THE JOURNAL OF BIOLOGICAL CHEMISTRY Vol. 268, No. 16, Issue of June 5, pp. 1201&12016,1993 Printed in U.S.A.

Brefeldin A Blocks the Response of Cultured Cells to Cholera Toxin IMPLICATIONS FOR INTRACELLULAR TRAFFICKING IN TOXIN ACTION*

(Received for publication, December 30, 1992, and in revised form, February 10, 1993)

Palmer A. Orlandi, Patricia K. Curran, and Peter H. Fishman4 From the Membrane Biochemistry Section, Laboratory of Molecular and Cellular Neurobiology, National Institute of Neurological Disorders and Stroke, National Institutes of Health, Bethesda, Maryland 20892

Cholera toxin (CT) consists of a pentameric B subunit which binds to ganglioside G M ~ on the cell surface and an A subunit which activates adenylylcyclase. The lat- ter process involves the reduction of A to the A1 peptide which ADP-ribosylates the stimulatory G protein, G. of adenylylcyclase. There is a distinct lag phase be- tween toxin binding and activation of adenylylcyclase. Little is known about the events during this lag includ- ing where A1 is generated and how it gains access to G. on the cytoplasmic side of the plasma membrane. We explored the effects of several inhibitors of intra- cellular trafficking on the response of human SK-N- MC neurotumor and Caco-2 intestinal tumor cells to CT. Whereas chloroquine or monensin had little or no effect on CT stimulation of cyclic AMP accumulation, brefeldin A (BFA) totally inhibited the response to CT in a time- and dose-dependent and reversible manner. BFA was effective when added at the same time as CT and had an ICao of 30 ng/ml. BFA did not alter cell surface Gml as cells treated with BFA for 30 min bound as much 1a61-CT as control cells. Furthermore, BFA inhibited CT stimulation of GM1-treated rat glioma C6 cells. BFA treatment did not affect &adrenergic ago- nist stimulation of cyclic AMP. In addition, adenylyl- cyclase was activated by AI peptide and NAD+ to the same extent in membranes from control and BFA- treated cells, or when BFA was added directly to the assay. Whereas control cells generated small amounts of Al from bound CT with time, no A, was detected in BFA-treated cells. BFA treatment did not prevent the internalization of CT but did inhibit its degradation. BFA is known to disrupt the organization of the Golgi complex, resulting in inhibition of protein transport from the endoplasmic reticulum and redistribution of Golgi enzymes to the endoplasmic reticulum. BFA also prevents the formation of non-clathrin-coated vesicles from Golgi membranes and thus vesicular transport between Golgi cisternae. We confirmed that BFA caused the morphological disruption of the Golgi ap- paratus in Caco-2 cells. The data support a role for a functional Golgi apparatus with its associated vesicu- lar routing in CT action.

Cholera toxin (CT)’ is the causative agent of the diarrheal

* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “adoertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

2 T o whom correspondence should be addressed: Bldg. 49, Rm. 2A28, National Institutes of Health, Bethesda, MD 20892. Tel.: 301- 496-1325; Fax: 301-496-8244.

The abbreviations used are: CT, cholera toxin; BFA, brefeldin A; G., stimulatory GTP-binding protein; ARF, ADP-ribosylation factor; IBMX, 3-isobutyl-1-methylxanthine; EMEM, Eagle’s minimal essen-

MDCK, Madin-Darby canine kidney cells; GMl, 113 Neu5Ac-Gg- tial medium; BSA, bovine serum albumin; ER, endoplasmic reticulum;

Ose,Cer; NBD-ceramide, N-[N-(7-nitro-2,1,3,-benzoxadiazol-4-yl)-~- aminohexanoyl]-D-erythro-sphingosine.

disease cholera, and mediates its effects by increasing CAMP (Finkelstein, 1973). Although the human small intestinal mucosal cell is the normal target of the toxin, CT is a ubiquitious activator of adenylylcyclase in most vertebrate cells (van Heyningen, 1983; Fishman, 1990). The structure of CT is well defined, especially as the crystal structure of the homologous Escherichia coli heat-labile enterotoxin has re- cently been published (Sixma et al., 1991). CT is composed of an A subunit and a homopentameric B subunit. Each subunit has a different function; the B subunit recognizes and binds to specific receptors on the cell surface which have been identified as the ganglioside GMl (Fishman, 1990). The A subunit, which consists of two peptides, A, and Az linked by a disulfide bond, activates adenylylcyclase. The A, peptide is an ADP-ribosyltransferase which catalyzes the transfer of ADP-ribose from NAD’ to the a subunit of the stimulatory G protein, G., resulting in the persistent activation of adeny- lylcyclase (Moss and Vaughan, 1979).

The intervening steps between binding and activation are less well understood. There is a characteristic lag period after CT binds to the cell surface and before an increase in adeny- lylcyclase activity is observed (Fishman, 1980). At the end of the lag period, small amounts of A1 peptide begin to be formed and increase in parallel with the increase in adenylylcyclase activity (Kassis et al., 1982). The site where A, is generated and the mechanism(s) involved are not known. How AI reaches G,, which is located on the cytoplasmic face of the plasma membrane also is unclear. Two major models have been proposed. The first is based on photolabeling studies (Wisnieski and Bramhall, 1981; Tomasi and Montecucco, 1981), and envisions the CT-A penetrating across the mem- brane bilayer, undergoing reduction, and releasing A1 at the cytoplasmic face of the membrane where it can activate G. (reviewed in Fishman (1990)). In the other model, the holo- toxin undergoes endocytosis through noncoated pits or invag- inations of the plasma membrane and appears first in non- coated vesicles and then in a tubulovesicular compartment (Tran et al., 1987). This latter model is supported by studies on the intoxification of rat hepatocytes both in vivo and in vitro (Houslay and Eliott, 1981; Janicot and Desbuquois, 1987; Janicot et al., 1988, 1991). In these studies, the acidotropic agent chloroquine, as well as the carboxylic ionophore monen- sin, were found to inhibit CT activation of adenylylcyclase and formation of A1, as well as increase the lag period for both processes. In addition, chloroquine reduces the amount of labeled CT found in endosomal fractions.

