A Dipeptide Metalloendoprotease Substrate Completely Blocks the Response of Cells in Culture to Cholera Toxin*

Prior exposure (15 min at 37 °C) of several cell types (Vero, SH-SY5Y neuroblastoma, human intestinal epithelial T84) to 3 mm N-benzoyloxycarbonyl-Gly-Phe-amide (Cbz-Gly-Phe-NH2), a competitive substrate for metalloendoproteases, completely suppressed cholera toxin (CT)-induced intracellular cAMP accumulation. The specificity of the inhibitory effect was demonstrated by the complete lack of effect of the dipeptide Cbz-Gly-Gly-NH2, an inactive analogue of Cbz-Gly-Phe-NH2. The effect was reversible and dose- (IC50 as low as 0.2 mm depending on the cell type) and time-dependent. Adding Cbz-Gly-Phe-NH2 during the lag phase caused a diminution of its inhibitory effect similar to that observed with brefeldin A (BFA). Whereas the dipeptide completely suppressed the CT-induced adenylate cyclase (AC) activity, a direct effect on AC is unlikely since the elevation of intracellular cAMP by forskolin was only slightly reduced. The A1 peptide of CT and NAD+ activated the AC to the same extent in membranes from control and Cbz-Gly-Phe-NH2-treated cells or when Cbz-Gly-Phe-NH2 was added directly to the assay. The inhibitory effects of suboptimal amounts of Cbz-Gly-Phe-NH2and BFA were not additive pointing to a similar mode of action of the two substances. However, Madin-Darby canine kidney cells of which the Golgi structure is BFA-resistant were not resistant to the inhibitory action of Cbz-Gly-Phe-NH2 on CT cytotoxicity. Several lines of evidence indicate that a perturbation of intracellular Ca2+ homeostasis by Cbz-Gly-Phe-NH2 is not responsible for the inhibitory effect of the dipeptide. The dipeptide had also no effect on the binding of 125I-CT to cells and even increased its intracellular internalization. In contrast with BFA, Cbz-Gly-Phe-NH2 did not completely suppress the formation of the catalytically active A1 fragment from bound CT. The data are compatible with a role of metalloendoprotease activity in the intracellular trafficking and processing of CT, although other mechanisms of action of Cbz-Gly-Phe-NH2cannot be excluded.

erae classical as well as El Tor biotypes, is the major causative agent of the acute diarrheal disease of humans. CT and the Escherichia coli heat-labile enterotoxin (LT), are structurally and immunologically highly homologous, belonging to the same enterotoxin family (1,2). Both are oligomeric proteins of the A-B type. CT is composed of one A or activating subunit (CT-A M r 27,400), which consists of two distinct polypeptide chains CT-A 1 (M r 22,000) and CT-A 2 (M r 5,400), linked by a single disulfide bridge and five identical B subunits (M r 11,600) arranged in a ring-like configuration (CT-B).
The subunits are arranged such that CT-A occupies the central channel of the CT-B pentamer extending well above the plane of the pentameric ring (3,4). The CT-A 2 peptide goes through the pore in the doughnut-like structure of the CT-B pentamer and protrudes on the side that binds cell surface receptors with its COOH-terminal KDEL sequence exposed. CT elicits a secretory response from intestinal epithelia by binding to the apical cell membrane through interaction between CT-B and the monosialoganglioside GM 1 followed by entry of polypeptide A 1 into the cell, where it is able to stimulate the basolateral adenylate cyclase by catalyzing the ADPribosylation of Arg-201 of the G s ␣ subunit of the stimulatory GTP-binding regulatory protein (1)(2)(3).
There is a distinct lag period between toxin binding and the activation of adenylate cyclase, during which the toxin must be internalized and processed. At the end of this lag period small amounts of CT-A 1 appear in the cells in parallel with activation of the cyclase (5). Internalization is believed to occur via a clathrin-independent mechanism (6, 7), whereby CT⅐GM 1 complexes cluster in caveolae-like detergent-insoluble subdomains (lipid rafts) of the plasma membrane (8,9). Invagination and internalization of these membrane domains result in the formation of smooth endocytotic vesicles, which fuse with endosomes, where sorting for further routing in the cell to the trans-Golgi network takes place (10 -12). Brefeldin A (BFA), a fungal metabolite that disrupts the structural and functional integrity of the Golgi apparatus (13) renders cells resistant to CT cytotoxicity and blocks intracellular formation of CT-A 1 (14 -16) suggesting the involvement of the Golgi apparatus.
