Anal. Chem. 2003, 75, 2256-2261
Ganglioside-Liposome Immunoassay for the Ultrasensitive Detection of Cholera Toxin Soohyoun Ahn-Yoon,† Thomas R. DeCory,† Antje J. Baeumner,‡ and Richard A. Durst*,†
Department of Food Science & Technology and Department of Biological & Environmental Engineering, Cornell University, Geneva, New York 14456-0462
An extremely sensitive bioassay has been developed for cholera toxin (CT) detection, using ganglioside-incorporated liposomes. Cholera is a diarrheal disease, often associated with water or seafood contamination. Ganglioside GM1 was used to prepare the liposomes by spontaneous insertion into the phospholipid bilayer. CT recognition and signal generation is based on the strong and specific interaction between GM1 and CT. In a sandwich immunoassay, CT was detected as a colored band on the nitrocellulose membrane strip, where CT bound to GM1liposomes can be captured by immobilized antibodies. The intensity of the band could be visually estimated or measured by densitometry, using computer software. The limit of detection (LOD) of CT in the assay system was found to be 10 fg/mL which is equivalent to 8 zmol in the 70-µL sample. The assay was also tested with water samples spiked with CT, providing a LOD of 0.1-30 pg/mL, which is much better than previously reported limits of detection from other assays. The assay could be completed within 20 min. These results demonstrate that the bioassay developed for CT is rapid and ultrasensitive, suggesting the possibility for detecting CT, simply and reliably, in field screening. Cholera toxin (CT) is a protein enterotoxin produced by Vibrio cholerae. Cholera is an epidemic disease characterized by severe watery diarrhea that can lead to rapid dehydration, acidosis, and even death in 3-4 h, if left untreated. Despite the understanding of its molecular mechanism and the development of detection methods for V. cholerae or CT, cholera still remains a major concern throughout many underdeveloped countries. Most cholera cases are reported in Southeast Asia, Africa, and South America, and it is estimated that cholera causes approximately 120 000 deaths annually.1 CT (85 kDa) is composed of a single A subunit (27 kDa) and five identical B subunits (11.6 kDa each). The A subunit has two peptides (A1 and A2) linked by a disulfide bond. The B subunits bind specifically to ganglioside receptors on the cell membrane, while the A subunit is an enzymatic protein responsible for adenylate cyclase activation.2 After CT binds to * Corresponding author. E-mail:
[email protected]. Phone: 315-787-2297. Fax: 315-787-2284. † Department of Food Science & Technology. ‡ Department of Biological & Environmental Engineering. (1) Popovic, T.; Olsvik, O.; Blake, P. A.; Wachsmuth, K. J. Food Prot. 1993, 56, 811-821. (2) Ganguly, N. K.; Kaur, T. Indian J. Med. Res. 1996, 104, 28-37.
2256 Analytical Chemistry, Vol. 75, No. 10, May 15, 2003
GM1 on the cell surface through the B subunit, the A subunit is translocated across the membrane and the disulfide bond is reduced to form the active A1 subunit. Inside the cell, the A1 subunit transfers the ADP-ribose moiety of nicotinamide adenine dinucleotide to the regulatory GSR protein, which controls the activity of adenylate cyclase. ADP-ribosylation of GSR protein by CT activates adenylate cyclase, resulting in cAMP accumulation in the cytoplasm. The cAMP accumulation causes osmotic imbalance across the membrane by increasing Cl- secretion, which leads to diarrhea.3 CT is a common contaminant of water and foods, and even though it has been eliminated in most countries, it can enter the food supply through imported foods contaminated with CTproducing V. cholerae. In addition, the possibility of bioterrorism utilizing biological toxins makes the development of rapid and sensitive methods for detecting CT of great importance. Bacterial toxins must first bind to the cell receptors to exert their toxicity in the cell. Gangliosides and carbohydrates existing as glycolipids or glycoproteins on the cell surface have long been implicated as major receptors for biological toxins, as well as for the