Carbohydrate-Binding Specificity of the Escherichia coli Cytolethal Distending Toxin CdtA-II and CdtC-II Subunits
http://www.100md.com
感染与免疫杂志 2005年第4期
Division of Cell Biology and Biophysics, School of Biological Sciences, University of Missouri—Kansas City, Kansas City, Missouri
ABSTRACT
Intoxication by cytolethal distending toxin depends on assembly of CdtB, the active A component of this AB toxin, with the cell surface-binding (B) component, composed of the CdtA-CdtC heterodimer, to form the active holotoxin. Here we examine the cell surface binding properties of Escherichia coli-derived CdtA-II (CdtA-IIEc) and CdtC-IIEc and their capacity to provide a binding platform for CdtB-IIEc. Using a flow cytometry-based binding assay, we demonstrate that CdtB-IIEc binds to the HeLa cell surface in a CdtA-IIEc- and CdtC-IIEc-dependent manner and that CdtA-IIEc and CdtC-IIEc compete for the same structure on the HeLa cell surface. Preincubation of cells with glycoproteins (thyroglobulin and fetuin), but not simple sugars, blocks surface binding of CdtA-IIEc and CdtC-IIEc. Moreover, CdtA-IIEc and CdtC-IIEc bind immobilized fetuin and thyroglobulin as well as fucose and to a lesser degree N-acetylgalactoseamine and N-acetylglucoseamine. Removal of N- but not O-linked carbohydrates from fetuin and thyroglobulin prevents binding of CdtA-IIEc and CdtC-IIEc to these glycoproteins. In addition, removal of N- but not O-linked surface sugar attachments prevents CDT-IIEc intoxication. To characterize the cell surface ligand recognized by CdtA-IIEc and CdtC-IIEc, lectins having various carbohydrate specificities were used to block CDT activity and the cell surface binding of CdtA-IIEc and CdtC-IIEc. Pretreatment of cells with AAA, SNA-I, STA, UEA-I, GNA, and NPA partially or completely blocked CDT activity. AAA, EEA, and UEA-I lectins, all having specificity for fucose, blocked surface binding of CdtA-IIEc and CdtC-IIEc. Together, our data indicate that CdtA-IIEc and CdtC-IIEc bind an N-linked fucose-containing structure on HeLa cells.
INTRODUCTION
Cytolethal distending toxin (CDT) is a heterotrimeric protein toxin produced by several pathogenic bacteria. CDT activity is dependent on the expression of three closely linked genes, cdtA, cdtB, and cdtC, encoding proteins with molecular masses of approximately 27, 29, and 20 kDa, respectively (1, 9, 19, 23, 25, 28). Several reports show that the CDT holotoxin is composed of all three proteins bound together to from a tripartite complex (9, 13, 15, 22, 24). Considerable evidence supports the notion that CdtB is the biologically active subunit. The direct introduction or expression of CdtB in mammalian cells results in all the cytotoxic effects associated with the CDT holotoxin (7, 10, 12). The CdtB subunit is a homolog of mammalian type I DNase (7, 12), and DNase activity associated with CdtB is required for biological activity. Several reports now document that chromosomal DNA is the target of CDT action (3, 4, 8, 11). Following cellular entry by clathrin-dependent receptor-mediated endocytosis, CDT-containing vesicles fuse with downstream compartments involving sites within the Golgi (2). Once free of the Golgi, nuclear translocation of CdtB is mediated by nuclear localizing sequences found at the carboxy-terminal end of Escherichia coli CdtB-II (CdtB-IIEc) (16) and near the amino-terminal end of the CdtB of Actinobacillus actinomycetemcomitans (CdtBAa) (18). CdtB-dependent chromosomal strand breakage triggers DNA damage-dependent mitotic arrest and/or apoptosis (3, 4, 7, 8, 10-12).
Accumulating evidence suggests that CdtA and/or CdtC is responsible for the binding and/or internalization of the CDT holotoxin. Deng and Hansen previously reported that the CdtA and CdtC subunits of Haemophilis ducreyi (CdtAHd and CdtCHd, respectively) form a complex that binds to the HeLa cell surface and blocks subsequent intoxication by CDT holotoxin (5). Unlike the CdtAC heterodimer, individual CdtAHd and CdtCHd subunits failed to block the action of holotoxin, suggesting that the CdtAC heterodimer is the functional cell surface-binding component of CDT (5). Recently, Lee et al. demonstrated that CdtA and CdtC of C. jejuni (CdtACj and CdtCCj, respectively) bound independently to HeLa cells and exhibited competitive binding for one another, suggesting that these subunits bind to the same cell surface structure (14). Competitive binding similar to that observed by Lee et al. (14) was also observed for CdtA-IIEc and CdtC-IIEc (16). Unlike CdtA-IIEc and CdtC-IIEc, CdtB-IIEc does not bind the cell surface except in the presence of both CdtA-IIEc and CdtC-IIEc (16). Taken together, these data support the hypothesis put forth by Lara-Tejero and Galan (13) that CDT is an AB type toxin in which CdtB is the active A component and CdtA and CdtC comprise the B (binding) component. Recent X-ray diffraction and three-dimensional structure modeling of the H. ducreyi CDT holotoxin suggests that CdtAHd and CdtCHd are ricin B chain-like lectin molecules that associate to form a scaffold for CdtBHd association and a binding domain for the cell surface (17).
This study was originated to examine the role that CdtA-IIEc and CdtC-IIEc play in CDT binding and intoxication and to characterize the interaction of these subunits with the cell surface. We report here that cell surface carbohydrates play a key role in CDT subunit binding and subsequent intoxication. Our findings indicate CdtA-IIEc and CdtCEc are carbohydrate-binding proteins that bind N-linked carbohydrate moieties on the cell surface and provide a scaffold for CdtB-IIEc binding. The characteristics of the putative CDT receptor are discussed.
MATERIALS AND METHODS
CDT nomenclature. General reference to CDT or the CDT subunits CdtA, CdtB, and CdtC will be as written. Specific reference to a particular CDT will include a subscript designating the bacterial species of origin as follows: CDTAa, A. actinomycetemcomitans; CDTCj, Campylobacter jejuni; CDTEc, E. coli, with sub-varieties IEc, IIEc, and IIIEc; and CDTHd, H. ducreyi.
Bacterial strains, plasmids, and culture conditions. The previously described plasmid pG3 containing the CDT-IIEc operon (7) was used in the expression of CDTEc holotoxin as described below. E. coli XL1 Blue (Stratagene, La Jolla, Calif.) was used for general cloning experiments and plasmid isolation. E. coli BL21 (DE3) (Invitrogen, Carlsbad, Calif.) was used for expression of CdtA-IIEc. E. coli TOP10 (Invitrogen) was used to express CdtB-IIEc-His6 and CdtC-IIEc. The arabinose-inducible expression vector pBAD/HisB containing the E. coli cdtB-II gene (pBAD-EcCdtB-II) was used as the source of His-tagged CdtB-IIEc (CdtB-IIEc-His6) (6). The expression vector pET16b containing the E. coli cdtA-II gene was used as the source for His-tagged CdtA-IIEc (16). The expression vector pBAD/gIII containing the E. coli cdtC-II gene was used as a source for CdtC-IIEc (16). Bacterial strains were grown on L agar plates or in L broth at 37°C containing the following antibiotics and chemicals when appropriate: carbenicillin (100 μg/ml), arabinose (0.02%), and IPTG (isopropyl--D-thiogalactopyranoside; 0.5 mM). HeLa cells (American Type Culture Collection, Manassas, Va.) were maintained in Dulbecco's minimal essential medium (DMEM) containing L-glutamine, 10% fetal calf serum, 100 mg of streptomycin per ml, and 100 IU of penicillin per ml at 37°C and 5% CO2.
Purification of CDT subunits. The expression and purification of CdtB-IIEc were performed as described by Elwell et al. (6). The expression and purification of CdtA-IIEc and CdtC-IIEc were performed as described previously by McSweeney and Dreyfus (16).
CDT activity and cell cycle distribution analysis. Unless specified, the CDT holotoxin used in all experiments was a polymyxin B extract of the periplasmic contents of E. coli XL1 Blue (pG3) (7). Holotoxin activity was assessed by DNA content-based cell cycle distribution analysis of CDT-treated HeLa cells as determined by flow cytometry. An aliquot of CDT was added to 5 x 105 HeLa cells in 100-mm-diameter culture dishes containing 5 ml of complete medium 24 h before assay. The amount of CDT required to cause a 50% block in the cell cycle after 24 h of incubation was designated 1 effective dose (ED). In most experiments, cells were treated with 3 EDs of CDT. This amount of toxin consistently blocked >95% of the cell population at the G2/M transition point after 24 h. After CDT treatment, HeLa cells were washed in phosphate-buffered saline (PBS) and removed from the culture dishes by treatment with trypsin. Cells were washed in PBS, fixed in 70% ethanol for 1 h on ice, and following removal of ethanol by washing in PBS, cells were stained with propidium iodide (50 μg of propidium iodide per ml, 1 mg of sodium citrate per ml, 0.3% NP-40, and 20 μg of RNase per ml) for 1 h at room temperature. Cellular fluorescence was analyzed by flow cytometry with a FACSCalibur cytometer (Becton Dickinson, San Jose, Calif.). The data from 104 cells were collected with the CellQuest acquisition software (Becton Dickinson). The data were then analyzed with the ModFitLT (Verity Software House, Inc., Topsham, Maine) software package to determine cell cycle distribution.
Enzyme and glycosylation inhibitors. HeLa cells (2 x 105) cultured in 100-mm-diameter dishes were treated with neuraminidase (0.2 U/ml) in 50 mM NaPO4, pH 6.0, peptide-N-glycosidase F (PNGase F) (10,000 U/ml) in 50 mM NaPO4, pH 7.5, or O-glycosidase (0.2 U/ml) in 50 mM NaPO4, pH 6.0, for 1 h at 37°C (Calbiochem, San Diego, Calif.). Control cell samples were incubated in the appropriate buffers but without glycolytic enzyme. Following enzyme treatments, HeLa cell cultures were washed with PBS to remove enzymes and incubated with CDT or PBS for 20 min at 37°C. Following the 20-min incubation, cells were washed twice with PBS, complete DMEM was added to the tissue culture plates, and the cells were incubated for 24 h at 37°C in an atmosphere containing 5% CO2. To inhibit the synthesis of N- or O-linked glycosylated proteins, HeLa cells were incubated with tunicamycin (10 μg/ml) or benzyl-GalNAc (10 μM) (Sigma-Aldrich, St. Louis, Mo.), respectively, in DMEM for 16 h at 37°C and 5% CO2. Following these treatments, the cells were washed in PBS and incubated with CDT for 20 min. Finally, the samples were washed with PBS to remove free CDT and incubated in fresh DMEM for 24 h. Cycle analysis of enzyme- and drug-treated cells was performed as described above.