In the present study, we explored the effects of chloroquine, monensin, and brefeldin A on the response of several cultured cell lines to CT. BFA-treated cells exhibit a block in the export,of proteins from the Golgi apparatus, a rapid dissembly of the Golgi apparatus as a distinct morphological compart- ment, and a redistribution of Golgi components into the ER (reviewed in Klausner et al. (1992)). BFA also blocks the

12010

Brefeldin A Blocks Cholera Toxin Action 12011

intoxification of cells by several plant and bacterial toxins which inhibit protein synthesis (Yoshida et al., 1991; Hudson and Grillo, 1991; Sandvig et al., 1991; Fishman and Curran, 1992). In addition to two neurotumor cell lines, we used Caco- 2 cells which were isolated from a human colonic adenocar- cinoma but resemble small intestinal enterocytes and thus are a more appropriate model for investigating the mechanism of action of CT. We found that BFA but not chloroquine or monensin blocked the response of these cells to CT.

EXPERIMENTAL PROCEDURES

Materials-CT and CT-A were obtained from List Biological Lab- oratories (Campbell, CA). Monensin was from Calbiochem. C6-NBD- ceramide was from Molecular Probes (Eugene, OR). Sigma was the source of (-)-isoproterenol, chloroquine, 3-isobutyl-1-methylxan- thine (IBMX), the reagents for the cyclase assay, and BFA. Samples of BFA also were obtained from Boerhinger Mannheim and Epicentre Technologies (Madison, WI); they were stored as stock solutions of 1 or 5 mg/ml in ethanol a t -20 "C. 1251-Protein A (8.3 pCi/pg) was from Du Pont-New England Nuclear. Antibodies to the A1 peptide were raised in rabbits. Briefly, after reducing CT-A, the A1 peptide was separated from As, and coupled to keyhole limpet hemocyanin (Calbiochem). The conjugate then was used as an immunogen.

Cells and Cell Culture-SK-N-MC, rat glioma C6, and Caco-2 cells were obtained from the American Type Culture Collection (Rockville, MD). Caco-2 cells were maintained in Eagle's minimal essential medium (EMEM) supplemented with nonessential amino acids, so- dium pyruvate, 2 mM glutamine, and 20% NuSerum IV (Collaborative Biomedical). SK-N-MC cells were maintained in the same medium but with only 10% NuSerum. SK-N-MC cells used for experiments were grown in a 1:l mixture of this medium and Ham's F-12 contain- ing 10% NuSerum which enhances cell attachment. Rat glioma C6 cells were cultured as described previously (Pacuszka and Fishman, 1990). For experiments involving cAMP accumulation and toxin binding, cells were grown in 24 X 16-mm clusters; for antibody binding, in 12 X 22-mm clusters; for assaying formation of AI peptide, in 6 X 35-mm clusters; and for preparing membranes and assaying adenylylcyclase, in 75- or 175-cm2 flasks. In some experiments, C6 cells were treated for 1 h with 0.3 p~ Gul in serum-free medium buffered with 25 mM Hepes (Pacuszka and Fishman, 1990).

Accumulation of CAMP and Activation of Adenylylcyclese-Cells were incubated at 37 "C in serum-free medium buffered with 25 mM Hepes and containing 1 mM IBMX and 0.01% BSA with 1 nM CT for 2 h unless otherwise indicated. The cells then were extracted with 0.1 M HC1, and the extracts assayed for cAMP by radioimmune assay (Zaremba and Fishman, 1984). Routinely, BFA or other drugs were added 30 min before adding the toxin. For the determination of adenylylcyclase activity, cells were incubated at 37 "C with and with- out BFA in EMEM/Hepes. The medium was replaced with ice-cold medium containing 1 nM CT and 0.01% BSA, and the cells were incubated at 4 "C for 30 min. The cells then were incubated at 37 "C for different times by replacing the medium with warm EMEM/ Hepes. With each medium change, BFA was added as required. The cells were washed with ice-cold Dulbecco's phosphate-buffered saline, and lysed in 1 mM Tris-HC1,2 mM EDTA, pH 7.4. Crude membranes were prepared (Zhou and Fishman, 1991), suspended in 10 mM Tris- HCl, pH 7.7, and portions assayed for adenylylcyclase activity. Briefly, membranes were incubated in 100 pl containing 50 mM Tris- HCI, pH 7.7, 1 mM EDTA, 2.4 mM MgC12, 0.1% BSA, 1 mM theoph- ylline, 50 p M GTP, 1 mM dithiothreitol, 1 p~ (-)-alprenolol, and 1 mM ATP for 10 min at 30 "C. The reactions were stopped by boiling and assayed for CAMP. Membranes also were incubated with 1 YM activated CT (10 pM incubated 10 min at 37 "C with 20 mM dithlo- threitol), 1 mM NAD+, and 100 pM GTP for 10 min at 30 "C and then assayed for adenylylcyclase activity as described above.

Staining of Cells with a Fluorescent Ceramide Analogue-Vital staining of the Golgi apparatus with Ce-NBD-ceramide was done by a modification of the method of Ktistakis et al. (1991), using cells grown in 8-well chamber slides (Lab-Tek from Nunc, Naperville, IL).

EMEM, Hepes, 0.34% defatted BSA containing 5 p~ Cs-NBD-cer- Briefly, the cells were washed to remove any serum, incubated in

amide for 30 min at 37 "C, washed, and incubated in the same medium (1 h, 37 'C) to allow the stain to accumulate in the Golgi. The cells then were incubated in the same medium with increasing concentra- tions of BFA (30 min, 37 "C) and incubated in two changes of the same medium plus BFA (15 min each, 25 "C) to remove any nonspe- cific stain. Finally, the cells were fixed in 0.5% gluteraldehyde in

calcium- and magnesium-free phosphate-buffered saline (15 min, 4 "C). After rinsing with phosphate-buffered saline, the slides were mounted with a coverslip and the cells were observed by fluorescence

APOCHROMAT 63X (1.4 NA) objective and photographed with microscopy using a Zeiss Axiophot microscope equipped with a Plan-

Kodak TMAX 400 film. Other Methods-Established methods were used to determine the

generation of AI peptide by intact cells (Kassis et al. 1982), the immunological detection of cell surface CT subunits (Fishman, 1982), the binding of Iz5I-CT to intact cells in situ (Spiegel, 1985), and the degradation of bound ls51-CT (Fishman, 1982). Unless otherwise indicated, values represent the mean f S.D. of triplicate determina- tions from one of at least three similar experiments. When error bars were less than 5% of the mean, they were not shown.

