(Received for publication, June 11, 1996, and in revised form, October 18, 1996)
From the Membrane Biochemistry Section, Laboratory of Molecular and Cellular Neurobiology, NINDS, National Institutes of Health, Bethesda, Maryland 20892-4440
A key step in the action of cholera toxin (CT) is
the reduction of its A subunit to the A1 peptide. The
latter is an ADP-ribosyltransferase, which activates the -subunit of
the stimulatory G protein of adenylyl cyclase. In this study, the
enzymatic reduction of membrane-bound CT in CaCo-2 human intestinal
epithelial cells was characterized. Whereas diphtheria toxin was found
to be reduced by a cell surface population of protein-disulfide
isomerase (PDI) and its cytotoxicity was inhibited by
p-chloromercuribenzenesulfonic acid, bacitracin, or
anti-PDI antibodies, these inhibitors had no effect on CT reduction or
activity in intact cells. In contrast, the reduction of CT in
vitro by either postnuclear supernatants (PNS) or microsomal membranes in the presence of Triton X-100 was significantly inhibited by p-chloromercuribenzenesulfonic acid and bacitracin.
Anti-PDI monoclonal antibodies likewise inhibited the in
vitro reduction of CT and also were effective in depleting
reductase activity from PNS. Since inhibition and depletion were not
observed in the absence of detergent, these results suggested that the
reductase activity was a soluble component localized to the lumen of
microsomal vesicles and correlated with the presence of
protein-disulfide isomerase. This was further confirmed by showing a
corresponding depletion of reductase activity and PDI in alkali-treated
microsomes. This activity was restored when purified bovine PDI was
added back to alkali-treated microsomes in a redox buffer that
reflected conditions found in the lumen of the endoplasmic reticulum
(ER). When the CT-related reductase activity was assayed in subcellular fractions of PNS-derived membranes isolated on a 9-30% Iodixanol gradient, the activity, as measured by CT-A1 peptide
formation localized to those fractions containing PDI. Likewise
CT-A1 peptide formed in intact cells co-localized to those
membrane fractions containing the majority of cellular PDI.
Furthermore, the banding density corresponded to a region of the
gradient containing ER-derived membranes. These results indicated that
CT was a substrate for PDI-catalyzed reduction in intact cells and
supported the hypothesis that CT reduction and activation occurs in the
ER.
Cholera toxin (CT)1 consists of a
pentameric B subunit that binds to specific cell surface receptors
identified as ganglioside GM1 and an A subunit that
activates adenylyl cyclase (see Ref. 1 for review). The A subunit
consists of A1 and A2 peptides linked by a
disulfide bond. The primary function of the A2 peptide is
to noncovalently attach the A subunit to the B pentamer (2-4). Since
its C terminus extends through and beyond the pore of the B pentamer
(2) and contains the endoplasmic reticulum (ER) retention signal, KDEL,
the A2 peptide may also function as a signal for the
intracellular transport of the toxin. The A1 peptide is an
ADP-ribosyltransferase that catalyzes the covalent modification of the
-subunit of the stimulatory G protein (Gs) and results in the persistent activation of adenylyl cyclase (for review, see Ref.
5). In target cells, the cytotoxic effect of CT is dependent on the
reduction of CT-A to release the A1 peptide (6, 7). There
is a distinct lag period between toxin binding and the activation of
adenylyl cyclase, during which the toxin must be internalized and
processed. At the end of this lag period, small amounts of
A1 peptide appear in the cells in parallel with the
activation of the cyclase. The mechanism of reduction and the
intracellular location at which it occurs have yet to be
identified.
Bound holotoxin is thought to undergo endocytosis through noncoated invaginations of the plasma membrane and appear first in noncoated vesicles and then in a tubulovesicular compartment (8). There is no indication that CT readily dissociates during this transit. Some insight into the intracellular pathway required for toxin activation may be provided by the recently described effects of brefeldin A (BFA) on CT action. The ability of BFA to disrupt the morphology of the Golgi apparatus and interfere with vesicular routing is well documented (9). In the human intestinal epithelial cell line, CaCo-2, as well as two different neurotumor cell lines, BFA completely inhibits the action of CT (10). Specifically, BFA prevents the reduction of CT to CT-A1 without any observable effects on toxin internalization. Thus, an intact, functional Golgi apparatus is essential for the intracellular activation of CT. Similar results were later reported by investigators using other cell lines (11-13). Several possible explanations for the profound effects of BFA on CT action were proposed (10): a direct inhibition of an unidentified reductase; the inability of internalized CT to be transported via a retrograde pathway to the site of reduction; or the relocation of the reductase to a new cellular compartment not accessible to the internalized toxin molecule. The identification and characterization of such a reductase may provide additional information about the intracellular processes involved in CT activation and those sites through which the toxin must pass.
