The full-length cDNA of rice tps46 [previously named as SSBis039 and shown to be induced by C. suppressalis larvae (Sun et al., 2011)] was cloned from “NPB” rice leaf tissues using the SMARTer^TM RACE cDNA Amplification Kit (Clontech, Palo Alto, CA, USA) according to the protocol described in Sun et al. (2014). Two 5′GSPs (5′GSP1 and 5′GSP2) and two 3′GSPs, (3′GSP1 and 3′GSP2) were designed using Primer 5.0 software based on the partial tps46 sequence obtained by our previous study (Sun et al., 2011), and primers were listed in Table 1.

Based on the sequence of tps46 (Genbank accession number: EU596452), the rice tps46 fragment was amplified using the RiTPS46F and RiTPS46R primers (Table 1). The 535 bps PCR product was digested with both SacI and SpeI or with KpnI and BamHI, gel purified, and then ligated into the same restriction sites within the RNAi ptck303 binary vector of rice (Wang et al., 2004). The ptck303-tps46 RNAi binary vector contained an antisense tps46 fragment, a rice intron, and sense tps46 fragment between the maize (Zea mays) ubiquitin promoter and the nos 3′-terminator in the ptck303 binary vector (Wang et al., 2004; Kong et al., 2006). For tps46 overexpression, the full-length CDS of tps46 was amplified using the OeTPS46F and OeTPS46R primers (Table 1). Then, the 1641bp of tps46 was ligated to a pCXUN vector with the TA cloning system (Chen et al., 2009). Rice callus was induced from the embryos of mature seeds of the Japonica cultivars “NPB” using an Agrobacterium tumefaciens-mediated method (Qu et al., 2006; Park et al., 2012). Finally, more than 15 T0 tps46-RNAi (Ri) rice lines and 15 T0 tps46-Overexpression (Oe) lines were obtained. The segregation lines of Ri and Oe transgenic plants were set as RWt or OWt controls.

Seven T1 Ri line (Ri1, 3, 4, 5, 8, 10, 11) and three T1 RWt line (RWt1, 2, 3) rice seedlings were first grown in a MS plus (75 mg L-1, Hygromycin B) selective liquid medium for 3 days in an incubator (28°C/26°C, 16 h of light/8 h of dark, and 50% humidity, respectively). Rice seedlings were grown individually in a cylindrical pot (25 cm and a height of 30 cm) in a greenhouse with temperature ranging from 24 to 32°C, 16 h of light/8 h of dark, and 50% humidity, where aphids, such as Rhopalosiphum padi (Hemiptera: Aphididae), Myzus persicae (Hemiptera: Aphididae), Brevicoryne brassicae (Hemiptera: Aphididae), and Aphis gossypii (Hemiptera: Aphididae), occurred naturally. The quantitative real-time PCR (qRT-PCR) measurements were conducted to investigate the target mRNA transcripts in leaves and sheaths of rice seedlings at the tillering stage. After qRT-PCR, every rice line as one treatment was random placed in a greenhouse to evaluate the insect infestation when exposed to aphids. All treatments in the experiment were repeated three times. This survey experiment was repeated twice. Insect infestation was recorded every day between 15:00 and 17:00 from tillering to the grain-filling stage of the rice plants. Aphid species were identified in insect taxonomy and toxicology laboratory, China Agricultural University.

Newly emerged wingless R. padi colonies were kindly provided by Professor Xi-wu Gao of the insect toxicology laboratory, China Agricultural University, which had been maintained in the absence of insecticide exposure more than 10 years (Lu et al., 2013). The different rice lines were cultivated as described in the previous section “insect population survey on T1 Ri lines.” Before bioassay, expression of tps46 in the Ri3, Ri5, Ri8, Ri10, RWt, Oe6, Oe7, Oe9, Oe11, and OWt rice lines at jointing-booting stage was evaluated with qRT-PCR measurements. Soon after, Ri3, Ri5, Ri8, Ri10, Oe6, Oe7, Oe9, and Oe11 were selected for indoor bioassays in an insect-free greenhouse (Figure 2 and Supplemental Figure S1). The temperature in the greenhouse was kept at 22–28°C and artificial infestation was performed between 13:00 and 17:00 PM. Each rice seedling in each pot was infested with 40 newly emerged wingless R. padi, and the number of R. padi on each rice seedling was recorded after 24, 48, and 72 h. Each treatment was replicated 3–5 times.

