Anal. Chem. 2004, 76, 98-104
An Analytical Device for On-Site Immunoassay. Demonstration of Its Applicability in Semiquantitative Detection of Aflatoxin B1 in a Batch of Samples with Ultrahigh Sensitivity Arindam Pal and Tarun K. Dhar*
Department of Immunobiology, Indian Institute of Chemical Biology, 4, Raja S. C. Mullick Road, Jadavpur, Calcutta, 700 032, India
A simple analytical device has been developed for performing noninstrumental immunofiltration-based assay on a batch of samples. The device consists of membrane strips, with antibody-immobilized zones, attached to a polyethylene card. A moist filter paper placed between the membrane and the polyethylene card acts as the absorbent body. The device was used to estimate very low concentrations of aflatoxin B1 (AFB1) present in food samples by using an improved catalyzed reporter deposition (Super-CARD) method of signal amplification involving biotinylated tyramine (B-T) and avidin-horseradish peroxidase conjugate. 4-Chloro-1-naphthol was used as the substrate for visualization. Semiquantitative results are obtained by visual comparison of the color intensity (inversely related to the analyte concentration) of a sample spot with those of reference standards. Quantitative estimation is possible by densitometric analysis (detection limit 0.25 pg/spot, 0.01 ng mL-1). Dilute samples can be assayed by in situ concentration with improved doseresponse characteristics. A batch of 12 extracted samples can be analyzed in a single test card within 12 min. Spiked and contaminated samples of groundnut, corn, wheat, cheese, and chilli were analyzed without sample cleanup. The matrix interferences were eliminated by using appropriate dilution of the aqueous methanol extracts. Mean recoveries from different food samples were between 91 and 104%. The values obtained for infected corn and groundnut samples correlated well (R2 ) 0.99) with the estimates by HPLC. The method is well-suited for visual screening of agricultural and food samples for AFB1 under field conditions. There is an increasing demand for rapid and reliable methods for assay of mycotoxins and other environmental pollutants under field conditions. Nonisotopic immunoassays, such as enzymelinked immunosorbent assays (ELISA), offer the required sensitivity and specificity for such assays and are widely used for analysis of batch of samples. However, most of these methods require wellequipped laboratories, trained personnel, and several hours to complete a test. During the past few years, numerous membrane* To whom correspondence should be addressed. Phone: +91-33-2473-3493/ 197. Fax: +91-33-2473-0284 /5197. E-mail:
[email protected].
98 Analytical Chemistry, Vol. 76, No. 1, January 1, 2004
based methods such as dip-sticks, immunochromatographic test strips and flow-through devices of various configurations and strategies have been developed for application at alternative testing sites, such as on farms or in storehouses and factories.1-4 However, most of these assays have low sensitivities and are not designed for analysis of a batch of samples at a time. Recently, substantial progress has also been made on development of biosensors in a variety of formats, including surface plasmon resonance, fiber optic probes, and microbead-based assays for onsite applications.5-11 In these assays also, only one sample can be assayed at a time. Limitations on the reuse of the immunosurface and higher reagent consumption are disadvantages of many of these methods.12 For rapid analysis of a batch of samples under field conditions, development of novel approaches are required to provide reliable semiquantitative results without instrumentation. This paper describes the development of a highly cost-effective analytical device for performing a noninstrumental immunofiltration-based assay on a batch of samples using aflatoxin B1 (AFB1) as a model analyte (patent pending). The easy to manufacture device consists of nitrocellulose membrane strips attached at one end to a polyethylene card. The membranes are marked into several zones, and each zone is used separately for immobilization of antibody. Thus, a number of immunochemical reactions may be performed simultaneously (Figure 1). The device differs from previously reported flow-through devices in several respects. The (1) De Saeger, S.; Van Peteghem. Appl. Envirn. Microbiol. 1996, 62, 18801884. (2) Sibanada, L.; De Saeger, S.; Van Peteghem, C.; Grabarkiewicz-Szczesna, J.; Tomczak, M. J. Agric. Food Chem. 2000, 48, 5864-5867. (3) Paek, S.; Lee, S.; Cho, J.; Kim, Y. Methods 2000, 22, 53-60. (4) Ho, J.-A.; A.; Wauchope, R. D. Anal. Chem. 2002, 74, 1493-1496. (5) Daly, S. J.; Keating, G. J.; Dillion, P. P.; Manning, B. M.; O’Kennedy, R.; Lee, H. A.; Morgan, M. R. A. J. Agirc. Food Chem. 2000, 48, 5097-5104. (6) Gonzalez-Martinez, M. A.; Morais, S.; Puchades, R.; Maquieira, A.; Abad, A.; Montoya, A. Anal. Chem. 1997, 69, 2812-2818. (7) Carlson, M. A.; Bargeron, C. B.; Benson, R. C.; Fraser, A. B.; Phillips, T. E.; Velky, J. T.; Groopman, J. D.; Strickland, P. T.; Ko, H. W. Biosens. Bioelectron. 2000, 14, 841-848. (8) Baumner, A. J.; Schmid, R. D. Biosens. Bioelectron. 1998, 13, 519-529. (9) Pulido-Tofino, P.; Barrero-Moreno, J. M.; Perez-Conde, M. C. Anal. Chim. Acta 2000, 417, 85-94. (10) Nasir, M. S.; Jolley, M. E. J. Agric. Food Chem. 2002, 50, 3116-3121. (11) Maragos, C. M. Adv. Exp. Med. Biol. 2002, 504, 85-93. (12) Gonzalez-Martinez, M. A.; Puchades, R.; Maquieira, A. Trends Anal. Chem. 1999, 18, 204-218. 10.1021/ac034694g CCC: $27.50
© 2004 American Chemical Society Published on Web 09/20/2003
Figure 1. Schematic diagram of the analytical device with four membrane strips (1-4) for on-site immunoassay. Each membrane strip contains four antibody-immobilized zones (a-d).
