interaction of clostridium difficile a with cultured cells ...rounding(c. fiorentini et al.,...

Vol. 58, No. 7INFECTION AND IMMUNITY, JUlY 1990, p. 2329-23360019-9567/90/072329-08$02.00/0Copyright C 1990, American Society for Microbiology

Interaction of Clostridium difficile Toxin A with Cultured Cells:Cytoskeletal Changes and Nuclear Polarization

C. FIORENTINI,' W. MALORNI,1* S. PARADISI,' M. GIULIANO,2 P. MASTRANTONIO,2 AND G. DONELLI'Departments of Ultrastructures' and Bacteriology and Medical Mycology,2 Istituto Superiore di Sanita,

Viale Regina Elena, 299, 00161 Rome, ItalyReceived 4 January 1990/Accepted 16 April 1990

Experiments done on in vitro-cultured cells exposed to toxin A from C. difficile showed a series ofcytopathologic changes leading to cell retraction and rounding accompanied by the marginalization of thenucleus, which localized at one pole of the cell. Cytoskeleton appeared to be strongly involved in suchmodifications. In particular, the microfflament system seemed to be involved in cell retraction, whilemicrotubule network integrity and function seemed to be necessary for the nuclear displacement. Thecarboxylic ionophore monensin completely blocked the cytopathic effect when added with the toxin. The serineprotease inhibitor chymostatin appeared to be protective also upon addition long after the end of the bindingstep. The Ca2+-dependent cytosolic protease inhibitors antipain and leupeptin were uneffective in protectingcells. Thus, our results suggest the involvement of an acidic compartment and the action of a serine proteasein toxin A-induced cytopathic effect.

Toxigenic strains of Clostridium difficile are recognized asthe major cause of antibiotic-associated pseudomembranouscolitis in humans (3). They produce at least two differenttoxins, designated toxins A and B, which are large, heat-labile cytotoxic proteins implicated in the pathogenesis ofthe disease (2). Toxin A elicits severe epithelial damageassociated with hemorrhage and fluid secretion when in-jected into rabbit ileal loops (38). These effects are notaccompanied by activation of adenylate cyclase (17), indi-cating that its mechanism of action differs from that choleratoxin (12) and Escherichia coli heat-labile enterotoxin, whichcause fluid secretion and stimulate intestinal adenylate cy-clase without producing an inflammatory infiltrate (32).Conversely, toxin B appears to be devoid of enterotoxicactivity, and its role in the pathogenesis of the disease is stillunclear (19).Both toxins A and B induce cytopathic effect (CPE) in

cultured cells (7, 16, 33) and belong to the group of intracel-lularly acting proteins which have to be internalized andprocessed to exert their cytotoxic activity (8, 11). In partic-ular, toxin B appears to modify the cytoskeleton, disaggre-gating actin filaments (40) and rearranging some actin-binding proteins (23). On the other hand, toxin A has beenshown to induce morphological changes in several culturedmammalian cells, mainly consisting of cell retraction androunding (C. Fiorentini et al., Microecol. Ther., in press). Ithas also been demonstrated to perturb the cytoskeleton (10)and to stimulate intracellular calcium release, directly alter-ing epithelial cell permeability without producing cell deathor disrupting the confluence of the monolayer (28).To better clarify the subcellular mechanisms underlying C.

difficile toxin cytotoxicity, we have planned an extensivestudy on in vitro-cultured cells different in origin and growthpattern. The present work focused on toxin A-induced CPEand was mainly concerned with an intracellular phenomenonnamed nuclear polarization. The mechanisms underlyingsuch a morphological change were investigated by immuno-cytochemical and electron microscopic approaches.

* Corresponding author.

MATERIALS AND METHODS

Cell cultures. Different human cell lines from cervix car-cinoma (HeLa), larynx carcinoma (HEp-2), mammary breastcarcinoma (CG5), epidermoid carcinoma (A431), and hepa-toma (Hep-G2) as well as a murine myogenic cell line (L8)were cultured at 37°C in Dulbecco modified Eagle mediumsupplemented with 10% fetal calf serum (Flow Laboratories,Inc., McLean, Va.), 1% nonessential amino acids, 5 mML-glutamine, penicillin (100 U/ml), and streptomycin (100,g/ml).For light microscopy, both control and treated cells were

grown on 13-mm-diameter glass cover slips in separatedwells (5 x 104 cells per well). For transmission electronmicroscopy, cells were subcultured in 25-cm2 Falcon plasticflasks at a density of approximately 104 cells per ml. Flasksand wells were placed in a 37°C incubator containing anatmosphere of 95% air and 5% CO2.Toxin purification. The procedure adopted for toxin puri-

fication was that described by Sullivan et al. (37). C. difficileVPI strain 10463 was grown in 3 liters of brain heart infusion(dialysis flasks) for 72 h at 37°C. The culture supernatant wascentrifuged at 8,000 x g for 10 min, filtered through amembrane filter (0.45-,um pore size; Millipore Corp., Bed-ford, Mass.), and concentrated by ultrafiltration at 4°C withan XM-100 membrane filter (Amicon Corp., Lexington,Mass.). The concentrated supernatant was loaded onto aDEAE Trisacryl M column (2.5 by 40 cm) which had beenequilibrated with 50 mM Tris hydrochloride (pH 7.5) andwas eluted with a 600-ml linear NaCl gradient (0.05 to 0.5 MNaCl).

