interactions with respect to trafficking dysfunction, we used the combined criteria of colony size threshold, negative GSS tstatistic and reproducibility (Materials and Methods) to define a set of 76 alleviating gene pairs out of a possible 1081. It is apparent that gene pairs that interact with respect to one phenotype do not always interact with respect to another, reinforcing the notion that phenotypic context is an important defining feature of genetic interactions. 3.3.5 Sub-classification of alleviating interactions with respect to trafficking dysfunction reveals novel masking relationships  Alleviating interactions can reflect multiple types of genetic interaction such as coequal and masking interactions [9]. Further mapping of genetic interactions into such classes increases the information gained from the network and helps to further reveal pathway relationships [9, 10]. Mapping can be accomplished by comparing and ordering the two parental single mutant phenotypes (Øx, Øy) with each other, and with the resulting double mutant phenotype (Øxy) (Figure 3.5b, Materials and Methods). Comparing these three phenotypes allowed us to map each of the 76 alleviating interactions into one of five modes: coequal, masking, partial masking, intermediate and suppressive. These genetic relationships can be expressed and analyzed as a network where nodes represent single gene deletions and their respective phenotypes, and edges reflect the type of alleviating interaction observed for that gene pair (Figure 3.6a). A network of these 76 edges involves 37 nodes. Interestingly, directed edges, including masking, partial masking and suppression, represented two thirds of the interaction categories observed (Figure 3.6). In a linear pathway, directed interactions can reflect the order of gene product function [39]. In the context of intracellular transport each vesicle transport event involves budding from one organelle and fusion with a specific downstream organelle, and in this simplified view, each pathway resembles a linear process. However, the mechanisms of vesicle transport and intracellular trafficking pathways are much more complex, and trafficking pathways themselves are cyclical [12, 40]. Furthermore, although pathways are specific, their interconnected nature can ultimately allow cargo to follow alternative routes, suggesting trafficking pathways themselves interact in complex ways. For these reasons, an analysis of directed interactions is expected to provide insight into the connections between Golgi and endosomal trafficking pathways.  83  Figure 3.6. Common masking, partial masking and suppressive genes are identified by network analysis. A. The mode of genetic interaction relating two genes can be depicted as a coloured edge (co-equal = green, masking = dark blue, partial masking = blue, intermediate = light blue, suppression = black). An arrow at the end of an edge refers to the direction or orientation of interaction, which is relevant for masking and suppressive interactions. The percent of each category that occurs for the 76 total alleviating GSS genetic interactions is shown in brackets. B. Network diagram of 76 GSS interactions involving 37 nodes. Single mutants are illustrated as nodes (circles) and their single mutant GSS phenotype is depicted with node colour (legend for GSS phenotype is shown with the colour bar). The magnitude of GSS t-statistic for each pair of genes is mapped to edge thickness. Considering directed edges only, the out-degree of each node was computed with Network Analyzer [41] and mapped to node size. C. A sub-network produced by extracting nodes with a degree greater than three focuses on interactions between the most interactive nodes. Interestingly, deletion of coats, GGA2, ENT3 and SNX4 are all masked or suppressed by the PI-3 kinase regulatory complex II, Vps30 and Vps38, and by the endosomal Na+,K+/H+ exchanger, Nhx1 (highlighted in yellow). D. Masking and suppressive 84  interactions observed for nhx1∆, vps30∆ and vps38∆ in high throughput reproduce in small scale. Single mutant GSS phenotypes are shown in the outermost columns (s), while each respective double mutant was plated at the point of intersection (d). Similar to the approach taken to interpret genetic interactions based on growth, we set out to identify network and pathway trends, which become apparent by focussing on common relationships and highly interactive nodes. Using Cytoscape and the Network Analyzer plugin to calculate the number of times a given node is a target or source (in-degree and out-degree respectively), we discovered highly connected masking, partial masking or suppressive nodes [41, 42]. A sub-network of nodes with a degree greater than 3 illustrates the most highly interactive genes (Figure 3.6c). From this network, one particularly striking set of relationships involves three of the core retrograde transport genes initially used as queries in our genetic interaction profiling step: SNX4, GGA2 and ENT3. These three genes, whose protein products are involved in sorting at endosomes and the Golgi, are commonly masked or suppressed by deletion of three regulators of endosomal transport NXH1, VPS30 and VPS38 (Figure 3.6d). Notably, VPS30, VPS38 and NHX1 are members of clusters 1 and 3 in the initial genetic interaction profiling analysis (Figure 3.3). The rediscovery of this group of alleviating relationships confirms our initial hypothesis that gene clusters from semi-quantitative genetic interaction profiling do indeed point to alleviating interaction partners of retrograde transport queries. Importantly the quantitative analysis accomplished here enables conclusive mapping of alleviating interactions into modes, in this case masking and suppression. It is interesting that the strongly defective GSS sorting strains snx4∆, gga2∆, and ent3∆ can be rescued by deletion of the regulatory gene products Vps30, Vps38 and Nhx1. To investigate these masking and suppressive relationships, we reconstructed double mutant strains and retested GSS phenotypes (Figure 3.6d). Single and double mutant strains were plated in blocks of 16 replicates and GSS phenotypes were analyzed. As shown in Figure 3.6d, the relationships observed in high-throughput reproduced on a small scale. It is difficult to determine whether the distinction between masking and suppressive modes of interaction in these 9 cases is biologically significant. Masking refers to the case where a double mutant phenotype is more similar to one parental single mutant than another, while suppression describes a phenotype that is unlike either single mutant and deviates toward the wild-type phenotype. This distinction is subtle when single and double mutant GSS phenotypes are visually compared (Figure 3.6d) and for this reason, it is prudent to consider these particular 85  masking and suppressive interactions as indistinguishable. Since these 9 double mutant phenotypes are all closer to the wild-type phenotype than their snx4∆, ent3∆ or gga2∆ parental phenotypes, the parental strains are suppressed. Therefore, from this quantitative analysis, we have identified a group of strong suppressive interactions that are intriguing with respect to regulation of retrograde transport. 3.3.6 Stimulation of recycling pathways may underlie suppressive interactions  These observations beg the question, what underlying biology reflects the genetic relationship between sorting proteins and their regulators? Gene deletions that are suppressed, (SNX4, ENT3, GGA2) encode vesicle coat proteins that are important for sorting at endosomes or the Golgi [17, 34, 38, 43-47]. In the absence of a vesicle coat protein, cargo such as GSS will be missorted, often to the vacuole. Gene deletions that suppress this defect encode regulators of endosomal transport, deletions of which exhibit defects in vacuolar delivery. For instance, Nhx1 is an endosome localized Na+,K+/H+ exchanger, important for maintaining pH gradients [48-50]. Deletion of NHX1 results in accumulation of cargo in an aberrant pre-vacuolar compartment which is described as ESCRT-like [51]. The ESCRT complex is required for forming and sorting cargo into the multivesicular body (MVB), a critical step for the degradation of many cargo proteins (reviewed in [52]). A defect in ESCRT machinery, and to a lesser extent nhx1∆, produces an aberrant endosomal compartment that accumulates cargo enroute to the vacuole [48, 51]. Given the phenotypes resulting from deletion of coat proteins and regulators, we hypothesized that double mutants will, like nhx1∆ and ESCRT mutants, have a defect in transporting cargo to the vacuole. Interestingly, ESCRT mutants also have a GSS missorting phenotype in which more GSS is present on the surface than wild-type cells [22]. Moreover ESCRT mutants are also hyper-efficient for recycling the uracil transporter Fur4 from endosomes to the cell surface [53]. We hypothesized that deletion of NHX1 may suppress the GSS recycling defect of vesicle coat proteins as a consequence of hyper-efficient recycling from a pre-vacuolar compartment to the cell surface. Deletion of VPS30 or VPS38 does not cause an obvious defect in MVB formation based on the wild-type localization of the protein GFP-CPS [26]. Moreover their vacuolar transport and morphology phenotype is distinct from ESCRT mutants [54]. However, due to their function in directing the synthesis of PI3P at endosomes, Vps30 and Vps38 are important for localization of the ESCRT complex, as well as endosomal coat proteins and other factors ([17, 26, 55], 86  reviewed in [56]). We hypothesized that deletion of VPS30 or VPS38 also suppresses mutations in vesicle coat proteins by enabling hyper-efficient recycling to the cell surface. To test these hypotheses, we examined the recycling kinetics of the uracil transporter Fur4 tagged with GFP (Fur4-GFP) in single and double mutants (Figure 3.7, and Appendix B Figure B2). Fur4 localization is sensitive to stress such as inhibition of protein synthesis, or nutrient deprivation, and is transported from the plasma membrane to the vacuole via the MVB. ESCRT mutants, which are defective in MVB formation, are capable of internalizing Fur4 but are hyper-efficient at recycling Fur4 back to the plasma membrane; this efficient recycling results in the rapid redistribution of Fur4 back to plasma membrane when severe stress, such as glucose deprivation, is relieved [53]. To examine the efficiency of recycling, strains carrying a GAL inducible form of Fur4-GFP were induced to express for 5 hours and were then transferred to glucose containing media to inhibit further expression. After twenty minutes (to allow stabilization) the localization of Fur4-GFP was imaged (t0). Wild-type, single and double mutant strains were all capable of localizing Fur4-GFP to the plasma membrane (Figure 3.7, Appendix B, t0). To stimulate internalization, strains were deprived of glucose for 100 minutes (carbon starvation, CS). Under these conditions, wild-type cells and single mutants of gga2∆, snx4∆ and ent3∆ predominantly localized Fur4-GFP to the vacuole lumen, whereas nhx1∆, vps30∆ and vps38∆ also accumulated punctate structures (Figure 3.7, Appendix B; CS). This is consistent with the previously described delay in transporting other MVB transiting cargo to the vacuole in nhx1∆ cells [48, 51]. Unlike single deletions of GGA2, SNX4 or ENT3, double mutants exhibited a highly punctate Fur4-GFP localization, suggesting transport to the vacuole is delayed in these strains. This apparent defect in transport to the vacuole is also consistent with our observation that double mutants secreted the vacuolar hydrolase carboxypeptidase Y (CPY) to a similar extent as their nhx1∆, vps30∆ or vps38∆ parent, while wild-type, snx4∆, gga2∆ or ent3∆ single deletion strains did not (data not shown).  87  Figure 3.7. Deletion of NHX1, VPS30 or VPS38 results in efficient recycling to the plasma membrane. Localization of the uracil permease Fur4-GFP was analyzed by fluorescence microscopy in wild-type BY4741, and nhx1∆, vps30∆, vps38∆, and gga2∆ single mutants, as well as nhx1∆gga2∆, vps30∆gga2∆ and vps38∆gga2∆ double mutants transformed with GALFur4-GFP (pFL38-GFP) after imposing and releasing a carbon stress (CS). Fur4-GFP expression was induced, and strains were transferred to minimal media containing glucose for 20 minutes to inhibit further expression and imaged by fluorescence microscopy (t0). To stimulate endocytosis of Fur4-GFP, cells were deprived of carbon for 100 minutes and imaged (CS). Note that wild-type and gga2∆ strains localize Fur4 to the vacuole while nhx1∆, vps30∆, vps38∆ and each of the double mutants also accumulate Fur4-GFP in punctate structures. After addition of carbon for 40 minutes (CA) cells were imaged. Fur4-GFP is efficiently localized back to the plasma membrane in all double mutants. Intensity levels were equally adjusted to visualize plasma membrane localization for all images. 88  After carbon starvation, cells were relieved of stress by growth in media containing glucose for 40 minutes (carbon addition (CA), Figure 3.7). While wild-type cells retained Fur4GFP in the vacuole, the ESCRT mutant vps37∆, and to varying degrees single mutants of vps30∆, vps38∆, nhx1∆, gga2∆ and snx4∆ were all capable of recycling Fur4 back to the plasma membrane. Strikingly, double mutants involving gga2∆ were all highly capable of transporting Fur4-GFP back to the plasma membrane (Figure 3.7, CA). Similarly, double mutants involving snx4∆ were also hyper-efficient at recycling Fur4 to the cell surface (Appendix B, Figure B.2). Double mutants involving ent3∆ were relatively less efficient at recycling; however, plasma membrane localization could be observed in ent3∆ double mutants but not single mutants. If mutants are hyper-efficient at recycling endogenous Fur4, more of the uracil transporter should be localized to the plasma membrane relative to wild-type cells. This difference can be observed by growing cells in the presence of the toxic uracil analogue, 5fluorouracil (5-FU). As a consequence of elevated plasma membrane localization, hyperefficient mutants will exhibit slow growth. Wild-type, single and double mutants were tested for sensitivity to 5-FU (Appendix B, Figure B.3). Wild-type, gga2∆, snx4∆, ent3∆, and nhx1∆ single mutants grew to a similar extent as wild-type, whereas vps30∆, vps38∆ and all double mutants were hypersensitive. The degree of sensitivity correlates with Fur4-GFP localization in that double mutants involving ent3∆ were less sensitive. While gga2∆ and snx4∆ single mutants were not hypersensitive to 5-FU, we did observe recycling and stabilization of Fur4-GFP by microscopy in these strains, suggesting deletion of GGA2 or SNX4 may also stimulate some degree of recycling. This discrepancy could reflect differences in carbon source as microscopy was performed with strains grown in the presence of galactose to induce Fur4-GFP expression but only in the presence of glucose for sensitivity to 5-FU. As sensitivity to 5-FU is most likely to reflect endogenous Fur4 localization it is likely that deletion of GGA2 or SNX4 does not appreciably affect Fur4 recycling. Taken together, these data suggest the suppressive interactions observed could be accounted for by hyper-efficient use of recycling pathways to the plasma membrane. Stimulation of recycling pathways as a consequence of defective transport through the MVB has also been described in mammalian cells, where loss of particular components of the ESCRT machinery leads to hyper-efficient recycling of epidermal growth factor receptor (EGFR) to the plasma 89  membrane [57]. Disrupting transport through endosomes by deletion of PEP12, an endosomal SNARE required for fusion with the late endosome, also causes efficient recycling to the surface, suggesting there are multiple compartments from which transporters such as Fur4 are capable of recycling [53]. Given the previously described roles of Snx4 and Gga2 in sorting from endosomes and the Golgi, our finding suggests that a general block in endosomal transport combined with delayed transport to the vacuole provides a synergistic opportunity for proteins to use recycling pathways to the cell surface. Clearly hyper-efficient recycling of receptors or transporters could have a detrimental effect on cellular homeostasis and could provide insight into mechanisms of disease.  