|Home | About | Journals | Submit | Contact Us | Français|
Endoplasmic reticulum (ER) α-1, 2-mannosidase (ERManI) contributes to ER-associated protein degradation (ERAD) by initiating the formation of degradation signals on misfolded N-linked glycoproteins. Despite its inferred intracellular location, we recently discovered that the mammalian homologue is actually localized to the Golgi complex. In the present study, the functional role of Golgi-situated ERManI was investigated. Mass spectrometry analysis and coimmunoprecipitation (co-IP) identified a direct interaction between ERManI and γ-COP, the gamma subunit of coat protein complex I (COPI) that is responsible for Golgi-to-ER retrograde cargo transport. The functional relationship was validated by the requirement of both ERManI and γ-COP to support efficient intracellular clearance of the classical ERAD substrate, null Hong Kong (NHK). In addition, site-directed mutagenesis of suspected γ-COP–binding motifs in the cytoplasmic tail of ERManI was sufficient to disrupt the physical interaction and ablate NHK degradation. Moreover, a physical interaction between NHK, ERManI, and γ-COP was identified by co-IP and Western blotting. RNA interference–mediated knockdown of γ-COP enhanced the association between ERManI and NHK, while diminishing the efficiency of ERAD. Based on these findings, a model is proposed in which ERManI and γ-COP contribute to a Golgi-based quality control module that facilitates the retrieval of captured ERAD substrates back to the ER.
The selective clearance of conformationally aberrant proteins is a fundamental process required for maintaining protein homeostasis in living cells (Balch et al., 2008 ). Endoplasmic reticulum (ER)-associated degradation (ERAD) is part of a proteostasis network that functions in the eukaryotic secretory pathway. It consists of functional modules that regulate distinct steps of the disposal process, including substrate tagging for dislocation into the cytoplasm and elimination by 26S proteasomes (Plemper and Wolf, 1999 ; Kincaid and Cooper, 2007a ; Bernasconi and Molinari, 2011 ).
ERAD machinery has been speculated to function primarily in the ER because 1) the vast majority of associated components are localized to this organelle (Christianson et al., 2012 ) and 2) many misfolded proteins are statically retained in this compartment prior to their retrograde transport into the cytoplasm for proteasomal degradation (Nehls et al., 2000 ; Spiliotis et al., 2002 ). However, in yeast, numerous studies have demonstrated that proteins carrying subtle conformational defects can escape the ER and travel through more distal compartments of the secretory pathway (Chang and Fink, 1995 ; Jenness et al., 1997 ). Also, in both fission yeast and mammalian cells, proteins that exhibit a severe misfolded conformation can travel to post-ER compartments (Spear and Ng, 2003 ; Kincaid and Cooper, 2007b ). These observations imply that the capacity of the ER to retain misfolded proteins is limited, requiring the involvement of post-ER surveillance systems as a means to prevent premature secretion and to ensure the elimination of escaped misfolded proteins.
Evidence for involvement of the Golgi complex as a component of ERAD has emerged from observations in both yeast and mammalian cells, in which a number of ERAD substrates undergo vesicular cycling through early Golgi compartments (Hammond and Helenius, 1994 ; Caldwell et al., 2001 ; Vashist et al., 2001 ; Kincaid and Cooper, 2007b ). Direct involvement of this organelle in ERAD has been further supported by a recent finding in which the three traditional mammalian Golgi-localized α-1, 2-mannosidases (IA, IB, and IC) contribute to the intracellular degradation of misfolded glycoproteins (Hosokawa et al., 2007 ). Irrespective of these observations, little is known regarding the mechanism by which the Golgi-localized machinery might contribute to the operation of a potential, possibly secondary, quality control module within an expanded ERAD network.
ER α-1, 2-mannosidase I (ERManI) is a type II transmembrane protein capable of initiating the ERAD of numerous misfolded N-glycosylated protein substrates (Jakob et al., 1998 ; Gonzalez et al., 1999 ; Tremblay and Herscovics, 1999 ; Cabral et al., 2000 ; Hosokawa et al., 2003 ; Wu et al., 2003 ). This role was first identified in Saccharomyces cerevisiae as a mannosidase that catalyzes the cleavage of a terminal α-1, 2-mannose unit from the middle B branch of the N-linked glycan Man9GlcNAc to generate Man8GlcNAc (Jelinek-Kelly et al., 1985 ). The corresponding gene cloned from Saccharomyces cerevisiae, designated MNS1 (Camirand et al., 1991 ), was implicated in ERAD, because its deletion was sufficient to severely impair the intracellular degradation of misfolded N-glycosylated protein substrates (Jakob et al., 1998 ). Following this lead, the mammalian orthologue was cloned and designated “ERManI” (Gonzalez et al., 1999 ; Tremblay and Herscovics, 1999 ). The inhibition of ERManI activity, either with the α-1, 2-mannosidase–specific inhibitor kifunensine or small interfering RNA (siRNA)-mediated knockdown, significantly hindered the intracellular degradation of luminal ERAD substrates such as α1-antitrypsin (AAT) variant null Hong Kong (NHK) (Cabral et al., 2000 , 2002 ; Hosokawa et al., 2003 , 2007 ; Wu et al., 2003 ; Karaveg et al., 2005 ; Avezov et al., 2008 ).