Cell surface binding assay. The cell surface binding of CdtA-IIEc, CdtB-IIEc, and CdtC-IIEc was examined by a flow cytometry protocol based on a method described by Warner et al. (26). HeLa cells were detached from 100-mm-diameter culture dishes by incubation for 5 min with 5 ml of 0.5 mM EDTA in PBS. After detachment, the cells were washed three times with PBS to remove the EDTA and enumerated. Equal molar amounts (1 nmol) of CdtA-IIEc, CdtB-IIEc, and CdtC-IIEc, alone or in various combinations, were combined with 5 x 105 HeLa cells in 1 ml of PBS. For samples containing multiple subunit combinations, the subunits were combined and allowed to incubate for 30 min at room temperature prior to the addition of HeLa cells. Following 1.5 h of incubation, the HeLa cells were washed twice in cold PBS to remove unbound CDT subunits and then incubated with either anti-CdtA-IIEc (1:1,000), anti-CdtB-IIEc (1:1,000), or anti-CdtC-IIEc (1:5,000) antibodies (16) for 1 h on ice. Following this incubation, cells were washed free of unbound antibody and then incubated for 1 h on ice with Alexa 488-conjugated chicken anti-rabbit immunoglobulin G (IgG; 1:1,000) (Molecular Probes, Eugene, Oreg.). Fluorescently labeled cells were then analyzed by flow cytometry. For lectin interference experiments, HeLa cells (5 x 105) suspended in PBS were incubated with 100 μg of the various lectins (described below) for 1 h prior to the addition of the CDT subunits. Following the addition of CDT subunits, binding was assessed by flow cytometry. A single population of cells, identified by forward- and side-scatter emission, was analyzed for fluorescence intensity on the fluorescein isothiocyanate emission channel of a FACSCalibur flow cytometer (BD Biosciences). Background was set with a cell suspension of approximately 106 cells per ml that were incubated with primary and secondary antibody but without the addition of CDT subunits. Gain and voltage settings were adjusted to position the background (control) population within in the first decade of fluorescence emission. For each experiment, a minimum of 104 cellular events were recorded.
Carbohydrate- and lectin-mediated inhibition of CDT activity. Fucose, lactose, ovalbumin, mannose, N-acetylgalactosamine (GalNAc), N-acetylglucosamine (GlcNAc), fetuin, asialofetuin, bovine submaxillary mucin (BSM), 1-glycoprotein, thyroglobulin, and transferrin receptor were purchased from Sigma Aldridge. The following lectins were purchased from EY Laboratories (San Mateo, Calif.): EEA, ECA, AAA, SNA-I, STA, UEA-I, GNA, NPA, ConA, CSA, WGA, MPA, and SBA. (For definitions of the lectin abbreviations, see Table 2.) For carbohydrate inhibition experiments, 3 EDs of native CDT holotoxin was combined with various concentrations of sugars or glycoproteins (dissolved in PBS) in a total volume of 1 ml. After 1 h of incubation at room temperature, the holotoxin-sugar or holotoxin-glycoprotein mixtures were added to 2 x 105 HeLa cells in six-well plates and incubated at 37°C for 20 min. The samples were washed twice with 5 ml of PBS to remove unbound toxin. Fresh DMEM was added to the cells that were then incubated for 24 h at 37°C in atmosphere containing 5% CO2. Cell cycle distribution analysis was performed as described above.
For lectin inhibition experiments, individual lectins (100 μg) were added to 2 x 105 HeLa cells in six-well plates and incubated for 1 h at room temperature. Unbound lectins were washed away with two 5-ml PBS washes. Following removal of unbound lectins, 3 EDs of CDT holotoxin were added to the lectin-treated cells, which were further incubated for 20 min at 37°C in atmosphere containing 5% CO2. The treated cells were washed twice with PBS to remove unbound toxin and suspended in 5 ml of complete DMEM followed by plating in 100-mm-diameter culture dishes. After 24 h of incubation at 37°C in atmosphere containing 5% CO2, cell cycle distribution analysis was performed as described above.
Binding of CdtA-IIEc and CdtC-IIEc to immobilized carbohydrates and glycoproteins. Agarose bead matrices coupled with lactose, mannose, fucose, GalNAc, GlcNAc, fetuin, thyroglobulin, and BSM were obtained from EY Laboratories. Binding assays were performed in PBS with 25 μg of either CdtA-IIEc or CdtC-IIEc and 100 μg of the carbohydrate gel in a final volume of 300 μl. Samples were mixed on a tube rotator at 4°C for 1.5 h. Following incubation, the tubes were centrifuged at 4,000 x g for 5 min at 4°C. The supernatant fraction was removed, and the pellets were washed twice with 1 ml of ice-cold PBS. The pellets were suspended in 300 μl of sodium dodecyl sulfate (SDS) loading buffer and boiled, and aliquots were loaded on a 12% polyacrylamide gel and separated by SDS-polyacrylamide gel electrophoresis (PAGE). The separated proteins were transferred to polyvinylidene difluoride (PVDF) membranes by electroblotting, blocked with 5% skim milk, and incubated for 2 h with anti-CdtA-IIEc, anti-CdtB-IIEc, or anti-CdtC-IIEc antibodies. Following incubation and subsequent washing to remove excess primary antibody, the PVDF membranes were incubated with alkaline phosphatase-conjugated goat anti-rabbit IgG (Bio-Rad, Hercules, Calif.), washed to remove excess secondary antibody, and then developed with 5-bromo-4chloro-3-indolyl phosphate and nitroblue tetrazolium (Bio-Rad).
RESULTS
Binding of CDT subunits to HeLa cells. Several lines of evidence implicate CdtB as the biologically active CDT subunit. CdtB alone, however, is not cytotoxic unless introduced into the cell by microinjection, electroporation, or expression from a transgene construct. The normal route for triggering the cytolethal activity of CDT requires the presence of all three CDT subunits. Recent reports indicate that CdtA and CdtC are the binding subunits of the CDT holotoxin (5, 14, 16). These data are supported by structural analysis of the CDTHd holotoxin, indicating that CdtAHd and CdtCHd possess structural folds that resemble those of the ricin B chain (17). To investigate the association of CDT subunits to HeLa cells, we developed a flow cytometry-based binding assay modeled after a previous study by Warner et al. (26). In this assay, live, unfixed HeLa cells are incubated with the CDT subunits, either alone or in combination. Cells are then incubated with the appropriate primary antibody followed by Alexa 488-conjugated secondary antibody and examined by flow cytometry. A subsequent increase in the surface fluorescence intensity above background is indicative of subunit binding. Cells incubated with CdtA-IIEc or CdtC-IIEc alone displayed an approximate 10-fold increase in fluorescence above the background (Fig. 1A and C). The background cell population was represented by control cells incubated in the presence of primary and secondary antibody but without CDT subunits (Fig. 1A and C). The levels of binding, or fluorescence intensity, observed with either CdtA-IIEc or CdtC-IIEc alone were assigned arbitrary scores of 3+, as shown to the right of Fig. 1A and C, respectively. In the case of the CdtA-IIEc binding, fluorescence was decreased when an equimolar amount of CdtC-IIEc was included in the reaction mixture (Fig. 1A). The same pattern was observed with the binding of CdtC-IIEc in the presence of CdtA-IIEc (Fig. 1C). When the binding of CdtA-IIEc was examined in the presence of CdtB-IIEc, the results were the same as those observed with CdtA-IIEc alone (binding pattern shown to the right of Fig. 1A). When the binding of CdtA-IIEc was examined in the presence of CdtB-IIEc and CdtC-IIEc, the pattern was identical to that of CdtA-IIEc in the presence of CdtC-IIEc (binding pattern to the right of Fig. 1A): that is to say, the fluorescence intensity signal for CdtA-IIEc was reduced to the 2+ level and CdtB-IIEc had no influence on the binding of CdtA-IIEc. These results were identical to those observed for the binding of CdtC-IIEc in the presence of CdtA-IIEc and/or CdtB-IIEc (binding pattern to the right of Fig. 1C). These observations suggested that CdtA-IIEc and CdtC-IIEc compete for the same binding site on HeLa cells. By comparison to the cell surface binding capacity of CdtA-IIEc and CdtC-IIEc, CdtB-IIEc bound poorly, if at all, to HeLa cells in the absence of CdtA-IIEc and CdtC-IIEc (Fig. 1B). The presence of either CdtA-IIEc or CdtC-IIEc alone had no influence on CdtB-IIEc binding; however, in the presence of both CdtA-IIEc and CdtC-IIEc subunits, the cell surface binding of CdtB-IIEc was equivalent to that of CdtA-IIEc and CdtC-IIEc (Fig. 1B).
Competitive binding of CdtA-IIEc and CdtC-IIEc subunits. To further explore the apparent competitive nature of CdtA-IIEc and CdtC-IIEc binding, we tested the binding of each subunit in the presence of increasing concentrations of the other subunit. The standard binding assay with either CdtA-IIEc or CdtC-IIEc added alone (1 nmol of subunit per 5 x 105 cells) was performed, and results identical to those shown in Fig. 1 were obtained (Fig. 2). The bold black histograms represent the binding of CdtA-IIEc and CdtC-IIEc individually, whereas, as before, the gray areas represent the background fluorescence of cells incubated with primary and secondary antibodies in the absence of CDT subunits (Fig. 2). In the case of CdtA-IIEc binding, the presence of one-, two-, and fourfold molar excess of CdtC-IIEc in the binding reaction (denoted by dashed gray histograms) steadily decreased the fluorescence intensity to the level of the background (Fig. 2). Identical results were obtained for the binding of CdtC-IIEc; that is, increasing amounts of CdtA-IIEc in the reaction mixture reduced the binding of CdtC-IIEc. These data support the notion that CdtA-IIEc and CdtC-IIEc compete for the same binding site or binding ligand on the HeLa cell surface.
Relationship of CdtA-IIEc and CdtC-IIEc to other known proteins. The binding patterns observed for CDT subunits suggested that CdtA-IIEc and CdtC-IIEc independently bind to the HeLa cell surface and that the binding of CdtB-IIEc is dependent upon both CdtA-IIEc and CdtC-IIEc. To further assess the specific function of CdtA-IIEc and CdtC-IIEc, PSI-BLAST analysis was performed on these protein sequences in an attempt to infer functionality based on sequence similarity. Results of a PSI-BLAST search suggested that CdtA-II-IIEc is related to a number of carbohydrate-binding proteins with either galactose or mannose specificity, including ricin B chain, abrin B chain, mistletoe lectin I B chain, and nigrin B chain. In addition, CdtA-IIEc displayed apparent similarity to carbohydrate-specific enzymes, including N-acetylgalactosaminyltransferase, -mannase, xylanase, and -galactosidase. All of the lectins and enzymes having similarity to CdtA-IIEc had final E-values of less than 1e–24. These data inferred that CdtA-IIEc may function as a carbohydrate-binding protein, the target for which may be a cell surface structure. The result of a PSI-BLAST analysis of CdtC-IIEc was less conclusive than that with CdtA-IIEc. After the final PSI-BLAST iteration, proteins identified with significant similarity to CdtC-IIEc were the CdtC and CdtA proteins from other bacterial species. Early iterations of the CdtC-IIEc PSI-BLAST analysis resulted in the same list of lectins and carbohydrate-modifying enzymes present in the final search analysis for CdtA-IIEc. These observations suggest that CdtA-IIEc and CdtC-IIEc may have a similar structural fold and/or the same function.
