Endoplasmic Reticulum Exit of Golgi-resident Defective for SREBP Cleavage (Dsc) E3 Ligase Complex Requires Its Activity

Background: Proteolytic activation of fungal SREBP requires the five-subunit Golgi Dsc E3 ligase.

Results: Dsc1 is an active E3 ligase; loss of Dsc E3 ligase activity leads to ER localization of the Dsc complex.

Conclusion: ER exit of the Dsc E3 ligase requires E3 ligase activity.

Significance: This is the first example of enzyme activity-dependent protein sorting in the secretory pathway.

Abstract

Layers of quality control ensure proper protein folding and complex formation prior to exit from the endoplasmic reticulum. The fission yeast Dsc E3 ligase is a Golgi-localized complex required for sterol regulatory element-binding protein (SREBP) transcription factor activation that shows architectural similarity to endoplasmic reticulum-associated degradation E3 ligases. The Dsc E3 ligase consists of five integral membrane proteins (Dsc1–Dsc5) and functionally interacts with the conserved AAA-ATPase Cdc48. Utilizing an in vitro ubiquitination assay, we demonstrated that Dsc1 has ubiquitin E3 ligase activity that requires the E2 ubiquitin-conjugating enzyme Ubc4. Mutations that specifically block Dsc1-Ubc4 interaction prevent SREBP cleavage, indicating that SREBP activation requires Dsc E3 ligase activity. Surprisingly, Golgi localization of the Dsc E3 ligase complex also requires Dsc1 E3 ligase activity. Analysis of Dsc E3 ligase complex formation, glycosylation, and localization indicated that Dsc1 E3 ligase activity is specifically required for endoplasmic reticulum exit of the complex. These results define enzyme activity-dependent sorting as an autoregulatory mechanism for protein trafficking.

Introduction

Sterol regulatory element-binding proteins (SREBPs)² are membrane-bound, basic helix-loop-helix leucine zipper transcription factors that regulate lipid homeostasis in mammalian cells (1). Inactive SREBP is retained in the ER. Upon stimulation, SREBP is transported to the Golgi where its cytosolic N-terminal transcription factor domain is released by proteolysis, allowing it to enter the nucleus and activate transcription. Fission yeast Schizosaccharomyces pombe has two SREBP homologs, Sre1 and Sre2, and the SREBP pathway is conserved in S. pombe with two key differences. First, Sre1 is an oxygen-responsive transcription factor required for hypoxic adaptation, and low oxygen stimulates Sre1 activation (2, 3). Second, fission yeast SREBP proteolytic activation requires the Golgi-localized, multi-subunit Dsc E3 ligase (4, 5).

The Dsc E3 ligase is a Golgi complex consisting of five integral membrane proteins, Dsc1–Dsc5 (see Fig. 1A). Dsc1 is a candidate RING E3 ligase homologous to Saccharomyces cerevisiae Tul1, which functions in Golgi quality control and uses Ubc4 as its cognate E2 ubiquitin-conjugating enzyme (5, 6). Dsc2 is a homolog of mammalian UBAC2, containing an N-terminal rhomboid pseudoprotease domain and a C-terminal ubiquitin-associated (UBA) domain (7, 8). Dsc5 is a UBX domain-containing protein with homology to mammalian UBXD8 (4), whereas Dsc3 and Dsc4 are less well characterized. Importantly, deletion of any dsc gene, mutation of the Dsc1 RING domain, or conditional inactivation of ubc4 prevents proteolytic activation of SREBP, establishing a genetic requirement for the Dsc E3 ligase in this pathway (4, 5, 9).

FIGURE 1.

Dsc1 is an active RING E3 ligase. A, model of Dsc E3 ligase complex. Black lines designate interactions among complex subunits, and functional domains are indicated in parentheses. B, S. pombe Dsc1 contains a C-terminal RING domain homologous to the S. cerevisiae Tul1 RING domain. Zinc-coordinating residues (*) and E2-interacting residues (●) are marked. Conserved residues Cys-634, Ile-636, His-668, and Leu-675 (boxed) were mutated to assay RING function. C, Coomassie Blue-stained gel of purified S. pombe E1 Uba1 and E2 Ubc4. D, purified, recombinant GST-RING, GST-RING mutants, or GST were mixed with E1 Uba1, E2 Ubc4, ubiquitin (Ub), and ATP and incubated for 1 h at 37 °C. Auto-ubiquitination of the GST-RING was assayed by immunoblotting.

The Dsc E3 ligase complex shares an overall architecture with mammalian gp78, an ER-localized, E3 ligase that functions in ER-associated degradation (8, 10). How these related enzymes achieve their different subcellular localizations is unknown. Here, we demonstrate that Dsc1 is a functional E3 ligase and that Dsc1 E3 ligase activity is required for ER exit of the ligase complex, thus establishing enzyme activity-dependent sorting as a new mechanism for protein trafficking.

