Complex, Two-way Traffic of Molecules Across the Membrane of the Endoplasmic Reticulum*

It has been known for many years that the endoplasmic reticulum (ER) is the site of assembly of polypeptide chains destined either for secretion or routing into various subcellular compartments. The N-glycosylation of these proteins, as well as their maturation assisted by certain resident luminal proteins, also occurs in the ER. Although many features of the import processes involved in synthesis of these glycoproteins have been elucidated, what is now becoming apparent is that the ER is not only involved in translocation and import, but it also functions in novel processes that mediate the export of a diversity of molecules, including unfolded or misfolded glycoproteins, glycopeptides, and oligosaccharides into the cytosol. Thus, it is clear that two-way traffic occurs, involving not only movement of molecules from the cytosol into the lumen of the ER but also out of the lumen into the cytosol. In this review the components involved in import of proteins into the ER, which have been reviewed elsewhere (1–3), are considered only in so far as they are implicated in the retrograde process, namely export out of the ER to the cytosol.

It has been known for many years that the endoplasmic reticulum (ER) 1 is the site of assembly of polypeptide chains destined either for secretion or routing into various subcellular compartments. The N-glycosylation of these proteins, as well as their maturation assisted by certain resident luminal proteins, also occurs in the ER. Although many features of the import processes involved in synthesis of these glycoproteins have been elucidated, what is now becoming apparent is that the ER is not only involved in translocation and import, but it also functions in novel processes that mediate the export of a diversity of molecules, including unfolded or misfolded glycoproteins, glycopeptides, and oligosaccharides into the cytosol. Thus, it is clear that two-way traffic occurs, involving not only movement of molecules from the cytosol into the lumen of the ER but also out of the lumen into the cytosol. In this review the components involved in import of proteins into the ER, which have been reviewed elsewhere (1)(2)(3), are considered only in so far as they are implicated in the retrograde process, namely export out of the ER to the cytosol.

Protein Import and N-Glycosylation of Proteins in the ER
Protein import from the cytosol to the lumen of the ER, the first step in the biosynthesis of luminal and/or secretory proteins, occurs by either a co-or post-translational process (1)(2)(3). In both cases, proteins are known to cross the membrane by a protein-conducting channel, the translocon, in which the Sec61p trimeric complex is believed to play a central role in mammalian cells. During the co-translocational insertion of proteins into the ER the enzyme oligosaccharyltransferase (OST) transfers an oligosaccharyl moiety, in most cases GlcNAc 2 Man 9 Glc 3 , from the dolichol intermediate to Asn residues located within the sequence -Asn-X-Ser/Thr-to form Nlinked glycans on the nascent polypeptide chain. The enzyme complex has been characterized both in mammals and yeast; four subunits have been identified in the mammalian system, and so far eight subunits have been found in Saccharomyces cerevisiae (4). The spatial relationship of this enzyme with respect to the translocon remains to be determined. Also still unknown is the process whereby the oligosaccharide lipid, Dol-PP-GlcNAc 2 Man 9 Glc 3 , that serves as donor of its oligosaccha-ride chain in the OST-catalyzed reaction is assembled. This assembly process is believed to involve translocation within the ER membrane, i.e. "flip-flop" of some of the intermediates leading to formation of Dol-PP-GlcNAc 2 Man 9 Glc 3 . The currently accepted model is that assembly of the lipid-linked oligosaccharide occurs at the membrane of the ER in a stepwise manner (5)(6)(7)(8), and three molecules are implicated in translocation from the cytosol to the lumen, namely Dol-PP-GlcNAc 2 Man 5 , Dol-P-Man, and Dol-P-Glc. However, so far putative "flippases" to facilitate this energetically unfavorable process have not been identified. In addition, there is no information on the topological orientation of Dol-P in the lipid bilayer of the ER.
Although much less is known about post-translational insertion of proteins into the ER, most of what is known has been elucidated in S. cerevisiae. In the post-translational pathway, the targeting of polypeptides is independent of ribosomes or the signal recognition particles. Instead, ER membrane proteins, namely the Sec62p and Sec63p subcomplex, are believed to be important. In addition, a heat shock protein (hsp70) with ATPase activity is thought to function in maintaining the polypeptide chain in a loosely folded state to facilitate the translocation (9). From in vivo and in vitro experiments, it has been concluded that an ER-resident protein, BiP, also has an important role in this process. In one model, it is speculated that BiP binds to a portion of the polypeptide chain as it protrudes through the Sec61p complex, thereby preventing the peptide chain from sliding back through the membrane of the ER (10). In a second model, BiP is thought to actively pull the polypeptide across the ER membrane (11).

