Stability of Herbicide Resistance over 8 Years of Coppice in Field-Grown, Genetically Engineered Poplars

by the publication of a high-quality genome sequence—only the third such sequence for any plant species (Tuskan et al. 2006). In spite of over 200 field trials of GE trees that have been reported in at least 35 countries, the only commercially deployed transgenic trees are insect-resistant Populus nigra and white poplar (clone 741) in China (Food and Agriculture Organization 2004), and virus-resistant papaya in Hawaii (Gonsalves and Ferreira 2003). Until the stability of GE traits, and means for reliable containment and social acceptance are established, further commercial use may be extremely limited in most parts of the world (Brunner et al. 2007). Our goal was to study the stability of transgene expression in trees under field conditions relevant to its practical use and propagation. We selected herbicide tolerance because it is both a trait of commercial value, and because it also makes such studies highly efficient, as the effects of the expression of a single gene can be readily observed without expensive molecular analyses. We report very high levels of stability of resistance over 8 years, as well as a high degree of predictability of HR level based on a simple enzyme-linked immunosorbent assay (ELISA) for the transgene-encoded protein. These results suggest that high stability of this trait, and perhaps transgenic traits generally, may be achievable with a modest, short-term screening effort. Materials and Methods The plasmid vector pTTM8 was used to transform two hybrid poplar clones: INRA 353-38 (P. tremula P. tremuloides) and 717-1B4 (P. tremula P. alba). The transferred DNA (T-DNA) region contains two chimeric genes conferring resistance to the antibiotic kanamycin and the herbicide glufosinate-ammonium, respectively, and another chimeric BARNASE gene, which can impart male sterility (Figure 1). The herbicide-resistance gene was driven by the promoter of a strongly photosynthesis-associated gene, the small subunit of ribulose bisphosphate carboxylase/oxygenase, derived from Arabidopsis. The construct was transformed into a rifampicinresistant derivative of Agrobacterium C58, under kanamycin selection. A total 19 independent transgenic events were produced for poplar clone 353-38, and 13 events were produced for clone 717-1B4. Twelve ramets for each of these 32 transgenic events, plus an equal number of nontransgenic control trees, were propagated and rooted in vitro, transplanted into soil in pots, and then planted in the field in Benton County, OR, in 1997. Two ramets of each transgenic event and control were randomly planted in each of six blocks. Over-the-top herbicide sprays were applied in three different years (1997, 2004, and 2005). Most plants were coppiced between growing seasons to keep them short enough for spraying; however, some reached heights up to approximately 5 m before cutting. In 1997, herbicide was sprayed on four of the six plots shortly after planting. Two of these plots were sprayed with a 1 concentration (1.5 lbs ha 1 glufosinate), whereas the other two plots were sprayed with a 2 concentration (3.0 lbs ha ). In 2004 and 2005, all plots were treated with the 2 concentration. All treatments were applied using a backpack sprayer. In 1997 and 2005, the degree of herbicide resistance was scored approximately 4 weeks after herbicide application and categorized into three classes: 0, no resistance (heavy necrosis or dead plants); 0.5, intermediate resistance (moderate necrosis); 1, high resistance (no or very slight necrosis). For protein (ELISA) analysis, the uppermost shoot tips of the plants from one of the six plots were taken in 2005, approximately 5 weeks after herbicide treatment in June and after all trees, even susceptible ones, had recovered and resumed growth. For protein extraction, approximately 50 mg of leaf tissue was ground in 400 l of extraction buffer (50 mM NaHPO4, pH 7.0; 10 mM EDTA) in a 1.5-ml microtube with a disposal pestle. The samples were centrifuged at 16,000g for 15 minutes at 4° C in a tabletop microcentrifuge. Approximately 200 l of supernatant was transferred to a new 1.5-ml tube, frozen in liquid nitrogen, and stored at 80° C until it was assayed. The total protein concentration was measured by the Figure 1. Schematic diagram of the T-DNA region of the plasmid vector pTTM8 used to transform the two hybrid poplar clones. OCSt, the 3 untranslated region from the octopine synthase gene; NEO, neomycin phosphotransferase II; pNOS, the promoter from the nopaline synthase gene; pTA29, the promoter from tobacco anther-specific gene TA29; NOSt, the 3 untranslated end of the nopaline synthase gene; pSSUARA-TP, the promoter from the atS1A ribulose-1,5-biphosphate carboxylase small subunit gene from Arabidopsis thaliana; G7t, the 3 untranslated fragment from the TL-DNA gene 7; RB, right border; LB, left border. T-DNA regions are not drawn to scale. Figure 2. Herbicide damage after herbicide application. Herbicide was sprayed with a hand applicator at a concentration of 3.0 lbs ha 1 active ingredient (glufosinate). Leaves on the nontransgenic control plants quickly desiccated and browned, whereas leaves on the highly tolerant transgenic plants remained green and free of lesions (only shadows from bright sunlight are visible). 