absorbent body is not fixed to the reaction membrane using compression or adhesive. It is provided separately and discarded after addition of the sample and enzyme conjugate. This permits the use of additional reagents for signal amplification to improve assay sensitivity. Second, the use of a wetted absorbent body keeps it in close contact with the membrane and facilitates the fast transfer of the added standards, samples or reagents to the absorbent body without lateral spreading. Third, the device allows semiquantitative analysis of a batch of samples in the presence of reference standards in a single test strip, and the number of membrane strips used can be varied depending on the number of samples to be assayed. The device was used for both conventional and ultrasensitive assays. In the conventional format, the assay can be performed within 5 min. However, the detection level is not sufficient to accurately measure AFB1 in foodstuffs. Moreover, concentration and cleanup of samples is required before analysis.13-16 To avoid the time-consuming sample cleanup step, we have used the improved catalyzed reporter deposition method of signal amplification, termed Super-CARD17,18 to increase the detection sensitivity. The method is similar to catalyzed reporter deposition (also referred to as “tyramide signal amplification”) method19 except for the use of synthetic electron-rich proteins containing multiple phenolic groups to block the vacant sites of the membrane. This markedly increases the deposition of B-T molecules by peroxidase activity. The net effect is high signal amplification and increased sensitivity compared to the tyramide signal amplification method. By this method, AFB1 could be detected with high sensitivity under field conditions within 12 min. EXPERIMENTAL SECTION Materials and Chemicals. Nitrocellulose membrane, 0.45 µm (Catalogue no. HAHY00010) was from Millipore Corporation, Bedford, MA. Filter paper (Whatmann no. 3) was from Whatman International Ltd. Maidstone, England. Semirigid polyethylene sheets, adhesive tape, and analytical grade buffer salts were (13) CHU, F. S. J. Anim. Sci. 1992, 70, 3950-3963. (14) Scott, P. M.; Kanhere, S. P.; Weber, D. Food Addit. Contam. 1993, 10, 381389. (15) Helrich, K. In Official Methods of Analysis, 15th ed.; AOAC: Gaithersburg, MD, 1990; Vol. 49, p 1184. (16) Daly, S. J.; Keating, G. J.; Dillon, P. P.; Manning, B. M.; O’Kennedy, R.; Lee, H. A.; Morgan, R. A. J. Agric. Food Chem. 2000, 48, 5097-5104. (17) Bhattacharya, D.; Bhattacharya, R.; Dhar, T. K. J. Immunol. Methods 1999, 230, 71-86. (18) Bhattacharya, R.; Bhattacharya, D.; Dhar, T. K. J. Immunol. Methods 1999, 227, 31-39. (19) Bobrow, M. N.; Shanghessy, K. J.; Litt, G. J. J. Immunol. Methods 1991, 137, 103-112.
