- Email: [email protected]
Real time biosensor analysis of Staphylococcal enterotoxin A in food
International Journal of Food Microbiology 49 (1999) 119–127
Real time biosensor analysis of Staphylococcal enterotoxin A in food a,b
Linda Rasooly , Avraham Rasooly
b,c ,
*
a
Department of Molecular Microbiology and Immunology, Johns Hopkins University, Baltimore, MD, USA b Department of Pathology, Johns Hopkins University, Baltimore, MD, USA c US Food and Drug Administration, Division of Microbiological Studies, 200 C St. SW, HFS 515, Washington, DC 20204, USA Accepted 14 April 1999
Abstract Currently there is no ‘real-time’ detection system to identify food borne toxins. In order to develop such a system, we have used a evanescent wave biosensor for real time detection of staphylococcal enterotoxin A (SEA) in foods. The approach used here is sandwich biosensor, a method utilizing two antibodies. The toxin binds initially to a capturing antibody which is bound covalently on the surface of the biosensor detector. The second antibody binds to the captured toxin. We were able to measure SEA in foods with little or no background interference, demonstrating that biosensor-based measurement of SEA was possible not only with purified SEA but also in complex food matrices such as hot dogs, potato salad, milk and mushrooms. Autoclaved samples of SEA did not evoke a positive response. With both purified SEA and SEA-spiked foods, the assay sensitivity is 10–100 ng / g depending on the material tested and the assay is rapid ( , 4 min) when a single antibody is used. 1999 Elsevier Science B.V. All rights reserved. Keywords: Biosensor; Foodborne toxins; Staphylococcal enterotoxins
1. Introduction Staphylococcal enterotoxins (SE) are a family of five major serological types (SEA through SEE) of heat stable (Tan et al., 1969; Denny et al., 1971; Humber et al., 1975; Lee et al., 1977) emetic enterotoxins. They are encoded by five genes that
*Corresponding author. Tel.: 11-202-205-4192; fax: 11-202401-7740. E-mail address: [email protected] (A. Rasooly)
share 50 to 85% homology at the predicted amino acid level (Betley et al., 1984; Betley and Mekalanos, 1985; Betley et al., 1986; Betley and Mekalanos, 1988; Singh et al., 1988). Staphylococcal enterotoxin A (SEA), a monomeric protein (Mr 27 000), is a heat stable, extremely potent gastrointestinal toxin (Archer and Young, 1988; Bergdoll, 1991; Bergdoll, 1995) and is one of the leading causes of gastroenteritis (vomiting and diarrhea) resulting from consumption of contaminated food (Archer and Young, 1988; Garthright et al., 1988). Its potency requires that methods to detect it in foods
0168-1605 / 99 / $ – see front matter 1999 Elsevier Science B.V. All rights reserved. PII: S0168-1605( 99 )00053-7
120
L. Rasooly, A. Rasooly / International Journal of Food Microbiology 49 (1999) 119 – 127
be sensitive enough to measure very low levels (nanograms per gram food) (Evenson et al., 1988). Immunological testing (especially ELISA), has been the method of choice for SEA detection since antibodies to the toxin are readily available (Miller et al., 1978; Freed et al., 1982; Thompson et al., 1986; Bergdoll, 1991; Park et al., 1996). Real time detection allows testers to obtain immediate interactive information on the sample. This is important in the diagnostics and detection of food borne toxins, especially during the food production process. A food contamination alert might allow the processor to take immediate corrective measures. Real time detection would also increase the efficiency of food quality testing since the current methodology takes 4–8 h to complete and is not interactive. Biosensors can be used as analytical devices to translate a biological response (e.g. presence of toxin) into an electronic output that can be analyzed. The most common biosensors are based on evanescent wave. These biosensors analyze interactions using evanescent waves, an electromagnetic wave that exists at the surface of many forms of optical waveguides, to measure changes in refractive index at the sensor surface. These changes can be measured as shifts in resonance angle as in the IAsys instrument used in this study. Other common biosensors are based on surface plasmon resonance (SPR). One of the most common biosensors is based on surface plasmon resonance (SPR) (Malmqvist and Karlsson, 1997) SPR is a quantum optical–electrical phenomenon based on the fact that energy carried by photons of light can be ‘coupled’ or transferred to electrons in a metal. Changes in the chemical properties of the environment within the range of the plasmon field (e.g. antibody–antigen binding) causes a change in the plasmon, which can be measured as change in the refraction index, a shift in the wavelength of light absorbed, or by light diffraction. In a biosensor, a specimen is tested for its adsorption to a covalently immobilized molecule (binding agent) by surface sensitive optical techniques (Malmqvist, 1993). The amount of adsorption is measured as a function of time, and results are generated in the form of a sensogram that shows the response units measured as a consequence of the adsorption. Biosensors can be integrated with on-line process monitoring schemes to detect and identify food contaminants and provide real-time information
about toxin presence at every step or time point in a process. Various binding molecules have been developed and utilized for the measuring compounds in clinical, medical, environmental and industrial process settings. Frequently, the biosensors use antibodies (Fagerstam et al., 1990; Malmqvist, 1993) because of their specificity. Other components of the immune system have also been used in biosensors including immunoglobulins, MHC class I (major histocompatibility complex molecules), MHC class II molecules, T-cell receptor, and receptor–ligand binding reactions (Khilko et al., 1993; Seth et al., 1994; Khilko et al., 1995). The sensitivity of biosensor techniques is on the order of 10 212 M, with a total assay time of several minutes (Olson et al., 1990). The kinetics of the binding between Staphylococcal enterotoxin B (SEB) and the T cell receptor (HA1.7, Vb3.1) have been studied using surface plasmon resonance (Olson et al., 1990; Seth et al., 1994). Although the technique was highly sensitive, it was not possible to measure SEA using the T-cell receptor because there the T-cell receptor does not bind SEA in the absence of MHC class II molecules (Seth et al., 1994). Biosensors may be very useful in environmental and food safety analysis. They do not require labeling of either molecule, nor does the analyte need to be purified as the background binding is usually low (Brigham-Burke et al., 1992). Recently, an effective, portable, fiber-optic biosensor method was used for SEB detection in food. This biosensor is very sensitive and detects |10 ng toxin / g various foods, using fluorescent-labeled antibodies (Tempelman et al., 1996). We have used a biosensor to measure SEA. The ligand, rabbit anti-SEA antibody, was covalently crosslinked onto an activated surface and reacted with different concentrations of SEA in solution. The amount of SEA binding to the antibody-coated surface was measured by shifts in resonance angle.
