Anthrax toxin component, Protective Antigen, protects insects from bacterial infections

Anthrax is a major zoonotic disease of wildlife, and in places like West Africa, it can be caused by Bacillus anthracis in arid nonsylvatic savannahs, and by B. cereus biovar anthracis (Bcbva) in sylvatic rainforests. Bcbva-caused anthrax has been implicated in as much as 38% of mortality in rainforest ecosystems, where insects can enhance the transmission of anthrax-causing bacteria. While anthrax is well-characterized in mammals, its transmission by insects points to an unidentified anthrax-resistance mechanism in its vectors. In mammals, a secreted anthrax toxin component, 83 kDa Protective Antigen (PA83), binds to cell-surface receptors and is cleaved by furin into an evolutionary-conserved PA20 and a pore-forming PA63 subunits. We show that PA20 increases the resistance of Drosophila flies and Culex mosquitoes to bacterial challenges, without directly affecting the bacterial growth. We further show that the PA83 loop known to be cleaved by furin to release PA20 from PA63 is, in part, responsible for the PA20-mediated protection. We found that PA20 binds directly to the Toll activating peptidoglycan-recognition protein-SA (PGRP-SA) and that the Toll/NF-κB pathway is necessary for the PA20-mediated protection of infected flies. This effect of PA20 on innate immunity may also exist in mammals: we show that PA20 binds to human PGRP-SA ortholog. Moreover, the constitutive activity of Imd/NF-κB pathway in MAPKK Dsor1 mutant flies is sufficient to confer the protection from bacterial infections in a manner that is independent of PA20 treatment. Lastly, Clostridium septicum alpha toxin protects flies from anthrax-causing bacteria, showing that other pathogens may help insects resist anthrax. The mechanism of anthrax resistance in insects has direct implications on insect-mediated anthrax transmission for wildlife management, and with potential for applications, such as reducing the sensitivity of pollinating insects to bacterial pathogens.


Introduction
Anthrax is a major zoonotic disease of wildlife caused by the Gram-positive bacterium Bacillus anthracis or by the closely related B. cereus biovar anthracis (Bcbva), which combines the chromosomal background of B. cereus with the toxin-encoding pXO1 and the capsule-encoding pXO2 plasmids of B. anthracis [1]. Bcbva causes "sylvatic anthrax", a prevalent and persistent cause of death in a broad range of mammalian hosts in the rainforest ecosystem of Taï National Park (TNP), Côte d'Ivoire. Sylvatic anthrax was responsible for as much as 38% of wildlife mortality observed over 26 years in TNP and is predicted to accelerate the decline and possibly result in the local extinction of chimpanzee populations [2, 3]. Moreover, viable spores of Bcbva were detected in blow flies sampled inside and outside the chimpanzee habitats in TNP [2]. In addition to TNP, Bcbva has been isolated from dead or sick mammals in other West and Central Africa regions, such as Cameroon, Central African Republic, and Democratic Republic of Congo [4], as well as in regions outside of Africa, such as Texas and Louisiana, USA [5].
In contrast, B. anthracis, which is distributed globally [6], causes "classical anthrax" that is most commonly observed in arid nonsylvatic ecosystems, such as African savannahs in Krüger National Park [7], South Africa and Etosha National Park, Namibia [8,9]. In these ecosystems, major anthrax outbreaks display strong seasonal variation, where the mortality coincides with rainfall cycles and the wet season [7][8][9]. In these nonsylvatic ecosystems, it has also been suggested that flies spread B. anthracis spores from carcasses through the environment, potentially contributing to anthrax transmission [10,11].
In nature, insects can serve as the vectors for mechanical anthrax transmission to humans, cattle, and the mammalian wildlife [2, 6,10,[12][13][14]. Biting flies have been shown to acquire anthrax-causing bacilli from infected animals and directly transmit them to other mammals [14]. Similarly, non-biting flies feed on the bodily fluids of infected carcasses and deposit contaminated feces or vomit on nearby vegetation, creating a reservoir for herbivores to potentially contract anthrax while grazing [7]. Other insects, such as mosquitoes and ticks, may also contribute to anthrax transmission [15][16][17], raising the interesting question as to how insect vectors tolerate anthrax-mediated toxicity.
Insects depend on their innate immunity for protection against pathogens. During an infection, insects produce antimicrobial peptides (AMP), while circulating hemocytes phagocytose sylvatic anthrax in tropical rainforests [1][2][3][4]46], presented an opportunity to test the effects of anthrax toxin components on the sensitivity of Drosophila to this Bcbva-related bacterium.
When feeding on anthrax-infected mammals, insects can be exposed to anthrax toxin components. We observed that the oral exposure of flies to a mix of purified components PA 83 , LF, and EF in the absence of the bacterial challenge did not alter their survival (S1 Fig). In the presence of the anthrax toxin components, flies showed increased survival time (quantified as the increased median survival time) to B. cereus infection by 25 hours, P < 0.0001 ( Fig 1A). While neither EF nor LF affected the sensitivity of flies to B. cereus (Fig 1B and 1C), PA 83 alone delayed the death of infected flies (delay of the median survival time by 26 hours, P < 0.0001) (Fig 1D). In addition to flies aged 4-5 days (Fig 1), the protective effect of PA 83  The effect of PA 1β 13 −1α 1 loop and PA 20 on insects' sensitivity to bacterial infections PA 83 contains four functional domains (Fig 2A): the host receptor-binding domains 2 and 4 (PA D2 and PA D4 ), the multimerization domain 3 (PA D3 ), and the PA 20 -containing domain 1 (PA D1 ) [29]. After furin cleavage, the portion of PA D1 that remains on PA 63 is called domain 1' (PA D1' ). To investigate the region within PA 83 responsible for the protection of infected flies, we recombinantly expressed all PA 83 domains (Fig 2B). We observed that PA 20 -containing
Like insects, plants rely only on innate immunity to protect them from pathogens. In plants, programmed cell death, or necrotic lesions, often occur at the site of infection to prevent the systemic spread of invading microorganisms [47]. Previously we observed that amino acids 181-200 of PA 83 caused necrotic lesions in Nicotiana benthamiana [48], suggesting that they trigger an innate immune response in plants. This peptide forms an unstructured loop (between β-sheet 13 and α-helix 1, referred to as the "1β 13 −1α 1 loop") within domain 1 of PA 83 [29], which contains the furin-cleavage site [49] (Fig 2A). The loop residues 181-192 remain on PA 20 after furin processing. We assessed the ability of the entire loop (residues 181-200) and the furin-cleaved portion (residues 181-192) of the 1β 13 −1α 1 loop to reduce the sensitivity of flies to bacterial infections. To some degree, both peptides delayed the death of B. cereusinfected flies (delay of the median survival time 11 hours, where some flies survive for 29 hours longer, P < 0.0001) (Fig 2D) while also not directly altering bacterial growth (S3 Fig). Therefore, the residues within the PA 20 1β 13 −1α 1 loop are, in part, responsible for the PA 83 -mediated reduction in the sensitivity of flies to bacterial infections.
When feeding on anthrax-infected mammals, insects can be exposed to PA in its native PA 83 state, as well as in the furin-cleaved state, PA 63 and PA 20 . Further testing revealed that within PA 83 , it was PA 20 and not PA 63 responsible for the reduction in fly sensitivity to B. cereus: PA 20 protected infected flies without and with PA 63 (delay of the median survival time by 24 hours, P < 0.0001) (Fig 2E).