Of the enzymes that catalyze thiol-disulfide interchange reactions,
protein-disulfide isomerase (PDI; EC 5.3.4.1) is the best known. Found
predominantly as a soluble protein within the lumen of the ER, the
chief function of PDI is considered to be in the proper folding and
processing of membrane and secretory proteins (14). PDI is also thought
to act as a molecular chaparone, since it is able to interact and bind
to a broad variety of peptides with no discernible requirement for
sequence composition (15). It has also been identified as a thyroid
hormone-binding protein and as the -subunit of prolylhydroxylase,
although its capacity in these regards is not fully understood (15,
16). This 57-kDa enzyme is present in extremely high concentrations
(for review, see Ref. 15), and its distribution within the ER is
thought to be maintained in part by its carboxyl-terminal KDEL sequence (14), which prevents it from being secreted. Its high level of
expression may influence its presence throughout the secretory pathway
and the plasma membrane. PDI has been described in association with the
Golgi apparatus (17), the trans-Golgi network (18), and the
plasma membrane of mammalian cells (19, 20).
The plasma membrane-associated population of PDI has recently been shown to facilitate the reduction and activation of diphtheria toxin (DT), another ADP-ribosylating toxin. As with CT, the activation of DT requires the cleavage of an interchain disulfide bond to exert its cytotoxic effects. Treatment of intact Chinese hamster ovary (CHO) cells with anti-PDI monoclonal antibodies, membrane-impermeant sulfhydryl blockers, or bacitracin prior to toxin exposure inhibits DT cytotoxicity (21). In contrast, neither the cellular reductase responsible for the activation of CT nor the site at which the reduction occurs in target cells has been fully established. While it had been demonstrated that purified bovine PDI increased the rate of CT activation in solution in the presence of dithiothreitol or glutathione (22), there has been no direct evidence to suggest that this enzyme is responsible for the cellular activation of CT. This study examines the characteristics of membrane-bound CT reduction and the cellular site of toxin reduction and explores the potential role of the intracellular population of PDI in this process.
CT, CT-A, and DT were obtained from List
Biological Laboratories (Campbell, CA). Sigma was the
source of 3-isobutyl-1-methylxanthine, para-chloromercuribenzenesulfonic acid (p-CMBS),
N-ethylmaleimide (NEM), and monoclonal antibodies against
-COP and the 58-kDa Golgi protein. Anti-KDEL antibodies were
purchased from Affinity BioReagents, Inc. (Golden, CO). Bacitracin and
forskolin were purchased from Calbiochem; Immobilon-P membranes were
from Millipore; protein A-Sepharose and the Mab Trap kit were from
Pharmacia Biotech Inc.; and purified bovine liver PDI was from Panvera
(Takara Biomedicals). Na125I, 125I-protein A
(9.5 µCi/µg), and [3,4,5-3H]leucine (152.2 Ci/mmol)
were purchased from DuPont NEN. Iodixanol (OptiprepTM) was
purchased from Life Technologies, Inc.
CaCo-2 cells were obtained from the American Type Culture Collection (Rockville, MD) and maintained in Eagle's minimal essential medium (EMEM) supplemented with nonessential amino acids, sodium pyruvate, 2 mM glutamine, and 20% NuSerum IV (Collaborative Biomedical, Bedford, MA) as described previously (7). For assaying cAMP accumulation and DT cytotoxicity, cells were grown in 24 × 16-mm clusters; for assaying the formation of CT-A1, they were grown in 6 × 35-mm clusters; and for preparing cellular fractions, they were grown in 75- or 175-cm2 flasks.
Toxin Cytotoxicity AssaysCells were washed once in serum-free EMEM buffered with 25 mM HEPES (EMEM/HEPES) and preincubated in the same medium containing 0.01% bovine serum albumin and p-CMBS or bacitracin where indicated for 60 min at 37 °C. Cells treated with anti-PDI monoclonal antibody RL90 were preincubated for 60 min at 4 °C. The exposure of cells to these inhibitors or antibodies was continuous throughout the experiment. The cells then were exposed to 1 nM CT or 1.7 nM DT for 2 h at 37 °C. For those cells treated with CT, 1 mM 3-isobutyl-1-methylxanthine was included in the culture medium throughout the experiment. Following exposure to CT, cells were assayed for cAMP by radioimmune assay as described previously (23). To control for the effects of p-CMBS or bacitracin on adenylyl cyclase activity, some of the cells treated with the inhibitors were exposed to 100 µM forskolin instead of CT for 30 min at 37 °C and assayed for cAMP. Those cells exposed to DT were washed with serum-free EMEM/HEPES, pulsed an additional 1 h at 37 °C with 1 µCi of [3H]leucine, and assayed for trichloroacetic acid-insoluble radioactivity as described previously (24).
Preparation of PNS and Membrane FractionsCaCo-2 cells were
grown to confluence and maintained in culture for at least 14 days
before harvesting. For labeling with 125I-CT, cells were
washed once with serum-free EMEM/HEPES and chilled. Cells then were
labeled with 125I-CT (106 cpm/ml; ~0.3
nM) in the same medium containing 0.01% bovine serum albumin for 30 min either (a) at 4 °C, washed, and
harvested immediately (t = 0) or (b) at 15 °C,
washed, and incubated for 60 min at 37 °C (t = 60)
prior to harvesting. When harvested, cells were rinsed once with
ice-cold PBS, scraped in PBS, and pelleted by low speed centrifugation.
