| To whom correspondence should be addressed: Kazuma Tanaka, Division of Molecular Interaction, Institute for Genetic Medicine, Hokkaido University Graduate School of Medicine, N15 W7, Kita-ku, Sapporo 060-0815, Japan. Tel: +81–11–706–5165, Fax: +81–11–706–7821 E-mail: k-tanaka@igm.hokudai.ac.jp Abbreviations: AP, adaptor protein; APLT, aminophospholipid translocase; Arf, ADP-ribosylation factor; CCV, clathrin-coated vesicle; CPY, carboxypeptidase Y; EM, electron microscopy; ER, endoplasmic reticulum; GAP, GTPase-activating protein; GEF, guanine nucleotide exchange factor; GFP, green fluorescent protein; GGA, Golgi-localizing, γ-adaptin ear homology domain, Arf-binding protein; 3HA, three tandem repeats of the influenza virus hemagglutinin epitope; LAT-A, latrunculin A; mRFP1, monomeric red fluorescent protein 1; TGN, trans-Golgi network; ts, temperature-sensitive. |
It is widely accepted that plasma membrane phospholipids are asymmetrically distributed in the two leaflets of the membrane bilayers in eukaryotic cells (Devaux, 1991). In this phospholipid asymmetry, the inner leaflet facing the cytoplasm are enriched in phosphatidylethanolamine and phosphatidylserine, whereas phosphatidylcholine, sphingomyelin, and glycolipids are predominantly found in the outer leaflet. The asymmetric distribution of phospholipids is generated and maintained by ATP-driven lipid transporters or translocases. The P4 subfamily of the P-type ATPases has been implicated in the translocation of aminophospholipids from the external to the cytosolic leaflet (Holthuis and Levine, 2005; Graham, 2004; Pomorski et al., 2004). In the yeast Saccharomyces cerevisiae, five members of this subfamily (Drs2p, Neo1p, Dnf1p, Dnf2p, and Dnf3p) are encoded by the genome (Hua et al., 2002; Catty et al., 1997). Drs2p is localized to the endosomal/trans-Golgi network (TGN) compartments (Saito et al., 2004; Pomorski et al., 2003; Hua et al., 2002; Chen et al., 1999), suggesting that Drs2p regulates the phospholipid asymmetry in these membranes. In fact, Golgi membranes isolated from a temperature-sensitive (ts) drs2 mutant lacking DNF1, DNF2, and DNF3 exhibited defects in the ATP-dependent transport of a fluorescent 7-nitrobenz-2-oxa-1,3-diazol-4-yl (NBD)-labeled analog of phosphatidylserine (Natarajan et al., 2004). Alder-Baerens et al. (2006) also demonstrated that post-Golgi secretory vesicles contained a Drs2p- and Dnf3p-dependent NBD-labeled phospholipid translocase activity and that the asymmetric phosphatidylethanolamine arrangement in these vesicles was disrupted in the drs2Δ dnf3Δ mutant.
We previously isolated CDC50, which encodes a conserved membrane-spanning protein, as a gene required for polarized cell growth (Misu et al., 2003). Cdc50p colocalized with Drs2p at endosomal/TGN membranes, and associated with Drs2p as a putative non-catalytic subunit (Saito et al., 2004). In the absence of Cdc50p, Drs2p was observed in the endoplasmic reticulum (ER) due to a failure of the protein to exit the ER; a lack of Drs2p conversely led to the retention of Cdc50p in the ER. Both the cdc50Δ and drs2Δ mutants exhibited growth defects and depolarization of cortical actin patches at lower temperatures (Saito et al., 2004; Misu et al., 2003). This actin depolarization phenotype may be due to defects in the endocytic recycling pathway, because actin patches seemed to assemble on endocytosed membranes in the cdc50Δ mutant cultured at the low temperature and Cdc50p-depleted erg3Δ (a mutation in the gene that codes for sterol C-5 desaturase Erg3p, which catalyzes a late step in the ergosterol biosynthetic pathway) mutant (Kishimoto et al., 2005, and our unpublished results). Thus, the functions of the Cdc50p-Drs2p complex in vesicle trafficking need to be investigated to understand how Cdc50p-Drs2p regulates the polarized organization of the actin cytoskeleton as well as polarized cell growth.
A mutation in DRS2 was shown to cause synthetic lethality with a null mutation in ARF1, a gene that codes for an ADP-ribosylation factor (Arf) small GTPase (Chen et al., 1999). ARF1 is part of an essential gene family that also includes ARF2, which produces a gene product that is 96% identical to Arf1p at the amino-acid sequence level, and ARF1 produces approximately 90% of total Arf1p/Arf2p (Stearns et al., 1990), suggesting that Arf1p plays a major role. Arf proteins cycle between an inactive GDP-bound form and an active, membrane-associated GTP-bound form. Conversion from the GDP-bound to the GTP-bound form is facilitated by an Arf guanine nucleotide exchange factor (Arf GEF), whereas GTP hydrolysis is induced by an Arf GTPase-activating protein (Arf GAP) (reviewed in Donaldson and Jackson, 2000). Gea1p, Gea2p, Sec7p, and Syt1p have been identified as Arf GEFs in S. cerevisiae (reviewed in Jackson and Casanova, 2000). Gea1p and Gea2p play important roles in the structure and function of the Golgi apparatus, and have overlapping functions in the Golgi-to-ER retrograde transport pathway (Peyroche et al., 2001; Spang et al., 2001). Sec7p is a Golgi-localized protein that is involved in the ER-Golgi and intra-Golgi transport pathways (Franzusoff et al., 1991, 1992). On the other hand, Gcs1p, Glo3p, Age1p, and Age2p have been identified as Arf GAPs in S. cerevisiae (Zhang et al., 2003; Poon et al., 2001; Dogic et al., 1999; Poon et al., 1996). Gcs1p functions redundantly with Glo3p in the Golgi-to-ER retrograde transport pathway (Poon et al., 1999), and with Age2p in transport from the TGN (Poon et al., 2001). Gcs1p is also involved in the organization of the actin cytoskeleton (Blader et al., 1999) and mitochondrial morphology (Huang et al., 2002).
Arf1p has been implicated in the formation of COPI-coated vesicles in the Golgi-to-ER retrograde transport pathway (Poon et al., 1999; Gaynor et al., 1998), and clathrin-coated vesicles (CCVs) in transport from the TGN (Chen and Graham, 1998). The drs2Δ mutation also causes synthetic lethality with a mutation in CHC1, which codes for the clathrin heavy chain (Chen et al., 1999), suggesting the involvement of Cdc50p-Drs2p in clathrin-associated vesicle transport. Indeed, the drs2Δ mutant exhibits TGN defects comparable with those exhibited by strains with clathrin mutations, and is defective in the formation of CCVs (Gall et al., 2002; Chen et al., 1999).
Clathrin adaptors may regulate the Arf-dependent formation of CCVs. These proteins link clathrin to membrane cargo, lipids, and accessory proteins that regulate coat assembly and disassembly. There are two known types of clathrin adaptors; the heteromeric adaptor protein (AP) complexes composed of four subunits, and the monomeric GGA (Golgi-localizing, γ-adaptin ear homology domain, Arf-binding protein) proteins. In S. cerevisiae, there are three AP complexes, AP-1, AP-2R, and AP-3 (Yeung et al., 1999), and two GGAs, Gga1p and Gga2p (Costaguta et al., 2001; Dell’Angelica et al., 2000; Hirst et al., 2000). Both AP-1 and Gga1p/Gga2p physically interact with clathrin (Costaguta et al., 2001; Yeung and Payne, 2001; Yeung et al., 1999). It has been suggested that Gga1p and Gga2p are involved in the formation of CCVs in the TGN-to-late endosome transport pathway (Costaguta et al., 2001; Black and Pelham, 2000; Dell’Angelica et al., 2000; Hirst et al., 2000), although the interaction with Arf1p is not sufficient for the Golgi-localization and function of Gga1p/Gga2p (Boman et al., 2002). On the other hand, Arf1p and AP-1 have been implicated in the retrieval of the Chitin synthase III subunit Chs3p from early endosomes to the TGN (Valdivia et al., 2002).
In this study, in order to identify vesicle transport routes that involve the Cdc50p-Drs2p putative aminophospholipid translocase (APLT) and the Arf signaling pathway, we searched for ARF-related genes that genetically interact with CDC50, resulting in the identification of GCS1. The Cdc50p-depleted gcs1Δ mutant exhibited severe defects in the early endosome-to-TGN transport pathway, but not in other vesicle transport pathways. Similar phenotypes were found in the gga mutants depleted of Cdc50p, and in a mutant carrying the gcs1Δ mutation in combination with a mutation in AP-1. Our results raise the possibility that changes in phospholipid asymmetry are involved in the Arf-dependent formation of TGN-targeted CCVs from early endosomes.
Unless otherwise specified, strains were grown in YPDA rich medium (1% yeast extract [Difco, Detroit, MI], 2% bacto-peptone [Difco], 2% glucose, and 0.01% adenine). Strains carrying plasmids were selected in synthetic medium (SD) containing the required nutritional supplements (Rose et al., 1990). When appropriate, 0.5% casamino acids (Difco) were added to SD medium without uracil (SDA-Ura). For induction of the GAL1 promoter, 3% galactose and 0.2% sucrose were used as carbon sources instead of glucose. Standard genetic manipulations of yeast were performed as described previously (Guthrie and Fink, 1991). Yeast transformations were performed using the lithium acetate method (Gietz and Woods, 2002; Elble, 1992). DH5α and XL1-Blue Escherichia coli strains were used for the construction and amplification of plasmids.