PA 20 protection is broad-spectrum and insect-directed
We tested the sensitivity of D. melanogaster to oral administration of the toxigenic Sterne strain of B. anthracis harboring the pXO1 plasmid, and the ΔSterne strain, a pXO1-cured derivative of the Sterne strain. While flies were sensitive to both B. anthracis strains in the vegetative state, surprisingly, the Sterne strain was less pathogenic than the toxin-negative ΔSterne strain (delay of the median survival time by 21 hours, P < 0.0001) ( Fig 3A). Moreover, we investigated the sensitivity of flies to spores of the same B. anthracis strains. Interestingly, while uninfected flies feeding solely on sucrose can survive for 10-14 days, the oral exposure to B. anthracis spores extended the longevity of flies, with the Sterne strain extending the life significantly longer than the ΔSterne strain (delay of the median survival time by 57 hours, P < 0.0001) ( Fig 3A). These data show that the toxin-containing strain of B. anthracis is less pathogenic to flies and extends the survival of flies, compared to the toxin-negative strain of bacteria.
When feeding on anthrax-infected mammals, insects can be exposed to pXO1-containing bacteria, such as Bcbva and B. anthracis cells, as well as secreted toxins components, including PA 20 . We show that exogenously added PA 20 protects flies from the Sterne strain ( Fig 3B) and ΔSterne strain (Fig 3C). PA 20 delayed the median survival time of Sterne strain infected flies by 19 hours (P < 0.0001) and of ΔSterne strain infected flies by 16 hours (P < 0.0001), although some flies were protected by as long as 25-70 hours.
Mosquitoes may also contribute to anthrax transmission [15,17] and were previously found to be insensitive to anthrax toxin [50]. We tested whether PA 20 -mediated protection occurs in B. anthracis-infected mosquitoes, Culex quinquefasciatus. Interestingly while mosquitoes were sensitive to the vegetative Sterne and ΔSterne strains of B. anthracis, in contrast to Drosophila, their sensitivity was similar to both strains ( Fig 3D). Moreover, in contrast to our fruit fly experiments, oral exposure to B. anthracis spores did not affect the longevity of mosquitoes ( Fig 3E). Nevertheless, PA 20 significantly protected Culex from both strains of bacteria and delayed the median survival time of Sterne strain infected mosquitoes by 102 hours (P < 0.0001) ( Fig 3D) and of ΔSterne strain infected mosquitoes by 96 hours (P < 0.0001) ( Fig 3D). Collectively, these results show that PA 20 causes the delay of B. anthracis-induced death in Drosophila and Culex.
The effects of PA 20 extend beyond bacteria of the B. cereus group, as it enhanced the survival of flies exposed to related B. subtilis and unrelated Gram-negative Serratia liquefaciens and Escherichia coli.   To test whether PA 20 exerted its protective effects by suppressing the growth of bacteria, we measured the growth rates of bacteria in the presence or the absence of PA 20 . PA 20 did not affect the growth of B. cereus and S. liquefaciens in liquid culture (Fig 4D and 4E) or on solid media (Fig 4F and 4G), suggesting that PA 20 reduces the sensitivity of flies to insecticidal bacteria through an insect-oriented mechanism.

