Characterizing the serotonin biosynthesis pathway upon aphid infestation in Setaria viridis leaves

Setaria viridis (green foxtail millet), a short life-cycle C4 plant in the Gramineae, serves as a resilient crop that provides good yield even in dry and marginal land. Although S. viridis has been studied extensively in the last decade, its defense responses, in particular the chemical defensive metabolites that protect it against insect herbivory, are unstudied. To characterize S. viridis defense responses, we conducted transcriptomic and metabolomic assays of plants infested with aphids and caterpillars. Pathway enrichment analysis indicated massive transcriptomic changes that involve genes from amino acid biosynthesis and degradation, secondary metabolites and phytohormone biosynthesis. The Trp-derived metabolite serotonin was notably induced by insect feeding. Through comparisons with known rice serotonin biosynthetic genes, we identified several predicted S. viridis Trp decarboxylases and cytochrome P450 genes that were up-regulated in response to insect feeding. The function of one Trp decarboxylase was validated by ectopic expression and detection of tryptamine accumulation in Nicotiana tabacum. To validate the defensive properties of serotonin, we used an artificial diet assay to show reduced Rhopalosiphum padi aphid survival with increasing serotonin concentrations. This demonstrated that serotonin is a defensive metabolite in S. viridis and is fundamental for understanding the adaptation of it to biotic stresses. HIGHLIGHT A combined transcriptomic and metabolomic profiling of Setaria viridis leaves response to aphid and caterpillar infestation identifies the genes related to the biosynthesis of serotonin and their function in defense.


INTRODUCTION 56
In nature, plants are continuously challenged by diverse insect herbivores. In response to 57 insect infestation, plants produce constitutive and inducible defenses to reduce damage and 58 enhance their own fitness (Zhou et al., 2015). Although many plant defenses are produced 59 constitutively during a specific developmental stage, regardless of insect attack, others are 60 inducible in response to insect damage. Examples of herbivore-induced defense strategies are the 61 accumulation of chemical defenses such as benzoxazinoids and glucosinolates, physical barriers 62 such as the increased density of thorns, spikes or glandular trichomes, as well as the synthesis of 63 protease inhibitors and proteases (Bennett and Wallsgrove, 1994;Boughton et al., 2005;van 64 Loon et al., 2006;Ahuja et al., 2010;Markovich et al., 2013). These inducible defenses usually 65 are present at low levels and become more abundant in response to insect feeding. The general 66 herbivore-induced defense strategies are mediated by signaling pathways, including those 67 mediated by jasmonic acid and salicylic acid (Kerchev et al., 2012), which allow plants to 68 conserve metabolic resources and energy for growth and reproduction in the absence of insect 69 herbivory. 70 The aphid family (order Hemiptera, family Aphididae), which comprises approximately 71 5,000 species distributed worldwide (Rebijith et al., 2017), causes massive yield losses due to 72 both direct and indirect crop damage (Guerrieri and Digilio, 2008). Aphids consume water and 73 nutrients from plants, while transmitting toxins through their saliva (Rabbinge et al., 1981;Bing 74 et al., 1991;Zhou et al., 2015;Tzin et al., 2015). Aphids also are responsible for transmission of 75 40% of all plant viruses, including the most harmful of plant viruses (Fereres et al., 1989;Nault, 76 1997). Another major pest for graminaceous plants are leaf-chewing insects such as lepidopteran 77 larvae. In response to the lepidopteran attack, plants massively modify their transcriptome and 78 synthesize chemical defense compounds (Oikawa et al., 2004;Niemeyer, 2009;Glauser et al., 79 2011). Many plant deterrent compounds are derived from catabolism of the aromatic amino acids 80 Trp (tryptophan), Tyr (tyrosine) and Phe (phenylalanine) (Zhou et al., 2015;Tzin et al., 2015Tzin et al., , 81 2017Wisecaver et al., 2017). In the Gramineae family, indole, and its precursor Trp, serve as 82 sources for several classes of defensive secondary metabolites including i) benzoxazinoids in 83 maize (Zea mays), wheat (Triticum aestivum), rye (Secale cereal), and wild barley species (Grün 84 et al., 2005;Ishihara et al., 2017;Niculaes et al., 2018;Zhou et al., 2018), ii) gramine in 85 cultivated