2. Materials and methods
2.1. Instrumentation The IAsys (Affinity Sensors, Paramus, NJ, USA) evanescent wave biosensor applied in this study was used and calibrated according to the manufacturer’s
L. Rasooly, A. Rasooly / International Journal of Food Microbiology 49 (1999) 119 – 127
instructions. The reproducibility of the instrument is 610% depending on the concentration material tested.
2.2. Immobilization of ligand Chemicals were obtained from Sigma (St. Louis, MO, USA) unless specified. Carboxymethyl-dextran (CMD) cuvettes (Affinity Sensors) were washed three times with PBS–Tween (10 mM sodium phosphate, 2.7 mM KCl, 138 mM NaCl, 0.05% Tween-20). A 1:1 (v / v) mixture of N-hydroxysuccinimide (NHS; 1.3% in H 2 O) and 1-ethyl-3-(3dimethylaminopropyl) carbodiimide (EDC) (7.6% in H 2 0) was prepared immediately prior to use. The mixture was applied to the cuvettes (activation step) for 7 min. Cuvettes were washed in PBS–Tween and immobilization buffer (10 mM sodium acetate, pH 5.0) was added to the cuvettes for 2 min. The ligand (5 ml of rabbit anti-SEA affinity purified IgG; Toxin Technology, Sarasota, FL, USA) was added into the immobilization buffer and left for 7 min to bind onto the activated surface. The cuvettes were washed in PBS–Tween and blocked with 1 M ethanolamine (pH 8.5) for 3 min, and then washed in PBS–Tween, followed by washing in regeneration buffer (10 mM HCl) for 2 min. After coating, cuvettes were stored with PBS1azide (0.05%) at 48C until used.
2.3. Preparation of food samples A 10-g amount of potato salad, canned mushrooms or hot dogs purchased from local stores was homogenized with 10 ml PBS. The suspensions were centrifuged at 1000 g for 10 min and supernatant was collected. Milk was prepared as 5% w / v of dry milk (Carnation, Glendale, CA, USA) in PBS. SEA (Sigma) was diluted with sterile distilled water to 1 mg / ml. Aliquots were stored at 2208C. The toxin was added to the homogenized sample before centrifugation. Food specimens were tested either directly, with different concentrations of SEA, or after autoclaving at 1218C for 20 min.
2.4. Testing of food samples A 100-ml aliquot of either a pure SEA solution (Sigma) or a food suspension spiked with SEA was applied to a coated cuvette for 10 min. The cuvette
121
was washed and the remaining binding activity was recorded. For sandwich biosensor detection, 200 ml PBS–Tween was added to the cuvette followed by 1 ml of rabbit anti-SEA affinity purified antibody (Toxin Technology). After a 10-min incubation, the cuvette was washed four times in PBS–Tween and the response was recorded.
3. Results
3.1. Measurements of SEA by sandwich biosensor In the first experiment our goal was to study the binding between SEA and a rabbit anti-SEA antibody bound to the IAsys biosensor cuvette. Our results are reported in terms of the absolute amount of SEA added to the cuvette. The total volume of the cuvette is 100 ml. Upon application of SEA (1 mg) to the cuvette, a strong response peak appeared after less than 1 min and reached equilibrium within 10 min (Fig. 1A,b). Fig. 1B demonstrates the kinetics of the binding. Significant binding (|90%) was complete within 4 min. This appears to be the minimum time required for measuring without significantly sacrificing sensitivity. However, longer incubation times did produce a higher signal. The cuvette was washed to eliminate nonspecific binding (Fig. 1A,c). Following the addition of a second antibody, a new peak appeared reflecting the binding between captured SEA (bound to the immobilized antibody in the cuvette) and the second antibody in solution (Fig. 1A,d). The response remained high even after washing (Fig. 1A,e), showing that SEA–second antibody interaction is specific. We compared the level of detection of SEA by the primary antibody alone to the sandwich biosensor method using serial dilutions of SEA (Fig. 2). As shown in Fig. 2B, the addition of a second antibody generally increased the level of response at all detectable SEA levels. The level of SEA detection of is ¯10 ng with the sandwich biosensor method. Like any other immunological method, the sensitivity of the biosensor assay depends heavily on the antibodies used. Therefore, the results of our studies do not indicate the absolute level of sensitivity of this method. Instead, the measurements reported here reflect, in part, the affinity of specific antibodies for the ligand.
122
L. Rasooly, A. Rasooly / International Journal of Food Microbiology 49 (1999) 119 – 127
Fig. 1. Sandwich biosensor detection of SEA. 1 mg of purified SEA was added to a cuvette coated with anti-SEA rabbit antibody and the response was measured. After 10 min, a second anti-SEA antibody was applied and the response was measured. (A) The interaction between SEA and the two antibodies: (a) Baseline of the buffer; (b) Reaction with SEA; (c) Residual binding of SEA after washing; (d) Reaction following application of a second anti-SEA antibody; (e) Residual binding of second antibody after washing; (f) Return to baseline after acid wash. (B) Detailed first 5 min of the interaction between SEA and the antibody coated cuvette: (a) Buffer baseline; (b) Reaction with 10 ng SEA.
3.2. Statistical analysis To analyze the linearity of the assay, curve fitting analysis ( SLIDEWRITE PLUS software) was performed on the biosensor data. The linear regression between the amount of toxin and the signal was very high, with and r 2 value of 0.92 and 0.94 for one and two antibodies, respectively. The best fit curve is a third order polynomial curve with r 2 of 0.99 for one or two antibodies. The equations for these curves are y 5 7.38 1 23.41x 1 8.51x 2 2 0.97x 3 for one antibody and y 5 32.61 1 60.27x 1 18.27x 2 2 2.36x 3 . These results suggest that there is very high, nonlinear, correlation between the amount of SEA and the biosensor signal.
3.3. Measurement of SEA in milk by sandwich biosensor To evaluate the ability of the biosensor and the sandwich biosensor method to measure toxin levels in a complex matrix such as food, we measured SEA contamination in milk. Milk did not interfere with binding between the rabbit anti-SEA antibody on the
cuvette and SEA in the milk. The response measured was proportional to the amount of toxin in the sample. The addition of a second antibody (in sandwich biosensor) increased the response and the sensitivity to 1 ng (Fig. 3). Thus, both methods can detect pure toxin as well toxin in a more complex matrix, such as milk.
