Introduction

Entomopathogenic fungi (EPF) that infect insects via cuticle penetration play critical roles in regulating insect populations1. Representative species such as the ascomycete Metarhizium anisopliae, M. acridum, and Beauveria species have been developed as ecofriendly mycoinsecticides2,3. To boost fungal virulence for cost-effective applications, different strategies have been explored by genetic engineering of EPF species4,5. For example, overexpression of endogenous protease or chitinase genes significantly increased the virulence of either Metarhizium or Beauveria to facilitate fungal penetration of insect cuticles6,7. Transgenic expression of insect peptide hormones such as the caterpillar diuretic hormone increased the virulence of B. bassiana against the wax moth Galleria mellonella by interfering with insect physiological homeostasis8. The use of insect-specific neurotoxin peptides from scorpions or spiders for gene synthesis and engineering Metarhizium could also substantially augment fungal performance by paralyzing insects before their death9,10,11. Genetic engineering of EPF expressing double-strand RNAs or miRNAs targeting insect immune defensive genes also boosted fungal virulence against different insects12,13. Alternative strategy remains to be determined to improve fungal biocontrol potency.

Insect gut microbiotas have been well demonstrated in connection with host development and health14,15. Similar to the protective roles of human skin and plant phyllosphere microbiomes16,17, recent works have indicated that insect cuticular bacteria can mediate defensive colonization resistance against fungal parasites18,19,20,21,22. To counterattack, EPF have evolved strategies such as the secretion of a defensin-like gene by Beauveria bassiana and the production of potent antibiotics by Metarhizium species to suppress host cuticular bacteria23,24. It is implicative therefore that genetic engineering of EPF with a potent antimicrobial peptide (AMP) gene may boost fungal infection performance.

In this study, we selected two antibacterial peptides for experiments, i.e., the artificially designed peptide guavanin 2 and silkworm (Bombyx mori) moricin, which have been reported with high bactericidal activities against both the Gram-positive (G+) and -negative (G−) bacteria25,26. After peptide syntheses and antibacterial assays, moricin peptide was selected for gene synthesis and transgenic transformation of M. robertsii. It was evident that the virulence of engineered strains was significantly increased against four insect species when compared with the wild-type (WT) strain. In addition to suppressing insect cuticular bacteria as designed, the engineered strains more severely disrupted insect gut microbiomes than the parental strain.

Results

Engineering Metarhizium with the moricin gene significantly increases fungal virulence against different insects

We first synthesized each mature peptide for testing their minimum inhibitory concentrations (MICs) against the G+/G− bacteria isolated from different insect cuticles. The results showed that moricin, but not guavanin 2, could efficiently suppress bacterial proliferations more or less similar to ampicillin. Both were non-active against the yeasts Saccharomyces cerevisiae and Candida albicans (Table S1).

The moricin gene was then synthesized to generate the overexpression mutants (OMs) by transformation of the WT strain of Metarhizium robertsii. Two independent mutants, OM2 and OM7, were selected for further experiments based on their higher expression levels of moricin than the other mutants (Fig. S1a). Consistent with the above non-antifungal activity of moricin, there was no phenotypic difference between WT and mutants (Fig. 1a). We next performed topical fungal infections against four insect species using the WT and engineered strains. Relative to the WT strain, both OM2 (log-rank test, p = 0.0009) and OM7 (p = 0.008) killed the spotted-wing pest Drosophila suzukii significantly more quickly than did the WT strain (Fig. 1b). The estimation of the median lethal time (LT50) indicated that both OM2 (LT50 = 132 ± 4.39 h) and OM7 (LT50 = 132 ± 6.81 h) exhibited a 27% faster killing speed than the WT (LT50 = 168 ± 15.26 h). Similarly, both mutants were able to kill the mosquito Culex sinensis (a 43% increase), the desert locust Locusta migratoria (17%), and the fruit fly D. melanogaster (22%) significantly faster (p < 0.01) than the WT strain (Fig. 1c–e). No statistical difference was detected between OM2 and OM7 in killing insects.

