Lipid Droplet Formation Is Dispensable for Endoplasmic Reticulum-associated Degradation*

Proteins that fail to fold or assemble in the endoplasmic reticulum (ER) are destroyed by cytoplasmic proteasomes through a process known as ER-associated degradation. Substrates of this pathway are initially sequestered within the ER lumen and must therefore be dislocated across the ER membrane to be degraded. It has been proposed that generation of bicellar structures during lipid droplet formation may provide an “escape hatch” through which misfolded proteins, toxins, and viruses can exit the ER. We have directly tested this hypothesis by exploiting yeast strains defective in lipid droplet formation. Our data demonstrate that lipid droplet formation is dispensable for the dislocation of a plant toxin and the degradation of both soluble and integral membrane glycoproteins.

Proteins that fail to fold or assemble in the endoplasmic reticulum (ER) are destroyed by cytoplasmic proteasomes through a process known as ER-associated degradation. Substrates of this pathway are initially sequestered within the ER lumen and must therefore be dislocated across the ER membrane to be degraded. It has been proposed that generation of bicellar structures during lipid droplet formation may provide an "escape hatch" through which misfolded proteins, toxins, and viruses can exit the ER. We have directly tested this hypothesis by exploiting yeast strains defective in lipid droplet formation. Our data demonstrate that lipid droplet formation is dispensable for the dislocation of a plant toxin and the degradation of both soluble and integral membrane glycoproteins.
Secreted and transmembrane proteins are co-translationally inserted into the endoplasmic reticulum (ER), 3 where they are folded, modified, and assembled into higher order complexes prior to transport to their final destinations. Proteins failing to reach their native conformation are recognized and delivered to the cytosolic ubiquitin-proteasome system for degradation by a process termed ER-associated degradation (ERAD). How substrates overcome the spatial segregation imposed by the ER membrane and are finally dislocated from the ER into the cytosol is a fundamental question that remains unresolved (1,2). Polytopic ER-resident membrane proteins including Sec61p (3), Derlins (4,5), and Hrd1p (6) have been suggested, largely on the basis of their polytopic membrane-spanning topology, their proximity to other membrane-associated components of the ERAD system, and the ability to co-precipitate or be crosslinked to putative dislocation intermediates, to be structural elements of a transmembrane channel that mediates substrate dislocation. However, the structure and composition of such a channel remain obscure, and its existence remains controversial (1,2,7). The ability of the ER dislocation apparatus to accommodate large bulky substrates such as glycoproteins with intact N-glycans (8), ligand-stabilized folded proteins (9), as well as toxins (10) and even intact viruses (11), which appear to hijack this pathway to enter the cytosol, disfavors the existence of such a pore. In response to these concerns, Ploegh (7) has proposed a provocative ER exit model in which substrate dislocation is coupled to the formation of lipid droplets (LDs), which are ER-derived organelles required for the storage and mobilization of neutral lipids. Relatively little is known about how LDs form, and Ploegh (7) hypothesized that generation of bicellar structures during LD formation could result in transient pores, or "escape hatches," in the ER membrane, allowing the escape of luminal ERAD substrates and/or the direct extraction of polytopic substrates associated with the LD membrane. This LD-based ER exit model remains untested.
Yeast strains lacking the acyltransferases (Dga1p, Lro1p, Are1p, and Are2p) necessary for the synthesis of triacylglycerides and steryl esters (referred to herein as LD⌬ strain) are devoid of cytoplasmic LDs (12)(13)(14)(15). Yeast has proven to be a powerful system in which to investigate the basic mechanisms that underlie substrate recognition and degradation by ERAD (2); most of the key components are well conserved between yeast and humans. In this study, we exploit the LD⌬ strain to examine the role of LDs in ERAD and provide definitive evidence that the formation of LDs is not required for ER protein dislocation.

