Agasicles hygrophila attack increases nerolidol synthase gene expression in Alternanthera philoxeroides, facilitating host finding

Herbivorous insects use plant volatile compounds to find their host plants for feeding and egg deposition. The monophagous beetle Agasicles hygrophila uses a volatile (E)-4,8-dimethyl-1,3,7-nonanetriene (DMNT) to recognize its host plant Alternanthera philoxeroides. Alternanthera philoxeroides releases DMNT in response to A. hygrophila attack and nerolidol synthase (NES) is a key enzyme in DMNT biosynthesis; however, the effect of A. hygrophila on NES expression remains unclear. In this study, the A. philoxeroides transcriptome was sequenced and six putative NES genes belonging to the terpene synthase-g family were characterized. The expression of these NES genes was assayed at different times following A. hygrophila contact, feeding or mechanical wounding. Results showed that A. hygrophila contact and feeding induced NES expression more rapidly and more intensely than mechanical wounding alone. This may account for a large release of DMNT following A. hygrophila feeding in a previous study and subsequently facilitate A. hygrophila to find host plants. Our research provides a powerful genetic platform for studying invasive plants and lays the foundation for further elucidating the molecular mechanisms of the interaction between A. philoxeroides and its specialist A. hygrophila.

Transcriptome sequencing, assembly, and annotation. Alternanthera philoxeroides individuals with similar growth were selected for sampling. The fourth and fifth pairs of leaves, the third and fourth internodes from the top, flowers, and roots were harvested and frozen in liquid nitrogen for further processing. Total RNA was isolated from each of the four tissues (leaves, stems, flowers, and roots) using TRIzol (Thermo Fisher Scientific, Waltham, MA, USA) and RNA from each tissue was mixed in equal amounts. The RNA was then submitted to Biomarker Technologies (Beijing, China) for transcriptome sequencing on a HiSeq 2500 system (Illumina, San Diego, CA, USA). The sequencing adaptors, primer sequences, and low-quality data (reads that were< 50 bp and contained more N-bases) were removed from the original data, after which the clean reads were assembled using Trinity (Broad Institute, Cambridge, MA, USA) 24 to obtain transcripts and unigenes. The unigenes were annotated by using them as a template for BLASTX-searches of multiple public repositories of known proteins, including RefSeq non-redundant proteins (NR), UniProtKB/Swiss-Prot 25 , Gene Ontology (GO; https ://geneo ntolo gy.org/) 26 , Clusters of Orthologous Genes (COG; https ://www.ncbi.nlm.nih.gov/COG/) 27 , EuKaryotic Orthologous Groups (KOG), Kyoto Encyclopedia of Genes and Genomes (KEGG; https ://www.kegg.jp/kegg/ kegg1 .html) 28 , Pfam (https ://pfam.xfam.org/) 29 , and other databases. The results were parsed to calculate the total number of positive identities per query-subject pair and further filtered using an E-value ≤ 1e−05. The KEGG Orthology results of the unigenes were predicted using KOBAS 2.0 (https ://kobas .cbi.pku.edu.cn).
Identification of terpene synthase genes. The identified nerolidol synthase FaNES1 was used to align to the unigene data from the A. philoxeroides transcriptome. The Unigenes in the results annotated as "terpene synthase" or "TPS" were then selected and aligned with the NCBI NR database using an online BLAST search (https ://blast .ncbi.nlm.nih.gov/Blast .cgi) 30 . The open reading frames (ORFs) were searched with ORFfinder (https ://www.ncbi.nlm.nih.gov/orffi nder/) and translated into amino acid sequences. DNAMan 7.0 (Lynnon Scientific Reports | (2020) 10:16994 | https://doi.org/10.1038/s41598-020-73130-z www.nature.com/scientificreports/ Biosoft, San Ramon, CA, USA) was used to analyze the sequence identity of the A. philoxeroides TPS protein sequences. After removing incomplete sequences shorter than 500 bp and those shorter than 150 aa of translated protein sequences, we used the Maximum Likelihood method and JTT + G + I model to construct a phylogenetic tree of the TPS protein sequences in A. philoxeroides together with other plant species (Table S1) by MEGA6.0 (https ://www.megas oftwa re.net/) 31 . The phylogenetic tree was verified using 1000 bootstrap replicates and was drawn by iTOL 5.6.1 (https ://itol.embl.de/) 32 .