cell-cell recognition and hormones.4 Gangliosides (Figure 1) contain both hydrophilic and hydrophobic regions and carry negative electric charges. The hydrophobic portion, ceramide, consists of a long-chain fatty acid linked to the amino alcohol sphingosine through the amide bond. The hydrophilic carbohydrate moiety is composed of hexoses, N-acetylated hexosamines, and at least one sialic acid.5 In the membrane, the ceramide portion is imbedded in the lipid bilayer, while the hydrophilic oligosaccharide chain is exposed to the outer environment. It has been reported by two research groups6,7 that CT binds specifically to the monosialoganglioside, GM1. Liposomes, spherical vesicles composed of a phospholipid bilayer surrounding an aqueous cavity, were originally developed to study cell membranes.8 However, because of their ability to carry different agents in the aqueous cavity, liposomes have been utilized in diagnostics, in drug delivery, and even by the cosmetics (3) Lencer, W. I.; Delp, C.; Neutra, M. R.; Madara, J. L. J. Cell Biol. 1992, 117, 1197-1209. (4) Eidels, L.; Proia, R. L.; Hart, D. A. Microbiol. Rev. 1983, 47, 596-620. (5) Fishman, P. H.; Brady, R. O. Science 1976, 194, 906-915. (6) Cumar, F. A.; Maggio, B.; Caputto, R. Mol. Cell Biochem. 1982, 46, 155160. (7) Fishman, P. H.; Pacuszka, T.; Orlandi, P. A. Adv. Lipid Res. 1993, 25, 165187. (8) Papahadjopoulos, D.; Portis, A.; Pangborn, W. Ann. N.Y. Acad. Sci. 1978, 308, 50-66. 10.1021/ac026428t CCC: $25.00
© 2003 American Chemical Society Published on Web 04/19/2003
Figure 1. Structure of the monosialoganglioside, GM1. GM1 is a natural receptor for cholera toxin. Abbreviations: Glc, glucose; Gal, galactose; GalNAc, N-acetylgalactosamine; NANA, N-acetylneuraminic acid.
and food industries. Liposomes used in detection assay systems mostly occur as immunoliposomes with antibodies bound to the surface and also as DNA- or analyte-tagged liposomes. Despite having specificity and strong affinity for biological toxins comparable to those of antibodies, gangliosides have not been used as receptors in liposome-based assays until recently.9,10 Gangliosideincorporated liposomes have advantages over immunoliposomes because of the amphiphilicity of the gangliosides. Gangliosides contain the hydrophobic ceramide, which can be spontaneously incorporated into a lipid bilayer structure, while antibodies require several chemical steps for covalent conjugation to the liposome structure. In this paper, we describe detection of CT, using GM1incorporated liposomes (GM1-liposomes) containing the red dye, sulforhodamine B (SRB). Antibodies against the toxins were coated in a narrow stripe onto plastic-backed nitrocellulose (NC) membrane sheets, which were later cut into the test strips. The assay system consisted of GM1-liposomes, buffer, test strips, and pure toxin samples or food samples. CT first binds to GM1 on the liposomes and then is captured by the antibodies on the test strip. When present in the samples, CT was observed as a colored band on the strip. The intensity of the band was measured by densitometry, using a computer scanner. The GM1 assay system showed promising results as an alternative method to ELISA or other conventional assays, having the advantages of sensitivity, speed, and simplicity. EXPERIMENTAL SECTION Materials. Dipalmitoyl phosphatidylcholine (DPPC) and dipalmitoyl phosphatidylglycerol (DPPG) were purchased from Avanti Polar Lipids, Inc. (Alabaster, AL). Monosialoganglioside (GM1), SRB, cholesterol, N-acetylneuraminic acid (NANA), and all other chemicals were purchased from Sigma Chemical Co. (St. Louis, MO). To avoid the biological hazard, commercial toxin subunits or toxoids (formaldehyde-inactivated toxins) were used, if available, in this study. Cholera toxin B subunit was purchased from Sigma Chemical Co. Tetanus toxoid, diphtheria toxoid, botulinum neurotoxin type A toxoid, and Escherichia coli heatstable toxin (STa) were obtained from List Biological Laboratories, (9) Pan, J. J.; Charych, D. Langmuir 1997, 13, 1365-1367. (10) Singh, A. K.; Harrison, S. H.; Schoeniger, J. S. Anal. Chem. 2000, 72, 60196024.
