A transgenic approach to control hemipteran insects by expressing insecticidal genes under phloem-specific promoters

The first generation transgenic crops used strong constitutive promoters for transgene expression. However, tissue-specific expression is desirable for more precise targeting of transgenes. Moreover, piercing/sucking insects, which are generally resistant to insecticidal Bacillus thuringiensis (Bt) proteins, have emerged as a major pests since the introduction of transgenic crops expressing these toxins. Phloem-specific promoters isolated from Banana bunchy top virus (BBTV) were used for the expression of two insecticidal proteins, Hadronyche versuta (Blue Mountains funnel-web spider) neurotoxin (Hvt) and onion leaf lectin, in tobacco (Nicotiana tabacum). Here we demonstrate that transgenic plants expressing Hvt alone or in combination with onion leaf lectin are resistant to Phenacoccus solenopsis (cotton mealybug), Myzus persicae (green peach aphids) and Bemisia tabaci (silver leaf whitefly). The expression of both proteins under different phloem-specific promoters resulted in close to 100% mortality and provided more rapid protection than Hvt alone. Our results suggest the employment of the Hvt and onion leaf lectin transgenic constructs at the commercial level will reduce the use of chemical pesticides for control of hemipteran insect pests.

promoter (352 bp), coat protein (CP) promoter (540 bp), Hvt (117 bp) and onion leaf lectin (333 bp). Constructs were cloned and sequenced. Homology of all of the sequences was checked by performing BLAST comparisons. The Hvt gene sequence showed 100% identity to H. versuta spider neurotoxin 54 and the lectin gene sequence showed 100% identity to the A. cepa mannose-binding insecticidal leaf lectin 61 .
Nicotiana tabacum leaves were infiltrated with GUS constructs to test promoter function. The Cauliflower mosaic virus 35S promoter was used as a positive control and empty-vector, pGreen0029, as negative control, to assess the BBTV-promoter expression. As the 35S promoter is a constitutive promoter, it showed GUS expression in whole inoculated patch (Fig. 1a). In contrast, the pGreen0029 (− ve control) did not show GUS gene expression (Fig. 1b). Both BBTV promoters, NSP and CP, showed GUS expression only in the vascular cells (Fig. 1c,d), but expression from the NSP promoter was stronger than that from the CP promoter (Fig. 2). These results indicate that expression of GUS from both CP and NSP promoters is phloem-limited, making these promotors suitable for phloem-specific expression of Hvt and onion leaf lectin insecticidal transgenes.  Transgene Analysis. Transgenic tobacco plants were confirmed to carry Hvt and lectin genes at the T 1 stage by PCR. PCR-positive plants were further processed for the extraction of RNA and cDNA synthesis. Quantitative reverse transcriptase-PCR (qPCR) showed a single peak in each of the melt curves for the genes of interest, Hvt and lectin, indicating that there was no non-specific amplification (see Supplementary data Figure S6). To allow later comparison to insect resistance traits, relative transgene expression levels were quantified in six independent transgenic lines, two single gene HVT constructs and four double gene constructs (DGC) with both Hvt and lectin (Fig. 3). In case of the Hvt gene, all of the plants showed good expression. DGC 15 was the most highly expressing line (relative expression = 1.00), expressing Hvt almost two-fold higher than the H 22 (0.682). DGC 6 is the second most highly expressing line (0.704), while in case of DGC 5, 9 and H 21 the gene was expressed at a very low level and relative expression was almost the same (0.234, 0.288 and 0.265 respectively) (Fig. 3a).
The results obtained from lectin qPCR are distinct from the Hvt results. In case of lectin, the highest level of expression was observed in DGC 6 (relative expression = 1.000). The lectin expression in DGC 6 was almost 2-fold higher than in the rest of the lines DGC 5, 9 and 15 (i.e. 0.609, 0.666 and 0.662 respectively) (Fig. 3b). The relatively high expression level was predicted to make the DGC 6 construct more toxic towards insect pests. Insect Bioassays. Bioassays were performed in order to assess the effectiveness of the toxin proteins against P. solenopsis, M. persicae, and B. tabaci, all of which are not only major threats to cotton but also feed on tobacco.
