Differential side-effects of Bacillus thuringiensis bioinsecticide on non-target Drosophila flies

Bioinsecticides based on Bacillus thuringiensis (Bt) spores and toxins are increasingly popular alternative solutions to control insect pests, with potential impact of their accumulation in the environment on non-target organisms. Here, we tested the effects of chronic exposure to commercial Bt formulations (Bt var. kurstaki and israelensis) on eight non-target Drosophila species present in Bt-treated areas, including D. melanogaster (four strains). Doses up to those recommended for field application (~ 106 Colony Forming Unit (CFU)/g fly medium) did not impact fly development, while no fly emerged at ≥ 1000-fold this dose. Doses between 10- to 100-fold the recommended one increased developmental time and decreased adult emergence rates in a dose-dependent manner, with species-and strain-specific effect amplitudes. Focusing on D. melanogaster, development alterations were due to instar-dependent larval mortality, and the longevity and offspring number of adult flies exposed to bioinsecticide throughout their development were moderately influenced. Our data also suggest a synergy between the formulation compounds (spores, cleaved toxins, additives) might induce the bioinsecticide effects on larval development. Although recommended doses had no impact on non-target Drosophila species, misuse or local environmental accumulation of Bt bioinsecticides could have side-effects on fly populations with potential implications for their associated communities.


Results
Btk formulations adversely impact the development of D. melanogaster. In a dose-response assay, emergence rates (ER) and developmental times (DT) of wild-type D. melanogaster Canton S flies exposed to doses up to 10 7 CFU/g of DELFIN A in a standard low-protein/high-sugar fly medium were similar to those of the control unexposed group (Fig. 1a,b; Table 1). At higher doses, both ER and DT were affected in a dosedependent manner: ER was reduced by 17% at 5 × 10 7 CFU/g (although not statistically significant), up to 100% at 10 9 CFU/g, at which no individual reached the pupal stage. The lethal dose 50 (LD50) was estimated between 5 × 10 7 and 10 8 CFU/g (Fig. 1a). DT was increased of about 0.5 day at 5 × 10 7 CFU/g (+ 4% versus controls) to up to 1.5 days (+ 14%) at 10 8 CFU/g ( Fig. 1b; Table 1). The sex-ratio at emergence (SR, proportion of males) was strongly biased towards males at 10 8 CFU/g, with 58% more males compared to the control (Supplementary information S2).
We observed no change in ER using the same dose range of the Btk Cry-free strain 4D22 (Fig. 1a,e; Table 1) and the non-pathogenic Bacillus subtilis (Fig. 1a, Table 1), two controls for the effect of ingestion of high loads of spores. In contrast, addition of the formulation of Bt var. israelensis VectoBac WG reduced ER by 89% only at 10 9 CFU/g (~ 2000 times the recommended dose; Fig. 1a; Table 1; Supplementary information S1). DT varied with the dose of Btk 4D22, the differences being mainly between doses but not with the control. DT increased by ~ 1.5 days at the highest dose of VectoBac WG ( Fig. 1b; Table 1) and showed a similar trend with B. subtilis (P = 0.06; Fig. 1b; Table 1). None of these treatments influenced dramatically the SR (Supplementary information S2).
To test whether these effects are generic to Btk formulations, the fly development was evaluated on two other formulations, DELFIN B (same brand) and Scutello DF (brand Dipel), at the critical doses 10 8 and 10 9 CFU/g. As DELFIN A, these formulations contain spores and Cry toxins such as Cry-1A as pro-toxins of ~ 130 kDa, activated toxins of ~ 60-70 kDa, but also as smaller fragments 20 (Fig. 1e, red asterisks). ER remained unchanged at 10 8 CFU/g whereas no individual reached pupation at 10 9 CFU/g on DELFIN B and very few individuals reached  www.nature.com/scientificreports/ Developmental exposure to Btk formulation does not strongly influence fitness-related traits in adults. Despite a large variation between the two independent sets of experimental replicates (Table 1), the longevity of adults reared on 5 × 10 6 CFU/g of DELFIN A in low-protein/high-sugar medium was similar to that of non-exposed controls (Fig. 3). Males and females which developed on the two highest doses showed a moderate longevity benefit, higher in females for 10 8 CFU/g (Fig. 3a,b,d,e; Table 1). Males generally survived better than females (Table 1) but their longevity benefit of developing on 10 8 CFU/g was only observed in one experiment (Fig. 3b,e). The number of offspring produced by the 15 females of each fly group during the longevity experiment varied depending on both the experiment and the DELFIN A dose (Table 1). In the 1st experiment, adults from larvae reared on 10 8 CFU/g had fewer offspring compared to the controls and to adults developed on the other doses whereas the total offspring number varied regardless of the DELFIN A dose in the 2nd experiment (Fig. 3c,f, Table 1). Table 1. Results of statistical analyses to assess the effect of the dose of formulation/spore production and its interaction with the treatment, the larval instar, the experiment, the sex, the fly strain and the fly species when appropriate. See figures for post hoc comparisons of the doses with the control dose. Significant statistical differences are indicated in bold strains (wild-type Nasrallah and Sefra, and double mutant YW1118) was not impacted at doses up to 10 7 CFU/g of DELFIN A in a high-protein/sugar-free medium. In contrast, the ER of each strain was greatly reduced and DT was increased at higher doses ( Fig. 4a,b, Table 1), with no individual reaching the pupal stage at 10 9 CFU/g (LD50 between 10 8 and 10 9 CFU/g). At 10 8 CFU/g, the magnitude of effects on Canton S flies was lower than that observed on the low-protein/high-sugar medium (see Fig. 1a,b). At this dose, the ER varied between strains, the largest reduction being observed for Sefra (Table 1). We observed no dose-dependent bias in SR (Supplementary information S3).