Yeast strains used in this study are listed in Table I. The yeast YEF473 background strains carrying complete gene deletions (CDC50, ARF1, ARF2, ARF3, GCS1, GLO3, GGA1, GGA2, GEA1, and GEA2); GAL1 promoter-inducible CDC50 tagged with three tandem repeats of the influenza virus hemagglutinin epitope (3HA); enhanced green fluorescent protein (EGFP)-tagged APL4, VPS10, and KEX2; monomeric red fluorescent protein 1 (mRFP1)-tagged SEC7; and the yeast S288C background strains carrying complete gene deletions (APL2, GCS1, GGA1, and GGA2) were constructed by polymerase chain reaction (PCR)-based procedures as described previously (Longtine et al., 1998). All constructs produced by PCR-based procedures were verified by colony-PCR amplification to confirm that the replacements had occurred at the expected loci. The yeast S288C background strains carrying complete gene deletions (arl1Δ, arl3Δ, age1Δ, age2Δ, syt1Δ, apl4Δ, apm1Δ, aps1Δ, apl5Δ, apl6Δ, apm3Δ, and aps3Δ were gifts from C. Boone (University of Toronto).
The plasmid pRS313-GCS1 was constructed by the gap-repair method. The 5'-upstream region from –820 to –353 and the 3'-downstream region from +1369 to +1893 of the GCS1 gene were amplified by the PCR using the oligonucleotide primers GCS1up-5 (5'-ATGGATCCCTACGTGAACCCTGGTGTCCTC-3') and GCS1up-3 (5'-TAGATATCTATGTGGGCCAGCAGGTACAGG-3'), and GCS1down-5 (5'-ATGATATCAGACCTGGGACAATCGTTATCC-3') and GCS1down-3 (5'-TACTCGAGCCGATAATGGCACCGTCTTTTG-3'), respectively. The 5'-upstream and the 3'-downstream PCR products were digested with restriction enzymes, and subcloned into the BamHI-EcoRV and EcoRV-XhoI gaps of pRS313, respectively. The resulting plasmid was digested with EcoRV and introduced into YEF473, and the gap-repaired plasmid was isolated from His+ colonies. The gcs1-R54A plasmid was generated with a QuikChange Site-Directed Mutagenesis Kit (Stratagene, La Jolla, CA) using pRS313-GCS1 as a template. The entire open reading frame of GCS1 was sequenced to confirm that only the desired mutation was introduced. The plasmids used in this study are listed in Table II. Schemes detailing the construction of all the plasmids are available on request. The appropriate references for GFP-Snc1p, GFP-Rer1p, Chs3p-GFP and mRFP1-Snc1p are listed in the right column of Table II. GFP-Tlg1p was functional, because YEplac181-GFP-TLG1 rescued the lethality of tlg1Δ mutant in the YEF473 strain background (our unpublished results).
Cells were observed with a Nikon ECLIPSE E800 microscope equipped with an HB-10103AF super high-pressure mercury lamp and a 1.4 NA 100x Plan Apo oil immersion objective lens with the appropriate fluorescence filter sets or differential interference contrast optics (Nikon Instec, Tokyo, Japan). Images were acquired with a digital cooled CCD camera (C4742-95-12NR; Hamamatsu Photonics K.K., Hamamatsu, Japan) using the AQUACOSMOS software package (Hamamatsu Photonics). Observations are based on the examination of at least 100 cells.
To visualize GFP- or mRFP1-tagged proteins in living cells, cells were grown in YPDA, harvested, and resuspended in SD medium. Cells were mounted on a glass microslide, followed by immediate observation using a GFP bandpass or G-2A filter set. When the effect of latrunculin A (LAT-A, Wako Pure Chemicals, Osaka, Japan) was examined, cells were treated with 100 μM LAT-A by the addition of a 20 mM stock solution in dimethyl sulfoxide (DMSO) to the medium as described previously (Ayscough et al., 1997).
Staining with the lypophilic stylyl dye FM4-64 (N-(3-triethylammoniumpropyl)-4-(p-diethylaminophenylhexatrienyl) pyridinium dibromide) was performed as described previously with minor modifications (Misu et al., 2003). Briefly, cells were grown in YPDA at 30°C for 6.5 h. Four OD600 units of cells were collected by centrifugation, suspended in 100 μl of YPDA, and labeled in 32 μM FM4-64 (Molecular Probes, Eugene, OR) for 30 min on ice. Cells were harvested by centrifugation, resuspended in 200 μl of fresh YPDA, and chased at 30°C for the indicated time periods. After the chase, cells were washed with SD medium, followed by immediate microscopic observation using a G-2A filter set. The vacuole lumen was visualized using CellTracker Blue CMAC (Molecular Probes) according to the manufacturer’s protocol.
Vacuolar sorting of carboxypeptidase Y (CPY) was examined by pulse-chase analysis and immunoprecipitation experiments as described previously with minor modifications (Misu et al., 2003; Rothblatt and Schekman, 1989). Briefly, cells were grown in low-sulfate SD medium at 30°C for 7.5 h, washed, resuspended in sulfate-free SD medium, and grown at 30°C for 30 min. Cells were labeled with 50 μCi of Trans35S-label (ICN Radiochemicals, Irvine, CA) for 10 min before they were chased for the indicated time periods. CPY was immunoprecipitated with rabbit antibodies against CPY (a gift from Y. Ohsumi, National Institute for Basic Biology, Okazaki, Japan), resolved by SDS-PAGE, and visualized with a phosphorimager system (Fuji Photo Film, Tokyo, Japan).
Secretion of invertase was also examined by pulse-chase analysis and immunoprecipitation experiments as described previously with minor modifications (Rothblatt and Schekman, 1989). Briefly, cells were grown in low-sulfate SD medium containing 5% glucose at 30°C for 7.5 h. Cells were then converted to spheroplasts with zymolyase, transferred to sulfate-free SD medium containing 0.1% glucose, and further grown at 30°C for 30 min to induce invertase expression. Cells were labeled with 50 μCi of Trans35S-label at 30°C for 4 min before they were chased. To terminate the chase, NaN3 was added to a final concentration of 10 mM. Cell suspensions were separated into spheroplasts and media by centrifugation to obtain intracellular and extracellular fractions, respectively. Intracellular and extracellular fractions were immunoprecipitated with anti-invertase antisera (a gift from A. Nakano, The University of Tokyo, Japan). The immunoprecipitated invertase was resolved by SDS-PAGE, and visualized using the Phosphorimager system.
Ultrastructural observation of cells by conventional EM was performed using the glutaraldehyde-osmium fixation technique as described previously (Wright, 2000; Banta et al., 1988). Briefly, cells were fixed in 2% glutaraldehyde, treated for 10 min in 1% sodium metaperiodate, and postfixed in 2% reduced osmium. Cells were then embedded in Q651 resin (Nissin EM, Tokyo, Japan). Thin sections (50–60 nm) were cut on an Ultracut microtome (Leica, Wetzlar, Germany) equipped with a Sumiknife (Sumitomo Electric Industries, Osaka, Japan), stained with 3% uranyl acetate and Reynold's lead citrate, and viewed using an H-7100 electron microscope (Hitachi, Tokyo, Japan) at 75 kV. Immuno-EM was performed using the aldehyde fixation/metaperiodate permeabilization method as described previously (Mulholland and Botstein, 2002), except glutaraldehyde was not included in the fixative. Cells were embedded in LR White resin (medium grade; London Resin Company, Berkshire, UK), and sectioned as described above. Mouse anti-HA antibodies (HA.11; BAbCO, Richmond, CA) were used as primary antibodies at a 1:500 dilution. Ten-nm gold-conjugated anti-mouse IgG antibodies (BBInternational, Cardiff, UK) were preadsorbed with fixed wild-type cells, and used as secondary antibodies at a 1:100 dilution. Samples were poststained with uranyl acetate and viewed as described above.
In accord with the previous study that identified DRS2 in a genetic screen for mutations that are synthetically lethal with the arf1Δ mutation (Chen et al., 1999), the arf1Δ mutant displayed a growth defect when Cdc50p was depleted by using the PGAL1-3HA-CDC50 allele, which employs the glucose-repressible GAL1 promoter to control the expression of Cdc50p (Fig. 1A). This synthetic genetic interaction with arf1Δ was specific, and was not observed with mutations in other genes encoding Arf and Arf-like proteins, including ARF2, ARF3, ARL1, and ARL3 (Fig. 2A). Several arf1-ts arf2Δ mutants were isolated and some of these mutations appeared to show synthetic lethality with drs2Δ (Yahara et al., 2001). Interestingly, the arf1-18 arf2Δ mutant, which did not exhibit impaired Golgi-to-ER transport, exocytosis, or endocytic transport to vacuoles at the restrictive temperature (Yahara et al., 2001), displayed a growth defect when Cdc50p was depleted (Fig. 1A). These results suggest that Arf and Cdc50p-Drs2p functionally overlap in a vesicular transport pathway that is distinct from those described above.
