Retention of Subunits of the Oligosaccharyltransferase Complex in the Endoplasmic Reticulum*

Membrane proteins of the endoplasmic reticulum (ER) may be localized to this organelle by mechanisms that involve retention, retrieval, or a combination of both. For luminal ER proteins, which contain a KDEL domain, and for type I transmembrane proteins carrying a dilysine motif, specific retrieval mechanisms have been identified. However, most ER membrane proteins do not contain easily identifiable retrieval motifs. ER localization information has been found in cytoplasmic, transmembrane, or luminal domains. In this study, we have identified ER localization domains within the three type I transmembrane proteins, ribophorin I (RI), ribophorin II (RII), and OST48. Together with DAD1, these membrane proteins form an oligomeric complex that has oligosaccharyltransferase (OST) activity. We have previously shown that ER retention information is independently contained within the transmembrane and the cytoplasmic domain of RII, and in the case of RI, a truncated form consisting of the luminal domain was retained in the ER. To determine whether other domains of RI carry additional retention information, we have generated chimeras by exchanging individual domains of the Tac antigen with the corresponding ones of RI. We demonstrate here that only the luminal domain of RI contains ER retention information. We also show that the dilysine motif in OST48 functions as an ER localization motif because OST48 in which the two lysine residues are replaced by serine (OST48ss) is no longer retained in the ER and is found instead also at the plasma membrane. OST48ss is, however, retained in the ER when coexpressed with RI, RII, or chimeras, which by themselves do not exit from the ER, indicating that they may form partial oligomeric complexes by interacting with the luminal domain of OST48. In the case of the Tac chimera containing only the luminal domain of RII, which by itself exits from the ER and is rapidly degraded, it is retained in the ER and becomes stabilized when coexpressed with OST48.

Membrane proteins of the endoplasmic reticulum (ER) may be localized to this organelle by mechanisms that involve retention, retrieval, or a combination of both. For luminal ER proteins, which contain a KDEL domain, and for type I transmembrane proteins carrying a dilysine motif, specific retrieval mechanisms have been identified. However, most ER membrane proteins do not contain easily identifiable retrieval motifs. ER localization information has been found in cytoplasmic, transmembrane, or luminal domains. In this study, we have identified ER localization domains within the three type I transmembrane proteins, ribophorin I (RI), ribophorin II (RII), and OST48. Together with DAD1, these membrane proteins form an oligomeric complex that has oligosaccharyltransferase (OST) activity. We have previously shown that ER retention information is independently contained within the transmembrane and the cytoplasmic domain of RII, and in the case of RI, a truncated form consisting of the luminal domain was retained in the ER. To determine whether other domains of RI carry additional retention information, we have generated chimeras by exchanging individual domains of the Tac antigen with the corresponding ones of RI. We demonstrate here that only the luminal domain of RI contains ER retention information. We also show that the dilysine motif in OST48 functions as an ER localization motif because OST48 in which the two lysine residues are replaced by serine (OST48ss) is no longer retained in the ER and is found instead also at the plasma membrane. OST48ss is, however, retained in the ER when coexpressed with RI, RII, or chimeras, which by themselves do not exit from the ER, indicating that they may form partial oligomeric complexes by interacting with the luminal domain of OST48. In the case of the Tac chimera containing only the luminal domain of RII, which by itself exits from the ER and is rapidly degraded, it is retained in the ER and becomes stabilized when coexpressed with OST48.
The endomemembrane system of eukaryotic cells consists of a series of distinct membrane-bound organelles that communicate with each other by means of vesicular or tubular interactions (for review see Refs. 1 and 2). Despite extensive anterograde and retrograde traffic, cellular organelles maintain the characteristic protein and lipid composition necessary to exe-cute their specific functions. Therefore, defined localization signals are required to establish the residency of a particular protein in a specific organelle or compartment (3). A paradigm for this model has been provided by the identification and characterization of a C-terminal KDEL motif. This motif is found at the C termini of several luminal proteins of the ER 1 and serves to mediate the retention of resident proteins of the ER lumen through retrieval from post-ER compartments ER via a transport mechanism involving vesicular and tubular elements (4,5). Another retention mechanism confers ER residency to membrane proteins that contain a cytoplasmically disposed C-terminal dilysine retention motif located close to the C terminus (6). This dilysine motif is recognized by coat protein I components, a membrane coat mainly concerned with retrograde transport from the Golgi apparatus back to the ER (5,(7)(8)(9).