Experimental Procedures

Materials

We obtained all fine chemicals from Fisher. Edinburgh minimal medium and amino acids were from Q-Biogen; oligonucleotides were from Integrated DNA Technologies; GelCode Blue, West Pico ECL, and Dura ECL kits were from Pierce; and HRP-conjugated, goat anti-rabbit, and goat anti-mouse antibodies were from Jackson ImmunoResearch Laboratories. Protease inhibitors were from Sigma (1000× stock; leupeptin (10 mg/ml), aprotinin (28 trypsin inhibitor units/ml), pepstatin A (5 mg/ml), and PMSF (0.5 m)). Protein sequence homology analysis was performed using ExPASy LALIGN.

Media and Growth Conditions

S. pombe cells were grown in YES (5 g/liter yeast extract, 30 g/liter glucose) or Edinburgh minimal medium, both supplemented with 225 μg/ml each of uracil, adenine, leucine, histidine, and lysine at 30 °C. Temperature-sensitive strains were grown at 25 °C and transferred to non-permissive 36 °C for the indicated times. For anaerobic growth, exponentially grown cells were pelleted, transferred to an Invivo₂ 400 work station (Biotrace), and resuspended in deoxygenated medium. Anaerobic conditions were maintained by 10% hydrogen gas with balanced nitrogen in the presence of palladium catalyst. Media were deoxygenated by incubation in an Invivo₂ 400 work station for at least 24 h.

Strains, Plasmids, and Protein Expression

Yeast manipulations and transformations were done using standard protocols. dsc1-integrating plasmid (pES215) (5) was subjected to site-directed mutagenesis to generate pES216 (dsc1-C634A), pES217 (dsc1-H668A), pSR14 (dsc1-I636D), and pSR15 (dsc1-L675D). These plasmids were linearized and transformed to strains for integration at the his3 locus. Transformants were selected for growth on Edinburgh minimal medium lacking histidine. Correct plasmid integration was confirmed by DNA sequencing of PCR-amplified genomic DNA, and protein expression was checked by Western blotting. ESY462 (h+ leu1-32 ura4-D18 ade6-M216 his3-D1::dsc2-6×GFP-his3⁺ Δdsc2-D1::kanMX6) was generated using dsc2-6×GFP plasmid (5). Strains expressing Dsc2-6×GFP and mCherry-tagged Ost1 were generated by standard mating and random spore analysis using ESY462 as a parent strain. Wild-type Dsc1 RING domain (aa 608–695) and the corresponding mutants (C634A, H668A, I636D, and L675D) were PCR-amplified from dsc1 integrating plasmids and cloned into XbaI and HindIII sites of pGEX-KG as N-terminal GST fusion proteins. Recombinant wild-type and mutant Dsc1 RING domains were expressed in Escherichia coli BL21 Codon Plus® (DE3)-RIPL (Stratagene), induced with 0.4 mm isopropyl-1-thio-β-d-galactopyranoside for 4 h and purified using GST-agarose beads (Sigma Aldrich). Briefly, E. coli cell pellets expressing the recombinant proteins were lysed by sonication in phosphate-buffered saline (pH 7.4) containing 1× protease inhibitors, 1% Triton X-100, and 0.1% Nonidet P-40. The lysate was cleared by centrifugation at 25,000 × g for 20 min and allowed to bind to GST-agarose beads for 1 h at 4 °C. Bound GST fusion proteins were eluted with 10 mm reduced glutathione. The purity of the protein was checked by SDS-PAGE and staining with GelCode Blue. S. pombe strains are listed in Table 1.

TABLE 1.

S. pombe strain list

The ade6-M21? allele is either ade6-M210 or ade6-M216. h? indicates that the mating type is either h⁻ or h⁺.

