Introduction

In recent years, new plant varieties developed by adopting modern biotechnology, including genetic transformation, have to a large extent replaced some conventional crops, such as corn, cotton, and soybean. As a consequence, food products containing genetically modified organisms (GMOs) are largely diffused, especially in the USA, where most of the food products in supermarkets contain GMOs [1]. However, genetically modified (GM) foods have not gained worldwide acceptance because of concerns about the environmental and public health safety issues of GMOs, such as potential gene flow to other organisms, destruction of agricultural diversity, allerginicity, induction of antibiotic resistance, and gastrointestinal problems [2–5]. Thus, in the late 1990s, some countries, in particular those in the EU, introduced legislation requiring mandatory labeling of GMO-derived food products (EU Regulations 258/97 and 1139/98). In terms of the accidental contamination of unmodified food with GMO materials, the EU Regulation 49/2000 established that any food containing more than 1% of GMO materials from EU-approved transgenic varieties must be labeled as a “GMOs containing” food. More recently, the EU Regulation 1829/2003 reduced this threshold to 0.9%, while zero tolerance was maintained for EU-unapproved GMO varieties. This regulation also allowed a tolerance of 0.5% for GMO varieties that have not been fully approved by the European Commission but have initial clearance from the European Commission Scientific Committee. In addition, the EU Regulation 1830/2003 established guidelines concerning the traceability of GMOs and GMO-derived food and feed products. Detection and traceability of GMOs in the food production chain thus represent very important issues in the EU [6]. The implementation and maintenance of actual and, in perspective, future regulations necessitates analytical methodologies that allow for the accurate quantitative determination of the content of GMOs in a food sample.

Current recognized methodologies for the quantitative analysis of GMOs are focused on either one of two targets, namely, the transgenic DNA inserted into the genetically modified organism or the novel protein(s) expressed, which are usually detected by means of polymerase chain reaction (PCR)-based assays and immunoassays, respectively [6, 7]. Other analytical approaches, such as those based on microarray technology, mass spectrometry, or surface plasmon resonance (SPR) devices have been proposed, but their potential for GMOs detection, identification, and quantification has still to be evaluated.

The measurement of transgenic gene copies by real-time PCR (rt-PCR) represents the method of choice for quantification of GMO constituents in food and feed [8, 9], and within recent years several rt-PCR methods for quantitative analysis of GMOs have been published [10–13]. The exponential increase of the targeted DNA sequence during PCR is the reason for the high sensitivity of these methods, which allow the detection of 0.1–0.01% GMO DNA (lower detection limits cannot be easily achieved in the analysis of real samples due to the intrinsic limitations of the technique [14]). However, rt-PCR techniques are expensive, time-consuming, and require sophisticated instrumentation and trained personnel. Thus, they are not adequate for the extensive high-throughput screening of crops and foodstuffs samples, as would be required for the implementation of EU regulations, nor can they be exploited for on-site GMO assays.

On the other hand, immunoassays allow both qualitative and quantitative measurement of the target proteins using simple experimental procedures and inexpensive instrumentation, and hundreds of samples can be simultaneously analyzed using high-throughput analytical formats, such as 384-well microtiter plates [15]. Immunoassays have already been developed for the detection and quantification of the most common GMO proteins expressed by biotechnology-derived crops in production, and many immunoassay kits are commercially available with detection limits in the order of 1.0% or less of GM material [16–19]. Recently, an immunoassay for the genetically modified maize line MON 810, developed and commercialized by Monsanto, has been the subject of a large international collaborative study and was demonstrated to quantitatively determine the concentrations of biotech crops in samples of ground grain [20]. Quantitative immunoassays for biotech crops are currently being used on a large scale in the USA and worldwide to manage the sale and distribution of grain and for controls purposes. Qualitative or semi-quantitative field testing is also possible by using immunoassays based on the lateral flow strip technology [21, 22]. However, use of immunoassays for the analysis of food samples is hampered by the different levels of protein expression in various tissues and the dependence of the expression level on the growing conditions, as well as by the reduced sensitivity due to protein denaturation during food processing. Therefore, an improvement in the analytical performance of immunoassays is desirable to increase the applicability of such assays, in particular for the detection of the small amounts of transgenic proteins resulting from contamination of unmodified food with GMO-derived materials at levels below the actual EU threshold values.

