Derlin-dependent accumulation of integral membrane proteins at cell surfaces

The degradation of misfolded proteins is essential for cellular and organism viability. Quality control mechanisms of protein folding involve multi-component systems, which include chaperones, ubiquitylation enzymes, and ultimately degradation by the proteasome. So far, quality control mechanisms have been described in the cytoplasm, the nucleus and endoplasmic reticulum (ER) (Bader et al., 2007; Goldberg, 2003; Hampton, 2002; Jarosch et al., 2003; Meusser et al., 2005; von Mikecz, 2006).

The recognition and degradation of misfolded proteins in the ER is called ER-associated degradation (ERAD) (Hampton, 2002; Jarosch et al., 2003; McCracken and Brodsky, 2003; Meusser et al., 2005; Richly et al., 2005; Sitia and Braakman, 2003). Membrane-spanning and secretory proteins are first transported into the ER in an unfolded state through the Sec61p complex (Matlack et al., 1998). Folding of these nascent polypeptides is assisted by a number of ER-resident chaperones. Translocated proteins also undergo modifications to support folding; these include N-terminal glycosylation and disulphide bond formation (Meusser et al., 2005; Schroder and Kaufman, 2005; Sitia and Braakman, 2003). In the ER, proteins that do not fold properly are retro-translocated to the cytoplasm. During retro-translocation, these misfolded proteins are ubiquitylated by several ER-specific E3 ubiquitin ligase complexes. A cytoplasmic ubiquitin-binding and multi-ubiquitylation enzyme complex further modifies these proteins and finally transports them to the proteasome for degradation.

The accumulation of misfolded proteins in the ER activates the unfolded protein response (UPR), which is required for cells to survive conditions of stress. The UPR is mediated by three ER transmembrane proteins, IRE1, PERK and ATF6, which get activated at least in part because of the dissociation of the ER chaperone BiP, to which they are normally bound and also because of their sequestration by misfolded proteins (Bertolotti et al., 2000; Cox et al., 1993; Harding et al., 1999; Haze et al., 1999; Iwawaki et al., 2001; Kimata et al., 2004; Lee et al., 2002; Mori et al., 1993; Okamura et al., 2000). IRE1, PERK and ATF6 function to decrease the load on the ER by reducing translation rate and activating the transcription of chaperones, ERAD proteins and other enzymes. UPR activation also results in increased biosynthesis of some lipids, the elaboration of the ER and increased secretion (Sato et al., 2002; Shaffer et al., 2004; Sriburi et al., 2004). A major downstream regulator of UPR is XBP1/HAC1. Upon activation of IRE1, the XBP1 mRNA is directly spliced by an endonuclease activity in the C-terminus of IRE-1; this splice variant of XBP1 functions as a potent transcriptional activator of several genes (Calfon et al., 2002; Cox and Walter, 1996; Sidrauski and Walter, 1997; Yoshida et al., 2001).

Derlin proteins are a conserved family that function in ERAD (Schekman, 2004). They have four transmembrane domains and are conserved in all eukaryotes. There are two members in Saccharomyces cerevisiae, Der1p and Dfm1p, two in C. elegans, F25D7.1 and R151.6, and three in humans, Derlin-1, Derlin-2 and Derlin-3 (DERL1-DERL3) (Lilley and Ploegh, 2004; Ye et al., 2004). Der1p in S. cerevisiae is required for the ERAD-mediated degradation of soluble, but not of membrane-spanning misfolded proteins (Hitt and Wolf, 2004; Knop et al., 1996; Taxis et al., 2003). By contrast, the HCMV-encoded US11 protein recruits major histocompatibility complex class I (MHCI) molecules to human Derlin-1, leading to retro-translocation and degradation of MHCI (Lilley and Ploegh, 2004; Ye et al., 2004). Derlin-1 and Derlin-2 function as part of a complex that includes other components of the ERAD machinery: VIMP/p97 and ubiquitylation-associated proteins such as SEL1, HRD1 and HERP (Lilley and Ploegh, 2004; Lilley and Ploegh, 2005; Schulze et al., 2005; Ye et al., 2005; Ye et al., 2004). Derlin-1, Derlin-2 and Derlin-3 form homo- and hetero-oligomers (Lilley and Ploegh, 2005; Ye et al., 2005) and have therefore been proposed to contribute to the channels through which ERAD substrates might translocate (Schekman, 2004).

In contrast to the ER, less is known about quality control mechanisms that monitor the structure of membrane proteins at the plasma membrane and endosomes. In yeast, misfolded Pma1p at the plasma membrane is targeted to the endosome or vacuole for degradation following ubiquitylation (Gong and Chang, 2001; Pizzirusso and Chang, 2004). In mammalian cells, several transmembrane proteins that are misfolded (or perceived as such) at the plasma membrane are degraded more rapidly than their properly folded counterparts; these misfolded proteins at the plasma membrane include unliganded major histocompatibility complex (MHC) Type I, and mutations in the cystic fibrosis transmembrane conductance regulator (CFTR) and the α2-adrenergic receptor (Benharouga et al., 2001; Ljunggren et al., 1990; Sharma et al., 2004; Wilson et al., 2001). Our results suggest that Derlin proteins function in the recognition and/or the degradation of proteins that are misfolded at the plasma membrane and/or within the endosomal system.

The cup-2 gene was identified on the basis of the ar506 mutation that results in decreased endocytosis in scavenger cells called coelomocytes in Caenorhabditis elegans (Fares and Greenwald, 2001a). cup-2(ar506) worms have a temperature-sensitive defect in endocytosis by coelomocytes (Fig. 1A,B). pmyo-3::ssGFP transgenic worms express GFP fused to a signal sequence and expressed in body wall muscles: this GFP is secreted into the body cavity and is endocytosed, and subsequently degraded, primarily by the coelomocytes (Fares and Greenwald, 2001a). Mutations that decrease endocytosis by coelomocytes result in the accumulation of GFP in the body cavity and a decrease in the size of the GFP-filled compartments in coelomocytes (Dang et al., 2004; Fares and Greenwald, 2001a; Patton et al., 2005). In addition to this temperature-sensitive endocytosis defect, cup-2(ar506) worms are sick or nonviable at 25°C, such that 90-95% of L4 larvae shifted from 20°C to 25°C die; the rest of the worms survive and lay eggs, most of which are nonviable.

