Mannose Trimming Targets Mutant α2-Plasmin Inhibitor for Degradation by the Proteasome

We have previously characterized the molecular and cellular mechanisms of α2-plasmin inhibitor (α2PI) deficiency. The mutant α2PI-Nara and α2PI-Okinawa proteins were found to be retained and degraded in cells stably expressing these mutant forms of α2PI. Degradation of the two mutant α2PI proteins, mediated by proteasomes, occurred after a lag time of 1.5 h during which glucose trimming took place. The mutant α2PI proteins were not ubiquitinated. Inhibition of mannosidase activity blocked the degradation of the mutant α2PI proteins without resulting in any changes in their binding to calnexin. Inhibition of glucose removal completely blocked the interaction between the α2PI proteins and the molecular chaperone calnexin. Under these conditions, mannose residues were removed from the oligosaccharides even when glucose residues were not processed. With mannose removal, the glucose-untrimmed mutant forms of α2PI, which failed to bind to calnexin, were degraded by proteasomes. The initiation of mannose trimming was a prerequisite for their degradation. Our findings show that modification of oligosaccharides of the mutant forms of α2PI determines their recognition by the degradation apparatus and that mannose trimming is important for targeting the mutant α2PI proteins for the degradation pathway.

␣ 2 -Plasmin inhibitor (␣ 2 PI) 1 is a plasma glycoprotein with an estimated four glycosylation sites and a molecular mass of 67 kDa, which contains about 11% carbohydrate (1). ␣ 2 PI belongs to the serine protease inhibitor superfamily. It is able to inhibit several different serine proteases, but its main function is to inhibit plasma-mediated fibrinolysis (2)(3)(4). The physiological importance of ␣ 2 PI was established by the discovery of individuals with congenital ␣ 2 PI deficiency in whom hemostatic plugs are dissolved prematurely before the restoration of injured vessels, resulting in a severe hemorrhagic tendency (5,6). Genetic abnormalities of ␣ 2 PI have been well characterized at the molecular level in two Japanese familial cases affected with the deficiency (7)(8)(9). One familial case of congenital ␣ 2 PI deficiency, designated as ␣ 2 PI-Nara, involves a frameshift mutation that results in the substitution of 178 amino acid residues for 12 carboxyl-terminal amino acid residues of the wild-type ␣ 2 PI (7). The other case, designated as ␣ 2 PI-Okinawa, involves a trinucleotide deletion that gives rise to the deletion of Glu-137 (9,10). The mutant molecules of ␣ 2 PI-Nara and ␣ 2 PI-Okinawa, which were expressed transiently in COS-7 cells, were retained for a prolonged period as endoglycosidase H (Endo H)-sensitive forms, and in each instance only a small proportion of the expressed proteins was secreted.
Retention within the endoplasmic reticulum (ER) has been reported for several other naturally occurring or genetically engineered mutant proteins that do not fold correctly. It has been shown that newly made secretory or membrane proteins do not move from the rough ER to the Golgi apparatus unless they fold into a native or near-native conformation (11). Misfolded proteins do not accumulate in the ER and are eventually removed from the ER by proteolysis. This rapid and selective degradation of proteins that are unable to reach the Golgi apparatus has been described as ER or pre-Golgi degradation. The process called ER degradation was generally assumed to occur inside the ER or a pre-Golgi compartment (12,13). However, the proteases present in the ER have not yet been identified. Several recent reports have suggested a role for the proteasome in the ER degradation of some membrane or luminal proteins (14,15). For example, misfolded cystic fibrosis transmembrane conductance regulator molecules are rapidly degraded in a process that requires covalent modification with ubiquitin and that is blocked by lactacystin, a specific proteasome inhibitor (16,17). Degradation of mutant ␣ 1 -antitrypsin (18), yeast carboxypeptidase Y (19), and major histocompatibility class I heavy chains have also recently been shown to require proteasome activity (20 -22).