Because the majority of ER membrane proteins do not have recognizable retrieval signals, it can be expected that other ER retention mechanisms exist. Misfolded or incompletely folded proteins may interact in the ER lumen with chaperones such as imunoglobulin heavy chain-binding protein, calnexin, or disulfide-isomerase, which by themselves carry retrieval signals (3, 10 -12). Retention may also be achieved by the functional interactions of ER membrane proteins with an organized matrix of luminal ER proteins or with cytoskeletal elements via their cytoplasmic domains (3). Another type of retention mechanism involves interaction of membrane proteins with each other to form oligomeric complexes or even larger assemblies. This may be the case for membrane proteins of individual Golgi stacks (13)(14)(15)(16)(17), and it may also provide a mechanism to retain in the ER components of the translocation apparatus, as well as closely associated membrane components, including the OST complex (1, 18 -21).
The mammalian oligosaccharyltransferase is composed of the four ER membrane proteins, ribophorin I and II (RI and RII), OST48, and DAD1, which form an oligomeric complex (22,23). Newly synthesized polypeptides may be N-glycosylated by this OST complex by transferring oligosaccharides, which were preassembled on the membrane-bound dolichol-pyrophosphate lipid carrier, to asparagine residues in the context of Asn-Xaa-(Ser/Thr). RI and OST48, and probably also RII, are type I transmembrane proteins. We have shown that the luminal domain of OST48 interacts with those of RI and RII and that the cytoplasmic domain of OST48 has affinity for the cytoplasmically exposed N-terminal tail of DAD1 (23). The OST complex interacts also with Sec61, the core component of the protein translocation apparatus (24), which also provides ribosomal binding sites (25). Results from experiments obtained over many years have demonstrated that components of * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  the translocation apparatus, as well as those involved in the cotranslational modification of the newly synthesized polypeptides, are part of a proteinaceous network that confines these membrane proteins to the rough domain of the endoplasmic reticulum (18 -20, 26, 27). Inclusion of proteins into this network would be an effective means of preventing their incorporation into vesicles that exit from the ER toward the Golgi apparatus. Our working hypothesis is that individual subunits of these oligomeric complexes carry by themselves ER localization motifs that function until the newly synthesized subunits are integrated into oligomeric complexes. Although OST48 contains a typical dilysine retrieval motif, RI, RII, and DAD1 do not carry established ER localization signals. We have previously shown that ER localization information in RII are contained independently in the cytoplasmic domain and in the transmembrane domain together with a short luminal flanking region. 2 Here, we report that in RI only the luminal domain contains ER localization information and that replacement of the two lysine residues by serine in the dilysine retrieval motif of OST48 results in the expression of this OST subunit at the cell surface. Coexpression of OST48ss with RI or RII prevents the exit of OST48ss from the ER. This effect is most likely due to interactions between the luminal domain of OST48 with those of RI or RII. This conclusion is also supported by the finding that a Tac chimera where the luminal domain is replaced by that of RII is not only prevented from leaving the ER, but it becomes protected from degradation when coexpressed with OST48.

MATERIALS AND METHODS
Plasmid Construction-The rat RI cDNA was cloned in this lab by Harnik-Ort et al. (28). The dog OST48 cDNA was a gift of Dr. Reid Gilmore (University of Massachusetts Medical School, Worcester, MA) (29). The human Tac antigen cDNA was obtained by Dr. Greg Pirozzi (Cytogen Corp., Princeton, NJ). To construct pMT-2/RI and pMT-2/ OST48, full-length cDNAs encoding the wild type RI and OST48 were generated by PCR using rat RI and dog OST48 cDNAs as templates. The primers used are those described in Tables I and II. As part of the primers, EcoRI sites were introduced and the cDNAs were subcloned into the EcoRI site of pMT-2 (30) using standard molecular cloning methods. To construct pMT-2/OST48ss, oligonucleotides encoding the two lysines in the antisense primer (see Tables I and II) were replaced by those encoding serine residues at Ϫ3 and Ϫ5 (TTT 3 ACT and CTT 3 ACT), and the resulting PCR products were ligated into pMT-2 at EcoRI sites. A cDNA encoding the Tac antigen with HindIII and KpnI sites at the 5Ј and 3Ј ends, respectively, was generated by PCR using the human Tac cDNA mentioned above as a template and the primers described in Tables I and II. The cDNA was subcloned into pExp1 (a gift of Dr. Herbert Samuels, New York University Medical Center, New York, NY). To construct plasmids containing the cDNAs encoding chimeras composed of domains of RI and the Tac antigen, I-T-T, T-I-T, and T-T-I (where I is RI, II is RII, and T is the Tac antigen), we generated cDNAs by using PCR for gene splicing by the overlap extension method as described previously (31). The primers used for PCR are described in Tables I and II. All three cDNAs have EcoRI sites. The transformation and preparation of plasmid DNA were carried out according to standard protocols. Schematic representation of the membrane disposition of all constructs are shown in Fig. 1.
