KDEL-cargo regulates interactions between proteins involved in COPI vesicle traffic: measurements in living cells using FRET.

How the occupied KDEL receptor ERD2 is sorted into COPI vesicles for Golgi-to-ER transport is largely unknown. Here, interactions between proteins of the COPI transport machinery occurring during a "wave" of transport of a KDEL ligand were studied in living cells. FRET between CFP and YFP fusion proteins was measured by multifocal multiphoton microscopy and bulk-cell spectrofluorimetry. Ligand binding induces oligomerization of ERD2 and recruitment of ARFGAP to the Golgi, where the (ERD2)n/ARFGAP complex interacts with membrane-bound ARF1. During KDEL ligand transport, interactions of ERD2 with beta-COP and p23 decrease and the proteins segregate. Both p24a and p23 interact with ARF1, but only p24 interacts with ARFGAP. These findings suggest a model for how cargo-induced oligomerization of ERD2 regulates its sorting into COPI-coated buds.

Functional roles and efficiencies of the thioredoxin boxes of calcium-binding proteins 1 and 2 in protein folding.

The rat luminal endoplasmic-recticulum calcium-binding proteins 1 and 2 (CaBP1 and CaBP2 respectively) are members of the protein disulphide-isomerase (PDI) family. They contain two and three thioredoxin boxes (Cys-Gly-His-Cys) respectively and, like PDI, may be involved in the folding of nascent proteins. We demonstrate here that CaBP1, similar to PDI and CaBP2, can complement the lethal phenotype of the disrupted Saccharomyces cerevisiae PDI gene, provided that the natural C-terminal Lys-Asp-Glu-Leu sequence is replaced by His-Asp-Glu-Leu. Both the in vitro RNase AIII-re-activation assays and in vivo pro-(carboxypeptidase Y) processing assays using CaBP1 and CaBP2 thioredoxin (trx)-box mutants revealed that, whereas the three trx boxes in CaBP2 seem to be functionally equivalent, the first trx box of CaBP1 is significantly more active than the second trx box. Furthermore, only about 65% re-activation of denatured reduced RNase AIII could be obtained with CaBP1 or CaBP2 compared with PDI, and the yield of PDI-catalysed reactions was significantly reduced in the presence of either CaBP1 or CaBP2. In contrast with PDI, neither CaBP1 nor CaBP2 could catalyse the renaturation of denatured glyceraldehyde-3-phosphate dehydrogenase (GAPDH), which is a redox-independent process, and neither protein had any effect on the PDI-catalysed refolding of GAPDH. Furthermore, although PDI can bind peptides via its b' domain, a property it shares with PDIp, the pancreas-specific PDI homologue, and although PDI can bind malfolded proteins such as 'scrambled' ribonuclease, no such interactions could be detected for CaBP2. We conclude that: (1) both CaBP2 and CaBP1 lack peptide-binding activity for GAPDH attributed to the C-terminal region of the a' domain of PDI; (2) CaBP2 lacks the general peptide-binding activity attributed to the b' domain of PDI; (3) interaction of CaBP2 with substrate (RNase AIII) is different from that of PDI and substrate; and (4) both CaBP2 and CaBP1 may promote oxidative folding by different kinetic pathways.

Factors involved in the cell density-dependent regulation of nuclear/cytoplasmic distribution of the 11.5-kDa Zn(2+)-binding protein (parathymosin-alpha) in rat hepatocytes.

Although the 11.5 kDa Zn(2+)-binding protein (ZnBP, parathymosin-alpha) possesses a functional bipartite nuclear localization signal it was found in most tissues in the cytoplasm. The cultivation of freshly isolated rat hepatocytes for 24 hours under standard conditions was associated with an almost complete translocation of ZnBP from the cytoplasm to the nuclei. Here we demonstrate, that this translocation is negatively correlated with cell density. The translocation of ZnBP to the nucleus can be inhibited or abolished by inhibitors of protein synthesis (cycloheximide) or transcription (actinomycin D). Moreover, cycloheximide can induce a relocation of ZnBP to the cytoplasm when applied after the appearance of ZnBP in the nuclei. DMSO, an inhibitor of dedifferentiation of cultured hepatocytes, abolishes also the translocation of ZnBP into the nucleus. Thinly seeded cells keep their ZnBP in the cytoplasm if they are co-cultured with plasma membranes from Morris MH7777 hepatoma cells or antibodies against E-cadherin indicating the involvement of cell adhesion proteins. We have enriched a protein from the cytosol of fresh hepatocytes which inhibits the translocation of ZnBP, but not that of albumin-NLS into the nucleus in a permeabilized cell system. Such an activity could not be found in the cytoplasm of permanent cell lines which harbor ZnBP only in the nucleus. A model for the regulation of the nuclear import of ZnBP is proposed.