barley (Grün et al., 2005), and iii) serotonin (5-hydroxytryptamine) in rice (Oryza 86 Serotonin, a well-studied neurotransmitter in mammals, also has been found at detectable 89 levels in more than 90 plant species (Kang et al., 2008;Erland et al., 2016;Alexandra et al., 90 2017). In plants, it has been implicated in adaptations to environmental changes and in defense 91 responses against pathogen infection and insect herbivory (Ishihara et al., 2008b;Kang et al., 92 2008;Park et al., 2009). Serotonin biosynthesis requires the sequential action of two enzymes: 93 first Trp decarboxylase (TDC), which produces tryptamine (Gill et al., 2003;Ishihara et al., 94 2011;Li et al., 2016;Welford et al., 2016), and second, tryptamine 5-hydroxylase (T5H), a 95 cytochrome P450 that catalyzes the conversion of tryptamine to serotonin (Fujiwara et al., 2010;96 Dharmawardhana et al., 2013;Lu et al., 2018) (Fig. 1). Both enzymes play an important role in 97 determining serotonin levels and are induced by both chewing insect and pathogen attacks (Kang 98 et al., 2009;Hayashi et al., 2015). Still, the function of serotonin in plants is inconclusive 99 (Alexandra et al., 2017). For example, ectopic expression of the Camptotheca acuminata TDC1 100 gene allowed sufficient tryptamine to accumulate in poplar and tobacco leaf tissues to 101 significantly suppress the growth of two moth species Malacosoma disstria and Manduca sexta, 102 respectively (Gill et al., 2003;Gill and Ellis, 2006). Similarly, ectopic expression of the Aegilops 103 variabilis TDC1 gene in tobacco plants increased resistance to the cereal cyst nematode, 104 Heterodera avenae (Huang et al., 2018). On the other hand, a recent study of the mutated 105 CYP71A1 (T5H) in rice, demonstrated that reduced serotonin levels caused higher susceptibility 106 to rice blast Magnaporthe grisea, but also conferred resistance to rice brown spot disease 107 (Bipolaris oryzae), and the rice brown planthopper (Nilaparvata lugens) (Lu et al., 2018). The 108 function of serotonin and tryptamine in defense against aphids has not yet been studied. 109 The Gramineae, a large plant family consisting of approximately 12,000 species and 771 110 genera (Soreng et al., 2015), includes staple crops such as rice, wheat, maize, barley, sorghum 111 and several millets (Brutnell et al., 2010;Soreng et al., 2015;Kokubo et al., 2016a;Doust and 112 Diao, 2017). Setaria, which serves as both a research model and a crop plant is a genus of 113 panicoid C4 grasses (Brutnell, 2015;, closely related to grasses such 114 as switchgrass (Panicum virgatum) and pearl millet (Panicum glaucum), as well as to maize and 115 sorghum (Li and Brutnell, 2011;Pant et al., 2016). The domesticated S. italica is an ancient 116 cereal grain crop from China, which expanded to India and Africa, excelling as a drought-and 117 low-nutrient-tolerant grain (Goron and Raizada, 2015;Nadeem et al., 2018), while its wild 118 ancestor, Setaria viridis (green foxtail millet), is a widespread weed across the globe (Hu et al., 119 2018). Both S. italica and, S. viridis genotypes have recently been sequenced, and their genomes 120 are publicly available (Bennetzen et al., 2012). Setaria species have thus recently emerged as 121 model plants for studying C4 grass biology and other agronomic traits (Brutnell et al., 2010). 122 Setaria viridis has many advantageous properties as a monocot genetic model system: i) a short 123 generation time of 6-8 months (illustrated in Fig. 2A); ii) a small sequenced diploid genome 124 containing approximately 395 Mb and 38,000 protein-coding genes; iii) transient and stable 125 transformation systems and established protocols; and iv) a small stature and simple growth 126 requirements (Li and Brutnell, 2011;Liu et al., 2016;Mei et al., 2016;Van Eck et al., 2017;Zhu 127 et al., 2018). Recent studies have exploited these attributes to achieve a comprehensive 128 understanding of the dynamic gene expression of multiple inflorescence stages (Zhu et al., 2018) 129 and different tissues and stages of development, under abiotic stresses (Martins et al., 2016;Saha 130 et al., 2016). 131 In this research, we integrated transcriptomic and metabolomic datasets to identify 132 changes that occur in response to herbivore feeding. We identified S. viridis genes that are 133 involved in serotonin biosynthesis and suggest that this pathway plays a role in plant defense 134 against herbivores. By the ectopic expression of a Trp-decarboxylase gene (TDC1), we 135 demonstrated that the encoded enzyme converts Trp into tryptamine. By exposing aphids to an 136 artificial diet supplemented with different concentrations of serotonin and tryptamine, we were 137 able to demonstrate toxic functions of these molecules relative to Trp. Overall, our results 138 indicate that serotonin biosynthesis is triggered by aphid feeding and that it is a deterrent 139 metabolite in S. viridis leaves. 140