3.4 Conclusion An evaluation of genetic interactions among targeted genes involved in endosomal and Golgi transport highlighted relationships and interdependencies of trafficking pathways. In an effort to enrich for alleviating interactions, we found that a scaled-down version of the E-MAP approach, using a combination of phenotypic screens and logically targeted queries, effectively identified alleviating interaction partners, revealing the most influential genes initially discovered from a targeted phenotypic screen. This strategy provides a framework to prioritize candidates discovered in a genetic screen by enriching for alleviating interactions. Genetic interaction analysis with a non-growth phenotype has been infrequently studied yet it is clear that genes interact in a context dependent manner; phenotype is an important element of that context. In this work, quantitative analysis of genetic interactions based on trafficking dysfunction enabled discovery and mapping of alleviating interactions into classical categories of interaction. Mapping uses the information rich nature of alleviating interactions, enabling discovery of masking and suppressive genetic interactions. Interrogation of the biological context of suppressive interactions between coat proteins and their putative regulators suggests enhanced use of recycling pathways to the cell surface. These observations reflect the complex interplay between the multiple pathways connecting endosomes, Golgi and the vacuole. In conclusion, performing quantitative genetic interaction analysis of a set of functionally related genes provides an unbiased method to uncover particularly influential genes due to an enrichment of alleviating genetic interactions.  90  3.5 Materials and Methods Strain Construction MATa and MATα yeast strains (BY4741 and BY4742) and their gene deletion derivatives were obtained from Open Biosystems (Huntsville, AL). Other strains used in this work are listed in Table 3.1. Query strains used for genomic screens of GSS were previously described [22]. Further gene disruption was accomplished by transforming a PCR product conferring NAT resistance (NATR) and 50-55bp homology to the 5’ and 3’ UTR of the open reading frame of interest. For each query strain constructed in this study, the natNT2 cassette was amplified from pFA6natNT2 and transformed into MDY519 ([58], Table 3.1). Deletions were confirmed by PCR and expression of GSS by fluorescence microscopy. NQY168, NQY169 and NQY170 are derivatives of BY4741 constructed as was described for query strains. Double mutant strains, NQY171 – NQY179, were constructed by transforming NQY180, NQY181 and NQY182 with PCR product to disrupt VPS30, VPS38 or NHX1 amplified from pFA6natNT2 [58]. For reassessment of GSS localization, NQY168 – NQY179 were transformed with Xba1 digested pCS30. Diploid strains, NQY183 - NQY185, were constructed by mating BY4741 expressing pKanMX4, ent3∆::KanMX4, or rvs167∆::KanMX4 with MDY519 expressing p NATR. Table 3.1 Yeast strains used in Chapter 3 Name BY4741 Y7043 MDY519 MDY47 MDY48 MDY46 MDY14 MDY34 MDY35 MDY13 MDY36 MDY43 MDY37 MDY44 MDY10 MDY39 MDY52 MDY33  Genotype MATa his3∆1 leu2∆0 met15∆0 ura3∆0 MATα his3∆1 leu2∆0 met15∆0 ura3∆0 can1::STE2pr-LEU2 lyp1 cyh2 Y7043 suc2::GFP-SNC1-SUC2::URA3 MDY519 apl4∆::NAT MDY519 apm1∆::NAT MDY519 apm2∆::NAT MDY519 atg20∆::NAT MDY519 ent1∆::NAT MDY519 ent2∆::NAT MDY519 ent3∆::NAT MDY519 ent4∆::NAT MDY519 inp52∆::NAT MDY519 prk1∆::NAT MDY519 sec28∆::NAT MDY519 snx4∆::NAT MDY519 vps5∆::NAT MDY519 vps13∆::NAT MDY519 yap1801∆::NAT  Source Open Biosystems C. Boone [22] [22] [22] [22] [22] [22] [22] [22] [22] [22] [22] [22] [22] [22] [22] [22] 91  Name MDY40 MDY16 MDY88 MDY87 MDY85 MDY129 MDY131 MDY83 MDY82 MDY127 MDY169 MDY271 MDY92 MDY270 MDY95 CSY57 MDY80 MDY267 MDY282 MDY171 MDY280 MDY133 MDY279 MDY277 MDY42 MDY265 MDY268 MDY45 MDY284 MDY286 MDY134 MDY281 MDY283 CSY56 NQY162 NQY149 NQY150 NQY161 MDY429 NQY133 NQY134 NQY164  Genotype MDY519 yap1802::NAT MDY519 ypt35∆::NAT MDY519 apl6::NAT MDY519 ent5∆::NAT MDY519 erg4∆::NAT MDY519 gga2∆::NAT MDY519 inp53∆::NAT MDY519 las21∆::NAT MDY519 lcb4∆::NAT MDY519 nhx1∆::NAT MDY519 snx41∆::NAT MDY519 vam3∆::NAT MDY519 vps4∆::NAT MDY519 vps9∆::NAT MDY519 vps54∆::NAT MDY519 vps68∆::NAT MDY519 ypt6∆::NAT MDY519 ypt31∆::NAT MDY519 age2∆::NAT MDY519 cka1::NAT MDY519 ckb1∆::NAT MDY519 fks1∆::NAT MDY519 gcs1∆::NAT MDY519 gef1∆::NAT MDY519 hul5∆::NAT MDY519 sak1∆::NAT MDY519 scy1∆::NAT MDY519 ubi4∆::NAT MDY519 vam7∆::NAT MDY519 vps17∆::NAT MDY519 vps21∆::NAT MDY519 vps51∆::NAT MDY519 vps8∆::NAT MDY519 vps55∆::NAT MDY519 aep1∆::NAT MDY519 alr1∆::NAT MDY519 arl3∆::NAT MDY519 bre5∆::NAT MDY519 cdc50∆::NAT MDY519 cog5∆::NAT MDY519 cog7∆::NAT MDY519 cox23∆::NAT  Source [22] [22] [22] [22] [22] [22] [22] [22] [22] [22] [22] [22] [22] [22] [22] [22] [22] This study [22] [22] [22] [22] [22] [22] [22] [22] [22] [22] [22] [22] [22] [22] [22] [22] This study This study This study This study This study This study This study This study 92  Name NQY145 NQY143 NQY148 NQY137 MDY383 NQY163 NQY138 NQY155 NQY151 NQY157 NQY156 NQY146 NQY154 NQY136 NQY164 NQY140 MDY475 NQY139 MDY278 NQY135 NQY153 NQY180 NQY181 NQY182 NQY168 NQY169 NQY170 NQY173 NQY176 NQY178 NQY172 NQY175 NQY177 NQY171 NQY174 NQY179  Genotype MDY519 cpr6∆::NAT MDY519 dcr2∆::NAT MDY519 gos1∆::NAT MDY519 pep8∆::NAT MDY519 ptc1∆::NAT MDY519 qri5∆::NAT MDY519 ric1∆::NAT MDY519 rps21b∆::NAT MDY519 rtg1∆::NAT MDY519 rtg2∆::NAT MDY519 rtg3∆::NAT MDY519 sec66∆::NAT MDY519 sro7∆::NAT MDY519 sso2∆::NAT MDY519 swf1∆::NAT MDY519 sys1∆::NAT MDY519 trs33∆::NAT MDY519 vps30∆::NAT MDY519 vps38∆::NAT MDY519 ygr016w∆::NAT MDY519 yor305w∆::NAT BY4741 snx4∆::KanMX4 BY4741 ent3∆::KanMX4 BY4741 gga2∆::KanMX4 BY4741 nhx1∆::natNT2 BY4741 vps30∆::natNT2 BY4741 vps38∆::natNT2 NQY180 vps30∆::natNT2 NQY180 vps38∆::natNT2 NQY180 nhx1∆::natNT2 NQY181 vps30∆::natNT2 NQY181 vps38∆::natNT2 NQY181 nhx1∆::natNT2 NQY182 vps30∆::natNT2 NQY182 vps38∆::natNT2 NQY182 nhx1∆::natNT2  Source This study This study This study This study [22] This study This study This study This study This study This study This study This study This study This study This study This study This study [22] This study This study Open Biosystems Open Biosystems Open Biosystems This study This study This study This study This study This study This study This study This study This study This study This study  93  SGA techniques Synthetic genetic array techniques were performed as described previously [22]. Briefly, lawns of query strains were grown overnight at 30˚C on appropriate selective media. Query strains were mated with tester strains (arrays of the KanMX4 deletion collection, Open Biosystems) by sequentially pinning query and tester colonies to the same YPD array in 384 format using a Virtek automated colony arrayer (Biorad, Hercules, CA). Mating was allowed to occur for 2-3 days at 30˚C, and arrays were then replicated to YPD + G418 +clonNAT plates and grown for another 2-3 days at 30˚C to select for diploid strains. Arrays were then replicated to YPD for a pre-sporulation step, grown overnight at 30˚C and replicated to sporulation media. At the transition from pre-sporulation to sporulation media, colonies were duplicated from 384 to 768-format. In this way, two sporulation events were allowed to occur for each double mutant cross. Strains were allowed to sporulate for 9-11 days and MATa or MATα haploid mutants were selected by replica-pinning colonies to appropriate selective media. MATa double mutant cells expressing GSS were selected on SC –leu, -ura +G418 +clonNAT plates. After 2-days growth at 30˚C, double mutants were replicated to two sets of YP-fructose (2% fructose, YPF). For each set, 384 colonies from the MATa double mutant source plate were replicated in quadruplicate to create a 1536-colony format. In this way, the two independent double mutant crosses for each strain were tested on two different assay plates. For assessment of double mutant growth phenotypes, haploid MATa arrays were replicated to both MATa haploid G418 selection, and MATa haploid double mutant selection (G418 + NAT). Strains were grown for 3days at 30˚C and images were scanned for analysis of colony size.  GSS/invertase assay Invertase assays were performed as described previously [22]. Briefly, colonies were pinned onto YPF and grown for one overnight (16-18h) at 30˚C. Plates were scanned to image colony size on YPF, then overlayed with 20mL of a warm solution of assay reagents (2.75% ultra-pure sucrose, 64mM NaOAc pH 5.5, 0.25mM N-ethyl-maleimide, 400 U horseradish peroxidase, 320 U glucose oxidase, 1.57mM o-dianisidine, 0.45% hot agar) prepared as described [59]. Plates were allowed to react for 35 minutes before scanning in colour on a flatbed scanner at a resolution of 150dpi. Images were converted to greyscale in Adobe Photoshop (San Jose, CA), and processed with National Institute of Health ImageJ software (NIH, http://rsb.info.nih.gov/ij/index.html) as follows: local normalization at sigma 50 (plugin) followed by a despeckle filter, and contrast enhancement (1% pixel saturation). Processed images were quantified using spot-finding microarray software (GridGrinder, gridgrinder.sourceforge.net) and 94  the spot median of each colony was extracted. In all high-throughput experiments, each respective mutant was plated in 4 positions. In this way 384 unique positions were replicated four times to form a density of 1536 colonies per assay plate. With the exception of edge and corner colonies, for which 2-3 replicates were discarded due to high noise at these positions, the median of four replicates were used in all downstream processing and analyses.  Data Pre-processing Normalization Results from screening the MATa and MATα haploid yeast deletion collections were normalized by expressing the pixel intensity assigned to each mutant GSS missorting score relative to its respective plate median. Since most mutants in each collection lack a GSS phenotype, the plate median did not tend to vary from plate to plate making this a suitable approach. To enable a quantitative measure of double mutant phenotypes across assay plates, 19 external controls were included on the parental array of MATa haploid yeast cells. Diploid yeast cells representative of a phenotype that is darker than wild-type (rvs167∆), lighter than wild-type (ent3∆) and wild-type cells were spotted onto the haploid array prior to the construction of double mutants by SGA. Since diploid cells will not mate with the query strain, the final SGA selection produced an array of MATa double mutants as well as 19 single mutants which acted as fixed points for normalization. Normalization techniques will be described elsewhere (R. White, manuscript in preparation). In preparation for computation of t-statistics, growth data was log-transformed.  Computation of t-statistics for GSS and Growth phenotypes Single mutant phenotypes ( deletion effect,  ,  ), were expressed as the wild-type phenotype plus the  or :  A linear model for the double mutant phenotype (  ), was constructed based on the double  mutant phenotype predicted by a model of neutrality (  ,  ) plus the interaction effect (  ).  Under an additive neutrality model: 95  , If there is a genetic interaction, the observed double mutant phenotype differs from Øxy,neut by the interaction effect (  :  or  Solving for the interaction effect (  :  [eq 1] Mean and standard error was computed from all wild-type (  ), single ( , ) and double (  )  mutant observations (growth n = 866, n=28 and n=4; GSS n = 970, n = 36, n = 4). Mean and standard error values were substituted into equation 1, to estimate the interaction effect and its error: –  A standard t-test was then applied to test the null hypothesis (  0) resulting in a t-statistic  for each combination of x and y. To avoid artefacts caused by combinations of x and y with unusually small error, a minimum variance (minVar) was applied to each dataset that is equal to the minimum variance observed for a single mutant invertase (minVar = 414 447.9), growth (minVar = 0.0003473057). For invertase data, we also applied a maximum expectation that reflects the average value of  in the upper 2-5% of the data (xyMax = 16 750).  Defining pairs of interacting genes Alleviating interactions in the GSS genetic interaction dataset were defined as those gene pairs for which the GSS t-statistic is less than the mean (-4.8476), and consistent growth on YPF was observed (threshold = 14 500). A consistent gene pair is one which passes the size  96  criterion in two independent reciprocal crosses: geneX∆::kanMX4 geneY∆::natNT2, geneY∆::kanMX4 geneX∆::natNT2.  Defining modes of genetic interaction The mode of genetic interaction was defined with inequalities that order single and double mutant phenotypes. To produce an inequality, two t-tests were computed for each double mutant, comparing its phenotype  with that of its two parental single mutants  and  (T1 and T2 respectively). Magnitudes greater than 3.5 were deemed significant. The resulting inequality was used to categorize gene pairs of x and y into one of five defined modes of genetic interaction: co-equal, masking, partial masking, intermediate and suppression. A sixth mode to account for any inequality that did not fit into the following five categories was also permitted, however we did not observe this case for the 76 alleviating interactions categorized. Modes were defined as follows. Coequal x and y were coequal if  , and  or  Masking Masking relationships were defined if either  and the double mutant phenotype  is equal to  or .  x masks y if  ,  y masks x if  ,  Partial Masking Partial masking accounted for cases where the double mutant phenotype was more similar to one single mutant than the other, but was not statistically equal to either single mutant phenotype. Similarity is determined by considering the ratio of t-statistics (T1 and T2), if the magnitude of both T1 and T2 were significant (>3.5). If T1/T2 > 1.5, x partial masks y; if T2/T1 > 1.5, y partial masks x. Intermediate A double mutant was no closer to one single mutant than the other, therefore there is no edge direction. |T1|/|T2| =<1.5 97  Suppression Suppression describes the case wherein a double mutant had an invertase value that was statistically smaller than either single mutant. This mode is named suppression since these phenotypes were always closer to the wild-type phenotype than either single mutant by definition. We do not however include the wild type phenotype in any of the t-statistic comparisons. Direction was inferred by qualitatively considering which single mutant phenotype was closer to wild-type.  Fur4 recycling and sensitivity to 5-FU Protocols to test Fur4-GFP recycling were performed as described previously with minor modification [53]. Strains, transformed with a plasmid to enable expression of pGAL-Fur4-GFP (pFL38-GFP [53], were grown overnight in selective media containing 2% dextrose. Strains were diluted in selective media containing 2% galactose to induce expression for 5 hours, and subsequently transferred to glucose containing media for 20 minutes at 25°C, to inhibit further expression and facilitate stabilization of Fur4-GFP at the cell surface, at which point cells were imaged (t=0). To deplete cells of carbon, strains were harvested, washed in dH2O, resuspended in selective media lacking a carbon source and allowed to incubate for 100 minutes at 25°C. Finally strains were released by growth in selective media containing 2% glucose for 40 minutes. At each time point cells were viewed using a 100x oil immersion lens and FITC optics with a Zeiss Axioplan2 fluorescence microscope (Thornwood, NY). Images were captured with a CoolSnap camera, exposed and scaled equally using Metamorph software (Universal Imaging, West Chester, PA). Sensitivity to 5-fluorouracil (5-FU) was assessed by spotting equal serial dilutions of logphase cells onto selective media lacking uracil plus or minus 5µM 5-FU. Plates were incubated at 30°C for 2-4 days.  98  3.6 References 1.  Boone C, Bussey H, Andrews BJ: Exploring genetic interactions and networks with yeast. Nat Rev Genet 2007, 8(6):437-449.  2.  Schuldiner M, Collins SR, Thompson NJ, Denic V, Bhamidipati A, Punna T, Ihmels J, Andrews B, Boone C, Greenblatt JF et al: Exploration of the Function and Organization of the Yeast Early Secretory Pathway through an Epistatic Miniarray Profile. Cell 2005, 123(3):507-519.  3.  