Integrative proteomic and metabolomic analysis reveals the 1 cotton plant defense mechanisms induced by insect 2 ( Adelphocoris suturalis Jakovlev ) feeding 3 4

43 Cotton (Gossypium hirsutum Linn.) is widely cultivated in China. The polyphagous insect 44 Adelphocoris suturalis (Jakovlev) is a serious insect pest in cotton growing regions. Plants have 45 evolved sophisticated systems to cope with herbivore attacks. However, the cotton defense 46 mechanisms induced by A. suturalis feeding have lagged behind. We carried out untargeted 47 proteomic analysis using the iTARQ technique and metabolomics based on LC-MS/MS analysis 48 of cotton leaves fed upon by A. suturalis. Proteomic analysis identified 775 upregulated proteins 49 and 477 downregulated proteins in plants that were infested by A. sututralis compared to the 50 controls. Metabolomic analysis identified 50 differentially expressed metabolites in the positive 51 ion mode and 14 in the negative ion mode compared to the controls. The tryptophan metabolism 52 pathway was significantly changed in both the positive and negative ion mode in the 53 metabolomics analysis. The alpha-linolenic acid pathway was significantly changed in both the 54 proteomic and metabolomics analyses. Furthermore, the result was validated by RT-qPCR 55 analysis of 5 related genes involved in alpha-linolenic acid pathway. These results indicate that 56 tryptophan metabolism and the alpha-linolenic acid pathway may be important in cotton defense 57 against herbivores and would enhance our understanding of plant defenses induced by A. sututrali 58 feeding. 59 60


Introduction
Plants have successfully colonized most environments where herbivores are common because they have evolved sophisticated systems to cope with herbivore feeding [1].Some plants can release volatile organic chemicals (VOCs) after herbivore infestation [2], while other plants have developed structural defenses, such as spines, trichomes, and thick, tough leaves [3].When plants receive physical and chemical cues from herbivorous insects, such as elicitors in oral secretions and compounds in oviposition fluids, they can alter their proteins and metabolites.Plant defense responses induced by herbivores can occur in both wounded and in undamaged regions [4].Herbivores, in turn, have responded to plant defenses by evolving counter adaptations that make plant defenses less effective or render them useless.This can lead to an evolutionary ''arms race'' between plants and their herbivore enemies [5,6].
Cotton (G.hirsutum) is an important cash crop worldwide.In China, the development and widespread adoption of transgenic Bt (Bacillus thuringiensis) cotton has led to a substantial reduction in the use of broad-spectrum insecticides.This in turn has led to frequent outbreaks of mirids [7][8][9].Two species in the Miridae, Adelphocoris suturalis and Apolygus lucorum, are emerging as the most destructive pests in the major cotton growing regions.These species are highly polyphagous and attack a broad range of cultivated crops, such as cotton, beans, alfalfa, vegetables, and fruit crops [10].Both the nymphs and adults of these species suck plant sap from cotton flower buds, tender shoots, and buds, resulting in abscission, wilting, abnormal growth, and losses in lint yield and quality [11].
The detection and quantification of multiple proteins and metabolites can be accomplished using techniques such as LC-Q/TOF-MS-based metabolomics [12][13][14][15].Quantitative iTRAQ-LC-MS/MS proteomics can be a sensitive method for high-throughput protein identification and quantification [16].It has been used in insects and plants, including the silkworm [17], brown planthopper [18], pine beetle [19], locust [20], and cotton [21,22].These techniques can provide a global view of dynamic proteomic and metabolomic variation and allow for the discovery of key proteins and metabolites that are essential for plant defenses.However, there have been few reports on cotton proteins and chemical metabolite changes after attacks by A. suturalis.In this study, we used a proteomic and metabolomic approach to study cotton plant defenses in response to A. suturalis feeding.Tryptophan metabolism and alpha-linolenic acid metabolism are important pathways that regulate plant development and defense responses induced by pathogens and insects [23][24][25][26].We evaluate the new data and discuss how the tryptophan metabolism and alpha-linolenic acid pathways play essential roles in cotton defenses.The present study provides a new insight into the molecular mechanism of plant defense induced by insect feeding.2. Results 2.1.iTRAQ identified the different proteins in the cotton treatment with A. suturalis infested for 48 h and the control without insect pressure Cotton proteomic analysis by iTRAQ was performed on three replicates as previously described [21,22,[27][28][29].A total of 371575, 374246, and 373127 spectra were generated in the treatments of P48E, and the two controls of P48C and P0C.P48E represents plants infested with insects for 48 h; P48C represents plants grown for 48 h without insect infestation; and P0C represents experimental cotton plants without insect infestation.We obtained 52750, 48103, and 52474 unique spectra, 17865, 16116, and 16483 identified peptides, including 12906, 11903, and 12135 unique peptides and 5520, 5210, and 5313 identified proteins from the cotton in P48E, P48C, and P0C, respectively, with a false discovery rate (FDR) of <1% (Table S1).And finally, a total of 8302 proteins were identified (Table S1).