Investigation of the short-term feeding behavior in A. hygrophila adults. Newly emerged adults of A. hygrophila were placed in two glass bottles separately by their genders. The genders of adults were distinguished by individual size and abdominal morphological characteristics. Specifically, the female adult was larger than the male. The female adult has a ventral flat, and two abdominal segments at the end of its abdomen exposed outside the elytra. The abdomen of the male adult is covered with the elytra, and an oblong cavity that contains retractable external genitalia is in the 5th abdominal sternite. Ten newly emerged healthy A. hygrophila adults (five males and five females) were taken out from the bottles and starved for 12 h, then placed in a cylindrical box 5 cm in diameter and 4 cm in height. At the beginning of the light period, fresh leaves with stem were placed into the containers. The feeding duration of each beetle was observed and recorded for 2 h, and the area consumed by each adult was evaluated after 1 and 2 h using scanning pixel method. Specifically, we scanned a piece of square paper 10 cm in side length, the same leaf before and after insects feeding with a scanner for grayscale scanning. To calculate the gray part pixels, Photoshop CS6 was used. According to the paper size and the number of pixels, the area size of the insect feeding part was calculated as following: Consumed area = (pixel of leaf − pixel of leaf after consumed)·centimeter/pixel ratio. Centimeter/pixel ratio was calculated by Photoshop CS6.0 using a scanned picture contains a square paper 10 cm in side length. Each experiment was repeated three times.
Analyses of NES expression. Plant treatments. Plants with similar levels of growth were selected and cultured in cylindrical glass tubes (3.5 cm diameter, 30 cm height) containing 50 ml water until the roots were grown. One of the fourth pair of leaves was covered with a gauze cage (3 cm diameter, 6 cm height). Six newly emerged A. hygrophila adults (three males and three females) were starved for 12 h before they were placed in the cage. Starting from the time adults touched the leaves, a 5-min insect contact treatment was recorded as long as no feeding or excretion occurred. After 1 h of feeding, A. hygrophila was removed, and the leaves were collected at 0, 15, 30, 60, 180, 360, and 720 min after insect feeding. For each time point, different individual plants were used. Mechanical wounding was used to distinguish the effect caused by A. hygrophila secretion or mechanical damage by A. hygrophila feeding. The area and duration of mechanical wounding affect the plants' volatile release 22 . Therefore, a 6-mm aperture puncher was used to mechanically injure the plant leaf margin for about 5 min until the area was similar to the insect feeding area, and the leaves were collected respectively from different individual plant at 0, 15, 30, 60, 180, 360, and 720 min after the mechanical wounding. At each point, a control was also performed. The leaves were frozen with liquid nitrogen for subsequent RNA extraction. Each treatment has five biological replicates, and each contained three leaves from different plant.
Sequence search and primer design. The sequences of the housekeeping and NES genes were identified in the transcriptome data based on their similarity with the sequences of these genes in other species stored in the NR protein database. The actin gene ACT , the α-tubulin gene TUA , the β-tubulin gene TUB, and the gene encoding ubiquitin-binding enzyme, UBC 33 , were selected as the reference genes. Real-time quantitative PCR (qPCR) primers (Table S2) were designed using Primer3 (https ://prime r3.ut.ee/) and synthesized by Sangon Biotech (Shanghai, China). To check the specificity of all primers, total RNA was reverse transcribed using the Prime-Script RT reagent kit with gDNA eraser (Takara Bio, Kusatsu, Japan). PCR was performed on the cDNA, and the PCR products were analyzed on a 1% agarose gel.