Btk formulation affects differently other Drosophila species.
For seven other Drosophila species from different phylogenetic clades that co-occur in the field [48][49][50][51] , doses up to 10 6 CFU/g of DELFIN A in a highprotein/sugar-free medium had no effect on ER and DT, whereas all individuals failed to reach the pupal stage at 10 9 CFU/g (Figs. 5, 6). The amplitude of development alterations at 10 7 and 10 8 CFU/g varied among species (Figs. 5, 6; Table 1). All species were affected at 10 8 CFU/g as was D. melanogaster (see Fig. 4a for comparison). D. simulans and D. busckii had unchanged ER, but DT was slightly increased for D. simulans (although slightly reduced at 10 7 CFU/g; similar results with a Japanese strain, data not shown) and strongly increased for D.

Development alterations may result from a synergy between formulation components.
Because some additives of commercial formulations might contribute to the observed effects, a DELFIN A suspension was dialyzed to remove low molecular weight additives, resuspended, and mixed with low-protein/high-sugar fly medium. At 10 7 CFU/g, the suspension did not affect ER and DT, while no individual pupated at 10 9 CFU/g ( Fig. 7a; Table 1). At 10 8 CFU/g, ER was not modified but DT increased in one experimental set by ~ 1 day, partially reproducing the changes observed without dialysis (Fig. 7a,b; see also Fig. 1a,b, Table 1; 3 independent experiments for ER, 2 independent experiments for DT). The Cry1A profiles of DELFIN A suspensions (dialyzed or not), included a band for the 130-kDa pro-toxins and a band at 60-70 kDa likely representing the activated toxins, but also smaller fragments resulting from the degradation of Cry1A (Fig. 7c). We further explored the respective roles of Btk toxin fragments and spores in the alterations of D. melanogaster development through dialysis experiments followed by successive centrifugations to remove most of the spores and toxin crystals. Despite variation between experiments, ER was strongly affected only in one of the three experiments while DT was always significantly increased in the presence of centrifuged supernatants (Supplementary information S6). Noteworthy, the D. melanogaster development was not impacted in the presence of a homemade production of the Btk strain 4D1 containing spores, toxins, but no additives, even at the highest dose (Supplementary information S7). www.nature.com/scientificreports/

Discussion
Our study tested the side-effects of ingestion of Bt bioinsecticide commercial formulations (mainly made of Bt kurstaki strains (Btk) but also of Bt israelensis (Bti)) during the development of eight non-target species of Drosophila naturally present in treated areas. Although the recommended doses for one formulation field spray did not affect the Drosophila development, those 10 and 50 times higher markedly induced mortality and/or developmental delay in at least two of the species tested. We can extrapolate from our data that these doses may affect six of the eight tested species and the four strains of D. melanogaster. The development alterations were already strong at these doses, suggesting an occurrence from lower ingested doses but not visible in our experimental set-up. In addition, in our experimental conditions, a single Drosophila larva could probably not process 1 g of medium during its development. Further analyses, maybe at molecular level, would be required to determine the minimal dose affecting the fly larva. Furthermore, all the tested species except D. simulans were strongly affected at a 100 times the field spray dose, and no or very limited fly development occurred at the highest tested dose, equivalent to 1000 times the maximum field dose but far below the acute intoxication doses classically used in numerous studies 5 . The recommended doses for each spraying of stabilized formulation are given for a homogeneous and dry area, without overlapping. In the field, recommended repeated sprays and post-spray rainfall washouts may increase the concentration of Bt spores and toxins in both space and time. While a dose 1000 times the recommendations would be hardly reached in the field, the minimum doses at which the fly development was impacted and the lower doses from which developmental changes appeared could be reached. Our data also identified a first developmental window of susceptibility to Btk formulation during the 1st larval instar mainly explaining the adverse