Proc. Nati. Acad. Sci. USA Vol. 86, pp. 6992-6996, September 1989 Cell Biology Targeting and processing of glycophorins in murine erythroleukemia cells: Use of brefeldin A as a perturbant of intracellular traffic (endoplasmic reticulum/fatty acylation/glycosylation/Golgi complex) JEFFREY B. ULMER* AND GEORGE E. PALADE Department of Cell Biology, Yale University School of Medicine, P.O. Box 3333, New Haven, CT 06510 Contributed by George E. Palade, June 1, 1989

ABSTRACT We previously showed that glycophorins are brane; the others are soluble or membrane-associated but not expressed in virus-transformed, murine erythroleukemia cells; protected against proteolysis (7). we detected four glycophorin precursors (two more than in Brefeldin A (BFA) is a fungal metabolite (10) that inhibits normal erythroblasts) and found that two of them are not protein secretion and intracellular transport in rat hepato- translocated or are inefficiently translocated across the endo- cytes (11, 12) and leads to the accumulation of proteins plasmic reticulum (ER) membrane. By using the drug brefeldin bearing high mannose-type oligosaccharides in mouse pitu- A to block intracellular transport ofproteins from the ER to the itary cells (13). This arrest in protein processing is coinci- Golgi complex, the translocated precursors were shown to dental with the disorganization and disperson of the Golgi accumulate in the ER, while the untranslocated forms were complex (14) and results in an apparent net membrane flow rapidly degraded with an intracellular half-life of 20 min. to the ER (15, 16). We have used this drug as a perturbant of Brefeldin A did not inhibit the synthesis or fatty acylation ofthe intracellular traffic in order to investigate the effects of BFA precursors but substantially delayed their acquisition of 0- on glycophorin synthesis, processing, and transport in MEL linked oligosaccharides, which indicates that murine glyco- cells. The effective block in protein transport caused by BFA phorins are fatty acylated in the ER and 0-glycosylated in the has enabled us to study the differential fates of the four Golgi complex. Even after 6 hr in brefeldin A, glycophorins glycophorin precursors, as well as to define the locations of were only partially glycosylated, resulting in the accumulation their fatty acylation and glycosylation. of apparently sialylated but lower in apparent molecular mass than mature glycophorins. Complete glyco- phorin processing resumed only after removal of the drug. In MATERIALS AND METHODS murine erythroleukemia cells, brefeldin A caused a rapid and Materials. BFA was a generous gift from Jennifer Lippin- extensive disorganization of the entire Golgi complex accom- cott-Schwartz and Richard Klausner (National Institutes of panied by the accumulation of membranes in a part of the ER Health); tissue culture supplies were purchased from closely associated with ER transitional elements. These findings GIBCO; [35S]methionine (1200 Ci/mmol; 1 Ci = 37 GBq) extend recently published results [Lippincott-Schwartz, J., was from Amersham; D-[6-3H]galactose (25 Ci/mmol) was Yuan, L. C., Bonifacino, J. S. & Klausner, R. D. (1989) CeU from ICN; [9,10-3H]palmitic acid (30 Ci/mmol) and [1,6- 56, 801-8131 and suggest that brefeldin A induces net mem- 3H]glucosamine hydrochloride (40-50 Ci/mmol) were pur- brane flow from the entire Golgi complex to the ER. chased from New England Nuclear; biotinylated wheat germ agglutinin and -agarose were obtained from Sigma; Three glycophorins have been identified in mouse erythroid protein A-Sepharose CL-4B was from Pharmacia; and fatty cells (1-3), and the amino acid sequence of the major form, acid-free bovine serum albumin was from Miles. gp3, has been deduced from recently isolated cDNA clones Metabolic Labeling. MEL cells (clone 745) were grown, (4, 5). Murine glycophorins are fatty acylated (3, 6), phos- harvested, and labeled as previously described (6). Pulse- phorylated (6), and exclusively 0-glycosylated (1, 6, 7). chase labeling experiments were carried out by using the Potential sites for these modifications and amino acid se- following protocol: 1 x 107 cells per ml were incubated for 5 quence comparison of gp3 with human (8) min at 370C in methionine-free RPMI 1640 medium supple- indicate that gp3 is a type I , possessing an mented with fetal bovine serum (10%, vol/vol), BFA (1 extracellular amino terminus and a single, hydrophobic, ,ug/ml), and [35S]methionine (200 ;Ci/ml). Chase conditions membrane-spanning domain. Unlike other proteins of this were initiated by adding unlabeled methionine (final concen- type, however, gp3 does not undergo signal sequence cleav- tration of 1.5 mg/ml; 150,000-fold excess) to the cells fol- age upon translocation (7). Moreover, it does not have at its lowed by incubation for up to 60 additional min. In contin- amino terminus a stretch of hydrophobic amino acids that uous labeling experiments, 5 x 105 cells per ml (20 ml) were could function as a signal sequence (4, 5). Apparently its incubated for up to 6 hr in the presence or absence of BFA hydrophobic stop-transfer sequence functions both as a (1 ,ug/ml) with [35S]methionine (25 ;XCi/ml) in methionine- signal and as an anchor sequence, as in the case of type II free RPMI 1640 medium or in complete medium in the membrane proteins (9). In murine erythroleukemia (MEL) presence of [3H]glucosamine hydrochloride (20 ,uCi/ml) or cells, we have detected four glycophorin precursors (two [3H]galactose (20 ,uCi/ml). For fatty acid incorporation, 2 X more than in normal erythroblasts) (6). Cell fractionation 107 cells per ml were incubated for 1 hr in complete medium studies showed that only two ofthese peptides are apparently containing [3H]palmitic acid (250 ,uCi/ml) and 0.05 mM fatty properly inserted in the endoplasmic reticulum (ER) mem- acid-free bovine serum albumin (mixed for 15 min prior to