Cell Line and Culture Condition-HeLa cells were grown as monolayer cultures at 37°C, 10% CO 2 in Dulbecco's modified Eagle's medium supplemented with 8% (v/v) fetal calf serum and 2 mM glutamine and penicillin/streptomycin. Transfection of cells, pulse-chase labeling, immunoprecipitation and Endo H treatment were done as described. 2 Antibodies-The polyclonal rabbit anti-RI antibody was raised against RI purified from rat liver microsomes by SDS-polyacrylamide gel electrophoresis (32). The monoclonal rat anti-RI antibody was raised against SDS-polyacrylamide gel electrophoresis-purified rat RI. It detects an epitope located on the luminal domain of RI (33). The polyclonal antibody directed against OST48 was raised in rabbits against a fusion protein containing the luminal domain of dog OST48 fused at its C terminus with glutathione S-transferase using an expression vector developed by Ron and Dressler (34). To prepare the antibody that detects the Tac antigen, the mouse myeloma cell line 7G7B6 was propagated in BALB/c 3T3 mice (35), and the IgG fraction was purified from the ascites fluid by passing the fluid over a protein A-Sepharose column.
Immunofluorescence Microscopy-Transfected cells were grown on coverslips at 50 -70% confluence (about 48 h after transfection), rinsed three times with 1ϫ cold phosphate-buffered saline, and then fixed with 3% paraformaldehyde for 20 min, followed by the same rinsing steps. The cells were treated with 0.2% Triton X-100 and incubated for 10 min if they had to be permeabilized. Permeabilized or nonpermeabilized cells were then incubated with 10 mM glycine for 10 -20 min. After incubation with blocking solution (5% dried milk in 1ϫ phosphatebuffered saline) for 30 min, cells were incubated at 37°C for 1 h with the primary antibody diluted in blocking solution. The monoclonal antibody ␣-RI was used at 1:200 dilution, the polyclonal antibody ␣-OST48 was used at 1:400 dilution, and the antibody against the Tac antigen, 7G7B6, was used at 1:100 dilution. After washing three times with blocking solution, the cells were incubated at 37°C for 1 h with either fluorescein isothiocyanate-conjugated donkey anti-rabbit IgG antibody or Texas Red-conjugated donkey anti-mouse IgG antibody di-   a For the structure of these constructs see Figure 1. b These oligonucleotide numbers are those listed in table 1A.
luted 1:100 in blocking solution, followed by three washing steps as described before. The coverslips were then mounted with citifluor and labeled proteins located either on the cell surface or in the ER were detected using the Axiphot microscope (Carl Zeiss, Inc., Thornwood, NY) equipped with 63ϫ and 100ϫ Plan-neofluar objective lenses. If double labeling was performed, a mixture of the two primary antibodies, RI and OST48, raised in different species such as mouse and rabbit respectively, was used to recognize two overexpressed proteins, followed by staining with a mixture of two secondary antibodies that includes fluorescein isothiocyanate-conjugated donkey directed against mouse and Texas Red-conjugated donkey directed against rabbit IgG, respectively. Immunofluorescence micrographs shown in Fig. 6 were obtained with the Nikon PCM 2000 laser scanning confocal microscope (Nikon Inc., Melville, NY) using the Simple 32 software (Compix Inc., Cranberry Township, PA). Images were prepared for publication using the Adobe Photoshop software.

