Transient expression of RhoA, -B, and -C GTPases in HeLa cells potentiates resistance to Clostridium difficile toxins A and B but not to Clostridium sordellii lethal toxin.

Abstract

The bacterial pathogen Clostridium difficle synthesizes two high-molecular-weight toxins (A and B), which exhibit toxic effects in vivo and in vitro. Here, we present evidence that the major intracellular targets of these two toxins are the Rho GTPases. Overexpression of RhoA, RhoB, or RhoC GTPases in transfected HeLa cells conferred an increased resistance to toxins A and B, indicating that these toxins cause their cytopathic effects primarily by affecting Rho proteins. In addition, toxin A and B treatment appeared to result in modification of Rho, since Rho isolated from toxin-treated cells had a decreased ability to be ADP-ribosylated by Clostridium botulinum C3 exoenzyme. In contrast, the lethal toxin (LT) of Clostridium sordellii, although structurally and immunologically related to C. difficile toxin B, appeared to induce cytopathic effects independently of the Rho GTPases. Overexpression of RhoA in transfected HeLa cells did not protect them from the effect of LT, and Rho isolated from lysates of LT-treated cells was not resistant to modification by C3. Immunofluorescence studies showed that LT treatment caused a cytopathic effect that was very different from those described for C. difficile toxins A and B, resulting in an increase in cortical F-actin, with a concomitant decrease in the number of stress fibers, and in the formation of numerous microvilli containing the actin-bundling protein fimbrin/plastin.