PGRP-SA binds to PA 20 and is necessary for PA 20 -mediated protection
Lys-type peptidoglycans are recognized by a circulating extracellular heterodimer receptor consisting of peptidoglycan-recognition protein-SA (PGRP-SA) and GNBP1 [19,51] (Fig 5A). This peptidoglycan binding triggers a proteolytic cascade that includes ModSP, Grass, Sphe, Psh, and SPE [18]. Ultimately, SPE cleaves pro-Spätzle to release the mature Toll ligand, Spätzle (Spz), and trigger the subsequent intracellular Toll signaling leading to the degradation of the inhibitor of κB, cactus, the activation of NF-κB (Dif), and the transcription of AMPs.
The five amino acid consensus sequence of lys-type peptides recognized by PGRP-SA is AQKA/SA/S [52]. The PA 20 portion of the 1β 13 −1α 1 loop contains a similar sequence: 186-KQKSS-190. To test whether the protective capacity of PA 20 might be due to its ability to bind to PGRP-SA (Fig 5A), we used purified recombinant D. melanogaster PGRP-SA and its human orthologue PGRP-S, as they share 60% sequence homology and the peptidoglycanbinding site is closely conserved between the two proteins [53,54]. Moreover, just like PGRP-SA, PGRP-S binds to lys-and dap-type peptidoglycans (dap is a derivative of lys) [54]. We performed an ELISA, where PRGP-SA or PGRP-S was added to plates coated with PA 20 or peptidoglycans. PRGP-SA ( Fig 5B) and PGRP-S ( Fig 5C) bound PA 20 and peptidoglycans and were detected with anti-PGRP antibodies.
To validate this observation, we analyzed the ability of PGRP-S to bind to PA 20 in a co-precipitation assay, which relied on the observation that dap-type peptidoglycan and PA 20 are partially insoluble. We incubated PGRP-S with either peptidoglycan or PA 20 , followed by centrifugation and washing of pellets. Western blot analysis demonstrated that PGRP-S bound to PA 20 and peptidoglycan, and thus, was pulled down into pellets ( Fig 5D). Moreover, PA 20 did not affect the sensitivity of PGRP-SA loss-of-function mutant flies to B. cereus, showing that this gene is necessary for the PA 20 -mediated phenomenon ( Fig 5E). Collectively, these results suggest that PA 20 may exert its protective effects in Drosophila by binding to PGRP-SA ( Fig 5A).

The Toll pathway is necessary, and the Imd pathway is sufficient for PA 20mediated protection
We investigated whether the Toll pathway, the Imd pathway, and the phagosomal ROS production are involved in PA 20 -mediated protection of infected flies. Toll pathway GNBP1, ModSP, Grass, Sphe, Psh, SPE, Spz, and cactus loss-of-function mutants were infected with B. cereus in the absence and the presence of PA 20 (Fig 6A-6H). PA 20 did not protect any of the Toll-pathway mutants infected with B. cereus. We confirmed the inability of PA 20 to protect SPE and Spz mutants challenged with S. liquefaciens (S5A and S5B Fig). We tested the ability without 1 μg/mL of PA 20 . Each data point shown indicates the mean ± SD value obtained in triplicate assays done in a representative experiment. (F-G) Agar diffusion susceptibility assay of B. cereus (F) and S. liquefaciens (G). LB and TSB plates were treated with 1 μL of toxins right after the spreading of B. cereus and S. liquefaciens cultures, respectively, and left to incubate overnight. Spots below a-g contain a 1 μL spot of: 1 μg/mL PA 83 (a), PA 63  We evaluated whether Imd is necessary for PA 20 -mediated protection by testing the ability of PA 20 to reduce the sensitivity of Imd-deficient mutant flies to bacterial infections. The results revealed that PA 20 was able to delay B. cereus-induced ( Fig 6J)