Fractions (5 ml) were collected and assayed for cytotox-icity with HEp-2 cells. For enterotoxicity, the rabbit ilealloop test was performed as previously described (19).

Cytotoxic fractions eluting at 0.1 M NaCl (toxin A) werefurther purified by acetic acid precipitation in acetate buffer(pH 5.5). When analyzed by nondenaturating gel electropho-resis, the toxin A preparation was homogeneous. The pro-tein concentration was 118 ,ug/ml as determined by themethod of Lowry et al. (18). The toxin was stored at -20°Cin 50 mM Tris hydrochloride (pH 7.5), and dilutions weremade in the same buffer.

2329

on April 1, 2021 by guest

http://iai.asm.org/

Dow

nloaded from

http://iai.asm.org/

2330 FIORENTINI ET AL.

Cell treatment. For light and electron microscopy, 24 hafter seeding, cell cultures were treated with 4 ,ug of toxin Aper ml. This toxin concentration caused the CPE in 100% ofepithelial cells (HeLa, HEp-2, CG5, A431) in 1 h and 100%of the other cell lines (Hep-G2, L8) in 3 h. Treatments wereperformed at different temperatures (4 or 370C).

Chemicals. Experiments with different chemical agentswere done with the maximum dose producing no cytopatho-logic changes in the cells; all the experiments were per-formed in triplicate. The following chemicals were used:trifluoperazine (TFP) (4.8 ,ug/ml), which has been demon-strated to be capable of antagonizing the calcium-bindingprotein calmodulin (29); drugs which interact with the cyto-skeletal components (31), such as cytochalasin B (0.48jig/ml), phalloidin (7.88 ,ug/ml), and demecolcine (0.26 ,ug/ml); a Na+/H+-exchanging ionophore, monensin (340 ,ug/ml)(25); Ca2+-dependent protease inhibitors (1, 9) leupeptin(100 ,ug/ml) and antipain (100 p,g/ml); and a serine proteaseinhibitor, chymostatin (400 pug/ml) (39). The stock solutionswere made in dimethyl sulfoxide, and monensin was dilutedin ethanol. For the experiments, final concentrations wereobtained directly in growth medium. Parallel treatments withvehicles alone (dimethyl sulfoxide and ethanol) were used ascontrols. All the chemicals were purchased from SigmaChemical Co. (St. Louis, Mo.).

Fluorescence microscopy. For microtubule-associated pro-teins (MAPs), Tau, filamin, vinculin, tubulin, and vimentindetection, cells grown on cover slips were fixed in 3.7%formaldehyde in phosphate-buffered saline (pH 7.4) for 10min at room temperature. After being washed in the samebuffer, the cells were permeabilized with 0.5% Triton X-100(Sigma) in phosphate buffer (pH 7.4) for 10 min at roomtemperature.For vinculin and a-tubulin labeling, monoclonal antibodies

directed against vinculin (Amersham Corp., ArlingtonHeights, Ill.) and a-tubulin (Amersham) were used. After 30min at 37°C, cells were washed in phosphate-buffered salineand then incubated with a sheep anti-mouse immunoglobulinG fluorescein-linked whole antibody (Sigma) for 30 min at37°C. For MAP and Tau detection, cells were incubated withrabbit anti-MAP and anti-Tau antibodies (Bioyeda), respec-tively, for 30 min at 37°C. After washing, an incubation witha goat anti-rabbit immunoglobulin G rhodamine-linkedwhole antibody (Sigma) was performed. For filamin andvimentin detection, cells were incubated with goat poly-clonal antifilamin (Sigma) and antivimentin (Sigma) antibod-ies at 37°C for 30 min. After washing, an incubation with arabbit anti-goat immunoglobulin G fluorescein-linked anti-body for 30 min at 37°C was performed.