A role for Golgi-ER retrograde transport in the processing of CT and LT is suggested by the observation that both CT-A and LT-A possess a carboxyl-terminal tetrapeptide sequence (KDEL in CT-A and RDEL in LT-A) that functions as an ER retrieval sequence in eukaryotic cells (17). Mutagenesis studies on CT and LT, however, indicated that the ER retrieval signal enhances the efficient delivery of the toxin but is not absolutely required for toxin action (12,18). At present, there are no experimental data to support the mechanism of membrane translocation of the active fragment of CT. It has been suggested that translocation involves the ER-associated protein degradation pathway and that the translocated active A 1 fragment escapes ubiquitin-mediated protein degradation because of the complete lack of internal lysine residues (19).
To explore further the role of retrograde Golgi-ER transport in CT action and its associated vesicular trafficking events, the effect of benzyloxycarbonyl (Cbz)-Gly-Phe-amide, a specific inhibitor of metalloendoproteases, was studied. Recently (20), it has been reported that Cbz-Gly-Phe-amide inhibits the BFAinduced retrograde transport of Golgi cisternae to the ER.

EXPERIMENTAL PROCEDURES
Materials-Highly purified CT and anticholeragenoid antibodies were obtained from List Biological Laboratories (Campbell, CA). CT was radiolabeled with 125 I using the chloramine-T method described by Cuatrecasas (21) and as modified by Janicot and Desbuquois (22). Unreacted [ 125 I]iodine was removed using gel filtration on a Sephadex G-50 minicolumn using the centrifugation procedure of Tuszynski et al. (23). Cbz-Gly-Phe-NH 2 , Cbz-L-Tyr-p-nitrophenyl ester, brefeldin A, forskolin, and 3-isobutyl-1-methylxanthine were from Sigma. Thapsigargin was from Calbiochem and Cbz-Gly-Gly-NH 2 from Bachem Bioscience Inc. Na 125 I and 125 I-protein A (30 Ci/g) were purchased from Amersham Pharmacia Biotech.
Cell Culture-Vero cells originally obtained from Flow Laboratories were cultured in Medium 199 with Earle's salts supplemented with 5% fetal calf serum. Human intestinal epithelial T84 cells (obtained from ATCC) were propagated in a 1:1 mixture of Ham's F12 medium and Dulbecco's modified Eagle's medium with 2.5 mM L-glutamine and 5% fetal bovine serum. Madin-Darby canine kidney (MDCK) cells (obtained from ATCC) were grown in Eagle's minimum essential medium with 2 mM L-glutamine and Earle's balanced salt solution adjusted to contain 1.5 g/liter sodium bicarbonate, 0.1 mM non-essential amino acids, and 1.0 mM sodium pyruvate with 10% fetal bovine serum. SH-SY5Y neuroblastoma cells (obtained from ATCC) were cultured in a 1:1 mixture of Eagle's minimum essential medium with non-essential amino acids and Ham's F12 medium supplemented with 10% fetal bovine serum. Cells were counted in a hemocytometer. Growth medium was changed twice a week, and cells were passed weekly (at confluency) using a 0.1% trypsin solution (Worthington) in Dulbecco's phosphate-buffered saline without Ca 2ϩ and Mg 2ϩ . Vital staining of the Golgi apparatus with C 6 -NBD-ceramide was done by the method of Ktistakis et al. (24) as modified by Orlandi et al. (14).
Preparation of Ca 2ϩ -depleted Vero Cells-Ca 2ϩ -depleted cells were prepared according to the method of Brostrom et al. (25). Briefly, monolayer cultures of Vero cells were washed twice with 10 ml of Earle's balanced salt solution supplemented with 1 mM EGTA and 25 mM Tris, pH 7.5. Another 10 ml of the above salt solution was added, and the cells were incubated at 37°C until they detached from the plastic surface. The cell suspension was centrifuged at 600 ϫ g for 3 min at room temperature and the supernatant fluid discarded. Cells were resuspended in the above solution and centrifuged. The pellet of the cells was resuspended in 10 ml of the buffered salt solution containing 1 mM MgCl 2 and 1 mM 3-isobutyl-1-methylxanthine and incubated at 37°C for 30 min.