In this paper we describe the development and validation of a highly sensitive sandwich chemiluminescent (CL) enzyme immunoassay for Cry1Ab protein. Truncated forms of such protein, produced by Bacillus thuringiensis (Bt), are expressed in several transgenic crops to confer protection against chewing insect pests. In particular, Bt maizes (including the MON 810 maize) express Cry1Ab to control the European corn borer [23], whereas the closely related Bt protein Cry1Ac is produced in current Bt cotton cultivars to protect the plants from bollworm, a major pest of cotton. In the developed CL assay, sample extracts are added to microtiter plate wells coated with anti-Cry1Ab antibodies, which bind the toxin. The bound toxin is then detected by adding a second anti-Cry1Ab antibody and an anti-antibody horseradish peroxidase(HRP)-labeled antibody, followed by the CL measurement upon addition of a suitable enzyme CL substrate. The CL detection of the bound enzyme allowed the development of a highly sensitive assay. In fact, provided a low degree of non-specific binding of the detection reagents occurs, the detectability of the enzyme label is one of the most important factors determining the detection limit of a sandwich-type non-competitive enzyme immunoassay, and it is known that CL detection of enzyme labels is more specific and sensitive than other spectroscopic techniques, such as spectrophotometry or fluorometry [24, 25]. The assay can be performed both in conventional 96-well and in high-throughput 384-well microtiter plate formats, the latter being suitable for large-scale screening purposes once optimized and automated, including the preliminary sample extraction steps. The analytical performance of the assays has been evaluated and compared with that of the 96-well microtiter plate assay with colorimetric detection. The suitability of the developed assays for the detection of GMOs in maize samples has been demonstrated by analyzing certified reference materials.

Experimental

Reagents and instrumentation

Mouse monoclonal antibody (capture antibody, catalog No. 490002) and rabbit polyclonal antibody (detection antibody, catalog No. 490012) specific for Cry1Ab protein, as well as Cry1Ab, Cry1Ac, Cry1F, Cry2A, and Cry9C protein standards, were purchased from Abraxis LLC (Warminster, PA, USA). Horseradish peroxidase-labeled anti-rabbit goat antibody (tracer antibody, catalog No. A-6154), bovine serum albumin (BSA), and Tween 20 were provided by Sigma (St Louis, MO, USA). Chromogenic HRP substrate (SureBlue TMB 1-Component Microwell Peroxidase Substrate) and chromogenic stop solution (TMB Stop Solution) were from KPL Inc. (Gaithersburg, MA, USA), while the enhanced HRP chemiluminescent system (luminol/peroxide/enhancer) SuperSignal ELISA Femto was purchased from Pierce (Rockford, IL, USA). Certified reference materials (maize powders containing 0%, 0.1%, 0.5%, 1%, and 2% MON 810) were acquired from Fluka (Fluka Chemie GmbH, Buchs, Switzerland). A maize sample containing 0.05% MON 810 was prepared by mixing equal amounts of the 0% and 0.1% reference materials. All other reagents used were of analytical grade. Deionized water was purified with a Milli-Q System (Waters, Milford, MA, USA).

Transparent 96-well microtiter plates (Nalge Nunc International, Roskilde, Denmark) and black 96- and 384-well microtiter plates (Thermo Electron Corporation, Waltham, MA, USA) were used for the development of the colorimetric and CL immunoassays, respectively. The reagents were dispensed in the microtiter plates using a Multidrop 384 microtiter plate reagent dispenser (Thermo Electron Corporation). Absorbance and CL measurements were performed using a Multiskan EX microtiter plate photometer (Thermo Electron Corporation) and a Luminoskan Ascent microtiter plate luminometer (Thermo Electron Corporation), respectively.