Collins SR, Miller KM, Maas NL, Roguev A, Fillingham J, Chu CS, Schuldiner M, Gebbia M, Recht J, Shales M et al: Functional dissection of protein complexes involved in yeast chromosome biology using a genetic interaction map. Nature 2007, 446(7137):806-810.  4.  Mani R, St Onge RP, Hartman JLt, Giaever G, Roth FP: Defining genetic interaction. Proc Natl Acad Sci U S A 2008, 105(9):3461-3466.  5.  Segre D, DeLuna A, Church GM, Kishony R: Modular epistasis in yeast metabolism. Nat Genet 2005, 37(1):77-83.  6.  Tong AHY, Evangelista M, Parsons AB, Xu H, Bader GD, Page N, Robinson M, Raghibizadeh S, Hogue CWV, Bussey H et al: Systematic Genetic Analysis with Ordered Arrays of Yeast Deletion Mutants. Science 2001, 294(5550):2364-2368.  7.  Kelley R, Ideker T: Systematic interpretation of genetic interactions using protein networks. Nature Biotechnol 2005, 23:561-566.  8.  Tong AHY, Lesage G, Bader GD, Ding H, Xu H, Xin X, Young J, Berriz GF, Brost RL, Chang M et al: Global Mapping of the Yeast Genetic Interaction Network. Science 2004, 303(5659):808-813.  9.  St Onge RP, Mani R, Oh J, Proctor M, Fung E, Davis RW, Nislow C, Roth FP, Giaever G: Systematic pathway analysis using high-resolution fitness profiling of combinatorial gene deletions. Nat Genet 2007, 39(2):199-206.  10.  Drees B, Thorsson V, Carter G, Rives A, Raymond M, Avila-Campillo I, Shannon P, Galitski T: Derivation of genetic interaction networks from quantitative phenotype data. Genome Biol 2005, 6(4):R38.  11.  Conibear E: An E-MAP of the ER. Cell 2005, 123(3):366-368.  12.  Bonifacino JS, Rojas R: Retrograde transport from endosomes to the trans-Golgi network. Nat Rev Mol Cell Biol 2006, 7(8):568-579.  99  13.  Schluter C, Lam KKY, Brumm J, Wu BW, Saunders M, Stevens TH, Bryan J, Conibear E: Global Analysis of Yeast Endosomal Transport Identifies the Vps55/68 Sorting Complex. Mol Biol Cell 2008, 19(4):1282-1294.  14.  Bowers K, Stevens TH: Protein transport from the late Golgi to the vacuole in the yeast Saccharomyces cerevisiae. Biochim Biophys Acta Mol Cell Res 2005, 1744(3):438-454.  15.  Bonangelino CJ, Chavez EM, Bonifacino JS: Genomic Screen for Vacuolar Protein Sorting Genes in Saccharomyces cerevisiae. Mol Biol Cell 2002, 13(7):2486-2501.  16.  Lewis MJ, Nichols BJ, Prescianotto-Baschong C, Riezman H, Pelham HRB: Specific Retrieval of the Exocytic SNARE Snc1p from Early Yeast Endosomes. Mol Biol Cell 2000, 11(1):23-38.  17.  Hettema EH, Lewis MJ, Black MW, Pelham HR: Retromer and the sorting nexins Snx4/41/42 mediate distinct retrieval pathways from yeast endosomes. EMBO J 2003, 22(3):548-557.  18.  Furuta N, Fujimura-Kamada K, Saito K, Yamamoto T, Tanaka K: Endocytic Recycling in Yeast Is Regulated by Putative Phospholipid Translocases and the Ypt31p/32pRcy1p Pathway. Mol Biol Cell 2007, 18(1):295-312.  19.  Sakane H, Yamamoto T, Tanaka K: The functional relationship between the Cdc50pDrs2p putative aminophospholipid translocase and the Arf GAP Gcs1p in vesicle formation in the retrieval pathway from yeast early endosomes to the TGN. Cell Struct Funct 2006, 31(2):87-108.  20.  Conibear E, Cleck JN, Stevens TH: Vps51p Mediates the Association of the GARP (Vps52/53/54) Complex with the Late Golgi t-SNARE Tlg1p. Mol Biol Cell 2003, 14(4):1610-1623.  21.  Robinson M, Poon PP, Schindler C, Murray LE, Kama R, Gabriely G, Singer RA, Spang A, Johnston GC, Gerst JE: The Gcs1 Arf-GAP Mediates Snc1,2 v-SNARE Retrieval to the Golgi in Yeast. Mol Biol Cell 2006, 17(4):1845-1858.  22.  Burston H, Maldonado-Baez L, Montpetit B, Davey M, Schulter C, Wendland B, Conibear E: Regulators of Yeast Endocytosis Identified by Systematic Quantitative Analysis. J Cell Biol 2009, In press.  23.  Dudley AM, Janse DM, Tanay A, Shamir R, Church GM, *equal: A global view of pleiotropy and phenotypically derived gene function in yeast. Mol Syst Biol 2005, 1.  24.  Collins S, Schuldiner M, Krogan N, Weissman J: A strategy for extracting and analyzing large-scale quantitative epistatic interaction data. Genome Biol 2006, 7(7):R63. 100  25.  Hillenmeyer ME, Fung E, Wildenhain J, Pierce SE, Hoon S, Lee W, Proctor M, St Onge RP, Tyers M, Koller D et al: The chemical genomic portrait of yeast: uncovering a phenotype for all genes. Science 2008, 320(5874):362-365.  26.  Burda P, Padilla SM, Sarkar S, Emr SD: Retromer function in endosome-to-Golgi retrograde transport is regulated by the yeast Vps34 PtdIns 3-kinase. J Cell Sci 2002, 115(20):3889-3900.  27.  Liu YW, Lee SW, Lee FJ: Arl1p is involved in transport of the GPI-anchored protein Gas1p from the late Golgi to the plasma membrane. J Cell Sci 2006, 119(Pt 18):3845-3855.  28.  Montpetit B, Conibear E: Identification of the novel TRAPP associated protein Tca17. Traffic 2008, In Press.  29.  Casey FP, Cagney G, Krogan NJ, Shields DC: Optimal stepwise experimental design for pairwise functional interaction studies. Bioinformatics 2008:btn472.  30.  Setty SR, Strochlic TI, Tong AH, Boone C, Burd CG: Golgi targeting of ARF-like GTPase Arl3p requires its Nalpha-acetylation and the integral membrane protein Sys1p. Nat Cell Biol 2004, 6(5):414-419.  31.  Behnia R, Panic B, Whyte JRC, Munro S: Targeting of the Arf-like GTPase Arl3p to the Golgi requires N-terminal acetylation and the membrane protein Sys1p. Nat Cell Biol 2004, 6(5):405-413.  32.  Setty SR, Shin ME, Yoshino A, Marks MS, Burd CG: Golgi recruitment of GRIP domain proteins by Arf-like GTPase 1 is regulated by Arf-like GTPase 3. Curr Biol 2003, 13(5):401-404.  33.  Panic B, Whyte JR, Munro S: The ARF-like GTPases Arl1p and Arl3p act in a pathway that interacts with vesicle-tethering factors at the Golgi apparatus. Curr Biol 2003, 13(5):405-410.  34.  Singer-Kruger B, Lasic M, Burger AM, Hausser A, Pipkorn R, Wang Y: Yeast and human Ysl2p/hMon2 interact with Gga adaptors and mediate their subcellular distribution. EMBO J 2008, 27(10):1423-1435.  35.  Wuestehube LJ, Duden R, Eun A, Hamamoto S, Korn P, Ram R, Schekman R: New mutants of Saccharomyces cerevisiae affected in the transport of proteins from the endoplasmic reticulum to the Golgi complex. Genetics 1996, 142(2):393-406.  36.  McNew JA, Coe JGS, Søgaard M, Zemelman BV, Wimmer C, Hong W, Söllner TH: Gos1p, a Saccharomyces cerevisiae SNARE protein involved in Golgi transport. FEBS Letters 1998, 435(1):89-95.  101  37.  Harsay E, Schekman R: A subset of yeast vacuolar protein sorting mutants is blocked in one branch of the exocytic pathway. J Cell Biol 2002, 156(2):271-286.  38.  Black MW, Pelham HRB: A Selective Transport Route from Golgi to Late Endosomes that Requires the Yeast GGA Proteins. J Cell Biol 2000, 151(3):587-600.  39.  Avery L, Wasserman S: Ordering gene function: the interpretation of epistasis in regulatory hierarchies. Trends Genet 1992, 8(9):312-316.  40.  Maxfield FR, McGraw TE: Endocytic recycling. Nat Rev Mol Cell Biol 2004, 5(2):121132.  41.  Assenov Y, Ramirez F, Schelhorn S-E, Lengauer T, Albrecht M: Computing topological parameters of biological networks. Bioinformatics 2008, 24(2):282-284.  42.  Shannon P, Markiel A, Ozier O, Baliga NS, Wang JT, Ramage D, Amin N, Schwikowski B, Ideker T: Cytoscape: a software environment for integrated models of biomolecular interaction networks. Genome Res 2003, 13(11):2498-2504.  43.  Duncan MC, Costaguta G, Payne GS: Yeast epsin-related proteins required for Golgi-endosome traffic define a [gamma]-adaptin ear-binding motif. Nat Cell Biol 2003, 5(1):77-81.  44.  Copic A, Starr TL, Schekman R: Ent3p and Ent5p Exhibit Cargo-specific Functions in Trafficking Proteins between the Trans-Golgi Network and the Endosomes in Yeast. Mol Biol Cell 2007, 18(5):1803-1815.  45.  Costaguta G, Duncan MC, Fernandez GE, Huang GH, Payne GS: Distinct roles for TGN/endosome epsin-like adaptors Ent3p and Ent5p. Mol Biol Cell 2006, 17(9):3907-3920.  46.  Chidambaram S, Zimmermann J, von Mollard GF: ENTH domain proteins are cargo adaptors for multiple SNARE proteins at the TGN endosome. J Cell Sci 2008, 121(3):329-338.  47.  Eugster A, Pecheur E-I, Michel F, Winsor B, Letourneur F, Friant S: Ent5p Is Required with Ent3p and Vps27p for Ubiquitin-dependent Protein Sorting into the Multivesicular Body. Mol Biol Cell 2004, 15(7):3031-3041.  48.  Brett CL, Tukaye DN, Mukherjee S, Rao R: The yeast endosomal Na+K+/H+ exchanger Nhx1 regulates cellular pH to control vesicle trafficking. Mol Biol Cell 2005, 16(3):1396-1405.  49.  Nass R, Cunningham KW, Rao R: Intracellular sequestration of sodium by a novel Na+/H+ exchanger in yeast is enhanced by mutations in the plasma membrane H+102  ATPase. Insights into mechanisms of sodium tolerance. J Biol Chem 1997, 272(42):26145-26152. 50.  Nass R, Rao R: Novel localization of a Na+/H+ exchanger in a late endosomal compartment of yeast. Implications for vacuole biogenesis. J Biol Chem 1998, 273(33):21054-21060.  51.  Bowers K, Levi BP, Patel FI, Stevens TH: The sodium/proton exchanger Nhx1p is required for endosomal protein trafficking in the yeast Saccharomyces cerevisiae. Mol Biol Cell 2000, 11(12):4277-4294.  52.  Hurley JH: ESCRT complexes and the biogenesis of multivesicular bodies. Curr Opin Cell Biol 2008, 20(1):4-11.  53.  Bugnicourt A, Froissard M, Sereti K, Ulrich HD, Haguenauer-Tsapis R, Galan J-M: Antagonistic Roles of ESCRT and Vps Class C/HOPS Complexes in the Recycling of Yeast Membrane Proteins. Mol Biol Cell 2004, 15(9):4203-4214.  54.  Raymond CK, Howald-Stevenson I, Vater CA, Stevens TH: Morphological classification of the yeast vacuolar protein sorting mutants: evidence for a prevacuolar compartment in class E vps mutants. Mol Biol Cell 1992, 3(12):13891402.  55.  Kihara A, Noda T, Ishihara N, Ohsumi Y: Two Distinct Vps34 Phosphatidylinositol 3Kinase Complexes Function in Autophagy and Carboxypeptidase Y Sorting in Saccharomyces cerevisiae. J Cell Biol 2001, 152(3):519-530.  56.  Katzmann DJ, Odorizzi G, Emr SD: Receptor downregulation and multivesicularbody sorting. Nat Rev Mol Cell Biol 2002, 3(12):893-905.  57.  Raiborg C, Malerød L, Pedersen NM, Stenmark H: Differential functions of Hrs and ESCRT proteins in endocytic membrane trafficking. Exp Cell Res 2008, 314(4):801813.  58.  Janke C, Magiera MM, Rathfelder N, Taxis C, Reber S, Maekawa H, Moreno-Borchart A, Doenges G, Schwob E, Schiebel E et al: A versatile toolbox for PCR-based tagging of yeast genes: new fluorescent proteins, more markers and promoter substitution cassettes. Yeast 2004, 21(11):947-962.  59.  Darsow T, Odorizzi G, Emr SD: Invertase fusion proteins for analysis of protein trafficking in yeast. Methods Enzymol 2000, 327:95-106.  103  Chapter 4: Discussion and Conclusions 4.1 Summary of Findings Many factors contribute to the specificity of trafficking machinery at both upstream and downstream compartments, including membrane recognition, cargo signals and posttranslational modifications. Our understanding of the components and mechanisms that confer the activity and specificity of each trafficking pathway has been advanced by the abundance of studies accomplished in model and higher organisms that use diverse and complementary techniques. In this work I have used a combination of molecular and genetic approaches to study components and pathways for retrograde transport to the Golgi. Chapter 2 explored questions relating to tethering complex specificity and membrane recognition by evaluating how the GARP tethering complex is able to tether vesicles derived from multiple endosomal compartments. The contribution of Vps54 to GARP complex function was investigated by studying N- and C-terminal truncation mutants. While the N-terminal half was found to be important for GARP complex assembly and stability, the C-terminal portion was localized to an endosomal compartment. This localization depended on conserved residues in Vps54-C, and suggested Vps54-C is involved in an upstream membrane recognition event. A function for the C-terminal region in sorting the v-SNARE Snc1 was only revealed when late endosome to Golgi transport, a retrieval pathway that Snc1 normally does not require, was simultaneously blocked. This observation suggests that Snc1, like other cargo, can follow alternative or bypass pathways illustrating the connectivity of retrograde transport from early and late endosomes. Chapter 3 describes an approach to study retrograde transport that involves a combination of genetic screens to discover influential genes involved in retrograde transport, and genetic interaction analyses to understand how pathways are connected. This method enriched for information-rich alleviating genetic interactions as measured in the context of their trafficking phenotype. These alleviating genetic interactions could be mapped into classical types of genetic interaction such as masking or suppressive interactions. Mapping allowed identification of unexpected suppressive interactions between coat proteins and regulators of endosomal transport. Molecular evaluation of these genetic relationships suggested that loss of transport from endosomes to both the Golgi and vacuole enhanced use of alternative recycling 104  pathways to the cell surface. This analysis illustrates the interconnected nature of endosomal transport pathways and demonstrates the value of contextual genetic interaction analyses. A role for retrograde transport in neurodegenerative disease has been suggested based on multiple findings in model organisms and patients. A deeper understanding of the machinery required for retrograde transport, their mechanisms of action, and their interactions with other pathways is required. Work in model organisms can provide valuable insights that can be applied to higher systems.  4.2 Implications and Significance 4.2.1 Recruitment of tethers by vesicle coats - How is GARP recruited to membranes? In Chapter 2, the model that GARP associates with endosomal transport vesicles before or while they are forming is proposed (Figure 2.10). This model is supported by two types of data. Overexpression of Vps54-C stimulated accumulation of tubular-vesicular structures that colocalize with proteins involved in early endosomal sorting (Figures 2.5; 2.6). Secondly, sorting of Snc1, an early-endosome-to-Golgi cargo, relies on late-endosome-to-Golgi retrograde transport when the C-terminal domain of Vps54 is removed, suggesting that Vps54-N is defective in early-endosome-to-Golgi retrograde transport (Figure 2.2; Appendix A). Taken together, these data suggest that GARP is capable of interacting with a factor that is enriched at early endosomes, and that this recognition event is important for transport of cargo from early endosomes to the Golgi. Coat proteins are increasingly found to interact with, or to recruit, tethering complexes. Recognition of coat proteins at upstream compartments has been observed for tethering complexes, such as EXOCYST, COG and HOPS as discussed in Chapter 2. Recently, the early Golgi tethering complex TRAPPI was found to interact with a component of the COPII coat in both yeast and mammalian cells [1]. The entire COPII coat, as well as the specific interaction between the TRAPPI component Bet3, and the COPII component Sec23, is required for vesicle tethering and fusion in vitro. Activation of the Golgi Rab protein, Ypt1, is also required to tether vesicles with the early Golgi. The TRAPPI complex was previously shown to activate the Golgi Rab protein Ypt1, which is also required for early Golgi tethering, but the subunit that enables GEF activity was not known [2]. Recent work shows that nearly all subunits of TRAPPI are required for GEF activity and hence Ypt1 activation [1], leading to the model that COPII coated 105  vesicles recruit TRAPPI which then activates Ypt1 on the early Golgi, thereby ensuring COPII vesicles are tethered to the correct compartment [1, 3, 4]. Activated Ypt1 also interacts with another early Golgi tether, Uso1 (p115 in mammals). Uso1, a long coiled-coil tether, is proposed to interact with and stabilize uncoated COPII vesicles after they are first docked by TRAPPI [1, 3, 4]. Uso1 may also dock vesicles independently of TRAPPI since acidic residues in the terminal head portion of the mammalian Uso1 homolog, p115, were shown to interact with the adaptor-like subunit of the COPI coat, βCOP [5]. The interaction between COPI and p115 is important for Golgi structure [5]. Furthermore, a recently discovered tethering complex that also acts in the early-Golgi, Grh1/Bug1, interacts with Sec23/24 component of COPII coats [6]. Hence, tether-coat interactions are emerging as a potential common mechanism to mediate tethering of specific subsets of transport vesicles with correct downstream compartments [3, 7]. If