qPCR and data analysis. The total RNAs of the collected materials were extracted using the TRIzol method outlined above. The RNAs were quantified using a BioPhotometer Plus (Eppendorf, Hamburg, Germany), and their quality was verified using electrophoresis on an agarose gel. A 1-μg aliquot of total RNA was reverse transcribed at 37 °C for 15 min with an oligo-dT primer using the PrimeScript RT reagent kit with gDNA eraser (Takara Bio, Kusatsu, Japan). The gDNA eraser was used to remove genomic DNA. Two independent reactions were performed for each RNA sample. The qPCR was performed using an Applied Biosystems 7500 Fast Real-Time Detection system (Thermo Fisher Scientific, Waltham, MA, USA) with the SYBR Green Realtime PCR Master Mix (Toyobo, Osaka, Japan). Each reaction was carried out according to the manufacturer's protocol in a 20-µl volume, which contained 2 µl tenfold diluted reverse-transcribed cDNA as a template and 500 nM of each primer. In all qPCR experiments, the transcript levels of three independent biological replicates were measured, with three technical replicates performed for each sample. A no-template control was included in each run for each gene. The qPCR reaction was performed using the following conditions: 95 °C for 2 min, followed by 40 cycles of 95 °C for 15 s, 55 °C for 15 s, and 72 °C for 30 s. Finally, a melting curve was generated by increasing the temperature from 65 °C to 95 °C to determine the specificity of the reactions.
The stability of the four candidate reference genes UBC, ACT , TUA , and TUB were evaluated in the control group, A. hygrophila contact group, A. hygrophila feeding group, and mechanical wounding group using the reference screening software geNorm 34 37 . For calculating the basal gene expression in the control group, ApTPS19 was used as the control gene and the basal expression of ApTPS19 was normalized equal to 1. Other NES gene basal expression levels in the A. philoxeroides control group leaves were calculated and compared as the folds of ApTPS19 basal expression. The transcript levels of the NES genes in the control leaves and each of those subjected to contact, feeding by A. hygrophila, and mechanical wounding were transformed with logarithmic transformation to ensure normal distribution and compared using independent two-sample Student's t-tests performed in IBM SPSS Statistics 22.0 (International Business Machines Corporation, Amonk, New York, USA). One-way ANOVA was used to analyze the basal expression of each NES gene. Two-way ANOVA was used to analyze the expression of each NES gene in each treatment group, with treatment methods and post-treatment time as the influential factors. Two-way ANOVAs were also used to analyze the expression of the NES genes in the four treatment groups at the same time point. Multiple comparisons were performed using Tukey's post-hoc test. Student's t-test was used to compare the expression levels of NES genes between two different treatments, after the homogeneity test.

Results
RNA sequencing, assembly, and annotation. To identify the terpene synthase genes involved in DMNT biosynthesis in A. philoxeroides, we performed transcriptome sequencing on the major tissues of this plant. A total of 9.91 Gb nucleotides and 40,984,113 raw reads were obtained from the A. philoxeroides sequencing, and the percentage of Q30 bases reached 91.14%. The original data were deposited in the NCBI SRA library under the accession number SRR10537333. In total, 95.97% of the clean reads were assembled into 4,904,951 contigs, 182,318 transcripts, and 83,179 unigenes ( Table 1).
Homologs of the unigenes were identified by BLASTX; each publicly available protein database was searched using the transcriptome data, which resulted in the successful annotation of 63,536 (76.4%) of the unigenes. Among them, 63,076 genes were annotated based on their homology with sequences in the NR protein database, which accounted for 99.28% of the total number of annotations (Table 1). 32.74% of A. philoxeroides annotated sequences were aligned to beet (Beta vulgaris subsp. vulgaris L.) (Fig. 1A).
We classified the predicted genes of A. philoxeroides using the GO, COG, KOG, KEGG, and other protein databases. A total of 42,386 unigenes were divided into three major categories and 52 functional groups (Fig. 1B); 23,984 unigenes were mapped to 277 KEGG pathways, including the glycolytic, gluconeogenesis, and pentose phosphate pathways, among which the numbers of unigenes associated with the carbon cycle (ko01200), amino acid synthesis (ko01230), and ribosome (ko03010) pathways were 1172, 1057, and 1016, respectively (Table 1; Fig. S1, S2).

Identification and analysis of the terpene synthase genes and NES genes in A. philoxeroides.
In the A. philoxeroides transcriptome data, a total of 34 terpene synthase genes were identified (Table S3). Their encoded protein sequences were compared with those of other species. The sequences of terpene synthase genes in A. philoxeroides had the highest similarity to those of B. vulgaris.