Phenacoccus solenopsis. Transgenic plants expressing toxins were exposed to P. solenopsis, with each plant receiving fifteen P. solenopsis individuals. The mortality data of the insects were collected every 24 h and insects were observed for the behavioral changes due to toxin ingestion. During the initial 24 h, insects became adapted to the new environment and settled to feed on transgenic plants. No remarkable changes were observed in the behavior of the insects. The typical effects of the toxin, in particular slower movement of P. solenopsis, were observed after 48 h. Almost 28% of the nymphs feeding on DGC 6 and 15 and 10% on DGC 5 and 9 were found dead, but no mortality was recorded on the plants expressing the single gene construct (SGC), plants 21 and 22. After 72 h, 65% of the insects feeding on DGC 6 and 15 were found dead, followed by 35% mortality on DGS 5 and 9. After feeding for 72 h, approximately 15% of the insects were dead on the SGC 21 and 22 plants. After 96 h, almost 100% of the insects were killed on DGC 6 and 15 (Fig. 4b), whereas 80% mortality was recorded on DGC 5 and 9. In the case of the SGC 21 and 22 plants, approximately 38% of the insects were found dead. After 120 h feeding, there was 100% insect mortality on DGC 5 and 9 followed by approximately 77% mortality on SGC 21 and 22 (Fig. 4a). The insects feeding on SGC lines were allowed to feed on plants for another 24 h, at which time 97% of the insects were found dead (Fig. 5). Only a few adult insects were found alive.  The insects feeding on control plants were found to be normal in all aspects. They fed on plants in the usual manner, gained weight, and multiplied enormously. They were normal in their body movements and spread over the whole plant when allowed to feed for longer time period (Fig. 4c).
Myzus persicae. After testing the transgenic plants with P. solenopsis, the same plants were exposed to the M. persicae, to determine whether there are similar toxic effects. To allow recovery from any possible stress due to insect feeding, plants were left uninfested for 3 to 4 weeks prior to M. persicae exposure. After 24 h feeding, behavioral changes, particularly very slow movement, were observed in M. persicae feeding on the transgenic plants DGC 5,9 and SGC 21,22. In the case of DGC 6 and 15, there was a remarkable difference in the behavior and survival of the insects. Approximately 40% of the individuals, mainly nymphs, were found dead on DGC 6 and 15. None of the nymphs feeding on DGC 6 and 15 survived until the next day. After 48 h feeding, almost 80% mortality was recorded on DGC 6 and 15. Almost 35% of the insects feeding on DGC 5 and 9 were found dead, whereas very low (7-8%) mortality rate were recorded on SGC plants. After 72 h feeding, 100% of the insects were dead on DGC 6 and 15, followed by 70% mortality on DGC 5 and 9 and approximately 40% mortality on SGC 21 and 22. Insects feeding on transgenic plants exhibited the typical effects of toxin uptake, including weight loss, lack of food uptake, lack of movement, jerky movements in the body, and inability to multiply (Fig. 6a,b). On the DGC 5 and 9 lines, 100% of the insects were found dead when they were allowed to feed on for 96 h, followed by almost 70% deaths on SGC 21 and 22. Almost 100% of the insects were dead when allowed to feed on SGC 21 and 22 for another 24 h (Fig. 7). On control plants, M. persicae fed normally, gained weight, nymphs moved to the next growth stage, and multiplied rapidly (Fig. 6c). A few exoskeletons were also found on the control plants showing that the insects had molted to the next instar stage (Fig. 6d).