 View Details | Fig. 1. The arf1 mutants exhibit defects in endocytic recycling of GFP-Snc1p. (A) Depletion of Cdc50p causes the growth defect in the arf1Δ and arf1-18 arf2Δ mutants. Wild-type (YKT38; a), PGAL1-3HA-CDC50 (YKT934; b), arf1Δ (YKT1275; c), arf1-18 arf2Δ (NYY18-2; d), PGAL1-3HA-CDC50 arf1Δ (YKT1276; e), and PGAL1-3HA-CDC50 arf1-18 arf2Δ (YKT1277; f) strains were streaked onto a plate containing galactose (YPGA) or glucose (YPDA) as a carbon source, followed by incubation at 30°C for 2 d. (B) Localization of GFP-Snc1p in the arf1 mutants. Wild-type (YKT38; WT), arf1-18 arf2Δ (NYY18-2), and arf1Δ (YKT1275) strains carrying pRS416-GFP-SNC1 or pRS416-GFP-SNC1 (pm) were grown to mid-log phase in YPDA medium at 30°C, incubated at 30°C or 37°C for 1 h, and observed immediately by fluorescence microscopy. An arrowhead indicates the ring structure that contain GFP-Snc1p. (C) Localization of Sec7p-mRFP1 in the arf1 mutants. Wild-type (YKT905; WT), arf1-18 arf2Δ (YKT1278), and arf1Δ (YKT1279) strains expressing Sec7p-mRFP1 were grown to mid-log phase in YPDA medium at 30°C, incubated at 30°C or 37°C for 1 h, and observed immediately by fluorescence microscopy. Bars, 5 μm. |
 View Details | Fig. 2. CDC50 exhibits a specific genetic interaction with GCS1 among Arf GAP genes. (A) Genetic interactions between CDC50 and genes encoding Arf, Arf-like proteins, or regulators of Arf proteins. The cdc50Δ mutant was crossed with the indicated mutants to generate diploids. Diploid cells were sporulated, and synthetic lethality was examined using tetrad analysis at 30°C. (B) Tetrad analyses of progeny derived from crossing the cdc50Δ mutant (YKT912) with the indicated Arf GAP mutants. Arrows indicate the positions of double mutant segregants. (C) Construction of a conditional cdc50Δ gcs1Δ mutant. Wild-type (YKT38; WT), PGAL1-3HA-CDC50 (YKT934), gcs1Δ (YKT1282), and PGAL1-3HA-CDC50 gcs1Δ (YKT1286) strains were streaked onto a YPGA or YPDA plate, followed by incubation at 30°C for 2 d. (D) Requirement of the Arf GAP activity of Gcs1p for growth of the PGAL1-3HA-CDC50 gcs1Δ mutant. pRS313 (vector), pRS313-GCS1 (pKT1613), or pRS313-gcs1-R54A (pKT1614) was introduced into the PGAL1-3HA-CDC50 gcs1Δ (YKT1286) mutant. Cells were streaked onto a YPDA plate, followed by incubation at 30°C for 2 d. |
We previously demonstrated the involvement of Cdc50p-Drs2p in endocytic recycling of GFP-Snc1p (Saito et al., 2004). Snc1p is an exocytic v-SNARE that normally cycles from the plasma membrane through the early endosome to the TGN and back to the plasma membrane (Lewis et al., 2000). We examined the localization of GFP-Snc1p in the arf1Δ and arf1-18 arf2Δ mutants. At the permissive temperature (30°C), GFP-Snc1p was primarily localized to the plasma membrane, and was concentrated within polarized growth sites such as buds or cell division sites in the arf1-18 arf2Δ mutant, as well as in the wild-type strain (Fig. 1B). At the restrictive temperature (37°C), however, GFP-Snc1p was diffusely distributed in the cytoplasm or accumulated as small punctate structures in the arf1-18 arf2Δ mutant. In the arf1Δ mutant, 58% (n=103) and 70% of the cells (n=102) accumulated GFP-Snc1p in intracellular structures at 30°C and 37°C, respectively (Fig. 1B). In addition, GFP-Snc1p was also localized to ring structures in 24% (n=103) and 21% of the cells (n=103) at 30°C and 37°C, respectively (Fig. 1B, arrowhead). In contrast, GFP-Snc1p (pm), an endocytosis-defective mutant form of Snc1p (Lewis et al., 2000), was exclusively localized to the plasma membrane in the arf1-18 arf2Δ mutant at the restrictive temperature and in the arf1Δ mutant, as well as in the wild-type strain (Fig. 1B). These data indicate that the intracellular accumulation of Snc1p-containing structures in the arf1 mutants was not caused by defects in the exocytotic pathway from the TGN to the plasma membrane, but was dependent on endocytosis. The TGN marker Sec7p-mRFP1 (Robinson et al., 2006) was localized to internal punctate structures in the arf1-18 arf2Δ and arf1Δ mutants as it was in wild-type cells, suggesting that the Snc1p-containing structures were not derived from the TGN (Fig. 1C). These results suggest that Arf1p and Arf2p are involved in the regulation of the endocytic recycling pathway.
Because Arf1p and Arf2p seem to be involved in multiple vesicle transport pathways, dissection of the defects in the arf1 cdc50Δ mutant may be difficult. Therefore, we investigated the genetic interaction between CDC50 and genes encoding regulators of Arf proteins (Arf GAPs and Arf GEFs). The cdc50Δ mutant was crossed with a null mutant or a ts mutant of the Arf regulators and the growth of double or triple mutants was examined using tetrad analysis (Fig. 2A). Single mutations in the Arf GEF genes (sec7-1, syt1Δ, gea1Δ, and gea2Δ) or the gea1-ts gea2Δ mutations (gea1-4 gea2Δ, gea1-6 gea2Δ, and gea1-19 gea2Δ) did not affect the growth of the cdc50Δ mutant at 30°C. Interestingly, among the mutations in the genes coding for Arf GAPs (gcs1Δ, glo3Δ, age1Δ, and age2Δ), the gcs1Δ mutation displayed synthetic lethality with the cdc50Δ mutation (Fig. 2B). The drs2Δ mutation exhibited the same genetic interaction pattern with the mutations in the genes coding for Arf GAPs (our unpublished results). Consistent with our results, Robinson et al. (2006) recently identified the synthetic lethal interaction between drs2Δ and gcs1Δ by synthetic genetic analysis. The Arf GAP activity-defective gcs1-R54A mutation (Yanagisawa et al., 2002) failed to rescue the Cdc50p-depleted gcs1Δ mutant from lethality (Fig. 2D), suggesting that the Gcs1p-mediated regulation of Arf1p was required for growth of the Cdc50p-depleted wild-type cells. We therefore decided to examine the defects in the vesicular transport pathways in the Cdc50p-depleted gcs1Δ mutant.
To investigate essential functions governed by Gcs1p and Cdc50p-Drs2p, we constructed the PGAL1-3HA-CDC50 gcs1Δ mutant. As shown in Fig. 2C, the PGAL1-3HA-CDC50 gcs1Δ mutant grew normally in the galactose-containing medium, but not in the glucose-containing medium. After 3-h incubation in the glucose-containing medium, the expression of 3HA-Cdc50p was not detected in the immunoblot analysis either by the anti-3HA or -Cdc50p antibodies (our unpublished results). Because complete growth arrest required incubation for at least 6 h in the glucose-containing medium (our unpublished results), the phenotypes of the PGAL1-3HA-CDC50 gcs1Δ mutant (the PGAL1-3HA-CDC50 allele is hereafter referred to as Cdc50p-depleted) were analyzed after incubation for more than 6 h at 30°C.
Endocytic internalization and transport to vacuoles was examined in the Cdc50p-depleted gcs1Δ mutant at 30°C with the fluorescent endocytic marker FM4-64. Cells were depleted of Cdc50p for 6.5 h, labeled with FM4-64, and chased. As in the wild-type cells, FM4-64 was internalized and delivered to vacuoles after a 1-h chase in the Cdc50p-depleted gcs1Δ mutant (Fig. 3A). These results also revealed that the vacuoles in the Cdc50p-depleted gcs1Δ mutant were somewhat fragmented. The endocytic marker Lucifer Yellow and the α-factor receptor Ste2p-GFP were similarly delivered to vacuoles in the Cdc50p-depleted gcs1Δ mutant (our unpublished results). These results indicate that endocytosis was not severely impaired in the Cdc50p-depleted gcs1Δ mutant.