For the detection of the surface antigens epidermal growthfactor receptor and p2-microglobulin, the light subunit ofhuman leucocyte antigen, the cells were incubated with thepolyclonal antibodies anti-epidermal growth factor receptor(Sigma) and anti-P2-microglobulin (Dako) at 4°C for 30 min.After being washed in phosphate buffer, cells were fixed in3.7% formaldehyde in the same buffer for 30 min at 4°C andincubated with a goat anti-rabbit immunoglobulin G rho-damine-linked antibody (Sigma) at 37°C for 30 min. Surfaceglycoproteins recognized by the lectin from Triticum vul-garis (wheat germ agglutinin) are mainly represented byproteins carrying N-acetylglucosamine carbohydrates. Spec-imen preparation with fluorescein-linked wheat germ agglu-tinin (direct fluorescence) was performed as stated above.

Finally, after washings, cover slips were mounted withglycerol-phosphate-buffered saline (2:1) and analyzed with aNikon Optiphot fluorescence microscope.

Transmission electron microscopy. Cells were fixed with2.5% glutaraldehyde in 0.05 M cacodylate buffer, postfixedwith 1% OS04 in the same buffer for 1 h, dehydrated throughgraded ethanols, and embedded in Agar 100 resin (AgarAids).

Ultrathin sections were stained with uranyl acetate andlead citrate and examined with a Zeiss EM 10 C electronmicroscope.

RESULTS

Morphological features of intoxicated cells. As a generalrule, morphological effects of toxin A on the different celllines considered here are represented by (i) cell retraction,(ii) cell rounding, and (iii) displacement of the nucleustoward one pole of the cell. In the present study, a dose oftoxin of 4 ,ug/ml showed a maximal effect of rounding up andnuclear polarization after 1 h in HeLa, HEp-2, CG5, andA431 cells and after 3 h in L8 and Hep-G2 cells. Taking intoaccount the strict overlapping among the results obtainedwith the different cell lines under different culture condi-tions, we report here only the data relative to HeLa cells,which can be assumed to be a representative model forcultured cell intoxication.When an ultrastructural analysis was performed, control

cells were characterized by a round or elongated nucleuswith two or more nucleoli and several mitochondria with anelectron-dense matrix (Fig. la). In toxin-treated cells under-going the retraction process (data not shown), no modifica-tion in the morphology of cell organelles was detected.However, at the final stage of the intoxication process, thenucleus appeared to be kidney shaped and polarized to onepole of the cell (Fig. lb). The rough endoplasmic reticulumwas slightly dilated, while mitochondria were localizedaround the Golgi apparatus, opposite to the nucleus, andshowed an unchanged general morphology but a rarefiedmatrix. The Golgi apparatus stalks appeared to be remark-ably increased, numerous vesicles being visible; bundles ofcytoskeletal filaments, probably microfilaments, were alsoobserved (Fig. lc). Moreover, when cells were maintained insuspension by continuous shaking, the exposure to the toxininduced nuclear polarization with the same ultrastructuralfeatures mentioned above.

Cytoskeletal changes induced by toxin treatment. The cy-toskeletal components underwent a different behavior aftertoxin A treatment. In the microfilament system, F-actin anda-actinin were localized only in the side opposite the nucleusin a patchily organized form, as previously described inHEp-2 cells (7). Filamin, normally localized together withat-actinin (Fig. 2a), displayed a diffuse positivity throughoutthe cytoplasm (Fig. 2b and c). Vinculin seemed to beunexpressed in intoxicated cells (Fig. 3b) compared withcontrol cells (Fig. 3a).The microtubule network, well visible in control cells (Fig.

4a), disappeared in the cytoplasm of treated cells. In fact, adiffuse positivity for a-tubulin was detectable in toxin-exposed cells (Fig. 4b), even though, by varying the focalplane of the microscope, a thin meshwork of microtubuleswas observed around the nucleus (Fig. 4c and d). Further-more, the MAPs and Tau, normally scattered in the cyto-plasm and mainly distributed near the nucleus, appeared tobe diffused throughout the cytoplasm of the roundish intox-icated cells (data not shown).

Finally, the intermediate filament vimentin, organized in acomplex network in untreated cells (Fig. 5a) was localized as

INFECT. IMMUN.


http://iai.asm.org/

Dow

nloaded from

http://iai.asm.org/

CYTOPATHOLOGIC EFFECTS INDUCED BY C. DIFFICILE TOXIN A

47

A.At- /(

l;.;.].-xp7*osi iX C *4k.-

Ici* a4rq4

FIG. 1. Transmission electron microscopy. Ultrastructural fea-tures of control (a) and toxin A-treated (b and c) HeLa cells. Nuclearpolarization appears evident after 1 h of incubation with 4 jig oftoxin A per ml (b). A kidney-shaped nucleus is localized at one poleof the cell and numerous mitochondria are scattered throughout thecytoplasm (b). Bundles of cytoskeletal elements and a well-devel-oped Golgi complex are visible at high magnification (c). Bars, 1 ,um.

a positive, condensed meshwork at the side opposite thenucleus in treated, rounded cells (Fig. 5b).