Functional categories of differentially expressed proteins
In this study, any proteins with a ≥ 1.2-fold difference and a q-value <0.05 were designated as significant differently expressed proteins (DEPs) [30][31][32][33][34].We got q-value through p-value corrected for false discovery rate (FDR) by Benjamini-Hochberg (BH).A total of 1015 (56%) proteins were upregulated and 796 (44%) were downregulated in P48E compared to the control P0C.A total of 775 (62%) proteins were upregulated and 477 (38%) were downregulated in P48E compared to the control P48C.There were 485 (48%) proteins that were upregulated and 520 (52%) were downregulated in P48C compared to the control P0C (Fig. 1A, Table S2).Venn diagram shows common or uniquely up-and down-regulated proteins in different experimental groups.(Fig. 1B, 1C).The remaining pathway enrichment categories are shown in Table 1.The results suggest that plants can alter their proteins when attacked by herbivorous insects, and then plant defense responses are induced.
For analysis of the metabolic pathways of the cotton plants infested by A. suturalis, DEPs were also investigated using the KEGG database (ver.81).We first compared the two control groups, P48C and P0C (Table S3), and then removed the pathway changes arising from plant growth.These changes were considered as background noise.Then, we compared experimental P48E with the control P48C, and found that the DEPs were enriched in alpha-linolenic acid metabolism (1.85%), fructose and mannose metabolism (2.53%), amino sugar and nucleotide sugar metabolism (4.09%), selenocompound metabolism (1.17%), protein digestion and absorption (0.49%).The remaining pathway enrichment categories are shown in Table 1.Those results indicated that metabolic pathway may play an important role in plant defense.2.3.Metabolome changes in cotton plant in response to A. suturalis feeding Reproducibility of the UPLC-Q-TOF-MS was determined from ten replicates with the same quality control (QC) sample interspersed throughout the analysis (Fig. 2A).In all of the QC samples, ions with relative standard deviation (RSD) > 30% were deleted.We then had 7513 positive ions (RSD ≤ 30%, 96.44%) of the total 7790, and 4053 negative ions (RSD ≤ 30%, 93.71%) of the total 4325.To investigate cotton metabolic changes in response to A. suturalis feeding, all of the observations, acquired in both positive and negative ion modes, were analyzed using two components principal component analysis (PCA) score (Fig. 2B).To best analyze the metabolic variations of the A. suturalis feeding groups, all of the observations acquired in both ion modes were analyzed using orthogonal partial least squares-discriminant analysis (OPLS-DA).The differential metabolites were selected according to the variable important for the projection (VIP) threshold (VIP > 1) in the OPLS-DA model with the q-value (q < 0.05) after FDR correction [32][33][34].The plots of the OPLS-DA model discriminated the insect feeding groups from their corresponding control groups, and exhibited satisfactory classification (Fig. 2C).Under this standard, there were 70 (18 upregulated and 52 downregulated) metabolites in the positive ion mode and 15 (all downregulated) in the negative ion mode (Table S4).A clear metabolite separation was observed between the A. suturalis feeding groups and the corresponding control groups as illustrated in a heat-map (Fig. 3, Table S4).The results suggest that metabolites were downregulated and this may induce cotton' defense against herbivorous insects.
To further analyze the cotton metabolic pathways, different metabolites were studied using the KEGG database.Comparing the insect feeding group with the control group, in positive ion mode, the top 6 pathways were metabolic pathways, biosynthesis of secondary metabolites, sesquiterpenoid and triterpenoid biosynthesis, tryptophan metabolism, isoquinoline alkaloid biosynthesis, tropane, piperidine and pyridine alkaloid biosynthesis.In the negative ion mode, the top 6 pathways were metabolic pathways, biosynthesis of secondary metabolites, alpha-linolenic acid metabolism, tryptophan metabolism, phenylpropanoid biosynthesis, isoquinoline alkaloid biosynthesis.The metabolites in metabolic pathways, biosynthesis of secondary metabolites, alpha-linolenic acid metabolism and tryptophan metabolism were significantly changed in both positive and negative ion modes (Table 2).