Those cells labeled with 125I-CT were harvested in the
presence of 1 mM NEM to prevent any further reduction of CT
(6). The cell pellets were suspended in 50 mM Tris-HCl, pH
7.4, containing 250 mM sucrose (Tris/sucrose buffer) and
homogenized on ice with 10-20 strokes of a Dounce homogenizer. The
homogenate was spun at 3,000 × g to pellet unbroken cells, cell debris, and nuclei. The PNS was removed, and a portion of
it was centrifuged at 100,000 × g for 1 h at
4 °C. The resulting microsomal pellet was suspended in Tris/sucrose
buffer, and both the PNS and microsomal membranes (~1 mg/ml) were
stored in 1-ml portions at 80 °C.
To assay the enzymatic reduction of CT, PNS or microsomal membranes (1-10 µg of protein) were incubated with 125I-labeled CT in 0.2 ml (106 cpm, ~1-2 nM) for 30 min at 4 °C and centrifuged to pellet the resulting CT-bound membranes. The latter were then suspended in 50 µl of 50 mM sodium acetate buffer, pH 5.5, and 1% Triton X-100 unless otherwise stated. Samples were incubated for 1 h at 37 °C and rapidly neutralized by diluting with 450 µl of ice-cold PBS, pH 7.4, containing 1% Triton X-100, 1 mM NEM, and 20% ethylene glycol. Reductase assays in the presence of PDI inhibitors or anti-PDI antibodies were conducted in a similar manner. When the effects of sulfhydryl blockers were examined on intact microsomal vesicles, membranes were incubated with the indicated inhibitor in Tris/sucrose buffer for 30 min at 37 °C, pelleted, washed in the same buffer, and assayed for reductase activity as described above with the exception that 125I-CT was added directly to the solubilized membranes. In some experiments, PNS were incubated overnight at 4 °C in PBS, 1% Triton X-100 with and without 500 µg/ml of anti-PDI antibodies. After the samples were absorbed with protein A-Sepharose for 1 h at 4 °C, they were centrifuged to pellet the beads, and the resulting supernatants were assayed for reductase activity. The resulting protein A-Sepharose beads were also analyzed by SDS-PAGE and Western blotting to verify the presence or absence of PDI. The amount of CT-A1 generated during reductase assays was determined by SDS-PAGE. Each sample (~25,000 cpm) was mixed with an equal volume of 0.2% SDS sample buffer, heated for 10 min at 37 °C, and separated on a 15% gel. The gels were then fixed, dried, and exposed to Kodak XAR-5 film to visualize the labeled products by autoradiography.
Subcellular Fractionation and Gradient CentrifugationSubcellular fractionation was always performed on
freshly prepared PNS or microsomal membranes from either unlabeled
cells or those labeled with 125I-CT. PNS or microsomal
preparations were adjusted to 35% Iodixanol, and 3 ml was layered
under a linear 10-ml gradient of 9-30% Iodixanol in Tris/sucrose
buffer. Samples were centrifuged in an SW 28 rotor at 52,000 × g for 90 min at 4 °C using a Beckman L8-70M
ultracentrifuge. Fractions (~0.5 ml) were collected from the top of
the gradient using a density gradient fractionator (ISCO). A 15-µl
portion of each gradient fraction was used to determine its refractive index () as measured on an Abbe MARK II digital refractometer. From
the refractive indices of standard Iodixanol solutions in Tris/sucrose
buffer and their corresponding densities (
), the following equation
was calculated to determine the densities of individual gradient
fractions and is in agreement with that reported by Graham et
al. (25):
= 3.4911
3.6664.
Gradient fractions from cells labeled with 125I-CT for 60 min at 37 °C were analyzed for CT-A1 peptide formation by immunoprecipitation of the individual fractions. Every two fractions along the gradient were pooled, treated with 1% Triton X-100, and immunoprecipitated with anti-CT-A1 antiserum overnight at 4 °C. Samples were analyzed by SDS-PAGE and autoradiography as described. Fractions from unlabeled cells were similarly pooled, precipitated with 10% trichloroacetic acid and 100 µg of bovine serum albumin as carrier, washed twice with 95% ice-cold ethanol, and solubilized with SDS-sample buffer. Samples were separated by SDS-PAGE, electroblotted onto Immobilon-P, and probed with anti-PDI antibodies. For the analysis of CT-related reductase activity in isolated gradient fractions, membranes from individual gradient fractions were diluted 4-fold with Tris/sucrose buffer and the membranes pelleted by a brief centrifugation in a TLA-100.2 fixed angle rotor at 200,000 × g for 15 min at 4 °C using a Beckman benchtop TL-100 ultracentrifuge. Isolated membrane fractions were then solubilized in 90 µl of 50 mM sodium acetate buffer, pH 5.5, containing 1% Triton X-100, and each was separated into two equivalent portions. One portion was subsequently treated with 1 mM NEM for 30 min at 37 °C. All sample volumes were then adjusted to 50 µl with the addition of 125I-CT and buffer, incubated for 60 min at 37 °C, and analyzed for the formation of CT-A1 as described above. A portion (25 µl) of these reaction mixtures was simultaneously probed for PDI by Western blotting as described above.
