De novo transcriptome assembly of the bamboo snout beetle Cyrtotrachelus buqueti reveals ability to degrade lignocellulose of bamboo feedstock

Background The bamboo weevil Cyrtotrachelus buqueti, which is considered a pest species, damages bamboo shoots via its piercing–sucking mode of feeding. C. buqueti is well known for its ability to transform bamboo shoot biomass into nutrients and energy for growth, development and reproduction with high specificity and efficacy of bioconversion. Woody bamboo is a perennial grass that is a potential feedstock for lignocellulosic biomass because of its high growth rate and lignocellulose content. To verify our hypothesis that C. buqueti efficiently degrades bamboo lignocellulose, we assessed the bamboo lignocellulose-degrading ability of this insect through RNA sequencing for identifying a potential route for utilisation of bamboo biomass. Results Analysis of carbohydrate-active enzyme (CAZyme) family genes in the developmental transcriptome of C. buqueti revealed 1082 unigenes, including 55 glycoside hydrolases (GH) families containing 309 GHs, 51 glycosyltransferases (GT) families containing 329 GTs, 8 carbohydrate esterases (CE) families containing 174 CEs, 6 polysaccharide lyases (PL) families containing 11 PLs, 8 auxiliary activities (AA) families containing 131 enzymes with AAs and 17 carbohydrate-binding modules (CBM) families containing 128 CBMs. We used weighted gene co-expression network analysis to analyse developmental RNA sequencing data, and 19 unique modules were identified in the analysis. Of these modules, the expression of MEyellow module genes was unique and the module included numerous CAZyme family genes. CAZyme genes in this module were divided into two groups depending on whether gene expression was higher in the adult/larval stages or in the egg/pupal stages. Enzyme assays revealed that cellulase activity was highest in the midgut whereas lignin-degrading enzyme activity was highest in the hindgut, consistent with findings from intestinal gene expression studies. We also analysed the expression of CAZyme genes in the transcriptome of C. buqueti from two cities and found that several genes were also assigned to CAZyme families. The insect had genes and enzymes associated with lignocellulose degradation, the expression of which differed with developmental stage and intestinal region. Conclusion Cyrtotrachelus buqueti exhibits lignocellulose degradation-related enzymes and genes, most notably CAZyme family genes. CAZyme family genes showed differences in expression at different developmental stages, with adults being more effective at cellulose degradation and larvae at lignin degradation, as well as at different regions of the intestine, with the midgut being more cellulolytic than the hindgut. This degradative system could be utilised for the bioconversion of bamboo lignocellulosic biomass. Electronic supplementary material The online version of this article (10.1186/s13068-018-1291-9) contains supplementary material, which is available to authorized users.


Background
Lignocellulosic biomass resources are abundant, renewable and environmentally compatible [1]. Therefore, they may become an ideal energy resource for humans. It has been estimated that terrestrial biomass can produce 130 million tonnes of dry wood per year [1][2][3]. However, the stable structure of lignocellulose leads to a high cost of transformation and processing, which greatly restricts industrialisation. Although lignocellulolytic activity was originally believed to be restricted to plants, bacteria and fungi, evidence has accumulated in recent years for the existence of animal lignocellulolytic enzyme activity (such as cellulases, hemicellulases and lignases), particularly in cellulose-feeding insects [4][5][6][7]. These natural biomass utilisation systems (NBUS) are environmentfriendly and cost-effective for lignocellulose degradation, and their underlying mechanism could provide the basis for high-efficiency bioconversion of lignocellulose [8].
Bamboo is considered as a suitable plant for energy utilisation [14][15][16], and some studies suggest that bamboo is a promising candidate industrial feedstock for lignocellulose biomass because of its high growth efficiency [17][18][19][20]. As a relatively recently identified lignocellulose biomass resource, bamboo has attracted increasing interest over the past 5 years as an energy crop [17][18][19][20][21][22][23][24]. However, energy utilisation of bamboo is still in its initial stages, with the main research direction being the cracking of bamboo lignocellulose and the utilisation of bamboo products [25]. Bioconversion offers a new idea for highly efficient conversion of bamboo lignocellulose biomass to fuel ethanol and biodiesel. This challenge now awaits a solution: how to achieve highly efficient bioconversion of bamboo lignocellulose biomass. The bamboo weevil C. buqueti, a bamboo plantation pest, causes severe damage to several bamboo species, including Phyllostachys pubescens, Neosinocalamus affinis, Bambusa textilis and Dendrocalamus farinosus [26,27]. This insect damages bamboo shoots via both its piercing-sucking mode of feeding and egg-laying [28]. On the basis of previous research on termites and other beetles that utilise lignocellulosic biomass [29][30][31][32][33][34], in the present study, we used RNA sequencing and quantification of lignocellulolytic enzyme activity to explore the possibility of bioconversion of lignocellulosic biomass of bamboo feedstock by C. buqueti.

