|Home | About | Journals | Submit | Contact Us | Français|
The capability of a recombinant mucin-like fusion protein, P-selectin glycoprotein ligand-1/mouse IgG2b (PSGL-1/mIgG2b), carrying Galα1,3Galβ1,4GlcNAc determinants to bind and inhibit Clostridium difficile toxin A (TcdA) was investigated. The fusion protein, produced by a glyco-engineered stable CHO-K1 cell line and designated C-PGC2, was purified by affinity and gel filtration chromatography from large-scale cultures. Liquid chromatography-mass spectrometry was used to characterize O-glycans released by reductive β-elimination, and new diagnostic ions to distinguish Galα1,3Gal- from Galα1,4Gal-terminated O-glycans were identified. The C-PGC2 cell line, which was 20-fold more sensitive to TcdA than the wild-type CHO-K1, is proposed as a novel cell-based model for TcdA cytotoxicity and neutralization assays. The C-PGC2-produced fusion protein could competitively inhibit TcdA binding to rabbit erythrocytes, making it a high-efficiency inhibitor of the hemagglutination property of TcdA. The fusion protein also exhibited a moderate capability for neutralization of TcdA cytotoxicity in both C-PGC2 and CHO-K1 cells, the former with and the latter without cell surface Galα1,3Galβ1,4GlcNAc sequences. Future studies in animal models of C. difficile infection will reveal its TcdA-inhibitory effect and therapeutic potential in C. difficile-associated diseases.
Clostridium difficile is an opportunistic Gram-positive pathogenic bacterium responsible for antibiotic-associated diarrhea and other gastrointestinal diseases. C. difficile infections, collectively known as C. difficile-associated disease (CDAD), range from mild cases of diarrhea to fatal pseudomembranous colitis (1, 2). Various treatment options for CDAD include antibiotics, fecal transplantation therapy, and, potentially, toxin-specific antibodies, vaccines, and replenishment of the patient microflora with oral probiotic therapy (3, 4). However, the emergence of strains with reduced susceptibility to antibiotics such as metronidazole and vancomycin and high rates of recurrent infection have limited the treatment options, necessitating more effective therapeutics that target the pathogenic mechanism of C. difficile (5 – 7).
Toxin A (TcdA) and toxin B (TcdB), with molecular masses of 308 kDa and 270 kDa, respectively, are the two primary virulence factors secreted by C. difficile, and can inactivate the Rho/Ras superfamily of GTPases by glucosylation (8 – 10). As Rho GTPases regulate a number of vital cellular processes, their functional inactivation results in the inhibition of cell migration, morphogenesis, division, and membrane trafficking. Glucosylation of Rho family GTPases causes breakdown of the actin cytoskeleton and activation of caspase-3, leading to apoptosis of intoxicated cells. Diarrhea and inflammation ensue, and ultimately, the control of intestinal epithelial barrier function is lost (11 – 15). Much of the pathology seen during a C. difficile infection is believed to be due to toxin effects. TcdA exhibits many pathobiological functions, such as cell rounding (cytopathic effect) and cell death (cytotoxicity) in a wide range of cell types, fluid accumulation in rabbit intestinal loops, intestinal fluid secretion with hemorrhage and necrosis of intestinal epithelia (enterotoxicity), agglutination of rabbit erythrocytes (hemagglutination), and lethality in many C. difficile infection models (16).
The functional domains of TcdA include an N-terminal glucosyltransferase domain, a cysteine protease domain, a hydrophobic pore-forming domain, and a receptor binding domain (RBD). As an initial step in pathogenesis, the toxin binds to the carbohydrates on colonic epithelial cells through the RBD that carry combined repetitive oligopeptides (CROPs) (17). The TcdA CROP domain is comprised of 32 short repeats (SRs) and 7 interspersed long repeats (LRs) that form the carbohydrate binding motifs (9, 11). Various glycoconjugates containing the terminal sequence Galα1,3Galβ1,4GlcNAc have been shown to bind specifically to TcdA (18, 19). Even though this trisaccharide determinant is not the native human ligand of TcdA, it has been reported that human I, Lewis X, and Lewis Y antigens as well as a human glycosphingolipid that carries a common type 2 chain (Galβ1,4GlcNAc) can bind to TcdA (20, 21). The cocrystal structure of a larger fragment (f2) in the TcdA CROPs complexed with a Galα1,3Galβ1,4GlcNAc derivative has revealed the carbohydrate receptor binding junctions that are formed between the LRs and SRs of TcdA (22). The key residues of the toxin that interacts with the Galα1,3Galβ1,4GlcNAc trisaccharide are conserved, and the possibility of seven putative glycan binding sites in TcdA suggests a mode of multivalent binding that can be exploited for the design of novel carbohydrate-based therapeutics (23).
In this study, we targeted the CROP region of TcdA with mucin-based receptor mimetics that can block the binding of TcdA to its native cell surface receptors. Previously, we have generated expression vectors encoding a recombinant mucin-type fusion protein by genetically fusing the extracellular portion of a mucin-like protein, P-selectin glycoprotein ligand-1 (PSGL-1), to the Fc portion of a mouse IgG2b (PSGL-1/mIgG2b), so that it is secreted as a dimer and can be easily purified with protein A/G. PSGL-1/mIgG2b carries 106 potential O-glycosylation and 6 N-glycosylation, sites making it a suitable scaffold for the multivalent display of a diverse repertoire of bioactive carbohydrate determinants (24, 25). Here, we have shown that C. difficile TcdA binds to PSGL-1/mIgG2b carrying the Galα1,3Gal determinant, a mucin-type fusion protein produced by coexpressing the porcine α1,3-galactosyltransferase (α3GalT) and the core 2 β1,6-N-acetylglucosaminyltransferase 1 (C2 β6GnT1) in CHO-K1 cells (26). The ability of this fusion protein to neutralize TcdA-mediated hemagglutination of rabbit erythrocytes and the cytopathic and cytotoxic properties of TcdA was assessed. We also report C-PGC2, a glyco-engineered CHO-K1 cell line with high TcdA sensitivity, as a novel reference cell line for in vitro TcdA cytotoxicity and neutralization studies.
