Cellular Interactions of the Cytolethal Distending Toxins from Escherichia coli and Haemophilus ducreyi*

Background: Cytolethal distending toxins (CDTs) produced by pathogenic bacteria are genotoxic. Results: CDTs exploit two different endocytic pathways to reach the nucleus. Conclusion: Individual members of the CDT superfamily interact with host cells by distinct mechanisms. Significance: Learning how CDTs interact with and modulate host cells and tissues is critical for understanding the strategies used by pathogenic bacteria during infection. The cytolethal distending toxins (CDTs) compose a subclass of intracellularly acting genotoxins produced by many Gram-negative pathogenic bacteria that disrupt the normal progression of the eukaryotic cell cycle. Here, the intoxication mechanisms of CDTs from Escherichia coli (Ec-CDT) and Haemophilus ducreyi (Hd-CDT), which share limited amino acid sequence homology, were directly compared. Ec-CDT and Hd-CDT shared comparable in vitro DNase activities of the CdtB subunits, saturable cell surface binding with comparable affinities, and the requirement for an intact Golgi complex to induce cell cycle arrest. In contrast, disruption of endosome acidification blocked Hd-CDT-mediated cell cycle arrest and toxin transport to the endoplasmic reticulum and nucleus, while having no effects on Ec-CDT. Phosphorylation of the histone protein H2AX, as well as nuclear localization, was inhibited for Hd-CdtB, but not Ec-CdtB, in cells expressing dominant negative Rab7 (T22N), suggesting that Hd-CDT, but not Ec-CDT, is trafficked through late endosomal vesicles. In support of this idea, significantly more Hd-CdtB than Ec-CdtB co-localized with Rab9, which is enriched in late endosomal compartments. Competitive binding studies suggested that Ec-CDT and Hd-CDT bind to discrete cell surface determinants. These results suggest that Ec-CDT and Hd-CDT are transported within cells by distinct pathways, possibly mediated by their interaction with different receptors at the cell surface.

The cytolethal distending toxins (CDTs) 5 compose a family of multisubunit bacterial genotoxins that modulate the eukaryotic cell cycle (1)(2)(3)(4)(5). CDT-mediated cell cycle arrest and the disruption of cytokinesis have been proposed to alter the normal barrier and immune functions of both epithelial cells and lymphocytes (3). Animal model studies as well as the association of the cdt gene carriage in disease-causing bacteria from human isolates both support the importance of CDTs for the virulence strategies of specific pathogens (6,7).
Most CDTs function as assembled complexes of three protein subunits, encoded by three contiguous genes (cdtA, cdtB, and cdtC) within an operon (8). Consistent with the canonical AB model of intracellularly acting toxins (9), CdtB appears to function as the enzymatic A-subunit. CdtB has DNase I-like activity (10,11) and localizes to the nucleus when expressed ectopically within the cytosol of mammalian cells (12)(13)(14), suggesting that this subunit directly causes DNA damage within the nuclei of intoxicated cells. CdtA and CdtC facilitate the delivery of CdtB into cells (15)(16)(17)(18), although the molecular details of how these subunits facilitate the cell-surface binding, uptake, and intracellular transport of CdtB remain poorly understood.
A diverse group of Gram-negative pathogenic bacteria that colonize distinct niches within the host have been identified to possess cdt genes (19). The AB 2 toxin architecture as well as a number of other key structural features appear to be generally conserved across the CDT family (20), suggesting that individual toxin members may interact with and intoxicate cells in a similar fashion. However, the cellular intoxication properties of CDTs produced by different pathogenic organisms are poorly understood. Recently, the sensitivity of several cell lines to CDTs from Aggregatibacter actinomycetemcomitans, Campylobacter jejuni, Escherichia coli, and Haemophilus ducreyi was demonstrated to be differentially affected by alterations in host glycans and membrane cholesterol (21), suggesting that host cell requirements for CDT intoxication of mammalian cells may not be universally conserved. However, it remains unclear whether the overall mechanism and molecular basis of toxin binding, uptake, and intracellular transport are broadly applicable to all members of the CDT family.
The objective of this study was to directly compare the cellular intoxication mechanisms employed by CDTs produced by E. coli and H. ducreyi, two pathogens that colonize highly divergent host niches (i.e. the intestinal and urogenital tracts, respectively). Notably, the CDTs from E. coli (Ec-CDT) and H. ducreyi (Hd-CDT) share only 22 and 19% sequence identity, respectively, in their CdtA and CdtC subunits, suggesting the possibility that these two toxins might interact with host cells in fundamentally different ways. These studies revealed differences in the cellular requirements for toxin intracellular trafficking. Moreover, Ec-CDT and Hd-CDT did not compete with each other for binding to the surface of cells, suggesting that these toxins may target and bind to discrete receptors. Overall, these studies suggest that Ec-CDT and Hd-CDT are transported within cells by distinct pathways, possibly mediated by their interaction with different receptors at the cell surface.

Cloning of cdt Genes and Preparation of Expression
Strains-The cloning of the genes encoding Ec-CDT and Hd-CDT in plasmids for recombinant expression in E. coli was described previously (21).
Expression and Purification of Recombinant Ec-CDT and Hd-CDT-Each recombinant protein was expressed and purified as described previously (21). Protein concentrations were quantified using the Bradford Protein Assay (Thermo Scientific, Rockford, IL). Recombinant proteins were used only when purified to at least 95% homogeneity, as estimated by resolving the proteins using SDS-PAGE and visualizing after staining the gels with Coomassie Brilliant Blue (Bio-Rad; data not shown). The purified, denatured subunits were stored at Ϫ20°C in 20 mM HEPES (Calbiochem), pH 7.5, containing urea (8 M) and NaCl (200 mM).
Ec-CDT and Hd-CDT holotoxins were prepared as described previously (22). Ec-CDT and Hd-CDT holotoxin integrity was evaluated using the dialysis retention assay, as described previously (17). Ec-CDT or Hd-CDT holotoxin (5-20 M, 1 ml) was dialyzed (100-kDa molecular mass cut-off tubing; Spectrum Laboratories) at 4°C against four 250-ml volumes of PBS, pH 7.4, containing 5% glycerol. After 24 h, the dialyzed proteins were evaluated using SDS-PAGE followed by staining with Coomassie Brilliant Blue. The gels were scanned with a CanonScan 9950F scanner (Canon, Lake Success, NY) using ArcSoft Photo Studio 5.5 software (ArcSoft, Fremont, CA). The integrity of the holotoxins was quantified by comparing the relative intensities of the bands corresponding to CdtA, CdtB, or CdtC before and after dialysis, as determined by using the UN-SCAN-IT program (Silk Scientific, Inc., Orem, UT). Individual CDT subunits, each of which has a molecular mass less than 35 kDa, were used as negative controls.