Tryptophan metabolism in cotton after feeding by A. suturalis
We integrated positive and negative ion mode data to analyze the tryptophan metabolism pathway in the treatment cotton feeding by A. suturalis compared to the control cotton.We found that there were 21 metabolites including 18 metabolites in positive ion mode and 3 metabolites in negative ion mode were downregulated compared to the control group; while only 4 metabolites in positive ion mode were upregulated compared to the control group; and there was 1 metabolite in negative ion mode was downregulated but it was upregulated in positive ion mode (Fig. 4A; Table 3).The results showed that among 26 changed metabolites in tryptophan metabolism pathway, 21 (80.8%)metabolites were downregulated and only 4 (15.4%)metabolites were upregulated, these results indicated that most of metabolites in tryptophan metabolism pathway were downregulated and this may be an important response for cotton to defend against herbivores, the results described here in Fig. 4B.2.5.Integrating proteomics and metabolomics data to analyses alpha-linolenic acid metabolism pathway Interestingly, we found that the alpha-linolenic acid metabolism pathway was significantly changed in both the proteomic and metabolomic analyses (Table 1, 2).We investigated the pathway enrichment of P48E vs P48C in both omics.Most of the metabolites were downregulated (Fig. 5, blue rectangular box), except for volicitin and 3-hexenol (Fig. 5, red rectangular box), while most related proteins were upregulated (Fig. 5, red elliptic box).These proteins included allene oxide synthase (AOS), cytochrome P450 allene oxide synthase (CytP450), phospholipase A (PLA), alcohol dehydrogenase (ADH), allene oxide cyclase 4 (AOC4), 12-oxophytodienoate reductase (OPR), AMP-dependent CoA ligase, acyl-CoA oxidase 4 (ACO4), acyl-CoA oxidase 3 (ACO3), and 3-ketoacyl-CoA thiolase 2(3-KAT-2) (Table S5).This finding indicated proteins as the upstream regulator increased led to the downstream metabolites deceased in alpha-linolenic acid metabolism pathway.The reconfiguration of proteins and metabolites may be one of the ways that plants cope with insect-feeding stress.There were several unknown proteins that may also play important roles in alpha-linolenic acid metabolism (Fig. 5).We selected 5 proteins including AOS, PLA1, AOC4, and two OPR proteins named OPR1 and OPR2, which were related to the alpha-linolenic acid metabolism pathway (Table S5), and performed quantitative reverse transcription PCR (RT-qPCR) analysis.Two housekeeping genes [35] (GhHis3 with GenBank accession AF024716 and GhUBQ7 with GenBank accession DQ116441) were used.The RT-qPCR analysis validated our sequencing results (Fig. 6).

