Trehalose and glucose levels regulate feeding behaviour in two color morphs of Acyrthosiphon pisum Harris

Trehalose serves multifarious roles in growth and development in insects. We have previously shown that trehalose regulates Acyrthosiphon pisum chitin metabolism. Accordingly, we hypothesize here that trehalose-regulated A. pisum responses in chitin metabolism might also include trehalose-regulated feeding behaviour to involve in chitin metabolism. When RNA interference (RNAi) of trehalose-6-phosphate synthase gene increased the percentage of E2 (i.e. phloem ingestion) waveform and decreased the percentage of F (i.e. stylet work) and G (i.e. xylem ingestion) waveforms compared with the control A. pisum. RNAi of trehalase gene did not affect the percentage of each waveform compared with the control A. pisum. The high trehalose and glucose diets increased the percentage of E2 waveform of red A. pisum. The multiple nonlinear regression shown that the both low trehalose and glucose levels increased the percentage of np (i.e. non-probing), E1 (i.e. phloem salivation), and E2 waveforms. The high or low trehalose:glucose (T:G) ratio decreased the percentage of np, E1, and E2. Interestingly, the percentage of C (i.e. probing), F and, G waveforms were increased at low, low, and high T:G ratio, respectively. The results provided strong evidence that the trehalose and glucose levels regulate A. pisum feeding behavior.


Background
Understanding insect feeding behaviours is important for insect pest management. Previous studies have shown that insect feeding behaviour is strongly in uenced by biotic and abiotic factors [1][2][3][4][5] , as well as by the change in its physiology status, host plant nutrition and species 6 , and resistance to pesticide 3,7 .
Aphids use their stylets to obtain nutrients from sieve tubes of plant tissue, and ingest passively on the phloem, driven by the pressure in the sieve tubes, and actively on the xylem, intercellular apoplastic and epidermal 8 . The stylet penetrates into the plant tissue and forms a stable food channel with secreting saliva to ingest plant sap 9 . The electrical penetration graph (EPG) has been used to monitor stylet activity, saliva excretion and food ingestion during aphid feeding recorded as different EPG waveforms associated with specific stylet tip positions and activities 3,7,10 . The EPG waveform np, C and E1, represent, respectively, non-probing, intercellular apoplastic stylet pathway and salivation into phloem sieve elements at the beginning of the phloem phase 4,7 . While the EPG waveform E2, G and F are correlated with passive phloem sap uptake from sieve element, active intake of xylem sap and derailed stylet mechanics, respectively 4,7 . Interestingly, it was reported that aphids aposymbiotics (endosymbiotic bacteria Buchnera aphidicola disruption), pesticides and pathogen Pandora neoaphidis affected feeding behaviours of piercing-sucking insects [3][4][5]7 . However, the studies of the effects of body sugar level on aphid feeding behaviours and EPG waveforms are very limited.
Sugar such as trehalose, a non-reducing disaccharide in insect hemolymph, is formed by two glucose molecules linked by an α-α bond and widely present in bacteria, fungi, insects, and plants [11][12][13] . It is mainly present in hemolymph of insects and typically occurs at a high concentration; whereas glucose may occur together with trehalose, but at a signi cantly lower concentration 14 . Trehalose plays important roles in the growth, development 15,16 , ight, feeding 17,18 , overwinter and diapause 19 of insects. Simpson and Raubenheimer (1993) 20 suggested that hemolymph trehalose level re ects the nutritional status of the insect and may serve a role in regulating food choice and nutrient consumption. Dietary nutrient levels on gluconeogenesis in Manduca sexta was positively correlated to hemolymph trehalose levels 21 , and the ratio of carbon to nitrogen from carbohydrate absorption affected the growth and development of Acyrthosiphon pisum 2,22 . A growing number of studies have found that elevation of trehalose rich has great agronomic potential to improve the stress tolerance of plants 23,24 . Trehalose serves multifarious roles in insects such as facilitating carbohydrate absorption, being a source of energy, and a component of a feedback mechanism regulating feeding behaviour and nutrient intake 13,25 . However, the feedback mechanism of rich trehalose on the feeding behaviours of piercing-sucking insects have not been clari ed yet.