Assay procedure

The assay was performed in transparent 96-well microtiter plates or black 96- and 384-well microtiter plates employing colorimetric and CL detection, respectively (the 384-well microtiter plate required lower volumes of samples and reagents, which are reported in parentheses below).

Coating of microtiter plates

One hundred μL (or 25 μL) of a 2 μg mL−1 solution of the anti-Cry1Ab mouse monoclonal antibody (capture antibody) in coating buffer (0.05 mol L−1 carbonate buffer, pH 9.6) were added to the wells of the microtiter plates and incubated overnight at 4°C. The solution was then removed and the plates were blocked with 150 μL well−1 (or 40 μL well−1) of blocking buffer (coating buffer containing 1.0% BSA) for 1 h at 37°C. After incubation, the plates were washed with distilled water, dried under vacuum, and stored at 4°C until use.

Sample extraction

Certified reference materials were subjected to protein extraction prior to analysis. Five hundred milligrams of powdered sample was added to 2 mL of assay buffer (0.05 mol L−1 phosphate buffer, pH 7.4, containing 0.1% BSA and 0.05% Tween 20). The suspension was incubated at room temperature for 2 h under gentle shaking, then centrifuged at 3,000 g for 5 min, and the supernatant was assayed by ELISA.

Sample analysis

One hundred μL (or 25 μL) of sample extract was added to the wells of the microtiter plate and incubated for 30 min (all incubation steps of the immunoassays were performed at 25°C). For the calibration curve, serial dilutions of the Cry1Ab standard protein in assay buffer, in the concentration range 0.008–2 ng mL−1, and a blank (assay buffer) were also dispensed into the plate. After incubation, the wells were washed three times with washing buffer (0.05 mol L−1 phosphate buffer, pH 7.4, containing 0.1% Tween 20), and 100 μL (or 25 μL) of a 2 μg mL−1 solution of the anti-Cry1Ab rabbit polyclonal antibody (detection antibody) in assay buffer was added and incubated for 30 min. The wells were washed three times with washing buffer, then 100 μL (or 25 μL) of a 0.2 μg mL−1 solution of the HRP-labeled anti-rabbit goat antibody (tracer antibody) in assay buffer was added and incubated for 30 min. The wells were washed again three times with washing buffer, then the HRP activity of the bound tracer antibody was measured using colorimetric or CL detection.

Colorimetric detection

The colorimetric detection of the peroxidase activity was performed by adding 100 μL of the chromogenic HRP substrate to each well. After incubation for 30–45 min in the dark at 25°C, the enzymatic reaction was stopped with 100 μL well−1 of TMB Stop Solution and the absorbance at 450 nm was recorded.

Chemiluminescent detection

The CL detection of the HRP activity was performed by adding 100 μL (or 25 μL) of the CL HRP substrate to each well. The CL signal, which reached a steady-state intensity within 2–3 min, was then recorded. Chemiluminescence intensities were expressed in relative light units (RLU).

Data analysis

Before data analysis, the blank signal was subtracted from all the sample and calibration curve readings (absorbance or CL intensity). The concentration of Cry1Ab protein in the samples was then determined by interpolation of the sample signals on the linear calibration curve run concurrently.

Results and discussion

The optimal concentrations of the anti-Cry1Ab capture antibody, anti-Cry1Ab detection antibody, and HRP-labeled anti-rabbit tracer antibody were determined (Fig. 1). For the immobilization of the anti-Cry1Ab capture antibody, physical adsorption on polystyrene microtiter plates was chosen because it is the simplest method and gives reproducible and stable immobilization. The concentration of the capture antibody was determined by coating 96-well microtiter plates using different capture antibody dilutions in the range 10–0.5 μg mL−1. A calibration curve for Cry1Ab standard protein samples in the range 0.008–2 ng mL−1 was then obtained in each plate using either the colorimetric or CL assay. The lowest capture antibody concentration (2 μg mL−1) giving a linear calibration curve extending up to 2 ng mL−1 Cry1Ab was selected as the optimal concentration. The same procedure was employed for the optimization of the anti-Cry1Ab detection antibody and the HRP-labeled anti-rabbit tracer antibody concentrations: linear dose–response curves in the range 0.008–2 ng mL−1 Cry1Ab were obtained until the concentrations of the anti-Cry1Ab detection antibody and the HRP-labeled anti-rabbit tracer antibody were decreased up to 2 and 0.2 μg mL−1, respectively.