and S. liquefaciensinduced (S5C Fig) mortality of Imd mutants.
We then evaluated whether the activity of the Imd pathway is sufficient for PA 20 -mediated protection. One of D. melanogaster's mitogen-activated protein kinase kinases, Dsor1, acts as a suppressor of the Imd pathway, and the downregulation of Dsor1 mimics the induction of the Imd pathway by microbes, even in the absence of an immune challenge [56]. Dsor1 blocks the Imd pathway by activating Pirk, which, in turn, interacts directly with PGRP-LC and PGRP-LE and disrupts their interaction with Imd [57] (Fig 5A). We tested the ability of PA 20 to reduce the sensitivity of Dsor1 mutant flies to B. cereus infection. We observed that these mutant flies are less sensitive to B. cereus infection in the absence of PA 20 , and that the addition of PA 20 has no effect on Dsor1 mutant fly sensitivity ( Fig 6K). We confirmed the inability of PA 83  Collectively, these data demonstrate that while Imd is not necessary for the PA 20 -mediated phenomenon, the constitutive activity of the Imd/NF-κB pathway in Dsor1 mutant flies is sufficient to negate the protective effects of PA 20 .
We examined whether PA 20 induced the production of reactive oxygen species (ROS) in Drosophila macrophage-like plasmatocytes, S2 cells. We measured the total ROS and separately superoxide radicals produced by S2 in the absence or the presence of B. cereus and various concentrations of PA 20 (0.25 to 20 μg/mL) (Fig 6L and S7 Fig). Although B. cereus increased the abundance of ROS in S2, the addition of PA 20 did not further alter the levels of these species. Our overall results demonstrate that the Toll pathway is necessary and the Imd pathway is sufficient but not necessary for the PA 20 -mediated phenomenon, and that PA 20 does not affect phagosomal ROS production.

Toxins of other bacterial pathogens may help flies resist anthrax
During the infection, anthrax-causing bacilli may co-exist with other environmental and commensal pathogenic agents. Consequently, anthrax-exposed insects can also be co-exposed to endocytosed with the help of clathrin. Upon entry into the cell, the PA pre-pore becomes the PA-pore. This change allows LF and EF to escape into the cell cytosol. Once in the cytosol, EF functions as an adenyl cyclase resulting in edema. Conversely, LF cleaves MAPKK and Nlrp1, causing caspase-mediated pyroptosis. Upon contact with anthrax infected animals or their carcasses, scavenging insects are likely exposed to PA 20 . PA 20 , once ingested by Drosophila, may interact with the Toll pathway to reduce the sensitivity of flies to bacterial challenge (center panel). Two pathways trigger the expression of antimicrobial peptides in Drosophila: the Toll and Imd pathways. During the bacterial challenge, Toll is activated by bacterial lys-type peptidoglycan (PGN) by binding to the PGRP-SA/GNBP1 heterodimer and undergoing enzymatic modification. This results in Toll activation by SPZ ligand binding. Concurrently or separately, Toll is also known to be activated by the circulating virulence factor detection protein, Psh, which also works through SPZ to induce Toll activation. The Imd pathway responds primarily to bacterial dap-type peptidoglycans such as that found in B. cereus and S. liquefaciens. Both Toll and Imd ultimately activate NF-κB transcription factors, which induce the transcription of AMPs that act in a broadspectrum manner. PA 20 , once ingested by Drosophila, works through Toll and likely PGRP-SA to reduce the sensitivity of flies to bacterial challenge. The effect of PA 20 on the Toll/NF-κB pathway may also occur on the homologous mammalian pathway (right panel). The vaccination with PA 83

PLOS PATHOGENS
Insects resist anthrax toxins of other bacteria. We investigated whether exotoxins of eight pathogenic bacteria affect the sensitivity of flies to B. anthracis Sterne strain (Fig 7A-7K). We observed that Clostridium septicum alpha toxin reduces the sensitivity of flies to B. anthracis Sterne strain infection (delay of the median survival time by 43 hours, P < 0.0001) (Fig 7A), demonstrating that toxins from other microorganisms may help flies resist anthrax.