It is well known that trehalose-6-phosphate synthase (TPS) and trehalase (TRE) can directly or indirectly affect trehalose content and feeding behaviour 16,26-28 . Knockdown of TRE genes increased trehalose content and reduced food intake of S. exigua 26 , while knockdown of TPS reduced trehalose content but did not affect feeding behaviour of Nilaparvata lugens 16 Bactrocera minax 28 and Leptinotarsa decemlineata 27 . We have recently shown that RNAi of ApTPS and ApTRE regulate the chitin metabolism of A. pisum 29 . However, the study on the TPS and TRE of the pea aphid and the detailed relationship between trehalose level and feeding behaviour is still unclear. Therefore, a more comprehensive study of trehalose level and feeding behaviour is necessary because sugars are the main components of plant sap that aphids feed on, and would lay a solid foundation for further investigation of molecular biology and physiology of this pest.
In this study, we report (1) the effect of high sugar diets and knockdown of TPS and TRE expressions on the body trehalose and glucose contents of red and green A. pisum, (2) the stylet activity thus feeding behaviour of these treated aphids, and (3) the relationships between A. pisum fooding behaviour and the level of body trehalose and glucose. The results help to provide a theoretical basis for further development of biological agents targeting the feeding behaviours against A. pisum.

Results
Effect of RNAi and high sugar diets on ApTPS and ApTRE expression. Figure 1 shows the effects of the RNAi and high sugar (trehalose and sugar) diets on the expression level of ApTPS and ApTRE genes in both red and green of the A. pisum at 24 h and 48 h on fresh leaves after the RNAi treatments relative to the expression levels in the A. pisum treated with the normal diet. The ApTPS expression was signi cantly decreased for both red and green A. pisum at 24 h and 48 h after the dsTPS RNAi treatment (Fig. 1A), and was decreased at 24 h but increased signi cantly at 48 h after the dsTRE RNAi treatment by more than 2 folds in both biotypes compared with that in dsGPF-treated A. pisum (Fig. 1A). However, the ApTRE expression was downregulated by both dsTPS-and dsTRE-treatment relative to that in the dsGFP-treated A. pisum (Fig. 1B).
The ApTPS expression was downregulated at 24 h and upregulated at 48 h on fresh leave after the high trehalose diet treatment for both biotypes (Fig. 1A). After the high glucose diet treatment, the ApTPS expression was upregulated only at 48 h on fresh leave for the green biotype (Fig. 1A). The ApTRE expression was downregulated by the high sugar diets in most cases (Fig. 1B), apart from in the high trehalose-treated red biotype A. pisum where the ApTRE expression was upregulated at 24 h on fresh leave after the treatment. Notably, unlike the red A. pisum, the survival rate of the green A. pisum was signi cantly decreased by the RNAi treatments (P < 0.001, Fig. S1). The reproduction (the total number of the offspring) was signi cantly decreased by the dsTPS and dsTRE treatments relative to that of the dsGFP-treated parents and by the high sugar diets relative to that of the untreated A. pisum (CK) (P < 0.001, Fig. S2). The expression of ApTPS and ApTRE, survival and reproduction had a similar trend between red and green A. pisum.
Trehalose and glucose contents. The effect of the RNAi-treatment and the high sugar diets on the trehalose and glucose contents were determined in the red and green A. pisum. There was no difference in both the trehalose and glucose contents between untreated control A. pisum (CK) and the dsGFPtreated A. pisum (dsGFP) (Fig. 2). The trehalose contents were decreased in the dsTPS-treated A. pisum but increased in the dsTRE-treated A. pisum in all cases compared with those in CK and dsGFP-treated A. pisum ( Fig. 2A). The glucose contents were decreased in the dsTPS-and dsTRE-treated A. pisum at both time points (24 h and 48 h) for both red and green A. pisum (Fig. 2B). The trehalose contents were not in uenced by the high trehalose diet but decreased by the high glucose diet ( Fig. 2A). However, both high sugar diets increased the glucose contents ( Fig. 2B). In addition, the content of trehalose and glucose had a similar trend between red and green A. pisum.