RESULTS
The Luminal Domain of RI Contains ER Retention Information-RI is a type I transmembrane protein of the ER that does not contain identifiable ER localization signals, such as a Cterminally located dilysine motif. From previous studies we know that truncated forms of RI that consist of luminal portions of RI but are not anchored to the membrane are retained in the ER (36). When the membrane anchored RI-Tac chimera I-T-T was expressed in HeLa cell, it was quantitatively retained in the ER as indicated by the typical ER staining patterns seen in immunofluorescence micrographs (Fig. 2B). On the other hand, the transmembrane and the cytoplasmic domains contain apparently no ER localization information because the Tac chimera T-I-T and T-T-I were expressed on the cell surface as detected by immunofluoresence of nonpermeabilized cells (Fig. 2, C and E). In the case of T-I-T, surface labeling could be detected even in permeabilized cells (Fig. 2D), suggesting that this chimera exits from the ER very efficiently, resulting in high concentrations of the chimera at the cell surface. It appears, therefore, that in RI only the luminal domain contains ER localization information.  After a chase period of 3 h, I-T-T remained sensitive to the Endo H digestion (Fig. 3, lanes c and d)

, whereas T-I-T (lanes g and h) or T-T-I (lane k and l) became at least partially Endo H-resistant (marked by an arrowhead). It is interesting to note that more of the chimera T-I-T became Endo H-resistant compared with T-T-I. These results are consistent with immunostaining experiments where cells expressing the chimera T-I-T
showed much stronger surface immunofluorescence than those expressing T-T-I (Fig. 2). It appears, therefore, that the cytoplasmic domain of RI contains weak ER retention information. This was also seen when T-T-I was coexpressed with OST48ss (see Fig. 6K).
The Dilysine Motif at the Cytoplasmic Domain of OST48 Functions as an ER Localization Signal-OST48 is a type I ER transmembrane protein that contains a typical dilysine motif close to the C terminus. This motif serves as a retrieval signal in other type I ER transmembrane proteins (6). To test whether the dilysine motif in OST48 behaves indeed as an ER localization signal as predicted, HeLa cells were transfected with a cDNA encoding either the wild type OST48 or a mutant form of OST48 in which the two lysine residues were replaced by two serine residues (OST48ss) (Fig. 1A). As expected, in permeabilized cells we observed immunostaining of OST48 as a lace-like pattern characteristic of the rough endoplasmic reticulum (Fig.  4B), whereas nonpermeabilized cells remained unstained (Fig.  4A). In HeLa cells expressing OST48ss, surface labeling was detected in nonpermeabilized cells (Fig. 4C), and in addition, ER fluorescence was detected when the cells were permeabilized (Fig. 4D). These results demonstrated that the overexpressed wild type OST48 was retained in the ER, whereas OST48ss was not only found in the ER but also on the cell surface, thus confirming that the dilysine motif in OST48 does function as an ER localization signal, as was previously shown for the dilysine motif in Wbp1p, a yeast homolog of OST48 (37).
OST48ss Is Retained in the ER When Coexpressed with RI and RII-RI, RII, and OST48 together with DAD1 form an oligomeric complex that has oligosaccharyltransferase activity (22,38). We have also shown that this oligomer is maintained by interactions between the luminal domain of OST48 with those of RI or RII (23). We therefore reasoned that when OST48ss is coexpressed with RI and/or RII, these subunits may form partial oligomeric complexes by interacting via their luminal domains. Two possibilities may be considered: either OST48ss, RI, and RII move together out of ER to the cell surface because by interacting with OST48ss the retention information of RI and RII is neutralized or OST48ss is retained in the ER because the retention signals of RI or RII are dominant. To test these two alternatives, OST48ss was coexpressed in HeLa cells with RI and/or RII, followed by coimmunostaining. When expressed alone, OST48ss was found both at the cell surface (Fig. 5A) and in the ER (Fig. 5B), whereas coexpression of OST48ss and RI resulted only in ER staining of both proteins (Fig. 5, D and F). When OST48ss with RII or OST48ss with both RI and RII were coexpressed, the same results were obtained (data not shown). Retention of OST48ss in the ER was also observed when it was coexpressed with the Tac chimera containing the luminal domain of RI and the transmembrane and the cytoplasmic domains of the Tac antigen (I-T-T; Fig. 6, A-D) or the chimera containing both luminal and transmembrane domains of RII and the cytoplasmic domain of the Tac antigen (II-II-T) (data not shown). 2 We have shown that the chimera T-I-T and T-T-I are expressed at the cell surface (Fig.  2, C and E, respectively). When coexpressed with OST48ss either one of the RI-Tac chimera will exit from the ER and they, as well as OST48ss can be detected at the cell surface (Fig. 6, E, G, I, and K). These results are best explained by assuming that the luminal domain of OST48ss interacts with those of RI or RII, as was recently demonstrated (23), and that these partial oligomeric complexes are retained in the ER when one of the partners has ER retention information.

