Ectropis grisescens eggs were obtained from stock cultures from the State Key Laboratory of Tea Plant Biology and Utilization, Anhui Agricultural University, Hefei, China (31.86°^°N, 117.27°^°E). The collected larvae were reared on tea leaves in transparent boxes in a controlled climate room (22°C ± 1°C; relative humidity 75 ± 10%; 16 h light:8 h dark photoperiod). The tea leaves were cut from branches of tea plants using scissors and inserted into floral foam for storage (Zhang et al., 2021).

Larvae were reared on tea leaves until they reached the fifth instar. Then, 250 healthy 5th-instar larvae were selected for starvation treatment for 0, 4, 8, 12, or 24 h.

Ten 5th-instar larvae were pooled to into biological replicates and five biological replicates were established per group. Larvae were surface-sterilized by dipping in 75% ethanol for 15 s and then rinsing twice with sterile water for 15 s each time. Dissecting scissors were used to cut laterally behind the head capsule, and the gut was removed from the cuticle with larval forceps. The whole gut, including contents, was collected and placed in a 2.0 ml microcentrifuge tube for DNA extraction.

Total genomic DNA was extracted from samples using a QIAamp DNA Stool Mini Kit (Qiagen, Hilden, Germany) (Mirsepasi et al., 2014). The quality of the DNA was assessed using electrophoresis in 1.2% (w/v) agarose gel prior to amplification and sequencing.

Gut bacteria were analyzed by sequencing the V3-V4 region of the 16S ribosomal RNA gene (16S rRNA) using the Illumina NovaSeq platform (Illumina, San Diego, CA, United States). Genomic DNA samples were subjected to PCR for amplification of the V3-V4 regions of the 16S rRNA using the universal primers 338F (5′-ACTCCTACGGGAGGCAGCAG-3′) and 806R (5′-GGACTACHVGGGTWTCTAAT-3′). All PCR reactions consisted of 15 μl of Phusion R High-Fidelity PCR Master Mix (New England Biolabs, Beverly, MA, United States), 2 μM forward and reverse primers, and 10 ng of template DNA. The thermal cycling conditions were as follows: 98°C for 1 min followed by 30 cycles at 98°C for 10 s, 50°C for 30 s, 72°C for 30 s, and 72°C for 5 min. The mixture of PCR products was then purified with a Gel Extraction Kit (Qiagen). After PCR amplification, the samples were sequenced on the Illumina NovaSeq platform and 250-bp paired-end reads were generated. All sequences are available as SRA files at the National Center for Biotechnology Information Sequence Read Archive database (NCBI-SRA) under bioProject PRJNA720281 (SRA accession numbers: SAMN18644841).

The sequences were analyzed using the QIIME software package and were used to compare the relative abundances of bacterial taxa (Caporaso et al., 2010b). Operational taxonomic units (OTUs) were assigned with a 97% similarity cutoff using UCLUST version 1.2.22 (Edgar, 2010). The representative sequence (the sequence with the highest relative abundance) for each OTU was used to build the overall OTU table. The taxonomic classifications of each microbial OTU were assigned using Ribosomal Database Project classifier PyNast and SILVA and UNITE as bacterial 16S rRNA databases (Caporaso et al., 2010). Abundances of OTUs were normalized using a standard value for the sample with the fewest sequences. Subsequent analyses of alpha diversity were performed using this normalized output.

The alpha- and beta-diversity indices were calculated using QIIME V1.7.0. The alpha diversity of the gut bacteria complexes was calculated as two diversity indices (Shannon-Wiener and Simpson’s) and two richness estimators (ACE and Chao 1 index). Rarefaction curves were used to verify the quality and depth of sampling. Principal coordinate analysis (PCoA) with weighted and unweighted UniFrac distance metrics was used to detect differences among microbial community structures (Lozupone and Knight, 2005).