3.4. Effect of toxin heat inactivation on biosensor detection of SEA SEA is affected by prolonged heat treatment. In order to evaluate the effect of heat treatment of SEA on biosensor detection, SEA was autoclaved (1218C for 20 min). Using a cell proliferation assay, we have determined that autoclaving for 20 min affects SEA. The response of the biosensor to heat-treated toxin, based on binding to a cuvette coated with anti-SEA antibody, is presented in Fig. 4; autoclaving totally abolished the biosensor response to SEA. Although this assay does not measure SEA activity directly, the lack of response to heat treated SEA suggests that this measurement may indicate SEA heat inactivation.
L. Rasooly, A. Rasooly / International Journal of Food Microbiology 49 (1999) 119 – 127
123
Fig. 2. Sandwich biosensor measurement of purified SEA at different concentrations. (A) Response peaks to decreasing levels of purified SEA followed by a second antibody: (a) Response to SEA alone (b) response after second antibody application. 151mg, 25100 ng, 3510 ng, 451ng, 550.1ng 650.01 ng. (B) Effect of enhancement of the response by a second antibody to SEA, standard curve of increasing SEA concentrations.
124
L. Rasooly, A. Rasooly / International Journal of Food Microbiology 49 (1999) 119 – 127
3.5. Detection of SEA in food In order to assess the ability of the assay to measure SEA in foods we tested four food products spiked with different levels of SEA. Each food item was homogenized, diluted and spiked with various levels of SEA. The suspensions were used for the biosensor detection assay with a single antibody. Although this method is less sensitive than the sandwich biosensor (Fig. 2), it is faster and more appropriate for real time detection. Upon the addition of 10 ng SEA to the food suspensions, a signal was measured in all foods. The signal was proportional to the amount of toxin added. In addition, autoclaving the food suspensions containing SEA abolished the biosensor response (Fig. 5). Fig. 3. Biosensor and sandwich biosensor detection of SEA in milk. Milk was spiked with increasing concentrations of SEA and applied onto the anti-SEA antibody coated cuvette. Milk samples were washed following 10 min incubation and the remaining binding was recorded.
Fig. 4. Effect of autoclaving on biosensor detection of SEA. Purified SEA was either directly applied or autoclaved for 20 min and then applied onto the anti-SEA coated cuvette.
4. Discussion We evaluated the use of evanescent wave biosensor for detection of SEA in food using a single binding antibody or sandwich biosensor employing two antibodies. The sandwich method uses a second antibody to amplify the signal. As shown here, the sandwich biosensor method increased the level of detection by increasing the response of the biosensor (Fig. 2). In addition, use of different primary and secondary antibodies could minimize false positives resulting from non-specific binding to the primary antibody. The only drawback of the sandwich method is that it adds a step, slowing the analysis and making real time monitoring more complicated. The sensitivity of the biosensor assay (like any other immunological method as ELISA and Western blotting), depends mainly on the antibodies used. Thus, our results do not indicate the absolute level of biosensor detection but the measurements reported here reflect, in part, the affinity of specific antibodies for the ligand. The sensitivity of our sandwich biosensor assay, i.e. 1–10 ng of total protein, using 100-ml cuvette is 10–100 ng / ml depending on the material tested, is lower than that of ELISA (generally in the |1 ng / g range depending on the material tested). Some of the lower sensitivity can be attributed to the antibodies used in this study, which are not the same as those used by commercial kits.
L. Rasooly, A. Rasooly / International Journal of Food Microbiology 49 (1999) 119 – 127
125
Fig. 5. Biosensor detection of SEA in four different foods and effect of autoclaving. Hot dogs, milk, mushrooms and potato salad were prepared as described in Section 2 and spiked with different levels of SEA. Food samples were either applied directly, or autoclaved for 20 min and then applied onto a cuvette coated with rabbit anti-SEA. Samples were incubated 10 min, washed and the remaining binding level recorded.
When developing a specific assay for a specific set of materials, use of alternative or multiple antibodies will affect the sensitivity of the method, and perhaps increase its detection power. Another parameter affecting the level of sensitivity is that the instrument used lacked a second reference cell. While binding conditions are tightly controlled, a second channel used as a reference cell (found in later models of the instrument) might decrease background, thus increasing sensitivity. Nevertheless, the sensitivity of the biosensor detection is in the general range of other immunological methods.
The only method with significantly higher sensitivity (at pg level) is the T-cell proliferation assay (Rasooly et al., 1997) which measures the ability of SEA to act as a superantigen. However, the assay is not specific for SEA; any superantigen in the sample will induce T-cell proliferation. Therefore, in order for this assay to be applied to food, the superantigen in the sample must be identified subsequently by Western immunoblot, ELISA or a biosensor. The assay detected only native SEA but not heat denatured, inactive SEA. Although this assay does not measure biological activity of the toxin, the loss
126
L. Rasooly, A. Rasooly / International Journal of Food Microbiology 49 (1999) 119 – 127
of immunogenic recognition after heat treatment may be a useful indicator loss of biological activity. The main advantage of biosensor detection of SEA (or other toxins) over other immunological methods is the speed of the assay. ELISA and Western immunoblots take an entire day and the T-cell proliferation takes 3 days to complete. The fiberoptic biosensor method (Tempelman et al., 1996) used for SEB detection in food (ham) is very sensitive, however, it requires secondary fluorescent labeled antibodies and it is not interactive. Evanescent wave biosensor analysis is suitable for continuous monitoring during food processing. Continuous monitoring allows food manufacturers to respond immediately to contamination. Potential end users of this general technology include food manufacturing plants, central field laboratories, and regulatory agencies. In addition, the technology may be applicable to the identification of other of toxic materials in the environment.
Acknowledgements Thanks to Dr. N. Rose for his support of this project. This work was funded in part by the FDA Food Safety Initiative funds and grant No. 9702102 from the USDA. LR is very grateful to the Affinity Sensors company (Paramus, NJ, USA) for the use of the biosensor.