S. pombe strains	Genotype	Reference or source	Figure
KGY425	h− leu1-32 ura4-D18 ade6-M210 his3-D1	ATCC	2, 3, 4, 7
PEY522	h− leu1-32 ura4-D18 ade6-M210 his3-D1Δsre1-D1::kanMX6	Hughes et al. (3)	2
PEY1474	h− leu1-32 ura4-D18 ade6-M21? his3-D1 dsc1-1	Stewart et al. (5)	2, 3
PEY1448	h− leu1-32 ura4-D18 ade6-M210 his3-D1 Δdsc1-D1::kanMX6	Stewart et al. (5)	2, 3, 4, 7
PEY1747	h? leu1-32 ura4-D18 ade6-M210 his3-D1::dsc1-his3+Δdsc1-D1::kanMX6	This study	2, 3, 4
PEY1743	h− leu1-32 ura4-D18 ade6-M210 his3-D1::dsc1L675D-his3+Δdsc1-D1::kanMX6	This study	2, 3, 4, 7
PEY1742	h− leu1-32 ura4-D18 ade6-M210 his3-D1::dsc1C634A-his3+ Δdsc1-D1::kanMX6	This study	2, 3, 4,
PEY1547	h− leu1-32 ura4-D18 ade6-M210 his3-D1Δdsc2-D1::kanMX6	Stewart et al. (5)	3, 4, 7
PEY1567	h− leu1-32 ura4-D18 ade6-M216 his3-D1 ubc4-P61S::ura4+	Seino et al. (23)	3, 4
PEY1078	h− leu1-32 ura4-D18 ade6-M216	Matynia et al. (27)	3
PEY1079	h− leu1-32 ura4-D18 ade6-M210 sar1-1	Matynia et al. (27)	3
PEY1735	h− leu1-32 ura4-D18 ade6-M216 his3-D1::dsc2-6 × mGFP-his3+ Δdsc2-D1::kanMX6 ost1-mCherry::ura4+	This study	5
PEY1736	h− leu1-32 ura4-D18 ade6-M216 his3-D1::dsc2-6 × mGFP-his3+ Δdsc1-D1::kanMX6 Δdsc2-D1::kanMX6 ost1-mCherry::ura4+	This study	5
PEY1737	h− leu1-32 ura4-D18 ade6-M216 his3-D1::dsc2-6 × mGFP-his3+ Δdsc3-D1::kanMX6 Δdsc2-D1::kanMX6 ost1-mCherry::ura4+	This study	5
PEY1738	h− leu1-32 ura4-D18 ade6-M216 his3-D1::dsc2-6 × mGFP-his3+ dsc1-1::kanMX6 Δdsc2-D1::kanMX6 ost1-mCherry::ura4+	This study	5
PEY1739	h+ leu1-32 ura4-D18 ade6-M216 his3-D1::dsc2-6 × mGFP-his3+ Δdsc4-D1::kanMX6 Δdsc2-D1::kanMX6 ost1-mCherry::ura4+	This study	5
PEY1740	h−leu1-32 ura4-D18 ade6-M216 his3-D1::dsc2-6 × mGFP-his3+ Δdsc5-D1::kanMX6 Δdsc2-D1::kanMX6 ost1-mCherry::ura4+	This study	5
PEY1741	h− leu1-32 ura4-D18 ade6-M216 his3-D1::dsc2-6 × mGFP-his3+ Δdsc2-D1::kanMX6 ubc4-P61S::ura4+	This study	6
PEY1746	h− leu1-32 ura4-D18 ade6-M216 his3-D1::dsc2-6 × mGFP-his3+ Δdsc2-D1::kanMX6	This study	6
PEY854	h+ leu1-32 ura4-D18 ade6-216 anp1-mCherry::ura4+	Vjestica et al. (39)	7
PEY1744	h− leu1-32 ura4-D18 ade6-216 anp1-mCherry::ura4+ Δdsc1-D1::kanMX6	This study	7
PEY1745	h+ leu1-32 ura4-D18 ade6-216 anp1-mCherry::ura4+ Δdsc2-D1::kanMX6	This study	7
PEY1761	h? leu1-32 ura4-D18 ade6-M216 his3-D1::dsc2-6 × mGFP-his3+ Δdsc1-D1::natMX6 Δdsc2-D1::kanMX6 anp1-mCherry::ura4+	This study	7

S. pombe uba1 (SPBC1604.21c) and ubc4 (SPBC119.02) were PCR-amplified from a cDNA library (11) and cloned into XhoI and NheI sites of pET28a (Novagen) as N-terminal His₆ fusion proteins. The plasmid harboring full-length uba1 was transformed into E. coli Rosetta DE3 competent cells (EMD Millipore) and induced overnight with 0.2 mm isopropyl-1-thio-β-d-galactopyranoside at 22 °C in the presence of 2% ethyl alcohol. The plasmid containing full-length ubc4 was transformed into E. coli BL21 Codon Plus® (DE3)-RIPL (Stratagene) and induced with 0.5 mm isopropyl-1-thio-β-d-galactopyranoside for 4 h. In both cases, E. coli cell pellets were lysed by sonication in buffer A (50 mm Na₂HPO₄ pH 8.0, 300 mm NaCl), and cell debris was removed by centrifugation at 20,000 × g for 20 min. The supernatant was allowed by bind to nickel-nitrilotriacetic acid-agarose beads (Qiagen) and washed with buffer A containing 20 mm imidazole. Bound protein was eluted with buffer A containing 250 mm imidazole. Protein purity was checked by SDS-PAGE and GelCode Blue.

Antibodies

Polyclonal antibodies against Sre1, Dsc1, Dsc2, Dsc3, Dsc4, and Dsc5 used in this study were described previously (3, 5). Dsc1 antiserum was affinity-purified by passing through a column containing Dsc1 N terminus (aa 1–300) coupled to agarose beads (AminoLink Plus immobilization kit, Thermo Scientific) according to the manufacturer's protocol. HRP-conjugated Dsc antibodies were prepared using the EZ-link Plus activated peroxidase kit (Thermo Scientific) following the manufacturer's instructions. Anti-GST antibody was from Covance, and anti-ubiquitin (P4D1) and anti-FLAG (M2) were from Sigma.

In Vitro Ubiquitination Assay

In vitro ubiquitination assays were performed as described (12). Briefly, purified S. pombe E1 (Uba1), E2 (Ubc4), wild-type Dsc1 RING domain (aa 608–695), and Dsc1 mutants (C634A, H668A, I636D, and L675D) were used in this assay. Purified RING domain (500 nm) was mixed with 125 nm E1, 400 nm E2, and 5 μm ubiquitin (S. cerevisiae, Boston Biochem) in the presence of 2 mm ATP in ubiquitination buffer (50 mm Tris-HCl, pH 7.4, 5 mm MgCl₂, 2 mm NaF, and 0.6 mm DTT). The reaction was incubated at 37 °C for 1 h and stopped by adding equal amount of 2× SDS-DTT buffer (100 mm Tris-HCl, pH 6.8, 4% SDS, 20% glycerol, 200 mm DTT, 0.2% bromphenol blue). After boiling for 5 min, proteins were separated using 12% SDS-PAGE, ubiquitination was detected using anti-ubiquitin antibody (P4D1), and expression of the proteins was determined using anti-GST antibody.