Plant growth and insect bioassays 142
The Setaria viridis A10.1 accession (Brutnell et al., 2010;Huang et al., 2014)  The Rhopalosiphum maidis (corn leaf aphid) aphid was reared on B73 maize seedlings as 146 described previously (Meihls et al., 2013). The Rhopalosiphum padi (bird cherry-oat aphid) was 147 reared on barley seedlings (Noga cultivar, Negev Seeds, Israel). For the Spodoptera exigua (beet 148 armyworm) eggs were purchased from Benzon Research (Carlisle, PA, USA), and kept for 48 h 149 in a 29 °C incubator. After hatching, first instar caterpillars were transferred to an artificial diet. 150 For transcriptomic analysis, ten adult R. maidis aphids or three 2 nd -3 rd instar S. exigua were 151 placed on 15-day old S. viridis plants for 96 h (five plants for each replicate), in addition to 152 untreated control plants, and covered with micro-perforated polypropylene bags (whole cage). 153 To identify defense metabolites and confirm the gene expression by using qRT-PCR, 15-day old 154 S. viridis plants were subjected to infestation with ten adult R. padi in a whole cage setup for 6, 155 24, 48 and 96 h, in addition to the untreated control (0 h). 156

Transcriptome sequencing and RNA-seq data analysis 157
Tissues from 4-6 plants were combined into one replicate, and four replicates were collected for 158 each treatment (aphid, caterpillar, and untreated control). Total RNA was extracted using an SV 159 Total RNA Isolation Kit with on-column DNase treatment (Promega) . For 160 transcriptomic analysis, strand-specific RNA-seq libraries were prepared (Zhong et al., 2011;161 Chen et al., 2012). The purified RNA-seq libraries were quantified, and 20 ng of each were used 162 for next-generation sequencing using an Illumina HiSeq2000 instrument at Weill Cornell 163 Medical School with a 101 bp single-end read length. For RNA-seq read quality, values were 164 checked using FASTQC. Then, by using fastq-mcf (https://github.com/ExpressionAnalysis/ea-165 utils/blob/wiki/FastqMcf.md), the adapters and low-quality sequences were trimmed and removed 166 with a minimum length of 50 bp and a minimum quality value of 30 . The 167 RNA-seq data were mapped to the Setaria viridis v2.1(Green foxtail) genome reference 168  Table S1). 174 The RNA was treated with DNase I and then purified further using an RNeasy binding column. 177 Single-strand cDNA was synthesized from 1 µg of total RNA using a Verso cDNA Synthesis 178 Kit (Thermo Fisher Scientific). Quantitative PCR reactions were performed using SYBR Green 179 Master Mix (Bio-Rad), according to the manufacturer's protocol. Primers for the TDC1 gene 180 (Sevir.6G066200) were designed using Primer 3 plus software with the following parameters: 181 product size range of 100-120 bp, primer size of 20 to 25 bp, primer Tm of 57 to 63 C, primer 182 GC% of 45 to 65 % and product Tm of 60 C. To quantitatively determine the steady-state 183 levels of transcripts, the 2 -∆∆Ct method (Livak and Schmittgen, 2001) was used. The expression 184 of genes was normalized against Setaria Eukaryotic Initiation factor 4A (eIF4A; 185 Sevir.4G254900) as an internal reference (housekeeping gene) . Each real-time 186 PCR sample was run in triplicate. For primer list used for the qRT-PCR see supplementary Table  187 S2. 188

Cloning of TDC1 gene and Agrobacterium mediated transformation 189
The full coding sequence of TDC1 (Sevir.6G066200) was retrieved from Phytozome 190 (https://phytozome.jgi.doe.gov) and cloned using GoldenBraid cloning vectors, including pUPD1 191 and p3α2, that were developed by Diego Orzaez (Sarrion-Perdigones et al., 2014). An internal 192 BsmBI site was substituted by a synonymous base, a 6xHA tag that was fused to the 3' and a 193 BsmBI site was added at both ends according to user guidelines (https://gbcloning.upv.es/). The 194 fused gene sequence was synthesized and cloned into pUC57 (acquired from Hylabs, Israel) The 195 3' overhang of the introduced BsmBI site on the TDC1 plasmid was mutated by site-directed 196 mutagenesis PCR to make it compatible for cloning into the pUPD1 vector. The domestication 197 