References Archer, D.L., Young, F.E., 1988. Contemporary issues: diseases with a food vector. Clin. Microbiol. Rev. 1, 377–398. Bergdoll, M.S., 1991. Symposium on microbiology update: old friends and new enemies. Staphylococcus aureus. J. Assoc. Off. Anal. Chem. 74, 706–710. Bergdoll, M.S., 1995. Importance of staphylococci that produce nanogram quantities of enterotoxin [editorial]. Int. J. Med. Microbiol. Virol. Parasitol. Infect. Dis. 282, 1–6. Betley, M.J., Lofdahl, S., Kreiswirth, B.N., Bergdoll, M.S., Novick, R.P., 1984. Staphylococcal enterotoxin A gene is associated with a variable genetic element. Proc. Natl. Acad. Sci. USA 81, 5179–5183. Betley, M.J., Mekalanos, J.J., 1985. Staphylococcal enterotoxin A is encoded by phage. Science 229, 185–187. Betley, M.J., Mekalanos, J.J., 1988. Nucleotide sequence of the type A staphylococcal enterotoxin gene. J. Bacteriol. 170, 34–41.
Betley, M.J., Miller, V.L., Mekalanos, J.J., 1986. Genetics of bacterial enterotoxins. Annu. Rev. Microbiol. 40, 577–605. Brigham-Burke, M., Edwards, J.R., O’Shannesy, D.J., 1992. Detection of receptor–ligand interactions using surface plasmon resonance: model studies employing the HIV-1 gp120 / CD4 interaction. Anal. Biochem. 205, 125–131. Denny, C.B., Humber, J.Y., Bohrer, C.W., 1971. Effect of toxin concentration on the heat inactivation of staphylococcal enterotoxin A in beef bouillon and in phosphate buffer. Appl. Microbiol. 21, 1064–1066. Evenson, M.L., Hinds, M.W., Bernstein, R.S., Bergdoll, M.S., 1988. Estimation of human dose of staphylococcal enterotoxin A from a large outbreak of staphylococcal food poisoning involving chocolate milk. Int. J. Food Microbiol. 7, 311–316. Fagerstam, L.G., Frostell, A., Karlsson, R., Kullman, M., Larsson, A., Malmqvist, M., Butt, H., 1990. Detection of antigen– antibody interactions by surface plasmon resonance. Application to epitope mapping. J. Mol. Recognit. 3, 208–214. Freed, R.C., Evenson, M.L., Reiser, R.F., Bergdoll, M.S., 1982. Enzyme-linked immunosorbent assay for detection of staphylococcal enterotoxins in foods. Appl. Environ. Microbiol. 44, 1349–1355. Garthright, W.E., Archer, D.L., Kvenberg, J.E., 1988. Estimates of incidence and costs of intestinal infectious diseases in the United States. Public Health Rep. 103, 107–115. Humber, J.Y., Denny, C.B., Bohrer, C.W., 1975. Influence of pH on the heat inactivation of staphylococcal enterotoxin A as determined by monkey feeding and serological assay. Appl. Microbiol. 30, 755–758. Khilko, S.N., Corr, M., Boyd, L.F., Lees, A., Inman, J.K., Margulies, D.H., 1993. Direct detection of major histocompatibility complex class I binding to antigenic peptides using surface plasmon resonance. Peptide immobilization and characterization of binding specificity. J. Biol. Chem. 268, 15425– 15434. Khilko, S.N., Jelonek, M.T., Corr, M., Boyd, L.F., Bothwell, A.L., Margulies, D.H., 1995. Measuring interactions of MHC class I molecules using surface plasmon resonance. J. Immunol. Methods 183, 77–94. Lee, I.C., Stevenson, K.E., Harmon, L.G., 1977. Effect of beef broth protein on the thermal inactivation of staphylococcal enterotoxin B1. Appl. Environ. Microbiol. 33, 341–344. Malmqvist, M., 1993. Surface plasmon resonance for detection and measurement of antibody– antigen affinity and kinetics. Curr. Opin. Immunol. 5, 282–286. Malmqvist, M., Karlsson, R., 1997. Biomolecular interaction analysis: affinity biosensor technologies for functional analysis of proteins. Curr. Opin. Chem. Biol. 3, 378–383. Miller, B.A., Reiser, R.F., Bergdoll, M.S., 1978. Detection of staphylococcal enterotoxins A, B, C, D, and E in foods by radioimmunoassay, using staphylococcal cells containing protein A as immunoadsorbent. Appl. Environ. Microbiol. 36, 421–426. Olson, J.D., Panfili, P.R., Armenta, R., Femmel, M.B., Merrick, H., Gumperz, J., Goltz, M., Zuk, R.F., 1990. A silicon sensorbased filtration immunoassay using biotin-mediated capture. J. Immunol. Methods 134, 71–79. Park, C.E., Warburton, D., Laffey, P.J., 1996. A collaborative study on the detection of staphylococcal enterotoxins in foods
L. Rasooly, A. Rasooly / International Journal of Food Microbiology 49 (1999) 119 – 127 by an enzyme immunoassay kit RIDASCREEN. Int. J. Food Microbiol. 29, 281–295. Rasooly, L., Rose, N.R., Shah, D.B., Rasooly, A., 1997. In vitro assay of Staphylococcus aureus enterotoxin A activity in food. Appl. Environ. Microbiol. 63, 2361–2365. Seth, A., Stern, L.J., Ottenhoff, T.H., Engel, I., Owen, M.J., Lamb, J.R., Klausner, R.D., Wiley, D.C., 1994. Binary and ternary complexes between T-cell receptor, class II MHC and superantigen in vitro. Nature 369, 324–327. Singh, B.R., Kokan-Moore, N.P., Bergdoll, M.S., 1988. Molecular topography of toxic shock syndrome toxin 1 as revealed by spectroscopic studies. Biochemistry 27, 8730–8735. Tan, P.L., Denny, C.B., Heilman, W.R., Bohrer, C.W., 1969. New
127
technique for rapid purification, entrapment, and recovery of enterotoxin A from a liquid chamber by polyacrylamide gel electrophoresis. Appl. Microbiol. 18, 1089–1090. Tempelman, L.A., King, K.D., Anderson, G.P., Ligler, F.S., 1996. Quantitating staphylococcal enterotoxin B in diverse media using a portable fiber-optic biosensor. Anal. Biochem. 233, 50–57. Thompson, N.E., Razdan, M., Kuntsmann, G., Aschenbach, J.M., Evenson, M.L., Bergdoll, M.S., 1986. Detection of staphylococcal enterotoxins by enzyme-linked immunosorbent assays and radioimmunoassays: comparison of monoclonal and polyclonal antibody systems. Appl. Environ. Microbiol. 51, 885– 890.