Assay of Cellular Cyclic AMP Content-Monolayer cultures of cells were washed twice with Earle's balanced salt solution without Ca 2ϩ and Mg 2ϩ and treated with 0.05% (w/v) trypsin solution containing 0.02% EDTA. After 1 min the trypsin solution was removed, and 10 -15 min FIG. 1. Effect of Cbz-Gly-Phe-NH 2 on the ability of CT to raise the concentration of cAMP in BFA-treated Vero cells. Vero cells in suspension were incubated in the absence and presence of the indicated amounts of Cbz-Gly-Phe-NH 2 for 30 min at 37°C. Brefeldin A (1 g/ml) was added, and the cells were incubated for an additional 30 min. The temperature was lowered to 4°C, and cells were further incubated for 1 h in the presence of CT (1 g/ml). After removal of unbound ligand the temperature of the medium (supplemented with the agents as indicated) was shifted to 37°C, and the cells were finally incubated for another 90 min and the intracellular concentration of cAMP measured. Values are the means Ϯ S.D. of triplicate assays from one of three similar experiments. Error bars are indicated when they were more than 5% of the mean. Addition of the different agents in the same order and at the same concentrations had no effect on basal cAMP accumulation. later cells were suspended in culture medium. Cells were collected by centrifugation at 750 rpm for 5 min and resuspended in 1 ml of serumfree medium containing 25 mM Hepes, 0.01% (w/v) bovine serum albumin, and 1 mM 3-isobutyl-1-methylxanthine. 1-ml aliquots of cell suspension (10 5 cells/ml) were divided over Eppendorf tubes and cooled to 4°C. Following addition of the indicated amounts of CT, suspensions were incubated for 1 h at 4°C. After washing the cells the temperature was shifted to 37°C, and cells were further incubated for the indicated times. Afterward, cell suspensions were put on ice, centrifuged for 6 min at 1000 rpm, and resuspended in 0.1 ml of 0.5 M sodium acetate buffer (pH 6.2). The suspensions were boiled for 10 min and then sonicated for 30 s. After centrifugation for 10 min at 10,000 rpm, 20 l of cell extract was taken for cyclic AMP (cAMP) assay. cAMP was assayed using a cAMP assay kit (Amersham Pharmacia Biotech) based on a competitive protein-binding method. Results for cAMP represent the mean of values from duplicate samples each assayed in triplicate.
Activation of Adenylate Cyclase-For the determination of adenylate cyclase, Vero cells (10 5 cells) were incubated in 1 ml of serum-free medium 199 buffered with 25 mM Hepes at 37°C with and without 3 mM Cbz-Gly-Phe-NH 2 for 15 min. The medium was replaced with icecold medium containing CT (2 g/ml), 0.01% BSA, and Ϯ3 mM Cbz-Gly-Phe-NH 2 , and the cells were further incubated at 4°C for 30 min. The cells were subsequently incubated at 37°C for different times by replacing the medium with warm Medium 199/Hepes Ϯ3 mM Cbz-Gly-Phe-NH 2 . The cells were washed with ice-cold Earle's balanced salt solution, lysed by subjecting to a freezing (Ϫ80°C) thawing cycle in 1 mM Tris-HCl, 2 mM EDTA, pH 7.4. Crude membranes, obtained by centrifugation of the lysate for 10 min at 16,000 ϫ g were washed and incubated for 10 min at 37°C in 100 l containing 25 mM Tris acetate buffer, pH 7.4, 5 mM creatine phosphate, 50 IU/ml creatine phosphokinase, 5 mM magnesium acetate, 0.5 mM ATP, 1 mM dithiothreitol, 0.1 mg/ml BSA, 0.01 mM GTP, and 1 mM 3-isobutyl-1-methylxanthine. The reaction was stopped by boiling the solution for 2 min. The tubes were cooled in ice, centrifuged (10,000 ϫ g), and triplicate 20-l samples removed for assay of cyclic AMP.
Membranes prepared (26) from control and Cbz-Gly-Phe-NH 2treated cells were also incubated with 5 g/ml activated CT (incubated 10 min at 37°C with 20 mM dithiothreitol), 1 mM NAD ϩ , 100 M GTP, with and without 3 mM Cbz-Gly-Phe-NH 2 for 10 min at 30°C and then assayed for adenylate cyclase activity as described above.
Assay of Surface Toxin-Vero cells were grown attached to 35-mm wells of multicluster dishes in Medium 199 with Earle's salts supplemented with 5% fetal calf serum. The cells were washed once with phosphate-buffered saline and incubated in serum-free medium buffered with 25 mM Hepes containing 0.01% BSA and CT (0.5 g/ml) for 30 min at 4°C. The cells were then washed three times with ice-cold phosphate-buffered saline, incubated in fresh serum-free medium in the presence and absence of dipeptide for the indicated times at 37°C,

FIG. 3. Time dependence of the inhibitory effect of Cbz-Gly-Phe-NH 2 and BFA on the CT-induced accumulation of cAMP in Vero cells.
Vero cells were exposed to 3 mM Cbz-Gly-Phe-NH 2 (E) and to 1 g/ml BFA (‚) for the indicated times before or after the addition of 1 g/ml CT and incubated for 90 min at 37°C after adding the toxin. Data points are the means of triplicate assays from one of three similar experiments. S.D. values were always lower than 7% of the mean.