Full text

Page 1

INFECTION AND IMMUNITY, Oct. 1995, p. 4063–4071
0019-9567/95/$04.00?0
Copyright ? 1995, American Society for Microbiology
Vol. 63, No. 10
Transient Expression of RhoA, -B, and -C GTPases in HeLa
Cells Potentiates Resistance to Clostridium difficile Toxins A
and B but Not to Clostridium sordellii Lethal Toxin
MURIELLE GIRY,1* MICHEL R. POPOFF,1CHRISTOPH VON EICHEL-STREIBER,2
AND PATRICE BOQUET1
Unite ´ des Toxines Microbiennes, Institut Pasteur, 75724 Paris cedex 15, France,1and Institut fu ¨r
Medizinische Mikrobiologie, Johannes-Gutenberg-Universita ¨t, D-55101 Mainz, Germany2
Received 30 March 1995/Returned for modification 11 May 1995/Accepted 13 June 1995
The bacterial pathogen Clostridium difficile synthesizes two high-molecular-weight toxins (A and B), which
exhibit toxic effects in vivo and in vitro. Here, we present evidence that the major intracellular targets of these
two toxins are the Rho GTPases. Overexpression of RhoA, RhoB, or RhoC GTPases in transfected HeLa cells
conferred an increased resistance to toxins A and B, indicating that these toxins cause their cytopathic effects
primarily by affecting Rho proteins. In addition, toxin A and B treatment appeared to result in modification
of Rho, since Rho isolated from toxin-treated cells had a decreased ability to be ADP-ribosylated by Clostridium
botulinum C3 exoenzyme. In contrast, the lethal toxin (LT) of Clostridium sordellii, although structurally and
immunologically related to C. difficile toxin B, appeared to induce cytopathic effects independently of the Rho
GTPases. Overexpression of RhoA in transfected HeLa cells did not protect them from the effect of LT, and Rho
isolated from lysates of LT-treated cells was not resistant to modification by C3. Immunofluorescence studies
showed that LT treatment caused a cytopathic effect that was very different from those described for C. difficile
toxins A and B, resulting in an increase in cortical F-actin, with a concomitant decrease in the number of stress
fibers, and in the formation of numerous microvilli containing the actin-bundling protein fimbrin/plastin.
Clostridium difficile produces two high-molecular-weight tox-
ins, called A and B, which are the cause of antibiotic-associated
pseudomembranous colitis (5, 6, 45). Toxins A and B are also
cytotoxic for cultured cells but have different potencies, with
toxin B being 1,000-fold more active than toxin A. Both C.
difficile toxins induce similar morphological changes in cul-
tured cells, including rounding of cell bodies (9, 14) and for-
mation of membrane arborization (15). The N termini of toxins
A (300 kDa) and B (250 kDa) show extensive similarities (50).
This similarity may indicate that these two toxins share a com-
mon intracellular activity, since the enzymatic domain (effec-
tomer) in many bacterial toxins is located near the amino
terminus of the protein (18).
Toxins A and B from C. difficile act on cultured cells pri-
marily by disrupting the F-actin cytoskeleton network without
interfering directly with F-actin formation (35, 46, 47). Re-
cently, it was shown that C. difficile toxin B may affect actin
microfilaments by causing a covalent modification of the Rho
GTPase, which presumably inactivates this regulatory protein
(23, 24).
It is now clear that the Ras-like Rho protein family, which in
mammalian cells includes RhoA, RhoB, RhoC, Cdc42Hs,
RhoG, Rac1, and Rac2 (11, 13, 30, 34, 44, 48, 52) is involved
mainly in the regulation of the F-actin cytoskeleton (reviewed
in reference 20). Direct evidence that Rho molecules are in-
volved in the assembly of F-actin structures such as stress fibers
(10) was obtained by treating cells with a specific rho ADP-
ribosylating bacterial enzyme (exoenzyme C3), isolated from
the culture supernatants of C. botulinum C and D (3, 43). This
observation was confirmed by microinjection into serum-
starved mouse Swiss 3T3 cells of the constitutively activated
form of RhoA GTPase (RhoA Val14) which resulted in the
formation of a dense network of actin stress fibers (36). The
role of Rho proteins was further characterized by Ridley and
Hall, who demonstrated that Rho GTPases regulate formation
of stress fibers in response to growth factor-stimulated signal
transduction pathways (40, 42). Recent observations have im-
plicated Rho GTPases in the regulation of actin assembly via
activation of different enzymes such as phosphatidylinositol
3-phosphate kinase (53), a kinase which phosphorylates the
focal adhesion kinase protein (p125FAK) (26), a tyrosine ki-
nase involved in the phosphorylation of certain adhesion
plaque proteins (41), and phospholipase D (7). It is therefore
of interest to demonstrate that C. difficile cytotoxins A and B
could act directly on the Rho protein family, since these toxins
can provide valuable tools for cell biologists to study how these
small GTPases are involved in the control of F-actin assembly
and disassembly.
In the present work, we have developed an assay to test the
possibility that C. difficile toxins A and B were active in vivo
either on a specific form of Rho or on different Rho members
of this family of small GTPases. The rationale of our experi-
ments was that if toxins A and B act on Rho proteins (by an
enzymatic covalent modification), one should be able to po-
tentiate the cellular resistance to these toxins by greatly in-
creasing the intracellular concentration of Rho GTPases. We
have transfected cells with RhoA, RhoB, and RhoC cDNAs
and shown that such cells have an increased resistance to C.
difficile A and B toxins. In contrast, C. sordellii lethal toxin
(LT), which is immunologically and structurally related to C.
difficile toxin B (37), did not act on Rho GTPases, since cells
transfected with the cDNAs of these proteins did not show any
resistance to LT toxin. Instead, LT toxin induced the formation
of numerous microvillus structures on the cell surface. These
structures contained the actin-binding and microfilament-bun-
dling protein fimbrin/plastin (8).
* Corresponding author. Mailing address: Unite ´ des Toxines Micro-
biennes, Institut Pasteur, 28 rue du Dr Roux, 75724 Paris cedex 15,
France.
4063

