Developmental Penalties Associated with Inducible Tolerance in Helicoverpa armigera to Insecticidal Toxins from Bacillus thuringiensis

Abstract

Exposure of insect larvae to sublethal concentrations of crystal toxins from the soil bacterium Bacillus thuringiensis (Bt toxins) causes the induction of immune and metabolic responses that can be transmitted to offspring by epigenetic inheritance mechanisms. Given that the elevated immune status carries significant developmental penalties, we wanted to establish the relationships between immune induction, tolerance to the toxin and developmental penalties. A laboratory culture of Helicoverpa armigera was induced by a sublethal bacterial suspension containing crystal toxin Cry1Ac in one generation and maintained in the presence of toxin, acquiring significant levels of tolerance to the toxin within 12 generations of continuous exposure. Comparing tolerant and susceptible insects, we show that the induction of larval immune response and the coincident alteration of development-related metabolic activities by elicitors in the larval gut (larval induction) differs from the elevated immune status transmitted by epigenetic mechanisms (embryonic induction). Because the damaging effects of larval induction processes are higher compared to embryonic induction, it is likely that overall developmental penalties depend on the relative contribution of the two induction processes. When insects are kept with the same amount of toxin in the diet for subsequent generations, the embryonic induction process increases its contribution compared to the larval induction, resulting in reduced overall developmental penalty, while tolerance to the toxin is maintained.

Bacillus thuringiensis endotoxins (referred to here as Bt toxins) are the most important biopesticides for the management of insect pests and disease vectors (4, 8). Bt toxins are extensively used in agriculture and natural environments because of their relatively target-specific mode of action, low risk to human health, and environmentally benign properties. Because of the extensive usage, the development of resistance to Bt toxin in insect pests and insect vectors of disease is a serious threat to human health and agricultural production (2, 8). One of the unintended outcomes of intensive pest control measures has been the emergence of new tolerance mechanisms in pest insects (4). For example, the selection pressure imposed on insect populations in transgenic crops expressing Bt toxins requires continuous monitoring to prevent or delay insect populations that exhibit genetic resistance (7, 15) characterized by mutations in coding sequences associated with protoxin activation or subsequent toxicity (e.g., toxin-receptor active sites [8, 14]). In contrast to genetic resistance based on target site mutations (which produces individuals resistant to high toxin concentrations [8]), it is plausible that inducible low-level resistance (which we refer to here as inducible tolerance) caused by gene and protein regulatory mechanisms could relate to the relative amounts and activities of immune and fitness-related metabolic components, thereby leading to sequestration or inactivation of the toxin.

We investigate here the induction of upregulated immune activity associated with a Bt toxin suspension. We also investigated potential developmental penalties (larval development time and mass) associated with markers of enhanced immune status to identify effects of transient and/or inducible mechanisms on the physiology of tolerant insects. We regard this as different from the term “fitness costs,” which is generally used in the context of mutational alterations in resistance genes (8) causing pleiotropic effects due to absence of resistance gene products or changes in the function of mutant gene products (9). We therefore use the term developmental penalty instead of fitness costs to reflect this distinction.

Although not much is known about inducible tolerance, a possible mechanism for the sequestration of toxins is the aggregation of immune components (procoagulants) around the toxin inside the gut lumen. In fact, endotoxins from B. thuringiensis have lectin-like properties (14), raising the possibility that the mature toxin can potentially interact with glycosylated immune components, such as procoagulants that are secreted into the gut lumen of immune-induced insects.

Given that lipophorin is a lipid carrier in insects and under certain conditions has procoagulant properties (12, 18), it is likely to have important functions in both lipid metabolism and cell-free defense. Since lipophorin and phenoloxidase are increased in tolerant larvae (21), the question is how tolerance levels correlate with developmental penalties. Although not much is known about lipid uptake from food, lipid carriers in the gut lumen are likely involved in lipid transport from the peritrophic membrane to gut cells. Another question is how the immune status correlates with tolerance and fitness (4) and whether melanization is a reliable indicator of the immune status. We show here that H. armigera larvae can become tolerant to Cry1Ac after continuous exposure for 12 generations to sublethal concentrations of the toxin. While tolerance to the toxin is correlated with an overall elevated immune and metabolic status, the associated developmental penalties are affected by other factors, such as how induction is performed and over how many generations the insects have been exposed to the toxin.