tethers recognize coat proteins, what type of coat could the GARP complex recognize at early endosomes? The sorting nexin Snx4, and its interaction partners Snx41, Snx42 are a putative early endosome coat complex that interact with PI3P at endosomes and bind the cargo protein Snc1 [8]. In Chapter 2, we excluded these as putative Vps54-C interaction partners due to their localization to both early and late endosome compartments, and the observation that in their absence Snc1 is predominantly missorted to the vacuole [8, 9]. Recently however, the function of sorting nexins, including Snx4, in mammalian cells has been further characterized revealing a dynamic role in forming transport carriers that are reminiscent of the Vps54-C bound compartment. Sorting nexins all contain a PI3P or PI(3,5)P2 binding PX domain, and a subset also contain a BAR dimerization domain, which forms a rigid curved structure that can associate with vesicles or narrow tubules [10-12]. Snx-BAR proteins are capable of driving and stabilizing the formation of tubules in vitro and in vivo, although the mechanisms are not known [12-14]. Snx1, the mammalian homolog of the yeast retromer component Vps5, and Snx4 are both BAR domain-containing sorting nexins, and in mammalian cells, they are found on different tubules that emanate from the same early endosomal compartment [10, 14]. In a highly dynamic manner, Snx4 positive structures form tubules that contain transferrin receptor (TfnR), a cargo protein that recycles to the cell surface by a slow route that passes through recycling endosomes, or a fast route directly to the surface (reviewed [15]). Due to its recycling itinerary, TfnR’s steady state perinuclear localization is a characteristic marker of recycling endosomes [15]. Similar to the sorting relationship between Snx4 and Snc1 in yeast, when Snx4 is depleted by RNAi, TfnR is largely degraded in the lysosome [14]. It is possible that Snx4 also forms tubules in yeast cells to sort Snc1 away from degradative pathways. The accumulation of 106  tubular-vesicular structures in vps54∆ yeast cells over-expressing Vps54-C could be caused by an accumulation of Snx4-bound tubular compartments with which Vps54-C interacts. A simple experiment to begin testing this hypothesis is to determine whether Snx4 colocalizes with the Vps54-C compartment. Although Snx4 does localize to both early and late endosomes in yeast, these interactions facilitate transport of distinct cargo [8, 16], and Snx4 could expose a unique Vps54-C binding site while in the context of forming Snc1 positive tubules with Snx41 and Snx42. Apart from Snx4/41/42, there is another endosomal coat that could interact with Vps54C. Components of the COPI coat, or coatomer, have been found to interact with Snc1 in vitro, and this interaction is proposed to contribute to early-endosome to late-Golgi retrieval of Snc1 [17]. The Snc1-coatomer interaction requires the Arf and ARL GAP Gcs1, which also interacts with Snc1 and Snx4 [17]. Taken together, the model that components of coatomer and Snx4 act with Gcs1 to retrieve Snc1 from early endosomes is proposed [17], and hence components of the COPI coat are also candidates for the Vps54-C receptor. The timing of tethering complex association with membranes is a significant question in the field. Although many tethers have been reported to interact with coat components, it is not clear whether this interaction occurs while the coat forms, or if the meta-stable coat persists long enough to be tethered [18]. Coat proteins are expected to shield vesicle bound cargo, including SNAREs, and must be removed before a SNARE complex between opposed membranes can form [3, 19]. If the tethering complex binds a coat component during vesicle formation, this implies the coat is stable until it reaches the correct compartment and is then destabilized by another presumably regulated event. Alternatively it has been proposed that the interaction of tethers with coats triggers coat disassembly [18]. Subsequent tethering to the correct compartment could be facilitated by tether-SNARE interactions, or with other tethers that function at the same organelle, as was described for Uso1 and TRAPPI [3, 7]. As no tether-coat interactions have been described that involve sorting nexins, or other putative endosomal coats, it is not clear whether these models apply to mechanisms of tethering vesicles or tubules derived from endosomal compartments. It remains to be determined whether the interaction of GARP with its putative endosomal receptor occurs during vesicle formation or post-budding. However, since cells expressing Vps54-N as the sole copy of Vps54 are dependant on retromer for retrieval of Snc1, it can be inferred that Snc1 is accessible to retromer at late endosomes,  107  suggesting the interaction of Vps54-C and its receptor occurs prior to, and may be required for, vesicle/tubule formation. 4.2.2 Recruitment of tethers by lipid domains - How is GARP recruited to membranes?  Lipids are also possible candidates for the Vps54-C interaction partner as endosomal compartments and Golgi sub-domains are known to be enriched for different lipid species. PI3P is a known endosomal lipid determinant and is a possible candidate that could be tested using in vitro binding assays. Since the mammalian endosomal tether EEA1 contains a PI3P binding motif, there is a precedent [20]. Other phospholipids and sterols are also important in Golgiendosomal traffic. In mammalian cells, glycerophospholipids and sterols are involved in sorting into recycling endosomes, a characteristic that is important for the retrograde transport of ShigaToxin [15, 21]. The role of these lipids in mammalian cells warrants determination of whether the Vps54-C bound compartment is enriched for sterols, an experiment that could be attempted with filipin staining, as it has an affinity for cholesterol and structurally related sterols [22]. If Vps54-C binds to sterol enriched membranes, this recognition event could occur on an uncoated vesicle or tubule, or prior to coat formation. Several phospholipid-translocases (PLT) are also involved in endosomal transport. PLT complexes actively establish and maintain phospholipid asymmetry, which is important for phosphatidylethanolamine (PE) and phosphatidylserine (PS). PLT complexes consist of a catalytic and non-catalytic partner, and in yeast these complexes have various subcellular localizations. Recently a mutant that is defective in all three non-catalytic subunits, as well as the enzymatic component Drs2, were found to accumulate Snc1 in a compartment with characteristics that are reminiscent of that bound by Vps54-C [23]. Like the Vps54-C bound compartment, it exhibits a polarized localization near the bud, co-localizes with Rcy1 and the lipid-dye FM4-64, but not the Golgi protein Sec7. While this mutant is defective in retrograde transport of Snc1 and other early endosome cargo, it is not affected in secretion, endocytosis, or late endosome-to-TGN transport [23]. Moreover, the endosomal-Golgi localized PLT (Cdc50/Drs2) was found to interact with the early endosome recycling protein Rcy1, which interacts with Snc1 and colocalizes with Vps54-C ([23-25], Chapter 2). PLTs may play a role in membrane deformation, acting with coats to form vesicles or tubules [23]. The similarity of this compartment with that bound by Vps54-C supports the concept that Vps54-C is bound to an  108  aberrant, normally dynamic early endosomal compartment. It would be interesting to know if Vps54-C also binds to the compartment formed in the PLT mutant. Current evidence in the literature suggests that tethering complex and coat interactions may be a general model to contribute to the specificity of trafficking pathways. Recent work on sorting nexin function in mammalian cells and yeast, and the role of lipids in endosomal sorting suggest several possible mechanisms by which GARP could associate with upstream compartments supporting implications of the work in Chapter 2. Identification of the interaction partner of Vps54-C is an important next question in understanding recruitment and timing of GARP complex association with membranes. 4.2.3 Relationship between GARP and ARL pathway  The Arf-like GTPase Arl1 interacts with GARP in a GTP-dependent manner, but the significance of this interaction is unknown [26]. Arl1 is localized to the Golgi by the action of an upstream GTPase, Arl3, whose localization requires the activity of an N-acetyltransferase complex (Mak3/Mak10) and a Golgi protein of unknown function, Sys1; collectively this is referred to as the ARL pathway [27-29]. Loss of the ARL pathway did not prevent Vps54-C localization to the polarized compartment suggesting that Arl1-GTP is not the interaction partner of Vps54-C (Chapter 2). Most of the known effectors of Arl1 in mammalian cells are long-coiledcoil tethers that bind to Arl1-GTP through their GRIP domain [30]. In yeast, Imh1 is the only GRIP domain containing protein and, like many GRIP domain proteins in mammalian cells, Imh1 is a coiled-coil tether that requires Arl1 for its localization [26, 27]. Genetic relationships between GARP and the ARL pathway suggest that they may both function to tether endosome derived vesicles with the Golgi [30]. Overexpression of ARL1 or IMH1 can suppress the trafficking defects of ypt6∆, the Rab protein required for GARP function at the Golgi [31, 32]. Deleterious mutations of GARP and ARL pathway genes are also synthetic sick ([33], Chapter 3, Figure 3.4, Appendix B). Single GARP and ARL deletion phenotypes differ in severity, which might indicate that the ARL pathway is not involved in all GARP related functions but rather a subset ([30], Figure 4.1). Moreover, localization of GARP to the Golgi is unaffected by deletion of the ARL pathway [26]. Recently, a mammalian ARL pathway effector GCC185 was found to interact with both Arl1 and Rab6a (Ypt6 homolog) simultaneously [34], which might suggest a similar relationship between GARP, Ypt6 and Arl1 in yeast. In a manner 109  similar to that proposed for TRAPPI, the COPII coat, Ypt1-GTP and the coiled-coil tether Uso1 described above, the coiled-coil tether Imh1 could also be involved in this model of tethering. That is, Ypt6-GTP and Arl1-GTP could simultaneously bind to GARP to tether a subset of endosome-derived vesicles [26, 35] that would first be recognized by the C-terminal portion of Vps54. Interaction of vesicle associated GARP with Ypt6 and Arl1 at the Golgi could disrupt the coat and facilitate interaction with the Arl effector, Imh1, and subsequent fusion.  Figure 4.1 Schematic model of endosomal transport and cell surface recycling in yeast. Similar to mammalian cells, endosomal transport in yeast is likely to involve multiple types of compartments that are related to each other though compartment maturation. Solid arrows illustrate trafficking pathways, and dotted lines illustrate maturation. One hypothesis for the interaction partner of Vps54-C is Snx4, which might interact at a recycling endosome (RE) like compartment in yeast. GARP may interact with both ARL and Ypt6 simultaneously at the Golgi to tether different types of vesicles. GARP may also act independently of ARL at the Golgi with Ypt6. In mammalian cells there are fast and slow recycling routes to the cell surface, the latter, slow route is used by transferrin receptor and transverses the recycling endosome. Here, machinery shown in purple are proposed to be involved in a putative slow recycling route to the cell surface in yeast cells. This route could be used by the GPI anchored protein, Gas1. The fast recycling route, shown in orange, may be used by Gap1 and Fur4. Defects in formation of the late endosome (LE), also called the multi-vesicular body (MVB), could stimulate the transport of cargo through the fast recycling route. For Gap1, recycling to the surface requires the Gse complex.  While GARP and Arl1 may participate in a common trafficking step at the Golgi, Arl1 has also been found to act independently of the GARP pathway, and may act at a recycling 110  endosome-like compartment in yeast. Plasma membrane localization of Gas1, a GPI-anchored ß-1,3-glucanosyltransferase that is important in cell wall integrity, requires the ARL pathway, but not Ypt6 [36]. GPI anchored proteins are associated with sterol-rich membrane micro-domains, and in yeast cells compromised in the ARL pathway, Gas1 accumulates in a compartment near the bud-neck that co-fractionates with the phospholipid translocase Drs2 ([36]). The authors propose a function for the ARL pathway in secretion of Gas1, but could not exclude an alternative role in retrograde transport [36]. In mammalian cells, GPI anchored proteins recycle between the plasma membrane and the sterol-rich recycling endosome [15]. Perhaps Arl1 functions in transport through the recycling endosome (or in yeast a recycling endosome equivalent) with other proteins implicated in early endosome recycling: Gcs1, Snx4, Rcy1 and Snc1 (Figure 4.1). Gcs1, an Arf-GAP that interacts with Snc1 and the coat Snx4, also has activity toward Arl1 [17, 37] and, like arl∆, gcs1∆ cells accumulate Gas1 in a polarized compartment [36]. Snc1 also interacts with Rcy1, the early endosome recycling protein that colocalizes with Vps54-C ([24], Chapter 2). In support of a role for the ARL pathway at recycling endosomes, inhibition of the ARL pathway in mammalian cells causes accumulation of the glycolipid binding protein, Shiga Toxin, and another recycling endosome cargo, TGN46, but does not impair recycling from late endosomes (reviewed in [30]). Taken together, it is possible that the ARL pathway, like GARP, is involved in multiple trafficking pathways, and is likely to affect transport between early endosomes, the Golgi and plasma membrane. In summary, these observations imply that the organization of endosomes in yeast cells is expected to be as complex and dynamic as endosomal sorting in mammalian cells. The relationships between pathways connecting endosomes, the Golgi and plasma membrane are not fully understood. 4.2.4 Comparison of yeast and mammalian GARP  There is evidence that GARP also interacts with upstream compartments in mammalian cells. Transport of cargo from the recycling endosome to the TGN (for example TGN46, and βsubunit of Shiga Toxin), and from later endosomes to the TGN (for example MPR) requires GARP [38]. Overexpression of Syntaxin10, a SNARE protein that interacts with GARP and is also required for endosome-to-TGN transport, recruits Vps52 to enlarged perinuclear structures that colocalize with TfnR [39] suggesting that GARP may be capable of associating with recycling endosomes. One difference between yeast and mammalian GARP is the 111  mislocalization phenotypes that result from GARP deletion or depletion. In yeast, loss of the GARP complex results in vacuolar degradation of its cargo Snc1, and Vps10, which are qualitatively similar to the phenotypes caused by a loss of their respective coat proteins Snx4/41/42 and retromer [8, 40-42]. Other cargoes however, such as Kex2 and A-ALP, are mislocalized to both the vacuole and to vesicle-like structures in yeast strains lacking GARP subunits [40]. In mammalian cells depletion of the GARP complex predominantly results in accumulation of cargo in light-density vesicles, whereas depletion of the coats Snx4 or retromer results in lysosomal degradation of TfnR and MPRs respectively [14, 38, 43]. This difference implies that yeast GARP may be involved at the earlier step of vesicle/tubule formation, whereas mammalian GARP may recognize transport carriers once they are formed. Alternatively, these differences in cargo localization could be due to differences in stability of the GARP complex when subunits are deleted or depleted in yeast versus mammalian cells. Depletion of Vps54 by siRNA did not destabilize Vps53 or Vps52 in cell lysates, while deletion of VPS54 in yeast destabilizes the complex [40], presenting the possibility that in the absence of Vps54, Vps52 and Vps53 could perform a function independent of Vps54 in mammalian cells. Interestingly, although the late endosome cargo, CI-MPR is stable in GARP depleted cells, the recycling-endosome-to-TGN cargo, TGN46 is partially degraded [38]. Taken together, this might indicate that GARP plays different roles at the two retrieval pathways: an early role in the recycling-endosome to TGN pathway, and a post-vesicle formation role in late-endosome to TGN transport; or that Vps52 and Vps53 can function in the early step in the absence of Vps54, but the whole complex is required for tethering at the TGN. It would be interesting to determine if Vps52 and Vps53 are localized to punctate structures when Vps54 is depleted. Vps54-GFP was found predominantly co-localized with TGN proteins, but was also consistently found on small puncta [38], which could reflect association of GARP with dynamic upstream compartments. Although some differences have been observed, the function of the GARP complex in tethering endosome derived vesicles with the Golgi is conserved. 