Discussion
While biting and non-biting flies have been known to be capable of transmitting B. anthracis since the 19th century [58][59][60], the mechanism of insect-mediated anthrax transmission is unknown. Recent studies emphasized that anthrax and anthrax-containing insects pose a significant ecological threat in West African ecosystems [2][3][4][7][8][9][10][11]. When flies feed on anthraxinfected or deceased mammals, they invariably ingest both B. anthracis and circulating anthrax toxin components, including PA 20 [37]. B. anthracis is known to remain in the insect's gut for an extended time or as part of the insect microbiota for life [7]. While the potential benefit of this co-existence has not been studied, we propose a mechanism by which the vector-borne pathogen may benefit. Bacillus species, which naturally reside in the soil, are relatively nonmotile and benefit from a suitable vector to assist bacterial distribution. The reduced insect sensitivity would benefit mammalian pathogens by allowing microbes to further their reach. Our study exposed insects to vegetative B. anthracis for days, while in previous studies blow flies, horn flies, house flies, mosquitoes were found to be insensitive to this pathogen after short exposures for several hours [11,15,17,39], possibly explaining the difference in the insect sensitivity between studies. Moreover, we demonstrate that Drosophila and Culex are insensitive to B. anthracis spores, which is similar to previous observations that B. anthracis spores are not lethal to flies, mosquitoes [11,15,17,39], and nematodes [61,62]. Our results revealed that flies and mosquitoes respond differently to spores. Spores from either the Sterne or ΔSterne strains extended the longevity of Drosophila, with toxin-containing Sterne strain extending the life longer than the ΔSterne strain, whereas the longevity of mosquitoes was not affected by spores of either strain (Fig 3). It is possible that in our experimental conditions, uninfected flies can succumb to environmental microorganisms after 10-14 days, and the oral exposure to B. anthracis spores extended the longevity of flies by stimulating their innate immunity. Since it was previously shown that PA is detected on the surface of B. anthracis spores [63,64], we hypothesize that the presence of this surface PA proteins could further activate the innate immunity of flies and extend their longevity, as we observed in this study. The reason for the lack of the extension of longevity in mosquitoes could be that the cause of death of mosquitoes may be different from flies or because spores do not affect the immunity the same way they do in flies. Moreover, our results revealed that unlike Drosophila, mosquitoes sensitivity was similar to Sterne and ΔSterne strains. We conjecture this could occur due to mosquitoes' comparatively superior immune system, but this observed difference was not further explored. Future studies should investigate the toxin-dependent and -independent interaction of B. anthracis with flies and mosquitoes.

Insects resist anthrax
This work describes a new function of PA 20 , the domain that previously has only been known to prevent self-assembly of PA 83 in solution. PA 20 has also been known to contribute to the adaptive immunogenicity of PA 83 in mammals, as it contains at least three immunogenic epitopes [32]. One such PA 20 epitope is the 1β 13 −1α 1 loop shown to affect insect and plant innate immunity in this and previous studies [48]. The results described above led us to propose a novel mechanism by which PA 20 reduces the sensitivity of flies to insecticidal bacterial pathogens by utilizing Drosophila's Toll pathway (Fig 5A). Both Bacillus and Serratia species of bacteria are known to possess dap-type peptidoglycans and thus are known inducers of the Imd pathway [52]. Toll pathway is required for PA 20 function, and PA 20 may directly interact with Drosophila PGRP-SA to activate Toll, while Bacillus bacteria activate Imd, which would result in the transcription of AMPs [23]. Independently and together, Toll and Imd pathways result in the expression of AMPs that are known to be broad-spectrum in their effects and may possess the ability to act additively and/or synergistically [23,65]. PA 20 may be providing insects with an added benefit, resulting in the extension in lifespan during the bacterial challenge and allowing pathogens to further their reach. Alternatively, as dap is a structural derivative of lys, and since both Toll and Imd pathways can detect lys-and dap-peptidoglycan [21,54], it is possible that PA 20 may bind to Imd-activating PGRP-LC and PGRP-LE, thus leading to the activation of both Toll and Imd pathways. Although we show that the Imd pathway is not necessary, its activity is sufficient to negate the protective effects of PA 20 .
Drosophila GNBP-1 and PGRP-SA form a functional heterodimer. Fly PGRP-SA binds predominantly to lys-type peptidoglycan, but also binds to dap-type peptidoglycan [19]. In contrast, GNBP-1 only binds to lys-type peptidoglycan [19]. GNBP-1 is known to hydrolyze Gram-positive peptidoglycan, while PGRP-SA binds peptidoglycan fragments (muropeptides). GNBP-1 presents a hydrolyzed form of peptidoglycan for sensing by PGRP-SA, and this tripartite interaction between these proteins and peptidoglycan fragments is essential for downstream signaling. Future studies should establish whether GNBP-1 and PGRP-SA form a tripartite interaction and whether this leads to the activation of the downstream Toll pathway.
In Drosophila, at least seven AMPs (plus their isoforms) have been described: Diptericins In addition to AMP and hemocyte responses, fly intestinal epithelia exposed to pore-forming toxins undergo an evolutionarily conserved process of thinning (purging) followed by the

difficile binary toxin subunit A (B) and toxins B (C) and A (D), C. perfringens epsilon toxin (E), pertussis toxin (F), botulinum neurotoxins A (G) and B (H) (heavy chains), cholera toxin (I), Pasteurella toxin (J)
, and diphtheria toxin (K). All toxins were tested orally at 1 μg/ml in a sucrose solution. The effect of each toxin on fly survival was tested in the absence of bacteria as a control. P as in Fig 1. https://doi.org/10.1371/journal.ppat.1008836.g007

