A collection of Bacillus thuringiensis isolates plausibly active against coleopteran insects was obtained from the Agricultural Research Service Culture Collection. Each isolate was cultured, spores and δ-endotoxin crystals were pelleted by centrifugation and lyophilized, and the resulting product was incorporated in insect diet for testing against Diaprepes abbreviatus neonates. A bioassay method was developed that utilized small amounts of insect diet and B. thuringiensis spores and δ-endotoxin to treat single neonates confined to 0.2-mL clear polymerase chain reaction (PCR) tubes. The method was less expensive in terms of labor and materials as compared to previous methods and reduced control losses due to burrowing and aggressive behaviors of D. abbreviatus larvae confined together. Of 19 B. thuringiensis isolates screened for activity against D. abbreviatus with a discriminating dose of 250 ppm spores and δ-endotoxin on diet, 5 were selected for further evaluation in dose-response experiments. Diaprepes abbreviatus larvae demonstrated a significant dose response to 4 of the 5 isolates tested. The most active isolates were those that expressed CryET33 and CryET34, or Cyt2Ca1 proteins. A wild-type B. thuringiensis strain that expressed Cyt2Ca1 generated the lowest LC50 value (50.7 μg/ml) and steepest slope (1.11) based on log10 probit analysis of the data. These B. thuringiensis δ-endotoxins may have utility in transgenic approaches to citrus rootstock protection from D. abbreviatus.
The invasive weevil species,Diaprepes abbreviatus (L.), has become one of the most damaging insect pests of citrus and nursery crops in Florida since it was first reported in 1964 (Woodruff 1964). Prior to its introduction into the continental United States, D. abbreviatus was known to be a serious pest of sugarcane in the Lesser Antilles and is considered the most important pest of agriculture, horticulture, and silviculture in Puerto Rico (Hantula et al. 1987). Efforts of researchers and pest managers to develop an effective long-term management strategy for this pest have been unsuccessful. Apparently, D. abbreviatus is not under effective biological control within its putative native range of Puerto Rico and the Lesser Antilles (Lapointe 2004). The lack of natural enemies, combined with the wide host range of this highly polyphagous weevil (Simpson et al. 1996) and its slow subterranean larval development (Lapointe 2000), makes D. abbreviatus particularly well adapted to semi-permanent, tropical, and subtropical agroecosystems such as citrus groves. Efforts to identify plant resistance to D. abbreviatus within sexually compatible citrus germplasm have been only marginally successful (Bowman et al. 2001; Lapointe & Bowman 2002; Shapiro et al. 2000).
The bacterial entomopathogen, Bacillus thuringiensis (Berliner), has been recommended for the control of other insect pests of citrus, particularly those in the order Lepidoptera (Shapiro et al. 1998; Stansly et al. 2006). Although use of B. thuringiensis as an applied biopesticide can be an effective control method for some insects, the subterranean feeding habits of D. abbreviatus larvae make them difficult to target with a biocontrol agent, such as B. thuringiensis, that must be ingested to be effective. It has been suggested that the most appropriate and economically viable method for control of D. abbreviatus will be the production of transgenic rootstocks engineered to express exogenous toxins (Lapointe 2004). Al-Deeb & Wilde (2005) reported that transgenic corn, expressing the Cry3bb1 toxin from B. thuringiensis, was protected from another root-feeding coleopteran, the western corn rootworm, Diabrotica virgifera virgifera LeConte. Transgenic crops that express B. thuringiensis proteins display resistance to some of the most devastating pests of agriculture, yet are virtually safe to non-target organisms (Betz et al. 2000). A transgenic approach that uses a genetically-engineered citrus rootstock to express a δ-endotoxin active against D. abbreviatus is a plausible solution.
