QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

Vol. 16, Issue 10, 4918-4930, October 2005

This Article Abstract Full Text (PDF) All Versions of this Article: E05-02-0101v1 16/10/4918    most recent Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Weinberger, A. Articles by Gerst, J. E. Search for Related Content PubMed PubMed Citation Articles by Weinberger, A. Articles by Gerst, J. E.

Control of Golgi Morphology and Function by Sed5 t-SNARE Phosphorylation

Adina Weinberger ^*, Faustin Kamena , Rachel Kama ^*, Anne Spang , and Jeffrey E. Gerst ^*

^* Department of Molecular Genetics, Weizmann Institute of Science, Rehovot 76100, Israel; Friedrich Miescher Laboratory, Max Planck Society, D-72076 Tuebingen, Germany

Submitted February 7, 2005; Revised July 25, 2005; Accepted August 2, 2005
Monitoring Editor: Howard Riezman

	ABSTRACT

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION ACKNOWLEDGMENTS REFERENCES

 
Previously, we demonstrated that the phosphorylation of t-SNAREs by protein kinase A (PKA) affects their ability to participate in SNARE complexes and to confer endocytosis and exocytosis in yeast. Here, we show that the presumed phosphorylation of a conserved membrane-proximal PKA consensus site (serine-317) in the Sed5 t-SNARE regulates endoplasmic reticulum (ER)-Golgi transport, as well as Golgi morphology. Sed5 is a phosphoprotein, and both alanine and aspartate substitutions in serine-317 directly affect intracellular protein trafficking. The aspartate substitution results in elaboration of the ER, defects in Golgi-ER retrograde transport, an accumulation of small transport vesicles, and the inhibition of growth of most cell types. In contrast, the alanine substitution has no deleterious effects upon transport and growth, but results in ordering of the Golgi into a structure reminiscent of mammalian apparatus. This structure seems to require the recycling of Sed5, because it was found not to occur in sec21-2 cells that are defective in retrograde transport. Thus, a cycle of Sed5 phosphorylation and dephosphorylation is required for normal t-SNARE function and may choreograph Golgi ordering and dispersal.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION ACKNOWLEDGMENTS REFERENCES

 
The secretory pathway in eukaryotes consists of distinct membrane-bound intracellular compartments that transfer proteins and lipids in bulk from one to the other. This occurs via transport structures that undergo fusion with subsequent compartments. Because these compartments are in a constant state of flux, due to the continuous exchange of membrane and protein, obligate protein sorting and retrieval mechanisms have evolved to maintain organelle identity and integrity. Moreover, there is a specific need for recreating intact and functional organelles after mitosis, via organelle inheritance and/or biogenesis. The Golgi apparatus consists of well-defined subcompartments (cisternae) arranged in an ordered structure in higher eukaryotes, called the Golgi stack, which displays an asymmetric distribution of specific processing enzymes (reviewed in Farquhar and Palade, 1998). Cargo molecules traverse the stack via the process of cisternal maturation, which involves the retrograde flow of enzymes and transport factors back to earlier compartments while allowing the cargo to proceed (reviewed in Pelham, 1998; Pelham and Rothman, 2000). This results in steady-state distribution of the protein processing and delivery factors while allowing cargo to advance and eventually exit the Golgi via transport vesicles.

The fusion of membranes along the secretory pathway involves SNAREs, which comprise three main families of conserved membrane-associated proteins (the VAMP/synaptobrevin, syntaxin, and SNAP-25/light chain families) that mediate vesicle docking and fusion (reviewed in Chen and Scheller, 2001). Conventionally, these families fall into two categories, the v-SNAREs that reside on vesicles and t-SNAREs that reside on the acceptor compartments. Both v- and t-SNAREs assemble into a four-helix bundle that bridges apposed membranes and leads to membrane fusion (reviewed in Chen and Scheller, 2001). The v-SNARE contributes one -helix to the SNARE complex, whereas three are contributed by the t-SNAREs. Notably, every intracellular trafficking step has one syntaxin t-SNARE that serves as an essential element of the fusion complex. In yeast, the syntaxin family member, Sed5, plays an essential role in protein transport from the endoplasmic reticulum (ER) to the Golgi as well as intra-Golgi transport (Hardwick and Pelham, 1992, Nichols and Pelham, 1998). Sed5 forms functional SNARE complexes with Sec22, Bet1, and Bos1 (Sogaard et al., 1994; Parlati et al., 2002) to mediate ER-Golgi transport and with other SNAREs, such as Sft1, Ykt6, Gos1, and Vti1 to mediate intra-Golgi and endosome-Golgi transport (reviewed by Pelham, 1999). Attempts to systematically define the number of possible Sed5-containing SNARE complexes revealed that it is promiscuous and forms numerous complexes in vitro (Tsui et al., 2001); however, only two have been shown to be functional using an in vitro fusion assay (Parlati et al., 2002). Nevertheless, the large number of complexes that Sed5 forms in vitro may reflect its importance in the maintenance of Golgi structure and function.

Sed5 localizes to the cis-Golgi (Hardwick and Pelham 1992) and cycles through the ER (Wooding and Pelham, 1998). However, it is not known how Sed5 is retained at steady state in the Golgi and what physiological purpose is served by recycling. In an attempt to define its retention signal, mutations were made in the transmembrane domain (TMD), which is known to play a role in the retention of other SNAREs (Lewis et al., 2000). However, the Sed5 localization signal is only partially determined by its TMD (Banfield et al., 1994), and thus an additional localization/retention mechanism exists. In addition to regulating transport to the Golgi, Sed5 and its orthologues play an important role in Golgi maintenance and structure. The loss of Sed5 function is characterized by the accumulation of small transport vesicles and an elaboration of ER membranes concomitant with a decrease in protein transport and cell viability (Hardwick and Pelham, 1992). ER expansion is indicative of a block in the exit of proteins from the ER, which may result from an inhibition in retrograde transport from the Golgi. Interestingly, the overproduction of Sed5 is also inhibitory to cell growth, resulting in the accumulation of intracellular membranes and secretion of an ER resident protein (Hardwick and Pelham, 1992). Moreover, Sed5 orthologues have prominent effects upon Golgi structure. For example, the overproduction of Drosophila Sed5 results in the loss of Golgi stacks in COS cells (Banfield et al., 1994), whereas that of the NH₂-terminal sequence of mammalian Syntaxin-5 disrupts the Golgi in Vero cells (Yamaguchi et al., 2002).

We have demonstrated a role for t-SNARE phosphorylation in the regulation of exocytosis and endocytosis in yeast. Dephosphorylation of the Sso and Tlg t-SNAREs was found to enhance SNARE assembly and to restore transport in certain secretory mutants (Marash and Gerst, 2001; Gurunathan et al., 2002; Weinberger and Gerst, 2004). In our studies on the role of signaling cascades in the control of membrane transport, we have now examined whether Sed5 is controlled by phosphorylation. We found that Sed5 is a phosphoprotein that harbors a highly conserved PKA phosphorylation site proximal to the TMD. Amino acid substitutions in this site, serine-317, have dramatic effects upon Golgi morphology and function. Expression of pseudophosphorylated Sed5, shown using an aspartate substitution, results in the accumulation of ER and transport vesicles and an inhibition in cell growth. In contrast, expression of a nonphosphorylated form of Sed5, using an alanine substitution, results in the accumulation of an ordered Golgi atypical to Saccharomyces cerevisiae. This structure was absent in sec21-2 cells, which are defective in Golgi-ER and intra-Golgi retrograde transport, suggesting a role for Sed5 recycling in its ability to induce ordered structures. Supportive of this idea, we found that the nonphosphorylated, but not pseudophosphorylated, form of Sed5 readily entered into transport COPI vesicles. These results suggest that Sed5 phosphorylation and dephosphorylation may play a crucial role in Golgi function and structure. In particular, it may allow for the Golgi to cycle between ordered and dispersed states, the latter being important for Golgi inheritance during mitosis (Shorter and Warren, 2002).

