Combined transcriptome and metabolome analyses to understand the dynamic responses of rice plants to attack by the rice stem borer Chilo suppressalis (Lepidoptera: Crambidae)

Background Rice (Oryza sativa L.), which is a staple food for more than half of the world’s population, is frequently attacked by herbivorous insects, including the rice stem borer, Chilo suppressalis. C. suppressalis substantially reduces rice yields in temperate regions of Asia, but little is known about how rice plants defend themselves against this herbivore at molecular and biochemical level. Results In the current study, we combined next-generation RNA sequencing and metabolomics techniques to investigate the changes in gene expression and in metabolic processes in rice plants that had been continuously fed by C. suppressalis larvae for different durations (0, 24, 48, 72, and 96 h). Furthermore, the data were validated using quantitative real-time PCR. There were 4,729 genes and 151 metabolites differently regulated when rice plants were damaged by C. suppressalis larvae. Further analyses showed that defense-related phytohormones, transcript factors, shikimate-mediated and terpenoid-related secondary metabolism were activated, whereas the growth-related counterparts were suppressed by C. suppressalis feeding. The activated defense was fueled by catabolism of energy storage compounds such as monosaccharides, which meanwhile resulted in the increased levels of metabolites that were involved in rice plant defense response. Comparable analyses showed a correspondence between transcript patterns and metabolite profiles. Conclusion The current findings greatly enhance our understanding of the mechanisms of induced defense response in rice plants against C. suppressalis infestation at molecular and biochemical levels, and will provide clues for development of insect-resistant rice varieties. Electronic supplementary material The online version of this article (doi:10.1186/s12870-016-0946-6) contains supplementary material, which is available to authorized users.


Background
To protect against attack by herbivorous insects, plants have evolved both constitutive and induced defense mechanisms [1]. Induced defenses include both direct and indirect responses, which are activated by herbivore feeding, crawling, frass, or oviposition [2]. Induced direct responses involve the production of secondary metabolites and insecticidal proteins, which can reduce herbivore development and survival [1,3]. While induced indirect responses mainly involve the release of volatile chemicals that can attract natural enemies of herbivores [1, 3,4].
Plant response against herbivory are associated with large-scale changes in gene expression and metabolism [5][6][7][8][9]. The integration of modern omics technologies such as transcriptomics, proteomics, and metabolics provides a great opportunity for a deeper understanding of the mechanisms of plant defence responses to herbivore feeding at molecular and cellular levels [7,[9][10][11]. Previous results have suggested that plant response to herbivore feeding is a dynamic process, and that the transcript patterns, protein and metabolite profiles are temporally and spatially regulated [1, 10,12]. This suggests that it is essential to investigate the dynamic at transcriptional, proteomic and metabolic changes associated to insect feeding [6,7,9,11]. Transcriptomic and proteomic studies are only able to predict changes in gene expression and the protein level, while metabolomic studies investigate the changed functions exerted by these genes or proteins. Therefore, the integration of transcriptomic, proteomic, and metabolic approaches can gain a better understanding of plant responses to herbivore feeding [10].
Rice (Oryza sativa L.) is the staple food for more than half of the world's population [13], but rice yield is frequently reduced by herbivorous insects [14]. Lepidopteran stem borers are chronic pests in all rice ecosystems, and the rice stem borer Chilo suppressalis is among the most serious rice pest in temperate regions of Asia [15]. C. suppressalis is particularly damaging in China because of the wide adoption of hybrid varieties. A better understanding of the genetic and molecular mechanisms underlying rice plant defense against insect pests is important for developing resistant rice varieties and other strategies for controlling pests [14]. The genetic basis of rice defense against piercingsucking planthoppers has been well elucidated, and several gene functions have been identified [16][17][18][19]. For example, Liu et al. [16] identified several lectin receptor kinase genes that confer durable resistance to the brown planthopper Nilaparvata lugens and the white back planthopper Sogatella furcifera. However, the defense response of rice plants to chewing insects, such as lepidopteran larvae, has rarely been studied, although a few studies have been conducted using microarray technology, in which a relatively small number of differentially expressed genes were identified [8,20,21]. In addition, the previous experiments were conducted with rice samples collected at only one time point after C. suppressalis infestation, and the data did not therefore reveal the dynamic response of rice plants to C. suppressalis feeding at transcriptional and metabolic levels.
In the current study, we combined transcriptome and metabolome analyses to investigate the dynamic responses of rice plants to attack by C. suppressalis, with the expectation to provide a better understanding of rice defense mechanisms to C. suppressalis infestation and clues for the development of rice pest control strategies.