Surface antigen distribution after toxin treatment. Theeffects of toxin A on the distribution of some surfaceantigens were also analyzed in Triton X-100-untreated cul-tured cells by immunocytochemical methods. Different sur-face markers were labeled: the receptor for epidermalgrowth factor, which is expressed by the A431 and CG5epithelial cell lines; the light subunit of human leukocyteantigen (,32-microglobulin), which was evaluated in all hu-man cell lines examined; and the glycoproteins of the N-acetyl-D-glucosamine group, labeled by the wheat germ

- agglutinin lectin. All these markers were uniformly distrib-uted on the surfaces of control and treated cells (data notshown).

Effects of various compounds on toxin-induced CPE. Ex-periments with several agents able to perturb cell structureand function were done on different cell lines. However, aspreviously explained, we report here only the results ob-tained with HeLa cells. Pretreatment with TFP, a compoundcapable of antagonizing the calcium-binding protein calmod-ulin, did not protect cells against intoxication. However, thecharacteristic retraction of the cell body and the changes inactin microfilaments were strongly delayed in the presenceof this drug (Fig. 6a and b), even though after 3 h oftreatment with the toxin, the CPE almost completely devel-oped.Drugs which interact with the cytoskeletal components

were also tested (Table 1). Actin-depolymerizing or -stabi-lizing compounds such as cytochalasin B and phalloidinwere uneffective in protecting cells from toxin A. On theother hand, when cells were treated simultaneously withtoxin A and the microtubule-depolymerizing agent demecol-cine (Table 1), they normally retracted the cell body andbecame round, but the nucleus did not migrate (95%) towardone pole of the cell (Fig. 7a and b).Monensin, which is able to raise the pH in intracellular

acidic vesicles, was highly protective when added with thetoxin at 37°C. In fact, only 5% of cells treated this wayunderwent rounding and nuclear polarization (Table 1).

..4t However, monensin was not protective when the cells weretreated with a mixture of monensin-toxin at 4°C (to allow

%*;* toxin binding but not entry) and then maintained in freshA medium at 37°C. In this case, 90% of cells showed the


44~~~~~~~~~~~~~~~~~~~~~~~~~4

Ailk~A

FIG. 2. Immunofluorescence. Control (a) and toxin A-treated (b and c) HeLa cells. The actin-binding protein filamin appears to be diffusedthroughout the cytoplasm in treated cells (b and c [corresponding brightfield]). Bar, 5 p.m.

statin indicated that this drug appeared to be protective until30 min after the binding step (Table 2).

DISCUSSION

The effects of toxin A from C. difficile on cultured cellsseem to be mainly represented by a cascade of eventsleading to cell rounding, nuclear polarization, and cell death.In fact, three different phases can be detected in toxin Aintoxication (Table 3). In the first phase, cell retraction,changes in the adhesion pattern and integrity of the cytoskel-etal apparatus were observed; in the second phase, cellrounding, nuclear polarization and peculiar cytoskeletalchanges were visible; finally, in the third phase, cells under-going necrosis showed two or more kidney-shaped nucleiand a disrupted cytoskeleton.

It is well known that the cell shape and the polarity of thecell as well as the positioning of intracellular organelles arestrongly dependent on the cytoskeletal components (31).Toxin A, like other bacterial toxins (6, 26, 27, 30, 34) and C.difficile toxin B (20, 23), has been described to perturb thecytoskeleton (Fiorentini et al., in press), causing changes inF-actin (7, 10) and a-actinin (7) distributions. Strikingly,experiments reported here with microfilament system-per-turbating agents such as cytochalasin B and phalloidin,which, respectively, depolymerize or stabilize F-actin fila-

ments (31), did not appear to be able to protect cells fromboth retraction and nuclear polarization. This could be dueto an effect of the toxin on actin-binding proteins playing arole in the dynamic modifications of microfilament systemrather than on F-actin itself. In fact, some plaque adhesionproteins like a-actinin, which acts as a cross-linking mole-cule, and vinculin, which connects stress fibers to theextracellular matrix, showed an altered pattern after toxintreatment. Moreover, cells seeded in toxin-containing me-dium failed to spread out but were still able to adhere to thesubstratum and displayed nuclear polarization. This canprobably be related to the dependence of cell spreading onactin microfilaments and on the organization of stress fibers(4, 36). Furthermore, the organization of actin in a three-dimensional network is strongly dependent on the distribu-tion of a-actinin and filamin, two gel-forming proteins (31).After exposure to toxin A, a-actinin appeared to be distrib-uted together with actin, while filamin showed a diffusepattern in the cytoplasm. These findings suggest an impair-ment of the relationships between actin and some actin-related proteins, which can also explain the homogeneoussurface distribution of the surface protein epidermal growthfactor receptor, known to be actin cytoskeleton related (41),in toxin-treated cells. In addition, the contraction-producingprotein myosin, which is associated with actin filaments in

FIG. 3. Immunofluorescence. Control (a) and toxin A-treated (b) HeLa cells after immunostaining with antivinculin antibodies. Theantigen appears to be unexpressed in intoxicated cells (b). Bar, 5 pum.