Glycoproteins but not simple sugars inhibit CDT activity. Based on the PSI-BLAST results for CdtA-IIEc and CdtC-IIEc and X-ray diffraction analysis of CDT holotoxin (17), it is reasonable to speculate that CdtA-IIEc and CdtC-IIEc bind to carbohydrate-containing receptors on the target cell surface. In this set of experiments, we examined the effect of various simple sugars and glycoproteins on the toxicity of CDT. CDT was preincubated with individual simple sugars or glycoproteins before the addition to HeLa cells for a brief exposure. The ligands tested were lactose, mannose, fucose, GalNAc, GlcNAc, fetuin, asialofetuin, thyroglobulin, ovalbumin, 1-glycoprotein, and transferrin receptor. Following preincubation of CDT with blocking sugars and glycoproteins, the mixture was added to HeLa cells and toxicity was assessed by an examination of the cell cycle distribution of treated and untreated cells following 24 h of incubation (Materials and Methods). None of the simple sugars tested inhibited CDT action at doses as high as 1 mg/ml (Table 1). In contrast, thyroglobulin, BSM, and, 1-glycoprotein inhibited the effects of CDT, with various half-maximal inhibitory concentrations ranging from 75 to 550 μg/ml (Table 1). Thyroglobulin, the strongest inhibitor of CDT activity, resulted in a 50% decrease in toxin activity at a dose of 75 μg/ml. BSM and 1-glycoprotein blocked CDT activity by 50% at concentrations of 350 and 550 μg/ml, respectively (Table 1). These data suggested that complex carbohydrates, such as those on glycoproteins, may inhibit CDT activity by interfering with the binding of CDT to the cell surface.
CdtA-IIEc and CdtC-IIEc binding to carbohydrates. Based on the above results, we examined the binding of CdtA-IIEc and CdtC-IIEc to immobilized carbohydrates and glycoproteins. The binding of CdtA-IIEc and CdtC-IIEc to immobilized sugar and sugar-containing ligands was assessed with the following sugar- or glycoprotein-conjugated agarose beads: lactose, mannose, fucose, GalNAc, GlcNAc, fetuin, thyroglobulin, and BSM. CdtA-IIEc and CdtC-IIEc were incubated with the carbohydrate- or glycoprotein-conjugated matrices, which were subsequently washed free of unbound CDT subunits. The gel matrices were then prepared for SDS-PAGE analysis, and CDT subunit binding was assessed by Western blot analysis. CdtA-IIEc and CdtC-IIEc bound the same set of ligands, including thyroglobulin, fetuin, and fucose (Fig. 3A) To a lesser extent, CdtA-IIEc and CdtC-IIEc also bound to GalNAc- and GlcNAc-agarose (Fig. 3A). Binding of CdtA-IIEc and CdtC-IIEc to the unmodified agarose matrix was negligible, as was the binding of these subunits to mannose and lactose (Fig. 3A).
Fetuin and thyroglobulin are glycoproteins containing both N- and O-linked complex carbohydrates (21, 27). In addition, the carbohydrate moieties on these glycoproteins may or may not contain terminal sialic acid residues (21, 27). In an attempt to identify the linkage specificity of the carbohydrate moiety bound by CdtA-IIEc and CdtC-IIEc, fetuin- and thyroglobulin-coupled agarose matrices were subjected to various enzyme pretreatments prior to incubation with the CDT subunits. The binding of CdtA-IIEc and CdtC-IIEc to fetuin- and thyroglobulin-coupled matrices was completely abolished by pretreatment with PNGase F to remove N-linked sugars (Fig. 3B). Pretreatment of the glycoprotein-conjugated matrices with O-glycosidase or neuraminidase to remove O-linked sugars and sialic acid residues, respectively, had no effect on CdtA-IIEc and CdtC-IIEc binding (not shown).
N-linked sugars on the HeLa cell surface mediate CDT-dependent intoxication. A hallmark of CDT action is disruption of cell cycle progression. Histograms depicting DNA content of HeLa cell following various treatments are shown in Fig. 4. Cells in G0 or G1 occupy the left-most population of cells in each histogram representing a DNA content of 2N (Fig. 4). Cells at the G2/M transition of the cell cycle possess 4N DNA content and thus occupy the right-most peak of each histogram (Fig. 4). Compared to untreated HeLa cells (Fig. 4), cells treated with CDT accumulated at the G2/M transition of the cell cycle (Fig. 4). To determine whether an interaction of CDT with cell surface carbohydrates was necessary for cellular intoxication, we first removed or prevented the formation of N- and O-linked surface glycan structures on HeLa cells prior to treatment with CDT. Treatment of HeLa cells with benzyl-GalNAc, an inhibitor of O-linked oligosaccharide glycoprotein attachments, had no effect on subsequent intoxication (Fig. 4). Likewise, pretreatment of cells with O-glycosidase to remove O-linked carbohydrate structures had no effect on subsequent CDT intoxication (Fig. 4). In contrast to these observations, cells pretreated with tunicamycin to block the surface expression of N-glycosylation completely blocked the action of CDT (Fig. 4). Likewise, treatment of HeLa cells with PNGase F to remove surface-N-linked oligosaccharides also blocked subsequent CDT intoxication (Fig. 4). Together, these data suggest that HeLa cells bear N-linked surface carbohydrate moieties involved in CDT intoxication. Presumably, the N-linked sugar residues are responsible for CDT binding to the HeLa cell surface.
In preliminary experiments, we noted that pretreatment of HeLa cells with neuraminidase to remove exposed sialic acid residues did not appear to inhibit CDT action. Instead, the data were suggested that neuraminidase pretreatment may enhance CDT activity. To verify this, HeLa cells were either pretreated, or not, with neuraminidase to remove terminal sialic acid residues from surface glycoproteins and glycolipids and then treated with 1 ED of CDT. By definition, 1 ED of CDT is that amount which induces a 50% G2/M cell cycle block in 24 h (Materials and Methods). As expected, approximately 50% of the CDT-treated cells not pretreated with neuraminidase were blocked at the G2/M transition (Fig. 4). Neuraminidase pretreatment followed by 1 ED of CDT, however, consistently resulted in >95% of the cells blocked at the G2/M interface (Fig. 4). Treatment with neuraminidase alone had no effect on the cell cycle distribution of HeLa cells (not shown).
Several carbohydrate-binding proteins identified by PSI-BLAST analysis as having similarity to CdtA-IIEc and CdtC-IIEc had binding specificities of galactose and/or mannose oligosaccharides. We examined the effect of CDT on HeLa cells following pretreatment with endo--N-acetylglucosaminidase H (endoH) and - or -galactosidase for removal of mannose and galactose residues, respectively. These pretreatments had no affect on the subsequent action of CDT (not shown).
Lectins block CDT activity on HeLa cells. In an attempt to define the cell surface carbohydrates recognized by CDT, HeLa cells were pretreated with lectins of known carbohydrate specificity and then assessed for sensitivity to CDT. Of the lectins tested (Table 2), EEA was the most effective in consistently blocking CDT activity. Other lectins that reduced CDT activity were AAA, SNA-I, STA, UEA-I, GNA, and NPA (Table 2). CDT activity on HeLa cells was not blocked following pretreatment with ECA, ConA, CSA, WGA, MPA, and SBA. The carbohydrate specificity for EEA is Gal(1,3)[fuc(1,2)]Gal. Of the lectins that partially blocked CDT activity, the carbohydrate specificities are as follows: -L-fucose, UEA-I; -L-fucose, AAA; NeuNAc(2,6)GalNAc or lactose, SNA-I; (1,4)GlcNAc oligomers, STA; -D-mannose, NPA; and mannose, GNA (Table 2). These data suggest that oligosaccharides containing fucose, galactose, and/or mannose may be responsible for the cell surface binding of CDT.
Effect of lectins on cell surface binding of CdtA-IIEc and CdtC-IIEc. If CdtA-IIEc and/or CdtC-IIEc is responsible for holotoxin binding to the cell surface, the inhibitory effect of certain lectins on CDT toxicity is likely the result of inhibition of CdtA-IIEc and/or CdtC-IIEc binding to the cell surface. We therefore investigated the capacity of various lectins to block the cell surface binding of CdtA-IIEc and CdtC-IIEc. HeLa cells were preincubated with the lectins that reduced or diminished CDT activity (AAA, ConA, EEA, VAA, NPA, SNA-I, SNA, and UEA-I). Following preincubation with lectins, HeLa cells were then incubated with CdtA-IIEc or CdtA-IIEc alone or with all three CDT subunits. The binding of individual CDT subunits was then assessed by flow cytometry. As with the data from previous binding experiments, the shaded histograms in each panel of Fig. 5 represent the background level of fluorescence obtained following incubation of HeLa cells with primary and secondary antibody without the addition of CDT subunits. A second control consisting of HeLa cells incubated with the various lectins followed by primary and secondary antibody was also performed. The results of these controls to assess the potential cross-reactivity between CDT subunit-specific antibody and the various lectins to be tested were identical to those with the standard background fluorescence controls (not shown). The bold black-traced histograms shifted to the right of background in each panel of Fig. 5 represent the binding of either CdtA-IIEc (left panels) or CdtC-IIEc (right panels) in the absence of lectin preincubation to HeLa cells and thus are identical to the results shown in Fig. 1. The light gray-traced histograms in each panel represent the fluorescence intensity yielded by subunit interaction with HeLa cells following preincubation with the various lectins shown to the left of each panel. Superimposition (or near superimposition) of the gray trace with the bold black trace, such as observed following preincubation with GNA, NPA, SNA, and STA lectins, indicated that lectin preincubation did not interfere with the capacity of HeLa cells to bind CdtA-IIEc or CdtC-IIEc. Lectin interference as observed with AAA, EEA, and UEA resulted in a leftward shift or downward shift in fluorescence intensity, indicating that the binding activity of CdtA-IIEc or CdtC-IIEc was blocked or inhibited by lectin preincubation. The consequences of lectin preincubation with HeLa cells were the same for CdtA-IIEc and CdtC-IIEc binding, suggesting again that CdtA-IIEc and CdtC-IIEc have binding specificity for the same HeLa cell surface structure. In addition to these results, we tested the effect of lectin preincubation on CdtB-IIEc binding in the presence of CdtA-IIEc combined with CdtC-IIEc. In each case, CdtB-IIEc binding was dependent upon CdtA-IIEc and CdtC-IIEc binding. That is to say, CdtB-IIEc binding to HeLa cells was observed when the cells were preincubated with noninterfering lectins (GNA, NPA, SNA, and STA) but not after preincubation with interfering lectins (AAA, EEA, and UEA). The carbohydrate specificity of both AAA and UEA-I is fucose, whereas, EEA recognizes fucose/galactose oligomers.
DISCUSSION
Accumulating evidence suggests that CDT is an example of the AB class of bacterial toxins in which CdtB is the active A component and the CdtAC heterodimer is the cell surface-binding B component. Although the role of CdtB in intoxication is well established, the participation of CdtA and CdtC in cell surface binding is less well characterized. Deng and Hansen previously reported that the CdtA and CdtC subunits of H. ducreyi form a complex that binds to the HeLa cell surface and blocks subsequent intoxication by CDT holotoxin (5). However, individual CdtAHd and CdtCHd subunits failed to block the action of the holotoxin, suggesting that the CdtACHd heterodimer is the functional cell surface-binding component of CDT. More recently, Lee et al. (14) and McSweeney and Dreyfus (16) demonstrated that CdtA and CdtC from C. jejuni and E. coli, respectively, independently bind the surface of target cells. Previously reported results that were extended here indicate that the cell surface binding of CdtA and CdtC is competitive, suggesting that these subunits recognize the same cell surface ligand (14) (16). In addition, CdtB-IIEc fails to bind to the cell surface except in the presence of both CdtA-IIEc and CdtC-IIEc (16).