vector, pUPD1, and the TDC1 PCR product were separately digested with BsmBI (Thermo 198 Fisher Scientific) at 37 °C overnight following the manufacturer's protocol. Digested products 199 were separated on a 1% agarose gel, purified and ligated using T4 DNA ligase (Thermo Fisher 200 Scientific). The ligated product was transformed into E. coli DH5α strain using the heat-shock 201 method at 42 °C for 1 min, and selected on Ampicillin, isopropylthio-β-galactoside (IPTG) and 202 5-bromo-4-chloro-3-indolyl-β-D-galactopyranoside (X-gal). The positive colonies (white) were 203 confirmed by PCR followed by sequencing. PUPD1 harboring TDC1, the Arabidopsis thaliana 204 Ubiquitin3 promoter, and THI4 terminator plasmids were assembled into destination vector p3α2 205 using BsaI restriction-ligation reaction. The restriction-ligation reaction was set up by mixing 75 206 ng of each plasmid, 1 ul BsaI, 1 ul T4 ligase, 2.5 ul 10x ligase buffer and 2.5 ug BSA. The 207 reaction was incubated for 10 min at 37 °C and 100 cycles of (3 min at 37 °C and 4 min at 16 208 °C), followed by 10 min at 50°C and 10 min at 80 °C. The ligated product was transformed into 209 E. coli DH5α competent cells as described above. Transformants were selected with Kanamycin, 210 IPTG, and X-gal, and the clones with correct assembly were confirmed by restriction with 211 HindIII. The primer sequences that were used for cloning are described in Supplementary Table  212 S3. 213

TDC1 protein expression in Nicotiana benthamiana plants and immunoblot analysis 214
For the transient expression of TDC1, a plasmid was transformed into Agrobacterium 215 tumefaciens strain GV3101 by heat-shock. Agroinfiltration was performed as described 216 previously (Wieland et al., 2006). In brief, overnight grown bacterial cultures were centrifuged 217 and the pellet was re-suspended in agroinfiltration medium including 10 mM of MES pH 5. Soluble protein extracts were separated from cell precipitates by centrifugation at 15,000 g for 10 227 min and boiled in protein sample buffer (Manela et al., 2015). Ponceau staining and de-staining 228 were performed using Ponceau dye (Sigma-Aldrich) according to the manufacturer's instructions. 229 The immunoblot was performed as previously described (Stepansky and Galili, 2003), using 230 monoclonal anti-HA antibodies (Sigma-Aldrich) and Clarity Western ECL Substrate (Bio-rad) 231 following the manufacturer's instructions and visualized in a chemiluminescence imager 232 (Chemi-DoC, Bio-rad). The protein activity was validated by using GC-MS as described below. 233 Metabolites were extracted using 400 mg of ground frozen plant tissue mixed with solvents 235 containing a ratio of methanol/water/chloroform of 55:23:22, v/v following the protocol as 236 previously described (Rosental et al., 2016). After phase separation, 300 Rapid Separation LC System attached to 3000 Ultimate diode array detector attached to Thermo 259 Q-Exactive mass spectrometer (Thermo Fisher Scientific). The samples were separated on a 260 Titan C18 7.5 cm×2.1 mm×1.9 μ m Supelco Analytical Column (Sigma-Aldrich), using the same 261 parameters as previously described (Handrick et al., 2016). Raw mass spectrometry data files 262 were processed with XCMS (Smith et al., 2006) and CAMERA (Kuhl et al., 2012), software 263 packages for R. Both negative and positive ionization data sets were transferred to Microsoft 264 Excel and presented in Supplementary Table S5.  265 266 Assessment of aphid performance on an artificial diet 267 A previously described artificial diet assay involving indole alkaloids from Gramineae and R. 268 maidis (Corcuera, 1984) was modified to measure the effects of Trp, tryptamine, and serotonin 269 on R. padi survival. Aphid artificial diet, composed of twenty different essential amino acids and 270 sucrose was adapted from Prosser and Douglas (1992)

Statistical analyses 280