Prediction of genes encoding carbohydrate-active enzymes in the developmental stage transcriptome of C. buqueti
The de novo developmental transcriptome of C. buqueti comprised 31,469,916, 36,773,825, 32,128,345, 33,070,448 and 31,434,121 clean reads in eggs, larvae, pupae, female and male imagos, respectively, with a total of 108,854 transcripts obtained and assembled into 83,115 unigenes [35].
The main enzymes related to lignocellulose degradation were CAZymes, which can be divided into the following six categories: glycoside hydrolases (GHs), glycosyltransferases (GTs), polysaccharide lyases (PLs), carbohydrate esterases (CEs), auxiliary activities (AAs) and carbohydrate-binding modules (CBMs) [36]. Consequently, we conducted a search to find all CAZyme genes in the developmental transcriptome. We predicted the total proteins of the C. buqueti transcriptome with an e value = 1e −5 . The results indicated that 806 unigenes had multiple domains that were assigned to CAZyme families, including 55 Table S1; Additional file 2: Table S2). Among these genes, only 99 genes belonged to microbial communities (Additional file 2: Table S2).
Six xylanase genes were present in the transcriptome. Numerous genes encoding proteins associated with hemicellulose degradation, such as mannosidase and galactosidase, were also detected (Table 1).  Furthermore, 22 β-galactosidases, 25 mannosidases,  17 xyloglucosyltransferases, 82 arylesterases and 75 acetylxylan esterases were identified in the transcriptome (data not shown). The CE10 family exhibited carboxylesterase and xylanase activities as well as mannosidase, galactosidase, xyloglucosyltransferase and acetylxylan esterase activities involved in hemicellulose degradation [37]. These findings indicate that C. buqueti has the ability to degrade xylan and other components of hemicellulose.

Co-expression network analysis of unigenes with weighted gene co-expression network analysis (WGCNA) at different developmental stages
WGCNA was used for analysing relationships and networks involving the various genes. To build a scale-free network, parameter analysis was performed (Fig. 1). An adjacency function in WGCNA was used to weigh different genes using the following formula: a ij = (S ij , β) = |S ij | β . As shown in Fig. 1a, we changed the value step-by-step to identify the optimal value, so that the average connectivity of the network was smooth. The value of β = 11 was ultimately determined on the basis of the diagnosis chart showing that the average number of co-expressed genes in the final network was 50 (Fig. 1b). As observed in the dendrogram (Fig. 2a), 19 unique module eigengenes were identified (Table 2; Additional file 3: Table S3). Each of the 19 eigengenes correlated with a particular tissue type and developmental stage (Fig. 2b). The three coexpression modules comprised genes that were highly expressed in the egg stage, four in the pupal stage, three in the larval stage, three in female imagos and two in male imagos (r > 0.8; Fig. 2b).
The gene expression patterns of the MEyellow module were divided into the following two types: egg and pupal stages clustered together with a decreasing gene expression level (Fig. 2c); the male, female and larval stages formed another cluster in which gene expression levels increased (Fig. 2c). The egg and pupal stages were dormant and had no obvious foraging activity, while there was a vigorous period of foraging activity in the female imago and larval stages. These findings suggest that the genes in this module may be involved in the life activities of C. buqueti, such as foraging.