Total RNA from the leaves and sheaths of rice seedlings (Oe, Ri, and Wt) was extracted by using the SV Total RNA Isolation System (Promega, Madison, WI, USA). One microgram of total RNA was used as a template for synthesizing the first-strand cDNA following the manufacturer’s protocol for MMLV Reverse Transcriptase (Promega, Madison, WI, USA). The cDNAs were treated with Ribonuclease H (TaKaRa, Tokyo, Japan) and quantified on a ND-1000 spectrophotometer (NanoDrop, Wilmington, DE, USA) at OD₂₆₀ nm. The qRT-PCR was performed using the SYBR Premix Ex Taq Kit (TaKaRa, Tokyo, Japan) on a Bio-rad iCycler detecting system with SYBR green fluorescent dye. The rice EF-1α gene (Genbank accession number: AK061464) was used as an internal control to normalize the transcript levels of tps46 in each experiment (Jain et al., 2006), and the specific primers of the target gene and reference gene are listed in Table 1. The qRT-PCR reactions were replicated five times each treatment and non-template control (NTC) reactions were performed in triplicate. The 2^-ΔΔCt method was used to evaluate the relative expression of the target gene (Livak and Schmittgen, 2001).

Rice lines, including Oe6, Oe7, OWt, Ri5, Ri8, and RWt at the beginning of jointing-booting stage were selected for volatile collection by an open head-space sampling system (Yu et al., 2010). Pots containing one rice plant were placed within a cylindrical type glass container (25 cm in diameter × 50 cm high). The container was sealed with a glass lid that had an air inlet and an air outlet. The container was tightly sealed with metal clamps on the lid. Air, purified by passage through an activated charcoal filter, was actively pumped through the container at a flow rate of 1 ml/min with a vacuum pump (Beijing Institute of Labor Instrument, Beijing, China). Volatiles were collected for 6 h on 60 mg of 60/80 mesh Tenax-TA (Shanghai ANPEL Scientific Instrument Company, Shanghai, China) in an 8-mm-diameter glass tube, which was directly connected to the outlet. All connections were made with Teflon tape. Three to five replicates were performed for each experiment (Huang et al., 2015). All collections were conducted from 11:00 AM to 17:00 PM. The collected samples were stored at 4°C for further analysis.

Volatiles from Ri5, Ri8, and RWt rice lines were extracted with 300 μl of HPLC-grade hexane (Fisher Scientific, Fairlawn, NJ, USA), and 21 ng n-octane (Sigma-Aldrich, Oakville, Canada) was individually added as an internal standard. The collected samples were analyzed by Agilent 6890 GC coupled with an Agilent 5973 MS detector on a HP-5MS dimethylpolysiloxane column (30 m × 0.250 mm × 0.25 μm, Agilent Technologies, Palo Alto, CA, USA). One microliter of the sample was used for injection by using a 10 μl syringe (Hamilton, USA). The GC oven temperature was maintained at 40°C for 2 min and increased to 100°C at a rate of 6°C/min (hold for 1 min), increased again at a rate of 5°C/min to 150°C (hold for 1 min), increased to 250°C at a rate of 10°C/min, and finally maintained at 250°C for 5 min. Helium was used as the carrier gas at 1.0 ml/min. Tentative identifications were made by comparison of mass spectra (a) with mass spectra libraries (NIST and Department of Chemical Ecology, Göteborg University, Sweden) and (b) with the mass spectra and retention times of authentic samples obtained from Fluka, Sigma¹.