Based on the aforementioned observations, ERManI was designated as a critical component in the mammalian glycoprotein ERAD network that contributes to the generation of a glycan-based component of a bipartite degradation signal, in which nonnative protein structure functions as the additional signal component (Sifers, 2010b ). The function of ERManI in cleaving the terminal mannose unit from the B branch appears to be synergistic with the catalytic activity of EDEM1 in unmasking the penultimate α-1, 6–linked mannose on the C branch of the N-glycan, which generates the glycan-based component of the degradation signal in both yeast and mammalian cells (Hirao et al., 2006 ; Olivari et al., 2006 ; Clerc et al., 2009 ; Hosokawa et al., 2009 , 2010 ). Subsequent recognition of the modified glycans by the ER-situated lectins OS9 and XTP3-B is suspected to initiate a series of events that culminates in the retrograde translocation of N-glycosylated ERAD substrates into the cytoplasm for ubiquitinylation and elimination by 26S proteasomes (Satoh et al., 2010 ). The catalytic activity of ERManI is relatively weak under basal conditions due to an extremely low level of the enzyme (Wu et al., 2003 ), implying that mannose units are preferentially removed from glycoproteins that have lingered for long periods in the early secretory pathway. Based on this series of experimental observations, ERManI was designated as a “timer” that initiates the ERAD of newly synthesized N-glycosylated proteins unable to attain native structure (Cabral et al., 2001 ).
Although ~50% homologous to the primary amino acid sequences of Golgi α-1, 2-mannosidases IA, IB, and IC (Gonzalez et al., 1999 ; Tremblay and Herscovics, 1999 ), human ERManI was initially predicted to function in the ER based on the localization of its yeast orthologue MNS1 (Burke et al., 1996 ). However, the subcellular localization of the mammalian orthologue has been controversial, mostly due to the poor specificity of available anti-ERManI polyclonal antibodies that were useful only for transient expression studies (Gonzalez et al., 1999 ; Avezov et al., 2008 ). Recently, after generating a panel of highly specific monoclonal antibodies against human ERManI, we made the discovery that the mammalian orthologue is actually localized to the Golgi complex, and several subsequent lines of evidence supported that conclusion (Pan et al., 2011 ). The spatial separation of this central ERAD component from the remaining ER-situated machinery raised a question as to where it functions as a participant in the mammalian ERAD system. In a subsequent set of experiments, directed localization of recombinant ERManI to the ER led to an enhanced rate of mannose removal, without accelerating ERAD substrate degradation over that of the wild-type molecule (Pan et al., 2011 ). These observations strongly argued against the previously accepted concept that ERManI functions as an enzyme in the ER and support a model in which ERManI's role in ERAD requires residence in the Golgi complex. Consistent with this notion, addition of the ER-to-Golgi recycling amino acid sequence KDEL at the carboxy terminus of NHK significantly hindered intracellular degradation of this molecule (Le et al., 1990 ).
A central goal for the present study was to experimentally investigate how Golgi-localized ERManI contributes to the proposed temporospatial expansion of the mammalian ERAD network (Pan et al., 2011 ). To this end, a direct interaction was identified between ERManI and γ-COP, a subunit of COPI coatomers that promote Golgi-to-ER retrograde vesicular transport. Additional supporting lines of evidence, including a detectable association between ERManI, γ-COP, and NHK, support a model in which the mannosidase contributes to the establishment of a multifunctional gatekeeper that recruits escaped ERAD substrates into COPI vesicles for retrograde transport to the ER.