4.2.5 Role of Vps54 in Amyotrophic Lateral Sclerosis (ALS)  The significance of retrograde transport to neurodegenerative diseases was illustrated by the discovery that a missense mutation in the C-terminus of mouse Vps54 causes motor neuron disease in the wobbler mouse model of ALS [44]. This initiated work by others to 112  sequence human VPS54 in sporadic and familial ALS patients (Appendix C, [45]). A rare variant of human VPS54 was discovered but found to have incomplete penetrance; the father of the patient also carried the allele but did not develop ALS. Similarly, mutation of conserved residues in yeast VPS54 did not alter GARP function in vivo (Appendix C). Other evidence however implicates defective retrograde transport as a contributing factor to the development of ALS, perhaps due to an increased sensitivity of motor neurons to defects in transport between the synapse and cell body [45, 46]. Continued investigations of the impact of GARP complex mutations to the occurrence of ALS are underway (personal communication, M. Meisler). A greater understanding of retrograde transport pathway components and relationships in model organisms such as yeast may facilitate identification of additional susceptibility loci, or reveal ways in which subtle defects in retrograde transport could be amplified by modifiers. The biochemical impact of the wobbler VPS54 mutation was also recently investigated in mammalian cells, and was found to have no detectable impact on GARP function in sorting recycling endosome cargo (Shiga-Toxin, TGN46) or late endosome cargo (CI-MPR) [38]. While this may explain viability of the Wobbler mouse [38], experiments were performed in HeLa cells, and thus do not rule out the possibility that motor neurons could be hypersensitive to subtle defects that are not apparent in HeLa cells. Given that motor neuron disease of the wobbler mouse could be rescued by transgenic expression of a wild-type copy of mouse VPS54 [44], one would expect some function of Vps54 to be inhibited in the wobbler mouse. Further work to determine the biochemical defects seen in this model system are required. 4.2.6 Trafficking pathway relationships and recycling to the cell surface  Trafficking pathways comprise a network, and although each is specific, their interconnected and cyclical nature allows cargo to follow alternative routes. This property adds to the complexity of an analysis of pathway components, and was seen throughout the thesis. In Chapter 2, the early endosome function of the C-terminal region of Vps54 was partially masked by the capacity for Snc1 to follow a late endosome-to-TGN retrograde transport pathway (Figure 2.2; Appendix A). In Chapter 3, GSS sorting defects caused by mutation of genes encoding coat proteins were suppressed by mutations that block exit from endosomes to the vacuole; the combined defect enabled use of recycling pathways to the cell surface (Figure 3.7; Appendix B). What is the mechanism of such bypass? Is bypass is a consequence of sorting inefficiency, or are there mechanisms by which cargo is sorted in the absence of signals? The complex 113  organization of endosomes in mammalian cells, combined with the high level of conservation of machinery [15, 47], suggests that both mechanisms play a role in the bypass pathways seen in this work. No trafficking pathway is expected to be 100% efficient; cargo is rather enriched in the compartments to which it is specifically transported [15]. Furthermore, endosomes are increasingly appreciated as a continuum of maturing organelles, progressing from an early sorting endosome that is capable of producing tubules for retrograde transport or recycling (sorting endosome), to a multivesicular body filled with intralumenal vesicles that is capable of fusion with the vacuole/lysosome ([48], Figure 4.1). In the framework of endosomal maturation, it is possible to envision how defects in one pathway can affect another. The observation that defects in two types of coats can activate recycling pathways to the surface when combined with a defect in vacuolar delivery, presents the question, what machinery is responsible for endosome to cell surface recycling? In mammalian cells there are at least two types of recycling pathways to the cell surface, a fast route that is thought to occur from the earliest stage of an endosome, and a slow route, in which cargo are transported through the perinuclear recycling endosome [15]. In yeast cells, there is some evidence for a recycling endosome-like compartment that may involve Rcy1, as discussed in Chapter 2 and section 4.2.3. Recycling directly from endosomes to the cell surface has been described for few cargoes in yeast and among these is the uracil transporter Fur4 [49]. Previous work established that a defect in MVB formation caused by mutation of the ESCRT machinery enhanced recycling of Fur4 to the cell surface [49]. Hyper-efficient recycling was unaffected by a simultaneous block in retrograde transport to the Golgi (GARP deletion strains) or by deletion of RCY1 [49], suggesting that Fur4 may recycle from an endosomal compartment with characteristics of the “fast-recycling” route in mammalian cells. Mutation of Class C Vps genes (VpsClC), which are required for tethering and fusion at endosomes and the vacuole, prevents hyper-efficient recycling of Fur4. Whether VpsClC plays a specific role in Fur4 recycling from endosomes remains to be established. In mammalian cells, sorting nexins are implicated in recycling to the surface [50]. In our data however, the combined loss of the PI3P kinase complex II (Vps30/Vps38) and the sorting nexin Snx4 enhanced recycling. PX domaincontaining sorting nexins are unlikely to be the machinery for recycling as these require PI3P for their membrane association. There is the possibility however that in these conditions other coat complexes like Gga2 substitute, and in the absence of Gga2, sorting nexins substitute. That is, each double mutant with enhanced recycling uses a different mechanism to recycle. Such a model would assume that Fur4 is not sorted but is incorporated into a default pathway to the 114  surface. In support of this, the VpsClC mutants that prevent Fur4 recycling are also defective in the bulk flow of membranes recycled to the surface [49]. There is evidence however for selective incorporation of cargo into endosome-to-cellsurface recycling pathways. The general amino acid permease (Gap1) was also found to have enhanced cell surface recycling in ESCRT mutants, however this could be inhibited by mutation of a newly characterized conserved complex called Gse ([51], Figure 4.1). Subunits of the Gse complex interact with a number of residues on the C-terminal tail of Gap1, and are required for Gap1 to reach the cell surface from endosomes [51]. Loss of Gse does not affect bulk membrane flow to the surface, and Gap1 recycling is also unaffected by mutations in VpsClC [51]. It would be interesting to know if mutation of Gse affects Fur4 or GSS recycling, which would help to determine if the Gse complex is cargo specific. If increased recycling to the cell surface is a general consequence of multiple defects in endosomal and vacuolar/lysosomal delivery, this would have many implications for disease. The mechanisms that enable retrograde transport and cell-surface recycling are still being defined [48]. It is apparent however that the complexity of endosomal transport in yeast cells is similar to that in mammalian cells. Tools that enable high-throughput combinatorial genetic analyses are readily available in yeast, enabling analysis and dissection of complex relationships between genes and pathways. The modification and application of these techniques to study endosomal transport provided insight into the complex relationships between endosomal transport pathways. 4.2.7 Systematic screens of genetic interaction  In Chapter 3, a genetic screen and genetic interaction analyses discovered retrograde transport genes, and defined some of the complex relationships between endosomal trafficking pathways. Cell surface localization of the GSS reporter provided a high-throughput, quantitative measure of trafficking phenotypes ([52], and Chapter 3). As discussed in Chapter 3, many of the genetic backgrounds with low levels of GSS activity were important for endosomal transport. Genetic interaction profiling with a panel of queries involved in trafficking pathways revealed which genes from the primary screen were most interactive with various trafficking pathways, and enabled discovery of a set of genes enriched for functions in endosomal transport (Figure 3.3). One of the implications of this finding is that genetic interaction profiling presents a general 115  unbiased method to determine which genes are highly influential to a given biological process after performing a genomic screen. The value of this minimal “E-MAP” approach to reveal important functional subgroups is supported by recent re-analysis of E-MAP data [53]. In a genomic screen it is expected that not all hits will be direct, or even relevant, so methods to fully use the data to evaluate and prioritize candidates are both valuable and needed [54]. Genetic interaction profiling revealed three clusters of genes that were enriched for direct roles in endosomal transport. Many mitochondrial genes were also included in trafficking clusters, which suggests the role of these genes to transport may be more significant than other pathways that were discovered in the screen, but did not cluster with trafficking genes (Figures 3.2, 3.3). The inclusion of mitochondrial genes may be due to the energy intensive nature of intracellular transport. Mutants of the secretory pathway accumulate different types of aberrant organelles depending on the concentration of glucose in the media [55]. Thus, performing a screen based on a quantitative trafficking phenotype, followed by genetic interaction profiling using the same trafficking phenotype, enriched for a group of highly relevant, functionally related genes. Genetic and underlying biochemical relationships between genes and pathways were assessed using quantitative measures of genetic interaction. A future improvement to genetic interaction profiling is to enable a precise measure of single and double mutant phenotypes from the initial analysis. To enable quantitative measure of genetic interactions in this work, diploid cells were added to control positions of the haploid array in order to facilitate downstream normalization across arrays. Final SGA selection will produce an array of double mutants; but, since the control position diploid cells would not have mated with the query strain, single mutants are produced at these positions. The common set of single mutants present on each double mutant array was used for normalization. The capacity to transform a genetic interaction, or more precisely an alleviating genetic interaction, into a category such as suppressive or coequal was essential to infer deeper meaning to the context of genetic interaction, highlighting the importance of quantitative genetic interaction analysis. Consideration of false positives (incorrect interactions) and false negatives (missed interactions) is also important in high-throughput data collection. In genome-wide synthetic genetic analyses the rate of false negatives was high (estimated at 17-41%) while false positives were minimized by conducting screens in replicate and using random spore analysis [33]. E-MAP analysis lowered the rates of false positives and negatives by focusing interaction analysis on a targeted subset of genes, measuring genetic interactions with reciprocal crosses, 116  and their use of a scoring method that is based on deviation a double mutant from a relevant average phenotype [61, 62]. In our work, we used techniques that are similar to the E-MAP approach and therefore expect the rate of false positives and negatives to be low. To help minimize false positives we also imposed criteria such as a minimum variance, maximum expectation, and a cut-off of reproducibility which increased the stringency of detected genetic interactions (Section 3.5). Nonetheless, when cut-offs are used to define interaction partners, false negatives are expected. A benefit of stringent criteria however was the increased confidence in defined interaction partners; this aided interpretation of genetic interactions and guided further molecular analyses. Interpretation of genetic interactions to appreciate the context of the underlying biochemical relationships required a consideration of many factors. Trafficking pathways are clearly interconnected and cyclical as multiple anterograde and retrograde pathways connect organelles. In this context, interpretation of masking and suppressive interactions does not necessarily imply pathway order. This was exemplified by the observation that suppressive interactions between endosomal coat proteins and regulators seem to reflect complex relationships between pathways rather than individual gene products. Two phenotypic measures of genetic interaction were also applied in this work: growth and trafficking dysfunction. Interpretation of the meaning of an interaction also required a consideration of each phenotype because, as might be expected, they are not entirely independent measures (Figure 3.5). Knowledge of this relationship however enabled an analysis of genetic interactions based on trafficking dysfunction that are relatively independent of growth. Non-growth phenotypes are a valuable way to assess the relationships between genes and pathways in the context of a biological function. Using electron microscopy to assess and quantify vesicle formation and morphology, mutants of the early secretory pathway could be ordered according to their role in vesicle formation or fusion [56]. Assessment of such a specific phenotype accurately related genes in the context of their biochemical function. In our work, construction of the GSS reporter enabled a compromise between highly detailed information, like cellular imaging provides, and high-throughput analysis as applied to trafficking dysfunction. In mammalian cells genetic screens using RNA interference apply many different types of nongrowth assays, and high-throughput imaging platforms are becoming widely used [57-59]. Analyses of genetic interactions in higher organisms using these phenotypes is not yet common, but thus far have provided valuable information about the relationships between gene 117  products [60]. Since many genetic interaction analyses have been accomplished using yeast growth or fitness as a phenotype, models of how to compute an expected neutral genetic interaction are based on fitness. The properties of a non-growth assay should be carefully considered when computing the neutral expectation of combining two genetic mutations. Such evaluations should also be applied to future analyses of genetic interactions using non-growth phenotypes. In the work described in Chapter 3, the predominant type of interaction observed at highstringency with respect to growth was aggravating, that is deletion of two genes resulted in a colony that was smaller than expected. Interpretation of this type of interaction can be complicated, but often reflects relationships between parallel, inter-dependant pathways [63]. Growth has been successfully used as a phenotype to identify both aggravating and alleviating interactions to infer relationships between genes and pathways [62, 64-66]. Quantitative measures of growth are of considerable importance for these analyses and different groups apply diverse techniques [61, 65, 67]. St. Onge and colleagues undertook an analysis of genetic interactions among genes involved in DNA damage and repair by measuring growth curves of single and double mutants grown with our without the DNA damage agent MMS [65]. Using these techniques, alleviating interactions were identified and could be mapped into categories of interaction such as masking and suppression. In other work, measures of the strength of alleviating and aggravating interactions are used to group genes using cluster analyses, or to create additional scores that reflect the relatedness between genes [62, 64, 66]. These largescale techniques are informative for relating genes of a common function and for broadly interpreting relationships between pathways. As was seen in our work and the work of others, the capacity to precisely order single and double mutant phenotypes enables further classification, increasing the amount of information gained about the context and biochemical meaning of interaction [65, 68]. Both strategies have strengths: the former, E-MAP-style approach enables very large scale measures of interaction between many genes, while the latter approach enables deeper analysis of genetic interaction. An ideal approach is to combine the strengths of each, but requires precise high-throughput estimates of single and double mutant phenotypes. Perhaps further development of other non-growth phenotypic assays in yeast could meet these requirements.  