Inc. (Campbell, CA). Mouse monoclonal antibodies to the B subunit of cholera toxin were obtained from Biodesign International (Saco, ME). NC membrane with plastic backing (pore size, 10 µm) was purchased from Millipore (Bedford, MA). Polycarbonate (PC) filter membranes of 0.2-µm pore size were obtained from Whatman International Ltd. (Maidstone, England). STa is an intact toxin, so it requires handling precautions. Appropriate laboratory attire should be worn, including a lab coat, gloves, and safety glasses. In case of exposure, the area of the body that comes into contact with STa should be washed thoroughly. STa can be inactivated by 0.04 mM dithiothreitol or 0.1 M β-mercaptoethanol. STa also can be inactivated by autoclaving at 121 °C and 15 psi11,12 (not recommended for personnel). Preparation of GM1-Liposomes. GM1-liposomes were prepared by the extrusion method, with repetitive freeze-thaw cycles,13 from a mixture of DPPC, DPPG, cholesterol, and GM1 in a molar ratio of 40.3:4.2:40.9:1.3. In a 100-mL round-bottom flask, the lipid mixture (86.7 µmol) was dissolved in 7 mL of a chloroform/methanol mixture (6:1, v/v). The lipid mixture was dried on a rotary evaporator to form a thin lipid film on the flask wall. To the dry lipid mixture, 4 mL of 150 mM SRB solution in 20 mM HEPES buffer (pH 7.5) containing 0.01% sodium azide was added. After gentle swirling, five cycles of freezing and thawing were performed by alternating placement of the flask in a dry ice/acetone bath and a 50 °C water bath. The hydrated lipids were extruded through a 0.2-µm-pore-sized PC filter membrane, using a miniextruder (Avanti Polar Lipids, Inc.). The resulting liposomes were gel-filtered through a 1.5 × 25 cm Sephadex G-50 column, equilibrated with Tris-buffered saline (TBS; 25 mM Tris base, 150 mM NaCl, pH 7.4) containing 0.01% sodium azide and sucrose for osmolality, to remove unencapsulated dye. The phospholipid concentration in the resulting liposomes was determined by phosphorus quantification using Bartlett’s phosphorus assay.14 The mean diameter of the liposomes was measured by laser diffraction particle size analysis, with an LS particle size analyzer (Coulter Scientific Instruments, Hialeah, FL). Gan(11) Dreyfus, L. A.; Frantz, J. C.; Robertson, D. C. Infect. Immun. 1983, 42, 539-548. (12) Thompson, M. R.; Luttrell, M.; Overmann, G.; Giannella, R. A. Anal. Biochem. 1985, 48, 26-36. (13) Mayer, L. D.; Hope, M. J.; Cullis, P. R. Biochim. Biophys. Acta 1986, 858, 161-168. (14) Bartlett, G. R. J. Biol. Chem. 1959, 234, 466-468.
Analytical Chemistry, Vol. 75, No. 10, May 15, 2003
2257
Figure 2. Test strip assay format. CT in the reaction mixture binds to the gangliosides on the liposome surface. The CT-GM1-liposome complex migrates through the nitrocellulose test strip by capillary action until it reaches the analytical zone, where toxins in the complex are captured by immobilized antibodies. This binding zone is shown as the dark band on the test strip.
glioside concentration in the liposomes was quantified by the method of Hikita et al..15 Liposome concentration, receptor concentration, and number of SRB molecules per liposome were determined as described previously.16 Preparation of Test Strips. Test strips were prepared as reported previously,16 with modifications. The NC membrane sheet with plastic backing was prewetted with 10% (v/v) methanol in phosphate-buffered saline (PBS; 20 mM KH2PO4, 150 mM NaCl, pH 7.4) containing 0.01% sodium azide and dried at room temperature in a vacuum oven. Antibodies to the CT B subunit (concentration, 0.65 mg/mL) in PBS were applied to the analytical zone (∼2 cm from one end of the membrane) of the NC membranes, using a Linomat IV TLC applicator (Camag Scientific, Wrightsville Beach, NC). The antibody-immobilized membranes were dried for 1.5 h under vacuum at room temperature and then immersed in the blocking solution (2% poly(vinylpyrrolidone) (PVP), 0.01% gelatin and 0.002% Tween 20 in PBS) for 1 h at room temperature with shaking and dried overnight under vacuum at room temperature. After drying, membranes were cut into test strips (5 × 50 mm or 5 × 80 mm), and a filter paper pad was attached to the top of the test strip to provide additional absorbency for the migration process. Assay Formats. The assay was performed at room temperature by adding 30 µL of the GM1-liposome stock solution, diluted with TBS containing 0.01% sodium azide, to 70 µL of the sample in a glass test tube (10 × 75 mm). Thus, the total volume of the test mixture was 100 µL. After the contents were mixed briefly, the test strips were inserted into the mixture and left in the tube until all of the solution was drawn from the bottom of the test tube. This capillary migration process takes approximately 15-20 min. The assay format is depicted in Figure 2. Detection and Quantification. The signal color on the test strips can be detected visually, or for quantification of the signal (15) Hikita, T.; Tadano-Aritomi, K.; Iida-Tanaka, N.; Toyoda, H.; Suzuki, A.; Toida, T.; Imanari, T.; Abe, T.; Yanagawa, Y.; Ishizuka, I. Anal. Biochem. 2000, 281, 193-201. (16) Siebert, S. T. A.; Reeves, S. G.; Durst, R. A. Anal. Chim. Acta 1993, 282, 297-305.
