Plants: the hexaploid wheat variety Triticum aestivum L. cv Zhengmai 7698 (ZM7698) was used in this study. A total of 30-35 wild-type and transgenic wheat plant seeds were sown in pots and were cultured in a climate chamber at 22°C under a 16-h photoperiod, and with a relative humidity of 40%-60%.

Insects: grain aphids, S. miscanthi were reared on two-leaf stage aphid susceptible wheat seedlings in a controlled chamber with similar conditions than for plant growing. Apterous adult grain aphids from a single clonal lineage were reared on wheat seedlings in a continuous culture for 24 hours to produce synchronized nymphs. After that, the adults were removed, and the offspring were used in subsequent experiments. All experiments were carried out in a climate chamber under the above-mentioned conditions.

Total RNA of pooled adults was extracted by using TransZol Up (TransGen Biotech, Beijing, China). cDNA was synthesized by using FastKing RT Kit (Tiangen, Beijing, China). The full length of the SmDSR33 gene was obtained using TransStart^® FastPfu DNA Polymerase (TransGen Biotech, Beijing, China) following the instructions. The DNA amplification products were sequenced by the Institute of Crop Sciences (Institute of Crop Sciences, Chinese Academy of Agricultural Sciences, Beijing, China). The theoretical isoelectric point (pI) and molecular weight (MW) of SmDSR33 were calculated through ExPASy (https://web.expasy.org/compute_pi/). The transmembrane region and putative signal peptide were predicted using TMHMM (http://www.cbs.dtu.dk/services/TMHMM/) and SignalP (http://www.cbs.dtu.dk/services/SignalP/), respectively. The counterparts of SmDSR33 of other aphid species were obtained by the Basic Local Alignment Search Tool (BLAST) against the aphidbase database (http://bipaa.genouest.org/is/aphidbase/). Phylogenetic trees of SmDSR33 in twelve aphid species were constructed using the nucleotide acid sequences as a matrix via MEGA X software (www.megasoftware.net). The branch strength was analyzed by using the maximum likelihood method and performing 100 bootstrap replications.

To amplify the 439 bp SmDSR33 target sequence, specific primers were designed. A 320 bp fragment of GFP were selected as a control in the aphid bioassay experiment. The amplified PCR products were recovered and inserted at inverted repetitions into the SpeI/EcoRV and SacI/HpaI sites of the pEasy-Blun-Zero-AdhI vector to construct the hairpin RNAi, Bzero-DSR33-adhI-DSR33. The vector of Bzero-DSR33-adhI-DSR33 was digested by Ssp I and BsrG I to obtain the expression cassette. The latter was recovered for bombardment. The RNAi fragment was driven by the maize Ubi promoter. Bombardment-mediated transformation was applied to immature embryos isolated from ZM7698. Somatic embryos were induced in tissue culture on medium, and whole plants were then regenerated and selected. Healthy seedlings were transplanted to soil to grow until maturity.

The CTAB method was used to extract genomic DNA from young T₃ plant leaf tissues as described by Sambrook et al. (1989). The restriction enzyme was used to digest 35 μg of genomic DNA overnight. The products were fractionated for 12-16 h at 60 V on a 0.8% agarose gel in 1×TBE buffer. The Hybond-N⁺ membranes were used for blotting (Amersham, UK). The digoxigenin (DIG) High Prime DNA Labeling and Detection Starter Kit I (Roche, Germany) was used for prehybridization, hybridization, washing, and detection of the membranes. The primer sets SmDSR33S-F/R were used to synthesized DNA probes ( Supplementary Table 1 ).

For the expression level of SmDSR33 at different development stages, total RNAs of grain aphids were isolated from the four nymphs and adults reared on susceptible wheat. For the expression level of SmDSR33 in aphids fed with different transgenic wheat and wild-type plants, the adult aphids were collected and used for total RNA extraction and further experiments.

The cDNA was synthesized following conventional procedures. A quantitative real-time PCR (qRT-PCR) assay was carried out using the SYBRH Green Real-time PCR Master Mix (Tiangen, Beijing, China) in an ABI 7300 Real Time PCR system. The aphid Actin gene and ribosomal protein S27 A (Rps27) gene were selected as internal controls, and SmDSR33 specific primers were designed for normalization ( Supplementary Table 1 ). All qRT-PCR experiments were performed in triplicate. The relative gene expression of each target gene was calculated by using the mean value of the reference genes through the 2^–ΔΔCT method (Livak and Schmittgen, 2001).