Bamisia tabaci. Plants were left uninfested for 30-40 days, before being infested with Bemisia tabaci. B. tabaci took very little time to acclimatize to the plant growth room conditions. After 24 h feeding, almost 50% of the insects feeding on DGC 6 and 15 were found dead. The rest of the insects showed notable behavior changes. In comparison to P. solenopsis and M. persicae, the toxin effects in B. tabaci were quite strong and easily noticed. B. tabaci were found with abnormal jerky and shaky movements, abnormal stretching in the body, and other symptoms after 48 h of feeding. There was 100% mortality rate on plants DGC 6 and 15 (Figs 8 and 9) after 48 h, followed by approximately 56% mortality on DGC 5 and 9 and 55% in SGC 21 and 22. The insects were allowed to feed for another 24 h and were found to be dead on all transgenic plants. In case of control plants, the insects remained alive and active, laid eggs, and grew in a regular manner.
All of the tested insects showed similar physical responses to plants expressing Hvt and lectin, including loss of body mass, browning of bodies over time, and stretching of legs inwards in case of DGC or outwards in case SGC, as has been observed previously for neonicotinoid insecticides 62,63 . Insects that spent more time on plants also showed some developmental changes as they moved from one growth stage to the next. In case of M. persicae, some individuals shed their exoskeletons and moved to the next growth stage, and exoskeletons were found on the plants. P. solenopsis stayed on the plants for a longer time, in particular the adult individuals. However, B. tabaci did not show such changes, perhaps due to the short time that they spent on the plants.

Discussion
We have presented results from the expression of two insecticidal proteins in a phloem-specific manner in tobacco. This approach was specifically used to control sap-sucking insects, which take up the toxic proteins when they are feeding. In the case of a constitutive promoter like the CaMV 35S promoter, there is expression in almost all cells of the plant during all developmental stages 64 . Although expression from this promoter is reported to be low in the phloem cells compared to the other plant tissues 64 , expression levels for our GUS constructs with the 35S and phloem-specific BBTV promoters was similar. Expression of transgenes from the 35S promoter can be used to control phloem-feeding insects 64 . However, a potential drawback is that continuous overexpression of insecticidal proteins in multiple tissue types is more likely to cause deleterious effects in the plants.
We used a gene pyramiding technique, both to limit the escape of the insects and for more efficient control. We found significant mortality, and the dead insects showed interesting physical changes. Insects on plants expressing a single protein (Hvt, a neurotoxin) had a low mortality rate when they were examined on a 24-hour basis. Dead insects showed decreased body mass and browning of the bodies, along with their legs being stretched outwards (Figs 4a, 6a and 8a). In contrast, the insects feeding on DGC plants also exhibited a mass decrease and brown bodies, were killed more efficiently, but exhibited different physical characteristics, with their legs folded inward (Figs 4b, 6b and 8b). The differing physical changes of dead insects, in particular the legs being stretched inwards or outwards, are likely due to the effect of the toxins expressed in the host plants. In case of the Hvt, an antagonist of a calcium gating neurotransmitter, the insect nervous system is affected and we observed the stretched insect body 50,65,66 . In the case of the double protein treatment, a neurotoxin along with a lectin, the lectin acts upon the digestive tract of the insect. That may explain why, along with jerky and shaky movement of  the insects, the legs were observed folded inward toward the body. The fecundity of insects was also affected due to feeding on toxin-producing plants 62,63 . Insects were observed multiplying on the control plants, but not on the toxin-producing ones. Some of the B. tabaci laid eggs on controls, whereas M. persicae and P. solenopsis were found to develop into the next instars.
In summary, our results show that expression of genes in target tissues and gene pyramiding are efficient techniques for controlling insect pests. Future research can be directed toward the identification of other promoters for tissue-specific expression and targeted control of other insect pest species.

