Pichia pastoris (SMD1168H strain) was transformed with genes encoding GNA (Raemaekers et al., 1999) or Hv1a/GNA (Pyati et al., 2014) and fermentation carried out in a Bio Console ADI 1025 (Applikon) fermenter (2 l vessels), with a continuous 50% glycerol feed. After expression, cultures were centrifuged at 7500 g for 30 min and the supernatant collected. Recombinant GNA was purified by hydrophobic interaction chromatography on a phenyl-sepharose resin packed into a Pharmacia XK16 column. Fractions containing GNA were reloaded onto a size-exclusion column (HiPrep^TM 16/60 Sephacryl S-100, GE-Healthcare). Following purification, recombinant proteins were dialyzed, freeze-dried and stored at -20°C. For His-tagged Hv1a/GNA purification, supernatants were diluted in binding buffer (0.02 M sodium phosphate, 0.4 M NaCl, pH 7.4). Samples were loaded onto a HisTrap^TM (GE Healthcare) column and then eluted with binding buffer containing 0.2 M imidazole. After purification, samples were extensively dialyzed in water and freeze-dried. The concentration of Hv1a/GNA was estimated by comparing band intensities with known amounts of GNA on SDS-PAGE, as described Down et al. (2006).

Myzus persicae were kept on Chinese cabbage plants (Brassica rapa) at 25°C, 16:8 (L:D), whereas S. avenae were reared on wheat (T. aestivum), at 20°C, 16:8 (L:D). Prior to bioassays, apterous adult aphids were transferred from plants to 90 mm diameter Petri dishes containing artificial diet (Febvay et al., 1988) in Parafilm sachets as described by Down et al. (1996), and allowed to reproduce for 24 h. Neonate aphids (ten per Petri dish) were collected and exposed to one of the four treatments in artificial diet: (i) artificial diet alone (negative control), (ii) 1 mg/ml GNA, (iii) 0.5 mg/ml Hv1a/GNA, or (iv) 1 mg/ml Hv1a/GNA. Mortality was recorded daily for 8 days and diets were changed every 48 h. Thirty aphids per treatment (in three Petri dishes) were used for S. avenae bioassays, and 70 aphids/treatment (in seven Petri dishes) were used for M. persicae bioassays.

Fecundity of M. persicae was evaluated by continuously feeding neonate aphids with Hv1a/GNA or GNA at 0.25 mg/ml for 9 days, as aphids do not reach adulthood when fed higher concentrations of fusion protein. Three cages containing 10 aphids were used for each treatment, and the cumulative number of nymphs produced/day/adult was recorded. For evaluating the effects of GNA or Hv1a/GNA on M. persicae development, three replicates of ten 2-days old aphids were given artificial diet alone, GNA, or Hv1a/GNA at 1 mg/ml of artificial diet. Aphid lengths (from head to cauda) were measured on the first 3 days by using a graticule. For all bioassays, environmental conditions were as stated above for rearing.

Neonate M. persicae and S. avenae were fed for 24 h on artificial diet containing Hv1a/GNA at 0.5 or 1 mg/ml. Insects (10–15) were either collected, flash frozen in liquid nitrogen and macerated in SDS sample buffer for protein extraction, or transferred to Petri dishes containing artificial diet without added proteins for a pulse-chase experiment. After 24 h, those aphids were collected and their proteins extracted as described above. Honeydew from each treatment was collected from the bottom of the Petri dishes, 24 h after the beginning of the assays. Samples were heat-denatured and separated in 15% SDS-PAGE. Proteins were then transferred to nitrocellulose membranes and the uptake of fusion proteins evaluated by western blot using anti-GNA antibodies (1:5000 dilution) and enhanced luminol-based chemiluminescent (ECL) substrate, as previously described Fitches et al. (2012).

A sequence coding for Hv1a/GNA was synthesized with Arabidopsis thaliana codon usage for optimal plant expression (ShineGene Molecular Biotech, Inc., Supplementary material). Primers containing attB1 and attB2 sites (Table 1) were used to amplify the gene via PCR (30 cycles of 98°C for 10 s, 55°C for 30 s, and 72°C for 30 s, with a final extension step of 7 min), which was then transferred to pDONR vectors using BP clonase reaction (Gateway^®, invitrogen^TM). The construct (Figure 1) was composed of a GNA precursor leader sequence (van Damme et al., 1991), followed by the venom toxin Hv1a (K34Q; Pyati et al., 2014), a linker region composed of three alanines, and GNA followed by its C-terminal extension (van Damme et al., 1991). The GNA precursor leader and the C-terminal extension sequences were added to the construct in order to provide correct folding and trafficking of the fusion protein to the phloem sap (Rao et al., 1998). Constructs were electroporated into Escherichia coli Top10 and plasmids extracted from positive colonies. In a subsequent step, the gene coding for the fusion protein was transferred from the pDONR to pK2GW7 vector (Karimi et al., 2002) via LR clonase using Gateway^® technology (Invitrogen^TM).