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION ACKNOWLEDGMENTS REFERENCES

 
Growth Tests
Yeast were grown in standard amino acid-rich medium (YPD) or selective synthetic media (SC).

Drop Tests. Cells were grown to log phase in liquid selective medium. Next, cells were diluted to 10⁶ cells/ml, followed by five serial dilutions of 10-fold each. Aliquots of the dilutions were applied as drops onto solid media, which were then incubated at various temperatures.

Yeast Strains
Strains are listed in Table 1.

Plasmids
Constructs for SED5 Expression. SED5 was amplified by PCR using genomic DNA as a template and primers encoding SalI and SacI sites at the 5' and 3' ends, respectively. The SalI-SacI fragment was inserted in-frame and downstream to the hemagglutinin (HA) epitope encoded by vector pAD54 (2µ, LEU2), to yield pADH-HASED5. SED5 is under the control of the constitutive ADH1 promoter. Next, a BamHI fragment containing ADH1-HASED5 was excised from pADH-HASED5 and cloned into plasmids pRS315 (CEN; LEU2) to yield pLADH-HASED5, pSE358 (CEN; TRP1) to yield pTADH-HASED5, and pRS426 (2µ, URA3) to yield pADHU-HASED5. Point mutations were made in SED5 to create the alanine-317 and asparate-317 substitutions, by PCR-based site-directed mutagenesis with PfuI polymerase (Stratagene, LaJolla, CA). Multicopy plasmids bearing either LEU2 or URA3 and single-copy plasmids bearing either LEU2 or TRP1, all of which express SED5, SED5^S317A, or SED5^S317D were created. A gene encoding green fluorescent protein (GFP) was introduced in-frame and upstream to SED5 by subcloning a SalI GFP fragment into the SalI site of the LEU2 multicopy SED5 plasmids. This created HA-GFP fusions with SED5, SED5^S317A, or SED5^S317D. Next, the gene fusions were excised as BamHI fragments and cloned into the BamHI site of pRS315 to yield the same fusions in single-copy plasmids. Constructs for the integration of HA-GFP tagged SED5, SED5^S317A, or SED5^S317D at the native SED5 locus were created by subcloning into plasmid pFA6a-His3MX6 (Longtine et al., 1998), which encodes a PCR amplification module for integration at target genes. HindIII and BamHI fragments bearing the different GFP-tagged forms of SED5 were subcloned from the pAD54 plasmids into plasmid pFA6a-His3MX6 via the HindIII and BamHI sites. After verification by sequencing, plasmids were amplified using a 62-bp forward chimeric oligonucleotide corresponding to the 5' untranslated region of SED5 and downstream HA-GFP, and a 60-bp reverse oligonucleotide corresponding to the promoter region of the Schizosaccharomyces pombe his5+ gene and 3' untranslated region of SED5. PCR amplified fragments were used to transform SP1 wild-type cells and were selected for on medium lacking histidine. Integration at the SED5 locus was verified by PCR. A plasmid expressing Sec22-myc-, pWB-GalA, was provided by H-D. Schmidt (University of Gottingen, Germany).

Metabolic Labeling In Vivo
Protein Phosphorylation In Vivo. Proteins were metabolically labeled in vivo either with [³²P]orthophosphate or [³³P]orthophosphate (0.25 and 0.65 mCi/10 O.D.₆₀₀ U, respectively; GE Healthcare, Piscataway, NJ), essentially as described previously (Marash and Gerst, 2001).

Pulse-Chase Analysis. Intracellular protein processing was monitored by pulse-chase analysis using [³⁵S]methionine (GE Healthcare), as described previously (Couve et al., 1995).

Immunoprecipitation and Subcellular Fractionation
Immunoprecipitation. Coimmunoprecipitation from lysates was performed as described previously (Marash and Gerst, 2001). When performed in sec18-1 lysates, the cell pellet was first washed with 10 mM NaN₃, and both 20 mM NaF and 1 mM N-ethylmaleimide were added to the lysis buffer. For the detection of phosphorylated Sed5 in Westerns or by autoradiography, the lysis buffer was supplemented with the following phosphatase inhibitors: 10 mM NaF, 20 mM NaPPi, 25 mM -glycerophosphate, and 0.5 mM sodium vanadate. Proteins were detected in immunoblots by chemiluminescence.

Subcellular Fractionation. Yeast were subjected to subcellular fractionation, as described previously (Lustgarten and Gerst, 1999).

Antibodies
Monoclonal anti-HA antibodies (gift of M. Wigler, Cold Spring Harbor Laboratory, Cold Spring Harbor, NY) were used for both immunoprecipitation (IP) (1 µl) and detection (1:5000). Protein detection in blots was performed using monoclonal anti-c-myc (1:1000; Santa Cruz Biotechnology, Santa Cruz, CA) and polyclonal anti-phosphoserine (1:1000; Zymed Laboratories, South San Francisco, CA), anti-Sed5 (1:3000; gift of H. Pelham, MRC Laboratory for Molecular Biology, Cambridge, United Kingdom), anti-Bos1 (1:500), anti-Kar2 (1:3000; gifts of C. Barlowe, Dartmouth University, Hanover, NH), anti-Sec22 (1:3000; gift of S. Ferro-Novick, Yale University, New Haven, CT), anti-Vti1 (1:3000; gift of G. Fischer von Mollard, University of Gottingen); anti-Emp47 (1:3000; gift of H. Riezman, University of Geneva, Switzerland), anti-Dpm1 (1:1000; Molecular Probes, Eugene, OR), and anti-Mnn1 antibodies (1:2000; gift of S. Emr, University of San Diego, San Diego, CA). IP antibodies for pulse-chase experiments included anti-Gas1 (gift of H. Riezman) and anti-CPY (gift of S. Emr).

Microscopy
Confocal Microscopy. Cells were grown to log phase and concentrated to 1 O.D.₆₀₀ U/100 µl. Samples were mixed (1:1) with a cooled solution of 2.6% low melting point agarose in medium and plated on slides. GFP and red fluorescent protein (RFP; DsRed) fluorescence were visualized using a Radiance 2000 confocal system (Bio-Rad, Hercules, CA).

Immunofluorescence Microscopy. Cells were fixed and permeabilized for immunofluorescence, as described previously (Lustgarten and Gerst, 1999). HA-Sed5^S317A and HA-GFP-Sed5^S317A were labeled with affinity-purified anti-HA monoclonal antibodies and CY3-conjugated goat anti-mouse antibodies (Jackson ImmunoResearch Laboratories, West Grove, PA).