MATERIALS AND METHODS

Insect strains.

An established Bt toxin-susceptible H. armigera laboratory strain (ANGR) was maintained on an artificial diet (modified from reference 22) in multiwell plastic trays, heat sealed with Mylar (heat-sealable polyester film) in the University of Adelaide, Waite Campus Insectary (25 ± 1°C; 75% ± 10% relative humidity with a photoperiod of 14:10 h [light/dark]), and without exposure to Bt toxin. Although the ANGR strain was kept in the laboratory without any overt inbreeding depression, the toxin exposure experiments were performed with an ANGR subpopulation mixed with susceptible insects (testing negative for resistance alleles) from the field (Narrabri, Australia). The field-derived strain was compared to the susceptible strain in terms of their relative tolerance to Cry1Ac crystal preparations from HD73 and respective ability to be induced by exposure to a sublethal concentration in artificial diet. Since there was no significant difference between susceptible and field strains (see Fig. S1 in the supplemental material), they were combined into a single susceptible population to provide additional genetic diversity and to prevent inbreeding depression. About 500 neonate larvae from the combined susceptible strains were then exposed to a sublethal concentration of Cry1Ac preparation (0.01 mg/ml) in artificial diet (10% lethal concentration [LC₁₀], 0.01 mg/ml; LC₅₀, 0.03 mg/ml; LC₉₀, 0.138 mg/ml). Under these conditions, the Cry1Ac suspension (potentially containing multiple elicitors such as spores and bacterial components) could induce the larval immune system without causing significant mortality. The toxin concentration was doubled to 0.02 mg/ml at generation 6 and to 0.1 mg/ml at generation 12.

Cry1Ac toxin suspension.

A bulk, crude, commercial bacterial suspension (B. thuringiensis strain HD73) containing Cry1Ac toxin (and control suspension of bacteria containing empty expression vector) were kindly supplied by John L. Reichelt (Bacterial Fermentation Pty., Ltd., Arundel, Queensland, Australia). Initially, bioassays were performed with egg-hatch neonate larvae and confirmed the relative toxicity of the Cry1Ac-containing suspension at different dilutions compared to the negative control preparation (not shown). Aliquots of this bulk preparation were used throughout the project.

Toxin bioassays.

Bioassays were performed using an artificial diet overlaid with a crude bacterial suspension containing Cry1Ac. In each bioassay, fresh artificial diet was poured into 45-well plastic trays (∼2 ml per well) and left to solidify in a fume hood for about 30 min. Because the exact concentration of toxin in each preparation was not known, a preliminary assay was conducted using a broad range of Cry1Ac suspension concentrations to determine the appropriate concentration for the formal assay (data not shown). Full bioassays were then conducted with 10 concentrations (plus a Milli-Q water control) and at least 90 larvae for each concentration. Toxin-containing stock solutions were diluted in Milli-Q water to specific concentrations, and 50-μl aliquots were spread evenly on top of the artificial diet in each well. Trays were left to dry in a fume hood for about 30 min. One neonate larva was placed in each well, and the tray sealed as described above. A total of 20 to 25 fine holes were then punched into the film to allow air exchange. Trays were placed in a ventilated room under standard culture conditions (see above), and the efficacy of the Cry1Ac treatments was assessed on day 7.

Developmental penalties.

To measure developmental penalties (i.e., reduced body weight and extended larval development), toxin bioassays were performed as described above with the modification that toxin suspension (or buffer only) was mixed through the diet prior to pouring into wells. This was performed to ensure constant toxin exposure over the assay period. Mortality and larval body weight were recorded after 10 days of exposure. The developmental time was recorded for each insect (egg-hatched neonates to pupa formation).

Melanization assay.

