Clostridial Glucosylating Toxins Enter Cells via Clathrin-Mediated Endocytosis

Clostridium difficile toxin A (TcdA) and toxin B (TcdB), C. sordellii lethal toxin (TcsL) and C. novyi α-toxin (TcnA) are important pathogenicity factors, which represent the family of the clostridial glucosylating toxins (CGTs). Toxin A and B are associated with antibiotic-associated diarrhea and pseudomembraneous colitis. Lethal toxin is involved in toxic shock syndrome after abortion and α-toxin in gas gangrene development. CGTs enter cells via receptor-mediated endocytosis and require an acidified endosome for translocation of the catalytic domain into the cytosol. Here we studied the endocytic processes that mediate cell internalization of the CGTs. Intoxication of cells was monitored by analyzing cell morphology, status of Rac glucosylation in cell lysates and transepithelial resistance of cell monolayers. We found that the intoxication of cultured cells by CGTs was strongly delayed when cells were preincubated with dynasore, a cell-permeable inhibitor of dynamin, or chlorpromazine, an inhibitor of the clathrin-dependent endocytic pathway. Additional evidence about the role of clathrin in the uptake of the prototypical CGT family member toxin B was achieved by expression of a dominant-negative inhibitor of the clathrin-mediated endocytosis (Eps15 DN) or by siRNA against the clathrin heavy chain. Accordingly, cells that expressed dominant-negative caveolin-1 were not protected from toxin B-induced cell rounding. In addition, lipid rafts impairment by exogenous depletion of sphingomyelin did not decelerate intoxication of HeLa cells by CGTs. Taken together, our data indicate that the endocytic uptake of the CGTs involves a dynamin-dependent process that is mainly governed by clathrin.

CGTs consist of at least four domains [5]. At the N-terminus, the glycosyltransferase domain is located [6], which modifies low molecular mass GTP-binding proteins of the Rho and/or Ras family by mono-O-glucosylation [7], [8] or mono-O-GlcNAcylation (atoxin) [9]. An adjacent cysteine protease domain releases the glucosyltransferase into the cytosol by autoproteolytic cleavage [10]. The middle portion of the toxins is considered to mediate membrane insertion during the translocation process and is probably responsible for pore formation in membranes. Finally, target cell binding is mainly mediated by the C-terminal domain, which is characterized by repetitive oligopeptides (CROPs) [11], [12].
CGTs enter cells by receptor-mediated endocytosis and require an acidic endosomal compartment for complete translocation of the enzyme moiety into the cytosol [13], [14], [15]. To date, only for toxin A binding sites at the cell surface have been described, namely carbohydrates, including the trisaccharide Gala1-3Galb1-4GlcNac or protein receptors like sucrase-isomaltase and the glycoprotein gp96 [16], [17], [18]. Far less is known about the endocytic mechanisms underlying the internalization of the clostridial glucosylating toxins.
Here we studied the endocytic processes that mediate cell internalization of the CGTs, using pharmacological substances and genetical approaches that impair certain endocytic pathways. We show that the route to intracellular compartments for this toxin family is mediated by a dynamin-dependent process governed by clathrin. Our study additionally excludes the involvement of lipid rafts during clathrin-dependent uptake of the CGTs.

