Immune tolerance to an intestine-adapted bacteria, Chryseobacterium sp., injected into the hemocoel of Protaetia brevitarsis seulensis

To explore the interaction of gut microbes and the host immune system, bacteria were isolated from the gut of Protaetia brevitarsis seulensis larvae. Chryseobacterium sp., Bacillus subtilis, Arthrobacter arilaitensis, Bacillus amyloliquefaciens, Bacillus megaterium, and Lysinibacillus xylanilyticus were cultured in vitro, identified, and injected in the hemocoel of P. brevitarsis seulensis larvae, respectively. There were no significant changes in phagocytosis-associated lysosomal formation or pathogen-related autophagosome in immune cells (granulocytes) from Chryseobacterium sp.-challenged larvae. Next, we examined changes in the transcription of innate immune genes such as peptidoglycan recognition proteins and antimicrobial peptides following infection with Chryseobacterium sp. PGRP-1 and -2 transcripts, which may be associated with melanization generated by prophenoloxidase (PPO), were either highly or moderately expressed at 24 h post-infection with Chryseobacterium sp. However, PGRP-SC2 transcripts, which code for bactericidal amidases, were expressed at low levels. With respect to antimicrobial peptides, only coleoptericin was moderately expressed in Chryseobacterium sp.-challenged larvae, suggesting maintenance of an optimum number of Chryseobacterium sp. All examined genes were expressed at significantly higher levels in larvae challenged with a pathogenic bacterium. Our data demonstrated that gut-inhabiting bacteria, the Chryseobacterium sp., induced a weaker immune response than other pathogenic bacteria, E. coli K12.


Results
Isolation of gut bacteria and analysis of the host cellular immune response against them. To isolate gut-inhabiting bacteria, we surface-sterilized ten fourth-instar larvae, and dissected and homogenized the alimentary canal (midgut and hindgut). Bacterial colonies exhibiting different morphologies with different growth rates were observed after 12 h-48 h incubation (Fig. 1A). Six of these colonies were randomly chosen and maintained in the laboratory for all further experiments. In order to identify the cultured bacterial species, sequences from their small subunit ribosomal RNA (16S rRNA) genes were obtained and blasted against NCBI database (Fig. 1B).
Next, to examine host innate cellular immune responses, the larval hemocoel was injected with Chryseobacterium sp., B. subtilis, A. arilaitensis, B. amyloliquefaciens, L. xylanilyticus, or B. megaterium (Fig. 1C,D). We then examined pathogenic-associated phagocytosis (including lysosomal formation and pathogen-related autophagosome vacuole formation) as an indicator of the cellular immune response against bacterial infection. As previously reported 14 , we found that a specific type of hemocyte (granulocytes) was associated with pathogenic lysosome formation and autophagy-related phagocytosis, which are efficient mechanisms for eliminating pathogens. Therefore, we used LysoTracker Red (a lysosome-selective stain) to identify acidified compartments in granulocytes, and green fluorescent staining-microtubule-associated protein 1A/1B-light chain 3 (LC3) to identify autophagosome formation (autophagosomes sequester materials intended for delivery to lysosomes). Granulocytes from bacteria-challenged larvae were stained with LysoTracker Red and green fluorescent-LC3 at 12 h post-infection and analyzed by flow cytometry. As shown in Fig. 1C,D, low levels of pathogen-associated lysosome-and autophagy-activity were observed in Chryseobacterium sp.