PLOS PATHOGENS
Insects resist anthrax rapid recovery of their initial thickness [66]. We argue that while PA is a pore-forming toxin component, the effect of PA 20 is independent of pore formation, as it lacks PA domain 3, which is located within PA 63 and is necessary for PA 63 -multimerization [67]. We propose that the mechanism of action by which the pore-forming Clostridium septicum alpha toxin provides flies protection from Bacillus anthracis Sterne strain infection is by inducing the purging of fly intestinal epithelia, as described by Lee et al. [66]. Future studies should explore this hypothesis.
The effect of PA 20 on the Toll/NF-κB pathway observed in this study is consistent with the impact PA 83 and PA 20 have on the homologous mammalian pathway (Fig 5A, right panel). PA 83 is the central antigen in the FDA-approved anthrax vaccine, BioThrax [68]. The vaccination with PA 83 [38] and human cell exposures to PA 83 [36] or PA 20 [37] lead to the Toll-like receptors-dependent activation of NF-κB and a subsequent upregulation in the expression of pro-inflammatory cytokines, especially Interleukin-6 (IL-6), IL-6 receptor, IL-1β, and Tumor necrosis factor alpha (TNF-α). Thus, in addition to an established paradigm of PA 83 activating adaptive immunity in mammals, this toxin component may also affect innate immunity. Bio-Thrax is approved for use as both a pre-exposure vaccination and post-exposure prophylaxis [69,70], which is consistent with the potential that its efficacy is partly based on an unrecognized effect on the human innate immune system, as presented here.
While our study may help understand the mechanism through which insects tolerate anthrax, resulting in their greater opportunity to transmit the anthrax-causing bacteria, it further suggests that PA 20 could be used beneficially in agriculture. Since fruit flies and mosquitoes are known to act as plant pollinators [71,72], PA 20 should be further evaluated for its ability to reduce the sensitivity of pollinating insects to bacterial pathogens, such as Serratia and Bacillus species [73][74][75].

PLOS PATHOGENS
Insects resist anthrax washed with DMF for three times. For the coupling reaction, the resin was added with Fmocprotected amino acid, Cl-HOBt, DIC, and NMP. The mixture was vortexed for 20 minutes at 50˚C. Afterward, the resin was washed with DMF once. The cycle of deprotection and coupling steps was repeated until the last amino acid residue was assembled. After the final Fmoc protecting group was removed, the resin was treated with 20 v% acetic Anhydride-NMP for 20 minutes. The resin was then washed with DMF, DCM, and dried with air. The peptides were cleaved using a TFA cocktail (95 v% TFA, 2.5 v% water, and 2.5 v% TIS) for three hours.  Strains of B. anthracis used included the toxigenic Sterne and toxin-negative ΔSterne strains. Spores were prepared and purified from solid agar (NBY-Mn) cultures of the strains and purified as described [76,77]. Medium components were from BD-Difco, and the NBY-Mn medium was composed of nutrient broth (8 g/L) (BD 23400); yeast extract (3 g/L) (BD 244020); MnSO4·H2O (25 mg/L) (Sigma-Aldrich, M7634); and agar (15 g/L) (Aldon Corp, AA0075). Spores were used only if >95% were refractile (ungerminated) as determined by phase microscopy and heat-resistance. Spore preparation included two centrifugations in density gradient medium (58 mL Hypaque-76™, Nycomed into 42 mL WFI) accompanied by extensive washing in sterile water for injection (WFI). The spores were activated by heating at 65˚C for 30 min just before use in assays [76].

Oral feeding survival assay
Flies or mosquitoes were infected according to the bacterial intestinal infection methods previously employed by Nehme et al. [78] with minor modifications. To prepare the Drosophila or Culex vials for infection, three pre-cut extra-thick Whatman blotting papers (Bio-Rad Laboratories, 1703965) were stacked at the bottom of either 25x95 mm diameter polystyrene Drosophila vials or 55x55x102 mm specimen bottles for fly and mosquito infections, respectively, following by capping with a foam plug. Overnight bacterial cultures were centrifuged, and the bacterial pellets were resuspended in either a 50 mM or 10% sucrose solution for fly or mosquito experiments, respectively. Insects were exposed orally to anthrax toxin components, domains, or peptides, each at a concentration of 1 μg/mL, which falls within the known 0.01-100 μg/mL range of the concentration of plasma-circulating toxin components in infected mammals during the late stage of anthrax infection [79][80][81][82][83][84][85][86][87][88]. All other bacterial toxins were tested at the same concentration.
Flies and mosquitoes were exposed orally to approximately 2.6 X 10 9 cells/mL, as previous studies have shown that the concentration of plasma-circulating bacteria in infected mammals during the late stage of anthrax infection is 10 9 −10 10 cells/mL [80,81,88]. Oral exposure was chosen because it represents a route by which flies are exposed to anthrax toxin components and bacilli in a natural environment [18]. All experiments were performed with 4-5 day old (post-eclosion) wild type male flies and mosquitoes to reduce any potential variability in sensitivity to bacterial infections due to confounding gender and age differences.
A final optical density at 600 nm (OD 600 ) was 3.3 for Bacilli species and E. coli (2.6 X 10 9 cells/mL), as well as OD 600 of 4.6 for S. liquefaciens (3.7 X 10 9 cells/mL). Depending on the experimental condition, toxin components and cortisone acetate were also added to bacterial sucrose solutions before adding to Drosophila vials and mosquito bottles. Finally, 2.5 or 10 mL of the respective solution was pipetted onto the Whatman paper in each Drosophila vial or mosquito bottle. Flies and mosquitoes were anesthetized using CO 2 , separated by gender, and placed ten at a time into their respective vials or bottles and incubated at 30˚C. The survival of insects was recorded a minimum of twice per day.

