Learn more: PMC Disclaimer | PMC Copyright Notice
Midgut bacteria required for Bacillus thuringiensis insecticidal activity
Abstract
Bacillus thuringiensis is the most widely applied biological insecticide and is used to manage insects that affect forestry and agriculture and transmit human and animal pathogens. This ubiquitous spore-forming bacterium kills insect larvae largely through the action of insecticidal crystal proteins and is commonly deployed as a direct bacterial spray. Moreover, plants engineered with the cry genes encoding the B. thuringiensis crystal proteins are the most widely cultivated transgenic crops. For decades, the mechanism of insect killing has been assumed to be toxin-mediated lysis of the gut epithelial cells, which leads to starvation, or B. thuringiensis septicemia. Here, we report that B. thuringiensis does not kill larvae of the gypsy moth in the absence of indigenous midgut bacteria. Elimination of the gut microbial community by oral administration of antibiotics abolished B. thuringiensis insecticidal activity, and reestablishment of an Enterobacter sp. that normally resides in the midgut microbial community restored B. thuringiensis-mediated killing. Escherichia coli engineered to produce the B. thuringiensis insecticidal toxin killed gypsy moth larvae irrespective of the presence of other bacteria in the midgut. However, when the engineered E. coli was heat-killed and then fed to the larvae, the larvae did not die in the absence of the indigenous midgut bacteria. E. coli and the Enterobacter sp. achieved high populations in hemolymph, in contrast to B. thuringiensis, which appeared to die in hemolymph. Our results demonstrate that B. thuringiensis-induced mortality depends on enteric bacteria.
Bacillus thuringiensis is an opportunistic insect pathogen that was discovered almost a century ago (1–7). The salient feature of this species is accumulation of crystalline parasporal inclusions during sporulation. These inclusions are composed of one or more protoxins, known as δ-endotoxins, each of which is specific primarily at the level of insect orders, particularly Lepidoptera, Diptera, and Coleoptera (6). In Lepidoptera, specificity is due in part to the extremely alkaline midgut environment that is required to solubilize the protoxin into the active form. Solubilized protoxins are activated by midgut proteases and bind to receptors on the epithelial surface. Then, by a process that remains unclear, the toxins appear to insert into the membranes of gut cells, where they form pores that lead to cell lysis (2). It has been proposed that disruption of the midgut epithelium results in a prolonged cessation of feeding and eventual death by starvation. An alternative proposed mechanism of killing is that extensive cell lysis provides spores access to the more favorable environment of the hemocoel, where they germinate and reproduce, leading to septicemia and death (2, 3). Although these two models have been cited many times in the literature, neither is entirely consistent with experimental observations. For example, the starvation model is not supported by the fact that B. thuringiensis kills insects much more rapidly than does starvation: B. thuringiensis toxin-induced mortality typically takes 2–5 days, whereas starvation-induced mortality often takes 7–10 days. In addition, rapid turnover of gut epithelial cells permits larvae to recover from toxin-induced feeding inhibition (8–10). The septicemia model is challenged by the observation that the toxin causes mortality when it is separated from the bacterial cells, which has been accomplished with purified toxin and transgenic plants that produce the toxin (1–4, 6).
Results
Effect of Antibiotics on Larval Susceptibility to B. thuringiensis.
In a previous study, we found that a number of treatments that altered the susceptibility of the gypsy moth to B. thuringiensis also affected the bacterial community in the larvae's midgut (11, 12). To determine whether there was a causal relationship between changes in this community and susceptibility to B. thuringiensis toxin, we explored the impact of antibiotics. We reared larvae from hatching on either sterile artificial diet or diet amended with a range of concentrations (8–500 μg/ml of diet) of an antibiotic mixture that consisted of penicillin, gentamicin, rifampicin, and streptomycin. Upon molting to the third instar, larvae from each treatment were fed with 0, 1, or 10 units of B. thuringiensis, and mortality was monitored for 7 days. B. thuringiensis-induced larval mortality was reduced by amendment of the larval diet with antibiotics in a dose-dependent fashion (Fig. 1). Mortality of insects fed B. thuringiensis was inversely proportional to the antibiotic concentration, with no mortality caused by B. thuringiensis among larvae fed 10 units and reared on antibiotic mixtures containing 125 μg/ml or more of each antibiotic. This reduction in mortality was accompanied by reduced populations of culturable Enterococcus and Enterobacter spp. from the midguts of larvae (Table 1). These species were previously identified as the dominant culturable members of the gypsy moth midgut (11). This result suggested that the microbial community in the midgut contributed to larval death due to B. thuringiensis treatment.