The requirement for both CdtA-IIEc and CdtC-IIEc in CDT intoxication coupled with the lack of binding of CdtB-IIEc to target cells in the absence of CdtA-IIEc and CdtC-IIEc is supported by the proposed model for the three-dimensional structure of the H. ducreyi holotoxin (17). The proposed structural model for the CDTHd holotoxin suggests that the CdtAHd and CdtCHd subunits form a dimer with two functional faces (17). One CdtAC face binds the cell surface, while the other CdtACHd face binds CdtBHd. Of particular interest to this report is the proposed cell surface-binding domain of the CdtAHd-CdtCHd dimer. The CdtAHd domain of the proposed cell surface binding face formed by the CdtACHd contact region contains a patch of aromatic residues highly conserved among CdtAs of all species of origin. Mutagenesis of four conserved residues (W91G, W98G, W100G, and Y102A) eliminated toxicity but preserved the ability of the CdtAHd mutant to form a holotoxin structure with CdtBHd and CdtCHd (17). The authors propose that this region defines the cell surface-binding of CDTHd. It is interesting to note that these conserved residues are all contained within an in-frame 43-amino-acid C. jejuni CdtA deletion mutant prepared by Lee et al. (14). The CdtACj deletion mutant still bound HeLa cells but was unable to participate in the formation of CDT holotoxin and thus was inactive. In addition, the CdtACj deletion mutant competed with CDT holotoxin for cell surface binding. These authors suggested that residues contained within the 43-amino-acid deletion were involved in the association of CdtACj and CdtCCj (14). The aromatic patch identified on the surface of CdtAHd by structural analysis fits a projected model for a carbohydrate-binding domain on a globular protein (20). The amino acids defining the binding activity of CdtA and CdtC, as well as the CdtAC dimer, although presently unknown, will undoubtedly lie within the surface area of the CdtAC contact region as defined for the H. ducreyi CDT by Nesic and Stebbins (17).
In this report, we examined the binding of CdtA-IIEc and CdtC-IIEc to HeLa cells, using a flow cytometry-based binding assay as a method to better define the binding activities of the CDT subunits. Our findings are consistent with the X-ray diffraction model for the CDT holotoxin that suggests carbohydrate-binding roles for CdtA-IIEc and CdtC-IIEc (17). Here we demonstrate for the first time functional lectin-like activity for both CdtA-IIEc and CdtC-IIEc. Both CDT subunits bound various immobilized carbohydrates, including fucose, GalNAc, and GlcNAc, to various degrees. We also examined the binding of CdtA-IIEc and CdtC-IIEc to fetuin and thyroglobulin, two model glycoproteins containing well-characterized carbohydrate linkages. Thyroglobulin contains 16 confirmed N-linked glycosylation sites, 8 of which are linked to complex oligosaccharide units containing fucose, galactose, mannose, and glucosamine (27). Other confirmed N-linkage sites are coupled to a mixture of high mannose, galactose, and glucosamine. N-linked sugars on fetuin are highly fucosylated and carry the fucose- and galactose-containing Lewis X or asialo-Lewis X epitopes (21). CdtA-IIEc and CdtC-IIEc both bound efficiently to each of these glycoproteins. The increase in CDT activity for HeLa cells observed following treatment of cells with neuraminidase suggests that removal of terminal sialic acid residues exposes additional binding sites for CdtA-IIEc and CdtC-IIEc. Release of the N-linked sugars from thyroglobulin and fetuin completely abolished their binding capacity for CdtA-IIEc and CdtC-IIEc. However, O-linked sugars were apparently not involved in CDT subunit binding, since removal of these linkages from the model glycoproteins had no effect on CdtA-IIEc or CdtC-IIEc binding. In addition to these data, treatment of HeLa cells with PNGase F to remove N-linked glycan structures or tunicamycin to prevent N glycosylation completely inhibited subsequent CDT intoxication. Treatments to block, or remove, O-linked sugars from HeLa cells had no effect on subsequent toxicity, suggesting again that O-linked sugars are not involved in the association of CDT with target cells.
Lectins with a number of different specificities including fucose, galactose, mannose GalNAc, and GlcNAc reduced or diminished subsequent CDT intoxication. These results were in contrast to the subunit blocking experiments, in which only three lectins, AAA, EEA, and UEA, blocked the binding of CdtA-IIEc and CdtC-IIEc to HeLa cells. This apparent discrepancy may indicate that lectins that block CDT activity, but do not block binding of individual CdtA-IIEc and CdtC-IIEc subunits, sterically interfere with holotoxin binding. The common ligand specificity of the three lectins that block individual subunits binding is fucose.
These data, coupled with the results of our lectin blocking experiments suggest that CdtA-IIEc and CdtC-IIEc bind N-linked fucose containing complex carbohydrates on the HeLa cell surface. A determination of the specific limits of the carbohydrate-binding specificity of CdtA-IIEc and CdtC-IIEc awaits verification by additional carbohydrate binding analyses. We are presently attempting to refine the carbohydrate specificity of the cell surface ligands bound by CdtA-IIEc and CdtC-IIEc and define the subunit residues responsible for the carbohydrate-binding activity described in this report.
ACKNOWLEDGMENTS
This research was supported by a grant from the National Institutes of Health (AI47999).
REFERENCES
1. Cope, L. D., S. Lumbley, J. L. Latimer, J. Klesney-Tait, M. K. Stevens, L. S. Johnson, M. Purven, R. S. Munson, Jr., T. Lagergrd, J. D. Radolf, and E. J. Hansen. 1997. A diffusible cytotoxin of Haemophilus ducreyi. Proc. Natl. Acad. Sci. USA 94:4056-4061.
2. Cortes-Bratti, X., E. Chaves-Olarte, T. Lagergard, and M. Thelestam. 2000. Cellular internalization of cytolethal distending toxin from Haemophilus ducreyi. Infect. Immun. 68:6903-6911.
3. Cortes-Bratti, X., T. Frisan, and M. Thelestam. 2001. The cytolethal distending toxins induce DNA damage and cell cycle arrest. Toxicon 39:1729-1736.
4. Cortes-Bratti, X., C. Karlsson, T. Lagergard, M. Thelestam, and T. Frisan. 2001. The Haemophilus ducreyi cytolethal distending toxin induces cell cycle arrest and apoptosis via the DNA damage checkpoint pathways. J. Biol. Chem. 276:5296-5302.
5. Deng, K., and E. J. Hansen. 2003. A CdtA-CdtC complex can block killing of HeLa cells by Haemophilus ducreyi cytolethal distending toxin. Infect. Immun. 71:6633-6640.
6. Elwell, C., K. Chao, K. Patel, and L. Dreyfus. 2001. Escherichia coli CdtB mediates cytolethal distending toxin cell cycle arrest. Infect. Immun. 69:3418-3422.
7. Elwell, C. A., and L. A. Dreyfus. 2000. DNase I homologous residues in CdtB are critical for cytolethal distending toxin-mediated cell cycle arrest. Mol. Microbiol. 37:952-963.
8. Frisan, T., X. Cortes-Bratti, E. Chaves-Olarte, B. Stenerlow, and M. Thelestam. 2003. The Haemophilus ducreyi cytolethal distending toxin induces DNA double-strand breaks and promotes ATM-dependent activation of RhoA. Cell. Microbiol. 5:695-707.
9. Frisk, A., M. Lebens, C. Johansson, H. Ahmed, L. Svensson, K. Ahlman, and T. Lagergard. 2001. The role of different protein components from the Haemophilus ducreyi cytolethal distending toxin in the generation of cell toxicity. Microb. Pathog. 30:313-324.
10. Hassane, D. C., R. B. Lee, M. D. Mendenhall, and C. L. Pickett. 2001. Cytolethal distending toxin demonstrates genotoxic activity in a yeast model. Infect. Immun. 69:5752-5759.
11. Hassane, D. C., R. B. Lee, and C. L. Pickett. 2003. Campylobacter jejuni cytolethal distending toxin promotes DNA repair responses in normal human cells. Infect. Immun. 71:541-545.
12. Lara-Tejero, M., and J. E. Galan. 2000. A bacterial toxin that controls cell cycle progression as a deoxyribonuclease I-like protein. Science 290:354-357.
13. Lara-Tejero, M., and J. E. Galan. 2001. CdtA, CdtB, and CdtC form a tripartite complex that is required for cytolethal distending toxin activity. Infect. Immun. 69:4358-4365.
14. Lee, R. B., D. C. Hassane, D. L. Cottle, and C. L. Pickett. 2003. Interactions of Campylobacter jejuni cytolethal distending toxin subunits CdtA and CdtC with HeLa cells. Infect. Immun. 71:4883-4890.
15. Mao, X., and J. M. DiRienzo. 2002. Functional studies of the recombinant subunits of a cytolethal distending holotoxin. Cell. Microbiol. 4:245-255.
16. McSweeney, L. A., and L. A. Dreyfus. 2004. Nuclear localization of the Escherichia coli cytolethal distending toxin CdtB subunit. Cell. Microbiol. 6:447-458.
17. Nesic, D., Y. Hsu, and C. E. Stebbins. 2004. Assembly and function of a bacterial genotoxin. Nature 429:429-433.
18. Nishikubo, S., M. Ohara, Y. Ueno, M. Ikura, H. Kurihara, H. Komatsuzawa, E. Oswald, and M. Sugai. 2003. An N-terminal segment of the active component of the bacterial genotoxin cytolethal distending toxin B (CDTB) directs CDTB into the nucleus. J. Biol. Chem. 278:50671-50681.
19. Pickett, C. L., D. L. Cottle, E. C. Pesci, and G. Bikah. 1994. Cloning, sequencing, and expression of the Escherichia coli cytolethal distending toxin genes. Infect. Immun. 62:1046-1051.
20. Rini, J. M. 1995. Lectin structure. Annu. Rev. Biophys. Biomol. Struct. 24:551-577.
21. Ritchie, M. A., A. C. Gill, M. J. Deery, and K. Lilley. 2002. Precursor ion scanning for detection and structural characterization of heterogeneous glycopeptide mixtures. J. Am. Soc. Mass Spectrom. 13:1065-1077.
22. Saiki, K., K. Konishi, T. Gomi, T. Nishihara, and M. Yoshikawa. 2001. Reconstitution and purification of cytolethal distending toxin of Actinobacillus actinomycetemcomitans. Microbiol. Immunol. 45:497-506.
23. Scott, D. A., and J. B. Kaper. 1994. Cloning and sequencing of the genes encoding Escherichia coli cytolethal distending toxin. Infect. Immun. 62:244-251.
24. Shenker, B. J., D. Besack, T. McKay, L. Pankoski, A. Zekavat, and D. R. Demuth. 2004. Actinobacillus actinomycetemcomitans cytolethal distending toxin (Cdt): evidence that the holotoxin is composed of three subunits: CdtA, CdtB, and CdtC. J. Immunol. 172:410-417.
25. Sugai, M., T. Kawamoto, S. Y. Peres, Y. Ueno, H. Komatsuzawa, T. Fujiwara, H. Kurihara, H. Suginaka, and E. Oswald. 1998. The cell cycle-specific growth-inhibitory factor produced by Actinobacillus actinomycetemcomitans is a cytolethal distending toxin. Infect. Immun. 66:5008-5019.
26. Warner, S., C. A. Hartley, R. A. Stevenson, N. Ficorilli, A. Varrasso, M. J. Studdert, and B. S. Crabb. 2001. Evidence that Equine rhinitis A virus VP1 is a target of neutralizing antibodies and participates directly in receptor binding. J. Virol. 75:9274-9281.
27. Yang, S. X., H. G. Pollock, and A. B. Rawitch. 1996. Glycosylation in human thyroglobulin: location of the N-linked oligosaccharide units and comparison with bovine thyroglobulin. Arch. Biochem. Biophys. 327:61-70.
