Symbiosis with Francisella tularensis provides resistance to pathogens in the silkworm

Francisella tularensis, the causative agent of tularemia, is a highly virulent facultative intracellular pathogen found in a wide range of animals, including arthropods, and environments. This bacterium has been known for over 100 years, but the lifestyle of F. tularensis in natural reservoirs remains largely unknown. Thus, we established a novel natural host model for F. tularensis using the silkworm (Bombyx mori), which is an insect model for infection by pathogens. F. tularensis established a symbiosis with silkworms, and bacteria were observed in the hemolymph. After infection with F. tularensis, the induction of melanization and nodulation, which are immune responses to bacterial infection, were inhibited in silkworms. Pre-inoculation of silkworms with F. tularensis enhanced the expression of antimicrobial peptides and resistance to infection by pathogenic bacteria. These results suggest that silkworms acquire host resistance via their symbiosis with F. tularensis, which may have important fitness benefits in natural reservoirs.


Results
Silkworm as a novel host model for F. tularensis symbiosis. F. tularensis is often found in arthropods such as ticks, but symbiosis between the bacteria and arthropods remains unclear. Thus, we established a novel host model for F. tularensis subsp. holarctica LVS symbiosis in silkworms. Silkworms were infected with F. tularensis, Escherichia coli, or Staphylococcus aureus via injection into the hemocoel. Silkworms infected with F. tularensis and E. coli were alive at 7 days after infection (Fig. 1a,b). No significant differences were observed in the body weight of the silkworms infected with F. tularensis and E. coli compared with uninfected silkworm (data not shown). By contrast, silkworms infected with S. aureus were dead within one day of infection (Fig. 1a,b). We also evaluated the growth of the bacteria in silkworms. We found that the bacterial numbers of F. tularensis did not show acute fluctuation for 6 days after infection (Fig. 1c). The bacterial number of E. coli decreased each day after infection (Fig. 1c). S. aureus exhibited slight growth in silkworms one day after infection but further data could not be obtained because of the decreased growth and death of the silkworms caused by the bacterial infection (Fig. 1c). Green fluorescent protein (GFP)-expressing F. tularensis were observed in the hemolymph isolated from silkworms until 5 days after infection (Fig. 1d). By contrast, GFP-expressing E. coli were not observed in the hemolymph at 5 days after infection (Fig. 1d). These results suggest that F. tularensis establish a symbiosis with the silkworms.
F. tularensis blocks the silkworm immune response trigger. In silkworms, melanization and nodulation are known to be the most common immune responses to bacterial infection 21 . To analyze the melanization and nodulation responses in the silkworms after bacterial infection, we collected hemolymph from the silkworms after bacterial infection and determined the optical density of the samples. Live F. tularensis did not induce melanization at 1 and 18 h after infection, whereas heat-killed F. tularensis significantly induced melanization at 18 h after inoculation (Fig. 2a,b). E. coli and S. aureus induced melanization after bacterial infection strongly ( Fig. 2a,b). To analyze nodulation, the dorsal vessel was observed by microscopy in silkworms after bacterial infection. E. coli and S. aureus strongly induced nodulation immediately after bacterial infection (Fig. 2c,d). Live F. tularensis did not induce nodulation at 1 and 18 h after infection, whereas heat-killed F. tularensis significantly induced nodulation at 18 h after inoculation (Fig. 2c,d). These results suggest that immune responses of silkworm against F. tularensis are inhibited during the early stage of the infection, and some activities of F. tularensis are involved in the inhibition.

Silkworm acquires host resistance to S. aureus infection after pre-inoculation with F. tularensis.
Melanization and nodulation were not induced by F. tularensis infection, so we investigated whether the melanization and nodulation responses induced by S. aureus were inhibited by pre-inoculation with F. tularensis. Silkworms were inoculated with live or heat-killed F. tularensis and incubated for 72 h at room temperature. After 72 h incubation, melanization and nodulation induced by heat-killed F. tularensis calmed down. The silkworms were then inoculated with PBS or S. aureus, and we measured the melanization and nodulation responses (Fig. 3a). The results showed that pre-inoculation with live F. tularensis inhibited the melanization and nodulation responses induced by S. aureus infection, whereas the heat-killed bacteria were not effective at immune inhibition ( Fig. 3b-e).