Uptake of CGTs into cells depends on dynamin
The GTPase dynamin is involved in the pinch-off of endocytic vesicles from the plasma membrane. Therefore, dynamin-dependency confines the endocytic uptake mechanism for a given molecule to clathrin-, caveolae-and RhoA-mediated pathways [20]. To test whether internalization of CGTs requires dynamin, dynasore, a potent cell-permeable inhibitor of dynamin [22], [23], was preincubated with HeLa cells, prior to addition of toxins. Diphtheria toxin that is endocytosed via clathrin-coated pits in a dynamin-dependent manner [24] was used as a positive control. Accordingly, the cytopathic effect of diphtheria toxin on HeLa cells was strongly inhibited when cells were pretreated with dynasore (Fig. 1A). Intoxication of HeLa cells with CGTs leads to cell rounding, due to the inactivation of Rho proteins, which regulate the actin cytoskeleton as well as microtubule-based structures [25]. Dynasore conferred resistance towards cell rounding in HeLa cells incubated with the prototypic member of the CGT family, C. difficile toxin B (Fig. 1B). Interestingly, cell rounding induced by toxin B, toxin A, lethal toxin and a-toxin was equally reduced to ,10% in dynasore-pretreated cells, when compared with non-pretreated cells (85-90% cell rounding) (Fig. 1C). The importance of dynamin in the uptake of CGTs was also tested with toxin B in the human colon adenocarcinoma cell line HT-29 (Cell Line Services, Eppelheim, Germany). Toxin B-induced intoxication was monitored by analysis of the glucosylation status of Rac1 in HT-29 cell lysates, using a specific anti-Rac1 antibody that recognizes only non-glucosylated Rac1. As expected, glucosylation of Rac1 in HT-29 cells treated with toxin B, was prevented by preincubation of cells with dynasore (Fig. 1D). Additional evidence about the importance of dynamin in the uptake of toxin B was obtained by intoxication of HeLa cells expressing dominant-negative dynamin (HA-dynamin K44A ). Toxin B-induced cell rounding was strongly inhibited in HA-dynamin K44A -transfected HeLa cells when compared to mocktransfected cells (Fig. 1E). Expression of HA-dynamin K44A was detected in HeLa cell lysates by the use of an anti-HA antibody (Fig. 1F). Taken together, data obtained in HeLa and/or HT-29 cells using dynasore or dominant-negative dynamin, strongly indicate the involvement of a dynamin-dependent endocytic pathway in the cellular uptake of CGTs.

Pharmacological inhibition of clathrin-coated pits formation prevents uptake of CGTs
Clathrin is a coat protein of specific endocytic vesicles, mediating the internalization of a wide range of transmembrane receptors and their ligands [26]. Since clathrin-dependent endocytosis requires dynamin, this endocytic pathway could be relevant for the uptake of CGTs that are sensitive to dynamin inhibition. To test the importance of clathrin in the endocytosis of CGTs, we used chlorpromazine, a drug that prevents formation of clathrin-coated pits at the plasma membrane [27]. HeLa cells preincubated with chlorpromazine showed strong resistance against cytopathic effects induced by the diphtheria toxin, which served again as a positive control ( Fig. 2A). This was not the case for CNF1, a Rho-deamidating toxin from pathogenic Escherichia coli leading to cell flatenning and polynucleation [28]. CNF1 enters cells by clathrin-independent mechanisms [29] and thereby confirms the selective inhibition of clathrin-dependent pathways by this drug (Fig. 2B). Importantly, chlorpromazine-pretreated HeLa cells were less sensitive to cell rounding induced by toxin B, toxin A, lethal toxin and a-toxin (,20-30% cell rounding), when compared with cells not pretreated with this drug (,75-85% cell rounding) ( Fig. 2C and 2D). The CGTs-inhibiting effect of chlorpromazine in HeLa cells was paradigmatically further verified with toxin B in the human epithelial colorectal adenocarcinoma cell line Caco-2 (Cell Line Services, Eppelheim, Germany), by measuring changes in the transepithelial resistance of confluent monolayers. Toxin B-induced drop of the transepithelial resistance in the Caco-2 cell monolayer was attenuated, when cells were preincubated with chlorpromazine (Fig. 2E). Taken together, the pharmacological tool chlorpromazine applied as an inhibitor of the formation of clathrin-coated pits, indicated an involvement of this endocytic compartment in the cellular internalization of CGTs.