Fig. 1: Transgenic expression of moricin substantially increases the virulence of M. robertsii against different insects.
figure 1

a No phenotypic difference was observed between the WT and mutants after growing on potato dextrose agar for 2 weeks. Bar, 1 cm. b–e Overexpression of moricin in M. robertsii significantly enhanced fungal virulence against D. suzukii (b), L. migratoria (c), C. sinensis (d), and D. melanogaster (e). The plotted values are the means ± SEMs. Log-rank tests were conducted to determine the difference between the WT and individual mutants: **p < 0.01; ***p < 0.001; ****p < 0.0001; ns, not significant. Control (CK) insects were treated with 0.05% Tween 20. f Fungal load difference between treatments assayed by qPCR analysis of fungal DNA copies. D. melanogaster females were topically infected by different strains for 108 h before analysis. One-way ANOVA followed by Tukey’s test was conducted to determine the significance of difference among samples with the p values of 0.01 (capital) and 0.05 (lowercase).

Full size image

Our probit estimation of the median lethal concentration (LC50) revealed that both OM2 (LC50 = 5.02 × 106 conidia/ml) and OM7 (LC50 = 5.37 × 106 conidia/ml) required fewer (>35%) spores than the WT did (LC50 = 7.26 × 106 conidia/ml) to cause 50% mortality of fruit flies. Further using D. melanogaster as a target, our quantification of fungal loads unveiled that the mutant strains colonized the flies more quickly (one-way ANOVA, p < 0.01) than the WT strain did at 108 h post-treatment (Fig. 1f). In contrast to the results of the above topical assays, both the injection assays to bypass fruit fly cuticles and topical infection of axenic flies did not result in any difference in survival rates between the WT and mutants (p > 0.05) (Fig. S1b, c).

Engineered strains exhibit a greater ability to outcompete bacteria for germination

The following isolation of cuticular bacteria showed that the mosquitos were dominantly inhabited by Enterococcus bacteria; the locusts were dominated by Pseudomonas; and the spotted-wing drosophila flies were dominated by Paenibacillus and Lactiplantibacillus (Table S2). Our scanning electron microscopy (SEM) observations confirmed the presence of bacterial biofilms on the cuticles of D. suzukii adults and locusts (Fig. 2a, b), which were similar to the cuticular inhabitation of bacteria on D. melanogaster27. Taking together with the isolated bacteria from D. melanogaster27, we performed the germination of fungal spores in the absence or presence of G+ or G− bacteria isolated from different insects. Both WT and mutant spores showed similar germination rates, reaching approximately 80% after 12 h in Luria-Bertani (LB) broth (Figs. 2c, S2). When germinating fungal spores in the presence of either G+ or G− bacterial cells, however, the OM2 and OM7 strains exhibited significantly higher germination rates (p < 0.01) than the WT strain (Figs. 2d, S2). For example, in the presence of the G+ Lactiplantibacillus plantarum cells (an initial concentration at 0.05 OD600), the WT spores only germinated at a rate of 26.7% whereas both OM2 and OM7 had a similar germination rate of approximately 50% at 12 h post-inoculation. In the presence of G− bacteria such as Pseudomonas oryzihabitans from locusts or Elizabethkingia miricola from mosquitos, both OM2 and OM7 also similarly germinated better (p < 0.01) than the WT strain did (Fig. 2d). The dual overlay assays confirmed that the transgenic strains could better curb the proliferation of L. plantarum than the WT strain (Fig. S3a). The data indicated therefore that the engineered strains were better than the WT in suppressing insect cuticular bacteria.

Fig. 2: Transgenic mutants showing the increased antagonistic activity against different bacteria isolated from insect cuticles.
figure 2

a, b SEM images showing the presence of abundant bacterial cells on the body surfaces of D. suzukii (a) and L. migratoria (b). Two panels were randomly selected and shown for D. suzukii abdomen (left) and leg (right) parts and for different L. migratoria abdomen parts. Bar, 2 μm. c No obvious difference between the WT and transgenic strains in spore germination in LB broth without bacterial cells. d Substantial increase of the M. robertsii WT and mutant spore germinations after co-culturing for 12 h in LB broth with different bacteria. Values are the mean ± SD. One-way ANOVA followed by Tukey’s test was conducted and different letters labeled above columns show the difference with the p values of < 0.01 (capital) and < 0.05 (lowercase). Different insects are Dmel D. melanogaster; Lmig L. migratoria; Csin C. sinensis; Dsus D. suzukii. OD600 values are indicated in parenthesis for the final use of bacterial cell amount in each competing assay.

