Discovery of a new highly pathogenic toxin involved in insect sepsis

ABSTRACT Insect sepsis is a severe consequence that arises from the invasion of the hemocoel by symbionts of entomopathogenic nematodes and bacteria. In the present study, we unveiled the heightened virulence of the entomopathogenic nematode Steinernema feltiae and the entomopathogenic bacteria Xenorhabdus bovienii, which operate symbiotically, against the wax moth Galleria mellonella. Maximum mortality was observed at 25°C while the optimal infestation efficiency was 20 nematodes per host. After infestation, G. mellonella displayed rapid darkening and softening, accompanied by an escalated esterase activity at 9 h. The X. bovienii, released by S. feltiae, underwent substantial proliferation and discharged toxins that attacked hemocytes, thus triggering extensive hemolysis and sepsis. The host G. mellonella was usually killed within 24 h due to disseminated septicemia. Additionally, X. bovienii infestation led to the upregulation of metabolites like 3-hydroxyanthranilic acid. Strikingly, we identified the perilous actinomycin D, generated through kynurenine metabolites, representing a novel biomarker of insect sepsis. Furthermore, a comprehensive transcriptomic analysis unveiled a noteworthy upregulation of gene expression associated with actinomycin D. Overall, X. bovienii induced apoptosis and sepsis through actinomycin D production, indicating its pivotal role in infestation activity. These findings open up new avenues for studying the mechanism of sepsis and developing innovative biotic pesticides. IMPORTANCE As a current biocontrol resource, entomopathogenic nematodes and their symbiotic bacterium can produce many toxin factors to trigger insect sepsis, having the potential to promote sustainable pest management. In this study, we found Steinernema feltiae and Xenorhabdus bovienii were highly virulent against the insects. After infective juvenile injection, Galleria mellonella quickly turned black and softened with increasing esterase activity. Simultaneously, X. bovienii attacked hemocytes and released toxic components, resulting in extensive hemolysis and sepsis. Then, we applied high-resolution mass spectrometry-based metabolomics and found multiple substances were upregulated in the host hemolymph. We found extremely hazardous actinomycin D produced via 3-hydroxyanthranilic acid metabolites. Moreover, a combined transcriptomic analysis revealed that gene expression of proteins associated with actinomycin D was upregulated. Our research revealed actinomycin D might be responsible for the infestation activity of X. bovienii, indicating a new direction for exploring the sepsis mechanism and developing novel biotic pesticides.

of agriculture (3).EPNs are highly pathogenic toward insects, species specific, and environmentally friendly, thereby benefitting humans (4).The infective juveniles (IJs) of Steinernema spp.are the only free-living nematodes that enter and invade insect hosts via surface penetration (5,6).Then, the EPNs release entomopathogenic bacteria (EPBs), tightly and longitudinally arranged in foregut vesicles, into the insect hemolymph (7).EPBs proliferate and generate a wide range of toxin molecules that kill the host within 48 h (8).The symbiosis of EPNs/EPBs induces physiological and biochemical responses in the host hemolymph (9).For example, esterases in the hemolymph of insects, which are a class of critical proteins that can affect the transport and metabolism of normal lipids and nerve conduction (10,11), are upregulated for breakdown of the cytomem brane and disruption of the immune system (11).Then, overexpression of esterases leads to dysfunction or direct destruction in host hemocytes (12).The EPB Xenorhabdus bovienii can function alone to kill insect hosts, although both Steinernema feltiae and X. bovienii are involved in the lethality of insects (13,14).To date, all Xenorhabdus species have exhibited phenotypic variation; they are usually isolated directly from EPNs in the primary form but some cells progress to the secondary form under nutrient and oxygen depletion during in vitro cultivation (15,16).The primary forms of EPBs can attach to the surface of specific tissues in the hemocoel via flagella, producing a variety of immunosuppressive factors, toxin proteins, and specialized metabolites during infection that allow the host to die rapidly (17)(18)(19)(20).However, the secondary forms of EPBs exhibit weaker motility and pathogenicity, but they have greater capacity for survival and reproduction, which allows them to better adapt to the in vitro environment (7,13,21,22).
We explored three research topics in this study.First, we studied the environmentdependent infestation by EPNs and their phenotypes associated with insect mortality.Second, we performed multiomics (e.g., metabolomics and transcriptomics) analysis of the virulence of the primary and secondary forms of EPBs, as well as examination of the synthetic pathways of potential biomarkers.Third, we addressed the phenotypic differences in EPBs and infestation mechanisms.Metabolomics was used to reveal the virulence effects of potential toxins via synthetic pathway analysis.We hope that these findings will be helpful in facilitating the study of the pathogenic mechanisms of EPNs.