Feeding behavior. Figure 3 shows the feeding activities recorded as EPG waveforms when the A. pisum probe into plants and presented as the percentage of each EPG waveform, and an overview of the representative EPG waveforms of treatment and control A. pisum on both time points is shown in Fig. S3-6. At 24 h on fresh leaves after the treatments, no change in any EPG waveform was found in the dsTPStreated and dsTRE-treated red A. pisum compared with the dsGFP-treated A. pisum, which was not different from those of the untreated A. pisum (CK) ( Fig. 3A; Tab. S2A). In the green A. pisum, the percentage of E2 waveform was increased by the dsTPS-treatment and decreased by the dsTREtreatment ( Fig. 3C; Tab. S2C). The high sugar diets decreased and increased the percentage of E2 waveform of the red and green A. pisum, respectively. The high sugar diets also increased the percentage of G waveform of the red A. pisum ( Fig. 3A; Tab. S2E) and the percentage of F waveform of the green A. pisum ( Fig. 3C; Tab. S2G).
After 48 h treatment, the dsTPS-treatment increased the percentage of E2 waveform and decreased the percentage of F and G waveforms compared with the CK (Fig. 3B and 3D; Tab. S2B and S2D), but the dsTRE-treatment did not affect the percentage of any waveform compared with the CK group ( Fig. 3B and 3D; Tab. S2B and S2D). The high sugar diets had little effect on the EPG waveforms ( Fig. 3B and 3D; Tab. S2F and S2H) of both the red and green A. pisum. In addition, the feeding behaviours had a similar trend between red and green A. pisum.
Relationships between feeding behaviour and physiological sugar level. To illustrate the relationshops of physiological sugar levels (trehalose and glucose levels) on the feeding behaviour, the relationships between the physiological sugar levels and the percentages of EPG waveforms were assayed using multiple nonlinear regression at 48 h. As Figure 3B and 3D shown that the EPG waveforms were huge difference between the each group at 48 h. These data shown that the percentage of np waveform was increased at the low both trehalose and glucose levels and the high both trehalose and glucose levels, while was decrased at the low trehalose level and high glucose level and at the high trehalose levels and low glucose level, and the regression equation was (R 2 = 0.371, Fig. 4A). Where z, x, and y are the percentage of the waveform, trehalose content and glucose content, respectively. The percentage of C waveform was increased with trehalose content elevalted, which was rstly increased and then decreased with glucose content rised, and the regression equation was (R 2 = 0.5302, Fig. 4B). The percentage of E1 waveform showed the same trend as did the percentage of np waveform, and the regression equation was and (R 2 = 0.347, Fig. 4C). Interestingly, the percentage of E2 waveform was increased at the level of low trehalose and low glucose, low trehalose and high glucose, high trehalose and low glucose, and high trehalose and high glucose, and the regression equation was (R 2 = 0.6975, Fig. 4D) and the minimum at the point (0.44, 0.090). The percentage of G waveform was rstly increased and then decreased with trehalose content rised, which was increased with glucose content rised, and the egression equation was Fig. 4E). The percentage of F waveform was rstly increased and then decreased with trehalose content or/and glucose content rised, and the egression equation was (R 2 = 0.6218, Fig. 4F).
Trehalose and glucose are the two main sugars in the insects hemolymph and they plays a important role in food-choice behaviour 30 . We assayed whether the balance of trehalose and glucose induce a foodchoice feeding behaviour in A. pisum. As was observed that the top or bottom contour each subgraph of Fig. 4. There was a small difference in the percentage of np, E1, and E2 waveforms occurred between high trehalose:glucose (T:G) ratio and low T:G ratio, but decreased the percentage of np, E1, and E2 ( Fig.  4A, 4C, and 4D). Interestingly, the percentage of C, F and, G waveforms were increased at low, low, and high T:G ratio (Fig. 4B, 4E, and 4F), respectively.