Full size image

Fly ecto- and endo-microbiome dysbioses caused by fungal infections

Next, we used the females of fruit flies to isolate cuticular bacteria and count the number of bacterial colony-forming units (CFUs) after topical fungal infections. As a result, fungal treatments significantly (p < 0.01) reduced cuticular bacterial numbers compared to the mock control, and the engineered strains suppressed the proliferation of fly surface bacteria significantly more (p < 0.01) than the WT strain did (Fig. 3a). We also examined fly gut bacteria by counting CFUs after fungal infections and found that, in contrast, the gut bacterial numbers were significantly increased (p < 0.01) in the treated flies than that of mock control (Figs. 3b, S3b).

Fig. 3: Dysregulation of cuticular and gut microbiomes of fruit flies after fungal challenges.
figure 3

a, b Bacterial number variations on cuticles (a) and in guts (b) of flies after fungal infections. Cuticular bacteria were isolated for CFU counting from D. melanogaster females 16 h post fungal inoculations, and gut bacteria were isolated from female flies 108 h post fungal infections. Mock control insects were treated with 0.05% Tween 20. c Quantitative microbiome analysis showing the decrease of fly cuticular bacterial loads after fungal treatments. d Quantitative microbiome analysis showing the increases of fly gut bacterial loads after fungal treatments. Differential reduction in fly cuticular (e) and gut (f) bacterial diversity after fungal infections. Venn diagram analysis showing the fly cuticular (g) and gut (h) bacterial taxa divergently shared among the treatments. The cuticular bacteria isolated from D. melanogaster females after topical infection for 16 h, and gut bacteria isolated from the flies after being treated for 108 h were used for analyses. There were six replicates for each treatment. For (a, b, e, and f) one-way ANOVA followed by Tukey’s test was conducted to determine the significance of difference among samples with the p values of 0.01 (capital) and 0.05 (lowercase).

Full size image

Considering that OM2 and OM7 were similar to each other in killing insects and outcompeting different bacteria, we further performed quantitative microbiome analyses using WT and OM7 strains. Treatments with WT or OM7 spores resulted in 2.3-fold and 2.8-fold reductions, respectively, in cuticular bacterial numbers compared to the mock control (Fig. 3c). In contrast, the number of fly gut bacteria increased by 40% following infection with the WT strain, while OM7 treatment caused a 5-fold increase in the number of bacteria. In particular, dominant increases of the G− Providencia bacteria were observed following WT (1.7-fold) and OM7 (7.2-fold) infections (Fig. 3d). Consistently, it was interesting to find that moricin was largely non-active against this bacterium (MIC > 128 μg/ml) by using an isolated Providencia strain for inhibition assay. Both the WT and OM7 infections substantially (p < 0.01) reduced the diversities of both cuticular and gut bacteria (Fig. 3e, f). Our Venn diagram analysis revealed that the cuticular bacterial taxa were mostly shared among treatments (Fig. 3g). In contrast, OM7 infection sharply reduced the abundance of gut bacterial species, resulting in fewer shared taxa across treatments (Fig. 3h). Overall, topical fungal infections caused the dysbioses of both the ecto- and endo-microbiomes of insects.

Fungal infections result in immune compromise of fly gut immunity

Drosophila evolved the diptericin genes DptA and DptB to mediate specific resistance against the Providencia and Acetobacter bacteria, respectively28. To corroborate the finding of a Providencia surge in fly guts, we performed a reverse transcription (RT)-quantitative PCR (RT-qPCR) analysis of different AMP genes in fruit-fly fat bodies (FBs) and guts after fungal infections. Largely consistent with the observation, the expression of DptA was not induced (p < 0.001) in the guts of the OM7-treated insects when compared to the WT strain 48 h post-infection (HPI), and its expression became similar in the WT- and OM7-treated fly guts 72 HPI (Fig. 4a). DptB expression in guts was lower (p < 0.05) in the OM7-treated insects than that of the WT-infected flies 48 HPI and then decreased to a similar level between WT and OM7 72 HPI (Fig. 4b). The antifungal drosomycin Drs and baramicin BaraA29 genes increased first but more substantially decreased (p < 0.05) in the guts after OM7 infection 72 HPI (Fig. 4c, d). In FBs, both DptA and DptB were substantially induced 48 HPI when compared to mock control and substantially downregulated 72 HPI (Fig. 4e, f). Otherwise, relative to mock controls, both Drs and BaraA were significantly upregulated (p < 0.01) in fly FBs after infections by both strains for up to 72 h. In comparison, OM7 triggered a stronger (p < 0.05) expression of Drs and BaraA than the WT strain did 72 HPI (Fig. 4g, h).

