| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
J. Biol. Chem., Vol. 280, Issue 30, 27613-27623, July 29, 2005
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
From the Department of Physiology and Biophysics, University of Arkansas for Medical Sciences, Little Rock, Arkansas 72205
Received for publication, April 27, 2005 , and in revised form, June 2, 2005.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
cog1–8 mutants. As a result, we created a model of the protein-protein interactions within the yeast COG complex and proposed that Cog1p is a bridging subunit between the two COG lobes. In support of this hypothesis, we have demonstrated that Cog1p is required for the stable association between two COG subcomplexes. |
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
The COG complex consists of eight subunits named COG1–8 (12, 27–31). Based on yeast genetic studies and on results from quick freeze/deep etch/rotary shadow electron microscopy (28), the COG subunits have been grouped into two lobes consisting of COG1–4 and COG5–8 apparently interconnected by thin rods and/or globules. A recently proposed model suggested that, in the mammalian COG complex, Cog4p acts as the bridging subunit between the two lobes of the complex (32). Mutation or deletion of the individual members of the COG complex gives rise to different phenotypes, suggesting that each member plays a distinct role within the complex. In yeast, deletion of COG1–4 results in severe growth defects (12, 13, 31, 33), accumulation of internal membranes (30), reduced glycosylation of secretory proteins (12, 30), and an altered distribution of SNARE proteins Sec22p (12) and Snc1p (30).
Yeast and mammalian COG3–6 and COG8 are structurally homologous. The remaining subunits COG1/LdlBp, COG2/LdlCp, and COG7 are not structurally related, but may represent functional counterparts (30). As in yeast, mutations in COG subunits (COG1–8) have been shown to affect the structure and function of the Golgi apparatus in Drosophila melanogaster sperm as well as in mammalian somatic cells (12, 16, 28, 31, 34, 35). Compromised COG function causes severe defects in glycosylation, intracellular protein sorting, secretion, and, in some cases, cell growth. For example, in Chinese hamster ovary cells mutated for COG1 (ldlB) or COG2 (ldlC), a phenotype of multiple dilated Golgi cisternae is shown (28). COG mutant cells exhibit pleiotropic defects in a number of medial- and trans-Golgi-associated glycosylation reactions affecting virtually all N-linked, O-linked, and lipid-linked glycoconjugates (12, 29). The activities of glycosylating enzymes depend on their proper intra-Golgi localization (36, 37). The diversity and heterogeneity of protein glycosylation defects suggest that the COG mutations affect the compartmentalization or activity of multiple Golgi glycosylation enzymes without substantial disruption of secretion and endocytosis. Thus, the COG complex may play a direct or indirect role in transport, retention, or retrieval to appropriate cisternae of resident Golgi proteins.
A role for the COG complex in membrane trafficking is supported by biochemical and genetic studies in yeast that have identified a large number of COG complex-interacting genes that encode proteins implicated in Golgi trafficking (12, 13, 33). The COG complex interacts genetically and physically with Ypt1p, intra-Golgi SNARE molecules, and the COPI coat complex. In addition, electron microscopy revealed that temperature-sensitive yeast cog2 and cog3 mutants accumulate vesicles at the non-permissive temperature (38). These findings led to the hypothesis that the COG complex acts as a tether that connects COPI vesicles with cis-Golgi membranes during retrograde intra-Golgi traffic (12, 16). In addition, the yeast COG complex has been proposed to function as a vesicle tether in anterograde endoplasmic reticulum-to-Golgi traffic (13, 33, 38). The COG complex is also involved in proper localization of yeast enzymes in the trans-Golgi network (39) and possibly in cargo sorting during exit from the endoplasmic reticulum (40). Whether all these functions are related to a tethering role remains unknown.
To better understand the architecture and molecular dynamics of the COG complex, we have applied three individual methods of investigation to study the direct protein-protein interactions between the eight subunits of the yeast complex. In this study, we describe the observed interactions and propose an organizational model of the complex with Cog1p acting as a bridging subunit between the two lobes of the COG complex.
|
EXPERIMENTAL PROCEDURES |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
80-COG1 plasmids were obtained by PCR amplification of either full-length yeast COG1 or COG1 without the first 80 amino acids and subcloning of the PCR products into the pRS415 vector together with the COG1 promoter. The Saccharomyces cerevisiae strains used in this work are listed in Table II.