RNAi-mediated gene silencing of clathrin heavy chain impedes toxin B intoxication
The involvement of clathrin in the uptake of CGTs was further analyzed by transfection of siRNA directed against the clathrin heavy chain to impair the formation of clathrin-coated pits. Two days after transfection of HeLa cells with siRNA, glucosylation of Rac1 upon addition of toxin B and incubation for 30 and 60 min was compared between siRNA-and mock-transfected cells. Reduced protein levels of clathrin heavy chain in lysates of siRNA-treated cells were confirmed by the use of an anti-clathrin heavy chain antibody (Fig. 3A). As expected, cells transfected with siRNA against the clathrin heavy chain retained higher amounts of non-glucosylated Rac after 30 and 60 min of toxin-treatment, when compared with mock-transfected cells (Fig. 3B). This result substantiates the importance of clathrin-mediated processes in the endocytic uptake of CGTs.
Clathrin-rather than caveolae-mediated endocytic mechanisms are implicated in the cellular internalization of toxin B To discriminate between the two major endocytic pathways that include dynamin (clathrin-and caveolae-pathway), plasmids encoding a dominant-negative variant of Cav-1 (Cav-1 DN, caveolae pathway) or Eps15 (Eps15 DN, clathrin pathway) were transfected in HeLa cells. Both proteins are fused to a GFP moiety and enable for the direct observation of the transfected cells by fluorescence microscopy. Importantly, addition of toxin B led to cell rounding predominantly in non-transfected and Cav-1 DNtransfected cells (81.9%, SD +/210.4%, of total cells and 82.1%, SD +/212.9%, of total cells, respectively), and to much lesser extent in Eps15 DN-transfected cells (26.2%, SD +/214.8%, of total cells) (Fig. 4A). The selective protection of cells expressing dominant-negative Eps15 against toxin B was confirmed by analyzing the Rac1 glucosylation status of Cav-1 DN-or Eps15 DN-transfected HeLa cells. For this purpose, transfected cells (expressing a GFP moiety fused to Cav-1 DN or Eps15 DN) were separated from non-transfected cells by applying fluorescenceassisted cell sorting, subsequently after preincubation with toxin B. Strikingly, Rac1 glucosylation was prevented exclusively in cells expressing dominant-negative Eps15 (Fig. 4B), but not in cells expressing dominant-negative Cav-1 (Fig. 4C) or non-transfected cells ( Fig. 4B and 4C). Conclusively, this result finally proves the predominant role of clathrin-rather than caveolae-mediated processes in the endocytic uptake of CGTs.

Lipid rafts are not implicated in the endocytic uptake of CGTs
Caveolae-mediated endocytic processes are exclusively taking place at microdomains of the plasma membrane denoted as lipid rafts [30]. But there is growing evidence for the implication of lipid rafts also in clathrin-dependent uptake mechanisms [31], [32], [33]. Therefore, we wanted to study the role of lipid rafts in the uptake of the CGTs. Sphingomyelin is a prominent component of lipid rafts and can be exogenously depleted from cell membranes by the addition of the enzyme sphingomyelinase (SMase) from Bacillus cereus [34]. In HeLa cells pretreated with SMase, no significant delay in cell rounding was observed upon addition of toxin B, toxin A, lethal toxin or a-toxin, when compared with nonpretreated cells (Fig. 5A). Control cells that were incubated with SMase alone did not show morphological alterations when compared with untreated cells (Fig. 5B). The exogenous depletion of sphingomyelin in HeLa cells was evaluated by using the vacuolating cytotoxin A (VacA) from Helicobacter pylori as a marker for sphingomyelin-/lipid rafts-dependent entry into host cells [35], [36]. As expected, vacuolization in HeLa cells upon addition of VacA was strongly reduced in sphingomyelin-depleted cells, when compared with control cells under equal conditions (85% reduction) (Fig. 5C). The lack of protection of HeLa cells from toxin B-induced cell rounding by SMase pretreatment was further confirmed in a time-dependent manner (Fig. 5D). In conclusion, lipid rafts seem not to be crucially involved in processes leading to the endocytic uptake of CGTs.