28. Young, V. B., K. A. Knox, and D. B. Schauer. 2000. Cytolethal distending toxin sequence and activity in the enterohepatic pathogen Helicobacter hepaticus. Infect. Immun. 68:184-191.(Leslie A. McSweeney and L)
ABSTRACT
Intoxication by cytolethal distending toxin depends on assembly of CdtB, the active A component of this AB toxin, with the cell surface-binding (B) component, composed of the CdtA-CdtC heterodimer, to form the active holotoxin. Here we examine the cell surface binding properties of Escherichia coli-derived CdtA-II (CdtA-IIEc) and CdtC-IIEc and their capacity to provide a binding platform for CdtB-IIEc. Using a flow cytometry-based binding assay, we demonstrate that CdtB-IIEc binds to the HeLa cell surface in a CdtA-IIEc- and CdtC-IIEc-dependent manner and that CdtA-IIEc and CdtC-IIEc compete for the same structure on the HeLa cell surface. Preincubation of cells with glycoproteins (thyroglobulin and fetuin), but not simple sugars, blocks surface binding of CdtA-IIEc and CdtC-IIEc. Moreover, CdtA-IIEc and CdtC-IIEc bind immobilized fetuin and thyroglobulin as well as fucose and to a lesser degree N-acetylgalactoseamine and N-acetylglucoseamine. Removal of N- but not O-linked carbohydrates from fetuin and thyroglobulin prevents binding of CdtA-IIEc and CdtC-IIEc to these glycoproteins. In addition, removal of N- but not O-linked surface sugar attachments prevents CDT-IIEc intoxication. To characterize the cell surface ligand recognized by CdtA-IIEc and CdtC-IIEc, lectins having various carbohydrate specificities were used to block CDT activity and the cell surface binding of CdtA-IIEc and CdtC-IIEc. Pretreatment of cells with AAA, SNA-I, STA, UEA-I, GNA, and NPA partially or completely blocked CDT activity. AAA, EEA, and UEA-I lectins, all having specificity for fucose, blocked surface binding of CdtA-IIEc and CdtC-IIEc. Together, our data indicate that CdtA-IIEc and CdtC-IIEc bind an N-linked fucose-containing structure on HeLa cells.
INTRODUCTION
Cytolethal distending toxin (CDT) is a heterotrimeric protein toxin produced by several pathogenic bacteria. CDT activity is dependent on the expression of three closely linked genes, cdtA, cdtB, and cdtC, encoding proteins with molecular masses of approximately 27, 29, and 20 kDa, respectively (1, 9, 19, 23, 25, 28). Several reports show that the CDT holotoxin is composed of all three proteins bound together to from a tripartite complex (9, 13, 15, 22, 24). Considerable evidence supports the notion that CdtB is the biologically active subunit. The direct introduction or expression of CdtB in mammalian cells results in all the cytotoxic effects associated with the CDT holotoxin (7, 10, 12). The CdtB subunit is a homolog of mammalian type I DNase (7, 12), and DNase activity associated with CdtB is required for biological activity. Several reports now document that chromosomal DNA is the target of CDT action (3, 4, 8, 11). Following cellular entry by clathrin-dependent receptor-mediated endocytosis, CDT-containing vesicles fuse with downstream compartments involving sites within the Golgi (2). Once free of the Golgi, nuclear translocation of CdtB is mediated by nuclear localizing sequences found at the carboxy-terminal end of Escherichia coli CdtB-II (CdtB-IIEc) (16) and near the amino-terminal end of the CdtB of Actinobacillus actinomycetemcomitans (CdtBAa) (18). CdtB-dependent chromosomal strand breakage triggers DNA damage-dependent mitotic arrest and/or apoptosis (3, 4, 7, 8, 10-12).
Accumulating evidence suggests that CdtA and/or CdtC is responsible for the binding and/or internalization of the CDT holotoxin. Deng and Hansen previously reported that the CdtA and CdtC subunits of Haemophilis ducreyi (CdtAHd and CdtCHd, respectively) form a complex that binds to the HeLa cell surface and blocks subsequent intoxication by CDT holotoxin (5). Unlike the CdtAC heterodimer, individual CdtAHd and CdtCHd subunits failed to block the action of holotoxin, suggesting that the CdtAC heterodimer is the functional cell surface-binding component of CDT (5). Recently, Lee et al. demonstrated that CdtA and CdtC of C. jejuni (CdtACj and CdtCCj, respectively) bound independently to HeLa cells and exhibited competitive binding for one another, suggesting that these subunits bind to the same cell surface structure (14). Competitive binding similar to that observed by Lee et al. (14) was also observed for CdtA-IIEc and CdtC-IIEc (16). Unlike CdtA-IIEc and CdtC-IIEc, CdtB-IIEc does not bind the cell surface except in the presence of both CdtA-IIEc and CdtC-IIEc (16). Taken together, these data support the hypothesis put forth by Lara-Tejero and Galan (13) that CDT is an AB type toxin in which CdtB is the active A component and CdtA and CdtC comprise the B (binding) component. Recent X-ray diffraction and three-dimensional structure modeling of the H. ducreyi CDT holotoxin suggests that CdtAHd and CdtCHd are ricin B chain-like lectin molecules that associate to form a scaffold for CdtBHd association and a binding domain for the cell surface (17).
This study was originated to examine the role that CdtA-IIEc and CdtC-IIEc play in CDT binding and intoxication and to characterize the interaction of these subunits with the cell surface. We report here that cell surface carbohydrates play a key role in CDT subunit binding and subsequent intoxication. Our findings indicate CdtA-IIEc and CdtCEc are carbohydrate-binding proteins that bind N-linked carbohydrate moieties on the cell surface and provide a scaffold for CdtB-IIEc binding. The characteristics of the putative CDT receptor are discussed.
MATERIALS AND METHODS
CDT nomenclature. General reference to CDT or the CDT subunits CdtA, CdtB, and CdtC will be as written. Specific reference to a particular CDT will include a subscript designating the bacterial species of origin as follows: CDTAa, A. actinomycetemcomitans; CDTCj, Campylobacter jejuni; CDTEc, E. coli, with sub-varieties IEc, IIEc, and IIIEc; and CDTHd, H. ducreyi.
Bacterial strains, plasmids, and culture conditions. The previously described plasmid pG3 containing the CDT-IIEc operon (7) was used in the expression of CDTEc holotoxin as described below. E. coli XL1 Blue (Stratagene, La Jolla, Calif.) was used for general cloning experiments and plasmid isolation. E. coli BL21 (DE3) (Invitrogen, Carlsbad, Calif.) was used for expression of CdtA-IIEc. E. coli TOP10 (Invitrogen) was used to express CdtB-IIEc-His6 and CdtC-IIEc. The arabinose-inducible expression vector pBAD/HisB containing the E. coli cdtB-II gene (pBAD-EcCdtB-II) was used as the source of His-tagged CdtB-IIEc (CdtB-IIEc-His6) (6). The expression vector pET16b containing the E. coli cdtA-II gene was used as the source for His-tagged CdtA-IIEc (16). The expression vector pBAD/gIII containing the E. coli cdtC-II gene was used as a source for CdtC-IIEc (16). Bacterial strains were grown on L agar plates or in L broth at 37°C containing the following antibiotics and chemicals when appropriate: carbenicillin (100 μg/ml), arabinose (0.02%), and IPTG (isopropyl--D-thiogalactopyranoside; 0.5 mM). HeLa cells (American Type Culture Collection, Manassas, Va.) were maintained in Dulbecco's minimal essential medium (DMEM) containing L-glutamine, 10% fetal calf serum, 100 mg of streptomycin per ml, and 100 IU of penicillin per ml at 37°C and 5% CO2.
Purification of CDT subunits. The expression and purification of CdtB-IIEc were performed as described by Elwell et al. (6). The expression and purification of CdtA-IIEc and CdtC-IIEc were performed as described previously by McSweeney and Dreyfus (16).
CDT activity and cell cycle distribution analysis. Unless specified, the CDT holotoxin used in all experiments was a polymyxin B extract of the periplasmic contents of E. coli XL1 Blue (pG3) (7). Holotoxin activity was assessed by DNA content-based cell cycle distribution analysis of CDT-treated HeLa cells as determined by flow cytometry. An aliquot of CDT was added to 5 x 105 HeLa cells in 100-mm-diameter culture dishes containing 5 ml of complete medium 24 h before assay. The amount of CDT required to cause a 50% block in the cell cycle after 24 h of incubation was designated 1 effective dose (ED). In most experiments, cells were treated with 3 EDs of CDT. This amount of toxin consistently blocked >95% of the cell population at the G2/M transition point after 24 h. After CDT treatment, HeLa cells were washed in phosphate-buffered saline (PBS) and removed from the culture dishes by treatment with trypsin. Cells were washed in PBS, fixed in 70% ethanol for 1 h on ice, and following removal of ethanol by washing in PBS, cells were stained with propidium iodide (50 μg of propidium iodide per ml, 1 mg of sodium citrate per ml, 0.3% NP-40, and 20 μg of RNase per ml) for 1 h at room temperature. Cellular fluorescence was analyzed by flow cytometry with a FACSCalibur cytometer (Becton Dickinson, San Jose, Calif.). The data from 104 cells were collected with the CellQuest acquisition software (Becton Dickinson). The data were then analyzed with the ModFitLT (Verity Software House, Inc., Topsham, Maine) software package to determine cell cycle distribution.
Enzyme and glycosylation inhibitors. HeLa cells (2 x 105) cultured in 100-mm-diameter dishes were treated with neuraminidase (0.2 U/ml) in 50 mM NaPO4, pH 6.0, peptide-N-glycosidase F (PNGase F) (10,000 U/ml) in 50 mM NaPO4, pH 7.5, or O-glycosidase (0.2 U/ml) in 50 mM NaPO4, pH 6.0, for 1 h at 37°C (Calbiochem, San Diego, Calif.). Control cell samples were incubated in the appropriate buffers but without glycolytic enzyme. Following enzyme treatments, HeLa cell cultures were washed with PBS to remove enzymes and incubated with CDT or PBS for 20 min at 37°C. Following the 20-min incubation, cells were washed twice with PBS, complete DMEM was added to the tissue culture plates, and the cells were incubated for 24 h at 37°C in an atmosphere containing 5% CO2. To inhibit the synthesis of N- or O-linked glycosylated proteins, HeLa cells were incubated with tunicamycin (10 μg/ml) or benzyl-GalNAc (10 μM) (Sigma-Aldrich, St. Louis, Mo.), respectively, in DMEM for 16 h at 37°C and 5% CO2. Following these treatments, the cells were washed in PBS and incubated with CDT for 20 min. Finally, the samples were washed with PBS to remove free CDT and incubated in fresh DMEM for 24 h. Cycle analysis of enzyme- and drug-treated cells was performed as described above.