Life history of S. feltiae
The long lifespan of EPNs includes the egg stage, four larval stages, and the adult stage within the ~13 days they spend in their insect hosts (Fig. 1A).A single female adult EPN can produce approximately 800 eggs, and an individual G. mellonella can release 41,056.67± 12,699.65 IJs into the surrounding environment (Fig. 1A).The body lengths of male and female EPNs were 1.32 ± 0.20 and 2.09 ± 0.25 mm, respectively (Fig. 1B).Additionally, the body length of IJs was nearly 30 times the greatest body width and the length was the most distinct feature of the IJs compared with the nematodes at other stages (Fig. 1C).

Insecticidal activity of S. feltiae
The increasing temperature could enhance the mortality of G. mellonella, which indicates that it takes 24 h to achieve 100% mortality post IJ injection at 25°C (Fig. 1D).The mortality of Bactrocera dorsalis similarly peaked at 71.9% at 25°C (Fig. S1A).Therefore, the optimal temperature for S. feltiae and X. bovienii infestation of insect prey was 25°C, with high insecticidal activity.The EPN/EPB symbiotic infestation of G. mellonella was significantly higher than that of B. dorsalis (F = 16.00,P < 0.001).When the same dose was instilled onto the body surface of B. dorsalis, virulence attenuation and hysteresis were observed (Fig. 1E; S1B).The optimum density of IJs was assayed in the range of 20-50 IJs/ larva, which showed significantly higher mortality at 48 h compared with 5-10 IJs/larva (F = 17.07,P < 0.001).These observations indicate that both temperature and IJ dosage could affect the infestation efficacy, which means an optimal infestation condition (20 IJs/larva at 25°C).

Esterase and hemocyte characteristics of infected G. mellonella
The infected G. mellonella turned gray or black, softened 15 h postinjection, and died within 24-48 h (Fig. 2A).Intriguingly, the infested G. mellonella had higher esterase activity in the hemolymph and peaked at 9 h postinjection (Fig. 2C).The hemocytes began to exhibit cell membrane lysis.A few X. bovienii were found in the hemolymph 12 h after IJ infection (Fig. 2B).In contrast, the number of hemocytes exhibited a slight increase at 9 h and then rapidly decreased, with a large number of lysed hemocytes present in the hemolymph (Fig. 2D).

Phenotypic variation characteristics and virulence of X. bovienii
The primary form of X. bovienii causes the insect cadaver to become flaccid and liquefied, while the secondary form causes rigidity (Fig. 3A).The secondary form of X. bovienii grew more rapidly than the primary form (F = 3.128, P < 0.001).Thus, the primary form exhibited small green colonies on NBTA, while the secondary form grew as red colonies with larger diameters (Fig. 3B).The primary form had significantly higher toxicity toward insect hosts than the secondary form during infestation (F = 0.235, P < 0.05 Fig. 3C).At 120 h, both the EPB alone and cell-free fermentation broth showed the same toxic activity, both at 100%, toward the insect host for the primary form of X. bovienii with no difference observed; although the secondary form exhibited relatively weaker toxicity, there was no significant difference between the EPB alone and the cell-free fermentation broth (F = 1.839,P ＞ 0.05) (Fig. 3D).Therefore, the active metabolites causing insect mortality were present in the cell-free fermentation broth of the primary form of X. bovienii.