The phylogenetic tree showed that the 12 terpene synthase proteins were divided into five subfamilies: TPSa, TPS-b, TPS-c, TPS-e, and TPS-g (Fig. 2). The TPS-a, TPS-b, and TPS-g subfamilies contained the terpene synthases specific to angiosperm species, which were isolated from the other TPS subfamilies. These angiospermspecific proteins were class-III terpene synthases, which was in accordance with the species classification of A. philoxeroides. ApTPS7, ApTPS9, ApTPS17, ApTPS18, ApTPS21, and ApTPS22 were classified into the TPS-a subfamily; ApTPS5 and ApTPS6 were classified into the TPS-b subfamily; and ApTPS16 and ApTPS19 were classified into TPS-g family. The TPS-e subfamily largely comprised kaurene synthase, which had a function similar to the annotated function of ApTPS23. The TPS-c subfamily mainly contained the monofunctional diterpene synthase, copalyl diphosphate synthase (CPS). ApTPS25, which was annotated as CPS, was also grouped into the TPS-c subfamily.  Table 2). ApTPS19 was the longest of these genes (2161 bp), and was predicted to have an ORF of 1791 bp and to encode a protein of 596 aa in length. Four NES proteins were removed due to their shorter sequences, leaving ApTPS16 and ApTPS19, both of which were classified into the TPS-g subfamily. These proteins were annotated as NES, and shared about a 40% similarity with the other TPS enzymes in A. philoxeroides.

Short-term feeding behavior of A. hygrophila.
In the present study, we examined the behavior of A.
hygrophila adults on the leaves of A. philoxeroides within the first two hours. These actions include contacting, eating and resting. The time between the male and female adults touching the leaves and the initiation of feeding was 5.9 min and 6.2 min, respectively, after a 12-h starvation. The duration of a single feed was 19.6 min for www.nature.com/scientificreports/ males and 26.5 min for females. No statistical differences were observed between the feeding behaviors of male and female adults ( Table 3). The interval between the first and second feeding periods varied from 1 to 90 min, during which the adults usually remained inactive. We, therefore, selected a 5-min period after placing the adults into the cage for use as the contact treatment in future experiments. In addition, the leaf areas consumed by one adult in 1 h and the areas consumed by different numbers of adults in 1 h were also evaluated (Tables 3, S4). The feeding of male and female adults resulted in different leaf areas consumed; therefore, the use of six adults (three males and three females) was chosen as the insect feeding group for the subsequent 1-h feeding experiment. The leaf area consumed was about 20% of the entire leaf using this treatment method, which provided a basis for the mechanical wounding treatment of the A. philoxeroides leaves.  Table S3. The number on the branches indicate the supporting bootstrap value of proteins on the branches (between 40 and 100%). The closer the bootstrap value is to 1, the higher confident the branch is. The proteins on the same branch mean they are the most related proteins, which may share high homology and have similar function.   S3), with the lowest Ct values observed for UBC (19)(20)(21)(22) and the highest Ct values observed for TUB (23)(24)(25)(26). The expression levels of each candidate reference gene were different when they were analyzed using GeNorm, Nor-mFinder, and BestKeeper. The GeNorm and NormFinder software analysis revealed that the M values of the four reference genes were all less than 1.5, and therefore these genes were considered to be relatively stable when expressed. Among them, UBC had the smallest M value and the best expression stability, whereas TUA had the largest M value and a relatively poorer stability. The BestKeeper software analysis revealed that the SD values of the Ct values for the four reference genes were < 1, indicating relatively stable gene expressions. Among them, UBC had the smallest SD value and the best expression stability, whereas TUA had the largest SD value and the worst stability (Table 4). Based on these results, the expression stabilities of the four candidate reference genes were ranked as follows: UBC > TUB > ACT > TUA . We therefore used UBC as the reference gene to investigate the expression of the NES genes in the leaves of A. philoxeroides under the different damage treatments.