The publication costs of this article were defrayed in part by page charge Abbreviations: BFA, brefeldin A; ER, endoplasmic reticulum; payment. This article must therefore be hereby marked "advertisement" MEL, murine erythroleukemia. in accordance with 18 U.S.C. §1734 solely to indicate this fact. *To whom reprint requests should be addressed.

6992 Downloaded by guest on October 3, 2021 Cell Biology: Ulmer and Palade Proc. Natl. Acad. Sci. USA 86 (1989) 6993 incubation). Cells were recovered by sedimentation, washed, and prepared for immunoprecipitation as in ref. 6. 1 2 3 BFA was added to the medium from a 5 mg/ml stock gp2- -43 solution in methanol to a final concentration of 1-10 ,ug/ml (1 gg/ml was found to be sufficient for at least 6 hr). Control cells were incubated with the corresponding concentration of methanol (0.02-0.2%). In recovery experiments, cells (col- lected by sedimentation) were resuspended in fresh complete RPMI medium and incubated at 370C in the absence of BFA gp3= E- 30 for an additional 1-6 hr. 27- Lectin Precipitation. Murine glycophorins were immuno- 26- precipitated from MEL cell lysates with a polyclonal rabbit 23- anti-mouse glycophorin serum, as previously described (6). In lectin precipitation experiments, a modification of the 21- -20 procedure of Schnitzer et al. (17) was used. Metabolically labeled MEL cells (1 x 107 cells) were lysed in 1 ml of FIG. 1. Effect of BFA on glycophorin biosynthesis. MEL cells phosphate-buffered saline containing 0.68 mM CaC12, 1% were incubated for 1 hr with [35S]methionine in the absence (lane 1) SDS, and 5% (vol/vol) Triton X-100 and left on ice for 1 hr. or presence (lane 2) of BFA (1 pg/ml). Aliquots of the BFA-treated by centrifugation cells were collected by sedimentation, washed, and incubated for an The lysates, cleared of particulate material additional 1 hr in complete medium (lane 3). Cells were then (6500 x g for 10 min), were heated in a boiling water bath for processed through immunoprecipitation, SDS/PAGE, and fluorog- 10 min, cooled to 40C, and then incubated (with gentle raphy. The mature glycophorins (gp2 and gp3) and their putative agitation) with anti-glycophorin serum (1:100 dilution) for 15 precursor forms (21, 23, 26, and 27 kDa) are indicated. The mobility min, followed by protein A-Sepharose CL-4B beads (10 mg) of molecular mass markers (in kDa) is shown on the right. for 45 min. The beads were recovered and washed as in ref. 6, and the bound immune complexes were solubilized in lysis the cells resumed glycophorin processing; =90% of the total buffer. The lysates were cooled to 40C and incubated with radioactivity incorporated into the precursors was recovered biotinylated wheat germ agglutinin (0.1 mg/ml), followed by in the cells, and =70% of it was found already converted to avidin-agarose (5 mg), each for 1 hr. The agarose beads were mature forms by 1 hr. These figures, which compare favor- recovered by sedimentation, washed twice with phosphate- ably with previously determined kinetics of glycophorin buffered saline containing 0.68 mM CaC12, 0.2% SDS, and 1% maturation in control cells (7), indicate that MEL cells (vol/vol) Triton X-100, and washed once with phosphate- recover rapidly and virtually quantitatively from BFA treat- buffered saline. Proteins were solubilized from the beads by ment. These data also show that BFA blocks, for at least 1 hr, heating for 5 min in a boiling water bath with 62.5 mM glycophorin processing, presumably by interfering with in- Tris-HCl (pH 6.8), 3% SDS, 1% (vol/vol) 2-mercaptoethanol, tracellular transport. The precursor accumulation of glyco- 2 M urea, and 2 mM EDTA. phorin during BFA treatment has allowed us to study (i) the Determination of Radioactivity. SDS/PAGE (12 or 15% fate of the untranslocated precursors in relation to the others