Screening and identification of insecticidal substances
A total of 158 differential metabolites (P < 0.05, fold change > 1.5, and CV < 20) were identified in the positive ion mode in the G. mellonella hemolymph metabolomics sample.These metabolites included organoheterocyclic compounds, phenylpropanoids, polyketides, benzenoids, organic acids and derivatives, organic oxygen compounds, organosulfur compounds, nucleosides, nucleotides, and their analogs.Significant differences were observed between the principal component analysis (PCA) scores of the IJ-injected group and those of the water-injected group and the untreated control group.In addition, the PCA plots revealed differences in metabolic substances among the three groups (Fig. 2E).In total, 392 chemicals were considerably upregulated in the treatment group relative to the negative control group, whereas 290 compounds were significantly upregulated in the treatment group close to the positive control group (Fig. 2F and G).Sixty-seven chemicals were discovered to be jointly upregulated (Fig. 2H).The pathway topology analysis revealed eight metabolic pathways: glycerophospholipid metabolism; phenylalanine (Phe), tyrosine (Tyr), and Trp biosynthesis; alpha-linolenic acid (ALA) metabolism; Trp metabolism; glycine (Gly), serine (Ser), and threonine (Thr) metabolism; cyanoamino acid metabolism; and arginine (Arg) biosynthesis (Fig. 2I).Through effect size analysis, 5-hydroxyindolealdehyde (5-HIAL), methionine (Met), and 5-hydroxytrypto phan (5-HTP) were identified as the top ranked among all the potential chemicals.
By analyzing differences in fermentation broth between the primary and secondary forms of X. bovienii, 107 and 57 metabolites were obtained in the positive and negative ion modes, respectively.These metabolites included alkaloids and their derivatives; benzenoids; hydrocarbon derivatives; lignans, neolignans, and related compounds; lipids and lipid-like molecules; nucleosides, nucleotides, and their analogs; organic acids and their derivatives; organoheterocyclic compounds; and phenylpropanoids and polyketi des.The PCA score plots showed that the primary and secondary forms were mainly separated along t2 (Fig. 3E).Volcano plots showed that 324 substances were significantly upregulated in the primary form of the EPB (Fig. 3F).Pathway analysis revealed that the substances with defined structures were mainly enriched in histidine (His) metabolism, Arg biosynthesis, glutathione (GSH) metabolism, nitrogen (NH) metabolism, D-glutamine (Gln) and D-glutamate (Glu) metabolism, aminobenzoate degradation, and Arg and proline (Pro) metabolism (Fig. 3H).The results of transcriptomic analyses showed that 4,523 genes were mapped to the reference genome, and the PCA score plots showed that the samples of the primary and secondary forms of the bacterium were well separated (Fig. 4A).The volcano plot showed that 1,597 genes were differentially expressed in the transcriptome, of which 363 were significantly upregulated and 1,234 were downregulated.The upregulated genes were enriched in metabolic pathways involving aromatic compounds; Phe metabolism; benzoate degradation; Tyr metabolism; inositol phosphate (IP) metabolism; valine (Val), leucine (Leu), and isoleucine (Ile) degradation; β-alanine (Ala) metabolism; and the tricarboxylic acid cycle (TCA) (Fig. 4B and C).Again, GO function enrichment showed that the significantly different genes had 95 annotations attributed to biological processes, 69 to cellular components, and 90 to molecular functions.Numerous activities linked to xenobiotic and toxin metabolism were enriched in the biological process category, while processes related to bacterial flagellum metabolism were the most abundant in the cellular component category.Additionally, many genes associated with type VI secretion system (T6SS) assembly, methyltransferases, and NRPSs were considerably upregulated; however, this pathway was not significantly enriched (Fig. 4D and E).Comparison of all the enriched pathways at the metabolomic and transcriptomic levels showed that the Arg and Pro metabolism pathway was present in both the bacterial transcriptome and the supernatant metabolome.

Toxicity of potential insecticidal substances
In total, 40 components, including amino acids and their derivatives, antibiotics, and macrolides, were finally screened for toxicity verification.Actinomycin D showed high insecticidal activity (F = 0.49, P < 0.001).Additionally, N-acetyl-L-Leu (F = 3.23, P < 0.001), N-acetyl-L-aspartic acid (Asp) (F = 8.91, P < 0.01), and glycocholic acid (F = 1.13,P < 0.001) had high hemocoelic toxicity toward the insect host G. mellonella.Actinomycin D exhibited the highest hematological toxicity, with a cumulative mortality rate of 86.49% at 72 h (Fig. 5).The mortality trend in B. dorsalis was the same as that of G. mellonella for each chemical, with actinomycin D (F = 5.67, P < 0.05) lethality ranked at the top again (Fig. S2).Interestingly, N-terminally acetylated amino acids, such as N-acetyl-L-Asp, Nacetyl-L-Met, and N-acetyl-L-Leu, showed higher insecticidal activity regardless of the method of exposure or injection.