purchased from a local market. Other chemicals were from Sigma, St. Louis, MO. AFB1-O-carboxymethyl oxime was synthesized according to a standard procedure.20 B-T and 3-(p-hydroxyphenyl) propionic acid-casein conjugate (p-OH-PPA-casein) were prepared according to the method described previously.17-18 Densitometry. Densitometric analysis was carried out with an image scanner (Amersham Pharmacia Biotech) using the instrument’s Magic Scan software (version 4.5) for image scanning and Image Master Total lab software (version 1.11) for quantitation. Preparation of Immunoassay Reagents. Polyclonal antibody was raised in New Zealand white rabbits using AFB1-O-carboxymethyloxime-BSA conjugate as immunogen. The antibody has already been characterized by ELISA17 and was found to be highly specific for AFB1. The relative cross-reactivity of aflatoxin B2, G1, and G2 were 13.5, 11.5, and 0.6%, respectively. For the present work, this antibody was purified by repeated precipitation with ammonium sulfate (50% saturation) followed by dialysis against phosphate buffered saline. It was passed through a BSA-Sepharose column to remove anti-BSA antibodies. The AFB1-horseradish peroxidase conjugate (AFB1-HRP) was synthesized by the NHS-ester method21 and purified by extensive dialysis against phosphate buffer (50 mM, pH 7.6). Substrate solution for the conventional assay was prepared by dissolving 3,3′-diaminobenzidine (DAB, 15 mg) and immidazole (200 mg) in 30 mL of Tris-HCl (50 mM, pH 8.0) and used after addition of 0.001% H2O2. In the ultrasensitive assay, 4-chloro-1naphthol (4-CN, 15 mg) dissolved in 5 mL of methanol was diluted to 30 mL with Tris-HCl (50 mM, pH 8.0) and used after addition of 0.001% H2O2. Preparation of Membrane Strips for Conventional Assay. Nitrocellulose membrane strips (14 × 70 mm) were marked with a pencil to give five 14 × 14 mm squares. The strips were soaked in blotting buffer (Tris-HCl, 20 mM, pH 8.0, containing 0.9% NaCl) for 5 min with gentle shaking. The strips were then semidried by gently shaking in the air. Anti-AFB1 antibody was diluted 100-fold with Tris-buffered saline (50 mM, pH 8.0) containing 10 µg/mL of BSA. It was applied in a volume of 5 µL at the center of each 14-mm square and spread to ∼7-mm diameter with the tip of the dispenser. The top square of the rectangular membrane was not spotted and kept free for attachment to the polyethylene base. After drying the membrane at room temperature for 15 min, it was further dried by incubating at 37 °C for 30 min. The vacant sites of the membrane were blocked with 0.4% casein in carbonate-bicarbonate buffer (50 mM, pH 9.6) and then washed three times with washing buffer (Tris-HCl buffer, 20 mM, pH 8.0, containing 2.9% NaCl and 0.05% Tween 20). The strips were dried at room temperature for 15 min and then at 37 °C for 30 min. Preparation of Membrane Strips for Ultrasensitive Assay. The membrane strips were prepared as described above, except that anti-AFB1 antibody containing 200 µg/mL of BSA was used at a dilution of 1:2500. The vacant sites of the membrane were blocked with 0.2% p-OH-PPA-casein conjugate in carbonatebicarbonate buffer (50 mM, pH 9.6). (20) Chu, F. S.; Hsia, M.-T. S.; Sun, P. J. Assoc. Off. Anal. Chem. 1977, 60, 791794. (21) Das Sharma, J.; Duttagupta, C.; Ali, E.; Dhar, T. K. J. Immunol. Methods 1995, 184, 1-6.
Analytical Chemistry, Vol. 76, No. 1, January 1, 2004
99
Assembly of the Analytical Device. A 5 × 70 mm tape having adhesive on both sides was fixed parallel to and ∼1 cm away from the shorter edge of a polyethylene card (140 × 120 mm). The unspotted top square of the membrane strip was then fixed on the tape. Up to four membrane strips can be fixed to one card with 2 mm gaps between adjacent strips. A rectangular piece of filter paper was used as the adsorbent body during the assay. The sizes of the absorbent body required for 1, 2, 3, and 4 strip devices were 90 × 56, 100 × 80, 110 × 100 and 120 × 120 mm, respectively. The assembled device can be stored in a desiccator for at least 6 months. Conventional Assay Procedure. The filter paper was wetted with a stream of distilled water from a wash bottle. After thoroughly shaking off excess water, it was placed between the membrane and the polyethylene card. Alternatively, the filter paper was placed between the membrane and the polyethylene card and wetted with a limited amount of water (1.5, 2.5, 3.5, and 4.5 mL for 1-, 2-, 3-, and 4-strip analytical devices, respectively). In both cases, the air entrapped between the membrane and the absorbent body was removed by repeated rolling (3-4 times) of a rimless tube or a glass tube under slight pressure over the membrane strips. Standards (0, 10, 25, 50, 75, and 100 pg) and samples (25 µL/spot) were applied at the center of the antibody spotted zones of the membrane. To each spotted area, 25 µL of AFB1-HRP conjugate (1:1000) in assay buffer (Tris-HCl, 50 mM, pH 8, containing 0.9% NaCl, 0.2% BSA, and 0.01% thimerosal) was added. The absorbent body was then removed, and the membrane was thoroughly washed with an even stream of wash buffer from a wash bottle. After removing the adhered buffer from the surface by gently sponging