 View Details | Fig. 3. Intracellular vesicle transport in the Cdc50p-depleted gcs1Δ mutant. Transport in various vesicle transport pathways was examined in the PGAL1-3HA-CDC50 gcs1Δ and control strains described in Fig. 2C. (A) Internalization and transport to the vacuole of FM4-64 in the Cdc50p-depleted gcs1Δ mutant. Cells were grown in YPDA at 30°C for 6.5 h, labeled in 32 μM FM4-64 for 30 min at 0°C, and then chased in fresh medium at 30°C for 1 h. (B) CPY processing in the Cdc50p-depleted gcs1Δ mutant. Cells were depleted of Cdc50p for 8 h at 30°C, labeled with Trans35S-label for 10 min, and chased for 0, 15, or 30 min. CPY was immunoprecipitated, resolved by SDS-PAGE, and detected by autoradiography. (C) Invertase secretion in the Cdc50p-depleted gcs1Δ mutant. After incubation in glucose-containing medium for 8 h at 30°C, wild-type (a), PGAL1-3HA-CDC50 (b), gcs1Δ (c) and PGAL1-3HA-CDC50 gcs1Δ (d) strains were labeled with Trans35S-label for 4 min and chased for 0 or 60 min. Secreted and intracellular invertase were separated into external and internal fractions, respectively. Invertase was recovered by immunoprecipitation, and visualized by SDS-PAGE and autoradiography. The band indicated by an arrowhead is non-specific. (D) Localization of GFP-Rer1p in the Cdc50p-depleted gcs1Δ mutant. The strains used in Fig. 2C were transformed with a single-copy plasmid containing GFP-RER1 (pSKY5RER1-0). Cells were grown in YPDA medium at 30°C for 8 h, and observed by fluorescence microscopy. Bars, 5 μm. |
The vacuolar protein sorting pathway was assessed in the Cdc50p-depleted gcs1Δ mutant by monitoring the maturation of the soluble vacuolar protein CPY. Pulse-chase experiments were performed after depletion of Cdc50p for 8 h in glucose-containing medium. During transport of CPY from the ER through the TGN to vacuoles, CPY undergoes a series of characteristic modifications. CPY is found as a 67-kDa precursor species (P1) and a fully glycosylated 69-kDa precursor form (P2) in the ER and the Golgi, respectively. It was previously reported that the kinetics of the maturation of CPY were delayed in the drs2Δ, cdc50Δ, and dnf1Δ drs2Δ mutants at lower temperatures (Misu et al., 2003; Hua et al., 2002; Chen et al., 1999). The conversion of CPY from P1 through P2 to the mature form (m), however, was completed in 15 min at 30°C in the Cdc50p-depleted wild-type cells, as well as in the wild-type strain and the gcs1Δ mutant (Fig. 3B). Although the Cdc50p-depleted gcs1Δ mutant accumulated a small amount of the P2 form, most of the CPY was converted to the mature form during the 30-min chase. Because defects in vacuolar protein sorting lead to secretion of CPY, we also examined CPY sorting by colony immunoblotting. Although CPY was secreted from the vps1Δ cells, it was not secreted from the Cdc50p-depleted gcs1Δ, Cdc50p-depleted wild-type, and gcs1Δ cells (our unpublished results). These results suggest that CPY transport was not severely impaired in the Cdc50p-depleted gcs1Δ mutant.
To investigate the secretory pathway in the Cdc50p-depleted gcs1Δ mutant, we examined the processing and secretion of the periplasmic enzyme invertase in pulse-chase experiments using PGAL1-3HA-CDC50 gcs1Δ cells that were depleted of Cdc50p for 8 h. After a 60-min chase, invertase was processed to a highly glycosylated form, which was predominantly found in the extracellular fraction in the experiments with the Cdc50p-depleted gcs1Δ strain, as well as in the experiments with the wild-type, Cdc50p-depleted wild-type, and gcs1Δ cells (Fig. 3C). These data suggest that the secretory pathway was not severely impaired in the Cdc50p-depleted gcs1Δ mutant.
To investigate the Golgi-to-ER retrograde transport pathway, we observed the localization of GFP-Rer1p in the Cdc50p-depleted gcs1Δ mutant. Rer1p is a membrane protein found in the cis-Golgi that serves as a retrieval receptor for ER-resident membrane proteins (Sato et al., 1997, 2001). It has been shown that when the Golgi-to-ER retrograde transport is blocked by impairment in the COPI-dependent pathway, the majority of GFP-Rer1p is transported to vacuoles (Sato et al., 2001). GFP-Rer1p was localized to internal punctate structures in the Cdc50p-depleted gcs1Δ mutant and control strains in a manner that resembled Golgi localization (Fig. 3D). These data suggest that the Golgi-to-ER retrograde transport was not impaired in the Cdc50p-depleted gcs1Δ mutant.
Taken together, our results suggest that the Cdc50p-depleted gcs1Δ mutant was defective in a more selective vesicle transport pathway. This is in contrast to previous observations that the Arf GAPs Gcs1p and Age2p provide essential functions for transport from the TGN (Poon et al., 2001), and that Gcs1p and Glo3p are required for retrograde transport from the Golgi to the ER (Poon et al., 1999). In the Cdc50p-depleted gcs1Δ mutant, it seems that Age2p and Glo3p compensated for these essential functions.
To investigate the retrieval pathway from late endosomes to the TGN in the Cdc50p-depleted gcs1Δ mutant, we examined the localization of two TGN resident membrane proteins, Kex2p and Vps10p, whose TGN localization is dependent on the late endosome-to-TGN retrieval pathway. Kex2p is required for maturation of the precursors of secreted peptides and proteins, including the α-mating factor (Fuller et al., 1988; Julius et al., 1984). Vps10p is a CPY sorting receptor, which dissociates from its cargo at the late endosome (Cooper and Stevens, 1996; Marcusson et al., 1994). Kex2p and Vps10p are recycled back from late endosomes to the TGN in a retromer-dependent manner (Seaman et al., 1998), whereas Kex2p is also transported to early endosomes and retrieved to the TGN by an unknown mechanism (Lewis et al., 2000).
Vps10p-GFP localized to punctate structures in the wild-type strain (Fig. 4A), whereas deletion of VPS26, a component of retromer, caused mislocalization of Vps10p-GFP to vacuoles as assessed by staining with CMAC (our unpublished results), suggesting that Vps10p-GFP is normally retrieved from late endosomes to the TGN by the retromer-dependent mechanism. The localization of Vps10p-GFP to internal punctate structures was normal in the Cdc50p-depleted gcs1Δ mutant (Fig. 4A). Kex2p-GFP localizes to punctate structures that appeared to be endosomes or TGN compartments in the wild-type strain (Chen et al., 2005), and it was not localized to vacuoles as assessed by staining with CellTracker Blue CMAC (Fig. 4B). In the Cdc50p-depleted wild-type and the gcs1Δ cells, Kex2p-GFP was also localized to punctate structures, although a fraction was mislocalized to vacuoles (Fig. 4A). In contrast, in the Cdc50p-depleted gcs1Δ mutant, Kex2p-GFP was almost mislocalized to the vacuolar compartments (Fig. 4, A and B). Proteins such as Kex2p and Vps10p that cycle via endosomes are mislocalized to the vacuole if their cycling is impaired (Conibear and Stevens, 2000; Cooper and Stevens, 1996; Wilcox et al., 1992). Our results imply that the Cdc50p-depleted gcs1Δ mutant was defective in the early endosome-to-TGN transport pathway, but not in the late endosome-to-TGN transport pathway.
 View Details | Fig. 4. The Cdc50p-depleted gcs1Δ mutant exhibits mislocalization of Kex2p-GFP, but not Vps10p-GFP. (A) Localization of Vps10p-GFP and Kex2p-GFP in the Cdc50p-depleted gcs1Δ mutant. A Vps10p-GFP- or Kex2p-GFP-expressing variants of the PGAL1-3HA-CDC50 gcs1Δ mutant and control strains described in Fig. 2C were constructed. Cells were grown in YPDA medium at 30°C for 8 h, and observed by fluorescence microscopy. The Vps10p-GFP- and Kex2p-GFP-expressing strains were YKT957 and YKT903 (wild-type; WT), YKT1287 and YKT1290 (PGAL1-3HA-CDC50; Cdc50p-depleted), YKT1288 and YKT1291 (gcs1Δ), and YKT1289 and YKT1292 (PGAL1-3HA-CDC50 gcs1Δ; Cdc50p-depleted gcs1Δ), respectively. (B) Kex2p-GFP is mislocalized to the vacuole in the Cdc50p-depleted gcs1Δ mutant. Wild-type (YKT903; WT) and PGAL1-3HA-CDC50 gcs1Δ (YKT1292; Cdc50p-depleted gcs1Δ) cells expressing Kex2p-GFP were grown in YPDA medium at 30°C for 7.5 h, labeled with CellTracker Blue CMAC at 30°C for 30 min, and observed by fluorescence microscopy. Bars, 5 μm. |
Defective retrieval of Kex2p from early endosomes in the Cdc50p-depleted gcs1Δ mutant is consistent with the results that arf1 mutants are defective in endocytic recycling of GFP-Snc1p (Fig. 1B). Thus, we investigated defects in the early endosome-to-TGN transport pathway in the Cdc50p-depleted gcs1Δ mutant. Robinson et al. (2006) reported that the gcs1Δ mutant accumulated GFP-Snc1p in intracellular compartments. In our gcs1Δ mutant derived from the YEF473 strain, such mislocalization of GFP-Snc1p was observed at 18°C (our unpublished results), whereas at 30°C GFP-Snc1p was primarily localized to the plasma membrane at polarized sites, and some of the protein were localized to internal punctate structures that appeared to be early endosomes or TGN compartments (Fig. 5A), as in the wild-type strain (Lewis et al., 2000). In 80% of the PGAL1-3HA-CDC50 mutant cells depleted of Cdc50p for 8 h at 30°C, the localization pattern of GFP-Snc1p was indistinguishable from that in wild-type cells, whereas in the remaining 20% of the cells, GFP-Snc1p was not localized to the plasma membrane and instead accumulated in intracellular structures (n=113) (Fig. 5A and our unpublished results). In 99% of the Cdc50p-depleted gcs1Δ mutant cells (n=104), however, GFP-Snc1p was not observed on the plasma membrane, and instead accumulated in aberrant membrane structures (Fig. 5A). These GFP-Snc1p-positive structures in the Cdc50p-depleted gcs1Δ mutant were distinct from vacuoles that were visualized by staining with CMAC (our unpublished results). The accumulation of GFP-Snc1p in aberrant membrane structures in the Cdc50p-depleted gcs1Δ mutant was observed even after 3-h incubation to deplete Cdc50p, suggesting that mislocalization of GFP-Snc1p in this mutant reflected primary defects caused by depletion of Cdc50p in the gcs1Δ mutant (our unpublished results).