Fig. 4: RT-qPCR analysis of AMP gene expressions in the guts and fat bodies of fruit flies after topical fungal infections.
figure 4

Expression of DptA (a), DptB (b), Drs (c), and BaraA (d) in fly guts after infections by the WT and OM7 strains of M. robertsii for different times. Expression of DptA (e), DptB (f), Drs (g), and BaraA (h) in fly fat bodies (FBs) after infections by the WT and OM7 strains of M. robertsii for different times. Values are the means ± SDs. There were three replicates for each sample. Pairwise two-tailed Student’s t-test was conducted between WT and OM7: ns, not significant; *p < 0.05; ***p < 0.001.

Full size image

Discussion

In this study, we aimed to increase EPF performance by engineering to express an exogenous AMP gene to suppress insect cuticular defensive microbiotas. As designed, the engineering of M. robertsii with the silkworm moricin gene substantially augmented fungal virulence against different insects during topical infections. No obvious difference was observed between the parental WT and engineered strains during injection of nonaxenic flies or during topical infection of axenic flies. Unexpectedly, our quantitative microbiome data revealed that topical fungal infections sharply dysregulated fly gut microbiomes. In particular, the flies infected by the transgenic strain resulted in a more extensive surge of the pathogenic bacteria Providencia spp. in fly guts than by the WT strain, which might additionally facilitate the engineered strains to kill insects.

Guavanin 2 was designed and derived from a guava glycine-rich peptide by showing potent bactericidal activity against the G+ and G− mammalian pathogens such as Listeria ivanovii and Pseudomonas aeruginosa25. It was unexpected to find that the synthetic peptide had no obvious activity against the bacterial species isolated from insect cuticles, suggesting that this AMP might have specificity against different bacteria. Cationic moricin has a rather broad bactericidal activity30. Both the MIC tests and fungal spore germination assays confirmed its antibacterial activity against different G+/G− bacteria isolated from insects. However, we found that this AMP was non-active against the G− Providencia, which could be due to the specific lipopolysaccharides produced by this bacterial genus31. It has also been found that the Drosophila diptericin and drosocin isoforms have specific activities against alternative bacterial species28,32. We found that in contrast to the altered gut bacterial communities, fly surface bacterial taxa were largely shared after the treatments by the WT and mutant strains. Apart from the function of the bacteriostatic antibiotic helvolic acid produced and accumulated in M. robertsii spores23, the selective antibacterial activity of moricin could be a plausible reason, however, which requires further investigations.

As expected and designed, the natural infection virulence of the engineered strains OM2 and OM7 expressing moricin was significantly increased against different insects. Taken together with the suppression of insect cuticular bacterial loads by topical fungal infections, the data supported the role of insect ectomicrobiotas in defending hosts against fungal parasite infections18,19. In addition to better-suppressing insect ectomicrobiomes, it was unexpected to find that the engineered strain caused a more severe dysbiosis of the infected fly gut microbiomes than the WT strain. The axenic and nonaxenic mosquitos (Anopheles stephensi) topically infected by B. bassiana showed that the former died much slower than those nonaxenic insects, which was largely due to the overgrowth of the opportunistic pathogenic bacterium Serratia marcescens in nonaxenic guts and its translocation into insect hemolymph33. The axenic and nonaxenic bark beetle (Dendroctonus valens) larvae infected by B. bassiana resulted in a similar finding that the latter died much quicker than those axenic insects due to fungal dysregulation of gut microbiota, e.g., the overgrowth of Erwinia bacteria34. Likewise, the dysbiosis of honeybee gut microbiota could be induced by the attack of the microsporidian parasite Nosema ceranae35. It is common therefore that fungal infection of insects can cause dysbiosis of gut microbiotas that further facilitates insect death.