Electron Microscopy. Cells were concentrated by centrifugation, drawn into cellulose capillary tubes (200-µm inner diameter), and frozen in a Bal-Tec HPM010 HPF machine. For procedure A, cells were freeze-substituted in a Leica AFS device in anhydrous acetone containing 0.01% osmium tetroxide for 3 d at -90°C and then warmed to 0°C over 24 h. Samples were washed two times with ethanol and infiltrated for 5-7 d at room temperature (RT) in a series of increasing concentrations of LR-White in ethanol as follows: 5, 10, 30, 50, 60, 80%, and six exchanges of pure resin. After polymerization at 52°C, 60- to 80-nm sections were stained with uranyl acetate and lead citrate and examined in an FEI Tecnai T12 electron microscope at 120 kV. Alternatively, in procedure B, the frozen capillary tubes were freed from extraneous hexadecene under liquid nitrogen and freeze-substituted in 2% osmium tetroxide in anhydrous acetone at -90°C for 32 h, and at -60 and -30°C for 4 h at each step in a Balzers FSU010 freeze-substitution unit. After washing with acetone, samples were transferred into an acetone-Epon mixture at -30°C, infiltrated at RT in Epon, and polymerized at 60°C for 48 h. Ultrathin sections stained with uranyl acetate and lead citrate were viewed using a Philips CM10 electron microscope at 60 kV.

Golgi Budding Assay
In vitro Golgi budding was performed as described by Spang and Schekman (1998) with modifications. Enriched Golgi membranes were incubated with 0.1 mM GTP, coatomer (250 µg/ml), and Arf1 (80 µg/ml) for 30 min at 30°C in a volume of 200 µl. After chilling on ice, samples were loaded on top of a Ficoll-sucrose gradient consisting of 0.4 ml 60%(wt/wt) sucrose, 0.8 ml 7.5% (wt/wt) Ficoll, 1 ml of 5% Ficoll, 1 ml of 4% Ficoll, 1 ml of 3% Ficoll, 0.8 ml of 2% Ficoll in 15% sucrose, 20 mM HEPES, pH 6.8, and 5 mM Mg(OAc)₂. Vesicles were separated from the Golgi by centrifugation for 2 h at 35,000 rpm (SW55 rotor; Beckman Coulter, Fullerton, CA). Fractions (400 µl) were collected from the top of the gradient. Fractions 5-7 were pooled, mixed with an equal volume of 80% Nycodenz in 20 mM HEPES, pH 6.8, 150 mM KOAc, 5 mM Mg(OAc)₂ (B150), and overlaid with 600 µl of 30, 25, 20, and 15% and 400 µl of 10% Nycodenz in B150. The gradient was centrifuged for 16h at 40,000 rpm (SW55 rotor). Fractions (300 µl) were collected from the top, trichloroacetic acid precipitated, and analyzed in immunoblots.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION ACKNOWLEDGMENTS REFERENCES

 
Sed5 Is Phosphorylated In Vivo
PKA is involved in the regulation of the Sso and Tlg t-SNAREs, which confer exo- and endocytic transport in yeast, respectively (Marash and Gerst, 2001; Gurunathan et al., 2002). In the case of the Sso t-SNAREs, phosphorylation inhibits t-t SNARE assembly and recruits a SNARE regulatory protein, Vsm1, to the autoinhibitory domain of Sso (Marash and Gerst, 2001; Marash and Gerst, 2003). Correspondingly, activation of a ceramide-activated protein phosphatase (CAPP) reduces Vsm1 binding, restores t-t SNARE assembly, and confers exocytosis (Marash and Gerst, 2001; Marash and Gerst, 2003). Because phosphorylation and dephosphorylation play a significant role in t-SNARE regulation on the late secretory pathway, we examined whether ER-Golgi transport is controlled in a similar manner.

We examined whether temperature-sensitive mutations in SNAREs functioning in the early pathway might be rescued for growth at restrictive temperatures by the exogenous addition of C2-ceramide, a CAPP activator, to the medium. We found that a temperature-sensitive mutation in SED5 (e.g., sed5-1), which encodes an essential Golgi t-SNARE of the syntaxin family, was partially rescued, whereas mutations in other SNAREs involved in ER-Golgi transport (i.e., bos1-1, bet1-1, and sec22-1) were either not affected or were inhibited (our unpublished data). Thus, Sed5 might be subject to regulation by phosphorylation.

We noticed two putative PKA phosphorylation sites present in Sed5: serine-8 and serine-317. The NH₂-terminal phosphorylation site (serine-8) resides proximal to the Habc domain (Figure 1A), which has a suggested regulatory role (Yamaguchi et al., 2002). However, alignments of Sed5 orthologues revealed that only the COOH-terminal phosphorylation site (serine-317) is highly conserved from yeast to humans (Figure 1B). This site lies in the region directly downstream of the SNARE domain (residues 249-311) and proximal to the TMD (residues 325-339; Hardwick and Pelham, 1992) (Figure 1A).

View larger version (56K): [in this window] [in a new window]   Figure 1. Sed5 is phosphorylated in bos1-1 and wild-type cells. (A) Schematic of Sed5 structure. A, B, and C, -helices of the putative autoinhibitory domain; SNARE, SNARE domain; TM, transmembrane domain. An arrow marks the location of serine-317. (B) Sequence alignment of S. cerevisiae Sed5 with Syntaxin-5 orthologues from other species. Only the last 36-37 residues of the alignment are presented. (C) HA-Sed5 is phosphorylated in bos1-1 cells. HA-tagged Sed5 was expressed from a multicopy vector (pADH-HASED5) in bos1-1 cells. Cells were harvested in log phase and labeled with [32P]orthophosphate for 3 h, before being either shifted to 37°C for 15 min or maintained at 26°C. Cells were lysed in the presence of phosphatase inhibitors and HA-Sed5 was precipitated from lysates with anti-HA antibodies. Immunoprecipitated proteins were detected either in westerns with anti-Sed5 antibodies or by autoradiography (autoradiogram). Samples of the total cell extract (TCL) are shown beneath the IP results. (D) HA-Sed5 in bos1-1 cells is recognized by anti-phosphoserine antibodies. Same as in C, except that unlabeled bos1-1 cells were also transformed with a control plasmid (-). Immunoprecipitated proteins were detected in Westerns with either anti-Sed5 or anti-phosphoserine antibodies. (E) Endogenous Sed5 is phosphorylated in wild-type cells. Wild-type (SP1) cells were grown to log phase and labeled with [33P]orthophosphate for 3 h before lysis and immunoprecipitation with (+) or without (-) anti-Sed5 antibodies (2 µl/10 O.D.600 U). Immunoprecipitated proteins were detected in Westerns with anti-Sed5 antibodies or by autoradiography (autoradiogram). (F) Serine-317 is a bona fide phosphorylation site. Wild-type (SP1) cells overproducing HA-Sed5 or HA-Sed5S317A (plasmids pADH-HASED5 or pADH-HASED5S317A) were lysed and proteins precipitated using anti-phosphoserine antibodies (10 µl/10 O.D.600 U/reaction) and detected with anti-HA antibodies in blots. TCL, total cell lysate (50 µg of protein).

Because exogenously expressed HA-Sed5 is phosphorylated in bos1-1 cells, it was necessary to determine whether endogenous Sed5 also undergoes phosphorylation. We examined the phosphorylation of Sed5 expressed from its genomic locus in wild-type cells using [³³P]orthophosphate and immunoprecipitation with anti-Sed5 antibodies (Figure 1E). In Westerns detected with anti-Sed5, we observed two Sed5 bands that migrated closely (39 and 41 kDa, respectively); however, autoradiography indicated that only the higher band incorporated the radiolabel (Figure 1E). Similar results might have been observed for HA-Sed5 (Figure 1, C and D), but its overexpression may have impeded resolution of the two bands that correspond to the phosphorylated and nonphosphorylated forms. The results obtained from both bos1-1 and wild-type cells indicate that Sed5 is phosphorylated like the Sso and Tlg t-SNAREs (Marash and Gerst, 2001; Gurunathan et al., 2002).