Regulation of sepsis in the hemolymph by actinomycin D
4-Methyl-3-hydroxyanthranilic acid (4-MHA), containing an o-aminophenol group, was upregulated in kynurenine (Kyn) metabolism in the Trp pathway; 4-MHA and the five amino acids with the peptide lactone rings are subsequently assembled into the 4-MHA pentapeptide lactone.Two 4-MHA pentapeptide lactones undergo dimerization to produce the phenoxazinone actinomycin chromophore (Fig. 6A).Multiple amino acids with pentapeptide lactones in the form of dipeptides were upregulated in the hemo lymph of infected G. mellonella (Fig. 6B).Some genes related to the methyltransferases responsible for converting 3-HAA to 4-MHA and NRPSs were found to be upregulated in the primary form of X. bovienii.XBW1_RS00535, XBW1_RS02860, XBW1_RS15570, and XBW1_RS15900 encode methyltransferases, and the rest are NRPS-associated genes (Fig. 6C).

DISCUSSION
Although it has been known for many decades that infestation with EPNs/EPBs leads to efficient insect sepsis, little is known about the mechanism and biomarkers produced by them (56).Our findings were consistent with previous results showing that EPBs were indeed capable of causing sepsis in insects and were highly virulent.Meanwhile, EPB invasion stimulates the immune response of host insects, such as by enhancing esterase activity and increasing the abundance of immune cells (11,57).We found that the change in esterase activity disrupted the balance of the insect immune system and the resulting fat body impairment prevented further resistance to dangerous EPB invasion (55).As in previous studies, the insect hemolymph infected by IJs showed a blurred nuclear boundary, then the hemocyte membrane was utterly broken, and the contents of the cells flowed out; eventually, the cells became apoptotic (58).It was speculated that the increase in the THC in the early stage might have been due to the rapid division of protohemocytes and the pinocytosis and phagocytosis of bacteria.Over time, the EPB aggravated the destruction of the host defense system and began to multiply in large numbers, causing hemocyte deformation, transformation, and disintegration.Finally, the number of hemocytes was reduced to a minimum.Interestingly, there were differences in the insecticidal activity of the two forms of the EPB.They also differed markedly in phenotype compared with that in which IJ injection led to insect sepsis (7).Other studies have shown that only when the EPB carried by the EPN was the primary form were the carcasses swollen and "floppy." In contrast, the body a secondary form for killing maintained a rigid exoskeleton (59).This indicates that the EPN at the IJ stage specifically expresses some products with activity toward insect tissues and the expression of these products might be triggered by only the primary form of the EPB (60).Furthermore, there was also a significant difference in the toxicity of the cell-free fermentation broth of the two forms of the EPB and the insecticidal activity of the secondary form was relatively weak.In conclusion, our results demonstrated that the symbiotic EPB in IJs was a highly toxic primary form and key to the septicemia and insecticidal activity of IJs.The critical biomarker was also present in its cell-free fermentation broth.
Recent studies have shown that bacterial sepsis might be associated with immune cell apoptosis; it plays a critical role in biological development, self-protection, and defense against interference from external factors (61).Many compounds have been found to initiate the apoptotic program of cells, such as actinomycins of the cytotoxic antibiotic drug family.Actinomycin D is the most common actinomycins, as it is an isoactinomycin composed of similar amino acids, including Thr, D-Val, Pro, sarcosine (Sar), and N-methylvaline (MeVal), where the source of sarcosine is Gly (62).Actinomycin D was detected in our results and exhibited extremely high insecticidal activity.Paramanathan et al. indicated that actinomycin D can block DNA-dependent mRNA synthesis by binding to DNA, thereby impeding protein synthesis, inhibiting cell division, and causing apoptosis (63).An essential precursor for actinomycin D synthesis, 4-MHA, is a unique compound derived from Trp and occurs in only the actinomycete family of bacteria; more common than 4-MHA in nature is its homolog 3-HAA (64).It is speculated that Trp, as the essential amino acid in organisms, is largely consumed and used to synthesize a variety of substances