Immobilization of Reducing Sugars as Toxin Binding Agents

Bioconjugate Chem. 1997, 8, 466−471

466

Immobilization of Reducing Sugars as Toxin Binding Agents U. J. Nilsson,† L. D. Heerze,‡ Y.-C. Liu,† G. D. Armstrong,‡ M. M. Palcic,† and O. Hindsgaul*,† Departments of Chemistry and Medical Microbiology and Immunology, University of Alberta, Edmonton, Alberta T6G 2G2, Canada. Received December 12, 1996X

A simple and economical procedure for the attachment of reducing sugars to aminated solid supports has been developed. Reaction of the amino groups on the solid support with p-nitrophenyl chloroformate, followed by 1,6-hexanediamine, yields a chain-extended amine to which reducing sugars can be attached while remaining accessible to macromolecules. Immobilization of the reducing sugars involves a simple incubation followed by trapping of the resulting glycosylamine with acetic anhydride and recovery of the unreacted sugar by filtration. This technique was used to immobilize lactose and sialyllactose onto silylaminated Chromosorb P, producing solid supports that effectively neutralized the activity of cholera toxin from Vibrio cholerae and heat-labile enterotoxin of enterotoxigenic Escherichia coli. The general applicability of such solid supports for toxin neutralization was further demonstrated by immobilization of the enzymatically synthesized RGal(1-3)βGal(1-4)Glc trisaccharide, which produced a support that efficiently neutralized toxin A of Clostridium difficile. The results from this study suggest that these solid supports have the potential to serve as inexpensive therapeutics for bacterial toxin-mediated diarrheal diseases.

INTRODUCTION

The recognition of oligosaccharide receptors on host cells by pathogenic microorganisms or their toxins is a crucial event in causing disease in humans (1). One potential therapeutic approach to prevent disease in humans is to inhibit the attachment of the pathogens or their toxins to carbohydrate receptors on host cells using oligosaccharide receptor analogs that have the capability of binding bacteria or toxin. This approach has been utilized in two gastroenteric applications in which synthetic oligosaccharide sequences were immobilized onto a nondegradable diatomaceous earth, silylaminated Chromosorb P (2). These glycosylated solid supports possess the ability to bind shiga-like toxin produced by enterohemorrhagic Escherichia coli (the causative agent of hemorrhagic colitis and hemolytic-uremic syndrome), as well as toxin A from Clostridium difficile, which plays a major role in causing antibiotic-associated diarrhea (35). While these affinity supports were efficient at toxin neutralization, their preparation involves multistep chemical synthesis of 8-(methoxycarbonyl)octyl glycosides followed by their immobilization using coupling procedures that are very labor intensive. If potential therapeutics for third-world diseases such as cholera are being considered, the solid supports become prohibitively expensive. This paper describes a simple and economical procedure for immobilization of commercially available reducing oligosaccharides onto an inert matrix to produce solid supports (termed SYNSORB’s), for use as toxin binding agents. Commercial Chromosorb P was conventionally silylaminated using 3-(triethoxysilyl)propylamine (2). Reducing oligosaccharides were then coupled to the silylaminated Chromosorb P via a chain-extended glycosylamide linkage using inexpensive reagents and simple chemistry. The resulting solid supports were then * Author to whom correspondence should be addressed [telephone (403) 492-4171; fax (403) 492-7705; e-mail [email protected]]. † Department of Chemistry. ‡ Department of Medical Microbiology and Immunology. X Abstract published in Advance ACS Abstracts, June 1, 1997.

S1043-1802(97)00060-8 CCC: $14.00

screened for their ability to bind cholera toxin (CT) from Vibrio cholerae, heat-labile enterotoxin (LT) from enterotoxigenic Escherichia coli, a common cause of bacteriainduced traveler’s diarrhea, and toxin A from C. difficile. The oligosaccharides used in this investigation represent components of the pentasaccharide ganglioside GM1 structure, the primary receptor for both CT and LT (6), and the trisaccharide recognized by toxin A of C. difficile, RGal(1-3)βGal(1-4)Glc (5). The RGal(1-3)βGal(1-4)Glc trisaccharide was obtained by enzymatic glycosylation of lactose using calf thymus R(1-3)-galactosyltransferase. EXPERIMENTAL PROCEDURES