Several ER-resident proteins that function as molecular chaperones have a role in ER quality control and contribute to the retention of misfolded proteins (23). Calnexin (CNX) and calreticulin preferentially interact with the nascent one-glucose-attached forms of N-linked oligosaccharides after glucose trimming and are thought to select proteins for degradation or secretion pathways. Their substrate-binding function is linked to a number of diseases with an ER-storage phenotype, such as cystic fibrosis, familial hypercholesterolemia, and ␣ 1 -antitrypsin deficiency (24,25). In these disease states, they serve as a part of the machinery that retains the misfolded glycoproteins in the ER. However, the selective mechanism that directs the misfolded proteins to the degradation pathway is not clear. Recent studies have indicated that the misfolded glycoproteins show prolonged association with CNX; removal of mannose from N-linked oligosaccharides resulted in misfolded glycoproteins being led to a subsequent degradation pathway (26 -28).
In this study, we demonstrate that mutant ␣ 2 PI proteins are degraded by proteasomes. We also examine whether oligosaccharide processing and association with CNX play a role in ER quality control in the case of the mutant ␣ 2 PI proteins.
Construction of Expression Vectors and Transfection-The plasmid pSV 2 PI, a wild-type expression construct, carries the normal ␣ 2 PI cDNA under the control of the SV40 promoter. The ␣ 2 PI-Nara expression vector, pSV 2 PN, and the ␣ 2 PI-Okinawa expression vector, pSV 2 PO, were constructed by replacing the normal ␣ 2 PI cDNA with the cDNA fragments carrying the mutations. Chinese hamster ovary (CHO) cells were maintained in F-12 Nutrient Mixture (Life Technologies, Inc.), and HepG2 cells were maintained in Dulbecco's modified Eagle's medium (DMEM, Nissui Pharmaceutical, Tokyo, Japan). Both media were supplemented with 10% fetal calf serum. For stable expression in CHO cells, both the expression vector DNA and pSVneo, which carries the G418 gene conveying neomycin resistance, were introduced into the cells by the LipofectAMINE-mediated transfection procedure, and neomycin was employed for selection.
Immunochemical Analysis of [ 35 S]Methionine-labeled Recombinant ␣ 2 PI-The CHO cells transfected with pSV 2 PI, pSV 2 PN, or pSV 2 PO were pulse-labeled for 15 min with [ 35 S]methionine and chased as described previously (7). CST, dMM, and LCT were added to the medium before the pulse-chase analysis. The inhibitors TLCK, TPCK, ALLN, ALLM, chloroquine, and leupeptin were added to the medium after pulse labeling. The ␣ 2 PI proteins radiolabeled in the course of synthesis were immunoprecipitated using an excess of the antibody and protein G from cell lysates or conditioned medium and were then subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) using a 10% acrylamide gel. Endo H digestion was performed as described previously (9). A nonstringent buffer, HBS containing 0.2 M NaCl and 50 mM HEPES, pH 7.5, with 2% CHAPS and 1 mM phenylmethylsulfonyl fluoride, was used in experiments examining the association of ␣ 2 PI proteins with CNX (29). After immunoprecipitation, by incubating the sample with goat anti-␣ 2 PI or rabbit anti-calnexin antibodies and protein G for 2 h, the precipitates were washed four times with lysis buffer. After incubation with HBS containing 1% SDS, the supernatants were diluted 20-fold in HBS containing 1% Nonidet P-40, and the ␣ 2 PI proteins were immunoprecipitated with anti-␣ 2 PI antibody and protein G.
Bands were detected by autoradiography using Kodak XAR-5 film, and the proteins were quantified on the basis of the intensity of the bands by means of a Fuji BAS 2000 Bio-imaging analyzer (Fuji, Tokyo, Japan).
Immunoblotting-The CHO cells expressing either pSV 2 PI, pSV 2 PN, or pSV 2 PO were preincubated with LCT (20 M) or ALLN (20 M) for 16 h. The cells were lysed in a lysis buffer (0.05 M Tris-HCl, 1% Nonidet P-40, 0.15 M NaCl, 1 mM phenylmethylsulfonyl fluoride, and 20 mM N-ethyl-maleimide), and the ␣ 2 PI was immunoprecipitated. The immunoprecipitates were subjected to SDS-PAGE, and Western blot analysis was performed using an anti-ubiquitin antibody for detection. The blot was developed using the ECL system (Amersham Pharmacia Biotech). In the detection of ubiquitinated p53 as the positive control, HepG2 cells, a human hepatoma cell line, were used. HepG2 cells were preincubated with LCT (20 M) for 16 h and lysed in the lysis buffer by freezing and thawing four times. p53 was immunoprecipitated using the monoclonal antibody to p53, and Western blot analysis was performed.