RESULTS

Effects of Inhibitors of Intracellular Trafficking on Response of Cells to CT-Human intestinal Caco-2 cells were highly sensitive to CT (Fig. 1). Half-maximal stimulation of cAMP accumulation occurred at 50 PM toxin, and the cells exhibited the characteristic lag between exposure to CT and the rise in CAMP. Human neurotumor SK-N-MC cells were similarly sensitive to CT (data not shown). Prior exposure of either cell line to chloroquine or monensin had little or no effect on their response to CT; even 0.4 mM chloroquine was ineffective (Fig. 1, Table I, and data not shown). By contrast, prior exposure of the cells to BFA caused a dramatic reduction in CT stimulation of cAMP with a 50% inhibition occurring at 30 ng/ml BFA (Fig. 2.4) . Similar results were obtained with

5 1 5

E c v

1.0 I

2 0 5

0

0 ,0"2 10-10 10-9 ,o-8

CHOLERA TOXIN (M) MINUTES

FIG. 1. Dose and time dependence of CT-stimulated cAMP response in Caco-2 cells: effects of chloroquine and monensin. Cells were incubated in EMEM/Hepes containing 1 mM IBMX and 0.01% BSA with no addition (O), 100 p~ chloroquine (m), or 10 p~ monensin (A) for 30 min at 37 "C. Then the cells were exposed to either increasing concentrations of CT for 2 h ( A ) or to CT (1 nM final) added at different times over a 2-h period ( B ) . After 2 h, the cells were assayed for cAMP and protein as described under "Exper- imental Procedures."

TABLE I Cholera toxin stimulation of cyclic AMP in control and chloroquine-

or monensin-treated cells Cells grown in 24-well clusters were incubated in 0.5 ml of EMEM/

Hepes containing 1 mM IBMX and 0.01% BSA for 30 min in the presence of the indicated concentration of chloroquine or monensin. Then CT (1 nM final) was added to some of the wells, and the cells were incubated an additional 2 h and assayed for cAMP and protein as described under "Experimental Procedures." Values are the means f S.D. of triplicate wells from one of two similar experiments.

Cyclic AMP accumulation

Treatment SK-N-MC cells Caco-2 cells

-Toxin +Toxin -Toxin +Toxin

p m l / m g protein p m l j m g protein None 9.4 f 0.9 3130 rt 158 31.6 f 5.9 2770 f 148 100 p M Chloroquine 10.8 f 0.9 3280 f 232 41.5 f 4.3 2870 f 393 400 p~ Chloroquine 7.8 f 1.3 3560 f 117 45.4 k 1.2 3430 f 191 10 p M Monensin 9.8 f 1.0 3320 f 68 47.1 f 2.4 2570 f 131

12012 Brefeldin A Blocks Cholera Toxin Action

1 :: , , , ~ ...t J :: , . , , , ";- c

'0 10' 10' lo2 10' IO4 Om - 3 0 -20 -10 0 10 20 30 [BREFELDIN A] (ng/mL) TIME BETWEEN BFA AND CT (min)

FIG. 2. Effect of BFA on CT stimulation of cAMP in Caco- 2 (0) and SK-N-MC (0) cells. A, cells were incubated in EMEM/ Hepes containing 1 mM IBMX and 0.01% BSA in the presence of increasing concentrations of BFA for 30 min, and then stimulated with 1 nM CT for 2 h. B, cells were exposed to 1 pg/ml BFA for the indicted times before or after the addition of 1 nM CT and incubated for 2 h after adding the toxin. Data points are the means f S.D. of triplicate wells from one of two similar experiments except for A, 0, which represents the means of three separate experiments.

TABLE I1 Effect brefeldin A on stimulation of rat glioma C6 cells by cholera

toxin Cells were incubated for 30 min at 37 "C in Hepes-buffered medium

in the absence and presence of 0.3 @M GMl; BFA (1 g / m l final) was added as indicated and the incubation was continued for 30 min. The cells then were rapidly washed and incubated in Hepes-buffered medium containing 1 mM IBMX and 0.01% BSA; BFA and CT (10 nM final) were added as indicated. After 90 min at 37 "C, the cells were assayed for cAMP and protein as described under "Experimental Procedures." Values are the means f S.D. of triplicate wells from one of two similar experiments.

GMl treatment BFA Cyclic AMP accumulation

-Toxin +Toxin pmollmg protein

- - 22.1 f 5.3 396 f 23 + 20.4 f 1.1 24.3 sf: 3.6 - 22.8 f 0.8 3040 k 124 + + 19.0 f 6.2 19.3 f 1.7

- +

BFA from three different sources (data not shown, but see "Experimental Procedures" for sources). The inhibitory effect of BFA was time-dependent and occurred very rapidly (Fig. 2B). Almost complete inhibition was observed even when BFA was added to the cells at the same time as CT. Adding BFA during the lag phase reduced its effectiveness as an inhibitor.

BFA treatment also inhibited the response of rat glioma C6 cells to CT (Table 11). These cells have very few CT receptors and thus respond poorly to the toxin; their response can be greatly increased by treatment with G M l (Fishman, 1980; Spiegel, 1985). BFA was able to block the stimulation of G M ~ - treated C6 cells by CT (Table 11). Taken together with the rapid effect of BFA, it is unlikely that the drug is acting through the receptors for CT. BFA has been reported to inhibit ganglioside biosynthesis (van Echten et al., 1990; Young et al., 1990), an effect not unexpected as ganglioside biosynthesis occurs in the Golgi apparatus which is disrupted by BFA. To further address this possibility, we directly as- sayed control SK-N-MC cells and cells treated 30 min with BFA, chloroquine, or monensin for lZ6I-CT binding and found no differences (data not shown).