Materials. Calf thymus was obtained from Pel-Freeze Biologicals, UDP-Gal and alkaline phosphatase were from the Sigma Chemical Co., and E. coli β-galactosidase was from Boehringer Mannheim. AG 1 X8 and Bio-Gel P-2 were obtained from Bio-Rad. R(1-3)-Galactosyltransferase was isolated from calf thymus glands by extraction and chromatography on a UDP-hexanolamine Sepharose column as described by Blanken and van den Eijnden (7) using sodium cacodylate buffer instead of Tris-maleate buffer. After chromatography, the enzyme was concentrated by ultrafiltration, dialyzed against 30 mM sodium cacodylate buffer, pH 6.5, containing 20 mM MnCl2 and 0.1% Triton X-100, and stored at 4 °C. Galactosyltransferase activity was monitored by incubation with 540 µM βGal(1-4)βGlcNAcO(CH2)8COOCH3, 1 mM UDP-Gal, 35 000 dpm UDP-[3H]-Gal, 1 mg/mL bovine serum albumin, 0.8% Triton X-100, 50 mM MnCl2, and 100 mM sodium cacodylate buffer, pH 6.1, in a total volume of 20 µL. After reaction for 30 min at 37 °C, products were isolated on a reversed phase C18 cartridge as previously described (8). Typical Procedure for Immobilizing Reducing Sugars on Silylaminated Chromosorb P. To silylaminated Chromosorb P (20 g) and p-nitrophenyl chloroformate (15 g, 75 mmol) in dry tetrahydrofuran (80 mL) and dry dichloromethane (80 mL) was added diisopropylethylamine (13.1 mL, 75 mmol). The mixture was shaken occasionally for 3 h, after which time the resulting © 1997 American Chemical Society

Immobilization of Reducing Sugars

Bioconjugate Chem., Vol. 8, No. 4, 1997 467

Scheme 1. Comparison of Traditional Organic Synthesis and Direct Coupling of Reducing Oligosaccharides to an Aminated Solid Support Using Lactose as the Example

a

Reagents: (a) Ac2O, NaOAc; (b) SnCl4, HO(CH2)8COOEt; (c) NaOMe/MeOH; (d) H2NNH2; (e) HONO, H2Nssolid support; (f) H2Nssolid support, 60 °C; (g) Ac2O.

iMeOH,

resin was filtered, washed with dichloromethane/tetrahydrofuran (1:1, 5 × 100 mL), and dried under vacuum. To the resulting dried resin was added 1,6-hexanediamine (8.7 g, 75 mmol) in dry dimethylformamide (200 mL) containing triethylamine (10.5 mL, 75 mmol). The reaction was allowed to proceed for 90 min with occasional shaking. The resin was then removed by filtration, washed successively with water (3 × 300 mL), dimethylformamide (3 × 300 mL), and dichloromethane/tetrahydrofuran (1:1, 5 × 100 mL), and dried under vacuum to give 22 g of resin. A portion of the resin (2.0 g), lactose (27.4 mg, 80 µmol), and acetic acid (40 µL) in dry methanol (6.5 mL) were heated to 60 °C in a sealed flask for 47 h. The mixture was then cooled on ice (∼5 °C) and acetic anhydride (2.1 mL) was added. The mixture was shaken occasionally for 12 h, and the resin was recovered by filtration and then washed with water (3 × 50 mL) and methanol (3 × 50 mL). Fine particles were removed by suspending the resin in methanol and decanting the supernatant until it became clear. Drying the resin under vacuum gave 1.95 g of SYNSORB 260. Analysis of the product according to the phenol-sulfuric acid assay indicated an oligosaccharide incorporation of 1.24 µmol/g of resin (9). Synthesis of rGal(1-3)βGal(1-4)Glc. A reaction mixture containing lactose (50 mg), UDP-Gal (20 mg), R(1-3)-galactosyltransferase (60 milliunits), alkaline phosphatase (20 units), 20 mM MnCl2, and 0.1% Triton X-100 in 50 mM sodium cacodylate buffer (3 mL) at pH 6.5 was incubated at 37 °C. Additional UDP-Gal was added to the mixture after 24 h (20 mg) and 48 h (50 mg). After 120 h, fresh R(1-3)-galactosyltransferase (20 milliunits) and UDP-Gal (10 mg) were added to the mixture, which was incubated for an additional 72 h, at which point very little unreacted lactose remained as estimated by TLC [SiO2, 0.1 M sodium borate/2-propanol 1:4, Rf(lactose) ) 0.60, Rf[RGal(1-3)βGal(1-4)Glc] )

