Intestinal Bacterial Diversity and Functional Analysis of Three Lepidopteran Corn Ear Worm Larvae

Simple Summary Corn is one of the most important food crops in the world and comprises a large proportion of crops in China. Pests are one of the important factors affecting corn yield. Conogethes punctiferalis, Ostrinia furnacalis and Helicoverpa armigera are three main pests in the ear stages of corn, which significantly affect the yield and quality of corn. The three co-occurring lepidopteran pests at the ear stage occur frequently. Recently, the amount of Conogethes punctiferalis increased, even becoming the most serious pest in the Huang-Huai-Hai summer corn region of China, which is the second-largest corn producing area in China. Gut bacteria play important roles in insect adaptation to various environments. This study aimed to compare the diversity and function of the intestinal bacteria in three co-occurring lepidopteran pests, and to explore the reason for their prevalence. The results might provide insight into the prevalence of corn earworm larvae from the perspective of gut microbiota and function prediction. Abstract Insects, as the most abundant animal group on earth, and their symbionts help their hosts to adapt to various environments. Conogethes punctiferalis, Ostrinia furnacalis and Helicoverpa armigera are three main pests co-occurring in the ear stage of corn, which significantly affect the yield and quality of corn. The purpose of this study was to compare the diversity and function of the intestinal bacteria of the three co-occurring lepidopteran pests, C. punctiferalis, O. furnacalis and H. armigera, and to explore the reason of their prevalence from the microbiota’s view. Our results showed the difference of diversity and abundance of the gut bacteria of three co-occurring lepidopteran pests at the ear stage. Proteobacteria and Firmicutes were the dominant phyla, and the Enterobacteriaceae and Enterococcaceae were the dominant families in the three pests. Compared with the other two pests, Bacteroidetes was found much more in C. punctiferalis. In addition, C. punctiferalis showed more correlation and similarity in bacteria composition with corn endophytic bacteria, as well as had obvious advantages in metabolic, environmental information processing, cellular processes and organic systems function pathways. Our findings may provide insight into the prevalence of corn earworm larvae from the perspective of gut microbiota and function prediction.


Introduction
Corn is the world's most widely grown and highest-yielding food crop. There are six major corn areas in China. The Huang-Huai-Hai summer corn region is China's secondlargest corn-producing area, comprising about 40% of the country's cultivated area and yields [1]. Despite many other factors influencing corn productivity, pests have always been an important component that affect corn production [2]. Due to increased corn planting area, industrialization, improved agricultural systems, global warming and other factors, corn pests have become more common in the 21st century, which has negatively impacted corn yields [3]. In China, there are more than 200 species of corn pests, with more than

DNA Extraction
Twenty healthy 4th to 5th instar larva of 3 different corn earworms were selected for dissection as a sample, and each pest had five replicates. After 24 h of starvation, all individuals were surface-sterilized in 75% ethanol for 2 min, followed by three rinses in sterile water. Midgut dissections were conducted in a clean Petri dish (90 mm in diameter) by an anatomical microscope under a clean bench. Healthy and full corn seeds were selected, soaked in 75% alcohol for 2 min, rinsed with sterile water 3 times, then ground with sterile water and put 1 mL abrasive solution into a 1.5 mL centrifuge tube for later use.
Microbial community genomic DNA was extracted from samples using the OMEGA-D5625-01 Soil DNA Kit according to manufacturer's instructions. The DNA extract was checked on with 1% agarose gel, and DNA concentration and purity were determined with NanoDrop 2000 UV-vis spectrophotometer (Thermo Scientific, Wilmington, NC, USA).

