Soil samples were collected from within and outside six fields where insect-resistant transgenic cotton (Bollgard) encoding the Bacillus thuringiensis Berliner (Bt) subsp. kurstaki cry1Ac gene had been grown and subsequently incorporated into soil by postharvest tillage for 3–6 consecutive years. The level of Cry1Ac protein in these samples (collected 3 mo after the last season’s tillage) was evaluated using both enzyme-linked immunosorbent assays (ELISA) and bioassays with a susceptible insect species, Heliothis virescens (F.), the tobacco budworm. Both methods revealed that no detectable Cry1Ac protein was present in any of the soil samples collected from within or outside the Bollgard fields. Based on the results from reference standards, the limit of detection for the ELISA was 3.68 ng of extractable protein per gram of soil, and that of the bioassay (measured by EC50) was 8 ng of biologically active protein per gram of soil. Together, these findings demonstrate that the amount of Cry1Ac protein accumulated as a result of continuous use of transgenic Bt cotton, and subsequent incorporation of plant residues into the soil by postharvest tillage, is extremely low and does not result in detectable biological activity.
Advancements in genetic engineering technology have enabled the introduction and expression of novel genes in plants to produce agronomically useful traits such as insect and disease resistance. Several species of crop plants (including cotton, corn, and potato) have been genetically modified to express genes of various subspecies of Bacillus thuringiensis Berliner (Bt) encoding insecticidal proteins (δ-endotoxins) (see Lewellyn et al. 1994, Persley 1996, Federici 1998). These insecticidal proteins confer protection to the plant from damage by insect herbivores. The improved delivery system of Bt toxins through transgenic plants has great potential to reduce the reliance on traditional chemical insecticides in insect pest control programs (e.g., Hoffmann et al. 1992). In addition, because Bt proteins are highly specific in their effects, the transgenic crop plants producing these proteins have advantages over broad-spectrum pesticides in facilitating integration of other environmentally benign pest control strategies such as biological control into integrated pest management (IPM) programs (e.g., Bolin et al. 1996, Mascarenhas and Luttrell 1997, Orr and Landis 1997, Schuler et al. 1999, Reed et al. 2001).
However, concerns have been raised over the environmental risks associated with the large-scale commercial release of transgenic crops (see review in Williamson 1992). These environmental concerns include (1) risk of enhanced selection pressure for Bt-resistance in target insect pests, and (2) possible impacts on nontarget organisms that are of ecological and/or economic interest. One important aspect to consider in evaluating these concerns is the possible accumulation and persistence of the plant-produced Bt proteins in soils where the crops are repeatedly grown and residues of the crop plants are incorporated into the soil by tillage or as litter.
Recent laboratory studies (Venkateswerlu and Stotzky 1992, Tapp et al. 1994, Tapp and Stotzky 1995, Crecchio and Stotzky 1998) have shown that insecticidal Cry proteins from B. thuringiensis subsp. kurstaki and subsp. tenebrionis are readily adsorbed at equilibrium and bound to clay minerals and humic acids, and that their insecticidal activity is maintained or enhanced in the soil-toxin complexes. Crecchio and Stotzky (1998) also showed that under laboratory conditions, insecticidal protein (Cry1Ac) from B. thuringiensis subsp. kurstaki, which had bound to soil humic acids degraded more slowly than free protein. Based on those laboratory findings, these researchers hypothesized that incorporation of Bt proteins into soil from repeated large-scale use of transgenic crop plants could exceed the rate of natural degradation and inactivation, thereby leading to an accumulation of the protein in the soil that could reach biologically active levels.
To date, however, this hypothesis has not been evaluated under field conditions. No studies have examined the levels of Bt protein in soils of agricultural fields where transgenic Bt crops have been repeatedly planted and residues of the crops incorporated into the soil. In the current study, we use both insect bioassays and enzyme-linked immunosorbent assays (ELISA) to evaluate levels of Cry1Ac protein in soils where transgenic Bt cotton producing Cry1Ac protein had been continuously grown, and subsequently incorporated into soil by postharvest tillage, for 3-6 yr.