C-PGC2 was generated by stably transfecting CHO-K1 cells (ATCC, Manassas, VA, USA) with expression plasmids encoding PSGL-1/mIgG2b, the core 2 β1,6-N-acetylglucosaminyltransferase 1 (C2 β6GnT1), and the porcine α1,3-galactosyltransferase (26, 27), the C-P55 cell line was established by stable transfection of CHO-K1 cells with the PSGL-1/mIgG2b plasmid (28), and C-PP1 was generated by stable transfection of CHO-K1 cells with cDNA encoding PSGL-1/mIgG2b, C2 β6GnT1and, the pigeon α1,4-galactosyltransferase (29).
C-PGC2 cells were cultured in Dulbecco's modified Eagle's medium (DMEM) (Invitrogen, Carlsbad, CA, USA) supplemented with 10% fetal bovine serum (FBS) (Invitrogen), 2 mM l-glutamine (Invitrogen), 25 μg/ml gentamicin (Invitrogen) together with the selection medium containing 2 μg/ml puromycin, 400 μg/ml G418, and 200 μg/ml hygromycin. Cells were maintained in a humidified incubator at 37°C and 5.0% CO2. For large-scale purification, C-PGC2 cells were adapted to serum-free conditions and was cultured in ProCHO-4 (Lonza Group Ltd., Basel, Switzerland) medium supplemented with 2 mM l-glutamine, 100 μg/ml dextran sulfate (Sigma-Aldrich, St. Louis, MO), and 25 μg/ml gentamicin together with the selection medium containing half the concentrations of the above-listed selection drugs. Cells were maintained as single-cell suspension cultures in 2-liter shaker flasks (Corning Inc., NY, USA) at 100 rpm, 37°C, and 5.0% CO2.
The frequency of PSGL-1/mIgG2b and αGal expression in stably transfected cells was analyzed by immunocytochemistry staining with fluorescein isothiocyanate (FITC)-conjugated antibodies. Cells were seeded in 6-well BD Biocoat plates (BD Biosciences) for 24 to 48 h of culture and then fixed in 30% (vol/vol) ice-cold acetone in methanol for 2 min at room temperature. Fixed cells were washed in phosphate-buffered saline (PBS) and blocked in 1% bovine serum albumin (BSA) in PBS for 30 min. PSGL-1/mIgG2b expression was detected by staining the cells with an FITC-conjugated goat anti-mouse IgG(Fc) antibody (Sigma-Aldrich) diluted 1:100. For αGal expression, the cells were incubated with human anti-Galα1,3Gal antibodies (840 μg/ml; purified from 50 ml human plasma [Sigma-Aldrich]) diluted 1:100 followed by a secondary FITC-conjugated goat anti-human IgG antibody (Sigma-Aldrich) diluted 1:100). All antibodies were diluted in 1% BSA in PBS, and the incubations were performed for 1 h at room temperature in the dark. The cells were washed in PBS after each antibody incubation. As a counterstain, the 4′,6-diamidino-2-phenylindole (DAPI) (Invitrogen) nucleic acid stain was used at a concentration of 300 nM diluted in PBS. Cells were incubated for 1 to 2 min and were analyzed in a fluorescence microscope (Olympus IX2; Olympus, Germany) after rinsing in PBS.
C-PGC2 cells were cultured in serum-free ProCHO-4 medium (Lonza) in 3-liter shaker flasks (Corning) at 100 rpm and 37°C with 5% CO2. The culture was harvested when the final cell density had reached 3 × 106 to 4 × 106 cells/ml and the viability had dropped to 70%. The cell culture supernatant containing PSGL-1/mIgG2bwith Galα1,3Gal determinants was cleared by centrifugation and further clarified by filtration.
All chromatographic procedures were carried out on an Äkta Explorer 100 controlled by the Unicorn software (v. 5.11) (GE Healthcare). The clarified supernatants were sterile filtered with a 0.22-μm polyether sulfone filter (Nalgene) before loading onto a MabSelect SuRe column (GE Healthcare) preequilibrated with PBS. The column was washed with 10 column volumes (CV) of PBS, and elution of recombinant fusion protein was achieved using 5 CV of 0.1 M sodium citrate, pH 3.0. After elution, selected fractions were pooled, neutralized with 300 μl per ml of 1 M Tris-HCl (pH 9.0), and then dialyzed extensively (12- to 14-kDa cutoff) against MilliQ water at 4°C. After dialysis, the samples were frozen, lyophilized, and stored at −80°C before further purification. Lyophilized samples were dissolved to approximately 5 mg/ml in gel filtration buffer (0.1 M sodium phosphate [pH 7.2] and 0.5 M sodium chloride). Gel filtration of the PSGL-1/mIgG2b was carried out on a preequilibrated HiPrep 26/60 Sephacryl S-300 HR column (GE Healthcare). Typically, 5 ml of sample was applied to the gel filtration column and eluted with a flow rate of 1 ml/min. Eluted fractions were kept at 4°C until pooling were done on the basis of Western blot analysis. Pooled fractions were then dialyzed as described above, frozen, lyophilized, and stored at −80°C. The concentrations of recombinant fusion protein in purified fractions were determined by a two-antibody sandwich enzyme-linked immunosorbent assay (ELISA) method as described previously (30).