The immune status of susceptible and tolerant larvae was determined by melanization assays. Eight to ten third-instar larvae were chilled on ice for 5 min, washed with ice-cold 70% ethanol, and then washed with ice-cold phosphate-buffered saline (PBS). Hemolymph was extracted by cutting a foreleg and bleeding larvae directly into 1.5 ml of ice-cold PBS. The solution was centrifuged for 5 min at 3,000 × g, and the supernatant was transferred to a quartz cuvette. The optical absorbance was measured (Varian DMS 100S spectrophotometer) at 280 nm to determine the relative protein concentration, and then the degree of melanization (i.e., the A₄₉₀) was measured every minute for 30 min. At least five replicates for each sample were used.

Statistical analyses.

Melanization, larval body weight, and developmental delays were analyzed by using Student t test paired analyses (SigmaPlot 10; Systat Software, Inc.). Mortality data were analyzed (POLO-PC software; LeOra Software, Berkeley, CA) to estimate the LCs. For each probit analysis, the mortality was corrected using Abbott's formula (1). Differences in susceptibility were considered significant when the 95% confidence intervals did not overlap at LC₅₀. The resistance ratio (RR) was expressed as the ratio of the LC₅₀ of the relevant sample to that of the susceptible insects.

RESULTS

Tolerance after continuous exposure to toxin.

Field-collected and susceptible laboratory (ANGR) insects did not show significant differences in mortality to Cry1Ac and associated developmental penalties (see below and Fig. S1 in the supplemental material). Before long-term toxin exposure was started, we merged the two susceptible populations to introduce field-related genotypes and increased genetic diversity in the susceptible experimental insects, given that the H. armigera can suffer from inbreeding depression. Populations of ca. 500 larvae were kept on artificial food containing 0.01 mg of total protein/ml and compared to control insects. The toxin concentration was increased over generations as described above. When the level of the tolerance of the larvae was assessed at generation 12, it had increased significantly (Table 1, T×T/S×S at LC₅₀, RR = 29.8). Transmission of the tolerance was likely due to an epigenetic mechanism showing a strong maternal effect (Table 1; compare RR of larvae from T×S to those from S×T). This is also evidenced by a probit analysis of larval mortality exhibited by a single generation of larvae derived from different crosses, when exposed to various concentrations of the crude toxin suspension (see Fig. S2 in the supplemental material).

TABLE 1.

Differential Cry1Ac toxicities to egg-hatched neonate larvae derived from susceptible strains (S×S), tolerant (T×T) strains, and reciprocal genetic crosses

Insect strain	Toxicitiya									Mean slope ± SE
	LC50			LC90			LC99
	Concn (mg/ml)	95% CI (mg/ml)	RR	Concn (mg/ml)	95% CI (mg/ml)	RR	Concn (mg/ml)	95% CI (mg/ml)	RR
S×S	0.04	0.02-0.06	1	0.01	0.06-0.33	1	0.23	0.11-1.75	1	2.88 ± 0.23
S×T	0.06	0.05-0.08	1.57	0.46	0.32-0.76	4.69	2.65	1.46-6.20	11.65	1.38 ± 0.14
T×S	0.32	0.24-0.43	9.26	2.29	1.54-3.86	23.1	11.23	6.19-25.67	48.48	1.51 ± 0.12
T×T	1.04	0.84-1.29	29.8	4.33	3.25-6.30	43.78	13.84	9.03-24.83	60.98	2.07 ± 0.18

^aNote that concentrations refer to total protein in the crude toxin preparation. 95% CI, 95% confidence intervals; RR, resistance ratio.

Developmental penalties in tolerant larvae.