For obtaining mechanistic insight about the functional significance of Golgi-localized ERManI in ERAD, endogenous immunocomplexes were isolated from MCF7 (a human breast epithelial cell line) cells with 1D6, a well-characterized anti-ERManI monoclonal antibody previously generated by our lab (Pan et al., 2011 ). MCF7 cells were chosen because of the relatively high abundance of ERManI in this cell line (Supplemental Figure S1). The immunocomplexes were resolved by SDS–PAGE and detected by silver staining (Figure 1A). Immunoprecipitated proteins were subjected to in-gel digestion followed by mass spectrometry analysis. ~100 proteins were identified to be specifically associated with ERManI (Supplemental Table S1). Among these, several Golgi-localized proteins (Table 1) were chosen for further validation using reciprocal coimmunoprecipitation (co-IP). As shown in Figure 1, B and C, immunocomplexes containing ERManI and γ-COP were readily detected using this methodology. A similar phenomenon was also observed when the MCF7 cells were replaced with HeLa cells (Figure S2A). Two of the additional candidate proteins, including GBF1 and β′-COP, could be detected in ERManI immunocomplexes, but no ERManI was detected in reciprocal IPs (Figure S2, B and C). Another candidate protein, COG1, could not be validated by IP in either direction (Figure S2D). These results implied that a stable interaction exists between ERManI and γ-COP, whereas weaker or indirect interactions exist between ERManI and β′-COP, GBF1, or COG1. To determine whether the ERManI/γ-COP interaction is direct, we performed an in vitro binding assay using glutathione S-transferase (GST)-ERManI and His-γ-COP purified from bacteria. As shown in Figure 1D, His-γ-COP was pulled down with GST-ERManI, but not by GST or an irrelevant protein Alix-MB (a truncated form of an irrelevant protein named Alix), indicating a direct interaction.
γ-COP (Stenbeck et al., 1992 ; Wegmann et al., 2004 ), β′-COP (Stenbeck et al., 1993 ), and GBF1 (Garcia-Mata and Sztul, 2003 ) are known Golgi-associated proteins that all contribute to the formation of COPI vesicles, which are responsible for the retrieval of Golgi-localized proteins back to the ER (Lowe and Kreis, 1998 ; Nickel et al., 2002 ). For example, γ-COP and β′-COP are the core subunits of COPI coatomers (Nickel et al., 2002 ), and a direct interaction between γ-COP and GBF1 initiates the formation of COPI-coated vesicles on the Golgi membrane (Deng et al., 2009 ). The detection of these proteins in the immunocomplexes implied that ERManI is localized to the Golgi membrane in proximity to where COPI-coated vesicles are formed. Coimmunostaining in HeLa cells was next performed to identify the intracellular location where ERManI and γ-COP interact. As shown in Figure 1E, ERManI staining decorated a typical Golgi structure, which overlapped with the staining for γ-COP. Besides the localization in the Golgi, γ-COP also exhibited punctate staining in the cytoplasm, the latter likely representing the ER–Golgi intermediate compartment in which COPI vesicles are often located (Bannykh et al., 1998 ; Lowe and Kreis, 1998 ; Marie et al., 2009 ). However, no ERManI was observed in these cytoplasmic punctate structures. Considering the presence of both GBF1 and γ-COP in ERManI immunocomplexes, the result strongly indicated that the interaction between ERManI and γ-COP occurs on the Golgi membrane, where γ-COP is recruited (Waters et al., 1991 ) through a process assisted by GBF1 (Szul et al., 2007 ). In addition, because COPI coatomers are recruited to the membrane en bloc (Hara-Kuge et al., 1994 ), ERManI is likely localized at, or near, sites on the Golgi membrane where COPI coats are formed.
Considering the involvement of ERManI in ERAD, the direct interaction identified above raised the possibility that γ-COP functionally participates in ERAD as well. For testing this hypothesis, HeLa cells were transfected separately with three siRNAs, each of which targets a distinct sequence on the γ-COP mRNA. Effects on ERAD were evaluated by monitoring changes in the concentration of the terminally misfolded genetic variant NHK of AAT, which is a representative N-glycosylated ERAD substrate. Transfections with two separate ERManI siRNAs served as positive controls. At 48 h posttransfection, both ERManI siRNAs resulted in ~80% knockdown of endogenous ERManI without altering γ-COP expression (Figure 2A, lanes 2 and 3). In contrast, the manipulation resulted in a twofold increase in the intracellular concentration of NHK and an ~20% increase in extracellular NHK, implying that impaired degradation led to enhanced secretion of the molecules (Figure 2A, compare lanes 2 and 3 with lane 1). Because NHK is missing amino acids required to achieve conformational maturation, these data imply that the intracellular retention system had been saturated to an even greater extent than under control conditions. Importantly, individual transfections with each of the three γ-COP–specific siRNAs led to a 60–70% of knockdown in γ-COP expression, resulting in a 1.7- to 2.3-fold increase in intracellular NHK and ~10% increase in extracellular NHK (Figure 2A, lanes 4–6). Such changes did not reflect induction of the unfolded protein response, as the level of GRP78/BiP (glucose-regulated protein, 78 kDa/immunoglobin-binding protein) was unaffected (Figure 2A, compare lanes 4–6 with lane 1). To further determine whether the increased intracellular and extracellular levels of NHK were due to the hindrance of its degradation, we cotransfected HeLa cells with NHK together with one of the γ-COP siRNAs and monitored the degradation of NHK by metabolic pulse–chase radiolabeling. As shown in Figure 2, B and C, 30% and ~20% of intracellular and extracellular NHK levels, respectively, remained at 5 h after synthesis in cells transfected with control siRNAs. In contrast, ~45% and ~40% of the intracellular and extracellular NHK, respectively, remained in response to transfection with γ-COP–specific siRNAs, indicating NHK degradation was significantly hindered as a result of γ-COP knockdown. These results, in addition to the direct interactions demonstrated above, indicate that ERManI and γ-COP functionally partner in orchestrating the ERAD of an N-glycosylated ERAD substrate.