II-T-T Is Retained and Stabilized in the ER by
Interacting with OST48 -As described before, when the chimera II-T-T was expressed in HeLa cells, it exited from the ER and was found at the cell surface, as well as in a lysosomal compartment. 2 It was rather quickly degraded (t1 ⁄2 ϭ ϳ2 h), and the addition of leupeptin, an inhibitor of lysosomal proteases, largely prevented its degradation. 3 Based on the coexpression experiments shown above, it was expected that II-T-T is retained by OST48 in the ER through the interaction between their luminal domains. We therefore transfected HeLa cells with the Tac chimera II-T-T alone or II-T-T together with OST48, followed by pulse-chase labeling and immunoprecipitation using an antibody directed against the lumimal domain of RII. Aliquots of the immunoprecipitates were then treated with Endo H.
As shown in Fig. 7 (lane a), immediately after the pulse period, the major band immunoprecipitated with the anti-RII 3 G. Pirozzi and G. Kreibich, unpublished observation.  H (ϩ) (lanes b, d, f, h, j, and l) or kept as untreated controls (Ϫ) (lanes a, c, e, g, i, and k). Samples were analyzed on a 12% SDS-polyacrylamide gel, followed by autoradiography. Arrowheads indicate the position of T-I-T and T-T-I that acquired Endo H resistance.  lanes c and d) corresponds to II-T-T molecules that have exited from the ER and received modification of the N-linked oligosaccharide that rendered it insensitive to Endo H digestion. When II-T-T and OST48 were coexpressed in HeLa cells, the band corresponding to II-T-T* o carrying a complex oligosaccharide is no longer seen (lanes g and h), indicating that the chimera no longer leaves the ER. Instead the protein carrying a high mannose oligosaccharide remains Endo H-sensitive even after a chase period of 3 h (lane h). Noteworthy is also the fact that the chimera is not only retained in the ER, but it becomes stabilized; the ratio of the intensity of the radioactive bands corresponding to the endogenous RII and the chimera II-T-T immunoprecipitated immediately after the pulse (0 h; lane f) is similar to that obtained after a chase period of 3 h (lane h). It appears, therefore, that through interactions of the luminal domains of RII and OST48 the chimera, II-T-T was not only retained in the ER, but it was also protected from proteolytic degradation. DISCUSSION This study is mainly concerned with an analysis of the polypeptide domains of RI and OST48 that contribute to their retention in the ER. Although native subunits, modified polypeptides, or Tac chimeras were overexpressed, the results obtained are expected to be functionally relevant with regard to the biosynthesis of the OST complex, because before integration into oligomeric assemblies, subunits of the OST complex exist as individual molecules, and similar retention mechanisms are expected to act on them as on overexpressed subunits and chimeras containing specific domains of the OST subunits. The retention of the OST complex in the rough portion of the ER is further complicated by the fact that in addition to retention mechanisms that affect the individual subunits or the oligomer, this complex is known to cosediment with the Sec61p complex (24), which is part of an oligomeric array that extents throughout the rough portion of the ER, thus interconnecting translocation sites and segregating them from the smooth ER Cells were also cotransfected with cDNAs encoding OST48ss and RI (C-F) and fixed (C and E), or they were fixed and in addition permeabilized (D and F) as described before. For double immunofluorescence labeling, antibodies directed against OST48 or RI were used (for details, see "Materials and Methods"). Results shown in C-F demonstrate that OST48ss is retained in the ER when coexpressed with RI. (1, 18, 39). Two retrieval signals, the KDEL (4, 40) and the dilysine motif (41,42), located in the ER lumen and the cytoplasm, respectively, are well characterized. In addition, proteins could be retained in the ER through interactions with chaperones (12,(43)(44)(45) or cytoskeletal elements (46), forming large aggregates by themselves (47) or by being incorporated into a proteinaceous network (1,18,39). By applying immunofluorescence microscopy on transfected cells and immunoprecipitation of labeled proteins, we have begun to characterize domains within the individual OST subunits that prevent them from leaving the ER.