-challenged larvae (LysoTracker Red-positive cells, 8.03% (C-1); green fluorescent-LC3-positive cells, 8.98% (D-1)). However, after infection with L. xylanilyticus, the percentage of cells showing pathogen-associated lysosome-and autophagy-activity increased to 38.84% and 42.01%, respectively . Infection with B. subtilis, A. arilaitensis, B. amyloliquefaciens, and B. megaterium led to moderate lysosome-and autophagosome-related immune responses (C-2; 18.95% and D-2; 19.60%, C-3; 14.19% and D-3; 30.42%, C-4; 12.25% and D-4; 11.52%, C-5; 38.84% and D-5; 42.01%, and C-6; 15.78% and D-6; 9.17% respectively). To better understand the evolutionary relationship between Chryseobacterium sp. and other Chryseobacterium lineages, we next performed phylogenetic analyses using MEGA 6.0 software (Fig. 1E) 34 . Neighbor joining analysis and construction of a phylogenetic tree revealed that this bacterial strain closely related to the genus Chryseobacterium. Indeed, this bacterial strain showed 99.6% sequence identity with Chryseobacterium sp. IMER-A-2-17, 98.9% with Chryseobacterium sp. MH48, 98.6% with Chryseobacterium sp. IMER-A-2-9, and 98.3% with Chryseobacterium sp. AU939 (Fig. 1E).  Larval survival and immune response to infection by Chryseobacterium sp. or E. coli K12. To investigate whether Chryseobacterium sp. is non-pathogenic in this insect, we analyzed host survival after infection with Chryseobacterium sp. After injection with either LB medium, Chryseobacterium sp., or E. coli K12, we then monitored survival every 12 h for 180 h ( Fig. 2A). Larvae were injected with 10 6 colony-forming units because this is the dose of E. coli K12 required to kill ~50% of larvae after 72 h. As shown in Fig. 2A, the survival rate of larvae injected with E. coli K12 was < 50% at 72 h post-infection; however, 75% and 90% of larvae injected with Chryseobacterium sp. or LB medium survived ( Fig. 2A). After 180 h, 65% of Chryseobacterium sp.-challenged larvae and 85% of LB medium-challenged larvae were alive as opposed to < 30% of E. coli K12-injected larvae ( Fig. 2A), respectively. There was better survival of Chryseobacterium sp.-challenged group compared that of the E. coli K12-challenged group (p = 0.01) but, there was no significant difference between LB medium-and Chryseobacterium sp.-challenged group (p = 0.154) ( Fig. 2A). In addition, microscopic analysis revealed no immune-related morphological changes in granulocytes from Chryseobacterium sp.-challenged larvae. However, granulocytes from E. coli K12-challenged larvae showed amoeba-like, lobopodia-like, and fan-like structures, which are indicative of immune activation 4 .

Figure 2. Larval survival and changes in hemocyte number in response to infection by
Host melanization in response to pathogen infection is an important indicator of immune or wound healing responses in vertebrates and invertebrates. Therefore, we examined melanization in an area injected with either LB medium, Chryseobacterium sp., or E. coli K12. Melanization in response to E. coli K12 was initiated at 12 h post-infection and spread over the entire body by 24 h post-infection; this was maintained at 72 h post-infection (Fig. 2B). Melanization in response to Chryseobacterium sp. or LB medium also occurred up until 24 h post-injection, but this gradually disappeared and was completely absent at 72 h post-injection ( Fig. 2B; injection site indicated by the red circle).

Lysosome formation in granulocytes in response to infection by Chryseobacterium sp. or E. coli K12.
To further explore whether the granulocytes from Chryseobacterium sp.-challenged larvae generated phagosomes or lysosomes in the cytoplasm, we stained granulocytes with LysoTracker Red and compared them with granulocytes from E. coli K12-challenged larvae (  Fig. 3E, compared to that of the Chryseobacterium sp.-challenged group, pathogen-associated lysosome activities in the E. coli K12-challenged group were significantly increased at 12 h or 24 h. These results reconfirmed the formation of phagosomal or lysosomal compartments within granulocytes from E. coli K12-challenged larvae; however, this was not the case in Chryseobacterium sp.-or LB medium-challenged larvae.