Endophilin I mediates synaptic vesicle formation by transfer of arachidonate to lysophosphatidic acid.

Endophilin I is a presynaptic protein of unknown function that binds to dynamin, a GTPase that is implicated in endocytosis and recycling of synaptic vesicles. Here we show that endophilin I is essential for the formation of synaptic-like microvesicles (SLMVs) from the plasma membrane. Endophilin I exhibits lysophosphatidic acid acyl transferase (LPAAT) activity, and endophilin-I-mediated SLMV formation requires the transfer of the unsaturated fatty acid arachidonate to lysophosphatidic acid, converting it to phosphatidic acid. A deletion mutant lacking the SH3 domain through which endophilin I interacts with dynamin still exhibits LPAAT activity but no longer mediates SLMV formation. These results indicate that endophilin I may induce negative membrane curvature by converting an inverted-cone-shaped lipid to a cone-shaped lipid in the cytoplasmic leaflet of the bilayer. We propose that, through this action, endophilin I works with dynamin to mediate synaptic vesicle invagination from the plasma membrane and fission.

The protein disulphide-isomerase family: unravelling a string of folds.

The mammalian protein disulphide-isomerase (PDI) family encompasses several highly divergent proteins that are involved in the processing and maturation of secretory proteins in the endoplasmic reticulum. These proteins are characterized by the presence of one or more domains of roughly 95-110 amino acids related to the cytoplasmic protein thioredoxin. All but the PDI-D subfamily are composed entirely of repeats of such domains, with at least one domain containing and one domain lacking a redox-active -Cys-Xaa-Xaa-Cys- tetrapeptide. In addition to their known roles as redox catalysts and isomerases, the last few years have revealed additional functions of the PDI proteins, including peptide binding, cell adhesion and perhaps chaperone activities. Attention is now turning to the non-redox-active domains of the PDIs, which may play an important role in all of the known activities of these proteins. Thus the presence of both redox-active and -inactive domains within these proteins portends a complexity of functions differentially accommodated by the various family members.

KDEL receptor (Erd2p)-mediated retrograde transport of the cholera toxin A subunit from the Golgi involves COPI, p23, and the COOH terminus of Erd2p.

A cholera toxin mutant (CTX-K63) unable to raise cAMP levels was used to study in Vero cells the retrograde transport of the toxin A subunit (CTX-A-K63), which possesses a COOH-terminal KDEL retrieval signal. Microinjected GTP-gamma-S inhibits the internalization as well as Golgi-ER transport of CTX-A-K63. The appearance of CTX-A-K63 in the Golgi induces a marked dispersion of Erd2p and p53 but not of the Golgi marker giantin. Erd2p is translocated under these conditions most likely to the intermediate compartment as indicated by an increased colocalization of Erd2p with mSEC13, a member of the mammalian coat protein II complex. IgGs as well as Fab fragments directed against Erd2p, beta-COP, or p23, a new member of the p24 protein family, inhibit or block retrograde transport of CTX-A-K63 from the Golgi without affecting its internalization or its transport to the Golgi. Anti-Erd2p antibodies do not affect the binding of CTX-A to Erd2p, but inhibit the CTX-K63-induced translocation of Erd2p and p53.

ERp28, a human endoplasmic-reticulum-lumenal protein, is a member of the protein disulfide isomerase family but lacks a CXXC thioredoxin-box motif.

We report on the isolation, sequence and a putative role of a human endoplasmic-reticulum-lumenal protein, ERp28. The protein has the C-terminal retention signal KEEL and localizes to the endoplasmic reticulum (ER) as seen by subcellular fractionation and immunofluorescence studies. The protein has significant sequence similarity to members of the protein disulfide isomerase (PDI) family, although it lacks the thioredoxin box (CGHC) motif. We propose, on the basis of sequence analysis, a model of the domain structure of PDI, representing a significant extension of previously proposed models. Our results are in partial agreement with recently published NMR data [Kemmink, J., Darby, J., Dijkstra, K., Nilges, M. & Creighton, T. E. (1997) Curr. Biol. 7, 239-245] and indicate that PDI contains, in addition to the two thioredoxin folds described in previous models, two thioredoxin folds within the domains previously defined as b and b'. The thioredoxin domain of ERp28 shares a higher degree of similarity with the corresponding active and inactive domains of PDI than with other members of the PDI family, indicating that ERp28 developed from an ancient form of PDI or a PDI precursor. In contrast to Ig-heavy-chain-binding protein, human ERp28 is not induced by metabolic stress (tunicamycin). In in vitro experiments, ERp28 and calnexin precipitate with overexpressed, wild-type hepatitis B small surface antigen and with a mutated ER-retained form. This indicates that ERp28, as calnexin, may be involved in the processing of secretory proteins within the ER.