Mutant ␣ 2 PI Proteins Are Degraded in the Pre-Golgi
Area-We examined the secretory process for normal and mutant ␣ 2 PI proteins in stably transfected CHO cell lines. Most of the wild-type ␣ 2 PI molecules with a molecular mass of 64 kDa disappeared from the cells within 2 h and were secreted into the culture medium with a molecular mass of 67 kDa (Fig. 1A). The secreted wild-type protein had complex oligosaccharides, which were resistant to Endo H digestion. The ␣ 2 PI molecules in the cell extract were Endo H-sensitive. In contrast, the ␣ 2 PI-Nara and ␣ 2 PI-Okinawa molecules were not detected in the culture medium even after a 6-h chase period, and they were retained as Endo H-susceptible forms within the cells (Fig. 1B). Because Endo H removes N-linked oligosaccharides from polypeptides during transit through the ER without any effect on the complex oligosaccharides already transferred to the medial stacks of the Golgi apparatus, the mutant molecules were retained in the ER. As observed in the case of the cells that transiently expressed these molecules, the retained mutant molecules did not accumulate but eventually disappeared. We tried to isolate misfolded proteins from the cells, but no insoluble fraction was detected. The mutant ␣ 2 PI-Nara molecules were degraded in the pre-Golgi area. The time course of degradation of the mutant ␣ 2 PI-Nara and ␣ 2 PI-Okinawa molecules is shown in Fig. 1C. The degradation of ␣ 2 PI-Nara started 1.5 h after protein labeling. The half-life of the mutant molecules was about 4 h. The degradation of ␣ 2 PI-Okinawa showed a similar pattern.
The Precursor Forms of ␣ 2 PI-Nara and ␣ 2 PI-Okinawa Are Decreased in Molecular Size in the ER-The retained ␣ 2 PI-Nara and ␣ 2 PI-Okinawa each showed a change in molecular size during the pulse-labeling and chase periods (Fig. 1B). The deglycosylated polypeptides treated with Endo H showed identical mobility. These findings indicate that the decrease in size of the retained proteins was the result of modification of the oligosaccharide moieties and not the result of a change in the radiolabeled polypeptides.
We determined whether the decrease in size of the retained mutant ␣ 2 PI molecules was a result of the hydrolysis of either glucose or mannose residues. The mobility of immunoprecipitated ␣ 2 PI-Nara molecules preincubated with CST was retarded compared with that of the molecules without CST treatment. Digestion of the ␣ 2 PI proteins with Endo H produced molecules with the same mobility (Fig. 2a). This finding indicates that glucose residues were co-translationally hydrolyzed from oligosaccharides of the ␣ 2 PI-Nara proteins. CST decreased the amount of expression of the proteins. The inhibition of glucose trimming might impair effective translation and translocation of the proteins. Treatment with dMM inhibited the reduction of the molecular size of ␣ 2 PI-Nara. The molecular size of ␣ 2 PI-Nara treated with dMM was smaller than that of the molecules at 0 h that were co-translationally hydrolyzed (Fig. 2b). The post-translational removal of glucose residues preceded mannose trimming. These findings indicate that the decrease in size of the retained ␣ 2 PI-Nara molecules was the result of post-translational hydrolysis of glucose and mannose residues from oligosaccharides of the mutant protein in the ER. Because the retained ␣ 2 PI molecules were localized to the ER, it is evident that mannose processing was attributable to dMMinhibitable mannosidase present in the ER (30 -32). Detailed analysis by SDS-PAGE showed that the mobility of molecules treated with dMM corresponded to that observed after a 1.5-h chase period (Fig. 2c). The initiation of mannose trimming occurred after the 1.5-h chase period during which glucose trimming had occurred. The oligosaccharides of ␣ 2 PI-Okinawa proteins were also found to undergo similar post-translational hydrolysis in the cells (data not shown).
