Bioconjugate Chem. 2007, 18, 1442−1449
1442
Biofunctionalized Surfactant Mesophases as Polyvalent Inhibitors of Cholera Toxin Anastasios Polyzos,*,† Malcolm R. Alderton,‡ Raymond M. Dawson,‡ and Patrick G. Hartley† CSIRO Molecular and Health Technologies, Bayview Avenue, Clayton South, Victoria 3169, Australia, and DSTO Melbourne, Defence Science and Technology Organisation, P.O. Box 4331, Melbourne, Victoria 3001, Australia. Received February 25, 2007; Revised Manuscript Received July 9, 2007
The cubic lyotropic mesophase composed of the ganglioside GM1 and the synthetic surfactant phytantriol has been employed as a scaffold to prepare a polyvalent inhibitor of cholera toxin (CT). Surfactant mixtures containing up to 20% (w/w) GM1/phytantriol afforded a hydrated insoluble gel-like material, which retained an inverse cubic phase (Q) structure in excess water and at 22 °C, as confirmed using small-angle X-ray scattering. The GM1functionalized mesophases bind up to 98.8% of CT from solution containing 100 ng/mL of CT with a dissociation constant (Kd) of 67-73 nM. The estimated IC50 values for the mesophase systems were 0.1-0.27 µM. The inhibitory effect of the mesophases may be enhanced through the high internal surface area of the inverse cubic phase in addition to the optimal self-orientation of GM1 ligand to match the distribution of binding sites on the toxin surface. As a result, polyvalent inhibitor materials manufactured using these mesophase structures do not require precise control of ligand distribution, which offers advantages with respect to complexity of synthesis and formulation when compared to more rigid and conformationally restricted materials. This approach provides a route to a unique class of polyvalent inhibitors, which should be broadly applicable to a range of bacterial and plant toxins.
INTRODUCTION Cholera toxin (CT) is a heterohexameric protein which is secreted by the Gram-negative bacteria Vibrio cholerae (1) and is the agent responsible for the diarrheal disease cholera. The toxin architecture consists of a catalytic A subunit that is surrounded by an axisymmetric arrangement of five identical B subunits (2), and it therefore belongs to the AB5 class of toxins. Within the lumen of the gastrointestinal tract, toxin endocytosis is mediated by the polyvalent recognition of the cell surface receptor lipid, ganglioside GM1 (Figure 1) (2), by the B pentamer (3). The enzymatic A subunit then separates from the B-pentamer and translocates into the cytosol, where it effects the stimulation of cyclic adenosine monophosphate (cAMP) production. This results in an active efflux of salts and water from the lumen, an accumulation of intestinal fluids in the gut, and diarrhea. An attractive approach to developing cholera prophylactics or therapeutics involves the development of receptor-binding antagonists that interfere with B-subunit recognition of the cell surface receptor. Crystallographic elucidation of the structure of CT has led to the extensive characterization of the recognition of GM1 by the binding epitopes in the pentameric B-subunits (4, 5). These studies reveal that the terminal galactose of the GM1 oligosaccharide binds to a shallow binding epitope on each B subunit (6). Hence, the development of molecular mimics of the terminal galactose and sialic acid groups on the GM1 oligosaccharide are of interest (7). CT receptor-binding antagonists, or inhibitors, fall into two classes. The first class comprises low molecular weight galactose derivatives, such as m-nitrophenyl galactose (MNPG) (8), that reversibly bind to one or more binding sites on the B-pentamer * Author to whom correspondence should be addressed. E-mail:
[email protected]. † CSIRO Molecular and Health Technologies. ‡ Defence Science and Technology Organisation.
of CT. This class of inhibitor, also known as a monovalent inhibitor, has proven only moderately successful due to their relatively low affinity (9), which can be attributed to the weak binding interaction with the receptor site. The second class of inhibitors are known as polyvalent inhibitors. Polyvalent inhibitors, which are generally macromolecules, employ multiple binding moieties (typically galactose functionalities) to take advantage of the pentameric nature of the B subunit-cell surface receptor interaction. Structure-based inhibitors that incorporate five or ten single-site inhibitors covalently coupled to a central core have been reported (10, 11), resulting in remarkable gains in CT inhibitor activity compared to the monovalent analogues (7, 12-15). A variety of polyvalent scaffolds using generic backbones have been developed including random-coil and cyclic polypeptide-based glycopolymers (16). Further, oligosaccharide-derivatized dendrimers (glycodendrimers), which are hyperbranched polymers with a narrow size distribution, have been used as polyvalent scaffolds (17, 18). The polyvalent display of mono- and oligosaccharides at the glycodendrimer surface has resulted in binding enhancements of up to 3 orders of magnitude compared with the monovalent analogue (19). The limitation associated with the rational design of polyvalent CT inhibitors is that their design is heavily reliant on crystallographic information, where the geometry and the dimensions of the polyvalent macromolecules must match the spatial distributions of binding epitopes on the B pentamer. As such, in pentameric and decameric inhibitors, the spacer length between the central core and pendant galactose groups must be optimized, using numerous structural derivatives to achieve optimal binding. Further, the display of galactose derivatives on polymeric scaffolds may not contain the requisite complimentary geometry to match the CT epitope surface, and these restrictions may often lead to modest binding affinities. We have recently introduced a new class of polyvalent scaffold based on lyotropic (solvent containing) mesophases (20). In the presence of water, these systems develop from the
10.1021/bc0700640 CCC: $37.00 © 2007 American Chemical Society Published on Web 08/24/2007
Biofunctionalized Surfactant Mesophases
Bioconjugate Chem., Vol. 18, No. 5, 2007 1443
Figure 1. The structures of GM1 (1) and phytantriol (2).