Page 2

MATERIALS AND METHODS
Materials. Escherichia coli TG1 was used as the host of recombinant plasmids.
Human RhoA cDNA was kindly provided by P. Madaule, Institut Pasteur, Paris,
France, and human RhoB and RhoC cDNAs were kindly provided by P. Char-
din, Institut de Pharmacologie Mole ´culaire et Cellulaire, Sophia Antipolis,
France. As vectors, we used pUC19, M13mp18, pKC3 (a eukaryotic expression
vector derived from the parental vector pSV-neo) containing the simian virus 40
early promoter, and pCB6 (a eukaryotic expression vector containing the cyto-
megalovirus promoter). PCR amplifications were performed with Taq poly-
merase (Beckman, Paris, France). DNA sequencing was carried out with the
Sequenase DNA-sequencing kit (version 2.0; U.S. Biochemical Co., Cleveland,
Ohio) as specified by the manufacturer.
Cell culture. HeLa (human epitheloid carcinoma) cells were grown in Dul-
becco’s modified Eagle’s medium (Gibco BRL, Eragny, France), supplemented
with 5% fetal calf serum (Gibco BRL), 100 U of penicillin per ml, and 100 ?g of
streptomycin per ml (Gibco BRL), at 37?C under a 5% CO2atmosphere. LL-
CPK1 cells were cultivated in Dulbecco’s modified Eagle’s medium supple-
mented with 10% fetal calf serum.
Antibodies and toxins. Monoclonal antibody P5D4 was raised against the
11-amino-acid peptide (YTDIEMNRLGK) contained in the vesicular stomatitis
virus glycoprotein G carboxy terminus (25). C. botulinum type D (strain 1873)
exoenzyme C3 (43), C. difficile toxins A and B (49), C2-like toxin from C.
spiroforme (strain NCTC 11493 [Rho ˆne-Me ´rieux collection no. 15991]) (38), and
C. sordellii LT from strain IP82 (37) were purified as previously described.
Constructions of epitope-tagged Rho proteins. Rho proteins were tagged with
oligonucleotides encoding 11 amino acids of the vesicular stomatitis virus (VSV)
glycoprotein G epitope (25). The epitope tag was fused to the N termini of the
Rho proteins. RhoA cDNA was subcloned into pUC19 digested with XmaI and
XbaI. Complementary oligonucleotides encoding the epitope tag were synthe-
sized with an EcoRI 5?-flanking site and a KpnI 3?-flanking site. The linker,
formed by annealing the two oligonucleotides, was introduced in frame upstream
of RhoA into pUC19-RhoA. The EcoRI-PstI fragment, encoding VSV G-RhoA,
was subcloned into the eukaryotic expression vector pKC3, and the insert was
verified by sequencing. RhoB cDNA was amplified by PCR to create a KpnI site
upstream of the initiation codon and a BamHI site downstream of the stop
codon. PCR products were digested with KpnI and BamHI and cloned into the
eukaryotic expression vector pCB6 (previously digested with the same enzymes).
Complementary oligonucleotides encoding the tag were synthesized with a BglII
5?-flanking site and a KpnI 3?-flanking site, and the corresponding linker was
introduced in frame upstream of RhoB into pCB6-RhoB. The BglII-BamHI
fragment encoding VSV G-RhoB was subcloned into pUC19 vector to allow the
subcloning of a larger EcoRI-XbaI fragment from pUC19 into the eukaryotic
expression vector pKC3. The sequence of this fragment was verified. The EcoRI-
BamHI fragment containing RhoC cDNA (subcloned into the M13mp18 vector)
was cloned into pCB6, and a linker encoding the VSV G epitope tag was
introduced in frame directly upstream of RhoC into pCB6-RhoC. This linker
shares a KpnI 5?-flanking site and an EcoRI 3?-flanking site. The subsequent
recombinant plasmid pCB6-VSV G-RhoC was digested. The BglII-BamHI DNA
fragment present in pCB6-VSV G-RhoC was subcloned into pUC19 to allow the
subcloning of a larger SacI-PstI DNA fragment into the eukaryotic expression
vector pKC3.
Transfection of HeLa cells and assay of toxin activity. Exponentially growing
HeLa cells were seeded on coverslips at 105cells per 35-mm culture dish 24 h
before transfection. DNA transfection was done by the classical calcium phos-
phate technique. The calcium phosphate-DNA complex was allowed to form
gradually for 18 h. Cells were then washed twice with Dulbecco’s modified
Eagle’s medium, refed, and incubated for an additional period of 24 h before
treatment with toxins. Preliminary experiments were performed with 96-well
tissue culture plates to determine the concentration of each toxin yielding 100%
of cells rounding within 3 h at 37?C. The doses found were 16 ?g of C. difficile
toxin A per ml, 500 pg of C. difficile toxin B per ml, 15 ?g of C. sordellii LT per
ml with addition of 100 ?M dithiothreitol (37), and 100 ng of component Sa plus
200 ng of component Sb of C. spiroforme toxin per ml. The concentration of
cytochalasin D used was 0.625 ?g/ml. Transfected cells were washed twice with
Dulbecco’s modified Eagle’s medium and incubated for 3 h with the indicated
toxins before fixation.
Immunofluorescence. The presence of the expressed epitope-tagged proteins
was determined by indirect immunofluorescence with monoclonal antibody
P5D4 directed against the VSV G epitope. After fixation by 4% paraformalde-
hyde, monolayers were washed three times with phosphate-buffered saline
(PBS), free aldehyde groups were quenched by incubation with 50 mM NH4Cl
