Fate of Bacillus thuringiensis subsp. israelensis in the Field: Evidence for Spore Recycling and Differential Persistence of Toxins in Leaf Litter

Abstract

Bacillus thuringiensis subsp. israelensis is a bioinsecticide increasingly used worldwide for mosquito control. Despite its apparent low level of persistence in the field due to the rapid loss of its insecticidal activity, an increasing number of studies suggested that the recycling of B. thuringiensis subsp. israelensis can occur under specific, unknown conditions. Decaying leaf litters sampled in mosquito breeding sites in the French Rhône-Alpes region several months after a treatment were shown to exhibit a high level of larval toxicity and contained large amounts of spores. In the present article, we show that the high concentration of toxins found in these litters is consistent with spore recycling in the field, which gave rise to the production of new crystal toxins. Furthermore, in these toxic leaf litter samples, Cry4Aa and Cry4Ba toxins became the major toxins instead of Cyt1Aa in the commercial mixture. In a microcosm experiment performed in the laboratory, we also demonstrated that the toxins, when added in their crystal form to nontoxic leaf litter, exhibited patterns of differential persistence consistent with the proportions of toxins observed in the field-collected toxic leaf litter samples (Cry4 > Cry11 > Cyt). These results give strong evidence that B. thuringiensis subsp. israelensis recycled in specific breeding sites containing leaf litters, and one would be justified in asking whether mosquitoes can become resistant when exposed to field-persistent B. thuringiensis subsp. israelensis for several generations.

INTRODUCTION

Bacillus thuringiensis-based insecticides are often considered environmentally safe and efficient alternatives to chemical insecticides for the control of pest populations, including lepidopteran, coleopteran, and mosquito larvae (36, 37). The modes of action of some B. thuringiensis subspecies together with their fate in the environment have been largely studied (6, 30, 45). B. thuringiensis is a natural larval pathogen (bacterium) that produces a toxic crystal during its sporulation (34), and its fate in the field can be monitored by using three different approaches. First, its insecticidal activity can be tested by performing bioassays using environmental samples (i.e., water, soil, or leaf litter samples, etc.) (9, 38, 44). Unfortunately, the sensitivity of this approach is often too low to detect slight modifications of toxicity, as B. thuringiensis usually does not exhibit a high level of residual activity (36). Second, spores, which usually persist longer than the larvicidal activity, can be detected in field samples by culturing them on petri dishes (18, 46) or by PCR and quantitative PCR, allowing the detection of potential recycling events (12, 17). Finally, the fate of Cry toxins, which are mainly responsible for the toxicity of B. thuringiensis, has often been studied by using enzyme-linked immunosorbent assays (ELISAs) (1, 29). Although it is widely used, such an approach does not guarantee that the native full protein is present, because antibodies react only with a specific part of the target protein. Nevertheless, focusing on toxins is the only approach that can confirm if the recycling of spores, detected by quantitative PCR or by plating, is combined with crystal production or not. Only the combination of these three approaches allows the precise monitoring of the persistence of B. thuringiensis in the field.

Bacillus thuringiensis subsp. israelensis, whose crystal is constituted of four main toxins (Cry4Aa, Cry4Ba, Cry11Aa, and Cyt1Aa), is being increasingly used worldwide for mosquito control due to its high specificity in targeting dipteran insects and due to the absence of firm cases of resistance detected in the field so far (24). As its larvicidal activity often quickly decreases after spraying, B. thuringiensis subsp. israelensis is often said to be low-level persistent in the field (24, 38). Even though spores have already been described to germinate and proliferate in mosquito cadavers (2, 23), they are not believed to be able to proliferate in the environment and to produce a large enough amount of crystals to kill mosquitoes (4). Nevertheless, an increasing number of studies suggested that B. thuringiensis subsp. israelensis can persist and even recycle (i.e., germination, vegetative growth, sporulation, and the production of toxins) under specific conditions in the environment (4, 11). Moreover, decaying leaf litter sampled in mosquito breeding sites several months after a treatment exhibited a high level of toxicity against mosquito larvae (7, 8) and contained large amounts of spores (43), suggesting a higher level of persistence than expected and a potential recycling of spores, leading to an unexpected high level of toxicity. The selection of an Aedes aegypti laboratory strain for 30 generations with these litter samples led to moderate resistance to B. thuringiensis subsp. israelensis (3.5-fold) but higher levels of resistance to separate Cry toxins (up to 60-fold) (41). This strain provides evidence that mosquitoes can become resistant when exposed to field-persistent B. thuringiensis subsp. israelensis. It is therefore essential to characterize the behavior of B. thuringiensis subsp. israelensis toxins in the field and to precisely quantify the concentrations of these toxins in toxic leaf litter.