that play a role in host insects.These findings prove that Trp metabolism, especially Kyn metabolism, plays a critical role in the interactions of pathogens and host insects and may have a regulatory effect on immune cells.Because the expression levels of related substances and enzymes of the Trp metabolic pathway in the two forms of X. bovienii were significantly different, which was probably also related to the synthesis and expression of actinomycin D, we concluded that the Trp pathway was the critical metabolic pathway influencing EPB toxicity and that actinomycin D was the biomarker responsible for the difference in insecticidal activity of the two forms of X. bovienii.
Moreover, other compounds in this study also showed some insecticidal activity.They may play a synergistic role in insect killing or be of great importance in other processes, such as bacteriostasis.These compounds included glycocholic acid, with fat-dissolving properties and high cytotoxicity, as well as some acetylated amino acids, and all of them exhibited some pathogenicity toward G. mellonella (65,66).Previous studies have shown that N-acetyltryptamine does have higher nematocidal, antibacterial, and insecticidal activities than tryptamine due to the presence of N-acetyl groups (67).However, whether X. bovienii can use the above compounds to synthesize substances that have insecticidal activity in insect-nematode-bacteria interactions, such as actinomycin D, and actinomycin D plays a role in causing insect sepsis remain unclear.Therefore, in-depth investigation of the metabolites of the mechanisms of action of the aforementioned compounds, as well as the metabolites of the Kyn pathway, could contribute to the development of novel insecticide lead compounds.
Taken together, after the EPN/EPB was delivered into the hemolymph of the host insect, the G. mellonella body gradually exhibited graying with black speckles and then rapidly disintegrated, leading to death.The actinomycin D content in the fermentation broth of the primary form of X. bovieniii is higher compared to that in the secondary form, which may contribute to the higher pathogenicity of the primary form bacteria.Through the combined analysis of transcriptomics and metabolomics of both forms of X. bovienii, several precursor substances and enzyme activities involved in actinomycin D synthesis were found to be significantly upregulated in the primary form X. bovienii.Considering the biological activity of actinomycin D in inducing cell apoptosis and exhibiting toxicity to insects, it can be speculated that actinomycin D produced by X. bovienii may contribute to insect sepsis.However, the specific mechanisms by which actinomycin D causes insect sepsis remain unclear.Therefore, in-depth investigation of the mechanisms of action of actinomycin D in insect sepsis would be beneficial for understanding its important role in biological control.

Conclusion
In this study, we discovered new biomarkers and revealed the EPN/EPB infection mechanism on insect hosts that intuitively interact using ecological observations and omics techniques.EPBs may cause host insect hemolymph septicemia by releasing a series of biomarkers (e.g., actinomycin D, glycocholic acid, and acetylated amino acids).By jointly analyzing the hemolymph metabolome, fermentation broth metabolome, and transcriptome, we found that actinomycin D might be the critical biomarker of X. bovienii and may cause hemolymph septicemia.In addition, the synthesis of actinomycin D in X. bovienii was also summarized based on the synthesis in Streptomyces spp.In conclusion, we found that both temperature and IJ dose significantly impacted the pathogenic capacity of EPNs and the host insects could undergo a large degree of sepsis both at very low IJ doses and in the cell-free fermentation broth of EPBs.We demonstrated that the primary form of X. bovienii is the key factor underlying host death and that these biomarkers with lethal activity were present in its fermentation broth.After EPB infection, the G. mellonella surface or the esterase and hemocytes in the body change dramatically, which is considered necessary for insect immunity.Therefore, these results will be helpful in exploring the biocontrol potential of EPNs/EPBs and biocontrol technology.This study may provide a new model for generating more effective, safe, and environmentally sustainable chemical insecticides to protect people and animals from the effects of hazardous synthetic insecticides.