Peptides Glycosylated in the ER Are Exported to the Cytosol
Following the demonstration that unfolded proteins and simple tripeptides containing -Asn-X-Thr/Ser-sequences could be glycosylated in vitro using microsomes (12,13), the fate of such glycopeptides when formed in vivo was studied with the view that they would serve as so-called markers for the secretory pathway. Although it was reported that in HepG 2 cells newly formed glycopeptides could be secreted and thereby serve as markers for non-selective, "bulk flow" through the ER (14), subsequent studies in other biological systems have not supported this idea. For example, studies using frog oocytes injected with glycosylatable peptides showed that the glycopeptides formed were not secreted but slowly degraded in a process inhibited by chloroquine; this observation suggested that lysosomes could be involved (15). Furthermore, in vitro studies in S. cerevisiae also showed that in this system glycopeptides were not exported like secretory proteins and that their exit from the ER (presumably to the cytosol) involved ATP and cytosol (16). There now is growing evidence that flow through the ER is by no means non-selective, and there is a mechanism that functions to concentrate cargo proteins during exit from the ER (17).
In contrast to the results using the yeast system, an initial attempt to detect glycopeptides exported from the ER to the cytosol using a mammalian system was not successful (18). This lack of success may have been due to the presence of a recently described peptide:N-glycanase (PNGase) activity in the soluble (cytosol) fractions of all mammalian cell lines and tissues studied (19,20). The action of this enzyme in the cytosol (21) would cleave the glycopeptides at the bond linking the oligosaccharide chain to the peptide, and it thereby would no longer be detectable as a glycopeptide. In this connection, it is interesting to note that in mammals there is a system for transport of small peptides across the membrane of the ER, which is critical for presentation of the antigenic peptide to the major histocompatibility class I complex. With respect to the peptide import mechanism, the transporter associated with antigen processing (TAP) has been identified (22). It is a member of the ABC (ATP-binding cassette) transporter family that requires ATP hydrolysis for transporter activity. It is also known that efflux of small peptides occurs in an ATP-dependent but TAP-independent manner (23,24), but the relationship of these processes to the glycopeptide transport system described above remains to be elucidated.