We determined that CUP-2 corresponds to open reading frame F25D7.1 based on phenocopy by RNA-mediated interference (RNAi), sequence analysis of the ar506 allele and transgenic rescue of the mutant phenotypes (Fig. 1C; supplementary material Fig. S1A; Fig. 2). The ar506 allele is a nucleotide change in the first exon that results in an early stop codon and therefore represents a predicted null mutant of cup-2 (Fig. 1C; supplementary material Fig. S1B). CUP-2 is one of two C. elegans Derlin proteins and shows the highest identity to human Derlin-1 (Fig. 1D; supplementary material Fig. S1B). Derlin proteins are conserved proteins that function in ERAD and have four predicted transmembrane domains, with both the N- and the C-termini being cytoplasmic (Fig. 1D, supplementary material Fig. S1B) (Hitt and Wolf, 2004; Knop et al., 1996; Lilley and Ploegh, 2004; Lilley and Ploegh, 2005; Ye et al., 2004). Consistent with this basic cell biological function, fusion of GFP to the CUP-2 promoter resulted in ubiquitous expression of GFP in all tissues, including coelomocytes (Fig. 1E).

We confirmed that CUP-2 functions in ERAD based on two criteria. First, cup-2(ar506) was synthetically lethal at 20°C with a null mutation in ire-1, the worm orthologue of the UPR sensor IRE1 (Fig. 1F) (Shen et al., 2001). Similarly, a yeast Δder1 Δire1 strain was nonviable at 37°C (Hitt and Wolf, 2004). cup-2; pek-1 and cup-2; atf-6 double mutants only showed slight increases in lethality (Fig. 1F). Second, in the cup-2(ar506) mutant, or after reducing CUP-2 levels by RNAi, the UPR was activated in most cells, including coelomocytes, as determined by the induction of GFP expression from an hsp-4::GFP reporter (Fig. 1G,H; supplementary material Fig. S1A). A similar effect of cup-2 RNAi on induction of hsp-4::GFP expression in intestinal cells has been reported previously (Ye et al., 2004). Unlike the endocytosis defect, the UPR induction due to cup-2(ar506) is not temperature sensitive. Furthermore, the activation of the UPR response is absent in an xbp-1 null mutant, indicating that it is dependent on the site-specific cleavage and activation of xbp-1 mRNA by IRE-1 (supplementary material Fig. S1A) (Calfon et al., 2002).

RNAi of R151.6, the second worm Derlin protein, did not result in detectable UPR activation in any cells of wild-type animals; it also did not affect endocytosis by coelomocytes (supplementary material Fig. S1A). However, CUP-2 and R151.6 exhibited partially redundant function in activating the UPR, because RNAi of R151.6 in cup-2(ar506) resulted in a dramatic increase in XBP-1-dependent hsp-4::GFP expression (supplementary material Fig. S1A).

To determine whether CUP-2 function is conserved, we expressed several Derlin proteins from other species in cup-2 mutant worms. The CUP-2 promoter drove Derlin homologue expression in these transgenic animals. We observed essentially two patterns of rescue (Fig. 2; supplementary material Fig. S2). C. elegans CUP-2 and R151.6, and human Derlin-1 and Derlin-3 rescued both endocytosis and UPR activation phenotypes. These results indicate that both activities of CUP-2 are also found in mammalian Derlin proteins. Yeast Der1p and Dfm1p, and human Derlin-2 did not rescue either defect. Furthermore, yeast Dfm1p exacerbated the UPR activation, but not the endocytosis defect, suggesting that it might function as a dominant-negative protein that interferes with R151.6 activity in the ER.

To determine whether the loss of CUP-2 also affects the endocytic trafficking of transmembrane proteins, we analyzed the levels of the Ca²⁺ pump CUP-7/MCA-3. MCA-3 was originally identified on the basis of mutations that disrupt endocytosis (Fares and Greenwald, 2001a). A functional GFP::MCA-3 fusion localizes to the plasma membrane (supplementary material Fig. S3A) (Bednarek et al., 2007). GFP::MCA-3 does not localize to the large vacuoles in a cup-5 mutant that is defective in lysosome-mediated degradation, indicating that it is not normally transported to the lysosome before or after endocytosis (supplementary material Fig. S3A) (Fares and Greenwald, 2001b; Treusch et al., 2004). By contrast, GFP::MCA-3 accumulates to the membranes of expanded mCherry::RAB-11-positive recycling endosomes in an rme-1 mutant, indicating that MCA-3 continuously cycles between the plasma membrane and endosomes (supplementary material Fig. S3A,B) (Grant et al., 2001; Lin et al., 2001). RME-1 is an EH-domain-containing protein that is required for the exit from recycling endosomes; loss of RME-1 results in expanded recycling endosomes (Grant et al., 2001; Lin et al., 2001).

cup-2(ar506) (or cup-2 RNAi) results in increased levels of GFP::MCA-3 at the plasma membrane as measured by quantitative microscopy (Fig. 3A,B; and data not shown). Western blot analysis of immunoprecipitated GFP::MCA-3 indicated that the increased GFP::MCA-3 signal is due to increased levels of GFP::MCA-3 protein in cup-2 mutant worms (Fig. 3C,D). From three independent immunoprecipitation experiments, there was a 1.73±0.06 increase in the levels of GFP::MCA-3 in the coelomocytes of cup-2 mutant worms relative to the wild type; this is in agreement with the measurements from microscopy (Fig. 3A,B). By contrast, GFP::MCA-3 does not accumulate at the plasma membrane of cup-5 mutants with strongly reduced lysosome function nor in a cup-4 mutant, which displays an even more severe general endocytosis defect than the cup-2 mutant (supplementary material Fig. S3A,C) (Patton et al., 2005; Treusch et al., 2004). Neither cup-2 nor cup-4 mutations completely block endocytosis in coelomocytes. This indicates that the accumulation of GFP::MCA-3 at the plasma membrane is not a general consequence of decreasing internalization rates. Furthermore the increase of GFP::MCA-3 at the plasma membrane is probably independent of the UPR activation defect of cup-2(ar506) because this increase is evident at 20°C and not at 15°C, which correlates with the endocytosis phenotype and not the UPR activation defect (Fig. 3A,B). Finally, we checked the levels of two other integral membrane proteins that are expressed under the control of the same coelomocyte promoter used to express GFP::MCA-3. Mannosidase II::GFP localizes to the Golgi complex whereas GFP::CUP-5 localizes to lysosomes in coelomocytes (Treusch et al., 2004). Neither protein showed increased levels in cup-2 mutants, indicating that the increase in MCA-3 levels is not due to an increase in general secretion or promoter activity (supplementary material Fig. S4A-D).

Therefore, at least one transmembrane protein, MCA-3, accumulates at the surface of coelomocytes in the absence of CUP-2. Given this result and the endocytosis defect, we assayed other markers of the endocytic pathway in coelomocytes.