The inhibitory effect of BFA on all three cell lines was reversible (Fig. 3A). In these experiments, the cells were exposed to 1 pg/ml BFA for 30 min, washed several times, and then stimulated with CT for 90-120 min. Exposure of cells to BFA followed by its removal, however, altered the time course of CT stimulation by doubling the lag phase (Fig.

I -CT +BFA/+CT

0 +CT +BFA/ -BFA/+CT

C 6 GM1-C6 CACO-2 SK-N-MC

r

MINUTES -~

MINUTES

FIG. 3. Reversibility of BFA-mediated inhibition of CT ac- tion. A, cells were incubated in the absence and presence of 1 pg/ml BFA for 30 min. Then the cells were washed twice, incubated with BFA and CT as indicated using the media, concentrations, and times described in Tables I and 11, and assayed for cAMP and protein. B, SK-N-MC cells were incubated in the absence (0) and presence (A) of 1 pg/ml BFA for 30 min, washed twice, and incubated with 1 nM CT for the indicated times. C, SK-N-MC cells were incubated in the absence (0) and presence (A) of 1 pg/ml BFA for 30 min at 37 "C, incubated in ice-cold medium f BFA with 2 nM CT for 30 min at 4 "C, washed twice, and incubated at 37 "C for the indicated times. Data points are the means -+ S.D. of triplicate wells from one of two similar experiments.

3B). To explore the possibility that CT might be trapped at some intracellular site in the BFA-treated cells, cells were treated with BFA, cooled to 4 "C, and incubated with CT. At this temperature, CT is able to bind to the cells but is unable to enter them, to be converted to AI peptide, or to activate adenylylcyclase (Fishman, 1980, 1982; Kassis et al., 1982). The cells then were washed free of BFA, incubated at 37 "C, and assayed for cAMP at different times (Fig. 3C). Again, the lag period was increased by prior treatment of the cells with BFA.

Effect of BFA on Actiuation of Adenylylcycluse and Genera- tion of AI Peptide-To explore the possibility that BFA may be directly inhibiting adenylylcyclase, we took advantage of the presence of @-adrenergic receptors on SK-N-MC and C6 cells (Table 111). Although BFA treatment blocked CT stim- ulation of cAMP in both cell lines, it did not inhibit the stimulation by isoproterenol. CT was able to activate adeny- lylcyclase in control SK-N-MC cells in a time-dependent manner but had no effect on the adenylylcyclase activity in BFA-treated cells (Fig. 4A). The A1 peptide, however, was able to activate adenylylcyclase in membranes from BFA- treated cells (Fig. 4B). Furthermore, adding BFA to the latter assay had no effect on the activation of the cyclase by the Al peptide (Fig. 4C).

Whereas small amounts of A1 were formed from '"I-CT bound to SK-N-MC cells with time, no increase in Al was detected in BFA-treated cells (Fig. 5A). Similar results were obtained with Caco-2 cells (Fig. 5B). By contrast, treatment of the latter cells with chloroquine or monensin did not interfere with Al generation.

Effect of BFA on Internalization and Degradation of CT-

Brefeldin A Block Cholera Toxin Action 12013

We next explored the possibility that BFA treatment of the cells prevented the internalization of CT-A. Control and BFA- treated cells were incubated with CT at 4 “C to allow binding, washed free of unbound CT, and shifted to 37 “C for increas- ing times in the absence and presence of BFA. The cells then were shifted to 4 “C and assayed for cell surface CT-A using antibodies raised against the Al peptide followed by lZ5I- protein A. As shown in Fig. 6, the proportion of immunoreac- tive CT-A remaining on the cell surface decreased with time and this decrease was not affected by BFA treatment of either SK-N-MC or Caco-2 cells. In separate experiments using lZ5I- CT, we confirmed that this disappearance was not due to dissociation of CT or CT-A from the cells (see below). The internalization of CT-A from the surface of Caco-2 cells was much more rapid than from the surface of SK-N-MC cells. The significance of this difference is unclear as both cell lines responded to CT with similar lag times (see Figs. 1 and 3). It is important to point out that in both cell lines, significant amounts of CT-A were internalized during the lag period and only a small amount of A, peptide has to be generated to activate adenylylcyclase (see Fig. 5, and Kassis et al. (1982)).

To assess the cellular degradation of CT, control and treated cells were incubated with lZ5I-CT at 4 “C, washed free

TABLE I11 Isoproterenol stimulation of cyclic AMP in control and brefeldin

A-treated cells Cells were incubated in 0.5 ml of Hepes-buffered medium contain-

ing 1 mM IBMX and 0.01% BSA in the presence and absence of 1 pg/ml BFA for 30 min. The cells then were stimulated with CT as described in the legends to Tables I and II; or stimulated with 1 p~ isoproterenol for the last 30 min of the incubations, and assayed for CAMP and protein as described under “Experimental Procedures.” Values are the means & S.D. of triplicate wells from one of two similar experiments.

Cyclic AMP accumulation Cells Brefeldin A

Basal ‘k’,1;”,’” Isoproterenol

pmol/mg protein SK-N-MC

+ - 9.4 f 1 3130 f 158 3130 f 38

5.2 f 1 116 f 2 3050 f 235 C, - 22.1 f 5 366 f 17 3760 f 315

+ 20.4 k 1 66 f 5 4840 f 287

so0 A CON

W n 4 200-

0 ‘ ‘ ~ n ’ B ’ ’ ~ ’ ~ *

0 10 20 30 4 0 50 60 MINUTES

of any unbound CT, and incubated in fresh medium contain- ing the different drugs for increasing times. The medium then was analyzed for trichloroacetic acid-soluble radioactivity (Fishman, 1982). Whereas both control SK-N-MC and Caco- 2 cells degraded over 30% of the lZ5I-CT initially bound by 6 h, BFA-treated cells degraded only 10% of the bound toxin compared to around 20% for cells treated with chloroquine or monensin (Table IV). CT degradation was time-dependent with no significant amounts of degradative products appear- ing in the medium until 1.5 h (Fig. 7 A ) . BFA was a more effective blocker of degradation than the other drugs at all times up to 6 h and it acted in a concentration-dependent manner as did chloroquine (Fig. 7 B ) . Only a few percent of the initially bound lZ5I-CT appeared in the medium as trichlo- roacetic acid-precipitable material and/or as trichloroacetic acid-soluble cell-associated radioactivity and these small amounts did not change with either time or treatment (data not shown). Thus, BFA (or the other drugs) did not appear to be causing a release of undegraded toxin from the cells or an accumulation of degraded toxin within the cells.

