From phyllosphere to insect cuticles: silkworms gather antifungal bacteria from mulberry leaves to battle fungal parasite attacks

Background Bacterial transfers from plants to insect herbivore guts have been well investigated. However, bacterial exchanges between plant phyllospheres and insect cuticles remain unclear, as does their related biological function. Results Here, we report that the cuticular bacterial loads of silkworm larvae quickly increased after molting and feeding on the white mulberry (Morus alba) leaves. The isolation and examination of silkworm cuticular bacteria identified one bacterium Mammaliicoccus sciuri that could completely inhibit the spore germination of fungal entomopathogens Metarhizium robertsii and Beauveria bassiana. Interestingly, Ma. sciuri was evident originally from mulberry leaves, which could produce a secreted chitinolytic lysozyme (termed Msp1) to damage fungal cell walls. In consistency, the deletion of Msp1 substantially impaired bacterial antifungal activity. Pretreating silkworm larvae with Ma. sciuri cells followed by fungal topical infections revealed that this bacterium could help defend silkworms against fungal infections. Unsurprisingly, the protective efficacy of ΔMsp1 was considerably reduced when compared with that of wild-type bacterium. Administration of bacterium-treated diets had no negative effect on silkworm development; instead, bacterial supplementation could protect the artificial diet from Aspergillus contamination. Conclusions The results of this study evidence that the cross-kingdom transfer of bacteria from plant phyllospheres to insect herbivore cuticles can help protect insects against fungal parasite attacks. Video Abstract Supplementary Information The online version contains supplementary material available at 10.1186/s40168-024-01764-6.


Background
Many insects are herbivores that feed on or inside plants.The cross-kingdom transfers or exchanges of bacteria from plant phyllospheres to insects or vice versa have been reported [1][2][3][4].The plant phyllosphere bacteria ingested by insects can colonize insect guts to facilitate insect detoxification, nutrient acquisition, and/ or survival [3][4][5].In addition to the gut microbiotas, the ectomicrobiotas inhabiting animal skins/cuticles or plant phyllosphere have been unveiled with different features and biological functions, especially the defensive roles against different diseases [6][7][8].For example, the protective roles of leaf microbiotas against bacterial or fungal pathogens have been evident in different plants such as Arabidopsis thaliana, rice, and citrus [9][10][11].The microbes inhabiting insect cuticles can protect hosts from the topical infection of entomopathogenic fungi (EPF) like Metarhizium anisopliae and Beauveria bassiana [12][13][14].Irrespective of their functional importance, plant or animal ectomicrobiome assembly is still elusive.The use of Drosophila melanogaster as a model for investigations has revealed that the ectomicrobiotas of adult flies were largely assembled from their fecal bacteria when insects were reared in vials to eat and excrete [15].Metamorphic insects frequently molt, indicating that their cuticular microbiotas have to be quickly re-assembled after ecdysis.The beetle (Lagria villosa) larvae contain unusual pouch structures in cuticles that can house ectosymbionts for seeding bacterial cells after ecdysis [16].The dynamic and function of insect cuticular microbiotas are still largely unclear, especially for those herbivorous insects.
The silkworm Bombyx mori is an important domesticated insect for the sericulture industry and a model animal for the research of caterpillar developments and life sciences [17,18].This oligophagous insect only feeds on mulberry (Morus alba) leaves and the change in diets, especially the use of artificial diet, could substantially change the gut microbiotas, growth, and physiology of silkworm larvae [19][20][21][22][23]. Silkworm mass rearing frequently encounters the threat of fungal, bacterial, virus, and microsporidian diseases [18].Different bacteria isolated from silkworm gut have been found with probiotic potentials to improve insect resistance against diseases or toxic compounds including insecticides [24][25][26][27].In contrast to the per os infection by pathogenic bacteria and viruses, fungal parasites infect insects via penetration of cuticles [28,29].Insect cuticular bacteria can thus mediate colonization resistance against EPF such as M. robertsii and B. bassiana by inhibiting fungal spore germination and or infection structure differentiation [6,30].The ectomicrobiotas of silkworm larvae are unclear and it is also unknown whether its cuticle bacteria could help improve insect survival against fungal parasite infections.
Co-existing microorganisms commonly engage in cross-inhibition and competition for resources by employing different strategies directed at each other.For example, to counteract the resistance of insect cuticular bacteria, B. bassiana secrets a defensin-like antibacterial peptide that can kill diverse bacteria, especially Grampositive bacteria [31].M. robertsii, however, produces different potent antibiotic compounds to combat different bacteria to facilitate fungal infection of insects or growth in diverse environments [30,32].Likewise, different bacteria can produce diverse antifungal metabolites or enzymes (e.g., lysozymes) [33].Thus, species like Bacillus spp.have long been used as biocontrol agents against plant diseases, especially fungal pathogens [34,35].Similar to the uses for human health and the food industry, beneficial bacteria have also been explored as probiotics for promoting health in mass-reared insects [22,36].Until this study, it is unclear whether there are any probiotic bacteria that protect beneficial insects like silkworms against fungal pathogens entering from cuticles.
In this study, we investigated the age-associated ectomicrobiomes of silkworm larvae and found a quick increase in bacterial loads on larval cuticles after molting.One of the dominant bacteria Mammaliicoccus sciuri (Ma.sciuri, previously Staphylococcus sciuri) [37] isolated from silkworm cuticles could completely inhibit the spore germination of fungal pathogens M. robertsii and B. bassiana largely due to its secretion of an antifungal chitinolytic lysozyme.Interestingly, we found that Ma. sciuri originates from mulberry leaves that can be gathered by silkworms as an ecto-and endosymbiont to protect the caterpillars from fungal parasite infections.

Fungal and bacterial strains
The wild-type strains of M. robertsii ARSEF 2575, B. bassiana ARSEF 2860, Aspergillus oryzae RIB 40, and A. flavus NRRL 3357 were maintained on potato dextrose agar (PDA; BD Difco) at 25 ℃ for 2 weeks for harvesting conidial spores.Fungi were also grown in Sabouraud dextrose broth (SDB, BD Difco) for different experiments.The isolated Ma. sciuri strain CGMCC 28001 was grown on a Luria-Bertani (LB) plate at 25 ℃ or 30 ℃.The Escherichia coli strains BL21 and pLysS were maintained on LB plates and used for protein expression analysis.