To better understand the nature of the endocytosis defect in cup-2 mutants we analyzed the morphologies of all major membrane bound organelles of the endocytic and secretory pathways in cup-2(ar506) coelomocytes using a set of GFP-tagged markers that we developed (Fig. 3D). There was no change in the localization of GFP-RME-1, a recycling endosome and plasma-membrane-localized EH-domain-containing protein required for endocytic traffic in coelomocytes (Fares and Greenwald, 2001a; Grant et al., 2001). By contrast, there was a significant increase in the levels of GFP-tagged clathrin heavy chain (CHC) at the surface of cup-2(ar506) coelomocytes (Greener et al., 2001). This increase in membrane CHC localization was not accompanied by a change in the absolute levels of CHC in cup-2(ar506) worms, indicating that the increased clathrin signal probably indicates increased clathrin assembly on membranes (Fig. 3E).

RAB-7, a marker for late endosomes, and CUP-5, a marker for lysosomes, both showed normal staining patterns, albeit of compartments that were reduced in size; this reduction in the size of endosomes or lysosomes is a general defect seen in mutations that block coelomocyte endocytosis (Grant and Hirsh, 1999; Patton et al., 2005; Treusch et al., 2004) (Fig. 3D). There was an elaboration and a dispersal of the ER in cup-2(ar506) coelomocytes (Fig. 3D). Both of these phenotypes are seen during hyperactivation of UPR in mammalian plasma cells and in other mutations that block coelomocyte endocytosis (Patton et al., 2005; Shaffer et al., 2004; Sriburi et al., 2004). Finally, there was no obvious change in the localization of mannosidase II, a transmembrane Golgi marker (Fig. 3D) (Patton et al., 2005; Rolls et al., 2002).

We also determined whether the loss of CUP-2 results in a delay in transport of endocytosed BSA-Rhodamine to lysosomes. This experiment was feasible because although the rate of endocytosis of soluble molecules was reduced in the cup-2 mutant, internalization was not blocked. Wild-type and cup-2 mutant worms that express the endosomal marker RME-8::GFP or the lysosomal marker GFP::CUP-5 were injected with BSA-Rhodamine in their pseudocoeloms. In wild-type and cup-2 mutant worms, all of the BSA-Rhodamine was found in RME-8-positive, CUP-5-negative endosomes after 5 minutes of uptake, as has been previously shown (Dang et al., 2004; Treusch et al., 2004). In both wild-type and cup-2 mutant worms, we first detected BSA-Rhodamine in RME-8-negative, CUP-5-positive lysosomes 10 minutes into the time course (arrowheads in Fig. 3F). Therefore, the loss of CUP-2 does not affect the progress of endocytosed solutes and membrane through the endo-lysosomal pathway.

Therefore, the only defect that we detect in the absence of CUP-2, so far, is an accumulation of CHC at the surface of coelomocytes. This accumulation has not been observed in other mutants that disrupt coelomocyte endocytosis and is therefore not due to a general reduction in internalization (Bednarek et al., 2007; Grant et al., 2001; Patton et al., 2005; Sato et al., 2005; Xue et al., 2003; Zhang et al., 2001).

Given the roles of Derlin proteins in degrading misfolded proteins in the ER, we hypothesized that misfolded protein accumulation could explain the increased MCA-3 levels at the surface of cup-2 mutant coelomocytes. We therefore did studies in cultured cells where this idea could be tested.

The rescue of cup-2 mutant phenotypes by mammalian Derlin proteins suggests functional conservation. We therefore initiated studies in mammalian cells to allow for pulse-chase and biochemical manipulations that are not currently feasible in worms. We chose murine RAW264.7 macrophages because they are analogous to coelomocytes and therefore allow for a smooth transition for comparative analysis between the worm and the mammalian work.

We made a stable Derl1 RNAi clone in RAW264.7 cells. In these cells, Derlin-1 levels were 8.2±0.9% and Derlin-2 levels were 133.5±14.8% of that in the wild type (Fig. 4A). Derlin-3 was not detectable in the wild type or Derlin-1 RNAi clone (Fig. 4A). The increase in Derlin-2 levels is consistent with previous results showing that both Derlin-2 and Derlin-3 are upregulated by the UPR, which has probably been activated because of the reduced Derlin-1 levels (Oda et al., 2006).

Previous studies have shown that at least some of the low-density lipoprotein receptor (LDLR) at the plasma membrane is degraded by the proteasome (Martin de Llano et al., 2006; Miura et al., 1996). We therefore checked the levels of the LDLR at the plasma membrane by incubating live cells with antibodies at 4°C, followed by fixation and imaging. We detected a sevenfold increase in the levels of LDLR at the plasma membrane of the Derl1 RNAi clones relative to RAW264.7 cells or to Mcoln1 RNAi stable clones that were used as a control (Fig. 4B,C). Mcoln1 encodes mucolipin 1, which is an endosomal/lysosomal-localized protein required for efficient lysosomal trafficking and lysosomal degradation of endocytosed proteins in RAW264.7 cells (Thompson et al., 2007). We also checked the levels of Fcgamma receptors (FcR) using an antibody that recognizes CD16, CD32 and CD64 (Unkeless, 1979). Although the trafficking itinerary from the plasma membrane of these proteins is not clear, we also detected a fourfold increase in the levels of FcR at the plasma membrane of the Derl1 RNAi clone relative to RAW264.7 cells or to an Mcoln1 RNAi stable clone that was used as a control (Fig. 4B,C). Therefore, similarly to worm cup-2 mutants, reduced Derlin-1 levels result in an increase in the plasma membrane levels of at least two integral membrane proteins.

The increase in LDLR levels at the plasma membrane of the Derl1 RNAi clone did not result in a increase in LDL binding at the plasma membrane. We incubated cells with fluorescent LDL at 4°C before washing and fixing the cells at 4°C. We detected similar levels of LDL binding to the surfaces of RAW264.7, Derl1 RNAi, and Mcoln1 RNAi cells (Fig. 4D,E). This suggests that the substantial number of LDLRs at the plasma membrane of Derl1 RNAi cells that are unable to bind to LDL probably represent LDLR molecules whose extracellular ligand-binding domain is misfolded.

We wanted to directly assay the fates of transmembrane proteins at the plasma membrane that become misfolded. We first tested a high-salt and low-pH buffer (741 mM citric acid, 258.7 mM sodium citrate, pH 3.5) that is predicted to affect the conformation of the extracellular domains of many receptors. Treatment of cells with this misfolding buffer for 15 minutes at 4°C resulted in approximately one-fifth of the LDL binding to the cell surfaces relative to D-PBS-treated samples (Fig. 4D,E). This indicates that this buffer, although not lethal to cells (data not shown), causes conformational changes in the extracellular domains of receptors.