Sre1 and Sre2 Cleavage Assays

Sre1 and Sre2 cleavage assays were performed as described previously (3, 9). Briefly, for Sre1 cleavage, cells were grown to exponential phase under normoxic conditions and then shifted to anaerobic chamber (Invivo₂ 400 work station) for the indicated times. Cells were collected and washed, and lysates were prepared under denaturing conditions as described (3). Processing of Sre1 was detected by Western blot analysis using anti-Sre1 antibody. Anti-Dsc5 antibody was used as a loading control. For quantification purposes, blots were developed using IRDye secondary antibodies and an Odyssey CLx imager from LI-COR. Signals for Sre1 were normalized to Dsc5, and data were quantified from three independent experiments using Image Studio software from LI-COR.

For Sre2 cleavage assays, strains expressing 3×FLAG-Sre2 (aa 423–793) from the cauliflower mosaic virus promoter (9) were grown at 30 °C to exponential phase, and whole cell lysates were assayed by immunoblotting using anti-FLAG antibody. Immunoblots were developed using chemiluminescence except where noted in the figure legends.

Deglycosylation Assay

Cells (5 × 10⁸) were grown to exponential phase and collected by centrifugation at 500 × g for 5 min. Cell pellets were washed once and resuspended in B88 buffer (20 mm HEPES, pH 7.2, 150 mm KOAc, 5 mm Mg(OAc)₂, 250 mm sorbitol) and supplemented with 1× protease inhibitors and with an additional 1× Complete, EDTA-free protease inhibitor (Roche Diagnostics). Cell lysis was performed by vortexing with glass beads (Sigma, 400–600 μm) for 10 min, and debris was removed by centrifugation at 500 × g. The resulting supernatant was subjected to 20,000 × g centrifugation for 20 min. The membrane pellet was resuspended in B88 buffer containing 1% Nonidet P-40 and nutated at 4 °C for 1 h followed by centrifugation at 20,000 × g for 20 min. Supernatant was collected and used as Nonidet P-40-solubilized membrane. Solubilized membrane proteins (30 μg) were denatured by adding SDS and 2-mercaptoethanol to final concentrations of 0.5 and 1%, respectively, in 15 μl and heating at 37 °C for 30 min. Peptide-N-glycosidase F (500 units) (New England Biolabs) was added to the denatured membrane protein and incubated at 37 °C for 1 h. The reaction was stopped by the addition of 5× SDS-PAGE loading dye (150 mm Tris-HCl, pH 6.8, 15% SDS, 25% glycerol, 0.02% bromphenol blue, and 12.5% 2-mercaptoethanol) and heating at 37 °C for an additional 30 min. Dsc1 mobility was assayed by immunoblotting using affinity-purified Dsc1 antiserum.

Cycloheximide Treatment

Temperature-sensitive sar1-1 or ubc4-P61S strains were grown overnight at permissive 25 °C to exponential phase. Cycloheximide (100 μg/ml) was added to the cells and shifted to non-permissive 36 °C for the indicated times before harvesting. Cell membranes were processed as described above or treated cells were directly used for live cell microscopy.

Microscopy

S. pombe cells expressing fluorescently tagged proteins were immobilized on 2% agarose as described previously (5) and imaged using a Marianas/Yokogawa CSU22 spinning disk confocal microscope (3i) equipped with Axio Observer (Zeiss), Photometrics Cascade II EM-CCD camera (Roper Scientific), and environmental chamber (Tokai HIT). Cells were imaged at room temperature, 25 °C, or 36 °C as indicated in the figure legends. Images were captured using 100× oil objective with 1.0 NA. We used SlideBook 5.0 for data acquisition. Fifteen Z-images (0.34-μm step size) were collected, and three-dimensional reconstitution of the confocal slices was performed by ImageJ (National Institutes of Health). All the images described in this study were acquired and processed in an identical manner.

Co-immunoprecipitation

Co-immunoprecipitation assays were performed as described previously (8). Briefly, cells (6 × 10⁸) from exponentially growing wild-type or mutant strains were collected and lysed using glass beads in digitonin lysis buffer (50 mm HEPES, pH 6.8, 1% (w/v) digitonin, 50 mm KOAc, 2 mm Mg(OAc)₂, 1 mm CaCl₂, 200 mm sorbitol, 1 mm NaF, 0.3 mm Na₃VO₂, supplemented with 1× protease inhibitors). Cellular debris was removed by centrifugation at 100,000 × g for 10 min, and 0.75 mg of lysate supernatant was incubated with 5 μl of Dsc2 antiserum for 15 min at 4 °C followed by the addition of 40 μl of protein A-agarose beads. Binding reaction was rotated overnight at 4 °C, and unbound proteins were removed by three washes with digitonin lysis buffer. The bound fraction was eluted by boiling 5 min in SDS-lysis buffer (10 mm Tris-HCl, pH 6.8, 100 mm NaCl, 1% SDS and 1 mm EDTA) and analyzed by Western blotting using HRP-conjugated Dsc antibodies.

