‘Real time’ genetic manipulation: a new tool for ecological field studies



Field experiments with transgenic plants often reveal the functional significance of genetic traits that are important for the performance of the plants in their natural environments. Until now, only constitutive overexpression, ectopic expression and gene silencing methods have been used to analyze gene-related phenotypes in natural habitats. These methods do not allow sufficient control over gene expression for the study of ecological interactions in real time, of genetic traits that play essential roles in development, or of dose-dependent effects. We applied the sensitive dexamethasone (DEX)-inducible pOp6/LhGR expression system to the ecological model plant Nicotiana attenuata and established a lanolin-based DEX application method to facilitate ectopic gene expression and RNA interference-mediated gene silencing in the field and under challenging conditions (e.g. high temperature, wind and UV radiation). Fully established field-grown plants were used to silence phytoene desaturase and thereby cause photobleaching only in specific plant sectors, and to activate expression of the cytokinin (CK) biosynthesis gene isopentenyl transferase (ipt). We used ipt expression to analyze the role of CKs in both the glasshouse and the field to understand resistance to the native herbivore Tupiocoris notatus, which attacks plants at small spatial scales. By spatially restricting ipt expression and elevating CK levels in single leaves, damage by T. notatus increased, demonstrating the role of CKs in this plant–herbivore interaction at a small scale. As the arena of most ecological interactions is highly constrained in time and space, these tools will advance the genetic analysis of dynamic traits that matter for plant performance in nature.


Experiments with transgenic plants in natural environments are often indispensable for ecological research, because the complex blend of abiotic and biotic factors can reveal plant phenotypes which might be absent under the coddled conditions of the laboratory and glasshouse (Izawa et al., 2011; Baldwin, 2012; Kaur et al., 2012; Dinh et al., 2013). Until now the genetic tools for ecological field studies have been mainly restricted to the use of mutants and constitutive silencing or overexpression technologies. These techniques allow only functional analysis of genes that do not cause strong developmental defects, since these confound the analysis of traits that are important for ecological interactions with other organisms. Constitutive techniques also do not allow restricted fine-scale transcriptional regulation in specific plant tissues or developmental stages that is necessary to address basic questions about the spatial dynamics of herbivore feeding or to study season-specific interactions with herbivores. Additionally they complicate the work with plant traits whose ecological functions are dose dependent or are tightly regulated in time. Expression systems using tissue-, stress- or developmental-specific promoters (Potenza et al., 2004; Moore et al., 2006), like the stress- and ontogeny-regulated SARK (senescence-activated protein kinase) promoter and the stress-inducible HVA22P promoter have been used in the field (Xiao et al., 2009; Qin et al., 2011), but are limited to specific tissues, stresses or developmental stages. Chemically inducible expression systems provide the flexibility required for studies of ecological interaction, since they allow immediate control over spatial, temporal and quantitative construct expression. To the best of our knowledge, such expression systems have only been used under controlled laboratory and glasshouse conditions in several plant species, and their use in ecological field research remains untested.

Various chemically inducible systems, which express the target construct only in the presence of specific compounds like estradiol, alcohol or dexamethasone (DEX), have been developed (Moore et al., 2006; Corrado and Karali, 2009). There are several reports of conditional basal expression (the alc system; Salter et al., 1998; Roslan et al., 2001), chimeric patterns (the Cre/loxP recombination system; Guo et al., 2003) and side-effects to the plant from the chemical elicitors (ethanol; Camargo et al., 2007) or the activated expression system itself (the GVG system; Kang et al., 1999). One of the most sensitive expression systems, which allows for the regulation of gene expression with minimal side-effects on the plant's physiology, is based on the DEX-inducible pOp6/LhGR system.

The pOp6/LhGR expression system was developed from the pOp/LhG4 system (Moore et al., 1998) by Craft et al. (2005) and Samalova et al. (2005). The system comprises a constitutively expressed chimeric transcription factor (LhGR) containing a high-affinity DNA binding lac-repressor domain, a Gal4 transcription activator region and the ligand-binding domain of a glucocorticoid receptor. The target construct is under control of a minimal CaMV promoter downstream of an array of lac-operator repeats. In the presence of DEX, the transcription factor dissociates from heat shock proteins (Picard, 1993), binds to the lac-operator array and activates the otherwise inactive minimal CaMV promoter, leading to expression of the target construct.

The system was tested for heterologous gene expression as well as for RNA interference (RNAi)-mediated gene silencing (Craft et al., 2005; Samalova et al., 2005; Wielopolska et al., 2005), for example to regulate the expression of the Agrobacterium tumefaciens isopentenyl transferases (ipt) coding gene Tumor morphology root (Tmr), which catalyzes the rate-limiting step in the biosynthesis of the cytokinin (CK) trans-zeatin (tZ; Heidekamp et al., 1983; Craft et al., 2005; Samalova et al., 2005; Ueda et al., 2012). Since minor CK changes already affect plant development (Medford et al., 1989; Bohner and Gatz, 2001), ipt represents a sensitive visual marker for analyzing dose-dependent induction. Gene silencing was tested by silencing phytoene desaturase (pds), which is involved in carotenoid biosynthesis, resulting in visually observable photobleaching (Chamovitz et al., 1993; Wielopolska et al., 2005; Zhang et al., 2010). In the traditional application methods, such as spraying or soil drenching (Aoyama and Chua, 1997; Samalova et al., 2005), the use of this steroid-based system has been restricted to the controlled environment of a laboratory (Moore et al., 2006; Corrado and Karali, 2009). The pOp6/LhGR system has found limited use for field studies, mostly due to the biological activity of DEX in humans (Walton, 1959) and other organisms (Miller et al., 1994), as well as the high risk of contaminating the environment.