We then used this misfolding buffer to determine the parameters of the degradation of misfolded LDLR at the plasma membrane. We pre-incubated cells for 2 hours in cycloheximide to block translation and treated them with D-PBS (control) or with misfolding buffer. These cells were placed back in medium containing cycloheximide alone, with the proteasomal inhibitor MG132, or with the lysosomal inhibitor leupeptin. Samples were taken every 2 hours for western blot assays to measure the rates of disappearance of the plasma-membrane-localized mature form of the LDLR (confirmed using surface biotinylation assays, data not shown). There are three main conclusions from this analysis (Fig. 4F,G). First, induction of misfolding of surface LDLR results in a sharp increase in the rate of its degradation in normal cells (red vs black solid lines in Fig. 4G). Second, misfolded LDLR is degraded primarily by the proteasome (orange solid line in Fig. 4G) and is less dependent on lysosomal function (green solid line in Fig. 4G). This is consistent with previous studies of plasma-membrane-localized LDLR turnover in CHO cells under normal conditions, where a proteasomal pathway degrades 70% of these receptors and the rest are degraded in lysosomes (Martin de Llano et al., 2006). Third, reducing Derlin-1 levels stabilizes plasma membrane LDLR under normal conditions (black dashed lines in Fig. 4G) and after misfolding (red dashed lines in Fig. 4G). Therefore, misfolding of LDLR at the plasma membrane results in its rapid turnover in a proteasome and Derlin-1-dependent manner.

Loss of CUP-2 results in MCA-3 accumulation at the surfaces of cells. Reduced Derlin-1 levels results in an accumulation of misfolded LDLR at the plasma membrane and in a reduction in its rate of degradation. One possible explanation for this data is that some Derlin molecules localize to the plasma membrane and/or endosomes where they regulate the fates of misfolded proteins at the cell surface. We therefore assayed the localization of Derlin proteins in more detail.

Der1p and Dfm1p in yeast, and Derlin-1 and Derlin-2 in humans, localize to the ER (Hitt and Wolf, 2004; Knop et al., 1996; Lilley and Ploegh, 2004; Lilley and Ploegh, 2005; Ye et al., 2004). To confirm the ER localization of CUP-2, we fused GFP to its C-terminus. Expression of this CUP-2::GFP fusion under the control of the coelomocyte promoter rescues both the endocytosis and the UPR activation defects of cup-2(ar506) in coelomocytes, indicating that the fusion protein is functional and that CUP-2 acts cell autonomously (Fig. 5A-C). In wild-type coelomocytes, CUP-2::GFP colocalizes extensively with the smooth ER marker cytochrome b5 and the rough ER marker TRAM, indicating that at steady state, the majority of CUP-2 molecules reside in the ER (Fig. 5D) (Rolls et al., 2002). However, in addition to the ER staining, we consistently detected CUP-2::GFP, but not the other ER markers, in peripheral organelles, suggesting that CUP-2 localizes to other compartments besides the ER (Fig. 5D, arrows). This peripheral staining was not of the Golgi complex that we visualized as discrete centralized puncta in coelomocytes (see Fig. 3D) (Bednarek et al., 2007; Dang et al., 2004; Patton et al., 2005; Treusch et al., 2004). At least a portion of the peripheral CUP-2 staining was endosomal because it appeared to colocalize with RAB-5, although the elaborate nature of the ER in coelomocytes precludes an unambiguous determination (Fig. 5D, arrows).

We determined the subcellular localization of endogenous mammalian Derlin proteins to confirm this extra-ER localization. We first used immunofluorescence comparing the localization of Derlin proteins to the ER-marker calreticulin. We observed significant colocalization of Derlin-1 with calreticulin and of Derlin-2 with calreticulin in RAW264.7 macrophages (Fig. 6A). However, there were vesicular structures that labeled for endogenous Derlin proteins but that did not contain calreticulin (arrows in Fig. 6A). We then transfected RAW264.7 cells with GFP- or YFP-tagged Rab5 (early endosomes), Rab7 (late endosomes) and Rab11a (recycling endosomes), that had been fixed, and immunostained to detect endogenous Derlin proteins and GFP or YFP. We saw significant colocalization between Derlin proteins and Rab5 and Rab7, indicating Derlin proteins are also found on endosomes (arrows in Fig. 6B). The Derlin proteins and Rab11a showed very limited, if any, colocalization in the perinuclear region (Fig. 6B). The colocalization of Derlin proteins with the Rab proteins is not an indirect consequence of Rab overexpression affecting ER integrity because we did not observe any colocalization between the Rab proteins and calreticulin in the same Rab-transfected cells (supplementary material Fig. S5).

We used immuno-electron microscopy to determine Derlin protein localization at a higher resolution. We first allowed cells to endocytose BSA-gold (15 nm) for 10 minutes to unambiguously label endosomal compartments (Fig. 6C). We then added anti-Derlin-1, anti-Derlin-2, or PBS (control) to cells before visualization using 6-nm-gold-conjugated secondary antibodies. RAW264.7 endosomes containing the 15-nm-gold particles showed peripheral staining for Derlin-1 (25/27 endosomes) and for Derlin-2 (24/25 endosomes). Only 1 out of 16 endosomes had 6-nm-gold particles in the control. Similar staining of the Derl1 RNAi cells revealed reduced staining for Derlin-1 (6/17 endosomes) but not for Derlin-2 (14/14 endosomes) (data not shown). Although we detected specific Derlin-1 and Derlin-2 labeling near the plasma membrane, we could not ascertain whether these represented actual plasma membrane labeling or labeling of subcortical organellar (ER, endosomes) membranes.

Finally, we fractionated the post-nuclear membrane fraction of murine RAW264.7 macrophages on a continuous iodixanol gradient to biochemically probe the presence of Derlin proteins in compartments besides the ER. Endogenous Derlin-1 and Derlin-2 was separated in two pools (boxed in Fig. 6D). Pool I (fractions 9-16) included ER membranes. Pool II (fractions 2-5) did not include ER membrane but did include Golgi complex, endosomes, and plasma membrane. Derlin-3 was not expressed in these cells (see Fig. 4A). These results indicate that Derlin proteins have a conserved localization to endosomal compartments in addition to the ER.

We describe the identification of a null mutation in cup-2, the C. elegans orthologue of Derlin proteins, on the basis of an endocytosis defect in specialized scavenger cells called coelomocytes. We show that in cup-2 mutant worms, the UPR is activated indicating that CUP-2 is required for ERAD. Rescue experiments indicate that human Derlin-1 and Derlin-3, but not human Derlin-2, yeast Der1p, or Dfm1p are able to substitute for CUP-2 in vivo. R151.6 and CUP-2 are functionally redundant, because reducing R151.6 activity further exacerbates the ERAD defect of cup-2(ar506) and overexpressing R151.6 rescues both the endocytosis and the ERAD defects of cup-2(ar506).