Results

The fission yeast Golgi Dsc E3 ligase complex contains five integral membrane proteins, Dsc1–Dsc5. Bioinformatic analysis identified S. pombe Dsc1 as a homolog of the S. cerevisiae Golgi Tul1 E3 ligase (6). Although Dsc1 and Tul1 have low sequence similarity (less than 30%), both are predicted to contain a large luminal N-terminal domain (about 300–400 aa long), followed by seven transmembrane segments, and a C-terminal, cytosolic RING domain. The RING domain is a well characterized zinc finger in which a total of eight cysteine and histidine residues coordinate two zinc atoms (13). Dsc1 contains an H2-type RING domain (C3H2C3) that differs from the classical RING domain (C3HC4) by having two histidines in the fourth and fifth positions (Fig. 1B) (14). RING domains interact directly with E2 ubiquitin-conjugating enzymes to ubiquitinate protein substrates (13, 15, 16). Although the Dsc1 RING domain contains residues required for E2 binding (Fig. 1B), not all predicted RING domains possess E3 ligase activity (13). To test whether the Dsc1 RING domain is a functional E3 ligase, we performed an in vitro ubiquitination assay. RING domains will auto-ubiquitinate, allowing assessment of E3 ligase activity in the absence of substrate (17, 18). Purified Dsc1 RING domain (aa 608–695) fused to GST was incubated with ubiquitin-activating enzyme E1 (Uba1) and cognate ubiquitin-conjugating enzyme E2 (Ubc4) (Fig. 1C). GST-Dsc1 RING was efficiently ubiquitinated in a reaction that required ATP, E1 and E2 enzymes, and Dsc1 RING domain (Fig. 1D). The Dsc1 RING domain lacks lysine residues, indicating that ubiquitination occurred on lysine(s) in GST or on non-lysine residues in Dsc1. The E3 ligase activity of Dsc1 RING required both zinc-coordinating residues (C634A and H668A) and E2-interacting residues (I636D or L675D) (Fig. 1D, lanes 6–9). These results demonstrate that Dsc1 is a functional RING E3 ligase.

Fission yeast SREBP proteolytic activation requires Dsc1 (5). Previously, we demonstrated that mutation of a zinc-coordinating histidine (Dsc1-H668A) or a partial truncation of the RING domain (Dsc1-Q673X) abrogated SREBP activation (5). These mutations may cause misfolding of the RING domain, so we further tested whether SREBP activation required a RING domain surface residue essential for E2 binding (Dsc1-L675D) and E3 ligase activity (Fig. 1D) (15). In addition, we tested a second zinc-coordinating residue (Dsc1-C634A). We assayed proteolytic cleavage of both fission yeast SREBPs, Sre1 and Sre2. Consistent with previous results (5), mutation of either residue blocked SREBP activation (Fig. 2, A and B), further demonstrating that SREBP activation requires Dsc1 E3 ligase activity.

FIGURE 2.

Dsc1 RING domain function is required for SREBP cleavage in fission yeast. A, wild-type, dsc1Δ, dsc1-1, dsc1-C634A, dsc1-L675D, and sre1Δ cells were shifted to low oxygen for 4 h to induce Sre1 activation. Sre1 cleavage was assayed by immunoblotting whole cell lysates with anti-Sre1 antibody. P and N denote precursor and cleaved nuclear form, respectively. Sre1 levels were normalized to the Dsc5 loading control, and quantification was performed with a LI-COR Odyssey CLx using three biological replicates. Error bars indicate S.E. AU, arbitrary units. B, strains carrying empty vector (−) or a plasmid expressing 3×FLAG-Sre2 (423–793) (+) were cultured at 30 °C. Sre2 cleavage occurs in the presence of oxygen and was assayed by immunoblotting cell lysates with anti-FLAG antibody. CaMV, cauliflower mosaic virus.

Quantification of Sre1 cleavage showed an ∼20-fold increase in Sre1N in wild-type cells shifted to low oxygen (Fig. 2A). Interestingly, we observed an ∼4-fold decrease in Sre1 precursor in dsc1Δ cells under normoxic conditions. This decrease likely results from reduced levels of the Sre1-binding protein Scp1 in dsc1Δ cells (19). Scp1 stabilizes Sre1 precursor, and in the absence of Scp1, Sre1 precursor is degraded via ER-associated degradation (20). In addition, Sre1 precursor decreased under low oxygen in dsc mutant cells. Low oxygen stimulates ER-to-Golgi transport of the Sre1-Scp1 complex for Sre1 cleavage in the Golgi. In dsc mutants, Sre1N is not made (Fig. 2A, lanes 3–12), the Sre1 precursor is likely degraded, and the supply of Sre1 precursor is not replenished by positive feedback activation of the sre1 promoter (21). Sre2 does not bind Scp1, and Sre2 cleavage is Scp1-independent (3, 5).