Silkworm rearing and surface bacterial isolation
The Nistari strain of B. mori larvae was conventionally reared (CR) with mulberry leaves at 25 ℃ and a relative humidity of 50%.Different instars/ages of larvae were then used for isolation of cuticular bacteria, including the end of the 4th instar larvae (L4E; stopping eating just before molting), and the newly molted (L5D0), three-(L5D3), and 6-day-old (L5D6) 5th instar larvae.To determine the resident or transient relationship of some bacteria, we also performed the diet switch experiments.Thus, silkworm larvae were fed with mulberry leaves to the end of the 4th instar, and the newly molted 5th instar insects were then fed with the bacterium-free Silkmate 2S fodder (Nosan Co., Yokohama, Japan) till the insect pupated.The collected insects were anesthetized on ice for 1 h, and three insects were used as a replicate by washing and vortexing in 50 ml phosphate buffer solution (PBS; pH, 7.0) for 30 s.The washing buffers were diluted properly (500 × for L4E and L5D6 samples; 100 × for L5D3 sample; no dilution for L5D0 sample) for inoculation of LB plates, Marine agar (Biofeng), and GYC plates (glucose, 50 g/l; yeast extract, 10 g/l; 0.5% CaCO 3 , 5 g/l; agar, 15 g/l) for 2 days at 30 ℃ to count bacterial colony forming units (CFUs), which were converted to the unit of CFUs per insect [31].The antibiotic specifically inhibiting the Gram-positive (G + ; Nafcillin, Topscience, Shanghai) and Gram-negarive (G − ; Aztreonam, Qisong Biotech, Beijing) bacteria were added to LB medium at a final concentration of 50 μg/ml to determine the numbers of the G + /G − bacterial CFUs [30], respectively.There were 10 replicates for each age of insects, and a two-tailed Student's t-test was conducted to compare the difference in CFU numbers between different ages of insects using the program GraphPad (ver.10.1.1).
Individual colonies formed on LB plates were selected based on their phenotypes and transferred to new plates.Each bacterium was labeled and used for colony PCR using the primers 27F and 1492R1 (Table S1) [15].PCR products were purified for Sanger sequencing and the obtained sequences were used for Blastn analysis to identify bacterial species.

Microbiome analysis
The washing buffers obtained above were concentrated to 200 μl each by centrifugation at 12,000 rpm and 4 ℃ for 20 min.Each sample containing cuticular bacteria was treated with an equal volume of lysis buffer (Mouse Tissue Direct PCR Kit, Tiangen Biotech, Beijing) for 30 min at 65℃ and then 5 min at 95 ℃ [15].The samples were then used as individual templates for PCR amplification of bacterial 16S rRNA genes with the universal primers 515F and 806R (Table S1) [25].The purified products were then used to generate the amplicon libraries for sequencing (PE 2 × 250 bp; Illumina HiSeq 2500) at the service company Biozeron (Shanghai).Sequencing reads were normalized and clustered into individual operational taxonomic units (OTUs) at a cutoff of 97% identity [15].The α-diversities of Shannon and Simpson indices were estimated based on the detected OTUs.

Bacterial inhibition assays
The species of silkworm cuticular bacteria obtained above (Table S2) were used to test their inhibition effects on spore germination of M. robertsii and B. bassiana [31].Each bacterial species was grown in the LB broth at 30 ℃ and 220 rpm overnight, and bacterial cells were concentrated by centrifugation.Bacterial cells (at a final OD600 = 0.01) and fungal spores (at the final concentration of 5 × 10 6 conidia/ml) were mixed in 2 ml LB, and samples were transferred to individual Petri dishes (60 mm in diameter) and incubated at 30 ℃ for 16 h to determine the germination of B. bassiana and for 12 h for estimation of M. robertsii spore germinations.There were ten replicates for each bacterial species, and at least 100 fungal spores were counted for each replicate under microscopy.
To further test the inhibition effect of the selected Ma. sciuri strain, we conducted the growth inhibition assays on a solid LB medium.Thus, the sterile filter-paper strips (3 × 0.5 cm) were cut and transferred upside down to the right side of the LB plates (90 mm in diameter).The cells of Ma. sciuri (20 μl; OD 600 = 5) were inoculated on each strip, and fungal spores (10 μl; 5 × 10 6 conidia/ml) were incubated on the left side of the same LB plates.The plates were incubated at 30 °C and photographed 8 and 16 days post-inoculations.
We analyzed and designed the specific primers V4F and V5R for targeting the 16S rDNA V4-V5 region of Ma. sciuri (Table S1).Primer specificity was verified by including the other isolated bacteria, such as Glutamicibacter mishrai (Actinomycetota, G +), Enterobacter asburiae (Gammaproteobacteria, G −), and Agrobacterium larrymoorei (Alphaproteobacteria, G −).After verification, primers were used to detect the presence or absence of Ma. sciuri on mulberry leaves and Silkmate fodder as well as the insect feces, guts, and cuticles of different age insects.Mulberry leaves were washed with sterile PBS buffer, and the washing buffers were concentrated by centrifugation and used for bacterial detection.The feces (20 each) were randomly collected into 1 ml PBS and suspended for 1 h before being used for bacterial detection.

Antifungal component analysis
After the shaking growth of Ma. sciuri at 25 ℃ and 220 rpm for 12, 24, and 48 h, 1 ml bacterial culture was collected and centrifuged at 5000 rpm for 5 min.The supernatants were filtered through the sterile Millex ® Syringe Filters (Merck) and diluted (1:20) into fresh LB broth to germinate fungal spores of M. robertsii and B. bassiana.Spore germinations in LB broth were included as mock controls.To determine the potential bioactive metabolite or protein/peptide factor in supernatants, we performed the batch growth of bacterial cells to a large volume (400 ml each) at 25 ℃ and 220 rpm for 24 h.After collecting the supernatant by centrifugation, an equal volume of ethyl acetate was used to extract small molecules by following a previous protocol [38].LB broth without bacterial growth was extracted as a mock control.The concentrated extracts were dissolved in 2 ml methanol, which was added into LB broth at a ratio of 1:20 to inhibit the spore germination of both fungal parasites.
We also performed the precipitation of proteins/ peptides from the aliquots of supernatants.Thus, the analytical ammonium sulfate was gradually added into the supernatant samples to a saturated level, and the precipitates were then collected by centrifugation.The samples were re-dissolved/suspended in 2.5 ml sterile PBS buffer and dialyzed in a dialysis tubing (1 kDa, Sigma-Aldrich) in 5 L PBS buffer three times by replacing the buffer every 8 h.The solutions were re-concentrated to 2.5 ml using a centrifugal evaporator (Martin Christ), and the samples were then loaded into individual Amicon ® Ultra 10 K (Merck) devices to separate proteins (≥ 10 kDa) and peptides (≤ 10kDa) by centrifugation.The samples were lyophilized, and the proteins/peptides were added into LB broth at the final concentration of 10 and 100 μg/ml to examine their inhibition of fungal spore germination.The difference in inhibiting spore germinations was compared by two-tailed Student's t test.