An Aspartate Substitution at Position 317 of Sed5 Inhibits the Growth of Secretory Mutants
To explore the significance of Sed5 phosphorylation in vivo, we mutated serine-317 by substitution either with alanine to mimic the nonphosphorylated state or with aspartate to mimic the phosphorylated state. To verify that serine-317 is indeed phosphorylated, we immunoprecipitated HA-tagged Sed5 proteins from wild-type cells using anti-phosphoserine antibodies. Even though protein expression was similar, we found a large reduction in the amount of HASed5^S317A that could be precipitated relative to HASed5 (Figure 1F). This is consistent with the idea that serine-317 undergoes phosphorylation in vivo.

Next, we examined the growth of yeast overexpressing SED5 or the SED5 mutants. We found that both SED5^S317A and SED^5S317D were functional in terms of conferring growth to cells lacking the SED5 gene (Figure 2A). The loss of Sed5 function is lethal (Hardwick and Pelham, 1992), and strains expressing a galactose-inducible form of SED5 remained viable on glucose if they expressed native SED5 or either mutant (Figure 2A). Similar results were observed with sed5-1 cells, which could be rescued at restrictive temperatures (35°C) by overexpression of either mutant, although SED5^S317A expression was less effective in conferring growth (Figure 2B). Interestingly, we noted that the overexpression of SED5 (or SED5^S317D) was inhibitory to the growth of wild-type (WT) yeast, as described previously (Hardwick and Pelham, 1992). However, this effect was not observed with SED5^S317A.

View larger version (57K): [in this window] [in a new window]   Figure 2. Effect of alanine and asparate substitutions at position 317 in Sed5 on the growth of early secretory mutants. (A) Sed5 mutants rescue sed5 cells. sed5 cells bearing a galactose-inducible SED5 were transformed with single-copy plasmids expressing SED5, SED5S317A, and SED5S317D (plasmids pTADH-HASED5, pTADH-HASED5S317A, and pTADH-HASED5S317D, respectively). Cells were grown on galactose-containing medium to ensure SED5 expression and then either shifted to glucose-containing medium overnight or maintained on galactose, before serial dilution and plating. Control, cells transformed with a control plasmid. Cells were grown on plates for >2 days. (B) Sed5 mutants rescue sed5-1 cells. SED5, SED5S317A, and SED5S317D were expressed from multicopy plasmids (plasmids pADH-HASED5, pADH-HASED5S317A, and pADH-HASED5S317D, respectively) in either WT or sed5-1 temperature-sensitive cells. Cells were grown on glucose-containing medium, diluted and plated, and grown as described in A at 26 and 35°C. The plated wild-type transformants were grown for 36 h, whereas the sed5-1 transformants were grown for >48 h. (C) SED5, SED5S317A, and SED5S317D were expressed from multicopy plasmids in the temperature-sensitive mutants listed. Plasmids used to transform bos1-1, sec22-2, and sec23-2 cells are as listed in B, whereas plasmids pADHU-HASED5, pADHU-HASED5S317A, and pADHU-HASED5S317D were used for ufe1-1 cells. Cells were grown at permissive temperatures (26°C), diluted and plated, and grown at either permissive (26°C) or semirestrictive temperatures (32-35°C), which varied from strain to strain.

It would seem that the nonphosphorylated and pseudophosphorylated forms of Sed5 have very different effects upon cell growth. The effect upon ufe1-1 cells by SED5^S317D was of particular interest, because Ufe1 functions in retrograde transport and does not cycle between compartments like Bos1 and Sec22 (Lewis and Pelham, 1996). Because ufe1-1 cells are exquisitely sensitive to SED5^S317D, but not SED5^S317A, it suggested that phosphorylated Sed5 might affect retrograde trafficking.

An Aspartate Substitution at Position 317 of Sed5 Induces Retrograde Trafficking Defects
SED5 was identified as a multicopy suppressor of erd2 cells, which lack the HDEL receptor and thus are unable to retrieve soluble HDEL-tagged ER resident proteins to the ER (Semenza et al., 1990; Hardwick and Pelham, 1992). Overproduction of Sed5 in this mutant restored retrograde trafficking back to the ER, possibly due to the enhanced vesiculation of Golgi membranes (Hardwick and Pelham, 1992). In wild-type cells, however, SED5 overexpression causes Kar2 (Semenza et al., 1990) to be secreted, perhaps due to Erd2 depletion within the Golgi (Hardwick and Pelham, 1992). Because the phosphorylation state of Sed5 affects a strain (ufe1-1) known to be deficient specifically in retrograde trafficking (Figure 2C), we looked for further evidence of retrograde transport defects in WT cells overproducing Sed5 or the Sed5 mutants (Figure 3). We first examined whether cells overproducing these proteins secrete Kar2, a lumenal ER resident protein secreted from yeast bearing defects in retrograde transport (Semenza et al., 1990) or retention. We found that cells overexpressing either SED5 or SED5^S317D secreted considerable amounts of Kar2 onto filters (Figure 3A, left), whereas control cells or cells overexpressing SED5^S317A secreted little or no Kar2. Because cell lysis might also account for this phenomenon, we examined the filters for the presence of cytosolic proteins (e.g., Sec1 and hexokinase). However, we found no evidence to suggest that cells overexpressing SED5 or SED5^S317D are more labile (Weinberger and Gerst, unpublished observations). Moreover, no difference in Kar2 expression was detected in cells overexpressing SED5 or the mutants (Weinberger and Gerst, unpublished observations). Finally, wild-type yeast expressing either GFP-SED5 or GFP-SED5^S317D from the SED5 locus were examined for Kar2 secretion (Figure 3A, right). Importantly, we found that cells expressing GFP-SED5^S317D secreted Kar2 to a level similar to that of ufe1-1. In contrast, cells expressing GFP-SED5 (Figure 3A, right) did not secrete significant amounts of the protein. Thus, an aspartate substitution at serine-317 alone enhances Kar2 secretion. Therefore, the lack of Kar2 retention in these cells is independent of Sed5 overproduction.