Currently, the few B. thuringiensis δ-endotoxins known to be active against coleopterans are far outnumbered by known lepidopteran-active toxins. In order to pursue this paradigm toward the development of a transgenic citrus rootstock, B. thuringiensis toxins that are active against D. abbreviatus larvae must first be identified. One strain of B. thuringiensis has been reported to cause mortality of D. abbreviatus larvae (Weathersbee et al. 2002), but otherwise B. thuringiensis has received minimal attention as a potential biocontrol agent for this pest. We assembled a collection of B. thuringiensis isolates that expressed novel δ-endotoxins putatively active against one or more representatives of Coleoptera. This paper presents the results of experiments that determined if any of these toxins were active against D. abbreviatus larvae.
Materials and Methods
Source and Culture of B. thuringiensis Isolates
Patent databases at the United States Patent and Trademark Office were searched to locate B. thuringiensis isolates potentially active against species of Coleoptera. Representative samples of 19 isolates, for which patents had issued (Bradfisch et al. 2005; Donovan et al. 2005; Narva et al. 2005; Rupar et al. 2003; Rupar et al. 2004; Soares et al. 1989), were obtained by request from the curator of the Agricultural Research Service Culture Collection, National Center for Agricultural Utilization Research, Peoria, Illinois (Table 1). Samples were received as dried pellets sealed in glass ampoules. Growing cultures of each isolate were established following instructions provided with the samples. Briefly, an ampoule was scored with a file and broken. The broken end of the ampoule was flame sterilized, the pellet was removed and then cultured in Luria-Bertani (LB) broth in an incubator shaker. Cultures were stored on LB agar slants at 4°C and in LB broth/glycerol stock solutions at -80°C until needed.
Starter cultures of each isolate were prepared by inoculating 10 mL of LB broth with a loopful of bacterial cells removed from storage. Cultures were grown overnight in 125-mL baffled Erlenmeyer flasks at 27°C and 150 rpm in an incubator shaker. Then 80 μl of starter culture were added to 100 mL of LB broth in a 500-mL baffled Erlenmeyer flask and grown in the incubator shaker until approximately 90% of the cultured cells had sporulated and autolyzed. Cultures were pelleted by centrifugation (15,000 G) for 15 min at 4°C, washed 3 times with phosphate buffered saline (PBS) containing 0.005% Triton X-100, lyophilized, and weighed. Lyophilized pellets, containing B. thuringiensis spores and δ-endotoxin crystals, were stored in 1.5-mL microcentrifuge tubes at -80°C until they were used in the experiments.
Insect Source and Rearing
Neonatal larvae of D. abbreviatus were obtained from a laboratory colony maintained at the U.S. Horticultural Research Laboratory, Fort Pierce, FL. Larvae were reared on a commercially-prepared insect diet (Product No. F1675, Bio-Serv, Frenchtown, NJ). Temperature and moisture content of the diet were optimized for larval development according to Lapointe (2000) and Lapointe & Shapiro (1999). Neonate larvae were surface sterilized with a solution of 0.31% sodium hypochlorite and individually inspected to insure only healthy and active larvae were used in the experiments.
A bioassay method was developed to test the effectiveness of B. thuringiensis isolates on individual D. abbreviatus neonates. A clear polymerase chain reaction (PCR) tube containing a small amount of insect diet in the lid was used to hold a single D. abbreviatus neonate that could be visually inspected for response to treatments. Monitoring of larvae had not been possible with older bioassay methods that used larger volumes of medium (soil or diet) because D. abbreviatus larvae burrow into the medium, complicating visual inspection.
All items used in the bioassay procedure were sterilized by autoclaving, filtering, or treating with 75% ethanol, and the procedure was conducted in a biological safety cabinet. A stock solution was prepared that contained 5% sucrose and 0.005% Triton X-100 in deionized distilled water. The previously prepared lyophilized pellets of each B. thuringiensis isolate were resuspended in the stock solution and diluted with stock to provide a discriminating dose of spores and δ-endotoxin in diet of 250 ppm (μg/mL) for screening experiments. Dose response assays were conducted with isolates that caused >50% mortality of neonates at the discriminating dose level. Concentrations of 300, 150, 75, and 32.5 ppm of spores and δ-endotoxin in diet were used in the dose-response experiments. Diet treatments for the controls received stock solution only.