Nicotiana attenuata is intensively used for ecological field experiments (e.g. Kessler et al., 2004, 2008; Allmann and Baldwin, 2010; Long et al., 2010; Weinhold and Baldwin, 2011; Meldau et al., 2012a; Schuman et al., 2012). The extensive use of transgenic N. attenuata plants in their native habitat in the analysis of plant–animal and plant–microorganism interactions has made it one of the best-characterized model organisms for understanding the genetic traits responsible for ecological performance under natural conditions.

Many ecologically important plant traits, such as defense responses against specific herbivores, are regulated by temporal and tissue-specific changes in plant hormones (Erb et al., 2012). Cytokinins, for example, mediate many developmental and stress-related processes (Werner and Schmülling, 2009). Yet the physiological responses to CKs are highly concentration-dependent and specific to different tissues and developmental stages. Unfettered manipulation of CKs induces severe developmental perturbations (Klee et al., 1987), which may confound the analysis of their role in the interactions of plants with native herbivores. A field-applicable pOp6/LhGR system, with locally restricted dose-dependent CK manipulations, would allow for rigorous tests of many hypotheses about the roles of CKs in plant ecological interactions (Giron et al., 2013).

Here we describe the application of the DEX-inducible pOp6/LhGR system for precisely controlled overexpression and gene silencing in N. attenuata and present an approach to DEX application that can be used in the field.


Establishment of the pOp6/LhGR system in N. attenuata

We tested the pOp6/LhGR system for field experiments in N. attenuata using two different pOp6 constructs (Figure 1). The i-irpds expresses a construct for RNAi-mediated gene silencing (Wesley et al., 2001) of the N. attenuata pds, leading to visible photobleaching, whereas the i-ovipt line ectopically expresses an ipt from A. tumefaciens (Heidekamp et al., 1983), leading to higher levels of CKs (Figure 1). Both lines contain a specific pOp6 construct (pOp6irpds for i-irpds and pOp6ipt for i-ovipt), as well as the LhGR construct (Figure 1 and Figure S1 in Supporting Information), which were combined by crossing plants homozygous for each of the individual constructs. All pOp6 and LhGR lines were independently screened (Figure S2) before the crossings. Separate screening allowed us to combine the most promising lines for both constructs, reuse the screened LhGR line and to combine the pOp6 constructs with inducer lines with tissue- or developmental-specific expression in the future (Moore et al., 2006). Hemizygosity also reduces possible insertion site effects by maintaining one wild-type (WT) allele. We modified the screening procedure optimized for N. attenuata (Figure S2; Gase et al., 2011) because of the high levels of hptII (Figure S3) silencing, which thwarted efficient hygromycin-based screening. After crossing the pOp6 with homozygous LhGR plants, we visually screened the resulting seedlings on agar plates containing DEX for photobleaching and CK over-accumulation phenotypes (Figure S4). The pOp6 and LhGR lines were selected by choosing the transformed lines that showed the highest inducibility. An example of line optimization is shown for LhGR in Figure S5 (line 92 was chosen for final crosses). Photobleaching in pds-silenced plants is only achieved in lines with high silencing efficiency. This allows efficient screening of functional LhGR lines by analyzing crosses with the corresponding i-irpds lines. Lines with insufficient phenotypes in the presence of DEX or with phenotypic changes in the absence of DEX were excluded. The final selected lines exhibit highly DEX-inducible seedling phenotypes, but are phenotypically normal in the absence of DEX (Figure S4).

Figure 1.

The pOp6/LhGR system for inducible construct expression. Constructs used for ectopic expression and gene silencing, their regulated transcripts and the expected phenotype. The color code (red, gene silencing; green, ectopic expression) is used consistently for all figures in the online version. ‘ir’ stands for inverted repeat, ‘pds’ for phytoene desaturase and ‘ipt’ for isopentenyl tranferase.

The DEX application procedure

We next established a DEX treatment procedure for mature plants for use under field conditions. We first tested the stability of DEX under conditions relevant for studies in the plant's natural environment (Table S1; Dinh et al., 2013). To test the effect of elevated temperature, a DEX-containing MeOH solution was incubated at 37°C for several days in the dark. Even after 6 days at 37°C only a small proportion of DEX was degraded (11% degradation; Figure S6a). However, when exposed for 6 days to the light environment of the glasshouse at 22°C, we found strong DEX degradation (63% degradation; Figure S6b), suggesting that DEX is sensitive to light. Since our field site in the Great Basin Desert (UT, USA) is characterized by light irradiation (Dinh et al., 2013) that is much higher than in the glasshouse, the stability of DEX was considered as a limiting factor under these conditions. Therefore DEX should be applied to areas of the plant that are protected from direct sunlight and repeatedly applied for long-term treatments. Spray application is not a useful field technique because wind increases the risks of DEX exposure to both the environment and the researcher. Lanolin is commonly used as a matrix for applying lipophilic substances to N. attenuata (Baldwin, 1996; Kessler and Baldwin, 2001; Meldau et al., 2011; Kallenbach et al., 2012). When the test tubes containing the DEX–MeOH solutions were covered with a thin layer of lanolin (about 1 mm), the degradation of DEX in the light environment of the glasshouse was significantly reduced (45% degradation; Figure S6b). Thus lanolin can help to increase the stability of DEX in high-light environments. To dissolve substances in lanolin, it is usually liquefied at 60°C and then mixed with the compound of interest. We tested whether heating leads to the breakdown of DEX. Since we were not able to purify DEX from lanolin, we analyzed its stability when dissolved in MeOH. Our results show that short exposures of DEX to 60°C did not affect its stability (Figure S6c). Therefore it is unlikely that degradation of DEX occurs during dissolution in lanolin. However, we cannot rule out that DEX has different stabilities in lanolin compared with in MeOH. To dissolve the DEX in lanolin, we used DMSO as the primary solvent. This is expected to improve absorption by the plant tissue (Williams and Barry, 2012), and DMSO has minimal side-effects on plant physiology at low doses when compared with other solvents such as ethanol (Samalova et al., 2005; Robison et al., 2006). Since we mainly manipulated the gene expression in leaves, most experiments described here involved the application of a thin layer of DEX-containing lanolin paste to the lower side of the petiole of the targeted leaves (Figure S7).