Autophagosome formation in granulocytes in response to infection by E. coli K12 or
Chryseobacterium sp. To determine whether autophagosomes were generated upon Chryseobacterium sp.
infection, we examined GFP-LC3 expression in hemocytes by fluorescence microscopy and flow cytometry. The accumulation of pathogen-related autophagosomes within granulocytes was examined by confocal microscopy (Fig. 4A). At 0 h after E. coli K12-injection, granulocytes did not emit a green fluorescent signal (LC3) (Fig. 4A-1). However, high levels of LC3 accumulated in granulocyte vacuoles at 24 h post-infection ( Fig. 4A-2,A-3); the in the insets in panels A-1 through to A-8. GR, granulocytes; PL, plasmatocytes. Scale bar = 20 μ m. At 12 h post-infection with E. coli K12, over 90% of granulocytes were strongly stained by LysoTracker Red (red in A-2 and A-3); however, staining was very faint after infection by Chryseobacterium sp. (A-6 and A-7). (B,C) Flow cytometry analysis of the total hemocyte population at 0~48 h post-infection with E. coli K12 or Chryseobacterium sp. At 12 h post-infection with E. coli K12, the percentage of stained granulocytes in the Lyso high region increased from 10.36% to 32.31%. This gradually fell to 11.08% at 48 h post-infection. However, there was no increase in the population of stained granulocytes from Chryseobacterium sp.-challenged larvae (10.07% at 0 h, 11.02% at 12 h, and etc.). (D) Low levels of pathogen-associated lysosome activity were always observed in LB medium-challenged larvae (< 5%). (E) Student's t-test analysis of flow cytometry results to compare differences between E. coli K12-and Chryseobacterium sp.-challenged group. The experiment was repeated for three times. Error bars indicate Mean± SEM. * P < 0.05 (t-test). NS, not significant. level decreased again at 48 h post-injection ( Fig. 4A-4,A-5). By contrast, granulocytes from larvae injected with Chryseobacterium sp. were negative for LC3 at 0~48 h post-injection ( Fig. 4A-6~A-10).
In flow cytometry analysis, at 24 h post-infection with E. coli K12, 27.90% of granulocytes emitted a fluorescent signal compared with 6.88% at 0 h (Fig. 4B). At 48 h post-injection, the LC3 signal decreased gradually (8.40% LC3 high cells). However, granulocytes from larvae infected by Chryseobacterium sp. or LB medium did not show increased emission of a fluorescent signal at any time point (Fig. 4C; 7.11~5.32% LC3 high and Fig. 4D). Compared to that of the Chryseobacterium sp.-challenged group, pathogen-associated autophagosome activities in the E. coli K12-challenged group were significantly increased at 12 h, 24 h, or 48 h (Fig. 4E).

Expression of immune-related genes in E. coli K12-and Chryseobacterium sp.-challenged larvae.
To examine the expression of host innate immune genes after infection with Chryseobacterium sp., we analyzed the expression profiles of a set of molecular pattern recognition proteins (which are usually activated upon interaction with microorganisms or microorganism-derived molecules) and AMPs (which are usually synthesized by the fat body in response to infection and secreted into the hemolymph). We previously isolated a number of PGRP orthologs and AMPs from P. brevitarsis seulensis, including PGRP-1, 2, SC2, SC3, defensin A, and coleoptericin 33 . Therefore, we used sequences derived from fragments of these genes (and the actin housekeeping gene) as reference genes for quantitative real-time polymerase chain reaction (qRT-PCR) analysis (Fig. 5A~F). Transcription of these genes was examined at 12, 24, and 48 h post-infection with LB-medium, E. coli K12, or Chryseobacterium sp. At 24 h post-infection with E. coli K12 or Chryseobacterium sp., PGRP-1 expression was higher than after LB-medium infection (Fig. 5A). However, expression fell significantly at 48 h post-infection, whereas that in E. coli K12-challenged larvae remained high until 48 h post-infection (Fig. 5A). PGRP-2 expression was also significantly higher than that after LB-medium infection at 48 h post-infection with E. coli K12 or Chryseobacterium sp. (Fig. 5B). As shown in Fig. 5C,D, expression of PGRP-SC2 and -SC3 was more than 30 times higher than in LB-medium challenged larvae at 24 h post-infection. Especially, that of PGRP-SC2 was also more than 20 times higher than in Chryseobacterium sp.-challenged larvae at 24 h post-infection. Similarly, expression of defensin A increased by 30-fold in E. coli K12-challenged larvae at 24 h post-infection (Fig. 5E). In addition, expression of coleoptericins increased by more than 100-fold at 24 h post-infection and by more than 3,000-fold at 48 h post-infection with E. coli K12 (Fig. 5F). However, both defensin A (6-fold) and coleoptericin (182-fold) were slightly or moderately expressed in Chryseobacterium sp.-challenged group at 24 h or 48 h post-infection (Fig. 5E,F).