Plants and growing conditions
The rice cultivar Minghui 63, an elite indica restorer line for cytoplasmic male sterility in China, was used in this study. Seeds were incubated in water for 2 day and sown in a seedling bed in a greenhouse (27 ± 3°C, 65 ± 10% RH, 16 L: 8 D). Fifteen-day-old seedlings were individually transplanted into plastic pots (630 cm 3 ) containing a mixture of peat and vermiculite (3:1). Plants were watered daily and supplied with 10 ml of nitrogenous fertilizer every week. Plants were used for the experiments four weeks after transplanting.

Insect colony
Specimens of C. suppressalis were retrieved from a laboratory colony that had been maintained on an artificial diet for over 60 generations with annual introductions of field-collected individuals. The colony was maintained at 27 ± 1°C with 75 ± 5% RH and a 16 L : 8 D photoperiod [22].

Insect bioassay
Potted rice plants were transferred to a climate control chamber (27 ± 1°C, 75 ± 5% RH, 16 L : 8 D photoperiod) for 24 h and were then infested with three 3 rd -instar C. suppressalis per plant. The larvae had been starved for 2 h before they were caged with the rice plants. The main rice stems, 4 cm above the area damaged by the larvae, were harvested after they had been exposed to C. suppressalis feeding for 0 (healthy, control rice plants), 24, 48, 72, and 96 h. Plant samples were immediately frozen in liquid nitrogen and stored at −80°C for later analyses. Four samples (replicates) were collected at each of the following time points and were used for transcriptome analysis: 0, 24, 48, and 72 h. Ten samples were collected at each of the following time points and were used for metabolome analyses: 0, 48, 72, and 96 h. The sampling time points differed for the transcriptome and metabolome analyses because the rice plants were expected to respond faster to insect feeding on the transcriptomic level than on the metabolomic level [1, 10].

RNA extraction
The total RNA from the rice stem samples was isolated using TRIzol reagent (Invitrogen, Carlsbad, CA, USA) according to the manufacturer's instructions. RNA quality was checked with a 2200 Bioanalyzer (Agilent Technologies, Inc., Santa Clara, CA, USA). The assessment showed that the RNA integrity number (RIN) of all samples was > 9.7.

Library preparation and RNA-sequencing
The sequencing library of each RNA sample was prepared using Ion Total RNA-sequencing (RNA-Seq) Kit v2 (Life Technologies, Carlsbad, CA, USA) according to the manufacturer's protocols. In brief, mRNA was purified from 5 μg of total RNA from each sample with oligo (dT) magnetic beads and was fragmented using RNase III (Invitrogen, Carlsbad, CA, USA). The fragmented mRNA was hybridized and ligated with Ion adaptor. The first-strand cDNA strand was synthesized using reverse transcription of random primers, which was followed by second-strand cDNA synthesis using DNA polymerase I and RNase H (Invitrogen, Carlsbad, CA, USA). The resulting cDNA fragments underwent an end repair process followed by phosphorylation and then ligation of adapters. These products were subsequently purified and amplified by PCR to create cDNA libraries. The cDNA libraries were processed and enriched on a OneTouch 2 instrument using Ion PI™ Template OT2 200 Kit (Life Technologies, Carlsbad, CA, USA) to prepare the Template-Positive Ion PI™ Ion Sphere™ Particles. After enrichment, the mixed Template-Positive Ion PI™ Ion Sphere™ Particles were finally loaded on the Ion PI™ Chip and sequenced using the Ion PI™ Sequencing 200 Kit (Life Technologies, Carlsbad, CA, USA). Bioinformatics data analyses of the RNA-seq libraries were performed by Shanghai Novelbio Ltd. as previously described [23].