View larger version (45K): [in this window] [in a new window]   Figure 3. Overexpression of Sed5 or Sed5S317D inhibits Golgi-ER retrograde transport. (A) Secretion of Kar2. Left, control (ufe1-1 and bet1-1) and wild-type (SP1) cells expressing SED5, SED5S317A, and SED5S317D from multicopy plasmids (pADHU-HASED5, pADHU-HASED5S317A, and pADHU-HASED5S317D) were grown on synthetic medium at 26°C, replica-plated onto nitrocellulose filters (BA-S 85; Whatman Schleicher and Schuell, Dassel, Germany), and grown for 24 h. Next, cells were washed off and the filters detected for secreted Kar2, using anti-Kar2 antibodies. Right, WT (SP1) cells bearing integrated (int) forms of GFP-tagged SED5 and SED5S317D (RKY1 and RKY3 cells, respectively), and control ufe1-1 cells were grown at 26°C, replica plated onto filters, and grown before detection with anti-Kar2 antibodies, as described above. (B) Gas1 processing. WT cells expressing SED5, SED5S317A, and SED5S317D from multicopy plasmids (pADH-HASED5, pADH-HASED5S317A, and pADH-HASED5S317D) were pulse-labeled with [35S]methionine for 3 min, and samples were taken at the indicated time points of the chase. Gas1 was precipitated from lysates using anti-Gas1 antibodies, resolved by SDS-PAGE, and detected by autoradiography. An asterisk indicates a cleavage product at 50 kDa. (C) Sec22-myc--factor processing in cells overexpressing Sed5 mutants. Cells overexpressing SED5, SED5S317A, or SED5S317D (plasmids pADH-HASED5, pADH-HASED5S317A, and pADH-HASED5S317D, respectively) were transformed with a plasmid expressing the Sec22-myc--factor fusion under a galactose-inducible promoter. The cells were grown to log phase on galactose-containing medium and transferred to glucose-containing medium. Aliquots were removed at the indicated times, lysed, and subjected to Western analysis. Sec22-myc--factor (Sec22-myc-) and Sec22-myc were detected using anti-myc antibodies, whereas Sed5 was detected using anti-HA antibodies. (D) Sed5 mutants form complexes with ER-Golgi SNAREs to the same extent as native Sed5. sec18-1 cells expressing HA-tagged SED5 (SED5), SED5S317A (S317A) or SED5S317D (S317D) from single copy plasmids (pLADH-HASED5, pLADH-HASED5S317A, and pLADH-HASED5S317D) were grown to log phase and either maintained at the permissive temperature (26°C) or shifted to the restrictive temperature (37°C) for 15 min, before processing for coimmunoprecipitation. Proteins were immunoprecipitated using anti-HA antibodies and detected in immunoblots with antibodies against Sed5, Bos1, Vti1, and Sec22. TCL, total cell lysate was detected with anti-Sed5 antibodies to demonstrate presence of HA-tagged and endogenous Sed5.

The processing of the vacuolar protease carboxypeptidase Y (CPY) is used to examine anterograde transport along the secretory pathway. The p1CPY precursor is synthesized in the ER, modified in the Golgi to the larger p2CPY form, and is transported to the vacuole were it is cleaved to yield the mature form (Stevens et al., 1982). We examined CPY processing in WT cells overexpressing SED5 or the mutants by pulse-chase analysis. Unlike Gas1 processing, however, we found that CPY processing and maturation was only slightly delayed in cells expressing SED5 or SED5^S317D (Weinberger and Gerst, unpublished observations). mCPY was observed within 30 min or less in all cell types; thus, the inhibition of transport exerted by SED5^S317D and native SED5 overexpression seems specific to certain cargo. This may explain why cells overproducing these proteins are growth inhibited but are not inviable (Figure 2). Because at least two classes of COPII vesicles mediate ER-Golgi transport (Muniz et al., 2001), it is possible that biogenesis or trafficking of the Gas1-containing class is more strongly influenced by Sed5 phosphorylation.

As anterograde transport of CPY is unaffected by SED5 overexpression, we looked for additional retrograde transport defects. We used a Sec22-myc--factor (Sec22-) reporter, which contains a Kex2 cleavage site downstream of the myc epitope (Ballensiefen et al., 1998). Cells defective in retrograde transport (i.e., sec21-1 and ufe1-1) fail to retrieve Sec22- to the ER, resulting in Kex2-dependent cleavage upon reaching the trans-Golgi and subsequent degradation in the vacuole (Ballensiefen et al., 1998). We followed the rate of disappearance of Sec22- and its cleavage product in WT cells expressing SED5 or the mutants (Figure 3C). Sec22- was first expressed under the control of a GAL promoter overnight and then "chased" by transferring the culture to glucose-containing medium for up to 2 h. Rapid cleavage of Sec22- was observed in cells overproducing Sed5 or Sed5^S317D and within 1 h even the full-length protein disappeared. However, in control cells and more so in cells expressing Sed5^S317A, both the full-length protein and the Kex2 cleavage product were observed for up to 2 h. Quantification revealed that 12.3 and 7.2% of the initial Sec22- signal remained after 1 h in control and SED5^S317A-expressing cells, respectively. In contrast, cells overexpressing SED5 or SED5^S317D showed only 0.5 and 1.1% of their initial signal. This makes it likely that Sec22- retrieval to the ER is inhibited only in cells expressing a phosphorylatable form of Sed5. We note that Sec22- undergoes degradation even in control cells, because its overexpression probably saturates the basal retrieval machinery (Ballensiefen et al., 1998). Thus, from three independent assays we can see severe defects in retrograde transport from the Golgi to the ER, which are inflicted upon SED5 overexpression and sustained by the phospho-mimetic form SED5^S317D. In contrast, cells expressing the nonphosphorylated form of Sed5 at position 317 have no such defects.

Sed5^S317A and Sed5^S317D Assemble into SNARE Complexes
Because the phenotype of SED5 overexpression is abrogated by mutating serine-317 to alanine, we speculated that the mutant might be unstable and undergo degradation. This alone could prevent the defects in retrograde transport observed upon native SED5 overexpression. To test this, we performed pulse-chase analysis with [³⁵S]methionine and immunoprecipitated Sed5 proteins. However, we found that Sed5^S317A was no less stable than native Sed5 for up to 60 min and beyond (Weinberger and Gerst, unpublished observations).

View larger version (40K): [in this window] [in a new window]   Figure 4. Sed5S317A localizes to an enlarged Golgi compartment. (A) Localization of GFP-Sed5 and GFP-Sed5 mutants in WT cells. WT (SP1) cells were transformed either with multicopy (pAD54 based: pADH-HAGFPSED5, pADH-HAGFP-SED5S317A, or pADH-HAGFP-SED5S317D) or single-copy (pRS315 based: pLADH-HAGF-PSED5, pLADH-HAGFP-SED5S317A, or pLADH-HAGFP-SED5S317D) plasmids that express GFP-SED5, GFP-SED5S317A, or GFP-SED5S317D or with constructs that result in the integration of either GFP-SED5, GFP-SED5S317A, or GFP-SED5S317D at the genomic SED5 locus. The latter gene fusions are under the control of the endogenous SED5 promoter. Cells were grown to log phase, and samples were taken for confocal microscopy. (B) Sed5S317A labeling is identical to that of GFP-Sed5S317A, as shown by immunofluorescence. WT cells overexpressing HA-tagged Sed5S317A or HA-tagged GFP-Sed5S317A were grown to log phase, fixed for immunofluorescence, and visualized using affinity-purified anti-HA antibodies and fluorescence microscopy. (C) BFA induces the dispersal of the GFP-Sed5S317A labeled compartment and redistribution to the ER. ise1 cells transformed with a multicopy plasmid (pADH-HAGFPSED5S317A) expressing GFP-SED5S317A were grown to log phase, incubated either with 100 µg/ml BFA (+BFA) in medium or with medium alone (-BFA) for 15 min, and visualized by fluorescence microscopy. (D) GFP-Sed5S317A colocalizes with a trans-Golgi marker, Sec7-RFP. WT cells containing an integrated copy of SEC7-dsRED (SEC7-RFP) were transformed with multicopy plasmids expressing GFP-SED5, GFP-SED5S317A, or GFP-SED5S317D (plasmids pADH-HAGFPSED5, pADH-HAGFP-SED5S317A, or pADH-HAGFP-SED5S317D). Cells were grown to log phase and visualized by fluorescence microscopy. Merge shows the overlap between GFP-Sed5 and Sec7-RFP.