polyacrylamide) was performed as in Laemmli (18), and and (ii) the sites of certain posttranslational modifications of radiolabeled proteins were visualized by fluorography (19). the glycophorins. Incorporation of isotopes into individual proteins was quan- The 26- and 27-kDa Glycophorin Precursors Undergo Turn- tified densitometrically by using a Kodak Visage 2000 image over. Previous studies have shown that the 21- and 23-kDa analyzer. Radioactivity in solution was determined by liquid precursors are cotranslationally inserted into ER mem- scintillation spectrometry in a Beckman L5-2000 scintillation branes, unlike the inefficiently translocated 26- and 27-kDa counter. forms (7). The latter are mostly found in the cytosol and on Electron Microscopy. Aliquots containing 1 x 107 cells the surface of intracellular membranes. BFA was used in were fixed in 2.5% glutaraldehyde in 0.1 M sodium cacody- conjunction with a pulse-chase labeling protocol to arrest a late buffer (pH 7.2) for 1 hr at 0°C. The cells were pelleted, cohort of radioactive glycophorin precursors in the ER or a washed with the same buffer, and postfixed in 2% OS04 in the pre-Golgi compartment. During the chase period, the levels same buffer for 1 hr at 0°C. The cells were then washed and of radioactivity in the 21- and 23-kDa precursors remained processed for transmission electron microscopy by standard nearly constant (Fig. 2), as would be expected for normally procedures. translocated proteins accumulating in the ER. In contrast, the 26- and 27-kDa precursors rapidly lost radioactivity with a calculated half-life of -20 min. These results show that the RESULTS untranslocated precursors undergo degradation in situ, per- Effect of BFA on Protein Synthesis in MEL Cells. BFA had haps by a cytosolic degradative pathway, and explain the no detrimental effect on protein synthesis for at least 6 hr; in modest accumulation of radiolabeled 26- and 27-kDa forms fact, at all time points tested, BFA at 1 ,ug/ml had a during BFA treatment in continuous labeling studies (Fig. 1). stimulatory (30-50%) effect on [35S]methionine incorporation The kinetics of turnover match those previously determined into acid-precipitable material (unpublished data). Glyco- for the 26- and 27-kDa precursors under normal conditions phorin biosynthesis was monitored by immunoprecipitation (7), indicating that BFA has no direct effect on this process. from metabolically labeled cells. In controls, precursors (21, Effect of BFA on Glycophorin Processing. Murine glyco- 23, 26, and 27 kDa) and mature forms (gp2 and gp3) were phorin precursors are normally 0-glycosylated to give rise to detected after 1 hr of labeling (Fig. 1). The minor 34-kDa gp2 and gp3 within 10 min of synthesis (3, 6, 7); in contrast, glycophorin described in ref. 6 was only observed after longer in BFA-treated MEL cells, glycophorin glycosylation was incubations (see Fig. 3). In BFA-treated cells, however, no completely inhibited for at least 1 hr (Fig. 1). However, by 2 mature forms were detected up to 1 hr, and all ofthe label was hr glycophorin forms of slightly lower apparent molecular found in the precursors. By quantifying the incorporated mass than gp2 and gp3 began to accumulate and reached a radioactivity, it was estimated that in BFA-treated cells there substantial level by 6 hr (=67% conversion of precursors in was a 5- to 10-fold enrichment of the 21- and 23-kDa precur- a 2-hr pulse/4-hr chase experiment) (Fig. 3), but mature sors, whereas the 26- and 27-kDa precursors were only glycophorins were not present. It was concluded that these slightly enriched. When BFA was removed from the medium, newly detected forms represented incompletely glycosylated Downloaded by guest on October 3, 2021 6994 Cell Biology: Ulmer and Palade Proc. NatL Acad Sci. USA 86 (1989)