with a tissue paper, the device was placed on a horizontal surface. The DAB substrate solution (2.5 mL/strip) was poured uniformly over the membrane surface. Under these conditions, there is no overflow of the substrate solution from the membrane strips. After incubation for 1 min, the device was lifted, allowed to drain, and washed with tap water to stop the reaction. The whole assay was performed at ambient temperature. The color intensities of the sample spots were compared with those of reference standards either visually or by densitometry. The calibration curve for AFB1 was constructed by plotting percentage of B/B0 values (x axis), obtained from densitometric analysis against the toxin concentration (y axis), where B0 is the density of the zero standard (no AFB1 added), and B is the density of the standard. The limit of detection (LOD) was estimated at 2 SD below the zero standard (n ) 10). Double of the LOD value was taken as the limit of quantitation (LOQ). Ultrasensitive Assay Procedure. The initial steps were the same as described above. After addition of standards (0, 0.5, 1, 2.5, 5, and 10 pg) or sample (25 µL/spot) to antibody-spotted areas of the membrane, 25 µL of AFB1-HRP conjugate was added at a dilution of 1:20 000. The filter paper was then removed, and the membrane strips were thoroughly washed with washing buffer. For amplification, B-T solution (200 µmol/l) was poured uniformly over the membranes (2.5 mL/strip) and incubated for 2 min. The strips were washed again with washing buffer and another piece of wetted filter paper was placed between the membranes and the solid support of the device. The entrapped air was removed as described earlier, and 25 µL of avidin-HRP conjugate (diluted 500-fold with assay buffer) was added to each spotted area. The 100
Analytical Chemistry, Vol. 76, No. 1, January 1, 2004
filter paper was then removed, and the membrane was thoroughly washed with washing buffer. Substrate solution (4-CN) was poured uniformly over the membrane, incubated for 2 min, and then washed with tap water to stop the reaction. The whole assay was performed at ambient temperature. AFB1 content of the samples were estimated by visual comparison of the color intensities of the sample spots with those of reference standards. The calibration curve was constructed from densitometric data as described above. CARD Assay Procedure. The assay was carried out as described above for the ultrasensitive assay except that anti-AFB1 antibody supplemented with 100 µg/mL BSA was used at a dilution of 1:100 for spotting the membrane strips, and the vacant sites were blocked with 0.2% casein instead of p-OH-PPA-casein. AFB1-HRP conjugate was used at a dilution of 1:10 000. Effect of Absorbent Body Size. Membrane strips were prepared, and the assays were then carried out in the absence of AFB1 by the conventional protocol described earlier with a single strip of membrane. Filter papers of four different sizes, 90 × 56, 80 × 40, 80 × 30 and 80 × 20 mm (∼50, 32, 24, and 16 cm2 respectively), were tested. The reagents were added to the four spots marked a, b, c, and d in the order a to d. The color intensities of spots were measured by densitometry. Flow rate was measured as the time required for complete absorption of the applied fluid from the spotted zones. The effects of filter paper size and the distance between the spots were studied by applying 50-µL (2 × 25 µL) portions of 5% copper sulfate solution (w/v) in place of the assay reagents and measuring their absorption areas in the filter paper. In Situ Concentration Assay. For this study, the AFB1 standards (0, 10, 50, and 100 pg) in assay buffer were added in different volumes ranging from 50 to 400 µL. Additions were done in 50-µL (2 × 25 µL) portions. Excess fluid from the absorbent body was extruded by rolling a rimless glass tube under slight pressure over the membrane strip after each 50-µL addition. The procedure was repeated until the required volume of standard had been added. The assay was then carried out by the conventional protocol described earlier. Infection of Seeds. Aspergillus parasiticus spores were suspended in water, and 5 mL of suspension was added to groundnut or corn seeds (20 g) in Petri dishes and incubated at 30 °C from 48 to 192 h. The control sets contained sterile water in place of spore suspensions. After the specified time period, the seeds were autoclaved and dried by blotting on a filter paper. Extraction of Samples. Seeds were milled in a coffee grinder and 2-g portions of each sample were soaked in 5 mL of methanol-water (80:20) for 15 min. The suspension was then centrifuged, and the clear supernatant was decanted. In the case of the groundnuts, a 2-mL portion of the extract was defatted by extracting three times with 500-µL portions of hexane. The aqueous phase was diluted 500-fold in assay buffer and used directly in the assay without any further cleanup. Spiked Samples. For preparation of spiked samples, appropriate volumes of AFB1 solution in methanol (10 ng/µL) was added to 2 g of milled noninfected seeds and incubated at 37 °C for 24 h. The dried samples were extracted using methanol/water as described before.