Volatiles from the Oe6, Oe7, and OWt rice plants adsorbed on Tenax-TA were eluted with 300 μl HPLC-grade hexane (Fisher, Fairlawn, NJ, USA), and 2.586 ng of ethyl caprate (Sigma-Aldrich, Oakville, Canada) was individually added as an internal standard. Comprehensive two-dimensional gas chromatography/time-of-flight mass spectrometry (GC x GC-ToF, Pegasus 4D, LECO, USA) was performed to analyze the volatile samples. The GC inlet and transfer line were held constant at 250°C. A 1 μl injection was made onto column 1 (Rxi-5Sil MS, 30 m × 0.25 mm i.d. × 0.25 μm, Restek). The effluent from column 1 was then transferred to column 2 (Rtx-200, 2 m × 0.180 mm i.d. × 0.2 μm, Restek). Column 1 was held at 50°C for 1 min followed by a two-step temperature increase, first to 150°C (at a rate of 5°C/min, hold for 2 min) and then to 250°C (at a rate of 10°C/min, hold for 2 min). Column 2 was held at 55°C for 1 min followed by a two-step temperature increase to 155°C (at a rate of 5°C/min with a 2 min hold) and then to 255°C (at a rate of 10°C/min, with a 2 min hold). The volatile products were identified by comparing their retention times and mass spectra areas to authentic standards (Gaquerel et al., 2009; Huang et al., 2015).

Differences of tps46 expression levels, R. padi numbers and volatile emission between different rice lines were analyzed using one-way ANOVA methods by SAS 9.0 software for Windows with Duncan’s new multiple range method (P < 0.05) (SAS 9.0 system for windows, 2002, SAS Institute Inc., Cary, NC, USA).

In our greenhouse, several aphid species on rice lines were observed and recorded, while only the significant difference in R. padi numbers between the Ri and RWt rice plants were observed. Other aphids tested included M. persicae, B. brassicae, and A. gossypii. In infestation trials, a small number of winged R. padi were found on rice plants in the late tillering stage. Then the wingless R. padi population sharply increased on Ri rice lines in the jointing-booting stage and the population reached a peak value (62 ± 22 aphid/ individual Ri plant) at 22 days into the jointing-booting stage (Figure 1). However, the R. padi population on the segregation lines of Ri transgenic plants (RWt) increased very little during the same stage and after 31 days was not significantly different the population size at the start (Figure 1). Also, the same tendency in this experiment was reproduced in another repeats (Figure 1 and Supplemental Figure S3). Moreover, R. padi populations on all seven tested Ri rice lines were significantly higher than on control rice lines (P < 0.05). In contrast, there was no remarkable difference in R. padi populations among the three RWt control rice lines (P > 0.05) (Figure 2A). Expression levels of tps46 were significantly lower in all the seven tested Ri rice lines than in control lines (P < 0.05), and there was no visible difference between the RWt control rice lines (P > 0.05) (Figure 2B). The R. padi population (83 ± 22 aphid/ individual Ri plant) on Ri5 rice plants among these Ri rice lines was significantly higher than on Ri1, Ri4 and Ri11 rice lines (P < 0.05) (Figure 2A). Interestingly, the expression of tps46 in the Ri5 line was only 6.81 ± 0.79% of the expression in control rice lines (Figure 2B). In addition, our observations have shown that there was no R. padi on control rice lines in the field or the greenhouse for many years until the Ri rice plants were cultivated.

The tps46 expression levels in different rice lines were analyzed by qRT-PCR before the R. padi infestation bio-assays. The expression of tps46 in all of the Oe rice lines was significantly higher than in the Ri and control rice lines (P < 0.05), whereas expression of tps46 in all of the Ri rice lines was obviously lower than expression in the control rice lines (P < 0.05). Additionally, there was no significant difference between the RWt and OWt control rice lines (P > 0.05) (Figure 3D).