Data for partial least-squares discriminant analysis (PLS-DA) plots were normalized as follows: 281 an average of each parameter was calculated across all samples (untreated, caterpillar and aphid 282 infestation), and each individual parameter was divided by its own average and subjected to a 283 Log 2 value (Shavit et al., 2018). The plots were drawn using MetaboAnalyst 3.0 software (Xia et 284 al., 2009 Table S1). The transcript levels were normalized, and partial least-squares 297 discriminant analysis (PLS-DA) was conducted ( Fig. 2A). Biological replicates from each 298 treatment, as well as the control, clustered together, highlighting the reproducibility of the 299 experiment. The PLS-DA plot showed that samples from the aphid and caterpillar treatments 300 clustered separately from the control samples, indicating that a substantial change in the 301 transcriptome occurred due to aphid and caterpillar feeding (components 1 and 2 explained 69.3 302 % of the variance). The differences between the total up-and down-regulated transcripts were 303 calculated for each treatment and are presented in Venn diagrams showing genes with significant 304 expression differences of P value To understand the metabolic changes which occur following herbivore infestation, the 312 same leaf samples were subjected to untargeted liquid chromatography-tandem mass 313 spectrometry (LC-MS/MS) in negative and positive ion modes (Supplemental Table S5). The 314 PLS-DA clustering pattern of the untargeted metabolomics analysis (both negative and positive 315 ion modes) showed that samples of each treatment were clustered separately from the control. 316 However, the aphid-infested samples were located between the control and caterpillar-infested 317 clusters, which indicates an intermediate effect on the aphid-infested plants (Fig. 2B). Mass 318 features with significant differences (P value ≤ 0.05) and at least 2-fold changes relative to the 319 controls were selected, resulting in a total of 99 and 383 mass features significantly altered by 320 aphids and caterpillars, respectively. A total of 224 mass features were increased, 15 by aphids, 321 193 by caterpillars and 16 by both insects. Additionally, a total of 199 mass features were 322 reduced, 25 by aphids, 131 by caterpillars and 43 by both insects (Fig. 2D). This too suggests a 323 common pattern of gene expression and a unique one for each individual insect. The caterpillar 324 feeding modified a larger number of transcripts and mass features than the aphids which suggests 325 a stronger effect of caterpillar infestation. The similarity of the transcriptomic and metabolomic 326 dataset clustering, suggests both overlap in plant responses to herbivore attack, as well as a 327 unique pattern for each herbivore. 328

Pathway enrichment of the transcriptome data 329
To elucidate the metabolic processes that are involved in each gene group shown in the 330 Venn diagrams (Fig. 2C), over-representation pathway enrichment analysis was performed using 331 MetGenMAP (Joung et al., 2009) to compare to rice orthologues (LOC gene ID; table  332   Supplemental Table S6). In Table 1 the significantly enriched pathways of up-and down-333 regulated genes are presented (Table 1A and 1B respectively). The pathways that were 334 significantly enriched by both caterpillar and aphid feeding were mainly associated with amine 335 and polyamine degradation, amino acid metabolism, and biosynthesis of carbohydrates, cell 336 structures, fatty acids, lipids, hormones and secondary metabolites. The genes related to amino 337 acid biosynthesis and their degradation into secondary metabolites of phenylpropanoids and 338 salicylic acid and cell structures biosynthesis (suberin) were associated with caterpillar feeding. 339 Additionally, genes related to secondary metabolites degradation (betanidin) and cofactor, 340 prosthetic group, electron carrier biosynthesis (chlorophyllide a biosynthesis) were uniquely 341 enhanced by aphid feeding. As presented in Table 1B, only a few pathways were significantly 342 enriched among the down-regulated genes by both insects, including the degradation of amino 343 acids, Tyr, Arg and beta-alanine, and citrulline biosynthesis as well as coenzyme A and D-lactate 344 fermentation. Down-regulated genes from the amino acids Val, Ile, Leu and Thr were enriched 345 by caterpillar feeding as well as cofactors, prosthetic groups, electron carrier biosynthesis, Acyl-346 CoA thioesterase, and triacylglycerol degradation and similar pathways were enriched by aphid 347 feeding. No significantly altered pathways were found in the group significantly up-regulated by 348 caterpillars and down-regulated by aphids (Table 1C). Together, these observations indicate that 349 there is a shift in primary metabolism processes including amino acids, carbohydrates, and fatty 350 acids and lipids and the production of secondary metabolites, cell structures and phytohormones 351 in response to aphid and caterpillar feeding on S. viridis. 