 View Details | Fig. 5. The Cdc50p-depleted gcs1Δ mutant intracellularly accumulates GFP-Snc1p. (A) Localization of GFP-Snc1p in the Cdc50p-depleted gcs1Δ mutant. The strains used in Fig. 2C were transformed with pRS416-GFP-SNC1. Cells were grown in YPDA medium at 30°C for 8 h, and observed by fluorescence microscopy. (B) Effect of the LAT-A treatment on the localization of GFP-Snc1p in the Cdc50p-depleted gcs1Δ mutant. Wild-type and PGAL1-3HA-CDC50 gcs1Δ cells carrying pRS416-GFP-SNC1 that were grown in YPDA medium at 30°C for 7 h, followed by an additional 1-h culture in the presence (LAT-A) or absence (DMSO) of 100 μM LAT-A, were observed by fluorescence microscopy. Note that a fraction of GFP-Snc1p was localized to the plasma membrane when the Cdc50p-depleted gcs1Δ mutant was treated with LAT-A (arrowhead). Bars, 5 μm. |
The localization of GFP-Snc1p to the plasma membrane was somewhat restored, when the Cdc50p-depleted gcs1Δ mutant cells were treated with LAT-A for 1 h, which sequesters actin monomers and thereby inhibits endocytosis, after 7-h incubation to deplete Cdc50p (Fig. 5B, arrowhead). These results suggest that GFP-Snc1p is delivered to the plasma membrane and then is endocytosed before accumulating in the intracellular structures in the Cdc50p-depleted gcs1Δ mutant. In the Cdc50p-depleted wild-type and the gcs1Δ cells, GFP-Snc1p was exclusively localized to the plasma membrane after the treatment with LAT-A for 1 h as in the wild-type strain (our unpublished results). Thus, endocytic recycling of GFP-Snc1p is slowed, but not blocked in the Cdc50p-depleted wild-type cells. In contrast, the intracellular GFP-Snc1p-containing structures were still observed in the Cdc50p-depleted gcs1Δ mutant, suggesting the severe impairment of endocytic recycling in this mutant.
Chs3p, a subunit of the cell wall biosynthetic enzyme chitin synthase III, is localized to the plasma membrane at mother-bud junctions and in punctate intracellular structures (Valdivia et al., 2002; Santos and Snyder, 1997; Chuang and Schekman, 1996; Ziman et al., 1996). Similar to Snc1p, this protein is thought to be transported through the endocytic recycling pathway (Lewis et al., 2000; Holthuis et al., 1998b). Thus, we observed the localization of Chs3p-GFP in Cdc50p-depleted gcs1Δ mutant cells that simultaneously expressed mRFP1-Snc1p (Robinson et al. 2006). In the Cdc50p-depleted wild-type, the gcs1Δ, and the wild-type cells, Chs3p-GFP was primarily localized to intracellular punctate structures that presumably corresponding to early endosomes or TGN compartments, and some of these structures colocalized with mRFP1-Snc1p (Fig. 6A). Only 3% of the Cdc50p-depleted wild-type cells (n=101) and 1% of the gcs1Δ cells (n=100) exhibited the accumulation of Chs3p-GFP in aberrant membrane structures. In contrast, in 96% of the Cdc50p-depleted gcs1Δ cells (n=103), Chs3p-GFP accumulated in aberrant membrane structures, which largely colocalized with mRFP1-Snc1p.
 View Details | Fig. 6. Chs3p-GFP and GFP-Tlg1p accumulate in the Snc1p-containing endosomal structures in the Cdc50p-depleted gcs1Δ mutant. The strains used in Fig. 2C were transformed with plasmids or reconstructed to express two different GFP- or mRFP1-fused proteins. Cells were grown in YPDA medium at 30°C for 8 h, and observed by fluorescence microscopy. Obtained images were merged to demonstrate the coincidence of the two signals. (A) Colocalization of Chs3p-GFP and mRFP1-Snc1p in the Cdc50p-depleted gcs1Δ mutant. The strains were cotransformed with pRS313-CHS3-GFP, pRS315-CHS7, and pRS416-mRFP1-SNC1. pRS315-CHS7 was used to avoid artifactual accumulation of Chs3p in the ER (Valdivia et al., 2002). (B) Colocalization of GFP-Tlg1p and mRFP1-Snc1p in the Cdc50p-depleted gcs1Δ mutant. The strains were cotransformed with YEplac181-GFP-TLG1 and pRS416-mRFP1-SNC1. (C) Colocalization of GFP-Tlg1p and Sec7p-mRFP1 in the Cdc50p-depleted gcs1Δ mutant. YEplac181-GFP-TLG1 was introduced into Sec7p-mRFP1-expressing strains: wild-type (YKT905; WT), PGAL1-3HA-CDC50 (YKT1293; Cdc50p-depleted), gcs1Δ (YKT1294), and PGAL1-3HA-CDC50 gcs1Δ (YKT1295; Cdc50p-depleted gcs1Δ). Bars, 5 μm. |
We also observed the localization of Tlg1p, a member of the syntaxin family of t-SNAREs that is recycled between early endosomes and the TGN (Lewis et al., 2000; Holthuis et al., 1998a), in Cdc50p-depleted gcs1Δ mutant cells that simultaneously expressed mRFP1-Snc1p. Similar to Chs3p-GFP, in 99% of the Cdc50p-depleted wild-type cells (n=115) and 99% of the gcs1Δ single mutant (n=100), GFP-Tlg1p was localized to intracellular punctate structures as in the wild-type strain (Fig. 6B). In 99% of the Cdc50p-depleted gcs1Δ mutant cells (n=103), however, aberrant membrane structures containing both GFP-Tlg1p and mRFP1-Snc1p were observed. These results suggest that Chs3p-GFP, GFP-Tlg1p, and mRFP1-Snc1p accumulated in the same membrane structures in the Cdc50p-depleted gcs1Δ mutant.
To examine whether the aberrant membrane structures seen in the Cdc50p-depleted gcs1Δ mutant were TGN compartments, we observed the localization of Sec7p-mRFP1, and compared it with that of GFP-Tlg1p. Consistent with the previous data that Tlg1p is localized to early endosomes and the TGN (Lewis et al., 2000; Holthuis et al., 1998b), GFP-Tlg1p partially colocalized with Sec7p-mRFP1 in the wild-type, Cdc50p-depleted wild-type, and gcs1Δ cells (Fig. 6C). In the Cdc50p-depleted gcs1Δ mutant, Sec7p-mRFP1 gave a similar punctate pattern, and did not colocalize with GFP-Tlg1p, suggesting that the GFP-Tlg1p-containing structures were independent of the TGN membranes.
Taken together, our results suggest that the Cdc50p-depleted gcs1Δ mutant was defective in the retrieval pathway from early endosomes to the TGN, and that GFP-Snc1p, GFP-Tlg1p, and Chs3p-GFP all accumulated in the same aberrant endosomal structures.
We next investigated genetic interactions between cdc50Δ and mutations in genes coding for proteins that have been implicated in Arf1-mediated or clathrin-coated vesicle budding. These include GGAs (gga1Δ and gga2Δ) (Costaguta et al., 2001; Dell’Angelica et al., 2000; Hirst et al., 2000), clathrin (chc1-521) (Chen and Graham, 1998), clathrin-associated AP complexes (apl2Δ, apl4Δ, apm1Δ, and aps1Δ for AP-1 and apl5Δ, apl6Δ, apm3Δ, and aps3Δ for AP-3) (Yeung et al., 1999), and COPI (sec21-1) (reviewed in Kreis et al., 1995; Letourneur et al., 1994). The cdc50Δ mutant was crossed with the respective mutants and the growth of double or triple mutants was examined using tetrad analysis (Fig. 7A). Consistent with previous observations with the drs2Δ mutation (Chen et al., 1999), the cdc50Δ mutation exhibited synthetic growth defects with the chc1-521 mutation, but not with the sec21-1 mutation. Interestingly, the gga1Δ gga2Δ mutation exhibited synthetic lethality with the cdc50Δ mutation. To examine the effects of the gga1Δ or gga2Δ single mutation on the growth of the cdc50Δ mutant more precisely, we constructed PGAL1-3HA-CDC50 ggaΔ mutants in the S288C background in which mutations in GGAs, AP-1, and AP-3 in Fig. 7A were constructed. It was previously reported that the gga2Δ mutation, but not the gga1Δ mutation, exhibited synthetic growth defects in combination with the deletion of APL2 (the gene encoding the β1 subunit of AP-1) or the chc1-521 mutation (Costaguta et al., 2001), suggesting that the Gga2p is the major Gga protein. Consistently, the Cdc50p-depleted gga2Δ mutant exhibited the slow growth, whereas the Cdc50p-depleted gga1Δ mutant grew normally (Fig. 7B).