Fig. 1
figure 1

Optimization of the immunoassay: dose–response curves obtained for the 96-well CL immunoassay using different concentrations of a capture antibody, b detection antibody, and c tracer antibody. Fixed antibody concentrations were: a detection antibody 5 μg mL−1 and tracer antibody 1 μg mL−1, b capture antibody 10 μg mL−1 and tracer antibody 1 μg mL−1, c capture antibody 10 μg mL−1 and detection antibody 5 μg mL−1

Full size image

Different incubation times and temperatures were also examined and the best performance in terms of linearity of the dose–response curve and limit of detection was obtained when each incubation step was performed for 30 min at room temperature (25°C). Longer incubation times did not produce any significant improvement in the performance of the immunoassay.

Detection limit and sensitivity

Figure 2 shows the dose–response curves obtained in optimized assay conditions for the 96-well microtiter plate assay formats using either the colorimetric or CL detection. In both assays there is a good linear correlation between the signal and the concentration of the Cry1Ab protein up to 2 ng mL−1. The limits of detection (LOD) of the methods were calculated by interpolation on the Cry1Ab standard curve, as the analyte concentrations corresponding to the mean signal of blank (absorbance or CL intensity) plus three times its standard deviation. The limits of quantification (LOQ) of the methods were defined using the same criteria, but considering the blank signal plus ten times its standard deviation. The LOD (or LOQ) values corresponded to 0.08 (or 0.26) ng mL−1 and 0.003 (or 0.005) ng mL−1 of Cry1Ab for the colorimetric and the CL immunoassay, respectively. Thus, the detection limit of the CL immunoassay is more than one order of magnitude lower than that of the colorimetric one. This can be ascribed to the higher detectability of the CL signal, which allowed a precise quantitative evaluation of the signal also in the presence of very small amounts of tracer (the higher precision of the CL measurement is evident from the comparison of the standard deviations of the signals shown in the insets of Fig. 2a,b).

Fig. 2
figure 2

Representative dose–response curves for a the colorimetric and b the CL immunoassays for Cry1Ab in the 96-well (▪) and 384-well (▴) format. Equations of the calibration curves were \(y = {\left( {0.768 \pm 0.007} \right)}x - {\left( {0.006 \pm 0.015} \right)}\) (colorimetric immunoassay), \(y = {\left( {259.3 \pm 3.7} \right)}x - {\left( {0.2 \pm 1.8} \right)}\) (CL immunoassay, 96-well microtiter plate format), and \(y = {\left( {109.4 \pm 0.9} \right)}x - {\left( {0.1 \pm 0.4} \right)}\) (CL immunoassay, 384-well microtiter plate format), where y represents the absorbance (in absorbance units) or the CL intensity (in RLU) after subtraction of the blank signal, and x is the concentration of Cry1Ab (in ng mL−1). The insets show enlarged views of the dose–response curves in the Cry1Ab concentration range 0–0.1 ng mL−1. Each point represents the mean value ± SD of four replicate determinations

Full size image

Moreover, the CL immunoassay is faster than the colorimetric one, thus ensuring an improved analytical throughput. In fact, in the colorimetric immunoassay the time required to accumulate enough colored product to achieve adequate absorbance values is about 30–45 min, while the CL signal originates within a few minutes after the addition of the CL HRP substrate [26]. In addition, the CL signal is proportional to the amount of analyte over a wider range of concentration, because the CL measurements are not limited by the physical principle of the absorbance measurements (Beer–Lambert law): potentially, the HRP enzymatic activity can be accurately evaluated by CL over 6 decades of concentration, while colorimetric detection only allows measurement of HRP concentration within 2–3 decades.