Discussion
Toxins of the family of clostridial glucosylating toxins are considered to enter cells by receptor-mediated endocytosis [37]. In contrast to toxins which finally reach their cytosolic substrates by a retrograde transport via the Golgi apparatus and the endoplasmatic reticulum (e.g. cholera toxin) [38], CGTs seem to require acidified early endosomes for the delivery of their enzymatic part into the cytosol [14], [15]. So far, the mechanisms underlying internalization of CGTs into endosomal compartments were not investigated in detail. Kushnaryov et al. provided evidence for endocytosis of C. difficile toxin A via coated pits by visualizing colloidal gold labeled toxin A in CHO cells by electron microscopy [37]. However, it remained unclear, whether these coated pits are functionally relevant for intoxication. We studied the internalization of CGTs in detail by using C. difficile toxin B as a model protein. At least for toxin B, three independent lines of evidence indicate the involvement of clathrin in its uptake into cells. First, pharmacological inhibition of clathrin-coated pits formation using chlorpromazine strongly reduced cytopathic effects of toxin B in HeLa and Caco-2 cells. Second, siRNA targeted against the clathrin heavy chain strongly reduced Rac1 glucosylation in HeLa cells by toxin B. Third, ectopic expression of dominant-negative Eps15 (participates in assembly of clathrin-coated pits) retained toxin B-induced cell rounding and Rac glucosylation in HeLa cells. CGTs are considered to use different receptor structures on the cell surface for binding and internalization, leading to endocytic uptake mechanisms that are possibly unique for each member of the CGT family. Since the pharmacological inhibition of the clathrin pathway with chlorpromazine also blocked intoxication of HeLa cells with toxin A, lethal toxin and a-toxin, we conclude that members of the CGT family at least share a common mechanism for cell entry.
Pinch-off of clathrin-coated vesicles from the plasma membrane requires the action of the GTPase dynamin. In addition, dynamin confers the constriction of caveolar vesicles from the plasma membrane [39]. Consistent with the finding that CGT uptake is mediated by clathrin, inhibition of dynamin activity with dynasore prevented CGT-induced cell rounding in HeLa cells and toxin Binduced Rac1 glucosylation in HT-29 cells. Inhibition of C. difficile toxin B intoxication of HeLa cells by expressing dominantnegative dynamin confirms the participation of this GTPase in the toxin uptake. The involvement of caveolin in the uptake of C. difficile toxin B could be excluded, since ectopic expression of dominant-negative Cav-1 in HeLa cells did not prevent toxin Binduced cell rounding and Rac1 glucosylation. The fact that cells, which do not express detectable levels of caveolin, e.g. HepG2 [40] or Caco-2 cells [41], can readily be intoxicated by toxin B (data not shown), argues also against a caveolae-mediated internalization of this toxin. Moreover, caveosomes are neutral and do not communicate with other acidic compartments [42]. Internalization in such endocytic vesicles is therefore unlikely, due to the fact that transport of the enzymatic part of C. difficile toxin B into the cytosol requires low pH [14], [15]. Even more, caveolae exclusively arise from membrane microdomains denoted as lipid rafts [30]. Since toxin B, toxin A, lethal toxin and a-toxin were not negatively influenced after disruption of lipid rafts by exogenous sphingomyelin depletion, this result provides additional evidence that cell entry of all CGTs does not occur through caveolae.
Our experimental data does not directly exclude a minor participation of other clathrin-independent endocytic pathways, such as the RhoA-or Cdc42-regulated endocytosis, in the uptake of CGTs. Even though both GTPases can be modified and inactivated by the clostridial glucosylating toxins, with the exception of the lethal toxin that does not modify RhoA, their requirement in initial stages of the intoxication process is still conceivable. However, Cheng and coworkers reported in vivo membrane binding of both GTPases to depend on the presence of sphingomyelin in the plasma membrane [43]. They showed that upon incubation of cells with SMase to hydrolyze sphingomyelin at the cell surface, RhoA-and Cdc42-dependent endocytosis of cargo proteins was blocked. Based on these findings and the fact that CGTs intoxication occurs also in sphingomyelin-depleted cells, it is unlikely that RhoA-or Cdc42-mediated endocytic pathways participate in cell entry of those toxins.
Our study provides an evidence for the clathrin-dependent endocytic uptake of clostridial glucosylating toxins. The uptake process might serve as a target for therapeutical intervention in the pathogenesis of clostridial infections. Based on our findings, it is also expected that CGTs may be used as novel probes for studying cell entry of other cargo proteins with unknown uptake mechanism.

Toxins and plasmids used in this study
Toxin B and lethal toxin used in this study were recombinantly purified from Bacillus megaterium with protocols as published by others previously for Clostridium difficile toxin A and toxin B [44].
Native toxin A and a-toxin were used in this study, purified from C. difficile (strain VPI 10463) and C. novyi (strain ATCC 19402), respectively, according to protocols described elsewhere [45], [46]. Toxin A was additionally purified by thyroglobuline affinity chromatography [47]. Recombinant E. coli CNF1 protein was allocated by G. Schmidt (University of Freiburg, Germany). Diphtheria toxin was ordered from Calbiochem (Darmstadt, Germany). Native VacA toxin (allelic type s1a/m1) from concentrated Helicobacter pylori (strain Hp402022A) culture supernatans, was kindly provided by M. Kist (National Reference Center for Helicobacter pylori, University of Freiburg, Germany). VacA was acid-activated and re-neutralized prior to usage in intoxication assays. Plasmids encoding dominant-negative Eps15, Cav-1 or dynamin were generous gifts of A. Benmerah (INSERM, Paris, France), J. Eggermont (University of Leuven, Belgium) and C. van Koppen (University of Essen, Germany), respectively.