Quantitative real-time PCR
The plant tissue samples for quantitative real-time PCR (qPCR) were collected from different plants of the same batch of rice plants that were sampled for RNA-seq experiments. In brief, 500 ng of total RNA was reverse transcribed using a first-strand cDNA synthesis kit (Promega, Madison, WI, USA), digested with DNase I (Thermo Fisher Scientific, Waltham, MA, USA), and then diluted 50X. The qPCR reaction was performed using SYBR Premix Ex Taq Ready Mix with POX reference dye (Takara Biotech, Kyoto, Japan) and an ABI 7500 Real-time PCR Detection System instrument (Applied Biosystems Foster City, CA, USA). The thermocycler setting was as follows: 30 s at 95°C, followed by 40 cycles of 5 s at 95°C and 34 s at 60°C. To confirm the formation of single peaks and to exclude the possibility of primerdimer and non-specific product formation, a melt curve (15 s at 95°C, 60 s at 60°C, and 15 s at 95°C) was generated by the end of each PCR reaction. Primer pairs were designed using Beacon Designer software (Premier Biosoft, version 7.0) and are listed in Additional file 1: Table S1. The relative fold-changes of gene expression were calculated using the comparative 2 −ΔΔCT method [24] and were normalized to the housekeeping gene ubiquitin 5 [25]. All qPCR reactions were repeated in three biological and four technical replications.

Analyses of differentially expressed genes (DEGs)
RNA-seq read quality values were checked using FAST-QC (http://www.bioinformatics.babraham.ac.uk/projects/ fastqc/). The reads were mapped to the reference rice genome of the Michigan State University (MSU) Rice Genome Annotation Project database (RGAP, V7.0) (http://rice.plantbiology.msu.edu/) [26] using MapSplice software [27]. The DEGSeq algorithm [28] was used to filter DEGs. Reads per kilobase of exon model per million mapped reads (RPKM) were used to explore the expression levels of the DEGs [29], and an upper quartile algorithm was applied for data correction. False discovery rate (FDR) was used for the correction of data occur in multiple significant tests [30]. Genes whose expression differed by at least two-fold (log 2 (fold change) > 1 or < −1, FDR < 0.05) were regarded as DEGs as determined with the R statistical programming environment (http://www.rproject.org). The DEGs in rice plants that had been fed by caterpillars for 24, 48, or 72 h were, respectively, compared to those that had never been fed using MapMan software to get an overview of the metabolism [31]. Venn diagrams were generated using these DEGs to identify common and unique genes affected by C. suppressalis among different time points [32]. Time Series-Cluster analysis, based on the Short Time-series Expression Miner (STEM) method (http://www.cs.cmu.edu/~jernst/stem/) [33], was used to identify the global trends and similar temporal model patterns of the expression of the total DEGs.

Phytohormone signature analyses
Hormonometer program analyses [34] (http://hormono meter.weizmann.ac.il/) was used to assess the similarity of the expression of rice genes induced by C. suppressalis with indexed data sets of those elicited by exogenous application of phytohormones to Arabidopsis as previously described [7]. The rice genes were blasted to the Arabidopsis thaliana genome. The Arabidopsis gene identifies (AGI) were converted to Arabidopsis probe set identifies using the g:Convert Gene ID Converter tool [35] (http://biit.cs.ut.ee/gprofiler/gconvert.cgi). Only genes included in RNA-seq containing Arabidopsis probe set identifies were kept for analyses. In some cases, there were two probe sets for one AGI, while in few cases there were two AGIs for one probe set. This indicates that lines were duplicated and sets were thus discarded.

Gene ontology (GO) and pathway enrichment analyses
DEGs belonging to different classes were retrieved for GO and pathway analysis. GO analysis was conducted using the GSEABase (gene set enrichment analysis base) package from BioConductor (http://www.bioconductor.org/) based on biological process categories (Fisher's exact test, FDR < 0.001). Pathway analyses were conducted to elucidate significant pathways of DEGs according to the Kyoto Encyclopedia of Gene and Genomes (KEGG) (http://www.genome.jp/kegg) databases. Fisher's exact test followed by Benjamini-Hochberg multiple testing correction was applied to identify significant pathways (P < 0.05).