PA domains expression and purification
A gene strand containing a pelB leader sequence, the amino-terminal of PA 20 (residues 30-192), a 6-His tag, and 20 bp vector overlaps at the 5' and 3' end was ordered from Eurofins Genomics. PA 83 amino acids 1-29 were not included in the expression, because they form a signal peptide that is cleaved off of the mature protein [89]. Digested pSX2 expression vector (Scarab Genomics) and the pelB-PA 20 -6His gene strand were assembled using Gibson assembly.
PA domains 2, 3, and 4 fusion proteins were designed to include the PA domain, a C terminal InaD PDZ domain (amino acids 2-98, GenBank accession no. 1IHJ_A, cys53ala mutation), a C-terminal 6xHis tag, and 3' and 5' overlaps for insertion into the pSX2. The gene strands were cloned into KpnI and SacI-digested pSX2 using the NEBuilder HiFi DNA Assembly Cloning Kit (New England Biolabs, E5520S).
The sequences for the His-tagged PA D1 and PA D1' -InaD fusion proteins were PCR amplified with vector overlaps to IPTG-inducible expression vector containing maltose-binding protein (MBP) with a C-terminal PreScission protease cleavage site (pET His6 MBP prescission LIC cloning vector (HMPKS, plasmid #29721)). The vector was digested using XhoI and SspI, and the PCR-amplified inserts were cloned in using the NEBuilder HiFi DNA Assembly Cloning Kit.
MBP was cleaved from the MBP-PA D1 , and -PA D1' fusion proteins in solution using the Pierce HRV 3C Protease Kit (Thermo Scientific, 88946). The HRV3C protease was removed using Pierce Glutathione Agarose (Thermo Scientific, 16100) and Pierce Centrifuge Columns (Thermo Scientific, 89898). MBP was removed using amylose resin (New England Biolabs, E8021S), followed by dialysis and concentration of proteins using Vivaspin Centrifugal Concentrators (Sartorius, VS0101). Protein concentration was determined using the Pierce BCA Protein Assay Kit with bovine serum albumin as a standard (Thermo Scientific, 23225).
All PA domains were analyzed via SDS-PAGE and anti-PA 83 Western blotting to confirm the molecular weight using polyclonal goat anti-PA 83 and the secondary rabbit anti-goat IgG antibodies. Chemiluminescence of bands and their relative intensities were revealed using Azure c500 (Azure Biosystems, Dublin, CA).

Bacterial growth assays
For the agar-containing media assays, the OD 600 of overnight cultures of bacteria was measured. The OD 600 values were converted to cells/mL according to McFarland's scale [90]. 6 X 10 8 bacterial cells were added to solid media on 25 cm Petri dishes. One μL of toxin component or cortisone acetate was then pipetted onto the plate surface. The plates were then incubated for 24 hours at the temperatures appropriate for each bacterium. One μL of 10 mM levofloxacin (Cayman Chemical, 20382) or phosphate-buffered saline (PBS) (Corning, 21-040-CV) were included as controls.
In the liquid media experiments, the bacterial overnight culture was resuspended in a new liquid bacterial medium to OD 600 of 0.1. One hundred μL of the bacterial solution were then added into wells of a 96-well plate, followed by the addition of PA 20 to specific wells to a concentration of 1 μg/mL. The bacteria were incubated in 96-well plates at appropriate temperatures with constant shaking. The OD 600 was determined every 610 seconds for 700 minutes by a microplate reader (Molecular Devices, Spectra Max 384 PLUS).

PRGP-S pull-down assay
This assay was performed according to the procedure of Yoshida et al. [91] with minor modifications. Approximately 1 mg of insoluble dap-type peptidoglycan from B. subtilis was dispersed in 1 mL of PBS and centrifuged at 12,600 x g for 5 min. This process was repeated three times, and the sedimented peptidoglycan was resuspended in 1 mL of PBS. Ten μL (0.5 mg/ mL) of recombinant human PGRP-S was mixed with 100 μg of peptidoglycan suspension or PA 20 , and incubated overnight at 4˚C. The mixture was centrifuged at 12,600 x g for 5 min, and the pellets were washed three times with 300 μL PBS and resuspended in 200 μL Laemmli sample buffer (VWR, 89230-104). Both the pellets and the supernatants were analyzed by Western blot using anti-PA 83 and anti-PGRP-S antibodies.