There is an increase in the levels of GFP::MCA-3 at the plasma membrane of cup-2 mutant C. elegans coelomocytes and in the levels of LDLR at the plasma membrane of Derl1 RNAi RAW264.7 macrophages. We propose two models for this observation. In the first model, reductions in Derlin protein levels lead to the accumulation of misfolded GFP::MCA-3 or LDLR in the ER and therefore an increase in the secretion of these misfolded proteins to the plasma membrane. In this model, the ERAD and/or the UPR activation defects lead to this observed accumulation of transmembrane proteins at the plasma membrane. In the second model, Derlin proteins have a second function in the endosomes of cells, where they target misfolded transmembrane proteins for degradation. In this model, transmembrane proteins at the plasma membrane that have become misfolded are endocytosed into early endosomes where they are recognized by a Derlin-dependent quality control system and targeted for proteasomal degradation. In both models, the accumulation of misfolded proteins at the plasma membrane and in endosomes would lead to defects in cellular functions, for example, a reduction in the rate of endocytosis in cup-2 mutant coelomocytes because of the hyper-recruitment or sequestration of clathrin heavy chain to membranes and a corresponding decrease of available soluble clathrin. This phenotype, which we have so far only seen in cup-2 and not in other cup mutants, might reduce rates of internalization and/or intracellular transport of integral membrane proteins, leading to further increases in their levels at the surfaces of cells. We note that this increased recruitment of clathrin heavy chain to membranes was the only defect that we observed in the endo-lysosomal pathway in coelomocytes.

There are several observations that are more consistent with the second model. First, there was no correspondence between the GFP:MCA-3 accumulation defects and the UPR activation defects of the cup-2 null mutant grown at 15°C or at 20°C. Second, we saw a sharp decrease in the half-lives of LDLR that are misfolded at the plasma membranes of cells: this can be rescued more efficiently using proteasomal than lysosomal inhibitors. Third, although the majority of Derlin proteins localize to the ER, Derlin proteins are also found in endosomes.

There are several transmembrane proteins at the plasma membrane that are thought to be degraded primarily via the proteasome and not by lysosomes. For example, although some CFTR mutants at the plasma membrane seem to be degraded primarily by lysosomes, other CFTR mutants are degraded primarily via the proteasome and do not require lysosomal function (Benharouga et al., 2001; Sharma et al., 2004). Under normal conditions, the degradation of some integral membrane proteins at the plasma membrane, for example LDLR and the IFN-γ receptor 2 (IFNGR-2), requires their endocytosis followed by their proteasome-dependent degradation (Curry et al., 2004; Martin de Llano et al., 2006; Miura et al., 1996). There is also substantial evidence that one pathway for the degradation of gap junctional connexins involves the proteasome following the phosphorylation and the ubiquitylation of connexins, although it is not clear where this proteasome-mediated degradation takes place (Berthoud et al., 2004). Our results are consistent with the model that Derlin proteins might recognize these, and other, proteins at the plasma membrane or endosomes to target them for degradation.

There are physiological arguments for the presence of a quality control mechanism at the plasma membrane or endosomes. Most integral membrane proteins have long half-lives and are efficiently recycled back to the plasma membrane as a result of signals in their transmembrane and/or cytoplasmic domains (Maxfield and McGraw, 2004; Zaliauskiene et al., 2000). Without a specific quality control mechanism that monitors the folding state of these surface proteins, the misfolding of the extracellular domains of these proteins would probably not affect their recycling back to the plasma membrane following their endocytosis. This would result in their accumulation at the plasma membrane and in endosomes. Therefore, the relative importance of such a proposed quality control mechanism in endosomes would depend on how efficiently a transmembrane protein is recycled to the plasma membrane compared with being targeted to lysosomes.

Our results are consistent with a model where Derlin proteins are part of a quality control mechanism that recognizes misfolded plasma membrane or endosomal proteins and targets them for degradation. Future studies will focus on elucidating the transport steps and mechanism of this novel cellular process.

Standard methods were used for growth and genetic analysis of worms (Brenner, 1974). Integration of plasmid DNA was done by microparticle bombardment into unc-119(ed3) worms as previously described, except that we used plasmids pHD134 or pHD137 as co-bombardment markers (Praitis et al., 2001). Markers used: cup-2(ar506) I (Fares and Greenwald, 2001a); atf-6(ok551) X; pek-1(ok275) X (Shen et al., 2001); ire-1(v33) II (Shen et al., 2001); xbp-1(zc12) III; rme-1(b1045) V (Grant et al., 2001). zcIs4[hsp-4::GFP] V integrants express GFP under the control of the hsp-4 promoter (Calfon et al., 2002); the arIs37[pmyo-3::ssGFP] I line used to visualize uptake by coelomocytes was previously described (Fares and Greenwald, 2001a); cdIs70[pcc1::GFP::MCA-3a; unc-1119(+)-pmyo-2::GFP] expresses GFP-MCA-3 in coelomocytes and GFP in the pharynx.

Percentage embryonic lethality is a determination of the number of eggs that hatched after being laid by an adult hermaphrodite. Five different hermaphrodites were assayed for each strain. The results are represented as means ± s.d. cup-2; ire-1 double mutants were identified from cup-2(ar506); ire-1(v33)/dpy-10(e128) unc-4(e120) balanced hermaphrodites: two-thirds of the progeny laid by nonDpy-nonUnc hermaphrodites of this strain gave fully viable progeny that segregated Dpy-Unc worms and one-third laid eggs that did not hatch (presumed to be the cup-2; ire-1 double mutants).

RNA-mediated interference was done by feeding (Timmons and Fire, 1998). Most of the clones were from an existing genome-wide RNAi library (Kamath and Ahringer, 2003). Genes that were missing from that library were PCR amplified from cDNA and cloned into the appropriate vector (Timmons and Fire, 1998).

Murine RAW264.7 macrophages (ATCC, Manassas, VA) were grown in Dulbecco's modified Eagle medium (DMEM) containing 2 mM Glutamax and supplemented with 10% fetal bovine serum, 100 U/ml penicillin, and 100 μg/ml streptomycin (Invitrogen, Carlsbad, CA) at 37°C in 95% air at 5% carbon dioxide. Transfections were done using the Lipofectamine LTX Reagent (Invitrogen, Carlsbad, CA).

Standard methods were used for the manipulation of recombinant DNA (Sambrook et al., 1989). Polymerase chain reaction (PCR) was done using the Expand Long Template PCR System (Boehringer Mannheim, Mannheim, Germany) according to the manufacturer's instructions. All other enzymes were from New England Biolabs (Ipswich, MA), unless otherwise indicated.

We sequenced the cDNA clones yk1619h02 and yk1406f08 in full. This confirmed the predicted splicing pattern of cup-2.