Discussion
Plants have evolved sophisticated systems to cope with herbivore attacks.When plants perceive herbivore-derived physical and chemical cues, they can dramatically reshape their transcriptomes, proteomes, and metabolomes.These responses involve specific changes in metabolism, gene expression, and in the pattern of plant growth and development [36][37][38].
Secondary metabolic compounds of plants are an important biochemical basis for plant resistance to insects [39].Research with many plant species has revealed a great variety of small molecules with toxic or antifeedant effects on insect herbivores.For example, many terpenoids, the most metabolically diverse class of plant secondary metabolites, play important roles in plant defenses.Alkaloids (e.g., caffeine, nicotine, morphine, strychnine, and cocaine) are also secondary plant metabolites that help protect plants from herbivores [40].Other well-studied classes of plant secondary metabolites with defensive properties include furanocoumarins, cardenolides, tannins, saponins, glucosinolates, and cyanogenic glycosides [41,42].Cotton plants, coping with the stress from insect feeding, have evolved numerous inducible defense mechanisms that help them respond to biotic stress, including the synthesis of volatile terpenes, phytoalexins, gossypol, tannins, tyloses, pathogenesis-related proteins, as well as lignification, and the release of active oxygen species [39,[42][43][44][45].
Tryptophan is the amino acid metabolic precursor of many important secondary metabolites.Tryptophan biosynthesis plays a direct role in regulating plant development, pathogen defense responses, and plant-insect interactions [25,[44][45][46][47].In plants, several important secondary metabolites are derived from tryptophan or its indolic precursors.These include the plant growth regulator indole-3-acetic acid (IAA) and the pathogen defense compounds indole glucosinolates (IGs) and indolic phytoalexins [45].In the present study, we examined the molecular responses of cotton to insect feeding stress using metabolomics based on LC-MS/MS analysis.Among changed metabolites associated with the tryptophan metabolism pathway we found that there were 5 upregulated and 22 downregulated.Tryptophan metabolism is mainly downregulated and this may promote the effectiveness of the plant immune system.Jasmonate (JA) modulates numerous physiological processes that are related to plant development and defense responses [24,25].Oxygenation of alpha-linolenic acid is the initial step in JA biosynthesis [25].Alpha-linolenic acid is a stress signal released by lipase activity on chloroplast membranes.It is the substrate for numerous oxygenated compounds collectively called oxylipins, including JA, which comprise JA, MeJA (methyl jasmonate), JA amino acid conjugates, and further JA metabolites [4,24].It can be released into the plastid under stress conditions [4,24].The related proteins in the alpha-linolenic acid metabolism pathway AOS, AOCs (AOC1, 2, 3, and 4), and OPR3(12-Oxophytodienoate reductase 3) are key protiens involved in the synthesis of JA.AOSs belong to the CYP74A enzyme family [48].The AOC is the subsequent enzyme of the AOS branch [49].OPR3 belong to a family of flavoproteins identified first with Warburg's old yellow enzyme (OYE), and then from 13-hydroperoxylinolenic acid (13-HPOTrE) [24].This fatty acid hydroperoxide is then dehydrated by AOS and cyclized by AOC to the cyclopentenone 12oxo-phytodienoic acid (12-OPDA) [24].OPDA is a potent gene regulator in the wound response and can protect plants against the attack of insect or fungal pathogens when JA is absent [50].In our present study, we integrated proteomic and metabolomic data of the alpha-linolenic acid metabolism pathways and we found that proteins, such as AOS, AOC4, OPR1, OPR2, were upregulated, and this is similar to the result which were found in the rice feeding by stem borers [51], while in metabolites, such as 13(s)-HpOTrE,9,10-EoTrE, colnelenic acid, O-OPDA,12-OPDA, and 9(s)-HpOTrE, were downregulated.This result indicated that proteins as the upstream regulators increased and led to a decrease in downstream metabolites.Volicitin (N-(17hydroxylinolenoyl)-L-glutamine) was first isolated from oral secretions of beet armyworm caterpillars [52].The linolenic acid derivative volicitin induces maize (Zea mays L.) seedlings to release volatile compounds (terpenoids and indole) that are similar to those released from plants damaged by caterpillar feeding [52].And in our present result, volicitin were upregulated in the alpha-linolenic acid pathway after feeding by A. suturalis, this suggest that volicitin may function the cotton defense.
We chose 5 proteins and performed RT-qPCR, and we found the expression of those genes validated our sequencing results.At the beginning of the insect infested time (3-6h), all of the genes are at a low expression level, and then at 24 h and 48 h, there is a peak expression level, after that, the expression begin to decrease (Fig. 6).This result suggest that plant can regulate their proteins to cope with insect feeding.Finally, a better understanding of why the plant reshapes its proteins and metabolites after insect attack is important to explain the relationship between the plant and the insect.Future research should focus on the unknown proteins in the alpha-linolenic acid metabolism pathway.