118  4.3 Future Directions 4.3.1 GARP function and mechanisms of membrane recognition  Future work for understanding how the GARP complex recognizes multiple membranes would benefit from identification of the interaction partner of Vps54-C. Candidates include sorting nexins, the COPI coat, or possibly lipids. Tether-coat interactions are increasingly being described in the literature and may provide a simple mechanism by which tethers can recognize the correct upstream compartment and vesicle. Snx4/41/42 and components of the COPI coat could be tested for interaction with Vps54-C by immunoprecipitation. Another possible approach is to purify Vps54-C and test for interaction partners from yeast cell lysate by in vitro binding and mass spectrometry. Expression and purification of GST-Vps54-C was attempted but requires optimization. Putative lipid interactions could be tested using PIP dot blots (nitrocellulose membranes spotted with numerous lipids), or liposome binding assays. Studies in yeast versus mammalian cells suggest GARP may have different properties including requirements for stability, subunit composition and whether GARP acts before or after vesicle budding. Further work is required to understand the mechanisms by which GARP functions in yeast and mammalian cells in order to resolve these differences. Truncating subunits and assessing GARP function is a possible approach to identify functional domains in yeast GARP, and in addition to the work described in Chapter 2 for Vps54, this approach has been applied to Vps52 by a past Conibear lab member. Identification of GARP interaction partners in both yeast and mammalian cells should help to address the mechanism and timing of membrane interaction. Moreover, structural work of an intact yeast GARP complex could provide much information about mechanisms for membrane recognition since in yeast GARP is expected to only act in the context of a full complex. 4.3.2 Cell surface recycling  Analyses of genetic interactions among endosomal transport genes pointed to a novel method to stimulate recycling to the cell surface. Many questions remain about the mechanism and implications of this finding. The machinery responsible for recycling, and the compartment from which recycling was observed are both unknown. One hypothesis is that the compartment is an aberrant late endosome that maintains features of an early endosome, enabling a capacity to recycle to the cell surface. This hypothesis is difficult to test since features of early 119  endosomes are not well described in yeast, and proteins that have previously been identified in recycling (eg. Rcy1) may not be involved in this type of recycling. Identifying the machinery responsible is also difficult due to possibilities of redundancy, or if recycling occurs in the absence of specific sorting signals. One candidate that should be tested is the Gse complex since it is important for recycling another amino-acid transporter Gap1, but has not been tested for activity towards Fur4 in the literature [51]. The observation that loss of putative endosomal coats and defects in degradation can enhance cell surface recycling, might imply that yeast endosomes undergo maturation and that multiple pathways for cell surface recycling exist in yeast as they do in mammalian cells. Further work to characterize the mechanisms by which recycling occurs in yeast are needed. 4.3.3 Genetic interaction analyses  Genetic interaction analyses using the GSS reporter as a read-out of trafficking dysfunction was a valuable tool to gain insight into the relationships between genes and pathways of endosomal transport. The assay itself however presents both strengths and weaknesses that could be improved in future experimental design. For instance, the assay is highly sensitive in that it is capable of measuring low levels of GSS available on the surface, however low levels of GSS activity may be difficult to measure with high precision. The work in Chapter 3 focussed on genetic interactions among genes that all presented low levels of GSS activity in a genomic screen. An alternative approach would be to perform genetic interaction analyses by crossing groups of genes with low GSS activity to those with high GSS activity. In this way the dynamic range of the assay would be dramatically improved and the capacity to map interactions between genes involved in endosomal transport with those involved in endocytosis could be performed. Results of such a study could reveal negative regulators of trafficking pathways as these are expected to have opposite phenotypes to their pathways of action. Development of alternative reporters and assays to provide a high-throughput measure of trafficking dysfunction, or perturbations in other biological pathways, are under development in the Conibear lab. Genetic interaction analyses performed with alternative phenotypes to GSS could benefit from some of the lessons learned, and strategies used in this work. For example the normalization techniques, and the methods used to score and categorize genetic interactions described in Chapter 3, could be applied to future studies. Another genetic 120  interaction screen that focuses on a specialized type of endosomal transport has also been performed in the lab. Future work to determine how to integrate information from each of these screens could prove very valuable for understanding the relationships between endosomal transport pathways.  4.4 Concluding Statement A combination of genetic and molecular techniques provided new insights into endosomal and retrograde transport from the perspectives of pathway specificity and pathway relationships. Determining how tethering complexes function and confer specificity is important to understanding the mechanisms of vesicle transport. Work with the GARP complex supports the growing model that tethers recognize features of upstream compartments. Genetic interaction analyses allowed an expansion of this view to appreciate the relationships between endosomal trafficking pathways, providing insight into the relatedness of endosomal compartments. Principles learned from each of these analyses can be applied to future studies of retrograde and endosomal transport.  121  4.5 References 1.  Cai H, Yu S, Menon S, Cai Y, Lazarova D, Fu C, Reinisch K, Hay JC, Ferro-Novick S: TRAPPI tethers COPII vesicles by binding the coat subunit Sec23. Nature 2007, 445(7130):941-944.  2.  Sacher M, Barrowman J, Wang W, Horecka J, Zhang Y, Pypaert M, Ferro-Novick S: TRAPP I implicated in the specificity of tethering in ER-to-Golgi transport. Mol Cell 2001, 7(2):433-442.  3.  Miller EA: Vesicle tethering: TRAPPing transport carriers. Curr Biol 2007, 17(6):R211-213.  4.  Haas AK, Barr FA: COP sets TRAPP for vesicles. Dev Cell 2007, 12(3):326-327.  5.  Guo Y, Punj V, Sengupta D, Linstedt AD: Coat-Tether Interaction in Golgi Organization. Mol Biol Cell 2008, 19(7):2830-2843.  6.  Behnia R, Barr FA, Flanagan JJ, Barlowe C, Munro S: The yeast orthologue of GRASP65 forms a complex with a coiled-coil protein that contributes to ER to Golgi traffic. J Cell Biol 2007, 176(3):255-261.  7.  Cai H, Reinisch K, Ferro-Novick S: Coats, Tethers, Rabs, and SNAREs Work Together to Mediate the Intracellular Destination of a Transport Vesicle. Dev Cell 2007, 12(5):671-682.  8.  Hettema EH, Lewis MJ, Black MW, Pelham HR: Retromer and the sorting nexins Snx4/41/42 mediate distinct retrieval pathways from yeast endosomes. EMBO J 2003, 22(3):548-557.  9.  Huh WK, Falvo JV, Gerke LC, Carroll AS, Howson RW, Weissman JS, O'Shea EK: Global analysis of protein localization in budding yeast. Nature 2003, 425(6959):686-691.  10.  Cullen PJ: Endosomal sorting and signalling: an emerging role for sorting nexins. Nat Rev Mol Cell Biol 2008, 9(7):574-582.  11.  Carlton J, Bujny M, Rutherford A, Cullen P: Sorting nexins--unifying trends and new perspectives. Traffic 2005, 6(2):75-82.  12.  McMahon HT, Gallop JL: Membrane curvature and mechanisms of dynamic cell membrane remodelling. Nature 2005, 438(7068):590-596.  13.  Carlton J, Bujny M, Peter BJ, Oorschot VM, Rutherford A, Mellor H, Klumperman J, McMahon HT, Cullen PJ: Sorting nexin-1 mediates tubular endosome-to-TGN 122  transport through coincidence sensing of high- curvature membranes and 3phosphoinositides. Curr Biol 2004, 14(20):1791-1800. 14.  Traer CJ, Rutherford AC, Palmer KJ, Wassmer T, Oakley J, Attar N, Carlton JG, Kremerskothen J, Stephens DJ, Cullen PJ: SNX4 coordinates endosomal sorting of TfnR with dynein-mediated transport into the endocytic recycling compartment. Nat Cell Biol 2007, 9(12):1370-1380.  15.  Maxfield FR, McGraw TE: Endocytic recycling. Nat Rev Mol Cell Biol 2004, 5(2):121132.  16.  Kama R, Robinson M, Gerst JE: Btn2, a Hook1 Ortholog and Potential Batten Disease-Related Protein, Mediates Late Endosome-Golgi Protein Sorting in Yeast. Mol Cell Biol 2007, 27(2):605-621.  17.  Robinson M, Poon PP, Schindler C, Murray LE, Kama R, Gabriely G, Singer RA, Spang A, Johnston GC, Gerst JE: The Gcs1 Arf-GAP Mediates Snc1,2 v-SNARE Retrieval to the Golgi in Yeast. Mol Biol Cell 2006, 17(4):1845-1858.  18.  Spang A: The life cycle of a transport vesicle. Cell Mol Life Sci 2008, 65(18):27812789.  19.  Bonifacino JS, Glick BS: The mechanisms of vesicle budding and fusion. Cell 2004, 116(2):153-166.  20.  Gaullier JM, Ronning E, Gillooly DJ, Stenmark H: Interaction of the EEA1 FYVE finger with phosphatidylinositol 3-phosphate and early endosomes. Role of conserved residues. J Biol Chem 2000, 275(32):24595-24600.  21.  Mallard F, Tang BL, Galli T, Tenza D, Saint-Pol A, Yue X, Antony C, Hong W, Goud B, Johannes L: Early/recycling endosomes-to-TGN transport involves two SNARE complexes and a Rab6 isoform. J Cell Biol 2002, 156(4):653-664.  22.  Smejkal GB, Hoppe G, Hoff HF: Filipin as a Fluorescent Probe of LipoproteinDerived Sterols on Thin-Layer Chromatograms. Anal Biochem 1996, 239(1):115-117.  23.  Furuta N, Fujimura-Kamada K, Saito K, Yamamoto T, Tanaka K: Endocytic recycling in yeast is regulated by putative phospholipid translocases and the Ypt31p/32pRcy1p pathway. Mol Biol Cell 2007, 18(1):295-312.  24.  Chen SH, Chen S, Tokarev AA, Liu F, Jedd G, Segev N: Ypt31/32 GTPases and their novel F-box effector protein Rcy1 regulate protein recycling. Mol Biol Cell 2005, 16(1):178-192.  25.  Saito K, Fujimura-Kamada K, Furuta N, Kato U, Umeda M, Tanaka K: Cdc50p, a Protein Required for Polarized Growth, Associates with the Drs2p P-Type ATPase 123  Implicated in Phospholipid Translocation in Saccharomyces cerevisiae. Mol Biol Cell 2004, 15(7):3418-3432. 26.  Panic B, Whyte JR, Munro S: The ARF-like GTPases Arl1p and Arl3p act in a pathway that interacts with vesicle-tethering factors at the Golgi apparatus. Curr Biol 2003, 13(5):405-410.  27.  Setty SR, Shin ME, Yoshino A, Marks MS, Burd CG: Golgi recruitment of GRIP domain proteins by Arf-like GTPase 1 is regulated by Arf-like GTPase 3. Curr Biol 2003, 13(5):401-404.  28.  Setty SR, Strochlic TI, Tong AH, Boone C, Burd CG: Golgi targeting of ARF-like GTPase Arl3p requires its Nalpha-acetylation and the integral membrane protein Sys1p. Nat Cell Biol 2004, 6(5):414-419.  29.  Behnia R, Panic B, Whyte JR, Munro S: Targeting of the Arf-like GTPase Arl3p to the Golgi requires N-terminal acetylation and the membrane protein Sys1p. Nat Cell Biol 2004, 6(5):405-413.  30.  Munro S: The Arf-like GTPase Arl1 and its role in membrane traffic. Biochem Soc Trans 2005, 33(Pt 4):601-605.  31.  Tsukada M, Gallwitz D: Isolation and characterization of SYS genes from yeast, multicopy suppressors of the functional loss of the transport GTPase Ypt6p. J Cell Sci 1996, 109 ( Pt 10):2471-2481.  32.  Bensen ES, Yeung BG, Payne GS: Ric1p and the Ypt6p GTPase function in a common pathway required for localization of trans-Golgi network membrane proteins. Mol Biol Cell 2001, 12(1):13-26.  33.  Tong AHY, Lesage G, Bader GD, Ding H, Xu H, Xin X, Young J, Berriz GF, Brost RL, Chang M et al: Global Mapping of the Yeast Genetic Interaction Network. Science 2004, 303(5659):808-813.  34.  Burguete AS, Fenn TD, Brunger AT, Pfeffer SR: Rab and Arl GTPase family members cooperate in the localization of the golgin GCC185. Cell 2008, 132(2):286-298.  35.  Graham TR: Membrane targeting: getting Arl to the Golgi. Curr Biol 2004, 14(12):R483-485.  36.  Liu YW, Lee SW, Lee FJ: Arl1p is involved in transport of the GPI-anchored protein Gas1p from the late Golgi to the plasma membrane. J Cell Sci 2006, 119(Pt 18):3845-3855.  37.  Liu YW, Huang CF, Huang KB, Lee FJ: Role for Gcs1p in regulation of Arl1p at trans-Golgi compartments. Mol Biol Cell 2005, 16(9):4024-4033. 124  38.  Perez-Victoria FJ, Mardones GA, Bonifacino JS: Requirement of the human GARP Complex for mannose 6-phosphate-receptor-dependent sorting of cathepsin D to lysosomes. Mol Biol Cell 2008, 19(6):2350-2362.  39.  Liewen H, Meinhold-Heerlein I, Oliveira V, Schwarzenbacher R, Luo G, Wadle A, Jung M, Pfreundschuh M, Stenner-Liewen F: Characterization of the human GARP (Golgi associated retrograde protein) complex. Exp Cell Res 2005, 306(1):24-34.  40.  Conibear E, Stevens TH: Vps52p, Vps53p, and Vps54p form a novel multisubunit complex required for protein sorting at the yeast late Golgi. Mol Biol Cell 2000, 11(1):305-323.  41.  Conibear E, Cleck JN, Stevens TH: Vps51p mediates the association of the GARP (Vps52/53/54) complex with the late Golgi t-SNARE Tlg1p. Mol Biol Cell 2003, 14(4):1610-1623.  42.  Seaman MN, McCaffery JM, Emr SD: A membrane coat complex essential for endosome-to-Golgi retrograde transport in yeast. J Cell Biol 1998, 142(3):665-681.  43.  Arighi CN, Hartnell LM, Aguilar RC, Haft CR, Bonifacino JS: Role of the mammalian retromer in sorting of the cation-independent mannose 6-phosphate receptor. J Cell Biol 2004, 165(1):123-133.  44.  Schmitt-John T, Drepper C, Muszmann A, Hahn P, Kuhlmann M, Thiel C, Hafner M, Lengeling A, Heimann P, Jones JM et al: Mutation of Vps54 causes motor neuron disease and defective spermiogenesis in the wobbler mouse. Nat Genet 2005, 37(11):1213-1215.  45.  Meisler MH, Russ C, Montgomery KT, Greenway M, Ennis S, Hardiman O, Figlewicz DA, Quenneville NR, Conibear E, Brown RH, Jr.: Evaluation of the Golgi trafficking protein VPS54 (wobbler) as a candidate for ALS. Amyotroph Lateral Scler 2008, 9(3):141-148.  46.  Ström A-L, Gal J, Shi P, Kasarskis EJ, Hayward LJ, Zhu H: Retrograde axonal transport and motor neuron disease. Journal of Neurochemistry 2008, 106(2):495505.  47.  Bock JB, Matern HT, Peden AA, Scheller RH: A genomic perspective on membrane compartment organization. Nature 2001, 409(6822):839-841.  48.  Bonifacino JS, Rojas R: Retrograde transport from endosomes to the trans-Golgi network. Nat Rev Mol Cell Biol 2006, 7(8):568-579.  49.  Bugnicourt A, Froissard M, Sereti K, Ulrich HD, Haguenauer-Tsapis R, Galan J-M: Antagonistic Roles of ESCRT and Vps Class C/HOPS Complexes in the Recycling of Yeast Membrane Proteins. Mol Biol Cell 2004, 15(9):4203-4214. 125  50.  Seaman MN: Endosome protein sorting: motifs and machinery. Cell Mol Life Sci 2008, 65(18):2842-2858.  51.  Gao M, Kaiser CA: A conserved GTPase-containing complex is required for intracellular sorting of the general amino-acid permease in yeast. Nat Cell Biol 2006, 8(7):657-667.  52.  