Reduction of CT by Purified Bovine PDIReduction of
membrane-bound CT was performed using alkali-treated membranes from
PNS. Alkali-treated membranes were prepared as described (26). Briefly,
100 µg of microsomal membranes were pelleted, suspended in 0.1 ml of
100 mM Na2CO3, pH 11.5, and
incubated at 4 °C for 30 min. Control membranes were treated in a
similar manner except that Tris/sucrose buffer was used throughout.
Membranes were then pelleted, rinsed, and suspended in 0.1 ml of
Tris/sucrose buffer containing 10 nM 125I-CT
(107 cpm/pmol). Toxin was allowed to bind at 4 °C for 60 min, and the membranes were pelleted. Assays for the reduction of
membrane-bound CT were then performed either in 50 mM
sodium acetate, pH 5.5, or 50 mM sodium phosphate buffer,
pH 7.0, in the presence of purified bovine PDI, GSH, and GSSG at
concentrations described in Fig. 6. No detergent was included in assays
involving alkali-treated membranes. Samples were incubated at 37 °C
for 30 min, neutralized, and prepared for SDS-PAGE.
Other Methods
Monoclonal antibodies RL77, RL90, and HP13 (27) were purified from either ascites (RL77, HP13) or conditioned culture supernatants (HP13) by protein G affinity chromatography. The generation of CT-A1 by intact cells was determined as described previously (6). CT was iodinated by a chloramine-T procedure (28) and routinely had a specific activity of ~15-20 µCi/µg. Protein was determined by the method of Lowry et al. (29), using bovine serum albumin as the standard. SDS-PAGE was performed according to Laemmli (30). Proteins separated by SDS-PAGE were electrophoretically transferred to Immobilon-P membranes (31). PDI was detected by immunoblotting with a purified IgG fraction (25 µg/ml) from RL77 ascites and 125I-labeled protein A or by chemiluminescence (Boehringer Mannheim). Densitometry of autoradiographs was performed using a Lacie Silverscanner II and quantified using NIH Image software, version 1.55.
When CaCo-2 cells containing bound 125I-CT
were incubated at 37 °C, the cells generated small amounts of
CT-A1 after a lag of at least 15 min (Fig.
1A). Similar lags have been described
previously (6) and are consistent with the time-dependent
internalization of membrane-bound toxin that precedes the activation of
adenylyl cyclase and subsequent accumulation of cAMP by these and other cells (6, 7, 10, 32). The delay in CT-A1 peptide formation reflects the possibility that endocytosed CT proceeds via a defined intracellular pathway that results in its reduction and subsequent activation of adenylyl cyclase. This is in direct contrast to the
described mechanism for DT activation in which reduction of the toxin
occurs at or near the plasma membrane (21, 33). This disulfide-exchange
reaction is catalyzed via a cell surface pool of PDI. To compare the
activation of both CT and DT in intact cells, the cytotoxic effects of
both toxins on CaCo-2 cells were examined in the presence of several
membrane-impermeant sulfhydryl blockers. Cells were pretreated with
p-CMBS, bacitracin, or a combination of the two reagents
prior to the addition of either toxin. Bacitracin has been shown to be
an effective inhibitor of PDI activity (34, 35). As shown in Table
I, both pCMBS and bacitracin inhibited the cytotoxic
effects of DT at concentrations reported to inhibit PDI activity
in vitro (21) and block DT cytotoxicity in CHO cells (33).
In contrast, neither reagent, either alone or in concert, affected the
ability of the cells to generate significant levels of
CT-A1 peptide from bound CT (Fig. 1B). This was
also reflected in the lack of any observable effects on CT activity as
measured by the toxin-mediated increase in intracellular cAMP (data not
shown). NEM, which is membrane-permeant, however, completely inhibited
the formation of CT-A1 peptide.
|
To further study the involvement of cell surface-associated PDI in the activation of both CT and DT, CaCo-2 cells were incubated in the presence of anti-PDI monoclonal antibody RL90 prior to the addition of either toxin. Affinity-purified anti-PDI antibodies at a concentration of 0.5 mg/ml inhibited DT cytotoxicity by 36.9% as compared with 3.8% in the presence of an irrelevant mouse IgG control (Table II). This is consistent with results of a previous study on DT activity in the presence of anti-PDI antibodies (21). RL90 antibodies, however, had no effect on the activity of CT under the same conditions (Table III). Comparable levels of cAMP were generated in control and antibody-treated cells. The results (Fig. 1 and Tables I, II, III) indicated that CT, unlike DT, was not reduced and activated by the cell surface-associated population of PDI (or another reductase) although this did not exclude the possibility that the reduction of CT occurred at a later stage in its intracellular transit through the tubulovesicular network (8) with the possible involvement of the intracellular pool of PDI. Indeed, the potent inhibitory effects of NEM on the activation of CT (Fig. 1B) may reflect not only the involvement of the intracellular pool of PDI but also the role of intracellular vesicular transport in general and the internalization of plasmalemmal-derived vesicles in particular (36-38).
|
|
To identify the cellular
reductase(s) responsible for the formation of CT-A1
peptide, we next assayed for the reduction of CT in a cell-free system
using either PNS or microsomal vesicles derived from PNS (Fig.