The survival rate of R. padi on Ri rice lines was significantly higher than that on Oe and control lines (P < 0.05). The survival rate of R. padi on the Oe lines was significantly lower than the survival rate on control rice plants (P < 0.05), and there was no obvious difference in the R. padi survival rate between RWt and OWt control rice lines (P > 0.05) (Figures 3A–C). The survival rate of R. padi on the two control lines was only approximately 20% after 72 h. The survival rate of R. padi on the Ri lines was higher than 50% during the same period and reached 67 ± 11.5% on Ri5 rice plants (Figure 3C). During the same period, the survival rate of R. padi on the Oe rice lines was below 10%t after 72 h, and was less than 5% on the Oe6 and Oe7 rice plants (Figure 3C). Interestingly, the expression of tps46 in the Oe6 and Oe7 rice plants was 837 ± 144 and 976 ± 178 times higher than in control lines. The expression was significantly higher compared to expression in the other tested rice plants (P < 0.05). However, the expression of tps46 in the Ri5 lines was only 5.1 ± 0.94% of that in control rice lines (Figure 3D).

Recombinant TPS46 protein can catalyze substrates to produce several terpene volatiles in vitro (Yuan et al., 2008). In this study, we further investigated the volatile compounds released from the Ri, Oe, and Wt rice lines. GC-MS analyses indicated that the amount of two volatile compounds, limonene and Eβf, decreased significantly in the Ri rice lines compared to the RWt (control) rice plants (P < 0.05) (Supplemental Figure S2 and Figure 4A). The amounts of limonene in 1 μL of extracts from Ri5, Ri8, and RWt rice headspace samples were 0.11 ± 0.04, 0.14 ± 0.05, and 0.65 ± 0.14 ng, respectively. The amounts of Eβf in Ri5, Ri8, and RWt rice headspace samples were 0.01 ± 0.002, 0.035 ± 0.005, and 0.77 ± 0.13 ng, separately (Figure 4A). As expected, the amount of limonene and Eβf increased in the Oe rice lines compared to the OWt plants (P < 0.05) (Figures 4A and 5). The amounts of limonene in 1 μL extracts from the Oe6, Oe7, and OWt plants was 1.87 ± 0.42, 1.90 ± 0.53, and 0.99 ± 0.31 ng, respectively. Meanwhile, the amount of Eβf in the Oe6, Oe7, and OWt plants was 10.62 ± 4.04, 12.70 ± 3.09, and 1.06 ± 0.26 ng, respectively (Figure 4A). There was no significant difference in the amounts of limonene and Eβf between the OWt and RWt rice lines (Figure 4A).

Six other volatile compounds linalool, (E)-β-caryophyllene, methyl salicylate, α-bisabolene, (E)-α-bergamotene, and α-humulene were also increased significantly in the Oe rice lines compared to the OWt lines (P < 0.05). The content of these six volatiles in the OWt rice lines was extremely low (Figure 6). Also, these six volatile compounds were not detected in the Ri and RWt rice plants (Supplemental Figure S2). Among the six volatile compounds, the amount of linalool in 1 μL of extract from the Oe6 and Oe7 rice lines was 4.97 ± 2.58 ng, and 6.10 ± 2.38 ng, respectively (Figure 4B). The amount of (E)-β-caryophyllene, α-bisabolene, (E)-α-bergamotene, α-humulene in the Oe6 rice lines was 1.54 ± 0.27, 1.80 ± 0.19, 1.61 ± 0.29, and 2.88 ± 0.58 ng, separately. The amount of (E)-β-caryophyllene, α-bisabolene, (E)-α-bergamotene, α-humulene in the Oe7 lines was 1.53 ± 0.37, 2.05 ± 1.24, 1.24 ± 0.44, and 1.82 ± 0.71 ng, respectively (Figure 4B). Moreover, among the six volatile compounds, there were significant differences in the amount of methyl salicylate between the Oe6 and Oe7 lines (P < 0.05), and the amount of methyl salicylate in the Oe6 and Oe7 rice lines was 1.27 ± 0.24, and 3.65 ± 1.04 ng, respectively (Figure 4B).