|
|
-D-thiogalactopyranoside at 37 °C. Cells were harvested (6000 x g for 15 min at 4 °C in a Sorvall SA600 rotor) and washed twice with Tris-buffered saline. Cells were lysed using 20 ml of BugBuster protein extraction reagent (Novagen) with addition of 20 µl of lysozyme (10 mg/ml). Protease inhibitor mixture (Roche Applied Science) was added at the end of the lysis reaction. Insoluble proteins were separated by centrifugation at 20,000 x g for 15 min. The pellet was resuspended in 10 ml of buffer A (20 mM Tris-HCl and 300 mM NaCl (pH 8.0)) and 1% Sarkosyl and subjected to sonication (3 x 1 min). Then, 10 ml of 1% Triton X-100 in buffer A were added; the mixture was spun at 20,000 x g; and the S2 supernatant was obtained. The S2 supernatant was passed through a 0.45-µm filter and loaded onto a TALON metal affinity resin column (Clontech). The bound proteins were eluted with 100 mM imidazole. Amicon ultracentrifugal filter devices (Millipore Corp.) were used to concentrate the eluted fractions and to reduce the imidazole concentration to 5 mM.
T7-tagged COG Protein Extraction—Rosetta DE3 cells expressing the pET24b plasmid with the COG1–8 open reading frames were grown in 500 ml of LB medium at 37 °C to A600 = 0.4 and induced with 1 mM isopropyl -D-thiogalactopyranoside overnight at 20 °C to improve COG protein solubility. Cells were harvested by centrifugation and washed twice with Tris-buffered saline. Cells were resuspended in 5 ml of buffer A with 20 µl of lysozyme (10 mg/ml) and incubated for 20 min at room temperature. After incubation, 15 ml of buffer A supplemented with protease inhibitor mixture, DNase, and phenylmethylsulfonyl fluoride were added. Cells were lysed by sonication (3 x 1 min). Lysates were centrifuged at 16,000 x g for 15 min; S1 supernatants were passed through a 0.45-µm syringe filter, distributed in aliquots, and snap-frozen in liquid nitrogen for further experiments.
Anti-Cog1p and Anti-Cog4–8p Antibody Production and Immunoblotting—Purified GST-His6-tagged COG proteins were used to raise rabbit polyclonal antibodies. The antibodies were affinity-purified using fusion proteins immobilized on CNBr-activated Sepharose 4B (Amersham Biosciences) and were eluted with 200 mM glycine (pH 2.7). Anti-Cog1p (1:1000), anti-Cog2p (1:500), anti-Cog3p (1:300), anti-Cog4p (1:100), and anti-Cog5–8p (1:300) antibodies were used at the indicated dilutions for Western blotting in 5% bovine serum albumin for 1 h at room temperature. To detect the T7-tagged COG proteins, horseradish peroxidase-conjugated anti-T7 antibodies (Novagen) were used at 1:5000 dilution. Affinity-purified antibodies against Sed5p (41) and sera against gp400/HSP150 (42) and Sec21p and Sec61p (a gift from R. Schekman, University of California, Berkeley, CA) were all used at a dilution of 1:2000. Horseradish peroxidase-conjugated goat anti-rabbit secondary antibodies (Jackson ImmunoResearch Laboratories) was used at a dilution of 1:5000 in 5% milk. Immunoblots were developed with the Chemiluminescence Reagent Plus kit (PerkinElmer Life Sciences).
Coexpression of His-Cog1p and GST-tagged Cog1–8p—Rosetta DE3 cells were transformed with the pHis-COG1 plasmid and later converted into competent cells for transformation with pGST-COG1–8. Colonies were screened for strains coexpressing both proteins. Cells were grown in 50 ml of LB medium supplemented with kanamycin and ampicillin to A600 = 0.4 and induced with 1 mM isopropyl -D-thiogalactopyranoside overnight at 20 °C. Cells were lysed as described above, and 1 ml of the S2 supernatant was obtained. 20 µl of the S2 supernatant were incubated with 40 µl (bed volume) of S protein-agarose (Novagen) for 1 h, and the final volume was adjusted to 200 µl with S protein-agarose buffer (200 mM Tris-HCl (pH 7.5), 1.5 M NaCl, and 1% Triton X-100). Beads were washed with buffer A, and proteins were eluted with SDS sample buffer (63 mM Tris-HCl, pH 6.8, 2% SDS, and 10% glycerol) and analyzed by Western blotting.