Discussion
Generally, the lumen of the insect gut is densely colonized by various commensal or symbiotic microorganisms, which comprise < 20~30 taxa according to 16S rRNA sequence analysis 35 . In addition, Yun et al. 25 reported that the microbiota within the guts of various insects were dominated by Proteobacteria, Firmicutes, Bacteroidetes, Actinobacteria, Tenericutes, and unclassified bacteria 25 . Therefore, the lumen of the insect gut is a complex natural habitat occupied by various microorganisms, and there are multiple and complex interactions between them that enable them to survive. Thus, it is difficult to investigate the immunological relationships between gut-inhabiting microorganisms and host immunity.
To overcome these challenges, we cultured bacteria from the intestine of Protaetia brevitarsis seulensis and randomly selected or isolated six species of bacteria, which are probably gut-inhabiting microorganisms. Among these, we identified Chryseobacterium sp., which is probably a commensal/symbiont adapted to life in the gut of various insects [36][37][38][39] . Chryseobacterium sp. is a Gram-negative, aerobic, non-motile, yellow-pigmented, rod-shaped bacterium that belongs to the family Flavobacteriaceae; most members of this family are found in soil and in fresh or seawater habitats 40 . Chryseobacterium was originally isolated by Vandamme et al. 41 ; since then, many species have been identified in soil, clinical samples, and the guts of insects 42,43 . Recently, Chryseobacterium sp. was identified in the gut of the wood-boring beetle Anoplophora chinensis, in which it has the ability to degrade cellulose and/or aromatic compounds 38 .
As shown in Fig. 1C, we found that five species of bacteria (not including Chryseobacterium sp.) moderately or strongly activated pathogen-associated lysosomal and pathogen-related autophagosome responses. This raised the question of why Chryseobacterium sp. alone induces only a weak immune response compared to other bacteria. Gut-inhabiting microorganisms can interact with host immune cells in different ways, and their survival in the host is dependent on how well adapted they are to the host 44,45 . As shown in Figs 3 and 4, most granulocytes from Chryseobacterium sp.-challenged larvae showed weaker lysosomal or autophagosome activity than that of E. coli K12 used as the infective agent. Activated granulocytes contain phagosomes, lysosomes, and autophagosomes, all of which clear pathogens and engulf potentially hazardous substances 4 . These immune-related GR, granulocytes; PL, plasmatocytes. Scale bar = 20 μ m. Many granulocytes were strongly stained by green fluorescent LC3 at 24 h post-E. coli K12 infection (A-2 and A-3), but staining was very faint in granulocytes after Chryseobacterium sp.