Materials and Methods
Soil samples were collected from six different sites where Bollgard cotton had been planted in each year for the previous 3-6 consecutive years. The history of Bollgard cotton growing at each sampling site is summarized in Table 1. In each year, all the Bollgard fields were cultivated one to two times during the growing season, and the plant stalks were first shredded with motor-driven shredders and then tilled into the soil using disk plows at the end of each season immediately following harvest of the cotton lint and seeds. All soil samples were collected in February 1999, 3 mo after the last season's tillage. At the time of sampling, a majority of plant residues incorporated into the soil in the previous seasons with shredders and disk plowings (≈7.6 cm deep) had mostly decayed; thus, all the soil core samples collected at this time contained fully decayed plant residues, as well as some partially decayed plant residues.
At each of the sampling sites, three core samples, each 15.2 cm deep and 7.6 cm in diameter, were randomly taken from the Bollgard cotton field using a bulb setter. The distance between each core sample within the same Bollgard field ranged from 30.5 to 45.7 m. One soil sample also was taken from a point ≈6.1 m outside of the Bt cotton field at each site and served as a negative control. These negative control sites were either noncultivated areas or cultivated field plots where no Bt crops had been planted. All core samples were placed into glass jars, which were then placed on dry ice and immediately sent to Monsanto (St. Louis, MO) for evaluation. Although the pH (negative logarithm of hydrogen iron concentration) of the soil samples was not measured, all of the samples were collected from cotton fields in the main cotton belt of the United States of America and the pH of these samples should be within the normal range for cotton growing regions.
Upon receipt at Monsanto, all soil samples were stored at -80°C before any analysis. Before analysis by ELISA and insect bioassay, each soil core sample was ground using a mortar and pestle and then homogenized using an electric motor driven homogenizer (at 10,000-40,000 revolutions per minute) for at least 15 min. These procedures allowed all components (such as soil particles and decayed plant residues) of each core sample to be fully mixed and homogenized for ELISA and insect bioassay.
Two different methods (ELISA and bioassays with a susceptible insect species) were used to determine the level of Cry1Ac protein in soil samples. While ELISA allows the quantification of both functional and nonfunctional Cry1Ac protein in soil based on levels of protein binding to an antibody, bioassays allow assessment of the level of bio-active protein in the soil. In addition, the efficiency of detection by ELISA depends on the amount of the Cry1Ac protein that can be extracted from the soil, whereas the level of biological activity detected by bioassay is determined by the insect's ability to extract Cry1Ac protein from the soil (Tapp and Stotzky 1995).
The ELISA used in this study involved two stages: protein extraction and quantification. To extract Cry1Ac protein from the soil samples, the buffer solution described by Palm et al. (1994), consisting of 50 milimolar sodium borate (pH = 10.5), 0.75 molar potassium chloride, 0.075% polyoxyethylenesorbitan monolaurate (Tween 20), and 10 mM ascorbic acid, was used. The potassium chloride and Tween 20 were used in the buffer solution to maximize extraction efficiency by discouraging potential ionic and hydrophobic interactions between soil particles and Cry1Ac protein (Palm et al. 1994). Soil samples were mixed with the buffer solution at a ratio of 1:2, and homogenized for 15 s using a Brinkmann Polytron PT3000 (Brinkmann Instruments, Westbury, NY). Immediately after homogenization, the soil-buffer mixture was centrifuged at 16,000 × g for 10 min. After centrifugation, an aliquot was removed from the supernatant and analyzed using a double-antibody sandwich ELISA as described by Fuchs et al. (1990) and Palm et al. (1994). The amount of soil extracted from each core sample for analysis ranged from 0.5052 to 0.6112 g.