Real time biosensor analysis of Staphylococcal enterotoxin A in food a,b
Linda Rasooly , Avraham Rasooly
b,c ,
*
a
Department of Molecular Microbiology and Immunology, Johns Hopkins University, Baltimore, MD, USA b Department of Pathology, Johns Hopkins University, Baltimore, MD, USA c US Food and Drug Administration, Division of Microbiological Studies, 200 C St. SW, HFS 515, Washington, DC 20204, USA Accepted 14 April 1999
Abstract Currently there is no ‘real-time’ detection system to identify food borne toxins. In order to develop such a system, we have used a evanescent wave biosensor for real time detection of staphylococcal enterotoxin A (SEA) in foods. The approach used here is sandwich biosensor, a method utilizing two antibodies. The toxin binds initially to a capturing antibody which is bound covalently on the surface of the biosensor detector. The second antibody binds to the captured toxin. We were able to measure SEA in foods with little or no background interference, demonstrating that biosensor-based measurement of SEA was possible not only with purified SEA but also in complex food matrices such as hot dogs, potato salad, milk and mushrooms. Autoclaved samples of SEA did not evoke a positive response. With both purified SEA and SEA-spiked foods, the assay sensitivity is 10–100 ng / g depending on the material tested and the assay is rapid ( , 4 min) when a single antibody is used. 1999 Elsevier Science B.V. All rights reserved. Keywords: Biosensor; Foodborne toxins; Staphylococcal enterotoxins
1. Introduction Staphylococcal enterotoxins (SE) are a family of five major serological types (SEA through SEE) of heat stable (Tan et al., 1969; Denny et al., 1971; Humber et al., 1975; Lee et al., 1977) emetic enterotoxins. They are encoded by five genes that
*Corresponding author. Tel.: 11-202-205-4192; fax: 11-202401-7740. E-mail address: [email protected] (A. Rasooly)
share 50 to 85% homology at the predicted amino acid level (Betley et al., 1984; Betley and Mekalanos, 1985; Betley et al., 1986; Betley and Mekalanos, 1988; Singh et al., 1988). Staphylococcal enterotoxin A (SEA), a monomeric protein (Mr 27 000), is a heat stable, extremely potent gastrointestinal toxin (Archer and Young, 1988; Bergdoll, 1991; Bergdoll, 1995) and is one of the leading causes of gastroenteritis (vomiting and diarrhea) resulting from consumption of contaminated food (Archer and Young, 1988; Garthright et al., 1988). Its potency requires that methods to detect it in foods
0168-1605 / 99 / $ – see front matter 1999 Elsevier Science B.V. All rights reserved. PII: S0168-1605( 99 )00053-7
120
L. Rasooly, A. Rasooly / International Journal of Food Microbiology 49 (1999) 119 – 127
be sensitive enough to measure very low levels (nanograms per gram food) (Evenson et al., 1988). Immunological testing (especially ELISA), has been the method of choice for SEA detection since antibodies to the toxin are readily available (Miller et al., 1978; Freed et al., 1982; Thompson et al., 1986; Bergdoll, 1991; Park et al., 1996). Real time detection allows testers to obtain immediate interactive information on the sample. This is important in the diagnostics and detection of food borne toxins, especially during the food production process. A food contamination alert might allow the processor to take immediate corrective measures. Real time detection would also increase the efficiency of food quality testing since the current methodology takes 4–8 h to complete and is not interactive. Biosensors can be used as analytical devices to translate a biological response (e.g. presence of toxin) into an electronic output that can be analyzed. The most common biosensors are based on evanescent wave. These biosensors analyze interactions using evanescent waves, an electromagnetic wave that exists at the surface of many forms of optical waveguides, to measure changes in refractive index at the sensor surface. These changes can be measured as shifts in resonance angle as in the IAsys instrument used in this study. Other common biosensors are based on surface plasmon resonance (SPR). One of the most common biosensors is based on surface plasmon resonance (SPR) (Malmqvist and Karlsson, 1997) SPR is a quantum optical–electrical phenomenon based on the fact that energy carried by photons of light can be ‘coupled’ or transferred to electrons in a metal. Changes in the chemical properties of the environment within the range of the plasmon field (e.g. antibody–antigen binding) causes a change in the plasmon, which can be measured as change in the refraction index, a shift in the wavelength of light absorbed, or by light diffraction. In a biosensor, a specimen is tested for its adsorption to a covalently immobilized molecule (binding agent) by surface sensitive optical techniques (Malmqvist, 1993). The amount of adsorption is measured as a function of time, and results are generated in the form of a sensogram that shows the response units measured as a consequence of the adsorption. Biosensors can be integrated with on-line process monitoring schemes to detect and identify food contaminants and provide real-time information
about toxin presence at every step or time point in a process. Various binding molecules have been developed and utilized for the measuring compounds in clinical, medical, environmental and industrial process settings. Frequently, the biosensors use antibodies (Fagerstam et al., 1990; Malmqvist, 1993) because of their specificity. Other components of the immune system have also been used in biosensors including immunoglobulins, MHC class I (major histocompatibility complex molecules), MHC class II molecules, T-cell receptor, and receptor–ligand binding reactions (Khilko et al., 1993; Seth et al., 1994; Khilko et al., 1995). The sensitivity of biosensor techniques is on the order of 10 212 M, with a total assay time of several minutes (Olson et al., 1990). The kinetics of the binding between Staphylococcal enterotoxin B (SEB) and the T cell receptor (HA1.7, Vb3.1) have been studied using surface plasmon resonance (Olson et al., 1990; Seth et al., 1994). Although the technique was highly sensitive, it was not possible to measure SEA using the T-cell receptor because there the T-cell receptor does not bind SEA in the absence of MHC class II molecules (Seth et al., 1994). Biosensors may be very useful in environmental and food safety analysis. They do not require labeling of either molecule, nor does the analyte need to be purified as the background binding is usually low (Brigham-Burke et al., 1992). Recently, an effective, portable, fiber-optic biosensor method was used for SEB detection in food. This biosensor is very sensitive and detects |10 ng toxin / g various foods, using fluorescent-labeled antibodies (Tempelman et al., 1996). We have used a biosensor to measure SEA. The ligand, rabbit anti-SEA antibody, was covalently crosslinked onto an activated surface and reacted with different concentrations of SEA in solution. The amount of SEA binding to the antibody-coated surface was measured by shifts in resonance angle.