Proteomic analysis
To identify the potential antifungal extracellular proteins, we conducted the proteomic analysis of the Ma.sciuri culture filtrates with and without co-culturing with fungi.The spores of M. robertsii and B. bassiana were inoculated in SDB and incubated at 25 ℃ and 220 rpm for 36 h.Mycelia were then collected by filtration, washed twice with sterile water, and used for co-culturing by sealing in dialysis tubes [39].After growing Ma.sciuri in 400 ml LB broth at 25 ℃ and 220 rpm for 24 h, the dialysis tube containing 3.5 g of the fresh mycelia of M. robertsii or B. bassiana were added into the bacterial cultures for incubation for another 24 h.The Ma. sciuri cultures without fungi were incubated and included as controls.There were three replicates for each treatment.After incubations, fungal mycelial bags were carefully removed and the samples were centrifuged at 5000 rpm for 5 min.The supernatants were then used for protein precipitation with ammonium sulfate as described above.After extensive dialysis, the samples were lyophilized and used for proteomic analysis using the Q-Exactive mass spectrometer (Thermo Fisher Scientific) by the service company Omicsolution (Shanghai, China).A protein was considered identifiable with at least one unique peptide being detected in at least two replicates of each sample.Protein intensities were estimated by normalizing the spectral index for each protein [40].The intensity data were processed by log 10 conversion and used for heat mapping analysis with GraphPad to show the differential expression of proteins among three treatments.

Protein expression and antifungal activity assays
After proteomic analysis, three bacterial proteins showed upregulations in Ma. sciuri after being co-cultured with fungi and potential antifungal activities were expressed for activity assays, including WP_088592153 (termed Msp1), WP_088592189 (Msp2), and WP_088592490 (MsP3).The coding region of each protein gene was amplified using the 2 × Phanta Flash Master MixDye Plus kit (Vazyme) and different primer pairs (Table S1).After trial analysis, the purified products of Msp1 and Msp3 were cloned into the pET-28b plasmid by fusion with an MBP (maltose binding protein)-His (6 ×) tag at the C-termini, and Msp2 was cloned into the pET-28b vector by fusion with a His (6 ×) tag at the C-terminus.The Msp1 and Msp3 expression vectors were transformed into the E. coli BL21 strain while the Msp2 plasmid was transformed into the E. coli pLysS cells.E. coli cells were grown in LB and induced with 0.1 mM isopropyl β-D-1thiogalactopyranoside (IPTG, Sigma-Aldrich) at 16 °C overnight.The fusion proteins were purified using the Ni-NTA beads (Smart-Lifescience).The MBP-His tag within the Msp1-MBP and Msp3-MBP fusion proteins was cleaved using a TEV protease.After cleavage, the product was purified again using the Ni-NTA beads, and the eluents were collected and dialyzed with a Tris buffer (pH = 8) to remove salts.The purified proteins were checked using a precast Omni-Easy ™ One-Step PAGE Gel Kit (15%, Epizyme) before use (10 and 100 µg/ml in LB) to examine their activities to inhibit M. robertsii spore germination.

Chitinase activity assay
To determine the potential chitinase activity of lysozyme Msp1, we used the purified protein for activity assay using a Chitinase Assay Kit (Macklin).The chitinase from Streptomyces griseus (Sigma-Aldrich, C6137) was used as a positive control [41].In addition, we tested whether Msp1 could hydrolyze colloidal chitin to form the hydrolytic zone in agar plates.Thus, chitin powder (5 g; Sigma-Aldrich) was first added to the concentrated HCl (37%; 60 ml) and stirred with a magnetic rod at 4 °C overnight.The samples were then washed with 95% ethanol twice followed by washing with sterile water before being embedded in water agars (5%, w/v).The plates were punched with holes (7.5 mm in diameter), and 20 μl PBS or Msp1 protein solution (100 μg/ml) were individually inoculated for 12 h.

Cell damage assays
To determine whether Msp1 could damage fungal cells, we germinated the spores of M. robertsii (5 × 10 6 conidia/ ml) in LB with the addition of Msp1 at a final concentration of 10 µg/ml in a Petri dish (60 mm in diameter).The mock group was included without the addition of Msp1.Each group had three replicates.After 12 h, the supernatants were carefully removed with a pipette, and fungal cells were fixed in 4% formaldehyde for 12 h.The samples were then dehydrated for observation using the Field-Emission Scanning Electron Microscope (Merlin Compact VP, Zeiss) [42].
We also performed ninhydrin staining assays after treating Metarhizium spores with Msp1 [43].The supernatants collected above were centrifuged at 12,000 rpm for 30 min, transferred, and added with PBS containing 2% (w/v) ninhydrin.A reference control group was included by only containing Msp1 protein (10 µg/ml).The samples were boiled in a water bath for 15 min, and immediately cooled on ice.Sample absorbance was measured at a wavelength of 570 nm with a spectrophotometer (GENESYS 50 ™ , Thermo Fisher Scientific) [43].