Degradation of the Mutant Proteins Is Proteasome-mediated-To examine whether specific proteases are involved in degradation of the mutant proteins, pulse-chase experiments were performed in the presence of inhibitors of several proteases and the proteasome. The lysosomal protease inhibitors used in this experiment had no effect on the total amount of radioactivity in the cell extracts after the 6-h chase period. The proteasome inhibitors ALLN and LCT markedly inhibited the degradation of the ␣ 2 PI-Nara proteins, but the other inhibitors had no effect (Fig. 3A). These proteins were not secreted from the cells even in the presence of the protease inhibitors or proteasome inhibitors. The mutant molecules that accumulated in cells treated with ALLN or LCT for 6 h had the same mobility as observed after a 6-h chase period in the absence of inhibitors, indicating the accumulation of mannose-trimmed forms (Fig. 3B). On the other hand, neither ALLN nor LCT had any effect on the secretion of wild-type ␣ 2 PI stably expressed in CHO cells. The intracellular degradation of ␣ 2 PI-Okinawa was also inhibited by ALLN and LCT (data not shown). These results indicate that the degradation of the mutant ␣ 2 PI proteins was mediated by proteasomes.
Mutant Proteins Are Not Ubiquitinated-If the proteasome actually degrades the mutant proteins, mutant ␣ 2 PI should be ubiquitinated before degradation. We examined whether the mutant proteins were conjugated with ubiquitin. Since ubiquitination of p53 has been reported (33), we used p53 as a positive control. Although ubiquitinated p53 was detected, the mutant proteins were not ubiquitinated even in the presence of the inhibitors (Fig. 3C).
CST Accelerates and dMM Inhibits the Degradation of the Mutant Proteins-As shown in Fig. 2b, the blocking of mannose trimming by the mannosidase inhibitor dMM changed the mobility on SDS-PAGE and also inhibited the degradation of the mutant proteins. In contrast, blocking of glucose trimming by CST seemed to have no such inhibitory effect on their degradation. Therefore, we examined the effect of oligosaccharide trimming on their degradation (Fig. 4, A and B). The degradation of the mutant proteins was rather accelerated as a result of CST treatment. On the other hand, dMM treatment markedly inhibited the degradation of the mutant proteins. The lag time of degradation seen in the case of the other inhibitors was not observed in the case of CST treatment. The ability of dMM to inhibit degradation of the mutant ␣ 2 PI proteins was comparable with that seen in the case of LCT. Exposure of the cells to a combination of dMM and LCT did not result in an additive effect (data not shown). Similar results were obtained in the case of ␣ 2 PI-Okinawa (data not shown).
dMM Also Inhibits Degradation of the Mutant Proteins Treated with CST-We examined the effect of the mannosidase inhibitor dMM on the degradation of the molecules when glucose trimming had been blocked by CST. Pre-incubation with CST blocked co-translational glucose trimming and accelerated the degradation of the mutant ␣ 2 PI proteins (Fig. 5). Comparing the molecular sizes of the mutant proteins at 0 and 6 h, the mobility changed during the chase period. dMM blocked both their decrease in size and their degradation. LCT markedly inhibited the degradation of the molecules, but not the size reduction. When the cells were incubated with CST only after radiolabeling, post-translational glucose trimming was blocked, and the molecules were degraded. dMM and LCT also inhibited their degradation. These results suggest that the processing of mannose residues in the molecules occurred even without the removal of glucose. Also, mannose trimming was important for the accelerated degradation of the mutant ␣ 2 PI-Nara molecules in cells treated with CST.