Only a few studies have investigated the fate of toxins of B. thuringiensis subsp. israelensis in the field (4, 13). Contrary to other B. thuringiensis subspecies, for which toxins are studied mainly in the context of genetically modified crops, which produce large amounts of a unique soluble Cry toxin (20, 29, 40), B. thuringiensis subsp. israelensis is sprayed as a suspension of spores and crystals. The crystals are constituted of a mixture of four toxins often found at very low concentrations in the environment even a few hours after a treatment, rendering them difficult to detect (24). This may explain why, while numerous ELISAs exist for a large range of different Cry toxins, no efficient tool is available so far to extract and detect B. thuringiensis subsp. israelensis toxins in field samples. Furthermore, most previous studies investigated the interaction between Cry toxins and soil, but there have been few studies on the behavior of Cry and Cyt toxins in leaf litter under aquatic conditions (42), which is one of the main constituents of several mosquito breeding sites and the main food source for larvae.

In the present article, we used a recently developed and patented extraction protocol followed by detection using an ELISA targeting each toxin of B. thuringiensis subsp. israelensis to determine the concentrations of the toxins and their relative proportions in several field-collected leaf litter samples (42). We then applied this test to leaf litter samples contaminated in the laboratory with different amounts of toxins in their crystal form to better understand how long toxins can persist in leaf litter and if they have different behaviors. We also tested whether the drying/watering episodes, often experienced by the mosquito breeding sites, have an impact on the persistence of toxins. All these results are discussed in regard to B. thuringiensis subsp. israelensis recycling in the field, treatment strategies, and resistance management in the environment.

MATERIALS AND METHODS

Leaf litter sampling.

All the leaf litter samples were taken from mosquito breeding sites in the French Rhône-Alpes region. Toxic leaf litter samples were collected in 2008 at four different sites 4 months after a treatment with B. thuringiensis subsp. israelensis (Vectobac WG) (3,500 international toxic units [ITU] · mg⁻¹) at the beginning of spring (43). Two samples (samples P23 and P26) were constituted mainly of Populus nigra leaves, and two other samples (samples Aul1 and Aul2) were constituted mainly of Alnus glutinosa leaves. Litter samples were left to dry at room temperature, powdered, and stored at −20°C until the extraction of toxins and ELISA quantification were performed.

Leaf litter samples used for further contaminations with B. thuringiensis subsp. israelensis toxins were collected in 2011 in mosquito breeding sites never treated with the bioinsecticide. These samples were constituted mainly of Quercus sp. leaves and some Alnus glutinosa leaves (42). Litter samples were left to dry and were stored at room temperature until the toxins were added for the experiments. The samples were tested by ELISAs prior to contamination, and no signal was recorded, indicating that no indigenous B. thuringiensis subsp. israelensis was present.

Production of individual B. thuringiensis subsp. israelensis toxins.

A crystal-negative strain of B. thuringiensis subsp. israelensis (4Q2-81) transformed with plasmids pHT606, pHT618, pWF53, and pWF45 for Cry4Aa, Cry4Ba, Cry11Aa, and Cyt1Aa, respectively, was used to produce each protoxin separately (10, 32, 49). Spores and crystal suspensions were produced as previously described (41, 42) and conserved in distilled water at −20°C until use.

Contamination of leaf litter samples and parameters tested.