 View Details | Fig. 7. Gga1p and Gga2p are required for growth of the cdc50Δ mutant. (A) Genetic interactions between CDC50 and genes coding for a protein that has been implicated in Arf1-mediated or clathrin-coated vesicle budding. The cdc50Δ mutant was crossed with the indicated mutants to generate diploids. Diploid cells were sporulated, and synthetic lethality was examined using tetrad analysis at 30°C. (B) Depletion of Cdc50p causes a growth defect in the ggaΔ mutants. Wild-type (KKT2; a), PGAL1-3HA-CDC50 (KKT127; b), gga1Δ (KKT299; c), gga2Δ (KKT300; d), PGAL1-3HA-CDC50 gga1Δ (KKT301; e), PGAL1-3HA-CDC50 gga2Δ (KKT302; f), gga1Δ gga2Δ (KKT303; g), and PGAL1-3HA-CDC50 gga1Δ gga2Δ (KKT304; h) strains were streaked onto a YPGA or YPDA plate, followed by incubation at 30°C for 2 d. (C) CPY processing in the Cdc50p-depleted ggaΔ mutants. The strains used in (B) were grown in YPDA for 8 h at 30°C to deplete Cdc50p, labeled with Trans35S-label for 10 min, and chased for 0, 15, or 30 min. CPY was immunoprecipitated, resolved by SDS-PAGE, and detected by autoradiography. The Cdc50p-depleted wild-type, gga1Δ, and gga2Δ cells exhibited wild-type rates of CPY processing (our unpublished results). |
Because Gga1p and Gga2p are involved in the TGN-to-late endosome transport of CPY (Zhdankina et al., 2001; Dell’Angelica et al., 2000; Hirst et al., 2000), we examined the mutants after an 8-h culture at 30°C to determine whether the depletion of Cdc50p affect the CPY sorting in the gga mutants. The maturation of CPY was normal in the gga1Δ, gga2Δ, Cdc50p-depleted wild-type, and Cdc50p-depleted gga1Δ cells as well as in the wild-type strain (Fig. 7C and our unpublished results), whereas the Cdc50p-depleted gga2Δ mutant exhibited a slight delay in the maturation. Additionally, a small amount of the P2 form of CPY accumulated in the gga1Δ gga2Δ mutant as described previously (Zhdankina et al., 2001; Dell’Angelica et al., 2000; Hirst et al., 2000), and the depletion of Cdc50p slightly exacerbated this phenotype (Fig. 7C). Although about 50% of CPY matured within 15 min, the depletion of Cdc50p slightly exacerbated this phenotype (Fig. 7C), indicating that the CPY transport pathway was impaired in the Cdc50p-depleted gga1Δ gga2Δ mutant.
To examine the endocytic recycling pathway in the Cdc50p-depleted ggaΔ mutants, we observed the localization of mRFP1-Snc1p and GFP-Tlg1p in these cells. mRFP1-Snc1p was primarily localized to the plasma membrane at polarized growth sites in the gga1Δ, gga2Δ, and Cdc50p-depleted gga1Δ mutants, as well as in the wild-type strain (Fig. 8 and our unpublished results). Consistent with a previous report (Black and Pelham, 2000), mRFP1-Snc1p was primarily localized to intracellular punctate structures in 81% of the gga1Δ gga2Δ cells (n=103). The defects in the endocytic recycling pathway in the gga1Δ gga2Δ mutant, however, seemed to be mild, because GFP-Tlg1p exhibited a normal punctate localization and partial colocalization with mRFP1-Snc1p in the mutant cells, as was observed in the gga1Δ, gga2Δ, Cdc50p-depleted gga1Δ, and wild-type strains (Fig. 8 and our unpublished results). In contrast, mRFP1-Snc1p and GFP-Tlg1p accumulated in aberrant membrane structures in most of the Cdc50p-depleted gga2Δ and Cdc50p-depleted gga1Δ gga2Δ mutant cells, as was observed with the Cdc50p-depleted gcs1Δ mutant, suggesting that the retrieval pathway from early endosomes to the TGN was severely impaired in these mutants. When these cells were treated with LAT-A at 30°C for 1 h after a 7-h incubation to deplete the cell of Cdc50p, the plasma membrane localization of GFP-Snc1p was observed in the gga1Δ gga2Δ, Cdc50p-depleted gga2Δ, and Cdc50p-depleted gga1Δ gga2Δ mutants (our unpublished results), suggesting that the accumulation of the aberrant Snc1p-containing structures was dependent on endocytosis. These data also imply that the late secretory pathway was not severely impaired in the Cdc50p-depleted ggaΔ mutants.
 View Details | Fig. 8. mRFP1-Snc1p largely colocalizes with GFP-Tlg1p in the Cdc50p-depleted ggaΔ mutants. The strains used in Fig. 7B were cotransformed with pRS416-mRFP1-SNC1 and YEplac181-GFP-TLG1. Cells were grown in YPDA medium at 30°C for 8 h, and observed by fluorescence microscopy. Obtained images were merged to demonstrate the coincidence of the two signals. The Cdc50p-depleted, gga1Δ, and Cdc50p-depleted gga1Δ mutants exhibited the wild-type localization pattern of mRFP1-Snc1p and GFP-Tlg1p (our unpublished results). Bar, 5 μm. |
Taken together, these data suggest that like the Cdc50p-depleted gcs1Δ mutant, the Cdc50p-depleted gga1Δ gga2Δ mutant was defective in the retrieval pathway from early endosomes to the TGN.
Because AP-1 and COPI have been implicated in the retrieval pathway from early endosomes to the TGN (Robinson et al., 2006; Cai et al., 2005; Valdivia et al., 2002; Lewis et al., 2000), we further examined genetic interactions between CDC50 or GCS1 and two genes encoding coat proteins. Interestingly, the apl2Δ mutation, but not the sec21-1 mutation (γ subunit of COPI), caused mild growth defects with the gcs1Δ mutation (Fig. 9A and our unpublished results), whereas the cdc50Δ mutation did not exhibit genetic interaction with these genes (Fig. 7A). This growth defect may be caused by defective early endosome-to-TGN transport: the apl2Δ and gcs1Δ mutants displayed a normal localization of GFP-Snc1p, whereas GFP-Snc1p accumulated in intracellular membrane structures of 79% of the apl2Δ gcs1Δ cells (n=105; Fig. 9B). These results are consistent with the idea that AP-1 is involved in the formation of CCVs from early endosomes (Valdivia et al., 2002). To examine whether AP-1 is localized to early endosomes, we constructed strains expressing the integrated version of APL4-GFP (the gene encoding the γ subunit of AP-1), as a sole copy of APL4. The APL4-GFP allele was functional, as assessed by the growth phenotype of the mutant with APL4-GFP in combination with the gga1Δ gga2Δ mutation, which causes synthetic growth defects with the apl4Δ mutation at the restrictive temperature (Costaguta et al., 2002 and our unpublished results). Apl4p-GFP was localized to internal punctate structures in the wild-type strain, in a manner that resembled endosomal/TGN compartments (Fig. 9C). Microscopic examination of wild-type cells co-expressing Apl4p-GFP and Sec7p-mRFP1 revealed that Apl4p-GFP partially colocalized with Sec7p-mRFP1 (Fig. 9C). Interestingly, 32% of the Apl4p-GFP-containing punctate structures (69 of the 217 puncta in 55 cells) did not colocalize with Sec7p-mRFP1 (Fig. 9C, arrowhead), suggesting that Apl4p-GFP was also localized to endosomal compartments. The early endosome could be visualized after brief incubation with FM4-64 (Vida and Emr, 1995). Wild-type cells expressing Apl4p-GFP were labeled with FM4-64 at 0°C, incubated for 3 min at 30°C, and immediately observed by fluorescence microscopy. Thirty-two percent of the Apl4p-GFP-containing structures (27 of the 84 puncta in 18 cells) were labeled with FM4-64 (Fig. 9D, arrowhead). These results suggest that AP-1 localizes to both early endosomes and the TGN. Localization of AP-1 to endosomal membranes and defects in the early endosome-to-TGN transport pathway in the Cdc50p-depleted gcs1Δ mutant prompted us to examine the localization of Apl4p-GFP in the Cdc50p-depleted gcs1Δ mutant. Apl4p-GFP was localized to intracellular punctate structures in the cdc50Δ and gcs1Δ mutants as well as in the wild-type strain (Fig. 9E). In contrast, in the Cdc50p-depleted gcs1Δ mutant cells, Apl4p-GFP was observed in a hazy pattern distributed throughout the cell, in addition to some punctate structures (Fig. 9E). These results suggest that Cdc50p and Gcs1p have redundant functions in the recruitment of AP-1 to endosomal/TGN membranes, and that defects in the early endosome-to-TGN transport in the Cdc50p-depleted gcs1Δ mutant may be partly due to a failure in the recruitment of AP-1 to early endosome membranes.