aggregation of lipid molecules into a liquid crystalline phase, forming a three-dimensional bilayer structure with a high degree of internal order and structural uniformity. These self-assembled structures include the lamellar phase (LR), which is comprised of sheets of stacked bilayer sheets and the inverse hexagonal phase (HII) which can be conceptualised as infinitely packed cylindrical rods with an aqueous interior. Of particular interest is the bicontinuous inverse cubic phase in which a surfactant bilayer is folded onto an infinitely periodic minimal surface (IPMS) (21) consisting of infinite interwoven arrays of saddle surfaces with zero mean curvature and constant negative Gaussian curvature (22, 23). The middle of the bilayer, defined by the terminal methyl groups of the intersecting hydrophobic moieties, forms the IPMS and the hydrophilic parts interface with the bulk water. This molecular architecture affords distinct aqueous regions that form two continuous water networks (or channels) throughout the cubic phase. Long-range crystallographic periodicity is found in such structures, with crystallographic space groups including the Im3m, Pn3m, and Ia3d having been reported (24, 25). The inverse cubic mesophase is thermodynamically stable and coexists in equilibrium with excess water over a broad temperature range. The resulting self-assembled material, in contrast to other aggregate systems such as micelles, is stable to dilution in water. These properties, and in particular the internal three-dimensional cubic nanostructure, have been exploited in biological applications such as protein entrapment and crystallization (26, 27) as well as drug delivery (28-33). We reasoned that the cubic mesophase provides the appropriate hydrophilic and hydrophobic microenvironments reminiscent of the cell membrane, which should be favorable for biological ligand-receptor recognition (31). We further reasoned that owing to the high surface area of the internal mesophase structure (up to 400 m2/g) (34), the inverse cubic phase provides an excellent scaffold for the immobilization of a high concentration of mobile monomeric antagonists which might facilitate self-arrangement of the ligands resulting in a polyvalent interaction. Such self-arrangement mitigates the requirement of a rigid molecular architecture currently employed in the design of polyvalent inhibitors. We have recently demonstrated the utility of this approach by incorporating synthetic galactosefunctionalized surfactants within cubic phases for the sequestration of the plant toxin ricin (20). Here, we demonstrate for the first time a cubic phase toxin inhibitor based on the naturally derived lipid receptor for cholera toxin, ganglioside GM1 (Figure 1). The interaction between GM1 and CT is well characterized and involves a very strong binding interaction (Kd, 0.7-7 nM) (35).
The matrix we have employed for supporting the bioactive GM1 is the bicontinuous cubic phase of the inverse type formed by the surfactant 3,7,11,15-tetramethyl-1,2,3-hexadecanetriol or ‘phytantriol’ (Figure 1). The inverse cubic phase of phytantriol exists over a broad temperature range (36) and importantly, is not susceptible to hydrolysis that other surfactants such as glycerol monoolein (GMO) may undergo. We have studied both the physical chemistry and binding efficacy of GM1 functionalized cubic phases. We believe that this approach demonstrates the potential for inverse cubic phases in either bulk or dispersed forms to act as generic scaffolds for the design of polyvalent inhibitors which target a large class of biotoxins (37).