Metabolome analyses
Samples were prepared using the automated Microlab STAR ® system (Hamilton Company, Bonaduz, Switzerland) and were analyzed using ultrahigh performance liquid chromatography-tandem mass spectroscopy (UHPLC-MS) and gas chromatography-mass spectrometry (GC-MS) platforms by Metabolon Inc. (Durham, North Carolina, USA). These platforms have been previously described [36,37]. In brief, a recovery standard was added before the first step in the extraction process for quality control purposes. Protein fractions of the samples were removed by serial extractions with methanol. The samples were subsequently concentrated on a Zymark TurboVap® system (KcKinley Scientific, Sparta, NJ, USA) to remove the organic solvent and then were vacuum dried. The resulting samples were divided into five fractions, and they were used for analyis by: i) UHPLC-MS with positive ion mode electrospray ionization, ii) UHPLC-MS with negative ion mode electrospray ionization, iii) UHPLC-MS polar platform (negative ionization), iv) GC-MS, and v) for being reserved for backup, respectively. Before the UHPLC-MS analysis, the subsamples were stored overnight under nitrogen. For GC-MS analysis, each sample was dried under vacuum overnight. UHPLC-MS and GC-MS analyses of all samples were carried out in collaboration with Metabolon Inc. as previous described [36,37].
For statistical analysis, missing values were assumed to be below the limits of detection, and these values were inputted with a minimum compound value [37]. The relative abundances of each metabolite was log transformed before analysis to meet normality. Dunnett's test was used to compare the abundance of each metabolite between different time points. Statistical analyses were performed using the SPSS 22.0 software package (IBM SPSS, Somers, NY, USA).

Global transcriptome changes in rice plants during Chilo suppressalis infestation
A total of 16 libraries (four biological replicates of four sampling times) were conducted, resulting in approximately 29-41 million clean reads; GC content accounted for 48-53% of these reads (Additional file 2: Table S2). The average number of reads that mapped to the rice reference genome was > 87%, and unique mapping rates ranged from 73 to 87% (Additional file 2: Table S2). The unique matching reads were used for further analysis. Gene structure analysis showed that most of the mapped reads (61-73%) were distributed in exons (Additional file 3: Table S3). RNA-seq data were normalized to RPKM values to quantify transcript expression. In total, 42,100 genes were detected in all samples (Additional file 4: Table S4). Only significantly changed genes with P < 0.05 (FDR) and fold-change > 2 or < 0.05 were considered to be differentially expressed genes (DEGs), resulting in a total of 4,729 DEGs at a minimum of two time points (Fig. 1, Additional file 5: Table S5 and Additional file 6: Table S6). A comparison of DEGs at the different time points relative to the control (24 h vs. 0 h, 48 h vs. 0 h, and 72 h vs. 0 h) revealed over one thousand genes with significantly altered expression levels, with more genes being up-regulated than down-regulated (Fig. 1a). MapMan analyses showed that the up-regulated DEGs in rice plants between different time-point (24, 48, or 72 h) and the control (0 h) were mainly involved in cell wall, lipid and secondary metabolism. While the down-regulated DEGs mainly involved in light reactions (Additional file 7: Figure S1). A Venn Diagram of this data set indicated that 1,037 genes were differently expressed at all 3 time points of 24, 48, and 72 h relative to 0 h (Fig. 1b). However, much lower number of DEGs detected between the time points of 24 h vs. 48 h, 24 h vs. 72 h, or 48 h vs. 72 h and there was no commonality of the DEGs occurred between two of three time points (Fig. 1a, c).
The expression patterns of selected genes were confirmed by qPCR using the rice stem samples from the same batch of rice plants that were used for RNA-seq. A total of 20 genes were selected related to the signaling of phytohormones, primary metabolism, and secondary metabolism. The expression profiles of most genes tested by qPCR were consistent with those analyzed by RNA-seq although only one housekeeping gene was used in qPCR analysis (Fig. 2), which indicated the validation of the results from our transcriptome experiment.