The N-terminal luminal domain of Dsc1 contains five potential N-glycosylation sites and glycosylation of three sites (Asn-52, Asn-115, and Asn-220) has been confirmed by mass spectrometry (22). Interestingly, complete glycosylation of Dsc1 requires Dsc2, Dsc3, and Dsc4 (8). In the absence of Dsc2, Dsc3, or Dsc4, Dsc1 migrated with an intermediate mobility as compared with the mature and deglycosylated forms (Fig. 3A, lane 2) (8). Unexpectedly, we observed that Dsc1 mutants lacking E3 ligase activity also displayed incomplete glycosylation (Fig. 3A, lanes 3 and 4). Changes in mobility were due to differential N-linked glycosylation because Dsc1 mutants showed the same mobility as wild type after treatment with peptide-N-glycosidase F (Fig. 3A, lanes 6–9). These data indicate that Dsc1 E3 ligase activity is required for complete Dsc1 glycosylation.

FIGURE 3.

Dsc1 glycosylation requires RING activity. A, detergent-solubilized membranes from wild-type, dsc2Δ, dsc1 RING mutants, and dsc1Δ were treated without or with the glycosidase peptide-N-glycosidase F (PNGase F) and then immunoblotted using Dsc1 antiserum. B, wild-type, dsc1Δ, and temperature-sensitive ubc4-P61S cells were grown at 25 °C to exponential phase and then shifted to 36 °C. Cells were harvested at the indicated times, and detergent-solubilized membranes were immunoblotted using anti-Dsc1 antibody. C, wild-type, temperature-sensitive sar1-1, and dsc1Δ cells were grown at 25 °C to exponential phase and then shifted to 36 °C in the absence or presence of cycloheximide (CHX, 100 μg/ml) for 1 h. Cells were harvested at the indicated times, and detergent-solubilized membranes were treated without or with peptide-N-glycosidase F and immunoblotted using anti-Dsc1 antibody. D, wild-type, sar1-1, and dsc1-L675D cells were cultured for 2 h at the indicated temperatures. Isolated membranes were immunoblotted with anti-Dsc1 serum. In each panel, differentially glycosylated forms of Dsc1 are denoted: mature (M), intermediate (I), deglycosylated (D).

Dsc1 E3 ligase activity requires the E2 enzyme Ubc4 in vitro (Fig. 1D), and SREBP cleavage activation requires ubc4, demonstrating that Ubc4 is the cognate E2 enzyme for Dsc1 (5). ubc4 is an essential gene in fission yeast (23); thus we used a temperature-sensitive mutant (ubc4-P61S) as an independent test of whether Dsc1 E3 ligase activity is required for Dsc1 glycosylation. The intermediate Dsc1 band was absent at permissive temperature, but appeared after 30 min at non-permissive temperature and accumulated with time (Fig. 3B, lanes 1–5). Wild-type cells showed no change in Dsc1 mobility upon temperature shift (Fig. 3B, lanes 7 and 8). To this point, these results indicate that Dsc1 ubiquitin E3 ligase activity is required for complete Dsc1 glycosylation.

In fungi, secretory proteins receive mannose-rich core glycosylation in the ER (24). Core glycosylated proteins move to the Golgi via COPII vesicles, and in the Golgi, enzymes sequentially add and modify sugar residues to create the mature glycosylated protein (24). The Sar1 GTPase initiates COPII coat assembly and is required for ER-to-Golgi transport (25, 26). To test whether complete Dsc1 glycosylation requires Golgi enzymes, we used a temperature-sensitive sar1-1 strain to block ER-to-Golgi transport of newly synthesized Dsc1 (27). Upon shifting to the non-permissive temperature, we observed accumulation of the intermediate form of Dsc1 in sar1-1, but not wild-type cells (Fig. 3C, lanes 4 and 5). Treatment with cycloheximide blocked the appearance of the Dsc1 intermediate form (Fig. 3C, lane 6), indicating that this species represents newly synthesized, incompletely glycosylated Dsc1. Treating membrane extracts with peptide-N-glycosidase F confirmed that the observed mobility differences were due to altered N-linked glycosylation. Importantly, the mobility of Dsc1 from sar1-1 cells grown at non-permissive temperature was identical to that seen in dsc1-L675D cells defective for Dsc1 E3 ligase activity (Fig. 3D). Together, these data indicate that complete Dsc1 glycosylation requires Dsc1 E3 ligase activity and ER-to-Golgi transport for modification by Golgi enzymes.

We speculated previously that the Dsc1 glycosylation defect in cells lacking Dsc subunits was due to either improper complex assembly or Dsc1 mislocalization (8). To assess Dsc E3 ligase complex assembly in strains lacking E3 ligase activity, we performed co-immunoprecipitation studies with digitonin-solubilized membranes from wild-type and RING domain mutant cells. When Dsc2 was immunopurified, all Dsc components (Dsc1, Dsc3, Dsc5, and Dsc5) co-purified in wild-type cells and in Dsc1 RING mutants that are truncated (Q673X), fail to bind E2 ubiquitin-conjugating enzymes (L675D), or do not coordinate zinc (C634A) (Fig. 4A, lanes 7–12), indicating that complex assembly is unaffected by RING domain mutations. We further investigated Dsc E3 ligase complex assembly in ubc4-P61S cells defective for the cognate E2 ubiquitin-conjugating enzyme Ubc4. We observed above that after shifting to non-permissive temperature, Dsc1 accumulates as a faster migrating, intermediate form in ubc4-P61S cells (Fig. 3B). When purified with Dsc2, the Dsc1 intermediate form co-purified with other Dsc subunits as in wild-type cells (Fig. 4B, lanes 7–12). These data demonstrate that the incompletely glycosylated Dsc1 assembles into the Dsc E3 ligase complex and that E3 ligase activity is not required for complex assembly.