for 10 min, and the monolayers were washed three times in PBS. Cells were then
permeabilized for 5 min at room temperature in PBS containing 0.2% Triton
X-100 and 0.2% bovine serum albumin (BSA) (buffer A). Monolayers were
incubated for 30 min at room temperature with the first appropriate antibody.
The coverslips were washed extensively in buffer A, and the primary antibody
binding was detected by a 30-min incubation at room temperature with Texas
red-conjugated sheep anti-mouse antibody (Amersham, Les Ulis, France), to-
gether with fluorescein isothiocyanate (FITC)-phalloidin (1 ?g/ml; Sigma Chem-
ical Co., L’Isle d’Abeau, France) to visualize F-actin. After three washes in buffer
A followed by three washes in PBS, the coverslips were mounted in Mowiol
(Calbiochem, La Jolla, Calif.) and observed for fluorescence.
ADP-ribosylation. Exponentially growing HeLa cells were seeded at 2 ? 105
cells per 35-mm tissue culture dish 24 h prior to treatment with toxins. After
incubation with different concentrations of toxins, leading to different stages of
morphological changes, the cells were washed twice with PBS and detached from
their support with a rubber policeman in 20 mM N-2-hydroxyethylpiperazine-
N?-2-ethanesulfonic acid (HEPES) buffer (pH 8.0) in the presence of protease
inhibitors (10 ?M leupeptin, 1 ?M pepstatin, 250 ?M phenylmethylsulfonyl
fluoride). Harvested cells were lysed by three cycles of freezing and thawing and
centrifuged at 10,000 ? g for 5 min at 4?C. The concentrations of the cytosolic
proteins were determined by the Bradford assay (Bio-Rad, Paris, France) with
BSA as the standard. ADP-ribosylation was carried out with samples containing
20 ?g of total proteins in a final volume of 15 ?l in 20 ?M HEPES buffer (pH
8.0), incubated for 1 h at 37?C with 5 ?l of a radioactive solution containing 2.5
?M [32P]NAD (20,000 d.p.m./pmol; NEN-Du Pont de Nemours, Dreieich, Ger-
many), 2 mM AMP, and 4 ?g of exoenzyme C3 per ml. The reaction was
terminated by addition of 5 ?l of sodium dodecyl sulfate-polyacrylamide gel
electrophoresis sample buffer. After being boiled for 5 min, the samples were
subjected to electrophoresis on a 15% polyacrylamide gel. The gel was stained
with Coomassie brilliant blue and dried. The results of [32P]ADP-ribosylation
were evaluated by autoradiography.
RESULTS
Effects of C. difficile toxin B on transfected RhoA, RhoB, and
RhoC HeLa cells. Microinjection of an activated form of RhoA
(RhoA Val14) into subconfluent mouse Swiss 3T3 cells induces
both a change in cell morphology and the formation of a dense
microfilament network (36). The intracellular localization of
RhoA, RhoB, and RhoC in Rat-2 and MDCK cells has been
studied by a nuclear microinjection approach with eukaryotic
expression vectors containing c-myc epitope-tagged RhoA,
RhoB, and RhoC cDNAs (1). From this work, it was demon-
strated that the majority of RhoA and RhoC was cytosolic but
that RhoB was associated with intracellular compartments
identified as early and late endosomes (36). In the present
work, we have used the VSV G epitope to label RhoA, RhoB,
and RhoC and have introduced, by calcium phosphate trans-
fection, eukaryotic expression vectors containing the different
VSV G-tagged Rho cDNAs. We found that the choice of the
eukaryotic vector was important to obtain transfected cells
with acceptable morphology. For instance, vectors in which the
Rho cDNAs were under the control of a very strong promoter
such as the cytomegalovirus promoter did not result in trans-
fected cells with interpretable morphologies (18a).
There is good evidence suggesting that C. difficile toxin B
inhibits Rho GTPases (23, 24). We therefore tested the effects
of C. difficile toxin B on HeLa cells transfected with cDNAs of
the different Rho GTPases. A 3-h incubation with 125 pg of
toxin B per ml induced a cytopathic effect of toxin B in non-
transfected HeLa cells (or HeLa cells transfected with the
parent vector pKC3) (Fig. 1A and B). In contrast, cells trans-
fected with VSV G-RhoA were unaffected by treatment with
500 pg of toxin B per ml for the same period. Companion
nontransfected RhoA cells, however, exhibited the classical
FIG. 1. Effects of C. difficile A and B toxins on HeLa cells transfected with RhoA. HeLa cells were transfected with RhoA cDNA, as described in Materials and
Methods. (A and C) Cells were stained for F-actin with FITC-phalloidin. (B and D) Cells corresponding to those in panels A and C, respectively, were stained with
the monoclonal antibody P5D4 followed by Texas red-labeled anti-mouse antibody. (A and B) Effects of 500 pg of toxin B per ml for 3 h on RhoA-transfected and
nontransfected HeLa cells. (C and D) Effects of 16 ?g of toxin A per ml on RhoA-transfected HeLa cells. Large arrows show the transfected cells which are resistant
to toxins A and B; small arrows show toxin-affected nontransfected cells; arrowheads show stress fibers in toxin-resistant cells. Thin arrows in panels C and D show a
cell transfected with RhoA expressing a small amount of the GTPase (D) and sensitive to toxin A (C).
4064 GIRY ET AL.INFECT. IMMUN.