Prepared insect diet was liquefied by reheating and 80 μl of diet were pipetted onto the inside surface of the lid of a 0.2-mL clear PCR tube. The diet pellets were dried for 15 min to remove approximately 20 μl of water. Bacillus thuringiensis treatments were applied in a volume of 20 μl by pipette to each diet pellet and the pellets were dried for an additional 5 min. Controls were treated equally with stock solution only. A #1 fine camel hair brush was used to place a single D. abbreviatus neonate into each PCR tube containing diet and the lid was affixed. The PCR tubes were inverted and placed in a tube rack, covered, placed in a sealed plastic bag with a moist paper towel, and stored in an incubator at 27°C. After 2 weeks, each larva was inspected with the aid of a dissecting microscope and mortality was recorded. There were 3 replications, each containing 30 larvae, for the initial screening of each isolate at 250 ppm. A minimum of 3 replications, each with 30 larvae, was used for each level of treatment in the dose-response experiments.
Data Analyses and Statistics
Data collected from the screening experiments were subjected to the Means Procedure (SAS Institute 1999) to determine means and standard errors for mortality of D. abbreviatus neonates exposed to the discriminating dose of each isolate. Paired t-tests were conducted using the T-test Procedure (SAS Institute 1999) to determine if means for mortalities in treated groups differed from those of control groups. A probability level of 5 percent (P ≤ 0.05) was considered significant.
Data from the dose-response experiments were adjusted for control mortality by the Abbott (1925) formula and transformed (arcsine) before analyses. Transformed data were analyzed by the General Linear Models Procedure, and differences among treatment level means were determined by Tukey’s studentized range test (SAS Institute 1999). Differences among means were considered significant at a probability level of 5 percent (P ≤ 0.05). Untransformed means are presented in the data tables. Data from isolates that elicited a significant response to treatment were subjected to log10 Probit analyses by the Probit Procedure (SAS Institute 1999) to generate LC50 values and slopes of probit lines.
Of 19 B. thuringiensis isolates screened in diet bioassays against D. abbreviatus neonates, 7 caused significantly greater (P ≤ 0.05, paired t-tests) mortality compared with the controls (Table 1). Isolates B-21365, B-21366, and B-21367 containing CryET33 and CryET34 toxins caused the highest observed mortalities (90, 88, and 91%, respectively). Isolates B-21582 and B-21583 containing the Cyt2Ca1 toxin provided 82 and 52% mortalities, respectively. These five isolates (B-21365, B-21366, B-21367, B-21582, and B-21583) provided meaningful levels of mortality (>50%) and were further evaluated in dose-response experiments.
A significant effect of spore and δ-endotoxin dose was observed for 4 of the 5 isolates that were subjected to dose-response experiments against D. abbreviatus neonates, including B-21365 (F = 15.52; df = 4, 28; P < 0.0001), B-21367 (F = 9.46; df = 4, 28; P < 0.0001), B-21582 (F = 33.63; df = 4, 8; P < 0.0001), and B-21583 (F = 56.60; df = 4, 16; P < 0.0001). The highest corrected mortality observed in the dose-response experiments was 81% provided by the wild-type isolate B-21582 at a dose of 300 μg spores and δ-endotoxin/ml diet (Table 2). Recombinant isolates B-21367 and B-21583 also elicited good dose-responses with greater than 60% mortality of D. abbreviatus larvae observed at the 300 ppm dose. The effect of spore and δ-endotoxin dose on larval mortality was not significant for isolate B-21366 (F = 1.71; df = 4, 8; P = 0.2406). The dose-response obtained with isolate B-21366 was inconsistent, the response data were variable compared to those of the other isolates, and larval mortality obtained at the highest dose remained below 50%. Consequently, isolate B-21366 was not included in subsequent probit analyses.