Effect of BFA on Golgi Morphology-Caco-2 cells were ex- posed to different concentrations of BFA, and then fixed and stained with Ce-NBD-ceramide which is a specific fluorescent marker for trans-Golgi cisternae (Pagan0 et al., 1989). Whereas control cells exhibited a well defined Golgi complex as evidenced by intense staining of perinuclear structures, the addition of BFA resulted in a disappearance of the Golgi complex as evidenced by diffuse staining throughout the cy- toplasm (Fig. 8). The effect was dependent on the concentra- tion of BFA and occurred with as little as 100 ng/ml. Thus, there seemed to be a close correlation with the concentrations required to inhibit the response of the cells to CT.

DISCUSSION

As was indicated in the Introduction, the remaining major aspect of the intoxification of cells by cholera toxin which is not fully understood is the cellular processing of the holotoxin to the AI peptide which occurs during the lag phase. The mechanism by which CT or CT-A is reduced to generate A1 and the site of reduction is unknown. It also is unclear how A1 gains access to GSa. To address some of these questions, we examined the effects of inhibitors of intracellular process-

CON B FA

FIG. 4. Effect of BFA on activation of adenylylcyclase by CT in intact SK-N-MC cells and membranes. A, cells grown in flasks were incubated in the absence (0) and presence (A) of 1 pg/ml BFA for 30 min at 37 ‘C, incubated in ice-cold medium f BFA with 1 nM CT for 30 min at 4 “C, and then in warm medium f BFA at 37 “C for the indicated times. After the cells were washed and lysed, membranes were prepared and assayed for adenylylcyclase activity as described under “Experimental Procedures.” B, membranes prepared from control and BFA-treated cells were incubated with AI and NAD+ and assayed for adenylylcyclase activity as described under “Experimental Procedures.” C, same as in panel B except control membranes were incubated with the indicated concentration of BFA.

12014 Brefeldin A

A

Blocks Cholera Toxin Action

B

Untreated Chloroquine Monensin Brefeldin A 060' 0' 60' 0' 60' 0 60'

I 2 3 4 5 6 7 8

0 10 20 30 40 50 60 MINUTES

FIG. 5. Effect of BFA on generation of CT A, peptide in intact cells. A, SK-N-MC cells were incubated in the absence (0) and presence (A) of 1 pg/ml BFA for 30 min a t 37 "C, incubated in ice-cold medium f BFA with 1 nM '=I-CT for 30 min at 4 "C, washed, and then incubated in warm medium f BFA a t 37 "C for the indicated times. Cells were analyzed for labeled A1 peptide by SDS-polyacrylamide gel electrophoresis and autoradiography as described under "Experimental Procedures." B, Caco-2 cells were incubated at 37 "C for 30 min in the presence of no addition (lanes 1 and 2 ) , 100 p~ chloroquine (lanes 3 and 4) , 10 p~ monensin (lanes 5 and 6), and 1 pg/ml BFA (lanes 7 and 8). The cells then were exposed to 1251-CT at 4 "C, incubated for 0 (odd lanes) or 60 min (euen lanes) a t 37 "C, and analyzed for labeled A1 peptide as described in panel A . Fresh drugs were included in all the incubations. Values are single determinations from one of two similar experiments.

a CON I A SFA

100 B

60

20 -

0- 0- 0 10 20 30 4 0 50 60

MINUTES 0 10 20 30 40 50

MINUTES

FIG. 6. Effect of BFA on internalization of CT-A by SK-N- MC ( A ) and Caco-2 ( B ) cells. Cells were incubated in the absence (0) and presence (A) of 1 pg/ml BFA for 30 min at 37 "C, incubated in ice-cold medium k BFA with 1 nM CT for 30 min at 4 "C, washed, and then incubated in warm medium f BFA at 37 "C for the indicated times. Cells were then assayed for cell surface CT-A using anti-A, antibodies followed by lZ51-protein A as described under "Experimen- tal Procedures."

ing and trafficking on the response of cultured cells to CT. We found that neither monensin nor chloroquine had any substantial effect on the ability of CT to stimulate SK-N-MC or Caco-2 cells. While these results agree with some previous studies on HeLa and rat glioma C6 cells (Pacuszka and Fishman, 1992), they are in marked contrast with studies using rat hepatocytes where both drugs are found to block the intoxification process (Houslay and Eliott, 1981; Janicot and Desbuquois, 1987; Janicot et al., 1988, 1991). In isolated hepatocytes, they inhibit the internalization of CT, the gen- eration of Al, and the activation of adenylylcyclase, and they cause an increase in the lag phase for the latter two processes. When lZ6I-CT is injected in uiuo, A, is found in endosomes. These differences may reflect differences between hepatocytes and other cells. In this regard, our studies with Caco-2 cells may be more germain to intoxification of the human small intestinal enterocyte by CT.

By contrast, BFA was a rapid and potent blocker of CT action on Caco-2 cells as well as human SK-N-MC and rat

TABLE IV Effect of drugs on degradation of cholera toxin by SK-N-MC and

Caco-2 cells Cells were incubated for 30 min at 37 "C in the presence of the

indicated drugs. The cells then were incubated with 1 nM '=I-CT for 30 min at 4 "C, washed, and incubated in fresh medium at 37 "C for 6 h. Fresh drugs were added with each change of medium. The cells and medium were assayed for trichloroacetic acid-soluble and -insol- uble radioactivity as described under "Experimental Procedures." Degradation was calculated as the ratio of trichloroacetic acid-soluble to total radioactivity X 100. SK-N-MC and Caco-2 cells initially bound 25,000 and 32,500 cpm of '''I-CT per well. Values represent the mean f range of duplicate wells.

