EDEM1 reveals a quality control vesicular transport pathway out of the endoplasmic reticulum not involving the COPII exit sites

Abstract

Immature and nonnative proteins are retained in the endoplasmic reticulum (ER) by the quality control machinery. Folding-incompetent glycoproteins are eventually targeted for ER-associated protein degradation (ERAD). EDEM1 (ER degradation-enhancing alpha-mannosidase-like protein 1), a putative mannose-binding protein, targets misfolded glycoproteins for ERAD. We report that endogenous EDEM1 exists mainly as a soluble glycoprotein. By high-resolution immunolabeling and serial section analysis, we find that endogenous EDEM1 is sequestered in buds that form along cisternae of the rough ER at regions outside of the transitional ER. They give rise to approximately 150-nm vesicles scattered throughout the cytoplasm that are lacking a recognizable COPII coat. About 87% of the immunogold labeling was over the vesicles and approximately 11% over the ER lumen. Some of the EDEM1 vesicles also contain Derlin-2 and the misfolded Hong Kong variant of alpha-1-antitrypsin, a substrate for EDEM1 and ERAD. Our results demonstrate the existence of a vesicle budding transport pathway out of the rough ER that does not involve the canonical transitional ER exit sites and therefore represents a previously unrecognized passageway to remove potentially harmful misfolded luminal glycoproteins from the ER.

Fig. 1.

EDEM1 is a heterogeneously glycosylated, soluble protein that comigrates with restricted membrane fractions on Optiprep gradients. (A) EDEM1 was translated in rabbit reticulocyte lysate in the absence or presence of rough ER microsomes (MS). The total lysates were either analyzed directly (IP-) or after immunoprecipitation with either preimmune serum (P) or purified anti-EDEM1 antibodies (E). (B) EDEM1 translated in the presence of rough ER microsomes and subjected to deglycosylation using the indicated concentrations of PNGase F. The number of glycans is designated. (C and D) Isolated canine pancreas rough ER microsomes (C) or membranes from HepG2 cells (D) were alkaline extracted. The total membranes (TM) along with the membrane pellet (P) and soluble supernatant (S) were deglycosylated using PNGase F or Endo H where indicated, resolved on reducing SDS/PAGE, and immunoblotted with EDEM1 antibody (α-EDEM1). (E) Postnuclear cellular membranes from HepG2 cells were loaded onto a discontinuous Optiprep gradient. Fractions collected in the order of increasing density and resolved on a reducing SDS/PAGE were probed for EDEM1 (SI Fig. 6 shows full size blot), the ER markers calnexin and Sec61β, the Golgi marker GM130 and the ERAD proteins Derlin-1 and Derlin-2. P denotes the pellet. (F) Thin section from an Optiprep fraction corresponding to fraction 10 in E shows vesicles and cisternal membrane profiles. (G) Vesicles are positive for EDEM1 by immunoperoxidase electron microscopy. (Scale bars, 500 nm in F and 200 nm in G.)

Fig. 2.

EDEM1 vesicles are formed outside of the transitional ER. (A–C) Confocal immunofluorescence, HepG2 hepatoma cells, staining for EDEM1 (A), calnexin (B), and overlay (C, codistribution is indicated by different shades of yellow). EDEM1 staining is essentially punctate along with some elongated structures, whereas that for calnexin is reticular (B). (C) EDEM1 and calnexin staining partially overlaps in elongated structures. (D and E) Ultrathin frozen sections from HepG2 cells with immunogold labeling for EDEM1 in the ER lumen (arrowheads) and ER-associated smooth vesicles (arrows). Clathrin-coated vesicles (cv in D) are unlabeled. (F–I) Immunoperoxidase labeling for EDEM1 reveals staining in the lumen of rough ER (F, arrowhead), in ER buds (G), and vesicles pinching-off the ER (H, arrowhead) or close to the ER (I, arrowheads). (K) Four serial sections from a HepG2 cell with luminal EDEM1 labeling (black dots) in parts of ER cisternae. Arrowhead in K2 points to ER membrane-associated EDEM1 staining and arrow to cytoplasmic EDEM1 staining. cp in K 1–3: clathrin-coated pit. (Scale bars, 10 μm in A–C; 60 nm in D and E; 80 nm in F; 95 nm in G; 155 nm in H; 130 nm in I; and 400 nm in K.)

Fig. 3.

Overexpressed EDEM1-FLAG, unlike endogenous EDEM1, is present throughout the ER. Immunoblots of postnuclear cellular membrane fractions collected in the order of increasing density of Optiprep gradients. (A–C) CHO cells stably overexpressing EDEM1-FLAG (A and B) and HepG2 cells transiently overexpressing EDEM1-FLAG (C) were resolved by reducing SDS/PAGE and immunoblotted for calnexin (A) or the FLAG tag (B and C). Both calnexin and the EDEM1-FLAG exhibited similar distribution. Confocal immunofluorescence reveals an ER pattern for overexpressed EDEM1-FLAG in CHO cells (D), which is in contrast with the punctate pattern of endogenous EDEM1 in mock-transfected CHO cells (E). (Scale bars, 10 μm in D and E.)

Fig. 4.

EDEM1 does not localize to COPII component and to ER exit sites. (A–C) Confocal immunofluorescence for EDEM1 (A) or Sec23p (B) in CHO cells and overlay (C) are presented. Lack of codistribution is indicated by absence of yellow color best seen at higher magnification in C Inset. (D) Serial sections of HepG2 cells with immunoperoxidase staining for EDEM1. An ER cisterna forms an exit site with coated buds (arrowheads in D 2 and 3) and a vesiculotubular cluster (VTC in D 3 and 4). EDEM1 labeling is evident in the ER lumen (black spots) but undetectable in the coated buds of the transitional ER and the VTC. Arrow in D points to cytoplasmic EDEM1 labeling. For peroxisomes (PO), note the absence of DAB reaction product. (Scale bars, 8 μm in A–C and 400 nm in D.)

Fig. 5.

EDEM1 vesicles contain the HK A1AT. Rat clone9 cells stably expressing HK A1AT were studied. (A and B) Confocal double immunofluorescence for EDEM1 (red), A1AT (green) and overlays (codistribution indicated by yellow-orange) are shown. Enlarged boxed fields in A and B reveal partial codistribution of EDEM1 and HK A1AT in punctate structures (some labeled with arrows). (C–E) By immunoperoxidase labeling, EDEM1 (C) and HK A1AT (D and E) are detectable in ER buds and vesicles. (Scale bars, 1 μm in A and B, 270 nm in C, 285 nm in D, and 255 nm in E.)