Experimental species
S. feltiae SF-SN, G. mellonella, and B. dorsalis were preserved and bred at the MOA Key Laboratory of Pest Monitoring and Green Management, China Agricultural University.
The EPNs were propagated as previously described and collected using the White trap technique with slight modifications, and nematodes at different stages within cadavers of G. mellonella were observed (68).The G. mellonella larvae were placed in Petri dishes (90 × 15 mm) containing a double layer of filter paper soaked with IJ suspension at 25°C.The newly emerged IJs from the cadavers were collected with a White trap, washed with sterile water three times, and stored in 250-mL tissue culture flasks at 13-15°C.

EPN virulence assay
The virulence of IJs at different temperatures and densities was determined by a mortality assay.Both B. dorsalis and G. mellonella larvae were surface sterilized using 70% (vol/vol) ethanol, washed with sterile water, and anesthetized on ice for insect pretreatment.Nematode suspensions of 20 IJs/larva were utilized for the virulence assays at 16, 19, 21, 25, 28, and 31℃.EPN densities of 5 IJs/larva, 10 IJs/larva, 20 IJs/larva, 30 IJs/ larva, 40 IJs/larva, and 50 IJs/larva were used for virulence assays at 25°C with phosphatebuffered saline (PBS) as a control.Approximately 10 µL of nematode suspension was added dropwise onto the body surface of B. dorsalis and injected into the body cavity of G. mellonella from the third pair of gastropodia using a sterilized microsyringe (Hamilton 1702 RN, 25 μL).All treated insects were placed individually into the wells of 12-well culture plates with a double layer of wet filter paper.Insect larvae treated with different densities of nematode suspension were placed separately in a circulating, intelligent, artificial climate chamber at 25°C, 70 ± 5% RH, and 14L:10D.Additionally, insect larvae treated with different temperatures were placed in the same consistent conditions.The insect larvae were considered dead from EPN infection when they did not respond to prodding with a needle tip and their bodies showed typical signs of darkening and softening.The cohort mortality of G. mellonella was checked every 6 h, and that of B. dorsalis was checked every 12 h.In total, 50 larvae of G. mellonella with five replicates and 110 larvae of B. dorsalis with 11 replicates were used to examine the mortality responses to EPN infestation.

Hemocyte observation
To determine hemocyte pathology, observation was performed using a larva of G. mellonella injected with 20 IJs.The G. mellonella abdominal segment proleg was cut with curved tissue scissors, and the hemolymph was allowed to ooze out on a clean, oil-free glass microscope slide.The hemolymph was quickly diluted 20-fold with 1-2% glacial Turek solution and transferred to a disposable cell counting plate for counting under a microscope.The preparations were discarded if the cells aggregated or were distributed unevenly within the chamber (57).The changes in the pathological and total hemocyte count (THC) of hemocytes of the defined infected insects and control hemolymph were recorded every 3 h within 24 hr.

Esterase activity assay
Fifty microliters of hemolymph was added to a mixed solution of PBS (0.45 mL, 0.04 mol/L) and α-naphthyl acetate solution (3 mL, 3 × 10 −4 mol/L).Then, 0.5 mL of a mixed solution (2:5, vol/vol) of fast blue RR salt solution (10 g/L) and SDS (15 g/L) was added for approximately 30 min to determine the OD 600 of the hemolymph culture to characterize the esterase activity.The OD 600 of the defined infected insects and control hemolymph was recorded every 3 h within 24 hr.