Yeast Two-hybrid System—Matchmaker Gal4 Two-Hybrid System 3 was purchased from Clontech. Yeast strain AH109 was used as the host for interaction studies. Yeast strain AH109 was cotransformed with cDNAs for COG1–8 expressed as a fusion with the Gal4 DNA-binding domain as well as another cDNA expressed as a fusion with the Gal4 DNA activation domain. Transformants were grown on synthetic complete medium/–Leu/–Trp. To determine protein-protein interactions as a function of histidine and arginine biosynthesis (HIS3 and ADE3), four independent colonies were streaked on synthetic complete medium/–Leu/–Trp/–Ade/–His. Growth of dense colonies was interpreted as a strong interaction and that of sparse colonies as a weak interaction. The -galactosidase activity was quantified by growing yeast colonies in liquid synthetic complete medium/–Leu/–Trp to mid-log phase with shaking at 30 °C to induce expression of fusion proteins. Yeast cells were harvested by centrifugation; washed with 60 mM Na2HPO4, 40 mM NaH2PO4, 10 mM KCl, and 1 mM MgSO4; and lysed by three cycles of freezing in liquid nitrogen and thawing to 37 °C. O-Nitrophenyl -D-galactopyranoside was used as a substrate. Cell debris was removed by centrifugation at 10,000 x g for 5 min. The -galactosidase activity was measured photometrically at 420 nm. Three independent transformants harboring each of the respective plasmids were tested in duplicate. 1 unit of -galactosidase is defined as the amount that hydrolyzes 1 µmol of O-nitrophenyl -D-galactopyranoside/min/cell. -Galactosidase activities ranged from 0 to 10 units. Activities of 1–2 and >2 units were considered as weak and strong interactions, respectively.
In Vitro Interaction Experiments Using Affinity Chromatography— Purified GST-His6-tagged COG proteins were bound to 100 µl (bed volume) of glutathione-Sepharose during 30 min of incubation at room temperature in buffer A. Unbound material was removed by three washes with buffer A, and the resin was distributed in 10-µl (bed volume) aliquots and further incubated with soluble (S1) fractions (20 µl) of T7-tagged COG1–8 for 2–3 h at 4 °C in buffer A. The resin was washed four times with buffer A, and protein complexes were eluted with 10 mM glutathione (pH 8.0). Sample buffer was added to the eluates, and proteins were separated on 11% SDS-polyacrylamide gel. Eluted proteins were analyzed both by Coomassie Blue staining and by Western blotting using anti-GST or anti-T7 tag antibodies and anti-Cog1–8p antibodies.
Analysis of Lysates and Cytosolic Versus Membrane Fractions of Yeast cog1–8 Mutants—Yeast cultures were grown in 200 ml of yeast extract/peptone/dextrose at 25 °C to A600 = 1–2. Cells were collected by centrifugation and washed twice with ice-cold distilled water. Cells were lysed in 1 ml of buffer containing 20 mM HEPES (pH 7.0) and 1 mM dithiothreitol, protease inhibitor mixture, and 1 mM phenylmethylsulfonyl fluoride using 1 ml of glass beads by vortexing (3 x 1 min) on ice. Cell debris was removed from lysates by centrifugation at 3000 x g for 5 min; the lysates were then transferred to Beckman ultracentrifuge tubes and centrifuged at 150,000 x g for 1 h to obtain the cytosolic and membrane fractions. Membrane fractions were resuspended in 1 ml of 20 mM Tris-buffered saline. Total lysates and equal volumes of cytosolic/membrane fractions were loaded onto SDS-polyacrylamide gel and analyzed by Western blotting with anti-COG antibodies.
Comparative Analysis of COG Subunit Stability at Various Temperatures in the cog1 Strain—Protease-deficient yeast wild-type and cog1 strains were grown in 20 ml of yeast extract/peptone/dextrose at 25 or 30 °C to A600 = 1.5. 200 µl of culture were pelleted, resuspended, and incubated in 1 ml of 20 mM NaOH for 20 min. Cells were pelleted after incubation, resuspended, and boiled in sample buffer. Proteins were resolved on 9% SDS-polyacrylamide gel and transferred to nitrocellulose membrane for Western blotting with anti-COG antibodies.
Gel Filtration of S100 Fractions—Gel filtration was performed essentially as described by Ram et al. (31) with some modifications. Yeast cultures were grown to exponential phase at 24 °C, after which 200 A600 units of cells were collected; disrupted by glass beads in 1.5 ml of 190 mM KCl, 25 mM Tris-Cl (pH 8.0), and 1 mM dithiothreitol (RK buffer); centrifuged at 3000 x g for 5 min to pellet unlysed cells; and then centrifuged again at 20,000 x g for 30 min. The supernatant was centrifuged at 100,000 x g for 1 h in a Beckman Coulter TLS55 rotor to yield the S100 supernatant. 200 µl of the S100 supernatant was applied to a Superose 6 HR 10/30 gel filtration column (Amersham Biosciences) pre-equilibrated with RK buffer. The column was run at 0.5 ml/min, and 1-ml fractions were collected. Aliquots were trichloroacetic acid-precipitated and analyzed by SDS-PAGE and immunoblotting. Protein levels were quantitated using ImageJ software.