EXPERIMENTAL PROCEDURES Preparation of GM1/Phytantriol Mixtures. Monosialoganglioside (GM1) (95%) obtained from bovine brain was purchased from Sigma Aldrich. 3,7,11,15-tetramethyl-1,2,3-hexadecanetriol (phytantriol) (96%) was purchased from Aldrich and was used without further purification. Dichloromethane was HPLC grade and used without further purification. The required quantity of GM1 was suspended in phytantriol and dissolved in a solution of 10 mL of 20% methanol/dichloromethane. The mixture was thoroughly dissolved until an optically transparent solution resulted. The solvent was then removed using rotary evaporation at a temperature of 40 °C for 25 min @ -68 kPa to afford a clear viscous oil. Phase Behavior Observations by Polarizing Microscopy. Each sample mixture was placed between a microscope slide and coverslip and placed into the slide holder of a Mettler FP82HT hot stage controlled by a FP90 central processor. Water was dispersed at the edge of the coverslip, and capillary action drew the water between the two glass surfaces to surround the surfactant mixture. The sample was heated from room temperature (20 to 22 °C) at 2 °C/min or less. The interaction of water with the GM1/phytantriol mixtures was viewed using an Olympus IX-70 microscope equipped with crossed polarizing lenses. Images were captured using a Canon EOS 300D digital camera. These ‘water penetration scans’ were usually performed in the temperature range 22-100 °C ( 0.5 °C. Small-Angle X-ray Scattering (SAXS). The lipid phases in the presence of excess water were used for SAXS. Samples were analyzed at 22 °C. The hydrated mesophase systems did not exhibit any signs of instability under the conditions described. All small-angle X-ray scattering analyses were performed using a Bruker Nanostar SAXS camera equipped with a position sensitive 2-D “Hi-Star” detector using point collimated copper KR radiation. Experiments were performed under a low-pressure
1444 Bioconjugate Chem., Vol. 18, No. 5, 2007
helium atmosphere to minimize background scattering. The gels were prepared with excess water and placed between two mica windows taped to a 1 mm stainless steel spacer. This was clamped to the analysis geometry by a stainless steel plate. Data acquisition was performed at a sample-to-detector distance of 23 cm which facilitated analysis of intensity behavior over the wavevector range: 0.0498 Å-1 e q e 0.950 Å-1. The scattering profiles of the empty sample geometries were subtracted from the final scattering profile of the sample (per-pixel subtraction) after first correcting for the intensity of the X-ray beam. Assay for Cholera Binding by Lyotropic Mesophases. Nunc-Immuno 96-microwell flat-bottom plates with a Maxisorp surface were obtained through Medos Company, Melbourne, Australia. Phosphate-buffered saline (PBS) was prepared by dissolving a tablet from Sigma-Aldrich in 200 mL of deionized distilled water. The resultant solution comprised 0.01 M phosphate, 0.137 M NaCl, and 0.0027 M KCl, pH 7.4 at 25 °C. PBS-Tween was prepared by dissolving the contents of a sachet from Sigma-Aldrich in 1 L of deionized distilled water. This gave a solution of 0.01 M phosphate, 0.138 M NaCl, 0.0027 M KCl, 0.05% Tween 20, pH 7.4, at 25 °C. Cholera toxin from Vibrio cholerae Inaba 569B (azide free) was obtained from List Biological Laboratories, Campbell, CA. The toxin was received as a powder, dissolved in 1.0 mL of deionized distilled water, and then stored at 2 °C. The reconstituted cholera toxin contained 1 mg of protein in 1.0 mL of 0.05 M Tris, 0.2 M NaCl, and 0.001 M Na2EDTA at pH 7.5. The solution was diluted with PBS-Tween immediately prior to assays. Purified polyclonal goat anti-cholera toxin was a product of Biogenesis, Poole, UK. All other materials were obtained from SigmaAldrich, Sydney, Australia. Monosialoganglioside GM1 from bovine brain was obtained as 1 mg of powder. It was dissolved in 1 mL of methanol, stored at -20 °C, and diluted just before use. The substrate solution for alkaline phosphatase was prepared by dissolving two p-nitrophenyl phosphate tablets and two TRIS buffer tablets from Sigma-Aldrich in 11 mL of deionized distilled water. Stock solutions (50 mg/mL or 23 mg/mL) of the surfactant mixtures were prepared in methylene chloride. Serial dilutions of the mixtures in dichloromethane were prepared to give the appropriate concentration. Ten-microliter aliquots of the serial dilutions were placed in glass test tubes, and the solvent was allowed to evaporate for 1-2 h in a fume cupboard at room temperature (20 °C). The movement of air in the fume cupboard facilitated the rapid evaporation of the solvent. Cholera toxin solution, of a concentration between 1.0 and 100 ng/mL in blocking buffer (PBS-Tween with 0.25% bovine serum albumin; 130 µL), was added to the tubes, and they were placed on a rocker platform and gently oscillated for 24 h at room temperature. Aliquots (100 µL) of the supernatants from these tubes were then added to immobilized GM1 in Nunc microtiter plates, and the cholera toxin assay was performed as described below. Additional runs were performed both without the mesophase solution and in the presence of pure phytantriol (without GM1) as controls. In the standard cholera toxin assay, all incubations were at room temperature (20-22 °C), with a volume per well of 0.1 mL except for the incubation with blocking buffer (0.36 mL). After each incubation, the microtiter plate was emptied and washed three times with PBS-Tween. An additional wash with distilled water was performed before addition of the alkaline phosphatase substrate. The wells of a microtiter plate received 100 ng/mL of GM1 in PBS and were left overnight (ca. 16 h) before rinsing with PBS-Tween and sequential treatment with blocking buffer, cholera toxin from the assay supernatant (100 µL in blocking buffer), goat anti-cholera toxin antibody (1:8000 dilution in blocking buffer), rabbit anti-goat alkaline phosphatase
Polyzos et al. Table 1. GM1/Phytantriol Liquid Crystalline Phases Assigned in the Presence of Excess Water temperature regime of phases (°C) mixture
LR
Q
HII
LII
100% phytantriol 10% GM1/phytantriol 20% GM1/phytantriol 30% GM1/phytantriol 40% GM1/phytantriol