The plasmid pHD137 carries both wild type unc-119 and pmyo-2::GFP and was made by subcloning the 2.7 kb SphI-ApaI fragment from pPD118.33 into the XhoI-ApaI sites of plasmid MM106B after blunt ending the ends with T4 DNA polymerase. The plasmid pHD134 carries both wild-type unc-119 and ttx-3::GFP and was made by subcloning the 2.7 kb HindIII-ApaI fragment from pPD95.75-prom5 into the same sites of plasmid MM106B (Altun-Gultekin et al., 2001; Praitis et al., 2001). Plasmid pHD43 contains a minimal coelomocyte-specific promoter and was made by PCR amplifying a 200 bp fragment (primers: 5′-CACACAGCATGCGTTGACACGCAGTTTC-3′ and 5′-CACACAGGTACCATTGTGAGCCCAATG-3′; template: pOLFGFP-2), restriction digesting with SphI and KpnI and ligating into the same sites of pPD117.01 (Fares and Greenwald, 2001).

To make pHD79, a translational fusion of CUP-2 to GFP under the control of the coelomocyte-specific promoter, we PCR amplified a 1 kb fragment (primers: 5′-CACACAGGTACCAAAAATGGATTTAGAAAATTTCC-3′ and 5′-CACACAGGTACCGAATTTCCTCCAAGTCTAGCTCC-3′; template: worm genomic DNA), restriction digested with KpnI and ligated into the same site of pHD43. To make pHD213, a transcriptional fusion of the CUP-2 promoter to GFP, we PCR amplified a 3.3 kb fragment upstream of the CUP-2 ATG (primers: 5′-CACACAGCATGCCTGCACACTGCTCCACGTCCACCATC-3′ and 5′-CACACAGGTACCCTGAAAATGTATTGATTTGCAGTG-3′; template: worm genomic DNA), restriction digested with SphI and KpnI and replaced the 200 bp SphI-KpnI fragment of pHD43. To make pHD264, a translational fusion of CUP-2 to GFP under the control of the CUP-2 promoter, the 1 kb KpnI fragment from pHD79 was ligated into the same site of pHD213.

To make pHD242, cup-2 cDNA under the control of the CUP-2 promoter, we PCR amplified a 738 bp fragment (primers: 5′-CACACAGGTACCAAAATGGATTTAGAAAATTTCCTTCTC-3′ and 5′-CACACAGCTAGCTTAATTTCCTCCAAGTCTAGCTC-3′; template: worm cDNA library), restriction digested with KpnI and NheI and replaced the 904 bp KpnI-NheI GFP-containing fragment of pHD213. To make pHD257, R151.6 cDNA under the control of the CUP-2 promoter, we PCR amplified a 693 bp fragment (primers: 5′-CACACAGGTACCAAAATGCCACCGGTAACTCGGTTCTAC-3′ and 5′-CACACAGCTAGCTTAATCATGTTGTTCTTGTTCC-3′; template: worm cDNA library), restriction digested with KpnI and NheI and replaced the 904 bp KpnI-NheI GFP-containing fragment of pHD213. To make pHD257, DER1 cDNA under the control of the CUP-2 promoter, we PCR amplified a 645 bp fragment (primers: 5′-CACACAGGTACCAAAATGGATGCTGTAATACTGAATC-3′ and 5′-CACACAGCTAGCTTAGGGTGTTTCAGTGTTGCGG-3′; template: yeast genomic DNA), restriction digested with KpnI and NheI and replaced the 904 bp KpnI-NheI GFP-containing fragment of pHD213. To make pHD249, DFM1 cDNA under the control of the CUP-2 promoter, we PCR amplified a 1026 bp fragment (primers: 5′-CACACAACCGGTAAAATGGCAGGCCCAAGGAATGTGC-3′ and 5′-CACACAGCTAGCTTAGGGTGTTTCAGTGTTGCGG-3′; template: yeast genomic DNA), restriction digested with AgeI and NheI and replaced the 901 bp AgeI-NheI GFP-containing fragment of pHD213. To make pHD243, Derl1 cDNA under the control of the CUP-2 promoter, we PCR amplified a 765 bp fragment (primers: 5′-CACACAGGTACCAAAATGTCGGACATCGGAGACTGG-3′ and 5′-CACACAGCCGGCTCACTGGTCTCCAAGTCGAAAG-3′; template: human cDNA clone), restriction digested with KpnI and NaeI and replaced the 908 bp KpnI-NaeI GFP-containing fragment of pHD213. To make pHD240, Derl2 cDNA under the control of the CUP-2 promoter, we PCR amplified a 728 bp fragment (primers: 5′-CACACAGGTACCAAAATGGCGTACCAGAGCTTGCGGCTG-3′ and 5′-CACACAGCTAGCTAACCTCCAAGCCGCTGGCCCTCAC-3′; template: human cDNA clone), restriction digested with KpnI and NheI and replaced the 904 bp KpnI-NheI GFP-containing fragment of pHD213. To make pHD239, Derl3 cDNA under the control of the CUP-2 promoter, we PCR amplified a 717 bp fragment (primers: 5′-CACACAGGTACCAAAATGGCGTGGCAGGGACTAGCGGC-3′ and 5′-CACACAGCTAGCTCACTGCTGCGGGGGTGGCAGATGG-3′; template: human cDNA clone), restriction digested with KpnI and NheI and replaced the 904 bp KpnI-NheI GFP-containing fragment of pHD213.

To make plasmid pHD276, a C31E10.7 (cytochrome b5)::mCherry fusion protein under the coelomocyte-specific promoter, we first PCR amplified a 1 kb fragment (primers: 5′-CACACAGGTACCAAAAATGATTTTCAGAATGGCCGATC-3′ and 5′-CACACAGGTACCGCAGCGATAAGATAATAAACAAGAGC-3′; template: worm genomic DNA), restriction digested with KpnI and ligated it to the same site of pHD43 to give plasmid pHD189. To make plasmid pHD276, we ligated the 1206 bp SphI-AgeI fragment from plasmid pHD189 to the 2970 bp SphI-AgeI fragment of pH246.

Plasmid pHD306 expressing the mouse Derl1 shRNA was made by annealing and ligating the two complimentary oligos 84692 (5′-GATCCCCCAGTATTCTACTCGGCTTGTTCAAGAGACAAGCCGAGTAGAATACTGTTTTTA-3′) and 84693 (5′-AGCTTAAAAACAGTATTCTACTCGGCTTGTCTCTTGAACAAGCCGAGTAGAATACTGGGG-3′) into BglII-HindIII cut pSUPER-neo (Oligoengine, Seattle, WA). This shRNA is expressed under the control of the histone H1 promoter and targets the sequence 5′-CAGUAUUCUACUCGGCUUG-3′ in the mouse Derl1 mRNA.

We took confocal images of coelomocytes in young adult hermaphrodites grown at the various temperatures to quantify the fluorescence from the zcIs4[hsp-4::GFP] transgene in various backgrounds. The images were taken at the same magnification and exposure and included a cross-section through the nuclei of the coelomocytes. Images were examined using Adobe Photoshop (Adobe Systems Incorporated, San Jose, CA). We selected the nuclei and the average intensity was determined. This number was divided by the total number of pixels of the selected region to normalize over the area.