Burston H, Maldonado-Baez L, Montpetit B, Davey M, Schulter C, Wendland B, Conibear E: Regulators of Yeast Endocytosis Identified by Systematic Quantitative Analysis. J Cell Biol 2009, In press.  53.  Casey FP, Cagney G, Krogan NJ, Shields DC: Optimal stepwise experimental design for pairwise functional interaction studies. Bioinformatics 2008:btn472.  54.  Yaffe MB: How to "Cell" a Genomic or Proteomic Screen. Sci Signal 2008, 1(47):eg9.  55.  Novick P, Ferro S, Schekman R: Order of events in the yeast secretory pathway. Cell 1981, 25(2):461-469.  56.  Kaiser CA, Schekman R: Distinct sets of SEC genes govern transport vesicle formation and fusion early in the secretory pathway. Cell 1990, 61(4):723-733.  57.  Wheeler DB, Carpenter AE, Sabatini DM: Cell microarrays and RNA interference chip away at gene function. Nat Genet 2005, 37 Suppl:S25-30.  58.  Carpenter AE, Sabatini DM: Systematic genome-wide screens of gene function. Nat Rev Genet 2004, 5(1):11-22.  59.  Moffat J, Sabatini DM: Building mammalian signalling pathways with RNAi screens. Nat Rev Mol Cell Biol 2006, 7(3):177-187.  60.  Bakal C, Linding R, Llense F, Heffern E, Martin-Blanco E, Pawson T, Perrimon N: Phosphorylation Networks Regulating JNK Activity in Diverse Genetic Backgrounds. Science 2008, 322(5900):453-456.  61.  Collins S, Schuldiner M, Krogan N, Weissman J: A strategy for extracting and analyzing large-scale quantitative epistatic interaction data. Genome Biology 2006, 7(7):R63.  62.  Schuldiner M, Collins SR, Thompson NJ, Denic V, Bhamidipati A, Punna T, Ihmels J, Andrews B, Boone C, Greenblatt JF et al: Exploration of the Function and Organization of the Yeast Early Secretory Pathway through an Epistatic Miniarray Profile. Cell 2005, 123(3):507-519.  126  63.  Boone C, Bussey H, Andrews BJ: Exploring genetic interactions and networks with yeast. Nat Rev Genet 2007, 8(6):437-449.  64.  Collins SR, Miller KM, Maas NL, Roguev A, Fillingham J, Chu CS, Schuldiner M, Gebbia M, Recht J, Shales M et al: Functional dissection of protein complexes involved in yeast chromosome biology using a genetic interaction map. Nature 2007, 446(7137):806-810.  65.  St. Onge RP, Mani R, Oh J, Proctor M, Fung E, Davis RW, Nislow C, Roth FP, Giaever G: Systematic pathway analysis using high-resolution fitness profiling of combinatorial gene deletions. Nat Genet 2007, 39(2):199-206.  66.  Decourty L, Saveanu C, Zemam K, Hantraye F, Frachon E, Rousselle J-C, FromontRacine M, Jacquier A: Linking functionally related genes by sensitive and quantitative characterization of genetic interaction profiles. Proceedings of the National Academy of Sciences 2008, 105(15):5821-5826.  67.  Jasnos L, Korona R: Epistatic buffering of fitness loss in yeast double deletion strains. Nat Genet 2007, 39(4):550-554.  68.  Drees B, Thorsson V, Carter G, Rives A, Raymond M, Avila-Campillo I, Shannon P, Galitski T: Derivation of genetic interaction networks from quantitative phenotype data. Genome Biol 2005, 6(4):R38.  127  Appendix A: Erratum The authors of “Domains within the GARP subunit Vps54 confer separate functions in complex assembly and early endosome recognition” (MBC 17(4): 1859-1870) would like to make the following correction. In Figure 8, we erroneously show that a quadruple point mutant of Vps54 (Vps54EWNS) displays the same defect in the early endosome sorting of Snc1 as a Vps54 truncation mutant (Vps54-N) lacking the C-terminal domain. We have been unable to reproduce the Snc1 sorting defect shown for Vps54EWNS; however, we find the Snc1 sorting defect of the Vps54 truncation mutant to be fully reproducible, supporting our conclusion that the C-terminal region of Vps54 is important for early endosome sorting. We have repeated and extended our observations using the Vps54 truncation mutant as shown below (Figure1). Snc1 localization was evaluated in vps54∆ and vps5∆ vps54∆ mutant strains expressing Vps54, Vps54-N or empty vector by blind scoring of at least 150 cells/strain. When wild type Vps54 was present, the localization of Snc1 to the plasma membrane did not require the VPS5-dependent late endosome recycling pathway. In contrast, when Vps54-N was the sole copy of Vps54, Snc1 was found at the plasma membrane (~50% of cells) only when the late endosome pathway was functional. This is consistent with the hypothesis that Snc1-GFP is missorted at early endosomes and aberrantly transported via late endosomes when the Cterminal region of Vps54 is missing. Given these results, our original conclusions regarding the early endosome-specific function of the Vps54 C-terminal domain are unchanged, and all other observations or conclusions in the paper remain valid. We offer sincere apologies for any inconvenience caused by this revision. Figure A.1  128  Appendix B: Supplemental Figures to Chapter 3  Figure B.1: Network of genetic interactions based on growth. Interactions shown are greater than 2 standard deviations from the mean growth t-statistic. Aggravating (red) and alleviating (green) interactions are shown. Those described previously and recapitulated here (dashed line) and discovered here (solid) are shown.  129  Figure B.2: Double mutants are hyper-efficient for recycling Fur4-GFP. Single and double mutant strains are induced to express Fur4-GFP and cells are viewed by fluorescence microscopy (t0). After carbon starvation (CS), Fur4-GFP localizes intracellularly. Relief of carbon stress (carbon activation, CA), enables partial restoration of plasma membrane localization in double mutant strains, like was previously observed for ESCRT mutants (vps37∆). Images are exposed and scaled equally; levels were selected based on visualization of plasma membrane Fur4-GFP resulting in over-exposure of some internal puncta.  130  Figure B.3: Double mutants are hypersensitive to the toxic uracil analogue 5-FU. Single and double mutant strains were grown to exponential phase in minimal media, and equal concentrations of strains were serially diluted onto minimal media (left) or minimal media plus 5µM F-FU.  131  Appendix C: Meisler et al. (2008) Evaluation of the Golgi trafficking protein VPS54 (wobbler) as a candidate for ALS  132  This article was downloaded by: [Canadian Research Knowledge Network] On: 10 April 2009 Access details: Access Details: [subscription number 770885181] Publisher Informa Healthcare Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK  Amyotrophic Lateral Sclerosis Publication details, including instructions for authors and subscription information: http://www.informaworld.com/smpp/title~content=t713656198  Evaluation of the Golgi trafficking protein VPS54 (wobbler) as a candidate for ALS Miriam H. Meisler a; Carsten Russ b; Kate T. Montgomery c; Matthew Greenway de; Sean Ennis e; Orla Hardiman f; Denise A. Figlewicz g; Nicole R. Quenneville h; Elizabeth Conibear h; Robert H. Brown JR b a Department of Human Genetics, University of Michigan, Ann Arbor, Michigan b Department of Neurology, Harvard Medical School, Boston c Harvard Partners Genome Center, Cambridge, Massachusetts d Department of Clinical Neurological Sciences, Royal College of Surgeons in Ireland, Dublin e National Centre for Medical Genetics, Dublin f Department of Neurology, Beaumont Hospital, Dublin, Ireland g Department of Neurology, University of Michigan, Ann Arbor, Michigan h Centre for Molecular Medicine and Therapeutics, University of British Columbia, Vancouver, Canada First Published:2008  To cite this Article Meisler, Miriam H., Russ, Carsten, Montgomery, Kate T., Greenway, Matthew, Ennis, Sean, Hardiman, Orla,  Figlewicz, Denise A., Quenneville, Nicole R., Conibear, Elizabeth and Brown JR, Robert H.(2008)'Evaluation of the Golgi trafficking protein VPS54 (wobbler) as a candidate for ALS',Amyotrophic Lateral Sclerosis,9:3,141 — 148 To link to this Article: DOI: 10.1080/17482960801934403 URL: http://dx.doi.org/10.1080/17482960801934403  PLEASE SCROLL DOWN FOR ARTICLE Full terms and conditions of use: http://www.informaworld.com/terms-and-conditions-of-access.pdf This article may be used for research, teaching and private study purposes. Any substantial or systematic reproduction, re-distribution, re-selling, loan or sub-licensing, systematic supply or distribution in any form to anyone is expressly forbidden. The publisher does not give any warranty express or implied or make any representation that the contents will be complete or accurate or up to date. The accuracy of any instructions, formulae and drug doses should be independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims, proceedings, demand or costs or damages whatsoever or howsoever caused arising directly or indirectly in connection with or arising out of the use of this material.  133  Amyotrophic Lateral Sclerosis. 2008; 9: 141Á148  ORIGINAL ARTICLE  Evaluation of the Golgi trafficking protein VPS54 (wobbler) as a candidate for ALS  MIRIAM H. MEISLER1, CARSTEN RUSS2, KATE T. MONTGOMERY3, MATTHEW GREENWAY4,5, SEAN ENNIS5, ORLA HARDIMAN6, DENISE A. FIGLEWICZ7, NICOLE R. QUENNEVILLE8, ELIZABETH CONIBEAR8 & ROBERT H. BROWN JR2 Downloaded By: [Canadian Research Knowledge Network] At: 18:39 10 April 2009  1  Department of Human Genetics, University of Michigan, Ann Arbor, Michigan, 2Department of Neurology, Harvard Medical School, Boston, 3Harvard Partners Genome Center, Cambridge, Massachusetts, 4Department of Clinical Neurological Sciences, Royal College of Surgeons in Ireland, Dublin, 5National Centre for Medical Genetics, Dublin, 6 Department of Neurology, Beaumont Hospital, Dublin, Ireland, 7Department of Neurology, University of Michigan, Ann Arbor, Michigan, and 8Centre for Molecular Medicine and Therapeutics, University of British Columbia, Vancouver, Canada  Abstract VPS54 is a component of the Golgi-associated retrograde protein (GARP) complex of vesicle sorting proteins. A missense mutation of Vps54 is responsible for motor neuron disease in the wobbler mouse, but the human gene on chromosome 2p14Á15 has not been evaluated as a disease gene. We completely sequenced the 22 coding exons from 96 individuals with sporadic ALS, 96 individuals with familial ALS, and 96 controls. Twenty-one novel SNPs were identified. The nonsynonymous variant, T360A, was observed in one patient and 0/910 controls. Several polymorphic non-synonymous SNPs were also observed in patients and controls. These initial data suggest that mutations in VPS54 are not a major cause of ALS. Key words: ALS, VPS54, wobbler mouse, GARP complex, mutation  Introduction ALS is a severe neurological disorder of largely unknown etiology characterized by degeneration of upper and lower motor neurons, insidious focal onset, and progression to death from respiratory paralysis in 3Á5 years. Approximately 90% of ALS is sporadic, and 10% is familial; identified genes account for B5% of cases. Since gene identification provides insight into disease pathology and new targets for therapy, it is a high priority to identify additional genetic causes of ALS. Motor neurons appear to be more sensitive than other neurons to mutations affecting trafficking of subcellular vesicles, possible due to their requirement for long-distance axonal transport of proteins and vesicles. ALS disease mutations have been identified in VAPB/ALS1 (OMIM 105400), an ER associated protein involved in intracellular transport (1), and alsin/ALS2, (OMIM 606352), which is  believed to regulate endosome function (2Á4). Mutations in the dynactin gene cause slowly progressive motor neuron disease (5), and mutations in maspardin/ACP33, a component of the endosomal/ trans-Golgi network, are responsible for hereditary spastic paraplegia (6). Most recently, a missense mutation of the vesicle trafficking protein VPS54 was identified in wobbler, a mouse model of motor neuron disease (7). VPS54 (GenBank AY444798) is a subunit of the Golgi-associated retrograde protein (GARP) complex required for retrograde transport of vesicles to the trans-Golgi network (TGN). The wobbler (wr) mouse exhibits early onset motor neuron degeneration with impaired righting reflex (8). Abnormal swelling of the ER in lower motor neurons is evident within two weeks after birth (9Á11). Motor neurons are reduced in number to half of wild-type values by 40 days of age (12). The wobbler mutation L967Q is located 10 residues upstream of the C-terminus  Correspondence: M. H. Meisler, The Department of Human Genetics, University of Michigan Medical School, 1241 E. Catherine St., Ann Arbor, MI 481090618, USA. Fax: 1 734 763 1053. E-mail: meislerm@umich.edu (Received 13 June 2007; accepted 17 January 2008) ISSN 1748-2968 print/ISSN 1471-180X online # 2008 Informa UK Ltd. (Informa Healthcare, Taylor & Francis AS) DOI: 10.1080/17482960801934403  134  142  M. H. Meisler et al.  A  ALS  wobbler  L967Q  T360A  1  coiled-coil  Downloaded By: [Canadian Research Knowledge Network] At: 18:39 10 April 2009  B  F  S  T360A  977  Y  C  Figure 1. A missense variant of VPS54 in a patient with sporadic ALS. (A) Location of the missense variant in an ALS patient and the mutation responsible for autosomal recessive motor neuron disease in the wobbler mouse (Schmitt-John et al., 2005). (B) Sequencing chromatogram identifying the heterozygous missense mutation T360A in genomic DNA from the patient with sporadic ALS. (C) Evolutionary conservation of VPS54 residue threonine 360. The sequence alignment was generated with T-coffee (21). Black shading indicates residue identity; gray shading indicates conservative amino acid substitution in comparison with the human sequence. Sequences include H. sapiens (NP_057600), M. musculus (AAH25012.1), G. gallus (XP_419349.1), T. nigroviridis (CAF89824.1), D. melanogaster (Q9VLC0), C. elegans (Q22639), S. cerevisiae (NP_010310).  (Figure 1A, 7). The mutation also causes male infertility due to defective sperm motility (8). A second mouse allele with very low expression results in embryonic lethality in homozygotes and motor neuron degeneration in some heterozygotes (7, and unpublished observations). The human homolog of VPS54 is located on chromosome 2p14Á15 and the human protein has been localized to the GARP complex (13). We describe an initial screen for mutations of human VPS54 in patients with ALS.  Methods Sporadic and familial ALS cases of European ancestry from the Massachusetts General Hospital collection were screened by complete sequencing of the 22 coding exons of VPS54 at the Harvard Partners Genome Center. Blood samples were obtained according to protocols approved by our Institutional Review Board. Patients fulfilled the El Escorial criteria for the diagnosis of this disease. The age of onset for SALS patients was 53915 years (mean9SD, n 079) (median 54 years), and for FALS patients was 55915 years (median 55 years) (n 062). The disease duration was 4.894 years for SALS and 3.493.2 years for FALS patients. The male:female ratio was 2:1 for SALS patients and  1:1.3 for FALS patients. The site of disease onset was 23% bulbar, 43% upper extremities, 28% lower extremities and 7% multiple sites for SALS patients, and 27% bulbar, 31% upper extremities, 37% lower extremities and 5% multiple sites for FALS patients. Most of these samples were not screened for mutations in SOD1, which account for approximately 20% of familial ALS and 2% of sporadic ALS. Controls were also of European ancestry. Ageand ethinically-matched spouses of ALS patients provided 92 controls. Another cohort of 111 controls was older than 60 years of age and lacked personal or family history of neurological disease (Ranier et al., 2006). In order to determine the population frequency of the A360T allele, an additional 266 patients and 814 controls of European ancestry from the National Centre for Medical Genetics, Dublin were analyzed by sequencing exon 8 only. Yeast Vps54 alleles were generated using the Quikchange technique (Stratagene, La Jolla CA). In brief, clone pLC104 (Vps54-3 )HA in pRS316, 14) was mutagenized by amplification with oligonucleotides harboring mismatches at codon T415A to generate clone pNQ09. pNQ09 and Snc1-HAGFP:pRS416 (gift of J. Gerst) were transformed into the MATa vps54 deletion strain (Open Biosys135  VPS54 and ALS tems, Huntsville, AL). Previously described methods were used for detection of CPY by colony overlay (14) and localization of Snc1-GFP (15). Cells were viewed using a Zeiss Axioplan2 fluorescence microscope (Thornwood, NY). Images were captured with a CoolSnap camera using MetaMorph software (Universal Imaging, West Chester, PA) and adjusted using Adobe Photoshop (San Jose, CA).  