FIG. 4. Reversibility of Cbz-Gly-Phe-NH 2 and BFA-mediated inhibition of CT action. Vero cells (A) and
SH-SY5Y neuroblastoma cells (C) were incubated for 15 min at 37°C in the absence (E, Ⅺ) and presence (q, Ⅺ) of Cbz-Gly-Phe-NH 2 (3 mM), washed three times, and incubated with CT (1 g/ml) with (Ⅺ) and without (E, q) Cbz-Gly-Phe-NH 2 (3 mM) for the indicated times. Vero cells (B) and SH-SY5Y neuroblastoma cells (D) were incubated for 30 min at 37°C in the absence (‚, f) and presence (OE, f) of BFA (1 g/ml), washed three times, and incubated with CT (1 g/ml) with (f) and without (‚, OE) BFA (1 g/ml) for the indicated times. Data points are the means of triplicate assays from one of three similar experiments. S.D. values were always lower than 7% of the mean. and assayed for surface CT according to the procedure described by Fishman (27) except that anti-choleragenoid antibodies were used.
Generation of CT-A 1 and Analysis by SDS Polyacrylamide Gel Electrophoresis-Vero cells were grown to confluency in small (inner diameter ϭ 3 cm) Petri dishes (250 ϫ 10 6 cells) and maintained in culture for at least 1 week before use. Cells were washed once with serum-free medium buffered with 25 mM Hepes and incubated with and without Cbz-Gly-Phe-NH 2 , Cbz-Gly-Gly-NH 2 (3 mM), or BFA (1 g/ml) for 30 min at 37°C. The medium was replaced with ice-cold serum-free medium/Hepes (total volume ϭ 3 ml) containing 125 I-CT (10 6 cpm/ml; ϳ 1 nM CT) and 0.01% BSA, and the cells were further incubated at 4°C for 30 min. The cells were then washed with ice-cold serum-free medium/ Hepes and incubated at 37°C for different times by replacing the medium with warm serum-free medium/Hepes. With each medium change, Cbz-Gly-Phe-NH 2 , Cbz-Gly-Gly-NH 2 , or BFA was added as required. The cell incubations were stopped by adding 1 ml of ice-cold N-ethylmaleimide (1 mM) in phosphate-buffered saline to prevent any further reduction of CT (5), scraped in phosphate-buffered saline, and pelleted by low speed centrifugation. Generation of CT-A 1 was determined by SDS polyacrylamide gel electrophoresis in 16% Tris glycine gels. Protein bands corresponding to CT-A 1 and CT-A were cut from the gels and analyzed for radioactivity. Fig. 1 prior treatment of Vero cells with BFA (2 g/ml) strongly suppressed the CT-induced cAMP accumulation which is in agreement with previous studies on other cell types (14 -16). However, pretreatment of cells with Cbz-Gly-Phe-NH 2 only slightly prevented the BFA-induced blockage of CT action, whereas the dipeptide on its own strongly suppressed CT action. Because the role of metalloendoproteases in CT action has not been addressed before, we studied the effect of metalloendoprotease substrates on CT action in somewhat more detail.

Effect of Cbz-Gly-Phe-NH 2 on the BFA-induced Inhibition of CT Cytotoxicity-As shown in
Effect of Cbz-Gly-Phe-NH 2 on the Response of Intact Cells to CT-As illustrated in Fig. 2 prior exposure (15 min at 37°C) of Vero, SH-SY5Y neuroblastoma, and human intestinal epithelial T84 cells to Cbz-Gly-Phe-NH 2 completely or almost completely suppressed the CT-induced cAMP accumulation in a dose-dependent way. Depending on the cell type IC 50 values varied from 0.2 to 0.5 mM. The extent of inhibition was similar to that observed with BFA, except that with T84 cells BFA only partially inhibited CT action. The specificity of the inhibitor effect was demonstrated by the complete lack of an effect of the dipeptide Cbz-Gly-Gly-NH 2 , an inactive analogue of Cbz-Gly-Phe-NH 2 which is not a substrate for metalloendoproteases (28).
Time Dependence of the Inhibitory Effect of Cbz-Gly-Phe-NH 2 -As depicted in Fig. 3 the inhibitory effect of Cbz-Gly-Phe-NH 2 was time-dependent. When Cbz-Gly-Phe-NH 2 was added at the same time as CT 80% inhibition was observed, whereas when Vero cells were exposed to 3 mM Cbz-Gly-Phe-NH 2 15 min before the addition of CT the activation was virtually abolished. Adding Cbz-Gly-Phe-NH 2 during the lag phase caused a diminution of the inhibitory effect.
In parallel experiments we also determined the time dependence of the inhibitory effect of BFA on the CT-induced cAMP accumulation in Vero cells. In agreement with a previous study with SK-N-MC cells (14), the inhibitory effect was time-dependent and adding BFA during the lag phase also reduced its effectiveness as an inhibitor. However, as shown in Fig. 3 there is a marked difference in the time dependence of the inhibitory action of Cbz-Gly-Phe-NH 2 and BFA. Adding BFA 15 min after the addition of CT did not reduce but rather increased the response of Vero cells to CT. With the dipeptide a similar phenomenon was only observed when it was added 45 min after the addition of CT.