100- 1 2 3

%, 80- gp2-

*. 60-

.° 40-

X 20- gp3= -w

0 15 30 45 60 Incubation, min FIG. 2. Fate of the glycophorin precursors during pulse-chase labeling in the continuous presence of BFA at 1 ,ug/ml. MEL cells FIG. 4. Effect of BFA on glycophorin glycosylation. MEL cells were incubated for 5 min with [35S]methionine, followed by the were incubated for 6 hr with [3H]galactose (lanes 1 and 2) or addition ofexcess unlabeled methionine and further incubation for 60 [3H]glucosamine (lanes 3 and 4) in the absence (lanes 1 and 3) or min. Aliquots taken at 5, 10, 15, 35, and 65 min were processed presence (lanes 2 and 4) of BFA (1 ,ug/ml) and then processed through immunoprecipitation, SDS/PAGE, and fluorography. Ra- through immunoprecipitation, SDS/PAGE, and fluorography. Ar- dioactivity in the precursors was quantified, normalized to the 5-min rowheads indicate the glycophorins labeled in BFA-treated cells. time point for each, and plotted against time as the sum in the 21- plus [3H]Glucosamine is expected to be converted into, and incorporated 23-kDa precursors (0) and the 26- plus 27-kDa precursors (o). as, [3H]galactosamine in O-linked . glycophorins, based on the following evidence: (i) they are morphologic evidence of Golgi complex reorganization in as indicated by the incorporation of BFA-treated cells even after 6 hr (unpublished observations). glycoproteins, [3H]ga- Unlike the substantial delay in glycophorin glycosylation lactose and [3H]glucosamine (Fig. 4); (ii) a protein that com- caused by BFA, fatty acylation of the precursors was not igrates with at least one of these forms was transiently found inhibited by the drug (Fig. 6). The 21- and 23-kDa precursors under normal conditions (Fig. 3); and (iii) they disappeared, incorporated [3H]palmitic acid during a 1-hr incubation in the although not entirely, upon removal of BFA, concomitantly presence of BFA, indicating that in MEL cells this modifi- with the appearance of gp2 and gp3 (Fig. 3). These putative cation takes place in the ER or a pre-Golgi compartment. The immature forms ofgp2 and gp3 were apparently sialylated, as lack of detectable incorporation of palmitate into the 26- and well, judged by their ability to bind to wheat germ agglutinin 27-kDa precursors may be related to their inefficient trans- (Fig. 5). On the surface, these results suggested that the block location across ER membranes or may reflect limits of in protein transport from the ER to the Golgi complex was not detection. complete or that there was a partial resumption in intracel- Effect of BFA on MEL Cell Morphology. The Golgi com- lular traffic because BFA was metabolized (14). However, plexes of MEL cells were extensively disorganized by BFA; this is probably not the case because of the following obser- the cisternal stacks were replaced by clusters of small vesi- vations: (i) medium taken from cells incubated with BFA for cles, most ofthem uncoated. The conversion was not entirely 6 hr retained the ability to inhibit glycophorin glycosylation synchronous within the cell population, but by 30 min the in fresh cells, (ii) supplementing the medium with fresh BFA process was complete (Fig. 7, compare A to B). The vesicular every hour did not prevent the appearance of the incom- clusters retained their association with the centrioles, a pletely glycosylated glycophorins, and (iii) there was no feature that facilitated their identification (Fig. 7B). The amount of membrane in these vesicles was, however, too little to account for the totality of Golgi membranes; there- 1 2 3 4 5 6 7 fore, membrane relocation probably occurred. Concomi- tantly with the disappearance ofthe Golgi cisternae, a section