Differential distribution of calcium stores in paramecium cells. Occurrence of a subplasmalemmal store with a calsequestrin-like protein.

We have analyzed in Paramecium cells the occurrence and intracellular distribution of the high capacity/low affinity calcium-binding proteins, calsequestrin (CS) and calreticulin (CR) using antibodies against CS from rat skeletal muscle and against CR from rat liver, respectively. As revealed by Western blots, a CS-like protein isolated by affinity chromatography from Paramecium cells comigrated with CS isolated from rat skeletal muscle. The immunoreactivity of this 53 kDa protein band was blocked when the antibodies had been preadsorbed with purified rat CS. A band of identical molecular size was shown to bind 45Ca in overlays. By immunofluorescence and immunogold labeling this CS-like protein was localized selectively to the extended subplasmalemmal calcium stores, the "alveolar sacs", which cover almost the entire cell surface. Concomitantly the 53 kDa 45Ca-binding band became increasingly intense in overlays as we increasingly enriched alveolar sacs. Antibodies against rat CR react with a 61 kDa band but do not cross-react with CS-like protein in Paramecium. These antibodies selectively stained intracellular reticular structures, identified bona fide as endoplasmic reticulum.

Characterization of brefeldin A induced vesicular structures containing cycling proteins of the intermediate compartment/cis-Golgi network.

Residence of luminal ER proteins is mediated by a cyclic process which involves binding of escaped proteins to a KDEL receptor in a post-ER compartment and redistribution of the ligand-receptor complex back to the ER. We examined the relocation of the KDEL receptor after treatment with the fungal metabolite brefeldin A and compared this with the retrograde transport of the KDEL receptor observed after ligand or receptor overexpression. Incubation with brefeldin A led to the formation of vesicular structures containing the KDEL receptor and ERGIC-53, a marker for the ER-Golgi intermediate compartment. Immunoelectron microscopy revealed that these structures are composed of tubulo-vesicular clusters. The brefeldin A induced vesicular structures were morphologically and biochemically distinct from the ER-Golgi hybrid compartment as demonstrated by double immunofluorescence microscopy and subcellular fractionation. Overexpression of the receptor itself or a lysozyme-KDEL construct led to a shift of the KDEL receptor together with ERGIC-53, an intermediate compartment marker to the ER but not to structures resembling BFA induced vesicular structures. Moreover, overexpression of the receptor resulted in the partial redistribution of marker proteins of the medial Golgi and the trans-Golgi network to ER-like structures. We conclude that the effects of brefeldin A on the redistribution of the KDEL receptor do not reflect physiological events occurring during increased occupancy of the receptor with ligands.

Reduction of protein disulfide bonds in an oxidizing environment. The disulfide bridge of cholera toxin A-subunit is reduced in the endoplasmic reticulum.

Following retrograde transport to the endoplasmic reticulum (ER) the A-subunit of cholera toxin (CTX-A) is partially cleaved into CTX-A1 and CTX-A2 by reduction of a disulfide bridge [Majoul et al. (1996) J. Cell Biol. 133, 777-789], although the redox state in the ER favors disulfide formation. We show here that the disulfide bridge of CTX-A is cleaved in vitro already at GSH/GSSG ratios between 1 and 3. Protein disulfide isomerase (PDI) exerts only a minor accelerating effect. Various mixed disulfide intermediates (CTX-A1-S-S-CTX-A1; PDI-S-S-A2; PDI-S-S-A1) appear during CTX-A reduction. These results indicate that in the ER protein disulfide formation and protein disulfide reduction can take place simultaneously.

Imaging the intracellular trafficking and state of the AB5 quaternary structure of cholera toxin.