Wild-type and Mutant ␣ 2 PI Proteins Associated with CNX-Because CNX has been shown to bind monoglucosylated Nlinked oligosaccharides of nascent proteins in the ER, it has been speculated that the repeated binding of glucose to aglucosylated oligosaccharides of proteins might play a role in the selection of properly folded proteins for secretion and in the retention of unfolded proteins in the ER. To examine the role of the interaction between CNX and the ␣ 2 PI proteins, pulselabeled cells were lysed, and these proteins were co-immunoprecipitated by anti-CNX antibody under nonstringent conditions. The intensity of each band was quantified by a Fuji BAS 2000 Bio-imaging analyzer. The ratio of the intensity of the band of ␣ 2 PI protein immunoprecipitated by anti-CNX antibody to that of the band of ␣ 2 PI protein immunoprecipitated by anti-␣ 2 PI antibody at each time point was calculated and expressed as a percentage. About 10% of the wild-type ␣ 2 PI molecules were associated with CNX for a short duration of 30 min (Fig. 6A) and then secreted from the cells. In the case of ␣ 2 PI-Nara, about 20% of the mutant proteins interacted with CNX for a longer duration of 90 min. Thereafter, CNX dissociated from the mutant ␣ 2 PI-Nara despite the fact that the proteins were retained within the cells. ␣ 2 PI-Okinawa also interacted with CNX. In the case of the hepatoma cell line HepG2, the cells are known to produce ␣ 2 PI, which remains associated with CNX for a short duration. Despite the increase in retained mutant proteins, the mannosidase inhibitor dMM had no effect on the association of the mutant ␣ 2 PI proteins with CNX (data not shown). On the other hand, treatment of the stable cell lines, or HepG2 cells, with CST blocked the interaction of the wild-type and mutant ␣ 2 PI proteins with CNX (Fig. 6B). Binding of another ER chaperone, calreticulin, which is a soluble homologue of CNX, and binding of Bip with the wild-type or mutant ␣ 2 PI molecules were not detected. DISCUSSION In this study, we have shown that both ␣ 2 PI-Nara and ␣ 2 PI-Okinawa mutant proteins are degraded by proteasomes without ubiquitination. We also have demonstrated that inhibition of mannose trimming blocked the degradation of the mutant ␣ 2 PI proteins with or without removal of glucose. Also, we characterized the difference in association of the wild-type and mutant ␣ 2 PI proteins with CNX after translation.
ER-associated degradation is well characterized for several transmembrane and luminal proteins (34). This degradation of newly synthesized proteins in the ER is referred to as quality control and is distinct from lysosomal degradation. It has been recognized that the degradation of mutant proteins is mediated by the proteasome in the cytoplasm. ER forms of the multimembrane-spanning cystic fibrosis conductance regulator have been shown to be degraded by proteasomes, with the generation of ubiquitinated intermediates (17). Misfolded forms of this complex protein are recognized by enzymes of the ubiquitin-conjugating system and then targeted for destruction. Our study shows that mutant ␣ 2 PI proteins also are degraded by the proteasome, but these mutants could not be detected in a ubiquitinated form. The proteasome has been shown to degrade ubiquitinated forms of proteins. However, several authors have reported that the proteasome is involved in degradation of proteins without ubiquitination. Ornithine decarboxylase, the best characterized substrate for regulated proteolysis, is degraded by a 26 S proteasome in a ubiquitinindependent fashion (35). Ornithine decarboxylase proteolysis depends on another targeting system for proteasomal degradation. It is unknown whether the mutant ␣ 2 PI molecules are really degraded by the proteasome in a ubiquitin-independent manner. There are several possibilities to be considered in explaining the absence of ubiquitinated forms of the ␣ 2 PI mutants. First, the ubiquitination of these proteins may vary depending on the cell lines used. Recent studies have shown that T-cell receptor-␣ subunit chains do not require ubiquitination for their degradation by proteasomes in non-T cells (36), whereas ubiquitinated T-cell receptor-␣ is detected easily when the proteasome function is inhibited in a T-cell line (27). Second, unlike other short-lived proteins, the half-life of the mutant ␣ 2 PI is about 4 h, which is a relatively slow degradation process. The steady-state level of ubiquitinated forms may be low compared with the total number of molecules, probably depending on the activity of the cellular de-ubiquitinating enzymes, in which case the ubiquitinated forms would be scarcely detected. Another possible explanation is that there may be another ER-resident protease sensitive to LCT and ALLN.