Similar Findings Indicate That Misfolded Proteins Also Leave the ER and Enter the Cytosol
Recent evidence indicates that the ER has "quality control" machinery (25) that differentiates between unfolded or misfolded proteins and correctly folded proteins so that the latter move from the ER to the Golgi complex. In this system, proteins that fail to correctly fold and/or oligomerize are retained in the ER and interact with a number of chaperones that serve to facilitate their acquisition of the correct conformation before they exit the ER by means of the secretory path. However, proteins that are misfolded (which could be deleterious to cells) are known to be degraded by a mechanism formerly called "ER degradation" (26). Until recently, the nature of this degradation process has remained unclear, because earlier attempts to identify the responsible proteolytic activities in the ER failed. Now it is evident that the site for the degradation is, in fact, the cytosol rather than the ER (27)(28)(29)(30) and that membrane or secretory proteins can be translocated from the ER into the cytosol, where they are degraded by proteasomes (Fig. 1). In some cases, it has been reported that the glycoproteins are deglycosylated by the action of PNGase prior to proteolytic degradation (31)(32)(33)(34)(35), although the precise subcellular location of this deglycosylation reaction remains unclear (see below). Furthermore, recent studies suggest that the Sec61p complex also participates in the retrograde translocation of proteins from the ER to cytosol (36 -38). This mechanism to route malformed proteins from the ER to cytosol was also shown to exist in S. cerevisiae (37)(38)(39)(40)(41)(42)(43)(44)(45)(46), suggesting that this quality control system may occur widely in nature. It should be noted that the retrograde movement of polypeptides across the ER, possibly by the action of a translocon, has been previously proposed (47,48). This route for movement of proteins from the ER to cytosol was suspected based on the finding that toxic proteins that enter the ER kill the cell by inactivating protein synthesis in the cytosol (49). Very recently this idea was confirmed by showing that a mutant ricin A-chain was transported from the ER to the cytosol in this retrograde manner (50). These findings suggest that the toxin molecules are routed to the cytosol by a preexisting transport mechanism that can also be used to export unfolded proteins.
In S. cerevisiae, considerable progress has been made in identifying ER components in addition to the Sec61p complex that are involved in this export and degradation process. Among these is Ubc6p, which is a ubiquitin-conjugating enzyme that catalyzes the covalent attachment of ubiquitin to specific proteolytic substrates (41). Recently, Cue1p, an ER membrane protein, has been shown to recruit soluble Ubc7p, another ubiquitin-conjugating enzyme to the ER surface. The action of these conjugation enzymes appears to be a prerequisite for retrograde transport out of the ER, suggesting a coupling of export and degradation (45). Another ER protein, Der1p, has been described (40); although its function is unknown, it is postulated to be necessary for export to the cytosol of soluble luminal proteins that are then degraded (26). The protein encoded by the HRD1/DER3 gene, originally believed to be involved in regulating the level hydroxymethylglutaryl-CoA reductase in the ER, seems to play an even broader role in degradation of both membrane and luminal proteins (44,46). Hrd3p is also believed to be an ER membrane protein involved in the transport of hydroxymethylglutaryl-CoA, although its precise function is not clear (44). With respect to chaperones in the lumen of ER of yeast, thus far, Cne1p (calnexin) (39) and Kar2p (BiP) (37) have been suggested to be involved in the export process. There also is preliminary evidence that Pdi1p (protein disulfide isomerase) is involved in the export of malfolded proteins. 2 In contrast to yeast, in higher eukaryotes little is known about components of the ER other than in cytomegalovirus-infected cells, in which two viral gene products, US11 and US2 proteins, have been shown to bind to newly synthesized class I heavy chains and thereby facilitate their rapid routing back to the cytosol (31,36). Given the recent identification of multiple components of the export and degradation machinery in yeast, it seems likely that in the future a variety of other components will be identified in higher eukaryotes.

Oligosaccharides Generated in the ER Also Are Exported to the Cytosol and Degraded
A series of earlier in vitro studies demonstrated that free oligosaccharides, presumably released from the lipid-linked oligosaccharide involved in the N-glycosylation process, were 2 K. Römisch, personal communication.