infection (A-7 and A-8). (B,C) Total hemocytes at 0 ~48 h post-infection with E. coli K12 or
Chryseobacterium sp. At 24 h post-infection with E. coli K12, the percentage of stained hemocytes in the LC3 high region increased from 6.88% to 27.90%, before falling again to 8.40% at 48 h post-infection. However, there were no observable changes in green fluorescence intensity in hemocytes from Chryseobacterium sp.-challenged larvae (7.11% at 0 h and 4.50% at 24 h). (D) Low levels of pathogen-associated autophagosome activity were always observed in LB medium-challenged larvae (< 7%). (E) Student's t-test of flow cytometry results to compare differences between E. coli K12-and Chryseobacterium sp.-challenged group. The experiment was repeated for three times. Error bars indicate Mean± SEM. * P < 0.05 (t-test). NS, not significant. morphological changes are also observed in phagocytes from other insects such as flies, mosquitoes, and beetles, although the type of professional phagocyte differs according to the insect species 3,5,12 . Macrophages or phagocytes in vertebrates and invertebrates are the front line of immune defense against virulent pathogens, and recognition of invading pathogens by these cells is the key step in generating immune responses 46,47 . Macrophages or phagocytes engulf potentially hazardous substances, a process that depends on the morphological shape of the pathogen or on expression of certain molecules on the cell membrane 4,12 . To rule out the possibility of a weak cellular immune response induced by Chryseobacterium sp. because their morphological characteristics were different from those of other microorganisms, we analyzed E. coli K12 and the Chryseobacterium sp. under a scanning electron microscope (SEM) (Fig. supplement 1A,B). We found that the shape and size of Chryseobacterium sp. were very similar to those of E. coli K12: a typical rod-shape of approximately 2 μ m in length. Therefore, we concluded that morphological differences were unlikely to affect the ability of granulocytes to engulf the bacteria. Expression was normalized to that in LB medium-challenged larvae. Values are expressed as the mean of three replicates ± SEM, each containing three larvae (nine larvae per condition). * P < 0.05 (t-test).
Scientific RepoRts | 6:31722 | DOI: 10.1038/srep31722 Chryseobacterium sp. may elicit a considerably low immune reaction against granulocytes by altering expression of pathogen-associated molecular patterns (PAMPs) such as outer membrane protein (OmpA), peptidoglycan, LPS, lipoproteins, β -1,3-glucan, bacterial flagellin, and various porins, all of which are key molecules detected by professional phagocytes. This down regulation of PAMPs is common to many bacteria 48 . Weiss et al. 23 demonstrated that OmpA expressed by virulent bacteria is a strong inducer of host immunity. Thus, it appears that differences between avirulent and virulent pathogens are due to differences in the expression of common membrane constituents 23,49 . Recently, Kim et al. 50 showed that the absence of the bacterial lipopolysaccharide O antigen in the cell envelope was responsible for the increased susceptibility of the gut inhabiting bacteria to host antimicrobial peptides 50 . PAMPs are recognized by a group of proteins known as PRRs (pattern recognition receptors), which include β GRPs, GNBPs, C-type lectins, scavenger receptors, and PGRPs, which are generally classified as short (S) or long (L) forms [51][52][53][54] . We recently reported that P. brevitarsis seulensis expresses orthologs of PGRP-1, -2, -3, -LB, -LC, -LE, -SC2, and -SC3 33 . Bacterial motility is also essential for the successful symbiosis between host insect and gut inhabiting bacteria 55,56 .