We hypothesized that pre-inoculation with live F. tularensis may have affected the immune response in silkworms, so we investigated the survival rate of silkworms after bacterial infection. We found that silkworms pre-inoculated with live F. tularensis survived S. aureus infection and they exhibited significant host resistance to bacterial infection compared with those pre-inoculated with heat-killed F. tularensis (Fig. 4a). Pre-inoculation with live F. tularensis significantly inhibited the growth of S. aureus in silkworms at 24 h after infection compared to PBS inoculated control (Fig. 4b). By contrast, heat-killed F. tularensis did not affect the survival rate of silkworms or the growth of S. aureus in the silkworms (Fig. 4a,b).
Antimicrobial peptides (AMPs) are well-known immune factors that combat pathogens in arthropods 22 . To investigate whether AMPs contribute to the host resistance caused by pre-inoculation with live F. tularensis, we analyzed the expression of genes for typical AMPs, i.e., cecropin B, lebocin, attacin, and moricin, at 72 h after inoculation with live or heat-killed F. tularensis. We found that the expression levels of these AMP genes were significantly induced by pre-inoculation with live F. tularensis, whereas the heat-killed bacteria had no effect (Fig. 4c). We also confirmed the time-course expression of cecropin B by immunoblotting (Fig. 4d). Live F. tularensis induced cecropin B expression and the induction was sustained at 72 h post infection. In contrast, the expression induced by heat-killed F. tularensis was reduced at 48 h and disappeared at 72 h post infection.

Discussion
Arthropods are involved in the life cycle of F. tularensis 4 . Therefore, the development of arthropod host models is useful for studying the mechanisms related to F. tularensis infection and symbiosis. In this study, we established a novel, symbiotic host model for F. tularensis using silkworms. The ecology of F. tularensis and the natural reservoirs of the bacterium in the environment are not fully understood. The wax moth (Gallaria mellonella) has been used as a mammalian infection model for F. tularensis 23 , but symbiosis between F. tularensis and insects is still unclear. However, we observed a symbiosis between F. tularensis and silkworms in the present study; therefore, some types of insect may be candidates as natural reservoirs. Since the silkmoths that hatched from F. tularensis-infected larva still retained the F. tularensis bacteria (data not shown), it is possible that the animals were infected with F. tularensis by eating insects and/or larvae containing the bacteria. Insects only possess innate immunity 24 ; therefore, insects are generally used as models to study the basis of innate immunity. Melanization and nodulation are known to be the first defensive responses to bacterial invaders in arthropods 25 . Melanin can seal off foreign organisms in the hemocoel and starve them of nutrients 26,27 . Melanin synthesis also results in the production of reactive oxygen and nitrogen intermediates, which are toxic Silkworms were infected with indicated bacteria, and hemolymph was collected at 1 and 18 h post-inoculation. The optical density (λ = 405 nm) of the hemolymph was measured using a spectrometer immediately after centrifugation to remove hemolymph cells. (c) Silkworms were infected with indicated bacteria, and nodule formation around dorsal vessel was observed. Arrowheads indicate nodule formation induced by infected bacteria. (d) Silkworms were infected with indicated bacteria, and the total area of nodule formation was calculated using the area measurement tool. The relative melanized area was shown compared to PBS-inoculated control group. (b,d) The data represent the averages from triplicate samples based on three identical experiments, and the error bars denote the standard deviations. Significant differences were accepted at P < 0.05 or P < 0.01, and they are indicated by asterisks (* ) or double asterisks (* * ), respectively.  After 72 h incubation, silkworms were inoculated with PBS or S. aureus, and melanization and nodulation were observed. (b) Hemolymph was collected from silkworms at 1 h post-inoculation with PBS or S. aureus, and the condition of melanization was decided by color. (c) The optical density (λ = 405 nm) of the hemolymph collected at 1 h post second inoculation was measured using a spectrometer immediately after centrifugation to remove hemolymph cells. (d) Nodule formation around dorsal vessel at 1 h post second inoculation was observed. Arrowheads indicate nodule formation. (e) The total area of nodule formation was calculated using the area measurement tool. The relative melanized area of nodule formation was shown compared to PBS-inoculated control group. (c,e) The data are presented as averages from triplicate samples based on three identical experiments, and the error bars denote the standard deviations. Significant differences were accepted at P < 0.05, and they are indicated by asterisks (* ).