Cell Cycle Phase Determination-The indicated cell lines were seeded (1.5 ϫ 10 5 cells per well) in 6-well plates (Corning Inc., Corning, NY). After 18 h, the cells were further incubated in complete medium with or without Ec-CDT or Hd-CDT at the indicated concentrations or were mock-incubated with PBS, pH 7.4. After 48 h, the cells were analyzed for arrest at the G 2 /M interface, as described previously (21). From these data, dose-response curves were generated by plotting cells in G 2 /M as a function of toxin concentration. From the dose-response curves, we determined cell cycle arrest dose 50 (CCAD 50 ) values, which we defined as the toxin concentrations required to induce G 2 /M arrest in 50% of the cell population not already in G 2 /M.
For both Ec-CDT and Hd-CDT, preliminary studies revealed nearly identical dose-response curves for G 2 /M arrest in the presence or absence of the hexa-histidine tags (supplemental Fig. S4). All subsequent studies were conducted with Ec-CDT or Hd-CDT assembled from subunits that retained their respective hexa-histidine tags.
Some studies were conducted in the presence of the indicated concentrations of brefeldin A (BFA; MP Biomedicals), ammonium chloride (NH 4 Cl, Sigma), bafilomycin A1 (LC Laboratories, Woburn, MA), monensin (Sigma), or nigericin (Sigma). These studies were conducted at approximately the CCAD 50 values determined for Hd-CDT and Ec-CDT to best assess whether each pharmacological agent inhibited, potentiated, or had no effect on toxin activity.
Flow Cytometry-Analytical flow cytometry-based assays were carried out as described previously (21).
DNase Assay-The relative in vitro DNase activities of either Ec-CdtB or Hd-CdtB were determined as described previously (10), using purified pUC19 as the substrate. Reactions were stopped at 4 h by the addition of EDTA (to a final concentration of 10 mM). DNA gel loading dye (six times; Promega) was added to each sample, and the samples were resolved by employing agarose (0.8%) gel electrophoresis using Tris acetate/EDTA (TAE, 40 mM Tris base (Fisher), 20 mM acetic acid (Fisher), 1 mM EDTA) electrophoresis buffer with constant voltage (100 V). The gels were stained with ethidium bromide (1 g/ml, Bio-Rad) and photographed using a Gel Doc EQ system (Bio-Rad), and the relative pixel densities of stained bands corresponding to supercoiled pUC19 were measured using UN-SCAN-IT. Relative DNase activity was calculated from the pixel densities of ethidium bromide-stained bands corresponding to supercoiled pUC19, using the relationship: relative DNase activity ϭ ((super-coiled pUC19 pixels from control reactions lacking Ec-CdtB and Hd-CdtB) Ϫ (super-coiled pUC19 pixels from reactions containing Ec-CdtB or Hd-CdtB))/(super-coiled pUC19 pixels from control reactions lacking Ec-CdtB and Hd-CdtB). A value of 1.0 corresponds to a complete loss of detectable supercoiled pUC19.
Biotinylation of CDTs-Ec-CDT or Hd-CDT (1 ml at 5-20 M) in PBS, pH 7.4, was incubated overnight at 4°C with EZ Link Sulfo-NHS-LC-Biotin (10-fold molar excess, Thermo Scientific). The labeling reaction was arrested by the addition of Tris, pH 8.0 (to a final concentration of 20 mM). To remove free label, the proteins were dialyzed at 4°C against four changes of PBS, pH 7.4 (250 ml each), containing 5% glycerol. Preliminary studies indicated that biotinylation reactions had no detectable effects on the capacity of either Ec-CDT or Hd-CDT to induce cell cycle arrest in G 2 /M.
Internalization Assay-Cells were seeded (2 ϫ 10 4 per well) in 8-well chamber slides (Nunc, Rochester, NY). After 18 h, the slides were incubated on ice for 30 min. The cells were washed two times with ice-cold PBS, pH 7.4, and then further incubated on ice with or without biotinylated Ec-CDT or Hd-CDT (at the indicated concentrations) in PBS, pH 7.4, with bovine serum albumin (BSA, 3%, Sigma) or mock-incubated in PBS, pH 7.4, with BSA (3%). After 30 min of toxin prebinding on ice, the cells were washed three times with ice-cold PBS, pH 7.4.
To monitor CDT binding, the cells were immediately fixed by incubating with ice-cold 2% formaldehyde (Sigma) and then further incubated at room temperature for 30 min. To monitor CDT internalization, after 30 min of toxin prebinding on ice, the cells were incubated with prewarmed (37°C) complete medium. After 10 min, the cells were washed with ice-cold PBS, pH 7.4, and fixed with ice-cold 2% formaldehyde. After fixing for 30 min at room temperature, the cells were permeabilized by incubating in PBS, pH 7.4, containing 0.1% Triton X-100 for 15 min, and blocked with 3% BSA for 30 min. To probe for biotinylated Ec-CDT or Hd-CDT, the fixed and permeabilized cells were incubated at room temperature with streptavidin conjugated with Alexa Fluor 488 (1:200 dilution in PBS 7.4; Invitrogen). After 1 h, the cells were washed three times with 0.1% Tween 20 (Fisher) in PBS, pH 7.4. The nucleus was stained by incubating with 4Ј,6-diamidino-2-phenylindole (DAPI; 300 nM in PBS, pH 7.4; Invitrogen) for 2 min. Slides were washed three times with 0.1% Tween 20 in PBS, pH 7.4, and air-dried. The slides were mounted with ProLong Gold antifade reagent (Invitrogen) and cured overnight. The cells were then analyzed using DIC/fluorescence microscopy.
DIC/Fluorescence Microscopy-Chamber slides were analyzed using a DeltaVision RT microscope (Applied Precision, Issaquah, WA), using an Olympus Plan Apo ϫ40 oil objective with NA 1.42 and working distance of 0.17 mm; DIC images were collected using a Photometrics CoolSnap HQ camera (Photometrics, Tucson, AZ). Images were processed using Softworx Explorer Suite (version 3.5.1, Applied Precision Inc.). Deconvolution was carried out using Softworx constrained iterative deconvolution tool (ratio mode) and analyzed using Imaris 5.7 (Bitplane AG, Zurich, Switzerland).