Msp1 gene deletion and antifungal assays
We deleted the intact Msp1 gene in Ma. sciuri using a CRISPR/Cpf1 technique for homologous recombination [44].In brief, the flanking regions of Msp1 were amplified using the primer pairs Msp1UF/Msp1UR and Msp1LF/Msp1LR (Table S1), respectively.The purified products were then fused into the Xho1 restriction site of the pCpfSA vector using the Gibson Assembly Master Mix kit (NEB Biolabs) to generate the plasmid pCpfSA-Msp1.Five CRISPR RNAs (crRNAs) were synthesized by containing different protospacer adjacent motifs (PAMs) (Fig. S1; Table S1).The double-strand crRNAs were individually integrated into pCpfSA-Msp1 in a reaction system containing the Bsal-HF enzyme (NEB Biolabs) and T4 DNA ligase (Vazyme) [44].The obtained plasmids were then used for individual transformation of the competent Ma. sciuri cells by electroporation at 1800-2100 kV (Bio-Rad MicroPulser #1652100).
The drug-resistant colonies were further transferred onto LB plates containing chloramphenicol (5 μg/ml) and verified by colony PCR.The deletion mutants were successfully obtained from crRNAs (Fig. S1) and used at different dosages (at final OD600 = 0.01, 0.05, 0.0025, and 0.00125, respectively) to examine together with the WT strain against the spore germination of M. robertsii.

Protection of silkworm survival against fungal pathogen challenges
After preliminary trials, the newly-molted 5th instar CR silkworm larvae (10-12 h post-ecdysis) were preimmersed in PBS buffer containing Ma. sciuri (OD 600 = 5) for 5 s.The control insects were immersed in PBS buffer without bacterial cells.After 24 h, insects were topically infected with the spore suspensions (5 × 10 7 conidia/ml in 0.05% Tween 20) of either M. robertsii or B. bassiana by immersion for 30 s.Additional insects were immersed in 0.05% Tween 20 without fungal spores for 30 s as controls.Axenic silkworm larvae were also prepared by following a previous protocol [24].In brief, the surface-sterilized silkworm eggs were hatched on and reared with the sterile diet of Silkmate 2S to the 5th instar.Food changings were conducted in a fume hood.Before using for gnotobiotic bioassays, individual larvae were homogenized with a sterile set of mortar and pestle.The samples were inoculated on LB plates to examine whether there was any bacterial growth.In addition, the homogenates were used as a template for PCR (2 × Phanta Flash Master Mix, Vazyme) examination of bacterial 16S rDNA using the conventional primers 518F and 1492R2 (Table S1).The confirmed axenic silkworm larvae were then used for gnotobiotic survival assays with the M. robertsii and B. bassiana spore suspensions similarly as described above.Both the CR and axenic silkworm larvae were also pretreated with the WT and ΔMsp1 of Ma. sciuri (OD600 = 5) for 24 h before immersion with the spore suspensions of M. robertsii.The mortality was recorded every 12 h.There were 50 silkworm larvae used for each group, and those that died accidentally within 24 h after immersion were excluded in further analysis.The experiments were repeated at least three times, and the representative results from the same batch of experiments were shown.The log-rank test was conducted to compare the survival differences between treatments [45].

Determination of Ma. sciuri on silkworm development and food protection
We soaked mulberry leaves in PBS containing the Ma.sciuri cells (OD600 value = 0, 0.01, 0.1, 1, and 5, respectively) for 5 s.After treatments, mulberry leaves were air dried to feed the newly molted 5th instar CR silkworm larvae till the insect was pupated.The sterile Silkmate diet was also mixed with 50 ml (per 100 g) of PBS containing the Ma.sciuris cells at the OD600 values of 0, 0.01, 0.1, 1, and 5, respectively.The fodders were then used to feed the axenic 5th instar larvae before they stopped eating for pupation.Each batch of the treated leaves or Silkmate was used within 2 days.Five days post-insect pupation, individual cocoons, and pupae were weighted to compare the differences between treatments by one-way ANOVA test using the Tukey method with GraphPad.
During the mass rearing of silkworms, artificial feeds can be frequently contained by Aspergillus fungi [20].The species A. oryzae and A. flavus were thus used in the inhibition of spore germination with Ma. sciuri (OD600 = 0.01) for 12 h.In addition, the sterile Silkmate fodder (50 g) was added with an equal volume (10 ml each) of Ma. sciuri (OD600 = 5) and Aspergillus spore suspension (5 × 10 6 conidia/ml).The samples without the addition of Ma. sciuri were included as controls.After transferring (10 g per plate) to the Petri dishes (120 mm in diameter), the samples were incubated at 30 ℃ for 10 days to determine the bacterial inhibition of aspergilli molding.

Quick reassembling ectomicrobiotas after silkworm molting
We examined the dynamics of silkworm cuticular microbiotas by washing the L4E, L5D0, L5D3, and L5D6 larvae that were conventionally reared (CR) with mulberry leaves (Fig. 1a).Plating washing buffer on LB agar and counting bacterial CFUs indicated that silkworm molting could substantially wipe off cuticular bacteria.However, bacterial loads could be quickly and increasingly re-assembled on cuticles after ecdysis (Fig. 1b, c).For example, the average bacterial numbers increased more than 70-fold in 3 days from 3.2 × 10 4 CFUs per L5D3 insect to 2.3 × 10 6 CFUs per L5D6 insect (t-test, P < 0.0001).The bacterial load differences could be similarly evident when plating the washing buffers on the marine agar (Fig. S2a) or GYC agar (Fig. S2b).After adding the G + (nafcillin) or G − (aztreonam) specific antibiotic in LB agar [15], bacterial growth and CFU counting revealed that the culturable G + bacteria largely inhabited silkworm cuticles (Fig. 1d).
The washing buffer samples were also used for the PCR amplification of the bacterial 16S rDNA, and the amplicons were used for the generation of sequencing libraries.The obtained microbiome data revealed that the bacterial OTUs were largely shared among different instars and ages of silkworms (Fig. 1e).Intriguingly, a rather similar number of OTUs was detected on the L5D0 silkworms and other ages of insects.However, the Shannon and Simpson diversity indices of the L5D0 insects were lower (P < 0.01) than those of the other age silkworms (Fig. S2c,d).It was unexpected to find that, except for a few specific OTUs for the L4E insects, the 5th instar larvae did not have any age-specific OTUs, indicating that bacterial number rather than bacterial species increase occurred on silkworm cuticles.The estimation of relative OTU abundance at the genus level indicated that the different ages of silkworm larvae had rather divergent abundance of bacteria (Fig. 1f ).For example, different from the above colony plating assays, the G − Klebsiella bacteria dominated on the L4E insect surfaces whereas the G − Pseudomonas bacteria largely inhabited the 5th instar larvae.Otherwise, the G + Staphylococcus/Mammaliicoccus bacteria were rich in each stage of insects, especially in the L5D6 larvae.