INFECT. IMMUN.


http://iai.asm.org/

Dow

nloaded from

http://iai.asm.org/


I. J^

%k'.

..

vI" ",~.':<

Ac 4dFIG. 4. Immunofluorescence. Control (a) and toxin A-treated (b-d) HeLa cells after immunostaining with anti-a-tubulin antibodies. The

antigen appears to be diffused in treated cells. However, a thin microtubular cage is observable in the perinuclear region (c and d[corresponding brightfield]). Bar, 5 ,um.

stress fibers (15), has been shown to be modified by toxin Aaction at an early step of the intoxication process (Fiorentiniet al., in press). This protein is known to be regulated by acalcium-calmodulin-dependent protein kinase (29, 31), andthe perturbating effect of TFP on cell retraction can supportthe hypothesis of an involvement of myosin in this step ofintoxication. Thus, our results seem to indicate that therelationships between the different components of the micro-

filament system are strongly involved in the early steps ofthe intoxication process.The microtubular apparatus has been shown not to be

involved in the early cell response to toxin treatment (7). Atfinal stages, it showed a diffuse pattern in the cytoplasm,which is consistent with the homogeneous distribution ofMAPs, proteins known to bind to microtubules and tocross-link them to other cell structures (31). On the other

FIG. 5. Immunofluorescence. Control (a) and toxin A-treated (b) HeLa cells after immunostaining with antivimentin antibodies. Aftertreatment, vimentin filaments appear to be localized in the opposite side with respect to the nucleus. Bar, 5 ,.m.

VOL. 58, 1990 2333

(ipI It* g -.

'I. -4

0-" 0. N

w. 'o..

t e %..? 4

-tl. II i.01fIlp- .1.: .1, ,


http://iai.asm.org/

Dow

nloaded from

http://iai.asm.org/


FIG. 6. Fluorescence microscopy. TFP-treated (a) and TFP- and toxin A-treated (b) Hep-G2 cells. Actin staining. No morphologicalmodifications are observable after 1 h of incubation with the toxin. Bar, 5 ,um.

hand, microtubules appeared to be still organized in a thinnetwork around the polarized nucleus, where Tau protein,known to promote the assembly of microtubules and addi-tionally to stabilize the resulting polymers (5), was mainlycondensed. However, the inhibitory effect on nuclear dis-placement exherted by demecolcine, a microtubule-depoly-merizing agent (42), suggested that integrity of the microtu-bular apparatus is required for the polarization of thenucleus.

TABLE 1. Effects of various compounds on CPE'

Compound Concn % RC % PN

Cytochalasin B 0.48 100 100Phalloidin 7.89 100 100Demecolcine 0.26 100 5Monensin 340 5 5Chymostatin 400 0 0Leupeptin 100 100 100Antipain 100 100 100

a Compounds were added together with the toxin at 37'C for 3 h. Theresults in both columns are expressed as percentages of rounding cells (RC)and cells with a polarized nucleus (PN) compared with the same effects incontrol cultures treated with the toxin but without the tested compounds. Incultures treated with toxin A alone, the CPE was observed after 1 h.

The role of intermediate filaments in the intoxicationprocess induced by toxin A remains to be defined. In fact,keratin did not show any change after toxin treatment anddisplayed a well-defined network in rounded polarized cells,probably because of its structural role (7), while vimentinappeared to be colocalized with actin and ot-actinin at theside opposite the nucleus. Accordingly, it has recently beenshown that intermediate filaments are packed near the nucleiof intoxicated CHO cells (14). Because of the known rela-tionships between intermediate filaments and the nuclearenvelope (13), this "crowding" of cytoskeletal componentscould be related to the observed modification in nuclearshape, with the nucleus becoming kidney shaped in almostall intoxicated cells. Furthermore, 11-nm filaments, probablyrelated to the intermediate filament-type lamins, have beendetected in the nucleoplasm of treated cells (14). The authors(14) suggested that the disorganization in lamins induced bytoxin A could eventually affect cell mitosis and lead to celldeath.

It has been demonstrated that an acidic compartment isneeded in the toxin A activation pathway from the cellexterior to the cytosol (11). In fact, monensin, its majoreffects being to dissipate proton gradients (25) and to protectcells against a number of endocytosed toxins (22), seemed toinhibit the CPE induced by toxin A. The time course for

,, A',7" .3l

.0...