For quantification of Cry1Ac protein in test samples, the wells of a microtiter plate (F-96 MaxiSorb, Nunc, Roskilde, Denmark) were first coated with a primary monoclonal anti-Cry1Ac antibody (Monsanto, lot number M19N4A6). A secondary polyclonal rabbit anti-Cry1Ac antibody (Monsanto, lot number MR129) and the test sample (aliquot extract) then were added to the plate, and the plate was incubated for 1 h at 37°C. The reaction volume was 250 μl for each well. Extracts were diluted at least 10 times with 0.5% (weight/volume) ovalbumin phosphate buffer solution to minimize potential matrix effects. After the 1-h incubation, the plate was washed using the phosphate buffer solution and then treated with a donkey anti-rabbit alkaline phosphate conjugate antibody. After the treatment, the plate was incubated again for 1 h and subsequently washed with the phosphate buffer saline. The plate then was treated with p-nitrophenyl phosphate, incubated at room temperature for about 30 min, and then analyzed by spectrophotometry at 405 nm, with a reference wavelength of 655 nm, in a Bio-Rad model 3350 microtiter plate reader (Bio-Rad, Hercules, CA). For purposes of quantification, a seven-point standard curve comprising purified Cry1Ac standard protein was included in each microplate. Absorbance readings and Cry1Ac standard protein concentrations were logarithmically transformed and fitted with a quadratic regression model for extrapolation and estimation of Cry1Ac protein levels in soils taken from within and outside of Bt cotton fields. The limit of detection was calculated as the mean background Cry1Ac protein level in soils from non-Bt control samples (two assays for each of the six soil samples) plus three standard deviations.
The Cry1Ac protein standard (lot number: 11498 BR 1098) employed in this study was produced at Monsanto using the Bacillus thuringiensis spp. kurstaki HD-73 strain according to procedures described in MacIntosh et al. (1990). Purified protein (containing 94% Cry1Ac) was trypsinized and solubilized in 100 mM sodium carbonate buffer, pH 10.5, before use in different assays. To determine the efficiency of Cry1Ac protein extraction from different soil samples, soil from non-Bt control samples was spiked with the Cry 1Ac standard protein at a rate of 100 ng/g of soil (ppb), and then analyzed using the ELISA procedures described above.
The larvae of tobacco budworm, Heliothis virescens (F.), were used in the bioassay because of their high susceptibility to Cry1Ac protein (Hofte and Whitely 1989; Sims and Holden 1996). Eggs of H. virescens were purchased from Ecogen (Langhorne, PA) and shipped to Monsanto by express mail in a Styrofoam cooler that contained cooling packs inside. Upon receipt of the shipment, eggs were immediately placed in a ventilated plastic box, which was then placed in a growth chamber (Percival, Boone, IA) with controlled climatic conditions (27 ± 1.5°C, 50 ± 10% RH) for incubation.
To assay for Cry1Ac protein, soil samples were incorporated into the artificial diet, and then presented to first-instar H. virescens. The bioassay procedure was similar to that previously described by Sims and Holden (1996). One gram of soil from each core sample was thoroughly slurried with 4 ml water in a 50-ml centrifuge tube on a Vortex Mixer, and then mixed with agar-based liquid diet to bring it up to a total volume of 20 ml. The diet used in the bioassay was a standard soybean-based multiple species lepidopteran diet (King and Hartley 1992) and was purchased from Southland Products (Lake Village, AR). The soil-diet mixture for each soil sample was added to 16 cells of a bioassay tray (C-D International, Pitman, NJ) at a rate of 1 ml per cell using a repeater pipettor. Each bioassay tray contained a total of 128 cells so that eight different test samples (each with 16 cells) were run simultaneously on each tray. After the soil-diet mixture cooled and solidified, one first-instar H. virescens (12-24 h old) was introduced into each of the cells of the bioassay tray using a camel's hair brush. The bioassay tray was sealed with vented transparent acetate covers, and placed in a growth chamber at 27°C and 30-40% RH for 7 d. After the 7-d incubation, survival and mass of H. virescens larvae for each test sample were determined.