effects, while a second event of mortality seemed to occur at the pupation period. The impacts of Btk formulations on the development of D. melanogaster are consistent with growing evidence suggesting partial specific targeting of Bt 12,26,27 .The consensus on the mode of action of Bt after ingestion by insects relied until recently on the key steps of the specific binding of proteolyzed Bt toxins to midgut epithelial cell receptors, defining targets for each Bt subspecies 12,15,17 . Several primary and secondary types of toxin receptors have been identified in the Lepidoptera and Diptera mosquitoes such as cadherin-like proteins, aminopeptidases, GPI-anchored alkaline phosphatases 8 , and more recently the ATP dependent binding cassette reporter C2 52 . No orthologues of the Lepidoptera cadherin-like Cry receptors were found in Drosophila 52 , supporting the idea of the lack of effect of Btk toxins on these flies. Yet, Drosophila flies may have other types of Cry receptors, therefore explaining the developmental impacts observed, but this remains to be investigated. In addition, the possible lack of solubilization of the protoxin crystals and of proteolytic activation of toxins by proteases in the fly gut, both required for Cry activity in insects' larvae 15 , would be possibly compensated by the substantial amounts of active Cry1A toxin fragments in Btk formulations. Other Btk-synthesized toxins present in the formulations could also be players in the observed cross-order activity since some, like Cry2A, have an insecticidal effect on both Lepidoptera and Diptera 53 .
Since ingestion of Bacillus subtilis or Btk Cry-free does not affect the development of D. melanogaster, the observed development alterations cannot result solely from a severe disturbance of digestion and nutrient absorption/competition due to the presence of high loads of spores/bacteria in the larval gut throughout development. This suggests a synergistic action of Btk spores and Cry toxins, consistent with the Bt action models on insect larvae, i.e. the breach of the intestinal epithelium allowing colonization of the hemocoele by the gut bacteria, including Bt spores 15,17,18 . The partial mimicry of mortality rates and developmental delays in preliminary dialysis assays would also support a contribution of diffusible low molecular weight compounds in Btk formulations (e.g. residues of culture media, salts, additives) to these development alterations. Furthermore, there is no impact on the development of D. melanogaster of the ingestion of homemade spores and Cry toxins of the Btk strain 4D1 used without additives even at the highest dose (or HD1, a reference strain used also as a control). Unlike commercial Btk formulations, Btk 4D1 culture contains few activated Cry toxins and smaller toxin fragments, advocating the possible contribution of such fragments to the cross-order activity of Btk formulations on Drosophila. Completion of these preliminary tests is required to further investigate the mechanisms of the harmful effects of Btk formulations on the development of Drosophila and unravel the respective roles of the synergy spores/toxins/crystals and of formulation additives.
As reported for D. suzukii exposed to Btk cultures 45 , D. melanogaster mortality on the Btk formulation occurred mainly during early development. Only ∼40% of the 1st and 2nd instar larvae died at the highest dose tested (Fig. 2) while no individual reached the pupal stage, the remaining mortality likely occurring during, or at the end of, the 3rd larval instar, possibly due to the delayed action of the gut accumulated Btk spores and toxins at the onset of pupation. Interestingly, developmental alterations (mortality, delayed emergence) mimicked those typically caused by nutritional stress in insect larvae 54,55 . Accordingly, the developmental alterations were partially rescued on a protein rich fly medium, probably by a compensatory protein intake, as in other arthropod species [55][56][57] . Also, the sex ratio of flies was strongly biased towards males after development on the Btk formulation www.nature.com/scientificreports/ dose affecting fly emergence and under protein restriction. This highlights the importance of nutritional conditions such as protein restriction, added to sex-specific differences in larval susceptibility to environmental stressors, here the accumulation of Btk formulation, as already reported previously in D. melanogaster 58 .