Previous studies have demonstrated that the overexpression of transfected recombinant ERManI accelerates the intracellular degradation of NHK (Hosokawa et al., 2003 ; Pan et al., 2011 ). For determination of whether this capacity requires the involvement of γ-COP, HeLa cells transfected with NHK cDNA, were cotransfected with either control siRNA or γ-COP–specific siRNA. The fate of newly synthesized NHK was monitored by 35S metabolic pulse–chase radiolabeling 48 h posttransfection. In the presence of γ-COP–specific siRNA, the increased intracellular degradation rate of NHK due to overexpressed ERManI was completely inhibited, and the amount of NHK secreted into the medium was significantly increased as compared with control (Figure 3, B and C). Although the total ERManI level was moderately diminished in the presence of γ-COP–specific siRNAs (Figure 3A, compare lane 3 with lane 2), this could not account for the complete inhibition of intracellular degradation of NHK, because the overall ERManI level was still greater than the endogenous protein detected in untransfected cells (Figure 3A, compare lane 3 with lane 1). These results indicated that the capacity of ERManI to accelerate ERAD of NHK is dependent on γ-COP.
γ-COP is known to directly interact with the cytoplasmic tails of type I transmembrane proteins via a C-terminal KKxx (x represents any amino acids) motif (Harter et al., 1996 ). Whether γ-COP directly interacts with a similar motif in the N-termini of type II transmembrane proteins is still unclear. Several studies have, however, demonstrated that dibasic motifs (RRxx or RKxx) within the N-terminal cytoplasmic tails of type II transmembrane proteins directs their COPI-mediated vesicular retrieval from the Golgi complex back to the ER (Schutze et al., 1994 ; Lowe and Kreis, 1998 ). In support of this notion, two suspected sets of dibasic residues at positions 7–8 and 67–68 were detected in the N-terminal cytoplasmic tail of ERManI (Figure 4A). For determining whether these might mediate direct binding to γ-COP, each set was mutated to di-alanine (AA) residues, and the effect on the ERManI/γ-COP interaction was determined by co-IP. As shown in Figure 4B, as compared with the amount of wild-type ERManI pulled down by γ-COP, neither of the individual mutations (ERManI7A8A, ERManI67A68A) disrupted the association. To determine whether disruption of the ERManI/γ-COP interaction requires mutations at both dibasic motifs, we mutated all four arginine residues into alanines, and the consequences were evaluated by co-IP. As shown in Figure 4C, the combination of mutations significantly inhibited the ERManI interaction with γ-COP (~80%), confirming their dual involvement. For subsequent experiments, the recombinant protein containing both mutations was designated ERManIγ-COP-def.
For determining whether the functional dependence between ERManI and γ-COP reflects their direct interaction, either wild-type ERManI or ERManIγ-COP-def cDNA was cotransfected, along with NHK, into HeLa cells. At 48 h posttransfection, the fate of NHK was monitored by [35S]methionine pulse-labeling. As compared with the effects generated by the wild-type ERManI, expression of the mutated protein led to a significant increase in NHK secretion (Figure 4D), resulting in a 50% increase of total NHK (intracellular plus extracellular) levels at 5 h of chase (Figure 4E), possibly reflecting impaired degradation and enhanced secretion as sequential events. Importantly, the results obtained through the expression of ERManIγ-COP-def were not observed when replaced with either single di-arginine motif mutant (Figure S3). These findings demonstrated that ERManI lost its capacity to accelerate NHK degradation in response to the mutation of both di-arginine groups in the cytoplasmic tail, indicating that direct binding of ERManI to γ-COP is required for the efficient disposal of this ERAD substrate.