'9-Cholera toxin degradation

SK-N-MC cells Caco-2 cells Treatment

% of total Control 30.1 f 0.34 32.0 f 0.7 Brefeldin A (1 pg/ml) 10.2 2 0.17 10.9 f 1.1 Chloroquine (0.1 mM) 19.8 f 0.51 18.6 2 0.1 Monensin (10 pM) 23.4 f 0.31 22.8 f 0.6

glioma C6 cells with an ICs0 of 30 ng/ml. When BFA was added to the cells at the same time as CT, it still inhibited toxin stimulation by over 95%. Addition of BFA during the lag phase resulted in a rapid dimunition of its inhibitory effect. Although BFA can cause the redistribution of certain Golgi-associated proteins within 30 s, redistribution of other Golgi proteins occurs more slowly and the complete morpho- logical disassembly of the Golgi apparatus may take up to 15 min (Donaldson et al., 1990; Alcalde et al., 1992). Although BFA treatment completely blocked the response of cells to CT, it had no effect on the cellular internalization of CT as measured by the disappearance of immunoreactive CT-A from the cell surface. BFA did inhibit the degradation of CT by SK-N-MC and Caco-2 cells but previous studies have shown that degradation is not required for activation of the toxin (Fishman, 1982; Kassis et al., 1982). In this regard, we were unable to detect any significant degradation until well after A1 was generated and adenylylcyclase was activated. Further-

Brefeldin A Blocks Cholera Toxin Action 12015 CON 4 BFA m CHLOROQUINE HONENSIN

0 1 2 3 4 5 6 ?W HOURS SFA (ug/ml) or CHLOROQUINE (mH)

FIG. 7. Degradation of 1261-CT by SK-N-MC cells. A , cells were incubated for 30 min at 37 "C in the presence of no addition (a), 1 pg/ml BFA (A), 0.1 mM chloroquine (W), or 10 pM monensin (+). The cells then were incubated with 1 nM '=I-CT for 30 min at 4 "C, washed, and incubated in fresh medium at 37 "C. Fresh drugs were added at each change of medium. At the indicated times, the cells and medium were assayed for degraded Iz6I-CT as described in the legend to Table IV. B, details are similar to those in panel A except the cells were treated with increasing concentrations of BFA (A) or chloroquine (W) and degradation was assayed at 6 h. Values are the ( A ) means k range of duplicate wells or ( B ) means S.D. of triplicate wells.

A B

C

i FIG. 8. Effect of BFA on morphology of Golgi apparatus of

Caco-2 cells. Cells were stained with Cs-NBD-ceramide, incubated with medium containing 0 (A), 10 ( B ) , 100 (C), or 1000 (D) ng/ml BFA for 30 min at 37 "C, and washed to remove excess label as described under "Experimental Procedures." The cells then were fixed, mounted, and visualized by fluorescence microscopy (630 X) and photomicrographs were made under the same conditions of exposure and printing.

more, 0.4 mM chloroquine was as effective as 1 pg/ml BFA at inhibiting CT degradation, yet the former had no effect on toxin action.

BFA appears to block CT action by preventing the gener- ation of A1 from CT. Adenylylcyclase in membranes from BFA-treated cells, however, was activated by A1 and NAD+. Furthermore, addition of BFA to this in vitro ADP-ribosyla- tion reaction did not prevent the activation of the cyclase. This was important to establish as Al-catalyzed ADP-ribo- sylation of G, requires a small GTP-binding protein known as ADP-ribosylation factor or ARF (reviewed in Bobak et al.

(1990)). ARF forms a complex with A, and is required for ADP-ribosylating activity. Members of the ARF family have recently been shown to be a component of the coat of Golgi- derived non-clathrin-coated vesicles and bind to Golgi mem- branes in a GTP- and BFA-dependent manner (Serafini et al., 1991; Donaldson et al., 1991). In addition, BFA recently has been found to inhibit Golgi membrane-mediated exchange of guanine nucleotides bound to ARF (Donaldson et al., 1992; Helms and Rothman, 1992).

BFA is known to cause the dissolution of the Golgi cisternae and a redistribution of Golgi components to the ER (Klausner et al., 1992). Using C6-NBD-ceramide, we confirmed that BFA had similar effects on Caco-2 cells and was effective at con- centrations required to block the intoxification by CT. Thus, the more likely explanations for this blockade are: either CT or CT-A must gain access to or pass through a functioning Golgi apparatus in order to be converted to A1; or some essential component required for CT activation must be proc- essed through the Golgi complex. At present we are unable to distinguish between these two possibilities. In agreement with the reversible effects on Golgi morphology (Lippincott- Schwartz et al., 1989), we found that upon removal of BFA, the cells recovered their sensitivity to CT. The lag phase, however, was increased which could be interpreted to support either possibility. Thus, upon removal of BFA, different Golgi membrane coat proteins and integral proteins recover their normal distribution at different rates (Donaldson et al., 1990; Robinson and Kreis, 1992; Alcalde et al., 1992).

Supporting the first possibility is the presence of the ER retention sequence KDEL (Pelhan, 1990) on the C terminus of the A2 peptide of CT (Duffy et al., 1981; Mekalanos et al., 1983). Internalized CT (or CT-A) could be directed through the Golgi to the ER by this retrograde pathway. The toxin could then undergo reduction in the ER to generate AI, which free of Az would be able to recycle back to the plasma mem- brane, possibly as a complex with ARF. One implication of this model is the requirement for recognition of the C-termi- nal KDEL sequence. Whether this recognition will necessitate the dissociation of the A subunit from the internalized holo- toxin is not known; however, recent evidence indicates that all of the KDEL sequence protrudes from the pore of the B pentamer (Sixma et dl., 1992).