Purified PSGL-1/mIgG2b produced in C-PGC2 cells was analyzed by SDS-PAGE under nonreducing conditions using 3 to 8% Tris-acetate gels and Tris-acetate SDS running buffer (Invitrogen). A precision protein standard (Hi-Mark; Invitrogen) was applied as a reference for protein molecular weight determination. PSGL-1/mIgG2b produced in C-P55 was used as a positive control for the IgG(Fc) and PSGL-1 staining, and Galα1,3Gal-HSA (human serum albumin) (Dextra Laboratories Ltd., Reading, United Kingdom) was the positive control for the Galα1,3Gal staining. To remove N-linked glycans, purified PSGL-1/mIgG2b was subjected to N-glycosidase F digestion (peptide-N-glycosidase F [PNGase F]; New England BioLabs, Beverly, MA, USA) under reducing conditions according to the manufacturer's instructions. After 1 h of digestion at 37°C, the samples were analyzed by SDS-PAGE and Western blotting. Separated proteins were electrophoretically blotted onto nitrocellulose membranes (Invitrogen) using iBlot (Invitrogen). The membranes were blocked with 3% BSA in PBS with 0.2% Tween 20 (PBS-T), which was also used for the dilution of antibodies. Western blot membranes were incubated at room temperature for 1 h with a peroxidase-conjugated anti-mouse IgG (Fc specific; Sigma-Aldrich) diluted 1:20,000, a mouse anti-PSGL-1 antibody (clone KPL-1; BD Pharmingen, San Diego, CA, USA) diluted 1:1,000, and a human polyclonal anti-Galα1,3Gal antibody purified in the lab from human AB serum (Sigma-Aldrich) and diluted 1:500. Secondary antibodies for detecting PSGL-1 and Galα1,3Gal determinants were peroxidase-conjugated goat anti-mouse IgG F(ab′)2 (Sigma-Aldrich) diluted 1:10,000 and goat anti-human IgM-horseradish peroxidase (HRP) (Sigma-Aldrich) diluted 1:10,000, respectively. Bound antibodies were visualized by chemiluminescence using the ECL kit according to the manufacturer's instructions (GE Healthcare).
For O-glycan analysis, PSGL-1/mIgG2b purified from C-PGC2 and C-PP1 by protein A affinity chromatography and gel filtration was subjected to reductive β-elimination in solution. The O-glycans were released from the purified proteins in a solution of 1.0 M NaBH4 and 100 mM NaOH for 16 h at 50°C Reactions were quenched with 1 μl of glacial acetic acid, and the sample was desalted and dried as previously described (31). Released O-glycans were treated with green coffee bean α-galactosidase (2 mU; Prozyme, Hayward, Canada) at 37°C overnight in 100 mM sodium citrate-phosphate, pH 6.0.
O-glycans were then analyzed by LC-MS using a 10-cm by 150-μm (inner diameter) column prepared in-house and containing 5 μm porous graphitized carbon (PGC) particles (Thermo Scientific, Waltham, MA). Glycans were eluted using a linear gradient from 0 to 40% acetonitrile in 10 mM ammonium bicarbonate over 40 min at a flow rate of 10 μl/min. The eluted O-glycans were detected using an LTQ ion trap mass spectrometer (Thermo Scientific) in negative-ion mode with an electrospray voltage of 3.5 kV, a capillary voltage of −33.0 V, and a capillary temperature of 300°C. Air was used as a sheath gas, and mass ranges were defined depending on the specific structure to be analyzed. The data were processed using the Xcalibur software (version 2.0.7; Thermo Scientific). Glycans were identified from their tandem MS (MS/MS) spectra by manual annotation. Specified ions were isolated for MSn fragmentation by collision-induced dissociation (CID) with the collision energy set to 30%. The annotated structures have been submitted to the UniCarb-DB database (http://unicarb-db.biomedicine.gu.se/) and will be included in the next release.
C. difficile TcdA (List Biologicals, Campbell, CA, USA) binding to PSGL-1/mIgG2b was assessed by incubating the nitrocellulose membrane with 1 μg/ml TcdA followed by anti-C. difficile toxin A chicken IgY (List Biologicals) diluted 1:1,000 and donkey anti-chicken IgY-HRP (Sigma-Aldrich) diluted 1:10,000. Visualization was performed as described in “SDS-PAGE and Western blotting” above. Bovine thyroglobulin (Sigma-Aldrich) was used as a positive control, and a fusion protein produced in C-P55 was used as a negative control.
TcdA (120 nM) was 2-fold serially diluted in PBS in V-bottom 96-well plates (Corning Inc., Corning, NY, USA) and gently mixed 1:1 with 2% (vol/vol) rabbit red blood cells (RBCs) (Novakemi AB, Handen, Sweden) in PBS. Plates were incubated at 4°C, and the erythrocytes were allowed to settle before visually scoring them for hemagglutination; RBC aggregates diffusely distributed in the bottom of the well were scored as hemagglutination, and wells where the RBCs had settled in the center of the well, giving a button-like appearance, were scored as no hemagglutination. The standard TcdA concentration used for the subsequent hemagglutination inhibition assay was 60 nM. For the inhibition assay, C-PGC2-produced fusion protein was serially diluted in the V-bottom 96-well plates, and an equal volume of TcdA diluted in PBS was added to each well so that a 60 nM TcdA concentration was attained in all the assay wells. Without any preincubation, a 2% suspension of rabbit RBCs was added to each well; the contents of the plates were mixed and were scored for hemagglutination after the erythrocytes were allowed to settle at 4°C.