90 WEST. J. APPL. FOR. 23(2) 2008 Bradford method (Protein Assay Kit, catalog no. 500-0001; BioRad, Hercules, CA) using a microtiter plate reader (Molecular Devices, Sunnyvale, CA) following the instructions provided by the manufacturer. The relative concentration of the phosphinothricin acetyltransferase enzyme (PAT), encoded by the BAR gene, was quantified using the commercially available LibertyLink PAT/bar ELISA kit (catalog no. AP013; Envirologix, Inc., Portland, ME) following the manufacturer’s instructions. Blanks and nontransgenic controls were also included in each assay plate. Duplicated wells were used for all tested samples. The optical density (OD) was read at a wavelength of 450 nm using a microtiter plate reader 20 minutes after the stop solution was added. The mean OD from the blank wells was subtracted from the samples and nontransgenic controls before data analysis. Results and Discussion Herbicide Resistance Levels and Stability over Time A total of 384 transgenic poplar plants, derived from 32 independent transformation events, were grown in the field over a period of 8 years. In 1997, the herbicide treatment was applied approximately 1 month after the establishment of the plants in the field, and herbicide resistance was scored approximately 4 weeks after the herbicide application. We found that the mean responses for the two herbicide levels used were not significantly different from each other (0.7 and 0.69, respectively), and therefore, all resistance data were pooled for further expression analysis. After averaging herbicide resistance scores over all ramets of each individual event, we redefined herbicide classes based on the distribution of the mean scores into what might be operational levels for breeding programs: no resistance (score 0); intermediate, noncommercial resistance (score 0.1–0.9); and high, potentially commercial resistance (score 1) (Figure 3). All the treated, nontransgenic plants showed severe herbicide damage for both years, which indicates that both herbicide applications were sufficient to cause damage to any major sectors of transgenic plants in which gene silencing occurred (Figure 2). Of 32 transgenic events studied in 1997, 13% of the events were highly susceptible, 28% were highly resistant, and 59% were intermediate (Table 1). All the events that belonged to the highly susceptible and highly resistant groups in 1997 remained in the same groups in 2005. However, nearly 50% of the events in the intermediate group in 1997 showed high resistance in 2005. As a result, the herbicide resistance levels in 2005 appeared to be less continuous, and a majority of the events (59%) showed high resistance. The correlation coefficient of herbicide resistance between the two years was 0.58 and highly statistically significant (P 0.0003). Events with intermediate resistance appeared to be more likely to be affected by the plant’s physiological condition and the concentration and efficiency of herbicide application, which can vary widely between years. The lower mean resistance observed in 1997 was not surprising, as herbicide had been applied shortly after planting, when the plants were small and, thus, more completely treated with herbicide. In addition, they also might have had a less developed cuticle, possibly facilitating herbicide entry into cells. Correlation of Visualized Herbicide Resistance and Measured Protein Levels To determine the extent to which a simple measure of transgene expression could predict herbicide resistance, we performed sandwiched protein ELISA assays on 30 transgenic events and nontransgenic controls. Of the 30 events studied, 5 showed no herbicide resistance, 6 showed intermediate resistance, and the remaining events exhibited complete resistance. We used OD as an indication of relative expression level of the protein encoded by the BAR gene. For each event, the mean herbicide resistance and optical density over two ramets were used for statistical analysis. The association between transgene expression and herbicide tolerance was strong. The relative protein concentration of events in the highly sensitive class was very low, close to that of nontransgenic controls, whereas those of the high resistance events showed much higher OD readings (Figure 4). The coefficient between herbicide Figure 3. Distribution of mean herbicide resistance of 32 transgenic events in 1997 (A) and in 2005 (B). Herbicide resistance was scored using three classes: 0, no resistance; 0.5, intermediate resistance; 1.0, high resistance. Mean scores averaged over all ramets per event were used to create the histogram. Table 1. Percentage of transgenic events in each of three herbicide resistance classes in 1997 and 2005. Year Herbicide resistance class None (0) Intermediate (0.1–0.9) High (1.0) 1997 13% 59% 28% 2005 13% 28% 59% Percent retention in 2005 100% 47% 100% a Percentage of events in each class that were originally categorized in 1997 and remained in the same class in 2005. WEST. J. APPL. FOR. 23(2) 2008 91 resistance and OD was 0.98 and highly statistically significant (P 0.0,001); thus, herbicide resistance could be predicted on the basis of measured OD (Table 2). All events in the lowest 10% of OD values were highly susceptible, and all events in the highest 50% of OD values were highly resistant. These results suggest that a simple measurement of expressed protein levels could be used in the greenhouse or during early field testing to identify the top-performing events, reducing field trial costs and/or increasing experimental precision.