The development on sublethal doses of Btk formulation did not dramatically affect the longevity of D. melanogaster adults, nor their lifetime offspring number. Exposure during development to doses of Btk formulation that slightly and strongly reduced the likelihood of reaching the adult stage even provided a dose-dependent longevity benefit to the surviving flies and tended to increase their offspring number. Exposure to the Btk formulation throughout development probably selected resistant and/or tolerant individuals, reminding the increased longevity of adult insects having survived a nutritional stress during development 59,60 , or withstood environmental stressors 61 .  www.nature.com/scientificreports/ simulans, D. hydei, D. immigrans, and D. busckii belong to the guild of cosmopolitan domestic Drosophila species, D. subobscura is a sub-cosmopolitan species, and D. busckii is an opportunistic frugivorous species 62,63,64 . They all coexist frequently and compete on the same discrete and ephemeral rotting fruit patches, with seasonal variations in the composition of the fly community [47][48][49]62 . Differences in species susceptibility to Btk formulations could modify the conditions of larval competition, therefore adding local and temporal variations in the composition of Drosophila communities. The potential side-effects of Bt sprays on non-target Drosophila communities would be hardly predictable as they would depend on spatial patterns of Bt accumulation. A formal mesocosm study of Drosophila community dynamics under exposure to Btk formulation, at least under semi-field conditions, would help to identify the consequences of Bt accumulation on species competition and community composition. The exposure to Btk formulation also impacted the development of the invasive D. suzukii, as recently reported 45 , this species being the most susceptible with effects already clearly detectable at only 10 times the recommended spray dose. Compared to the other tested species living on rotten fruits, D. suzukii threatens fruit production since it feeds and lay eggs inside healthy ripening soft fruits [63][64][65] , colonizing orchards and vineyards earlier during the fruit season. The higher susceptibility of D. suzukii to the accumulation of Btk formulation in the environment might reduce the possible ecological burden of its invasion for local communities of fruit-eating Drosophila in orchards. Alternatively, since D. suzukii attacks on fruits can accelerate their decomposition, its increased susceptibility may reduce the number of fruits available for the rotting fruit-eating Drosophila species.
Overall, our data show that the ingestion of Btk bioinsecticides above the recommended spray doses can potentially impact non-target Drosophila flies, with an effect amplitude depending on both the formulation and the fly species. Although our study was carried out under controlled laboratory conditions, which may considerably differ from those of the field (e.g. temperature, pH, humidity, food availability, presence of predators/ parasites/pathogens, etc.…), standard laboratory strains and flies derived from recently collected populations exhibited similar patterns of developmental alterations, suggesting our results are likely generalizable. Recent studies have reported similar adverse side-effects due to repeated spraying of the Bti formulation on non-target organisms 25 , and indirectly on predators via food webs 66 . From these studies and our data here, care should clearly be taken when using Bt bioinsecticides to avoid, or at least minimize, potential side-effects on non-target organisms and therefore on biodiversity. At last, D. melanogaster could serve as a model species to identify the mechanisms underlying these side effects and/or the potential emergence of resistance to these bioinsecticides. All the flies were maintained at controlled densities (150-200 eggs/40 ml of fly medium) under standard laboratory conditions (25 °C, or 20 °C for recently collected species, 60% relative humidity, 12:12 light/dark cycle), on a high-protein/sugar-free fly medium (10% cornmeal, 10% yeast, 0% sugar). In our laboratory, the D. melanogaster Canton S was also reared on a standard low-protein/high-sugar medium (8% cornmeal, 2% yeast, 2.5% sugar). For each experiment, eggs, larvae and flies were maintained under standard conditions.

General method of intoxication and dose-response assay.
For the dose-response assays, formulations and spore productions were serially diluted in buffer, and 100 µl of each dilution was homogenized thoroughly with 1 g of fly medium (100 µl/g doses). Eggs and defined larval instar were collected from stock vials and transferred to the intoxication vials and dishes as described below. They were then reared until the fly emergence, and, for larval susceptibility tests, until the desired development stage was reached, or for 24 h for early larvae of the 1st and 2nd instars. Equivalent control groups were transferred on fly medium homogenized with the same volume of buffer alone.

Development-related traits and larval survival. Developmental traits upon intoxication throughout
the entire development of the D. melanogaster strains and the other Drosophila species were evaluated on a precise number of viable eggs collected from mass oviposition and transferred to intoxication vials containing fly medium (high sugar/low protein or high protein/sugar free as indicated) mixed with formulations or spore productions at doses ranging from 1 × 10 5 or 5 × 10 5 CFU/g (mean equivalent to the manufacturer recommendations; Supplementary information S1) to 10 9 CFU/g. Eggs were let to develop until fly emergence. Egg density was set at 8-10 eggs/g of medium (10 eggs/g on 2 g medium in small vials Ø 3.3 cm, surface ~ 8.5 cm 2 , 0.24 g/cm 2 for tests on D. melanogaster Canton S; 8 eggs/g on 6 g medium in large vials Ø 4.6 cm, surface ~ 16 cm 2 , 0.37 g/ cm 2 for strains and species comparisons), except for D. hydei, D. suzukii and D. immigrans for which the egg density was reduced by half because of their reproductive biology (5 eggs/g on 6 g). Numbers and sex of emerging flies were recorded once a day until the day on which the first pupae of the next generation forms. The emergence rate (proportion of flies emerged from the initial eggs), the developmental time (mean number of days for development completion) and the sex-ratio (proportion of male flies) were calculated for each replicate vial.