Functional enrichment analyses of genes in the MEyellow module
To understand the foraging behaviour of C. buqueti, we focussed on the MEyellow module. KEGG pathway and GO enrichment analyses were performed for this model, whereby all genes and hub genes in the MEyellow coexpression module, the first 10% of all genes, were used. According to the GO analysis, all genes in the MEyellow co-expression module were highly enriched in biological processes, such as carbohydrate metabolism, starch metabolism, sucrose metabolism, lipid glycosylation and cellulose catabolism; those enriched in the KEGG pathways were associated with starch and sucrose metabolism, protein digestion and absorption, carbohydrate digestion and absorption, fructose and mannose metabolism and other glycan degradation. Hub genes were mainly enriched in biological processes, such as carbohydrate metabolism, starch metabolism, sucrose metabolism, lipid catabolism, glycogen biosynthesis and cellulose catabolism; those enriched in the KEGG pathways were associated with carbohydrate digestion and absorption and protein digestion and absorption ( Table 3).
The 50 most highly connected hub genes in the MEyellow co-expression module were used for analysing gene expression and co-expression networks. Gene expression showed that the expression level in imagos and larval stages was higher than that in egg and pupal stages (Fig. 3a). Co-expression networks showed two core hub genes, namely c85857_g1 and c54229_g1 (Fig. 3b). The c54229_g1 gene belongs to the tetraspanin family, whereas the c85857_g1 gene is of unknown function. Remarkably, in this module, the hub gene c47220_g1 was annotated to the glycoside hydrolase 48 gene family (GH48), which is an important glycoside hydrolase. The GH48 gene family also encodes cellulose exonuclease, which degrades cellulose by the formation of a multienzyme cellulosome complex with other glycoside hydrolases or free enzyme systems.
Cqcbh5 belongs to GH48 and encodes a cellulose exonuclease that acts to degrade cellulose. Phylogenetic analysis revealed that Cqcbh5 was closely related to the exoglucanase of four phytophagous insects Dendroctonus ponderosae, Rhynchophorus ferrugineus, Leptinotarsa decemlineata and Anoplophora glabripennis, as well as to that from some cellulolytic bacteria (Fig. 3c, d). This finding suggests that Cqcbh5 has a function similar to that of insect and bacterial exoglucanases, which is involved in cellulose degradation. Moreover, the mRNA level of Cqcbh5 was higher in imago and larval stages than in egg and pupal stages (Fig. 3a), suggesting that the insect can utilise the cellulose of bamboo shoots during these developmental stages.

Expression of CAZyme family genes in sub-modules
We screened all CAZyme family genes in the MEyellow module. The MEyellow module contained 41 GHs, 16 GTs, 9 CBMs, 24 CEs and five AAs, whereas PLs were absent. A reads per kilobase per million reads expression heat map for each family of CAZymes in the MEyellow module was generated according to gene expression during development. The expression patterns of these CAZyme family genes were divided into two categories: one for eggs and pupae and another for female and male larvae. Expression levels of CAZyme family genes in adult and larval stages were higher than those in egg and pupal stages in the MEyellow module ( Fig. 4a-e). Lignocellulose degradation is mainly associated with the action of proteins encoded by CAZyme family genes [49]. In this study, many CAZyme family genes exhibited higher expression levels in adult and larval stages than in egg and pupal stages. Via its piercing-sucking mode of feeding, C. buqueti mainly eats bamboo shoots, which are enriched in carbohydrates, sugars and lignocellulose (Additional file 4: Table S4). These findings indicate that larvae and adults have the ability to convert lignocellulose in bamboo shoots into nutrients and energy for growth. In the presented summaries of expression patterns of all CAZyme genes in the transcriptome, genes that were not expressed in most samples have been deleted. As shown in Fig. 4S, 391 genes, namely 103 GHs, 132 GTs, 73 CBMs, 55 CEs and 28 AAs, remained for analysis. The expression patterns of GH can be divided into two main categories: one with no obvious differences between the developmental stages and one in which expression is higher in adult and larval stages than in pupal and egg stages (Additional file 5: Fig. S1a). The expression patterns of GT can be grouped into three categories: one with no obvious differences across development, one in which expression is higher in adult and larval stages than in pupal and egg stages and a third in which the expression pattern differs from category two (Additional file 5: Fig. S1b). The expression pattern of CBM was similar to that of GT (Additional file 5: Fig. S1c), whereas the expression pattern of CE was similar to that of GH (Additional file 5: Fig. S1d). AA gene expression did not show significant differences across developmental stages (Additional file 5: Fig. S1e).