Cell surface binding assay. The cell surface binding of CdtA-IIEc, CdtB-IIEc, and CdtC-IIEc was examined by a flow cytometry protocol based on a method described by Warner et al. (26). HeLa cells were detached from 100-mm-diameter culture dishes by incubation for 5 min with 5 ml of 0.5 mM EDTA in PBS. After detachment, the cells were washed three times with PBS to remove the EDTA and enumerated. Equal molar amounts (1 nmol) of CdtA-IIEc, CdtB-IIEc, and CdtC-IIEc, alone or in various combinations, were combined with 5 x 105 HeLa cells in 1 ml of PBS. For samples containing multiple subunit combinations, the subunits were combined and allowed to incubate for 30 min at room temperature prior to the addition of HeLa cells. Following 1.5 h of incubation, the HeLa cells were washed twice in cold PBS to remove unbound CDT subunits and then incubated with either anti-CdtA-IIEc (1:1,000), anti-CdtB-IIEc (1:1,000), or anti-CdtC-IIEc (1:5,000) antibodies (16) for 1 h on ice. Following this incubation, cells were washed free of unbound antibody and then incubated for 1 h on ice with Alexa 488-conjugated chicken anti-rabbit immunoglobulin G (IgG; 1:1,000) (Molecular Probes, Eugene, Oreg.). Fluorescently labeled cells were then analyzed by flow cytometry. For lectin interference experiments, HeLa cells (5 x 105) suspended in PBS were incubated with 100 μg of the various lectins (described below) for 1 h prior to the addition of the CDT subunits. Following the addition of CDT subunits, binding was assessed by flow cytometry. A single population of cells, identified by forward- and side-scatter emission, was analyzed for fluorescence intensity on the fluorescein isothiocyanate emission channel of a FACSCalibur flow cytometer (BD Biosciences). Background was set with a cell suspension of approximately 106 cells per ml that were incubated with primary and secondary antibody but without the addition of CDT subunits. Gain and voltage settings were adjusted to position the background (control) population within in the first decade of fluorescence emission. For each experiment, a minimum of 104 cellular events were recorded.
Carbohydrate- and lectin-mediated inhibition of CDT activity. Fucose, lactose, ovalbumin, mannose, N-acetylgalactosamine (GalNAc), N-acetylglucosamine (GlcNAc), fetuin, asialofetuin, bovine submaxillary mucin (BSM), 1-glycoprotein, thyroglobulin, and transferrin receptor were purchased from Sigma Aldridge. The following lectins were purchased from EY Laboratories (San Mateo, Calif.): EEA, ECA, AAA, SNA-I, STA, UEA-I, GNA, NPA, ConA, CSA, WGA, MPA, and SBA. (For definitions of the lectin abbreviations, see Table 2.) For carbohydrate inhibition experiments, 3 EDs of native CDT holotoxin was combined with various concentrations of sugars or glycoproteins (dissolved in PBS) in a total volume of 1 ml. After 1 h of incubation at room temperature, the holotoxin-sugar or holotoxin-glycoprotein mixtures were added to 2 x 105 HeLa cells in six-well plates and incubated at 37°C for 20 min. The samples were washed twice with 5 ml of PBS to remove unbound toxin. Fresh DMEM was added to the cells that were then incubated for 24 h at 37°C in atmosphere containing 5% CO2. Cell cycle distribution analysis was performed as described above.
For lectin inhibition experiments, individual lectins (100 μg) were added to 2 x 105 HeLa cells in six-well plates and incubated for 1 h at room temperature. Unbound lectins were washed away with two 5-ml PBS washes. Following removal of unbound lectins, 3 EDs of CDT holotoxin were added to the lectin-treated cells, which were further incubated for 20 min at 37°C in atmosphere containing 5% CO2. The treated cells were washed twice with PBS to remove unbound toxin and suspended in 5 ml of complete DMEM followed by plating in 100-mm-diameter culture dishes. After 24 h of incubation at 37°C in atmosphere containing 5% CO2, cell cycle distribution analysis was performed as described above.
Binding of CdtA-IIEc and CdtC-IIEc to immobilized carbohydrates and glycoproteins. Agarose bead matrices coupled with lactose, mannose, fucose, GalNAc, GlcNAc, fetuin, thyroglobulin, and BSM were obtained from EY Laboratories. Binding assays were performed in PBS with 25 μg of either CdtA-IIEc or CdtC-IIEc and 100 μg of the carbohydrate gel in a final volume of 300 μl. Samples were mixed on a tube rotator at 4°C for 1.5 h. Following incubation, the tubes were centrifuged at 4,000 x g for 5 min at 4°C. The supernatant fraction was removed, and the pellets were washed twice with 1 ml of ice-cold PBS. The pellets were suspended in 300 μl of sodium dodecyl sulfate (SDS) loading buffer and boiled, and aliquots were loaded on a 12% polyacrylamide gel and separated by SDS-polyacrylamide gel electrophoresis (PAGE). The separated proteins were transferred to polyvinylidene difluoride (PVDF) membranes by electroblotting, blocked with 5% skim milk, and incubated for 2 h with anti-CdtA-IIEc, anti-CdtB-IIEc, or anti-CdtC-IIEc antibodies. Following incubation and subsequent washing to remove excess primary antibody, the PVDF membranes were incubated with alkaline phosphatase-conjugated goat anti-rabbit IgG (Bio-Rad, Hercules, Calif.), washed to remove excess secondary antibody, and then developed with 5-bromo-4chloro-3-indolyl phosphate and nitroblue tetrazolium (Bio-Rad).
RESULTS
Binding of CDT subunits to HeLa cells. Several lines of evidence implicate CdtB as the biologically active CDT subunit. CdtB alone, however, is not cytotoxic unless introduced into the cell by microinjection, electroporation, or expression from a transgene construct. The normal route for triggering the cytolethal activity of CDT requires the presence of all three CDT subunits. Recent reports indicate that CdtA and CdtC are the binding subunits of the CDT holotoxin (5, 14, 16). These data are supported by structural analysis of the CDTHd holotoxin, indicating that CdtAHd and CdtCHd possess structural folds that resemble those of the ricin B chain (17). To investigate the association of CDT subunits to HeLa cells, we developed a flow cytometry-based binding assay modeled after a previous study by Warner et al. (26). In this assay, live, unfixed HeLa cells are incubated with the CDT subunits, either alone or in combination. Cells are then incubated with the appropriate primary antibody followed by Alexa 488-conjugated secondary antibody and examined by flow cytometry. A subsequent increase in the surface fluorescence intensity above background is indicative of subunit binding. Cells incubated with CdtA-IIEc or CdtC-IIEc alone displayed an approximate 10-fold increase in fluorescence above the background (Fig. 1A and C). The background cell population was represented by control cells incubated in the presence of primary and secondary antibody but without CDT subunits (Fig. 1A and C). The levels of binding, or fluorescence intensity, observed with either CdtA-IIEc or CdtC-IIEc alone were assigned arbitrary scores of 3+, as shown to the right of Fig. 1A and C, respectively. In the case of the CdtA-IIEc binding, fluorescence was decreased when an equimolar amount of CdtC-IIEc was included in the reaction mixture (Fig. 1A). The same pattern was observed with the binding of CdtC-IIEc in the presence of CdtA-IIEc (Fig. 1C). When the binding of CdtA-IIEc was examined in the presence of CdtB-IIEc, the results were the same as those observed with CdtA-IIEc alone (binding pattern shown to the right of Fig. 1A). When the binding of CdtA-IIEc was examined in the presence of CdtB-IIEc and CdtC-IIEc, the pattern was identical to that of CdtA-IIEc in the presence of CdtC-IIEc (binding pattern to the right of Fig. 1A): that is to say, the fluorescence intensity signal for CdtA-IIEc was reduced to the 2+ level and CdtB-IIEc had no influence on the binding of CdtA-IIEc. These results were identical to those observed for the binding of CdtC-IIEc in the presence of CdtA-IIEc and/or CdtB-IIEc (binding pattern to the right of Fig. 1C). These observations suggested that CdtA-IIEc and CdtC-IIEc compete for the same binding site on HeLa cells. By comparison to the cell surface binding capacity of CdtA-IIEc and CdtC-IIEc, CdtB-IIEc bound poorly, if at all, to HeLa cells in the absence of CdtA-IIEc and CdtC-IIEc (Fig. 1B). The presence of either CdtA-IIEc or CdtC-IIEc alone had no influence on CdtB-IIEc binding; however, in the presence of both CdtA-IIEc and CdtC-IIEc subunits, the cell surface binding of CdtB-IIEc was equivalent to that of CdtA-IIEc and CdtC-IIEc (Fig. 1B).
Competitive binding of CdtA-IIEc and CdtC-IIEc subunits. To further explore the apparent competitive nature of CdtA-IIEc and CdtC-IIEc binding, we tested the binding of each subunit in the presence of increasing concentrations of the other subunit. The standard binding assay with either CdtA-IIEc or CdtC-IIEc added alone (1 nmol of subunit per 5 x 105 cells) was performed, and results identical to those shown in Fig. 1 were obtained (Fig. 2). The bold black histograms represent the binding of CdtA-IIEc and CdtC-IIEc individually, whereas, as before, the gray areas represent the background fluorescence of cells incubated with primary and secondary antibodies in the absence of CDT subunits (Fig. 2). In the case of CdtA-IIEc binding, the presence of one-, two-, and fourfold molar excess of CdtC-IIEc in the binding reaction (denoted by dashed gray histograms) steadily decreased the fluorescence intensity to the level of the background (Fig. 2). Identical results were obtained for the binding of CdtC-IIEc; that is, increasing amounts of CdtA-IIEc in the reaction mixture reduced the binding of CdtC-IIEc. These data support the notion that CdtA-IIEc and CdtC-IIEc compete for the same binding site or binding ligand on the HeLa cell surface.
Relationship of CdtA-IIEc and CdtC-IIEc to other known proteins. The binding patterns observed for CDT subunits suggested that CdtA-IIEc and CdtC-IIEc independently bind to the HeLa cell surface and that the binding of CdtB-IIEc is dependent upon both CdtA-IIEc and CdtC-IIEc. To further assess the specific function of CdtA-IIEc and CdtC-IIEc, PSI-BLAST analysis was performed on these protein sequences in an attempt to infer functionality based on sequence similarity. Results of a PSI-BLAST search suggested that CdtA-II-IIEc is related to a number of carbohydrate-binding proteins with either galactose or mannose specificity, including ricin B chain, abrin B chain, mistletoe lectin I B chain, and nigrin B chain. In addition, CdtA-IIEc displayed apparent similarity to carbohydrate-specific enzymes, including N-acetylgalactosaminyltransferase, -mannase, xylanase, and -galactosidase. All of the lectins and enzymes having similarity to CdtA-IIEc had final E-values of less than 1e–24. These data inferred that CdtA-IIEc may function as a carbohydrate-binding protein, the target for which may be a cell surface structure. The result of a PSI-BLAST analysis of CdtC-IIEc was less conclusive than that with CdtA-IIEc. After the final PSI-BLAST iteration, proteins identified with significant similarity to CdtC-IIEc were the CdtC and CdtA proteins from other bacterial species. Early iterations of the CdtC-IIEc PSI-BLAST analysis resulted in the same list of lectins and carbohydrate-modifying enzymes present in the final search analysis for CdtA-IIEc. These observations suggest that CdtA-IIEc and CdtC-IIEc may have a similar structural fold and/or the same function.
Glycoproteins but not simple sugars inhibit CDT activity. Based on the PSI-BLAST results for CdtA-IIEc and CdtC-IIEc and X-ray diffraction analysis of CDT holotoxin (17), it is reasonable to speculate that CdtA-IIEc and CdtC-IIEc bind to carbohydrate-containing receptors on the target cell surface. In this set of experiments, we examined the effect of various simple sugars and glycoproteins on the toxicity of CDT. CDT was preincubated with individual simple sugars or glycoproteins before the addition to HeLa cells for a brief exposure. The ligands tested were lactose, mannose, fucose, GalNAc, GlcNAc, fetuin, asialofetuin, thyroglobulin, ovalbumin, 1-glycoprotein, and transferrin receptor. Following preincubation of CDT with blocking sugars and glycoproteins, the mixture was added to HeLa cells and toxicity was assessed by an examination of the cell cycle distribution of treated and untreated cells following 24 h of incubation (Materials and Methods). None of the simple sugars tested inhibited CDT action at doses as high as 1 mg/ml (Table 1). In contrast, thyroglobulin, BSM, and, 1-glycoprotein inhibited the effects of CDT, with various half-maximal inhibitory concentrations ranging from 75 to 550 μg/ml (Table 1). Thyroglobulin, the strongest inhibitor of CDT activity, resulted in a 50% decrease in toxin activity at a dose of 75 μg/ml. BSM and 1-glycoprotein blocked CDT activity by 50% at concentrations of 350 and 550 μg/ml, respectively (Table 1). These data suggested that complex carbohydrates, such as those on glycoproteins, may inhibit CDT activity by interfering with the binding of CDT to the cell surface.