FIG. 7. Light microscopy. Demecolcine-treated (a) and demecolcine- and toxin A-treated (b) cells. When intoxication followeddemecolcine pretreatment, cell retraction occurred but without nuclear displacement. Bar, 5 ,um.

INFECT. IMMUN.

R; :1. .- nr!;!:.


http://iai.asm.org/

Dow

nloaded from

http://iai.asm.org/


TABLE 2. Time course for prevention of toxin-induced CPE bymonensin and chymostatina

Time Monensin Chymostatininterval(min) % RC % PN % RC % PN

0 20 20 0 010 89 89 0 420 95 95 3 930 98 98 7 3345 100 100 94 9660 100 100 95 98

a Cells were treated with the toxin (4 pLg/ml) for 30 min at 4°C. The toxinsolution was then replaced by growth medium, and cells were incubated at37'C. Monensin and chymostatin were added after the time intervals indicatedabove. The percentages of rounding cells (RC) and cells with a polarizednucleus (PN) were evaluated ater 3 h of treatment. Cells treated with toxin Awithout the addition of chemicals displayed 100%o RC and PN and wereconsidered controls.

prevention of toxin-induced CPE by monensin is consistentwith an involvement of Golgi apparatus in the internalizationprocess (24) as previously hypothesized by Henriques et al.(11).

Furthermore, several findings indicate that at some stageof the intoxication process, a cytoplasmic protease is essen-tial for the activity of toxin A. This toxin could possess, likeClostridium botulinum C2 (21) and Clostridium spiroforme(26, 35) toxins, a subunit which is required for its penetrationin the cytosol and which must undergo limited proteolysis tobe functionally active. Chymostatin, a serine protease inhib-itor (39), seemed to block the action of toxin A. It appearedto be capable of inhibiting the binding of the toxin to the cellsurface, and it was protective upon addition long after theend of the binding step. In the first case, chymostatin couldinhibit the action of a membrane-bound protease whichactivates the toxin and allows it to penetrate into the cell.Henriques et al. (11) showed that membrane-bound toxin Acan be transferred to the cytosol in active form directlyacross the membrane when exposed to a low pH. The lowpH could alter the conformation of the toxin, allowing amembrane-bound protease to activate the toxin. On theother hand, the inhibitory effect induced by chymostatinafter the binding step could be explained by an involvementof an enzyme which is present on the cell surface and insome subcellular compartment too. In fact, a number ofmembrane-bound proteases could be found in the Golgicomplex, which seems to be involved in the toxin A inter-nalization process (11). However, Henriques et al. (11) also

TABLE 3. Different phases of morphological modificationsinduced by toxin A

Characteristic Phase 1 (cell Phase 2 (cell Phase 3 (cellor system retraction) rounding) death)

General morphology Polygonal Roundish RoundMultinucleation Rare Rare FrequentBlebbing-apoptosis Absent Absent AbsentAdhesion pattern Altered AlteredSurface morphology Unchanged Unchanged SmoothMitochondria Unchanged Unchanged DestroyedRough endoplasmic Unchanged Dilated Swollen

reticulumGolgi complex Unchanged Increased SwollenNucleus Unchanged Polarized PolarizedNuclear shape Unchanged Unchanged Kidney shapedCytoskeleton Rearranged Deranged Disrupted

found that chymostatin did not have a protective effect ontoxin A-treated fibroblasts. This discrepancy with our resultscould be due to differences among the different cell types andin particular could depend on differences in inhibition oflysosomal and cytosolic proteases, on the different relation-ships that different cell lines establish with the substrate, aswell as on differences in the organization of the cytoskele-ton.

In conclusion, it appears that cell rounding and nuclearpolarization induced by toxin A from C. difficile are separatephenomena which could depend on the microfilament sys-tem and on the microtubular apparatus, respectively. Fur-thermore, an acidic compartment and the action of a serineprotease seem to be essential for the activity of toxin A.Presently, it is unknown how toxin A causes its irreversibleeffect leading to cell death. However, the cytotoxicity in-duced by the toxin seems not to involve a nonlysosomalproteolytic system in that the Ca2"-dependent cytosolicprotease inhibitors, leupeptin and antipain, did not show anyprotective effect against the toxin. Eventually, toxin A couldrepresent a useful tool for cell biology studies in the samemanner as cholera and pertussis toxins. The understandingof its mode of action can give further information either onbasic cytoskeletal functions or on the pathobiology ofpseudomembranous colitis.

LITERATURE CITED1. Baldassarre, J. J., S. Bakshian, M. A. Knipp, and G. J. Fisher.

1985. Inhibition of fibrinogen receptor expression and serotoninrelease by leupeptin and antipain. J. Biol. Chem. 260:10531-10535.