It is noteworthy that the failure of recombinant ERManIγ-COP-def to accelerate NHK degradation is mainly reflected by enhanced secretion as compared with cells expressing the wild-type recombinant protein (Figure 4, D and E). The latter observation indicated that the uncoupling of ERManI from ERAD, via loss of γ-COP–binding motifs, was sufficient to exert a dominant-negative effect on ERAD substrate retention, thereby demonstrating the interconnected nature of substrate retention and retrieval. To confirm that our observation of ERAD substrate retention was not restricted to NHK, we used metabolic pulse–chase radiolabeling and IP in HeLa cells to examine the consequences of endogenous ERManI knockdown on the fate of another AAT variant, designated ATZ. An ~90% knockdown of endogenous ERManI led to a breach in the intracellular retention of radiolabeled ATZ as judged by a threefold-enhanced secretion into the medium as compared with untreated cells (Figure 5, A and B). We concluded that loss of the actual ERManI molecules, rather than the inhibition of ERAD, had led to enhanced ATZ secretion, because this phenomenon has never been observed (in any cell line) in response to the inhibition of the essential α-1, 2-mannosidase activity (Cabral et al., 2000 , 2002 ; Hosokawa et al., 2003 ; Wu et al., 2003 ; Avezov et al., 2008 ).
Results generated from the preceding set of experiments are consistent with the hypothesis that ERManI functions as part of the Golgi-based retention system that precedes the incorporation of captured ERAD substrates into COPI vesicles en route to the ER. γ-COP is a major component of COPI coatomers that mediate the retrograde transport of protein cargo. The functional interdependence of ERManI and γ-COP during ERAD therefore raised the possibility that the ERAD substrates recognized by ERManI in the Golgi complex can be retrieved back to the ER via COPI vesicles. By localizing at COPI vesicle assembly sites and being endowed with γ-COP–binding motifs, ERManI could contribute to the retention system in a manner that loads ERAD substrates into COPI vesicles. As an initial step in the testing of this critical hypothesis, it was important to determine whether a physical association between ERManI, γ-COP, and ERAD substrates exists. To this end, NHK was transfected into HeLa cells, and its association with the endogenous candidate proteins was examined by a series of co-IP experiments. Consistent with the hypothesis, NHK was identified in the ERManI immunocomplexes (Figure 6A), and both ERManI and NHK were detected in γ-COP immunocomplexes (Figure 6B), indicating their physical associations.
On the positioning of ERAD substrates in close proximity to COPI vesicles, ERManI itself could either be recruited into the vesicles or remain in the Golgi complex. To distinguish between these two possibilities, we transfected NHK into HeLa cells and performed coimmunostaining of ERManI, NHK, and γ-COP. A distinct Golgi localization of ERManI was observed, irrespective of the presence or absence of NHK (Figure 6C), supporting the latter possibility. Furthermore, much of the NHK was dispersed throughout the cytoplasm in a reticular pattern, with only a small portion colocalizing with ERManI and γ-COP in the Golgi complex. This observation implied that only a small portion of NHK escapes beyond the ER at any one time, consistent with previous observations that misfolded proteins are primarily retained in the ER (Nehls et al., 2000 ; Spiliotis et al., 2002 ). Finally, if ERManI detaches from NHK after delivering it to COPI vesicles, then eliminating the latter event should result in more ERManI associated with NHK. For testing this possibility, HeLa cells were transiently transfected with NHK together with an siRNA targeting γ-COP or a scrambled siRNA as control. At 48 h posttransfection, NHK was pulled down, and the amounts of ERManI and γ-COP in the immunocomplexes were examined by Western blotting and quantified by densitometry. As shown in Figure 6D, an ~50% knockdown of γ-COP resulted in an approximately fivefold increase in the amount of ERManI associated with NHK. These results imply that γ-COP is required for the release of ERAD substrates from ERManI-containing complexes during the loading of COPI vesicles en route to the ER.