The quality control of mutant ␣ 2 PI proteins is associated with oligosaccharide trimming and the assembly with molecular chaperones. It was found that degradation of the mutant ␣ 2 PI molecules took place after a lag time of 1.5 h in the chase period. The end point of the lag time corresponded to the initiation of mannose trimming. N-linked oligosaccharides are co-translationally added to luminal asparagine residues of proteins as Glc 3 Man 9 GlcNAc 2 (31). The three terminal glucose residues are rapidly cleaved in the ER, and ER and cis-Golgi mannosidases carry out trimming of these high mannose chains (37). During the glucose trimming, the carbohydratebinding luminal chaperone CNX preferentially associates with monoglucosylated oligosaccharides and facilitates protein fold- ing (38). Unfolded glycoproteins undergo a continuous cycle of binding to CNX in which assembly of the complex is facilitated by re-glucosylation of the asparagine-linked oligosaccharide by the ER-resident glycoprotein glucosyltransferase (39,40). The length of the lag time was also consistent with the duration of association of the mutant proteins with CNX. The mutant ␣ 2 PI was found to be associated with CNX for a 3-fold longer time than that in the case of the wild-type ␣ 2 PI molecules, which were led to the normal secretory pathway. The 1.5-h lag time before degradation of the mutant ␣ 2 PI is a period during which the mutant proteins are provided multiple opportunities to fold properly.
After the lag time, degradation of the mutant ␣ 2 PI began with mannose trimming. The initiation of degradation corresponded to the time when the mutant ␣ 2 PI molecules became dissociated from CNX and mannose trimming was initiated. Hammond et al. predict that permanent dissociation from the binding cycle would occur when hydrolysis of mannose residues generates an oligosaccharide unable to participate as a glucose acceptor, thereby leading to disposal of the unfolded glycoproteins. Blocking of mannose trimming by dMM inhibited the degradation of the mutant ␣ 2 PI. In vitro analysis showed that CNX preferentially bound to Glc 1 Man 9 GlcNAc 2 . Furthermore, a decrease in mannose residues results in a poor affinity in binding to CNX (41). Modification of mannose resides in the CNX cycle is an important step in the initiation of degradation of the mutant ␣ 2 PI proteins.
The inhibitory effect of treatment with the mannosidase inhibitor was comparable with that observed in the case of proteasome inhibitors, but the combination of both inhibitors did not show an additive effect in blocking the degradation of the mutant ␣ 2 PI proteins. Mannosidases and proteasomes function in the same pathway to degrade mutant proteins.
Recent studies have shown that mannosidase activity has a role in determining the fate of several proteins including yeast pro/pre-␣ factor, mutant ␣ 1 -antitrypsin, and T-cell receptor subunit CD 3-␦ (26 -28, 42). These studies show that the removal of mannose residues from N-linked oligosaccharides plays a role in the selection of mutant proteins to enter the proteasome-dependent degradation pathway. Liu et al. (26,28) proposed a model in which, after dissociation from CNX, mannose residues of mutant ␣ 1 -antitrypsin were trimmed and then the mannose-trimmed molecules targeted for disposal by proteasomes became bound to CNX again. Furthermore, Qu et al. (18) have reported that ubiquitination of CNX associated with variant ␣ 1 -antitrypsin precedes proteasomal degradation of the molecules in a cell-free system. In our study, it was shown that mannose trimming of the mutant ␣ 2 PI proteins that have dissociated from CNX is a very important process in targeting the mutant molecules for the proteasome degradation pathway. Although the association of mutant ␣ 2 PI proteins with CNX was transient in this experiment, it is unknown whether the mutant ␣ 2 PI proteins after mannose trimming bind to CNX prior to being degraded.