Expression constructs were finally electroporated into Agrobacterium tumefaciens C58C1, and antibiotic resistance was used to screen transformed colonies. Arabidopsis thaliana (var. Columbia) were transformed with A. tumefaciens following the floral dip method described by Clough and Bent (1998). Seeds were harvested, surface-sterilized and spread on plates with Murashige-Skoog medium containing 50 μg/mL kanamycin. Plates were kept at 4°C for 48 h in order to break seed dormancy and then transferred to environmentally controlled growth rooms (16:8 h L:D, 22°C day and 17°C night). Putative transformed plantlets were transferred to plastic pots containing soil (John Innes No. 2). Transformation was confirmed via PCR using the same conditions described above and by western blots. Protein expression was estimated by macerating a known amount of leaf tissue in 1.5x SDS loading buffer containing 2-mercaptoethanol (1 mg/10 μl). Samples were macerated, boiled for 5 min and centrifuged at 13,000 g for 2 min. Supernatants (20 μl) and GNA standards (25, 50, and 100 ng), used to estimate Hv1a/GNA concentrations, were loaded onto 15% SDS-PAGE. After electrophoresis, proteins were transferred to nitrocellulose membranes. Fusion proteins and GNA standards were probed with anti-GNA antibody as described above.

Transgenic F₃ Arabidopsis plants homozygous for the gene expressing Hv1a/GNA were used in bioassays with M. persicae only, as S. avenae does not feed on crucifers. Leaves from two homozygous transgenic lines (1.2a and 1.3b) and non-transgenic Arabidopsis (negative control) were detached from approximately five-week-old plantlets (ca. 30 plants/line were used, ensuring that leaves taken were of comparable age). Their petioles were immersed in 0.5% agar contained in 1.5 ml plastic tubes, which were then individually placed in 450 ml plastic boxes. Six replicates of five aphids were used for each treatment. Aphids were kept at 25°C, 16:8 (L:D). Leaves were replaced every 2 days and the number of alive aphids recorded daily for 6 days. Survival analysis was carried out as described below (see Statistical Analyses).

Log-rank Kaplan–Meier survival analyses with pairwise comparisons were carried out using Sigmaplot 11 (2008).

Data recorded for length and fecundity of M. persicae exposed to GNA or Hv1a/GNA via artificial diet (see Artificial Diet Bioassays) were evaluated by one-way ANOVA. Post hoc pairwise multi-comparisons were carried out using Holm–Sidak method. The median lethal concentrations (LC₅₀) of Hv1a/GNA against M. persicae and S. avenae were calculated by plotting log dose (0, 0.5, 1, and 2 mg/ml) vs. probit of corrected mortalities (Abbott, 1925; Miller and Tainter, 1944; Randhawa, 2009).

The toxicity of Hv1a/GNA was assayed using neonate (<24 h) M. persicae nymphs fed recombinant fusion protein Hv1a/GNA at 0.5 or 1 mg/ml of artificial diet. Results are shown in Figure 2. Aphids presented increased mortality on fusion protein treatments from the second and third days after the start of experiments. Survival curves differed from each other (p< 0.001), and pairwise multiple comparisons showed significant differences between all treatments (p < 0.05). Hv1a/GNA showed higher levels of toxicity toward M. persicae than that of GNA alone, demonstrating its increased toxicity against this species. When fed at a concentration of 1 mg/ml, the fusion protein Hv1a/GNA resulted in more than 90% decrease in survival after 8 days, whereas GNA alone at 1 mg/ml resulted in less than 35% reduction (Figure 2A). Subsequently, a dose/response assay was carried out using five different protein concentrations of either GNA or Hv1a/GNA. When continuously feeding on diets with test proteins, aphids were once more shown to be significantly more susceptible to the fusion protein than to GNA (data not shown, p < 0.05), with an estimated LC₅₀ for the fusion protein of 1.81 mg/ml after 2 days. It was not possible to reliably calculate the LC₅₀ for GNA alone with the concentrations used, as 50% mortality was not achieved at the doses fed, and mortalities did not always increase linearly with increased concentrations of the lectin.