Plant Material and insect infestation
Potted cotton plants (G.hirsutum) were soil grown in controlled-environment chambers under a regime of either a 10-h (short-day) or 16-h (long-day) light period at 25°C and 65% relative humidity, unless otherwise indicated.After 3 weeks of growth, cotton plant was transferred into one 30-cm square cage, and 9 adult bugs were released on the cotton leaves.Our preliminary experiment indicated that 24 h and 48 h were the optimal termination time and this result was similar to the rice feeding by stem borers [51].After 48 h of insect infestation, all of the insects were removed.The leaves were immediately frozen in liquid nitrogen and stored at -80°C until use.For proteomic sequencing, each treatment had three experimental replicates, with P48E representing the plants that were infested with insects for 48 h.The P48C control plants were not infested but simply grown for 48 h.The P0C controls were the experimental cotton plants without insect infestation.And for RT-qPCR analysis, we selected 5 proteins and performed 6 experiments, P3E, P6E, P12E, P24E, P48E and P72E, which represent 3 h, 6 h,12 h, 24 h, 48 h, and 72 h of insect infestation, respectively.The control plants were not infested but simply grown for 3 h, 6 h,12 h, 24 h, 48 h, and 72 h, respectively.The main leaf and cotyledon were immediately frozen in liquid nitrogen, respectively, and stored at -80°C until use.For metabolomic sequencing, each treatment had ten experimental replicates, with group 1 representing plants that were infested with insects for 48 h and group 2 representing plants without insect infestation but grown for 48 h as the controls.

Protein preparation and iTRAQ labeling
The iTRAQ analysis was conducted at BGI (Shenzhen, China).Total proteins were extracted from cotton plants using a previously reported phenol extraction procedure [54].Protein concentrations were determined using the Bradford method [55].Three independent biological replicates were performed in the experiment.A total of 25 μg total protein from each sample was used for each experiment.Protein was digested by sequencing grade trypsin (Promega) at a ratio of 1:10 (w:w) for12 h at 37°C, and then labeled using iTRAQ 8-plex kits (AB Sciex Inc., MA, USA) according to the manufacturer's instructions.The samples were labeled with iTRAQ tags 113 (P48E, 48 h experiment), 119 (P48C, 48 h CK), and 121 (P0C, 0 h CK), respectively.4.4.LC-MS/MS analysis LC-MS/MS analysis was performed as described previously [21,22].After labeling all of the samples mixing, HPLC separation, and LC-MS/MS analysis.For MS, balance group can show the same m/z no matter which report ion label peptide.In MS2, neutral loss happened to the balance group, the intensity of the report ion can reflect the relative abundance of the peptides.Data was collected with the AB SCIEX Triple TOF 5600 System (Concord, USA) fitted with a Nanospray III source (Concord, USA) with a pulled quartz tip as the emitter (New Objectives, Woburn, USA).The MS was operated with an RP greater than or equal to 30,000 FWHM for TOF MS scans.

iTRAQ protein identification and quantification
Protein identification and quantification were performed with the Mascot 2.3.02search engine (Matrix Science, Boston, MA).The protein mass is predicted by website (http:// www.expasy.ch/tools/)based on the protein sequences.Search settings were used as described previously [22].The Searches were made against database from the Institute of cotton research of Chinese Academy of Agriculture Science website (http://cgp.genomics.org.cn/page/species/index.jsp).The search parameters were in Table S6.To demonstrate the reproducibility of the replicates, protein abundances between various biological replicates were compared, and ratios for each protein in each comparison were normalized to 1.For quantitative changes, a 1.2-fold cutoff was set to determine upregulated and downregulated significant proteins, with q-value < 0.05 (FDR corrected by BH) present in at least one replicate [30][31][32][33][34].

Bioinformatic analysis of proteins
Functional annotation of proteins was conducted using the Blast2GO program against the non-redundant protein database (NR; NCBI).The KEGG database (http://www.genome.jp/kegg/)and the Clusters of Orthologous Groups (COG) database (http://www.ncbi.nlm.nih.gov/COG/) were used to classify and group identified proteins.GO and pathway enrichment analysis were performed to determine which functional subcategories and metabolic pathways were overrepresented by the differentially accumulated proteins.