CDT Localization to the ER-Cells were plated in an 8-well chamber slide to 30 -40% confluency. After overnight incuba-tion, cells were washed two times with PBS, pH 7.4, followed by addition of 100 l of Opti-MEM (Invitrogen), and further incubated for 2 h before transfection. Approximately 30 min prior to addition to cells, the appropriate dilutions of pDsRed2-ER (0.4 g/200 l/well; Clontech), which encodes red fluorescent protein fused to the ER-targeting sequence of calreticulin, the ER retention sequence KDEL, and the Lipofectamine 2000 reagent (1 l/200 l/well; Invitrogen) complex were prepared according to the manufacturer's instructions and allowed to incubate at room temperature. After 30 min, Opti-MEM was removed, and the plasmid DNA/transfection mixture (100 l) was added to each well in a dropwise fashion with gentle agitation, and the cells were immediately incubated at 37°C under 5% CO 2 . After 4 h, the transfection mixture was removed from the monolayers, and the cells were incubated in complete medium at 37°C and under 5% CO 2. After overnight incubation, we typically detected 80 -90% of the cells within the monolayer to be transfected based on the percentage of cells with red fluorescent protein fluorescence, as determined using an Olympus CKX 41 fluorescence microscope (Olympus America Inc., Center Valley, PA).
Cells that had been transfected, as described above, with pDsRed2-ER, were preincubated for 30 min in the absence or presence of NH 4 Cl (20 mM) at 37°C and under 5% CO 2 . The cells were then further incubated with biotinylated Ec-CDT or Hd-CDT (200 nM), and the cells were stained for each toxin as described above under "Internalization Assay." The cells were further stained for actin by incubating with Alexa Fluor 647labeled phalloidin (1 unit/200 l/well; Invitrogen). Images were collected using DIC/fluorescence microscopy and deconvoluted, as described above. For each cell, images were collected from an average of 30 z-planes, each at a thickness of 0.2 m. Localization analysis was conducted by using the co-localization module of the DeltaVision SoftWoRx 3.5.1 software suite. Results were expressed as the localization index, which was derived from calculating the Pearson's coefficient of correlation values, which in these studies was a measure of localization of the indicated CDT to the ER in each z-plane of the cell. In these studies, a localization index value of 1.0 indicates 100% localization of Ec-CDT or Hd-CDT to the ER, whereas a localization index of 0.0 indicates the absence of Ec-CDT or Hd-CDT localization to the ER. The localization index was calculated from the analysis of a total of 50 images collected over three independent experiments.
Monitoring Nuclear Localization of Ec-CdtB or Hd-CdtB within the Nuclear Fraction of CDT-intoxicated Cells-Cells were seeded in 100-mm culture dishes (BD Biosciences). After overnight incubation, the monolayers were treated with bafilomycin A1 (20 nM) for 30 min and further incubated with Ec-CDT or Hd-CDT (at the indicated concentrations). After 30 min, the cells were further incubated with fresh medium (not containing toxin) for an additional 210 min. After 240 min of total internalization of Ec-CDT or Hd-CDT, the cells were harvested by scraping, lysed using hypotonic buffer, and fractionated as per the manufacturer's protocol (nuclear extract kit, Active Motif, Carlsbad, CA). The fractionated samples were evaluated by Western blotting, using rabbit polyclonal antibodies against Ec-CdtB (1:20,000 dilution; generated by Thermo Fisher Scientific Inc., Rockford, IL) or rabbit polyclonal antibodies against Hd-CdtB (1:10,000; generated by Immunological Resource Center, University of Illinois, Urbana, IL), cytosolic marker GAPDH (1:200; Abcam, Cambridge, MA), microsomal marker calnexin (1:2000; Abcam), nuclear marker p84 (1:2000; Abcam), and cell lysate loading control actin (1:1000; NeoMarkers, Fremont, CA). Anti-mouse (1:2000) and anti-rabbit (1:5000) secondary antibodies were purchased from Pierce Protein Biology Products (Thermo Fisher Scientific Inc.). Relative amounts of CdtB were determined by densitometric analysis of the blot bands using the UN-SCAN-IT program.
H2AX Activation Assay-Cells were transfected with pDsRed-Rab7-DN (0.4 g/200 l/well), using the general transfection procedure as described above under "CDT Localization to the ER." The transfected monolayers were incubated at 37°C with Ec-CDT (200 nM) or Hd-CDT (200 nM). After 8 h, the cells were fixed and permeabilized as described above under "Internalization Assay." To monitor activation of H2AX by CDT, samples were stained overnight with a rabbit polyclonal anti-p-H2AX antibody (1:5000 dilution; Invitrogen) at 4°C, followed by incubation with goat anti-rabbit Alexa Fluor 488 antibody (1:1000 dilution; Invitrogen) at room temperature for 2 h. The cells were further counter-stained for actin by incubating with Alexa Fluor 647-labeled phalloidin (1 unit/200 l/well) and for nuclei by incubating with DAPI (1 ng/200 l; Invitrogen) for 30 min. The slides were mounted with ProLong Gold Antifade Reagent (25 l/well; Invitrogen). Images were collected using DIC/fluorescence microscopy and deconvoluted, as described above. For each cell, images were collected from an average of 30 z-planes, each at a thickness of 0.2 m. H2AX activation analysis was conducted by using the DeltaVision SoftWoRx 3.5.1 software suite. Percentages of H2AX-activated cells in each group of functional and dysfunctional Rab7 cells were calculated from ϳ50 cells from each group over three independent experiments.
Monitoring Nuclear Localization of Ec-CdtB or Hd-CdtB Using Fluorescence Microscopy-Cells were transfected with pDsRed-Rab7-DN (0.4 g/200 l/well), using the general transfection procedure as described above under "CDT Localization to the ER" The transfected monolayers were prechilled for 30 min on ice and further incubated with Ec-CDT (200 nM) or Hd-CDT (200 nM) on ice for 30 min. After 60 min of postinternalization, the cells were fixed and permeabilized as described above under "Internalization Assay." To monitor Ec-CdtB or Hd-CdtB localization to the nucleus, cells were incubated with rabbit polyclonal anti-Ec-CdtB or anti-Hd-CdtB antibodies (1:2000 dilution) at 4°C overnight, followed by incubation with goat anti-rabbit Alexa Fluor 488-labeled antibody (1:1000 dilution; Invitrogen) at room temperature for 2 h. The cells were further counter-stained for the nucleus by incubating with DAPI (1 ng/200 l; Invitrogen) for 30 min at room temperature. The slides were mounted with ProLong Gold Antifade Reagent (25 l/well; Invitrogen). Images were collected using DIC/fluorescence microscopy and deconvoluted, as described above. For each cell, images were collected from an average of 30 z-planes, each at a thickness of 0.2 m. Nuclear localization analysis was conducted by using the DeltaVision SoftWoRx 3.5.1 software suite. Percentages of CdtB localization into the nucleus in each group of functional and dysfunctional Rab7 cells were calculated from ϳ30 cells from each group over three independent experiments.