0.51]. The reaction mixture was filtered through a 0.2 µm Nalgene nylon filter, the filtrate was applied to a BioRad AG 1X8 column (Cl- form, 2.5 × 20 cm, 0.6 mL/min), and the column was eluted with water. Saccharide fractions were combined and lyophilized. The dry residue was dissolved in 50 mM potassium phosphate buffer, pH 7.5, β-galactosidase (150 milliunits) was added to the mixture to destroy unreacted lactose, and the sample was left at ambient temperature (24 °C) for 18 h. The mixture was boiled for 2 min, filtered though a 0.2 µm filter, and divided into three portions, each of which was loaded onto a C18 silica gel column (20 g). The columns were washed with water (200 mL), and the aqueous eluents were concentrated to dryness under reduced pressure. The residue was dissolved in water (5 mL) and applied to a Bio-Gel P-2 column (2.5 × 100 cm, H2O, 0.2 mL/min). Fractions that contained the trisaccharide were combined and lyophilized, to give 10.5 mg of RGal(1-3)βGal(1-4)Glc trisaccharide: 1H NMR (500 MHz, D2O) δ 5.22 (d, 0.36 H, J ) 3.6 Hz, H-1R), 5.14 (d, 1 H, J ) 3.0 Hz, H-1′′), 4.66 (d, 0.64 H, J ) 8.0 Hz, H-1β), 4.51 (d, 1 H, J ) 8.0 Hz, H-1′). Assay of Toxin Activity Using Tissue Culture Cells. The cytotonic activity of CT and LT was measured by using Chinese hamster ovary cells (CHO) maintained in Hams F12 medium supplemented with 10% fetal bovine serum (FBS) in an atmosphere of 5% CO2 at 37 °C. Toxin samples were diluted 1:5 in Hams media and filter sterilized through 0.22 µm syringe filters, the sterilized samples were serially 5-fold diluted in media, and 100 µL of each dilution was added to wells with confluent monolayers of CHO cells and incubated for 24 h at 37 °C/5% CO2. Each sample was analyzed two times. Cytotonic effects were readily visible after 24 h of incubation by comparing wells with controls that did not contain toxin. After 24 h, the cells were fixed with 95% methanol and stained with Geimsa stain. Toxin-contain-

468 Bioconjugate Chem., Vol. 8, No. 4, 1997

Nilsson et al.

Table 1. SYNSORB Derivatives Used in Toxin Neutralization Experiments

ing samples from neutralization experiments were treated in an analogous fashion except that the percent neutralization was determined by comparing the endpoint dilutions of samples with and without SYNSORB.

Screening of Immobilized Lactose, Sialyllactose, Maltose, and Cellobiose for the Ability To Neutralize CT and LT Activity. A solution containing purified CT or LT (Sigma, 2 µg in 1 mL of PBS) was added to

Immobilization of Reducing Sugars

Bioconjugate Chem., Vol. 8, No. 4, 1997 469

various SYNSORB derivatives (20 mg, SYNSORBs 16 and 89 were provided by D. Rafter, SYNSORB Biotech. Inc., Calgary, AB, Canada) in 1.5 mL microcentrifuge tubes and incubated at room temperature for 1 h on an end-over-end rotator. After incubation, the SYNSORB was allowed to settle to the bottom of the tubes and the supernatants were carefully removed by aspiration. Serial 5-fold dilutions of the supernatants were prepared and the cytotonic endpoints determined as described above. The extent of reduction in the endpoint in the presence of SYNSORB was determined by comparing with controls in which SYNSORB was not added. C. difficile Toxin A SYNSORB Neutralization Assays. Toxin A was purified from a toxin producing strain of C. difficile (ATCC 43255, VPI strain 10463) as described (10). Solutions containing purified toxin A (1 mL) were added to 20 mg samples of either SYNSORB 364 (prepared by chemical and enzymatic synthesis), 374, 376, or 90 (provided by D. Rafter, SYNSORB Biotech. Inc.) in 1.5 mL microcentrifuge tubes and processed as described for CT and LT. Hemagglutination Assays Using Rabbit Erythrocytes. Fresh rabbit erythrocytes were washed once in phosphate-buffered saline (PBS) and resuspended at a concentration of 2% (v/v) in cold PBS. Serial 2-fold dilutions (50 µL) of toxin A containing solutions were made in cold PBS in U-shaped microtiter wells. An equal volume (50 µL) of rabbit erythrocytes was then added to each well, and the microtiter plate was mixed gently. After the plate had been incubated for 4 h at 4 °C, the hemagglutination titer was assessed visually. All assays were done in duplicate. RESULTS AND DISCUSSION