Basal expression of various NES genes in healthy leaves. We investigated the basal expression level
of the NES genes in the healthy leaves because of the large amount of DMNT from healthy A. philoxeroides plants. The results revealed statistical differences in the basal expression of the six NES genes in the leaves of the control plants (P < 0.05). Among them, ApTPS19 had the lowest basal expression, while ApTPS15 had the highest basal expression, which was 2058.9 times higher than that of ApTPS19 (Fig. 3). Table 4. Stable values of data analyses performed using GeNorm, NormFinder, and BestKeeper. M dictates the stability index: the larger the M value, the worse the expression stability of the gene. SD is standard deviation of the Ct value. CV is coefficient of variation of the Ct value. If SD < 1, the gene was considered to be relatively stable; therefore, the smaller the standard deviation and coefficient of variation, the more stable the gene expression.  The relative expression level of each NES is shown in fold relative to that of ApTPS19 which had the lowest transcription level that was taken as "1". The values are means ± SD of three biological replications. Different letters indicate statistical differences between the genes, revealed using a one-way ANOVA and a Tukey's posthoc test (P < 0.05).

Induction of NES genes in A. philoxeroides prior to feeding by A. hygrophila. The early feed-
ing behavior of A. hygrophila adults was investigated, revealing that the time between the flea beetles touching the leaves and the initiation of feeding was typically longer than 5 min. The changes in NES gene expression were evaluated in the leaves touched by A. hygrophila for 5 min. The expression levels of ApTPS10, ApTPS12, ApTPS16, and ApTPS19 increased significantly in A. philoxeroides leaves touched by A. hygrophila (P < 0.05), with the expression of ApTPS16 being 6.27 times higher than that of the control group. By contrast, ApTPS15 expression decreased significantly in the leaves touched by A. hygrophila (P < 0.05), whereas ApTPS14 expression was not statistically different between the control and A. hygrophila contact group (Fig. 4).

A. hygrophila feeding strongly induces NES gene expression in A. philoxeroides. The changes in
the NES gene expression upon A. hygrophila feeding were also analyzed in a time series. All of the six NES genes in the leaves were significantly up-regulated immediately after 1 h insect-damaged (P < 0.05) (Fig. 5).
ApTPS14 expression showed the greatest up-regulation, reaching a level 42.7 times higher than that of the control plants (P < 0.05). The expression of ApTPS19 showed the smallest up-regulation by insect feeding, reaching a level only 7.3 times higher than that of the control plants (P < 0.05). The four other NES genes showed several-fold up-regulation of their expression following A. hygrophila feeding, and were significantly different from the control plants (P < 0.05). Fifteen minutes after feeding ended, the expression of the six NES genes remained statistically higher than in the control plants (P < 0.05), but as time progressed their expression levels gradually decreased, returning to their normal levels within 15 to 30 min. ApTPS12, ApTPS14, ApTPS15, and ApTPS16 also showed a small peak in expression 1 h after feeding ended (P < 0.05) (Fig. 5).

Mechanical wounding induces delayed and low levels of NES gene expression. Upon mechani-
cal wounding, all of six NES gene expression showed no statistical difference from that in the leaves of the control group (Fig. 5).
After the mechanical wounding, the expression levels of the differentially expressed NES genes first increased to a peak at around 30-60 min, then gradually decreased to a level similar to that of the control plants after 1-3 h. After 30 min, ApTPS10, ApTPS12, ApTPS15, and ApTPS16 were significantly differentially expressed in the leaves subjected to mechanical wounding in comparison with the control plants (P < 0.05). After 1 h, ApTPS14 expression remained significantly different in the two groups (P < 0.05). However, at several time points, ApTPS19 expression was similar between the two groups (Fig. 5).

Discussion
Here, we showed that the presence of A. hygrophila affects NES gene expression in A. philoxeroides even before feeding and short-term feeding (1 h), and by A. hygrophila strongly and rapidly induces the up-regulation of the NES genes. This up-regulation of NES expression was likely the main reason why chewed leaves released more DMNT, in turn attracting more A. hygrophila to feed on these plants.