Identification of an antifungal bacterium from silkworm cuticles
While counting the CFUs mentioned above, individual colonies were transferred to new plates, and the 16S rDNAs of 29 randomly selected colonies were sequenced (Table S2).The non-redundant bacteria were then used in the inhibition of spore germination assays against both M. robertsii and B. bassiana by coculturing with the bacterial cells in the LB broth [15].It was found that Ma. sciuri could completely inhibit the spore germination of both fungi (Figs.S3a, b and  S4a).A strain of Serratia ureilytica could completely suppress the germination of B. bassiana spores but not those of M. robertsii (Fig. S3a, b).We further performed the confrontation tests on LB agars between the selected bacteria and two fungi, which indicated again that Ma. sciuri but not the other bacteria could considerably suppress the growth of both M. robertsii and B. bassiana (Fig. S4b).
Previous studies have shown that S. sciuri (now Ma. sciuri) is widely present in soils, plant leaves, and humans as a commensal bacterium [46,47].This bacterium has also been detected in silkworm gut microbiomes [20,21,24].We designed the specific primers for detecting Ma. sciuri (Fig. S5a), and found that this bacterium was present on mulberry leaves used for feeding silkworms.Unsurprisingly, Ma. sciuri was present on the cuticles and in the gut and feces of silkworm larvae fed with mulberry leaves (Fig. S5b).This bacterium was absent in the artificial Silkmate diet, and the fodder-fed silkworms were therefore not inhabited by Ma. sciuri (Fig. S5c).We also performed the diet-switching experiments after feeding insects with mulberry leaves to the end of the 4th instar, the newly molted 5th instar insects were then fed with Silkmate fodder.It was found that Ma. sciuri could still be detected from the feces, guts, and cuticles of the 5th instar silkworms before pupation (Fig. S5d).The data indicated therefore that this plant-derived bacterium could be a resident rather than a transient member on and in silkworms.

Extracellular protein(s) of Ma. sciuri mediates the inhibition of fungal spore germination
We next aimed to determine the antifungal factor(s) of Ma. sciuri by growing the bacterium in LB for 12, 24, and 48 h.The supernatants were then diluted (1:20) with fresh LB broth to test the spore germination of both fungi.It was found that the diluted culture supernatants could completely inhibit fungal spore germination (Fig. 2a, b), indicating that the potent antifungal component(s) was produced and secreted by Ma. sciuri into cultures.We then grew Ma.sciuri to a large volume for metabolite extraction and extracellular protein precipitation.The precipitated proteins were further separated into the aliquots of < 10 kD and > 10 kD using centrifugal filter units.The following experiments revealed that the ethyl acetate extracts could not inhibit the spore germination of both M. robertsii and B. bassiana (Fig. S3c, d).In contrast, the precipitated total proteins, small peptides (< 10 kD), or bigger proteins (> 10 kD) could similarly reduce the spore germination rate of M. robertsii (P < 0.0001) by about 50% and that of B. bassiana (P < 0.0001) by about 40% when compared with the mock control (Fig. 2c, d).The results indicated that the protein(s) instead of small molecules secreted by Ma. sciuri largely mediated the antifungal activity.

A chitinolytic lysozyme produced by Ma. sciuri can damage fungal cells
To further determine the antifungal protein(s), we conducted the parallel growth of Ma. sciuri (Ms) without or with the mycelial cultures of M. robertsii (Ms + Mr) and B. bassiana (Ms + Bb), respectively.The total proteins of each sample were then precipitated, dialyzed, and subjected to mass spectrometry analysis.Different bacterial proteins (25 in total) were detected as well as three Metarhizium proteins from the Ms + Mr sample and ten Beauveria proteins from the Ms + Bb sample (Dataset S1).Differential expression of bacterial proteins was detected among three samples (Fig. 2e).Three proteins, i.e., WP_088592153 [Msp1; signal peptides (SP), 1-27 amino acids (aa)], WP_088592189 (Msp2; SP, 1-25 aa), and WP_088592490 (Msp3; SP, 1-24 aa), were selected for further activity assays based on the features of their expressions in Ma. sciuri pure culture, and upregulation by bacterium when being co-cultured with either fungus as well as the containing of putative antifungal/antibiotic domains.These three proteins each contain a SP region.Msp1 is similar (58% identity) to the lytic transglycosylase SceD of S. aureus [48], which contains a C-terminus lysozyme-like catalytic domain.Both Msp2 and Msp3 are the putative cysteine, histidine-dependent amidohydrolases/peptidase (CHAP) domain-containing proteins (Table S3).CHAP proteins also have lysozyme activities [49].
These three proteins were then heterologously expressed without their SPs in E. coli, and purified for antifungal activity assays (Fig. 2f ).The use of two dosages (10 and 100 μg/ml) of each protein in LB for inhibition assays revealed that Msp1 but not Msp2 and Msp3 could significantly (P < 0.0001) inhibit the germination of the M. robertsii spores when compared with the mock control (Fig. 2g-i).Our further scanning electron microscopy (SEM) analysis revealed that Msp1 treatment could damage the cells of M. robertsii (Fig. 3a).Thus, the leakage of intracellular protein contents was detected by ninhydrin staining assays (Fig. 3b, c).
Considering that the lysozyme domain of lytic transglycosylase shows a chitinase activity [50], we tested the chitinolytic activity of Msp1.It was found that the addition of Msp1 into the colloidal chitin agar could result in the formation of a hydrolytic zone (Fig. 3d).Activity assay confirmed that Msp1 has a chitinase activity, however, which was lower (P < 0.0001) than that of a commercial chitinase purified from Streptomyces griseus (Fig. 3e).

Deletion of Msp1 reduced the antifungal ability of Ma. sciuri
To further verify the Msp1 antifungal activity, we performed the deletion of Msp1 gene in Ma. sciuri using a CRISPR/Cpf1 technique [44], and the independent mutants were verified (Fig. S1).The subsequent uses of the wild-type (WT) and mutant isolates of Ma. sciuri for the inhibition of the M. robertsii spore germination revealed that ΔMsp1 and WT could similarly inhibit spore germination when being co-cultured with fungal spores at a dosage of 0.01 OD600.However, the null mutant showed impaired antifungal inhibition ability at lower dosages compared to the WT treatments (Fig. 3f ).We also used the culture broths of the WT and ΔMsp1 of Ma. sciuri for assaying the inhibition of fungal spore germination.As a result, in contrast to the WT samples, the ΔMsp1 culture broths of different fermentation times could not completely inhibit the germination of the M. robertsii spores until being cultured up to 48 h (Fig. 3g).Taken together, the data confirmed that Msp1 of Ma. sciuri plays an essential role in attacking fungal spores, and additional antifungal factor(s) could be produced by Ma. sciuri to synergistically suppress fungal spore germination.