Methodology
Promoter identification and cloning. To express the insecticidal genes, phloem-limited promoters were cloned from a Pakistani BBTV isolate 40 . Two viral components, the capsid protein (CP) and the nuclear shuttle protein (NSP) were chosen for amplifying the promoters. The intergenic region (IR) was considered as the promoter and sequence-specific primers ( Table 1 in supplementary data) were used to amplify the whole IR by PCR. The deleted part of the NSP intergenic region was used as reported 45 . The PCR reaction mixture was prepared  using the Thermo Scientific, USA Taq polymerase kit, containing 1X reaction buffer, MgCl 2 , 2 mM dNTPs, 10 pmol of each primer, 10 ng of template and 0.5 units of Taq polymerase. The PCR reaction was run with the following temperature profile: initial denaturation temperature 94 °C for 5 min, 1 cycle; followed by 35 cycles of 94 °C for 30 sec, 50 °C for 30 sec, 72 °C for 45 sec; and final extension at 72 °C for 10 min. SacI and HindIII restriction sites were used for cloning into the pJIT166 expression vector 67 , replacing the 2X 35S promoter upstream of the GUS reporter gene. In the next step, the complete expression cassette from pJIT166 was excised and cloned into the pGreen0029 binary vector 68 at restriction sites SacI and XhoI, which makes the construct ready for further cloning of genes under this promoter.
Agrobacterium transformation. One μ g of the cloned and confirmed plasmid was transformed into Agrobacterium tumefaciens strain GV3101 by electroporation 69 . Transformed cells carrying the construct were selected on LB agar plates containing antibiotics (rifampicin 25 μ g/mL, tetracycline 10 μ g/mL and kanamycin 50 μ g/mL). Different colonies were picked and the presence of the promoter was confirmed by PCR. A single-colony culture was prepared from a PCR-positive colony in 50 mL LB containing antibiotics (rifampicin 25 μ g/mL, tetracycline 10 μ g/mL and kanamycin 50 μ g/mL) in a 250 mL conical flask. The medium was incubated at 28 °C and 1.8 RCF (relative centrifugal force) for 48 h. After 48 h, the culture was pelleted by centrifugation in a 50 mL Falcon tube. The pellet was resuspended in 10 mM MgCl 2 , 20 μ M acetosyringone was added, and the sample was kept overnight at 4 °C. Nicotiana tabacum (tobacco) plants were infiltrated with the activated culture and kept under controlled conditions of 28 ± 2 °C and 16L/8D for 4-5 days.
Histochemical staining assay. GUS (β -glucuronidase) was used as a reporter gene to check promoter efficiency. Five days post infection (dpi) leaves were detached and vacuum infiltrated with 50 mM sodium phosphate buffer pH 7.0, containing ethylene-diaminetetraacetic acid (EDTA), Triton-X100 and X-Gluc (a substrate for β -glucuronidase) 70 . Leaves were kept in buffer overnight and incubated at 37 °C. Afterwards the incubation buffer was discarded and leaves were bleached with 100% ethanol until all the chlorophyll was removed and blue GUS staining was clearly visible. GUS expression in leaves was quantified by using ImageJ 1.5 software 71,72 . The leaf area showing GUS expression was selected and subjected for the quantification.
Insecticidal gene constructs. The Hvt (GenBank accession number AJ938032) gene was amplified by PCR using gene specific primers (Table S2 in the supplementary data) and cloned under the NSP promoter of BBTV in the pGreen0029 vector, using restriction sites HindIII and XbaI.
For double gene constructs, two individual cassettes were synthesized by Bio Basic Inc., Canada. One cassette contained the NSP promoter, Hvt, and the CaMV terminator, whereas the other consisted of the CP promoter, lectin (GenBank accession number DQ255944), and the CaMV terminator. Both cassettes were cloned in pGreen0029 using restriction sites SacI and XhoI.