As shown in Fig. 2A,B, larvae injected with E. coli K12 increased melanin production, which reduced their survival after infection. However, although larvae injected with Chryseobacterium sp. or LB medium did show slightly elevated levels of melanization at 24 h post-injection, this gradually and completely disappeared by 72 h (Fig. 2B). Melanization generally plays a role in thermoregulation, UV resistance, tolerance to desiccation, and resistance to abrasion 57 . Melanization in the cuticle is also another powerful immune mechanism that produces melanin at the site of wounds and provides resistance to infection; this type of melanization is visible as dark brown deposits 58 . Here, we found that expression of PGRP-1 transcripts increased by more than 10 times at 24 h post-infection with Chryseobacterium sp. before decreasing at 48 h post-infection. These results were also coincident with melanization in Fig. 2B, which almost completely disappeared after about 72 h post-infection with Chryseobacterium sp. or LB medium. By contrast, we observed a continuous increase in E. coli K12-challenged larvae for up to 48 h (Fig. 5A). Although the level of PGRP-1 transcripts need to be examined for more than 48 h, it seems likely that an increase of PGRP-1 transcripts might be maintained since melanization induced by infection with E. coli K12 remained high at 72 h (Fig. 2B). We also observed that PGRP-2 expression was kept constant higher during E. coli K12 infection (Fig. 5B). PGRP-1 and 2 activate PO (melanization) in flies, silk moths, and beetles 53,54,59 . In particular, Hd-PGRP-1 from Holotrichia diomphalia (Korean black chafer) recognizes β -1,3-glucan and induces prophenoloxidase (PPO) activity via serine proteases, which are clip domain enzymes that generate phenoloxidase (PO) 55,60 . In addition, we previously showed that melanotic encapsulation of pathogens mainly occurs in granulocytes activated only after infection with E. coli K12 4 . Therefore, PO-related melanization is adhesive elements to pathogen surface including hemocyte surfaces 61 .
As shown in Fig. 5C, expression of PGRP-SC2 increased 30-fold at 24 h post-infection with E. coli K12; however, expression of PGRP-SC2 was very low in Chryseobacterium sp.-challenged larvae. Although further confirmation is required, the expression of PGRP-SC2 could possibly be linked with bactericidal activity. PGRP-SC2 and -SC3 from this insect were more than 90% identical with PGRP-S2 and PGRP-S3 from Anopheles gambiae; both PGRP-S2 and PGRP-S3 are known as secreted bactericidal amidases 16 . Hence, these results suggest that Chryseobacterium sp., which does not induce PGRP-SC2 expression, is non-pathogenic or well adapted in this insect.
Finally, we examined expression of AMPs in response to E. coli K12 and Chryseobacterium sp. infection. AMPs are the last proteins to be secreted by the innate immune pathway. These proteins directly bind and kill bacteria in insects. Defencin A and coleoptericin act as AMPs in this insect 33,62 and their expression is indicative of an AMP response. As shown Fig. 5E,F, defensin transcripts were upregulated in E. coli K12-challenged larvae (by more than 30-fold). In addition, expression of coleoptericin-like transcripts increased in E. coli K12-challenged larvae (by more than 3,000-fold), but were expressed weakly or moderately in Chryseobacterium sp.-challenged larvae. Anselme et al. 63 showed that several immune genes, including coleoptericin, are highly overexpressed in the bacteriome, which is a specialized organelle that hosts endosymbiotic bacteria in some insects 63 . This AMP-driven selective pressure on endosymbiotic bacteria might be required to maintain the optimum number of microorganisms in the insect gut. Therefore, the very weak or moderate expression of defensin and coleoptericin transcripts in Chryseobacterium sp.-challenged larvae is probably due to the lack of a strong immune response, unlike the response to E. coli K12.
In summary, we cultured gut-inhabiting bacteria in vitro and examined host cellular and humoral immune responses against the bacteria. Compared to that of the E. coli K12-challenged larvae, we observed weaker or more moderate immune activities (e.g. lysosome and autophagosome formation in granulocytes or expression of immune-related genes) in Chryseobacterium sp.-challenged larvae. These results suggested that Chryseobacterium sp. may not have strong effector(s) or well-adapted to host immune system. It is very difficult to study the host immunological relationship between single bacterial taxa in the gut due to complex interactions between microorganisms that enable them to survive. The strategies used in this study can contribute to the ongoing investigation of the interactions between host immune system and single bacterial taxa inside of gut.

Methods
Insects, isolation and identification of intestine-adapted bacteria. The white-spotted flower chafers, Protaetia brevitarsis seulensis (Coleoptera: Cetoniidae) were reared and maintained as previously described 64 . In addition, the handling procedures of larvae including sample preparation, infection, and image analysis were also performed as previously described 4,12 .