A single clonal lineage of apterous adult grain aphids was reared on wheat seedlings in cages for 24 hours to produce nymphs. The newborn nymphs produced during the period of 24 hours were transferred to fresh transgenic wheat plants.

T₃ homozygous wheat plants with SmDSR33-dsRNA expression were selected to evaluate the effects on aphid survival and fecundity. At the 3-4 leaf stage, 20 neonatal first instar nymphs of S. miscanthi were placed on the leaf of each plant. The mortality of aphids was recorded every day. Ten plants from each line were used in every experiment. The experiment was repeated three times.

Life cycle parameters were calculated as follows: the net reproductive rate, R ₀= ∑^​ l _x ·m _x , the mean generation time, T=∑^​ xl _x m _x /∑^​ L _x m _x the intrinsic rate of increase, r _m =(lnR ₀)/T , and the finite rate of increase, λ =e ^{r _m} . In the equations, l _x is the surviving rate to a specific age x , and m _x is the number of new-born nymphs produced by per live adult for a specific age x (Biondi et al., 2013).

The Giga-8 DC EPG amplifier (EPG-Systems, Wageningen, Netherlands) and a Faraday cage was used to record the probing and feeding behaviors of apterous adult aphids on wheat. Firstly, synchronous adult aphids were inoculated on 33-592 transgenic plants and control plants for two days, respectively. Then, the aphid was starved for 2 h. After that, water-soluble silver conductive paint was used to attach each aphid to a flexible gold wire (18 μm diameter×2 cm length) through the dorsal thorax individually. The aphids were placed onto the adaxial side of a leaf from transgenic and wild-type wheat plants at the three-leaf stage, and the opposite ends of the gold wires (2 mm in diameter×3 cm length) were connected to copper wire with conducting silver glue, which was connected to a DC amplifier. The plant electrode was inserted into the soil. Under light conditions, the EPG signal of each individual was continuously monitored for 8 h. We monitored 12 behavioral recordings for each treatment. The software Stylet+a (EPG-Systems) was used to analyze EPG signals. According to the method described by Tjallingii (Tjallingii, 1985; Tjallingii, 1994), the different waveforms were correlated with feeding behavior. Non-probing (np) waveform, which reflects stylet external to wheat leaf tissue. Pathway phase contains two waveforms, waveform C, which reflects the intercellular stylet pathway, potential drops (pd), which reflects intracellular punctures during intercellular pathway. Waveform G (xylem phase) is the only waveform that reflects active sap ingestion from xylem elements. Phloem phase can be divided into two phases: E1 always occurs at the start of the phloem phase and reflects saliva secretion into the sieve element, E2 reflects passive phloem sap ingestion. EPG data was analyzed using the EPG-Excel data workbook provided by Sarria et al. (2009).

The two-tailed Student’s t-test was used to evaluate the differences between wild-type and transgenic wheat lines. For all comparisons, significance (P value) was calculated at the 1% or 5% level. The standard error of the mean (SEM) for each treatment was calculated using three biological replicates. For the EPG experiments, means and standard errors of variables were calculated from recordings per individual aphid, and differences were analyzed by Student’s t-test. All data represents means ± SEM.

We identified a candidate gene SmDSR33, which encoding a putative salivary protein in grain aphid, based on transcriptomic profiling and dsRNA feeding assay (Wang et al., 2015a). The full-length cDNA sequence of SmDSR33 was 534 bp in length, encoding a 177 amino acid putative salivary protein. The SmDSR33 protein was predicted to have an Mw of 19.376 kDa and a pI of 6.26, possess a signal secretion peptide with a predicted cleavage site between amino acid residues 20 and 21 and have one predicted transmembrane helix, suggesting that SmDSR33 was a secreted protein ( Figure 1A , Supplementary Figure S1 ).

To clarify the evolutionary relationships of this gene in different insect species, sequences of SmDSR33 counterparts in pea aphid (A. pisum), soybean aphid (Aphis glycines), cotton aphid (A. gossypii), Russian wheat aphid (Diuraphis noxia), sugarcane aphid (Melanaphis sacchari), black cherry aphid (M. cerasi), peach aphid (M. persicae), banana aphid (Pentalonia nigronervosa), corn aphid (Rhopalosiphum maidis), bird cherry-oat aphid (Rhopalosiphum padi), and yellow sugarcane aphid (Sipha flava) were obtained by BLAST against aphidbase database and NCBI. The phylogenetic tree of SmDSR33 was constructed via MEGA X software. Phylogenetic analysis demonstrated SmDSR33 was more closely related to its orthologs in the pea aphid (A. pisum) ( Figure 1B ).