Discussion
Sugar metabolism plays a critical role in aphids' adaptation under various environmental conditions. Our results showed that high trehalose-diet feeding did not increase A. pisum hemolymph trehalose content, but increased glucose content (Fig. 2). The trehalose contents were decreased by the high glucose diet ( Fig. 2A) and the glucose contents were increased by both high sugar diets (Fig. 2B), suggesting that glucose may be crucial to the A. pisum growth and development. The A. pisum may utilize lower glucose level in the hemolymph to regulate survival, reproduction and feeding behavior, and so are sensitive to the change in the hemolymph glucose level, while trehalose is stored as an energy resource. The insects' body uid balance is also in uenced by intrinsic and extrinsic factors such as food, gluconeogenesis, and glycogenolysis 6,14,21 . It is possible that, when the trehalose content in A. pisum's hemolymph is very high, it would be hydrolyzed to produce glucose. Gluconeogenesis contributed greatly to hemolymph sugar in insects maintained on a low carbohydrate diet, but on a high carbohydrate diet, the hemolymph sugar was derived mainly from dietary carbohydrate, whereas the generation of amino acids was regulated post-ingestively 21,31 .
It was also observed that the high sugar diets did not only affect the survival rate but also reduced the reproduction of red and green A. pisum (Fig. S1 and S2). It was reported that high trehalose negatively affected food intake of L. decemlineata and Spodoptera exigua 27,32 . High sucrose diet decreased the consumption rate of A. pisum, and a low sucrose diet increased food ingestion of Ceratitis capitata female by 35% compared with the control 22,33 . Moreover, the trehalose rich reduced reproduction and growth in insects 27,32,34 , it was found that over-accumulation of trehalose reduced L. decemlineata survival 27 and Drosophila melanogaster adaptation 34 .
The RNAi of ApTPS decreased trehalose content of the A. pisum as in B. minax and L. decemlineata 27,28 . However, the trehalose contents were increased in the dsTRE-treated A. pisum in all cases compared with those in CK and dsGFP-treated A. pisum ( Fig. 2A). This is contradictory to the report in S. exigua larvae 32 . Thus, the effects of RNAi of TRE on the glucose content may be different in different insect species. The glucose contents were decreased in the dsTPS-and dsTRE-treated A. pisum at both time points (24 h and 48 h) for both red and green A. pisum (Fig. 2B), further con rming the sensitive regulation of the glucose level in the A. pisum.
The EPG technique is a useful tool to detect the feeding behaviour of piercing-sucking insects 35 . The high sugar diets and the RNAi of ApTRE did not signi cantly change the percentage of each EPG waveform. The difference in the percentage of EPG waveforms between treatment and control groups was observed only at 48 h ( Fig. 3; Tab. S2). RNAi of ApTPS increased the percentage of E2 waveform and decreased the percentage of F and G waveforms. Overall, the A. pisum spent more time on E2 waveform (phloemfeeding) (Fig. 3). This is consistent with the phloem-feeding activity of the A. pisum for nutrients, such as sucrose and amino acid contained in their host plant 36 .
The measurements of trehalose and glucose contents and the feeding behaviours of A. pisum treated with RNAi and high sugar diets provided an unique opptunity to analyse the relationships of the physiological sugar levels with the feeding behaviours of A. pisum. The analysis showed that the low both trehalose and glucose levels increased the activity of non-probing phase (np) and phloem phase (E1 and E2) but the high both trehalose and glucose levels also increased non-probing phase and phloem phase (Fig. 4A , 4C, and 4D). Low physiological sugar level decreased np increased A. pisum phloemfeeding to obtain more carbohydrates in agreement with the analysis by the EPG recording ( Fig. 3B and  3D). The increase of phloem-feeding time under the low hemolymph sugar levels is a sign that the A. pisum needed more carbohydrates to maintain its homeostasis. However, the high hemolymph sugar levels increased phoem-feeding may be due to the range of the model. Interestingly, the low T:G ratio increased the activity of probing phase (C) and stylet work phase (F; Fig. 4B and 4E) , indicating that A. pisum spent more time feeding in the cell walls, intercellular spaces of vascular tissue, and the mesophyll as when A. pisum feed on resistant plants 4,37-39 . On the other hands, the increased F waveform of aphids was also reported to restore the stylets bundle 4 , or probably due to the differences in salivary components 38 . The high T:G ratio increased the activity of xylem ingestion phase (F) to intaken more water to avoid dehydration (Fig. 4F).