The subcellular localization and corresponding quaternary state of fluorescent labelled cholera toxin were determined at different time points after exposure to living cells by a novel form of fluorescence confocal microscopy. The compartmentalization and locus of separation of the pentameric B subunits (CTB) from the A subunit (CTA) of the toxin were evaluated on a pixel-by-pixel (voxel-by-voxel) basis by measuring the fluorescence resonance energy transfer (FRET) between CTB labelled with the sulfoindocyanine dye Cy3 and an antibody against CTA labelled with Cy5. The FRET efficiency was determined by a new technique based on the release of quenching of the Cy3 donor after photodestruction of the Cy5 acceptor in a region of interest within the cell. The results demonstrate vesicular transport of the holotoxin from the plasma membrane to the Golgi compartment with subsequent separation of the CTA and CTB subunits. The CTA subunit is redirected to the plasma membrane by retrograde transport via the endoplasmic reticulum whereas the CTB subunit persists in the Golgi compartment.

Definition of the lectin-like properties of the molecular chaperone, calreticulin, and demonstration of its copurification with endomannosidase from rat liver Golgi.

Calreticulin was identified by immunochemical and sequence analyses to be the higher molecular mass (60 kDa) component of the polypeptide doublet previously observed in a rat liver Golgi endomannosidase preparation obtained by chromatography on a Glc alpha 1 --> 3Man-containing matrix. The affinity for this saccharide ligand, which paralleled that of endomannosidase and was also observed with purified rat liver calreticulin, suggested that this chaperone has lectin-like binding properties. Studies carried out with immobilized calreticulin and a series of radiolabeled oligosaccharides derived from N-linked carbohydrate units revealed that interactions with this protein were limited to monoglucosylated polymannose components. Although optimal binding occurred with Glc1Man9GlcNAc, substantial interaction with calreticulin was retained after sequential trimming of the polymannose portion down to the Glc1Man5GlcNAc stage. The alpha 1 --> 6-mannose branch point of the oligosaccharide core, however, appeared to be essential for recognition as Glc1Man4GlcNAc did not interact with the calreticulin. The carbohydrate-peptide linkage region had no discernible influence on binding as monoglucosylated oligosaccharides in N-glycosidic linkage interacted with the chaperone to the same extent as in their unconjugated state. The immobilized calreticulin proved to be a highly effective tool for sorting out monoglucosylated polymannose oligosaccharides or glycopeptides from complex mixtures of processing intermediates. The copurification of calreticulin and endomannosidase from a Golgi fraction in comparable amounts and the strikingly similar saccharide specificities of the chaperone and the processing enzyme have suggested a tentative model for the dissociation through glucose removal of calreticulin-glycoprotein complexes in a post-endoplasmic reticulum locale; in this scheme, deglucosylation would be brought about by the action of endomannosidase rather than glucosidase II.

Transport of an external Lys-Asp-Glu-Leu (KDEL) protein from the plasma membrane to the endoplasmic reticulum: studies with cholera toxin in Vero cells.

The A2 chain of cholera toxin (CTX) contains a COOH-terminal Lys-Asp-Glu-Leu (KDEL) sequence. We have, therefore, analyzed by immunofluorescence and by subcellular fractionation in Vero cells whether CTX can used to demonstrate a retrograde transport of KDEL proteins from the Golgi to the ER. Immunofluorescence studies reveal that after a pulse treatment with CTX, the CTX-A and B subunits (CTX-A and CTX-B) reach Golgi-like structures after 15-20 min (maximum after 30 min). Between 30 and 90 min, CTX-A (but not CTX-B) appear in the intermediate compartment and in the ER, whereas the CTX-B are translocated to the lysosomes. Subcellular fractionation studies confirm these results: after CTX uptake for 15 min, CTX-A is associated only with endosomal and Golgi compartments. After 30 min, a small amount of CTX-A appears in the ER in a trypsin-resistant form, and after 60 min, a significant amount appears. CTX-A seems to be transported mainly in its oxidized form (CTX-A1-S-S-CTX-A2) from the Golgi to the ER, where it becomes slowly reduced to form free CTX A1 and CTX-A2, as indicated by experiments in which cells were homogenized 30 and 90 min after the onset of CTX uptake in the presence of N-ethylmaleimide. Nocodazol applied after accumulation of CTX in Golgi inhibits the appearance of CTX-A in the ER and delays the increase of 3',5'cAMP, indicating the participation of microtubules in the retrograde Golgi-ER transport.

The role of protein kinase C in carbachol-induced and of cAMP-dependent protein kinase in isoproterenol-induced secretion in primary cultured guinea pig parotid acinar cells.