GFP-Sed5^S317A Labels an Exaggerated Brefeldin A (BFA)-dissociable Compartment
Because serine-317 is proximal to the TMD of Sed5 and given that Sed5 cycles between the ER and Golgi (Wooding and Pelham, 1998), it is possible that this residue could regulate Sed5 localization. To examine this, we generated GFP-Sed5 chimeras using native Sed5 and both the alanine and aspartate substitution mutants to follow their localization in vivo by fluorescence microscopy. All three GFP-Sed5 chimeras exerted similar effects as their nonchimeric Sed5 counterparts, when expressed in sed5-1 temperature-sensitive yeast (our unpublished data). Both GFP-Sed5 and GFP-Sed5^S317D had a random punctate distribution, typical of yeast Golgi markers (Figure 4A). Surprisingly, GFP-Sed5^S317A was distributed differently, occurring in large aggregates that are predominantly adjacent to the bud neck in wild-type cells. This was observed using both multicopy and single-copy expression plasmids as well as by genomic integration of GFP-SED5^S317A at the SED5 locus (Figure 4A) and was reconfirmed by immunofluorescence studies using antibodies against HA-tagged Sed5^S317A lacking the GFP fusion (Figure 4B).

Next, we verified that GFP-Sed5^S317A labels Golgi membranes and is not mislocalized to other cellular compartments. We examined GFP-Sed5^S317A labeling in erg6/ise1 cells that are permeable and sensitive to BFA, a fungal inhibitor that induces disassembly of the Golgi (Vogel et al., 1993; Chardin and McCormick, 1999). Surprisingly, we were unable to transform erg6/ise1 cells with plasmids expressing GFP-SED5 or GFP-SED5^S317D (our unpublished data) but were able to express GFP-SED5^S317A. GFP-Sed5^S317A labeling in this strain (Figure 4C, left) was similar to that seen in WT cells (Figure 4, A and B) and could be dispersed to the ER by treatment of the cells with BFA for 15 min (Figure 4C, right). Thus, the large compartments labeled by GFP-Sed5^S317A seem to represent an exaggerated Golgi apparatus.

View larger version (166K): [in this window] [in a new window]   Figure 5. Sed5S317A overexpression leads to an ordered Golgi stack, whereas Sed5S317D overexpression induces ER elaboration and vesicle accumulation. HA-tagged SED5, SED5S317A, and SED5S317D were overexpressed from multicopy plasmids (pADH-HASED5, pADH-HASED5S317A, and pADH-HASED5S317D) in WT cells. Cells were harvested in log phase, fixed by HPF, and visualized by thin sectioning and electron microscopy. (A) Control WT cell. (B) Cell overexpressing SED5. Black arrowheads indicate elaborated ER. (C) Cell overexpressing SED5S317A. White arrowhead indicates stacked Golgi-like structure. Inset shows example of ordered Golgi from a different cell. (D) Immunogold labeling of a cell overexpressing SED5S317A. White arrowheads indicate ordered Golgi. Inset shows magnification of one stack; note the presence of gold particles indicating the localization of HA-tagged Sed5S317A within the stack. (E) Cells overexpressing SED5S317D. Black arrowheads indicate ER elaboration. White arrowheads indicate clusters of small vesicles. Top left, cell prepared by HPF using procedure B; bottom left, magnification of the same cell. Bottom right, example of vesicle cluster from a different cell; top right, example of cell fixed by HPF using procedure A. Bars, 0.2 µm.

GFP-Sed5^S317A Expression Results in a Stacked Mammalian-like Golgi, Whereas GFP-Sed5^S317D Expression Results in Vesiculation and ER Elaboration
To examine the intracellular morphology of cells overproducing Sed5 and the Sed5 mutants, we used electron microscopy. We overexpressed SED5, SED5^S317D, and SED5^S317A in WT cells and performed rapid fixation using high-pressure freezing (HPF). Interestingly, the overexpression of SED5 and SED5^S317D led to cells having abundant and elaborated ER membranes distributed throughout the cytoplasm (Figure 5, B and E). In addition, clusters of small vesicles (40-50 nm) were readily apparent in cells overexpressing SED5^S317D (Figure 5E). Notably, similar clusters have been shown in cells depleted of Sed5 (Hardwick and Pelham, 1992).

More surprisingly, overexpression of SED5^S317A resulted in the appearance of stacked membranes reminiscent of the Golgi observed in higher eukaryotes (Figure 5, C and D). Immunogold staining of these thin sections using antibodies against the HA-tagged protein revealed the presence of Sed5^S317A within the stacks (Figure 5D). Thus, expression of both the nonphosphorylated and pseudophosphorylated forms of Sed5 has dramatic effects upon intracellular morphology in yeast.

Golgi Stacking Does Not Change the Distribution of Golgi Markers
To determine whether the structural changes elicited by the two forms of Sed5 alter protein distribution throughout the endomembrane system, we examined the effects of Sed5^S317A and Sed5^S317D expression by crude subcellular fractionation. We expressed HA-tagged forms of Sed5 and the mutants in WT cells and prepared two membrane fractions: a medium-speed pellet (P10) that is thought to contain the majority of the ER membranes and a high-speed pellet (P100) that includes the bulk of Golgi membranes. We then examined the distribution of various proteins, including Emp47, which localizes at steady state to the cis-Golgi, but recycles between the ER and Golgi (Lewis and Pelham, 1996) like Sed5 (Wooding and Pelham, 1998). We also examined the distribution of a medial-Golgi marker, Mnn1, and an ER marker, Dpm1. First, we noticed that native Sed5 localized to the P10 fraction, along with Dpm1, implying that the t-SNARE associates either with an ER-associated structure, perhaps the transitional ER, or that the cis-Golgi sedimented in this fraction (Figure 6A). Next, we found that the overproduced HA-tagged Sed5 and Sed5^S317D proteins were also mainly in the P10 fraction, unlike HA-tagged Sed5^S317A, which was evenly divided between the P10 and P100 fractions (Figure 6A). Emp47 and Mnn1 distributed primarily to the P10 fraction in cells overexpressing SED5 and SED5^S317D, unlike in control cells or cells expressing SED5^S317A where the Golgi markers were present in both the P10 and P100 (Figure 6A). Thus, cells bearing the nonphosphorylated form of Sed5, which seem to have an ordered Golgi (Figure 5), show no alteration in the distribution of Golgi markers vis à vis control cells expressing SED5 at native levels. In contrast, the redistribution of Golgi markers to the P10 fraction in cells overproducing either native Sed5 or its pseudophosphorylated form implies that t-SNARE phosphorylation can alter Golgi constituency.

View larger version (33K): [in this window] [in a new window]   Figure 6. SED5 and SED5S317D overexpression induces the redistribution of Golgi markers, but they themselves are retained in the Golgi. (A) SED5 and SED5S317D overexpression redistributes Golgi markers. HA-tagged Sed5, Sed5S317A and Sed5S317D were overproduced in WT cells from multicopy plasmids (pADH-HASED5, pADH-HASED5S317A, and pADH-HASED5S317D, respectively). Cells were lysed in a detergent-free buffer, and the lysates were subjected to differential centrifugation to obtain a medium-speed pellet (P10) and high-speed pellet (P100) (10,000 and 100,000 x g, respectively). Samples of equal volume from the fractions were solubilized with SDS (1% final concentration); resolved by SDS-PAGE; and detected in blots using anti-Mnn1, anti-Emp47, anti-Dpm1, and anti-HA antibodies. (B) Sed5S317A is efficiently incorporated into COPI vesicles. Cells overproducing HA-tagged Sed5, Sed5S317A, and Sed5S317D were grown to log phase; lysed in the absence of detergent; and a Golgi-enriched fraction prepared by differential centrifugation. The formation of COPI-coated vesicles was induced by the addition of purified coatomer, Arf1, ATP, and GTP. Vesicles were fractionated by density gradient centrifugation, and samples from the fractions were precipitated, resolved by SDS-PAGE, and detected in blots using anti-Sed5 and anti-Emp47 antibodies. Ctrl-Golgi, a sample (30 µg) of the Golgi membranes used as the source for vesicles.