Overall, the ndings from this study suggested that the level and balance of trehalose and glucose A. pisum food-choice behavior. This is the rst reported using the EPG technique to study the link of A. pisum physiological sugar level and feeding behaviour. It provides strong evidence that the feeding behaviour of the A. pisum is in uenced by the level and balance of trehalose and glucose in the body.
In conclusion, this study shows that RNAi of ApTPS and high sugar diets can affect the trehalose and/or glucose content in the body of A. pisum (Fig.1). This allows to analyse the relationships between sugar contents and feeding behaviours under A. pisum physiological conditions. Both low trehalose and glucose levels increased the time of non-probing phase and phloem feeding phase, the low T:G ratio increased the feeding time of probing phase and stylet work, and decreased xylem feeding time. Future research is now required to validate the mechanism of physiological sugar level regulated feeding behaviour.

Materials And Methods
Insect and culture conditions. Clones of red and green morphs of A. pisum were established from single virginiparous females. Samples were collected in 2017 from same Alfalfa plant Medicago sativa in eld, Lanzhou, China, and reared on the fava bean Vicia faba in the laboratory. All plants and A. pisum cultures were reared in an arti cial climate incubator at 20 ± 1℃, 70 ± 10% relative humidity, with a photoperiod of 16 h L: 8 h D. Mature A. pisum were put on a fava bean leaf for 12 h and the resulting neonate nymphs, 0-12 h old, were used for experiments throughout this study.
RNA isolation and rst-strand cDNA synthesis. Total RNA was isolated using TRizol reagent (BBI Life Sciences, Shanghai, China) following the manufacturer's instructions. The total quantity of extracted RNA was assessed using a micro-volume UV spectrophotometer (Quawell Q5000, Quawell, USA). The RNA integrity was con rmed further by 1% formaldehyde agarose gel electrophoresis. Total RNA was dissolved in 50 mL DEPC-water and stored at -80℃. The rst-strand cDNA was synthesized using a First-Strand cDNA Synthesis kit (BioTeke, Beijing, China) and stored at -20℃ for subsequent experiments.  Table S1. The components of the PCR reaction mixture included 1.0 mL of the template (1 ng/mL), 12.5 mL 2× Power Tap PCR MasterMix (BioTeke, Beijing, China), 1.0 mL of each primer (10 mmol/mL), and 9.5 mL Rnase-free H 2 O concentration for a nal volume of 20 mL. The PCR reaction conditions were predenatured at 95℃ for 5 min, followed by 35 cycles of 95℃/45 s for denature 55℃/45 s for annealing and 72℃/1 min for extension, and then 10 min at 72℃ for a nal extension. PCR products were subjected to 1.0% agarose gel electrophoresis and puri ed by DNA gel extraction kit (BioTeke, Beijing, China). The puri ed DNA was ligated into the pMD18-T vector (TaKaRa, Dalian, China) and sequenced by Tsing Ke Biological Technology (Tsing Ke Biological Technology, Beijing, China) using the dideoxynucleotide method. The lengths of the resulting ApTPS, ApTRE, and GFP genes were 421 bp, 416 bp, and 688 bp, respectively. dsRNA synthesis. Three pairs of primers (dsTPS-F/R, dsTRE-F/R and dsGFP-F/R), with the T7 RNA promoter sequence anking the 5'-end of each gene, were designed and synthesized (Tab. S1), and used to make the templates for in vitro dsRNA transcription via PCR. The dsRNAs were synthesized using the TranscriptAid T7 High Yield Transcription Kit (Thermo Scienti c, Wilmington, DE, USA) according to the manufacturer's protocol 40 . The size of the dsRNA products was con rmed by electrophoresis on a 1.5% agarose gel and the concentration was assessed using a micro-volume UV spectrophotometer.
dsRNA and high sugars diet treatments. The arti cial diet bioassay was performed according to the following procedure 41 . A liquid arti cial diet was prepared as described previously 42,43 , ltered through a 2 mm membrane, dispensed in 1.0 mL aliquots, and stored at -20℃ before assays. The testing diets were prepared by adding either each of dsRNA (dsTPS, dsTRE and dsGFP) or each of sugar (trehalose and glucose) to the 1.0 mL arti cial diet for a nal concentration of 400 ng/mL (dsRNA) and 100 mg/mL (sugar). The diet containing nuclease-free water was used as control of the high sugar diet treatments and diet containing dsGFP was used as control of the RNAi treatments. There was a total of 6 treatments including two controls for either red or green A. pisum.