Page 3

VOL. 63, 1995Rho AND LARGE CLOSTRIDIAL CYTOTOXINS4065

Page 4

4066 GIRY ET AL.INFECT. IMMUN.

Page 5

morphologies of toxin B-treated cells. An identical result was
found for cells transfected with RhoB and RhoC (Fig. 2). We
must point out that cells transfected with the Rho GTPases
were not ‘‘resistant’’ to toxin B. Indeed, after a prolonged
incubation in the presence of toxin B (usually 18 h with 500 pg
of toxin B per ml), cells transfected with the different Rho
GTPases were finally affected by toxin B. Also, a low level of
the GTPase expression in transfected cells did not fully protect
them against toxin B, as shown in Fig. 2C and D. Therefore,
transfection of Rho GTPases into HeLa cells decreased the
sensitivity of these cells to toxin B, as expected if Rho was the
intracellular target of this bacterial protein.
Toxin B treatment of cells appears to result in an unidenti-
fied covalent modification of Rho proteins. Lysates of HeLa
cells treated with increasing amounts of toxin B contained Rho
proteins that were increasingly resistant to ADP-ribosylation
by exoenzyme C3 (Fig. 3).
Effects of C. difficile toxin A on transfected RhoA, RhoB, and
RhoC HeLa cells. C. difficile toxin A induces morphological
effects in cultured cells, and these effects are close to those
caused by C. difficile toxin B (15). However, doses of toxin A
required for cytotoxicity are much higher than those of toxin B
(reviewed in reference 16). The lowest dose of toxin A found
to be sufficient to induce a cytopathogenic effect on HeLa cells
within 3 h was 4 ?g/ml. In contrast, HeLa cells transfected with
RhoA and subsequently incubated with 16 ?g of C. difficile
toxin A per ml did not show cytopathic effects (Fig. 1C and D).
RhoB and RhoC were also effective in protecting HeLa cells
against toxin A (data not shown). As shown in Fig. 1C and D,
a low level of RhoA expression in HeLa cells did not induce
any protection against toxin A. As observed for toxin B, pro-
longed incubation (18 h with 16 ?g of toxin A per ml) did affect
transfected Rho cells (data not shown).
As shown in Fig. 3, Rho proteins present in cytosolic extracts
of toxin A-treated HeLa cells were ADP-ribosylated by exoen-
zyme C3 to a much lesser extent than were those present in
extracts from untreated cells.
We then verified the effects of cytochalasin D and C. spiro-
forme toxin on the F-actin cytoskeleton (38) in rho-transfected
cells. These two agents are known to disrupt the F-actin-asso-
ciated cell structures by a Rho-independent mechanism. Cy-
tochalasin D is a drug which caps the barbed ends of F-actin
filaments, thus decreasing the rate of filament assembly (39),
whereas C. spiroforme toxin, like C. botulinum C2 toxin, ADP-
ribosylates nonmuscular G-actin, thereby blocking F-actin fil-
ament formation (2, 51). As shown in Fig. 4, Rho GTPases did
not afford protection against the effects of cytochalasin D and
C. spiroforme toxin.
Effects of C. sordellii LT on the HeLa cell cytoskeleton. C.
sordellii produces two toxins, LT and hemorrhagic toxin (4),
which have some similarities to the toxins elaborated by C.
difficile (31). Like toxins A and B, LT induces rounding of
cultured cells, with, however, a morphological aspect different
from the one shown by C. difficile toxin B (37). It was therefore
of interest to test the effects of LT on our system of transfected
cells to determine a possible role of the Rho GTPases in the
