EDEM As an Acceptor of Terminally Misfolded Glycoproteins Released from Calnexin

See allHide authors and affiliations

Science  28 Feb 2003:
Vol. 299, Issue 5611, pp. 1394-1397
DOI: 10.1126/science.1079181


Terminally misfolded proteins in the endoplasmic reticulum (ER) are retrotranslocated to the cytoplasm and degraded by proteasomes through a mechanism known as ER-associated degradation (ERAD). EDEM, a postulated Man8B-binding protein, accelerates the degradation of misfolded proteins in the ER. Here, EDEM was shown to interact with calnexin, but not with calreticulin, through its transmembrane region. Both binding of substrates to calnexin and their release from calnexin were required for ERAD to occur. Overexpression of EDEM accelerated ERAD by promoting the release of terminally misfolded proteins from calnexin. Thus, EDEM appeared to function in the ERAD pathway by accepting substrates from calnexin.

The endoplasmic reticulum (ER) possesses a quality-control mechanism that discriminates correctly folded proteins from misfolded proteins (1). The misfolded proteins are dislocated to the cytosol through the translocon and degraded by the ubiquitin-proteasome system known as ER-associated degradation (ERAD) (1–4). It has been established that formation of the mannose–N-acetylglucosamine Man8GlcNAc2 isomer B of N-linked oligosaccharides is the critical luminal event as a signal for ERAD (5–9). EDEM is an ER stress-inducible membrane protein with homology to α-mannosidase but lacking mannosidase activity. Overexpression of EDEM greatly accelerates ERAD, most likely through recognition of the substrate-bound Man8B structure, because the mannosidase inhibitor kifnensine abolished the effects (5). However, the precise role of EDEM in the quality-control machinery is unknown.

In an attempt to search for proteins that work in concert with EDEM, we focused on calnexin (CNX), an ER protein with a potential role in the ERAD pathway (10–12). Although CNX is an abundant transmembrane component with an affinity to monoglucosylated oligosaccharides (13,14), its role in ERAD remains controversial (15–17). We first assessed whether CNX can physically interact with EDEM. Immunoprecipitation of cell lysates with antibodies to CNX and hemagglutinin (HA) revealed that pulse-labeled EDEM-HA was coprecipitated by the antibody to CNX (Fig. 1A). Similarly, immunoprecipitation followed by immunoblotting of cells expressing EDEM-myc also showed that EDEM-myc bound to CNX (Fig. 1B). The observed association was not a result of the chaperone function of CNX, because (i) EDEM remained in the CNX complex even after 1 hour of chase (Fig. 1A), and (ii) EDEM was still coprecipitated with CNX even after cells were treated with castanospermine, an inhibitor of glucosidase, which is known to prevent the binding of substrate proteins to CNX.

Figure 1

EDEM interacts with CNX. (A) HEK 293 cells transfected with EDEM-HA or a mock vector were pulse-labeled for 30 min with [35S]-methionine and chased for 1 hour in the presence or absence of castanospermine. Cell lysates were immunoprecipitated with the antibody to HA or CNX. (B) Whole-cell extract and α-CNX precipitates of cells expressing EDEM-myc were subjected to immunoblotting with antibody to myc or CNX. WB, western blot.

Next, we tested whether EDEM interacts with calreticulin (CRT), a soluble homolog of CNX that is present in the ER lumen (13,14) (Fig. 2A). EDEM-HA was not coprecipitated with the antibody to CRT (Fig. 2B), suggesting that there was no interaction between EDEM and CRT. Although CNX and CRT appear to share similar functions, CNX contains a transmembrane region that CRT lacks (Fig. 2A), suggesting that the transmembrane region might be responsible for the binding of EDEM to CNX. CRT that anchored to the ER membrane by fusing it with the membrane region of CNX (CRT-CNXtm) was revealed to coprecipitate with EDEM-myc (Fig. 2C). On the contrary, soluble CNX without its cytoplasmic and transmembrane segments (FS-CNX) was not coprecipitated with EDEM-myc (Fig. 2D). EDEM was not coimmunoprecipitated with the CNX mutant consisting of the ER luminal region of CNX fused to the membrane region of influenza HA (CNX-HAtm) (Fig. 2E), indicating that the transmembrane domain of CNX is crucial in EDEM-CNX complex formation.