Reversibility of Cbz-Gly-Phe-NH 2 Inhibition-The inhibitory effect of Cbz-Gly-Phe-NH 2 on Vero cells was reversible (Fig. 4). Exposure of cells to Cbz-Gly-Phe-NH 2 followed by its removal did not affect the extent of CT-induced cAMP accumulation and only slightly altered the time course of CT stimulation (Fig. 4) by causing a small increase in the lag phase. In contrast, no shift in time course was observed following a similar treatment of Vero cells with BFA (2 g/ml). With SH-SY5Y cells the inhibitory effects of Cbz-Gly-Phe-NH 2 (3 mM) and BFA (2 g/ ml) were less reversible. In both cases (Fig. 4) the extent of the CT-induced cAMP accumulation was somewhat reduced and the shift in the time courses was more pronounced.
Similar results were obtained when after preincubation of the cells with either Cbz-Gly-Phe-NH 2 (3 mM) or BFA (2 g/ml) for 15 and 30 min, respectively, the cells were cooled to 4°C, incubated with CT (2 g/ml) for 30 min at 30 min at 4°C, washed three times, and finally further incubated at 37°C for the indicated times.
Additivity of Inhibitory Effects of Cbz-Gly-Phe-NH 2 and BFA-Because of some similarities in the inhibitory action of Cbz-Gly-Phe-NH 2 and BFA we explored the effects of the combined addition of suboptimal amounts of these agents on the CT-induced cAMP accumulation in Vero cells.
As illustrated in Fig. 5 the inhibitory effects of suboptimal amounts of Cbz-Gly-Phe-NH 2 and BFA were not additive, indicating a similar mode of action. However, Madin-Darby canine kidney (MDCK) cells of which the Golgi structure is BFAresistant (29), and as a consequence are resistant to the inhibitory effect of BFA on CT action, were not resistant to the inhibitory effect of Cbz-Gly-Phe-NH 2 . With these cells prior exposure to Cbz-Gly-Phe-NH 2 (3 mM) also completely suppressed (IC 50 ϭ 0.5 mM) CT action (data not shown).
Effect of Cbz-Gly-Phe-NH 2 on Adenylate Cyclase Activity-To ascertain whether Cbz-Gly-Phe-NH 2 had any direct effect on AC activity, we determined the effect of the dipeptide on the forskolin-induced intracellular accumulation of cAMP in Vero cells. As illustrated in Table I, the elevation of intracellular cAMP by forskolin was only slightly suppressed by the dipeptide indicating no direct effect on AC activity.
CT was able to activate AC in control Vero cells in a timedependent manner, resulting in a 3-fold increase after 60 min of incubation but had almost no effect on AC activity in cells exposed to Cbz-Gly-Phe-NH 2 (data not shown). The A 1 peptide, however, was able to activate AC in membranes from Cbz-Gly-Phe-NH 2 -treated Vero cells (Fig. 6). Furthermore, adding Cbz-Gly-Phe-NH 2 to the latter assay had only a minor inhibitory effect on the activation of AC by the A 1 peptide.

Is the Inhibitory Effect of Cbz-Gly-Phe-NH 2 -mediated by a Perturbation of Ca 2ϩ
Homeostasis?-It has been previously reported (30) that Cbz-Gly-Phe-NH 2 and certain related analogues perturb intracellular Ca 2ϩ homeostasis by the mobilization of intracellular Ca 2ϩ from both high (ER) and low affinity (mitochondria) compartments. To explore this possibility we studied the effect of agents known to be able to mobilize Ca 2ϩ from intracellular stores.
We first determined the effect of Cbz-L-Tyr-p-nitrophenyl ester, a more hydrophobic tyrosine analogue possessing three aromatic rings, which has been shown to be 80-fold more potent than Cbz-Gly-Phe-NH 2 in inhibiting Ca 2ϩ accumulation by the ER and which is not recognized as a metalloendoprotease inhibitor (30).
Addition of Cbz-L-Tyr-p-nitrophenyl ester at concentrations up to 50 M (at higher concentrations the agent became insoluble) caused only a minor reduction (Ͻ15%) of the CT and forskolin (1 mM)-induced cAMP accumulation in Vero cells.