We cannot rule out the second possibility that BFA-induced disruption of the Golgi apparatus and the anterograde trans- port from ER to post-Golgi compartments (see Lippincott- Schwarz et al. (1990)) interferes with the processing or distri- bution of a component required for CT activation. One poten- tial candidate could be the thio1:protein disulfide oxidoreduc- tase, which has been shown to catalyze the reduction of CT- A in vitro and to be localized to the plasma membrane, to smooth vesicles adjacent to the plasma membrane, and to the smooth ER (Moss et al., 1980). A further consideration is the recent observation that BFA forms conjugates with glutathi- one and cysteine, that these reactions are catalyzed by gluta- thione S-transferase, and that cells appear to use these reac- tions to detoxify BFA (Briining et al., 1992). In addition, BFA covalently binds to glutathione S-transferase and other en- zymes containing reactive sulfhydral groups (Briining et d., 1992). Because we found that BFA inhibited the action of CT rapidly, reversibly, and at low concentrations, it is unlikely that BFA is mediating its effects by reducing the cellular levels of reductants such as cysteine and glutathione or by covalently inactivating an enzyme.

We also have to consider the possibility that BFA blocks CT action via a non-Golgi compartment. There have been several reports that BFA induces the formation of tubular structures from endosomes, lysosomes, and the trans-Golgi network (reviewed in Klausner et al. (1992) and Pelham

12016 Brefeldin A Blocks Cholera Toxin Action

(1991)). This is particularly evident in cells which are resist- ant to the Golgi-disrupting effects of BFA. Thus, MDCK cells, when exposed to BFA, retain their Golgi stacks but exhibit more tubular endosomes (Hunziker et al., 1991; Sand- vig et al., 1991; Prydz et al., 1992). BFA inhibits basolateral to apical surface transcytosis of dimeric IgA in these polarized kidney cells (Hunziker et al., 1991). Paradoxically, BFA stim- ulates transcytosis of transferrin (Wan et al., 1992) as well as ricin and horseradish peroxidase (Prydz et al., 1992) in MDCK cells. BFA also stimulates the apical but not basolateral endocytosis of ricin by these cells (Prydz et al., 1992). This may be related to the fact that BFA actually sensitizes MDCK cells to ricin, whereas BFA protects most other cells from the plant toxin (Hudson and Grillo, 1991; Sandvig et al., 1991; Fishman and Curran, 1992). These include Caco-2 and SK- N-MC cells. Transport of ricin to the Golgi is important for its action (van Deurs et al., 1986; Hudson and Grillo, 1991) and BFA does not inhibit this process in MDCK cells (Prydz et al., 1992). Thus, in the case of ricin, BFA-mediated protec- tion against toxicity appears to relate to its effects on the Golgi and not on non-Golgi compartments. This also may be the case with CT.

Finally, it is of interest to compare the effects of these drugs on CT with their effects on two other ADP-ribosylating toxins, diphtheria toxin and Pseudomonas exotoxin, both of which inhibit protein synthesis via ADP-ribosylation of elon- gation factor 2. After binding to cell surface receptors, both toxins appear to internalize via coated pits and endosomes, undergo proteolytic processing and reduction, and release their active A fragments into the cytosol (Olnes and Sandvig, 1985; Ogata et al., 1990). In contrast to their lack of effect on CT, chloroquine and monensin protect cells from diphtheria toxin and Pseudomonas exotoxin (Leppla et al., 1980; Marnell et al., 1982; FitzGerald et aL, 1980; Sundan et al., 1984). Whereas BFA does not interfere with diphtheria toxin, it does protect cells from Pseudomonas exotoxin (Yoshida et al., 1991). We were able to confirm the latter with SK-N-MC and C6 cells (Fishman and Curran, 1992). The C terminus of Pseudomonas exotoxin has the sequence REDLK and mu- tants with the sequences RDEL and KDEL are also toxic (Ogata et al., 1990). Thus, CT and Pseudomonas exotoxin may share some common sites in the intracellular trafficking pathway, particularly those involving the Golgi and ER, and their interconnecting pathways.

The apparently conflicting effects of monensin and chlo- roquine on the different toxins may best be explained by the observations of Tran et al. (1987), who found that CT binds to noncoated invaginations of the plasma membrane and with time appears in noncoated vesicles and then in a tubulovesic- ular network. When cells are co-labeled with a*-macroglobulin which enters cells via coated pits and endocytic vesicles, both ligands eventually appear in the same tubulovesicular struc- tures. Thus, toxins such as diphtheria toxin and Pseudomonas exotoxin which enter cells through coated pits will be inhib- ited by chloroquine and monensin, whereas those such as CT which enter through noncoated pits will not. Toxins such as Pseudomonas exotoxin and CT whose processing requires an intact Golgi will be inhibited by BFA even though they enter the cell through different structures.

Note Added in Proof-We recently found that MDCK cells were resistant to the inhibitory effects of BFA on CT action. Thus, in the presence of up to 1 pg/ml BFA, CT stimulation of cyclic AMP accumulation was the same as in control MDCK cells. These results

are consistent with those reported for the action of ricin on these cells and support the involvement of the Golgi compartment in the action of both toxins.

REFERENCES Alcalde, J., Bonay, P., Roa, A., Vilaro, S., and Sandoval, I. V. (1992) J. CeU

Biol. 116.69-83 Bobak, D. A., Tsai, S.-C., Moss, J., and Vaughan, M. (1990) in ADP-ribosylating

Toxins and G Proteins: Insights into Signal Transduction (Moss, J., and Vaughan, M., eds) pp. 439-456, American Society for Microbiolow, Wash- ington, D. C. ~ ~

Wieland. F. T. (1992) .I Biol. ChPm. 267. 7726-77.18

" .

Briining, A., Ishikawa, T., Kneusel, R. E., Matern, U., Lottspeich, F., and ~~~~ ~~~~

Donaldson: J. G., Liouincott ."", - -

-Schwa&, J., Bloom, G. S., Kreis, T. E., and

Donaldson. J. G.. Kahn. R. A,. Liuuincott-Schwartz. J.. and Klausner. R. D.