2). Because toxin subunit interactions are susceptible
to pH (39, 40), the effects of pH on CT-A1 peptide
formation were first examined. 125I-CT was incubated with
CaCo-2-derived PNS at 4 °C, and the toxin-bound membranes were
washed and pelleted by centrifugation. The membranes were then
solubilized in buffers ranging from pH 7.5 to 3.0 containing 1% Triton
X-100 and incubated at 37 °C for 60 min. Under these conditions,
holotoxin was most efficiently reduced in a slightly acidic environment
(Fig. 2A). Maximum reduction of CT to CT-A1 peptide occurred between pH 5.5 and 6.0. It was these parameters (50 mM sodium acetate, pH 5.5, containing 1% Triton X-100)
that were used to further examine the reduction of membrane-bound CT in vitro. When the CT-related reductase activity in PNS was
compared with that of the derived microsomal and cytosolic fractions,
the majority of the activity was found in the microsomal fraction (Fig.
2B, lane 3). This distribution correlated with
the distribution of PDI as assessed by immunoblotting of these same
samples with anti-PDI antibodies (Fig. 2B, lanes
7-9).
Since PDI is a soluble protein that resides in the lumen of the ER (15), we next determined whether the endogenous reductase activity had similar properties. Intact microsomal vesicles were pretreated with NEM, p-CMBS, and bacitracin for 30 min at 37 °C to examine their effects on the in vitro reduction of CT. Microsomal vesicles were then washed, solubilized in reaction buffer and 125I-CT, and incubated for 60 min at 37 °C. As was observed when intact cells were exposed to these inhibitors (Fig. 1B), neither of the membrane-impermeant sulfhydryl blockers inhibited CT reduction (Fig. 2B, lanes 5 and 6). Only NEM inhibited the formation of A1 peptide. In contrast, when membranes were first solubilized with reaction buffer containing 1% Triton X-100 and treated with these same inhibitors prior to the addition of 125I-CT, A1 was not formed (Fig. 2C). The inhibition of CT reduction in the presence of p-CMBS and bacitracin was concentration-dependent with IC50 values of 10 µM and 3 mM for p-CMBS and bacitracin, respectively (data not shown). These concentration ranges closely paralleled those shown to inhibit the PDI-dependent reduction of disulfide-containing macromolecules inserted into the plasma membrane of CHO cells (21, 41). We also tested the effects of BFA on CT reduction in vitro. No inhibition of CT reduction was detected when 1 µg/ml BFA was added directly to the assay (Fig. 2C, lane 6) or when PNS was prepared from cells treated with 10 µg/ml BFA for 60 min at 37 °C (Fig. 2C, lane 7). The results shown in Fig. 2 demonstrated that without Triton X-100 to solubilize the vesicles, the reductase was not accessible to membrane-impermeant inhibitors such as p-CMBS and bacitracin. It was found that in the absence of nonionic detergent only 19% of CT-A was reduced to the A1 peptide using untreated microsomes compared with 58% in the presence of 1% Triton X-100 (data not shown). These results were consistent with the properties of PDI, the vast majority of which is found as a soluble protein in the lumen of the ER (15).
PDI-Specific Monoclonal Antibodies Inhibit the in Vitro Reduction of CTCaCo-2 PNS were incubated with anti-PDI monoclonal antibody
RL77 or RL90 in the presence of 1% Triton X-100, and any PDI-antibody complexes were then absorbed with protein A-Sepharose. Both antibodies were raised against rat liver PDI and previously shown to cross-react with human PDI as well as inhibit its activity in vitro
(27). When the absorbed extracts were assayed for CT reductase
activity, it was found that RL77 and RL90 reduced the level of
CT-A1 generated by 55 and 80%, respectively when compared
with control extracts of PNS that were absorbed with protein
A-Sepharose alone. Immunoblotting confirmed that PDI was depleted from
the PNS extracts and recovered in the immunoabsorbates (data not
shown). When added directly to the reductase assay, both antibodies
inhibited the reduction of 125I-CT by PNS in a
concentration-dependent manner (Fig. 3). A
third anti-PDI monoclonal antibody, HP13, raised against human placenta PDI, gave similar results, whereas an irrelevant mouse IgG was ineffective.
Subcellular Co-localization of PDI and A1 Peptide Formation
To identify the intracellular site(s) of CT activation
and the reductase(s) involved, PNS-derived microsomes were analyzed by
subcellular fractionation on 9-30% Iodixanol gradients (25). CaCo-2
monolayers were either labeled with 125I-CT for 30 min at
4 °C, washed, and homogenized immediately to prepare the PNS
fraction (t = 0), or they were labeled for 30 min at
15 °C, washed, and incubated an additional 30 min (t = 30) or 60 min (t = 60) at 37 °C prior to PNS
preparation. Microsomes were then prepared, and the membranes were
separated on the Iodixanol gradient. The 125I-CT density
profiles for the subcellular fractionation of t = 0, t = 30, and t = 60 microsomes are shown
in Fig. 4A. Bound toxin is not internalized
at or below 15 °C (32); therefore, the t = 0 profile
represents the distribution of bound toxin to plasma membrane elements.