Statistical and Bioinformatics Analysis
In order to get high-quality readings, each sample were spliced through overlap by FLASH (Fast Length Adjustment of Short Read, version 1.2.11), a read pre-processing software, which assembled and merged the paired-end reads from fragments and generated >10 bp overlapped, with a maximum mismatch rate of the overlap region of 0.2 (https://ccb.jhu.edu/software/FLASH/index.shtml, (accessed on 12 August 2021)). Fastp (version 0.19.6, https://github.com/OpenGene/fastp, (accessed on 24 September 2021) was used for quality control of original sequencing to obtain optimized data. The nonrepetitive sequences were extracted from the optimized sequences, which could reduce the computational complexity of the intermediate process. The single sequences without duplication were removed, and operational taxonomic units (OTUs) based on the non-repetitive sequences (excluding the single sequence) were clustered according to 97% similarity by using UPARSE (version 7.1, http://drive5.com/uparse/,(accessed on 27 September 2021). During the clustering process, chimeric sequences were removed, and the representative sequences of OTUs were obtained. After the above steps, the optimized sequence can be obtained for further analysis. All the optimized sequences were mapped to the OTU representative sequence, and the sequences with more than 97% similarity were selected to generate the OTU table. To accurately assess the diversity of the microbial communities, all samples were rarefied to the same depth based on the minimum sequence number. Sample data were homogenized using rarefaction by the "vegan" package in R (version 3.3.1). The subsequent analyses conducted in this study were based on normalized data. To obtain the information of the species corresponding to each OTU, the RDP classifier (version 2.11, https://sourceforge.net/projects/rdp-classifier/,(accessed on 17 October 2021) Bayesian algorithm was used to analyze 97% similarity of the OTU representative sequence against the Silva (version 138, https://www.arb-silva.de/, (accessed on 20 October 2021) ribosomal RNA gene database using a confidence threshold of 70%. The community composition of each sample was analyzed at each classification level. Majorbio's cloud platform was used to extract the original data. The chloroplasts and mitochondria sequences were then removed by the platform and annotated the obtained sequences for species.
To count all OTUs and the number of OTUs shared and unique in multiple samples, Venn diagrams were made by the "venn diagram" package in R (version 3.3.1), and the sparse curves and other richness and diversity indices (ACE, Chao, Shannon and Simpson) of bacterial communities were estimated by Mothur (version 1.30.2, https: //www.mothur.org/wiki/Download_mothur, (accessed on 15 November 2021). The Circos-0.67-7 (http://circos.ca/, (accessed on 21 December 2021) was used to make the Circos sample and species relationship map. Clustering was performed using the "vegan" package of R language according to the similarity of abundance among species or samples to make a community Heatmap diagram and a community Pie diagram (PIE diagram). Difference tests were performed on the multiple groups of samples, and the Kruskal-Wallis H test was used as well as the "stats" package of R (version 3.3.1) and the "scipy" package of Python for mapping. Qiime (version 1.9.1, http://qiime.org/install/index.html, (accessed on 19 January 2022) was used to calculate the Beta diversity distance matrix for Hierarchical clustering analysis. A UPGMA algorithm was used to construct the tree structure, using the "pheatmap" package in R (version 3.3.1) for plotting. Principal coordinate analysis (PCoA) based on the Bray-Curtis distance was applied to reveal the differences in bacterial communities between groups. LEfSe (http://huttenhower.sph.harvard.edu/ galaxy/root?tool_id=lefse_upload, (accessed on 16 February 2022) was used for analysis and mapping, and Linear discriminant analysis (LDA) was used to screen the biomarkers for statistical differences between groups with LDA scores greater than 2. Use Networkx (version 2.1) network analysis kit to obtain the relative information of species and samples within or between groups and build a species correlation network. The Pathway (Level 1, Level 2, level 3) information in KEGG database was obtained by PICRUSt2 (version 2.2.0, https://github.com/picrust/picrust2/, (accessed on 26 March 2021) [33], and the function information of COG database and MetaCyc database was compared to predict the function of the assumed microbial community comprehensively. At the same time, a cluster heat map of the metabolic pathway abundance table was made by GraphPad Prism 8.0.2. Alpha analysis, Circos analysis, network analysis and function prediction were all calculated by each sample data, and the mean values of all samples of each insect species were then used for mapping and analysis. Raw sequencing data are available on the NCBI Sequence Read Archive under BioProject accession number PRJNA819209.
Differences were considered significant when * p < 0.05 and extremely significant when ** p < 0.01. SPSS23.0 software was used for statistical analysis.