To serve as a reference standard, non-Bt soil samples (from Loxley Agricultural Center, Loxley, AL) were spiked with a series of 10-fold dilutions of pure Cry1Ac protein. The series of concentrations used as a reference standard were prepared by thoroughly mixing 1 g of the non-Bt soil with 2 × 101, 2 × 102, 2 × 103, 2 × 104, or 2 × 105 ng Cry1Ac in 4 ml water, which then was mixed with agar-based liquid H. virescens diet to bring it up to a total volume of 20 ml. The final concentrations of Cry 1Ac in the soil-diet mixture were 1 × 100, 1 × 101, 1 × 102, 1 × 103, and 1 × 104 ng per ml of the mixture. The soil-diet mixtures for the reference standard were assayed with H. virescens using the procedures described previously for test soil samples. Non-Bt soil collected from the same site and not spiked with Cry1Ac protein was used as a negative control.
Analysis of variance (ANOVA) was used to analyze for differences in survival and mass of H. virescens exposed to soil samples collected within and outside the Bt cotton fields at different locations. Because only one soil sample was collected from outside the Bt field at each of the sites, the interaction between sample site and Bt treatment was not examined. All statistical analyses were performed with JMP Statistical Discovery Software (SAS Institute 1995). Probit analysis was used to analyze the dose-response relationships between the standard Cry1Ac concentrations spiked in soil and the mortality of the test H. virescens larvae (SAS Institute 2000), while a three-parameter logistic regression model (Sims and Holden 1996) was used to analyze the dose-response relationships between the standard Cry1Ac concentrations spiked in the soil and the mass of surviving H. virescens (SAS Institute 1995).
Quantification by ELISA.
No Cry1Ac protein was detected by ELISA in any of the soil samples collected either within or outside the Bollgard fields at the six sites. When soil from outside the Bt field (i.e., non-Bt treatment) at each site was spiked with a known amount (100 ng) of pure Cry1Ac protein, a mean (±SD) of 31.76% (±6.72) of the originally spiked Cry1Ac protein was detected by ELISA (Table 2). Additional tests with the negative control samples further established that the mean (±SD) background level of Cry1Ac protein in the six non-Bt soil samples was 2.36 (±0.44) ng of extractable protein per gram soil. The limit of detection for Cry1Ac protein in soil based on three standard deviations was 3.68 ng of extractable protein per gram of soil. Based on the average extraction (or recovery) rate of 31.76% (Table 2), the level of Cry 1Ac protein that could be detected by ELISA would be ≈11.57 ng/g of soil.
Biological Activity Determined by Bioassay.
The dose-response relationships between the standard concentration of Cry1Ac protein spiked in soil and mortality and growth of H. virescens larvae are presented in Fig. 1 (A and B). Probit analysis of dose-mortality relationship indicates that the concentration of Cry1Ac protein spiked into soil required to kill 50% of H. virescens larvae (LC50) was 755 ng/g of soil (with the 95% confidence interval being from 312 to 1,863 ng/g). Based on the logistic regression model, the concentration required to reduce larval growth by 50% (EC50) (relative to the negative control) was estimated to be 8 ng/g of the soil (with the 95% confidence interval being from 7 to 9 ng/g). Thus, any concentrations of Cry1Ac protein in soil at or above 8 ng/g of soil (corresponding to 0.4 ng/ml of diet of soil mixture) should be effectively detected by the H. virescens bioassay.