- of the ER next to the underwent gp2 * - _ Golgi complex pronounced r.- 1 2

34- .4 gp2- -I ..4 gp3= 1_t 'I - ' " =

_ -1:;|," A_

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~.... _ -

gp3=- 3 __- _ _ 9p_._

FIG. 3. Effect of prolonged exposure to BFA on glycophorin processing. MEL cells were incubated with [35S]methionine for 1 hr (lanes 1 and 2), for 2 hr (lanes 3 and 4), or for a 2-hr pulse followed by a 4-hr chase (lanes 5 and 6) in the absence (lanes 1, 3, and 5) or presence (lanes 2, 4, and 6) of BFA (1 mg/ml). An aliquot of the 6-hr labeled, BFA-treated cells was collected by sedimentation, washed, FIG. 5. Lectin precipitation of glycophorins from control and and incubated a further 6 hr in complete medium (lane 7). Cells were BFA-treated MEL cells. Cells were incubated for 6 hr with processed through immunoprecipitation, SDS/PAGE, and fluorog- [35S]methionine in the absence (lane 1) or presence (lane 2) of BFA raphy. Dashes on the left denote the 21-, 23-, 26-, and 27-kDa (1 yg/ml) and then processed sequentially through immunoprecip- precursors; those on the right indicate gp2 and gp3. The arrowheads itation and lectin precipitation. The lectin precipitates were sepa- in lane 6 indicate the glycophorins immunoprecipitated from BFA- rated by SDS/PAGE and visualized by fluorography. Arrowheads treated cells; one of them comigrates with a form transiently seen indicate the -bearing glycophorins present in BFA-treated under normal conditions (see open arrowhead, lane 3). cells. Downloaded by guest on October 3, 2021 Cell Biology: Ulmer and Palade Proc. NatL. Acad. Sci. USA 86 (1989) 6995