The cytopathic effect of TcdA on CHO-K1, C-PGC2, and C-PP1 cells was evaluated in the cell rounding assay by observing the morphological changes of cells using phase-contrast microscopy. Cells were seeded in 96-well, flat-bottom microtiter plates at a density of 4 × 105 cells/ml for 24 h, allowing the formation of confluent layers. TcdA was diluted to four different concentrations (6.6, 3.3, 1.6, and 0.3 nM) in cell culture medium with 2% FBS and was added to the assay wells. Each concentration was assessed in triplicate, and the assay was performed twice. Cells were incubated at 37°C and were observed every 30 min for 3 h. The cytopathic effect was estimated by calculating the number of rounded cells as a percentage of the total cell number at the indicated time intervals. Results are expressed as the mean ± standard error of the mean (SEM) for 2 random fields counted in three separate wells.
CHO-K1 cells were seeded in 12-well BD Biocoat cell culture plates (BD Biosciences) with coverslips and were grown to subconfluence. Various concentrations of Galα1,3Gal-substituted PSGL-1/mIgG2b were preincubated with 1.6 nM TcdA (a cytopathic dose that induced 100% cell rounding in 18 h) in the cell culture medium with 2% FBS for 1 h and were added to the respective wells. Toxin-induced cell rounding was monitored after 18 h, and the F-actin of the CHO-K1 cells was visualized by staining the treated cells with phalloidin conjugated with Alexa Fluor 488 (Invitrogen). For actin staining, cells were washed with PBS and fixed with warm 3.7% formaldehyde solution in PBS for 5 min. The cells were permeabilized with 0.1% Triton X-100 for 3 min after washing twice with PBS. The cells were then blocked with 1% BSA in PBS for 30 min after washing three times in PBS. The Alexa Fluor 488 phalloidin staining solution was prepared according to the manufacturer's instructions, and the cells were stained for 20 min at room temperature. The cells were further washed three times with PBS, and as a counterstain, DAPI (4′,6′-diamidino-2-phenylindole) (Invitrogen) nucleic acid stain was used at a concentration of 300 nM in PBS. The cells were mounted using ProLong Gold antifade mounting reagent (Invitrogen), and the immunofluorescence staining was analyzed using a Zeiss LSM 700 instrument on the inverted Axio Observer.Z1 (Carl Zeiss AG, Oberkochen, Germany). The images were edited in Image J software.
CHO-K1 and C-PGC2 cells were seeded in sterile 96-well microtiter plates at a density of 4 × 105 cells/ml for 20 h, allowing for the formation of confluent monolayers. Initially, a dose-response experiment was conducted to find the minimum concentration of TcdA which induced 100% cell death at 72 h after toxin addition. For that, a 3-(4,5-dimethyl-2-thiazolyl)-2,5-diphenyl-2H-tetrazolium bromide (MTT)-based colorimetric cytotoxicity assay was performed using a cell proliferation assay kit (Roche Applied Science, Penzberg, Germany) according to the manufacturer's instructions. Various concentrations of TcdA, serially diluted in 2-fold dilutions starting from 6.6 nM, were added to the cells. Each concentration was used in triplicate, and the assay was performed twice. Absorbance was measured at 570 nm using a 96-well ELISA plate reader. Cell viability was expressed as the percentage of survival compared to that of control cells incubated in the absence of TcdA. For the subsequent neutralization assay, 6.6 nM TcdA for CHO-K1 cells and 1 nM TcdA for C-PGC2 cells were used. Serially diluted PSGL-1/mIgG2b carrying Galα1,3Gal determinants was premixed with the respective concentrations of TcdA in cell culture medium with 2% FBS, incubated for 1 h, and then added to the cell culture. After 72 h, the cytotoxicity assay was performed using the cell proliferation assay kit. The percentage of cell survival in the neutralization assay was generated after normalizing the data to the control wells which were devoid of TcdA. Each data point was determined in triplicate.
The previously reported CHO cells designated C-PGC2 and expressing PSGL-1/mIgG2b substituted with Galα1,3Gal-terminated O-glycans were reanalyzed by immunocytochemistry to confirm the expression of PSGL-1/mIgG2b and the Galα1,3Gal carbohydrate determinant (Fig. 1A and andC).C). All DAPI-stained cells were also positive for staining with FITC-conjugated goat anti-mouse IgG(Fc) antibodies (Fig. 1A) and human polyclonal anti-Galα1,3Gal antibodies purified from human AB serum (Fig. 1C). False-positive staining was ruled out by staining nontransfected CHO-K1 cells as negative controls (Fig. 1B and andDD).