CdtA-IIEc and CdtC-IIEc binding to carbohydrates. Based on the above results, we examined the binding of CdtA-IIEc and CdtC-IIEc to immobilized carbohydrates and glycoproteins. The binding of CdtA-IIEc and CdtC-IIEc to immobilized sugar and sugar-containing ligands was assessed with the following sugar- or glycoprotein-conjugated agarose beads: lactose, mannose, fucose, GalNAc, GlcNAc, fetuin, thyroglobulin, and BSM. CdtA-IIEc and CdtC-IIEc were incubated with the carbohydrate- or glycoprotein-conjugated matrices, which were subsequently washed free of unbound CDT subunits. The gel matrices were then prepared for SDS-PAGE analysis, and CDT subunit binding was assessed by Western blot analysis. CdtA-IIEc and CdtC-IIEc bound the same set of ligands, including thyroglobulin, fetuin, and fucose (Fig. 3A) To a lesser extent, CdtA-IIEc and CdtC-IIEc also bound to GalNAc- and GlcNAc-agarose (Fig. 3A). Binding of CdtA-IIEc and CdtC-IIEc to the unmodified agarose matrix was negligible, as was the binding of these subunits to mannose and lactose (Fig. 3A).
Fetuin and thyroglobulin are glycoproteins containing both N- and O-linked complex carbohydrates (21, 27). In addition, the carbohydrate moieties on these glycoproteins may or may not contain terminal sialic acid residues (21, 27). In an attempt to identify the linkage specificity of the carbohydrate moiety bound by CdtA-IIEc and CdtC-IIEc, fetuin- and thyroglobulin-coupled agarose matrices were subjected to various enzyme pretreatments prior to incubation with the CDT subunits. The binding of CdtA-IIEc and CdtC-IIEc to fetuin- and thyroglobulin-coupled matrices was completely abolished by pretreatment with PNGase F to remove N-linked sugars (Fig. 3B). Pretreatment of the glycoprotein-conjugated matrices with O-glycosidase or neuraminidase to remove O-linked sugars and sialic acid residues, respectively, had no effect on CdtA-IIEc and CdtC-IIEc binding (not shown).
N-linked sugars on the HeLa cell surface mediate CDT-dependent intoxication. A hallmark of CDT action is disruption of cell cycle progression. Histograms depicting DNA content of HeLa cell following various treatments are shown in Fig. 4. Cells in G0 or G1 occupy the left-most population of cells in each histogram representing a DNA content of 2N (Fig. 4). Cells at the G2/M transition of the cell cycle possess 4N DNA content and thus occupy the right-most peak of each histogram (Fig. 4). Compared to untreated HeLa cells (Fig. 4), cells treated with CDT accumulated at the G2/M transition of the cell cycle (Fig. 4). To determine whether an interaction of CDT with cell surface carbohydrates was necessary for cellular intoxication, we first removed or prevented the formation of N- and O-linked surface glycan structures on HeLa cells prior to treatment with CDT. Treatment of HeLa cells with benzyl-GalNAc, an inhibitor of O-linked oligosaccharide glycoprotein attachments, had no effect on subsequent intoxication (Fig. 4). Likewise, pretreatment of cells with O-glycosidase to remove O-linked carbohydrate structures had no effect on subsequent CDT intoxication (Fig. 4). In contrast to these observations, cells pretreated with tunicamycin to block the surface expression of N-glycosylation completely blocked the action of CDT (Fig. 4). Likewise, treatment of HeLa cells with PNGase F to remove surface-N-linked oligosaccharides also blocked subsequent CDT intoxication (Fig. 4). Together, these data suggest that HeLa cells bear N-linked surface carbohydrate moieties involved in CDT intoxication. Presumably, the N-linked sugar residues are responsible for CDT binding to the HeLa cell surface.
In preliminary experiments, we noted that pretreatment of HeLa cells with neuraminidase to remove exposed sialic acid residues did not appear to inhibit CDT action. Instead, the data were suggested that neuraminidase pretreatment may enhance CDT activity. To verify this, HeLa cells were either pretreated, or not, with neuraminidase to remove terminal sialic acid residues from surface glycoproteins and glycolipids and then treated with 1 ED of CDT. By definition, 1 ED of CDT is that amount which induces a 50% G2/M cell cycle block in 24 h (Materials and Methods). As expected, approximately 50% of the CDT-treated cells not pretreated with neuraminidase were blocked at the G2/M transition (Fig. 4). Neuraminidase pretreatment followed by 1 ED of CDT, however, consistently resulted in >95% of the cells blocked at the G2/M interface (Fig. 4). Treatment with neuraminidase alone had no effect on the cell cycle distribution of HeLa cells (not shown).
Several carbohydrate-binding proteins identified by PSI-BLAST analysis as having similarity to CdtA-IIEc and CdtC-IIEc had binding specificities of galactose and/or mannose oligosaccharides. We examined the effect of CDT on HeLa cells following pretreatment with endo--N-acetylglucosaminidase H (endoH) and - or -galactosidase for removal of mannose and galactose residues, respectively. These pretreatments had no affect on the subsequent action of CDT (not shown).
Lectins block CDT activity on HeLa cells. In an attempt to define the cell surface carbohydrates recognized by CDT, HeLa cells were pretreated with lectins of known carbohydrate specificity and then assessed for sensitivity to CDT. Of the lectins tested (Table 2), EEA was the most effective in consistently blocking CDT activity. Other lectins that reduced CDT activity were AAA, SNA-I, STA, UEA-I, GNA, and NPA (Table 2). CDT activity on HeLa cells was not blocked following pretreatment with ECA, ConA, CSA, WGA, MPA, and SBA. The carbohydrate specificity for EEA is Gal(1,3)[fuc(1,2)]Gal. Of the lectins that partially blocked CDT activity, the carbohydrate specificities are as follows: -L-fucose, UEA-I; -L-fucose, AAA; NeuNAc(2,6)GalNAc or lactose, SNA-I; (1,4)GlcNAc oligomers, STA; -D-mannose, NPA; and mannose, GNA (Table 2). These data suggest that oligosaccharides containing fucose, galactose, and/or mannose may be responsible for the cell surface binding of CDT.
Effect of lectins on cell surface binding of CdtA-IIEc and CdtC-IIEc. If CdtA-IIEc and/or CdtC-IIEc is responsible for holotoxin binding to the cell surface, the inhibitory effect of certain lectins on CDT toxicity is likely the result of inhibition of CdtA-IIEc and/or CdtC-IIEc binding to the cell surface. We therefore investigated the capacity of various lectins to block the cell surface binding of CdtA-IIEc and CdtC-IIEc. HeLa cells were preincubated with the lectins that reduced or diminished CDT activity (AAA, ConA, EEA, VAA, NPA, SNA-I, SNA, and UEA-I). Following preincubation with lectins, HeLa cells were then incubated with CdtA-IIEc or CdtA-IIEc alone or with all three CDT subunits. The binding of individual CDT subunits was then assessed by flow cytometry. As with the data from previous binding experiments, the shaded histograms in each panel of Fig. 5 represent the background level of fluorescence obtained following incubation of HeLa cells with primary and secondary antibody without the addition of CDT subunits. A second control consisting of HeLa cells incubated with the various lectins followed by primary and secondary antibody was also performed. The results of these controls to assess the potential cross-reactivity between CDT subunit-specific antibody and the various lectins to be tested were identical to those with the standard background fluorescence controls (not shown). The bold black-traced histograms shifted to the right of background in each panel of Fig. 5 represent the binding of either CdtA-IIEc (left panels) or CdtC-IIEc (right panels) in the absence of lectin preincubation to HeLa cells and thus are identical to the results shown in Fig. 1. The light gray-traced histograms in each panel represent the fluorescence intensity yielded by subunit interaction with HeLa cells following preincubation with the various lectins shown to the left of each panel. Superimposition (or near superimposition) of the gray trace with the bold black trace, such as observed following preincubation with GNA, NPA, SNA, and STA lectins, indicated that lectin preincubation did not interfere with the capacity of HeLa cells to bind CdtA-IIEc or CdtC-IIEc. Lectin interference as observed with AAA, EEA, and UEA resulted in a leftward shift or downward shift in fluorescence intensity, indicating that the binding activity of CdtA-IIEc or CdtC-IIEc was blocked or inhibited by lectin preincubation. The consequences of lectin preincubation with HeLa cells were the same for CdtA-IIEc and CdtC-IIEc binding, suggesting again that CdtA-IIEc and CdtC-IIEc have binding specificity for the same HeLa cell surface structure. In addition to these results, we tested the effect of lectin preincubation on CdtB-IIEc binding in the presence of CdtA-IIEc combined with CdtC-IIEc. In each case, CdtB-IIEc binding was dependent upon CdtA-IIEc and CdtC-IIEc binding. That is to say, CdtB-IIEc binding to HeLa cells was observed when the cells were preincubated with noninterfering lectins (GNA, NPA, SNA, and STA) but not after preincubation with interfering lectins (AAA, EEA, and UEA). The carbohydrate specificity of both AAA and UEA-I is fucose, whereas, EEA recognizes fucose/galactose oligomers.
DISCUSSION
Accumulating evidence suggests that CDT is an example of the AB class of bacterial toxins in which CdtB is the active A component and the CdtAC heterodimer is the cell surface-binding B component. Although the role of CdtB in intoxication is well established, the participation of CdtA and CdtC in cell surface binding is less well characterized. Deng and Hansen previously reported that the CdtA and CdtC subunits of H. ducreyi form a complex that binds to the HeLa cell surface and blocks subsequent intoxication by CDT holotoxin (5). However, individual CdtAHd and CdtCHd subunits failed to block the action of the holotoxin, suggesting that the CdtACHd heterodimer is the functional cell surface-binding component of CDT. More recently, Lee et al. (14) and McSweeney and Dreyfus (16) demonstrated that CdtA and CdtC from C. jejuni and E. coli, respectively, independently bind the surface of target cells. Previously reported results that were extended here indicate that the cell surface binding of CdtA and CdtC is competitive, suggesting that these subunits recognize the same cell surface ligand (14) (16). In addition, CdtB-IIEc fails to bind to the cell surface except in the presence of both CdtA-IIEc and CdtC-IIEc (16).