The pOp6/LhGR system under glasshouse conditions

Before performing experiments in the field, we tested the system under glasshouse conditions. To quantify DEX-induced silencing, we measured irPDS and PDS transcript accumulations, while for DEX-induced heterologous expression we analyzed IPT transcript accumulation and CK levels. The i-irpds plants showed 10-fold increases in irPDS transcripts 12 h after treatment with 100 μm DEX (Figure 2a) and this increased to more than 100-fold 1 day after DEX application (Figure S8a). Accumulation of the PDS transcript was only reduced by approximately 10% at both time points (Figures 2a and S8a). The i-ovipt plants increased IPT transcript abundance 250-fold after 12 h, leading to increases of more than 15-fold in tZ levels (Figure 2d). After 1 day, IPT transcripts increased more than 1000-fold and tZ by nearly 35-fold. Three days after treatment of a single petiole with 100 μm DEX, the bleaching in i-irpds plants and growth effects due to CK overproduction in i-ovipt plants were clearly visible (Figure S8b,d). The corresponding changes in transcript and metabolite level were found to be similar in the basal and apical part and the midvein of the leaves (Figure 2b,e). Three days after DEX treatment, the irPDS transcripts accumulated 10–40-fold in i-irpds plants compared with control levels, leading to reductions of 60–80% in PDS transcript levels (Figure 2b). Even if the photobleaching was only visible in newly established leaf tissue, completely green leaf parts showed strong pds silencing (apical parts in Figure 2b; compare Ruiz et al., 1998). The IPT transcript abundance increased 4000-fold 3 days after DEX application, leading to increases in tZ levels of 60–200 fold (Figure 2e). The application of different DEX concentrations showed a strong dose-dependent response with a high dynamic range (1–100 μm DEX) for gene silencing and for ectopic expression (Figure 2c,f,g). In addition to the clearly observable growth phenotypes (Figure 2c,f), tZ-type CKs increased 5–15-, 15–30-, 45–100- or 250–2500-fold 12 days after the application of 1, 5, 20 or 100 μm DEX, respectively (Figure 2g). We also observed that treatment of single leaves did not affect adjacent leaves, indicating that the pOp6/LhGR system can be used to manipulate spatially restricted responses in plants (Figure S9). To test if the DEX system can also be used to manipulate responses in roots, plants were grown in hydroponic cultures and DEX was applied directly to the hydroponic solutions. The induction of i-irpds plants by treating roots with DEX induced bleaching in the leaves (Figure S10a). Treatment of i-ovipt plants with DEX through the roots reduced root growth and changed the growth of the shoots (Figure S10b,c).

Figure 2.

pOp6/LhGR system in Nicotiana attenuata under glasshouse conditions.
(a) Inverted repeat PDS (irPDS) and PDS transcript level in i-irpds plants 12 h after application of lanolin paste containing dexamethasone (DEX) to the midvein.
(b) The irPDS and PDS transcript levels in the basal and apical parts of the leaf lamina, as well as in the midvein of i-irpds plants 3 days after application of lanolin paste containing DEX.
(c) The i-irpds plants 6 days after application of lanolin paste containing DEX.
(d) IPT transcript and trans-zeatin [ng per g fresh mass (g FM–1)] levels in i-ovipt plants 12 h after application of lanolin paste containing DEX to the midvein.
(e) IPT transcript and trans-zeatin (ng gFM–1) levels in the basal and apical parts of the leaf lamina, as well as in the midvein of i-ovipt plants 3 days after application of lanolin paste containing DEX.
(f) i-ovipt plants 12 days after application of lanolin paste containing DEX.
(g) trans-Zeatin (tZ), trans-zeatin riboside (tZR), trans-zeatin riboside O-glycoside (tZROG) and trans-zeatin 7-glycoside (tZ7G) levels in i-ovipt plants after 12 days of treatment with lanolin paste containing different concentrations of DEX.
Lanolin paste was applied every 3 days. Asterisks indicate significant differences between DEX-treated samples and the corresponding control (0 μm) (independent samples t-test: *P < 0.05; ***P < 0.001). irPDS, IPT transcript and trans-zeatin data, 12 h after DEX treatment were log10-transformed before t-test analysis. Small letters indicate significant differences between samples (one-way anova, Turkey honestly significant difference, P < 0.05). Error bars show standard errors (a, b, n = 6; d, n = 5; e, g, n ≥ 3). FM, fresh mass.

The pOp6/LhGR system under field conditions

After establishing the pOp6/LhGR system under glasshouse conditions, we tested its utility for fieldwork. When i-irpds plants were treated with 100 μm DEX in the field, the first signs of photobleaching were visible after 3 days (Figure 3a), leading to strongly bleached plants after 2 weeks (Figure S11). We tested if gene expression in single branches of pOp6/LhGR lines could be silenced without affecting the other branches. After decapitation, plants developed several equally sized side branches, which were treated individually with different DEX concentrations (Figure S12a). The application of DEX to side branches of LhGR plants did not influence plant growth (Figure S12b,c), indicating minimal side-effects of the DEX treatment under field conditions. Notably, treatments with 0.5 μm DEX were sufficient to induce visible photobleaching in the field (Figure 3b). The application of 0, 5, 20 or 100 μm DEX to different side branches of i-irpds plants induced strong concentration-dependent photobleaching (Figure 3c). The irPDS and PDS transcript abundances mirrored the visual photobleaching patterns (Figure 3d). Twelve days of treatment with 100 μm DEX induced 2000-fold induction of irPDS transcript abundance and more than 80% pds silencing (Figure 3d). No photobleaching was observed on the control branch which was directly adjacent to the branch treated with 100 μm DEX (Figure 3c). Single treatments of complete plants also showed reliable, dose-dependent photobleaching phenotypes (Figure 3e). The observed visible bleaching was also correlated with a strong decrease in chlorophyll content, as expected for pds silencing (Figure S13; Qin et al., 2007). To evaluate the risks of contamination for surrounding plants, adjacent untreated plants and control plants were also monitored, but no signs of cross-contaminations were found.