Cog2p, Cog3p, and Cog8p Co-immunoprecipitation Experiment— Cog2p immunoprecipitations were done by a variation of the protocol that we used before (12). LY400 (wild type) and LY241 (cog1) cells were grown to A600 = 1.0. For each experiment, 30 A600 units were collected by centrifugation, washed with ice-cold distilled water, and resuspended in 1 ml of lysis buffer (20 mM HEPES, 1% Triton X-100, 150 mM NaCl, protease inhibitor mixture, and 1 mM phenylmethylsulfonyl fluoride). Cells were lysed with glass beads by vortexing (3 x 1 min) on ice. The following procedures were performed at 4 °C. Cell debris was removed by centrifugation at 3000 x g for 5 min. Lysates were centrifuged at 20,000 x g for 10 min. To remove nonspecifically bound proteins, supernatants were incubated with 100 µl of Sephadex G-25 (Amersham Biosciences) for 1 h; the Sephadex beads were pelleted; and precleared lysates were incubated overnight with 30 µl of anti-Cog2p antibodies. Lysates were centrifuged at 20,000 x g for 1 min to remove aggregated immune complexes and incubated for 1 h with protein A-Sepharose beads. The beads were washed five times with lysis buffer, and immune complexes were eluted with 40 µl of sample buffer and analyzed by Western blotting. To minimize nonspecific binding of anti-rabbit secondary antibodies to the denatured anti-Cog2p IgG antibodies, the rabbit IgG TrueBlot set (eBioscience, Inc., San Diego, CA) was used.
COG Complex Recovery by Tandem Affinity Purification (TAP) Pulldown with Human IgG-Agarose—Yeast cultures were grown and lysed as described previously (12) in the presence of 1% Triton X-100, 150 mM KCl, and 5 mM MgCl2. (For the TAP-Cog8p strain expressing the Cog1p fragment, cells were grown at 30 °C in yeast extract/peptone/dextrose or yeast extract/peptone/galactose.) Lysates were centrifuged at 20,000 x g for 10 min. The following procedures were all performed at 4 °C. To remove nonspecifically bound proteins, supernatants were incubated with 200 µl of Sephadex G-25 for 1 h; the Sephadex beads were pelleted; and the supernatant was incubated overnight with 30 µl of human IgG-agarose (Sigma). The beads were then washed four times with 20 mM HEPES (pH 7.0) and protease inhibitor mixture and boiled in 1x sample buffer. Eluted proteins were resolved on 9% SDS-polyacrylamide gel and transferred to nitrocellulose membrane for Western blotting with anti-COG antibodies. To minimize nonspecific binding of rabbit antibodies to the TAP tag, the membrane was blocked with purified human IgG (1 mg/ml) for 30 min.
|
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
A summary of the in vitro interactions from GST pull-down experiments is presented in Fig. 1C, illustrating the putative interacting subunits within either Lobe A or B. Two interaction groups were formed between COG3, COG2, and COG4 and between COG5, COG7, COG6, and COG8, leaving Cog1p standing aside because it demonstrated weak interactions with subunits from both groups. This was possibly due to the low solubility of bacterially expressed Cog1p (only 
10% was soluble), which resulted in relatively low concentrations of Cog1p in soluble fractions from bacterial lysates used in the binding assay. We therefore decided to further investigate the potential partners of Cog1p. To address this question, we coexpressed His-Cog1p and GST-Cog1–8p in E. coli and showed by GST pull-down that His-Cog1p interacted with GST-Cog2p, GST-Cog3p, GST-Cog5p, and, to a lesser extent, GST-Cog6p (Fig. 1D and supplemental Fig. S2).
The in vitro protein-protein interaction technique provided an essential start for our studies. Unfortunately, a number of COG subunits did not fold properly in bacteria, causing formation of insoluble inclusion bodies (for both Cog1p and Cog7p) and/or rapid protein degradation of expressed proteins (for both Cog4p and Cog6p).
COG Subunit Interactions in the Two-hybrid System—To overcome the difficulties encountered in the in vitro assay and to explore a more native cellular environment, we used the yeast two-hybrid system as a method of investigation to screen for the various interacting partners between the eight subunits of the complex. The respective COG1–8 cDNAs were subcloned into the yeast pGal4-AD and pGal4-BD expression vectors. The strength of interactions between the COG subunits was tested both by nutrition selection experiments and by determining the LacZ activity in liquid cultures of plasmid-transformed yeast. The results from the two-hybrid screening (Fig. 2) showed essentially three blocks of interaction: the first block between Cog1p and all subunits except Cog4p and Cog7p, the second block between Cog2p and Cog4p and between Cog2p and Cog3p, and the third block between Cog7p and Cog5p and between Cog7p and Cog6p. Because we examined interactions between subunits of a protein complex from yeast, we wondered whether "direct" interactions detected in the two-hybrid assay could actually be mediated by one or more endogenous subunits acting as bridge. We do not believe that this is the case; hence, in our studies, we did not detect any interactions between Cog5p and Cog6p, whereas both subunits interacted with Cog7p. Similarly, Cog3p did not show interaction with Cog8p, whereas both subunits interacted with Cog1p. These results were in good agreement with the results from our in vitro protein-protein interaction studies and supported the two-lobe model of the COG complex with Cog1p serving as a bridging subunit between the two lobes of the complex.