1 2 ries, since it appears to be a type I membrane protein with a relatively large ectodomain and without an amino-terminal gp2- signal sequence (4, 5, 7). If its stop-transfer sequence also functions as a signal sequence and if -70 amino acids must be added to the nascent peptide before the signal recognition particle can bind to the signal sequence (25), then elongation ofthe glycophorins may be terminated before effective signal gp3= _NOW recognition particle binding can take place, since the pre- dicted cytoplasmic tail ofgp3 is only =40 amino acids long (4, 5). This may result in inefficient arrest of elongation and loss -23 of translocation competence (26). Moreover, translocation _-21 may be further compromised by the relatively lengthy extra- cellular domain ofgp3 (=92 amino acids) acquiring a tertiary structure before the signal sequence emerges from the ribo- FIG. 6. Effect of BFA on fatty acylation of glycophorins. MEL some. The involvement of cytosolic factors, such as heat cells were incubated for 1 hr with [3H]palmitic acid in the absence shock proteins, to maintain a translocation-competent state (lane 1) or presence (lane 2) of BFA (1 gg/ml) and then processed (for review see ref. 27) may be particularly important for through immunoprecipitation, SDS/PAGE, and fluorography. Pal- proteins like murine glycophorins. Situations such as these mitate was incorporated into glycophorin mature forms (gp2 and gp3) could explain the presence of some glycophorin precursors in and two of the precursors (21 and 23 kDa). the cytosol and on the cytosolic aspect ofmembranes in MEL cells (7). However, structural differences in the 26- and expansion. It consisted of a network of dilated tubules 27-kDa precursors, compared to the 21- and 23-kDa forms, provided with few attached ribosomes but associated with must also play a role in determining the efficiency of their extensive accumulations of viral proteins (Fig. 7C). This insertion into ER membranes. It will be illuminating to characteristic compartment apparently includes transitional compare the primary structures of the glycophorin precur- ER elements and is in continuity with the rest of the ER. A sors in order to ascertain the characteristics that preclude the pre- (or juxta) Golgi compartment of the type described is normal translocation of the 26- and 27-kDa precursors. Even occasionally seen in MEL cells in the absence ofBFA, but its a subtle change in the leader peptide of the maltose-binding dimensions are smaller (20). Its rapid expansion in BFA- protein causes a defect in its export in Escherichia coli (28). treated cells may be explained by inflow ofGolgi membranes. However, regions ofthe carboxyl and amino ends ofproteins also play a role in proper targeting to the ER (29, 30). Hence, differences in many regions of the glycophorin precursors DISCUSSION could potentially affect their insertion into ER membranes. Type I membrane proteins (i.e., proteins with an extracellular The results of this study demonstrate that the untranslocated amino terminus and a single membrane-spanning domain) precursors (26 and 27 kDa) undergo rapid turnover, presum- have an amino-terminal signal sequence that is cleaved by a ably by means of a cytosolic degradative pathway. In con- signal peptidase whose active site is on the luminal surface of trast, the other precursors (21 and 23 kDa) are cotranslation- the ER (21). In contrast, the membrane-anchor segment of ally inserted into the ER and subsequently processed to give type II membrane proteins (which have an intracellular amino rise to mature glycophorins. It may be relevant that the terminus) can function as both an uncleaved signal sequence predicted amino acid sequence ofgp3 contains regions rich in and a stop-transfer sequence (9, 22). The few proteins that proline, glutamic acid, serine, and threonine, which is com- have a type I orientation but lack an amino-terminal signal mon to many rapidly degraded proteins (31). sequence (type III by the nomenclature in ref. 23) usually Membrane proteins are exported from the ER to the Golgi have only a short stretch of amino acids between the amino complex and pass successively through each cisterna of the terminus and the membrane anchor domain (23). It is inter- Golgi stack before arriving at their final destination (32). esting to note that one example of this type of membrane During their transit through these compartments, many pro- protein is human (24). The major murine teins are modified by fatty acylation and glycosylation. The glycophorin, gp3, may not fall into any of the above catego- mechanisms and topology of N-linked glycosylation are well