to some pathogens 28 . Nodule formation is a rapid response that removes microorganisms from the hemocoel. Granulocytes release sticky material after bacterial infection, and the hemocytes and bacterial cells clump together, thereby resulting in the formation of nodules 29 , which comprise aggregations of hemocytes and microorganisms that are subsequently subjected to melanization 30 . By contrast, AMPs might work during a later stage of infection because their production and concentration in the hemolymph both increase after bacterial infection 31,32 . In this study, we demonstrated that live F. tularensis inhibited melanization and nodulation, but (b,c) The data represent the averages from triplicate samples based on three identical experiments and the error bars denote the standard error of the mean (n = 9). Significant differences were accepted at P < 0.05 and they are indicated by asterisks (* ). not heat-killed F. tularensis, suggesting that some biological activities of F. tularensis may inhibit the immune responses. F. tularensis posses type VI secretion system which is important for intracellular growth in host cells. This secretion system contribute to control immune system in silkworm. Indeed, the type VI secretion system is reported to be involved in intracellular growth in mosquito cell line 33 . Thus, symbiosis between F. tularensis and silkworms may be established by inhibiting the silkworm immune responses during the early stage of infection. The over-activation of immune reactions can damage the host animal itself 34 . S. aureus has a very rapid growth rate and it causes high mortality in silkworms 20 . However, although F. tularensis remained at similar bacterial numbers to S. aureus in silkworms, they never died. Thus, the over-activated immune reactions caused by S. aureus infection may lead to silkworm death in the early stage of infection.
A key insect-based immunological study discovered Toll in Drosophila, which led to the identification of mammalian Toll-like receptors 35,36 . Insect Toll functions as a receptor for an endogenous ligand, which relays signals to transcription factors that produce AMPs 37 . The silkworm has 14 Toll isotypes, some of which are expressed several hours after bacterial infection 38,39 . These receptors mediate the induced immune responses that are known as pathogen-associated molecular patterns, such as those to lipopolysaccharide and peptidoglycan 40,41 . We found that F. tularensis inhibited the silkworm immune responses in the early stage of infection, but the production of AMPs was enhanced in the later stage of infection. F. tularensis is also sensitive to some AMPs 42 , but the bacterium may escape from the effects of AMPs by endosymbiosis within host cells. Therefore, silkworms engaged in symbiosis with F. tularensis exhibited resistance to S. aureus infection.
F. tularensis can resist degradation in the phagosome and replicate inside mammalian macrophages 43 . We did not demonstrate the phagocytosis activity of hemocytes directly in this study, but intracellular bacteria were clearly observed in the hemocytes at 5 days after infection. Hemocytes may be one of the targets for bacterial invasion and F. tularensis may control the immune response in the silkworm to take advantage of suitable conditions for symbiosis. The intracellular signaling pathway related to the expression of AMP in silkworms is still unclear. However, Ishii et al. showed that the phosphorylation of p38 MAPK protein conferred protection against S. aureus infections in silkworms by up-regulating the expression of AMP genes, but it did not affect the melanization activity 44 . In Drosophila, the activation of p38 induces host protection from various bacterial pathogens and fungi because of the up-regulated expression of genes for stress response factors and specific AMPs (cecropin B and attacin) 45 . Thus, F. tularensis may induce specific signaling pathways that are affected by intracellular bacteria. Therefore, these signaling pathways may be a possible target for disrupting the lifecycle of Francisella in the environment.