The Nonphosphorylated Form of Sed5 Efficiently Incorporates into COPI Vesicles
Because Sed5^S317D overproduction inhibits retrograde transport and induces elaboration of the ER and the accumulation of vesicles, we examined its ability to enter into Golgi-derived COPI-coated vesicles thought to mediate retrograde transport. We used an in vitro Golgi budding assay (Spang and Schekman, 1998) to generate COPI vesicles in the presence of Arf1, GTP, and coatomer. COPI vesicles were generated from isolated Golgi and were fractionated by density gradient centrifugation (Figure 6B). Although Sed5^S317A was efficiently incorporated into vesicles along with Emp47, both Sed5 and Sed5^S317D were far less able to undergo packaging therein. This suggests that nonphosphorylated Sed5 may be better recruited into COPI vesicles and thus may undergo more efficient retrieval (either to the ER, within the Golgi, or both) in contrast to either the pseudophosphorylated or phosphorylatable forms of the t-SNARE.

Formation of Ordered Golgi Structures Requires Golgi-ER Retrograde Transport but Not the Yeast GRASP65 Homolog
Because Sed5^S317A induces the formation of an enlarged and ordered Golgi (Figures 4 and 5) and is readily able to enter into COPI vesicles unlike Sed5 and Sed5^S317D (Figure 6), it suggested that the ability of Sed5 to recycle, either to the ER or within the Golgi, might be an important influence upon Golgi morphology. To test this, we examined whether ordered structures could be observed in yeast defective in retrograde Golgi-ER transport. We examined the localization of GFP-Sed5^S317A in sec21-2 cells, which are defective in COPI-mediated retrograde transport at restrictive temperatures (Figure 7). Interestingly, we were unable to find the enlarged Golgi aggregates typically observed in cells overproducing GFP-Sed5^S317A at any temperature (Figure 4). Instead, the labeling of GFP-Sed5^S317A in sec21-2 cells (Figure 7) was similar to that observed for GFP-Sed5 in wild-type cells (Figure 4); 92% of cells (n = 100) expressing GFP-Sed5^S317A did not show Golgi aggregates. In contrast, the enlarged Golgi puncta were readily observed in sec23-2 cells, which bear defects in anterograde ER-Golgi transport, at permissive temperatures, whereas labeling of the ER was observed at the restrictive temperature, as expected (Figure 7). Thus, Sed5 recycling is likely to be important for the formation of ordered Golgi structures.

View larger version (46K): [in this window] [in a new window]   Figure 7. Retrograde Golgi-ER transport is required for formation of the enlarged Golgi compartment. Yeast strains (sec21-2, sec23-2, and grh1) were transformed with a plasmid overexpressing GFP-SED5S317A (pADH-HAGFPSED5S317A) and examined by confocal fluorescence microscopy. Cells were examined at a temperature permissive for growth (26°C) or were shifted to the restrictive temperature (37°C), before examination.

Next, we examined whether the yeast orthologue of GRASP65, a protein known to function in the postmitotic reassembly of the Golgi in mammalian cells (Shorter and Warren, 2002), is necessary for formation of an ordered Golgi. We overproduced GFP-Sed5^S317A in cells lacking GRH1, but we found no obvious defects in formation of the fluorescent puncta (Figure 7). This suggests that this orthologue is not necessary for Golgi stacking per se

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION ACKNOWLEDGMENTS REFERENCES

 
Because t-SNARE phosphorylation and dephosphorylation seem to play a significant role in regulation of the exo- and endocytic processes (Weinberger and Gerst, 2004), we examined whether ER-Golgi transport is controlled in a similar manner. Here, we show that an essential Golgi t-SNARE, Sed5, is a phosphoprotein and that phosphorylation controls the morphology and apparent functioning of the Golgi apparatus. Sed5 overproduction inhibits the growth of yeast (Hardwick and Pelham, 1992; Figure 2), which is recapitulated in an even stronger manner by the replacement of a conserved serine residue (serine-317) with aspartate (Figure 2, B and C). In contrast, this phenotype is abrogated by introduction of an alanine residue in place of the relevant serine (Figure 2, B and C). Thus, both phosphorylation and dephosphorylation of this residue seem to play a critical role in Sed5 functioning.

Although both the aspartate and alanine substitutions confer growth to yeast lacking a functional Sed5 (Figure 2), the aspartate-317 substitution results in a marked decrease in the growth of bos1-1, sec23-2, and ufe1-1 cells (Figure 2C). However, only ufe1-1 cells are specifically defective in Golgi-ER retrograde transport (Lewis and Pelham, 1996). Other postulated retrograde transport defects, such as Kar2 secretion, an inhibition of Gas1 processing, and the pronounced degradation of a Sec22--factor chimeric protein, were all observed in wild-type cells expressing SED5^S317D (Figure 3). Similar results were obtained with overproduced native Sed5, which can undergo phosphorylation, but not with the nonphosphorylatable alanine substitution at position-317 (Figure 3). Importantly, these transport defects are not consistent with a general deficiency in protein export from either the ER or the Golgi. For example, Kar2 and Sec22--factor either reach post-Golgi compartments or are exported (Figure 3, A and C). Next, GFP-Sed5 and GFP-Sed5^S317D give punctate labeling that is typical of the Golgi and not the ER (Figure 4A), indicating that they can exit early compartments. Finally, CPY maturation is not blocked in cells overexpressing Sed5^S317D (Weinberger and Gerst, unpublished observations). Although Gas1 maturation does seem inhibited (Figure 3B), this export defect has also been shown in cells defective in retrograde transport (Sutterlin et al., 1997). Thus, our observations implicate serine-317 phosphorylation as a potential regulator of Golgi-ER retrograde transport.

Supportive of this view is electron microscopy data showing ER elaboration and the accumulation of small clustered transport vesicles in cells expressing the aspartate-317 substitution (Figure 5E). Because a similar phenotype was observed in cells lacking Sed5 (Hardwick and Pelham, 1992), it suggests that constitutive phosphorylation of serine-317 can have an effect analogous to decreased Sed5 function in some assays. However, the overexpression of native SED5 also inhibited cell growth (Figure 2C) and induced both retrograde sorting defects and ER elaboration (Figures 3, A-C, and 5) as well as SED5^S317D, but it could not induce vesicle accumulation (Figure 5). Thus, although the phospho-mimetic mutant is clearly functional (Figures 2A and 3D), it can exert effects unlike that those of the native t-SNARE. Thus, the effects of SED5 and SED5^S317D overexpression are comparable but not necessarily identical. This may explain why the aspartate substitution mutant is more potent in some assays (Figures 2C, 3A, and 5). Although the mechanism for vesicle accumulation in cells expressing SED5^S317D is unclear, it suggests defects in either the fusion of Golgi-derived COPI vesicles with the ER or COPII vesicles with the Golgi. The former possibility is more likely, because defects in retrograde transport invariably block the budding of ER-derived transport vesicles and lead to the accumulation of ER membrane. More work will be required to verify the nature of these vesicles.