Precision and accuracy

Precision and accuracy of the methods were measured by analyzing spiked maize samples. The negative certified reference sample was subjected to extraction, and the extract was then spiked with Cry1Ab protein at three different final concentration values (0.2, 1.0, and 2.0 ng mL−1). The Cry1Ab-fortified samples were analyzed in four replicates and in four independent assays to evaluate the intra- and inter-assay coefficients of variation, respectively. The obtained results are reported in Table 1. The precision of the assays was satisfactory, with coefficients of variation below 10% even in the inter-assay study. The spiked samples were also employed to assess the accuracy of the assays by comparing the theoretical and experimental amounts of Cry1Ab found in the samples. As shown in Table 2, recovery ranged from 83% to 97% over the 0.2–2.0 ng mL−1 Cry1Ab range, without any significant difference between the colorimetric and CL assays.

Table 1 Precision of the colorimetric and CL immunoassays for Cry1Ab protein (values are the mean of four replicates)
Full size table
Table 2 Accuracy of the colorimetric and CL immunoassays for Cry1Ab protein (values are the mean of four independent determinations)
Full size table

Cross-reactivity

As reported by the manufacturer, even if the monoclonal and polyclonal antibodies employed in the immunoassays were raised against the Cry1Ab toxin, they also strongly bind the structurally similar Cry1Ac protein (indeed, Cry1Ab immunoassay kits are usually commercialized as Cry1Ab/Cry1Ac assays), while showing a much lower affinity with other Bt proteins. We have verified the cross-reactivity of the CL immunoassay to Cry1Ac and other Bt endotoxins (Cry1F, Cry2A, and Cry9C) by producing dose–response curves for these proteins. The cross-reactivity to a given Bt protein has been estimated via the comparison of the slope of the calibration curves obtained for this protein and for Cry1Ab (Table 3). As expected, the Cry1Ac protein showed a significant cross-reactivity, with a CL signal even stronger than that of Cry1Ab, while other Bt proteins should not severely interfere in the immunoassay, unless they are present in large amounts. The high cross-reactivity against the Cry1Ac protein suggested that, providing suitable Cry1Ac calibration standards, the CL immunoassay could also be used for quantification of the Cry1Ac toxin with a detection limit similar to that achieved for Cry1Ab.

Table 3 Cross-reactivity against Cry1Ac and other common Bt proteins, evaluated with the 96-well microtiter plate format CL immunoassay
Full size table

Robustness

The stability of the antibody-coated microplates was investigated as follows. The coated plates were stored at 4°C in sealed plastic bags and dose-response curves for Cry1Ab protein were produced immediately and after storage times of either two and four weeks. The comparison of dose-response curves demonstrated that storage of coated microtiter plates for up to four weeks did not significantly alter the performance of the assay in terms of limit of detection and sensitivity (data not shown).

High-throughput chemiluminescent assay

The CL immunoassay for Cry1Ab has also been developed in the high-throughput 384-well microtiter plate format. Because of the smaller volume of the wells, use of the 384-well microtiter plate format permits a significant reduction of the amount of reagents and sample required for the analysis. In addition, the productivity of the method is increased because in this format about 180 samples (i.e., four times the number of samples that can be analyzed in the 96-well microtiter plate based assay) can be analyzed in duplicate in the same analytical session. On the other hand, due to the lower sample amount, highly sensitive detection systems, such as CL, are required to achieve a low detection limit.