Finally, to validate that our findings were not restricted to NHK and/or HeLa cells, the human hepatoma cell line Huh7 was transiently transfected with the secretion-impaired AAT variant designated ATZ (Graham et al., 1990 ; Le et al., 1990 ). Its subcellular localization was determined by costaining with antibodies against AAT and ERManI. Unlike NHK, which mainly displayed a reticular pattern throughout the cytoplasm, the localization of ATZ varied in response to its expression level. In cells expressing high levels of ATZ, the majority of the molecules were localized in the cytoplasm as large puncta, possibly reflecting ATZ's polymerized form, as previously reported (Le et al., 1992 ; An et al., 2005 ; Sifers, 2010a ). In these cells, only a small portion of ATZ colocalized with the endogenous ERManI in the Golgi complex. As compared with the cytoplasm-localized ATZ, the staining of Golgi-localized molecules was less intense and more homogenous, indicating a subpopulation of ATZ can escape the ER and travel to the Golgi complex, where ERManI is localized. In cells expressing low levels of ATZ, the majority of the ATZ molecules were localized to the Golgi complex, with the staining pattern overlapping that of ERManI. On the other hand, ERManI exhibited a distinct Golgi-localization pattern, irrespective of the ATZ expression level (Figure 7A). These observations support the notion that a portion of the ATZ population can escape the ER and travel to the Golgi complex, where it can be recycled.
The traditionally accepted role for ERManI in ERAD is that it functions as a mannosidase in the ER, contributing to the generation of degradation signals in response to the persistent retention of misfolded glycoproteins in the early secretory pathway of mammalian cells (Wu et al., 2003 ; Lederkremer, 2009 ). However, our recent discovery of Golgi-localized ERManI, and the fact that a portion of misfolded glycoproteins can escape the ER (this study; Sifers et al., 1989 ; Hosokawa et al., 2007 ) and are recycled through the Golgi complex prior to proteasome-mediated intracellular disposal in both yeast and mammalian cells (Hammond and Helenius, 1994 ; Caldwell et al., 2001 ; Vashist et al., 2001 ; Kincaid and Cooper, 2007b ), demanded a revision of the functional mechanism in which ERManI operates in the Golgi complex as a component of ERAD.
Currently, COPI-mediated vesicle trafficking is the only known route for the recycling of Golgi-situated proteins back to the ER. Theoretically, this routing system would also serve as the prominent mechanism by which ERAD substrates are retrieved from post-ER compartments. The identification of a direct interaction between ERManI and γ-COP supports this idea, and served to validate the notion that γ-COP plays a functional role in the intracellular disposal of the classical ERAD substrate, NHK. The identification of γ-COP–binding sites in the ERManI cytoplasmic tail, plus the consequences of mutating them, provided additional validation for the functional partnership. Our study provides, for the first time, experimental evidence to support this hypothesis.
The incorporation of protein cargo into COPI vesicles is a selective process. The selection of luminal cargo is mediated by groups of transmembrane proteins designated as cargo receptors (Nickel et al., 1998 , 2002 ). In this study, we discovered that the ERManI/γ-COP complex associates with the luminal ERAD substrate NHK. Such an association likely brings ERAD substrates in close proximity to the COPI coatomers, thus allowing their selective recruitment into COPI vesicles. In this regard, ERManI functions similar to a cargo receptor that retains escaped proteins in the Golgi complex prior to facilitating their loading into COPI vesicles. In support of this notion, in the absence of ERManI, the retention of NHK is impaired, allowing its enhanced secretion. The interconnected and transient natures of the aforementioned events were further validated by the enhanced secretion of NHK in response to the mutation of γ-COP-binding sites in the ERManI cytoplasmic tail. Also, RNAi-mediated knockdown of endogenous γ-COP enhanced the association between NHK and ERManI, implying that their dissociation either precedes vesicle loading or is coupled to delivery of the associated ERAD substrate to the ER. Enhanced secretion, possibly caused by saturation of the retention process in the absence of dissociated ERManI, implies that the freed mannosidase is recycled to the Golgi complex where it can functionally support the retention process. Unlike classic cargo receptors recruited with their cargo into COPI vesicles, the movement of ERManI from the Golgi complex has not been detected in our studies. On the basis of these findings, we propose a model in which ERManI contributes to a Golgi-based quality control module that captures ERAD substrates that have escaped from the ER and facilitates their loading into COPI vesicles via a dynamic process driven by an association with γ-COP (Figure 7B). However, considering the detection limits of our methodology, especially after proteins are diluted upon recycling back to the ER, our findings do not exclude the possibility that a trace amount of ERManI actually recycles in COPI vesicles. In fact, a previous proteomics study has indicated that ERManI is present in COPI vesicles (Gilchrist et al., 2006 ), supporting this notion.