|
cog Mutants—Proteins assembled in a complex often require the presence of their partners for proper integration into the complex and for protection against various protease activities (43). To test whether deletion of an interacting partner would cause a subunit to be degraded, we analyzed and quantified the amounts of COG proteins in the total lysates and cytosolic versus membrane fractions of the cog1–8 mutants. We developed affinity-purified anti-COG antibodies to all COG subunits (Fig. 3A, Wt lane) and were therefore set to perform these comprehensive tests. Special attention was paid to the behavior of Cog1p and the cog1 mutant because it was suspected to interact with several proteins from both lobes of the complex. Cog2p, Cog3p, and Cog4p are essential for yeast growth; subsequently, for these experiments, we used yeast strains in which deletion of the corresponding COG subunit was rescued by overexpression of Ypt1p (13, 33) or expression of the pSLY1-20 plasmid (12).
In the first set of experiments performed with cells grown at 30 °C, we found that the protein levels of the Lobe B subunits were reduced in the cog1 mutant strain (Fig. 3A, -1 lane), indicating the importance of Cog1p for their stability. The levels of Cog1p were diminished in all the cog mutants, but most dramatically in cog2 (35% from the wild-type level), cog3 (13%), and cog4 (18%). Cog2p was almost undetectable in the cog4 mutant strain (20%), which correlates with in vitro data that Cog4p is a direct partner of Cog2p. Reciprocally, a reduction of Cog4p levels was found solely in the cog2 mutant (32%), again pointing to a direct protein-protein interaction. Cog5p was degraded in the cog1 mutant (<5%) as well as in the cog6 (12%) and cog7 (11%) mutants, where it was almost undetectable, and, to a lesser extent, in the cog3 mutant (54%). The cellular level of Cog6p was significantly affected by deletion of Cog1p (<5%) and was less affected by deletion of Cog5p (45%), Cog7p (56%), and Cog8p (33%). Significant depletion of Cog7p levels was observed in cog1 (<5%), cog5 (13%), and cog6 (17%). The most noticeable reduction of Cog8p was observed in the cog1 mutant (<5%) and also in the cog6 strain (22%).
We also noticed that the cellular level of COG subunits was slightly elevated in cog2 (rescued by pSLY1-20) and cog4 (LY314; rescued by pYPT1) or cog4 (LY586; rescued by pSLY1-20). Although the same stabilization effect was observed in the wild-type strain, expression of either Sly1-20p or Ypt1p did not change the protein level of individual COG proteins in other tested cog mutants (supplemental Fig. S3). Moreover, expression of the pSLY1-20 plasmid in the cog7 and cog4 mutant strains was insufficient to prevent degradation of Cog5p and Cog2p, respectively.
These in vivo results corroborated our observations made using the in vitro and two-hybrid systems. They further implicated Cog1p as a central subunit interacting with subunits from both lobes of the complex. We decided to test whether growth at lower temperatures in a protease-deficient strain could prevent the massive degradation of COG subunits observed in the cog1 mutant grown at 30 °C. In a parallel experiment in which both protease-deficient wild-type and cog1 cells were grown at different temperatures, Western blot analysis of the total lysates revealed no changes in COG protein levels in cog1 mutant cells grown at 25 °C and some decrease in Lobe B subunits at 30 °C (Fig. 3B).
Deletion of Individual Subunits Does Not Affect the Interaction of Other COG Subunits with Membranes—To determine how deletion of each subunit would affect the interaction of other COG subunits with membranes, we prepared total lysates and cytosolic and membrane fractions from cog mutant strains grown at 25 °C and analyzed them by Western blotting with antibodies against all eight COG subunits. Analysis of the fractions revealed that >90% of the COG proteins were found in pelletable (membrane) fractions in wild-type cells (Fig. 4). In addition, the COG complex comigrated with Golgi membranes on a sucrose/D2O flotation gradient (supplemental Fig. S4) and co-localized with the cis-Golgi marker Sed5p by immunofluorescence (supplemental Fig. S5). Surprisingly, no significant dissociation of individual COG subunits from the membranes in the cog mutants was observed. The redistribution of Cog7p to the soluble fraction (25%) was noticeable in the cog2 mutant, indicating the possible existence of a free cytoplasmic pool of Cog7p. Our data indicate that more than one subunit of the COG complex is responsible for interaction with membranes.