Ten grams of leaf litter samples from untreated sites was put into crystallizing glass dishes (capacity, 300 ml) containing 60 ml of tap water. Litter samples were contaminated with two different concentrations of one of the four toxins, a low dose (LD) and a 4-fold-higher concentrated dose (high dose [HD]). For each dose and each toxin, two conditions of watering/drying were also tested: either the water level was maintained by adding 4 ml of water every day, or the litter was left to dry until it was totally dried after 17 days. For each condition, each dose, and each toxin, the litter was sampled after six durations of exposure: immediately after toxin addition (T₀), 72 h, 7 days, 14 days, 1 month, and 2 months. For each duration of exposure, a negative control not exposed to toxins was also sampled. After sampling, litter samples were left to dry overnight in order to powder them efficiently, a necessary step for the subsequent reproducible toxin extraction, and were stored at −20°C until use. All the conditions tested were performed in triplicate. Litter samples were left at 27°C with 80% relative humidity with a 14 h–10 h light/dark period.

Toxin extraction from leaf litter samples and detection by ELISA.

For the quantification of toxins extracted from leaf litter samples, calibration curves were performed by using purified toxins. Crystal toxins were solubilized by using an alkaline buffer (pH 10.8) containing Na₂CO₃ (50 mM) and dithiothreitol (DTT) (10 mM) and were incubated at 60°C for 1 h. Solubilized toxins were then purified by using a HiPrep 16/10 DEAE FF ion-exchange column (Amersham Biosciences) and desalted by using a HiPrep 26/10 desalting column (Amersham Biosciences). Toxins were lyophilized and stored at −20°C until use. For each ELISA, lyophilized toxins were resuspended in distilled water, and their concentrations were determined by a Bradford assay, using bovine serum albumin (BSA) as a standard (5). Two aliquots of each toxin were then used in duplicate as standards for each ELISA at concentrations ranging from 2 to 500 ng/ml.

Toxins were extracted from leaf litter samples by using a previously described extraction procedure (42), adapted and modified from references 19, 20, and 29. The procedure is described in pending French patent no. FR1160365. This protocol is highly reproducible for each toxin and with a large range of concentrations, suggesting that the extraction procedure in itself has no impact on the integrity of the toxins (42). All the extractions were performed in duplicate. After extraction, toxins were detected by sandwich ELISAs using anti-Cry4 (Cry4Aa and Cry4Ba), anti-Cry11 (Cry11Aa), and anti-Cyt (Cyt1Aa) antibodies, as previously described (28, 42). Each extracted sample was tested in triplicate by an ELISA. For each toxin and each dose, the time needed to lose 50% of the ELISA signal detected at T₀ (T_1/2) was calculated by performing an exponential decay analysis using XLSTAT v.2009.4.06 software (Addinsoft).

For the toxic leaf litter samples, the concentration of each toxin was compared to the concentration obtained by testing a commercial B. thuringiensis subsp. israelensis strain at the operational dose (on the basis of 1 kg of formulated product/hectare of B. thuringiensis subsp. israelensis Vectobac WG). Moreover, the proportion of each toxin among all the toxins and among Cry toxins was also calculated. These values were compared to the proportion obtained by testing a solubilized commercial B. thuringiensis subsp. israelensis strain (“B. thuringiensis subsp. israelensis solubilized”) directly and by testing B. thuringiensis subsp. israelensis added to leaf litter for 3 h and extracted according to the protocol described above (“B. thuringiensis subsp. israelensis extracted”) (42). For each toxin or group of toxins, the statistical differences in their proportions between the different conditions/litters were tested by using a Fisher exact test performed by using R 2.8.1 software (33).

RESULTS

The proportions of the different toxins calculated after extraction and quantification by ELISAs were conserved among the four toxic leaf litter samples tested regardless of their leaf compositions (Table 1). In the toxic leaf litter samples, the proportion of the Cyt1Aa toxin decreased by about 75% compared to the proportion of solubilized B. thuringiensis subsp. israelensis tested by ELISAs (Table 1). While a brief exposure of B. thuringiensis subsp. israelensis to leaf litter (“B. thuringiensis subsp. israelensis extracted”) did not induce modifications of the proportions of Cry toxins among them, Cry4 toxins were overrepresented among the Cry toxins and became the main toxin in the field-collected toxic leaf litter samples (Table 1). When the concentrations of toxins in the toxic leaf litter samples were compared to those in leaf litter samples where commercial B. thuringiensis subsp. israelensis had been added, the amount of Cry4 toxins was 26% higher in the field-collected toxic leaf litter samples (calculated as the means of the four toxic leaf litter samples) than in litter samples contaminated in the laboratory with B. thuringiensis subsp. israelensis at the operational dose (37.92 and 30.07 ng/ml, respectively). Conversely, the proportions of Cry11Aa and Cyt1Aa were 18- and 1,557-fold lower in the field-collected toxic leaf litter samples than in litter samples where B. thuringiensis subsp. israelensis had been added, respectively (111.79 and 278.16 ng/ml for Cry11Aa and 1.28 and 13,845.51 ng/ml for Cyt1Aa, respectively).