Results obtained for isolates B-21365, B-21367, B-21582, and B-21583 were examined further by log10 probit analyses to model the effects of spore and δ-endotoxin dose on mortality of D. abbreviatus larvae (Fig. 1). The calculated LC50 for larvae exposed to B-21365 in diet was 258.3 (95% FL = 130.5-2779) ppm [AI]. The slope of the probit line was 0.65 (SE = 0.23) (χ2= 8.11; df = 1; P = 0.0044) (Fig. 1A). The LC50 for larvae exposed to B-21367 was 115.3 (95% FL = 40.4-269.5) ppm [AI] and the slope of the probit line was 0.93 (SE = 0.31) (χ2= 8.79; df = 1; P = 0.0030) (Fig. 1B). The LC50 for larvae exposed to B-21582 was 50.7 (95% FL = 28.3-72.1) ppm [AI] and the slope was 1.11 (SE = 0.21) (χ2= 27.10; df = 1; P ≤ 0.0001) (Fig. 1C). The calculated LC50 for larvae exposed to B-21583 was 174.1 (95% FL = 114.6-361.2) ppm [AI] and the slope was 0.79 (SE = 0.0.19) (χ2= 17.96; df = 1; P ≤ 0.0001) (Fig. 1D).
The subterranean habit, and aggressive behavior when confined together, complicate the evaluation of control measures for D. abbreviatus larvae (Lapointe & Shapiro 1999) and result in high levels of control mortality (Schroeder & Sieburth 1997; Quintella & McCoy 1997; Weathersbee et al. 2002). The bioassay method used here provided an efficient means for screening B. thuringiensis isolates against D. abbreviatus neonates. It was less expensive in terms of labor and materials than methods used in the past. The current method avoids these problems by confining larvae singly on a nominal amount of diet, thereby reducing losses in the control group. Control mortality in these experiments did not exceed 20% and was most often maintained below 10%.
The highest levels of mortality of D. abbreviatus in the screening experiments (90, 88, and 91%) occurred when neonates were fed diet containing spores and δ-endotoxin of isolates B-21365, B-21366, and B-21367, respectively. Donovan et al. (2005) has shown these isolates to be active against the larvae of other species of Coleoptera, including the red flour beetle, Tribolium castaneum (Herbst), and the Japanese beetle, Popillia japonica Newman. Isolate B-21365 is a wild-type B. thuringiensis strain that contains genes for CryET33 and CryET34 toxins. Isolate B-21367 is a recombinant strain that was engineered to express CryET33 and CryET34 toxins. This isolate was derived from a parent B. thuringiensis strain that was crystal negative (Cry-). Isolate B-21366 also is a recombinant strain engineered to express CryET33 and CryET34 toxins, but the parent strain was wild-type and naturally expressed the Cry3Bb2 toxin. Although B-21366 expressed an additional δ-endotoxin compared to B-21365 and B-21367, it performed no better against D. abbreviatus than those isolates in the screening experiments, and its performance was inconsistent in the dose-response experiments. Perhaps expressing the CryET33 and CryET34 toxins in addition to Cry3Bb2 added a burden that affected overall toxin expression and the virulence of this strain. Because this isolate produced erratic results it was not further evaluated.
The wild-type isolate B-21365, expressing the CryET33 and CryET34 toxins, exhibited the shallowest dose-response curve and highest LC50 value of those tested, probably because the highest concentration of B-21365 tested provided 53% larval mortality. The recombinant isolate B-21367 expressing the same toxins displayed a steeper response curve, a lower LC50 value, and provided better predictions in the 50% response range. It is unclear why recombinant strain B-21367 apparently produced a more virulent product than the wild-type B-21365 since both expressed the same toxins. Perhaps B-21367 invests less energy into other processes such as spore formation and more into toxin production. Nonetheless, the dose responses displayed by the recombinant strain B-21367 and the wild type B-21365 against D. abbreviatus were not statistically different based on overlap of standard errors.