FIGURE 4.

Dsc complex assembly does not require Dsc1 E3 ligase activity. A, Dsc E3 ligase complex was immunoprecipitated (IP) from digitonin-solubilized membranes of wild-type, dsc1Δ, dsc1-1, dsc1-L675D, and dsc1-C634A cells using Dsc2 antiserum. Binding of Dsc proteins was assayed by immunoblotting. B, digitonin-solubilized membranes from wild-type, ubc4-P61S, and dsc2Δ cells either grown at 25 °C or grown after shifting to 36 °C for 2 h were subjected to immunoprecipitation using Dsc2 antiserum. Dsc proteins in bound fraction were assayed by immunoblotting.

Alternatively, incomplete Dsc1 glycosylation could result from a failure of Dsc1 to interact with Golgi glycosylation enzymes. The Dsc E3 ligase complex has a hierarchical organization where Dsc2-Dsc3-Dsc4 define the core, and Dsc1 and Dsc5 are peripheral subunits (Fig. 1A) (8). To assay Dsc E3 ligase localization, we determined the localization of the core subunit Dsc2 in mutant strains using live cell fluorescence confocal microscopy. As reported previously in wild-type cells, Dsc2 localized to Golgi puncta, distinct from the ER glycosyltransferase Ost1 (Fig. 5, top panels) (5). Deletion of dsc1, dsc3, or dsc4, but not dsc5, caused Dsc2 to localize to the ER (Fig. 5). These results correlated with the presence of incompletely glycosylated Dsc1 observed previously in dsc3Δ and dsc4Δ cells, but not dsc5Δ cells (8). In addition, Dsc1 is incompletely glycosylated in dsc2Δ cells (Fig. 3A). Thus, assembly of subunits Dsc1–Dsc4 is required for proper Golgi localization of the Dsc E3 ligase, and Dsc5 is not required for ER exit. Often, failure of a multi-protein complex to assemble or exit the ER results in subunit degradation by quality control pathways like ER-associated degradation or autophagy (28). Notably, Dsc complex subunits were stable when retained in the ER, for example, in dsc1Δ cells (Fig. 4A, lane 2) (8).

FIGURE 5.

Dsc complex localizes to ER in the absence of Dsc1 E3 ligase activity. Wild-type, dsc1Δ, dsc1-1, dsc3Δ, dsc4Δ, and dsc5Δ cells expressing dsc2-6×GFP and the ER marker ost1-mCherry were grown at 30 °C and imaged at room temperature using a spinning disk confocal microscope (3i). Scale bar = 10 μm.

Interestingly, Dsc2 also accumulated in the ER in dsc1-1 cells that assemble the Dsc complex normally (Fig. 4A), but lack Dsc E3 ligase activity due to a RING domain truncation (Fig. 4) (5). This result suggested that Dsc1 E3 ligase activity itself might be required for Golgi localization. To test this directly, we examined Dsc E3 ligase localization in ubc4-P61S cells conditionally defective for Dsc E3 ligase activity. Importantly, the Dsc E3 ligase is not mutated in these cells and assembles normally at the non-permissive temperature when E3 ligase activity is inhibited (Fig. 3B). As expected, incompletely glycosylated Dsc1 accumulated in ubc4-P61S cells (Fig. 6A, lanes 1–3). Cycloheximide treatment specifically blocked production of the Dsc1 intermediate form upon Ubc4 inactivation, indicating that Dsc1 fails to receive Golgi-dependent carbohydrate modifications in the absence of Ubc4 activity (Fig. 6A).

FIGURE 6.

Dsc1 E3 ligase activity is required for ER exit of the complex. A, temperature-sensitive ubc4-P61S dsc2-6×GFP cells were grown to exponential phase at 25 °C and then shifted to non-permissive temperature (36 °C) for 2 h in the absence (−) or presence (+) of cycloheximide (CHX, 100 μg/ml). Detergent-solubilized membranes were immunoblotted to assay Dsc1 glycosylation. In each panel, differentially glycosylated forms of Dsc1 are denoted: mature (M) and intermediate (I). B, temperature-sensitive ubc4-P61S dsc2-6×GFP cells were grown to exponential phase at 25 °C and imaged for Dsc2 at permissive temperature (25 °C, left panel). Alternatively, cells were shifted to the non-permissive temperature (36 °C) for 2 h and imaged at 36 °C in the absence (−CHX, middle panel) or presence of cycloheximide (+CHX, right panel). Imaging was performed using a 3i spinning disk microscope with an environmental chamber. The magenta arrowhead indicates cells showing only Golgi localization of Dsc2, and the green arrowhead indicates cells showing both ER and Golgi localization. The percentage of cells (160–170 cells/condition) showing either Golgi only or mixed ER and Golgi localization under different experimental conditions was calculated from two biological replicates. Error bars represent ± S.E. calculated using comparison of proportions. C, wild-type dsc2-6×GFP cells were imaged as in B. Scale bars = 10 μm.