2. Materials and methods
2.1. Instrumentation The IAsys (Affinity Sensors, Paramus, NJ, USA) evanescent wave biosensor applied in this study was used and calibrated according to the manufacturer’s
L. Rasooly, A. Rasooly / International Journal of Food Microbiology 49 (1999) 119 – 127
instructions. The reproducibility of the instrument is 610% depending on the concentration material tested.
2.2. Immobilization of ligand Chemicals were obtained from Sigma (St. Louis, MO, USA) unless specified. Carboxymethyl-dextran (CMD) cuvettes (Affinity Sensors) were washed three times with PBS–Tween (10 mM sodium phosphate, 2.7 mM KCl, 138 mM NaCl, 0.05% Tween-20). A 1:1 (v / v) mixture of N-hydroxysuccinimide (NHS; 1.3% in H 2 O) and 1-ethyl-3-(3dimethylaminopropyl) carbodiimide (EDC) (7.6% in H 2 0) was prepared immediately prior to use. The mixture was applied to the cuvettes (activation step) for 7 min. Cuvettes were washed in PBS–Tween and immobilization buffer (10 mM sodium acetate, pH 5.0) was added to the cuvettes for 2 min. The ligand (5 ml of rabbit anti-SEA affinity purified IgG; Toxin Technology, Sarasota, FL, USA) was added into the immobilization buffer and left for 7 min to bind onto the activated surface. The cuvettes were washed in PBS–Tween and blocked with 1 M ethanolamine (pH 8.5) for 3 min, and then washed in PBS–Tween, followed by washing in regeneration buffer (10 mM HCl) for 2 min. After coating, cuvettes were stored with PBS1azide (0.05%) at 48C until used.
2.3. Preparation of food samples A 10-g amount of potato salad, canned mushrooms or hot dogs purchased from local stores was homogenized with 10 ml PBS. The suspensions were centrifuged at 1000 g for 10 min and supernatant was collected. Milk was prepared as 5% w / v of dry milk (Carnation, Glendale, CA, USA) in PBS. SEA (Sigma) was diluted with sterile distilled water to 1 mg / ml. Aliquots were stored at 2208C. The toxin was added to the homogenized sample before centrifugation. Food specimens were tested either directly, with different concentrations of SEA, or after autoclaving at 1218C for 20 min.
2.4. Testing of food samples A 100-ml aliquot of either a pure SEA solution (Sigma) or a food suspension spiked with SEA was applied to a coated cuvette for 10 min. The cuvette
121
was washed and the remaining binding activity was recorded. For sandwich biosensor detection, 200 ml PBS–Tween was added to the cuvette followed by 1 ml of rabbit anti-SEA affinity purified antibody (Toxin Technology). After a 10-min incubation, the cuvette was washed four times in PBS–Tween and the response was recorded.
3. Results
3.1. Measurements of SEA by sandwich biosensor In the first experiment our goal was to study the binding between SEA and a rabbit anti-SEA antibody bound to the IAsys biosensor cuvette. Our results are reported in terms of the absolute amount of SEA added to the cuvette. The total volume of the cuvette is 100 ml. Upon application of SEA (1 mg) to the cuvette, a strong response peak appeared after less than 1 min and reached equilibrium within 10 min (Fig. 1A,b). Fig. 1B demonstrates the kinetics of the binding. Significant binding (|90%) was complete within 4 min. This appears to be the minimum time required for measuring without significantly sacrificing sensitivity. However, longer incubation times did produce a higher signal. The cuvette was washed to eliminate nonspecific binding (Fig. 1A,c). Following the addition of a second antibody, a new peak appeared reflecting the binding between captured SEA (bound to the immobilized antibody in the cuvette) and the second antibody in solution (Fig. 1A,d). The response remained high even after washing (Fig. 1A,e), showing that SEA–second antibody interaction is specific. We compared the level of detection of SEA by the primary antibody alone to the sandwich biosensor method using serial dilutions of SEA (Fig. 2). As shown in Fig. 2B, the addition of a second antibody generally increased the level of response at all detectable SEA levels. The level of SEA detection of is ¯10 ng with the sandwich biosensor method. Like any other immunological method, the sensitivity of the biosensor assay depends heavily on the antibodies used. Therefore, the results of our studies do not indicate the absolute level of sensitivity of this method. Instead, the measurements reported here reflect, in part, the affinity of specific antibodies for the ligand.
122
L. Rasooly, A. Rasooly / International Journal of Food Microbiology 49 (1999) 119 – 127
Fig. 1. Sandwich biosensor detection of SEA. 1 mg of purified SEA was added to a cuvette coated with anti-SEA rabbit antibody and the response was measured. After 10 min, a second anti-SEA antibody was applied and the response was measured. (A) The interaction between SEA and the two antibodies: (a) Baseline of the buffer; (b) Reaction with SEA; (c) Residual binding of SEA after washing; (d) Reaction following application of a second anti-SEA antibody; (e) Residual binding of second antibody after washing; (f) Return to baseline after acid wash. (B) Detailed first 5 min of the interaction between SEA and the antibody coated cuvette: (a) Buffer baseline; (b) Reaction with 10 ng SEA.
3.2. Statistical analysis To analyze the linearity of the assay, curve fitting analysis ( SLIDEWRITE PLUS software) was performed on the biosensor data. The linear regression between the amount of toxin and the signal was very high, with and r 2 value of 0.92 and 0.94 for one and two antibodies, respectively. The best fit curve is a third order polynomial curve with r 2 of 0.99 for one or two antibodies. The equations for these curves are y 5 7.38 1 23.41x 1 8.51x 2 2 0.97x 3 for one antibody and y 5 32.61 1 60.27x 1 18.27x 2 2 2.36x 3 . These results suggest that there is very high, nonlinear, correlation between the amount of SEA and the biosensor signal.
3.3. Measurement of SEA in milk by sandwich biosensor To evaluate the ability of the biosensor and the sandwich biosensor method to measure toxin levels in a complex matrix such as food, we measured SEA contamination in milk. Milk did not interfere with binding between the rabbit anti-SEA antibody on the
cuvette and SEA in the milk. The response measured was proportional to the amount of toxin in the sample. The addition of a second antibody (in sandwich biosensor) increased the response and the sensitivity to 1 ng (Fig. 3). Thus, both methods can detect pure toxin as well toxin in a more complex matrix, such as milk.