Stimulation of secretion by muscarinic agonists in guinea pig parotid or pancreatic acini is accompanied by a translocation of protein kinase C (PKC) from the cytosol to the particulate fraction [Machado-De Domenech and Söling (1987) Biochem. J. 242, 749-754] and by a PKC-mediated phosphorylation of the ribosomal protein S6 [Padel and Söling (1985) Eur. J. Biochem. 151, 1-10]. In order to decide whether PKC is directly involved in the secretory process, the effect of down regulation of PKC by phorbol 12-myristate 13-acetate (PMA) was studied in primary cultured guinea pig parotid acinar cells. These cells secrete in response to carbachol and isoproterenol. Only the carbachol response is associated with an increase in cytosolic calcium. Carbachol plus isoproterenol lead to an over-additive stimulation of secretion, an effect which depends completely on the presence of external calcium. Down regulation of PKC by about 90% did not significantly affect carbachol-induced exocytosis, whereas isoproterenol-stimulated secretion was almost doubled. The secretory response to permeable cAMP analogues was also enhanced in PKC-down-regulated acini, indicating a post-receptor effect. The increased response to isoproterenol was also observed in the absence of external calcium. The isoproterenol effect was significantly inhibited by the relatively specific cAMP-dependent protein kinase inhibitor H-89, which had only a minor effect on carbachol-induced exocytosis. Although down regulation of total PKC by up to 90% did not significantly affect the secretory response to carbachol, RO 31-8220, a relatively specific inhibitor of PKC, abolished carbachol-induced secretion in normal as well as in PMA-down-regulated cells. This indicates that a PKC isoform resistant to down regulation by PMA is involved in carbachol- but not in cAMP-mediated secretion.

Variable nuclear cytoplasmic distribution of the 11.5-kDa zinc-binding protein (parathymosin-alpha) and identification of a bipartite nuclear localization signal.

The 11.5-kDa zinc-binding protein (ZnBP, parathymosin-alpha), a potent inactivator of 1-phosphofructokinase, is found only in the cytoplasm of most tissues despite the presence of the putative nuclear localization signal PKRQKT. Recent reports on nuclear uptake of ZnBP could not exclude the participation of unspecific diffusion. We show here that wild-type ZnBP overexpressed in COS cells accumulates exclusively in the nucleus but that ZnBP with a mutated or deleted PKRQKT motif appears both in the nucleus and in the cytoplasm. In contrast, fusion proteins between ZnBP and parts of the endoplasmic reticulum protein calreticulin required the intact PKRQKT motif for nuclear import. The motif RKR, located nine amino acids upstream of the PKRQKT motif, is also involved in the active nuclear import of ZnBP. In contrast to rat hepatocytes and kidney cells in situ, which have ZnBP almost exclusively in the cytosol, we find ZnBP in Reuber H35 hepatoma cells and normal rat kidney cells only in the nuclei. Freshly isolated rat hepatocytes translocate their ZnBP to the nucleus in < 24 h during standard cell culture conditions.

Sorting and budding of constitutive secretory vesicles in hepatocytes and hepatoma cells.

We have isolated a population of post-TGN secretory vesicles from hepatocytes. These vesicles of 100-150 nm diameter carry heparan sulfate proteoglycans. Secretory proteins (albumin, apo-lipoprotein E, fibrinogen) are sorted into different post-TGN secretory vesicles. A member of the ARF family of small GTP-binding proteins is associated with these vesicles. A unique peripheral membrane protein of these vesicles (VAPP14) was shown to exist also on the TGN. Brefeldin A leads to a dissociation of VAPP14 from the TGN. Antibodies against VAPP14 inhibit budding of proteoglycan containing vesicles from the TGN in a cell-free system. Inhibition occurred also in the presence of GTP-gamma-S. The same type of vesicles exists in H35 Reuber hepatoma cells.

Enhanced cAMP production mediates the stimulatory action of pituitary adenylate cyclase activating polypeptide (PACAP) on in vitro catecholamine secretion from bovine adrenal chromaffin cells.