Symbiosis is considered to provide benefits to both symbionts. Various insects possess intracellular bacteria within specialized cells known as bacteriocytes, the sole function of which appears to be the housing and maintenance of bacteria 46 . Insect immune effectors have been implicated in the regulation of the bacteria found in the bacteriocytes of the weevil Sitophilus. Sitophilus bacteriocytes express a cationic AMP, coleoptericin A, at high levels, and this AMP plays an important role in controlling the maintenance of the symbiont 13 . Wolbachia pipientis is an obligate intracellular bacterium and a common endosymbiont of insects 47 . Drosophila melanogaster flies infected with W. pipientis are less susceptible to the induction of mortality by a range of RNA viruses 48 . Wolbachia also inhibits the ability of a range of pathogens, such as Plasmodium, dengue virus, and Chikungunya virus, to infect Aedes aegypti 49 . The depressed vector competence of Wolbachia-infected mosquitoes may be caused by an enhanced immune function, including the induction of AMPs, melanization, and reactive oxygen species 50 . We showed that silkworms in symbiosis with F. tularensis were protected from death caused by S. aureus infections. Thus, symbiosis with F. tularensis may provide fitness benefits for insects, and the human pathogen F. tularensis may have an important role in protecting natural reservoirs, such as arthropods, from pathogenic invaders. F. tularensis has been isolated from deer flies, horse flies, ticks, and mosquitoes 5 . Francisella-like endosymbionts have also been reported in various tick species [51][52][53] . In this study, our results suggest that F. tularensis can infect and survive in endosymbiosis with silkworms; therefore, many other insect species may also be vectors of tularemia. Thus, Francisella may be distributed in more arthropod species than considered at present.
Infection using silkworm larva. Day 2 fifth instar larva was inoculated in the hemocoel with 50 μ L of bacterial solution containing 1 × 10 8 CFU/mL in PBS using a 1-mL syringe equipped with a 30-gauge needle (Terumo Inc., Tokyo, Japan). After inoculation, the silkworms were incubated at room temperature with food. To obtain bacterial counts (as CFU/mL), the infected silkworm larvae were weighed and placed in disposable 15-mL centrifuge tubes, before homogenizing with a Biomasher SP (Funakoshi Co., Ltd, Tokyo, Japan) and suspending in 3 mL of PBS. The suspension was subsequently centrifuged at 300 × g for 30 s and solid tissues were separated from the concentrated suspension. Using appropriate dilutions, the suspension samples were spread onto agar plates and the numbers of colonies were counted. To calculate the counts (CFUs), the summed volumes of the hemolymph and tissues were estimated together (1 g = 1 mL).
In vivo melanization analysis. Day 2 fifth instar larva was inoculated with 50 μ L of bacterial solution containing 2 × 10 8 CFU/mL in PBS. Control groups were injected with PBS or an equal volume of 75 °C/30-min heat-killed F. tularensis. After 1 h and 18 h, the hemolymph was collected from the caudal horn and placed in a pre-chilled 1.5-mL tube on ice to prevent further melanization because of exposure to the air. The hemolymph samples were centrifuged at 6000 × g for 5 min at 4 °C to remove the hemolymph cells. The optical density (λ = 405 nm) of each supernatant fraction was measured using a spectrometer immediately after centrifugation.
In vivo nodule formation analysis. Melanized nodules precipitated around the dorsal vessel were observed, as described previously 29 . Photographs were taken from the seventh to ninth segments under the same conditions using a stereoscopic microscope. To quantify the formation of nodules, each photograph was imported into Image J 1.42l software (NIH, USA; http://rsbweb.nih.gov/ij/), and the total amount of pixels in each sample was calculated with the ImageJ area measurement tool. The relative melanized area was calculated as the ratio of each group relative to that of the control.