Plant transformation. The final construct in the binary vector pGreen0029 was transformed into
Agrobacterium tumefaciens strain GV3101 by electroporation. A single colony culture, after confirmation by PCR, was prepared in 50 mL LB containing antibiotics (50 μ g/mL kanamycin, 10 μ g/mL tetracycline, and 25 μ g/mL rifampicin) at 28 °C and 1.8 RCF for 48 h. Nicotiana tabacum was transformed with the leaf disc method, using a modified version of a published protocol 73 . For selecting transformants, MS medium was supplemented with 50 mg/L kanamycin, 250 mg/L cefotaxime, 250 μ g/L 1-napthalene acetic acid (NAA) and 1 mg/L 6-benzylaminopurine (BAP). The putative transformants were shifted to soil for further analysis.
Transgene Analysis. Seeds collected from T 0 plants were germinated on kanamycin so all of them have at least one copy of transgene. These plants were not tested for homozygosity. DNA was extracted from fresh leaves of transgenic plants at the T 1 stage using the cetyltrimethylammonium bromide (C-TAB) method 74 and confirmed for the presence of toxin genes and promoters by PCR using sequence specific primers and also by sequencing.
RNA extraction and cDNA synthesis. Fresh samples were taken from transgenic plants after washing with diethylpyrocarbonate (DEPC) and kept in liquid nitrogen. Total RNA from transgenic plants at the T 1 stage was extracted by using Plant RNA Purification Reagent (Invitrogen catalog number 12322-012) following a standard published protocol 75 . Freshly extracted RNA was quantified using a Nanodrop (Thermo Scientific, USA) and was used to synthesize cDNA with the cDNA Synthesis Kit (Thermo Scientific, USA). Two μ g of total RNA was used as a template for the synthesis of cDNA.
Real-time PCR for transgene analysis. Real-time PCR or quantitative PCR (qPCR) was performed to check the relative expression of transgenes. Here, SYBER Green technology was used to study the gene expression in transgenic plants. All of the optimization and experiments were carried out using iQ TM 5 iCycler (BioRad) equipment. The 18S ribosomal RNA gene was used as a reference gene and its expression was compared with the expression of transgenes. Finally, the six best lines were selected for analysis by qPCR, four from double gene construct (DGC) transgenes and two from single gene construct (SGC) or Hvt transgenes. Three plants from each line were selected for gene expression analysis.
The reaction mixture (25 μ L) consisted of 12.5 μ L of 1X iQ SYBER Green Supermix, 50 ng of sample cDNA, and gene specific primers (Table S2 in the supplementary data). The PCR reactions were carried out in a 96-well optical plate in an iQ TM 5 iCycler (BioRad) and all of the samples were run in triplicate. PCR was carried out using the profile: 1 cycle at 94 °C for 5 min followed by 40  Insect rearing. Three sucking pests, P. solenopsis, M. persicae and B. tabaci were used to check the effect of toxic proteins. Insects were collected from the field and were allowed to feed and multiply on cotton plants in a glass house under controlled conditions of 37 ± 2 °C, 16L/8D and ≈ 60% humidity. The insects settled very well on cotton and multiplied enormously.
Insect bioassays. Transgenic plants were grown in a growth room under controlled conditions of 26 ± 2 °C and 16L/8D. After confirmation by PCR, 8 lines of the DGC and 5 lines of the SGC were grown under controlled conditions in the T 1 stage. Four lines of the DGC and two lines of the SGC with the strongest transgene expression were used for insect bioassays. Five plants of each line at the 3-4 leaf stage were used for bioassays. Different phloem feeding insects infested on transgenic plants under controlled conditions. Fifteen individuals of P. solenopsis, 20 individuals of M. persicae and 20 of B. tabaci of mixed populations, consisting of both nymphs and adults, were allowed to feed on plants to check the effects of toxic proteins. Equal numbers of control plants (wild type N. tabacum) with equal numbers of insects were also used to get comparable mortality data. The percent mortality was calculated by the following formula: (number of insect pests found dead on the plant/(number of the pests allowed to feed on the plant − number of insects that absconded)) × 100. Statistical Analysis. ANOVA and Tukey's SD test were used to assess significance in the mortality data using the statistical software program SPSS-20 76 .