View larger version (35K): [in this window] [in a new window]   Figure 8. Model for Golgi morphology as a consequence of Sed5 phosphorylation. Sed5 in its nonphosphorylated (at position 317) state results in a stacked mammalian-like Golgi structure (Ordered Golgi). In contrast, upon phosphorylation at position 317 Sed5 induces the dispersal of Golgi markers, the accumulation of transport vesicles and ER, and inhibits cell growth (Dispersed Golgi). Thus, Golgi structure may alternate between structured and unstructured states as a consequence of t-SNARE phosphorylation.

Our other major observation is that the nonphosphorylated form of Sed5, Sed5^S317A, results in the accumulation of an ordered Golgi reminiscent of the stacked apparatus found in higher eukaryotes (Figure 5, C and D). Although cells expressing Sed5^S317A show no defects in protein trafficking or growth (Figure 2); nonetheless, the dispersed Golgi typical of S. cerevisiae seems absent. Thus, formation of an ordered Golgi from a disordered state seems to have no debilitating effects upon the growth of yeast per se. Indeed, other yeast, such as Pichia pastoris, favor an ordered state (Glick, 1996). The differences between these yeast are thought to result from the number of sites available for COPII vesicle budding, which bud from fixed sites in P. pastoris and throughout the ER in S. cerevisiae (Rossanese et al., 1999). This may be responsible for the formation of either polarized or scattered transitional ER, respectively, which eventually coalesce to form the Golgi (Rossanese et al., 1999). Our results imply that a lack of phosphorylation at position 317 of Sed5 induces an ordered state of the Golgi (Figure 5, C and D, and model in Figure 8). In contrast, pseudophosphorylation induces a dispersed state, if the vesicles that accumulate therein are COPI vesicles (Figure 5E and model in Figure 8). Thus, a dynamic cycle of t-SNARE phosphorylation and dephosphorylation may be necessary for the Golgi to maintain structural integrity even in S. cerevisiae. Changes in the ability of Sed5 to cycle (via phosphorylative control) can also be expected to play a role in the de novo formation of the early Golgi from the transitional ER (Rossanese et al., 2001).

How can the phosphorylation state of a SNARE influence organelle integrity so dramatically? One clue is that Sed5 acts upon multiple Golgi transport pathways, including anterograde transport to the cis-Golgi (using the Bos1, Bet1, and Sec22 SNAREs), intra-Golgi transport (using the Sft1, Got1, and Ykt6 SNAREs) as well as endosome-to-Golgi transport (using Tlg1, Vti1, and Gos1 SNAREs). Therefore, Sed5 localization must be controlled such that it maintains a steady-state distribution to the cis-Golgi (Hardwick and Pelham, 1992; Wooding and Pelham, 1998), but nevertheless can transverse the cisternae to interact productively with its other SNARE partners or to be retrieved to the ER.

We propose that Sed5 transport and retrieval through the Golgi is regulated by its phosphorylation state. The alanine substitution mutant, whose expression results in an ordered Golgi (Figures 4 and 5), and that efficiently incorporates into COPI vesicles (Figure 6B), implies that enhanced Sed5 retrieval back through the cisternae might induce stacking. This is supported by studies in sec21-2 cells, which are defective in retrograde transport, and do not allow for formation of ordered Golgi structures (Figure 7). However, we do note that Sed5^S317A colocalizes with Sec7, a trans-Golgi marker, implying a steady-state presence in the trans-Golgi (Figure 4D). This, as well, might account for the ordering phenomenon and therefore requires further study. It is note-worthy to add that stacking could also result from the alanine mutant independently of its enhanced packaging into COPI vesicles or retrograde transport. In contrast to Sed5^S317A, the aspartate substitution mutant, which is inefficiently incorporated into COPI vesicles (Figure 6B) and whose expression results in ER elaboration and vesicle accumulation (Figure 5), may reside too efficiently in the cis-compartment such that it inhibits retrograde transport events and therefore cell growth (Figures 2 and 3). The mechanism for Sed5 retention is also not well understood, but previous studies suggested that it is determined by the cytoplasmic domain and not by the TMD (Banfield et al., 1994), in contrast to other SNAREs such as Snc1 (Lewis et al., 2000). Serine-317, which is adjacent to the TMD, is therefore a strong candidate for a Sed5 Golgi retention signal upon phosphorylation.

Finally, we note that stacked Golgi-like structures have been observed in other secretory mutants of S. cerevisiae; however, these are seen at temperatures where Golgi export is blocked (i.e., in sec7, sec14, and ypt31 ypt32-1 cells; Novick et al., 1981; Franzusoff et al., 1991; Jedd et al., 1997), whereas the ordered structures observed in Sed5^S317A-expressing cells occur under conditions conducive for growth (Figures 2 and 5). Thus, ordering per se has no deleterious effects although the alanine mutant is not as effective for the rescue of sed5-1 cells (Figure 2B). An ordered organization of the Golgi, as seen in higher eukaryotes, may prove useful where ER exit sites are more tightly controlled, leading to the polarized accumulation of transitional ER (and subsequent Golgi coalescence). This may be of particular importance as cell size increases, allowing for more efficient transport between cisternae as well as to target compartments. The fact that a t-SNARE mutation can bring about such morphological changes suggests that a controlled mechanism for ordering and dispersal exists. This is likely to be important for Golgi assembly and disassembly during the mitotic cell cycle in higher eukaryotes, which also involves cycles of protein phosphorylation and dephosphorylation. In particular, the functioning of several Golgi structural proteins (i.e., GM130, GRASP65, and p115) during mitosis has been shown to be modulated by phosphorylation and in a manner that correlates with Golgi fragmentation and dispersal (reviewed in Rossanese and Glick, 2001; Shorter and Warren, 2002).

	ACKNOWLEDGMENTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION ACKNOWLEDGMENTS REFERENCES

 
We thank D. Banfield, C. Barlowe, S. Emr, G. Fischer von Mollard, B. Glick, M. Lewis, P. Novick, H. Pelham, H. Riezman, R. Schekman, H-D. Schmidt, G. Warren, and M. Wigler for the generous gifts of reagents. We also gratefully thank V. Shindler (Weizmann Institute of Science) and H. Schwarz (Max-Planck-Institute for Developmental Biology, Tübingen, Germany) for electron microscopy work. This work was supported by grants to J.E.G. from the Israel Science Foundation (374/02-1) and the M. D. Moross Institute for Cancer Research; A. S. from the Deutsche Forschungsgemeinshaft and Max Planck Gesellschaft; and to both J.E.G and A. S. by the Minna James Heineman/Minerva Foundation, Germany. J.E.G. holds the Henry Kaplan Chair in Cancer Research. A. S. is a European Molecular Biology Organization Young Investigator.

	Footnotes

 
This article was published online ahead of print in MBC in Press (http://www.molbiolcell.org/cgi/doi/10.1091/mbc.E05-02-0101) on August 10, 2005.

Address correspondence to: Jeffrey E. Gerst (jeffrey.gerst{at}weizmann.ac.il).

	REFERENCES

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION ACKNOWLEDGMENTS REFERENCES

 
Ballensiefen, W., Ossipov, D., and Schmitt, H. D. ((1998). ). Recycling of the yeast v-SNARE Sec22p involves COPI-proteins and the ER transmembrane proteins Ufe1p and Sec20p. J. Cell Sci. 111, , 1507-1520.