The requirement for both CdtA-IIEc and CdtC-IIEc in CDT intoxication coupled with the lack of binding of CdtB-IIEc to target cells in the absence of CdtA-IIEc and CdtC-IIEc is supported by the proposed model for the three-dimensional structure of the H. ducreyi holotoxin (17). The proposed structural model for the CDTHd holotoxin suggests that the CdtAHd and CdtCHd subunits form a dimer with two functional faces (17). One CdtAC face binds the cell surface, while the other CdtACHd face binds CdtBHd. Of particular interest to this report is the proposed cell surface-binding domain of the CdtAHd-CdtCHd dimer. The CdtAHd domain of the proposed cell surface binding face formed by the CdtACHd contact region contains a patch of aromatic residues highly conserved among CdtAs of all species of origin. Mutagenesis of four conserved residues (W91G, W98G, W100G, and Y102A) eliminated toxicity but preserved the ability of the CdtAHd mutant to form a holotoxin structure with CdtBHd and CdtCHd (17). The authors propose that this region defines the cell surface-binding of CDTHd. It is interesting to note that these conserved residues are all contained within an in-frame 43-amino-acid C. jejuni CdtA deletion mutant prepared by Lee et al. (14). The CdtACj deletion mutant still bound HeLa cells but was unable to participate in the formation of CDT holotoxin and thus was inactive. In addition, the CdtACj deletion mutant competed with CDT holotoxin for cell surface binding. These authors suggested that residues contained within the 43-amino-acid deletion were involved in the association of CdtACj and CdtCCj (14). The aromatic patch identified on the surface of CdtAHd by structural analysis fits a projected model for a carbohydrate-binding domain on a globular protein (20). The amino acids defining the binding activity of CdtA and CdtC, as well as the CdtAC dimer, although presently unknown, will undoubtedly lie within the surface area of the CdtAC contact region as defined for the H. ducreyi CDT by Nesic and Stebbins (17).
In this report, we examined the binding of CdtA-IIEc and CdtC-IIEc to HeLa cells, using a flow cytometry-based binding assay as a method to better define the binding activities of the CDT subunits. Our findings are consistent with the X-ray diffraction model for the CDT holotoxin that suggests carbohydrate-binding roles for CdtA-IIEc and CdtC-IIEc (17). Here we demonstrate for the first time functional lectin-like activity for both CdtA-IIEc and CdtC-IIEc. Both CDT subunits bound various immobilized carbohydrates, including fucose, GalNAc, and GlcNAc, to various degrees. We also examined the binding of CdtA-IIEc and CdtC-IIEc to fetuin and thyroglobulin, two model glycoproteins containing well-characterized carbohydrate linkages. Thyroglobulin contains 16 confirmed N-linked glycosylation sites, 8 of which are linked to complex oligosaccharide units containing fucose, galactose, mannose, and glucosamine (27). Other confirmed N-linkage sites are coupled to a mixture of high mannose, galactose, and glucosamine. N-linked sugars on fetuin are highly fucosylated and carry the fucose- and galactose-containing Lewis X or asialo-Lewis X epitopes (21). CdtA-IIEc and CdtC-IIEc both bound efficiently to each of these glycoproteins. The increase in CDT activity for HeLa cells observed following treatment of cells with neuraminidase suggests that removal of terminal sialic acid residues exposes additional binding sites for CdtA-IIEc and CdtC-IIEc. Release of the N-linked sugars from thyroglobulin and fetuin completely abolished their binding capacity for CdtA-IIEc and CdtC-IIEc. However, O-linked sugars were apparently not involved in CDT subunit binding, since removal of these linkages from the model glycoproteins had no effect on CdtA-IIEc or CdtC-IIEc binding. In addition to these data, treatment of HeLa cells with PNGase F to remove N-linked glycan structures or tunicamycin to prevent N glycosylation completely inhibited subsequent CDT intoxication. Treatments to block, or remove, O-linked sugars from HeLa cells had no effect on subsequent toxicity, suggesting again that O-linked sugars are not involved in the association of CDT with target cells.
Lectins with a number of different specificities including fucose, galactose, mannose GalNAc, and GlcNAc reduced or diminished subsequent CDT intoxication. These results were in contrast to the subunit blocking experiments, in which only three lectins, AAA, EEA, and UEA, blocked the binding of CdtA-IIEc and CdtC-IIEc to HeLa cells. This apparent discrepancy may indicate that lectins that block CDT activity, but do not block binding of individual CdtA-IIEc and CdtC-IIEc subunits, sterically interfere with holotoxin binding. The common ligand specificity of the three lectins that block individual subunits binding is fucose.
These data, coupled with the results of our lectin blocking experiments suggest that CdtA-IIEc and CdtC-IIEc bind N-linked fucose containing complex carbohydrates on the HeLa cell surface. A determination of the specific limits of the carbohydrate-binding specificity of CdtA-IIEc and CdtC-IIEc awaits verification by additional carbohydrate binding analyses. We are presently attempting to refine the carbohydrate specificity of the cell surface ligands bound by CdtA-IIEc and CdtC-IIEc and define the subunit residues responsible for the carbohydrate-binding activity described in this report.
ACKNOWLEDGMENTS
This research was supported by a grant from the National Institutes of Health (AI47999).
REFERENCES
1. Cope, L. D., S. Lumbley, J. L. Latimer, J. Klesney-Tait, M. K. Stevens, L. S. Johnson, M. Purven, R. S. Munson, Jr., T. Lagergrd, J. D. Radolf, and E. J. Hansen. 1997. A diffusible cytotoxin of Haemophilus ducreyi. Proc. Natl. Acad. Sci. USA 94:4056-4061.
2. Cortes-Bratti, X., E. Chaves-Olarte, T. Lagergard, and M. Thelestam. 2000. Cellular internalization of cytolethal distending toxin from Haemophilus ducreyi. Infect. Immun. 68:6903-6911.
3. Cortes-Bratti, X., T. Frisan, and M. Thelestam. 2001. The cytolethal distending toxins induce DNA damage and cell cycle arrest. Toxicon 39:1729-1736.
4. Cortes-Bratti, X., C. Karlsson, T. Lagergard, M. Thelestam, and T. Frisan. 2001. The Haemophilus ducreyi cytolethal distending toxin induces cell cycle arrest and apoptosis via the DNA damage checkpoint pathways. J. Biol. Chem. 276:5296-5302.
5. Deng, K., and E. J. Hansen. 2003. A CdtA-CdtC complex can block killing of HeLa cells by Haemophilus ducreyi cytolethal distending toxin. Infect. Immun. 71:6633-6640.
6. Elwell, C., K. Chao, K. Patel, and L. Dreyfus. 2001. Escherichia coli CdtB mediates cytolethal distending toxin cell cycle arrest. Infect. Immun. 69:3418-3422.
7. Elwell, C. A., and L. A. Dreyfus. 2000. DNase I homologous residues in CdtB are critical for cytolethal distending toxin-mediated cell cycle arrest. Mol. Microbiol. 37:952-963.
8. Frisan, T., X. Cortes-Bratti, E. Chaves-Olarte, B. Stenerlow, and M. Thelestam. 2003. The Haemophilus ducreyi cytolethal distending toxin induces DNA double-strand breaks and promotes ATM-dependent activation of RhoA. Cell. Microbiol. 5:695-707.
9. Frisk, A., M. Lebens, C. Johansson, H. Ahmed, L. Svensson, K. Ahlman, and T. Lagergard. 2001. The role of different protein components from the Haemophilus ducreyi cytolethal distending toxin in the generation of cell toxicity. Microb. Pathog. 30:313-324.
10. Hassane, D. C., R. B. Lee, M. D. Mendenhall, and C. L. Pickett. 2001. Cytolethal distending toxin demonstrates genotoxic activity in a yeast model. Infect. Immun. 69:5752-5759.
11. Hassane, D. C., R. B. Lee, and C. L. Pickett. 2003. Campylobacter jejuni cytolethal distending toxin promotes DNA repair responses in normal human cells. Infect. Immun. 71:541-545.
12. Lara-Tejero, M., and J. E. Galan. 2000. A bacterial toxin that controls cell cycle progression as a deoxyribonuclease I-like protein. Science 290:354-357.
13. Lara-Tejero, M., and J. E. Galan. 2001. CdtA, CdtB, and CdtC form a tripartite complex that is required for cytolethal distending toxin activity. Infect. Immun. 69:4358-4365.
14. Lee, R. B., D. C. Hassane, D. L. Cottle, and C. L. Pickett. 2003. Interactions of Campylobacter jejuni cytolethal distending toxin subunits CdtA and CdtC with HeLa cells. Infect. Immun. 71:4883-4890.
15. Mao, X., and J. M. DiRienzo. 2002. Functional studies of the recombinant subunits of a cytolethal distending holotoxin. Cell. Microbiol. 4:245-255.
16. McSweeney, L. A., and L. A. Dreyfus. 2004. Nuclear localization of the Escherichia coli cytolethal distending toxin CdtB subunit. Cell. Microbiol. 6:447-458.
17. Nesic, D., Y. Hsu, and C. E. Stebbins. 2004. Assembly and function of a bacterial genotoxin. Nature 429:429-433.
18. Nishikubo, S., M. Ohara, Y. Ueno, M. Ikura, H. Kurihara, H. Komatsuzawa, E. Oswald, and M. Sugai. 2003. An N-terminal segment of the active component of the bacterial genotoxin cytolethal distending toxin B (CDTB) directs CDTB into the nucleus. J. Biol. Chem. 278:50671-50681.
19. Pickett, C. L., D. L. Cottle, E. C. Pesci, and G. Bikah. 1994. Cloning, sequencing, and expression of the Escherichia coli cytolethal distending toxin genes. Infect. Immun. 62:1046-1051.
20. Rini, J. M. 1995. Lectin structure. Annu. Rev. Biophys. Biomol. Struct. 24:551-577.
21. Ritchie, M. A., A. C. Gill, M. J. Deery, and K. Lilley. 2002. Precursor ion scanning for detection and structural characterization of heterogeneous glycopeptide mixtures. J. Am. Soc. Mass Spectrom. 13:1065-1077.
22. Saiki, K., K. Konishi, T. Gomi, T. Nishihara, and M. Yoshikawa. 2001. Reconstitution and purification of cytolethal distending toxin of Actinobacillus actinomycetemcomitans. Microbiol. Immunol. 45:497-506.
23. Scott, D. A., and J. B. Kaper. 1994. Cloning and sequencing of the genes encoding Escherichia coli cytolethal distending toxin. Infect. Immun. 62:244-251.
24. Shenker, B. J., D. Besack, T. McKay, L. Pankoski, A. Zekavat, and D. R. Demuth. 2004. Actinobacillus actinomycetemcomitans cytolethal distending toxin (Cdt): evidence that the holotoxin is composed of three subunits: CdtA, CdtB, and CdtC. J. Immunol. 172:410-417.
25. Sugai, M., T. Kawamoto, S. Y. Peres, Y. Ueno, H. Komatsuzawa, T. Fujiwara, H. Kurihara, H. Suginaka, and E. Oswald. 1998. The cell cycle-specific growth-inhibitory factor produced by Actinobacillus actinomycetemcomitans is a cytolethal distending toxin. Infect. Immun. 66:5008-5019.
26. Warner, S., C. A. Hartley, R. A. Stevenson, N. Ficorilli, A. Varrasso, M. J. Studdert, and B. S. Crabb. 2001. Evidence that Equine rhinitis A virus VP1 is a target of neutralizing antibodies and participates directly in receptor binding. J. Virol. 75:9274-9281.
27. Yang, S. X., H. G. Pollock, and A. B. Rawitch. 1996. Glycosylation in human thyroglobulin: location of the N-linked oligosaccharide units and comparison with bovine thyroglobulin. Arch. Biochem. Biophys. 327:61-70.
28. Young, V. B., K. A. Knox, and D. B. Schauer. 2000. Cytolethal distending toxin sequence and activity in the enterohepatic pathogen Helicobacter hepaticus. Infect. Immun. 68:184-191.(Leslie A. McSweeney and L)