Figure 2

Interaction of EDEM with CRT and CNX mutants. (A) Schematic illustration of the constructs (26,27). Cells were transfected with EDEM constructs (EDEM-HA, EDEM-myc, or EDEM) with cotransfection of CNX or CRT mutants. Immunoprecipitation was performed with the antibodies indicated on the top of each panel. (B) EDEM did not interact with CRT. (C) EDEM-myc interacted with CRT-CNXtm. The asterisk indicates an unknown protein coimmunoprecipitated with CRT-CNXtm. (D) EDEM-myc did not interact with soluble CNX. (E) EDEM did not interact with CNX-HAtm. The arrow indicates EDEM and the arrowhead indicates each CNX or CRT mutant.

To investigate the role that EDEM-CNX complexes play in the ERAD pathway, we used the genetic variant null Hong Kong (NHK) of α1-antitrypsin (A1AT) as a substrate for ERAD (18). NHK is misfolded in the ER and subsequently degraded by the proteasome in the cytosol. The stability of NHK and wild-type A1AT (WTAT) was examined by pulse-chase analyses. Correctly folded WTAT was transported to the Golgi apparatus and its carbohydrates were modified into complex-type glycans as marked by the asterisk (Fig. 3A). In contrast with the wild type, there was a prolonged presence of NHK in the ER during the chase period. Immunoprecipitation with the antibody to CNX revealed that WTAT transiently interacted with CNX only in the pulse period, but that NHK interacted with CNX even after a 30-min chase (Fig. 3A). Because the half-time (t 1/2) of monoglucose in the nascent glycoproteins is about 2 min (19), NHK should be repeatedly reglucosylated by UDP-glucose:glycoprotein glucosyltransferase (UGGT) (13, 14), resulting in association with CNX. Neither WTAT nor NHK interacted with CRT (Fig. 3A).

Figure 3

Effects of CNX on NHK degradation. (A) NHK bound to CNX for a longer time than WTAT. Cells expressing WTAT or NHK were pulse-labeled for 15 min and chased. Aliquots of cell lysates were immunoprecipitated with antibody to A1AT, CRT, or CNX. The asterisk indicates WTAT that contains complex-type glycans. (B) Castanospermine inhibited the EDEM-accelerated degradation of NHK, suggesting that there was involvement of CNX in the EDEM-ERAD pathway. (C) The relative radioactivity of NHK at different chase periods was plotted from the results shown in (B). Error bars indicate standard deviations from four independent experiments.

We next examined whether NHK degradation would be reduced if binding to CNX was inhibited by castanospermine. We confirmed that castanospermine prevented NHK from binding to CNX (20), though it had a modest inhibitory effect on NHK degradation in cells transfected only with NHK (Fig. 3B), which is consistent with previous reports (21, 22). In contrast, castanospermine suppressed the acceleration of EDEM-mediated degradation in cells transfected with NHK and EDEM (Fig. 3B). The presence of castanospermine prolonged thet 1/2 of NHK from about 1 hour to about 2 hours, which is almost equivalent to the t 1/2 in cells lacking EDEM transfection (Fig. 3C). Thus, CNX is involved in the degradation of misfolded proteins mediated by EDEM.

To examine the effects of CNX on the degradation of NHK, we cotransfected cells with NHK and CNX. Degradation of NHK was observed in both NHK and NHK + EDEM transfected cells, albeit much faster in the latter (fig. S1). However, overexpression of CNX suppressed NHK degradation in both cells. The t 1/2 of NHK was prolonged from 1 hour to 2 hours by overexpression of CNX in cells cotransfected with EDEM (fig. S1), suggesting that overexpressed CNX retained NHK and prevented its degradation (23). Next we examined pulse-labeled NHK in cells to which butyl-deoxynojirimycin had been added immediately after pulse-labeling. The butyl-deoxynojirimycin inhibited glucosidase II, which prevented the release of NHK that was bound to CNX. Under these conditions, NHK remained bound to CNX after chase periods (fig. S2), and the degradation of NHK was slowed down either in the absence (fig. S2A) or in the presence (fig. S2B) of EDEM. These results suggest that the release of substrates from CNX is necessary for ERAD, which is consistent with a previous report (24).