Table 1.

Proportion of each toxin (Cry4Aa, Cry4Ba, Cry11Aa, and Cyt1Aa) among all the toxins of B. thuringiensis subsp. israelensis or among Cry toxins^a

Sample	Proportion (%) of each toxin among:
	All toxins			Cry toxins
	Cry4Aa and Cry4Ba	Cry11Aa	Cyt1Aa	Cry4Aa and Cry4Ba	Cry11Aa
B. thuringiensis subsp. israelensis solubilizedb	1.0*	8.1*	90.9*	11.1*	88.9*
B. thuringiensis subsp. israelensis extractedc	12.6†	61.7†	26.7†	17.2*	82.8*
P26 extractedd	59.7‡	28.3‡	11.9‡	67.8†	32.2†
P23 extractedd	56.7‡	27.4‡	15.9‡	67.4†	32.6†
Aul1 extractedd	68.9‡	19.1‡	12.0‡	78.3†	21.7†
Aul2 extractedd	55.8‡	24.7‡	19.6‡	69.3†	30.7†

^aFor each toxin or group of toxins, statistical differences (P < 0.05 by Fisher's exact test) of toxin proportions among the conditions are indicated by different symbols (*, †, and ‡). Data obtained for solubilized and extracted B. thuringiensis subsp. israelensis have been reproduced from reference 42.

^bThe proportion of each toxin was determined by ELISAs using a solubilized commercial B. thuringiensis subsp. israelensis strain (Vectobac WG) (3,500 ITU · mg⁻¹).

^cThe proportion of each toxin was determined by ELISAs after toxin extraction from leaf litters where B. thuringiensis subsp. israelensis had been added.

^dThe proportion of each toxin was determined by ELISAs after toxin extraction from toxic leaf litter samples. Samples P23 and P26 correspond to toxic leaf litter samples constituted of mainly Populus nigra leaves, and samples Aul1 and Aul2 correspond to toxic leaf litter samples constituted of mainly Alnus glutinosa leaves.

For each toxin, a rapid decline in the concentration of detectable toxin during the first 14 days of contact with the leaf litter followed by a slower decrease was observed (Fig. 1). The drying of leaf litter samples did not modify the detectability of the toxins, regardless of the dose and the toxin considered (Fig. 1). When toxins were added at higher doses, the concentration of detectable toxins became the same as the concentration detected for lower doses after 7 days (Cry11Aa and Cyt1Aa), 14 days (Cry4Aa), or 1 month (Cry4Ba) of contact with the litter (Fig. 1). Cry11Aa and Cyt1Aa toxins exhibited the highest decrease of detectability, with the T_1/2 being 2.2- to 23-fold lower than those of Cry4Aa and Cry4Ba (Table 2). Even after 2 months of contact, the Cry4Aa, Cry4Ba, and Cry11Aa toxins remained detectable at a basal level (23%, 30%, and 15% of the ELISA signal detected at T₀ when toxins were added to the litter samples at high doses and 34%, 21%, and 29% when toxins were added at low doses, respectively), while the Cyt1Aa toxin was undetectable after only 1 week of contact (Fig. 1).

Fig 1.

Detectability of Cry4Aa (A), Cry4Ba (B), Cry11Aa (C), and Cyt1Aa (D) toxins for the duration of contact with leaf litter, determined by using an ELISA after extraction. Each toxin was inoculated at a low dose (LD) or at a high dose (HD), and the leaf litter was left to dry (W−), or the water level was maintained during the entire incubation period (W+). Standard errors are indicated.