Bacterial viability and virulence assay
Stable primary and secondary forms of X. bovienii were purified by differential separation by color on NBTA selective medium (5 g/L beef peptone, 10 g/L tryptone, 5 g/L NaCl, 0.025 g/L bromthymol blue, 0.04 g/L triphenyl tetrazolium chloride, and 15 g/L agar) (71).The viability of the primary form and secondary form of X. bovienii on NBTA was characterized using colony diameter.A sterilized toothpick was dipped in the bacterial culture solution in the mid-logarithmic phase, and the toothpick was used to inoculate the center of the NBTA medium; the colony diameter was measured every 12 h for a total of 60 h.
Single colonies for the primary and secondary forms of X. bovienii from each NBTA selective medium were selected and grown overnight in 5 mL of LB broth supplemented with 0.1% (wt/vol) sodium pyruvate (LBP) to the mid-logarithmic phase.One milliliter of each bacterial culture was centrifuged at 12,000 rpm for 5 min to obtain cell-free fermentation broth and EPB cells.Bacterial cells were washed with PBS three times by centrifugation, and the sample was diluted five times with a 1:10 serial dilution to obtain a bacterial cell suspension of lower concentration.Ten microliters of inoculum was used for each form at the test concentration was injected into G. mellonella (72).PBS without fermentation broth and EPB cells was used as a control.The cohort mortality was checked every 6 h for 120 h.A total of 60 larvae, divided into six groups for each form at the test concentration, were used, and the other experimental settings were the same as those described above.

Hemolymph metabonomics
The hemolymph of G. mellonella was collected after injection of 20 IJs and transferred to a 1.5-mL centrifuge tube.The untreated hemolymph sample from G. mellonella was used as a negative control, and to eliminate any errors caused by injection methods such as piercing, sterile water injection was used as a positive control.In each test, G. mellonella was separated into six groups and 50 µL of hemolymph was collected from each group.Each sample was added to precooled MeOH-ACN (1:1, vol/vol), mixed well by shaking, and centrifuged at 12,000 rpm and 4℃ for 10 min.Then, the supernatant was vacuum dried and 40 µL of MeOH-H 2 O (1:1, vol/vol) was added for reconstitution.The fermentation broth sample was shaken for 30 min, centrifuged at 12,000 rpm and 4℃ for 10 min, filtered through a 0.1-µm membrane, and then used for analysis.A UPLC-HRMS system (UPLC, ACQUITY UPLC I-Clas Bio, Waters; MS, Q-Exactive Focus, Thermo Scientific) equipped with a heated electric spray ionization (HESI) source was used for hemolymph extraction, and MS analysis was performed in the positive ion mode.The analysis program setting was as previously described with slight modification (73).

Cell-free fermentation broth metabolomics
The cell-free fermentation broth of X. bovienii was collected and stored on dry ice with eight replicates.The mixture was shaken slowly at 4°C, and an appropriate amount of fermentation broth was added to precooled MeOH-ACN-H 2 O (2:2:1, vol/vol), vortexed, ultrasonicated at 4°C for 30 min, and centrifuged at 14,000 rpm and 4°C for 20 min.The fermentation broth was dried in vacuo, reconstituted in 100 µL of ACN-H 2 O (1:1, vol/vol), vortexed, and centrifuged at 14,000 rpm at 4°C for 15 min.Fermentation broth samples were retained for metabolite profiling using an ultra-high performance liquid chromato graph (UHPLC ) (1290 Infinity LC, Agilent Technologies) coupled to a time-of-flight (TOF) quadrupole platform (AB Sciex Triple-TOF 6600) in positive and negative ion modes.Xcalibur 4.0 software (Thermo Fisher) was used for data acquisition and to calculate peak areas, and SIEVE 2.1 software (Thermal Scientific) was used for optimization of peak alignment and component extraction from the original data.Metabonomics databases (HMDB, KEGG, LIPID, MAPS, MassBank, MeSH, METLIN, and PubChem) were used for metabolite identification, and then, the fragment ion spectrum was matched with the candidate compound by MS/MS spectral database matching.Finally, MetaboAnalyst 3.0 was used to screen and analyze the correlations and pathways of differential metabolites.

Transcriptome sequencing
The difference in virulence between primary and secondary forms of X. bovienii was represented by the differential expression of genes measured by transcriptomic analysis.The fermentation broth in the mid-logarithmic phase was centrifuged to collect bacterial cells for transcriptomic sequencing on the Illumina HiSeq platform (Illumina, Inc., San Diego, CA, USA).The sequencing results were assembled using Rockhopper according to the reference genome (X.bovienii strain CS03) and compared with the annotated gene model.For RNA-seq analysis, the gene expression level was estimated by counting the sequence reads located in the genome region or gene exon region.Then, DESeq2 was used to carry out gene expression difference analysis, study the functions of differentially expressed genes, and visualize the results of differential expression analysis.