Stability of Sed5p and Protein Glycosylation Are Severely Impaired in the cog1,cog6 Double Mutant—We (12) and others (31) have observed previously that purification of the COG complex from yeast cytosol often results in recovery of the incomplete complex that contains all Lobe A subunits and Cog6p. This raised the possibility that Cog6p directly interacts with Lobe A most likely through Cog1p. To test this idea, we investigated COG complex stability and the resulting cellular defects in strains bearing either a single deletion of Cog1p or Cog6p or a double deletion of both Cog1p and Cog6p. Indeed, we found that the cellular levels of all remaining Lobe B subunits were diminished to greater degree in the cog1,6 double mutant strain compared those in cog1 and cog6 cells (Fig. 5A, 1, 6, and 1,6 lanes). The remaining subunits in Lobe A were unaffected by this double deletion. These results were quantified using ImageJ software and are presented in Fig. 5B. As for the integration of the remaining COG subunits into the membrane-bound complex in the cog1,6 mutant, we found that they remained membrane-bound despite the double deletion (Fig. 4, COG1,6 lanes). The cog1,6 double mutant was also characterized by a noticeable degradation of the COG complex partner target SNARE protein Sed5p (32% compared with wild-type levels) (Fig. 5C) (12) and by an impaired outer chain protein glycosylation of the secreted O-mannosylated heat shock protein HSP150 (Fig. 5D). Although glycosylation of the HSP150 protein was compromised in the cog1 mutant as well, the degree of impairment was greater in the cog1,6 mutant strain, where a putative unglycosylated form of the protein could be found (Fig. 5D, asterisk).
|
|
|
|
cog4 strain. We performed a native immunoprecipitation of the COG complex using affinity-purified antibodies to Cog2p (12). The level of Cog2p was drastically reduced in the cog4 strain (23% of Cog2p remained) (Fig. 7A, Total panel, cog4 lane). We also noted that only 30% of total Cog3p was extractable by Triton X-100 in cog4 cells. As a result of the reduced level of Cog2p, Cog3p recovery in the Cog2p immunoprecipitation was also reduced to 19% (Fig. 7A, Cog3p row, cog4 lanes). Notable, the recovery of Cog8p with anti-Cog2p antibodies (35%) was proportional to the level of Cog2p in the cog4 strain (Fig. 7B, Cog8p row, cog4 lanes). This shows that Cog4p is required for the stability of Cog2p, but is not essential for the interaction between Lobes A and B in the yeast COG complex.
Cog1p Is Essential for the Interaction of the COG Lobes—The proposed model of protein-protein interactions between the COG subunits predicted that deletion of Cog1p would affect the association of Lobes A and B of the complex. To test this hypothesis, we used the TAP tagging system (44), which has been previously used by us (12) and others (45, 46) to isolate a number of native protein complexes from yeast. For our experiments, we designed a cog1 (LY389) mutant expressing TAP-tagged Cog2p. In the eluates from human IgG-agarose incubated with the lysate of the wild-type strain, all eight COG subunits were detected by Western blot analysis with anti-COG antibodies (Fig. 8A, Eluates panel, Wt lane). As expected, the recovery of both Cog3p and Cog4p with TAP-Cog2p was not affected by deletion of Cog1p, whereas Lobe B subunits were almost undetectable in the eluates from the cog1 mutant lysate (Fig. 8A, Eluates panel, cog1 lane). The results show that Cog1p is required for the interaction of Lobes A and B.
It has been suggested previously that the N terminus of Cog1p is sufficient for its activity in the cell at most physiological temperatures (31). To test whether this part of the molecule is necessary for the interactions of COG subcomplexes, we transformed a TAP-Cog2p-expressing cog1 strain with a plasmid expressing either full-length Cog1p (pCOG1) or Cog1p without the first 80 amino acids (p80-COG1). In the eluates from human IgG-agarose incubated with the lysates of the pCOG1-transformed cog1 strain, we detected the presence of all eight COG subunits (Fig. 8A, Eluates panel, cog1 + pCOG1 lane). In the eluates from human IgG-agarose incubated with the lysates of the p80-COG1-transformed cog1 strain, the levels of Lobe B subunits were almost undetectable (Fig. 8A, Eluates panel, cog1 + p80-COG1 lane). The recovery of Cog3p and Cog4p was also affected, but to a much lesser degree. Although truncated Cog1p (Cog1–80p) was present in the lysates of the p80-COG1-transformed cog1 strain (Fig. 8A, Lysates panel, cog1 + p80-COG1 lane), its presence was undetected in the eluates. These results show that Cog1p can restore the interactions between the two lobes of the complex in a cog1 strain. They also demonstrate that the N terminus (first 80 amino acids) of Cog1p is essential for its proper bridging function.
|
|
cog1 cells. To maximally preserve the intactness of the COG complex, we prepared lysates of the logarithmically growing yeast cells by rapid glass bead disruption, removed membranes by centrifugation, and then loaded the resulting S100 supernatant onto the Superpose 6 column. This S100 supernatant was mostly depleted from the membranes and therefore contained <20% of the total cellular COG complex.