Images were collected using DIC/fluorescence microscopy and deconvoluted, as described above. For each cell, images were collected from an average of 30 z-planes, each at a thickness of 0.2 m. Localization analysis was conducted by using the co-localization module of the DeltaVision SoftWoRx 3.5.1 software suite. Results were expressed as the localization index, which was derived from calculating the Pearson's coefficient of correlation values, which in these studies was a measure of localization of the indicated CDT to the Rab9 in each z-plane of the cell. In these studies, a localization index value of 1.0 indicates 100% localization of Ec-CdtB or Hd-CdtB to Rab9, whereas a localization index of 0.0 indicates the absence of Ec-CdtB or Hd-CdtB localization to the Rab9. The localization index was calculated from the analysis of a total of 50 images collected over three independent experiments.
Cell Binding Assays-Mammalian cell binding assays were conducted as described previously (17,23). Cells were seeded (2 ϫ 10 4 /well) in 96-well plates (Fisher). After 18 h, the plates were incubated on ice. After 30 min, the cells were washed two times with ice-cold PBS, pH 7.4, and then further incubated on ice with or without biotinylated Ec-CDT or Hd-CDT (at the indicated concentrations) and BSA (3%) in PBS, pH 7.4, or mock-incubated with BSA (3%) in PBS, pH 7.4. After 1 h on ice, the cells were washed three times with ice-cold PBS, pH 7.4, then fixed by adding ice-cold 2% formaldehyde and 0.2% glutaraldehyde, and then further incubated at room temperature. Preliminary studies indicated that maximal binding occurred between 30 and 60 min (not shown). After 15 min, the plate was washed three times with PBS, pH 7.4, at room temperature and then incubated at room temperature with streptavidin-HRP conjugate (1:15,000, GE Healthcare) in PBS, pH 7.4,. After 30 min, the cells were washed five times with PBS, pH 7.4, and then incubated at room temperature with TMB Ultra (100 l, Thermo Scientific). After 30 min, the supernatant from each well was removed and added to an equal volume of sulfuric acid (2 N, Mallinckrodt Baker Inc., Paris, KY) at room temperature.
The absorbance at 450 nm (A 450 nm ) was measured using a Biotek Synergy 2 plate reader (Biotek Instruments Inc., Winooski, VT). The A 450 nm measured for supernatants collected from wells containing cells incubated with PBS, pH 7.4, alone (background absorbance) was subtracted from the A 450 nm measured for supernatants collected from wells containing cells that had been incubated with Ec-CDT or Hd-CDT. Relative binding was determined by dividing the A 450 nm minus background at each toxin concentration by the A 450 nm minus background at the highest concentration of toxin used in these studies (200 nM). The dissociation constants (K d ) were calculated by nonlinear regression of the curve generated from plotting relative binding as a function of toxin concentration, using GraphPad Prism (Version 4.03, GraphPad Software, La Jolla, CA). Competitive binding assays were conducted as described above, except in the absence or presence of 100-fold molar excess of the specified nonbiotinylated proteins.
Effects of Lectin Binding on CDT Binding-Cell monolayers were prepared as described above under "Cell Binding Assays." Prechilled cells were pretreated on ice with Euonymus Europaeus Agglutinin lectin (EEA, EY Labs, San Mateo, CA; 40 g/ml). After 30 min, the cells were further incubated on ice with or without biotinylated Ec-CDT (100 nM) or Hd-CDT (100 nM) and BSA (3%) in PBS, pH 7.4, or mock-incubated with BSA (3%) in PBS, pH 7.4, in the presence or absence of 40 g/ml EEA. After 30 min on ice, the cells were washed, fixed, and analyzed as described above under "Cell Binding Assays." Statistics-Unless otherwise indicated, each experiment was performed at least three independent times, each time in triplicate. Statistical analyses were performed using Microsoft Excel (Version 11.0, Microsoft, Redmond, WA) or GraphPad. The Q test was performed to eliminate data that were statistical outliers (24). Error bars represent standard deviations. All p values were calculated with the Student's t test using paired, two-tailed distribution. A p value of less than 0.05 indicated that differences in the specified data were considered statistically significant. (21) reported that considerably lower concentrations of Hd-CDT rather than Ec-CDT were required to induce phosphorylation of the histone protein H2AX, a marker of DNA damage, in HeLa, CHO-K1, Balb/3T3, Y-1 cells, OT-1, NIH/3T3, IC-21, and Raw 264.7 cells. To evaluate the relationship between these results and the capacity of Ec-CDT or Hd-CDT to induce G 2 /M cell cycle arrest, one of several possible downstream consequences of H2AX activation, we measured cell cycle progression in HeLa and CHO-K1 cells, which have been commonly used as models for studying the function of bacterial toxins, including CDTs (25)(26)(27)(28)(29)(30)(31)(32)(33)(34)(35)(36), as a function of Ec-CDT or Hd-CDT concentrations. From the dose-response curves, we determined CCAD 50 (i.e. cell cycle arrest dose 50 ) values, which we defined as the toxin concentrations required to induce G 2 /M arrest in 50% of the cell population not already in G 2 /M. These studies revealed that substantially lower concentrations of Hd-CDT than Ec-CDT were required to induce G 2 /M cell cycle arrest in HeLa cells (supplemental Fig. S1, A and B). For HeLa cells, CCAD 50 values of 4 pM and 100 nM were determined for Hd-CDT (supplemental Fig. S1A) and Ec-CDT (supplemental Fig. S1B), respectively, indicating that Hd-CDT is ϳ2.5 ϫ 10 4 -fold more potent toward HeLa cells than Ec-CDT. In contrast, CHO-K1 cells required only 70-fold higher concentrations of Ec-CDT than Hd-CDT, with CCAD 50 values of 7 and 0.1 nM, respectively (supplemental Fig. S1, C and D). These results are consistent with those of the previous study (21), which reported that HeLa and CHO-K1 cells required ϳ1.0 ϫ 10 4 -and 50-fold higher concentrations, respectively, of Ec-CDT than Hd-CDT to induce H2AX phosphorylation.

Evaluating Relative Sensitivities of Mammalian Cell Lines to Ec-CDT and Hd-CDT-A recent study
We also compared the relative sensitivities toward Hd-CDT and Ec-CDT using two intestinal cell lines (Caco-2 and INT-407 cells), which had not been evaluated in the previous study (21). These studies revealed that Caco-2 (supplemental Fig. S1, E and F) and INT-407 cells (supplemental Fig. S1, G and H) were both arrested at the G 2 /M interface at considerably lower concentrations of Hd-CDT than Ec-CDT. However, these cell lines again varied in their relative susceptibilities to the two toxins. Caco-2 cells required ϳ3 ϫ 10 2 -fold higher concentrations of Ec-CDT (CCAD 50 ϭ 3 nM) than Hd-CDT (CCAD 50 ϭ 10 pM), whereas INT-407 cells required ϳ7 ϫ 10 3 -fold higher concentrations of Ec-CDT (CCAD 50 ϭ 40 nM) than Hd-CDT (CCAD 50 ϭ 6 pM).