352 pathways are differentially regulated by aphid and caterpillar feeding (Table 1) are required by Hormonometer (a total of 9,017 orthologous genes were used; Supplemental 364 Table S7). As presented in Fig. 3, the most common insect-induced S. viridis gene expression 365 changes were associated with jasmonic acid, abscisic acid and auxin-dependent signaling. 366 Salicylic acid-responsive genes were induced by both insects, with higher induction by aphids. 367 Also, there was an overall positive correlation between herbivore-induced genes and those that 368 were induced within 60 and 180 min of cytokinin and brassinosteroid-dependent signaling and 369 39 h of gibberellin treatment. Ethylene-responsive genes showed a general pattern of negative 370 correlation with herbivore feeding from S. viridis leaves, except for the first treatment of 30 min 371 that was positive in aphid feeding. Overall, the hormone patterns induced by aphids and 372 caterpillars were similar, with some differences in the time of hormone treatments. 373

TDC1 ectopic expression in planta 417
To understand the function of TDC1 in planta, its entire coding region was amplified and 418 fused into an expression plasmid using the GoldenBraid system (Sarrion-Perdigones et al., 419 2014). The gene was fused to an HA-tag, and the TDC1-HA construct was agroinfiltrated into 420 tobacco leaves. Five days after leaf infiltration, the tobacco leaves were harvested, and 421 expression of the TDC1 protein was quantified by immunoblot. A 63 kDa protein was detected 422 in the extract from transfected tobacco. No band appeared in control with an empty vector, p3α2 423 ( Fig. S1). To validate the function of this protein, metabolites were extracted from tobacco 424 leaves and analyzed by GC-MS analysis. This analysis reveals the appearance of tryptamine in 425 TDC1 transfected leaves, while tryptamine was not detected in the mock-transfected and empty 426 vector leaves (Fig. 5). Overall this suggests that TDC1 can catalyze decarboxylation of Trp to 427 produce tryptamine. 428

Identification of serotonin pathway metabolites in response to aphid feeding 429
To trigger the biosynthesis of defensive metabolites, we performed a time course 430 experiment of infesting S. viridis leaves with R. padi aphids for 0, 6, 24, 48 and 96 h. We used 431 GC-MS, to detect the levels of the shikimate-derived metabolites including shikimic acid, Trp, 432 tryptamine and serotonin (Fig. 6). The analysis revealed that shikimic acid is accumulated in the 433 leaves after 24 and 48 h, and serotonin is accumulated after 24 and 96 h. No significant change 434 was detected in the levels of Trp and tryptamine. This is the first indication that serotonin is 435 present in S. viridis plants and accumulates in response to aphid feeding. For more information 436 about other primary metabolites including amino acids, nucleosides, organic acids, sugars, and 437 sugar alcohols, that were modified in response to aphid feeding see Supplemental Table S4. 438 To obtain a more detailed understanding of the contrasting roles of tryptamine and 439 serotonin in aphid resistance, we conducted in vitro feeding bioassays with purified compounds. 440 Tryptamine and serotonin were added to an aphid artificial diet at concentrations similar to a 441 previous study (Corcuera, 1984). Trp was used as a control. As shown in Fig. 7, the artificial diet 442 bioassays indicated that, after exposure for two days to the compounds, serotonin was more toxic 443

DISCUSSION 446
In order to investigate the defense mechanisms of plant responses to herbivore feeding, it 447 is necessary to consider multiple herbivores from different feeding guilds. For example, 448 transcriptomic analysis of Arabidopsis thaliana leaves infested by either leaf chewing 449 caterpillars or piercing/sucking aphids identified only a small set of genes that were commonly 450 up-or down-regulated (Appel et al., 2014). However, many comparative studies indicated that 451 chemical defense mechanisms and phytohormones such as jasmonic-and salicylic acid were 452 modified in response to these herbivores (Erb et al., 2012;Kroes et al., 2016Kroes et al., , 2017Pandey et 453 al., 2017). Therefore, we infested S. viridis leaves with either S. exigua caterpillars or R. maidis 454 aphids to unravel the differences in transcriptomic and metabolic effects the two herbivores 455 induce in plants. Comparing the overexpression pathway enrichment and hormonal signature 456 suggested both a unique, as well as a common pattern of gene expression for each herbivore (Fig.  457   6 and Table 1). 