Immunoassays by western blot analysis of aphids fed on artificial diet containing Hv1a/GNA demonstrated that fusion proteins were rapidly digested by M. persicae. Anti-GNA antibodies recognized a single band of around 10 kDa (Figure 2B) in extracts from whole aphids fed with Hv1a/GNA in a pulse-chase experiment, 24 h after exposure. Furthermore, the ∼10 kDa band was also detected in the honeydew, suggesting that the fusion protein is cleaved in the gut, and no evidence of intact Hv1a/GNA was observed. Although GNA and fusion proteins are internalized by homopterans (Down et al., 2006), it was not transmitted to nymphs descended from aphids feeding on Hv1a/GNA.

Myzus persicae nymphs were significantly smaller than controls (p < 0.001) following 2 days continuously feeding on diet containing Hv1a/GNA at 1 mg/ml, although they presented similar sizes at the beginning of the experiments (p = 0.98). After 3 days, insects fed on 1 mg/ml Hv1a/GNA or GNA were approximately 30 and 20% smaller than controls, respectively (p < 0.001; Figure 3A). Additionally, when compared to controls, the cumulative number of nymphs produced per adult was significantly reduced on aphids fed with GNA (ca. 69%, p= 0.002) or Hv1a/GNA at 0.25 mg/ml (>90% reduction, p < 0.001) after 9 days from the start of the experiment (Figure 3B). It was not possible to test the effects of this recombinant protein at a higher dose of 1 mg/ml, as no nymphs reached adulthood (Figure 2).

A bioassay with a semi-specialist aphid species, the grain aphid Sitobium avenae, was carried out to test the efficacy of Hv1a/GNA against this important pest when fed in liquid diet. Following Kaplan–Meier Survival analysis, significant differences between survival curves were found with bioassays using S. avenae (p < 0.001). Pairwise multiple comparisons (Holm-Sidak) revealed non-significant differences between GNA and control treatments (p = 0.317), while Hv1a/GNA at 0.5 and 0.1 mg/ml differed from all other treatments (p < 0.05; Figure 4A). These results demonstrate that S. avenae is more susceptible to Hv1a/GNA than M. persicae, and while GNA alone did not significantly affect survival, the fusion protein rapidly induced mortality, with LC₅₀ of 0.73 mg/ml after 2 days, in contrast with a 2.4-fold higher LC₅₀ for M. persicae (1.81 mg/ml).

Western blot analyses show that the grain aphid, as opposed to M. persicae, does not readily cleave the fusion protein. Intact Hv1a/GNA was detected in whole grain aphids feeding on fusion protein and also in their honeydew. Limited proteolysis is suggested by the appearance of additional immunoreactive bands of lower molecular mass than that of intact fusion protein in Hv1a/GNA fed aphid and honeydew samples. Only after 24 h, as shown in the chase experiment, is the fusion protein completely cleaved (Figure 4B).

Transgenic Arabidopsis thaliana plants harboring the pK2GW7 vector carrying the sequence for Hv1a/GNA under the control of the CaMV 35S promoter were generated using the Agrobacterium tumefaciens-mediated floral dip technique. After selection of T₀ seeds on plates containing kanamycin, a transformation efficiency of 2.67 ± 0.46% (average number of kanamycin-resistant seeds ± SEM) was obtained from seven independent events. Integration of the transgene cassette was investigated by PCR (Figure 5A) and positive plants were self-pollinated in order to generate homozygous lines for the Hv1a/GNA fusion protein.

Western blot of leaf extracts from plants carrying Hv1a/GNA gene demonstrate that the fusion protein was expressed in T₀ and homozygous F₃ plants (Figures 5B,C, respectively). The ∼25 kDa band corresponding to the intact fusion protein is detected along with another lower molecular weight protein that also reacts with anti-GNA antibody. The lower molecular weight cleavage product was also present when the fusion protein is expressed in P. pastoris. This result indicates that the plant cleaves Hv1a/GNA following translation, and further improvements and alterations to the peptide structure would benefit its expression in heterologous systems. Quantification of expression was carried out by comparing intensity of Hv1a/GNA bands from known amounts of leaf extracts compared to GNA standards in western blots. It was estimated that the fusion protein was being expressed at 25.6 ± 4.1 ng/mg fresh weight (F.W.) leaf tissue.

In order to test the efficacy of fusion proteins expressed in plants against aphids, a bioassay with transgenic Arabidopsis was set. Two homozygous lines (designated 1.2a and 1.3b) from independently transformed plants were assayed for aphid resistance. Leaves were detached from plants and their petioles immersed in 0.5% agar. When compared to non-transformed controls, aphids feeding on both events showed similar survival patterns, with significantly increased levels of aphid mortality (K-M, p = 0.014; control vs. 1.2a, p= 0.01; control vs. 1.3b, p= 0.003; 1.2a vs. 1.3b, p = 0.691); aphid survival was reduced to around 60% after 7 days (Figure 6). The corrected mortality using Abbott’s formula for 1.2a was 29.6 and 37% for 1.3b.