Figure 3.

Inducible gene-silencing in Nicotiana attenuata in its native habitat.
(a) i-irpds plants 3 days after application of lanolin paste containing dexamethasone (DEX).
(b) i-irpds plants 6 days after application of lanolin paste containing DEX on different side branches of the same plant.
(c) i-irpds plant 12 days after application of lanolin paste containing DEX on different side branches of the same plant.
(d) PDS (solid bars) and inverted repeat PDS (irPDS, open bars) transcript levels in leaves of side branches of i-irpds plants treated with lanolin paste containing DEX.
(e) i-irpds plants 12 days after a single application of lanolin paste containing DEX.
Lanolin paste was applied every 3 days. Experiments were performed in the Great Basin Desert, UT, USA. Error bars show standard errors (n = 6).

We also analyzed if DEX treatments of different branches of i-ovipt plants led to branch-specific changes in the levels of CKs or CK-related phenotypes. Different side branches were treated with 0, 0.05 and 0.1 μm DEX-containing lanolin paste, which was refreshed every 3 days. The treatment increased tZ levels only in DEX-treated branches but not after lanolin treatments (Figure 4a). Previous work has shown that external application of CKs to poplar leaves increases wound-induced levels of oxylipins (Dervinis et al., 2010). We found wound-induced levels of the oxylipin 12-oxo-phytodienoic acid (OPDA) to be significantly higher in leaves of branches treated with DEX, compared with the controls (Figure 4b).

Figure 4.

Tissue-specific elevated cytokinin levels increase Tupiocoris notatus damage in Nicotiana attenuata in its native habitat and in the glasshouse.
(a) trans-zeatin levels in leaves of different side branches of i-ovipt plants 15 days after application of lanolin paste containing different concentrations of dexamethasone (DEX).
(b) The 12-oxo-phytodienoic acid (OPDA) levels in untreated control leaves and 60 min after wounding. Samples were taken from different side branches of i-ovipt plants 15 days after application of lanolin paste containing different concentrations of DEX.
(c) Leaf damage caused by T. notatus on different side branches of i-ovipt plants treated with lanolin paste containing different concentrations of DEX for 12 days.
(d) Between-plants choice assays. Leaf damage caused by T. notatus on i-ovipt plants after a single application of lanolin paste containing either 0 (numbered white leaves) or 5 μm (numbered black leaves) DEX. Paired plants were exposed to T. notatus for 14 days.
(e) Within-plant choice assays. Leaf damage caused by T. notatus on i-ovipt plants after a single alternating application of lanolin paste containing either 0 (numbered white leaves) or 5 μm (numbered black leaves) DEX. Plants were exposed to T. notatus for 14 days.
Lanolin paste was applied every 3 days. Experiments were performed under field conditions in the Great Basin Desert, UT, USA (a–c) or in transparent boxes under glasshouse conditions (d, e). Asterisks indicate significant differences between DEX-treated samples and the corresponding control (0 μm) (paired samples t-test: *P < 0.05; **P < 0.01). Damage by T. notatus under glasshouse conditions was analyzed by a mixed-effects model. Error bars show standard errors (a, n = 8; b, n ≥ 5; c, n ≥ 15; d, n = 7; e, n = 10). FM, fresh mass.

Using the pOp6/LhGR system to study plant–herbivore interactions

Since we were able to subtly increase CK levels in specific plant tissues, we analyzed if this influences the resistance of the plant to native herbivores, as has been suggested (Smigocki et al., 1993, 2000; Mujer and Smigocki, 2001; Dervinis et al., 2010; Erb et al., 2012; Meldau et al., 2012b; Giron et al., 2013). We quantified herbivore damage to lanolin- and DEX-treated branches of i-ovipt plants. We found that one of the most abundant herbivores in the field, the mirid bug Tupiocoris notatus, caused significantly more damage to the DEX-treated side branches of i-ovipt plants than branches treated with lanolin (Figure 4c). The growth of the analyzed i-ovipt plants was not altered by these gentle increases in CK levels (Figure S14), excluding possible developmental perturbations as an explanation for the observed increase in susceptibility to mirids. To examine potential direct effects of DEX on plant susceptibility, we measured levels of leaf damage on LhGR plants treated with high concentrations (up to 20 μm) of DEX and found no difference in mirid damage (Figure S15).

We repeated the experiment under glasshouse conditions, using a highly accurate method for quantifying mirid damage (Figure S16). We compared the T. notatus-inflicted leaf damage between DEX-treated and control i-ovipt plants in a paired design (Figure 4d). Additionally, a within-plant leaf choice assay was conducted by exposing i-ovipt plants with alternating DEX-treated and control leaves to T. notatus (Figure 4e). In both experiments, the DEX-treated leaves with elevated CK levels showed greater damaged by T. notatus (Figure 4d,e). Interestingly, the interaction between treatment and leaf position significantly affected the mirid damage in the between-plant choice assays, indicating that the effects of DEX and/or cytokinins on mirid damage are dependent on leaf position (Figure 4d).

We also tested whether DEX itself influences the performance of the specialist lepidopteran herbivore Manduca sexta and did not find significant effects (Figure S17).