We subsequently studied the effect of thapsigargin, a tumorpromoting sesquiterpene lactone, which selectively inhibits the Ca 2ϩ -ATPase responsible for Ca 2ϩ accumulation by the ER and causes a loss of Ca 2ϩ from the lumen of this compartment (31). Since under these conditions the Ca 2ϩ -transporting systems especially the Ca 2ϩ pumps in the plasma membrane remain functional, it is believed that Ca 2ϩ levels in the ER and cytosol equilibrate at a low level (30). As shown in Fig. 7 prior treatment of Vero cells with thapsigargin did not reduce but rather increased the CT-induced accumulation of cAMP in a dose-dependent way. At a concentration of 0.5 M the CT-induced increase in cAMP level was enhanced by 30%. As also illustrated in Fig. 7 pretreatment of Vero cells with thapsigargin

FIG. 6. Effect of Cbz-Gly-Phe-NH 2 on the CT-A 1 -induced activation of adenylate cyclase in membranes of Vero cells. A, membranes prepared from control (open bars)
and Cbz-Gly-Phe-NH 2 (3 mM)-treated (hatched bars) cells were incubated with and without CT-A 1 , NAD ϩ , and GTP and assayed for adenylate cyclase activity as described under "Experimental Procedures." B, membranes from control cells were incubated with and without CT-A 1 , NAD ϩ , GTP, and the indicated amounts of Cbz-Gly-Phe-NH 2 and assayed for adenylate cyclase. Values are the means Ϯ S.D. of triplicate assays from one of three similar experiments. Error bars are indicated when they were more than 5% of the mean.

FIG. 7. Effect of thapsigargin on the ability of CT to raise the concentration of cAMP in Cbz-Gly-Phe-NH 2 -treated Vero cells.
Vero cells were incubated in the presence and absence of thapsigargin (0.5 M) for 30 min at 37°C. Some of the cells were then incubated with Cbz-Gly-Phe-NH 2 (3 mM) for 30 min before CT (1 g/ml) was added as indicated. After another 90 min of incubation the intracellular concentration of cAMP was measured. Values are the means Ϯ S.D. of triplicate assays from one of three similar experiments. Error bars are indicated when they were more than 5% of the mean. Addition of the different agents in the same order and at the same concentrations had no effect on the basal cAMP accumulation.
(0.5 M) only partially prevented the BFA-induced blockage of CT action. In a third experiment we studied the inhibitory effect of Cbz-Gly-Phe-NH 2 on the CT-induced increase in cAMP level in Ca 2ϩ -depleted Vero cells as a function of extracellular Ca 2ϩ concentration. In this experiment Ca 2ϩ -depleted Vero cells, prepared by EGTA treatment as described under "Experimental Procedures," were equilibrated for 30 min in Earle's balanced salt solution containing different concentrations of Ca 2ϩ . As shown in Fig. 8 EGTA treatment makes the Vero cells somewhat less sensitive to the action of Cbz-Gly-Phe-NH 2 , but addition of increasing concentrations of extracellular Ca 2ϩ did not affect its inhibitory action.
Effect of Cbz-Gly-Phe-NH 2 on the Binding and Internalization of CT-Preincubation of Vero cells with Cbz-Gly-Phe-NH 2 (3 mM) for 15 min at 37°C did not affect its ability to bind 125 I CT (Table II). Neither the binding capacity nor affinity appeared to be affected (data not shown). We next explored the possibility that Cbz-Gly-Phe-NH 2 treatment of Vero cells prevented the internalization of CT. To this end control and Cbz-Gly-Phe-NH 2 -treated cells were incubated with CT at 4°C to allow binding, washed free of unbound CT, and shifted to 37°C for increasing times in the absence or presence of the dipeptide. The cells were cooled to 4°C and assayed for cell surface CT using anticholeragenoid antibodies followed by 125 I-protein A treatment. As shown in Fig. 9, the proportion of immunoreactive CT-B remaining on the cell surface decreased with time, and this decrease was not reduced but rather enhanced by Cbz-Gly-Phe-NH 2 treatment. Since Cbz-Gly-Phe-NH 2 did not affect the binding of 125 I-CT to Vero cells it is unlikely that this increased disappearance of CT was due to an enhanced dissociation of CT from cells.
Effect of Cbz-Gly-Phe-NH 2 on the Generation of the Catalytically Active CT-A 1 Fragment-When Vero cells incubated with 125 I-CT and washed were shifted from 4 to 37°C, treated with N-ethylmaleimide (to reduce nonspecific reduction of CT-A) for different times, small amounts of CT-A 1 were generated after a delay of about 20 min (Fig. 10).