The wild-type S100 supernatant was applied to a Superose 6 gel filtration column, and fractions were analyzed by immunoblotting with antibodies to individual COG subunits. The major peak of Cog1–8p eluted from the Superose 6 column before the 670-kDa thyroglobulin marker (Fig. 8B), confirming the previously estimated approximate molecular mass of 800 kDa for the cytosolic COG complex (31, 33). We also found that 10–15% of the COG subunits eluted in a void volume fraction, most likely associated with small membrane fragments. The membrane protein Sed5p eluted in the same fraction (supplemental Fig. S6A). About 5% of both Cog5p and Cog7p were detected in fraction 16 before the 158-kDa marker, indicating the presence of the Cog5p/Cog7p dimer. A small amount of Cog7p was also detected in fraction 18, most likely representing the Cog7p monomer. Anti-Cog4p antibodies were not sensitive enough to detect Cog4p in fractions from the Superose 6 column.
|
cog1 cells gave a completely different distribution of COG subunits on the gel filtration column (Fig. 8C). The major pool of all COG subunits eluted in a void volume fraction and most likely associated with small membranes. The remaining COG proteins eluted in three major peaks: Cog6p and Cog8p were preferentially found in fraction 14; Cog2p and Cog3p peaked in fraction 15; and finally, Cog5p and Cog7p eluted in fraction 16. These results corroborated our data from crude fractionation assays (Fig. 4) and indicate that, in cog1 cells, the soluble COG complex is dissociated into at least three distinct subcomplexes: one containing Lobe A subunits Cog2p and Cog3p, the second representing protein partners Cog5p and Cog7p, and the third enriched with Cog6p and Cog8p. These data support our model of the protein-protein interactions between the eight subunits of the COG complex and directly demonstrate that Cog1p is essential for the stability of the soluble COG complex and association between Lobes A and B. |
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
In the in vitro approach, we have used affinity chromatography to investigate binary interactions between different COG subunits expressed in bacteria. We uncovered several interactions (Cog1p-Cog2p, Cog1p-Cog3p, Cog1p-Cog5p, Cog1p-Cog6p, Cog2p-Cog3p, Cog2p-Cog4p, Cog5p-Cog7p, Cog6p-Cog7p, and Cog6p-Cog8p). Three of these protein-protein interactions (Cog1p-Cog6p, Cog2p-Cog3p, and Cog2p-Cog4p) have been observed previously in the two-hybrid assay (33, 48–51), validating our in vitro approach. The data obtained from the in vitro binding experiments are in concordance with the previously proposed bilobed model of the mammalian COG complex observed by quick freeze/deep etch/rotary shadow electron microscopy (28). Several studies have shown that Cog1–4p are essential for growth (12, 13, 31, 33), whereas the absence of Cog5–8p does not exhibit a similar effect, and these proteins were therefore labeled as nonessential subunits (30). The interpretation of these results suggested that Lobe A (essential) would be composed of Cog1–4p and that Lobe B (nonessential) would contain Cog5–8p. The question as how these two lobes were connected together remained unanswered. The fact that only Cog1p demonstrated interactions with subunits from both lobes posed the question of whether it could be the "bridging" subunit. Indeed, Cog1p coexpressed together with individual COG subunits in bacteria demonstrated a strong interaction with subunits from both lobes (Fig. 1D). A comprehensive yeast two-hybrid test confirmed the results from the in vitro experiments. This test alleviated the difficulties caused by the improper folding and/or degradation of certain COG subunits expressed in bacteria and allowed us to identify new intra-COG interactions between Cog1p and Cog8p. Cog5p was shown to form homodimers in both in vitro binding and two-hybrid assays and thus may be responsible for formations of high hierarchy COG multimeric complexes.
To further characterize the protein-protein interactions in the COG complex, we have employed affinity-purified antibodies to individual COG subunits to determine COG protein levels in cog mutants. We hypothesized that the absence of one subunit would affect either the integration into the complex or the stability of the interacting partner. Indeed, we have found that Lobe A subunits were unstable in cog2, cog3, and cog4 mutants, whereas Lobe B subunit levels were significantly reduced in cog1 and cog5–8 mutants.
We have also found that the absence of one subunit did not significantly affect the membrane association of the others. This suggests that the association of the COG complex with the Golgi membrane is essential for its function and is achieved independently through several subunits that may interact with different membrane partners. Indeed, we demonstrated previously that the COG complex interacts with both SNAREs and the Rab protein Ypt1p (12). In addition, the interaction between Cog5p and phosphatidylinositol 4-monophosphate was demonstrated by probing the yeast proteome chip with phosphatidylinositol 4-monophosphate ammonium (52).