Genetic tools for ecological research

In addition to constitutive and stress-, tissue- or developmentally regulated manipulations, chemically inducible expression technologies are increasingly being used in basic research as well as agriculture (Corrado and Karali, 2009). As pointed out by Corrado and Karali (2009), for studying molecular processes in plants, an inducible expression system should show fast, strong and concentration-dependent activity after induction but have insignificant activity in the absence of the inducer. The inducer should also have no pleiotropic effects and be applied in a flexible manner. In contrast, most techniques used in commercial and field-based applications have been selected by considerations of cost efficiency and lack of impact on the ecosystem, but lack the necessary precision required for molecular biology research. Here, we have developed a system that provides both the precision needed for surgical manipulation of gene expression and the robustness required for fieldwork.

Adaptation of the pOp6/LhGR system to ecological field research

The pOp6/LhGR system is one of the most popular inducible expression systems used in plant molecular biology, but its use has been restricted to the laboratory, since treatments with steroids require controlled environments (Moore et al., 2006; Corrado and Karali, 2009). The usual application methods for DEX include incorporation into the growth medium and spraying, as well as watering with aqueous DEX solutions (Aoyama and Chua, 1997; Samalova et al., 2005). Since DEX is a glucocorticoid known for its anti-inflammatory activities in humans (Walton, 1959), strategies for the application of DEX should avoid the uncontrolled formation of aerosols as well as direct skin contact or ingestion. The influence of DEX on other study organisms inhabiting natural environments is also a consideration, as DEX can suppress the immune responses of insects like M. sexta, by inhibiting phospholipase A2 and thereby eicosanoid biosynthesis (Miller et al., 1994). Since field experiments are often done under unpredictable conditions with limited technical support, the method used to apply DEX should be simple and work under a variety of environmental conditions without posing risks to personnel and the environment.

The field site used in this study is located in the Great Basin Desert, UT, USA, which is a natural habitat for N. attenuata and is characterized by high light irradiation, high temperatures, drought and intense wind (Table S1; Dinh et al., 2013), conditions which make the application of DEX by spraying or watering untenable. Many studies with N. attenuata have shown that lanolin can be used to apply substances to the plant (Baldwin, 1996; Kessler and Baldwin, 2001; Steppuhn et al., 2004) in its natural environment. A lanolin-based application method for DEX was also mentioned by Borghi (2010), but no data were reported. We therefore applied DEX-containing lanolin paste to the plants, which prevents aerosol formation, remains locally restricted and is expected to protect DEX from light-mediated degradation (Figure S6b). Lanolin-based applications are also expected to continuously supply DEX to the plant, which ensures stable long-term effects. To reduce the exposure of herbivorous insects to DEX, we only applied a thin layer of DEX-containing lanolin paste to the lower side of the petioles (Figure S7), which represents a shaded position and is less frequently visited by herbivores. This application also allows the DEX to access the leaf vasculature, leading to optimal distribution of DEX in the attached leaf. The application of DEX also did not affect the performance of T. notatus and M. sexta (Figures S15 and S17); however, control experiments should be performed when working with other herbivores. Since the DEX-treated plants were harvested at the end of the field season, DEX contamination of the environment is unlikely. Personal contact with DEX can be avoided by basic protection measures (protective clothes and gloves).

The efficiency of the pOp6/LhGR system in N. attenuata

We used the silencing of pds (i-irpds) and the heterologous ipt expression (i-ovipt) to evaluate the suitability of the pOp6/LhGR system for field studies. pds is highly expressed in leaf tissues and pds silencing results in the bleaching of newly developed above-ground tissues (Figures S8b and S11). Since the consequences of silencing are readily observable in vivo, it represents an ideal marker with which to track the effectiveness of silencing under field conditions. We used ipt-mediated changes in CK levels to test the sensitivity of the system, because even relatively small increases in ipt expression can lead to changes in plant development (Medford et al., 1989; Bohner and Gatz, 2001). As such, ipt expression is a very sensitive marker for ‘leaky’ pOp6 activity. In addition to visual observations and transcript analysis, the CK levels of i-ovipt plants were quantified by mass spectrometry.

In the absence of DEX, no physiological changes were observed in any of our selected lines, indicating negligible background activity, while the application of DEX induced transcript expression within hours (Figure 2a,d). Detailed analysis of different leaf parts revealed that petiole treatments are sufficient to regulate gene expression in the entire leaf (Figure 2b,e). The induced changes could be regulated in a temporal, spatial and quantitative manner and the dynamic range of the system was excellent (Figures 2, S8 and S9).

Phenotypes that are obtained in the field should be verified under controlled conditions in the glasshouse and the lab, and hence an inducible expression system should work with similar efficiencies under both conditions. Figure 3 shows that these requirements can be fulfilled. One major challenge for experiments with transgenic plants in natural habitats is that only a limited number of plants can be grown. When appropriate, eliciting the growth of equal-sized lateral branches by decapitation allows for branch-specific transcriptional manipulation (Figure S12a,b) and comparisons between DEX-induced and control side branches from the same plant (Figures 3b–d and 4a). In addition to reducing the number of replicate plants required, the use of the same individual plant for treatments and control also helps to compensate for environmental variation between the plants, as well as for insertion side-effects of the construct. Clearly, this approach should be used with caution when manipulating systemically transmitted traits.