To evaluate the possibility that differences in holotoxin stability might underlie the highly divergent cellular potencies of Ec-CDT and Hd-CDT, we investigated the assembly of the heterotrimeric complexes using the dialysis retention assay (17,37). These studies revealed that ϳ60 -99% Ec-CDT or Hd-CDT was assembled into holotoxins, as indicated by their retention within dialysis tubing with a molecular mass cutoff of ϳ100 kDa (supplemental Fig. S2), whereas CdtC (supplemental Fig. S2) or CdtA or CdtB (data not shown) was not retained. Even in the worst case scenarios (e.g. 60% holotoxin assembly), the large differences in potencies robustly measured for these two toxins (Ͼ10 4 -fold greater activity for Hd-CDT than Ec-CDT on HeLa cells) cannot readily be attributed to the inability of the holotoxins to assemble.
Relative in Vitro DNase Activities of Ec-CdtB and Hd-CdtB-CDT-dependent G 2 /M arrest of intoxicated cells has been attributed to the activation of cellular DNA repair pathways (38,39) in response to the DNase I-like enzymatic activity associated with the CdtB subunit of several CDTs, including Ec-CDT (10, 11) and Hd-CDT (40). To evaluate whether or not differences in the sensitivities of HeLa cells to Ec-CDT and Hd-CDT might be associated with the intrinsic catalytic properties of the CdtB subunits of each toxin, the in vitro DNase activities of the purified recombinant Ec-CdtB and Hd-CdtB subunits (in the absence of CdtA and CdtC) were compared. These studies revealed that both Ec-CdtB and Hd-CdtB demonstrated similar DNase activity in a dose-dependent manner (Fig. 1). Consistent with previous reports (11,17,40), Ec-CdtA, Ec-CdtC, Hd-CdtA, or Hd-CdtC did not yield detectable DNase activity (data not shown). These results support the idea that the disparity in cellular potencies of Ec-CDT and Hd-CDT are not likely due to intrinsic divergence in the DNase activities between the catalytic subunits of these toxins.
Both Ec-CDT and Hd-CDT Are Taken Up from Plasma Membrane into Cells within 10 min-Several studies have reported that ectopic expression of CdtB within mammalian cells is sufficient to induce cell cycle arrest in G 2 /M, even in the absence of CdtA and CdtC (11,13,38,41), strongly supporting a model that CdtB acts from an intracellular location. To evaluate the possibility that differences in cell sensitivity to Ec-CDT and Hd-CDT may be due to large disparities between the times required for toxin uptake from the cell surface, we examined the internalization of both toxins into HeLa cells using DIC/ fluorescence microscopy. These experiments revealed that under conditions nonpermissive for cell entry (4°C), both Ec-CDT and Hd-CDT were visible at the cell surface (supplemental Fig. S3, A and C), indicating that both toxins were bound to the plasma membrane. When the temperature was raised to 37°C to induce conditions permissive for uptake, both toxins were visible within the cell after just 10 min (supplemental Fig.  S3, B and D), indicating that Ec-CDT and Hd-CDT were both efficiently internalized from the plasma membrane into cells.
Effect of BFA on Toxin-induced G 2 /M Cell Cycle Arrest-Hd-CDT-mediated cell cycle arrest has been demonstrated to be sensitive to the action of BFA, which disrupts retrograde protein transport from the Golgi complex to the ER (42,43), suggesting that this toxin is trafficked by a retrograde mechanism (9). Because the importance of an intact Golgi complex for intoxication of cells with Ec-CDT had not been previously studied, toxin-mediated cell cycle arrest of HeLa cells was evaluated in the presence or absence of BFA and at approximately the respective CCAD 50 values of each toxin (e.g. 5 pM for Hd-CDT and 100 nM for Ec-CDT). These studies revealed that for Ec-CDT, as well as for Hd-CDT, toxin-mediated G 2 /M cell cycle arrest was blocked by BFA (Fig. 2, A and B), which supports a model that both toxins are trafficked to the ER via the Golgi complex.
Effects of Endosomal Acidification Inhibitors on Toxin-induced G 2 /M Cell Cycle Arrest-The finding that Ec-CDT and Hd-CDT-mediated G 2 /M cell cycle is blocked in the presence of BFA prompted us to further evaluate host cell requirements associated with transport from the cell surface to the ER. Transport of a subset of intracellularly acting bacterial toxins within the cell requires the lowering of pH within the lumen of endocytic compartments (9). Although an earlier study (42) reported that Hd-CDT-mediated G 2 /M cell cycle arrest in HeLa cells was inhibited by several agents that block endosome acidification, the requirement for endosome acidification associated with Ec-CDT-mediated G 2 /M cell cycle arrest had not previously been reported. As demonstrated previously (42), Hd-CDT-mediated G 2 /M cell cycle arrest in HeLa cells was inhibited (Fig. 3, A and  B) by the lysosomotropic amine, ammonium chloride (44). In contrast, Ec-CDT-mediated cell cycle arrest was not blocked in the presence of NH 4 Cl (Fig. 3, A and B). Moreover, bafilomycin A1, which blocks acidification of intracellular vacuoles by an alternative mechanism, inhibiting the action of vacuolar ATPases (45), also blocked Hd-CDT-mediated but not Ec-CDT-mediated G 2 /M cell cycle arrest (Fig. 3, C and D). Finally, the polyether ionophore monensin, which is known to block intracellular protein transport by collapsing proton gradients as a sodium/proton antiporter (46), and the potassium/ proton carboxylic ionophore nigericin (44) blocked Hd-CDT-FIGURE 1. In vitro DNase activities of Ec-CdtB and Hd-CdtB. The DNase activities of Ec-CdtB (filled triangles) and Hd-CdtB (filled circles) were determined as described under "Experimental Procedures" and plotted as a function of toxin concentration. The rendered data were combined from three independent experiments, each conducted in triplicate. Error bars indicate standard deviations. Statistical significance was calculated for the differences in relative DNase activities at the indicated concentrations of Ec-CdtB and Hd-CdtB. * indicates p Ͻ 0.05.

FIGURE 2. Effects of BFA on Ec-CDT-or Hd-CDT-induced cell cycle arrest.