Analysis of 16S rDNA Sequencing Results
A total of 18 samples, including 15 insect gut samples and 3 corn samples, were sequenced by Illumina Miseq PE300 and obtained 1,423,375 pairs of reads (Table 1). After splicing, quality control and redundancy removal, clustering analysis (based on 97% sequence similarity), and chimerism resulted in the removal of the above data. To accurately assess the diversity of the microbial communities, all samples were rarefied to the same depth based on the minimum sequence number. Sequence numbers were normalized to 12,358 for all samples of bacteria. Cluster analysis (based on 97% sequence similarity) obtained 176 OTUs, including 7 phyla, 16 classes, 31 orders, 67 families, 113 genera and 140 species. The Shannon-Wiener curves eventually flattened, indicating that the sequencing depth was sufficient to meet the requirements of the subsequent data analysis ( Figure S1). Among the 176 OTUs, 106 were found in the intestinal of C. punctiferalis, 101 OTUs in the intestinal of O. furnacalis and 75 OTUs in the intestinal of H. armigera (Table S1), whereas 72 OTUs were the endophytic corn bacteria (Table S2). C. punctifer-alis showed more OTU numbers than the other two insects. Compared with H. armigera, C. punctiferalis and O. furnacalis were closer in OTU numbers. The above data were all quality controlled and spliced. Sample\Info is the Sample name, Seq_num is the sequence number, Base_num is the base number and Mean_length is the average sequence length of all samples. Min_length is the shortest sequence length in the sample, and Max_length is the longest sequence length in the sample.
In the Venn diagram of the three lepidopteran pests, 42 OTUs were shared, whereas 29, 23 and 18 OTUs were specific to C. punctiferalis, O. furnacalis and H. armigera, respectively ( Figure S2). Of note, C. punctiferalis exhibited the most specific OTUs. C. punctiferalis and the O. furnacalis shared more OTU numbers than the comparisons between the others.

Comparison of the Gut Microbiota
The alpha diversity index indicated differences in intestinal flora of the three lepidopteran insects. The ace and chao indexes showed that the value of C. punctiferalis was the highest, followed by O. furnacalis, and H. armigera was the lowest. The values of C. punctiferalis and O. furnacalis in the Chao index were significantly different from those of H. armigera ( Figure 1A,B). The Shannon index showed that C. punctiferalis had the highest value, followed by O. furnacalis, whereas H. armigera had the lowest value ( Figure 1C). The Simpson index showed that C. punctiferalis had the lowest value, followed by O. furnacalis, and H. armigera had the highest value ( Figure 1D). All the indexes above showed that the richness and diversity of the microbial community in C. punctiferalis were the highest.
Cladogram from phylum to species was drawn to fully understand the distribution of these different taxa at various taxonomic levels ( Figure 3). It was found that the community differences of the three lepidopteran pests were mainly concentrated in Proteobacteria, Actinobacteria and Saccharibacteria, and Proteobacteria caused the most significant difference. One taxon differed significantly in the gut microbiota of H. armigera, whereas 4 and 25 taxa differed significantly in C. punctiferalis and O. furnacalis ( Figure S10). Figure 3. Cladogram of bacterial biomarkers, from the phylum (innermost ring) to species (outermost ring) level, with an LDA score >2. Each small circle at different taxonomic levels represents a taxon at that level, and the diameter of the circle is proportional to the relative abundance. The coloring principle is to color the species with no significant difference as yellow and the other different species as the group with the highest abundance of the species. Different colors represent different groups, and nodes with different colors represent the communities that play an important role in the group represented by the color. HA, H. armigera; CP, C. punctiferalis; OF, O. furnacalis.

Beta Diversity Analysis
At the genus level, the three pests were clustered together from the hierarchical clustering tree based on the Bray-Curtis distance algorithm, although the clustering form was chaotic. However, the corn samples were clustered in the same branch, which means that the composition similarity of the flora was higher than that of the three pests. The C. punctiferalis samples were the most complex, with some samples clustered with H. armigera or O. furnacalis and some samples clustered on one branch alone ( Figure S11).
The PCoA analysis based on the Bray-Curtis distance algorithm was used to compare the community similarities between samples. The PCoA scatter plot showed that the abscissa and ordinate represent the two characteristic values contributing to the largest differences between the samples. Their influence degrees were 45.58% and 31.15%, respectively ( Figure 4). The corn endophytic bacteria flora composition at the genus level differed from that of the pests. The corn endophytic bacteria flora was gathered together, whereas the bacterial flora of three pests covered more ( Figure 4). Performing subsequent correlation network analysis based on the selected samples with the sample abundance greater than 10, the collinear network analysis was carried out on three pests and corn. In addition to the common bacteria shared by the corn and three pests, the C. punctiferalis and corn samples were associated with Bacteroidetes at the phylum level ( Figure S12A). They were also associated with Pseudomonadaceae, Flavobacteriaceae and Xanthomonadaceae at the family level ( Figure S12B). At the genus level, they were associated with Chryseobacterium, Pseudomonas (Figure S12C), and they shared OTU103 at the OTU level ( Figure S12D).