LC-MS-based metabolomics
The LC-MS-based targeted metabolomic analysis was performed according to a previously described protocol [34,51].Unbiased metabolomic profiles of cotton samples were obtained using HPLC-MS.All of the cotton samples were rapidly flash-frozen in liquid nitrogen and stored at −80°C until processing.A 25-mg sample was ground in liquid nitrogen, and transferred into a 1.5ml polypropylene tube.Then, 800 μL of chilled methanol/water (1:1) buffer solution, and two small balls were added to each tube.A tissue Lyser was set at the frequency of 60 Hz and the tube contents were shaken for 5 min followed by centrifugation at 25000 rpm for 10 min at 4°C.We used QC samples to assess the reproducibility and reliability of the LC-MS system.Each tube was centrifuged with 200 μL supernatant (sample code + up) and we then centrifuged another 200 μL supernatant mixed for QC mark.We added 200 μL supernatant to each tube as a QC sample and then dried it via vacuum freezing.We removed all of the supernatant and dried precipitate, then added 800 μL ice-cold mixture of dichloromethane/methanol (3:1) The tissue Lyser was set at a frequency of 60 Hz, shaken for 5 min, and the mixture was centrifuged at 25000 rpm for 10 min at 4°C.Each tube centrifuged 200 μL supernatant (sample code + down) and then centrifuged another 200 μL supernatant mixed for QC mark.Each tube had 200 μL supernatant added as the QC sample and then dried by vacuum freezing.Samples-up and samples-down were dissolved with 50% methanol, shaken for 1 min, centrifuged at 25000 rpm for 10 min at 4°C.We added 100 μL supernatant to 96-well plates.
Sample analysis was conducted in both positive electrospray ionization (ESI+) and negative ion (ESI−) modes.The test instruments were the 2777C UPLC system (Waters, USA) for liquid chromatography and SYNAPT G2 XS QTOF (Waters, USA) for mass spectrum analysis.Nitrogen was used as the dry gas and cone gas with the parameters described in Table S6.The separation of all of the samples (injection volume 10 μL) was performed on a ACQUITY UPLC BEH C18 column (Waters, USA) (dimension 100 × 2.1 mm, 1.7 μm particle size).Liquid Chromatographic column parameters with mobile phase A (water), mobile phase B (acetonitrile), and the speed 0.4 mL/min, and the gradient of mobile phase were 0~2 min with 100% A-100% A, 2~12 min with 100% A-0% A, 12~14 min with 0% A-0% A, 14~15 min with 0%-100% A.