Exposure to the toxin preparation within their diet had physiological implications for larvae, which exhibited a reduction of food digestion and activation of the immune system. Because growth and developmental effects are often a result of immune activation and physiological changes imposed by the gut-derived toxin, we sought to determine how these penalties are correlated with the presence of toxin in offspring of immune-induced insects. We first examined developmental penalties in susceptible larvae that were exposed to the toxin for the first time. Prior to forming our experimental susceptible population from the two susceptible (ANGR and field) subpopulations, we first examined developmental penalties in the two subpopulations separately to rule out any inherent differences. After exposure of neonates to a sublethal concentration of Cry1Ac preparation, the time required to reach pupation was significantly increased compared to insects kept on toxin-free diet (T = 32.55, degrees of freedom (df) = 140, and P < 0.0001 for ANGR and T = 20.62, df = 179, and P < 0.0001 for the field population) without any overt differences between the susceptible ANGR and field-derived populations (T = 0.05, df = 218, and P = 0.961 for control and T = 2.40, df = 101, and P = 0.019 for Cry1Ac exposed; see Fig. S1 in the supplemental material). Likewise, when larval body mass was compared after 10 days of exposure, the body mass of toxin-exposed larvae was significantly reduced compared to control larvae (T = 1.58, df = 287, and P = 0.115 for control and T = 2.66, df = 339, and P = 0.008 for Cry1Ac exposed) without any overt differences between the susceptible strains (T = 36.48, df = 351, and P < 0.0001 for ANGR and T = 35.57, df = 275, and P < 0.0001 for the field population). After merging the two susceptible populations, one larval cohort was exposed to toxin suspension, and a control cohort was kept on toxin-free diet. In the first generation after toxin application, the toxin-exposed population responded with developmental penalties similar to the two separate populations, i.e., with developmental delays (data not shown) and reduced body weight (T = 0.71, df = 426, and P = 0.476 for control and T = 0.04, df = 510, and P = 0.967 for Cry1Ac exposed; Fig. 1). These measures were significantly different from the population kept on toxin-free diet (T = 35.49, df = 314, and P < 0.0001). However, after five generations of continuous exposure to this level of toxin, the mean larval weight increased such that they were significantly different (T = 23.72, df = 295, and P < 0.0001) to those in the first generation after exposure (Fig. 1, compare tolerant F₁ and F₅). In contrast, mean body mass of the control and toxin-exposed populations were not significantly different in the fifth generation (T = 1.19, df = 140, and P = 0.236), although the tolerant population showed more variation (Fig. 1). This suggests that exposure to the toxin causes significant developmental penalties, which are at least partly reversed after continuous exposure at the same concentration.

FIG. 1.

Developmental penalties (reduced body weight) in 10-day-old susceptible field and induced larvae when egg-hatched neonates were exposed to a sublethal concentration of Cry1Ac suspension (0.01 mg/ml) and compared to larvae not exposed to the toxin (Control). ANGR and field-collected larvae were equally susceptible to the toxin and combined to form a population that was continuously exposed to 0.01 mg of Cry1Ac/ml over five generations. Results are shown for the newly induced population (ANGR+Field)F₁, in the third generation (ANGR+Field)F₃, and in the fifth generation (ANGR+Field)F₅. Each bar represents measurements from 300 to 500 individual larvae. Note that the difference in the induced larvae becomes less pronounced over five generations.

Developmental penalties after toxin increase.

We examined how developmental penalties in the tolerant population respond to an incremental increase in dietary toxin exposure. After the toxin concentration was doubled to 0.02 mg/ml in generation 6 (and maintained at this level until generation 12), we compared the tolerant and susceptible populations. We also compared the developmental penalties in tolerant insects in generations 6 and 12, with larvae being either not exposed to toxin or exposed to increased concentrations of toxin. Developmental penalties (reduced body weight) in the control larvae (never exposed to the toxin) was similar to that of tolerant insects that were not exposed (T = 2.66, df = 116, and P = 0.009 at generation 6 and T = 1.49, df = 278, and P = 0.137 at generation 12; Fig. 2). The developmental penalties of tolerant insects were significantly reduced in subsequent generations when the toxin level was maintained (T = 15.67, df = 146, and P < 0.0001 at generation 6 for a 0.01-mg/ml exposure and T = 13.88, df = 328, and P < 0.0001 at generation 12 for a 0.02-mg/ml exposure), but the increase in toxin concentrations imposed additional developmental penalties, which were again reduced in subsequent generations (Fig. 2). Similarly, delays in the development of surviving susceptible and tolerant (F₁₂, 0.02 mg/ml) larvae exposed to increasing concentrations of Cry1Ac were significantly reduced in the tolerant strain (Fig. 3).