The broad application of immobilized oligosaccharides as bacterial toxin binding agents is hampered by the requirement of labor-intensive synthesis to produce complex oligosaccharides carrying a functionalized spacer arm that is suitable for coupling to a solid support. Immobilization of a synthetic lactoside typically involves five chemical reaction steps (11), even beginning with the commercially available inexpensive disaccharide (Scheme 1). Blomberg et al. (12) have, however, described an attractive alternative involving the coupling of reducing sugars to immobilize amines via a glycosylamine linkage to yield a stable glycosylamide. To determine the feasibility of the immobilization strategy based on reducing sugars to form materials suitable for use in gastroenteric applications, lactose was immobilized onto silylaminated Chromosorb P. The procedure, adapted from that of Blomberg et al., involves simply incubating the resin with a methanolic solution of lactose at 60 °C for 2 days. Addition of acetic anhydride, followed by filtration, then yielded the corresponding lactosyl acetamide (SYNSORB 343, Table 1). The resulting lactose conjugate was compared with SYNSORB 16, which carries lactose coupled via a chemically synthesized 8-(methoxycarbonyl)octyl linker, for the ability to neutralize CT and LT activity (13). SYNSORB 16 was used as the control, since we have previously demonstrated CT and LT binding to this support (unpublished results). The results from these experiments (Figure 1) indicate that lactose immobilized directly onto silylaminated Chromosorb P (SYNSORB 343) provided poor inhibitors of CT and LT binding as compared to SYNSORB 16. This is probably due to the close proximity of the sugar with the support, making it less accessible for toxin binding. To increase the accessibility of the sugar, an extended spacer carrying a primary amine at

Figure 1. Neutralization of purified heat-labile toxin (A) and cholera toxin (B) cytotonic activity using a panel of SYNSORB derivatives shown in Table 1.

the distal end was therefore introduced onto the silylaminated Chromosorb P as shown in Scheme 2 (14). Silylaminated Chromosorb P (1) was treated with p-nitrophenyl chloroformate to give the Chromosorb P (2) activated as a p-nitrophenyl urethane. This activated Chromosorb P (2) was then treated with 1,6-hexanediamine to give the Chromosorb P (3) carrying a primary amine on an extended spacer. Oligosaccharides were immobilized on this derivatized Chromosorb P (3) by simply stirring the reducing sugar and silylaminated Chromosorb P in methanol containing 0.6% acetic acid at 60 °C and then trapping the glycosyl amine as a glycosyl acetamide (4) with acetic anhydride. The presence of acetic acid significantly increased the incorporation yield of the reducing sugar when compared to the original procedure (12). The amount of reducing sugar in the immobilization step had to exceed 40 µmol of oligosaccharide/g of 3 to give a high incorporation of oligosaccharide. The excess reducing sugar could easily be recovered after removal of the resin by filtration and concentration of the mother liquor. Using this method, several commercially available oligosaccharides (Table 1) that resemble components of the ganglioside GM1 structure were immobilized onto silylaminated Chromosorb P, giving incorporation yields ranging from 0.37 to 2.4 µmol of sugar/g of Chromosorb P.

470 Bioconjugate Chem., Vol. 8, No. 4, 1997

Nilsson et al.

Scheme 2a

a (a) p-Nitrophenyl chloroformate, diisopropylethylamine, THF, CH Cl ; (b) 1,6-hexanediamine, Et N, DMF; (c) MeOH, AcOH, 2 2 3 lactose, 60 °C; (d) Ac2O.

The results from the toxin neutralization experiments (Figure 1) indicated that the extent of neutralization with SYNSORBs 260, 366, 368, and 370 (Table 1) were comparable with control SYNSORBs 16 and 89. The maltose and cellobiose containing SYNSORBs (374 and 376) were not efficient at neutralizing either CT or LT, indicating carbohydrate-specific recognition. The conclusions from the toxin neutralization experiments are that the 8-(methoxycarbonyl)octyl linker arm can be replaced with a much more cost-effective alternative. In addition, simple commercially available oligosaccharides readily isolated from natural sources can serve as receptor analogs that effectively bind CT and LT activity. To further demonstrate the generality of this method, it was applied to toxin A from C. difficile, which is unrelated to LT and CT. Toxin A from C. difficile recognizes the trisaccharide RGal(1-3)βGal(1-4)Glc (5) which, unlike lactose, is not readily available from natural sources. We chose to enzymatically galactosylate lactose in one step using calf thymus R(1-3)-galactosyltransferase to yield the reducing trisaccharide RGal(13)βGal(1-4)Glc on a 10 mg scale. Immobilization of this trisaccharide as described for lactose then gave SYNSORB 364 (Table 1). SYNSORB 364 neutralized the

Figure 2. Neutralization of purified toxin C. difficile A hemagglutination activity using SYNSORBs 364 and 90 (n ) 5). SYNSORBs 374 (maltose) and 376 (cellobiose) did not neutralize toxin A activity.