Fifty 4th-instar larvae were collected and divided into five groups. Each group was starved for 0, 4, 8, 12, or 24 h, and then immediately soaked in 70% ethanol for 3 min to remove surface bacteria. The dissected guts of the larvae were divided among five grinders with 1 ml of sterile water each. The homogenate was added to 100 ml enrichment medium (Yeast extract 0.20 g/L, NaCl 0.50 g/L, Na₂HPO₄ 3.50 g/L, KH₂PO₄ 1.50 g/L, MgSO₄⋅7H₂O 0.50 g/L, Olive oil 10 ml/L) and then incubated at 200 rpm overnight at 37°C. Twenty-four hours later, 1 ml of the enriched liquid was incubated in new enrichment medium for a second round of enrichment. This step was repeated twice. Finally, 100 μl of enriched liquid was coated on a Rhodamine B tablet (Jette and Ziomek, 1994) and cultured at 30°C for 3 days. The grown colonies were purified three times. The transparent circles from cultured strains were observed by irradiating plates with UV light at 350 nm (Kouker and Jaeger, 1987), and the diameter of each fluorescence circle and colony was measured using Vernier calipers. Strains with certain fat degradation ability were selected based on their H/C value, which represent the ratio of hydrolysis circle diameter (H) to colony diameter (C).

Lipase producing bacteria isolated from E. grisescens were identified using 16S rDNA gene sequencing and a series of physiological and biochemical tests. A phylogenetic tree of those 16S rDNA sequences was constructed using MAGA 7.0.

The isolated strain, EG-Q3, was incubated in LB medium overnight at 200 rpm and 37°C. The bacterial suspension was then centrifuged to remove the LB medium. The pellet of bacteria cells was then washed with ddH₂O to further remove residual medium. The bacterial cells were diluted with ddH₂O to a concentration of OD₆₀₀ = 1.0. Ten test tubes containing 5 ml MSM medium (NaCl 1.00 g/L, (NH₄)₂SO₄ 1.00 g/L, K₂HPO₄ 1.50 g/L, KH₂PO₄ 0.50 g/L, MgSO₄⋅7H₂O 0.50 g/L, pH 7.0–7.5) were divided into three experimental groups (i, ii, and iii) and a blank control. Each experiment was replicated three times. Fat bodies dissected from five E. grisescens larvae were mashed in a beaker containing 20 ml ddH₂O. Then, 1 ml fat body homogenate was aspirated into MSM mediums of groups i and iii using pipettes. The above EG-Q3 suspension (OD₆₀₀ = 1.0) was inoculated into sterilized MSM mediums of groups ii and iii at 4% inoculum volume. All of the test tubes were incubated for 3 days at 200 rpm and 37°C.

Bacterial suspension of EG-Q3 (OD₆₀₀ = 1.0) was applied to the surface of tea leaves evenly with a brush and allowed to air dry. Some tea leaves were treated at this step using sterile water instead of bacteria suspension. This treatment was repeated three times. One hundred and thirty-eight 2nd-instar larvae with similar body weight (mean ± SD; 11.021 ± 0.512 mg) were divided into control (I) and experimental (II) groups, and were placed in disposable Petri dishes with tea leaves and fed for 3 days. The tea leaves were renewed daily. In order to observe the colonization of bacteria, nine larvae were randomly selected from each group and their guts were diluted and applied to MYP medium (the appropriate medium for B. cereus) (Peng et al., 2001). Then, the remaining 120 larvae were subjected to starvation stress. Each treatment contained five biological replicates with ten larvae per replicate. The body mass of larvae from the two groups was measured after 4 and 12 h starvation using an electric microbalance.

The Illumina NovaSeq sequencing of the bacterial 16S rRNA amplicons from E. grisescens with after 0, 4, 8, 12, and 24 h starvation yielded 2,244,492 raw reads in total. After quality filtering and read merging, a total of 1,853,092 high-quality sequences remined. Rarefaction curves clearly demonstrated that the sampling efforts were adequate to fully represent the richness of the gut microbial communities (Figure 1).