Initially, second-generation transcriptome sequencing was performed on the roots, stems, leaves, and flowers of A. philoxeroides to provide genetic and protein data that were lacking for this species. Compared with the previously reported transcriptomes of A. philoxeroides roots 38 and A. philoxeroides individuals growing in ponds and uplands 33 , our sequencing provided novel genetic data for the flowers, which are an important organ for the release of volatile compounds 39,40 . Also, the annotation of the A. philoxeroides genes and proteins was improved. Along with the full-length transcriptome sequencing of A. hygrophila 41 , we provided valuable data Nerolidol synthase gene expression levels in the control group, A. hygrophila feeding group, mechanical wounding group of A. philoxeroides plants. In the A. hygrophila feeding group, the A. philoxeroides leaves were sampled at the indicated time points following a 1-h feeding experiment. In the mechanical wounding group, the A. philoxeroides leaves were sampled at the indicated time points following their mechanical damage. The values are means ± SD (normalized that the expression level in the control group equal to 1) of three biological replications. Different letters indicate statistical differences between different time treatments in control group, as determined using a one-way ANOVA and a Tukey's post-hoc test (P < 0.05). Different letters with single quotation mark indicated statistical differences between different time treatments after insect feeding, as determined using a one-way ANOVA and a Tukey's post-hoc test (P < 0.05). Different letters with double quotation marks indicate statistical differences between different time treatments after mechanical wounding, as determined using a one-way ANOVA and a Tukey's post-hoc test (P < 0.05). Asterisks indicate statistical differences between every two treatments in the same time treatment (*P < 0.05; **P < 0.01; Student's t-test).
Scientific Reports | (2020) 10:16994 | https://doi.org/10.1038/s41598-020-73130-z www.nature.com/scientificreports/ for the comprehensive study of the molecular mechanisms underlying the specificity between A. philoxeroides and A. hygrophila. Few studies have reported sequencing data for other species in the Alternanthera genus; the genetic and protein data obtained from sequencing other related species in this family are limited to several edible and medicinal crops, such as Amaranthus caudatus L. 42 , Amaranthus hypochondriacus L. [43][44][45] , and Achyranthes bidentata Blume 46 . The transcriptome data of A. philoxeroides has therefore enriched the genetic data available for the Amaranthaceae, especially the Alternanthera genus. It also enriched the genetic data available for invasive plants, providing powerful genetic support for exploring their invasive properties, and their interactions with herbivorous insects. In total, 34 TPS genes were predicted in the A. philoxeroides transcriptome (Table S3). In plants with published genomes, the number of TPS genes ranges from 19 to 152 16 ; for example, there are 40 TPS genes in Arabidopsis thaliana, of which 32 have known or predicted functions 47 . Based on the materials and methods used for this transcriptome sequencing, we speculate that the A. philoxeroides genome may contain more TPS genes, which should be further explored using genome sequencing. The discovery of these TPS genes will enhance our understanding of the TPS genes in the Amaranthaceae as a whole.
Feeding by the specialist insect A. hygrophila affects A. philoxeroides, causing the release of DMNT 7 . The NES genes, which belong to TPS-g subfamily, are related to DMNT biosynthesis. Six NES genes were screened in this transcriptome. Following the contact or feeding by A. hygrophila, the expression pattern of A. philoxeroides NES genes was further studied. The NES genes were up-regulated before feeding by A. hygrophila. In the 5 min before feeding, when adults merely touched the leaves of A. philoxeroides, the initial contact caused significant up-regulation of most of the NES genes. After injury, plants usually activated a transcription cascade, leading to the expression of genes involved in the release of volatile organic compounds; however, it generally takes a few hours to a day (24 h) for the compounds to be released [48][49][50][51][52] . Compared with the Lilium 'Siberia' terpene synthase gene LoTPS1, which is activated in response to wounding, NES genes in A. philoxeroides responded to damage much faster 48 .
The rapid up-regulation of these genes following contact with A. hygrophila may have two explanations. In the first case, when the beetle touches the leaf surface, it might cause slight mechanical damage to the leaf, which in turn up-regulates NES expression. The tarsus of an insect typically has claws, which can cause fine scratches on leaf blades and break their trichomes as the insect crawls along the leaf surface. This has been confirmed in tobacco (Nicotiana tabacum L.), soybean (Glycine max (L.) Merr.) 53,54 , and tomato (Solanum lycopersicum L.) 55 . Also, the minor mechanical damage causes the release of plant terpenoid volatiles, which has been documented in a touch study of potato (Solanum tuberosum L.) 56 . However, the expression level of NES in the A. hygrophila contact group was significantly higher than that in the mechanical wounding group. The other possibility is the chemical substances secreted by insects may generate a chemical "footprint" on the leaf and up-regulate the expression of the NES genes. In other species, insect chemical footprints have been shown to mediate the transmission of the insect's intraspecific and interspecific information. For example, some pollinators has been showed to transmit flower information to the same species by chemical "footprint" 57,58 , some parasitoids have located host insects by recognition of the chemical traces that host insects deposited after crawling 59 . It is not yet clear whether these secretions can trigger plant responses 60 , a hypothesis that will need to be tested in future studies.