For test samples, the mean survival (±SE) (across all sample sites) was 97.7% (±0.20) for soil samples collected from Bt cotton fields, and 98.0% (±0.30) for samples from outside the Bt cotton fields (Fig. 2A). The mean mass (±SE) of surviving H. virescens larvae was 107.8 mg (±2.23) and 99.8 mg (±4.23) for soil samples collected from within and outside Bt cotton fields, respectively (Fig. 2B). ANOVA detected no significant differences in either survival or mass of H. virescens larvae exposed to soil samples from within and outside Bt cotton fields, nor any significant differences among different sampling sites (Tables 3 and 4 and 4). If anything, there was a nonsignificant trend toward larger masses for larvae exposed to the soil samples from within the Bt cotton fields. When the dose-response relationships between the standard Cry1Ac protein concentrations spiked in soil and mortality or growth (mass) of H. virescens larvae are considered (Fig. 1 A and B), these results indicate that levels of Cry1Ac protein in the soil from Bt cotton fields are not high enough to be biologically active against even a highly susceptible insect, H. virescens.
Square root transformation was applied to data prior to analysis.
Because the non-Bt soil samples assayed in this study were collected from near (6.1 m away) Bt cotton fields, one might argue that the non-Bt soil samples could be contaminated with Cry1Ac proteins from the Bt cotton fields. However, mean larval survival (97.7-98.0%) and mass (98.8-107.8 mg) of H. virescens larvae assayed for both non-Bt and Bt soil treatments in this study were within the normal ranges for the survival (95-100%) and growth (94-157 mg) of 7-d old H. virescens larvae established previously using the same artificial diets with no soil or Bt protein present (J.J.D. and J.W.M., Monsanto, unpublished data).
Results from both ELISA and insect bioassay indicated that repeated agricultural use of transgenic Bt cotton (Bollgard) expressing Cry1Ac protein did not result in detectable levels of Cry1Ac protein in soil. The amount of Cry1Ac protein, if any, accumulated from 3 to 6 yr of repeated use of the transgenic cotton and subsequent incorporation of the plant residues into the soil by tillage was below the limit of detection for both ELISA and bioassay. Thus, these findings do not support the assertions that repeated and large-scale use of transgenic crop plants could lead to accumulation and/or persistence of Bt proteins in soil at levels that would result in biological impacts (Tapp and Stotzky 1995, Crecchio and Stotzky 1998).
The limits of Cry1Ac detection by both ELISA and bioassay used in this study were determined with soil samples spiked with pure Cry1Ac protein, and were not validated with transgenic Bt cotton plant residues. However, previous studies by Palm et al. (1994,1996) demonstrated that the protein extraction procedure and ELISA (comparable to that used here) worked effectively to detect Cry1Ac either spiked in soil or as a component of transgenic Bt plant tissues incorporated into soil. The efficiency of Cry1Ac extraction reported by Palm et al. (1994) ranged from 27 to 60.2% for protein spiked in different types of soil and 17-70% for protein as a component of different cultivars of transgenic Bt cotton leaf tissues. In the study reported here, the rate of pure Cry1Ac protein recovered from spiked soil samples was 24.34-38.89% with a mean of 31.76% (Table 2). Based on this range of extraction efficiencies and the detection limit of 3.68 ng extractable protein per gram of soil, the level of Cry1Ac protein that could have been detected by ELISA in soil samples collected from the six different field sites ranged from 9.46 to 15.14 ng/g of soil (with a mean of 11.57 ng/g).
A previous study (MacIntosh et al. 1990) showed that H. virescens larvae were extremely susceptible to pure Cry1Ac incorporated in the larval diet, and the EC50 against H. virescens larvae was 1.3 ng of Cry1Ac protein per ml of insect diet. The bioassay in the study reported here showed that the EC50 for pure Cry1Ac spiked in soil was 8 ng/g of soil (corresponding to 0.4 ng of Cry1Ac protein per milliliter of diet-soil mixture). These findings indicate that the H. virescens bioassay is highly sensitive in detecting the biological activity of Cry1Ac protein incorporated with their diet at the EC50 level. Although the efficiency of Cry1Ac extraction by H. virescens larvae from various diet mixtures has never been quantified partly because of the technical difficulty, there have been no published studies suggesting that H. virescens larvae would differentially extract Cry1Ac protein from spiked soil versus transgenic Bt plant materials when those materials are incorporated into the larval diet. In fact, previous studies strongly suggested that lepidopteran larvae (including H. virescens) show comparable responses to pure Bt proteins spiked in soil (MacIntosh et al. 1990, Sims and Holden 1996, Crecchio and Stotzky 1998) and Bt protein as a component of transgenic Bt plant tissues (Ream et al. 1994).