Functional Prediction of Gut Microbiota
To understand the function of gut microbes of the three panicle pests, a functional assessment through PICRUSt2 was performed to analyze the KEGG pathway. Compared with the other two pests, C. punctiferalis showed higher abundance at three different levels. At Level 1, the abundance of metabolism pathways was higher than other pathways, such as environmental information processing, cellular processes, organismal systems, etc. Metabolism plays an important part in organisms. At Level 2 under metabolism Level 1, amino acid metabolism, metabolism of cofactors and vitamins, lipid metabolism and xenobiotics biodegradation and metabolism and metabolism of other amino acids all were superior in numbers in C. punctiferalis to that in H. armigera and O. furnacalis by the Homogeneity of variance test. Using the Homogeneity of variance test, similarly, at Level 3 under xenobiotics biodegradation and metabolism Level 2, drug metabolism-other enzymes, chloroalkane and chloroalkene degradation, aminobenzoate degradation and naphthalene degradation all showed predominance and significant difference in C. punctiferalis than the other pests ( Figure 5). From biofilm formation-vibrio cholerae, biofilm formation-pseudomonas aeruginosa, under Level 2 of cellular community-prokaryotes, and Level 2 of environmental adaptation, C. punctiferalis showed an obvious advantage over the other pests. In addition, C. punctiferalis showed significant differences in microbial metabolism in diverse environments (Level 3), fatty acid metabolism (Level 3), sulfur metabolism (Level 3), bacterial secretion system (Level 3), necroptosis (Level 3) and insect hormone biosynthesis (Level 3). However, they were not significant at corresponding Level 2 ( Figure 5). Furthermore, C. punctiferalis revealed a high abundance in carbohydrate metabolism (Level 2), membrane transport (Level 2), biosynthesis of other secondary metabolites (Level 2), glycan biosynthesis and metabolism (Level 2), transport and catabolism (Level 2), metabolism of terpenoids and polyketides (Leve l2), biosynthesis of other secondary metabolites (Level 2), energy metabolism (Level 2), digestive system (Level 2) and the excretory system (Level 2). Compared with H. armigera, the abundance of some functions in O. furnacalis was closer to the C. punctiferalis. It even exceeded the C. punctiferalis at translation (Level 2), transcription (Level 2) and cellular Community-eukaryotes (Level 2) ( Figure S13).