Improving silkworm survival against fungal infections by Ma. sciuri
We next performed silkworm survival assays with or without the pretreatment of insects using the Ma.sciuri cells for 24 h before the topical infections.It was found that the pretreatment of the CR larvae with bacterial cells could significantly increase insect survival against M. robertsii (χ 2 = 23.7,P < 0.0001) and against B. bassiana (χ 2 = 36.8,P < 0.0001) (Fig. 4a, b).We also generated the axenic silkworm larvae for gnotobiotic survival assays (Fig. S5e-g).It was found that the pretreatment of the germ-free silkworms with Ma. sciuri could also substantially (P < 0.0001) improve silkworm survival against either M. robertsii or B. bassiana infection when compared with those without the pretreatment of bacterial cells (Fig. 4c,d).
After the pretreatment of the CR and axenic silkworms with the WT and ΔMsp1 of Ma. sciuri, comparative survival assays revealed that the deletion of Msp1 could considerably attenuate the bacterial protection ability examined with both CR (P < 0.01) and axenic silkworms (P < 0.05) against M. robertsii (Fig. 4e,f ).However, relative to the insects without pretreatments, the use of the ΔMsp1 cells could still protect the CR (P < 0.05) and axenic (P < 0.01) silkworms against fungal infections.It is noteworthy that the sole treatment of silkworms with the Ma.sciuri cells had no negative effect on the survival of both the CR and axenic silkworms (Fig. 4).

Probiotic effect of Ma. sciuri on protecting silkworm fodder
To further determine the probiotic effect of Ma. sciuri, we used the different dosages of bacterial cells to pretreat mulberry leaves and Silkmate fodder before use for feeding silkworm larvae.It was found that the pupae and cocoon weights of the CR and axenic insects had no obvious difference between the mock control and those reared with the diets pretreated with different amounts of bacterial cells (Fig. S6).
Having found that the artificial diets could be frequently contaminated by aspergilli molds, we tested and found that Ma. sciuri could also completely inhibit the germination of the A. flavus spores and substantially suppress (P < 0.0001) the germination of A. oryzae spores (Fig. 5a).We then tested the inoculation of the Silkmate fodder with Aspergillus spores and found that both Aspergillus species could outgrow on the fodders (Fig. 5b).However, the co-inoculation of fodder with aspergilli spores and Ma.sciuri cells could prevent diet molding by both fungi (Fig. 5c).The data indicate that the use of Ma. sciuri can additionally benefit the protection of silkworm diets against moldy fungi.