Changes in the expression of carbohydrate metabolism, fatty acid metabolism, protein metabolism and energy metabolism genes in the developmental transcriptome
Bamboo shoots are rich in various nutrients (Additional file 4: Table S4), containing abundant carbohydrates, sugars, fats and proteins. It is not clear whether C. buqueti can utilise these nutrients or whether their energy metabolism changes after feeding on bamboo shoots. To determine whether C. buqueti can efficiently utilise bamboo shoot biomass, we analysed the expression patterns of genes associated with the metabolism of carbohydrates, fatty acids, proteins and energy in the developmental transcriptome. We also assessed whether the expression changed across development and whether any such changes agreed with the feeding habits of the insect. The expression levels of most genes involved in these pathways in the MEyellow co-expression module were higher in imago and larval stages than in egg and pupal stages ( Fig. 5a-d). These findings indicate that metabolic pathways operate at a higher rate in adults and larvae and might relate to the ability of adults and larvae to digest carbohydrate, lipids and proteins from bamboo shoots. Fig. 1 Determination of power beta value based on the adjacency matrix using WGCNA. The adjacency matrix from co-expression data was weighted by the power of correlation data between different genes; i.e. a ij = |S ij |β. The weighted parameter power beta value was determined from the scale-free topology criterion. To ensure that the average connectivity of the network was smooth, we chose β = 11 based on both charts: a for topology fitting results and b for mean connectivity

Prediction of carbohydrate-active enzyme gene expression in the imago transcriptome
In a previous study, we conducted RNA sequencing of the digestive system, reproductive system and muscle tissue of imagos collected in the cities of Leshan and Chishui [42]. There are clear differences in the C. buqueti population sizes between the two cities [43]. Our analysis of genes related to lignocellulose degradation in the transcriptomes of these two populations demonstrated that 843 genes had multiple domains assigned to CAZyme families, namely 249 GHs, 244 GTs, 133 CEs, 9 PLs, 87 enzymes with AAs and 121 CBMs; 106 of these proteins also contained signal peptides that were predicted to be extracellular proteins (Fig. 6).
In the transcriptomes of C. buqueti in Muchuan and Chishui, there were 19 cellulase genes [including 4 endoglucanase (GH8) genes], 4 exoglucanase (GH10) genes and 11 β-glucosidase genes (GH1, 3) (Table 4). Among the cellulases, seven protein sequences, including endoglucanases, β-glucosidases and exoglucanases, exhibited potential secretion signals. However, there were only three GHs containing CBM domains, which were surprisingly unrelated to cellulose degradation. Three endoglucanase genes (Cqeng1, c20964_g1_i1; Cqeng2, c31184_g1_i1 and Cqeng3, c63642_g1_i1), four exoglucanase genes (Cqcbh1, c23242_g1_i1; Cqcbh2, c29519_ g1_i1; Cqcbh3, c49080_g1_i2 and Cqcbh4, c49080_g1_i1) and seven β-glucosidase genes (Cqbgln1, c31266_g1_i2; The parameter deepSlip = 4 was set in the WGCNA analysis, which provides high sensitivity to cluster splitting. We additionally required each gene module to contain ≥ 50 genes. In total, 10,789 genes were grouped into 19 modules, which are presented as different colours. The top five modules ordered by the number of genes were turquoise with 1952 genes, blue with 1701 genes, brown with 1333 genes, yellow with 1092 genes and green with 654 genes. The grey colour in the left of the figure represents the seven genes that were not associated with any module. b Module-tissue associations. Each row corresponds to a module. Each column corresponds to a specific tissue. The colour of each cell at the row-column intersection indicates the correlation coefficient between the module and developmental stage. A high degree of correlation between a specific module and developmental stage is indicated by dark red or dark green colour. c The gene expression patterns in MEyellow module gln2, c31266_g2_i1; Cqbgln3, c31266_g2_i2; Cqbgln4, c31732_g6_i1; Cqbgln5, c31732_g6_i2; Cqbgln6, 31852_ g1_i5 and Cqbgln7, c31852_g1_i2) were used in a phylogenetic analysis including termites and other beetles to assess the evolutionary relationships with these species (Additional file 6: Fig. S2a-c).
Many genes encoding enzymes potentially involved in lignin degradation were identified in the C. buqueti transcriptome (Fig. 6). Among them, genes for two laccases Cqlac1 (c27827_g1_i1) and Cqlac2 (c28149_g1_i1) were used in the phylogenetic analysis. The analysis revealed that Cqlac1 and Cqlac2 were closely related to the laccase (Lac) genes of Monochamus alternatus and D. ponderosae (Additional file 6: Fig. S2c), indicating that insects from different geographical areas exhibit many CAZyme family genes.