Preincubation with the glucosidase inhibitor CST blocked the interaction between CNX and the mutant ␣ 2 PI proteins and accelerated the degradation of ␣ 2 PI mutants by the proteasome. In the presence of CST, the degradation of the mutant molecules took place without a lag time. This immediate degradation might be because of lack of association with CNX, which may partly explain the rapid degradation of the mutant molecules in the presence of CST. Our findings also suggest that the processing of mannose residues occurred even when the glucose residues were not processed. After the trimming of three glucose residues, processing of mannose residues takes place in general (30). When added to cultures of various types  ). B cells were treated with LCT or ALLN for 6 h. In the pulse-chase analysis, the molecules observed in the chase period without any inhibitor are indicated (*). C, CHO cells expressing the wild-type or mutant ␣ 2 PI proteins and HepG2 cells expressing p53 were incubated for 16 h in the absence (Ϫ) or presence (ϩ) of LCT. The cells were lysed, and the mutant ␣ 2 PI proteins or p53 were immunoprecipitated with anti-␣ 2 PI antibody or anti-p53 antibody, respectively. Western blot analysis of immunoprecipitates was performed using anti-␣ 2 PI antibody, anti-p53 antibody, or anti-ubiquitin antibody for detection. To void detection of the antibody heavy chain (50 -55 kDa), the membrane was cut at a position corresponding to 60 kDa. The ubiquitinated forms of p53 (p53-Ub) were recognized by both antibodies. of animal cells, CST was found to prevent the processing of normal glycoproteins and therefore to cause the production of N-linked glycoproteins of the Glc 3 Man 7-9 GlcNAc 2 type (43). In CHO cells, glycoprotein containing Glc 3 Man 7 GlcNAc 2 oligosaccharide was produced upon treatment with a glucosidase inhibitor (44). Mannose processing can occur even without the removal of glucose resides. Although the secretion of these glycoproteins was not blocked, the location of mannosidase was not clear. Our results serve as indirect evidence, but alternate mannose processing may exist in the ER. It is unknown whether the process of the removal of mannose from the oligo-saccharides of the CST-treated mutant ␣ 2 PI proteins was specific to the cells used or specific to the proteins themselves. Mannose trimming may be a prerequisite for mutant ␣ 2 PI molecules to be degraded by the proteasome.
In our study, mannose trimming was found to be an important process in targeting the mutant ␣ 2 PI molecules for the proteasome degradation pathway. Further studies are necessary to clarify the features of the quality control system for mutant proteins. FIG. 5. Inhibition of the CST-accelerated degradation of ␣ 2 PI-Nara by a mannosidase inhibitor. a, CHO cells preincubated with CST (100 g/ml) were pulse-labeled with [ 35 S]methionine for 15 min and then treated with nonradiolabeled methionine for 6 h (the "chase" period) in the presence of the inhibitor. dMM (0.5 mM) or LCT (50 M) was added to the medium at the start of the chase period. G indicates the size of the molecules at 0 h in which glucose trimming was blocked. M indicates the size of the molecules after the 6-h chase period in which size reduction occurred. b, after the 15-min pulse-labeling period, the cells were incubated with CST (100 g/ml), and dMM (0.5 mM) or LCT (50 M) was added to the medium at the start of the 6-h chase period. g indicates the molecules at 0 h in which post-translational glucose trimming was inhibited. m indicates the molecules after the 6-h chase period in which reduction of the molecular weight occurred.
FIG. 6. Association of the wild-type ␣ 2 PI and mutant ␣ 2 PI-Nara proteins with CNX. A, in the pulse-chase analysis, these proteins were immunoprecipitated from cell extracts with goat anti-␣ 2 PI or rabbit anti-calnexin antibodies after the indicated chase period. The immunoprecipitates were incubated with anti-␣ 2 PI antibody again. Aliquots of the immunoprecipitates of molecules released from a ␣ 2 PI immunoprecipitate (A3A) and from a CNX precipitate (CNX3A) were applied to SDS-PAGE using a 10% acrylamide gel. The proteins were quantified on the basis of the intensity of the bands obtained upon SDS-PAGE analysis using the Fuji BAS 2000 image analyzer. The ratio of the intensity of the band of the ␣ 2 PI molecules immunoprecipitated by ␣ 2 PI antibody to that of the ␣ 2 PI molecules immunoprecipitated by the CNX antibody was calculated as the percentage of ␣ 2 PI molecules associated with CNX. B, the effect of CST treatment on the binding of the ␣ 2 PI proteins to CNX was examined in a hepatoma cell line, HepG2, and in cells stably expressing wild-type ␣ 2 PI, ␣ 2 PI-Nara, or ␣ 2 PI-Okinawa. CST was added 1 h before radiolabeling. Immunoprecipitates (IP) recovered from ␣ 2 PI (A) or CNX (CNX3A) precipitates were analyzed.