Discussion
In the corn ear stage, C. punctiferalis, H. armigera and O. furnacalis are three co-occurring pests that seriously affect the corn yield [34][35][36]. Recently, the C. punctiferalis has gradually aggravated and surpassed the O. furnacalis and H. armigera as the most important pest, especially in the Huang-Huai-Hai summer corn region [1]. In this study, we investigated the bacterial diversity and community composition of the three pests and the corn endophytic bacteria. According to our findings, C. punctiferalis gut bacteria composition and activity and its close relationship with corn endophytic bacteria may explain its prevalence.
Previous studies have shown that Proteobacteria and Firmicutes are the dominant phyla in intestinal samples of many insects [37,38]. They were also reported to be dominant in lepidopteran insects [30], as we reported in this study. The most significant difference among these three lepidopteran pests was Bacteroidetes, which were much more abundant in the peach moth than in the other two insects. The dominant family of these three lepidopteran pests was Enterobacteriaceae and Enterococcaceae, whereas Enterococcus and Klebsiella were the main genera, which are consistent with the intestinal flora of H. armigera, C. punctiferalis and O. furnacalis studied before [31,32,39].
Previous studies showed that diet [16] and taxonomy could influence insect gut bacterial communities [38,40]. Although there were some differences between the three pests at various bacterial levels, the majority were not significant, most likely because of their similar diets and feeding times. Two Pyralidae insects, C. punctiferalis and O. furnacalis [41], showed a more similar number and abundance of bacteria families, genera and OTUs than the Noctuidae insect H. armigera [42], which means that the phylogeny plays an important role in shaping insect gut microbiota.
Correlation network analysis showed more relationships between C. punctiferalis and corn kernel. Because they shared more OTU, genus and family numbers from Proteobacteria and Bacteroidetes, the C. punctiferalis intestinal microflora had more correlation and similarity in bacteria composition with corn samples, which could reflect the C. punctiferalis intestinal microflora's better adaptability to corn.
Among the three lepidopteran pests, OTU105 (Enterococcus casseliflavus), OTU151 (Klebsiella pneumoniae) and OTU49 (Serratia marcescens) were found in large amounts and accounted for a large proportion of the three lepidopteran pests. OTU105 (E. casseliflavus) was reported to have a complete L-tryptophan pathway in the silkworm intestine and could produce L-tryptophan, which is an essential aromatic amino acid for animal growth and development [43]. OTU151 (K. pneumoniae) may play important roles in fitness. It was reported K. pneumoniae could produce active molecules with effective antibacterial properties in cockroach intestines [44]. OTU49 (S. marcescens) is usually harmful and has strong pathogenicity and virulence [45,46]. Serratia was a pathogen in H. armigera [47]. In our study, C. punctiferalis showed the least Serratia, which is estimated to be related to the inhibition and antagonism of intestinal microorganisms.
Conogethes punctiferalis had a substantially higher abundance in functional prediction than the other two pests. On Level 1, the metabolism, environmental information processing, cellular processes and organismal systems were most outstanding in C. punctiferalis. Firstly, some intestinal bacteria of C. punctiferalis have been proved to have the ability to decompose and degrade the lignin, cellulose and hemicellulose [48,49]. In addition, they can use a variety of sugars, which is important for insects to overcome the defense mechanism of plants and adapt to complex environments and survive well. Proteobacteria exist widely in Lepidopteran insects and play important roles in function. Cedecea lapagei (OTU2), Pseudomonas mosselii (OTU103), Pseudomonas hibiscicola (OTU24) and Acinetobacter pittii (OTU38) all belong to Proteobacteria. OTU38 belongs to the Acinetobacter genus, which was reported to be able to degrade cellulose [50] and showed high total cellulase hydrolysis activity. Moreover, A. pittii was also found to have crude xylanase and magnetic xylanase (CLEA), which could convert xylanase form powdery rice straw (over 45%) and corn cob (over 60%) to xylo-oligosaccharide [49]. Secondly, some intestinal bacteria of C. punctiferalis could make good use of sugars. For example, P. mosselii could use a variety of sugars [51]. The A. pittii showed high endoglucanase hydrolytic activity [49]. All of the above functions enable insects to decompose better and digest food more easily, which perfectly corresponded with our predicted function higher in C. punctiferalis, glycan biosynthesis and metabolism (Level 2), transport and catabolism (Level 2) and the digestive system and excretory system (Level 2). Thirdly, the intestinal bacteria of C. punctiferalis have also shown adaptation and drug resistance. Chruseobacterium cucumeris (OTU200) belongs to Bacteroides. C. cucumeris and is involved in the biosynthesis of several compounds. It also contains genes for sodium/proton antiporter, glutathione, superoxide dismutase and cold shock proteins, which help it survive osmotic, oxidative and cold shock stress [52]. Moreover, P. mosselii, P. hibiscicola, A. pittii and C. sediminis could also degrade heavy metals, polycyclic aromatic hydrocarbons, long-chain alkanes and polychlorinated biphenyls [53][54][55][56][57][58]. It is speculated that they can help the host adapt to more complex environments and resist harsh ones. C. lapagei (OTU2), Achromobacter ruhlandii (OTU1) and P. hibiscicola (OTU24) were also reported to have resistant genes [59][60][61][62]. Similarly, the higher environmental adaptability and resistance of intestinal bacteria from C. punctiferalis were confirmed by functional analysis, such as xenobiotics biodegradation and metabolism (Level 2), drug metabolism-other enzymes (Level 3), chloroalkane and chloroalkene degradation (Level 3), aminobenzoate degradation (Level 3), naphthalene degradation (Level 3), environmental adaptation (Level 2), metabolism of terpenoids and polyketides (Level 2).
Furthermore, the intestinal bacteria of C. punctiferalis showed obvious bacteriostasis. C. cucumeris (OTU200) showed broad-spectrum antimicrobial activity [60]. P. mosselii (OTU103) and P. hibiscicola (OTU24) were reported to have an antagonistic effect on a variety of bacteria, including some pathogenic bacteria [63,64], which was speculated to better protect the C. punctiferalis from the invasion of some pathogenic microorganisms. Less Serratia observed in the stomach of C. punctiferalis compared to the other two pests may be explained by the presence of the above bacteria, implying that C. punctiferalis can readily withstand several pathogenic bacteria from the external environment, which is consistent with the functional prediction. C. punctiferalis was predicted with more xenobiotics biodegradation and metabolism (Level 2) and biosynthesis of other secondary metabolites (Level 2) functions.
In addition, C. punctiferalis gut bacteria also showed powerful metabolic ability. P. mosselii (OTU103) can produce bioactive secondary metabolites [65] and change the epithelial permeability of different cells to destroy the sytoskeleton [66]. A. pittii (OTU38) is also involved in unsaturated fatty acid synthesis, osmosis, membrane protein and expression of some genes in sulfur metabolism [56]. C. punctiferalis was predicted with more amino acid metabolism (Level 2), metabolism of other amino acids (Level 2), carbohydrate metabolism (Level 2), membrane transport (Level 2), lipid metabolism (Level 2), glycan biosynthesis and metabolism (Level 2), biosynthesis of other secondary metabolites (Level 2), metabolism of terpenoids and polyketides (Level 2) and transport and catabolism (Level 2) functions.
The intestinal bacterial composition and predicted functions of the three corn pests at the ear stage may explain the prevalence of C. punctiferalis. The increase amount of C. punctiferalis might be related to the abundant metabolic functions of P. mosselii (OTU103), C. lapage (OTU2), A. ruhlandii (OTU1), P. hibiscicola (OTU24) and other gut bacteria. However, metagenomic and meta-transcriptomic analysis will be needed in the future to elucidate host-microorganism interactions.