No further optimization was required for the development of the CL immunoassay in the 384-well microtiter plate format. The same reactant concentrations and experimental procedure used in the 96-well microtiter plate format CL assay were employed, except that reagents and sample volumes were reduced by a factor of four, according to the smaller volume of the wells. A comparison of the dose–response curve obtained for the CL immunoassay in the two microtiter plate formats is shown in Fig. 2b. As expected, the CL immunoassay in the 384-well microtiter plate format is characterized by a lower sensitivity due to the reduced amount of sample employed in the assay. However, in spite of the reduced intensity of the CL signal the detection and quantification limits of the assay (0.005 and 0.016 ng mL−1 Cry1Ab, respectively) remained lower than those of the colorimetric assay. Regarding precision and accuracy, the high-throughput CL assay showed slightly lower intra-assay and inter-assay precision (the coefficients of variation were up to 15%), whereas its accuracy was comparable to that of the 96-well microtiter plate assays (Tables 1 and 2). However, the precision of the assay can still be considered satisfactory for a screening assay.

Method validation

The methods were validated by analyzing commercially available certified reference samples. Each sample was subjected to four replicate extractions following the procedure described in the “Experimental” section and analyzed. The results, reported in Fig. 3a,b, showed that the CL immunoassays were able to unambiguously detect the presence of the Cry1Ab protein in the certified reference sample containing 0.1% MON 810, whereas the colorimetric method only allowed detection of Cry1Ab in the 1% MON 810 sample. Most interestingly, both the 96- and 384-well microtiter plate CL immunoassays showed the same discriminating ability, in spite of the reduced amount of sample employed in the latter assay.

Fig. 3
figure 3

Analysis of MON 810 maize certified reference samples using a the colorimetric and b the CL immunoassays for Cry1Ab in the 96-well CL (light gray) and the 384-well (dark gray) format. Each bar represents the mean signal value ± SD, after subtraction of the blank signal, of four determinations performed upon separate sample extractions. The dashed lines represent the signals corresponding to the limit of detection (defined as three times the standard deviation of the blank signal) for each immunoassay format

Full size image

It should be noted that even if the certified reference samples used for assay validation were sold as powders, the field-testing of maize samples will require a preliminary grinding step, which could be performed manually or using suitable mechanical grinding devices.

Conclusions

The developed CL sandwich enzyme immunoassay allowed analysis of Cry1Ab protein in maize samples after simple extraction procedure and fulfilled the standard requirements of precision and accuracy. Use of CL detection presented undoubted advantages in terms of detection limit and rapidity of the assay with respect to conventional colorimetric techniques. Thanks to the high detectability of the CL signal, the detection limit of the 96-well microtiter plate CL immunoassay is 3 pg mL−1 Cry1Ab, which is more than one order of magnitude lower than that obtained with the colorimetric immunoassay. This detection limit is also comparable or lower than those reported for previously developed colorimetric Cry1Ab immunoassays. In addition, CL detection speeds up the assay procedure because no incubation time is required for the development of the CL signal. The advantages of the CL detection are clearly exploited in the analysis of certified reference materials: the CL immunoassay allows detection of Cry1Ab in maize samples containing 0.1% MON 810, whereas the colorimetric immunoassay only detected 1% MON 810. The 384-well microtiter plate CL immunoassay showed analytical performance similar to that of the 96-well microtiter plate assay (detection limit 5 pg mL−1 Cry1Ab), while allowing a significant increase in the analytical throughput and a reduction in the consumption of samples and reagents. This simple and sensitive CL enzyme immunoassay for Cry1Ab protein could thus represent a convenient, low-cost screening method for detecting the presence of MON 810 maize (or other GMOs expressing the Cry1Ab or Cry1Ac proteins) in food below the actual regulation levels. In addition, this assay may also be suitable for field testing: thanks to the availability of sensitive single-tube or microtiter strip portable luminometers, the development of a CL immunoassay kit for on-site application, including pre-coated assay tubes or microtiter strips and ready-to-use reagents, could be envisaged. Further improvements in the sensitivity and rapidity of the immunoassay could derive from the use of dispersed supporting solid phases, such as micro- or nanospheres, which will increase the surface area available for antibody immobilization, thus allowing the binding of a higher amount of Cry1Ab toxin, and lead to faster reaction kinetics, thus reducing the overall assay time.