SDS-PAGE and Western blot analyses of PSGL-1/mIgG2b purified from C-PGC2 cells by protein A affinity chromatography and gel filtration revealed a protein of 250 to 350 kDa under nonreducing conditions (Fig. 2A to toC).C). Both anti-mIgG (mouse IgG Fc specific) (Fig. 2A) and anti-CD162 (which recognizes PSGL-1) (Fig. 2B) stained the purified PSGL-1/mIgG2b produced in C-PGC2 cells (lanes 1) and C-P55 cells (CHO-K1 cells transfected with cDNA encoding PSGL-1/mIgG2b only) (lane 2). The presence of core 2 O-glycans on the fusion protein following expression of C2 β6GnT1 was verified by an increase in size of the PSGL-1/mIgG2b produced in C-PGC2 cells (Fig. 2A and andB,B, lanes 1) compared to C-P55 cells (Fig. 2A and andB,B, lanes 2). The presence of Galα1,3Gal determinants on the fusion protein was verified using anti-Galα1,3Gal antibodies. This antibody, which recognizes the terminal Galα1,3Galβ1,4GlcNAc carbohydrate structure, strongly stained the C-PGC2-produced PSGL-1/mIgG2b (Fig. 2C, lane 1), as well as the positive control Galα1,3Gal-HSA (Fig. 2C, lane 3), but not the C-P55-produced PSGL-1/mIgG2b (Fig. 2C, lane 2). That the Galα1,3Gal determinants were carried on O-linked glycans on PSGL-1/mIgG2b was confirmed by digesting the fusion protein with N-glycosidase F (PNGase F), an enzyme known to cleave N-linked glycans. SDS-PAGE run under reducing conditions and followed by Western blotting analyses with anti-Galα1,3Gal antibodies verified a mobility shift of PSGL-1/mIgG2b caused by the N-glycan cleavage (Fig. 2D, lanes − and +). There was no significant reduction in the anti-Galα1,3Gal antibody staining intensity, suggesting that the majority of this carbohydrate determinant is carried by O-linked glycans.
The O-glycans of recombinant PSGL-1/mIgG2b produced in C-PGC2 cells were analyzed by LC-MS/MS in negative-ion mode (Fig. 3 and Table 1). In the base peak chromatogram, one major peak with the composition NeuAc1Hex3HexNAc2 (m/z 1,202, 31% of total ion intensity) and one minor peak with the composition Hex3HexNAc2 (m/z 911, 0.4% of total ion intensity) were identified that could contain a terminal Hex-Hex sequence (Fig. 3A and Table 1). The other major base peaks were annotated as core 1 or core 2 O-glycans based on the MS/MS spectra and our previous studies (32). Trace amounts of structures (m/z of 530 [2 structures], 821, 1,081, 1,622, 1,913) not described in CHO cells before and not easily explained by virtue of the enzymes expressed were also detected (Table 1). To confirm the presence of α1,3Gal on PSGL-1/mIgG2b produced in C-PGC2 cells, released O-glycans were treated with green coffee bean α-galactosidase. For comparison, O-glycans released from PSGL-1/mIgG2b produced in C-PP1 cells and carrying O-glycans terminated with a Galα1,4Gal sequence were used (29). As can be seen in Fig. 3B, only the base peaks at m/z 749 (Hex2HexNAc2), 911, 1,040 (NeuAc1Hex2HexNAc2), and 1,202 were selected to be shown. On comparing the base peaks generated in PSGL-1/mIgG2b produced in C-PGC2 and C-PP1 cells, the base peaks containing terminal α1,4Gal have a shorter retention time than those terminated with α1,3Gal. All αGal-containing O-glycans were sensitive to α-galactosidase treatment, indicating the presence of terminal αGal on O-glycans at m/z 911 and 1,202. MS/MS spectra of glycans at m/z 911 (Fig. 3C) and m/z 1,202 (doubly charged ions at m/z 601, [M-2H]2−) (Fig. 3D) indicated two terminal hexose residues: fragmentation ions at m/z 587 and 569 (Y2α/Z2α) for glycans at m/z 911 and fragmentation ions at m/z 878 and 860 (Y2α/Z2α) for glycans at m/z 1,202. However, cross-ring cleavages (A ions) or glycosidic cleavages (B ions) supporting the loss of two terminal hexoses residues were less abundant, making it hard to distinguish between the loss of a Hex-Hex terminal motif and loss of one Hex from each arm of the core 2 O-glycan. In order to resolve this, the MS3 spectra of the tentative B ions at m/z 526 of Galα1,3Galβ1,4GlcNAc (Fig. 3E) and Galα1,4Galβ1,4GlcNAc from the fusion protein produced in C-PP1 cells as a reference (Fig. 3F) were compared. In the MS3 spectra (Fig. 3E and andF),F), the presence of Y1 and Y2 ions at m/z 202 and 364 suggest the presence of a terminal Hex-Hex motif. In comparison with Galα1,4Gal, Galα1,3Gal yielded high-intensity ions at m/z 263 but lacked 0,2A (m/z 281) and 2,4A (m/z 221) fragmentation ions of the penultimate Gal residue. Together with fragmentation ions at m/z 235 (3,5A3/Y2) (Fig. 3E), we annotate ions at m/z 263 as 0,2A3-H2O/Y2. These cross-ring cleavages can be used as diagnostic ions to distinguish Galα1,3Gal from Galα1,4Gal. In summary, over one-third (32.3% of the total) (Table 1) of the O-glycans carry terminal Galα1,3Gal on PSGL-1/mIgG2b produced in the C-PGC2 cell line.