Ec-CDT-mediated (100 nM) or Hd-CDT-mediated (5 pM) arrest of HeLa cells in G 2 /M was determined in the absence or presence of BFA (0.2 g/ml). Data are rendered as individual histograms representative of those collected during three independent experiments. Histograms indicate the number of cells (y axis, same scale for each histogram) at a given propidium iodide fluorescence intensity (x axis, same scale for each histogram), with, as indicated in the top left histogram, the left peak representing cells in G 0 /G 1 phase (designated as G 1 ) of the cell cycle; the right peak representing cells in G 2 /M (designated as G 2 ), and the area between the peaks representing cells in S phase (designated as S). The results are rendered as bar graphs generated from data combined from three or more independent experiments that compare the percentage of cells arrested in G 2 /M in untreated cells (white bars), cells treated with either Ec-CDT or Hd-CDT, as indicated, in the absence of BFA (black bars), or cells treated with BFA in the absence or presence of Ec-CDT or Hd-CDT as indicated (gray bars). Error bars represent standard deviations. Statistical significance was calculated for the differences between cell populations incubated in the absence or presence of BFA. mediated but not Ec-CDT-mediated G 2 /M cell cycle arrest (Fig.  3E). The sensitivity of cells to Ec-CDT was modestly, but reproducibly, elevated in the presence of agents that block acidification of endosomal compartments (Fig. 3), but we do not currently understand the basis for the slight enhancement in Ec-CDT cellular activity.
Essentially identical results were obtained when using CHO-K1 cells in the presence of NH 4 Cl, bafilomycin A1, monensin, or nigericin (data not shown), indicating that the disparate effects of agents that block endosomal acidification on Ec-CDT-or Hd-CDT-mediated cell cycle arrest in G 2 /M are not idiosyncratic to HeLa cells. Additional studies confirmed that neither the cellular binding nor the uptake of either Ec-CDT or Hd-CDT into cells from the plasma membrane was affected by NH 4 Cl, bafilomycin A1, monensin, or nigericin (data not shown).
To more quantitatively assess whether the presence of the His tag on the amino terminus of the recombinant CDT subunits might alter CDT intracellular trafficking, we compared the dose-response curves of the His-tagged and non-Histagged forms of Ec-CDT and Hd-CDT in the presence or absence of NH 4 Cl or bafilomycin A1, both of which inhibited Hd-CDT-mediated G 2 /M cell cycle arrest but not Ec-CDT-mediated G 2 /M cell cycle arrest. These studies revealed that the dose-response curves of the His-tagged and non-His-tagged forms of Ec-CDT and Hd-CDT were essentially identical in the presence or absence of NH 4 Cl or bafilomycin A1 (supplemental Fig. S4). Taken together, these data indicated that Ec-CDTmediated G 2 /M cell cycle arrest does not require acidification of endosomal compartments, suggesting that Ec-CDT and Hd-CDT are transported from the cell surface to the ER by different pathways.
Effects of Inhibiting Endosomal Acidification on Toxin Transport to the ER-Although an earlier study indicated that Hd-CDT-mediated G 2 /M cell cycle arrest in HeLa cells was inhibited by several agents that block endosome acidification (42), the importance of endosome acidification for localization of Ec-CDT or Hd-CDT to the ER, the organelle from which CdtB subunits have been proposed to be translocated to the cytosol (5), has not been reported. To evaluate whether acidification of endosomal vesicles is required for intracellular toxin transport to the ER, we used fluorescence microscopy to determine whether Ec-CDT or Hd-CDT localization to the ER is altered in the presence of NH 4 Cl. These studies revealed that Ec-CDT localization to the ER is not visibly altered in the presence of NH 4 Cl (Fig. 4, A, B, and E). In contrast, Hd-CDT localization to the ER was significantly reduced in the presence of NH 4 Cl (Fig.  4, C-E). Several attempts to evaluate CDT localization specifically to the ER using biochemical fractionation approaches were inconclusive due to the inability to satisfactorily resolve the ER from other membrane-containing fractions, with the exception of the nucleus. These results indicate that transport to the ER of Hd-CDT, but not Ec-CDT, requires acidification of endosomal vesicles and, moreover, suggests that Ec-CDT and Hd-CDT may be transported to the ER via different trafficking pathways.
Effects of Endosomal Acidification Inhibitors on CdtB Localization to the Nucleus-Previous reports that CdtB possesses in vitro DNase I-like activity (10, 11) and localizes to the nucleus when expressed ectopically within the cytosol of mammalian cells (12)(13)(14) support a model that this CDT subunit functions within the nucleus of intoxicated cells. To evaluate whether acidification of endosomal vesicles was required for intracellular toxin transport to the nucleus, we fractionated cells intoxicated for 60 min with Ec-CDT or Hd-CDT in the absence or presence of bafilomycin A1, and we used Western blotting to determine whether the localization of the CdtB subunit was altered when endosomal acidification was prevented. These studies revealed that most of the internalized Ec-CdtB or Hd-CdtB was associated with the nuclear fraction (Fig. 5), whereas neither CdtB subunit was associated with either the cytosolic or microsomal fractions. In contrast to the CdtB subunits, neither the CdtA nor CdtC subunits of Ec-CDT or Hd-CDT were detected within the nuclear fractions (data not shown), which is consistent with the model that only the A fragments of retrograde-trafficked bacterial toxins are translocated out of the ER (9) and, more recently, with immunofluorescence studies of the intracellular trafficking of CDT from A. actinomycetemcomitans (47).
Ec-CdtB was detected within the nuclear fractions prepared from cells that had been incubated with Ec-CDT holotoxin in the presence or absence of bafilomycin A1 (Fig. 5). In contrast, Hd-CdtB was not detected in either the nuclear fraction or whole cell lysates prepared from cells that had been incubated with Hd-CDT holotoxin in the presence of bafilomycin A1 (Fig. 5). We speculate that in the presence of bafilomycin A1, Hd-CdtB is either degraded within the endolysosomal system or, alternatively, recycled back to the cell surfaces and released, although we did not investigate either of these possibilities further. These results indicate that the transport of Hd-CdtB, but not Ec-CdtB, to the nucleus requires acidification of endosomal vesicles.
Comparing the Importance of Late Endosome-targeted Carrier Vesicle Biogenesis for Ec-CDT-and Hd-CDT-mediated H2AX Phosphorylation-One of the consequences resulting from inhibition of the vacuolar ATPase-mediated pH drop is inhibition of endosomal carrier vesicle formation, which facilitates transport between endosomal compartments (48). To more directly evaluate the importance of late endosome-targeted carrier vesicle transport for the intracellular activity of Ec-CDT or Hd-CDT, we compared toxin-dependent phosphorylation of H2AX in cells expressing a DN form of Rab7, a small GTPase required for late endosome-targeted carrier vesicle biogenesis (49). Transiently transfected HeLa cells expressing Rab7 (T22N) fused to DsRed, (DN-DsRed-Rab7 (T22N)), which is defective in nucleotide exchange and has a reduced affinity for GTP (50), were incubated with either Ec-CDT or Hd-CDT. After 8 h, the monolayers were examined using fluorescence microscopy to quantify the number of cells with phosphorylated H2AX (p-H2AX). Because ϳ40 -60% of the cells in any well were discovered to be expressing DN-DsRed-Rab7 (T22N), H2AX phosphorylation could be monitored within the same well in both transfected and nontransfected cells. These studies revealed significantly fewer DN-DsRed-Rab7 (T22N)expressing cells with p-H2AX activation than in nontransfected cells (Fig. 6), supporting the idea that toxin trafficking from early to late endosomal compartments is important for the biological activity of Hd-CDT but not Ec-CDT.