We took cross-sectional confocal images of at least 20 coelomocytes in young adult hermaphrodites grown at the various temperatures to quantify the fluorescence from the GFP::MCA-3 at the plasma membrane in various backgrounds. The images were taken at the same magnification and exposure. Images were opened using Metamorph software (Molecular Devices, Sunnyvale, CA). We used the `Line Scan' tool to determine the maximum intensity per pixel of GFP::MCA-3 at various points on the plasma membrane from various coelomocytes. The line intersected with two points of the plasma membrane at any level of the image and the peak intensity per pixel was noted for each point. For all quantifications, the reported values reflect the average from at least 60 measurements ± s.d.

Western analysis of worms was done on synchronized adult hermaphrodites grown at 20°, as previously described (Grant et al., 2001).

Approximately 2-4 ml of packed adult hermaphrodites were resuspended in 20 ml of buffer D (20 mM Tris-HCl pH 8.0, 10% Glycerol, 150 mM NaCl) after washing the worms with this buffer. β-mercaptoethanol was added to a final concentration of 7 mM and 200 μl of a solution of protease inhibitors was added per 10 ml of worm suspension (Sigma). Worms were homogenized by french pressing three times at 1000-1400 p.s.i. (ratio `high') and then spun at 5000 r.p.m. for 10 minutes. The supernatant was spun at 125,000 g for 2 hours and the pellet was resuspended in 10 ml of buffer D with β-mercaptoethanol and protease inhibitors and adjusted to 1% Triton X-100. The pellet was homogenized using a Dounce homogenizer, stirred for 2 hours at 4°C, spun down at 35,000 g for 30 minutes, and the supernatant was dialyzed twice, 2 hours each time, in 2 liters IPP50 (20 mM Tris-HCl pH 8.0, 150 mM NaCl, 0.05% Triton X-100). 10 μl of antibody, and 200 μl of beads (washed with IPP150) were added to 4 ml of this solution for each immunoprecipitation. Protein A/G beads were used for immunoprecipitation (Pierce Biotechnology, Rockford, IL). After overnight binding at 4°C, the beads were washed four times with 50 mM HEPES 7.7, 150 mM NaCl, 1% NP-40. To elute, beads were incubated in 100 μl of 1× SDS buffer (50 mM Tris-HCl pH 8.5, 1% SDS, 2 mM DTT) for 5 minutes at 95°C. The beads were centrifuged at 2000 r.p.m. for 1 minute and 80 μl were recovered. This procedure was repeated with 70 μl SDS buffer added to the beads. This time, 70 μl was recovered and mixed with the previous 80 μl. 150 μl of 2× western loading buffer was added, and the mixture was boiled before loading on a gel.

Equal volumes of samples were run on gels and probed with the goat anti-GFP antibody. Since the GFP::MCA-3 transgene also expresses GFP in the pharynx, we used it as an internal control. In each lane, we measured the intensity of the GFP::MCA-3 band and divided it by the intensity of the GFP band to give us a normalized measurement. Samples were run three different times to ensure reproducibility. Measurements were done using ImageJ software (N.I.H., Bethesda, MD).

BSA-Rhodamine was diluted to 1 mg/ml in M9 medium and injected into the pseudocoelomic space of adult hermaphrodites, as previously described (Dang et al., 2004; Treusch et al., 2004). Immediately following the injections, worms were placed in 6 μl ice-cold 1 mM levamisole in 1× PBS on a 2.2% agarose pad on a glass slide and covered with a coverslip. Confocal images of the same coelomocyte were taken at different times after the worms were shifted to room temperature.

We identified a stable transfectant, LS8, using plasmid pHD306. To determine the efficiency of the RNAi, we ran 10 μg total protein from RAW264.7 and LS8 cells on a gel. The blot was probed with antibodies against Derlin-1, Derlin-2 or Derlin-3, and actin. The same filter was probed to detect Derlin proteins (28 kDa) and actin (42 kDa). The intensity of the bands was quantified using ImageJ software. To determine the levels of Derlin proteins in LS8 relative to RAW264.7 cells, we divided the intensity of the Derlin band by that of actin to get a `relative level' in each strain. This value from LS8 was divided by that from RAW264.7 cells and multiplied by 100 to get percentage change of Derlin protein levels. The protein isolation and western blots were repeated three times to calculate means ± s.d.

For conventional immunofluorescence, cells that were grown on coverslips were fixed for 20 minutes in 4% paraformaldehyde in PBS at RT or in 100% methanol (kept at –20°C) for 15 minutes at –20°C. Cells were washed three times with PBS at RT, 5 minutes each time. Paraformaldehyde-fixed cells were incubated in 50 mM NH₄Cl in PBS for 10 minutes at RT and washed twice more with PBS. Blocking was done for 30 minutes in blocking buffer (1% BSA, 0.1% saponin in PBS). Cells were then incubated in primary antibodies diluted in blocking buffer for 2 hours at RT, washed three times with PBS, incubated in Cy2- or Cy3-labeled secondary antibodies (Jackson ImmunoResearch Laboratories, West Grove, PA) diluted 1:200 in blocking buffer for 1 hour at RT, washed three times with PBS, and mounted in Slowfade mounting medium (Invitrogen) on slides for viewing.

For surface staining, cells that were grown for 24 hours were first washed three times with ice-cold PBS and were then incubated with primary antibody diluted in 1×PBS with 1 mg/ml BSA at 4°C for 2 hours. Cells were then washed three times, five minutes each, with 1×PBS with 1 mg/ml BSA at 4°C and were then incubated in secondary antibody diluted in 1×PBS with 1 mg/ml BSA at 4°C for 1 hour. Cells were washed again and then fixed in 1% formaldehyde for 1 hour at 4°C before washing with PBS and loading on slides. The confocal images of these cells were acquired using the same exposure and magnification. The intensity of the cell surface staining was determined using Adobe Photoshop.