The bacterial isolation from the gut of insects was performed as previously described 65 . Briefly, 10 last instar larvae were sterilized for 10 seconds in 95% ethanol and rinsing with sterile distilled water prior to dissection. The whole gut was aseptically removed under a stereomicroscope using insect pins and placed in 1.5 ml micro centrifuge tube containing 0.5 ml of phosphate buffer saline (NaH 2 PO 4 , Na 2 HPO 4 and NaCl (pH = 7.2)). The dissected Scientific RepoRts | 6:31722 | DOI: 10.1038/srep31722 gut were squeezed several times using glass pestle, 1 ml of PBS was added, and 3,000 rpm at 4 °C for 15 seconds to separate the microorganisms from the gut wall. Ten-fold dilution of the bacterial suspension was filtered in 0.22 μ m filter and the dilutions of the gut suspension were cultured immediately on SOC starch agar plates (2% tryptone, 0.5% Yeast extract, 8.56 mM NaCl, 2.5 mM KCl, dd H 2 O to 1000 mL, 10 mM MgCl 2 , 10 mM MgSO 4 , and 30 mM glucose). Then, six selected bacterial species including E. coli K12 were grown in LB media (although they grew more slowly) for 12 to 24 h at 28 °C and 0.5 ml liquid bacterial isolates were stored in liquid media containing 20% glycerol in cryovials at − 80 °C for further processing. All the isolated colonies were observed under microscope to check their colony morphologies i.e. color, shape, and pigmentation. Bacterial colonies were randomly selected and singularly plated on new agar plate. For 16S rRNA gene analysis of gut-inhabiting bacteria, the genomic DNA extracts from each bacteria. Next, PCRs were performed to amplify 1.4 to 1.5 kb of the 16S rRNA gene from all the genomic DNA samples using 27F (5′ -AGAGTTTGATCMTGGCTCAG-3′ ) and 1492R (5′ -TACGGYTACCTTGTTACGACTT-3′ ). The PCR conditions comprised: 3 minutes at 94 °C, followed by 30 cycles of 94 °C for 30 seconds, 58 to 48 °C for 30 seconds (the temperature was decreased by 1 °C every cycle for 10 cycles and then held at 48 °C for 20 cycles), 72 °C for 1 minute, followed by a final extension step at 72 °C for 20 minutes. The purified PCR products were sequenced at Macrogen, Korea, using 785F (5′ -GGATTAGATACCCTGGTA-3′ ) and 907R (5′ -CCGTCAATTCMTTTRAGTTT-3′ ). We obtained 16S rRNA sequences data (1. 47  Bacterial injection, hemocyte collection and counting, and statistical analysis. For injection, the last instar larvae were cold-anesthetized and a finely pulled glass needle was shallowly inserted into the dorsal vessel; around 40 μ l (with 10 6 colony-forming unit (CFU)) of cultured bacteria were injected into the hemocoel. For survival curve, the hemocoel of 60 larvae was injected and the larvae were divided into three groups: a Chryseobacterium sp.-injected group, an E. coli K12-injected group (a well-known pathogenic bacterium in this insect) [positive control], and an LB medium-injected group [negative control]. Hemocytes and fatbodies were collected and analyzed by direct puncture of the dorsal blood vessel or following dissection of larva abdominal cuticles of each larva at 0 h, 12 h, 24 h, and 48 h post injection to evaluate hemocyte morphology and the expression of immune-related genes. About 0.3 ml of hemolymph was collected in 1.5 ml micro centrifuge tube containing anti-coagulant solution (98mM NaOH, 186mM NaCl, 17 mM EDTA, and 41 mM citric acid, (pH 4.5)). After centrifugation (1,000 g for 10 min at 4 °C), the plasma were removed. For the Total Hemocyte Counts (THC), hemocytes were placed in a sterile disposable hemocytometer slide (Ncubauer Improve, iNCYTO C-Chip DHC-N01. www.incyto.com) (10 μ l capacity). The THC were counted in four squares using a light microscope and the percentages were determined per individual larva (Leica DMI 3000B; 40 × objective). The THC was calculated at 0, 12, 24, and 48 h post-infection. Statistical tests were performed using the Minitab software package. Significant difference of means was calculated using Student's two-tailed t-test or one-way ANOVA at a probability (P) value of less than 5%. The 16S rRNA gene sequences were compared with the National Center for Biotechnology Information Genbank database using the BLASTn (http://www.ncbi.nlm.nih.gov/BLAST/) and the Ribosomal Database Project II (RDP II) (http://rdp.cme.msu.edu). A phylogenetic study was performed by program MEGA 6.0 and a phylogenetic tree was constructed with the computer package ClustalW and the relationships between isolates were determined by the neighbor joining method 66,67 . A bootstrap analysis was performed to evaluate the topology of the phylogenetic tree (Bootstrap = 1,000).