Discussion
In this study, we found that the conventionally-reared (CR) silkworm larvae could quickly reassemble their ectomicrobiotas after molting including the gathering of mulberry leaf bacteria.Microbiome analysis revealed that the cuticles of the newly molted and different ages of insects largely shared the OTUs of different bacteria, and the Pseudomonas, Mammaliicoccus, and Staphylococcus bacteria quickly expanded on cuticles when silkworms fed on mulberry leaves.Taken together with the gnotobiotic survival assays, it was interesting to find that the silkworm cuticular bacterium Ma. sciuri that originated from mulberry leaves could confer defense against fungal parasite infections by secreting a chitinolytic antifungal lysozyme.The cross-kingdom transfer of the plant phyllosphere bacteria to herbivore cuticles can thus benefit insects in battling the infection of fungal pathogens (Fig. 5d).
It was unsurprising to find that the bacterial loads sharply dropped on the newly formed cuticles after The newly molted 5th instar larvae were immersed in Ma. sciuri (Ms) cells (OD600 = 5) for 5 s, and after 24 h the insects were topically treated with fungal spore suspensions (5 × 10 7 conidia/ml in 0.05% Tween 20).Log-rank tests were conducted between the survival curves: *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001.Mock control insects were treated with 0.05% Tween 20.The insects only treated with Ma. sciuri (Ms) were also included as controls.The number of insects finally included in analysis is shown in parentheses silkworm molting.However, intriguingly, a roughly equal number of bacteria OTUs could be detected in the L5D0 insects.In contrast to the beetle larvae [16], the pouchlike structure is absent on silkworm larvae for housing bacteria during insect molting.It is possible but remains to be determined that the bacteria hiding on the trachea of silkworm larvae may contribute to the seeding of cuticular bacteria after insect molting [51].Similar to the ingestion of plant bacteria into insect guts [5], the body surfaces of silkworm larvae can be "contaminated" with the mulberry leaf phyllosphere bacteria during insect feeding and crawling.Additional bacteria other than Ma.sciuri could be transferred to silkworm cuticles.It is known that the endo-or ecto-microbiotas of animals, especially the latter, could be contaminated with environmental bacteria [52].In addition to the previous detection of S. sciuri (now Ma. sciuri) in silkworm gut microbiomes [20,24], we found that Ma. sciuri were present on mulberry leaves, silkworm cuticles, guts, and feces.In contrast, Ma. sciuri was absent on and in silkworms after being fed with the bacterium-free fodder.The diet switched from mulberry leaves to an artificial diet without Ma.sciuri for the 5th instar larvae did not lead to the disappearance of this bacterium from silkworm cuticles and guts.Taken together, our data confirmed that silkworms gathered the antifungal Ma. sciuri from mulberry leaves as a symbiont to combat fungal pathogen infection (Fig. 5d).Silkworm has been domesticated for more than 5000 years [53], it is still elusive when silkworms evolved the ability to deploy this bacterium for disease resistance.
Phyllosphere microbiomes play an essential role in protecting plant health by antagonizing pathogenic bacteria and fungi [8,54].Apart from Ma. sciuri, we found that the isolated cuticular bacterium Serratia ureilytica could completely inhibit the germination of B. bassiana spores but not those of M. robertsii.Even the underlying mechanism remains to be determined, it is implicative that there would be additional cuticular bacteria other than Ma.sciuri involved in protecting silkworms against fungal pathogens.In support of this, different G + and G − bacteria isolated from Drosophila adult surfaces have been tested with varied but significantly inhibitory effects on the germination of M. robertsii and B. bassiana spores [15].Likewise, the dominant G − Pseudomonas and Serratia bacteria isolated from the Japanese pine sawyer beetle (Monochamus alternatus) cuticles could completely inhibit the germination of B. bassiana spores [12].The reciprocal evidence has shown that the deletions of the defensin-like antimicrobial peptide gene in B. bassiana and the antibiotic biosynthetic genes in M. robertsii impaired fungal ability to counterattack insect cuticular bacteria and therefore fungal virulence against insects during topical infections [30][31][32].The protective effect of insect ectomicrobiotas against fungal pathogens would therefore be common [6], which might have facilitated the arms-race co-evolution among insect hosts, insect ectomicrobiotas, and fungal pathogens.It remains to be investigated in other insects in terms of the employment of the defensive microbes from phyllospheres by insect herbivores, especially those devastating agriculture pests, against EPF since which may impede the biocontrol efficacy of mycoinsecticides.
Antagonisms commonly occur between microbes while the strategies may vary from either side.To counterattack fungal competition, bacteria can produce antifungal compounds or peptides/enzymes [55,56].We found that a chitinolytic lysozyme Msp1 secreted by Ma. sciuri demonstrated an antifungal activity by damaging fungal cell walls.Since chitin is one of the critical cell wall components of fungi [57], it is thus not surprising to find that Ma. sciuri or its culture broth could similarly inhibit the spore germination of aspergilli fungi.Different lysozymes with chitinolytic activity have also been identified from other bacteria such as Ralstonia and Streptomyces spp., which have the potential to produce N-acetyl chitooligosaccharides from chitin materials [58,59].Otherwise, the antibacterial and or antifungal lysozymes can be produced by different plants and animals for immune defenses against different pathogens, some of which have been used as antibiotic agents in the food industry [60].We examined and found that the chitinase activity of Msp1 was significantly lower than that of the commercial Streptomyces chitinase, which could explain, at least in part, why the deletion of Msp1 did not completely disable the antifungal activity of Ma. sciuri.It was found that a serine protease produced by the biocontrol bacterium Bacillus amyloliquefaciens had a broad spectrum of antifungal activity [61].Serine proteases were also detected in the culture broth of Ma. sciuri.It remains to be determined whether these proteases and or other enzyme(s)/ factors are used by Ma. sciuri to maintain the synergistic antifungal activity.For example, interestingly, it was found that the volatile compounds produced by S. sciuri had an antifungal activity and potential against the strawberry anthracnose fungi [62].
Consistent with the antifungal activity of Ma. sciuri, our fungal infection assays confirmed that the monoassociation of bacterial cells with either the CR or axenic silkworm larvae could significantly improve silkworm survival against the infection of M. robertsii or B. bassiana.Similarly, a lactic bacterium (Lactobacillus paraplantarum) was found with a probiotic potential after feeding to promote silkworm immunity against the pathogenic bacterium Pseudomonas aeruginosa [63].We also found that the pretreatment of mulberry leaves or artificial fodder with Ma. sciuri had no negative effect on insect development, which was consistent with a previous finding that a gut-isolated strain of S. sciuri did not affect silkworm growth [24].Taken together with the probiotic effect of Ma. sciuri on diet protection, our results indicate that the use of this bacterium may benefit the industrial and mass rearing of silkworms using artificial diets.
In conclusion, we report that the deployment of mulberry leaf bacteria to cuticles by silkworms can benefit the insects to combat fungal parasite infections.We also unveil that the antifungal bacterium Ma. sciuri can use an extracellular chitinolytic lysozyme to damage fungal cells.In addition to revealing the cross-kingdom transfer of beneficial bacteria from plants to insects, the beneficial bacterium identified in this study has the potential to be used for safer sericulture.S1.The bacterial species Glutamicibacter mishrai (G.mi), Enterobacter asburiae (E.as), and Agrobacterium larrymoorei (A.la) were included as negative controls.b Verification of the presence of Ma. sciuri on mulberry leaves and leaf-fed silkworms.c Verification of the absence of Ma. sciuri in Silkmate fodder and fodder-fed silkworms.L4S, slough of the 4 th instar larvae.d PCR verification of the symbiotic presence of Ma. sciuri in the 5 th instar silkworms fed with the bacterium-free fodder.The insects were fed with mulberry leaves till the end of the 4 th instar.e Images showing the silkworms conventionally reared (CR) with mulberry leaf and axenically reared with sterile Silkmate fodder.f, g Verification of the obtained axenic silkworm larvae by PCR of bacterial 16S rDNA (f ) and plating of insect homogenates for bacterial colony formation (g).The CR silkworms were used as a positive control.Figure S6.Pretreatment of silkworm diets using the Ma.sciuri cells has no negative effect on insect development.a, b Feeding of the 5 th instar larvae with the mulberry leaves soaked in Ma. sciuri cells (from 0.01 -5 OD600 in sterile BPS buffer) has no obvious negative effect on the pupa (a) and cocoon (b) weight of silkworms.c, d Feeding of the 5 th instar larvae with the artificial Silkmate fodder added with different amount of Ma. sciuri cells has no obvious negative effect on the pupa (c) and cocoon (d) weight of silkworms.Oneway ANOVA analysis was conducted: ns, not significant.Table S1.Primers used in this study.Table S2.Isolation of cuticular bacteria from silkworm larvae.Table S3.Mass spectrometry detection of culture filtrate proteins secreted by Ma. sciuri.
Additional file 2: Dataset S1.LC-MS proteomic analysis of the supernatant proteins isolated from Ma. sciuri (Ms) after co-culturing without or with M. robertsii (Mr) or B. bassiana (Bb).