Effect of antibiotic concentration on toxicity of B. thuringiensis to third-instar Lymantria dispar. Larvae were reared on sterile artificial diet that was unamended or amended with an antibiotic mixture containing up to 500 μg each of penicillin, gentamicin, rifampicin, and streptomycin per ml. Curves were fit by using PROC NLIN in SAS 9.1.3 (28). All are of the general form y = ae−bx. Dose 0 IU, a = 0.0862, b = 0.0108, F2,6 = 5.47, P > 0.05; dose 1.0 IU, a = 0.3125, b = 0.0608, F2,6 = 8.12, P > 0.02; dose 10.0 IU, a = 0.5351, b = 0.0474, F2,6 = 41.12, P > 0.0003.
Table 1.
Effect of antibiotic mixture concentration on the culturable midgut bacteria of third-instar L. dispar
| Antibiotic concentration, μg/ml of diet* | Enterococcus sp., cfu per gut | Enterobacter sp., cfu per gut |
|---|---|---|
| 0 | 4.80 × 106 | 5.21 × 105 |
| 8 | 1.17 × 105 | 6.08 × 107 |
| 16 | nd | 1.13 × 108 |
| 31 | 1.00 × 101 | 1.67 × 101 |
| 63 | 1.30 × 105 | 2.50 × 104 |
| 125 | 5.33 × 101 | nd |
| 250 | nd | nd |
| 500 | nd | nd |
nd, not detected.
*Larvae were reared on sterile artificial diet amended with an antibiotic mixture composed of 500 μg each of penicillin, gentamicin, rifampicin, and streptomycin per ml.
Impact of Enterobacter sp., a Midgut Inhabitant, on Larval Susceptibility to B. thuringiensis.
To test the possibility that midgut bacteria were contributing to B. thuringiensis-induced mortality directly, a resident member of the midgut microbial community, Enterobacter sp. NAB3 (11), was reestablished in larvae reared on antibiotic-amended medium. After feeding of the Enterobacter sp. to newly molted third-instar gypsy moth larvae, this bacterium established populations of 9.0 × 102 to 2.1 × 104 per midgut and restored the insecticidal activity of B. thuringiensis to nearly that of larvae not treated with antibiotics (Fig. 2). Other members of the gut community, Enterococcus casseliflavus and Staphylococcus xylosus, did not restore insecticidal activity to B. thuringiensis when fed to larvae in the same manner. These results suggest that the Enterobacter sp. is largely responsible for the septicemia associated with B. thuringiensis-induced mortality in gypsy moth larvae.
Restoration of B. thuringiensis toxicity by an Enterobacter sp. after elimination of detectable gut flora and B. thuringiensis activity by antibiotics. “Antibiotic cocktail” L. dispar larvae were reared until the third instar on sterile artificial diet amended with 500 μg each of penicillin, gentamicin, rifampicin, and streptomycin per ml. Each bar represents the mean mortality ± SEM of 48 larvae (four replications of 12 larvae each). Values at the bottom represent the sizes of the populations of the Enterobacter sp. as detected by culturing. nd, not detected.
In Vitro Growth of Enterobacter sp. NAB3 and B. thuringiensis in Larval Hemolymph.
To cause septicemia, bacteria must be able to enter and multiply in the hemolymph, so we compared the growth of B. thuringiensis and an Enterobacter sp. in vitro in larval hemolymph. We introduced ≈105 bacterial cells per ml of hemolymph withdrawn from fifth-instar gypsy moth larvae. The Enterobacter sp. reached a population density of 1.3 × 109 cfu/ml after 6 h, which was equivalent to its growth in tryptic soy broth. In contrast, B. thuringiensis populations dropped to below detectable limits (200 cfu/ml) in hemolymph after 6 h and remained undetectable throughout the 18.5-h assay, even though B. thuringiensis grew as quickly as the Enterobacter sp. in tryptic soy broth (Fig. 3).
Growth of B. thuringiensis, Enterobacter sp. NAB3, and E. coli ECE52 in tryptic soy broth (Left) and L. dispar hemolymph (Right). The detection limit was 200 cfu/ml; samples in which no colonies were detected were assigned this value and are indicated by open symbols. Each point represents the mean cfu/ml ± SEM of three replicate cultures.