Comparison of enzyme activities at different developmental stages and in different intestinal tissues from imagos or larvae
We detected the activity of several lignocellulolytic enzymes in imagos and larvae. Cellulase activities differed across the various developmental stages, with each enzyme exhibiting different activity patterns. Activity of exoglucanase (CBH), which reached 584.753 ± 91.215 U/g in the foregut of adult females, was higher than that of β-glucosidase (CB) and endoglucanase (EG) in adult females, which was 27.639 ± 9.401 U/g in the hindgut and 235.814 ± 59.925 U/g in the midgut, respectively. In adult males, EG exhibited the highest enzyme activity and CB exhibited the lowest. The enzyme activity pattern in larvae was similar to that in adult males (Table 5). Furthermore, enzyme activity differed between various regions of the intestine. The overall highest CBH activity was observed in the midgut (with the exception of highest activity in the female foregut), whereas the highest CB activity was in the hindgutparticularly the male hindgut (214.597 ± 54.711 U/g). EG activity was highest in the midgut, peaking in males (1744.8271 ± 50.604 U/g). In summary, these results showed that cellulase activities differed according to both developmental stage and intestinal region. These findings suggest that the different aspects of cellulose degradation in C. buqueti are performed at different developmental stages and in different parts of the intestine.
Lignin-degrading enzyme activity also differed according to the developmental stage and intestinal region. Laccase (Lac) activity was highest in the midgut, reaching 5.101 ± 1.171 U/g in larval midgut, which was notably higher than in the adult midgut of adults or other intestinal regions. Manganese peroxidase (MnP) activity was highest in males, by adult females and larvae, whereas whole-intestine enzyme activity was highest in larvae (0.893 ± 0.428 U/g). Among the different parts of the gut, MnP activity was highest in the hindgut of adult females (0.558 ± 0.257 U/g) and males (1.372 ± 0.421 U/g) and in the midgut of larvae (2.162 ± 0.997 U/g). Lignin peroxidase (LiP) activity was highest in the hindgut of males (1.453 ± 0.636 U/g). These results indicate that both adults and larvae of C. buqueti have the ability to degrade lignin and that this ability differs according to the developmental stage and intestinal region.