For the larval susceptibility tests, survival was measured on 20 eggs/larvae of D. melanogaster Canton S at a suitable instar collected from a 4-h mass oviposition and transferred to small dishes (Ø 3 cm, surface ~ 7 cm 2 ) containing 1 g of high-protein/sugar-free fly medium (less limiting for early larval development) homogenized with DELFIN A doses ranging from 10 5 CFU/g to 10 9 CFU/g. First and 2nd instar larvae were used since growth is exponential during these two instars 39,69 and larvae are more likely to be heavily exposed to the bioinsecticide. Proportion of surviving larvae was measured at the indicated developmental stage for the cumulative survival test, or 24 h later. For cumulative survival, unhatched eggs were discarded from the counting. The pH of the fly medium was measured for the Btk dose-responses; neither the presence of Btk formulation nor the dose altered it ( Supplementary Information S4).
Adult fitness-related traits. For longevity and total offspring number measurements, D. melanogaster Canton S eggs from mass oviposition were transferred to several vials with low-protein/high-sugar medium mixed with 5 × 10 6 , 5 × 10 7 or 10 8 CFU/g of DELFIN A. For each dose, flies from the vials were pooled 2 days after emergence and groups of 15 males and 15 females were transferred on the same medium without DELFIN A. Flies were transferred to a new vial every 3-4 days and previous vials were incubated for the offspring to develop. Mortality and sex of dead flies were recorded daily until the last fly died. Offspring numbers were counted from the first emergence until pupae of the next generation appeared. Two experimental blocks were set. Due to dif-Scientific RepoRtS | (2020) 10:16241 | https://doi.org/10.1038/s41598-020-73145-6 www.nature.com/scientificreports/ ferences in the duration of the experiment, the offspring numbers of all vials of each dose of DELFIN A were summed for each experimental block.
Dialysis and cry toxin analysis. For some products, additives can be more harmful than the active ingredient 70 . To eliminate low molecular weight additives present in the formulation, a DELFIN A suspension at 2 × 10 10 CFU was dialyzed against PBS (KH 2 PO 4 1 mM, Na 2 HPO 4 (2H 2 O) 3 mM, NaCl 154 mM, pH 7.2), overnight at 4 °C, using an 8-10 kDa cut-off membrane (ZelluTrans, Roth). CFUs of the dialyzed suspension were estimated as above. The effects on the emergence rate (ER) and developmental time (DT) of D. melanogaster Canton S were analysed on 20 eggs, at 10 7 , 10 8 and 10 9 CFU/g of dialyzed and also centrifuged suspension mixed with 2 g of low-protein/high-sugar fly medium. The dialyzed suspension was subject to a 12.5% SDS-PAGE and compared to the non-dialyzed suspension after silver staining. The presence of Cry1A pro-toxins, activated toxins and toxin fragments was probed by Western-blot using an in-house anti-Cry1A rabbit polyclonal antibody.
Data analysis. Data were analysed with mixed-effects models with replicates as random effects. Dose of Btk formulation/spore production, D. melanogaster strain, Drosophila species or developmental stage, experimental block where necessary, and appropriate 2-way interactions between these factors, were included as fixed effects. Main fixed effects and their interactions were tested with log-likelihood ratio tests. Post hoc pairwise comparisons were made for D. melanogaster strains, formulation/spore treatments, and between the control and the other doses. Data of emergence rate, sex-ratio, and larval survival were analysed with generalized linear models with binomial distribution and logit link; for emergence rate data, the model was also bias-corrected to correct for multiple 0 values, and with replicate as random effect. To run the post-hoc analysis, the same model including replicates as fixed effect was applied to the data of emergence rate and provided similar results. Developmental time (1/x transformed) and offspring number were analysed with linear models. Adult longevity data were analysed with proportional hazard Cox regression models, including fly sex and formulation dose as fixed effects, and replicate as a random effect. Analyses were performed in R 71 using the packages lme4 72 , brglm 73 , multcomp 74 , survival 75 , and coxme 76 .