 View Details | Fig. 9. AP-1 is implicated in the early endosome-to-TGN transport. (A) The apl2Δ gcs1Δ mutant exhibits a slow growth phenotype. Wild-type (KKT2; a), apl2Δ (KKT305; b), gcs1Δ (KKT306; c), and apl2Δ gcs1Δ (KKT307; d) strains were streaked onto a YPDA plate, followed by incubation at 30°C or 37°C for 1 d. (B) GFP-Snc1p accumulates in intracellular membrane structures in the apl2Δ gcs1Δ mutant. Wild-type (KKT2), apl2Δ (KKT305), gcs1Δ (KKT306) and apl2Δ gcs1Δ (KKT307) strains were transformed with pRS416-GFP-SNC1. Cells were grown to mid-log phase in YPDA medium at 30°C and observed by fluorescence microscopy. (C) Localization of Apl4p-GFP to TGN-independent structures. Wild-type cells co-expressing Apl4p-GFP and Sec7p-mRFP1 (YKT1306) were grown to mid-log phase in YPDA medium at 30°C, and observed by fluorescence microscopy. An arrowhead indicates the Apl4p-GFP structure that does not contain Sec7p-mRFP1. (D) Staining Apl4p-GFP-expressing cells with FM4-64. Wild-type cells expressing Apl4p-GFP (YKT1302) were grown in YPDA to late-log phase at 30°C, labeled in 32 μM FM4-64 for 30 min at 0°C, chased in SD medium for 3 min at 30°C, and observed immediately by fluorescence microscopy. An arrowhead indicates the Apl4p-GFP structure that is labeled with FM4-64. (E) Localization of Apl4p-GFP in the Cdc50p-depleted gcs1Δ mutant. Wild-type (YKT1302; WT), cdc50Δ (YKT1303), gcs1Δ (YKT1304), and PGAL1-3HA-CDC50 gcs1Δ (YKT1305; Cdc50p-depleted gcs1Δ) strains expressing Apl4p-GFP were grown in YPDA medium at 30°C for 8 h, and observed by fluorescence microscopy. In (C) and (D), obtained images were merged to demonstrate the coincidence of the two signals. Bars, 5 μm. |
EM has previously revealed the accumulation of large abnormal double-membrane structures with crescent- or ring-shaped morphologies in the cdc50Δ and drs2Δ mutants grown at lower temperatures (Misu et al., 2003; Chen et al., 1999; our unpublished results) and in the Cdc50p-depleted erg3Δ mutant grown at 30°C (Kishimoto et al., 2005). When grown in YPDA medium at 30°C for 8 h to deplete the cells of Cdc50p, the PGAL1-3HA-CDC50 gcs1Δ mutant cells accumulated a large number of similar structures [12.5 abnormal membrane structures (>200 nm in diameter)/10 μm2, n=32 sections]. In contrast, the PGAL1-3HA-CDC50 mutant cells accumulated a little (2.3 structures/10 μm2, n=30 sections), and the gcs1Δ mutant cells did not (Fig. 10A). To examine whether these abnormal structures were the same as the Snc1p-containing structures shown in Fig. 5A, we performed immuno-EM on Cdc50p-depleted gcs1Δ cells expressing 3HA-tagged Snc1p (Robinson et al. 2006). As shown in Fig. 10B, the abnormal double-membrane structures were labeled with immunogold particles in the Cdc50p-depleted gcs1Δ mutant, but not in control Cdc50p-depleted gcs1Δ cells expressing untagged Snc1p (our unpublished results).
 View Details | Fig. 10. Accumulation of abnormal membrane structures in the Cdc50p-depleted gcs1Δ mutant. (A) Electron microscopic observation of the Cdc50p-depleted gcs1Δ mutant. Strains used in Fig. 2C were grown in YPDA medium at 30°C for 8 h, fixed with glutaraldehyde-osmium, and processed for electron microscopic observation. For the Cdc50p-depleted gcs1Δ mutant, higher magnification photographs of the areas surrounded by dashed lines are presented to the right side of the panels. Bars, 1 μm and 400 nm for the lower and higher magnification images, respectively. (B) Immunoelectron microscopic observation of the Cdc50p-depleted gcs1Δ mutant expressing 3HA-Snc1p. The PGAL1-3HA-CDC50 gcs1Δ mutant (YKT1286) harboring pRS416-3HA-SNC1 was grown in YPDA medium at 30°C for 8 h, fixed, and processed for immunoelectron microscopic observation. Mouse anti-HA antibodies and 10-nm gold-conjugated anti-mouse IgG antibodies were used as primary and secondary antibodies, respectively. Arrows indicate representatives of the 3HA-Snc1p-positive membrane structures. Bar, 200 nm. |
The Cdc50p-depleted gga1Δ gga2Δ and apl2Δ gcs1Δ mutants were also examined for the accumulation of abnormal membrane structures by EM. When the cells were depleted of Cdc50p for 8 h, crescent- or ring-shaped membrane structures accumulated in the Cdc50p-depleted gga2Δ and Cdc50p-depleted gga1Δ gga2Δ mutant cells, as was observed in the Cdc50p-depleted gcs1Δ mutant, whereas these structures were rarely observed in the gga1Δ, gga2Δ, and gga1Δ gga2Δ mutant cells (Fig. 11A and our unpublished results). Crescent- or ring-shaped membrane structures also accumulated in the apl2Δ gcs1Δ mutant cells, although this phenotype was mild; abnormal membrane structures (>200 nm in diameter) were fewer (1.5 structures/10 μm2, n=32 sections) compared with the Cdc50p-depleted gcs1Δ mutant (12.5 structures/10 μm2, n=32 sections) and the Cdc50p-depleted gga1Δ gga2Δ mutant (5.7 structures/10 μm2, n=30 sections) (Fig. 11B and our unpublished results). Thus, it seems that the severity of the growth defect in these mutants is correlated with the extent to which abnormal membranes are intracellularly accumulated.
 View Details | Fig. 11. Accumulation of intracellular membrane structures in the Cdc50p-depleted ggaΔ and apl2Δ gcs1Δ mutants. (A) Electron microscopic observation of the Cdc50p-depleted ggaΔ mutants. The gga1Δ gga2Δ (YKT1300) and PGAL1-3HA-CDC50 gga1Δ gga2Δ (YKT1301) mutant cells were grown in YPDA medium at 30°C for 8 h, fixed with glutaraldehyde-osmium, and processed the electron microscopic observation. (B) Electron microscopic observation of the apl2Δ gcs1Δ mutant. The apl2Δ gcs1Δ (KKT307) mutant cells were grown to early logarithmic phase in YPDA medium at 30°C, and processed as described in (A). Bars, 1 μm. |
Taken together with the results obtained by fluorescence microscopy, these results imply that the Cdc50p-depleted gcs1Δ, Cdc50p-depleted gga1Δ gga2Δ, and apl2Δ gcs1Δ mutants are defective in the formation of vesicles destined for the TGN from early endosomes.
It was previously suggested that Drs2p is involved in Arf1p- and clathrin-dependent vesicle formation; DRS2 was identified as a synthetic-lethal mutation with arf1Δ (Chen et al., 1999; Chen and Graham, 1998), and fewer CCVs were isolated from drs2Δ cells than from wild-type cells (Chen et al., 1999). Some of these CCVs may have been post-Golgi secretory vesicles, because the drs2Δ mutation reduced the number of vesicles that accumulated when the actin cytoskeleton was disrupted (Gall et al., 2002). On the other hand, the involvement of Arf1p and Arf2p in the endocytic pathways has been suggested (Yahara et al., 2001; Gaynor et al., 1998). The effects of the cdc50Δ and drs2Δ mutations on the formation of endocytic vesicles, however, cannot be assessed by disruption of the actin cytoskeleton, because endocytosis is dependent on the actin cytoskeleton in yeast (Engqvist-Goldstein and Drubin, 2003). Snc1p-containing membrane structures accumulated in the arf1Δ and arf1-18 arf2Δ mutants in an endocytosis-dependent manner. Ultrastructural analyses revealed that the arf1Δ and some arf1-ts arf2Δ mutants contained unusual membranous structures, such as discontinuous ring-like or Golgi stack-like structures (Yahara et al., 2001; Gaynor et al., 1998), which may have been derived from endosomes. Thus, the accumulation of abnormal structures in the Cdc50p-depleted gcs1Δ mutant that likely were a result of excessive early endosomal membranes may have been due to defective formation of CCVs or COPI vesicles (see below) from early endosomes. The involvement of clathrin in the endocytic recycling pathway has been demonstrated in mammalian cells (van Dam et al., 2002), and suggested in yeast; clathrin and AP-1 act to recycle Chs3p from early endosomes to the TGN (Valdivia et al., 2002). A mutant carrying the chc1-521 allele, which is synthetically lethal with cdc50Δ or drs2Δ (Chen et al., 1999; our unpublished results), exhibited an intracellular accumulation of Snc1p, and this defect was exacerbated by depletion of Cdc50p (our unpublished results).