When A. hygrophila was in contact with the plant for prolonged periods of time, the beetles began to feed. Short-term feeding (1 h) by A. hygrophila caused rapid and intense up-regulation of NES genes, which is consistent with the previous study showing release of large amounts of DMNT after feeding by A. hygrophila 7 . Plants responded quickly to insects' damage, including the rapid biosynthesis of monoterpenes and sesquiterpenes 61 . For example, the feeding behavior of insect herbivores such as Spodoptera littoralis (Boisduval) 62 , Lymantria dispar L. 63 , and Hylobius abietis (L.) 64 also induced the expression of plant terpene synthase genes and the release of terpene volatiles. For insects with chewing mouthparts, such as A. hygrophila, the feeding behavior not only causes mechanical wounding to the leaves but also deposes oral secretions, such as saliva, in the leaf incisions. Both of these can induce plant volatile production through special transcriptional pathways 65 . In this study, we used a puncher simulating similar feeding areas to rule out the effect of mechanical wounding on gene expression changes. However, NES expression in the A. hygrophila feeding group was still significantly greater than that in the mechanical wounding group. In our previously work, we measured the DMNT released from A. philoxeroides following mechanical wounding and feeding by A. hygrophila. The DMNT released due to insect feeding was 240 units ng·h −1 ·10 g −1 ·FW −1 ; by contrast, the DMNT release by mechanical wounding was only 13 units ng·h −1 ·10 g −1 ·FW −17 . The magnitudes of NES expression caused by insect feeding and mechanical wounding in this study reflected the magnitudes of DMNT in response to these two treatments, respectively 7 . In Li 7 , the DMNT was collected over a 12 h period which covered the time frame used in this study. If there would be any DMNT released by mechanical wounding, it would be very minimal compared to the large quantity of DMNT caused by insect feeding. This suggests that the chemicals secreted by A. hygrophila might trigger the up-regulation of NES, which leads to the release of a large amount of DMNT after A. hygrophila feeding 7 .
In damaged A. philoxeroides leaves, high expression of NES genes causes a large amount of DMNT to be released, which attracts more specialist A. hygrophila 7 . The release of plant volatiles has a significant attraction to insects (herbivore and their parasitoids) and may provide signals for conspecifics to find host plants or increase aggregation. For example, the root-feeding larvae of a specialized insect Bikasha collaris (Baly) can induce its host plant Triadica sebifera (L.) Small to produce volatiles, attracting aboveground conspecific adults 66 . Also, Spodoptera littoralis caterpillars could be attracted by volatiles emitted by maize (Zea mays) damaged by conspecific larvae, and thereby find a less suitable but easily detectable host plant 67 . We speculated that the specialist A. hygrophila might use the chemicals it secretes to affect the release of volatiles from the host plants and facilitate population development by making it easier for conspecifics to locate their host plants A. phyloxeroides. These questions will need to be explored in future research.

Conclusions
In this study, we identified six NES genes from the A. philoxeroides transcriptome. When the leaves were contact by A. hygrophila, NES gene expression increased rapidly, and was strongly up-regulated after A. hygrophila feeding. Our results suggested that the rapid and strong increase of NES transcript levels promotes the large-scale release of DMNT. The released DMNT helps conspecific species of A. hygrophila locate host plants. Our results also suggested that chemicals secreted by A. hygrophila could be responsible for the rapid and strong increase of NES transcript levels. Our results represent an initial explanation of the molecular mechanisms underlying the interaction between A. philoxeroides and its natural enemy A. hygrophila and provide a molecular basis for understanding the relationships between obligate herbivorous insects and their host plants.