Killing Ability of Escherichia coli ECE52 and in Vitro Growth in Hemolymph.
Population densities of B. thuringiensis were similar in midguts of larvae that had been reared on antibiotic-amended and unamended sterile artificial diet, suggesting that the reduction in insect mortality was not due to direct elimination of B. thuringiensis. To test this possibility more directly, we assayed the effect of E. coli strain ECE52, which produces the B. thuringiensis toxin (13), and compared its toxicity to larvae reared on antibiotic-amended versus unamended diet. E. coli ECE52 killed larvae that had been reared on antibiotics as effectively as larvae containing the native gut community (Fig. 4). When the E. coli cells were heat-killed before feeding, mortality was equivalent to treatment with live E. coli in larvae reared on an unamended diet. However, in larvae reared on antibiotics, there was no mortality due to toxin alone in the absence of live E. coli cells (Fig. 4). E. coli lacking the crystal toxin genes (JM103) did not induce significant mortality in either group of larvae (0–4%). In hemolymph, E. coli ECE52 reached a population density of >109 cfu/ml after 14.5 h (Fig. 3). These results suggest that the B. thuringiensis toxin alone is not sufficient to cause larval mortality, because an enteric bacterium (a member of the Enterobacteriaceae family) such as E. coli or Enterobacter must also be present to induce septicemic death.
Insecticidal activity of a B. thuringiensis toxin-producing E. coli strain (ECE52) is abolished by heat-killing in the absence of normal gut microbiota. Larvae were fed live ECE52 with or without antibiotics, or they were fed heat-killed ECE52 with or without antibiotics, as indicated. Each bar represents the mean mortality ± SEM of 36 third-instar L. dispar (three replications of 12 larvae each).
Discussion
We present three lines of evidence that indicate that the resident midgut bacteria of gypsy moth larvae contribute to mortality associated with consumption of B. thuringiensis. First, insecticidal activity was abolished by eliminating the detectable midgut bacterial community. Second, insecticidal activity was restored by reestablishing an Enterobacter sp., a member of the normal midgut community. Third, a live E. coli strain that produces the B. thuringiensis toxin killed larvae whether or not they contained other bacteria in their midguts, but if the E. coli was heat-killed before feeding, it killed only those larvae that contained the normal gut community. These results, coupled with the observation that the Enterobacter sp. and E. coli grew rapidly in larval hemolymph, whereas B. thuringiensis did not, indicate that the enteric bacteria are responsible for septicemia associated with B. thuringiensis toxicity. The enteric bacteria alone do not induce mortality, suggesting that B. thuringiensis enables them to reach the hemocoel by permeabilizing the gut epithelium.
These results suggest that B. thuringiensis toxicity depends on an interaction with microorganisms of the normal gut community. This finding contrasts sharply with established models that assume that B. thuringiensis itself induces mortality through starvation or direct septicemia. Thus, much of the research on B. thuringiensis since its discovery in 1918 and its development for commercial use in the 1950s has been predicated on a mechanism that is quite different from the one we present here. Studies of the mode of action of B. thuringiensis in the last 25 years have clarified the molecular specificity of toxin binding and pore formation but have not addressed the subsequent events that lead to insect death (1–3). The pervasive acceptance of the models of starvation and B. thuringiensis-induced septicemia over the last decade appears to be associated with two review articles that have been cited >800 times since 1992; each article postulates the models, but neither provides empirical evidence to support either one.
A number of published results that are inconsistent with the starvation and B. thuringiensis-induced septicemia models are easily reconciled with the gut microbial community model that is presented here. B. thuringiensis spores, for example, are typically absent from the hemocoel until very late in the infection process and frequently do not appear until well after the death of the insect (14–16). This finding challenges the idea that B. thuringiensis induces death by septicemia but is consistent with the model that proposes that septicemia is initiated by the enteric bacteria in the insect gut. In addition, some studies indicate that B. thuringiensis is unable to grow in the hemocoel, which is consistent with our evidence that it does not grow in hemolymph (17, 18) and also makes it difficult to ascribe septicemia to B. thuringiensis alone. Most studies have not included an examination of dead larvae to verify the septicemic agent, thereby obscuring the contribution of the gut bacteria. Despite the prevailing models, there have been hints of the involvement of enteric bacteria in B. thuringiensis killing. For example, Dubois and Dean (19) demonstrated that some species of forest epiphytic bacteria enhance the toxicity of B. thuringiensis proteins, although these researchers and others fall short of concluding that the gut bacteria are required for B. thuringiensis killing activity, as we suggest here. Furthermore, a number of studies use a procedure known as “immunizing” larvae and hemolymph with enteric bacteria to enable B. thuringiensis to grow, casting doubt on the ability of B. thuringiensis to proliferate in this environment independently (20). This latter result suggests that the enteric bacteria might remove an immunological barrier, such as defensive enzymes or antimicrobial peptides that prevent growth of B. thuringiensis.