The ability of Galα1,3Gal-substituted PSGL-1/mIgG2b to interact with the TcdA of C. difficile was analyzed using SDS-PAGE and Western blot analysis. One microgram of fusion protein was run on an SDS-polyacrylamide gel and blotted onto a nitrocellulose membrane, which was subsequently probed with TcdA at a concentration of 1 μg/ml (Fig. 4). PSGL-1/mIgG2b produced in the C-PGC2 cell line reacted strongly with TcdA (Fig. 4, lane 1), whereas no binding to the C-P55-produced PSGL-1/mIgG2b (Fig. 4, lane 2), which carries mainly mono- and disialylated core 1 O-glycans (33), was observed. Tetrameric bovine thyroglobulin (2 μg) of 660 to 690 kDa, which has been previously reported to bind TcdA through its Galα1,3Galβ1,4GlcNAc determinant (18), was used as a positive control for this binding assay (Fig. 4, lane 3).
The repeating subunits (CROPs) of the TcdA receptor binding domain interact with Galα1,3Galβ1,4GlcNAc sequences on rabbit erythrocyte membranes, leading to cross-linking and subsequent hemagglutination (18, 19). Upon titration of the toxin concentration, complete hemagglutination was recognized by a diffuse distribution of red blood cell (RBC) aggregates in the bottom of the well (Fig. 5) and was obtained following incubation of the 2% (vol/vol in PBS) rabbit RBC concentrate in ≥30 nM TcdA. In contrast, no agglutination and a button-like appearance of the RBCs (Fig. 5) were seen at TcdA concentrations below 4 nM. For the hemagglutination inhibition assay, 60 nM TcdA was selected as the standard toxin concentration and was mixed with the C-PGC2-produced PSGL-1/mIgG2b serially diluted in 2-fold dilutions. Toxin with or without fusion protein was added together with rabbit erythrocytes to the wells, and toxin A-induced hemagglutination was scored as shown in Fig. 5. PSGL-1/mIgG2b substituted with Galα1,3Gal determinants could partially inhibit hemagglutination at a concentration as low as 15 nM. At 500 nM and above, the Galα1,3Gal-carrying fusion protein blocked hemagglutination completely, and at around 62 nM, a moderate inhibition was observed. This demonstrates that the fusion protein can block the interaction between TcdA and its cellular receptor on rabbit erythrocytes. C-PP1-produced PSGL-1/mIgG2b harboring O-linked Galα1,4Galβ1,4GlcNAc (blood group P1) sequences failed to inhibit TcdA-induced hemagglutination of rabbit erythrocytes even at a very high concentration (4 μM). This suggests that the α1,3 linkage between the two galactoses is important for TcdA binding.
TcdA induces retraction and rounding of all treated mammalian cells, which is referred to as the cytopathic effect of the C. difficile toxin (34). The rounding of cells serves as a measure of toxin potency in cytotoxicity assays. Even though CHO cells lack the typical Galα1,3Gal TcdA receptors, they are widely used as a reference cell line in studies assessing the cytopathic effect of TcdA. The cytopathic effects of TcdA on the glyco-engineered C-PGC2 cells expressing the Galα1,3Gal C. difficile receptor (Fig. 6A), wild-type CHO-K1 cells (Fig. 6B), and the glyco-engineered C-PP1 cells expressing the blood group P1 determinant (Fig. 6C) were compared. The cytopathic effect was measured at four different TcdA concentrations: 0.3, 1.6, 3.3, and 6.6 nM (Fig. 6). It can be clearly observed that C-PGC2 cells were highly sensitive (Fig. 6A), almost 20-fold more sensitive than the wild-type CHO-K1 cells (Fig. 6B). An effect on the C-PGC2 cells was noted within 15 min following exposure to 6.6 nM TcdA. The accumulation of rounded cells following incubation of CHO cells in 6.6 nM TcdA was comparable to that observed for C-PGC2 cells exposed to 0.3 nM TcdA. In order to confirm that this effect was due to the expression of Galα1,3Gal determinants and not to the transfection and cell selection per se, another glyco-engineered CHO-K1 cell line, C-PP1, was used as target cells for TcdA. These cells express the blood group P1 determinant (Galα1,4Galβ1,4GlcNAc) that has an α1,4-linked galactose instead of the α1,3-linked galactose found on the C-PGC2 cells (29). The cytopathic effects of TcdA on C-PP1 cells resembled those seen on wild-type CHO-KI cells (Fig. 6C). This confirms that the increased TcdA sensitivity of C-PGC2 cells is due to the expression of the Galα1,3Gal toxin receptor on their cell surface.
The capability of Galα1,3Gal-substituted PSGL-1/mIgG2b to neutralize the cytopathic effect of the C. difficile toxin was assessed by staining the actin filaments of CHO-K1 cells after exposure to 1.6 nM TcdA preincubated with various concentrations of PSGL-1/mIgG2b. The cytopathic dose of TcdA that induced 100% cell rounding 18 h after toxin addition was found to be 1.6 nM in CHO-K1 cells. The cytopathic effect of the clostridial glucosylating toxins has been attributed to the disappearance of actin stress fibers and disruption of its focal adhesion sites, which finally leads to cell rounding. The actin filaments of native CHO-K1 cells (Fig. 7A and andD),D), CHO-K1cells treated with 1.6 nM TcdA (Fig. 7B and andE),E), and CHO-K1 cells exposed to 1.6 nM TcdA preincubated with 1 μΜ C-PGC2-produced PSGL-1/mIgG2b (Fig. 7C and andF)F) were visualized by confocal microscopy following staining with Alexa Fluor 488-labeled phalloidin. Native CHO-K1 cells exhibited an organized F-actin architecture, while exposure to TcdA at a concentration of 1.6 nM completely disrupted the normal F-actin organization. The nuclei of TcdA-treated cells were polarized and appeared to be fragmented with an irregular shape (Fig. 7B). While exposure to TcdA preincubated with the fusion protein carrying Galα1,3Gal determinants did not alter the intracellular actin architecture, cell elongation was observed (Fig. 7C and andF).F). Cells exposed to TcdA preincubated with 250 nM and higher concentrations of PSGL-1/mIgG2b had reduced percentages of cell rounding. Instead, elongated cells appeared (Fig. 7C and andF).F). This type of morphological change, described as the cytotonic activity of TcdA, has been previously reported in CHO-K1 cells at low or noncytotoxic concentrations of TcdA (35). This implies that it is not possible to completely inhibit the cytopathic effect of TcdA by targeting the CROP region with our Galα1,3Gal-carrying recombinant mucin type fusion protein.