Fig. 1
Fig. 1 Quick re-assembly of ectomicrobiotas on silkworm cuticles after insect ecdysis.a Phenotype of the different ages of silkworm larvae used in experiments.The insects were conventionally reared with mulberry leaves.Bar, 1 cm.b Colony-forming patterns of the bacteria washed off from the different ages of silkworm larvae.c Comparison of the silkworm cuticular bacterial CFUs formed on LB agars among the different ages of silkworm larvae.Two-tailed Student's t-test was conducted between samples: *, P < 0.05; ***, P < 0.001; ****, P < 0.0001.Ten independent replicates (three insects per replicate) were included for each sample.d The G + bacteria largely inhabiting silkworm larvae body surfaces.LB agars were added with the antibiotic nafcillin for suppressing the G + bacteria whereas the addition of aztreonam for inhibiting the G − bacteria.e Venn diagram analysis showing the overlap of bacterial OTUs among samples.f Variation of the OTU relative abundance at the bacterial genus level among silkworm ectomicrobiotas.L4E, the end of the 4th instar larvae; L5D0, the newly molted 5th instar larvae; L5D3 and L5D6, the third-day and sixth-day old of the 5th instar larvae

Fig. 2
Fig. 2 Inhibition of fungal spore germination by bacterial extracellular proteins.a, b Inhibition of M. robertsii (a) and B. bassiana (b) spore germination by the Ma.sciuri culture filtrates.Ma. sciuri was grown in the LB broth for the indicated times before being centrifuged, and the supernatants were used for germinating fungal spores.Germination of fungal spores in LB was included as mock controls.c,dThe precipitated extracellular proteins of Ma. sciuri could inhibit the M. robertsii (c) and B. bassiana (d) spore germination.TP, total protein; Peptides (< 10 kD), and proteins (> 10 kD) were prepared using centrifugal filter units.e Heat mapping shows the differential expression of the Ma.sciuri extracellular proteins in different treatments.Ms, the proteins extracted from the pure Ma. sciuri culture.Mr + Ms, the proteins extracted from the M. robertsii and Ma.sciuri co-culture; Bb + Ms, the proteins extracted from the B. bassiana and Ma.sciuri co-culture.The protein intensity was obtained from three independent replicates by Log10 conversion.The proteins highlighted in bold were upregulated in challenging with fungi and selected for E. coli expression and antifungal activity assay.f Gel analysis of three expressed proteins.g-i The expressed Msp1 (g) but not Msp2 (h) and Msp3 (i) proteins could significantly inhibit the germination of M. robertsii spores.Panels c, d, g two-tailed Student's t-test was conducted between each sample and mock control: ****, P < 0.0001.ns, not significant.Ten independent replicates were included for each sample

Fig. 3
Fig. 3 Disruption of fungal spores by Msp1 with a chitinase activity.a Disruption of M. robertsii spores by Msp1.The conidia of M. robertsii were incubated with or without Msp1 (at a final concentration of 10 μg/ml) in LB for 12 h before being fixed for SEM observation.Cells with lytic lesions are arrowed.Bar, 5 μm.b Examination of spore leakage after Msp1 treatment by ninhydrin staining.The conidia of M. robertsii (Mr, 5 × 10. 6 conidia/ ml) were suspended in sterile PBS buffer with or without Msp1 (10 μg/ml) for 12 h.The supernatant was then collected and treated with ninhydrin (2%, w/v).c Comparison of A570 absorbance after sample reaction with ninhydrin.d Formation of a chitin hydrolytic zone by Msp1.The water agar was prepared by containing 5% (w/v) colloidal chitin, and 20 μl of Msp1 (100 μg/ml) was loaded for 12 h.e Activity assays showing the chitinase activity of Msp1.The commercial chitinase isolated from S. griseus was used as a positive control.f, g Comparison of the inhibition of M. robertsii spore germination among the different dosage of WT Ma. sciuri (Ms) and ΔMsp1 cells (f), and among LB culture filtrates (g).Panels c, e-g Two-tailed Student's t-test was conducted between samples: ns, not significant; *, P < 0.05; ***, P < 0.001; ****, P < 0.0001

Fig. 5
Fig. 5 Probiotic effect of Ma. sciuri on diet protection and schematic transfer of phyllosphere bacteria to insects.a Differential inhibition of Aspergillus spore germination by Ma. sciuri (Ms).The bacterial cells were added at a dosage of OD600 = 0.01 for 12 h.b Molding of the sterile Silkmate fodders after inoculation of A. flavus and A. oryzae for 10 days.c Pretreatment of the sterile Silkmate fodders with the Ma.sciuri cells inhibits the growth of two Aspergillus fungi.The sterilized fodder was included as a mock control.The photos were taken 10 days post-inoculations.d Schematic transfer of phyllosphere bacteria to herbivorous insect cuticles that can protect insects against fungal parasites.EPF, entomopathogenic fungi; PPF, plant pathogenic fungi; FDS, fungal disease spot; Ms, Ma. sciuri isolated from silkworm surfaces for antifungal activity.a, b Inhibition or non-inhibition of M. robertsii (a, for 12 h) and B. bassiana (b, for 16 h) spore germination by different bacteria isolated from silkworm cuticles.Fungal spores (5 × 10 6 conidia/ml) were germinated in LB with the addition of bacterial cells each at 0.01 OD600.c, d The ethyl acetate extracts of Ma. sciuri have no effect on inhibiting M. robertsii (c) and B. bassiana (d) spore germination.Ma. sciuri was inoculated in LB for 24 h and the cultures were centrifuged, and both the supernatant and bacterial cells were extracted and used for inhibition assays.

Figure S4 .
Inhibition of fungal spore germination and growth by Ma. sciuri.a Microscopic images showing the inhibition or non-inhibition of Metarhizium and Beauveria spore germination by different bacteria.Fungal spores (5 × 10 6 conidia/ml) were co-incubated with bacterial cells (each at a final value of OD600 = 0.01) in LB broth for 12 h (M.robertsii) or 16 h (B.bassiana) prior to imaging.Spore germination in LB was used as a mock control.Ungerminated fungal spores are arrowed.Bar, 5 μm.b Confrontation test of bacterial inhibition or non-inhibition of fungal growth for different times.Fungi were inoculated with 1 μl of spore suspensions (1 × 10 6 conidia/ml each).The bacterial strips were prepared by soaking the filter-paper strips in bacterial cells (OD600 = 5) for 30 sec before inoculation.The strips without bacterial cells were used as mock controls.DPI, day post inoculation.Figure S5.PCR verification of the presence of different bacteria.a Verification of the specific PCR primers for detecting Ma. sciuri (Ma.sc).Different isolates of Ma. sciuri are as shown in Table