FIG. 1. Traffic in and out of the ER.
Gn 2 -OS represents for simplicity oligosaccharides with N-acetylglucosamine at the reducing terminus whose complete structure is not defined. Solid lines represent processes that have already been described, whereas broken lines represent reactions that either are controversial or undocumented. The subcellular site for the deglycosylation by PNGase is still unclear, although the occurrence of this enzyme in both cytosol and in the ER has been reported (see the text).
found to be entrapped in the lumen of microsomes prepared from a variety of sources (Fig. 2). More recently, the source, structure, and fate of these oligosaccharides have been studied by several groups (51). Oligosaccharides of the polymannose type containing either one or two N-acetylglucosamine residues at the reducing end, as well as chains with or without "capping" glucose residues at the non-reducing end, have been described. The current model for the mechanism of formation of free oligosaccharides is more complicated than had been anticipated in early studies; four possible reactions leading to free oligosaccharides are shown below. For simplicity, the oligosaccharides are depicted as OS-Gn 1 , OS-Gn 2 , or OS-Gn 2 -P, where Gn represents GlcNAc at the reducing terminus and OS represents the remainder of the oligosaccharide chain. One reaction postulated to be involved in their formation in the ER is a hydrolytic process yielding OS-Gn 2 from oligosaccharide-lipid (I). It has been postulated that OST could function in this hydrolytic process in the absence of -Asn-X-Ser/ Thr-acceptor sites (52), although there is no direct biochemical evidence that OST can catalyze transfer of oligosaccharide chains to water instead of to Asn side chains. The second enzyme that could be responsible for the formation of free oligosaccharide is a pyrophosphatase that would release OS-Gn 2 -P from oligosaccharide-lipid (II). Although this activity has been reported to be cytosol-oriented in mammalian cells (53), it has been detected at the luminal face of S. cerevisiae microsomes (54). Nothing is known about the fate of the oligosaccharide-phosphate produced in this manner. A third route for oligosaccharide production (III) was hypothesized to be the result of the action of an endo-␤-N-acetylglucosaminidase (EN-Gase) acting upon N-linked glycoproteins to give rise to OS-Gn 1 (55). More recently, a fourth reaction (IV) involving formation of OS-Gn 2 from glycoproteins by the action of a PNGase, which cleaves the amide bond between the glycosylated Asn residue and the innermost GlcNAc residue, has been proposed (56,57). Indeed, the occurrence of both ENGase and PNGase in an ER-enriched fraction has been reported (58 -60). It is clear that oligosaccharides formed in the lumen of the ER exit from this organelle into the cytosol. Although there are uncertainties in terms of the detailed structure of the oligosaccharide that is translocated, it is apparent that the process is ATP-dependent (61). Once in the cytosol, the oligomannosides may be further processed by the previously described cytosolic ␣-mannosidase (57), ENGase (62), and/or chitobiase (63) to generate GlcNAc 1 Man 5 , which is then taken up by lysosomes and degraded (64,65). At present it is not known if the cytosolic ENGase/PNGase and their ER-associated equivalents actually are the same proteins; so far no clear difference has been described in terms of enzymatic properties, and it is not clear if the catalytic site of the ER-associated enzymes is really in the lumen or exposed on the cytosolic face of the ER membrane.

Key Questions That Remain To Be Answered
It should be clear in this brief overview that the ER is an incredibly dynamic organelle, with molecules traveling in opposite directions across the lipid bilayer. In the case of the oligosaccharide-lipids involved in N-glycosylation, this may involve flip-flop of saccharide lipids generated at the cytosolic face across the membrane to the luminal face of the ER, although we have no idea of the mechanism of this process. We also do not know why there is translocation of glycoproteins, glycopeptides, and oligosaccharides in the other direction, i.e. to the cytosol, rather than transport of these molecules through the endomembrane system followed by either secretion or degradation after routing to the lysosomes. In the case of unfolded proteins the conventional route may not be taken because excessive accumulation of these macromolecules within the lumen might lead to their aggregation and precipitation, thereby blocking the secretory pathway. Interestingly, observations made on calreticulin, a luminal glycoprotein that is FIG. 2. Pathways for formation of free oligosaccharides in the ER. OS-Gn 2 and OS-Gn 1 represent for simplicity oligosaccharides with N-acetylglucosamine at the reducing terminus whose complete structure is not defined.
heat-inducible in Chinese hamster ovary cells (66), are consistent with this idea. When these cells were subjected to heat stress, calreticulin was found to redistribute; it disappeared from the ER and appeared in the cytosol (67). Free oligosaccharides and small glycopeptides also are moved out of the cell by the secretory pathway. In this case, export from the ER to the cytosol may be necessary in order to prevent their interaction and possible inhibition of other glycosylation-related processes that occur in the ER or the Golgi complex (61).
With respect to the channel involved in the export of oligosaccharides, glycopeptides, and misfolded proteins, little is known other than the reported involvement of the Sec61p complex in the retrograde transport of misfolded proteins (36 -38). It is of interest that transport of all three classes of compounds appear to have the common feature of requiring ATP hydrolysis (21,36,39,61). However, it also should be noted that there are obvious differences in at least some aspects of these transport processes. For instance, under conditions whereby free oligosaccharides can be transported efficiently from the ER to cytosol, glycotripeptide cannot be transported from the ER in permeabilized HepG 2 cells (61). Finally, it is important to note that many different systems are under study, and at this stage of our understanding it is by no means certain that all of these will exhibit the same features. Clearly, understanding the mechanisms by which movement of these different classes of molecules, both in and out of the ER, is regulated offers even greater challenges than envisioned just a few years ago.