Isolates B-21582 and B-21583 provided 82 and 52% mortalities, respectively, in the screening experiments. Rupar et al. (2004) demonstrated these isolates were active against the larvae of representative species of Siphonaptera: including the cat flea, Ctenocephalides felis (Bouché); and Coleoptera: including the southern corn rootworm, Diabrotica undecimpunctata howardi Barber; the western corn rootworm, Diabrotica virgifera virgifera LeConte; the Colorado potato beetle, Leptinotarsa decemlineata (Say); the red flour beetle; and the Japanese beetle. Isolate B-21582 is a wild-type B. thuringiensis strain that contains the gene for the Cyt2Ca1 toxin, while recombinant strain B-21583 originally was a Cry - B. thuringiensis that was engineered to express the Cyt2Ca1 toxin. The Cyt2Ca1 protein is a δ-endotoxin that fits into a second category, aside from the Cry toxins, known as cytolitic (Cyt) toxins. The Cyt proteins, also known as hemolytic toxins, cause damage to the insect midgut through pore formation and cell lysis much like the Cry toxins. Guerchicoff et al. (2001) provides a discussion of the Cyt gene family, including similarities and differences with Cry genes.
The wild-type isolate B-21582, expressing the Cyt2Ca1 toxin, displayed the steepest dose-response curve and lowest LC50 value of those tested with the highest confidence in predictions. Though recombinant strain B-21583 also expressed Cyt2Ca1, the probit response was not as steep as that for B-21582, the LC50 value was greater, and the confidence in predicted values was lower. It appeared in this case that the wild-type isolate B-21582 produced a more virulent product against D. abbreviatus than did the recombinant strain B-21583 expressing the same toxin. Perhaps isolate B-21582 produces another, yet undetected product that works in conjunction with the Cyt2Ca1 protein to induce mortality in D. abbreviatus larvae.
Bacillus thuringiesis products have been widely accepted in agriculture for the control of many insect pests, but only in recent years have strains been discovered that control coleopteran pests. Control strategies that rely on formulated B. thuringiensis applications are used world-wide in many crops. Transgenic plant varieties that express B. thuringiensis δ-endotoxins have been used in the U.S. for several years and are now gaining international acceptance (Betz et al. 2000). Unfortunately, transgenic approaches to plant improvement are not currently being exploited in some crops that could benefit most from this technology, such as citrus, where a genetically engineered rootstock could be used to alleviate damage caused by D. abbreviatus.
These experiments demonstrated that there are B. thuringiensis δ-endotoxin genes currently available that could be used to transform citrus for protection against D. abbreviatus. The CryET33, CryET34, and Cyt2Ca1 genes could be expressed together or separately in a citrus rootstock. An appropriately engineered citrus rootstock, if properly managed, has the potential to offer resistance to D. abbreviatus throughout the life of the crop.
Because D. abbreviatus has a broad host range and is known to feed on other plants within and around citrus groves (Lapointe 2003), the presence of toxin in citrus roots within a grove may be expected to result in increased feeding by larvae on alternative food sources (either wild or intentionally planted) and thus reduce the likelihood of rapid resistance development to the toxin. The use of transformed citrus rootstocks would also avoid concerns associated with the possible effect of pollen from transformed plants, particularly if the gene inserted into the rootstock is constructed to be expressed only in root tissue.
Development of this technology for citrus should not be delayed since introduction of genetically modified crops requires investments of time and money, and years to complete, particularly for slow-maturing crops like citrus. Moreover, an effective control strategy for D. abbreviatus is long overdue and alternatives to the proposed genetically-modified citrus rootstock have not been presented.
Table 1. Screening of B. thuringiensis isolates against D. abbreviatus neonates with a discriminating dose (250 μg/ml) of lyophilized, sporulated cultures in diet.
Table 2. Mortality of D. abbreviatus neonates exposed to diet treated with differing rates of lyophilized, sporulated cultures of B. thuringiensis isolates.