The OTU richness (ACE and Chao 1 index) of the gut bacteria did not differ significantly among groups (Figures 2A,B). However, the OTU diversity (Shannon index) of the gut bacteria of larvae exposed to starvation for 24 h (TFh; 1.52 ± 0.27) differed from that of the gut bacteria of larvae exposed to starvation for 8 h (Eh; 2.98 ± 0.68; t = 3.99; df = 8; p < 0.001) and 12 h (Th; 2.71 ± 1.14; t = 2.02; df = 8; p < 0.5). In addition, the OTU diversity (Simpson index) of the gut bacteria of TFh (0.38 ± 0.10) differed from that of Eh (0.65 ± 0.17; t = 2.66; df = 8; p < 0.5) (Figures 2C,D).

The PCoA analysis using distances based on weighted and unweighted UniFrac showed that the gut bacterial communities of E. grisescens larvae were not distinctive (Figures 3A,B).

Three hundred sixty-four OTUs were shared by all groups, and accounted for 86.87, 81.79, 79.30, 80.35, and 81.80% of the total OTUs in CK, Fh, Eh, Th, and TFh, respectively (Figure 4A). The dominant bacterial phyla in each group were Firmicutes (50.89–78.12%), Proteobacteria (12.27–25.01%), and Bacteroidetes (0.38–14.95%) (Figure 4B). The dominant bacterial genera in each group were Enterococcus (64.19–79.90%) and Wolbachia (2.13–13.11%) (Figure 4C). The relative abundance of Bacillus in CK (0.01 ± 0.17%) differed significantly from that in Fh (0.64 ± 0.14%; t = 8.51; df = 8; p < 0.0001), Eh (0.32 ± 0.12%; t = 5.06; df = 8; p < 0.001), and Th (0.16 ± 0.14%; t = 2.01; df = 8; p < 0.05) (Figure 4D).

The gut contents of larvae subjected to starvation for different amounts of time were cultivated on Rhodamine B selective medium, and five potential lipase-producing strains were identified based on colony characteristics (Figure 5A). These strains were named EG-Q1, EG-Q2, EG-Q3, EG-Q4, and EG-Q5.

Isolates of the five strains were inoculated on selective medium again in order to compare their abilities to degrade lipids. Although EG-Q1 and EG-Q2 could be cultured on the selective medium, obvious hydrolysis circles were not observed. The H/C value of EG-Q3 (1.66 ± 0.05) was significantly higher than that of the other strains (Figure 5B). Therefore, EG-Q3 was selected for further in vivo and in vitro tests.

The three isolates producing transparent circles (EG-Q3, EG-Q4, and EG-Q5) were rod shaped Gram-positive bacteria (Supplementary Figure). Through combination with the phylogenetic tree constructed from 16S rDNA sequences of the five strains (Figure 6), EG-Q3 was identified as B. cereus. The genome sequence of the isolated strain was registered in GenBank (the Accession Number: MZ497322.1).

In in vitro tests of fat body degradation (Figure 7), we found that fat bodies floating in suspension were quite obvious and unbroken in group i. In group ii, although EG-Q3 was inoculated into the MSM medium, the turbidity of the medium did not change, and resembled group CK. However, the fat bodies in group iii were broken and had nearly disappeared, and the MSM medium became significantly turbid. This indicated that EG-Q3 can use fat bodies for proliferation.

To further determine whether bacteria EG-Q3 could proliferate in the larval gut, tea leaves coated with EG-Q3 were fed to E. grisescens. After 3 days of supplementation, EG-Q3 could be detected in the guts of larvae (Figure 8). There was no difference in the body weights of the two groups of insects before starvation (Figure 9A). After confirming that EG-Q3 can colonize larval guts, the larvae were starved and weighed regularly. We found that in group I, the body mass of larvae fed with EG-Q3 decreased significantly after 4 h of starvation (from 67.014 ± 5.318 mg to 62.514 ± 5.519 mg), and the fat-lowering ratio after 4 h of starvation was 6.76%, nearly twice that of group II (3.96%) (Figure 9B). With the extension of starvation time, the weight of the larvae further decreased. The fat-lowering ratio increased to 13.85% after 12 h starvation (control group 11.05%).