Finally, we analyzed the role of CNX-EDEM complexes in the retrotranslocation of NHK. This question was assessed by pulse-chase analyses in the presence or absence of the proteasome inhibitor lactacystin. In cells expressing NHK, lactacystin treatment increased the recovery of NHK that was bound to CNX during chase periods (Fig. 4A), suggesting that a portion of accumulated NHK interacted with CNX. In contrast, when EDEM was overexpressed, lactacystin treatment did not increase the recovery of NHK bound to CNX throughout the chase period (Fig. 4B). Instead, most of NHK during the chase periods was found in association with EDEM in cells treated with lactacystin (Fig. 4B). To confirm these data, we performed sequential immunoprecipitation by first using the antibody to CNX or EDEM and then the antibody to A1AT at various chase points for quantification. In cells without EDEM transfection, NHK was retained to CNX during chase periods in the presence of lactacystin, whereas it was gradually released from CNX in the absence of lactacystin (Fig. 4C). When EDEM was cotransfected, NHK that bound to EDEM increased rapidly (within 15 min) in the presence of lactacystin, concomitant with rapid release of NHK from CNX (Fig. 4D). These observations together suggest that NHK is transferred from CNX to EDEM.

Figure 4

NHK is transferred from CNX to EDEM. Pulse-chase analysis was performed as described in Fig. 3 in the presence or absence of lactacystin. (A) Cell lysates were immunoprecipitated with antibody to A1AT or CNX. (B) Cells cotransfected with NHK and EDEM were pulse labeled and chased in the presence of lactacystin, followed by immunoprecipitation with antibody to A1AT, CNX, or EDEM. (C) Association of NHK with CNX was determined by sequential immunoprecipitation in the presence or absence of lactacystin. Precipitates with the antibody to CNX were solubilized and subjected to a second round of immunoprecipitation using the antibody to A1AT for quantification of the amount of NHK. (D) Association of NHK with CNX and EDEM in the presence of lactacystin in cells that were cotransfected with NHK and EDEM. Sequential immunoprecipitation was performed as in (C).

Thus, EDEM participates in ERAD as an acceptor of terminally misfolded glycoproteins released from CNX. This transfer is secured by the formation of EDEM and CNX complexes, which may ensure that intrinsically unstable misfolded proteins are not liberated into the ER lumen. Mannose trimming to the Man8B form is also a prerequisite for EDEM-mediated degradation of misfolded proteins. When substrates are released from CNX by glucosidase II, they may be transferred to one of two pathways: EDEM for degradation or UGGT for reentry into the CNX cycle (13, 14). Terminally misfolded substrates retained by CNX were released and transferred to EDEM when EDEM was overexpressed in cells. Thus, terminally misfolded substrates enter the EDEM route for degradation rather than the UGGT-CNX cycle after being released from CNX. EDEM is up-regulated that combine glucose (Glc), mannose (Man), and N-acetylglucosamines by the unfolded protein response pathway (5), particularly through the XBP1-pathway (25), whereas in contrast, CNX is a housekeeping gene. Because CNX functions as a platform in the disposal pathway as well as in the productive folding pathway, this differential gene expression suggests that misfolded proteins may accumulate on EDEM but not on CNX, and thus EDEM expression is selectively induced. Further studies are required to understand how misfolded proteins that are bound to EDEM are delivered to the Sec61 channel for retrotranslocation.

Supporting Online Material

Materials and Methods

Figs. S1 and S2


  • * Present address: Institute of Biomedical Sciences, Fukushima Medical University, Fukushima 960–1295, Japan.

  • To whom correspondence should be addressed. E-mail: nagata{at}


View Abstract

Navigate This Article