On top of the above understanding, it was intriguing to find that the moricin-engineered strain triggered a more extensive proliferation of bacteria in fly guts than the WT M. robertsii did, especially the overgrowth of Providencia. Providencia spp. are pathogenic to Drosophila and resistant to most host AMPs28,31. In addition to the evident non-effectiveness of moricin to Providencia, we found that the expression of DptA (being able to inhibit Providencia) was not induced in the guts of flies infected by the engineered strain, which would facilitate the surge of Providencia in the OM7-treated fly guts. Providing that EPF like M. robertsii infects and colonize insects by propagation in host body cavities36, moricin expressed by the engineered strains might not be able to enter host guts to inhibit different bacteria. Taken together, the observation would help explain why the engineered strain had a better performance than the WT in killing insects apart from the additive suppressing of cuticular bacteria. Since the engineered strains could be comparably quicker in killing locusts, mosquitos, and spotted-wing drosophila, a similar trend of bacterial overgrowth in guts might also occur in those insects infected by the WT and engineered strains, however, which requires a further investigation.

The study in Drosophila showed that fly gut immune responses against bacteria are regulated by the IMD and JAK-STAT pathways but not the Toll pathway37. Our examination of the selected AMP genes in fly guts and FBs demonstrated that the antifungal genes Drs and BaraA were also upregulated in guts after topical fungal infections, suggesting that the fungus-induced Toll pathway might also contribute to maintaining gut microbiomes. Unlike the expression pattern of antifungal genes in FBs, the antibacterial DptA and DptB were substantially induced in FBs at the early stage of fungal infection (i.e., 48 HPI) and then quickly reduced at 72 HPI. It was reported that the fungus-induced antibacterial AMPs facilitated M. rileyi to outcompete the translocated bacteria in the hemocoel of the cotton bollworm (Helicoverpa armigera) larvae38. In this respect, the early upregulations of DptA and DptB in fly FBs might involve in eliminating hemolymph bacteria. As evidenced at 72 HPI, antifungal genes were downregulated in guts but upregulated in FBs would suggest that fungal infection triggered a compromise in local (intestinal) immunity that can facilitate the dysbiosis of gut microbiomes.

In summary, we report that engineering M. robertsii with a caterpillar antibacterial gene significantly enhanced fungal potency against different insects by disrupting both the cuticular and gut microbiotas. Besides the technical advance, our data highlight that both the ecto- and endo-microbiomes are functionally essential in protecting insect hosts against fungal parasite infections.

Methods

Fungal and bacterial strains

The wild-type ARSEF 2575 and its derived mutant strains of M. robertsii36 were maintained on potato dextrose agar (PDA; BD Difco, Franklin Lakes, NJ) at 25 °C. The fungi were also cultured in Sabouraud dextrose broth (SDB; BD Difco) for different experiments. Other fungi like Saccharomyces cerevisiae AH109 and Candida albicans SC5314 were cultured in YPD (yeast extract, 10 g/l; peptone, 20 g/l, and dextrose, 20 g/l) broth for MIC assays. Bacterial species isolated from the body surface of C. sinensis, L. migratoria, and D. suzukii (Table S2) and those previously isolated from D. melanogaster27 were subjected to different assays. The isolated bacterial strains were cultured in LB broth at 37 °C or in de Man, Rogosa, and Sharpe (MRS) broth (Oxoid, Hampshire, UK) at 30 °C.

Microscopic examination of insect cuticular bacteria

We observed the microbes on the body surfaces of D. suzukii and L. migratoria using a Field-Emission Scanning Electron Microscope (Merlin Compact VP, Zeiss, Köln, Germany)39. The adults of D. suzukii 6 days post molting and the 3rd instar nymphs of L. migratoria 2 days post molting were freeze-killed and subjected to SEM analysis as described before27.