The implications of this work encompass both applications and fundamental knowledge. The role of enteric bacteria in insect mortality by B. thuringiensis presents opportunities for designing approaches to manage insect pests by harnessing their indigenous communities or combining B. thuringiensis with bacteria that can induce septicemia. This work also raises the possibility that the genes encoding the B. thuringiensis toxins could be deployed more effectively in transgenic crops by exploiting the role of insect-borne bacteria that enhance insecticidal activity. Additionally, research on both the efficacy and nontarget impacts of B. thuringiensis has yielded a perplexingly wide range of results with regard to the environmental persistence of activity (21). Recent studies show that the composition and populations of gut communities can vary widely within insect species (11, 12); hence, a more thorough understanding of gut communities could help clarify these issues. Recent interest in the role of intestinal bacteria in sepsis invites a parallel between human disease and insect killing by B. thuringiensis (22). A number of human pathogens are similar to B. thuringiensis in that they deploy toxins that produce pores in the wall of the host gut. The possibility that normal gut microbiota contribute to human diseases that involve toxins such as colicin A, diphtheria toxin, and Pseudomonas aeruginosa exotoxin (23–25) bears exploration. Thus, the cooperative relationship between B. thuringiensis and the indigenous gut microbiota might provide a model for studying polymicrobial diseases in animals that are not as experimentally accessible as the gypsy moth.
Methods
Bacterial Strains.
The B. thuringiensis used was a commercial formulation of B. thuringiensis subspecies kurstaki (DiPel DF, Valent Biosciences, Libertyville, IL) consisting of proteins and spores. Enterobacter sp. NAB3 (11) was originally cultured from the midguts of gypsy moth larvae feeding on sterile artificial diet. E. coli strain ECE52 was provided by the Bacillus Genetic Stock Center (Columbus, OH). For assays with heat-killed E. coli ECE52, a washed overnight culture was heated to 70°C for 15 min before assay use.
Insects.
Gypsy moth egg masses were obtained from colony NJSS at the U.S. Department of Agriculture–Animal and Plant Health Inspection Service laboratory at Otis Air National Guard Base (Cape Cod, MA) and reared as described in ref. 26, except that larvae reared on the antibiotic mixture were fed sterile artificial diet amended with 500 μg each of rifampicin, gentamicin, penicillin, and streptomycin per ml.
Mortality Assays.
B. thuringiensis mortality assays were performed as described in ref. 26. All assays were performed on newly molted third-instar larvae. Within a treatment, the same dose of 1, 10, or 100 units of B. thuringiensis was applied each day. All treatments were applied in a volume of 1 μl to a standard diet disk (4-mm diameter and 1-mm height) and fed to the larvae on 2 consecutive days. Mortality was recorded every 24 h for 7 days.
Bacterial Growth Curves in Hemolymph.
Hemolymph was collected from ≈250 larvae by the method of Orchard and Goodrich-Blair (27), and bacterial growth in hemolymph was measured by dilution plating on tryptic soy agar. Approximately 105 cfu from an overnight culture grown in 1/10-strength tryptic soy broth was inoculated into 1 ml of hemolymph and grown at 25°C with shaking for 18.5 h. Subsamples (10 μl) were taken every 1.5 h for dilution plating.
Acknowledgments
We thank Peter M. Crump for assistance with statistical analysis, Matthew McMahon and Michelle Preston for maintenance of the gypsy moth colony, the U.S. Department of Agriculture for providing gypsy moth eggs, and the Bacillus Genetic Stock Center for providing E. coli ECE52. This work was supported by the University of Wisconsin College of Agricultural and Life Sciences Hatch Project, Valent Biosciences, and the Howard Hughes Medical Institute.
Footnotes
The authors declare no conflict of interest.
This paper was submitted directly (Track II) to the PNAS office.