Under the conditions used to lyse and dissolve the cells in SDS sample buffer, the labeled toxin was effectively dissociated in its subunits. The initial fraction (at time zero) of cell-associated CT-A converted to CT-A 1 varied between 0.7 and 1.2% and, even in the presence of N-ethylmaleimide, is probably due to nonspecific reduction of CT-A (5). These values were therefore subtracted from the increasing amounts of CT-A 1 gener-ated with time. The proportion of CT-A converted to CT-A 1 after 1 h at 37°C attained a value of 5%. Whereas preincubation of Vero cells with BFA (1 g/ml) for 30 min at 37°C completely suppressed the generation of CT-A 1 from bound toxin with time, a similar treatment with Cbz-Gly-Phe-NH 2 (3 mM) or Cbz-Gly-Gly-NH 2 (3 mM), respectively, strongly suppressed but did not completely block or not affect the formation of CT-A 1 (Fig. 10). DISCUSSION Prior incubation of Vero cells with Cbz-Gly-Phe-NH 2 (3 mM) only slightly prevented the BFA-induced block of CT action. More important, the dipeptide itself almost completely suppressed the CT-induced cAMP accumulation in all cell types used in this study. The inhibitory effect of the dipeptide was dose-dependent, and IC 50 values varied between 0.2 and 0.5 mM correlating with affinities of metalloendoproteases for this compound (32). The specificity of the inhibitory effect was demonstrated by the complete lack of an effect of the dipeptide Cbz-Gly-Gly-NH 2 , an inactive analogue of Cbz-Gly-Phe-NH 2 , which has been often employed to identify nonspecific or toxic effects of Cbz-modified dipeptide amides on intact cells (28).
Upon removal of Cbz-Gly-Phe-NH 2 , the cells recovered their sensitivity to CT as would be expected for a substrate acting to inhibit competitively a metalloendoprotease.
As previously shown for BFA (14), the inhibitory effect of Cbz-Gly-Phe-NH 2 is time-dependent. Addition of Cbz-Gly-Phe-NH 2 during the lag phase caused a rapid diminution of its inhibitory effect. In contrast with BFA, addition of the dipeptide after the lag phase still suppressed CT action. This could indicate that the dipeptide affects CT trafficking or processing beyond the BFA-sensitive step. It is of interest to note that  addition of BFA as well as Cbz-Gly-Phe-NH 2 , respectively, 15 and 40 min after the addition of CT actually enhanced the stimulation of cAMP accumulation in Vero cells. This has not been reported before and could indicate that once the toxin has reached a certain intracellular compartment (cis-Golgi cisternae) the BFA-induced retrograde transport of Golgi cisternae may facilitate further intracellular routing of the toxin. In the case of Cbz-Gly-Phe-NH 2 this effect is less clear since, in contrast to BFA, metalloendoprotease substrates have not been reported to affect the structural integrity of intracellular or-ganelles. On the contrary, millimolar concentrations of Cbz-Gly-Phe-NH 2 are able to prevent the BFA-induced disassembly of the Golgi apparatus (20). In addition, pretreatment of Vero cells with Cbz-Gly-Phe-NH 2 and staining of fixed cells with C 6 -NBD-ceramide, a specific fluorescent marker for Golgi cisternae (33), did not affect the staining pattern of the perinuclear structures (data not shown).
Although Cbz-Gly-Phe-NH 2 treatment completely blocked the response of Vero cells to CT, it did not affect CT binding and did not completely suppress the generation of CT-A 1 from CT. However, the total amount of reduced toxin may not reflect toxin that can be translocated to the cytosol; for instance in the presence of Cbz-Gly-Phe-NH 2 , part of the reduced toxin may be trapped in a compartment where translocation to the cytosol is less efficient.
At present, one can only speculate on the mechanism behind the Cbz-Gly-Phe-NH 2 -induced inhibition of CT action. Since Cbz-Gly-Phe-NH 2 strongly inhibits the formation of CT-A 1 , it may act by preventing the delivery of toxin into the ER of sensitive cells by a mechanism that differs from that of BFA or, as indicated by the different "apparent" time courses of BFA and Cbz-Gly-Phe-NH 2 action, by interfering with a further step in CT trafficking and processing. Taken together, our data are compatible with the involvement of metalloendoproteases in CT action. However, because of other cellular effects of metalloendoprotease substrates (see further), one should be cautious in drawing conclusions.
Apart from secreted enzymes, metalloendoproteases have been found in the cytoplasm, plasma membranes of the brush border in the gut, mitochondria, and the rough ER (32).