Conclusions
Our results showed differences in the gut bacteria communities of three co-occurring lepidopteran pests at the ear stage. Proteobacteria and Firmicutes were the dominant phyla, whereas the Enterobacteriaceae and Enterococcaceae were the dominant family in C. punctiferalis, H. armigera and O. furnacalis. Bacteroidetes were substantially more abundant in C. punctiferalis than in the other two insects, indicating a significant difference. C. punctiferalis had more correlation and similarity in bacteria composition with corn samples. According to the function prediction, the metabolism, environmental information processing, cellular activities, and organic systems dominated C. punctiferalis.

Supplementary Materials:
The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/insects13080740/s1, Figure S1: Shannon rarefaction curves for all samples; Figure S2: Venn diagram at OTU level; Figure S3: Circos (A) and distance heatmap (B) based on phylum level; Figure S4: Circos (A) and distance heatmap (B) based on family level; Figure S5: Pie chart of the intestinal and corn microbial community at phylum level; Figure S6: Pie chart of the intestinal and corn microbial community at family level; Figure S7: Pie chart of the intestinal and corn microbial community at genus level; Figure S8: Pie chart of the intestinal and corn microbial community at OTU level; Figure S9: Significance difference test based on Kruskal-Wallis rank sum test method at family level (A), genus level (B) and OTU level (C); Figure S10: Linear discriminant analysis (LDA) of three pests; Figure S11: Clustering analysis based on Bray-Curtis distance algorithm at genus level; Figure S12: Co-occurrence network diagram of all samples at phylum level (A), family level (B), genus level (C) and OTU level (D); Figure S13: Pathway heatmap of Level2; Table S1: Summary of multistage annotation information for three insect samples; Table S2  Institutional Review Board Statement: Not applicable.

Data Availability Statement:
The data presented in this study are available in supplementary material here.

Conflicts of Interest:
The authors declare no conflict of interest.