A significant amount of toxin was found in light fractions near the top
of the t = 0 gradient. The peak fraction in this region
was found to be rich in caveolin and suggested that a portion of the
plasma membrane-bound toxin localized in caveolae.2 This is consistent with the
observations of Tran et al. (8), who showed that CT bound to
smooth invaginations on the cell surface. These have since been
identified as caveolae and shown to be enriched in ganglioside
GM1 (42).
As shown in Fig. 1, when cells were warmed to 37 °C for 30 and 60 min, the internalized toxin was slowly reduced to generate minute amounts of the A1 peptide. The 125I-CT density profile of microsomal membranes prepared from these cells showed a significant shift in the distribution of the toxin (Fig. 4A) with the t = 30 profile reflecting an intermediate profile between the t = 0 and t = 60 samples. When fractions along the t = 30 and t = 60 profiles were analyzed for CT-A1 peptide, the majority of A1 present in each was distributed within a density range of 1.122-1.127 g/ml (Fig. 4B). Western blot analysis for the distribution of PDI using unlabeled microsomal membranes from a similar preparation detected PDI within the same density range of 1.121-1.129 g/ml (Fig. 4C).
We next examined the in vitro reduction of CT using
membranes isolated from the subcellular fractionation of PNS on
Iodixanol gradients. For these studies, we used freshly prepared PNS
without first spinning down the microsomal membranes to avoid any
perturbation of the subcellular organelles. Gradient fractions also
were probed for the presence of PDI. As shown in Fig.
5A, maximum reduction of CT occurred using
membranes isolated in the density range of 1.108-1.115 g/ml. These
fractions also contained the majority of PDI when immunoblots of these
fractions were probed with antibodies against PDI (Fig. 5B).
To distinguish these fractions from those of the Golgi, we probed the
same fractions with monoclonal antibodies against -COP and the Golgi
marker, g58. As shown in Fig. 5, the Golgi markers appeared at a
lighter density (1.077-1.08 g/ml). In contrast, anti-KDEL antiserum
(used to further establish ER-derived fractions) detected only those
proteins that co-localized to fractions containing PDI. Taken together,
the results shown in Figs. 4 and 5 indicated that the endogenous
CT-related reductase and PDI co-localized in the same ER-derived
subcellular fractions and that these same fractions contained the bulk
of the A1 peptide generated by intact cells after 1 h.
In Vitro Reduction of Membrane-bound CT by Purified Bovine PDI
Purified bovine PDI was tested for its ability to reduce membrane-bound CT in reconstitution experiments using membranes devoid of PDI. To examine PDI reductase activity on CT reduction, assay parameters also were altered to more closely reflect the redox environment and pH that are thought to exist within the lumen of the ER. CaCo-2 membranes were first treated with 100 mM sodium carbonate for 30 min at 4 °C. Such treatment causes membrane vesicles to open, flatten, and release their lumenal contents (26). Alkali treatment of PNS-derived membranes effectively removed all endogenous PDI as judged by Western blot analysis and resulted in a corresponding lack of CT reduction (Fig. 6). 125I-CT was then bound to the alkali-treated membranes for 60 min at 4 °C and repelleted. The reduction of membrane-bound 125I-CT was then monitored as a function of exogenously added PDI and glutathione. Assays were performed at pH 5.5 and pH 7.0 for comparison with previous results. Glutathione (GSH, reduced form) concentrations within the ER have been estimated to range between 0.5 and 1 mM (43). Similarly, PDI is found in such high concentrations within the lumen of the ER (0.4% of liver protein and over 2% of liver microsomal protein) as to suggest a direct participation in the redox buffering capacity of the ER (44). When experiments were conducted at pH 7.0 to examine the concentration-dependent effects of GSH on the reduction of membrane-bound CT in the absence of PDI, very little CT-A1 peptide was formed (3-6% reduction was routinely observed at concentrations of GSH between 0.5 and 1.0 mM). As demonstrated by Moss et al. (22), for the PDI-catalyzed reduction of CT in solution, reduction of membrane-bound CT at a neutral pH in the presence of 1 mM GSH was dependent on the presence of PDI and increased significantly with increasing concentrations of PDI added (Fig. 6A). The formation of CT-A1 peptide was increased ~13-fold at pH 7.0 and 56 µM PDI (3 mg/ml). No observable PDI-dependent reduction of CT occurred at neutral pH values in the absence of any exogenously added thiol. In contrast, when assays were conducted at pH 5.5, the reduction was increased nearly 27-fold at pH 5.5 and 19 µM PDI (1 mg/ml). In general, whereas the presence of PDI was effective in substantially decreasing the concentration of GSH necessary to generate a comparable level of CT-A1 peptide (data not shown), only at pH 5.5 were small amounts of CT-A1 peptide still formed by PDI in the absence of any exogenous thiol. The lowering of the dependence for GSH on the PDI-catalyzed reduction of CT at pH 5.5 may explain our earlier observations of more efficient CT reduction with PNS under slightly acidic conditions. It has likewise been suggested that reduction of disulfide bonds by PDI may also be dependent on protein conformation and accessibility of the disulfide bond to the enzyme.2,3 A subtle change in the conformation of CT-A at pH 5.5 may provide better access to the disulfide bridge and allow PDI to act even in the absence of optimal thiol concentrations.