PGRP-SA and PRGP-S binding ELISAs
PA 20 , washed dap-type peptidoglycan, or washed lys-type peptidoglycan were diluted to 10 μg/ mL in bicarbonate buffer, and 50 μL was added to flat-bottom, high-binding, half-area 96-well plates (Corning, 29442-318) in triplicates and incubated overnight at 4˚C. The next day, the wells were rinsed with PBST and PBS, then blocked with 100 μL blocking buffer (5% non-fat milk in PBS (PGRP-SA) or 5% BSA [Roche, 10738328103] in PBS (PGRP-S)) for 2 hours at room temperature. Wells were washed two times with PBST, and two times with PBS. Fifty μL of PRGP-SA with a T7 tag or PGRP-S diluted in PBS to 200 ng/μL was added to each well, and the plates were incubated for 2 hours at room temperature. The wells were washed as previously described, and 50 μL of anti-T7 monoclonal antibody (Sigma) diluted to 6.7 μg/mL in PBST and incubated overnight at 4˚C. For the PGRP-S plate, anti-PRGP-S polyclonal antibody diluted to 0.5 μg/mL in 5% BSA added to each well and incubated for 2 hours at room temperature. The wells were washed as above, and horseradish peroxidase-conjugated anti-mouse secondary antibody, diluted in PBST for PGRP-SA and 5% BSA for PGRP-S, was added to each well and incubated for 2 hours at room temperature. Following PBST and PBS washes, 50 μL of o-Phenylenediamine dihydrochloride (Sigma, P1526) dissolved in deionized water was added to each well, and the plate was incubated for 30 minutes at room temperature. The absorbance at 450 nm was measured using a microplate reader (Molecular Devices, Spectra Max 384 PLUS).

Measurement of reactive oxygen species in S2 cells
Reactive oxygen species (ROS) consisting of hydrogen peroxide, peroxynitrite, hydroxyl radicals, nitric oxide, and peroxy radicals, and separately superoxide radicals were measured using ROS-ID Total ROS/Superoxide Detection Kit (Enzo Life Sciences, ENZ-51010). S2 cells (Expression Systems, 94-005F) were collected by centrifugation at 400 x g for 5 minutes and resuspended in ESF 921 medium (Expression Systems, 96-001-01). Cells were seeded at 500,000 cells per well in 50 μL of resuspension medium into clear 96-well plates with black chimneys. Dilutions of PA 20 were performed in separate 12 channel basins with LB or LB containing B. cereus overnight culture at 1.6 OD 600 . PA 20 was tested at 22, 2, 1, 0.5, and 0.25 μg/ mL. A condition with no PA 20 was included with every assay. The oxidative stress detection reagent and the superoxide detection reagent were reconstituted in anhydrous DMF (VWR, 97064-586) to yield 5 mM stock solutions and stored at -20˚C. Detection reagents were added at 0.04% of sample preparation. Fifty μL of sample preparation reagent was added to the 96-well containing S2 cells and then incubated at 27˚C in the dark for 30 minutes. Fluorescence was measured with bottom reading for two different wavelengths. Total ROS was measured at excitation 488 nm, cutoff 515 nm, and emission 520 nm. Superoxide detection was measured at excitation 550 nm, cutoff 610 nm, and emission 610 nm (Molecular Devices, SpectraMax Gemini XPS/EM Microplate Reader).

Data analysis
Data analysis was conducted using GraphPad Prism software. All P-values reported are products of the respective positive control to a single experimental condition using two statistical analyses: the Log-rank (Mantel-Cox) and the Gehan-Breslow-Wilcoxon tests. An alpha of 0.05 was deemed the threshold for significance. We report P values adjusted by the Bonferroni correction. The delay in median survival was reported. Since the chance of dying in a small-time interval was not the same early in the study and late in the study, the values for the 95% CI of the ratio of median survivals were not meaningful and were not reported. Flies that died within the first 24 hours of the assay were censored from statistical analysis and considered to have died due to non-infective causes. Each insect experiment shown is representative of at least three independent experiments. Titration assays revealed that 20 mM cortisone acetate added to the feeding medium was the minimum concentration sufficient to immunosuppress Drosophila (Fig 6I), as 10mM did not alter the sensitivity of Drosophila to B. cereus. Wild type flies were fed a 50 mM sucrose solution. Some conditions included B. cereus, which was resuspended in 50 mM sucrose solution, or a condition containing an additional 10 mM cortisone acetate. Flies were maintained at 30˚C and monitored for death a minimum of twice daily and expressed as percent survival.

PLOS PATHOGENS
Insects resist anthrax