Even though TcdA-induced apoptosis (cytotoxic effect) depends on the glucosylation of Rho GTPases, it is not triggered by the destruction of the actin cytoskeleton (cytopathic effect) (36). Furthermore, the rounding of cells, which serves as an indicator of the TcdA effect, is not immediately followed by cell death. Also, the elongated CHO-K1 cells seen as a consequence of the cytotonic effect of TcdA can be metabolically inactive. Therefore, we assessed the neutralizing activity of the Galα1,3Gal-carrying fusion protein in an MTT-based cytotoxicity assay and compared the effect on both C-PGC2 (expressing Galα1,3Galβ1,4GlcNAc receptors) and CHO-K1 cells. Initially, a dose-response experiment was conducted to find the concentration of TcdA capable of inducing 100% cell death or no cell survival in CHO-K1 and C-PGC2 cells (Fig. 8A). As seen in the cytopathic assay, C-PGC2 cells showed higher TcdA sensitivity than wild-type CHO-K1 cells. The minimum concentration of TcdA which induced 100% cell death at 72 h after toxin addition in the CPGC2 cells was found to be 0.9 nM, whereas it was 6.6 nM for wild-type CHO-K1 cells. When PSGL-1/mIgG2b carrying Galα1,3Gal determinants was preincubated with TcdA at a concentration of 1 nM, we could observe a maximum of 48% cell viability in C-PGC2 cells (Fig. 8B). For CHO-K1 cells, which lack the Galα1,3Gal TcdA receptors, the fusion protein exhibited only a partial neutralization, with 34% cell survival at a fusion protein concentration of 1 mM and a TcdA concentration of 6.6 nM.
Because of the global spread of antibiotic resistance, there is an urgent need for novel antibacterial treatment strategies (37). One such approach to combat bacterial infections is targeted toward the initial adhesion event facilitating attachment of bacteria or their toxins to the cell surface. In this context, the potential of carbohydrates as new and viable bacterial and bacterial toxin antiadhesive therapeutics is widely being explored (38 – 40). The pathogenicity of C. difficile is primarily linked to the secretion of two homologous glucosylating exotoxins, TcdA and TcdB, that play an important role in disease pathogenesis. Even though recent studies suggest that majority of the intestinal damage is caused by TcdB, while TcdA is responsible for more superficial and limited damage, neutralization of both toxins is required to prevent CDAD (41, 42). It is a challenge to produce therapeutics conferring protection against recurrent CDAD because of the complex multifunctional domain structures of C. difficile toxins. Although a few cellular receptors for TcdA and TcdB have been recently identified, the Galα1,3Galβ1,4GlcNAc carbohydrate sequence has been the most studied receptor that binds to the CROP domain of TcdA (17, 43 – 45). However, TcdA also displays its cytotoxic activity on cells that lack Galα1,3Galβ1,4GlcNAc receptors. Less is known about these alternate TcdA binding structures, which makes the Galα1,3Galβ1,4GlcNAc sequence the best candidate for designing multivalent carbohydrate-based inhibitors. In this study, we describe the identification and characterization of a recombinant mucin-type fusion protein, PSGL-1/mIgG2b, that can bind to TcdA and inhibit its pathobiological functions. The mucin part of PSGL-1/mIgG2b provides the scaffold for multivalent display of core 2 O-glycans carrying Galα1,3Galβ1,4GlcNAc carbohydrate determinants and thus contributing to its overall binding strength (29, 30, 33).
The stable CHO cell line C-PGC2, which expresses Galα1,3Galβ1,4GlcNAc sequences, was shown to be sensitive to the cytotoxic effects of TcdA even at very low concentrations. CHO cells, which are widely used as a cell-based model for cytotoxicity and neutralization assays, lack Galα1,3Galβ1,4GlcNAc sequences and thus the typical TcdA receptor (46). Compared to wild-type CHO-K1 cells, the glyco-engineered C-PGC2 cells were 20-fold more sensitive to TcdA. The cytopathic effect of TcdA was visible after 15 min of TcdA incubation, and even at picomolar TcdA concentrations, cell rounding was observed. The results of the MTT-based cytotoxicity assay also correlated with the results of the cytopathic assay, as 100% cell death was observed in C-PGC2 cells with concentrations as low as 1 nM, whereas at least 10 nM TcdA was required to see an equivalent effect on the CHO-K1 parental cell line. The sensitivity of cells to TcdA has previously been mapped to the CROP domain (47 – 49). Therefore, the observed difference in TcdA sensitivity between C-PGC2 and CHO-K1 cells may be explained by the greater density in C-PGC2 cells of cell surface receptors binding the CROP region of TcdA. Furthermore, the crucial step for pathogenicity of the toxins is the translocation of the catalytic domain into the cytosol of target cells, a process for which TcdA CROPs are necessary (48, 50).