hemagglutination of rabbit erythrocytes by toxin A from C. difficile (Figure 2) as efficiently as SYNSORB 90 [carrying the RGal(1-3)βGal(1-4)Glc trisaccharide syn-

Bioconjugate Chem., Vol. 8, No. 4, 1997 471

Immobilization of Reducing Sugars

thesized in a traditional manner and linked via a 8-(methoxycarbonyl)octyl spacer]. Neither SYNSORB carrying maltose (374) nor SYNSORB carrying cellobiose (376) neutralized the hemagglutination activity of toxin A of C. difficile (data not shown). The results from this study suggest that immobilization of simple oligosaccharides on a chain-extended inert support using inexpensive reagents provide cost-effective SYNSORB alternatives that could serve as potential therapeutics for diarrheal diseases that affect third-world countries. Additional investigations are underway to further refine the production of these toxin binding agents to a point where cost of production would be favorable for developing therapeutics for diseases such as cholera. ACKNOWLEDGMENT

This work was supported by an industrial contract from SYNSORB Biotech. Inc., 201, 1204 Kensington Rd. NW, Calgary, AB, Canada T2N 3P5. We thank Sheila Hubscher for technical assistance. LITERATURE CITED (1) Karlsson, K.-A. (1989) Animal glycosphingolipids as membrane attachment sites for bacteria. Annu. Rev. Biochem. 58, 309-350. (2) Lemieux, R. U., Baker, D. A., Weinstein, W. M., and Switzer C. M. (1981) Artificial antigens. Antibody preparations for the localization of Lewis determinants in tissues. Biochemistry 20, 199-205. (3) Armstrong, G. D., Fodor, E., and Vanmaele R. (1991) Investigation of shiga-like toxin binding to chemically synthesized oligosaccharide sequences. J. Infect. Dis. 164, 11601167. (4) Armstrong, G. D., Rowe, P. C., Goodyer, P., Orrbine, E., Klassen, T. P., Wells, G., MacKenzie, A., Lior, H., Blanchard, C., Auclair, F., Thompson, B., Rafter, D. J, and McLaine, P. N. (1995) A phase I study of chemically synthesized verotoxin

(Shiga-like toxin) Pk-trisaccharide receptors attached to Chromosorb for preventing hemolytic-uremic syndrome. J. Infect. Dis. 171, 1042-1045. (5) Heerze, L. D., Kelm, M. A., Talbot, J. A., and Armstrong G. D. (1994) Oligosaccharide sequences attached to an inert support (SYNSORB) as potential therapy for antibioticassociated diarrhea and pseudomembranous colitis. J. Infect. Dis. 169, 1291-1296. (6) Fishman, P. H., Pacuszka, T., and Orlandi, P. A. (1993) Gangliosides as receptors for bacterial enterotoxins. Adv. Lipid Res. 25, 165-187. (7) Blanken, W. M., and Van den Eijnden, D. H. (1985) Biosynthesis of terminal GalR1-3Galβ1-4GlcNAc oligosaccharide sequences on glycoconjugates. J. Biol. Chem. 260, 12927-12934. (8) Palcic, M. M., Heerze, L. D., Pierce, M., and Hindsgaul, O. (1988) The use of hydrophobic synthetic glycosides as acceptors in glycosyltransferase assays. Glycoconjugate J. 5, 4963. (9) Dubois, M., Gilles, K. A., Hamilton, J. K., Rebers, P. A., and Smith, F. (1979) Colorimetric method for determination of sugars and related substances. Anal. Chem. 28, 350-356. (10) Sullivan, N. M., Pellet, S., and Wilkins T. D. (1982) Purification and characterization of toxin A and B from Clostridium difficile. Infect. Immunol. 35, 1032-1040. (11) Banoub, J., and Bundle, D. R. (1979) Stannic tetrachloride catalysed glycosylation of 8-ethoxycarbonyloctanol by cellobiose, lactose, and maltose octaacetates; synthesis of R- and β-glycosidic linkages. Can. J. Chem. 57, 2085-2090. (12) Blomberg, L., Wieslander, J., and Norberg, T. (1993) Immobilization of reducing oligosaccharides to matrices by a glycosylamide linkage. J. Carbohydr. Chem. 12, 265-276. (13) Lemieux, R. U., Bundle, D. R., and Baker, D. A. (1975) The properties of a “synthetic” antigen related to the human bloodgroup Lewis a. J. Am. Chem. Soc. 97, 4076-4083. (14) Hutchins, S. M., and Chapman, K. T. (1995) A strategy for urea linked diamine libraries. Tetrahedron Lett. 36, 2583-2586.

BC9700603