Potential biomarker virulence assay
The differential metabolites were dissolved in sterile water at 1 mg/mL.Ten microliters of the final solution was injected into the body of G. mellonella and dripped onto the body surface of B. dorsalis to verify its toxicity.The other treatments were the same as those described above.

Statistical analysis
The corrected mortality (M c ) was calculated using the following formula: M c =(M tre − M ctrl )/(100 − M ctrl ) (39), where M tre and M ctrl are the mean mortality treatment and control, respectively.
ImageJ software 2 (Wayne Rasband, National Institutes of Health, Bethesda, MD, USA) was used to visualize and measure nematode eggs, body length, body width, etc. One-way ANOVA was used to analyze all obtained data using SPSS software 25.0 (SPSS Inc., Chicago, IL, USA).Origin software 2022 (OriginLab Inc., Massachusetts, USA) was used to perform logistic analysis, nonlinear fitting, multivariate statistical analyses, PCA, mortality mapping, and preparing heatmaps and other graphs.

FIG 2
FIG 2 Pathologic phenotype and metabonomic analysis of the hemolymph.(A) Infestation processes of the EPN/EPB symbionts (S. feltiae on G. mellonella).(B) Hemocoel response of G. mellonella to EPN/EPB infestation.(C) Esterase activity of G. mellonella.(D) Changes in the THC after injection of S. feltiae.(E) PCA score plot for S. feltiae injection (EPN), untreated samples used as negative control (NC), and sterile water injected as a positive control (PC).(F) Volcano plots (EPN vs. NC).The horizontal coordinate indicates the change in gene expression [log 2 (fold change)], and the vertical coordinate indicates the significance level [−log 10 (P value)].Red and green dots represent the upregulated genes and downregulated genes, respectively.(G) Volcano plots (EPN vs. NC).(H) Venn diagram analysis of significantly differentially abundant compounds in EPN vs. NC and EPN vs. PC.(I) Results of pathway analysis.(J) Results of effect size analysis.

FIG 3
FIG 3 Pathologic phenotype and metabolomic analysis of X. bovienii.(A) The colony phenotype of the primary (PRY) and secondary (SEC) forms of X. bovienii on NBTA.(B) X. bovienii survival capability analysis in vitro.(C) The effects of X. bovienii for the primary and secondary forms on infestation and the mortality of G. mellonella.(D) Effect of the two forms of X. bovienii, EPB cells alone and cell-free fermentation broth (FB) alone on pathogenic ability.(E) PCA score plot of the primary and secondary EPB fermentation broths.(F) Volcano plots of PRY vs. SEC.(G) Heatmap of differential compound expression in the fermentation broth of the primary and secondary forms.Differential expression is shown as log 2 (fold change).(H) The enriched Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways of the differential metabolites in PRY compared with SEC.

FIG 4
FIG 4 Transcriptomic analysis of the primary and secondary forms of X. bovienii.(A) PCA score plot of the primary (PRY) and secondary (SEC) forms of EPB.(B) Volcano plots.The horizontal coordinate indicates the change in gene expression (PRY vs. SEC) [log 2 (fold change)], and the vertical coordinate indicates the significance level [−log 10 (P value)].The upregulated genes and downregulated genes are represented by red and green dots.(C) The enriched KEGG pathways of differentially expressed genes in the PRY form compared with the SEC form.(D) Gene function classification (GO).(E) Heatmap of the expression of the T6SS and toxin protein-associated genes in the PRY and SEC forms of the bacteria.Differential expression is shown as the log 2 TPM of the PRY form of X. bovienii/TPM of the SEC form of X. bovienii.

FIG 5
FIG5 Toxicity of potential biomarkers in G. mellonella.In total, 40 substances from metabolome analysis were tested in the survival analysis.

FIG 6
FIG 6 Biosynthesis of actinomycin D in the primary form of X. bovienii.(A) Biosynthetic pathway for actinomycin D involved in sepsis caused by EPB invading G. mellonella.(B) The abundance of related substances in fermentation broth metabolomics.(C) Expression of methyltransferase-and NRPS-related genes in the two forms of bacteria.*P < 0.05, ***P < 0.01.