Although the CROP region serves as the receptor binding domain, TcdA can display its cytopathic action in the absence of these repeats. It has been shown that removing the CROPs from TcdA only reduces the cytopathic potency, suggesting that an additional binding activity may be encoded in the region preceding the C-terminal repeats (48, 51). While removal of the CROPs has reduced the potency of toxin in many cells, CHO cells exhibited identical susceptibility for both the truncated and full-length TcdA. Furthermore, binding studies with TcdA CROPs alone showed no binding to CHO cells (48). This indicates that the Galα1,3Galβ1,4GlcNAc determinants could be the main receptor for TcdA CROPs and that the intermediate and transmembrane domains, which have additional binding activity, interact with alternate cell surface structures. Even though the role of CROPs in the cytotoxic effect of the toxin is not clear, hemagglutination and enterotoxicity mediated by TcdA seem to be solely dependent on the interaction of CROPs with their receptor (16). We showed that Galα1,3Galβ1,4GlcNAc-carrying PSGL-1/mIgG2b could completely neutralize the hemagglutination of rabbit erythrocytes mediated by TcdA, supporting this hypothesis. In an early study by Clark and coworkers, it was shown that the GlcNAc residue present in the terminal Galα1,3Galβ1,4GlcNAc sequence is required for carbohydrate-specific recognition by TcdA, whereas we show that PSGL-1/mIgG2b carrying the Galα1,4Galβ1,4GlcNAc sequence failed to inhibit the hemagglutination reaction (19). In the cytopathic and cytotoxicity neutralization assays, PSGL-1/mIgG2b carrying Galα1,3Galβ1,4GlcNAc could not completely neutralize the cytotoxic effects of TcdA. This lack of complete neutralization of TcdA can be due to the involvement of the intermediate regions of TcdA in host cell surface binding and cellular entry. The capability of C-PGC2-produced PSGL-1/mIgG2b to partially neutralize TcdA may be explained by steric interference of the intermediate domain, preventing it from binding to its native ligands present on the CHO-K1 cell surface. It is also possible that the binding of CROPs to Galα1,3Galβ1,4GlcNAc-terminated O-glycans alters the conformation of TcdA intermediate domains, preventing ligand binding (48, 50). In case of C-PGC2 cells, the neutralization effect of PSGL-1/mIgG2b was higher, as it could completely block the TcdA internalization dependent on Galα1,3Galβ1,4GlcNAc receptors. Blocking of the CROP domain with the mucin-type fusion protein, multivalently substituted with Galα1,3Gal carbohydrate determinants, reduced the potency of TcdA in C-PGC2 cells. Further, premixing TcdA with the Galα1,3Galβ1,4GlcNAc-substituted fusion protein before addition to the CHO-K1 cells abolished its cytopathic effect. Instead a cytotonic effect was detected, most likely due to the decreased TcdA potency (35). The cytotonic effect on CHO-K1 cells, reflected in their elongation, was not observed on the glyco-engineered C-PGC2 cells. This may be due to the fact that even very low TcdA concentrations can cause the rounding of C-PGC2 cells, perhaps explained by a much faster internalization process of TcdA in cells expressing the surface receptor for CROPs. However, many studies have clearly separated the cytopathic effect of TcdA in cell cultures in vitro from the intestinal effects seen with TcdA in vivo (47, 52 – 54). The monoclonal antibody PCG-4, which binds to the same pocket on TcdA as Galα1,3Galβ1,4GlcNAc, did not inhibit the cytopathic effect of TcdA on CHO-K1 cells. However, it could neutralize the enterotoxic activity of TcdA (52). Further, it completely inhibited the fluid accumulation induced by TcdA in rabbit small intestinal segments and blocked the intestinal tissue damage (54). This suggests that our Galα1,3Galβ1,4GlcNAc-substituted mucin-type fusion protein could also inhibit the enterotoxicity of TcdA. However, further studies are necessary to assess the ability of this mucin to neutralize the enterotoxic activity of C. difficile TcdA in vivo.
We have created a platform in which recombinant mucin-type immunoglobulin fusion proteins are used as scaffolds for multivalent expression of O-glycans with diagnostic or therapeutic potential (28, 29, 33, 55). Because the glycosyltransferases involved in the biosynthesis of many O-glycan core structures and determinants are known, host cells can be glyco-engineered to express multiple copies of a desired carbohydrate determinant on the mucin scaffold (24, 25, 32). The C-PGC2-produced fusion protein was proven to be an efficient adsorber of anti-pig antibodies due to its multivalent expression of Galα1,3Gal determinants (26, 27). In this study, we have provided evidence for the usefulness and efficacy of this PSGL-1/mIgG2b in blocking the CROP region of TcdA. We believe that further investigations can provide a novel therapeutic approach for preventing the toxic effects seen on the large intestine during a C. difficile infection, perhaps by providing an enema containing high concentrations of the mucin carrying the receptor for C. difficile TcdA.
We thank the staff at the Centre for Cellular Imaging, Sahlgrenska Academy, for their technical assistance and valuable suggestions.
This work was supported by the Swedish Research Council (K2011-65X-3031-01-6 to J.H. and 621-2010-5322 to N.G.K.) and the County Council of Västra Götaland (ALF) to J.H. The mass spectrometer was obtained by a grant from the Swedish Research Council (342-2004-4434).