Fluorescence microscopy. GFP-expressing bacteria were used to inoculate fifth instar day 2 larvae, which were then incubated at room temperature with food. At 1, 24, 72, and 120 h post-inoculation, hemolymph was collected from the caudal horn and added to a 24-well tissue culture plate, before diluting up to 500 μ L with IPL-41 Insect Medium (Sigma-Aldrich, St Louis, MO) supplemented with 10% heat-inactivated fetal bovine serum. The plates were then centrifuged for 5 min at 900 × g and incubated for 15 min at room temperature. After washing twice with PBS, the samples were fixed with 4% paraformaldehyde (Wako, Osaka, Japan) in PBS for 15 min at room temperature. Subsequently, the samples were washed twice with PBS. Fluorescent images were obtained using a FluoView FV100 confocal laser scanning microscope (Olympus).
Pre-inoculation of silkworms. We injected 50 μ L of PBS, 1 × 10 8 CFU/mL live F. tularensis, or the same amount of 75 °C/30-min heat-killed F. tularensis into each silkworm during pre-inoculation (Fig. 3a). After incubation for 72 h at room temperature with food, the silkworms were inoculated with 50 μ L of PBS suspension containing 2 × 10 8 or 1 × 10 7 CFU/mL S. aureus to obtain the survival curve or internal CFU measurements, respectively. For the in vivo melanization and nodule formation analyses, each silkworm was injected with 50 μ L of 1 × 10 8 CFU/mL S. aureus suspension or PBS as a control to determine the effects of pre-inoculation.

RNA isolation and qPCR analysis of AMPs.
To analyze the expression of AMP genes, we collected the fat bodies from silkworms dissected at 72 h after pre-inoculation. The total RNA was isolated from the fat body using NucleoSpin RNA (Macherey-Nagel, Düren, Germany). The RNA was quantified by absorption at 260 nm using a NanoDrop 2000 (Thermo Fisher Scientific Inc., MA). Reverse transcription was conducted using ReverTra Ace qPCR RT Master Mix (Toyobo Co. Ltd, Osaka, Japan), and cDNA samples were stored at − 30 °C prior to use. Next, qPCR was performed with the StepOne Real-Time PCR system (Applied Biosystems, CA, USA) using KOD SYBR qPCR Mix (Toyobo). The primer sets were described previously 57 . The Actin A3 amplicon was used as an internal control to normalize all of the data. The relative expression levels of the AMPs were calculated using the relative quantification method (∆ ∆ Ct).
Immunoblotting. Day 2 fifth instar larva was inoculated with 50 μ L of PBS, live or 75 °C/30-min heat-killed F. tularensis solution containing 2 × 10 8 CFU/mL in PBS. After 1 h, 24 h, 48 h and 72 h, the hemolymph was collected from the caudal horn and centrifuge (6000 × g, 4 °C) for 10 min to isolate hemolymph plasma. The proteins in 1.5 μ L of hemolymph plasma were separated by SDS-PAGE with 4-12% Bis-Tris Gel (Thermo Fisher Scientific Inc.), and transferred onto polyvinylidene difluoride membranes (Millipore, Billerica, MA). After blocking with 5% nonfat dry milk in Tris-buffered saline (TBS) at room temperature for 2 h, the membranes were incubated overnight with anti-cecropin B antibody (1:1000; ab27571; Abcam plc, Cambridge, UK) at 4 °C. After washing with TBS containing 0.02% (v/v) Tween 20, the membranes were incubated for 1 h with horseradish peroxidase-conjugated secondary antibody (0.01 μ g/mL) at room temperature and immunoreactions were visualized using the enhanced chemiluminescence detection system (GE Healthcare Life Science, Little Chalfont, UK).

Statistical analysis.
Statistical analyses were performed using one-way ANOVA with the post hoc Tukey-Kramer test. Statistically significant differences between groups were accepted at P < 0.05 or P < 0.01. The survival curves were estimated with the Kaplan-Meier method and the log-rank test was used to determine significant differences between the live and heat-killed F. tularensis pre-inoculated groups (P < 0.05).