Based on results of our experiments, we have designed a model of direct protein-protein interactions among the subunits from Lobes A and B of the complex with Cog1p serving as a bridging subunit (Fig. 6). Indeed, we have found that, in the cog1 mutant, all subunits from Lobe B of the COG complex were severely degraded at 30 °C. These data are in good agreement with the previously detected instability of Lobe B subunits in mammalian cog1 mutant cells (53) and suggest that Cog1p is essential for the stability of the Lobe B subunits in both yeast and mammals. In cog1 mutant cells, Lobe B subunits were not coprecipitated with Lobe A. Interestingly, even at the low temperature, all COG subunits in cog1 mutant cells were preferentially membrane-bound. Small amounts of soluble subunits in this strain were found in three distinct subcomplexes: one containing Lobe A subunits Cog2p and Cog3p, the second enriched with Cog6p and Cog8p, and the smallest containing Cog5p and Cog7p. At 30 °C, the levels of Cog1p were diminished in all cog mutants, but more visibly so in cog2, cog3, and cog4 cells, denoting their importance for the stability of Cog1p and probably a direct interaction between them and Cog1p. We have also demonstrated that only a combined Cog1p-Cog6p deletion could affect the stability of the Lobe B subunits at 25 °C.
While this manuscript was in preparation, a model of the binary interacting network of the mammalian COG complex was described using the data obtained from the cotranslation of several COG proteins in the reticulocyte translation system (32). Despite the fact that the published data were about the mammalian COG complex, several similarities to our findings were noticed: notably, the interactions between Cog1p and Cog3p, Cog1p and Cog4p, Cog2p and Cog4p, Cog5p and Cog7p, Cog7p and Cog6p, and Cog6p and Cog8p. However, the proposed model suggested that Cog4p would act as the bridging subunit between the two lobes of the complex. In our studies of the yeast COG complex, we did not find any evidence supporting this model. Moreover, the fact that deletion of Cog1p in both the mammalian and yeast cells caused a massive degradation of the subunits from Lobe B suggests that Cog1p acts as the bridging subunit. Indeed, when we directly tested the interactions between the COG subunits in a cog1 mutant, we found that Lobes A and B were dissociated. Barely detectable amounts of Cog5p and Cog8p were still associated with Lobe A possibly through weak interactions between Cog5p and Cog2p (two-hybrid system) or between Cog8p and unidentified Lobe A subunit(s). Also, the slight dissociation of Cog4p from Lobe A points to the fact that Cog1p facilitates the integration of Cog4p into the complex. This confirms the proposed organizational model of the COG complex in which Cog1p serves as bridging subunit between the two lobes of the complex. It has been shown previously that a mutation in the COG1 gene (Sec36-1) causes severe growth defects and that the N terminus (first 198 amino acids) of Cog1p is sufficient for its activity in the cell at most physiological temperatures (31). This critical role of the N terminus of Cog1p triggered us to investigate in detail whether it is evolutionarily conserved in higher species because no homologies have been described previously. Surprisingly, we found that the first 85 amino acids of Cog1p from 23 species show some sequence homology (Fig. 9). This finding suggests that the N terminus of Cog1p is an important evolutionarily conserved domain involved in the proper functioning of Cog1p as a bridging subunit in the COG complex. Indeed, we have shown that overexpression of the Cog1p N-terminal domain interfered with stable association of Lobe A and B subunits (supplemental Fig. S7) and that deletion of this N-terminal domain abolished COG lobe association (Fig. 8A) and compromised Cog1p function (supplemental Fig. S8). Finally, we note that both the proposed key role of Cog1p and the yeast COG subunit interaction map agree quite well with a subunit interaction map, generated using different methods, for the mammalian COG complex.2 Further investigations consisting of the systematic analysis of interacting COG subunits will help identify the domains of interactions and test our proposed model of organization and protein-protein interactions of the yeast COG complex.
|
FOOTNOTES |
|---|
The on-line version of this article (available at http://www.jbc.org) contains supplemental Figs. S1–S8.
To whom correspondence should be addressed: Dept. of Physiology and Biophysics, University of Arkansas for Medical Sciences, Biomed 261-2, Mail Slot 505, 200 South Cedar St., Little Rock, AR 72205. Tel.: 501-603-1170; Fax: 501-686-8167; E-mail: vvlupashin{at}uams.edu.
1 The abbreviations used are: SNAREs, soluble N-ethylmaleimide attachment protein receptors; COG, conserved oligomeric Golgi; GST, glutathione S-transferase; TAP, tandem affinity purification.
2 D. Ungar and F. M. Hughson, personal communication.
|
ACKNOWLEDGMENTS |
|---|
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
-globulin, and 44-kDa ovalbumin size standards from the Superose 6 column and the void volume are indicated. C, shown are the profiles of COG proteins from 