The pOp6/LhGR system for studies of plant–herbivore interaction

Cytokinins are important targets for crop improvement as they influence plant traits such as leaf senescence (Richmond and Lang, 1957; Gan and Amasino, 1995), drought resistance (Werner et al., 2010; Qin et al., 2011) and resistance against pathogens (Choi et al., 2010; Großkinsky et al., 2011). The influence of CKs on traits important for resistance against insect herbivores in the natural environment remains unstudied. Constitutive ipt expression leads to abnormal development, which confounds the analysis of natural plant–herbivore interactions. Here we used the i-ovipt plants to subtly elevate CK levels without affecting plant development and analyzed their role in resistance to natural herbivores in the field. We analyzed leaf damage after elevating CK levels in particular side branches of decapitated plants. As intended, only mild changes in CK levels were measured (Figure 4a). These subtly higher CK levels significantly increased the accumulation of jasmonate (JA) precursors after wounding (Figure 4b). Since the JA pathway is known to regulate the levels of defense metabolites involved in the resistance of N. attenuata to herbivores (Halitschke and Baldwin, 2003), increased CK levels in leaves were expected to correlate with an enhanced resistance to herbivores. However, the DEX treatment of the i-ovipt plants increased the leaf damage by T. notatus in both the field and the glasshouse (Figure 4c,d,e). Since mirids are only slightly affected by most JA-mediated defense responses, except diterpene glycosides (Dinh et al., 2013), increased OPDA levels might even induce metabolic changes that are favorable for mirids. Changes in OPDA were observed after mechanically wounding leaves, and it is not clear how far OPDA levels are affected by mirid feeding itself. The within-plant choice assays (Figure 4e) indicate that regulating CK levels are sufficient to shape the patterns of herbivore damage in a plant, which is an important aspect of ecological theories such as the optimal defense theory (Meldau et al., 2012b). Since CKs regulate source–sink relationships (Kuiper, 1993; Ehness and Roitsch, 1997; Lara et al., 2004), they likely enhance the nutritional quality of a tissue, which in turn attracts T. notatus feeding, resulting in higher levels of damage. Interestingly, Figure 4(d) indicates that the influence of changes in the CK levels on the mirid damage were dependent on leaf position, which might correlate with the inhomogeneous distribution of CKs between different plant parts (Hewett and Wareing, 1973; Ori et al., 1999). Future experiments will reveal the mechanism behind leaf-specific, CK-mediated plant susceptibility to insects and their contributions to herbivore resistance.

Potential future applications of the DEX-inducible pOp6/LhGR system

Since plants are the foundation for most food chains on the planet, the timing of their activities profoundly orchestrates most ecological interactions. Not surprisingly, many plant traits are not constitutively expressed throughout a plant's body, but are restricted to specific tissues or ontogenic stages and particular times. The expression of these traits is likely to reflect an evolutionary adaptation to the spatial and temporal activity patterns of interacting organisms (Figure 5). However, the tools available to ecologists for manipulating these interactions have been too ham-fisted to disentangle these sophisticated environmental interactions. ‘Real-time’ genetic tools such as the pOp6/LhGR system allow for the manipulation of such conditionally expressed traits in ecologically relevant situations (Figure 5). Many traits important for interactions with pollinators, such as nectar production and scent emission, are only transiently expressed at particular stages of flower development. Using the DEX system to surgically manipulate these traits can prevent off-target effects in other tissues, which may confound interactions with non-target organisms.

Figure 5.

‘Real-time’ genetic manipulations for studying plant ecology.
‘Real-time’ genetic manipulations can be used to study traits involved in biotic interactions of plants in different growth stages during seasonal progression. Black broken lines separate different plant growth stages. The shaded (blue) boxes illustrate the manipulation of tissue-specific target traits in different growth stages. Differentially colored ovals are examples of season-specific biotic interactions of plants.

The DEX system can also be applied to manipulate the spread of an induced defense, as its elicitation by herbivory. Analyzing the herbivore's feeding behavior on such plants will shed light on the role of complex spatio-temporal changes that are induced throughout a plant's body. This is especially valuable for the analysis of defense and signaling components whose constitutive manipulations do not yield viable plants (Meldau et al., 2011).

The pOp6 lines also offer the possibility for crosses with other activator lines with stress-, tissue- or developmentally regulated promoters, generating various functional combinations (Moore et al., 2006). This provides the possibility to broaden application to specific cell types, such as trichomes. Additionally the use of constitutive activators (Moore et al., 1998) might be beneficial for the analysis of tissues, such as roots, which are difficult to treat with DEX under field conditions.

The presented method could also be applied for field experiments with other transformable plant species like petunia (Kessler et al., 2013), peanut (Qin et al., 2011) or rice (Xiao et al., 2009). Next-generation sequencing technologies allow us to gain the genetic information required for the preparation of species-specific RNAi constructs from many plant species. This genetic information is not required for DEX-mediated ectopic expressions. We hope that the possibilities offered by the pOp6/LhGR system will encourage researchers to develop transformation systems for non-model plant species.

Conclusion and outlook

Flexible control over gene expression is highly desirable for ecological experiments in natural environments. Due to its ease of use and the experimental flexibility, affording precise manipulation of gene expression, we predict that the DEX-pOp6/LhGR system described here will allow scientists to revisit hypotheses about the function of traits whose analyses were previously thwarted by the inflexibility of the available technologies.

Experimental Procedures

Vector construction

Genomic DNA from transgenic tobacco plants harboring the A. tumefaciens tmr gene for isopentenyltransferase (IPT) under the control of the DEX-inducible LhGR/pOp6 promoter system was used as template to amplify the LhGR cassette and the pOp6-ipt cassette by PCR with primer pairs described in Table S3. After digestion with SalI the LhGR fragment was cloned in vector pSOL8DC2 cut with SalI and EcoRV, yielding pSOL9LHGRC (GenBank JX185747; Figure S1a). The pOp6-tmr cassette was cloned after digestion with SacI and HindIII in the vector pVKH18 digested with the same enzymes, yielding pPOP6IPT (GenBank JX185749, Figure S1b). The pPOP6IRPDS vector (GeneBank JX185750; Figure S1c) was constructed by replacing the 0.7-kb SalI-partial-BamHI fragment of pPOP6IPT with a 0.3 kb inverted repeat of a part of the N. attenuata pds gene (GeneBank JX185751), separated by an intron.