Table 2.

Time to loss of 50% of the ELISA signal detected at T₀ for all the toxins at a low dose or at a high dose in leaf litter samples, and R² values of the corresponding exponential decay curves

Condition	Cry4Aa		Cry4Ba		Cry11Aa		Cyt1Aa
	T1/2 (h)	R2	T1/2 (h)	R2	T1/2 (h)	R2	T1/2 (h)	R2
Low dose	140	0.78	212	0.95	63	0.77	86	0.77
High dose	308	0.83	543	0.87	24	0.93	53	0.95

DISCUSSION

Even if B. thuringiensis subsp. israelensis is usually said to be low-level persistent in the field, an increasing number of studies have shown that it can persist for a long time and suggest that it can recycle under specific conditions in the environment (4, 11). The high toxicity and the large quantity of spores found in the toxic leaf litter samples suggested that such recycling events occurred in mosquito breeding sites (8, 43). Here we demonstrated that the concentration of Cry4 toxins detected in the toxic leaf litter samples by ELISAs was higher than expected after a treatment. The most obvious explanation for this is that a recycling of the spores led to a large amount of spores and toxins in the litters, inducing a high toxicity to mosquito larvae. Further experiments are needed to characterize the specific conditions (physicochemical parameters and the presence of larvae) that allowed B. thuringiensis subsp. israelensis to recycle in the field. The lower concentrations of Cry11Aa and Cyt1Aa toxins observed in toxic leaf litter samples are consistent with the differential persistence of the toxins observed under laboratory conditions, with the Cry4Aa and Cry4Ba toxins being more persistent than Cry11Aa and Cyt1Aa. It has to be noted that the larger the protein, the more it persists: Cry4 toxins (130 kDa) have the highest molecular mass and are more persistent than Cry11Aa (70 kDa) and Cyt1Aa (28 kDa). This could be due to differences in their solubilities in water and/or to their degradability by organisms naturally present in the leaf litters. The parameters increasing or decreasing the persistence of each toxin (pH, salinity, and temperature) will be investigated to better understand their fate in the field.

Cry4 toxins became the main toxins among all the toxins present in the toxic leaf litter samples, together with a dramatic decrease in the concentration of Cyt toxins, which are known to delay the development of resistance in mosquitoes (16, 47, 48). This may explain why the LiTOX strain, selected with the toxic leaf litters, exhibited a high level of resistance to the Cry4Aa toxin (up to 60-fold) compared to the Cry4Ba and Cry11Aa toxins (about 9-fold) (41). As no antibody yet exists to distinguish between the Cry4Aa and Cry4Ba toxins due to their highly conserved amino acid sequences, it is impossible to know whether only Cry4Aa was overrepresented in the toxic leaf litter samples or if both Cry4Aa and Cry4Ba were dominant among all the toxins. The development of more specific antibodies is needed to characterize more precisely the toxin proportions in toxic leaf litter and to better understand the selective pressures exerted on mosquitoes in the field.

When toxins were added to leaf litter samples, a nonlinear decrease in the detectability of toxins was observed: all the toxins had a fast decrease in detectability in the first days, followed by a slower decrease. This profile of decreased detectability of toxins is congruent with previous studies which investigated the fate of soluble Cry3 and Cry1 toxins in different soils (1, 19, 21, 30). It seems that the heterogeneous structure of the leaf litters is close to the structure of soils, generating the same toxin adsorption profiles (25). In the present work, we used toxins in their crystal form instead of soluble toxins, and we obtained similar results. Interactions between crystals and litters have to be investigated further to better understand why the B. thuringiensis subsp. israelensis toxins have different behaviors in leaf litters. The same experiments have to be performed with other constituents of mosquito breeding sites, like soils, to determine if crystal toxins exhibit the same pattern of persistence.