Sample Cleanup by Solid-Phase Extraction. Sample extracts (100 µL) in aqueous methanol were diluted (30-fold) with water and passed through a Sep-Pak C18 cartridge (Catalogue no. 051910, Waters Assoc., U.S.A.). The cartridge was washed with 5 mL of water, followed by 10% methanol in water, and then eluted slowly with 2 mL of methanol. The eluates were evaporated on a steam bath and reconstituted with 1 mL of assay buffer. High-Performance Liquid Chromatography. HPLC was carried out on a model 626 liquid chromatograph (Waters Corporation, Milford, CT) equipped with a model 2487 variable wavelength absorbance detector. Conditions of HPLC were as follows: column, Novapak-C18, 150 mm × 3.9 mm i.d; particle size, 5 µm; mobile phase, methanol-water (50:50, vol/vol); flow rate, 0.4 mL/min; detector wavelength, 363 nm; sensitivity, 0.01 absorbance units, full scale. Standards and samples were dissolved in the mobile phase, and 10 µL was injected with the instrument’s Rheodyne 9725 injector. Under these conditions, good resolution of all the four aflatoxins was obtained, with a retention time of 9.5 min for AFB1. Quantitation was performed by measuring the peak area, which was proportional to the amount of AFB1 applied (1-20 ng). Safety Precautions. The solutions were prepared under a vented hood, and all of the work areas and glassware in contact with the mycotoxins were cleaned with 5% sodium hypochlorite solution. Waste was disposed in bleach. RESULTS AND DISCUSSION Immobilization of Antibody to the Membranes. We compared the assay response of nitrocellulose membranes (0.45-µm pore size) from Millipore (Catalog No. HAHY00010), Schleicher & Schuell (Catalog No. 10401197), Pierce (Catalog No. 88018), Sigma (Catalog No. N-8392), Osmonics (Catalog No. EP4HY00010), and Bio-Rad (Catalog No. 162-0148). Under identical conditions, all the membranes gave very low background color in the unspotted areas. However, the Millipore and Schleicher & Schuell membranes produced distinctly higher spot intensities of zero standard, leading to higher sensitivity. The Millipore membrane was used in all subsequent experiments. The overall performance of the assay is dependent on several key factors, such as the dilutions of anti-AFB1 antibody, the enzyme conjugate, and the B-T reagent, and the period of incubation in the amplification steps. The various parameters affecting the assay response were quantitatively investigated and optimized by using densitometric scanning of the spots. Because detection in the present method depends on visualization of the intensity of the color spots, the uniformity of spotting antibody over the membrane surface is important. The spotted areas also need to be sufficiently large to cover the volume of reactants added subsequently. We have applied the antibody in a volume of 5 µL and spread it with the tip of the dispenser to give spots of ∼7mm diameter. Smaller or larger volumes of antibody may also be used, but we have not investigated it in detail. The minimum spot distance required between the center of adjacent spots was found to be 12 mm from experiments carried out with copper sulfate. It also showed spot size on the membrane of ∼6 mm diameter. Decreasing this distance to 10 mm resulted in a decrease in the flow rate and merger of the absorption areas of adjacent spots in the filter paper. We used a 14-mm spot distance for convenience.
Table 1. Influence of Absorbent Body Size on Spot Intensity in Absence of AFB1 absorbent body size (mm) 90 × 56 80 × 40 80 × 30 80 × 20
spot density, arbitrary units (1 × 104) a b c d 10.92 11.65 11.50 10.70
11.04 11.36 10.73 9.01
11.46 11.40 9.75 8.20
11.14 11.52 9.15 7.44
Size of the Membrane and Polyethylene Card. The size of the membrane strip is not critical; however, its breadth should be larger than the diameter of the spotted area. We used a breadth of 14 mm because our spotted area had a diameter of ∼7 mm. The size of the polyethylene card should be slightly larger than the filter paper for convenience during the assay. Cards of alternative material with adequate mechanical strength and compatible with the chemistry of the reagents may also be used. We used up to four membrane strips in a single test device, but more than four strips can be used. The effect of distance between the strips on the assay was investigated using AFB1 zero standard and was found to have no effect in assay performance. We used a distance of 2 mm for convenience. Influence of Absorbent Body Size on Reproducibility of Spot Intensity. The effects on the intensities of the four spots (a, b, c, and d) of a strip obtained in the absence of AFB1 by using filter papers of different sizes is shown in Table 1. The results showed the use of absorbent bodies of smaller sizes (80 × 30 and 80 × 20 mm) resulted in a gradual decrease in spot intensities in the order a to d. The maximum decrease (30%) was in the d zone of the 80 × 20 mm size. The flow rate measured for a, b, c, and d spotted zones in this filter paper was ∼9, 9, 13, and 15 s for standards and 18, 28, 40, and 54 s for the enzyme conjugate, respectively. It may be pointed out that during the assay, the additions of standards and the enzyme conjugate to the spotted zone were in the order a, b, c, and d. To investigate