2258
Analytical Chemistry, Vol. 75, No. 10, May 15, 2003
intensity, gray scale densitometry can be used. The test strips were scanned using a color image scanner, and the scanned images were converted into gray scale readings. The intensities of each signal were quantified with Scan Analysis densitometry software (Biosoft, Ferguson, MO). Detection of CT in Water Samples. Tap water and spring water (commercial bottled spring water) were spiked with various concentration of CT and stored at room temperature. The assay was performed periodically with the spiked water samples over 3 days. The signal was determined as described above. RESULTS AND DISCUSSION Characterization and Optimization of GM1-Liposomes. GM1-incorporated liposomes encapsulating SRB were prepared by the extrusion method with repetitive freeze-thaw cycles. The amount of GM1 was maintained at approximately 1-2 mol % of total lipids in the lipid mixture. When the amount of GM1 in the resulting liposomes was measured by the DMB assay, it was found that ∼40% of the GM1 in the lipid mixture was incorporated into the GM1-liposomes. The mean diameter of GM1-liposomes was measured by a particle size analyzer to be 208 nm. The liposomes were extruded through a 0.2-µm pore-sized PC membrane filter, and thus, the size of liposomes can be controlled by the pore size of membrane filter used during the extrusion. With the assumption that the thickness of the lipid bilayer is 4 nm,17 the internal volume of a single liposome was calculated from the diameter. On the basis of the average size and lipid concentration in the liposomes, all other characteristics could be calculated as described previously.16 The characteristics of the liposomes used in these studies are shown in Table 1. Size and concentrations of liposomes and SRB were comparable to those previously reported from another test strip assay.18 The number of ganglioside molecules expressed on the liposome surface (1.7 × 104 molecules/liposome) was very similar to that reported by Singh et al.10 Since the fluorescence (17) Israelachvili, J. N.; Mitchell, D. J. Biochim. Biophys. Acta 1975, 389, 1319. (18) Park, S.; Durst, R. A. Anal. Biochem. 2000, 280, 151-158.
Table 1. Characteristics of GM1-Liposomes mean diameter (nm) volume (µL) liposome concentration (liposomes/mL) SRB concentration (mM) number of SRB molecules per liposome number of GM1 molecules per liposome
208 ( 43 4.2 × 10-12 8.4 × 1011 150 3.8 × 105 1.7 × 104
Figure 3. Scanned images of representative nitrocellulose test strips. Strips were run at room temperature, as described in the Experimental Section. Each strip was inserted into the test tube with the reaction mixture of GM1-liposomes and CT at the indicated concentrations (µg/mL): A, negative control; B, 1 × 10-8; C, 1 × 10-7; D, 1 × 10-6; E, 1 × 10-5; F, 1 × 10-4; G, 1 × 10-3; H, 1 × 10-2; I, 1 × 10-1; J, 1; K, 10.
of SRB in the liposomes is self-quenched, the stability of the liposomes could be determined by measuring the increase in the fluorescence of SRB resulting from leakage out of the liposomes during storage. No significant changes in liposome stability were observed over 9 months. The liposomes were stored at 4 °C and were suspended in TBS containing 0.01% sodium azide with the same osmolality as the encapsulated SRB in order to prevent osmotic pressure-related swelling or crenation. Development of a Test Strip Assay for CT Detection. The prototype format for the test strip assay is shown in Figure 2. The assay is based on ligand-receptor binding between CT and GM1 on the liposome surface, capillary migration on a nitrocellulose strip, and detection in an analytical zone. When the GM1liposome solution is added to the CT-containing sample, CT-GM1liposome complexes will form by strong and specific binding. These complexes migrate through the nitrocellulose membrane strip by capillary action and are captured by anti-CT antibodies immobilized in the analytical zone of the test strip, forming a colored band in the zone. The intensity of the band is proportional to the amount of toxin in the sample, as shown in Figure 3. After applying the capture antibodies to the analytical zone on the nitrocellulose membrane, the membrane is blocked to minimize the nonspecific binding of liposomes. Of the commonly used blocking agents, PVP, bovine serum albumin (BSA), and gelatin showed consistently low backgrounds. However, because of the slower migration rates in BSA- or gelatin-coated membranes, PVP was chosen as the main blocking agent, with only a small amount of gelatin added. Tween-20 was also added to the blocking solution to aid in the uniform migration of liposomes, but its concentration was only 0.002% to prevent lysis of the liposomes.