Peptide synthesis and MIC assay

The mature peptides of guavanin 2 (RQYMRQIEQALRYGYRISRR)25 and silkworm moricin (AKIPIKAIKTVGKAVGKGLRAINIASTANDVFNFLKPKKRKH)26 were synthesized by a company (Sangon Biotech, Shanghai, China) and dissolved in sterile phosphate buffer as stock solutions (128 μg/ml). MIC assays were performed as described before by serial dilutions in LB or MRS media39. The fresh bacterial cells were suspended in media at the concentration of 0.01 OD600 (~106 CFUs/ml). Equal volume (50 μl each) of different concentrations of AMPs and bacterial suspensions were mixed in 96-well plates. The plates were incubated at 30 °C in a rotary shaker at 200 rpm for 18 h to determine the MICs of moricin and guavanin 2 against each selected bacterium and yeast. The wells containing blank media were included as controls.

Peptide gene synthesis and transgenic expression in M. robertsii

Since silkworm moricin has better antibacterial activities than guavanin 2 against both G+ and G− bacteria, the former was synthesized (Sangon Biotech) by fusion with the Mcl1 mRNA 5′-untranslated region and its signal peptide that we used before10. The product was amplified with the primers TransMorF/R (Table S3), purified, and cloned into the binary vector pDHt-Bar-tef 40 for the transformation of the M. robertsii WT strain. The drug-resistant colonies were subsequently transferred to new PDA plates for single spore isolation and verification via PCR. To screen the high-expression transformants, we prepared the cDNA samples after growing the isolates in SDB for 36 h for RNA isolations. RT-qPCR was conducted with the ReverTra Ace quantitative PCR RT master mix (Toyobo, Japan). The β-tubulin gene of M. robertsii was used as a reference41. Two independent mutants (OM2 and OM7) were selected for further analysis.

Insect survival assays

Following the obtaining of the Metarhizium transgenic mutants, four insect species were then used for comparative virulence assays of the WT and OM mutants, including D. suzukii (male adults), the mosquito C. sinensis (female adults), L. migratoria (the 3rd instar nymphs) and D. melanogaster W1118 (female adults). After trial experiments, different concentrations of spore suspensions were prepared in 0.05% Tween 20 for topical infection of D. suzukii (3 × 106 conidia/ml) and D. melanogaster (5 × 106 conidia/ml) by immersion for 30 s. Spraying spore suspension was used to treat C. sinensis (3 × 107 conidia/ml) and L. migratoria (5 × 106 conidia/ml). The females of D. melanogaster were also used for injection assays (50 nl per insect; 1 × 107 conidia/ml) using a microinjector (Nanoject III; Drummond, Broomall, PA). The axenic D. melanogaster adults were also prepared for topical infections27. Thus, the serial concentrations of the WT and mutant spore suspensions, i.e., 5 × 105, 1 × 106, 5 × 106, 1 × 107, and 5 × 107 conidia/ml, were prepared for topical infection of D. melanogaster (6 days post eclosion) to determine the LC50 values of the WT and mutant strains. Except for the injection assays, the topically treated insects were maintained under a high humidity condition [relative humidity (RH) > 95%] for 16 h post treatments42. The data were analyzed by probit estimation of LC50 values. Insect survival was recorded every 12 h. At least 60 insects were used for each treatment, and the experiments were repeated twice. Representative results from one experimental batch were used.

After topical infection of D. melanogaster, the flies were collected (30 flies per replicate) at 108 h post-inoculation to determine fungal loads. The samples were homogenized in liquid nitrogen for DNA extraction using a DNeasy Blood & Tissue kit (QIAGEN). Quantitative PCR analysis was conducted using the primers for amplification of the Rpl32 gene of M. robertsii by referring to the Rpl32 gene of D. melanogaster43. There were six replicates for each sample.

Antagonistic activity assays against insect cuticular bacteria

To corroborate the virulence increase of transgenic fungus, we performed the antagonistic activity assays of the WT and mutant strains of M. robertsii in competition with different bacteria. In addition to the previously isolated cuticular bacteria from D. melanogaster27, body-surface bacteria were also isolated from the adults of D. suzukii and C. sinensis, and from the 3rd instar nymph of L. migratoria by washing with sterile phosphate buffer solution (PBS) and plating. Individual colonies were transferred for PCR amplification of the bacterial 16S rRNA genes using the universal primers 27F/1492R (Table S3). Different G+ or G− bacterial species were then randomly selected to examine the ability of the WT and mutants of M. robertsii to counteract bacterial inhibitions.