. . . . . . . . . . - - . , . . - . . - - Klausner, R. D.' (199il) J. Cell Biol. 111,2295-2306

Donaldson. J. G.. (1991) sciem s 228, i iw-1199 "

I ,

Finazzi, D., and Klausner, R. D. (1992) Nature 360,350-352 Duffy, L. K., Peterson, J. W., and Kurosky, A. (1981) J. Biol. Chem. 2 5 6 ,

12252-12256 Finkelstein, R. A. (1973) Crit. Rev. Microbial. 2,553-623 Fishman, P. H. (1980) J. Membr. Biol. 54,61-72 Fishman, P. H. (1982) J. Cell. Biol. 93,860-865 Fishman, P. H. (1990) in ADP-ribosyluting Toxins and G Proteins: Insights into

Signal Transduction (Moss, J., and Vaughan, M., eds) pp. 127-140, American Society for Microbiology, Washington, D. C.

Fishman, P. H., and Curran, P. K. (1992) FEBS Lett. 314,371-374 FitzGerald, D., Morris, R. E., and Saelinger, C. B. (1980) Cell 21,867-873 Helms, J. B., and Rothman, J. E. (1992) Nature 360,352-354 Houslay M. D. and Eliott K. R. F. (1981) FEBS Lett. 128,289-292 Hudson,'T. H.,'and Gril10,'F. G. (1991) J. Biol. Chem. 266,18586-18592 Hunziker, W., Whitney, J. A., and Mellman, I. (1991) Cell 67,617-627 Janicot, M. and Desbuquois, B. (1987) Eur. J. Biochem. 163,433-442 Janicot, M.: Clot, J.-P., and Desbuquois B. (1988) Biochem. J. 2 6 3 , 735-743 Janicot, M., Fouque, F., and Desbuquoii B. (1991) J. Bwl. Chem. 266,12858-

Kassis, S., Hagmann, J., Fishman, P. H., Chang, P. P., and Moss, J. (1982) J.

Klausner, R. D., Donaldson, J. G., and Lippincott-Schwartz, J. (1992) J. Cell

Ktistakis, N. T., Roth, M. G., and Bloom, G. S. (1991) J. Cell Biol. 113,1009-

Leppla, S. H., Dorland, R. B., and Middelbrook, J. L. (1980) J. Biol. Chem.

Lippincott-Schwartz, J., Yuan, L. C., Bonifacino, J. S., and Klausner, R. D.

Lip incott-Schwartz, J. Donaldson, J. G. Schweizer, A., Berger, E. G., Hauri,

Marnell, M. H., Stookey, M., and Draper, R. K. (1982) J. Cell Biol. 93,57-62 Mekalanos, J. J., Swartz, D. J., Pearson, G. D. N., Harford, N., Groyne, F., and

Moss, J and Vaughan M. (1979) Annu. Reu. Biochem. 48,581-600 Moss. J: Stanlev. S. j.. Morin. J. E.. and Dixon. J. E. (1980) J. Biol. Chem.

12865

B~X. them. 267,12148-12152

Bwl. 116,1071-1080

1023

256,2247-2250

(1989) Cell 66,801-813

I-f P., Yuan, L. C., d d Klausner, R. D.'(1990) Cell 60,821-836

de Wilde, M. (1983) Nature 306,551-557

2 6 6 , i1085-1io87 '

c&. 2 6 5 , 2 0 6 7 a i o w

, . . .

Ogata M. Chaudhary V. K., Pastan, I., and FitzGerald, D. J. (1990) J. Biol.

Olnes, S., and Sandvi , K (1985) in Receptor-mediated Endocytosis (Pastan, I., and Willingham, d. C:, eds) pp. 195-234, Plenum Publishing Corp., New York

~~

Pacuszka T. and Fishman P. H. (1990) J. Biol. Chem. 2 6 6 , 7673-7678 Pacuszka' T.' and Fishman' P. H. (1992) Biochemistry 31,4773-4778 Pagano, R. E., Sepanski, M. A,, and Martin, 0. C. (1989) J. Cell Biol. 109,

Pelham H. R. B. (1990) Trends Biochem. Sci. 15,483-486 Pelham' H. R. B. (1991) Cell 67,449-451 Prydz, K., Hansen, S. H., Sandvig, K., and van Deurs, B. (1992) J. Cell Biol.

Robinson, M. S., and Kreis, T. E. (1992) Cell 69,129-138 Sandvig, K., Prydz, K., Hansen, S., and van Deurs, B. (1991) J. Cell Biol. 115 ,

~~

2067-2079

119,259-272

971-981 Serafini, T., Orci, L., Amherdt, M., Brunner, M., Kahn, R. A., and Rothman,

"- "_ .I. E. 11991) Cell 67. 2.19-253

~ "", - . .. - - , - - - -. . Sixma, T. K., Pronk S. E., Kalk, K. H. Wartna, E. S., van Zanten, B. A. M.,

Sixma, T. K., Pronk, S. E., Kalk, K. H., van Zanten, B. A. M., Beghuis, A. M.,

Spie el S.'(1985) Biochemistry 24,5947-5952 Sun&; A. Sandvig K. and Olsnes S. (1984) J. Cell. Physiol. 119.15-22

Tran, D., Carpentier, J.-L., Sawano, F., Gordon, P., and Orci, L. (1987) Proc. Tomasi: M.', and M<nte&cco, C. (1481) J. Biol. Chem. 266,11177-11181

Van Deurs. R.. Tonnessen. T. I.. Petersen. 0. W.. Sandvie. K.. and Olsnes. S.

Witholt, B., and Hol, W. G. J. (1991) hature 361,371-377

and Hol W. G. J. (1992) Nature 365,561-564

Natl. Acad. SCL U. S. A. 84,7957-7961 ~~~~ ~ ." (1986) J: E d BGi.-102;'37-47

Y . .

van Echten. G.. mer. H.. Stoz. H.. Takatsuki. A.. and Sandhoff. K. (1990) Eur. J . Cell Sibl. d l , 135-139 ' '

. . . . .

van Heyningen, S. (1983) C u m Top. Membr. Trans 18,445-471 Wan, J., Taub, M. E., Shah, D., and Shen, W.4 . 6992) J. Bwl. Chem. 267 ,

Wisnieski, B. J., and Bramhall, J. S. (1981) Nature 289,319-321 Yoshida. T., Chen, C.. Zhang, M., and Wu, H. C. (1991) Exp. CeU Res. 1 9 2 ,

13446-13450

P&. Natl. Acad. Sci. U. S. A.