Glass vials (2.5 cm in diameter) were sterilized for the aphid arti cial double-membrane feeding assay and one opening was completely sealed with para lm. Seventy microliters of the testing diet were placed on the para lm and covered with para lm. So the testing diet was sandwiched between two layers of the para lm membrane at one opening of the glass vials 44 . The control group was fed with only the arti cial diet without dsRNA or sugars.
Fifteen 3-day-old A. pisum were introduced into one vial, and the vial was closed with a piece of sterilized gauze as one of bioassays. The arti cial diet was replaced every other day to prevent dsRNA degradation. After 4 days, all surviving A. pisum were transferred to fresh bean leaf discs.
Quanti cation of gene expression levels after RNAi treatments. Seven A. pisum were collected from fresh bean leaf discs at 24 h and 48 h after the 4-day treatment with the testing diet containing each of dsRNAs. A. pisum were immediately frozen in liquid nitrogen and three replicates were carried for each treatment. Total RNA was isolated from the seven pooled whole A. pisum bodies. The rst-strand cDNA was synthesized from total RNA using a First-Strand cDNA Synthesis kit (BioTeke, Beijing, China). The RT-qPCR analysis was carried out in 96-well 0.1-mL block plates using a QuantStudio TM 5 system (Thermo Scienti c, Wilmington, DE, USA). Each reaction contained 1.0 mL of the cDNA template, 10.0 mL 2×Plus SYBR real-time PCR mixture (BioTeke, Beijing, China), 0.5 mL of each primer (10 mmol/mL), 8 mL EDPC-ddH 2 O, and 0.5 mL 50 × ROX Reference Dye concentration for a nal volume of 20 mL. The RT-qPCR reaction conditions were pre-denatured at 94℃ for 2 min, by 40 cycles of 94℃/15 s, and 55-62℃/30 s for annealing. After each reaction, a melting curve analysis (denatured at 95℃ for 15 s, annealed at 60℃ for 1 min, and denatured at 95℃ for 15 s) was conducted to ensure consistency and speci city of the ampli ed product. Three biological replicates and three technical replicates were set for each treatment in the RT-qPCR analysis. Quanti cation of the transcript level was conducted according to the method 45 , and the ribosomal protein L27 gene (rpL27) was used as a reference gene 46 .
Trehalose and glucose content assays after high sugar diet treatments. Ten A. pisum were collected from fresh bean leaf discs at 24 h and 48 h after the 4-day treatment with the testing diet containing each of sugars. A. pisum were immediately frozen in liquid nitrogen and three replicates were carried for each treatment. The trehalose content assay was conducted according to the method described by Yang et al. 16 . Brie y, ten whole A. pisum bodies were ground in phosphate-buffered saline (PBS: 130 mM NaCl; 7 mM Na 2 HPO 4 ·2H 2 O; 3 mM NaH 2 PO 4 ·2H 2 O; pH 7.0), and then a 25 mL of tissue was taken and uniformly mixed with 25 mL of 1% sulfuric acid. The mixture was incubated at 90℃ for 10 min and placed in ice for 3 min, and then 25 mL of 30% potassium hydroxide solution was added into the sample and mixed uniformly. The resultant mixture was incubated at 90℃ for 10 min and then in ice for 3 min. Finally, 500 mL of 0.2% anthrone reagent was added to the sample and incubated at 90℃ for 10 min and then in ice for 3 min. The trehalose content was assayed by measuring the absorbance of the nal reaction mixture at 630 nm. The glucose content was determined using the glucose assay kit (Solarbio Biochemical Assay Division, Beijing, China) according to the manufacturer's protocols. Evaluation of A. pisum feeding behavior. The probing behaviour was evaluated with the electrical penetration graph (EPG) using an 8-channel DC-EPG device (Wageningen University, the Netherlands). Eight plants were placed in a faraday cage, and wingless A. pisum were placed on the abaxial side of the second fully expanded leaf from the top. Before exposure A. pisum to the plant, a 6 to 8 cm long gold wire (diameter 18 mm) was conductively glued (water-based silver glue) to A. pisum dorsum as the recording electrode. The other end of the gold wire was attached to a 3 cm long copper wire (diameter 0.2 mm) which was connected to the rst head stage on the DC-EPG ampli er with the setting of 1 Giga-Ohm input resistance and 50×gain. The reference electrode was inserted into the soil and connected to the plant voltage output of the DC-EPG device. A. pisum from each treatment were randomly distributed during recording. For each treatment, only the A. pisum that showed activities in an 8 h recording period were considered as valid replicates.