The fresh bacterial cells were collected from the overnight LB cultures, washed twice with sterile water, and re-suspended in LB. Fungal spores were collected from the 2-week-old PDA plates and suspended in 0.05% Tween 20. After trial experiments, fungal spores (at a final concentration of 5 × 106 conidia ml−1) were co-inoculated in the LB medium (10 ml) with the cells of different bacteria (Acetobacter persici, Lactiplantibacillus plantarum, Pseudomonas oryzihabitans, Paenibacillus glucanolyticus, Enterococcus thailandicus, and Elizabethkingia miricola each at a final OD600 = 0.05; Lactobacillus brevis and Glutamicibacter arilaitensis at a final OD600 = 0.01). The mixed samples were incubated in a rotary shaker at 28 °C and 200 rpm for 12 h to determine the spore germination rates39. There were six replicates for each sample. We also performed dual-culture overlay experiments using agar strips (2 × 0.5 cm each) inoculated with L. plantarum cells and transferred to a new plate (two strips per plate). Each plate was subsequently overlaid with 10 ml malt extract soft agar (BD Difco) containing the WT or mutant spores at a final concentration of 5 × 106 conidia/ml. The plates were incubated at 30 °C for 3 days to determine the size of the bacterial inhibition zone against Metarhizium growth27.

Cuticle and gut microbiome analysis

To compare the effect of moricin transgenesis on the fungal manipulation of insect ectomicrobiome structures, we used D. melanogaster females (6 days post eclosion) for topical infection with the WT and mutant spore suspensions (each containing 5 × 106 conidia/ml in 0.05% Tween 20). Treatment of flies with 0.05% Tween 20 was included as a mock control. After being maintained at an RH > 95% for 16 h, the files (10 files per sample) were anesthetized with CO2 and washed with sterile PBS (pH, 7.4). The wash buffers were individually diluted, plated on LB agar, and incubated at 37 °C for 2 days for counting the bacterial CFUs. The guts of the flies topically treated above after 108 h were dissected in sterile PBS and collected (5 guts per sample) in 1.5 ml microtubes containing sterile microbeads and 50 μl PBS. The samples were homogenized at 40 Hz for 30 s, and diluted for plating on LB agars for CFU counting.

The collected cuticular and gut bacteria were also concentrated by centrifugation at 12,000 rpm and 4 °C for 20 min for quantitative microbiome analysis27. To this end, PCR products were obtained from bacterial cells using the 16S universal primers 515F/806R (Table S3). After quality checks, the products were purified and used to generate libraries for amplicon sequencing (HiSeq 2500, Illumina, San Diego, CA) by a service company (Biozeron; Shanghai, China). Sequencing reads were normalized with the program Mothur (ver. 1.45.3)44, and clustered into individual operational taxonomic units (OTUs) with a cutoff value of 97% identity. The α-diversity of the Shannon H index was estimated based on the detected OTU taxa with the program Mothur. There were six replicates for each treatment.

Quantitative analysis of AMP gene expression

We performed the RT-qPCR analysis of different AMP genes in fruit-fly FBs and guts after fungal infection for different durations. Thus, the females of D. melanogaster were collected 6 days post molting for topical infections with the spore suspensions of WT and OM7 (5 × 106 conidia/ml each in 0.05% Tween 20). After treatment for 48 and 72 h, the flies were collected (10 each) to dissect FBs and guts for RNA extractions with TRIzol reagent (Thermo Fisher Scientific, Waltham, MA), respectively40. The flies newly treated with 0.05% Tween were used as a mock control. Four AMP genes were selected for examinations including the antifungal drosomycin Drs and baramicin BaraA29,45 and the antibacterial diptericins DptA and DptB28. There were three replicates for each sample and the experiments were repeated twice.

Statistics and reproducibility

Different statistical tests were conducted for different experiments. The log-rank test of difference in insect survival was performed with the program GraphPad Prism (ver. 10.1.1). One-way ANOVA analysis was conducted with the Turkey test using GraphPad Prism to determine the significance of differences in CFU numbers, α-diversity index, and fungal spore germinations between treatments or samples. Two-tailed Student’s t-tests were performed to determine the difference in AMP expression in fly FBs or guts between WT and OM7 after the same treatment duration. Sample sizes, animal numbers, replicate and or repeat numbers are indicated in relevant experiments.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.