We next examined the concentration effects of various GSH/GSSG buffers on the modulation of PDI-dependent reduction. The lumen of the ER is highly oxidizing with a GSH/GSSG ratio on the order of 3:1 to 1:1, whereas the overall cellular environment is considerably more reducing (30:1 to 100:1) (42). When the GSH/GSSG ratios were altered to more closely parallel the conditions of the ER, the PDI-dependent reduction of CT was decreased as the redox environment became more oxidizing. However, even when GSH/GSSG ratios ranged between 1:1 and 3:1, PDI still enhanced the reduction of CT from ~8- to 14-fold above levels generated in the presence of thiol only (Fig. 6B).
The intoxication of cells by CT requires the formation of
CT-A1 inside the cells. After a defined lag phase, small
amounts of CT-A1 are generated by the cells that are
sufficient to activate adenylyl cyclase by ADP-ribosylation of
Gs. Although it is well established that CT treated with
millimolar concentrations of reducing agents can catalyze the
ADP-ribosylation of Gs
and activation of adenylyl
cyclase in membranes (45), up to now little has been known about the
cellular reduction of CT. In the present study, it was shown by several
different methods that in CaCo-2 cells and their subcellular fractions,
the reduction of CT was catalyzed by a reductase with the properties of
intracellular PDI. These included the characterization and localization
of the CT reductase, the effects of PDI-specific monoclonal antibodies and inhibitors on this reductase, and the co-localization of
CT-A1 formed by intact cells in a subcellular compartment
containing PDI. Additionally, this compartment contained markers
characteristic of the ER with minimal overlap of those characteristic
of the Golgi apparatus.
As a soluble enzyme, PDI can readily be released from its lumenal domain by the addition of detergent to membrane preparations. We have shown that the reductase activity found in microsomal vesicles had only limited access to membrane-bound CT unless a nonionic detergent such as Triton X-100 was present in the assay mixture. Similarly, in the absence of detergent, intact vesicles shielded the lumenal-resident reductase from the inhibitory effects of p-CMBS or bacitracin. In contrast, most of the reductase activity was lost from microsomal membranes following alkali treatment, which causes the release of lumenal and peripheral proteins (25). This loss closely correlated with the loss of PDI content in those same treated vesicles. The ability of several anti-PDI monoclonal antibodies to deplete the CT reductase from CaCo-2 PNS by immunoabsorption as well as to inhibit the in vitro reduction of CT confirmed the identity of the reductase acting on CT in vitro as PDI. Further, where assayed under conditions designed to reflect the redox conditions present in the lumen of the ER, purified bovine PDI was able to reduce CT bound to membranes depleted of cellular reductase(s).
Studies on the reductive cleavage of membrane-bound disulfide-linked macromolecules have shown that reduction can occur at two distinct stages within the cell (41), both of which appear to involve the activity of PDI. When 125I-labeled, disulfide-containing conjugates are bound to intact CHO cells and then chased into the cell, reduction is observed both at early times (within the first 15 min without a lag period) and at later times (beginning at ~30 min into the chase). Only the early stage of reduction is inhibited by membrane-impermeant sulfhydryl blockers such as p-CMBS and the PDI-specific inhibitor, bacitracin. Hence, this phase of reduction occurs at the cell surface or during the earliest stages of endocytosis, presumably with the involvement of cell surface sulfhydryl groups. The latter stage of reduction appears to occur in an unidentified, nonlysosomal compartment.
The present study provided direct evidence that the reduction of CT by target cells occurred at an intracellular location rather than at or near the plasma membrane. This is in contrast to several earlier reports based on photolabeling techniques (46, 47), which presented a model for toxin activation at the plasma membrane. While a population of PDI exists on the cell surface of mammalian cells (19, 20), our results indicated that it was the intracellular pool of PDI that catalyzed CT reduction. Neither p-CMBS, bacitracin, nor anti-PDI monoclonal antibody RL90 exhibited any inhibitory effects on CT activity. CaCo-2 cells exposed to these agents were able to generate levels of CT-A1 similar to those found in untreated cells, and there were no appreciable differences in cAMP accumulation.
In contrast, these same membrane-impermeant reagents were effective inhibitors of DT in CaCo-2 cells, in agreement with earlier work on CHO cells (21, 33). Their inhibition of early stage reduction is coincident with the time during which DT is endocytosed and translocated across the membranes of acidic endosomes and indicated that surface sulfhydryls contributed by the cell surface pool of PDI were factors in the reductive cleavage and activation of DT. While reduction of DT may not necessarily occur at the plasma membrane, it may be catalyzed by PDI, which is originally present on the plasma membrane and, through endocytosis, becomes oriented on the lumenal side of early endosomes. Although small amounts of the sulfhydryl blockers present in the medium may be taken up by the cell, it is highly unlikely that they could proceed further than the early stages of the endosomal system. That these agents had no observable effect on the action of CT suggested that an intracellular pool of PDI (presumably that distributed within the ER and early Golgi complex) mediated the thiol-disulfide interchange reaction. Also, the distinct lag period prior to the CT-dependent activation of adenylyl cyclase coincides with the time frame of the later phase of PDI-dependent reduction of disulfide-containing conjugates described by Feener et al. (41).