The function of the dilysine retrieval signal found in Wbpl, a yeast homolog of OST48, had been previously investigated (37). We hypothesize that after synthesis, some of the unassembled OST48 molecule may exit from the ER, and they would be retrieved from a post-ER compartment via coat protein I-coated vesicles. The retention mechanisms that affect the luminal domain of RI or the cytoplasmic and transmembrane domain of RII remain to be determined. It is very unlikely, however, that in normal cells, where the OST subunits are expressed in the proper stoichiometric ratio, RI or RII are retained in the ER by retrieval from post ER compartments. They do not acquire glycosylation patterns typical for post-ER compartments (41,48), despite the fact that both membrane glycoproteins can acquire complex N-linked oligosaccharides and O-linked sugars when exposed to Golgi glycosyl transferases (49). When OST subunits or chimera such as I-T-T are overexpressed compared with OST48, it is even less likely that they are retrieved from post-ER compartments, because only a small fraction would be able to oligomerize with OST48, which is the only AST subunit that carries an established ER retrieval signal. We had previously investigated the retention of luminal domains of RI by expressing in HeLa cells two C-terminally truncated variants of RI, RI 467 and RI 332 , where RI 467 is still membrane anchored but lacks part of the cytoplasmic domain, and RI 332 consists of a partial luminal domain of RI (36). Both RI 467 and RI 332 were rapidly degraded by a nonlysosomal mechanism. Interestingly, RI 332 was degraded in a biphasic fashion and was not found to be associated with imunoglobulin heavy chain-binding protein (36). Recent studies have provided evidence that this so called nonlysosomal ER degradation is in fact mediated by cytoplasmically located proteasomes (50,51). It is possible that in an unassembled monomer, such as I-T-T, the luminal domain functions as a true retention signal even after partial complexes are formed with OST48ss. It is more likely, however, that the luminal domain in I-T-T and also in RI is not properly folded because it lacks the other subunits of the OST complex as interacting partners. In this state of incomplete folding they are expected to be recognized by and bound to the transmembrane chaperone calnexin that is known to interact with incompletely assembled ER glycoproteins (12,43,44). Alternatively, they may form aggregates by themselves that are large enough to be excluded from transport vesicles budding off the ER.
Kelleher and Gilmore (22) showed that RI, RII, OST48, and DAD1 can be isolated as an oligomeric complex, and we have determined the interacting domains within the OST subunits. Specifically it was found that the luminal domain of OST48 interacts with those of RI and RII (23). It may therefore be expected that through this type of intermolecular interactions, polypeptides related to the OST subunits may be indirectly retained in the ER. It was indeed found that when OST48ss was coexpressed with RI or RII it was no longer able to exit from the ER. One can therefore assume that as soon as OST48 is integrated into the OST complex, it is retained in the ER through interactions with the luminal domains of RI and RII. As may be expected, the formation of links between OST subunits are steps toward stabilization of the individual subunits, which protects them from "ER degradation" of unassembled subunits (52). This is in agreement with a demonstration that a mutation in the DAD1 subunit leads not only to the degradation of DAD1, but it also destabilizes the other three subunits (53). We have also shown that the chimera II-T-T whenever expressed by itself is able to exit from the ER and it accumulates at the cell surface and in lysosomes where it is degraded. 2 However, when this chimera is coexpressed with OST48, both polypeptides are retained in the ER and protected from lysosomal degradation.
It is interesting to note that, in contrast to the native RII, which is found in about equimolar amounts as N-glycosylated and nonglycosylated forms (Ref. 54 and Fig. 7, lane a), the newly synthesized Tac chimera II-T-T was almost quantitatively N-glycosylated. This finding cannot be easily explained if one assumes that N-glycosylation is a strictly cotranslational process, in which growing nascent polypeptide chains are presented to the luminally oriented oligosaccharide transferase in an identical fashion, irrespective of the type of transmembrane or cytoplasmic domain of the mature polypeptide. Alternatively, N-glycosylation of the growing nascent chain may not occur in a strictly cotranslational manner (55), which is also suggested by the fact that the acylated tripeptide Asn-Tyr-Ser can be posttranslationally N-glycosylated (38,56). The extensive N-glycosylation of Asn 84 in II-T-T may therefore reflect their extended accessibility to the N-glycosylation machinery, possibly because of their inability to become incorporated into the oligomeric structure associated with the translocation apparatus in the rough endoplasmic reticulum (see also Ref. 49).