458 The over-representation pathway enrichment analysis demonstrated that genes associated 459 with amine and polyamine degradation, amino acid metabolism, and biosynthesis of 460 carbohydrates, cell structures, fatty acid, and lipids, hormones, and secondary metabolites were 461 significantly enriched by aphid and caterpillar feeding (Table 1). These observations indicate that 462 there is a shift in primary metabolism and secondary metabolites. The time point experiment of 463 the exposure of S. viridis plants to aphid feeding revealed that the metabolic changes occur 464 mainly in the first few hours after infestation ( Fig. 6 and Supplemental Table S4). 465 High throughput methods like RNA-seq analysis are widely used to identify the 466 differential expression of genes (Huang et al., 2018). Hence, RNA-seq was performed to find the 467 differential expression of genes in S. viridis leaves following 96 h of herbivore feeding. This 468 analysis showed that putative serotonin biosynthesis genes; TDC and T5H were upregulated 469 (Table 2). In a previous experiment, it was observed that tryptamine accumulation is preceded by 470 transient induction of TDC, indicating that enhanced Trp production is linked to the formation of 471 serotonin from Trp via tryptamine (Kang et al., 2007;Ishihara et al., 2008b). In our study, the 472 expression levels of TDC1 (Sevir.6G066200) of this gene was upregulated in S. viridis infested 473 with either aphids or caterpillars. This suggested that these genes, which are highly expressed in response to insect feeding, are involved in serotonin biosynthesis. The TDC1 gene was cloned 475 and ectopically expressed in tobacco by the Agrobacterium-mediated transformation. The 476 expression and activity of TDC was demonstrated by accumulation of tryptamine. Serotonin was 477 not detected in the TDC1-transfected tobacco leaves, which suggests that tobacco lacks this 478 pathway. The function of the candidate cytochrome P450 T5H genes were not shown yet and 479 requires further research. 480 Trp-related secondary metabolites such as tryptamine and serotonin function as defensive 481 metabolites against pathogens and insect feeding in Panicoideae. For instance, rice plants 482 accumulate these compounds in response to herbivore attacks (Ishihara et al., 2008b). The main 483 finding of our research is that S. viridis leaves accumulate serotonin 24 and 96 h after aphid 484 infestation (Fig. 6). Serotonin is a neurotransmitter hormone that plays a key role in mood in 485 mammals. However, in plants, it has physiological roles in flowering, morphogenesis and, most 486 importantly, adaptation to environmental changes (Kang et al., 2007). Nevertheless, the effects 487 of serotonin on herbivore growth and feeding behavior is not yet clear. For example, previous 488 experiments using mutants of CYP71A1, which encodes to T5H enzyme, showed that lack of 489 serotonin increased resistance to planthopper (Nilaparvata lugens) (Lu et al., 2018), while 490 exposure to the striped stem borer (Chilo suppressalis) induced serotonin synthesis in wild type 491 rice plants (Ishihara et al., 2008a). This suggests that the function of serotonin in defense is 492 species-specific. Hence, we evaluated the effects of an artificial diet, supplemented with Trp, 493 tryptamine, and serotonin at different concentrations, on aphid survival. The artificial diet 494 experiment showed a reduction in aphid survival at 0.28, 0.56, and 1.1 mM serotonin. The aphid 495 survival rate was reduced on the first day in all three supplemented metabolites. The survival of 496 aphids on both Trp or tryptamine supplemented diet was not significantly modified on the second 497 day. Tryptamine and its derivatives are neuroactive substances that have been found to act as 498 insect oviposition-deterring and antifeedant agent or inhibitor of larval and pupal development 499 (Gill et al., 2003). The artificial diet supplemented with tryptamine also revealed that aphid 500 survival rate was decreased as tryptamine concentration increased, but with no significant 501 difference relative to control at 0.28 and 0.56 mM on the second day. The adverse effects of 502 tryptamine on feeding (Gill and Ellis, 2006) caused the aphids to die from starvation.    confidence intervals. C) and D) Venn diagrams illustrating the number of individual transcripts and mass spectrometry features that were significantly up-or down-regulated by each treatment, respective to untreated control. P value < 0.05, and fold change greater than 2 or less than 0.5.