FIG. 2.

Developmental penalties (reduced body weight) in 10-day-old susceptible and tolerant larvae after egg-hatched neonate larvae were exposed to increasing Cry1Ac concentrations. Tolerant larvae were exposed to a 0.01-mg/ml crude Cry1Ac suspension for five generations (F₅) and to a 0.02-mg/ml crude Cry1Ac suspension in subsequent generations (F₁₂). Each bar represents measurements from 300 to 500 individual larvae. Note that the concentration refers to total protein within the crude preparation.

FIG. 3.

Delay in larval development (from egg-hatched neonates to onset of pupation) of surviving susceptible and tolerant larvae (F₁₂, 0.02-mg/ml Cry1Ac suspension) exposed to increasing concentrations of Cry1Ac. No bars indicate that all insects died during larval stages. Note that the concentration refers to total protein within the crude preparation.

Melanization in cell-free hemolymph.

Although prophenoloxidase (PPO) is not directly involved in the tolerance to Bt toxins (10, 21), we sought to determine whether, and under what conditions, PPO-dependent melanization in cell-free hemolymph (plasma) can be used as a quick and reliable measurement of insect immune status. Having an insect strain that is tolerant to Cry1Ac, where the maternal immune status is transmitted to offspring by epigenetic processes, we first examined whether larval immune induction by gut-derived elicitors (larval induction) differs from transmissible induction passed on to offspring (embryonic induction). Because a comparison of the rate of melanization in plasma was difficult in induced and noninduced susceptible larvae due to variable growth and development, we compared similar aged or similar-sized larval offspring from insects maintained for 10 generations on toxin-containing diet. As described for other species (19), the rate of melanization in tolerant larvae (which at the time of measurement were kept on toxin-free diet) was significantly higher than in susceptible larvae (see Fig. S3 in the supplemental material) and a direct reflection of the elevated immune status. Because tolerant larvae showed a significant reduction in developmental penalties even in the presence of the toxin (Fig. 2), we could compare similar sized/aged larvae even if exposed to toxin at the time of measurement. Unexpectedly, when tolerant larvae were reared on a toxin-containing diet at the time of measurement, melanization reactions were inhibited almost to the level of susceptible larvae (see Fig. S3 in the supplemental material).

DISCUSSION

We examined some developmental penalties associated with inducible tolerance to Cry1Ac in a toxin-exposed H. armigera population under laboratory conditions. Since long-term toxin exposure may result in the emergence of recessive target site mutations (8, 15, 17), we exposed larvae to a Bt toxin preparation at concentrations where lethality was low, which enabled us to focus on inducible metabolic effects that are reversible. To rule out that the observed increase in larval survival was based on semidominant preexisting recessive alleles, we performed Cry1Ac bioassays in single female lines in toxin-exposed populations, which did not uncover highly resistant homozygous survivors among the F₂ offspring (not shown).

Instead, our observations suggest that insect larvae respond to gut-derived Cry1Ac toxin with an elevated immune response in the hemolymph (larval induction), which can be measured by the rate of melanization in cell-free plasma, only if larvae were kept on a toxin-free diet prior to plasma isolation. This confirms that toxin exposure causes an elevation of cell-free immune components, such as PPO in hemolymph of H. armigera, which is transmitted to offspring and persists for at least one generation in the absence of toxin. Under these conditions, melanization is a quick and reliable assay of the immune status of larvae as described previously (20, 21). In contrast, hemolymph derived from toxin-containing larvae contains a multitude of factors that among other effects inhibit melanization (5, 7), while amounts of PPO and other immune/metabolic components were increased (data not shown).