Based on an average expression rate of 20 μg of Cry1Ac protein per gram of dry Bollgard cotton plant tissue (Kollwyck and Hamilton 1999), a conservative estimate of the level of Cry1Ac protein added to the top three inches of soil in Bollgard fields (with 60,000 plants per acre, each plant ≈250 g dry weight) for each growing season would be ≈650 ng/g of dry soil. If all of the Bt protein accumulated in the soil without any degradation and/or inactivation, the level of Cry1Ac protein after 6 yr (seasons) of continuously planting Bollgard cotton would be 252 and 195 times higher than the limits of detection by ELISA and bioassay used in this study, respectively. Furthermore, this estimate does not include the amount of the Cry1Ac protein (if any) that may be released into soil by growing plants through root exudates. Recently, Saxena et al. (1999) reported that transgenic corn plants expressing the cry1Ab gene release Cry1Ab protein into soil through root exudation. No matter what the source of Cry1Ac, because the soil samples in the study reported here were taken 3 mo after the last growing season, the absence of detectable Cry1Ac protein in all of the soil samples suggest both little or no accumulation over years and rapid breakdown after each season (i.e., reduced to undetectable levels in 3 mo). Future studies should examine Cry1Ac protein dissipation in Bt crop fields at different times in the growing season, as well as immediately after incorporation of plant residues into the soil after harvest. Such studies will assist us in assessing the potential impact of Bt plant residues on soil dwelling nontarget organisms.
The persistence of Bt proteins in the environment is a function primarily of (1) the concentration added, and (2) the rate of inactivation and degradation by both biotic and abiotic factors; when the rate of addition is faster than the rate of inactivation and/or degradation, the toxin accumulates (Venkateswerlu and Stotzky 1992, Tapp and Stotzky 1995, Crecchio and Stotzky 1998). Palm et al. (1996) showed that, when incubated in soil under laboratory conditions, Cry1Ac protein contained in transgenic Bt cotton plant residues degraded in soil at a rate similar to or greater than pure Cry1Ac proteins. Using insect bioassays, Ream et al. (1994) estimated that, at room temperature (24°C), the half-life of Cry1Ac protein was 9.3-20.2 d when spiked into silt loam soil, and 41 d when added to soil as transgenic Bt cotton plant tissues. Studies of the environmental fate of other Bt proteins, including those used in transgenic Bt potato and Bt corn (Cry3Aa and Cry 1Ab), have found that the half-lives of these proteins in soil are generally less than 20 d (Ream et al. 1994, Sims and Holden 1996, Palm et al. 1996). Although degradation of Bt Cry protein in soil does not follow a first order process and the rate of degradation could become slower as the amount of the Cry protein becomes less in soil (e.g., Palm et al. 1996), results from the current study demonstrate that the amount of Cry1Ac protein accumulated as a result of the continuous use of transgenic Bt cotton and subsequent incorporation of the plant residues into the soil by postharvest tillage in multiple seasons, does not result in detectable immunological and biological activity. The apparently rapid breakdown of other Bt proteins in soil, at rates comparable to that measured for Cry1Ac (Ream et al. 1994, Palm et al. 1996, Sims and Holden 1996), suggests that these proteins also will not be accumulating at biologically significant levels.
We thank Gregg Dixon, Zachary Shappley, and Alan Weir for assistance in providing us soil samples. Tom Nickson and Mike Mckee (Ecological Technology Center, Monsanto) provided helpful comments on an earlier draft of the manuscript. Changjian Jiang (of the Environmental Science Center, Monsanto) provided assistance in statistical analysis.