The 38 amino acid peptide pituitary adenylate cyclase activating polypeptide (PACAP) induced a dose dependent increase of catecholamine secretion in cultures of bovine chromaffin cells. This secretagogue activity of PACAP was strictly dependent on the presence of calcium in the culture medium. If calcium was omitted from the medium no effect of PACAP on catecholamine secretion could be detected during an incubation of 20 min. Preincubation of cells with 1 nM PACAP for 5 min facilitated the subsequent nicotine stimulated catecholamine secretion during a 20 min incubation without addition of the peptide. PACAP induced catecholamine secretion was clearly accompanied by a dose dependent increase of intracellular cAMP concentrations. The percentage of cells responding to PACAP with increased catecholamine secretion was assessed by immunocytochemistry of the transient appearance of dopamine-beta-hydroxylase, associated with the membranes of the chromaffin granules on the cell surface during the secretory process. About 70% of adrenal medullary cells responded to 100 nM PACAP with enhanced secretory activity. Though PACAP stimulated catecholamine secretion, we did not observed major effects on intracellular free calcium concentrations ([Ca2+]i) as determined with fura-2 by single cell fluorescence microscopy. In maximally 20% of the cells a rise in [Ca2+]i in response to a challenge with 500 nM PACAP was observed. Lower concentrations of PACAP were without an effect on [Ca2+]i. These data indicate that the stimulatory action of PACAP on in vitro catecholamine secretion from bovine chromaffin cells is linked to a rise of intracellular cAMP.

Two resident ER-proteins, CaBP1 and CaBP2, with thioredoxin domains, are substrates for thioredoxin reductase: comparison with protein disulfide isomerase.

Protein disulfide-isomerase (PDI) is the best known representative of a growing family of enzymes with thioredoxin domains. Two such proteins with thioredoxin (Trx) domains, CaBP1 and CaBP2 (ERp72), have previously been isolated from rat liver microsomes. Here we report that they, like PDI are substrates for thioredoxin reductase and will catalyze NADPH-dependent insulin disulfide reduction. The activity of CaBP1 and CaBP2 in this assay was higher than that of PDI but lower than that of E. coli Trx. Furthermore, as isolated the thioredoxin domains of CaBP1 and CaBP2 were in disulfide form as judged by stoichiometric oxidation of 2 and 3 mol of NADPH in CaBP1 and CaBP2, respectively. The redox potential of the active site disulfide/dithiol was estimated from the equilibrium with a mutant E. coli Trx, P34H Trx, with a known redox potential (-235 mV). This showed that CaBP1 and CaBP2, like PDI, have a much higher redox potential than wild type thioredoxin (-270 mV) in agreement with a role in formation of protein disulfide bonds. In conclusion, in vitro CaBP1 and CaBP2 share catalytic properties in thiol disulfide-interchange reactions with PDI. Thus, the well known activity of PDI is not unique in the endoplasmic reticulum and CaBP1 and CaBP2 may be regarded as functional equivalents.

Localization of the Lys, Asp, Glu, Leu tetrapeptide receptor to the Golgi complex and the intermediate compartment in mammalian cells.

The carboxyl-terminal Lys-Asp-Glu-Leu (KDEL), or a closely-related sequence, is important for ER localization of both lumenal as well as type II membrane proteins. This sequence functions as a retrieval signal at post-ER compartment(s), but the exact compartment(s) where the retrieval occurs remains unresolved. With an affinity-purified antibody against the carboxyl-terminal sequence of the mammalian KDEL receptor, we have investigated its subcellular localization using immunogold labeling on thawed cryosections of different tissues, such as mouse spermatids and rat pancreas, as well as HeLa, Vero, NRK, and mouse L cells. We show that rab1 is an excellent marker of the intermediate compartment, and we use this marker, as well as budding profiles of the mouse hepatitis virus (MHV) in cells infected with this virus, to identify this compartment. Our results demonstrate that the KDEL receptor is concentrated in the intermediate compartment, as well as in the Golgi stack. Lower but significant labeling was detected in the rough ER. In general, only small amounts of the receptor were detected on the trans side of the Golgi stack, including the trans-Golgi network (TGN) of normal cells and tissues. However, some stress conditions, such as infection with vaccinia virus or vesicular stomatitis virus, as well as 20 degrees C or 43 degrees C treatment, resulted in a significant shift of the distribution towards the trans-TGN side of the Golgi stack. This shift could be quantified in HeLa cells stably expressing a TGN marker. No significant labeling was detected in structures distal to the TGN under all conditions tested. After GTP gamma S treatment of permeabilized cells, the receptor was detected in the beta-COP-containing buds/vesicles that accumulate after this treatment, suggesting that these vesicles may transport the receptor between compartments. We propose that retrieval of KDEL-containing proteins occurs at multiple post-ER compartments up to the TGN along the exocytotic pathway, and that within this pathway, the amounts of the receptor in different compartments varies according to physiological conditions.