The plants were kindly provided by Bretislav Brzobohaty (Masaryk University, Brno, Czech Republic) and the pVKH18 vector by Ian Moore (Oxford University, UK).

Plant transformation and growth

Plants from the 30th inbred generation of the inbred ‘UT’ line of N. attenuata (Torr. ex S. Wats.) were transformed with the vectors mentioned above. The transformation with A. tumefaciens (strain LBA 4404), seed germination and growth under glasshouse conditions were done as described by Krügel et al. (2002). Field experiments were conducted under the US Department of Agriculture Animal and Plant Health Inspection Service permission number 11-350-101r for the LhGR, i-irpds and i-ovipt plants. For field experiments, plants were grown as described by Schuman et al. (2012). The LhGR and i-ovipt plants for herbivore damage analysis were grown in pairs on a field plot located at latitude 37.141 and longitude 114.027 (Figure S11).

Application of DEX

The DEX was dissolved in DMSO and diluted to 100 times the final concentration of the lanolin paste or 2000 times that of the GB5 medium and hydroponic medium, respectively. Aliquots were stored at −20°C. The final DEX concentration for GB5 medium was 20 μm and for hydroponic solution 1 μm. Lanolin was liquefied at 60°C, DEX was added, and after thorough mixing the lanolin paste was taken up by a syringe (1 ml, Omnifix), in which the lanolin solidifies after cooling. These syringes can be directly used for plant treatments. As control treatments, the plants were treated with the corresponding amount of DMSO in lanolin without DEX. For phenotypic analyses of seedlings, the seeds were germinated on GB5 medium containing DEX. Applications of DEX to the hydroponic medium were performed 1 week after plants were transferred to the pots. Lanolin treatments were performed at the earliest 3 weeks after germination, when plants were already transferred to pots. The time of treatments varied according to the experiment. If not stated otherwise, for glasshouse experiments plants were treated in the early rosette stage of growth. For T. notatus experiments and under field conditions, plants were treated during the early flowering stages. Side branches of decapitated plants were treated after exceeding lengths of 3–5 cm. If not stated otherwise, lanolin paste was applied to the lower side of a petiole (Figure S7). Depending on leaf size, between 10 and 30 μl of lanolin paste was applied per petiole. In contrast, for the short-time experiments (12 and 24 h) the entire midveins of the leaves were treated.

The DEX analysis

For DEX degradation experiments a 50 mm DEX stock solution in DMSO was dissolved in pure MeOH to a concentration of 20 μm DEX. Aliquots were placed in clear 1.5-ml glass vials (Machery Nagel, http://www.mn-net.com/) and were incubated at 37°C in the dark, in a 60°C water bath or under glasshouse conditions with or without a lanolin layer of about 1 mm covering the vial. Samples were taken at indicated time points. The samples were diluted 1:20 with MeOH and stored at –80°C. Before measurement, 200 ng of [9,10-2H]dihydro-JA was added per sample as an internal standard for relative quantification. Samples were analyzed by liquid chromatography-tandem mass spectrometry (LC-MS/MS) on a Varian 1200 Triple-Quadrupole-LC-MS system (Varian, http://www.varian.com/). Separation was done on a Kinetex C18 column (50 × 2.10 mm, 2.6 μm, 100 A; Phenomenex, http://www.phenomenex.com/). The mobile phase comprised solvent A (water, 0.05% HCOOH, 0.1% acetonitrile) and solvent B (MeOH) used in a gradient mode: 0–1 min, 95% A; 1–8 min 5–98% B in A; 8–15.5 min 98% B; 15.5–17 min, 2–95% A; 17–20 min, 95% A; with a flow of time/flow (ml min−1): 0–0.5 min, 0.4 → 0.2 ml min–1, 0.5–15.5 min, 0.2 ml min−1, 15.5–16 min, 0.2 → 0.4 ml min−1, 17–20 min, 0.4 ml min−1. For detection, the mass spectrometer was operated in negative mode and multireaction monitoring (MRM) was performed to monitor analyte parent ion → product ion (Table S4). Varian MS Workstation Version 6.6 software was used for data acquisition and processing.

Cytokinin analysis

Cytokinins were analyzed by extraction of plant tissue in acidified aqueous MeOH followed by two solid-phase extraction (SPE) steps and subsequent measurement with LC-MS/MS. The methodology was adapted according to Dobrev and Kamı́nek (2002) with the modifications by Kojima et al. (2009).

In brief, 100 mg of ground frozen plant tissue was extracted twice with 800 μl MeOH:H2O:HCOOH (15:4:1) at −20°C. Labelled internal standards were supplemented in the first extraction step. Extraction and SPE were performed in 96-well BioTubes (1.1 ml individual tubes; Arctic White LLC, http://www.arcticwhiteusa.com/) and Nunc 96-well Deep Well Plates (Thermo Scientific, http://www.thermoscientific.com/). The first SPE step was performed on a Multi 96 HR-X column (96 × 25 mg) (Macherey-Nagel) conditioned with extraction buffer. The flow-through was collected and the MeOH was evaporated under constant nitrogen flow in an Evaporator system (Glas-Col, http://www.glascol.com/) at 42°C. After replenishment with 850 μl 1 m HCOOH, the samples were loaded on a Multi 96 HR-XC column (96 × 25 mg) (Macherey Nagel), conditioned with 1 m HCOOH. After washing with (i) 1 ml 1 m HCOOH, (ii) 1 ml MeOH and (iii) 1 ml 0.35 m NH4OH, (iv) the CK-ribosides, free bases and glucosides were eluted with 1 ml 0.35 m NH4OH in 60% MeOH. The SPE were performed using a Chromabond Multi 96 vacuum chamber (Macherey-Nagel). After evaporation, samples were reconstituted in 50 μl 0.1% acetic acid.