this phenomenon, we carried out experiments with copper sulfate solution. Results showed increased spreading of the spotted zones in the order a, b, c and d with diameters of ∼8, 8.5, 9.5, and 10.5 mm, respectively. The most likely explanation is the addition of a reagent in one spot decreases the void volume of the filter paper and thereby decreases the flow rate of the subsequent spot, leading to more spreading of the spots. For all subsequent spots, the effect would be cumulative. When absorbent bodies of larger sizes (90 × 56 and 80 × 40 mm) were used, good reproducibility of intensities in the four spots, a, b, c, and d, was obtained. The time required for complete absorption of the added reagents in all the spotted zones for the standards was ∼9 to 10 s in both the filter papers. The corresponding time for the enzyme conjugate was ∼13 to 14 s for the 90 × 56 mm and 16-17 s for the 80 × 40 mm. It was observed that the optimum water content in the 80 × 40 mm was in the range of 0.9-1.1 mL, as compared to 1.4-1.8 mL in the 90 × 56 mm size. Further increase of water content markedly influences the flow rate and causes lateral spreading. Therefore, the use of a larger size absorbent body is required to maintain reproducibility of the assay. We used the larger absorbent body of 90 × 56 mm (50 cm2) size in the single strip analytical device; however, the 80 × 40 mm (32 cm2) size can also be used Analytical Chemistry, Vol. 76, No. 1, January 1, 2004
101
by an experienced operator. We then determined the area of absorbent body to be used with the 2-, 3-, and 4-membrane strips. Because taking 50 cm2 for the first strip and adding 50 cm2 for each additional strip would have made the absorbent body sizes quite a bit larger, we added an alternative 32 cm2 area for each additional strip. On the basis of these calculations, we tested 100 × 80, 110 × 100 and 120 × 120 mm absorbent bodies for 2, 3, and 4 strips, respectively, by assaying the zero standard. The reproducibility of the spot intensities were found to be excellent, the coefficient of variation for 2-, 3-, and 4-strip devices being 2.6 (n ) 8), 1.6 (n ) 12) and 2.2% (n ) 16), respectively. Role of Prewetted Absorbent Body. The wetted absorbent body facilitates the adherence of the membrane strips to the absorbent body and maintains the continuity of capillary channels with fluid receiving zone. It is important that there is no air bubble between the membrane and the absorbent body and that both are in intimate contact with each other, because air bubbles interfere with the fluid flow through the membrane. The void volume of the wetted absorbent body is sufficient to absorb the additional fluid introduced during the assay. As a result, the applied fluid is efficiently absorbed by the absorbent body through antibody-immobilized zones within a few seconds without application of any force. Because there is very little lateral diffusion, costly labeled reagents can be used efficiently. Optimal wetting of the absorbent body is essential because it has considerable influence on the sensitivity of the assay. The water content can be controlled by using either of the methods described in the Experimental Section and needs to be between 1.4 and 1.8 mL for a 90 × 56 mm filter paper (∼1.4-1.8 mL/g) for good results. The water content of such a filter paper prepared by rolling a test tube by four different operators was consistently found to be within this range. If excess water is removed, the contact between the membrane and the absorbent body becomes insufficient and nonuniform. Excess wetting (2 mL/g) of the absorbent body reduces the flow rate, thereby allowing the fluid to spread laterally, resulting in a gradual decrease in spot intensities in the order a to d. Assay Characteristics Using the Analytical Device. To evaluate the performance of the analytical device, a calibration curve for AFB1 for both the conventional and ultrasensitive assay was constructed. The LOD using conventional and ultrasensitive method was 5 and 0.25 pg/spot (0.2 and 0.01 ng mL-1), respectively, with corresponding B/B0 values of 95 and 96%. Because the LOD dose is seldom practically achievable, LOQ doses of 10 and 0.5 pg/spot (0.4 and 0.04 ng mL-1) were taken as the lowest standard for the conventional and ultrasensitive methods, respectively. Accordingly, the mean percentage of B/B0 values for standards 10, 25, 50, 75, and 100 pg obtained in the conventional method were 87 ( 2, 60 ( 2.4, 44 ( 3, 30 ( 2, and 22 ( 2.5, respectively (n ) 10). In the ultrasensitive method, at the level of 0.5, 1, 2.5, 5,and 10 pg, the mean percentage of B/B0 values were 90 ( 2.5, 80 ( 2.0, 52 ( 1.6, 30 ( 1.6, and 8 ( 2 respectively (n ) 10). The AFB1 concentrations required to achieve 50% inhibition using the conventional and ultrasensitive methods were 42.5 and 2.7 pg/spot. Under identical conditions, the CARD method19 (membranes blocked with casein) did not produce any color. However, a calibration curve in the range of 0-50 pg was obtained by increasing the antibody and enzyme 102 Analytical Chemistry, Vol. 76, No. 1, January 1, 2004
Figure 2. The effect of applied standard volume on the doseresponse curve. The applied volumes were 25 ([), 50 (]), 100 (2), 200 (4), 300 (b), and 400 µL (O).