Expression analysis of lignocellulase genes in different intestine regions in imagos and larvae
To help elucidate the expression patterns of lignocellulolytic enzyme-encoding genes in the gut of C. buqueti, qRT-PCR was conducted on 10 such genes, namely three endoglucanase genes (Cqeng1, Cqeng2 and Cqeng3), two β-glucosidase genes (Cqbgln5 and Cqbgln7), two exoglucanase genes (Cqcbh1 and Cqcbh2), one xylanase gene (Cqxyn1) and two Lac genes (Cqlac1 and Cqlac2), with the EF1-ɑ gene acting as the reference gene, using the primers listed in Table 4. Expression of these 10 genes was detected in the mouthparts, foregut, midgut, hindgut and whole gut of adult females, adult males and larvae. Seven cellulase genes were mainly expressed in the foregut and midgut, with higher expression levels in the midgut. Expression patterns of the two β-glucosidase genes Cqbgln5 and Cqbgln7 were midgut > foregut > hindgut > mouthparts and midgut > foregut > mouthparts > hindgut, respectively. The Cqbgln7 β-glucosidase gene was highly expressed in the intestines of both adults and larvae, with the highest expression level occurring in the midgut, which was higher than that of the other β-glucosidase gene Cqbgln5. Of the three endoglucanase genes, Cqeng1 and Cqeng2 showed high expression in all samples whereas the expression of Cqeng3 was far lower but peaked in the midgut. The exoglucanase gene Cqcbh1 was more highly expressed in all samples than the other exoglucanase gene Cqcbh2. The xylanase gene Cqxyn1 exhibited low expression levels in all samples, except for the gut of larvae. The two Lac genes Cqlac1 and Cqlac2 were mainly expressed in the hindgut and their expression was higher in larvae than in adults (Fig. 7). These findings show that cellulase genes were expressed at the highest level in the midgut of C. buqueti, with highest expression in endoglucanase genes followed by the β-glucosidase and exoglucanase genes, with expression being higher in adults than in larvae. Lignin degradation genes were expressed at the highest level in the hindgut, with higher expression levels in larvae than in adults. These findings indicate that the expression of endoglucanase and β-glucosidase genes primarily occurred in the adult midgut, whereas that of Lac primarily occurred in the hindgut of larvae. These results are in accord with those of the enzyme activity assays. Moreover, correlation analyses between enzyme activity data and qRT-PCR data were performed, which revealed that only the expression pattern of Cqeng2 was significantly correlated to enzyme activity (Additional file 4: Table S4; Additional file 7: Table S5).  exocellulases and generate individual monosaccharides [45,46]. In this study, we used the 1,3-dinitrosalicylic acid (DNS) assay method to determine cellulase activities in the intestines of adult females, adult males and larvae. In adult females, endoglucanase activity was highest in the midgut and foregut, followed by the mouthparts and the hindgut. In males, endoglucanase activity was also highest in the midgut and foregut and was comparable to activity in larvae. CB activity was highest in the hindgut of adult females, in the hindgut and midgut of adult males and in the foregut and midgut of larvae. EG activity was highest in the midgut of adult females, in the mouthparts, foregut and midgut of adult males and in the midgut of larvae. Insect cellulase was identified in the intestines of termites and cockroaches [47,48] and subsequently detected in other insects [49][50][51][52][53][54][55][56][57][58][59]. Endogenous cellulase activity exists in insects and at least seven orders, comprising 28 species, have been found to contain a cellulase gene [6,60]. Jiang [61] compared cellulase activities among three species belonging to different subfamilies of Cerambycidae. Duan [62] compared cellulase activity between Monochamus alternatus and Cipangopaludina chinensis, whereas Shi [63] compared xylanase and cellulase activities between three orders and three species of insects, including one member of Cerambycidae. Oppert [7] investigated cellulase activities in 68 species from eight orders of phytophagous insects, whereas Su [64] studied intestinal cellulase activities in 54 species of seven insect orders. Li [65] determined cellulase activities in 15 beetles. Taken together, these studies demonstrate that cellulase activity is limited by many factors, including substrate concentration and reaction time, and that results depend on the protein quantification methods used and other factors. Hence, it is difficult to directly compare the results from the current study with those from other reports [7]. Therefore, the purpose of this study was to determine the activities of lignocellulase enzymes in C. buqueti and to compare the activities of the different lignocellulolytic enzymes at different developmental stages and in different intestinal regions within C. buqueti.
We also determined activities of individual lignin degradation enzymes, such as Lac, LiP and MnP. Lac and MnP activities were highest in larvae, particularly in the midgut and hindgut. Lac was mainly distributed in the midgut of adult females, the hindgut of adult males and the midgut of the larvae. MnP activity was highest in the hindgut of adult females, mouthparts of adult males and hindgut and midgut of larvae. LiP was mainly distributed in the midgut of adult females, hindgut of adult males and midgut of larvae. Ander and Eriksson [66] noted that although Lac could efficiently degrade lignin, LiP had higher catalytic oxidoreduction potential and could catalyse a range of lignin compounds, including phenol, aromatic ethers, methoxy benzene, methyl alcohol and polycyclic aromatic compounds [67,68]. The mechanism of action of MnP was similar to that of LiP [69]. These results indicate that C. buqueti has the ability to biodegrade lignin and cellulose, and that characterisation of its degradation system will be useful for using bamboo lignocellulose to produce biofuels.
In this study, biochemical techniques were used to demonstrate that lignocellulolytic enzyme activities in C. buqueti are highest in the midgut. These highly efficient enzymes could be introduced into microbes by synthetic biology to increase the yield of cellulase. Efficient lignocellulose degradation mechanisms by termites and other natural systems have provided important information on how lignocellulose can be exploited. By combining physical and chemical treatments with a natural enzyme system, it will be possible to achieve efficient hydrolysis of all carbohydrates in biomass under normal temperatures and pressure [89]. Sun [90] simulated a termite biotransformation system using properly comminuted biomass, adding specific glycosylhydrolases and lignin oxidase and separating aerobic and anaerobic reaction zones to achieve efficient lignocellulosic biomass biodegradation [90]. By contrast, research into the bamboo lignocellulose degradation mechanism in C. buqueti digestive is still in its infancy. Many significant biological problems must be resolved before rapid and effective bamboo lignocellulose degradation can be achieved, such as the mechanisms and biological functions of intestinal symbiotic bacteria.