The subcellular fractionation studies not only support the role of PDI in the reduction of CT but also provided additional insight into the intracellular transport of the toxin. Thus, the endogenous CT reductase activity co-localized to the same subcellular fractions as PDI and most probably represented the ER compartment. A second ER marker, the KDEL receptor, was also present in these same fractions, and on Iodixanol density gradients both ER markers appeared to band in a density range of 1.10-1.13 g/ml. This is in close agreement with values reported by Graham et al. (25) for the ER marker enzyme NADPH-cytochrome c reductase. This location was also quite distinct from CT-bound plasma membrane elements (as determined by CT binding at 4 °C). Golgi-associated membranes have been reported to band at a density of 1.06-1.09 (25), where we found the Golgi marker, gp58. Subcellular fractionation of CT-labeled cells showed that CT-A and the formation of CT-A1 peptide were localized to the same intracellular compartment as PDI. That these elements were coincident suggested that the intracellular trafficking of CT prior to its activation involved transit of CT-A through the ER. In showing the relationship between the toxin and an ER-resident reductase such as PDI, we further illustrated the importance of those organelles associated with the secretory pathway in the activation of CT.
We have previously provided evidence that an intact Golgi is required for the reduction of CT and the formation of CT-A1 peptide (7). Cells treated with BFA are unable to reduce CT to form the A1 peptide. While the effects of BFA on the morphology of the ER and Golgi complex, the redistribution of endogenous enzymes, and the inhibition of vesicular transport have been well studied (9), it was not clear how these effects contributed to the inhibition of toxin reduction. Whereas CaCo-2 cells pretreated with as little as 0.1 µg/ml of BFA are unable to reduce CT to the A1 peptide and are completely resistant to the action of the toxin (7), BFA had no effect on the reduction of CT in vitro. Likewise, PNS prepared from cells treated with 10 µg/ml of BFA for 60 min was as effective as PNS from untreated cells in reducing CT in vitro. These findings eliminated the possibility that BFA directly inhibited the reductase involved in CT reduction in intact cells. More study will be necessary to understand the exact relationship between the effects of BFA on Golgi organization and vesicular transport and to understand its ability to block toxin reduction. It is interesting to note, however, that PDI expresses the ER retention sequence, KDEL, at its C terminus (14). Its distribution in the ER and early Golgi would make this enzyme susceptible to the actions of BFA. Redistribution of ER- and Golgi-resident enzymes could contribute to the inability of BFA-treated cells to reduce internalized CT. Although some PDI has been reported to be present in the Golgi apparatus (17) and the trans-Golgi network (18), we have not been able to detect any significant levels of PDI and CT-A1 peptide co-localizing in these lighter fractions derived from CaCo-2 cells. Thus, the more likely possibility is that BFA treatment blocks the retrograde translocation of CT to the ER, where it is reduced by PDI.
Finally, the results of our in vitro experiments with purified PDI reflected those obtained with PNS and indicated that the two activities were the same. Additionally, PDI remained an effective catalyst for toxin reduction even under such suboptimal conditions as lower pH and thiol concentration (as defined for the oxidizing activity of PDI). Even under the oxidizing environment associated with the lumen of the ER, PDI-dependent reduction of CT was still favorable.
Although we cannot determine with absolute certainty that CT reduction occurs in the ER, all of the data are consistent with such a mechanism. The results of a recent study by Majoul et al. (48) support the conclusions drawn from our study. As in this work, their investigation demonstrated that small amounts of CT-A can be detected within the ER after 30 min. These levels were found to increase with time and were accompanied by the formation of minute amounts of CT-A1 peptide. In this regard, the intracellular processing of CT shares some similarities with another ADP-ribosylating toxin, Pseudomonas exotoxin A (see Refs. 7, 49, and 50). Both toxins need to be reduced in order to be activated; both toxins have C termini that end with sequences recognized by the KDEL receptor; the action of both toxins is blocked by BFA treatment; and the active fragments of both toxins have to undergo translocation from an extracytosolic compartment into the cytosol in order to reach their targets. In the case of exotoxin A, it is translocated out the ER into the cytosol via the ER protein-exporting mechanism (50). Whether CT uses the same mechanism remains to be determined.
I thank Dr. Peter H. Fishman for guidance and helpful suggestions. I also thank Drs. Charlotte S. Kaetzel (University of Kentucky College of Medicine) and H. J. P. Ryser (Boston University School of Medicine) for helpful discussions on PDI and appreciate Dr. Kaetzel's generosity in providing the anti-PDI monoclonal antibodies and parent hybridoma cells.