Series-cluster and enrichment analyses
To refine the sets of genes that were differently expressed at a minimum of two time points, we used the STEM method, which is commonly used for the cluster of gene expression in transcriptomic studies [33]. The 4,729 DEGs were clustered into 26 possible model profiles ( Fig. 3; Additional file 6: Table S6). Based on the expression dynamics of these DEGs, their expression patterns were assigned to five classes (Additional file 6: Table S6). Class I included 2,122 genes that showed a trend of up-regulated expression during the 72-h of larval feeding. Class II contained 1,318 genes showing a trend of down-regulated expression. Class III contained 873 genes that were up-regulated at early stage, but down-regulated at later stage. Class IV included 222 genes that were down-regulated at early stage but up-regulated at late stage. Class V contained the remaining 194 genes with irregular expression profile. GO analyses indicated that the number of significant GO terms with biological process categories in the five classes were 85, 47, 48, 2, and 5, respectively (Additional file 8: Table S7). This indicates that most DEGs involved in the response to C. suppressalis damage contained in the first three classes. More details of the GO analyses for these DEGs are provided in Additional file 8: Table S7. Pathway enrichment analyses showed that genes in class I are mainly related to pathways of biosynthesis of plant secondary metabolites, plant hormone signal transduction, nitrogen metabolism, galactose, and terpenoid (Table 1). Genes in class II are mainly involved in primary metabolism such as nucleotide metabolism and photosynthesis, which may indicate the repressed activity of photosynthesis and the increased catabolism of nucleic acids. Genes in class III are mainly involved in pathways of biosynthesis of secondary metabolites including glucosinolate and phenylpropanoids and the metabolism of carbohydrates such as galactose, fructose, and mannose. The genes in class IV are mainly related to the metabolism of starch and sucrose, and to the biosynthesis of photosynthesis-antenna proteins, flavone, and flavonol. The genes in class V are mostly involved in secondary metabolism.

Phytohormone-related DEGs
A total of 9,221 Arabidopsis orthologs of rice genes were included in the Hormonometer analyses (Additional file 9: Table S8). Changes in gene expression induced by C. suppressalis in rice were positively correlated with those induced by SA (salicylic acid), JA (jasmonic acid), ABA (abscisic acid), and auxin treatments in Arabidopsis (Fig. 4). The changes in gene expression were negatively correlated with genes associated with cytokinin (CTK) signatures. These patterns were generally supported by GO analyses of the five classes (Additional file 8: Table S7).

Metabolome composition analyses
A total of 151 known metabolites were detected and quantified in rice plants during the 96 h of larval feeding (Additional file 11: Table S10). By mapping the general biochemical pathways based on KEGG and plant metabolic network (PMN), we divided the metabolites into seven classes, of which amino acids were the most prevalent (33% of the metabolites), followed by carbohydrates (29%) (Additional file 12: Figure S2). The secondary metabolites accounted for 7% (Additional file 11: Table S10; Additional file 12: Figure S2).

Integrated analyses of the transcriptomic and metabolic data sets Biosynthesis of aromatic amino acids, salicylic acid, and phenylpropanoids
The shikimate pathway is a major pathway in plants and is responsible for the biosynthesis of the aromatic amino acids Phe, Tyr, and Trp, as well as of auxin, SA, lignin, and phenylpropanoid [39]. Integration of the transcriptomic and metabolic data revealed that transcriptional up-regulation of the genes was accompanied by the elevation of the main metabolites in the pathways (Fig. 5; Additional file 13: Table S11). For example, all of the genes encoding the crucial enzymes in the shikimate pathway that accumulated throughout the 72 h of larval feeding belong to class I containing up-regulated DEGs (Fig. 5).

Chilo suppressalis-induced changes in carbohydrate metabolism
As products of photosynthesis, carbohydrates are the main source of stored energy in plants. Most DEGs involved in carbohydrate metabolism were up-regulated (Fig. 6b), with an exception of the genes encoding trehalose 6-phosphate synthase (TPS) and 4-alpha-glucanotransferase (AGLS). Consistently, metabolic analysis showed that except for oligosaccharides and galactinol, all monosaccharides (orbitol, galactitol, glucose, fructose, and xylose) increased over time ( Fig. 6c; Additional file 11: Table S10).