In this study we used commercial (nicked) CT; therefore, the inhibitory effect of Cbz-Gly-Phe-NH 2 cannot be ascribed to an inhibition of a cell-associated protease that is able to cleave the peptide loop connecting the A 1 and A 2 polypeptide chains of CT-A at residue Arg-192, a process important for the generation of a fully active toxin (34,35). Proteolytic processing of CT by a metalloendoprotease, on the other hand, may offer an explanation for the observation that a serine protease-resistant CT mutant (CTR 192H) where Arg-192 has been replaced by a His residue, inactivating the nicking site connecting the A 1 and A 2 polypeptides of CT-A, was still able to elicit a secretory response in polarized T84 cells (35). We assume that the holotoxin-GM 1 complex is transported to the ER, which would fit with the marked stability of this complex (36, 37). Proteolytic FIG. 9. Effect of Cbz-Gly-Phe-NH 2 on the internalization of CT by Vero cells. Vero cells were incubated in the absence (E) and presence of 1 (‚) and 3 mM (OE) Cbz-Gly-Phe-NH 2 with 0.5 g/ml CT/ml for 30 min at 4°C, washed, and then incubated in warm medium Ϯ Cbz-Gly-Phe-NH 2 at 37°C for the indicated times. Cells were subsequently assayed for cell surface CT using anti-choleragenoid antibodies followed by 125 I-protein A. Data points are the means of triplicate assays from one of three similar experiments. S.D. values were always lower than 7% of the mean.
FIG. 10. Effect of Cbz-Gly-Phe-NH 2 , Cbz-Gly-Gly-NH 2 , and BFA on the generation of CT-A 1 in intact Vero cells. Vero cells were preincubated with (q) and without (E) Cbz-Gly-Phe-NH 2 , with (OE) Cbz-Gly-Gly-NH 2 (3 mM) or with (‚) BFA (1 g/ml) at 37°C for 30 min. The medium was replaced with ice-cold serum-free medium/Hepes containing 125 I-CT and 0.01% BSA, and cells were further incubated at 4°C for 30 min. The cells were then washed and incubated at 37°C for different times by replacing the medium with warm serum-free medium/Hepes. With each medium change, Cbz-Gly-Phe-NH 2 , Cbz-Gly-Gly-NH 2 , or BFA was added as required. Cells were analyzed for the formation of CT-A 1 by SDS polyacrylamide gel electrophoresis as described under "Experimental Procedures." Data points are the means of triplicate assays from one of three similar experiments. S.D. values were always lower than 7% of the mean. cleavage of CT-A, in combination with the reduction of the disulfide bridge linking CT-A 1 , might assist in the release of an active CT-A fragment. The A 2 polypeptide chain, which consists of a nearly continuous ␣-helix going through the central pore of the CT-B pentamer, contains peptide bonds that are potential cleavage sites not shielded by the CT-B pentamer. For instance, the long amino-terminal A 2 helix (residues 196 -228) which lies in a shallow groove that extends from the corner of the CT-A 1 /CT-B pentamer interface contains a -Ser-Phe-sequence at positions 205-206 (4), which corresponds with the reported specificity of metalloendoproteases (32). Typically, metalloendoproteases hydrolyze peptide bonds with an uncharged aromatic amino acid or a large aliphatic amino acid on the amino terminus and a serine or threonine on the carboxylterminal side of the hydrolyzed peptide bond. However, other potential peptide cleavage sites are present. Proteolytic cleavage of the A 1 polypeptide chain is not likely since an intact A 1 fragment is formed at the end of the lag period.
It has been reported (30,32,39) that millimolar concentrations of Cbz-Gly-Phe-NH 2 inhibit several important cellular processes such as the synthesis, processing, and secretion of proteins, whereas the synthesis of glucose-regulated stress protein 78 is induced. Mobilization of ER sequestered Ca 2ϩ is thought to be responsible for these effects (30,32,39). A perturbation of intracellular Ca 2ϩ homeostasis is, however, unlikely to be responsible for the inhibitory effect of the dipeptide as indicated by the effect of other Ca 2ϩ -perturbing agents such as Cbz-L-Tyr-p-nitrophenyl ester and thapsigargin. Furthermore, since thapsigargin also inhibits the synthesis, processing, and secretion of proteins (40), whereas it rather stimulates CT action in agreement with a previous study (10), it is reasonable to assume that perturbation of these processes by Cbz-Gly-Phe-amide is not responsible for the inhibitory effect on CT cytotoxicity.
Evidence has been presented (30) that the release of Ca 2ϩ from intracellular organelles by Cbz-Gly-Phe-NH 2 is not the result of a direct inhibition of metalloendoprotease activity but is rather due to an interaction with an ion pore that is not the inositol 1,4,5-trisphosphate or ryanodine receptor but is probably the translocon (Sec 61), the channel through which nascent proteins destined for modification by ER enzymes are inserted cotranslationally (41). Recent studies on the suppression of major histocompatibility complex class I expression by viruses (38) indicated that retrotranslocation of misfolded proteins across the ER bilayer may involve the same channel which further led to the idea that AB toxins such as CT may use this channel to enter the cytosol (19). Therefore, assuming that translocation of the A 1 fragment of CT to the cytosol involves Sec 61, one should also consider the possibility that Cbz-Gly-Phe-NH 2 blocks CT action by interfering with this process.