Figure 4. Dose-response curve for CT, generated from test strip assays using GM1-liposomes. The solid line represents the thirdorder polynomial curve fit, with an R2 value of 0.997. The straight horizontal line shows the limit of detection, defined as the color intensity 3 times higher than the standard deviation of the background signal. Each point represents four replicates of the gray scale value of the analytical zone.
The analytical sensitivity and detection limit of the test strip assay for CT detection was determined from a dose-response curve (Figure 4). Dose-response data were obtained by scanning densitometry of test strips, which were run at various concentrations of CT. The limit of detection (LOD) was defined as the lowest concentration of toxin producing a signal intensity 3 times higher than the standard deviation of the intensity of the sample without toxin (negative control). By this definition, the instrumental (densitometry) LOD of the current assay for CT was estimated to be 10 fg/mL. The visual detection limit was 100 fg/mL. In the dose-response curve for CT, the intensity of the binding signal increases with increasing concentration of CT in the sample, showing a dynamic analytical range between 101 and 106 fg/mL. At higher concentrations of CT (>107 fg/mL), the intensity of signal reaches saturation, followed by a decrease with increasing CT concentration. This “hook effect” occurs commonly in sandwich assays in the presence of high concentrations of analyte.19 The LOD of the current assay using GM1liposomes is much more sensitive than other detection assays reported to date. While other research groups have used GM1liposomes for CT detection,10,20 those liposomes had dye markers or enzymes on the liposome surface, not in the interior cavity. In this study, dye-encapsulating liposomes were used, which, because of the much larger number of dye molecules contained in each liposome, produced a much higher signal intensity, thereby resulting in higher sensitivity. The assay can be completed within 20 min, including the time needed for the sample preparation, in contrast to the previously reported methods, which take several hours or more. The specificity of the bioassay was studied by performing cross-reactivity studies with (19) Fernando, S. A.; Wilson, G. S. J. Immunol. Methods 1992, 151, 47-66. (20) Alfonta, L.; Willner, I. Anal. Chem. 2001, 73, 5287-5295.
Analytical Chemistry, Vol. 75, No. 10, May 15, 2003
2259
Figure 5. Specificity of the CT detection assay system using GM1liposomes. A 10 µg/mL aliquot of each toxin was dissolved in TBS and used in the test strip assay. For safety reasons, commercially available toxoids were used in this study, except STa. The data points are an average from three replicates. Abbreviations: BT, botulinum toxin; CT, cholera toxin; DT, diphtheria toxin; STa, E. coli heat-stable toxin; TT, tetanus toxin.
various toxins also recognized by ganglioside receptors. For this purpose, diphtheria toxin (DT) from Corynebacterium diphtheriae, E. coli heat-stable toxin), botulinum neurotoxin type A (BT) from Clostridium botulinum, and tetanus toxin (TT) from Clostridium tetani were used. For safety purposes, commercially available toxoids were used in these studies, except for STa. Each toxin was added to the assay system in high concentration (10 µg/mL), and the binding signal was measured, as described in the Experimental Section. As shown in Figure 5, the detection assay using GM1-liposomes showed strong positive signals only to CT, which suggests high specificity of the assay for CT detection. The high specificity of the detection assay provides an advantage over other immunological detection methods, which often show false-positive signals from cross reactivity, especially in sandwich-type assays. Application of the Bioassay to Detection of CT in Water Samples. Most cholera cases are associated with contaminated water sources. For that reason, many methods have been developed to detect V. cholerae in water. Since contamination of water could affect large numbers of people, a rapid detection method is highly desired. In this study, water samples (tap and spring water) were spiked with CT, and the detection assay was run periodically over 3 days. Initially, the LOD in tap and spring water samples was 30 pg/mL and 100 fg/mL, respectively. Figure 6 shows the dose-response curve of CT detection in the water samples. The higher LOD in tap water (pH 7.5; conductivity