Inhibitor assays for investigation of toxin uptake
Prior to addition of toxins, cells were preincubated for 45 min at 37uC in medium lacking FCS and with substances specifically inhibiting endocytic pathways. HeLa cells were preincubated with 80 mM dynasore or 12.5 mM chlorpromazine; HT-29 cells with 80 mM dynasore or 30 mM chlorpromazine and Caco-2 cells with 30 mM chlorpromazine. Exogenous sphingomyelin depletion in HeLa cells was achieved by direct addition of 100 mU/ml sphingomyelinase from Bacillus cereus in culture medium and incubation for 1 h at 37uC. All compounds were purchased from Sigma-Aldrich (Deisenhofen, Germany).

Microscopy
Mophological changes of intoxicated cells were directly analyzed in wells using an inverted microscope (Axiovert 25, Carl Zeiss Jena GmbH, Germany). Fluorescence signals of cells were analyzed using a fluorescence microscope (Axiophot, Carl Zeiss Jena GmbH, Germany) with a Neofluar 40 object lens and filter set 43H for phalloidin/TRITC or 65 HE for GFP. Images were obtained with an inbuilt digital camera (AxioCam HRc or HRm) and contrast was adjusted manually using AxioVision software (version 3.1.2.1).

FITC-phalloidin staining of F-actin in cultured cells
Fluorescein isothiocyanate (FITC) conjugated phalloidin was used to label F-actin in HeLa cells. Briefly, cells were grown on cover slips and fixed with 4% (v/v) paraformaldehyde for 15 min prior to permeabilization with 0.1% Triton X-100 for 1 min. Cells were then incubated with 1% (v/v) TRITC-conjugated phalloidin (Invitrogen, Karlsruhe, Germany) for 45 min in the dark. Fixed samples were analyzed with fluorescence microscopy.
Transfection studies for investigation of toxin uptake Clathrin heavy chain targeting siRNA, according to HC oligo I from reference [48], was obtained from Dharmacon (Lafayette, USA) and applied for transfection in HeLa cells by using Lipofectamine 2000 (Invitrogen) following the manufacturer's recommendations. The same transfection procedure was used for transfection of a plasmid encoding a dominant-negative dynamin mutant (pRK5/HA-dynamin-1aa K44A ; [49]) in HeLa cells. Plasmids encoding a dominant-negative form of Eps15 (pEGFP-C2/Eps15DN; [50]) or Cav-1 (pCINeo/Ires-GFP/cav-1DN; [51]) were transfected in HeLa cells either by using Lipofectamine (for subsequent microscopic analysis) or by addition of a mixture of Opti-MEMH I Reduced-Serum Medium (Invitrogen, Karlsruhe, Germany), including 0.005% w/v polyethylenimine and 0.2 mg/ ml plasmid DNA, to DMEM medium containing 10% FCS and 1% NEA (for subsequent fluorescence-assisted cell sorting). Following an incubation of 5 h at 37uC, cells were washed with DMEM and further incubated at standard conditions. The effects of dominant-negative Eps15 or Cav-1 expression and siRNAmediated clathrin gene silencing were assayed one and two days after transfection, respectively.

Fluorescence-activated cell sorting (FACS)
HeLa cells transfected with a plasmid encoding a GFP-tagged, dominant-negative form of Eps15 or Cav-1 were separated from non-transfected cells by using a MoFlo high speed cell sorter (Dako, Hamburg, Germany). Prior to cell sorting, mocktransfected cells were used to define the parameters for restrictive sorting of single and living cells. An argon ion laser, emitting light at 488 nm (blue), was used for excitation of GFP signals in transfected cells. Data were analyzed with the flow cytometry analysis program FlowJo.

TER measurement
For analysis of transepithelial resistence (TER) of Caco-2 cell monolayers, cells were grown on filters (Millicell; Millipore) to confluency. TER was measured at indicated time points after toxin-treatment by use of a volt-ohm meter (World Precision Instruments, USA) equipped with a chamber for filter insertion. Only filters with cell monolayers showing an initial resistance of at least 200 V per cm 2 were used.