The absence of obvious interaction domains in the crystallized three-dimensional structure of the mannosidase (Dole et al., 1997 ) suggests that the association between ERManI and luminal ERAD substrates likely takes place within the context of a much larger complex. In this regard, Cormier et al. (2009 ) reported the existence of a glycan-independent interaction between EDEM1, an evolutionary relative of ERManI, and NHK. Because the human orthologues of ERManI and EDEM1 share 33% sequence identity and 49% similarity (Kanehara et al., 2007 ), it is possible that a similar interaction might exist between ERManI and NHK. Currently, intensive immunoaffinity purification/proteomic investigations are underway to test the validity of this hypothesis and/or to both identify and characterize the components of a much larger complex. An additional prediction, which must be revisited as the subject of future investigations, is that degradation signals initiated by the enzymatic removal of terminal α-1, 2–linked mannose units diverts the recycled molecules away from additional rounds of folding events in response to their recognition by ER-situated lectins (Kanehara et al., 2007 ; Yoshida and Tanaka, 2010 ), all of which precedes their dislocation into the cytoplasm for proteasomal degradation.
Currently, we are not aware of the precise mechanism responsible for regulating the concentration of Golgi-situated ERManI, although previously published experiments have indicated that a significant fraction of the newly synthesized molecules are eliminated by lysosomal proteases (Wu et al., 2007 ). It is apparent, however, that the intracellular residence of the mannosidase does not rely on its direct binding to γ-COP, because ERManIγ-COP-def continued to localize to the Golgi complex (Figure S4A). Also, unlike the capacity of γ-COP knockdown to diminish the steady-state level of the wild-type mannosidase, the ERManIγ-COP-def did not exhibit a diminished intracellular concentration (Figure 4C), implying that the capacity of γ-COP to regulate the concentration of the mannosidase does not rely on a direct interaction. It should be noted, however, that the RNAi-mediated knockdown of γ-COP did result in the dispersal of endogenous ERManI throughout the cytoplasm (Figure S4B). In support of the notion that this phenomenon reflected fragmentation of the Golgi complex (Lippincott-Schwartz et al., 1990 ; Beller et al., 2008 ), we observed that incubating cells in media that contained brefeldin A (BFA), an antibiotic that fragments the Golgi complex by interfering with COPI complex assembly (Peyroche et al., 1999 ; Mossessova et al., 2003 ; Renault et al., 2003 ), led to a significantly diminished steady-state level of ERManI (Figure S4C). However, in contrast to that notion, incubation with nocodazole, a potent drug that is able to induce Golgi disassembly via the depolymerization of microtubules (Lippincott-Schwartz et al., 1990 ; Dinter and Berger, 1998 ), did not alter the steady-state level of ERManI (Figure S4C). Altogether, our observations support the idea that the steady-state level of ERManI is likely regulated by protein trafficking in the secretory pathway, rather than by the structural integrity of the Golgi apparatus.
In summary, our study indicates that Golgi-localized ERManI apparently serves as a linchpin in an expanded mammalian ERAD network in which functional modules are linked by the vesicular transport of cargo. This arrangement provides the temporospatial distancing of essential proteostasis events that would otherwise directly compete with one another if performed in the same compartment. In this regard, ERManI contributes to the establishment of a multifunctional gatekeeper.
Generation of ERManI cDNA was described previously (Pan et al., 2011 ). The di-arginine motif mutations were generated using the QuikChange site-directed mutagenesis kit purchased from Stratagene (La Jolla, CA), following the manufacturer's instructions. The mutagenic primers were designed using the QuikChange primer design program from Stratagene, and the sequences are available upon request. All cDNA constructs were verified by nucleotide sequencing. cDNA constructs of AAT NHK and ATZ were described previously (Wu et al., 2003 ). MGC (mammalian gene collection) human clones containing γ-COP cDNA (IMAGE:6046736) were purchased from the American Type Culture Collection (Manassas, VA). For generation of C-terminal His-tagged γ-COP cDNA plasmid, γ-COP cDNA was amplified by PCR using a forward primer (5′-CCGGATCCGTTGAAGAAATTCGACAAGAAG-3′ and a reverse primer 5′-TTCTCGAGTCCCACAGATGCCAAGATGAT-3′. The inserts were then cloned into pET23b vector using BamHI and XhoI restriction sites. GST-tagged ERManI plasmid was generated previously using pEGX4T1 vector (Pan et al., 2011 ). GST-tagged AlixMB plasmid was generated previously using pEGX4T3 vector (Pan et al., 2008 ). ERManI siRNAs used in this study were purchased from Ambion (Austin, TX), and their sequences were described previously (Pan et al., 2011 ). The three γ-COP siRNAs were purchased from Sigma-Aldrich (St. Louis, MO), and the target sequences are 1) GAGATGTTACCCAGTATCT, 2) CTTGTGAGAGGTCAGACAA, and 3) CTTGTAATCTGGATCTGGA.