Next, we examined Dsc E3 ligase localization in ubc4-P61S cells. At permissive temperature, the majority of cells showed Dsc2 localization only in Golgi puncta (88%) (Fig. 6B, left panel, magenta arrowhead), and a small number of cells showed both ER and Golgi localization (green arrowhead). Upon incubation at non-permissive temperature for 2 h, Dsc2 localization changed with 82% of cells showing mixed ER and Golgi localization and 18% displaying only Golgi localization (Fig. 6B, middle panel). Treatment with cycloheximide when shifting to the non-permissive temperature prevented ER accumulation of Dsc2 (88% Golgi only, 12% ER and Golgi) (Fig. 6B, right panel). Dsc2 localized to the Golgi in wild-type cells under each of the three conditions, showing that ER localization required Ubc4 inactivation (Fig. 6C). Taken together, the data in Fig. 6 demonstrate that the Dsc E3 ligase complex fails to exit the ER in the absence of Ubc4 activity. Given that inactive Dsc E3 ligase complex assembles normally when retained by ER in both dsc1-1 and ubc4 cells, we conclude that failure to exit the ER is due to the absence of E3 ligase activity and not structural alterations in the Dsc E3 ligase.

To test whether Dsc E3 ligase activity is required specifically for Dsc complex ER exit, we assayed secretory pathway function in dsc mutant strains. ER-to-Golgi protein trafficking is essential for cell viability (29). We hypothesized that growth at elevated temperature will stress cells and may reveal dsc-dependent defects in ER-to-Golgi transport. We compared growth of wild-type and dsc mutant cells at 37 °C and observed no growth difference in exponentially growing cells (Fig. 7A). Next, we looked at the localization of cis-Golgi resident, mannosyltransferase subunit Anp1. Inhibition of COPII-dependent vesicle formation results in Anp1 mislocalization to the ER (5). Anp1-GFP localized to Golgi puncta in wild-type, dsc1Δ, and dsc2Δ cells (Fig. 7B), indicating that COPII vesicle transport does not require Dsc E3 ligase activity. In an independent experiment, we co-localized Dsc2 and Anp1 in dsc1Δ cells, confirming that Golgi-resident Anp1 localizes normally under conditions in which the Dsc E3 ligase is mislocalized to the ER (Fig. 7C).

FIGURE 7.

Cell growth and Golgi morphology does not require Dsc E3 ligase activity. A, overnight grown cells from wild-type, dsc1Δ, dsc2Δ, and dsc1-L675D strains were diluted in rich medium and incubated at 37 °C. Cell growth was monitored by measuring OD₆₀₀. Error bars indicate standard error. B, localization of the Golgi mannosyltransferase subunit Anp1-GFP was determined in indicated strains by imaging cells at room temperature using a 3i spinning disk confocal microscope. C, localization of Dsc2-6×GFP and Anp1-mCherry in dsc1Δ cells as in B. Scale bar = 10 μm.

Discussion

Multiple lines of evidence demonstrate that Dsc1 E3 ligase activity is required for ER exit of the Dsc complex, thus ensuring that only fully functional Dsc E3 ligase moves to the Golgi where it acts in SREBP activation. Although E3 ubiquitin ligases are known to regulate their abundance through auto-ubiquitination (30–32), in this instance, enzyme activity controls protein localization. Multi-subunit membrane complexes such as MHC class I and Kir potassium channels require proper assembly prior to ER exit (33–35). However, even when the Dsc E3 ligase complex assembles properly, ER exit requires E3 ligase activity. To our knowledge, this represents the first example of enzyme activity-dependent protein sorting in the secretory pathway.

How ligase activity controls ER exit is unknown, but potential mechanisms exist. First, the Dsc E3 ligase complex may assemble with an ER retention protein whose removal and possibly degradation require Dsc1-dependent ubiquitination. Alternatively, the Dsc1 E3 ligase may ubiquitinate one or more COPII components whose modification is required for Dsc E3 ligase ER exit. Indeed, packaging of pro-collagen requires CUL3-KLHL12-mediated mono-ubiquitination of the COPII component Sec31 (36), and mono-ubiquitination of Sec23 has been detected in budding yeast (37). Interestingly, Dsc1 E3 ligase utilizes the E2 Ubc4 that preferentially adds mono-ubiquitin to substrates (38). Finally, a Dsc subunit could be a direct substrate, and the addition of ubiquitin may alter Dsc complex conformation or recruit an effector protein to facilitate ER exit. Although the mechanistic details are under investigation, these findings add to the growing complexity of ER cargo sorting and open up an opportunity to investigate the role of ubiquitination in this process.

Lastly, these findings have important implications for the current model of SREBP activation in fungi in which we proposed that SREBP activation requires its ubiquitination. The fact that the Dsc E3 ligase mislocalizes to the ER when inactive necessitates additional studies of the role for SREBP ubiquitination in its proteolytic activation.