Because Arf1p is likely involved in multiple vesicular transport pathways (Yahara et al., 2001), the cdc50 arf1 mutant should exhibit defects in various pathways, making the interpretation of the observed phenotypes complicated. To circumvent this problem, we explored genetic interactions between CDC50 and regulators of ARF1, which are presumably involved in a more specific transport pathway. Gcs1p and Age2p provide overlapping essential function for transport from the TGN (Poon et al., 2001). Interestingly, the cdc50Δ mutation did not cause synthetic growth defects with the age2Δ mutation, suggesting that Gcs1p is specifically involved in the function of Cdc50p-Drs2p. The Cdc50p-depleted gcs1Δ mutant did not exhibit apparent defects in post-Golgi vesicular transport pathways, presumably because Age2p compensates for the function of Gcs1p in these pathways. Recently, the Arf-like protein Arl1p was shown to be a potential substrate of Gcs1p (Liu et al., 2005), raising the possibility that the synthetic growth defect of the Cdc50p-depleted gcs1Δ mutant is caused by deregulation of Arl1p. The arl1Δ mutation, however, did not cause synthetic growth defects with the cdc50Δ mutation (our unpublished results). In addition, Arl1p has been implicated in vesicle fusion with TGN membranes (Panic et al., 2003), whereas we did not observe a discernible increase in the number of vesicles using EM sectioning of the Cdc50p-depleted gcs1Δ mutant (our unpublished results). Thus, it seems that the synthetic phenotypes in the Cdc50p-depleted gcs1Δ mutant are caused by deregulation of Arf1p.
In the current study, we have not identified a mutation in a gene encoding an Arf GEF that caused synthetic growth defects with the cdc50Δ mutation. It was previously reported that Drs2p physically interacts with Gea2p (Chantalat et al., 2004). GFP-Snc1p, however, was localized to the plasma membrane in the gea1-4 gea2Δ mutant even at the restrictive temperature (our unpublished results). Thus, this Drs2p-Gea2p interaction may play a specific role in the TGN function of Cdc50p-Drs2p as suggested previously (Chantalat et al., 2004).
Among the vesicular transport pathways examined in this study, only the retrieval pathway from early endosomes to the TGN was severely impaired in the Cdc50p-depleted gcs1Δ mutant. Mislocalization of the TGN resident protein Kex2p in the Cdc50p-depleted gcs1Δ mutant could be explained by a default transport system that moves this protein to the vacuole when the transport pathway from early endosomes to the TGN is defective (Lewis et al., 2000; Wilcox et al., 1992). It is possible that this defect was indirectly caused by defective TGN-to-early endosome transport, which would affect the function of early endosomes. However, we believe this was unlikely, since Tlg1p, which is recycled between early endosomes and the TGN (Lewis et al., 2000), accumulated in the GFP-Snc1p-containing structures, but not in the Sec7p-positive structures, implying that the TGN-to-early endosome transport was not impaired in the Cdc50p-depleted gcs1Δ mutant. Our recent results suggest that CDC50 and DRS2 are actually involved in the retrieval pathway from early endosomes to the TGN. Genetic data suggest that CDC50 and DRS2 are functionally very similar to RCY1, which is required for the endocytic recycling pathway, but not for vacuolar protein sorting or the secretory pathway (Wiederkehr et al., 2000). Moreover, Cdc50p-Drs2p was coimmunoprecipitated with Rcy1p (Furuta et al., our unpublished results). In addition, it has recently been suggested that Gcs1p is involved in the COPI vesicle formation at early endosomes through an interaction with Snc1p and subsequently promoting the Arf1p-GTP recruitment (Robinson et al., 2006). Taken together, it seems that the observed defects in the early endosome-to-TGN transport in the Cdc50p-depleted gcs1Δ mutant were due to synthetic effects in the same pathway, rather than cumulative defects in multiple pathways. Thus, the Cdc50p-Drs2p putative APLT and Gcs1p may cooperate in vesicle formation at early endosomes.
We identified GGA1 and GGA2 in our search for genes encoding clathrin adaptors that when mutated cause synthetic growth defects with the cdc50Δ mutation. Similar to the Cdc50p-depleted gcs1Δ mutant, the Cdc50p-depleted gga1Δ gga2Δ mutant intracellularly accumulated Snc1p-containing structures, suggesting that the Cdc50p-depleted gga1Δ gga2Δ mutant is defective in the retrieval pathway from early endosomes to the TGN. Unlike GCS1, however, in which null mutation does not affect the rate of CPY maturation, Gga1p and Gga2p are involved in the TGN-to-late endosome transport pathway with clathrin (Costaguta et al., 2001). In the gga1Δ gga2Δ mutant, the late endosomal protein Pep12p was redirected to early endosomes (Black and Pelham, 2000). Thus, delivery of late endosomal or vacuolar proteins to early endosomes may perturb the normal function of early endosomes. This defect caused by the gga1Δ gga2Δ mutation may exacerbate defects in the early endosome-to-TGN transport pathway in the Cdc50p-depleted wild-type strain, although it is also possible that Gga1p and Gga2p are involved in vesicle budding at early endosomes, as has been suggested in mammalian cells (He et al., 2005).
AP-1/clathrin and Arf1p have been implicated in the retrieval of Chs3p from early endosomes to the TGN (Valdivia et al., 2002). The apl2Δ mutation as well as the cdc50Δ mutation caused synthetic defects with the gcs1Δ mutation in growth and endocytic recycling of Snc1p, although the defects in the apl2Δ gcs1Δ mutant were mild compared to those in the Cdc50p-depleted gcs1Δ mutant. In addition, like the cdc50Δ mutation, the apl2Δ mutation causes synthetic growth defects with the gga1Δ gga2Δ mutation (Costaguta et al., 2001), but not with the cdc50Δ mutation (our unpublished results). These results suggest that Cdc50p-Drs2p and AP-1 may be involved in a common step in the early endosome-to-TGN transport pathway. Very recently, the apl2 gga1 gga2 mutant has been shown to exhibit defects in the early endosome-to-TGN retrieval of a model TGN protein consisting of the cytosolic domain of Ste13p fused to the transmembrane and luminal domains of alkaline phosphatase (Foote and Nothwehr, 2006). Furthermore, the direct interaction of a cytosolic region of Ste13p with AP-1 suggests that Ste13p is recruited into AP-1/clathrin vesicles at early endosomes (Foote and Nothwehr, 2006). We demonstrated that Apl4p-GFP was localized to endosomal/TGN compartments in a manner dependent on Cdc50p and Gcs1p (Fig. 9). Thus, the Cdc50p-Drs2p putative APLT may promote the formation of AP-1/clathrin vesicles at early endosomes. Localization of Apl2p-GFP to punctate structures was completely abolished by treatment with brefeldin A, an inhibitor for the Arf1 activation (Fernandez and Payne, 2006), consistent with the idea that Gcs1p contributes to the localization of AP-1 to endosomal membranes via regulation of the Arf1p activity.
The cdc50 null mutant intracellularly accumulates Snc1p (Saito et al., 2004), whereas the apl2Δ single mutant exhibits an almost wild-type phenotype, implying that Cdc50p-Drs2p is involved in the formation of another type of vesicle. COPI vesicles may be candidates for these vesicles. Several lines of evidence suggest that COPI is also involved in the transport of Snc1p from early endosomes to the TGN (Lewis et al., 2000; Cai et al., 2005; Robinson et al., 2006). In vitro studies have suggested that the physical interaction between Snc1p and Gcs1p results in the efficient recruitment of GTP-Arf1p and coatomer to Snc1p (Robinson et al., 2006). It is known that Arf GAPs act to dissociate the protein coat from membranes by stimulating GTP hydrolysis and converting Arf1p-GTP to Arf1p-GDP (Tanigawa et al., 1993). Recent studies about the COPI-mediated transport pathway, however, suggest that Arf GAPs function as subunits of coat proteins rather than simply Arf inactivators, and that Arf GAP activity is required for the packaging of cargo proteins into COPI vesicles (for review, see Nie and Randazzo, 2006). Therefore, Gcs1p may be directly involved in the formation of COPI vesicles at early endosomes.
It has been proposed that APLTs locally generate phospholipid asymmetry to recruit proteins that promote vesicle formation or to assist membrane deformation for vesicle budding (Graham, 2004). Interestingly, Gcs1p has an ArfGAP1 Lipid Packing Sensor (ALPS) motif that is thought to recognize curved membranes (Bigay et al., 2005). Snc1p physically interacts with Gcs1p (Robinson et al., 2006) and Rcy1p (Chen et al., 2005), and Rcy1p interacts with the Cdc50p-Drs2p complex (Furuta et al., our unpublished results), raising the possibility that these proteins are components of the machinery that initiates vesicle budding at early endosomes. Local phospholipid asymmetry generated by these protein interactions might be utilized to recruit AP-1/clathrin or COPI coatomer to the nascent vesicle budding site.
We wish to thank Akihiko Nakano, Catherine Jackson, Charles Boone, Gregory Payne, Randy Schekman, Roger Tsien, and Yoshinori Ohsumi for yeast strains, plasmids, and antibodies, and Eriko Itoh for her technical assistance. We wish to thank Mamiko Satoh for her instructions for electron microscopy and Masahiko Watanabe for the microtome. We wish to thank the members of the Tanaka Lab for valuable suggestions over the course of these experiments. This work was supported in part by Grants-in-Aid for Scientific Research from the Japan Society for the Promotion of Science and the Ministry of Education, Culture, Sports, Science, and Technology of Japan to T. Y. and K. T.
|
Index