Anti-ERManI monoclonal antibodies were generated previously in our group (Pan et al., 2011 ). Anti–γ-COP and anti-His polyclonal antibodies were purchased from Santa Cruz Biotechnology (Santz Cruz, CA). Anti-actin polyclonal antibodies were purchased from Sigma-Aldrich. Anti-BiP monoclonal antibody was purchased from BD Biosciences (San Jose, CA). Anti–human AAT antibodies were purchased from MP Biomedicals (Solon, OH).
HeLa cells were cultured in DMEM (Mediatech, Manassas, VA) supplemented with 10% fetal bovine serum (FBS; Gemini Bio-Products, West Sacramento, CA) and 1% ampicillin/streptomycin (Invitrogen, Carlsbad, CA). MCF7 cells were cultured in RPMI 1640 (Mediatech) supplemented with 10% FBS and 1% ampicillin/streptomycin.
The day prior to transfection, cells were plated onto 60-mm dishes and allowed to reach 80% confluence by the time of transfection. The cDNA plasmids (10 μg) or 20 nM siRNA (20 μl) were transfected into each well with Lipofectamine 2000 (Invitrogen), following the manufacturer's instructions. At 24 h posttransfection, cells were replated into six-well plates, and their culture was continued for an additional 24 h prior to protein extraction or pulse–chase assay. For co-IP, cells were lysed in buffer containing 50 mM Tris-HCl, 150 mM NaCl, 1 mM Na3VO4 (Sigma-Aldrich), 10 mM NaF (Sigma Aldrich), 0.5% CHAPS (Sigma-Aldrich), 2 mM phenylmethylsulfonyl fluoride (Sigma-Aldrich), and additional protease inhibitors (GenDepot, Houston, TX). After centrifugation, the cell extracts were incubated overnight with primary antibodies conjugated to protein G agarose beads. After stringent washing, proteins associated with the beads were eluted and resolved on SDS–PAGE, which was followed by Western blotting, as described previously (Pan et al., 2011 ). For immunofluorescence staining, cells cultured on 18-mm glass coverslips were fixed and stained, and the images were captured following the same procedure described previously (Pan et al., 2011 ).
pET23b-γ-COP, pGEX4T1-ERManI (Pan et al., 2011 ), pGEX4T1, and pGEX4T3-AlixMB (Pan et al., 2008 ) plasmids were each transfected into BL-21 Escherichia coli cells and cultured to ~0.6 of A600 before induction with 0.1 mM isopropyl 1-thio-β-d-galactopyranoside at 37°C for 4 h. The cells were then harvested by centrifugation at 5000 × g for 10 min. The His-γ-COP recombinant proteins were purified under native conditions using QIAexpress Ni-NTA columns (Qiagen, Valencia, CA), following the manufacturer's instructions. GST, GST-ERManI, and GST-AlixMB were purified following procedures described previously (Pan et al., 2006 ).
Fifteen micrograms of purified GST, GST-ERManI, or GST-AlixMB recombinant proteins was conjugated to 15 μl glutathione agarose beads (Sigma-Aldrich) by rotating at 4ºC for 4 h before being mixed with 5 μg γ-COP-His and rotated overnight at 4ºC. The beads were washed with native lysis buffer (Qiagen) five times, and the bound proteins were eluted using Laemmli sample buffer. After being separated by SDS–PAGE, the bound proteins were subjected to Western blotting using anti-His antibodies.
MCF7 cell pellets (~1 ml) were used for large-scale IP with 10 μg of either mouse immunoglobulin G (IgG) or anti-ERManI monoclonal antibodies, and the isolated proteins were identified using mass spectrometry, following the same procedure as described previously (Pan et al., 2011 ).
Cells cultured for at least 24 h were starved in methionine- and cysteine-free medium for 1 h, subjected to metabolic pulse-radiolabeling with [35S]methionine for 20 min, and chased for different time points following methods described previously (Pan et al., 2011 ). Cells and media were collected from each time point and were used for IP followed by autoradiography, as described previously (Pan et al., 2011 ).
This work was supported by National Institutes of Health Grants RO1 DK064232 (to R.N.S.), RO1 AI080656 (to Mary K. Estes), and R01 DK075322 (to Kelley W. Moremen), plus grant R06-06 from the Alpha1-Foundation (to R.N.S.), a postdoctoral research grant (to S.P.) from the Alpha-1 Foundation, and a Pilot/Feasibility Grant as part of grant P30 DK56338 from National Institute of Diabetes and Digestive and Kidney Disease Center. We thank the Baylor College of Medicine Mass Spectrometry Core for protein identification analysis.
This article was published online ahead of print in MBoC in Press (http://www.molbiolcell.org/cgi/doi/10.1091/mbc.E12-12-0886) on February 20, 2013.