Due to its synergistic effect on Cry toxins, Cyt toxins are considered key proteins in the cocktail of toxins produced by B. thuringiensis subsp. israelensis, allowing a delay in the development of resistance (16, 48). Our results show that Cyt is the less persistent toxin, as it was underrepresented in the field-collected toxic leaf litter samples and was undetectable after only 1 week of contact with leaf litter samples contaminated in the laboratory. A previous study showed that contact with leaf litter induced a strong decrease of the synergistic effect of Cyt on Cry toxins, leading to a drastic decrease of the toxicity of B. thuringiensis subsp. israelensis (42). As an increasing number of recombinant B. thuringiensis strains containing Cyt toxins are developed with the goal of obtaining a more efficient insecticidal product than B. thuringiensis subsp. israelensis (14, 15), it is necessary to better characterize the fate of Cyt toxins in the environment.

No effect of litter drying on the persistence of the toxins was observed, whatever the dose and the toxin considered. These results indicate that the watering/drying events often experienced by mosquito breeding sites are not responsible for the low level of persistence of B. thuringiensis subsp. israelensis in the environment. Moreover, higher doses of toxins did not led to a higher level of persistence of the toxins. These results are congruent with data from previous studies, indicating that an increase of the quantity of B. thuringiensis subsp. israelensis sprayed does not result in the increased persistence of the toxins and of the insecticidal activity (27). Our results confirm the importance of respecting the operational doses of insecticides for mosquito control for economic and ecological purposes.

Toxic leaf litters represent a unique opportunity to better understand how B. thuringiensis subsp. israelensis can persist and increase its toxicity in the field. Based on the results obtained in the present article and in previous studies, we propose a hypothetical scenario of the sequential events that led to the formation of toxic leaf litters in the field (Fig. 2). B. thuringiensis subsp. israelensis is sprayed into mosquito breeding sites as a suspension of spores and crystals (Fig. 2, step 1). Crystals and spores quickly settle in the bottom of the mosquito breeding site (24) (Fig. 2, step 2). This settling is partly responsible for the rapid decrease of the toxicity of B. thuringiensis subsp. israelensis, notably for surface-feeding mosquitoes like Anopheles (3). After contact with leaf litter, B. thuringiensis subsp. israelensis drastically loses its toxicity, also explaining a part of the toxicity loss, essentially for bottom-feeding mosquitoes like Aedes larvae (42) (Fig. 2, step 3). Spores can recycle under specific, unknown conditions (this work), as suggested by previous works (11, 43) (Fig. 2, step 4). Numerous environmental factors (the presence of organic matter, pollution, UV light, temperature, and solubilization) are known to strongly influence the residual activity of B. thuringiensis subsp. israelensis; therefore, their influence on the persistence of the toxins in the field must be investigated (22, 26, 38, 39). Moreover, the status of B. thuringiensis as an exclusive larval pathogen was recently confirmed, indicating that the presence of mosquito cadavers must be essential for recycling to occur (2, 34, 35). After proliferation, bacteria will sporulate, leading to a large amount of spores and to the liberation of crystals, inducing a high level of toxicity for mosquito larvae (7, 8, 43) (Fig. 2, step 5). At this step, toxins will exhibit the same proportions in this recycled B. thuringiensis subsp. israelensis strain as those in the commercial product. Toxins will then persist differentially, leading to modified proportions of toxins in the toxic leaf litters compared to the proportions of commercial B. thuringiensis subsp. israelensis (this work). Such an example shows that under specific, unknown conditions, B. thuringiensis subsp. israelensis can recycle and achieve a high level of toxicity for mosquito larvae. Considering that mosquitoes can become resistant when selected with these toxic leaf litters in the laboratory (31, 41), it can be expected that mosquito larvae exposed to field-persistent B. thuringiensis subsp. israelensis in some breeding sites may already be developing resistance to at least some toxins (Fig. 2, step 6), even if bioassays with B. thuringiensis subsp. israelensis still have not given evidence of it, probably due to their sensitivity being too low to be able to detect arising resistance in field mosquitoes. An understanding of the conditions that led to the recycling of B. thuringiensis subsp. israelensis and the molecular basis of mosquito resistance is therefore of high interest to ensure the long-term use of B. thuringiensis subsp. israelensis, one of the main larvicidals authorized at present for mosquito control.

Fig 2.

Hypothetical scenario of Bacillus thuringiensis subsp. israelensis recycling in mosquito breeding sites containing leaf litters.