Comparing Importance of Late Endosome-targeted Carrier Vesicle Biogenesis for Hd-CdtB Trafficking to the Nucleus-Because the biological activity of CDTs is generally considered to require their CdtB subunits to function within the nucleus of intoxicated cells (5), we hypothesized that DN-DsRed-Rab7 (T22N) blocked the cellular activity of Hd-CDT by inhibiting the transport of Hd-CdtB to the nucleus. To evaluate this hypothesis, the nuclear localization of Ec-CdtB and Hd-CdtB was compared within the transiently transfected HeLa cells expressing DN-DsRed-Rab7 (T22N) to nontransfected HeLa cells. These studies revealed significantly less Hd-CdtB localized to the nucleus of cells expressing DN-DsRed-Rab7 (T22N)  MARCH 15, 2013 • VOLUME 288 • NUMBER 11 than in nontransfected cells (Fig. 7). In contrast, Ec-CdtB was localized to the nucleus to approximately the same extent in cells expressing or not expressing DN-DsRed-Rab7 (T22N) (Fig. 7), indicating that late endosome-targeted carrier vesicle biogenesis is important for intracellular trafficking of Hd-CdtB but not Ec-CdtB.

Distinct Cell Interactions of E. coli and H. ducreyi CDTs
Comparing Localization of Ec-CDT-and Hd-CDT to Rab9enriched Vesicles-Because late endosome-targeted carrier vesicle biogenesis is required to transport cargo from early to late vesicles within the endolysosomal system, we hypothesized that Hd-CdtB but not Ec-CdtB is transported through late endosomal vesicles. To evaluate this hypothesis, we used fluorescence imaging to investigate the localization of Ec-CdtB and Hd-CdtB to vesicles enriched with the small GTPase Rab9, which contributes to the generation and maintenance of late endocytic compartments (51). These experiments revealed that after 30 min visibly more Hd-CdtB co-localized with Rab9-enriched puncta than Ec-CdtB (Fig. 8), supporting the hypothesis that Hd-CdtB, but not Ec-CdtB, is transported through late endosomal vesicles. Taken together with the data presented above indicating the importance of late endosome-targeted carrier vesicle biogenesis for Hd-CDT but not Ec-CDT, these results suggest that the catalytic CdtB subunits of these two toxins are trafficked along distinct pathways.
Relative Cell Binding of Ec-CDT and Hd-CDT-Analogous to all intracellularly acting bacterial exotoxins (9), CDTs must bind to the surface of target cells prior to internalization (16,52). Studies to evaluate the binding of Ec-CDT and Hd-CDT to HeLa cells as a function of toxin concentration revealed that both Ec-CDT and Hd-CDT yielded saturable binding curves (Fig. 9, A and B). Moreover, the presence of a 100-fold molar excess of nonbiotinylated toxins inhibited the dose-dependent cell association of biotinylated toxins, indicating that the binding of both Ec-CDT and Hd-CDT was largely specific, with K d values for specific binding of 172 (Ϯ27) nM (Fig. 9A) and 169 (Ϯ35) nM (Fig. 9B), respectively.
Despite binding to the cell surface with similar properties, it was not clear whether Ec-CDT and Hd-CDT were binding to the same or discrete plasma membrane receptors. To differentiate between these possibilities, we evaluated whether Ec-CDT and Hd-CDT bind competitively to the surface of HeLa cells.
These studies revealed that the binding of biotinylated Ec-CDT to the cell surface was inhibited to a greater extent by 100-fold molar excess of unlabeled Ec-CDT than a 100-fold molar excess of unlabeled Hd-CDT (Fig. 10A). In a similar manner, cell surface binding of biotinylated Hd-CDT was found to be inhibited to a greater extent by 100-fold molar excess of unlabeled Hd-CDT than a 100-fold molar excess of unlabeled Ec-CDT (Fig.  10B). These data suggest that although the interactions of both Ec-CDT and Hd-CDT with host cells are largely specific, the two toxins may bind to distinct cell surface components. In further support of this idea, the cell surface binding of Ec-CDT, but not Hd-CDT, was partially inhibited in the presence of EEA, a fucose-specific lectin that was previously reported to antagonize Ec-CDT binding to HeLa cells (Fig. 10C) (15).

DISCUSSION
The studies presented here revealed differences in the manner in which CDTs from two unrelated pathogenic bacteria intoxicate mammalian cells. Hd-CDT induces G 2 /M cell cycle arrest at substantially lower concentrations than Ec-CDT in all tested cell lines (supplemental Fig. S1), consistent with a recent report demonstrating that Hd-CDT induces H2AX activation within cells at lower concentrations than Ec-CDT (21). The high level of dissimilarity between the protein sequences of the CdtA and CdtC subunits (22 and 19% sequence identity, respectively) does not readily reveal the reasons underlying the divergent potencies of Ec-CDT and Hd-CDT, but it is consistent with the idea that these two toxins may interact with and intoxicate cells by disparate mechanisms. Indeed, although both Ec-CDT and Hd-CDT possess similar in vitro DNase activities exhibited by their catalytic CdtB subunits (Fig. 1), both are efficiently taken up by cells (supplemental Fig. S3), and both require an intact Golgi complex to induce G 2 /M cell cycle arrest (Fig. 2), divergence was discovered in the cellular requirements associated with intracellular toxin transport (Figs. [3][4][5][6][7][8]. Hd-CDT-mediated G 2 /M arrest and ER localization were inhibited by agents that prevent acidification of endosomal  compartments, although Ec-CDT-mediated G 2 /M arrest and ER localization were not affected. How might the requirement for acidification of endosomal compartments relate to the cyclomodulatory activities of Ec-CDT and Hd-CDT? Intoxication of cells by a subset of intracellularly acting bacterial toxins requires a drop of the luminal pH within endosomal compartments (9, 53-56), which is a normal step in vesicle maturation carried out by the proton-pumping vacuolar ATPase (57)(58)(59). For diphtheria, anthrax, or botulinum toxins, the pH drop induces conformational changes in the toxin structure required for insertion into the endosomal membrane and translocation of the catalytic fragment into the cytosol (9, 60). However, preventing acidification of endosomal compartments can also stall the trafficking of proteins along the endosomal pathway, such as the mannose 6-phosphate receptor that is normally transported from late endosomal vesicles to the Golgi complex (61,62). Inhibition of the vacuolar ATPasemediated pH drop inhibits trafficking by blocking the formation of endosomal carrier vesicles, which facilitate transport between endosomal compartments (48). The finding that the presence of NH 4 Cl or bafilomycin A1 significantly inhibited localization of Hd-CDT to the ER (Fig. 4) or nucleus (Fig. 5), respectively, suggests that Hd-CDT intoxication requires an intracellular transport pathway involving late endosomal compartments. This idea is further supported by the importance of Rab7, which regulates the biogenesis at early endosomes of transport vesicles targeted for late endosomal compartments (Figs. 6 and 7), as well the co-localization of Hd-CdtB with Rab9 (Fig. 8).