conjugate concentrations. The sensitivity obtained was 10-fold less, as compared to the Super-CARD method. Assay with In Situ Concentration of Samples. Before performing the calibration studies of AFB1 for the preconcentration assay, the effect of applied sample volume on the spot intensities was studied using the zero standard. The spot intensities obtained by using 50- or 100-µL volume were ∼25 and 65% less, respectively, compared to the control (25 µL) spot. A possible explanation is that the excess wetting of the absorbent body with the zero standard spreads the added enzyme conjugate solution and also reduces its flow rate. This effect was more pronounced when a higher volume (100 µL) was used. However, when 100 µL of zero standard was added in four 25-µL portions followed by removal of excess fluid from the absorbent body by pressing a rimless glass tube over the membrane strip after addition of each 50 µL, spot intensities comparable to those of the control were found. This showed that the analytical device could be used for the assay of diluted samples by preconcentration. The effect of the applied volume of standards on the doseresponse curve is shown in Figure 2. For this, each of the four AFB1 standards (0, 10, 50, and 100 pg/spot) was applied in 25-, 50-, 100-, 200-, 300-, and 400-µL volumes. The results showed that as the applied volume increased, the displacement curves became steeper, with a remarkable decrease in percentage B/B0 value at higher concentrations. With volumes of 25, 50, 100, 200, 300, and 400 µL, the percentage B/B0 values at 100 pg/spot of AFB1 were 22.5, 16.7, 11.6, 8.3, 6.7, and 4.8, respectively. The corresponding I50 values, that is, AFB1 concentration required to achieve 50% inhibition, were 42.5, 39.5, 38, 35, 34, and 33 pg/spot. The spot intensity of the zero standard was not affected at all the volumes tested. Analysis of Food Samples. Because the sensitivity of the ultrasensitive assay method was sufficient to meet the requirements of regulatory agencies, both spiked and contaminated food samples were analyzed in a batch of 12 extracted samples. The total time required for placing the absorbent body in the device, addition of standards and samples, enzyme conjugate, and washing after each step was ∼6 min. The method has two incubation steps of 2 min each after the addition of the B-T and substrate solutions. The total assay time required in our laboratory for analyzing 12 samples in a single test card was within 12 min (sample extraction time not included).
Table 2. Matrix Effect of Aqueous Methanol Extract of Noninfected Food Samples extract dilution
groundnuta
corn
% B/B0 wheat
1:25 1:50 1:100 1:250 1:500 1:1000 1:2000
44.7 (49.6) 58.3 (72.7) 71.8 (84.6) 86.0 (95.8) 96.4 (102.2) 107.5 (105.9) 108.0 (106.0)
22.5 37.3 60.6 86.6 108.9 109.5 109.8
29.2 48.5 64.3 101.5 103.2 102.8 103.6
a
Table 3. Recovery of AFB1 Spiked to Some Food Samples AFB1 (µg/kg)
cheese 97.4 125.1 126.1 110.0 107.4 113.8 128.6
chilli
28.6 82.3 95.0 104.5
Values in parentheses denote extracts treated with hexane.
matrix
added
founda
% recovery
groundnut
20 40 80 160
24 ( 1.6 36 ( 1.3 72 ( 3.4 156 ( 2.5
120 90 90 97 99
25 100 250
27 ( 1.6 100 ( 9.0 244 ( 7.5
108 100 98 102
50 250
51 ( 1.0 250 ( 2.0
102 100 101
25 100 150
21 ( 2.4 91 ( 6.5 147 ( 3.6
85 91 98 91
50 100 300
60 ( 2.0 100 ( 2.0 280 ( 3.5
120 100 93 104
mean corn mean wheat
To assess the possible interference of sample constituents other than aflatoxin, both hexane-treated and nontreated aqueous methanol extracts of noninfected groundnut, corn, wheat, chilli, and cheese were diluted with assay buffer (25-2000-fold) and assayed using the assay buffer as the control. It was found that hexane treatment of the extracts was not necessary except for groundnut. As can be seen from Table 2, the matrix interference can be avoided if aqueous methanol extracts of groundnut, corn, wheat, cheese, and chilli are diluted 1:500, 1:500, 1:250, 1:50, and 1:2000, respectively. After sample cleanup by solid-phase extraction, the corresponding dilutions required were 1:250, 1:250, 1:125, 1:25, and 1:500, respectively. These dilutions were selected for the subsequent studies. At low dilutions, all the extracts gave slightly higher color intensities than the negative control. Similar observations have also been reported earlier and were explained as due to the presence of endogenous peroxidases in the extracts.22 To study the recovery of AFB1 during extraction, noninfected samples of groundnut, corn, wheat, cheese, and chilli were spiked with AFB1 at different levels, extracted, diluted in assay buffer, assayed, and quantitated by densitometry. The recovery was found to be in the range of 85-120% (Table 3). To evaluate the accuracy of the method, extracts of infected groundnut and corn were spiked with different levels of AFB1 and analyzed in quadruplicate. Results summarized in Table 4 show excellent accuracy. The intra- and interassay variability was determined by assaying three groundnut extracts of infected seeds containing low, medium, and high concentrations of AFB1. The mean AFB1 concentrations in the three extracts were 14.6, 75.4, and 204 µg/kg with intra-assay CV of 17.8, 8.0 and 1.4% (n ) 12), respectively. The interassay variations were 20.5, 8.6, and 1.6% (n ) 3), respectively. The AFB1 levels of 22 samples of groundnut and corn infected with Aspergillus parasiticus were then compared with the values obtained from HPLC. Linear regression analysis (using the statistical analysis module from Analyze-It Software Ltd., U.K.) on the data yielded an excellent correlation (R2 ) 0.99, slope ) 1.043, intercept ) 3.228, n ) 22) between the methods. The stability of the reagents stored at ambient temperature was examined up to six months by periodically assaying AFB1 standards. The results showed that the coated membrane, B-T solution, and other reagents, except the enzyme conjugate, are stable for more than 6 months at ambient temperature. The (22) Laamanen, I.; Veijalainen, P. Food Addit. Contam. 1992, 9, 337-343.