Chromatography was performed on an Agilent 1200 HPLC system (Agilent Technologies, http://www.home.agilent.com/). For separation a Zorbax Eclipse XDB-C18 column (50 × 4.6 mm, 1.8 μm, Agilent Technologies) was used. The mobile phase comprised solvent A (water, 0.05% formic acid) and solvent B (acetonitrile) with the following elution profile: 0–0.5 min, 95% A; 0.5–5 min, 5–31.5% B in A; 5.01–6.5 min 100% B and 6.51–9 min 95% A, with a flow rate of 1.1 ml min–1. The column temperature was maintained at 25°C. The liquid chromatograph was coupled to an API 5000 tandem mass spectrometer (Applied Biosystems, http://www.appliedbiosystems.com/absite/us/en/home.html) equipped with a Turbospray ion source. For detection the mass spectrometer was operated in positive ionization mode (MRM modus) to monitor analyte parent ion → product ion (Table S5). Settings were as follows: ion spray voltage, 5500 eV; turbo gas temperature, 700°C; nebulizing gas, 70 p.s.i.; curtain gas, 25 p.s.i.; heating gas, 60 p.s.i.; collision gas, 6 p.s.i. Both Q1 and Q3 quadrupoles were maintained at unit resolution. analyst 1.5 software (Applied Biosystems) was used for data acquisition and processing. tZ, tZR, tZROG and tZ7G were quantified by using deuterated internal standards (Table S5; Olchemim, http://www.olchemim.cz/).

The OPDA analysis

The OPDA was extracted and analyzed by LC-MS/MS as described by Kallenbach et al. (2010).


Methanol was purchased from Merck (http://www.merck.com/), HCOOH for ultra-performance LC by Fisher Scientific (http://www.fisher.co.uk/), otherwise by Riedel-de Haën (http://www.riedeldehaen.com/), DEX by Enzo Life Sciences (http://www.enzolifesciences.com/), lanolin, as well as methyl JA and DMSO, by Sigma-Aldrich (http://www.sigmaaldrich.com/), GB5 by Duchefa (http://www.duchefa-biochemie.nl/), acetic acid by Roth (http://www.carlroth.com/) and CK standards by Olchemim. The [9,10-2H]dihydro-JA was synthesized by saponification and deuteration of methyl JA.

Transcript analysis

The RNA extraction was done with TRIzol (Invitrogen, http://www.invitrogen.com/), according to the manufacturer's instructions. Complementary DNA was synthesized by reverse transcription using oligo(dT) primer and RevertAid reverse transcriptase (Invitrogen). Quantitative (q)PCR was performed using actin as standard on a Stratagene Mx3005P qPCR machine using a SYBR Green containing reaction mix (Eurogentec, http://www.eurogentec.com/; qPCR Core kit for SYBR Green I No ROX). The primer sequences are summarized in Table S2.


Data were analyzed with SPSS Statistics 17.0. Either an independent sample t-test, paired samples t-test or one-way anova followed by Tukey's honestly significant difference test were used as indicated. Use of data transformation is indicated. R 3.0.1 was used for statistical analysis of mirid-damage data obtained in the glasshouse experiments (Figure 4d,e). A mixed-effects model was applied with cage and plant as random factors and treatment (DEX application), leaf position and their interaction as fixed factors. The model was simplified by stepwise elimination of fixed factors. To achieve the influence of the different fixed factors, a maximum likelihood ratio test of the different models was done. Mirid damage data were arc sin square-root transformed.

Tupiocoris notatus damage

Tupiocoris notatus damage in the field was estimated as a percentage of the total leaf area. Glasshouse choice assays were carried out on matured N. attenuata plants with flowers removed. Plants were kept in 25 × 25 × 50 cm (length × width × height) sealed glass cages, with 50–100 T. notatus adults and older nymphs at the start of experiment. The T. notatus were obtained from an in-house colony. Damage was quantified after 14 days. The inter-plant choice assays were carried out with two plants per cage; one treated with 0 and the other with 5 μm DEX (Figure 4d). Damage to the first eight stem leaves was determined. The intra-plant choice assays were carried out with leaves alternately treated with 0 or 5 μm DEX. Odd-numbered leaves, beginning with the oldest stem leaf, were DEX induced, while even-numbered leaves served as controls (Figure 4e). To determine the rate of damage we took pictures of the leaves and used Adobe Photoshop CS5 (http://www.adobe.com/) for picture analysis. We quantified the area of each leaf and manually marked damaged parts and calculated the percentage of the damaged area. Leaf area was marked as damaged if the leaf showed clear signs of T. notatus feeding, such as locally bleached chlorotic spots or mirid frass (Figure S16).

Manduca sexta

For M. sexta performance, freshly hatched neonates were placed on early rosette-stage LhGR plants pre-treated for 1 day with 0 or 100 μm DEX. Lanolin paste was applied every 3 days. Caterpillar mass was measured at the indicated time points. The M. sexta larvae were obtained from in-house colonies.


We thank Mario Kallenbach and Meredith Schuman for helpful scientific comments; Mario Kallenbach, Matthias Schöttner, Antje Wissgott, Susanne Kutschbach, Wibke Kröber and Eva Rothe for technical assistance; Tamara Krügel, Andreas Weber and Andreas Schünzel from the glasshouse team for plant cultivation; Grit Kunert for help with the statistical analysis; and Brigham Young University for the use of their Lytle Preserve field station. MS, KG, MR and IB are funded by the Max-Planck-Society and SM and CB are funded by Advanced Grant no. 293926 of the European Research Council to IB.