2. Banno, Y., T. Kobayashi, H. Kono, K. Watanabe, K. Ueno, andY. Nozawa. 1984. Biochemical characterization and biologicactions of two toxins from C. difficile. Rev. Infect. Dis. 6:11-20.

3. Bartlett, J. G., A. B. Onderdonk, R. L. Cisneros, and D. L.Kasper. 1977. Clindamycin-associated colitis due to a toxin-producing species of Clostridium in hamsters. J. Infect. Dis.136:701-705.

4. Connor, C. G., R. C. Brady, and B. L. Brownstein. 1981.Trifluoperazine inhibits spreading and migration of cells inculture. J. Cell. Physiol. 108:299-307.

5. Drubin, D. G., and M. W. Kirschner. 1986. Tau protein functionin living cells. J. Cell Biol. 103:2739-2746.

6. Fiorentini, C., G. Arancia, A. Caprioli, V. Falbo, F. M. Ruggeri,and G. Donelli. 1988. Cytoskeletal changes induced in HEp-2cells by the cytotoxic necrotizing factor of Escherichia coli.Toxicon 26:1047-1056.

7. Fiorentini, C., G. Arancia, S. Paradisi, G. Donelli, M. Giuliano,F. Piemonte, and P. Mastrantonio. 1989. Effects of Clostridiumdifficile toxins A and B on cytoskeleton organization in HEp-2cells: a comparative morphological study. Toxicon 27:1209-1218.

8. Florin, I., and M. Thelestam. 1983. Internalization of Clostrid-ium difficile cytotoxin into cultured human lung fibroblasts.Biochim. Biophys. Acta 763:383-392.

9. Grinde, B. 1983. Role of Ca for protein turnover in isolated rathepatocytes. Biochem. J. 216:529-536.

10. Hecht, G., C. Pothoulakis, J. T. LaMont, and J. L. Madara.1988. Clostridium difficile toxin A perturbs cytoskeletal struc-ture and tight junction permeability of cultured human intestinalepithelial monolayers. J. Clin. Invest. 82:1516-1524.

11. Henriques, B., I. Florin, and M. Thelestam. 1987. Cellularinternalization of Clostridium difficile toxin A. Microb. Pathog.2:455-463.

12. Kaslow, H. R., V. E. Groppi, M. E. Abood, and H. R. Bourne.1981. Cholera toxin can catalyze ADP-ribosylation of cytoskel-etal proteins. J. Cell Biol. 91:410-413.

13. Katzuma, Y., S. H. H. Swierenga, N. Marceau, and S. W.French. 1987. Connections of intermediate filaments with thenuclear lamina and the cell periphery. Biol. Cell. 59:193-204.

VOL. 58, 1990 2335


http://iai.asm.org/

Dow

nloaded from

http://iai.asm.org/


14. Kushnaryov, V. M., and J. J. Sedmak. 1989. Effect of Clostrid-ium difficile enterotoxin A on ultrastructure of Chinese hamsterovary cells. Infect. Immun. 57:3914-3921.

15. Langanger, G., M. Moeremans, G. Daneels, A. Sobieszek, M. DeBrabander, and J. De Mey. 1986. The molecular organization ofmyosin in stress fibers of cultured cells. J. Cell Biol. 102:200-209.

16. Lima, A. A. M., D. M. Lyerly, T. D. Wilkins, D. J. Innes, andR. L. Guerrant. 1988. Effects of Clostridium difficile toxins Aand B in rabbit small and large intestine in vivo and on culturedcells in vitro. Infect. Immun. 56:582-588.

17. Lonnroth, I., and S. Lange. 1983. Toxin A of C. difficile:production, purification and effect in mouse intestine. ActaPathol. Microbiol. Scand. 91:395-400.

18. Lowry, 0. H., N. J. Rosebrough, A. L. Farr, and R. J. Randall.1951. Protein measurement with the Folin phenol reagent. J.Biol. Chem. 193:265-275.

19. Lyerly, D. M., D. E. Lockwood, S. H. Richardson, and T. D.Wilkins. 1982. Biological activities of toxins A and B of Clos-tridium difficile. Infect. Immun. 35:1147-1150.

20. Mitchell, M. J., B. E. Laughon, and S. Lin. 1987. Biochemicalstudies on the effect of Clostridium difficile toxin B on actin invivo and in vitro. Infect. Immun. 55:1610-1615.

21. Ohishi, I., and M. Miyaki. 1985. Binding of the two componentsof C2 toxin to epithelial cells and brush border of mouseintestine. Infect. Immun. 48:769-775.

22. Olsnes, S., and K. Sandvig. 1985. Entry of polypeptide toxinsinto animal cells, p. 195-234. In I. Pastan and M. C. Willingham(ed.), Endocytosis. Plenum Publishing Corp., New York.