We the used qRT-PCR to investigate the SmDSR33 expression level in grain aphids at different developmental stages. Results revealed that SmDSR33 transcription was accumulated throughout the developmental phases at different levels ( Figure 1C ). The SmDSR33 expression pattern peaked in the adult aphid and was about 1.6-fold higher compared to first instar nymphs.

To investigate the function of SmDSR33, a 439 bp fragment of SmDSR33 gene was selected as a template for RNAi target ( Figure 2A ). We used BLAST against the NCBI database to evaluate the specificity of the SmDSR33 fragment. At the nucleotide acid level, no continuous three 21-nt matches were detected between the selected 439 bp fragment and aphid natural enemies or humans (data not shown), which implied that the selected dsSmDSR33 fragment would not pose potential risks to non-target organisms (Bachman et al., 2013). Then, the RNAi vector harboring SmDSR33-hairpin DNA was constructed ( Figure 2B ). After transformed into wheat immature embryos of wheat variety cv ZM 7698, we obtained 8 independent transgenic wheat lines, among which, we randomly selected 3 of them for further analysis. Southern blot analysis indicated that the expression cassette of SmDSR33-dsRNA had been successfully integrated into the wheat genome with two to twelve copies ( Figure 2C ).

To further investigate whether the expression of the target SmDSR33 gene in aphids was inhibited when feeding on transgenic wheat plants. The individual synchronous one-day-old nymphs were transferred to wild-type and transgenic wheat plants, respectively. The relative expression levels of SmDSR33 were detected in adult aphids. The relative expression levels of SmDSR33 in grain aphids decreased significantly upon feeding on three transgenic wheat lines (P< 0.01, Figure 2D ).

Fitness parameters including life cycle and mortality of aphids upon feeding on different transgenic lines were further investigated to evaluate the silencing impact of SmDSR33. The mortality rates of aphids fed on transgenic wheat lines significantly increased when compared to that of aphis fed on wild-type plants at 9 days after feeding (DAF), reaching more than 60% at 18 DAF ( Figure 3A ). We also monitored the development duration of aphids from the nymphal to imago stage. The adult preoviposition period (APOP) and the total preoviposition period (TPOP) of aphids showed no significant difference between host plant lines ( Figure 3B ). The aphid longevity fed on transgenic wheat lines was significantly shorter than on wild-type plants. Similarly, the adult longevity and reproductive period of aphids significantly decreased than wild-type (P<0.01) ( Figure 3C ). Consequently, in comparison with the wild-type plants, the aphid total production significantly decreased when fed on all three transgenic wheat lines (P <0.01) ( Figure 3D ), and the daily fecundity of aphids fed on 33-592 transgenic line decreased at a significant level (P <0.01) ( Figure 3D ).

All of the population parameters, including the net reproductive rate (R₀), mean generation time (T), the intrinsic rates of increase (r_m) and doubling times of the population (DT), showed differences between grain aphids fed on transgenic lines and those on wild-type plants ( Table 1 ). For example, the net reproductive rates (R₀) of aphids were significantly lower when fed on transgenic wheat lines. The mean generation time (T) of aphids fed on 33-592 line was significantly decreased than wild-type (P <0.01) ( Table 1 ).

To investigate the feeding behavior of aphids, transgenic wheat line 33-592 was selected to perform electropenetrography (EPG) assays. As shown in Figures 4A-F , there was no difference between the aphids fed on SmDSR33 and dsGFP wheat plants at time point of first probe activity. The number of non-probing (np) waveforms of SmDSR33-silenced aphids fed on 33-592 line was significantly higher than on wild-type plants. Furthermore, the total duration of np waveforms and C phases of SmDSR33-silenced aphids was significantly increased compared to control. Finally, there was no difference in the duration of E1 waveforms, but did of phloem ingestion (E2) with a significant reduction for aphids on control plants. These results indicated that the feeding behavior of grain aphids was affected after SmDSR33 silencing.

Newborn nymphs produced in a parallel experiment were used to investigate potential transgenerational RNAi effects of SmDSR33. The expression levels of SmDSR33 in the offspring of aphids fed on transgenic and wild-type plants was investigated subsequently. SmDSR33 expression in grain aphids was suppressed in their offspring fed on wild-type plants ( Figure 5 ). Aphid relative expression levels reached 77.71%, 70.04%, 74.63%, and 61.80% of control level in successive first to fourth generations ( Figure 5A ). Even after switching to wild-type plants, aphid offspring still exhibited higher mortality rates ( Figure 5B ).