In contrast, our data support a model that Ec-CDT transport to the ER occurs by a mechanism independent of late endosome-mediated trafficking. Notably, several reports have indicated that the retrograde transport mechanisms of other toxins, including Shiga toxin (63), cholera toxin (64,65), or ricin (66), are also insensitive to agents that prevent endosomal acidification, suggesting that these toxins are transported from early endocytic vesicles to the ER through a pathway that bypasses late endosomal compartments. Additional experimental work will be required to more completely define the Ec-CDT and Hd-CDT intracellular trafficking mechanisms, but we speculate that the overall pathway of Ec-CDT transport to the ER may be more similar to that used by Shiga toxin, cholera toxin, or ricin than the pathway used by Hd-CDT.
The molecular basis underlying the putative targeting of Ec-CDT and Hd-CDT to discrete intracellular trafficking path- FIGURE 8. Evaluating the co-localization of Ec-CdtB or Hd-CdtB with Rab9. HeLa cells that had been prechilled on ice for 30 min were incubated on ice with either Ec-CDT (100 nM) or Hd-CDT (100 nM), each of which had also been prechilled on ice. After 30 min, the cells were washed once with PBS, pH 7.4, and then further incubated at 37°C. After 30 min, the cells were fixed and imaged using fluorescence microscopy. A and B, images were representative of those collected from three independent experiments. The data were rendered as a single z-plane (ϳ5 m depth within each cell). As labeled, green puncta indicate either Ec-CdtB or Hd-CdtB; red puncta indicate that Rab9 and Ec-CdtB or Hd-CdtB co-localized with Rab9 are indicated by yellow puncta. White bars indicate 10 m. The solid white boxes are digitally enlarged images of the smaller dashed white boxes. C, results are rendered as a bar graph generated from data combined from three independent experiments that compare the Pearson's correlation coefficient for Ec-CdtB or Hd-CdtB co-localized with Rab9 in cells treated with Ec-CDT or Hd-CDT, as indicated. Error bars represent standard deviations. Statistical significance was calculated for differences in the Pearson's correlation coefficient between cell populations incubated with Ec-CDT or Hd-CDT.  (1-200 nM) in the absence (indicated on the graphs as total binding, empty circles (A) or squares (B)) or presence (indicated on the graphs as nonspecific binding, filled triangles (A) or diamonds (B)) of 100-fold molar excess of nonbiotinylated Ec-CDT (A) or Hd-CDT (B). Specific binding (indicated on the graphs as specific binding, filled circles (A) or squares (B)) was computationally derived by subtracting the nonspecific binding data from the total binding data. The normalized data from three independent experiments, each conducted in triplicate, were combined. K d values were calculated using nonlinear regression and indicated directly on the graphs, with the error bars indicating standard deviations.
ways is unclear. However, based on the inability of Ec-CDT or Hd-CDT to competitively bind to the surface of cells (Fig. 10, A  and B), we hypothesize that the two toxins utilize distinct cell surface receptors. Inhibition of the cell surface binding of Ec-CDT, but not Hd-CDT, by the fucose-specific lectin EEA (Fig.  10C) further supports this idea. Receptor recognition and binding is critical for the function of intracellularly acting toxins because the receptor, in part, functions as a molecular "carrier" whose normal uptake and trafficking through the cell directly impacts the final cellular destination of the toxin (9). A cell surface receptor for Hd-CDT has not yet been identified, but a genetic screen using the KBM7 chronic myeloid leukemia cell line revealed that a putative G protein-coupled receptor, TMEM181, contributes to cellular binding and sensitivity to Ec-CDT (67). Our competition studies (Fig. 10) suggest that Hd-CDT and Ec-CDT do not share the same receptor, but we cannot rule out the possibility of a role for TMEM181 in conferring cell sensitivity to Hd-CDT. Consistent with the idea that Ec-CDT and Hd-CDT may bind to different receptors on the surface of sensitive cells, a recent study reported that a mutant CHO cell line, characterized by abbreviated glycan sequences on membrane glycoproteins and glycolipids, was hypersensitive to Hd-CDT but demonstrated the same sensitivity as the parental CHO cells to Ec-CDT (21).
Comparing the molecular and structural basis for Ec-CDT and Hd-CDT receptor recognition is currently challenging because, even though the crystal structure has been solved for Hd-CDT (22), high resolution structural data are not yet available for the Ec-CDT holotoxin. The highly dissimilar protein sequences of the CdtA and CdtC subunits of Ec-CDT and Hd-CDT are consistent with the idea that these two toxins may interact with and intoxicate cells by disparate mechanisms. Additional studies will also be required to determine whether or not differences in intracellular trafficking pathways between Ec-CDT and Hd-CDT directly contribute to the disparity in the potencies exhibited by the two toxins toward sensitive cells. Nonetheless, we speculate that differential cell surface requirements for toxin association is critical for targeting Ec-CDT and Hd-CDT to their respective uptake and intracellular trafficking pathways.
Multiple factors could potentially contribute to the differences in cellular potencies of Ec-CDT and Hd-CDT. For example, discrepancies in the cytosolic localization of the respective CdtB subunits, which is widely thought to precede localization to the nucleus, may contribute to differences in toxin potencies. In studies to compare the levels of Ec-CdtB and Hd-CdtB within the cytosol of intoxicated cells, we could not detect Ec-CdtB and Hd-CdtB by either fluorescence microscopy or cellular fractionation and Western blot analysis (data not shown). We speculate that the levels of Ec-CdtB and Hd-CdtB within the cytosol in these experiments were below our detection limits. A recent study reported the detection of a fluorescent version of CdtB from A. actinomycetemcomitans within the cytosol and nucleus (47). However, in another recent study (68), the authors suggested that Hd-CdtB localization to the nucleus does not require that this subunit first be translocated to the cytosol.
In summary, we have identified differences in the intoxication pathways used by CDTs from pathogens that colonize two distinct niches. Because CDTs are generally believed to disrupt the normal functions of epithelial and immune cells comprising the mucosal barrier (3,4), we speculate that Ec-CDT and Hd-CDT may have evolved divergently in response to the specific tissue and cell tropisms of the pathogenic microbes that produce these toxins. Finally, these results provide experimental evidence that caution must be applied when extrapolating the properties of individual CDTs to the entire family of these toxins.