Effects of Chilo Suppressalis feeding on amino acids, organic acids, and nitrogen metabolism
Our analyses showed that genes encoding enzymes such as glutamate decarboxylase (GAD), N-carbamoylputrescine amidase (CPA), ornithine decarboxylase (ODC), and Laspartate oxidase (LASPO) were up-regulated; while those encoding adenylosuccinate lyase (ASL), and delta-1-pyrroline-5-carboxylate synthetase (P5CS) were down-regulated over time. As expected, the contents of metabolites ornithine, gamma-aminobutyrate and putrescine increased, while the levels of aspartate and spermidine decreased in rice plants during C. suppressalis feeding due to action of the enzymes mentioned above (Fig. 7a, b). In addition, we also detected increased levels of other amino acids such as Pro, Ala, and Asn (Fig. 7c).

Discussion
The current study describes the first effort to combine transcriptomic and metabolic techniques for the comparative analyses of the genes and the metabolites involved in rice plant responses to damage caused by C. suppressalis larvae. The results increase our understanding of the mechanisms underlying the dynamic responses of rice plants to caterpillar feeding.
Gene expression analyses revealed that more DEGs were up-regulated than down-regulated in response to feeding by C. suppressalis larvae. This is consistent with   4 Hormonometer analysis of differential gene expression in rice in response to Chilo suppressalis feeding. The response in gene expression in rice to Chilo suppressalis feeding (for 0, 24, 48, or 72 h) treatments was compared with that of Arabidopsis at 30, 60, and 180 min, or 3, 6, and 9 h after hormone application. Red shading indicates a positive correlation between the rice response to a C. suppressalis treatment and the Arabidopsis response to a hormone treatment; blue shading indicates a negative correlation. MJ, methyl jasmonate; ACC, 1-aminocyclopropane-1-caroxylic acid (a metabolic precursor of ethylene); ABA, abscisic acid; IAA, indole-3-acetic acid; GA3, gibberellic acid 3; BR, brassinosteroid; and SA, salicylic acid previous findings concerning aphid-infested maize [7] and maize that was mechanically wounded and then treated with the oral secretions of Mythimna separata [9]. Similarly, more DEGs were up-regulated than downregulated when Arabidopsis plants were individually infested with Myzus persicae, Brevicoryne brassicae, Spodoptera exigua, or Pieris rapae [40], or when cotton was damaged by the chewing insects Helicoverpa armigera or Anthonomus grandis [41]. However, there were also studies reporting that more DEGs were down-regulated than up-regulated, or the numbers of up-and down-regulated DEGs were equivalent when rice plants were damaged by C. suppressalis [8] or the brown planthopper N. lugens [42,43], or when cotton plants were infested with the whitefly Bemisia tabaci or the aphid Aphis gossypii [6,44]. This variability might be explained by differences in herbivore species, plant species, plant tissues infested, the duration of infestation, and the techniques used for the detection of gene expression [40].
As the key regulators of transcription, TFs are important in plant responses to herbivory [5,8,[45][46][47]. In our transcriptome analyses, we identified 385 TF genes that responded to C. suppressalis feeding, suggesting that the induced defense response is complex and involves a substantial change in rice metabolism. The TF families whose expression was most altered by C. suppressalis feeding were AP2-EREBP and WRKY. Evidence increasingly indicates that WRKYs play significant roles in plant development and in responses to biotic and abiotic stresses [5,8,[45][46][47], and members of the AP2-EREBP family mediate defense against biotic and/or abiotic stress [45]. For example, it was recently found that OsWRKY70 mediates the prioritization of defense over growth by positively regulating cross-talk between JA and SA when rice is attack by C. suppressalis [47], and OsWRKY53 is a negative regulator of plant growth and an early suppressor of induced defenses [46], both of which belong to WRKY family. The function of TFs in Phytohormones play important roles in a complex regulatory network that is essential for herbivore-induced response as previously reported [1,4,48] and as also indicated by our Hormonometer analysis. Our results showed that C. suppressalis elicited the expression of genes associated with JA and SA, which is consistent with a previous study [8]. In turn, exogenous application of methyl JA or JA to rice plants reduced the performance of two root herbivores, the cucumber beetle Diabrotica balteata and the rice water weevil Lissorhoptrus oryzophilus [49], and induced the release of volatiles that attract parasitoids [50]. SA, which is a central phytohormone in the shikimate pathway, plays an importance role in the defense against biotrophic pathogens and piercing/sucking insects [1]. Our data showed that a number of rice SArelated genes were up-regulated by C. suppressalis larval feeding (Fig. 5b). Although studies have reported that crosstalk between JA and SA is negative in Arabidopsis [51], and that JA-dependent defense may be hampered by SA and vice versa [5,19], our findings are consistent with the evidence that SA and JA can have overlapping or even synergistic effects in rice [8,51].