(N~~~~~~~~~~~~~W FIG.7Mohlyocno nBAradM cesMEcl e nbtdo miiteaec()openeBn

'.-, "' '. ':.5,,i tB : : >< s > X

C) of BFA (1 ,ug/ml) and then prepared for electron microscopy. Note that, compared to controls, the BFA-treated cells (i) lack Golgi stacks, (ii) have a preponderance of small, uncoated vesicles near the centriole (see arrowhead in B), and (iii) have incomplete virus particles accumulated on predominantly smooth ER membranes (C). (A, x18,000; B, x22,000; C, x16,000.) Downloaded by guest on October 3, 2021 6996 Cell Biology: Ulmer and Palade Proc. Natl. Acad. Sci. USA 86 (1989) documented (33), but the sites of addition of O-linked oligo- manuscripts. This work was supported by a program project grant saccharides and fatty acyl residues are not. Recent studies (no. 1 P01 CA46128-01) from the National Cancer Institute. have shown that the first step in O-glycosylation (i.e., addi- 1. Sarris, A. H. & Palade, G. E. (1979) J. Biol. Chem. 254, tion of N-acetylgalactosamine) occurs posttranslationally in 6724-6731. a pre-Golgi compartment (34, 35). However, the site of 2. Sarris, A. H. & Palade, G. E. (1982) J. Cell Biol. 93, 583-590. initiation of this process appears to vary in different cell 3. Dolci, E. D. & Palade, G. E. (1985) J. Biol. Chem. 260, types, ranging from the Golgi complex in rat hepatocytes (36) 10728-10735. to possibly as early as cotranslationally in pancreatic islet 4. Matsui, Y., Natori, S. & Obinata, M. (1989) Gene 77, 325-332. cells (37). Subsequent addition of galactose and sialic acid is 5. Dolci, E. D., Martin-Vasallo, P., Emanuel, J. R., Palade, G. E. & Levenson, R. (1988) J. Cell Biol. 107, 560a (abstr.). assumed to occur in a trans Golgi compartment (33). Based 6. Ulmer, J. B., Dolci, E. D. & Palade, G. E. (1989) J. Cell Sci. on the observation that BFA, which blocks transport of 92, 163-171. proteins out of the ER (11-16), completely inhibited 0- 7. Ulmer, J. B. & Palade, G. E. (1989) J. Biol. Chem. 264, glycosylation of murine glycophorins for at least 1 hr, we 1084-1091. conclude that this modification is initiated in a Golgi com- 8. Tomita, M. & Marchesi, V. T. (1975) Proc. Nadl. Acad. Sci. partment in MEL cells. However, we cannot rule out the USA 72, 2964-2968. 9. Spiess, M. & Lodish, H. F. (1986) Cell 44, 177-185. possibility that some minor processing occurs in the ER 10. Harmr, E., Loeffler, W., Sigg, H. P., Straehelin, H. & Tamm, without resulting in a change in electrophoretic mobility or C. (1963) Helv. Chim. Acta 46, 1235-1243. detectable incorporation of hexose precursors. The site of 11. Oda, K., Hirose, S., Takami, N., Misumi, Y., Takatsuki, A. & fatty acylation of nascent membrane proteins has not been Ikehara, Y. (1987) FEBS Lett. 214, 135-138. defined. Recent work has shown that palmitoylation can take 12. Misumi, Y., Misumi, Y., Miki, K., Takatsuki, A., Tamura, G. place in the ER (38), even on nascent chains during transla- & Ikehara, Y. (1986) J. Biol. Chem. 261, 11398-11403. tion (39). Other studies assume a Golgi location (for review, 13. Perkel, V. S., Liu, A. Y., Miura, Y. & Magner, J. A. (1988) see ref. 40). Our results demonstrate that murine glycophor- Endocrinology 123, 310-318. 14. Fujiwara, T., Oda, K., Yokota, S., Takatsuki, Y. & Ikehara, Y. ins are fatty acylated while still in the ER, as indicated by the (1988) J. Biol. Chem. 263, 18545-18552. fact that BFA did not inhibit the incorporation of[3H]palmitic 15. Lippincott-Schwartz, J., Yuan, L. C., Bonifacino, J. S. & acid into the accumulated precursors. Klausner, R. D. (1989) Cell 56, 801-813. The eventual galactosylation and sialylation ofglycophorin 16. Doms, R. W., Russ, G. & Yewdell, J. W. (1989) J. Cell Biol. precursors indicates that processes normally associated with 109, 51-72. the Golgi complex occur on proteins in the ER after pro- 17. Schnitzer, J. E., Carley, W. W. & Palade, G. E. (1988) Proc. longed exposure to BFA. A possible explanation for these Natl. Acad. Sci. USA 85, 6773-6777. results comes from two recent studies (15, 16), which con- 18. Laemmli, U. K. (1970) Nature (London) 227, 680-685. 