3.4. Effect of toxin heat inactivation on biosensor detection of SEA SEA is affected by prolonged heat treatment. In order to evaluate the effect of heat treatment of SEA on biosensor detection, SEA was autoclaved (1218C for 20 min). Using a cell proliferation assay, we have determined that autoclaving for 20 min affects SEA. The response of the biosensor to heat-treated toxin, based on binding to a cuvette coated with anti-SEA antibody, is presented in Fig. 4; autoclaving totally abolished the biosensor response to SEA. Although this assay does not measure SEA activity directly, the lack of response to heat treated SEA suggests that this measurement may indicate SEA heat inactivation.
L. Rasooly, A. Rasooly / International Journal of Food Microbiology 49 (1999) 119 – 127
123
Fig. 2. Sandwich biosensor measurement of purified SEA at different concentrations. (A) Response peaks to decreasing levels of purified SEA followed by a second antibody: (a) Response to SEA alone (b) response after second antibody application. 151mg, 25100 ng, 3510 ng, 451ng, 550.1ng 650.01 ng. (B) Effect of enhancement of the response by a second antibody to SEA, standard curve of increasing SEA concentrations.
124
L. Rasooly, A. Rasooly / International Journal of Food Microbiology 49 (1999) 119 – 127
3.5. Detection of SEA in food In order to assess the ability of the assay to measure SEA in foods we tested four food products spiked with different levels of SEA. Each food item was homogenized, diluted and spiked with various levels of SEA. The suspensions were used for the biosensor detection assay with a single antibody. Although this method is less sensitive than the sandwich biosensor (Fig. 2), it is faster and more appropriate for real time detection. Upon the addition of 10 ng SEA to the food suspensions, a signal was measured in all foods. The signal was proportional to the amount of toxin added. In addition, autoclaving the food suspensions containing SEA abolished the biosensor response (Fig. 5). Fig. 3. Biosensor and sandwich biosensor detection of SEA in milk. Milk was spiked with increasing concentrations of SEA and applied onto the anti-SEA antibody coated cuvette. Milk samples were washed following 10 min incubation and the remaining binding was recorded.
Fig. 4. Effect of autoclaving on biosensor detection of SEA. Purified SEA was either directly applied or autoclaved for 20 min and then applied onto the anti-SEA coated cuvette.
4. Discussion We evaluated the use of evanescent wave biosensor for detection of SEA in food using a single binding antibody or sandwich biosensor employing two antibodies. The sandwich method uses a second antibody to amplify the signal. As shown here, the sandwich biosensor method increased the level of detection by increasing the response of the biosensor (Fig. 2). In addition, use of different primary and secondary antibodies could minimize false positives resulting from non-specific binding to the primary antibody. The only drawback of the sandwich method is that it adds a step, slowing the analysis and making real time monitoring more complicated. The sensitivity of the biosensor assay (like any other immunological method as ELISA and Western blotting), depends mainly on the antibodies used. Thus, our results do not indicate the absolute level of biosensor detection but the measurements reported here reflect, in part, the affinity of specific antibodies for the ligand. The sensitivity of our sandwich biosensor assay, i.e. 1–10 ng of total protein, using 100-ml cuvette is 10–100 ng / ml depending on the material tested, is lower than that of ELISA (generally in the |1 ng / g range depending on the material tested). Some of the lower sensitivity can be attributed to the antibodies used in this study, which are not the same as those used by commercial kits.
L. Rasooly, A. Rasooly / International Journal of Food Microbiology 49 (1999) 119 – 127
125
Fig. 5. Biosensor detection of SEA in four different foods and effect of autoclaving. Hot dogs, milk, mushrooms and potato salad were prepared as described in Section 2 and spiked with different levels of SEA. Food samples were either applied directly, or autoclaved for 20 min and then applied onto a cuvette coated with rabbit anti-SEA. Samples were incubated 10 min, washed and the remaining binding level recorded.
When developing a specific assay for a specific set of materials, use of alternative or multiple antibodies will affect the sensitivity of the method, and perhaps increase its detection power. Another parameter affecting the level of sensitivity is that the instrument used lacked a second reference cell. While binding conditions are tightly controlled, a second channel used as a reference cell (found in later models of the instrument) might decrease background, thus increasing sensitivity. Nevertheless, the sensitivity of the biosensor detection is in the general range of other immunological methods.
The only method with significantly higher sensitivity (at pg level) is the T-cell proliferation assay (Rasooly et al., 1997) which measures the ability of SEA to act as a superantigen. However, the assay is not specific for SEA; any superantigen in the sample will induce T-cell proliferation. Therefore, in order for this assay to be applied to food, the superantigen in the sample must be identified subsequently by Western immunoblot, ELISA or a biosensor. The assay detected only native SEA but not heat denatured, inactive SEA. Although this assay does not measure biological activity of the toxin, the loss
126
L. Rasooly, A. Rasooly / International Journal of Food Microbiology 49 (1999) 119 – 127
of immunogenic recognition after heat treatment may be a useful indicator loss of biological activity. The main advantage of biosensor detection of SEA (or other toxins) over other immunological methods is the speed of the assay. ELISA and Western immunoblots take an entire day and the T-cell proliferation takes 3 days to complete. The fiberoptic biosensor method (Tempelman et al., 1996) used for SEB detection in food (ham) is very sensitive, however, it requires secondary fluorescent labeled antibodies and it is not interactive. Evanescent wave biosensor analysis is suitable for continuous monitoring during food processing. Continuous monitoring allows food manufacturers to respond immediately to contamination. Potential end users of this general technology include food manufacturing plants, central field laboratories, and regulatory agencies. In addition, the technology may be applicable to the identification of other of toxic materials in the environment.
Acknowledgements Thanks to Dr. N. Rose for his support of this project. This work was funded in part by the FDA Food Safety Initiative funds and grant No. 9702102 from the USDA. LR is very grateful to the Affinity Sensors company (Paramus, NJ, USA) for the use of the biosensor.