Conclusions
Cyrtotrachelus buqueti, a bamboo shoot snout beetle, is considered a pest by the bamboo industry. Using transcriptome analysis to dissect the mode-of-action of lignocellulose degradation in C. buqueti, this work provides a theoretical basis for the development of bamboo as a bioresource for the biofuel and bioenergy industries. Because larvae and adults feed mainly on bamboo shoots containing abundant lignocellulose, we hypothesised that C. buqueti utilise bamboo lignocellulose for development and growth. WGCNA was used to analyse the diversity of lignocellulose degradation enzymes, including CAZymes, during C. buqueti development. The results showed that CAZymes genes in the MEyellow module in larval and adult stages (when bamboo feeding takes place), rather than in egg and pupal stages, were consistent with the eating habits of C. buqueti. Of the three cellulases, enzyme activity assays showed that the activity of

Insect collection
Larvae and adults of C. buqueti were collected in July 2017 from the bases of bamboo plants at a bamboo plantation in Muchuan City, Sichuan Province, China (N103°98′, E28°96′). All adults were used in the experiment 3 days after emergence [28]. Adults and larvae were reared in the laboratory at 25 °C ± 1 °C and 70% ± 10%

Transcriptome data from Cyrtotrachelus buqueti Guérin-MéNeville and CAZyme family analysis
Transcriptomes from five different C. buqueti developmental stages, namely eggs, larvae, pupae, adult male and adult female, were used [35]. We downloaded raw data from the National Centre for Biotechnology Information (NCBI) (https ://www.ncbi.nlm.nih.gov/) and focussed on genes associated with the lignocellulose degradation pathway. To identify genes involved in lignocellulose degradation, coding sequences were analysed using the dbCAN CAZyme annotation algorithm, which gives the hidden Markov model index files of various carbohydrate enzyme domains by hmmscan [91]. Weighted correlation network analysis (WGCNA) was used to analyse the 10,789 genes in the transcriptome [92].