The EPG signal was recorded by the Stylet+d software and the EPG waveforms were recognized and labeled using Stylet+av01.30 software (EPG Systems, Wageningen, Netherlands). The EPG parameters were calculated for each A. pisum treatment using the Excel workbook for automatic parameter calculation of EPG data 4.4.3 47,48 and then the means and standard errors of the mean (SEM) were calculated for each treatment at 24 h and 48 h on fresh bean leaf discs after 4-day treatments.
Survival and reproduction assays. A. pisum were reared on fresh bean leaf discs after the treatments in an arti cial climate incubator at 20 ± 1℃, 70 ± 10% relative humidity, with a photoperiod of 16 h L: 8 h D. Survival and reproduction assays were conducted for the control and treated A. pisum. The daily numbers of adult A. pisum deaths and newborn nymphs per adult A. pisum were recorded until they no longer produced nymphs, once per day starting from the rst day after the treatments.
Multiple nonlinear regression. The 48 h trehalose and glucose content data of the A. pisum obtained in 2.7 and the percentage of EPG waveforms data of the A. pisum obtained in 2.8. The relationshops between the percentage of EPG waveforms under each treatment (z) and the corresponding physiological trehalose content (x) and glucose content (y) were then analyzed as by nonlinear curve tting with the softwave 1stOpt 15.0 (7D-Doft High Technology lnc, China), where a 1 , a 2 , a 3 , a 4 , and a 5 : coe cient; b: constant.
Statistical analysis. All statistical analyses were performed using 1stOpt 15.0, SPSS 19.0, and Origin 8.5 were used to construct the histograms. The RT-qPCR and sugar data were analyzed by Student's t-test. The EPG data (Tab. S2) and the total reproduction data (Fig. S2) were analyzed using one-way analysis of variance (ANOVA) followed by the Tukey's post hoc test. The survival data were subjected to a Kaplan-Meier survival log-rank analysis (Fig. S1) 49 . A p-value <0.05 was considered statistically significant. Figure 2 The physiological content of trehalose (A) and glucose (B) in red and green biotypes of the A. pisum. The content of trehalose and glucose were presented as Means ± SEM of three replicates. A. pisum treated with RNAi of GFP gene (dsGFP), A. pisum treated with RNAi of TPS gene (dsTPS), A. pisum treated with RNAi of TRE gene (dsTRE), A. pisum treated with high trehalose diet (T100), A. pisum treated with high glucose diet (G100) and A. pisum fed with normal diet (CK). All data were analyzed using Student's t-test.

Declarations
The asterisk indicates signi cant differences between treatment and control (*P < 0.05).  Relationship between feeding behaviour and the physiological sugar levels of A. pisum. The percentage of each waveform is the mean of three measurements at 48 h after each treatment (dsTPS, dsTRE, dsGFP, T100, G100 and normal diet). Z-axis: the percentage of EPG waveform by multiple nonlinear regression, where x is trehalose contents; y is glucose contents; X-axis: trehalose content; Y-axis: glucose content; The np waveform (A), C waveform (B), E1 waveform (C), E2 waveform (D),F wavefor (E), and G