Cells were grown to 95% confluency in 35 mm plates on 25 mm diameter Thermanox coverslips (Electron Microscopy Sciences, Hatfield, PA). BSA conjugated to 15 nm colloidal gold (Electron Microscopy Sciences) was washed once in DMEM medium (no serum) and added to cells at a final OD₅₂₀ of 5. After a 10 minute incubation at 37°C, cells were washed three times with PBS/BSA (10 mg/ml) and fixed in 2% formaldehyde, 0.5% glutaraldehyde, in 0.1 M PBS using three cycles, each cycle comprised of 2 minutes in the microwave (160 W) followed by 2 minutes cooling at room temperature. Following 0.1 M PBS rinses, cells were post-fixed in 0.5% osmium tetroxide, 2% potassium ferricyanide using two cycles, each cycle comprised of 10 seconds in the microwave (160 W) followed by 20 seconds of cooling at room temperature. Following rinses with ddH₂O, cells were dehydrated using step-wise incubations in ethanol, each with 40 seconds in the microwave, and embedded in Spurrs epoxy resin. 60-nm-thick sections were cut on a diamond knife and mounted on Nickel grids. Sections were blocked in 0.5% Tween-20 in PBS for 2.5 minutes at RT. Rabbit anti-Derlin-1, rabbit anti-Derlin-2, or PBS (control) was then added at a 1:10 dilution in PBS to sections and the samples were microwaved for 2.5 minutes. Following rinses with PBS, samples were incubated with 6-nm-gold-conjugated donkey anti-rabbit IgG (Jackson ImmunoResearch Labs, West Grove, PA) at a 1:30 dilution and the samples were microwaved for 2.5 minutes. Following rinses with PBS, samples were post-fixed in 4% formaldehyde, 1% glutaraldehyde, washed with ddH₂O, and images were acquired using a JHEOL 100 CX II TEM.

Density gradient centrifugation of RAW264.7 cell membranes was performed essentially as described previously (Frank et al., 1998; van't Hof and Resh, 1997). Briefly, cells harvested from ten 100 mm semi-confluent plates were lysed in lysis buffer (0.25 M sucrose, 1 mM EDTA, 10 mM HEPES-NaOH pH7.4, Roche Complete Inhibitor) using 15 strokes of a Dounce homogenizer followed by three 15-second sonications at low power. The sample was centrifuged at 3000 g for 10 minutes at 4°C and the supernatant was then centrifuged at 100,000 g for 40 minutes at 4°C. The pellet was resuspended in 1 ml lysis buffer and layered on top of a 35 ml continuous iodixanol (Optiprep, Invitrogen) density gradient (2.5-25% w/v). This was spun at 130,000 g for 3 hours at 4°C in a swing-bucket rotor and 2 ml fractions were removed starting from the top. For western blotting, fractions were first concentrated by trichloroacetic acid precipitation with 25 μg/ml BSA as carrier. Equal volumes of fractions were run in each lane for western analysis.

Cells on coverslips were incubated in serum-free DMEM at 37°C for 4 hours. Cells were then moved to the cold room and washed twice with ice-cold serum-free DMEM. Some coverslips were then treated with ice-cold D-PBS for 15 minutes at 4°C (no misfolding) and others were treated with ice-cold misfolding Buffer (741 mM citric acid, 258.7 mM sodium citrate, pH 3.5) for 15 minutes at 4°C (misfolding). Cells were then washed three times with ice-cold D-PBS, twice with ice-cold serum-free DMEM, and incubated for 1 hour at 4°C in 15 μg/ml Bodipy FL-LDL (Invitrogen) diluted in serum-free DMEM. Cells were then washed three times with ice-cold serum-free DMEM, fixed (1% formaldehyde, 1 mM CaCl₂, D-PBS) for 1 hour at 4°C, washed three times with D-PBS, and loaded on slides. The confocal images of all cells were acquired using the same exposure and magnification. The intensity of the cell surface staining was determined using Adobe Photoshop.

Approximately 12×10⁷ cells were incubated for 2 hours at 37°C in suspension in 10 ml of DMEM-FBS medium containing 25 μg/ml cycloheximide (Sigma). Cells were spun down (1000 g for 7 minutes) and incubated either in 10 ml of ice-cold D-PBS (no misfolding) or in 10 ml of ice-cold misfolding buffer (741 mM citric acid, 258.7 mM sodium citrate, pH 3.5) for 15 minutes at 4°C. Cells were then spun down as before at 4°C, washed once with ice-cold D-PBS, and resuspended in 15 ml of ice-cold DMEM/FBS + 25 μg/ml cycloheximide. The 15 ml cell suspension was divided between three tubes kept on ice. Nothing further was added to one. The second culture was supplemented with 50 μM MG-132 (Sigma). The third culture was supplemented with 0.1 mM Leupeptin (Sigma). Cells were shaken at 37°C and 1 ml aliquots were taken from each culture at 0, 2, 4 and 6 hours. The 1 ml aliquots were spun down at 4°C and the cell pellets were resuspended in 150 μl western loading buffer (50 mM Tris-HCl pH 6.8, 10% glycerol, 4% SDS, 10 mM DTT) supplemented with Complete Inhibitor (Roche). The protein concentration was determined for each sample (BioRad DCProtein Assay, Hercules, CA) and 10 μg of each sample was loaded on gels for western blotting. Quantification of the bands was done using ImageJ software (National Institute of Mental Health, Bethesda, MD) and the graphs represent the data from three separate experiments. There was no change in the viability of cells over the time-course of this experiment as determine by the MTT assay (Cory et al., 1991; Mosmann, 1983).

Confocal images were taken with a Nikon PCM 2000, using HeNe 543 excitation for the red dye and argon 488 for the green dye. Epifluorescence images were taken with a Nikon Eclipse E800 microscope. Deconvolution Images were acquired (×100, 1.4 NA objective, 0.2 mm z-sections) on a DeltaVision RT system (Applied Precision, LLC, Issaquah, WA) using a Series 300 CCD camera (Photometrics, Tucson, AZ) and deconvolved using DeltaVision software. For worm imaging, young adult hermaphrodites were paralyzed in 10 mM levamisole.

Antibodies used were goat anti-GFP polyclonal conjugated to HRP for Western blots (Research Diagnostics, Concord, MA), rabbit anti-GFP polyclonal for immunoprecipitation (Abcam, Cambridge, UK), rabbit anti-RME-1 polyclonal (Grant et al., 2001), rabbit anti-Derlin-1 (MBL International, Woburn, MA), rabbit anti-Derlin-2 (MBL International), goat anti-Derlin-3 (Santa Cruz Biotechnology, Santa Cruz, CA), rat anti-mouse CD16/CD32 clone 2.4G2 (BD Biosciences, San Jose, CA), rabbit anti-LDL receptor (Zhang et al., 2007) for surface staining, rabbit anti-LDL receptor (Abcam) for western blotting, rabbit-anti-EEA1 (Abcam), rabbit-anti-α-glucosidase-II (ABR, Golden, CO), rabbit-anti-GM130 (Abcam), rabbit-anti-IFNGR-2 (Santa Cruz Biotechnology), and chicken anti-Calreticulin (CRT) (Genway, San Diego, CA).

For antigen detection, we used Goat anti-Rabbit IgG, Rabbit anti-Chicken IgY, and Swine anti-Goat IgG secondary antibodies conjugated to HRP (1:50,000) and the SuperSignal West Dura Extended Duration Substrate (Pierce).

The Student's t-test was used to compare average measurements from two samples using a two-tailed distribution (Tails=2) and a two-sample unequal variance (Type=2).