EXPERIMENTAL PROCEDURES
Yeast Strains and Plasmids-All yeast strains used are listed in supplemental Table S1. The plasmids encoding UPRE-GFP (kind gift from Jonathan Weissman), HA-tagged CPY* (kind gift from Jonathan Weissman), HA-tagged Pdr5* (kind gift from Randolph Hampton), and ER-targeted E177A mutant ricin A chain (kind gift from Lynne Roberts) were described previously (10, 16 -18). The fragment encoding the UPRE-GFP reporter was subcloned into pRS314 by ligation of the ClaI-SacI fragment (18,19). Gene deletion was performed using standard PCR targeting methods. Yeast were transformed by the lithium acetate method (20).
Cycloheximide Chases-Cycloheximide chases were performed as described previously (17). Briefly, yeast were grown to logarithmic phase (OD ϳ0.5-1.0). 1 optical density unit of yeast was removed at each time point after the addition of 200 g/ml cycloheximide, pelleted by centrifugation at 20,000 ϫ g, and frozen in liquid nitrogen. CPY*-expressing yeast were solubilized by the addition of SDS loading dye, boiled for 5 min, centrifuged, separated by SDS-PAGE, and analyzed by immunoblotting. Pdr5*-expressing yeast were resuspended in lysis buffer (1% SDS, 8 M urea, 10 mM MOPS, pH 6.8, 10 mM EDTA) and vortexed with glass beads for 10 min. Soluble extract was removed, mixed with a 2ϫ loading dye, separated by SDS-PAGE, and analyzed by Western blotting. Densitometry quan-* This work was supported, in whole or in part, by National Institutes of Health Grant GM074874 (to R. R. K.). □ S The on-line version of this article (available at http://www.jbc.org) contains supplemental tification of Western blots was performed using National Institutes of Health ImageJ software version 10.2.
Fluorescence Microscopy-Yeast were grown to stationary phase, and vital staining was performed by the addition of 10 g/ml BODIPY 493/503 (Invitrogen). After 30 min, stained cells were mounted on poly-L-lysine-coated coverslips and immediately analyzed by fluorescence microscopy using a Zeiss Axiovert 200M microscope outfitted with a 100ϫ oil immersion objective and the appropriate filters.
Flow Cytometry-Yeast strains expressing the UPRE-GFP reporter, which encodes a GFP reporter gene driven by four repeats of the kar2 unfolded protein response elements (UPRE), were grown to logarithmic phase (OD ϳ0.5-1.0), collected by centrifugation, and resuspended in PBS. GFP fluorescence was measured using a FACSCalibur flow cytometer (BD Biosciences).
Ricin-based Viability Assay-Yeast viability assays and the RTA E177A expression plasmid were described previously (10). Briefly, RTA containing an E177A substitution in the active site, to reduce its toxicity, was directed to the ER by fusion with the Kar2 signal sequence and expressed under the control of the GAL1 promoter in a pRS316-derived plasmid. For viability assays, strains expressing RTA E177A were grown overnight and diluted to an OD of 1, and 10-fold serial dilutions were spotted onto URA-plates supplemented with either 2% glucose or 2% galactose. Images shown were taken after 5 days of growth.

RESULTS AND DISCUSSION
Immunofluorescence analyses of the BODIPY 493/503stained LDs demonstrated that LDs were present in wild-type (WT) yeast and the strain lacking the ERAD ubiquitin ligase Hrd1p (hrd1⌬) but were absent from the LD⌬ strain (Fig. 1A). In agreement with previous studies (12, 13), we confirmed that loss of LD formation results in the induction of the unfolded protein response (UPR), a pathway responsible for the up-regulation of ER protein folding and degradation machinery in response to stress (supplemental Fig. S1), consistent with a potential role for LD formation in ERAD. However, the UPR can also be induced by perturbations in cellular processes that are not directly involved in ER protein quality control (21).
To directly examine whether the LD⌬ strain exhibits deficits in ERAD, we employed translation shutoff experiments to assess the degradation of CPY*, a well characterized luminal glycosylated ERAD substrate (22). Despite nearly complete stabilization of CPY* in the hrd1⌬ strain, we observed that the kinetics of CPY* degradation in WT and LD⌬ strains were indistinguishable (Fig. 1, B and C). We considered the possibility that LD formation may contribute to the ERAD of specific topological classes of substrates, which are degraded by separate pathways utilizing distinct sets of machinery (1,2). For example, deformation of the ER lipid bilayer by deposition of neutral lipids into the ER membrane during LD formation could selectively destabilize polytopic substrates, thereby promoting their extraction from the membrane by ERAD machinery. However, our results indicate that Pdr5*, a polytopic ERAD substrate predicted to have a 12-pass transmembrane domain (23), was degraded at the same rate in both the WT and the LD⌬ strains but was stabilized in the hrd1⌬ strain (Fig. 1, D and E). These results demonstrate that LD formation is not required for the dislocation and degradation of both luminal and polytopic ERAD substrates.
In addition to the degradation of terminally misfolded secretory proteins, the ERAD pathway is also known to play a critical role in the dislocation of some A/B-type toxins (10). To examine whether the dislocation of luminal proteins is generally independent of LD formation, we employed a growth-based assay utilizing an ER-targeted attenuated ricin A chain variant (RTA E177A ), a potent plant toxin that co-opts the ERAD system to exit from the ER (10). Growth of WT yeast was severely impaired by galactose-induced expression of RTA E177A ; this growth impairment was substantially alleviated by the loss of Hrd1p (Fig. 2), confirming an essential role for this ubiquitin ligase in RTA E177A dislocation. By contrast, the growth of LD⌬ yeast was unaffected by RTA E177A expression, indicative of efficient dislocation.
These data establish that LD formation is not essential for protein dislocation from the ER but do not preclude a role for LD formation in the dislocation of viruses (11) or specific ERAD substrates (24,25) or a role for the ERAD machinery in LD biology. Indeed, associated with the LD phospholipid monolayer are at least three proteins functionally linked to ERAD: UBXD8 (26), AUP1 (27), and UBE2G2 (27). Their presence might signify the existence of an LD protein degradation complex that contributes to the turnover or remodeling of the droplet proteome.
The dislocation of large bulky substrates, such as glycoproteins, ligand-stabilized folded proteins, and even intact viruses, is unlikely to be mediated by narrow, protein-conducting pores, such as the Sec61p translocon (1,7,28,29). It is conceivable, however, that protein dislocation may resemble the process of protein import into peroxisomes, which appears to involve the formation of a cargo-activated channel that exhibits dynamic flexibility in size to accommodate the translocation of fully folded and even oligomerized proteins (1,28,30,31). Further studies on the structure and functional organization of ER membrane protein complexes, in particular those containing polytopic integral proteins that have been linked to dislocation, such as Derlins and Hrd1p, are clearly warranted.