mean cheese mean chilli mean a
Mean of three determinations (SD.
Table 4. Analytical Recovery of AFB1 from Spiked Sample Extracts AFB1 (µg/kg) matrix
endogenous
added
calcd
founda
% recovery
groundnut
17.2 ( 1.0
20 100 200 20 100 200 20 100 200 25 125 250 25 125 250 25 125 250
37.2 117.2 217.2 57.6 137.6 237.6 138.0 218.0 318.0 47.5 147.5 272.5 112.5 212.5 337.5 273.0 373.0 498.0
36 ( 1.5 126 ( 4.0 212 ( 2.5 64 ( 1.0 160 ( 4.0 264 ( 2.0 140 ( 3.0 220 ( 1.2 336 ( 3.0 49.3 ( 1.2 135 ( 3.0 257 ( 3.0 110 ( 5.0 207 ( 6.0 343 ( 7.6 257 ( 3.0 332 ( 3.0 505 ( 5.0
96.8 107.5 97.6 111.1 116.3 111.1 101.5 101.0 105.7 103.8 91.5 94.3 97.8 97.4 101.6 94.1 89.0 101.4
37.6 ( 1.6 118 ( 4.6 corn
22.5 ( 0.3 87.5 ( 2.0 248.0 ( 3.0
a
Mean of quadruplicate determination (SD.
enzyme conjugate solutions lost 20% of their activity after 30 days at ambient temperature but were stable for more than one year at 4 °C. CONCLUSION We have demonstrated a new approach for performing immunofiltration-based immunoassay for the detection of AFB1 on a batch of samples. The method is noninstrumental and does not require a pump to create vacuum to pull reagents through the membrane. The prewetted absorbent body allows focused absorption of the applied fluid through antibody-immobilized zones within a few seconds. To our knowledge, the use of such a prewetted Analytical Chemistry, Vol. 76, No. 1, January 1, 2004
103
absorbent body has not been reported earlier. Additionally, dilute samples can also be assayed by preconcentration with improved dose-response characteristics. Adaptation of the Super-CARD amplification technology in the analytical device was found to be a simple and efficient option to increase the sensitivity of AFB1 assay. Further increase in sensitivity should also be possible using other amplification systems. The reliable quantitation limit (0.5 pg/spot, 0.02 ng mL-1) obtained by the present method for AFB1 is much higher than a recently described4 test strip procedure (18 ng), surface plasmon resonance-based immunoassay5 (3 ng mL-1) and immunoaffinity fluorometric biosensor7 (0.1 ng mL-1). Rapid detection of total aflatoxin in extracted food samples has also been described using a fluorescence polarization method.10 The data reported here indicate that the method is capable of producing acceptable results to analyze AFB1 in a variety of foodstuffs at a level of regulatory relevance, with accuracy and precision. A single test card is used to analyze 12 samples in the
104
Analytical Chemistry, Vol. 76, No. 1, January 1, 2004
presence of standards. An additional advantage of this method is the ability to determine AFB1 in crude sample extracts, which has several practical advantages. The simplicity of the approach should make it suitable for use in the detection of many other toxins and environmental pollutants. Although we have limited our investigations to the determination of small molecules, it is obvious that this method could also be utilized for macromolecules. ACKNOWLEDGMENT The authors acknowledge financial support from the Department of Biotechnology, New Delhi, Grant BT/PR 0405/PID/14/ 013/96.
Received for review June 26, 2003. Accepted August 22, 2003. AC034694G