Reconstruction of a scale-free co-expression network using WGCNA
We used the co-expression network approach to reconstruct the scale-free co-expression network for C. buqueti and then built and mined the gene co-expression network. Using the WGCNA package [92], we first built a similarity matrix between all gene pairs using bi-weight mid-correlation based on normalised fragments per kilobase per million reads (FPKMs).

Identification of functional modules
To identify functional modules in our reconstructed coexpression network, the adjacency matrix was further transformed to a topological overlap matrix using the WGCNA package. By setting the deepSplit parameter from 0 to 4 with the dynamic TreeCut package version 1.62, we found the optimal value to generate smaller clusters; a final deepSplit value of 4 was chosen and resulted in 19 modules (Fig. 2a). The relationship between modules was summarised by the eigenvalue 'eigengene' , which represents the expression profile with weighted genes for each module [93].

Pathway enrichment analysis and network analysis
We performed pathway enrichment analysis on the genes of interest, including enrichment in predefined pathways from the Kyoto Encyclopaedia of Genes and Genomes (KEGG) and Gene Ontology (GO) using the Cytoscape software platform (version 3.4) [94]. We used the degree of node metric to represent the number of connections for one node to the other nodes in the network and to identify the shortest path, represented by the fewest number of steps from one node to another [95].

Assays of lignocellulolytic enzyme activity
A total of 165 female imagos, 165 male imagos and 165 larvae of C. buqueti were sampled to determine the activity of lignocellulolytic enzymes. First, the digestive system was dissected into mouthparts, foregut, midgut, hindgut, total intestine, and mouthparts + total intestine. Next, tissues were ground into 1 ml PH 5.6 PBS extraction buffer, the crude extract was centrifuged at 13,000× for 10 min at 4°C and the supernatant was collected. The supernatant represented the crude enzyme solution. Each replicate sample contained tissues from at least five insects and five biological replicates were conducted for each treatment.
The crude enzyme solution was used for the assays to determine lignocellulolytic enzyme activity. The assay method for endoglucanase (EC 3.2.1.4) and exoglucanase (EC 3.2.1.91) was performed as described by Ghose et al. [96], and β-glucosidase (EC 3.2.1.21) activity was assayed as described by Parry et al. [97]. Carboxymethyl cellulose (CMC), microcrystalline cellulose (MCC) and salicin were used as substrates for determination of endoglucanase, exoglucanase and β-glucosidase, respectively. First, 2 ml 1% CMC, MCC or salicin was added to a 25 ml test tube and preheated at 50 °C for 2-3 min. Second, 0.5 ml crude enzyme solution was added and incubated for 30 min at 50 °C. Next, 2.5 ml DNS was added and incubated for 5 min at 100 °C to immediately terminate the reaction. Finally, 25 ml PH 5.6 PBS was added and the optical density value was determined at a wavelength of 540 nm.
Lignin peroxidise (LiP)-like activity was measured according to the method by Shi et al. [98]. Briefly, veratryl alcohol (VA) was used as the substrate and the reaction was performed at PH 5.6 PBS. LiP activity was measured by monitoring the oxidation of VA at 310 nm. Laccaselike activity was measured according to the method used by Nakagawa et al. [99], in which 2, 2′-azino-bis (ABTS) was used as the substrate and enzyme activity was measured by monitoring oxidation of ABTS at 420 nm. Manganese peroxidise (MnP)-like activity was measured by monitoring oxidation of 2,6-dimethyl phenol (2,6-DMP) to coerulignone at 469 nm (ε469 = 49,600/mol cm) [98]. All assays were performed with five replicates.
Tissue RNA extraction and qRT-PCR of lignocellulolytic enzyme genes in the C. buqueti digestive system Thirty females, 30 males and 30 larvae that had been starved for 24 h were subjected to qRT-PCR assays. The five tissues (mouthparts, foregut, midgut, hindgut and intestine) of the three developmental stages (larvae, male and female) were rapidly extracted. The RNAprep Pure Table 6 The primer sequence of qRT-PCR