23. Ottlinger, M. E., and S. Lin. 1988. Clostridium difficile toxin Binduces reorganization of actin, vinculin and talin in culturedcell. Exp. Cell Res. 174:215-229.

24. Pastan, I., and M. C. Willingham. 1985. The pathway ofendocytosis, p. 29-37. In I. Pastan and M. C. Willingham (ed.),Endocytosis. Plenum Publishing Corp., New York.

25. Pohlmann, R., S. Kruger, A. Hasilik, and K. Von Figura. 1984.Effect of monensin on intracellular transport and receptor-mediated endocytosis of lysosomal enzymes. Biochem. J. 217:649-658.

26. Popoff, M. R., F. W. Milward, B. Bancillon, and P. Boquet.1989. Purification of the Clostridium spiroforme binary toxinand activity of the toxin on HEp-2 cells. Infect. Immun.57:2462-2469.

27. Popoff, M. R., E. J. Rubin, D. M. Gill, and P. Boquet. 1988.Actin-specific ADP-ribosyltransferase produced by a Clostrid-ium difficile strain. Infect. Immun. 56:2299-2306.

28. Pothoulakis, C., R. Sullivan, D. A. Melnick, G. Triadafilopoulos,A. S. Gadenne, T. Meshulam, and J. T. La Mont. 1988. Clos-

tridium difficile toxin A stimulates intracellular calcium releaseand chemotactic response in human granulocytes. J. Clin.Invest. 81:1741-1745.

29. Nayler, W. G. 1982. Calcium antagonists: classification andproperties, p. 1-16. In R. G. Rahwan and D. T. Witiak (ed.),Calcium regulation by calcium antagonists. American ChemicalSociety, Washington, D.C.

30. Reuner, K. H., P. Presek, C. B. Boschek, and K. Aktories. 1987.Botulinum C2 toxin ADP-rybosylates actin and disorganizes themicrofilament network in intact cells. Eur. J. Cell Biol. 43:134-140.

31. Schliwa, M. 1985. The cytoskeleton: an introductory survey, p.14-66. Springer-Verlag, Vienna.

32. Shafer, D. F., W. D. Lust, B. Sircar, and N. D. Goldberg. 1970.Elevated concentration of adenosine 3':5'-cyclic monophos-phate in intestinal mucosa after treatment with cholera toxin.Proc. Natl. Acad. Sci. USA 67:851-856.

33. Shahrabadi, M. S., L. E. Bryan, and P. W. K. Lee. 1984.Interaction of Clostridium difficile toxin A with L cells inculture. Can. J. Microbiol. 30:874-883.

34. Sharpe, A. H., L. B. Chen, J. R. Murphy, and B. N. Fields. 1980.Specific disruption of vinculin filament organization in monkeykidney CV-1 cells by diphtheria toxin, exotoxin A, and cyclo-heximide. Proc. Natl. Acad. Sci. USA 77:7267-7271.

35. Simpson, L. L., B. G. Stiles, H. Zepeda, and T. D. Wilkins. 1989.Production by Clostridium spiroforme of an iota-like toxin thatpossesses mono(ADP-ribosyl)transferase activity: identificationof a novel class of ADP-ribosyltransferases. Infect. Immun.57:255-261.

36. Spooner, B. S. 1973. Cytochalasin B: toward an understandingof its mode of action. Dev. Biol. 35:13-18.

37. Sullivan, N. M., S. Pellet, and T. D. Wilkins. 1982. Purificationand characterization of toxins A and B of Clostridium difficile.Infect. Immun. 35:1032-1040.

38. Taylor, N. S., G. M. Thorne, and J. G. Bartlett. 1981. Compar-ison of two toxins produced by Clostridium difficile. Infect.Immun. 34:1036-1043.

39. Troll, W., R. Wiesner, and K. Frenkel. 1987. Anticarcinogenicaction of protease inhibitors. Adv. Cancer Res. 49:265-284.

40. Wedel, N., P. Toselli, C. Pothoulakis, B. Faris, P. Oliver, C.Franzblau, and T. La Mont. 1983. Ultrastructural effects ofClostridium difficile toxin B on smooth muscle cells and fibro-blasts. Exp. Cell Res. 148:413-422.

41. Wiegant, F. A. C., F. J. Blok, L. H. K. Defize, W. A. M.Linnemans, A. J. Verkley, and J. Boonstra. 1986. Epidermalgrowth factor receptors associated to cytoskeletal elements ofepidermoid carcinoma (A431) cells. J. Cell Biol. 103:87-94.

42. Wilson, L. 1975. Action of drugs on microtubules. Life Sci.17:303-310.

INFECT. IMMUN.


http://iai.asm.org/

Dow

nloaded from
http://iai.asm.org/

interaction of clostridium difficile a with cultured cells ...rounding(c. fiorentini et al.,...

Documents