Hemocytes staining, fluorescence-activated cell sorting (FACS) analysis, and survival analysis. Hemocyte labeling, FACS, and visualization including image processing and microscopy analysis in this insect were performed as previously described 4,12 . Briefly, hemocyte morphology and phagocytic activities including lysosomal formation and pathogenic related autophagosome were observed with an Olympus FV1000 confocal microscope and Olympus image application software. Especially, DAPI (4′ -6-diamidino-2phenylindole; 5 μ g/ml) for nucleus, fluorescently-conjugated phalloidin (F-actin cytoskeleton) (6.6μ M; Molecular probes), LysoTracker Red (7.5 nM; Molecular Probes) for lysosomes, and a Cyto-ID Autophagy Detection Kit (Enzo) for autophagy protein, LC3 were used. Hemocytes were stained with each dye for 30 min, washed three times with PBS, fixed with 4% paraformaldehyde for 15 min, washed again three times with PBS, and mounted. 10,000 hemocytes per sample (three larvae) was determined by flow cytometry (a BD FACSCanto flow cytometer, BD Bioscience; San Jose, CA) and sample analysis was performed according to protocols developed for this application using FACSDiva software from BD Biosciences. The channels for green fluorescence and for red fluorescence were FL1 (530/30 band-pass) and FL3 (610/20 band-pass). The survival analysis was replicated on three consecutive days. On each day 10 larvae of each group were hemocoelic injected by LB medium, Chryseobacterium sp., or E. coli K12. Mortality of each group (LB medium, Chryseobacterium sp., or E. coli K12 injection group) were analyzed every 12 h until 180 h following injection. Kaplan Meier survival estimates and log rank test for comparison were used for time to death analysis for LB medium-, Chryseobacterium sp.-, or E. coli K12-challenged group. A significant P value from the log rank test was set at < 0.05. Data were analyzed using IBM SPSS statistics 22.0.

Quantitative real-time RT-PCR.
To investigate the expression levels of immune-related genes, total RNA was extracted from each larva fatbodies at 12, 24, and 48 h post injection (Promega Corp., Madison, WI, USA) and first-strand cDNA was synthesized using SuperScript III First-Strand Synthesis SuperMix (Invitrogen, Karlsruhe, Germany). For real-time quantitative RT-PCR (qPCR), reactions were performed using 72-well Rotor with in the Rotor-Gene-Q (Qiagen, USA) with 5 μ l SYBR GreenI master mix (Elpis-biotech, Korea), 0.2 μ l of cDNA, and 0.4 μ M each of gene-specific primers (Table supplement 1). The qPCR reactions were done in triplicate (40 cycles of 95 °C for 10 sec, 60 °C for 10 sec and 72 °C for 10 sec with a single 470 nm green channel fluorescence measurement). Expression was normalized to β-actin and calibrator (control group) was LB medium-challenged larvae. This replicate was assigned a relative concentration of 1. The statistical analyses of the qPCR data were performed using Prism 5.0 (GraphPad Software, San Diego, CA, USA). Significant difference of means was calculated using paired t-test (*p < 0.05).