The simplest explanation for this observation was that immune and associated metabolic components are induced in the hemolymph by the presence of toxin preparation in the gut. Although the molecular bases of the Bt toxin-mediated induction are not known, possible causes may involve signaling functions of Bt toxin (23) or toxin-mediated damage to gut lining, and contact of gut-derived elicitors with hemolymph, including enterobacteria (3), which enhance the induction and toxicity but are not obligatory to toxicity (13). Whatever the mechanism, gut-derived toxin has multiple effects in hemolymph, such as the reduction of hemolymph proteins and an increase in PPO and the number of circulating hemocytes (6), as well as a reduction in hemocyte phagocytosis (5) and an increase in PO-inhibiting proteins (7). These data suggest that induction of the immune defense by gut-derived elicitors generates a wide range of metabolic responses that includes defense reactions against an intruding toxin or pathogen and also downregulation of the insect's own defense responses, such as coagulation and melanization, that are potentially damaging if allowed to spread within the hemocoel. An increase in protease inhibitors in particular, may explain the observed inhibition of melanization reactions in induced larval plasma (see Fig. S3 in the supplemental material). Such inhibition may have other effects; for example, toxin-exposed susceptible larvae negatively affect predators (16), with some protease inhibitors having insecticidal activities (11). In addition to this larval induction process, the elevated immune status is transmitted mainly by induced tolerant mothers to offspring, apparently by epigenetic means (embryonic induction). Whatever the reasons for the difference in melanization reactions, our data clearly show that induction of the larval immune system by gut-derived toxins differs from the embryonic induction responsible for the transmission of the elevated immune status to offspring. Because females are the heterogametic sex in Lepidoptera, this transmission inconsistent with sex linkage but consistent with maternal inheritance. The maternal effect is also reflected in the probit analysis (see Fig. S2 in the supplemental material), which shows a clear increase in tolerance of larvae derived from tolerant mothers relative to those derived from tolerant fathers. The exact mechanism of this transmission is not known. In contrast to the gut-derived larval induction process, the embryonic induction is less damaging to the insect. In particular, the rate of melanization in cell-free plasma is more reproducible in offspring from induced larvae that have been kept on toxin-free diet before the melanization assays.

Likewise, the developmental penalties imposed by the two immune induction processes differ considerably. Whereas larvae exposed to toxin preparation in their gut are delayed in growth and development, the transmission of the elevated immune status to offspring is much less costly, affecting growth and development less than in their toxin-containing parents, while elevated immune and tolerance levels are maintained. In larvae that were continuously exposed to toxin, developmental penalties became significantly less within a few generations. Our data suggest that developmental penalties are heavily influenced by the toxin concentration, as well as the presence of toxin in the gut. This raises the important question as to whether the reduction in developmental penalties on subsequent generations are the result of an metabolic adaptation process under selection pressure or due to changes in the relative contributions of the two induction processes (larval and embryonic). Given that the two induction processes impose different levels of developmental penalties, we propose that, in insects that were continuously exposed to the same toxin concentration, the initial induction is exclusively larval induction. In contrast, in subsequent generations, an increasing ratio of embryonic induction is responsible for the reduction in developmental penalties over subsequent generations, while tolerance to the toxin is maintained or increased.

When susceptible larvae were exposed to the toxin for the first time, developmental delays and reduced body size/mass were direct indicators of the developmental penalties accrued by the larval induction of immune activities. While the elevated immune status was maintained for at least two subsequent generations, the overt signs of developmental penalties decreased within two generations in the presence of toxin (Fig. 3) and even faster in the absence of the toxin (data not shown). This suggests that the induction is reversible, probably to prevent the excessive expenditure of internal resources if toxin is no longer present. In addition, our data suggest that developmental penalties are reduced over subsequent generations regardless of the levels of continuous toxin exposure.

In summary, we have shown that insects are induced by gut-derived toxins to acquire a tolerance through elevated immune status (larval induction) that is transmitted to offspring by an epigenetic mechanism showing a maternal effect (embryonic induction). The elevated immune status is correlated with tolerance to the toxin, while overall developmental penalties depend on the ratio of contributions from the two induction processes. Continuous exposure of insects to toxin reduces the overall developmental penalties due to an increasing contribution of embryonic induction, which was much less costly than larval induction. This suggests that insects can become tolerant to Bt toxin over time due to increased immune induction without suffering high developmental penalties. This information may have direct impacts on the management of Bt toxin usage and resistance in cropping systems.