Data processing and statistics
Insect-infested samples and their corresponding control groups were prepared as described above.The MS original data was analyzed by Progenesis QI (ver.2.2) software to obtain the peak (mz) retention time and ion area.The normalized data were introduced to SIMCA-P V11.0 (Umetrics, Sweden) for PCA and for OPLS-DA analysis.We analyzed the QC samples with PCA, and their TIC (total ion current) map (Fig. S1).TIC map was used to determine the status of the instrument, the greater the overlap of the QC sample replicates, the more stable of the instrument.We then conducted univariate analysis by t test and the p-value in the t test after FDR correction then produced q-values.The results were considered to be significant when the q-value was less than 0.05.OPLS-DA was carried out to investigate and visualize the pattern of metabolite changes.The differential metabolites were selected when the statistically significant threshold of VIP values obtained from the OPLS-DA model was larger than one.Log2 fold change (FC ≥ 1.2 or ≤ 0.8) was used to show how these selected differential metabolites varied between groups [32][33][34].The related pathways of each metabolite were also listed by searching the KEGG pathway database (http://www.genome.jp/kegg/),and the metabolite molecular formula of matched metabolites was further identified by isotopic distribution measurement.
Metabolomics data have been deposited to the EMBL-EBI MetaboLights database with the identifier MTBLS573.The complete dataset can be accessed here： http://www.ebi.ac.uk/metabolights/MTBLS573 4.9.Data analysis We used Microsoft Excel 15. 37, Mac Preview (8.1), Adobe Photoshop (2017.0.0) and Prims 6 to analyze the data and prepare the figures.We used the R package (Ver.3.2.3).for heat map analysis.We used the online Venn Diagram Generator (http://www.pangloss.com/seidel/Protocols/venn.cgi) for Venn map analysis.4.10.Real Time Quantitative PCR (RT-qPCR) The plant samples were collected followed by the previously description.Total RNA from each sample was extracted using the RNAprep Pure Plant Kit (TIANGEN, China) according to the manufacturer's protocol.Approximately 1 μg RNA was reverse transcribed to cDNA using a PrimeScriptTM RT reagent kit (perfect real time) (TaKaRa, Dalian, China) following the manufacturer's protocol.The RT-qPCR was carried out using GoTaq Qpcr Master Mix (Promega, USA) on an Eppendorf Mastercycler eprealplex 2.2 (Germany) with three biological replicates and three technical replicates.The thermal cycle conditions used in the RT-qPCR were 95°C for 2 min, followed by 40 cycles of 95°C for 15 s and 60°C for 1 min.The relative gene expression was processed using the 2 − △△ Ct method [57], and two housekeeping genes [35](GhHis3 with GenBank accession AF024716 and GhUBQ7 with GenBank accession DQ116441) were used in the RT-qPCR analysis.The three hours control group was set as the reference sample for data normalization.All of the primer pairs used for RT-qPCR were designed using the online PrimerQuest Tool (http://sg.idtdna.com/PrimerQuest/Home/Index)and are listed in Table S6.Differences in expression level were tested for significance by a one-way ANOVA with means separation using Tukey's HSD (      The RT-qPCR analysis of proteins related to the alpha-linolenic acid metabolism pathway.Cotton samples were collected as previously described, and total RNA was extracted for RT-qPCR analysis.A, C, E, G, I represent relative expression of different genes in cotton main leaf, respectively.B, D, F, H, J represent relative expression of different genes in cotton cotyledon, respectively.AOS: allene oxide synthase (CotAD_35840), PLA1: phospholipase A1llgamma-like (CotAD_52791), AOC4: allene oxide cyclase 4 (Cotton_D_gene_10007844), OPR1: 12-oxophytodienoate reductase 1 (CotAD_59461), OPR2: 12-oxophytodienoate reductase 2 (Cotton_D_gene_10037325).The GhHis3 and GhUBQ7 gene were used as the reference gene.Three biological replicates were performed.The three hours control group was set as the reference sample for data normalization.Significant differences between treatments and their corresponding control groups were identified by a one-way ANOVA with means separated using Tukey's HSD, and two levels ("*" p< 0.05 and "**" p< 0.01) were adopted to judge the significance of difference.

Fig 2 .
Fig 2. PCA and OPLS-DA score plot.(A) PCA of QC sample.(B) PCA score plot of A. suturalis infected and control groups.(C) OPLS-DA score plot of A. suturalis infected and control groups (with R 2 =0.861,Q 2 =0.626 in positive ion mode, and R 2 =0.904,Q 2 =0.690 in negative ion mode).Group1 represents the plants that were infested with insects for 48 h, group 2 represents the plant that were not infested but simply grown for 48 h.ESI+ represents positive ion mode, ESI-represent negative ion mode.

Fig 3 .
Fig 3. Heat maps comparison of metabolites of the A. suturalis infected group and control group.Heat maps represents the highly significantly differential variables between the insect infected groups and the corresponding control groups using R package (Ver.3.2.3).ESI+ represents positive ion mode, ESI-represents negative ion mode.E represents the insect infected group, C represents the corresponding control group.

Fig 4 .Fig 5 .
Fig 4. Tryptophan metabolism pathway analysis.(A) Metabolites changed in tryptophan metabolism pathway.Red represents upregulated, blue represents downregulated, yellow represents upregulated in positive mode, down regulated in negative mode, white represents unchanged metabolites in both ion mode.Rectangular box represents metabolites changed in positive ion mode, elliptic box represents metabolites changed negative ion mode, regular hexagon box represents changed metabolites in both ion mode.(B) A schematic representation showing how the metabolites in tryptophan metabolism pathway affect plant defense attacked by insect.(The photo was taken by the author Hui Lu )

Fig 6 .
Fig 6.The RT-qPCR analysis of proteins related to the alpha-linolenic acid metabolism pathway.Cotton samples were collected as previously described, and total RNA was extracted for RT-qPCR analysis.A, C, E, G, I represent relative expression of different genes in cotton main leaf, respectively.B, D, F, H, J represent relative expression of different genes in cotton