Downloaded By: [Canadian Research Knowledge Network] At: 18:39 10 April 2009  Results The 22 coding exons of VPS54, exon 2 through exon 23, were amplified from genomic DNA using the primers in supplementary Table I. All 22 exons were sequenced from a panel of 293 DNA samples including 88 sporadic ALS, 109 familial ALS, and 96 control samples. Four non-synonymous variants were identified in this initial screen (Table I). Three of these, M353V, S561C, and M912I, were present in patients and controls and appear to be neutral polymorphisms of VPS54. The fourth non-synonymous variant, T360A, was identified in exon 8 from one patient with no family history of disease (Figure 1A, B). The male proband experienced onset of weakness in the upper left extremity at 44.2 years of age. Diagnosis at 44.5 years identified both upper and lower motor neuron findings affecting arms, legs and bulbar function. The patient, who had no history of smoking or military service, survived for 27 months after diagnosis. Genotyping of the parents revealed that patient’s father is also heterozygous for T360A. The father has remained disease-free into his 80s. Evolutionary comparison indicates that threonine residue 360 is highly conserved among vertebrate VPS54 genes (Figure 1C). The two exceptions, frog and fish, contain the polar residues serine or methionine at this position, in contrast to the hydrophobic alanine residue in the patient. To determine the frequency of the T360A variant in a larger sample, we sequenced exon 8 from a large number of  143  individuals including 234 additional patients with sporadic ALS, 28 patients with familial ALS, and 814 controls (Table I). No additional occurrence of T360A was detected, demonstrating a combined allele frequency in patients plus controls of 50.0005. To evaluate the effect of the T360A variant on VPS54 function, we introduced the corresponding mutation Vps54T415A into yeast and tested for defects in two different retrograde transport pathways. The Vps54T415A allele was expressed at wildtype levels, suggesting the mutation does not affect protein stability (Figure 2C). No differences were observed between wild-type Vps54 and the mutant Vps54T415A with respect to retrograde transport from either late endosomes (Figure 2A) or early endosomes (Figure 2B), indicating that basic function of VPS54 is retained despite the presence of the T415A missense change. Among the 288 individuals that were sequenced, we did not observe any nonsense mutations, altered splice sites, or failures of exon amplification. Twenty-six individuals were homozygous for all of the 34 SNPs located within the 10 kb of amplified exon fragments that were sequenced, which span a distance of 125 kb in genomic DNA. These individuals are potential carriers of genomic deletions, and included 1/96 FALS patient, 15/96 SALS patients and 10/96 controls. Allele frequencies for the 34 SNPs are presented in supplementary Table II.  Discussion VPS54 was originally identified in a genetic screen in S. cerevisiae for mutations causing mis-sorting of the vacuolar protease carboxypeptidase Y (14). The yeast Vps54 protein is associated with Vps51, Vps52 and Vps53 in a tetrameric protein complex referred to as the GARP (Golgi-associated retrograde protein) complex (16). The GARP complex mediates tethering of specific membrane-bounded vesicles to the membranes of the trans-Golgi net-  Table I. Exonic coding variants of human VPS54. Exons 2 to 23 were amplified and sequenced from genomic DNA. cDNA nucleotide positions from Genbank NM_016516. Heterozygote frequency Exon exon 8 exon 8 exon 12 exon 22  Variant  FALS  SALS  Controls  Overall allele frequency, minor allele  M353V c.A1220G T360A c.A1241G S561C c.C1845G M912I c.G2899T rs17619976  1/124  2/330  3/340  0.007  0/124  1/330  0/910  Á  1/88  0/109  1/96  0.003  22/88  16/109  18/96  0.11*  *For SNP rs17619976 the calculation of minor allele frequency included one homozygous FALS patient, one homozygous SALS, and two homozygous controls. FALS: familial ALS; SALS: sporadic ALS.  136  M. H. Meisler et al.  Downloaded By: [Canadian Research Knowledge Network] At: 18:39 10 April 2009  144  Figure 2. The T415A mutation in yeast Vps54. (A) The Vps54T415A mutant is competent for late endosome recycling. vps54 knockout strains (vps54D) were transformed with plasmids expressing wild-type VPS54, the mutant VPS54T415A allele or vector alone and tested for secretion of carboxypetidase Y (CPY) by colony overlay assay. (B) Vps54T415A is functional for retrograde transport from early endosomes. Strains described in A were transformed with Snc1-HA-GFP. Snc1-HA-GFP is localized to the plasma membrane at sites of polarized growth in vps54D strains expressing either wild-type VPS54 or the VPS54T415A allele. (C) Vps54T415A is expressed at the same level as wildtype Vps54. HA-tagged Vps54 protein and Snc1-HA-GFP were each detected by an anti-HA Western blot of strains described in B.  work, conferring specificity to the process of retrograde transport to the Golgi (17,18). This cellular function appears to be conserved in the mammalian GARP proteins (13,19). The identification of the wobbler mutation in VPS54 revealed that motor neurons are particularly dependent upon the VPS54 protein for survival. To evaluate the role of human VPS54 in motor neuron disease, we screened several hundred patients with sporadic and familial ALS and a larger number of controls. In the 197 ALS samples that were completely sequenced, only four coding variants were observed; three of these were seen in equal frequency in affected and control  individuals and thus appear to be neutral polymorphisms. The rare variant T360A was observed in one heterozygous ALS patient and his unaffected father, and was not detected in more than 900 control samples. The data are consistent with a potential role for this variant in susceptibility to motor neuron disease, with incomplete penetrance, but a functional test for the vertebrate protein will be required to assess its pathogenicity. Introduction of the corresponding mutation into the yeast gene did not impair endosomal trafficking, but this does not preclude an adverse, neurotoxic property of the 137  VPS54 and ALS mutant protein. For example, the majority of SOD1 mutants that cause ALS have been found to retain dismutase activity, but toxic effects of the mutated protein lead to motor neuron death (20). The results of this initial survey indicate that exonic and splice site mutations in VPS54 are unlikely to be a major cause of ALS, since only one disease-specific non-synonymous mutation was identified among 197 cases. It will be of interest in the future to extend this study to evaluation of copy number variation, since in the mouse there is a small incidence of motor neuron disease in heterozygotes for the null allele of VPS54.  Downloaded By: [Canadian Research Knowledge Network] At: 18:39 10 April 2009  Acknowledgements Support was provided by the Office of the Executive Vice President for Research, University of Michigan School of Medicine and by grants from the ALS Association (MM) and the Health Research Board of Ireland (MG, OH). DAF is supported by NIEHS and the Muscular Dystrophy Association. EC is supported by funding from CIHR, the BCRICWH and by a MSFHR Scholar Award. NRQ is supported by a CFRI Graduate Studentship. Also supported by the NINDS, the NIA, the ALS Association, Project ALS, the Angel Fund, the Pierre L. de Bourgknecht ALS Research Foundation, and the Al-Athel ALS Foundation. We express thanks to Raymond Ruddon, Julie M. Jones and Matthew Avenarius for valuable contributions. References 1. Nishimura AL, Mitne-Neto M, Silva HC, Richieri-Costa A, Middleton S, Cascio D, et al. A mutation in the vesicletrafficking protein VAPB causes late-onset spinal muscular atrophy and amyotrophic lateral sclerosis. Am J Hum Genet. 2004;75:822Á31. 2. Hadano S, Hand CK, Osuga H, Yanagisawa Y, Otomo A, Devon RS, et al. A gene encoding a putative GTPase regulator is mutated in familial amyotrophic lateral sclerosis 2. Nat Genet. 2001;29:166Á73. 3. Yang Y, Hentati A, Deng HX, Dabbagh O, Sasaki T, Hirano M, et al. The gene encoding alsin, a protein with three guanine-nucleotide exchange factor domains, is mutated in a form of recessive amyotrophic lateral sclerosis. Nat Genet. 2001;29:160Á5. 4. Topp JD, Gray NW, Gerard RD, Horazdovsky BF. Alsin is a Rab5 and Rac1 guanine nucleotide exchange factor. J Biol Chem. 2004;279:24612Á23.  145  5. Puls I, Jonnakuty C, LaMonte BH, Holzbaur EL, Tokito M, Mann E, et al. Mutant dynactin in motor neuron disease. Nat Genet. 2003;33:455Á6. 6. Simpson MA, Cross H, Proukakis C, Pryde A, Hershberger R, Chatonnet A, et al. Maspardin is mutated in mast syndrome, a complicated form of hereditary spastic paraplegia associated with dementia. Am J Hum Genet. 2003;73:1147Á56. 7. Schmitt-John T, Drepper C, Mussmann A, Hahn P, Kuhlmann M, Thiel C, et al. Mutation of Vps54 causes motor neuron disease and defective spermiogenesis in the wobbler mouse. Nat Genet. 2005;37:1213Á5. 8. Boillee S, Peschanski M, Junier MP. The wobbler mouse: a neurodegeneration jigsaw puzzle. Mol Neurobiol. 2003;28:65Á106. 9. Duchen LW, Strich SJ. A hereditary motor neuron disease with progressive denervation of muscle in the mouse: the mutant ‘wobbler’. J Neurol Neurosurg Psychiatry. 1968; 31:535Á42. 10. Andrews JM. The fines structure of the cervical spinal cord, ventral root and brachial nerves in the wobbler (wr) mouse. J Neuropathol Exp Neurol. 1975;34:12Á27. 11. Mitsumoto H, Bradley WG. Murine motor neuron disease (the wobbler mouse): degeneration and regeneration of the lower motor neuron. Brain. 1982;105:811Á34. 12. Pollin MM, McHanwell S, Slater CR. Loss of motor neurons from the median nerve motor nucleus of the mutant mouse ‘wobbler’. J Neurocytol. 1990;19:29Á38. 13. Liewen H, Meinhold-Heerlein I, Oliveira V, Schwarzenbacher R, Luo G, Wadle A, et al. Characterization of the human GARP (Golgi- associated retrograde protein) complex. Exp Cell Res. 2005;306:24Á34. 14. Conibear E, Stevens TH. Vps52p, Vps53p, and Vps54p form a novel multi-subunit complex required for protein sorting at the yeast late Golgi. Mol Biol Cell. 2000;11:305Á23. 15. Quenneville NR, Chao TY, McCaffery JM, Conibear E. Domains within the GARP subunit Vps54 confer separate functions in complex assembly and early endosome recognition. Mol Biol Cell. 2006;17:1859Á70. 16. Conibear E, Cleck JN, Stevens TH. Vps51p mediates the association of the GARP (Vps52/53/54) complex with the late Golgi t-SNARE Tlg1p. Mol Biol Cell. 2003;14:1610Á23. 17. Oka T, Krieger M. Multi-component protein complexes and Golgi membrane trafficking.J Biochem. (Tokyo) 2005; 137:109Á14. 18. Lupashin V, Sztul E. Golgi tethering factors. BBA Á Molecular Cell Research. 2005;1744: 325Á39. 19. Walter L, Stark S, Helou K, Flugge P, Levan G, Gunther E. Identification, characterization and cytogenetic mapping of a yeast Vps54 homolog in rat and mouse. Gene. 2002; 285:213Á20. 20. Pasinelli P, Brown RH. Molecular biology of amyotrophic lateral sclerosis: insights from genetics. Nat Rev Neurosci. 2006;7:710Á23. 21. Notredame C, Higgins DG, Heringa J. T-coffee: a novel method for fast and accurate multiple sequence alignment. J Mol Biol. 2000;302:205Á17.  138  146  M. H. Meisler et al.  Supplemental Table I. Primers for amplification of the 22 coding exons of VPS54. Numbering is based on the gene model in the UCSC genome browser, mouse build 35 (www.genome.ucsc.edu). Lower case sequence represents the M13 tags that permit sequencing the amplified exons with universal M13 primers. Due to the high GC content around the non-coding exon, exon 1, it was not included in the study.  Downloaded By: [Canadian Research Knowledge Network] At: 18:39 10 April 2009  primer name MMA_VPS54-EX2.F MMA_VPS54-EX2.R MMA_VPS54-EX3.F MMA_VPS54-EX3.R MMA_VPS54-EX4.F MMA_VPS54-EX4.R MMA_VPS54-EX5.F MMA_VPS54-EX5.R MMA_VPS54-EX6.F MMA_VPS54-EX6.R MMA_VPS54-EX7.F MMA_VPS54-EX7.R MMA_VPS54-EX8.F MMA_VPS54-EX8.R MMA_VPS54-EX9.F MMA_VPS54-EX9.R MMA_VPS54-EX10.F MMA_VPS54-EX10.R MMA_VPS54-EX11.F MMA_VPS54-EX11.R MMA_VPS54-EX12.F MMA_VPS54-EX12.R MMA_VPS54-EX13.Fa MMA_VPS54-EX13.Ra MMA_VPS54-EX14.F MMA_VPS54-EX14.R MMA_VPS54-EX15.F MMA_VPS54-EX15.R MMA_VPS54-EX16.F MMA_VPS54-EX16.R MMA_VPS54-EX17.F MMA_VPS54-EX17.R MMA_VPS54-EX18.F MMA_VPS54-EX18.R MMA_VPS54-EX19-20.F MMA_VPS54-EX19-20.Fa MMA_VPS54-EX19-20.R MMA_VPS54-EX21.F MMA_VPS54-EX21.R MMA_VPS54-EX22.F MMA_VPS54-EX22.R MMA_VPS54-EX23.F MMA_VPS54-EX23.R  primer sequence tgtaaaacgacggccagtCTTGGATTTTGATTTGCGGT aacagctatgaccatgTCTCGATCTCCTGACCTCGT tgtaaaacgacggccagtGCTGCCAGCTGAAGGTTATT aacagctatgaccatgTGCAACCCAGGACAGTACAA tgtaaaacgacggccagtTGTTGCAAAACACTTTCATG aacagctatgaccatgGGGACTACACACATGTGCCA tgtaaaacgacggccagtTTGGGGACCACTGAGATAGC aacagctatgaccatgGCCAAATGAAATCTGCTTGT tgtaaaacgacggccagtCCATCTCCATTGGTAAAGGC aacagctatgaccatgTCCATTGAACCAAAATGCTG tgtaaaacgacggccagtGCCCAATTTCAAGGATCTCA aacagctatgaccatgCAAATGATCAATCTGGAAGAAAAA tgtaaaacgacggccagtTGATTGCACTGGTGTTTGGT aacagctatgaccatgACAATCATGGTGCTTGCTCA tgtaaaacgacggccagtCGAAAACCAGCTTCATTGGT aacagctatgaccatgAATCTTTTGGCTGGTAAGGC tgtaaaacgacggccagtCCTTGCCAGCACCTGTTATT aacagctatgaccatgCCTACACCCTGCTCACACCT tgtaaaacgacggccagtTATTGATTGGCCCAGAGACA aacagctatgaccatgAAAAATCACGAGAGGAAATATTACTG tgtaaaacgacggccagtTGCACAGAATTCTCTTTTGGG aacagctatgaccatgTGCTCACCTCCTTACTTCTCA tgtaaaacgacggccagtCTTTTTATTGATAGGTCACCTGTG aacagctatgaccatgATCTCCCTAGTAGACCACATCAG tgtaaaacgacggccagtTGCTCAATTTCTGGGTCTGA aacagctatgaccatgCACTGGCCAATTAGGTTTCC tgtaaaacgacggccagtTTGACTTTTGTTCCCACTCAAA aacagctatgaccatgAACATGGTCCAGTGTCTGGC tgtaaaacgacggccagtTCCAAATCTCTCCAACTTCCA aacagctatgaccatgGAAAGAGCCTGGGATAGCAA tgtaaaacgacggccagtTGTTGTGCAAGCTGATCCTG aacagctatgaccatgAGGCAGCCTGTTTCCAATAA tgtaaaacgacggccagtGGCCATTGAATGAGAGGAAA aacagctatgaccatgCAGATAGCTTTCACCCCCAA tgtaaaacgacggccagtATGTTGCCTTCTGCCATAGG CTACTCCCTGGATCTTCAGCTC aacagctatgaccatgCTTGGTCACAGCACATGGTC tgtaaaacgacggccagtGCGCAGATCACAGTTGAAAA aacagctatgaccatgTGGTCTGTCACTGCAAGGAG tgtaaaacgacggccagtGTTGATCCCACTTACGGGAA aacagctatgaccatgACAGCAGAAAAGGATGGCAC tgtaaaacgacggccagtTTTGGACACTTGGGCACTAA aacagctatgaccatgTCCCACTGAATCCAGTTTCC  139  VPS54 and ALS  147  Supplemental Table II. SNPs in introns and exons of human VPS54. NA: not available.  Downloaded By: [Canadian Research Knowledge Network] At: 18:39 10 April 2009  Number of individuals SNP  Genotype  Control  SALS  FALS  IVS2'123 GT rs13025640  G G/T T  66 21 7  74 17 3  84 10 0  IVS4-267 T A  T T/A A  52 35 7  51 37 6  43 44 4  Y212Y (Ex 7) rs17028275  T T/C C  89 5 0  88 6 0  60 NA 0  M353V (Ex 8)  A A/G G  91 3 0  93 1 0  94 0 0  T360A (Ex 8)  A A/G G  94 0 0  93 1 0  94 0 0  IVS8-40 C T  C C/T T  73 21 0  63 26 4  41 NA 5  IVS11'239 A G  A A/G G  90 2 2  86 8 0  91 3 0  S561C (Ex 12)  C C/G G  93 1 0  94 0 0  94 0 0  IVS12'91 C T  C C/T T  86 8 0  90 4 0  93 1 0  IVS13'52 T C  T T/C C  89 5 0  88 6 0  87 7 0  IVS13-73 C T rs1440002  C C/T T  88 2 4  93 0 1  65 NA 0  IVS13-53 A T  A A/T T  94 0 0  93 1 0  94 0 0  IVS14-8 T G  T T/G G  90 3 0  94 0 0  94 0 0  IVS15'71 T G  T T/G G  93 0 0  90 4 0  90 4 0  IVS15-42 GA rs10496103  G G/A A  94 0 0  93 1 0  92 2 0  IVS16'13'14 TT GG, TTGA  TT TT/GG TT/GA  94 0 0  94 0 0  90 1 3  IVS16'14 T A  T T/A A  91 2 0  91 3 0  94 0 0  IVS16'150 A G  A A/G G  60 25 8  48 37 9  59 35 0  IVS17'67 5bp del TCTTA  TCTTA TCTTA/DEL DEL/DEL  94 0 0  93 1 0  94 0 0  140  148  M. H. Meisler et al.  Supplemental Table II (Continued)  Downloaded By: [Canadian Research Knowledge Network] At: 18:39 10 April 2009  Number of individuals SNP  Genotype  Control  SALS  FALS  IVS17'73 indel CTCAA  CTCCA CTCCA/ DEL DEL/DEL  93 1  94 0  94 0  0  0  0  IVS18'56 T A rs10207847  T T/A A  3 24 66  7 30 57  2 30 60  IVS18'133 ins AA  A Á/AA AA/AA  94 0 0  94 0 0  93 1 0  IVS18'135 ins AC  A Á/AC AC/AC  94 0 0  94 0 0  93 1 0  IVS18-106 TA rs2302809  T T/A A  46 18 4  50 25 7  53 15 23  IVS18-39 T A rs2302807  T T/A A  46 20 4  59 30 1  79 10 2  IVS19'16 T A  T T/A A  94 0 0  93 1 0  93 1 0  IVS19'44 T A  T T/A A  94 0 0  94 0 0  93 1 0  IVS20'9 T A  T T/A A  93 1 0  90 4 0  93 1 0  IVS20-60 GT  G G/T T  52 38 4  52 35 7  77 17 0  IVS20-50 T C rs3815616  T T/C C  51 35 8  61 29 4  83 11 0  IVS20-33 A T  A A/T T  93 1 0  93 1 0  94 0 0  IVS21-169 A G rs17620012  A A/G G  60 27 7  61 29 4  77 16 1  ATGATT (Ex 22)  G G/T T  72 20 2  69 23 2  71 22 1  141  

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.24.1-0067127/manifest

Comment

Related Items