References Archer, D.L., Young, F.E., 1988. Contemporary issues: diseases with a food vector. Clin. Microbiol. Rev. 1, 377–398. Bergdoll, M.S., 1991. Symposium on microbiology update: old friends and new enemies. Staphylococcus aureus. J. Assoc. Off. Anal. Chem. 74, 706–710. Bergdoll, M.S., 1995. Importance of staphylococci that produce nanogram quantities of enterotoxin [editorial]. Int. J. Med. Microbiol. Virol. Parasitol. Infect. Dis. 282, 1–6. Betley, M.J., Lofdahl, S., Kreiswirth, B.N., Bergdoll, M.S., Novick, R.P., 1984. Staphylococcal enterotoxin A gene is associated with a variable genetic element. Proc. Natl. Acad. Sci. USA 81, 5179–5183. Betley, M.J., Mekalanos, J.J., 1985. Staphylococcal enterotoxin A is encoded by phage. Science 229, 185–187. Betley, M.J., Mekalanos, J.J., 1988. Nucleotide sequence of the type A staphylococcal enterotoxin gene. J. Bacteriol. 170, 34–41.
Betley, M.J., Miller, V.L., Mekalanos, J.J., 1986. Genetics of bacterial enterotoxins. Annu. Rev. Microbiol. 40, 577–605. Brigham-Burke, M., Edwards, J.R., O’Shannesy, D.J., 1992. Detection of receptor–ligand interactions using surface plasmon resonance: model studies employing the HIV-1 gp120 / CD4 interaction. Anal. Biochem. 205, 125–131. Denny, C.B., Humber, J.Y., Bohrer, C.W., 1971. Effect of toxin concentration on the heat inactivation of staphylococcal enterotoxin A in beef bouillon and in phosphate buffer. Appl. Microbiol. 21, 1064–1066. Evenson, M.L., Hinds, M.W., Bernstein, R.S., Bergdoll, M.S., 1988. Estimation of human dose of staphylococcal enterotoxin A from a large outbreak of staphylococcal food poisoning involving chocolate milk. Int. J. Food Microbiol. 7, 311–316. Fagerstam, L.G., Frostell, A., Karlsson, R., Kullman, M., Larsson, A., Malmqvist, M., Butt, H., 1990. Detection of antigen– antibody interactions by surface plasmon resonance. Application to epitope mapping. J. Mol. Recognit. 3, 208–214. Freed, R.C., Evenson, M.L., Reiser, R.F., Bergdoll, M.S., 1982. Enzyme-linked immunosorbent assay for detection of staphylococcal enterotoxins in foods. Appl. Environ. Microbiol. 44, 1349–1355. Garthright, W.E., Archer, D.L., Kvenberg, J.E., 1988. Estimates of incidence and costs of intestinal infectious diseases in the United States. Public Health Rep. 103, 107–115. Humber, J.Y., Denny, C.B., Bohrer, C.W., 1975. Influence of pH on the heat inactivation of staphylococcal enterotoxin A as determined by monkey feeding and serological assay. Appl. Microbiol. 30, 755–758. Khilko, S.N., Corr, M., Boyd, L.F., Lees, A., Inman, J.K., Margulies, D.H., 1993. Direct detection of major histocompatibility complex class I binding to antigenic peptides using surface plasmon resonance. Peptide immobilization and characterization of binding specificity. J. Biol. Chem. 268, 15425– 15434. Khilko, S.N., Jelonek, M.T., Corr, M., Boyd, L.F., Bothwell, A.L., Margulies, D.H., 1995. Measuring interactions of MHC class I molecules using surface plasmon resonance. J. Immunol. Methods 183, 77–94. Lee, I.C., Stevenson, K.E., Harmon, L.G., 1977. Effect of beef broth protein on the thermal inactivation of staphylococcal enterotoxin B1. Appl. Environ. Microbiol. 33, 341–344. Malmqvist, M., 1993. Surface plasmon resonance for detection and measurement of antibody– antigen affinity and kinetics. Curr. Opin. Immunol. 5, 282–286. Malmqvist, M., Karlsson, R., 1997. Biomolecular interaction analysis: affinity biosensor technologies for functional analysis of proteins. Curr. Opin. Chem. Biol. 3, 378–383. Miller, B.A., Reiser, R.F., Bergdoll, M.S., 1978. Detection of staphylococcal enterotoxins A, B, C, D, and E in foods by radioimmunoassay, using staphylococcal cells containing protein A as immunoadsorbent. Appl. Environ. Microbiol. 36, 421–426. Olson, J.D., Panfili, P.R., Armenta, R., Femmel, M.B., Merrick, H., Gumperz, J., Goltz, M., Zuk, R.F., 1990. A silicon sensorbased filtration immunoassay using biotin-mediated capture. J. Immunol. Methods 134, 71–79. Park, C.E., Warburton, D., Laffey, P.J., 1996. A collaborative study on the detection of staphylococcal enterotoxins in foods
L. Rasooly, A. Rasooly / International Journal of Food Microbiology 49 (1999) 119 – 127 by an enzyme immunoassay kit RIDASCREEN. Int. J. Food Microbiol. 29, 281–295. Rasooly, L., Rose, N.R., Shah, D.B., Rasooly, A., 1997. In vitro assay of Staphylococcus aureus enterotoxin A activity in food. Appl. Environ. Microbiol. 63, 2361–2365. Seth, A., Stern, L.J., Ottenhoff, T.H., Engel, I., Owen, M.J., Lamb, J.R., Klausner, R.D., Wiley, D.C., 1994. Binary and ternary complexes between T-cell receptor, class II MHC and superantigen in vitro. Nature 369, 324–327. Singh, B.R., Kokan-Moore, N.P., Bergdoll, M.S., 1988. Molecular topography of toxic shock syndrome toxin 1 as revealed by spectroscopic studies. Biochemistry 27, 8730–8735. Tan, P.L., Denny, C.B., Heilman, W.R., Bohrer, C.W., 1969. New
127
technique for rapid purification, entrapment, and recovery of enterotoxin A from a liquid chamber by polyacrylamide gel electrophoresis. Appl. Microbiol. 18, 1089–1090. Tempelman, L.A., King, K.D., Anderson, G.P., Ligler, F.S., 1996. Quantitating staphylococcal enterotoxin B in diverse media using a portable fiber-optic biosensor. Anal. Biochem. 233, 50–57. Thompson, N.E., Razdan, M., Kuntsmann, G., Aschenbach, J.M., Evenson, M.L., Bergdoll, M.S., 1986. Detection of staphylococcal enterotoxins by enzyme-linked immunosorbent assays and radioimmunoassays: comparison of monoclonal and polyclonal antibody systems. Appl. Environ. Microbiol. 51, 885– 890.