We found that changes in gene expression induced by C. suppressalis in rice were positively correlated with changes induced by ABA treatment in Arabidopsis, which agrees with previous results in several plant-insect systems [5,7,9,40,44]. The role of ABA in regulating defense against pathogens in rice has been well documented [51], Fig. 6 Expression patterns of Chilo suppressalis-induced genes and metabolites involved in typical carbohydrate metabolism. a Typical carbohydrate metabolism pathway schematic. Uppercase letters are genes that encoded enzymes. Metabolites shaded in green were measured. Solid arrows represent established biosynthesis steps, while broken arrows indicate the involvement of multiple enzymatic reactions. RFS, raffinose synthase; GAL, alpha-galactosidase; BF, beta-fructofuranosidase; AGL, alpha-glucosidase; SUS, sucrose synthase; TREH, alpha, alpha-trehalase; PMI, mannose-6-phosphate isomerase; TPS, trehalose 6-phosphate synthase; PFK, 6-phosphofructokinase 1; PFPA, pyrophosphate-fructose-6-phosphate 1-phosphotransferase; FBA, fructose-bisphosphate aldolase, class I; AGLS, 4-alpha-glucanotransferase. b Heatmap of relative expression levels of the genes involved in the schematic pathway. The heatmap was generated from the RPKM data using MeV (V4.9.0). c Metabolite abundance after C. suppressalis infestation; values are means ± SE (n = 10). *, P < 0.05 by Dunnett's test relative to uninfested controls but its role in resistance to insects is much less understood. Our results suggest that ABA signature may also play a vital role in rice defense against insect herbivores, although researchers recently reported that applying ABA to rice roots did not affect the performance of D. balteata and L. oryzophilus [49]. We supposed that ABA may function in other ways in rice plant defense against herbivory, but further studies are needed for clarifying this hypothesis. In contrast, we found a negative correlation between CTK-induced and C. suppressalis-induced gene expression (Fig. 4). This negative correlation, which has been also observed in other plant species [7,34,52], may reflect the decrease in growth rate of rice plants caused by C. suppressalis infestation.
Insect infestation causes many changes in both primary and secondary metabolism, and the reconfiguration of metabolism is a common defense strategy [11,48,53]. Our MapMan analyses and GO and pathway enrichment analyses indicate that rice plants reprogram both primary and secondary metabolism in response to C. suppressalis feeding (Table 1; Additional file 7: Figure S1 and Additional file 8: Table S7). Reductions in photosynthesis, as indicated by down-regulation of photosynthesis-related genes, is a common response to Fig. 7 Expression patterns of Chilo suppressalis-induced genes and metabolites involved in the metabolism of amines and polyamines and amino acids from the glutamate and aspartate family. a Pathway schematic of amino acid metabolism. Uppercase letters are genes that encoded enzymes. Metabolites shaded in green were measured. Solid arrows represent established biosynthesis steps, while broken arrows indicate the involvement of multiple enzymatic reactions. GDH, glutamate dehydrogenase; GAD, glutamate decarboxylase; GS, glutamate synthase; ODC, ornithine decarboxylase; PAO, polyamine oxidase; CPA, N-carbamoylputrescine amidase; ASL, adenylosuccinate lyase; ADH, aldehyde dehydrogenase; LASPO, L-aspartate oxidase; and P5CS, delta-1-pyrroline-5-carboxylate synthetase. GABA, gamma-Aminobutyric acid; GGS, L-glutamate gammasemialdehyde. b Heatmap of relative expression levels of the genes involved in the schematic pathway. The heatmap was generated from the RPKM data using MeV (V4.9.0). c Metabolite abundance after C. suppressalis infestation; values are means ± SE (n = 10). *, P < 0.05 by Dunnett's test relative to uninfested controls insect feeding [5,8,11,40,53] what was also confirmed in the current study. The down-regulation of photosynthetic genes accompanied by the up-regulation of defenserelated genes may allow rice plants to redirect resources toward defense.