19. Bonner, W. M. & Laskey, R. A. (1974) Eur. J. Biochem. 46, cluded that BFA induces a rapid redistribution ofcis/middle 83-88. Golgi enzymes to the ER, concomitantly with the partial 20. Walter, R. J. & Tandler, B. (1989) J. Submicrosc. Cytol. disappearance of Golgi stacks. In our case, the gradual Pathol. 21, 93-101. acquisition of O-linked oligosaccharides by MEL cell glyco- 21. Evans, E. A., Gilmore, R. & Blobel, G. (1986) Proc. Natl. phorin precursors is apparently the result of a similar but Acad. Sci. USA 83, 581-585. more extensive mixing of Golgi and ER membrane proteins. 22. Mize, N. K., Andrews, D. W. & Lingappa, V. R. (1986) Cell In these cells, BFA caused the disorganization and disap- 47, 711-719. pearance of all Golgi stacks, simultaneously with the appear- 23. von Heijne, G. & Gavel, Y. (1988) Eur. J. Biochem. 174, ance of numerous small vesicles near the centrioles (Fig. 7). 671-678. 24. High, S. & Tanner, M. J. A. (1987) Biochem. J. 243, 277-280. In addition, viral coat proteins accumulated and budded 25. Wiedmann, M., Kurzchalia, T. V., Bielka, H. & Rapoport, intracisternally in a distinct part of the ER (close to transi- T. A. (1987) J. Cell Biol. 104, 201-208. tional elements), rather than in the entire ER system. BFA 26. Siegel, V. & Walter, P. (1988) EMBO J. 7, 1769-1775. prevented membrane protein sialylation in refs. 15 and 16, 27. Meyer, D. I. (1988) Trends Biochem. Sci. 13, 471-474. but allowed it (albeit slow and incomplete) in MEL cells, 28. Liu, G., Topping, T. B., Cover, W. H. & Randall, L. L. (1988) suggesting that BFA effects vary in extent from one cell type J. Biol. Chem. 263, 14790-14793. to another. In MEL cells (for instance), the disorganization 29. Lipp, J. & Dobberstein, B. (1988) J. Cell Biol. 106, 1813-1820. of the Golgi complex is complete and galactosyl- and sialyl- 30. Andrews, D. W., Perara, E., Lesser, C. & Lingappa, V. R. transferases to the ER; in other cells, (1988) J. Biol. Chem. 263, 15791-15798. apparently migrate 31. Rogers, S., Wells, R. & Rechsteiner, M. (1986) Science 234, trans Golgi elements are spared and galactosylation and 364-368. sialylation of proteins in the ER are not detected. This 32. Farquhar, M. G. (1985) Annu. Rev. Cell Biol. 1, 447-488. assumption remains to be checked by following the distribu- 33. Kornfeld, R. & Kornfeld, S. (1985) Annu. Rev. Biochem. 54, tion of trans Golgi markers during BFA treatment. 631-644. The effects of BFA on processing in MEL 34. Tooze, S., Tooze, J. & Warren, G. (1988) J. Cell Biol. 106, cells are consistent with the premise that there is an ongoing 1475-1487. recycling of vesicles between the ER and Golgi complex. In 35. Pathak, R. K., Merkle, R. K., Cummings, R. D., Goldstein, this way, lipids transported to the Golgi complex (41), as well J. L., Brown, M. S. & Anderson, R. G. W. (1988) J. Cell Biol. as luminal ER proteins that have been sorted in a post-ER 106, 1831-1841. 36. Abeijon, C. & Hirschberg, C. B. (1987) J. Biol. Chem. 262, compartment (42), could be recovered by the organelle. 4153-4159. Further study of these putative recycling vesicles could be 37. Patzelt, C. & Weber, B. (1986) EMBO J. 5, 2103-2108. facilitated by their isolation and characterization from BFA- 38. Rizzolo, L. J. & Kornfeld, R. (1988) J. Biol. Chem. 263, treated cells. 9520-9525. 39. Slomiany, A., Tsukada, G. & Slomiany, B. L. (1988) Int. J. We thank Drs. J. Lippincott-Schwartz and R. Klausner for gen- Biochem. 20, 1381-1390. erously providing a sample of BFA, R. O'Keefe for assistance with 40. Sefton, B. M. & Buss, J. E. (1987) J. Cell Biol. 104, 1449-1453. the electron microscopy, and P. Male for performing the image 41. Pfeffer, S. R. & Rothman, J. E. (1987) Annu. Rev. Biochem. 56, analysis. We also thank Drs. J. Lippincott-Schwartz, R. Klausner, 829-852. R. Doms, and M. Obinata for sending copies of their unpublished 42. Pelham, H. R. B. (1988) EMBO J. 7, 913-918. Downloaded by guest on October 3, 2021