LT mechanism of action. As shown in Fig. 5, the cytopathic
effect induced by LT in HeLa cells consists of the rounding of
the cell body with reorganization of F-actin structures into
numerous cell surface microvilli and a loss of F-actin stress
fibers. Increasing the intracellular concentration of RhoA did
not protect the cells from the cytopathic effects of LT (Fig. 5).
We wondered whether the microvilli induced by LT contain
an actin-binding and bundling protein, fimbrin/plastin, which is
normally found in these structures (8). Using a polyclonal
rabbit antibody which recognizes the different isoforms of fim-
brin/plastin, we observed that a considerable amount of fim-
brin/plastin was present in microvilli induced by LT (Fig. 5).
Incubation of HeLa cells with concentrations of C. sordellii
LT or cytochalasin D that induced cytopathic effects did not
modify the subsequent in vitro ADP-ribosylation of Rho pro-
teins (Fig. 3).
DISCUSSION
Transfection of HeLa cells with cDNAs encoding epitope-
tagged RhoA, RhoB, and RhoC allowed us to establish an
experimental system for the study of the putative role of these
GTPases in the intracellular activities of several bacterial pro-
tein toxins.
Cells transfected with RhoA, RhoB, and RhoC GTPases
were then treated with several large clostridial cytotoxins: C.
difficile A and B toxins (47), C. sordellii LT (37), and C. spiro-
forme iota-like toxin (38) or cytochalasin D (12), known to act
on the F-actin cytoskeleton. Cells transfected with expression
vectors containing cDNAs coding for RhoA, RhoB, and RhoC
were able to maintain F-actin structures in the presence of high
doses of toxin A or B from C. difficile. Although we cannot
totally rule out that the intracellular overexpression of the
GTPases did not alter the ability of cells to take up toxin A or
B by endocytosis, this seem unlikely since C. spiroforme C2-like
binary toxin and C. sordellii LT were equally active on control
FIG. 3. Effect of C. difficile toxins A and B, C. sordellii LT, and cytochalasin
D on ADP-ribosylation of Rho. HeLa cells were incubated for 3 h with the
indicated amounts of the following toxins: control untreated cells (lane 2), toxin
B (lanes 3 to 5), toxin A (lane 6 and 7), C. sordellii LT (lane 8), and cytochalasin
D (lane 9). ADP-ribosylation of each cell lysate containing identical amounts of
proteins was then performed as described in Materials and Methods. ADP-
ribosylation of 100 pg of purified rhoC by exoenzyme C3 is shown in lane 1.
Molecular masses are given in kilodaltons.
FIG. 2. Effects of C. difficile toxin B on HeLa cells transfected with RhoB or RhoC. HeLa cells were transfected with the RhoB and RhoC cDNAs as described
in Materials and Methods. (A, C, and E) Cells were stained for F-actin with FITC-phalloidin. (B, D, and F) Cells were stained with monoclonal antibody P5D4 and
then with Texas red-labeled anti-mouse antibody. (A and B) Cells transfected with RhoB cDNA treated with 500 pg of toxin B per ml for 3 h. (C and D) Cells transfected
with RhoB cDNA with a moderate expression of the GTPase (thin arrows). The small arrow in panel C shows a nontransfected cell affected by toxin B, stained for
F-actin. Indeed, in the same cell, no staining can be seen with the anti-VSV G monoclonal antibody (D). (E and F) RhoC toxin-resistant transfected cells treated with
500 pg of toxin B per ml for 3 h.
VOL. 63, 1995Rho AND LARGE CLOSTRIDIAL CYTOTOXINS4067