We describe pOp/LhGR, a dexamethasone-inducible derivative of the pOp/LhG4 transcription activation system, and its use in tobacco to regulate expression of uidA (encoding β-glucuronidase; GUS) and the cytokinin-biosnythetic gene ipt. The pOp/LhGR system exhibited stringent regulation and strong induced phenotypes in soil and tissue culture. In conjunction with an improved target promoter, pOp6, that carries six copies of an optimized lac operator sequence the pOp6/LhGR system directed induced GUS activities that exceeded those obtained with pOp/LhG4 or the CaMV 35S promoter but without increased uninduced activity. A single dose of dexamethasone was sufficient to direct cytotoxic levels of ipt expression in soil-grown plants although uninduced plants grew normally throughout a complete life cycle. In vitro, induced transcripts were detectable within an hour of dexamethasone application and 1 nm dexamethasone was sufficient for half maximal induction of GUS activity. Various methods of dexamethasone application were successfully applied under tissue culture and greenhouse conditions. We observed no inhibitory effects of dexamethasone or LhGR on plant development even with the highest concentrations of inducer, although tobacco seedlings were adversely affected by ethanol used as a solvent for dexamethasone stock solutions. The pOp/LhGR system provides a highly sensitive, efficient, and tightly regulated chemically inducible transgene expression system for tobacco plants.
Although the pOp/LhG4 system offers tissue-specific control over transgene expression through use of tissue-specific promoters, it provides minimal temporal control. To increase the versatility of this system we brought the existing pOp/LhG4 system under chemical control by adding the ligand-binding domain (LBD) of the rat glucocorticoid receptor (GR) to LhG4 creating a glucocorticoid-dependent transcription factor LhGR. The activity of many intracellular proteins can be hormonally regulated post-translationally when they are fused to the LBD of steroid receptors (Picard et al., 1988). This approach has been successfully applied to plant transcription factors by exploiting the LBD of glucocorticoid or oestrogen receptors (Aoyama et al., 1995; Lloyd et al., 1994; Sablowski and Meyerowitz, 1998; Simon et al., 1996). Neither GR LBD nor the commonly used inducing ligands such as dexamethasone have any reported detrimental consequences for plant development, although a synthetic transcription factor that incorporates the GR LBD can be detrimental when activated in Arabidopsis (Kang et al., 1999) and other species (Andersen et al., 2003; Ouwerkerk et al., 2001).
The principle of GR-based transcriptional systems is that in the absence of the steroid ligand, a transcription factor is trapped in an inactive complex via interaction between the GR LBD and heat-shock protein HSP90. This complex is believed to interfere with the function of the bound fusion protein by steric hindrance, blocking interactions with other proteins or with DNA target sites (Picard, 1993). The binding of dexamethasone to the LBD mediates dissociation of the fusion protein from HSP90. This model predicts that the effectiveness of the control exerted by GR LBD on a heterologous protein depends on how components of the HSP90 complex are positioned relative to at least one critical functional domain in the heterologous moiety. Domains that either face away from the HSP90 complex or are too distant would not be regulated (Picard, 1993). As hormone-binding domains form the C-terminus of steroid receptors, hormone-binding domains have usually been fused to the C-terminus of heterologous proteins, however the hormone-binding domain works at other positions (Böhner et al., 1999; Braselmann et al., 1993; Bruce et al., 2000; Eilers et al., 1989).
In the accompanying paper we describe the properties of three topologically different fusions between LhG4 and GR LBD in Arabidopsis (Craft et al., 2005). In this report we show that these components can be combined with improved target promoters (Craft et al., 2005) to provide a tightly regulated and efficient transcriptional activation system for tobacco. As reporter genes we used Escherichia coliuidA, encoding β-glucuronidase (GUS), and the Ti-plasmid derived ipt gene encoding isopentenyltransferase which catalyses an early step in cytokinin synthesis (Heidekamp et al., 1983). The GUS reporter provides a quantitative assay of gene expression while the ipt gene, which is active at very low expression levels (Böhner and Gatz, 2001; Gatz and Lenk, 1998), allowed us to determine whether the GR LBD is sufficient for practically complete inhibition of the transcriptional activity of LhGR in the absence of dexamethasone.
To determine the effectiveness of the pOp/LhGR system in tobacco we tested three different fusions between LhG4 and the LBD of a rat GR. These fusions differed in the position of the GR domain relative to the DNA-binding and activation domains of LhG4 (Craft et al., 2005). The GR domain was fused either at the N-terminus (LhGR-N), the C-terminus (LhGR-C) or internally, between the DNA-binding and transcription activation domains (LhGR-I) of LhG4. All constructs were placed under control of the CaMV 35S promoter (Figure 1a).
The pOp/LhGR system induces ipt phenotypes only in the presence of inducer
To test the activity of the three LhGR fusions, we used transgenic line pOp-ipt-S which carries the ipt gene of Agrobacterium tumefaciens under control of the pOp promoter (Lexa et al., 2002). The ipt gene encodes isopentenyl transferase which catalyses the first committed step in cytokinin biosynthesis (Åstot et al., 2000; Heidekamp et al., 1983). ipt expression results in a range of cytokinin-related phenotypes depending on its expression level (Böhner and Gatz, 2001; Estruch et al., 1991; Faiss et al., 1997; Gan and Amasino, 1995; van Loven et al., 1993; McKenzie et al., 1998; Medford et al., 1989; Schmülling et al., 1989; Smart et al., 1991; Smigocki, 1991) providing a sensitive assay of uninduced ‘leaky’ expression and an indicator of induced expression efficiencies. To confirm the utility of the pOp-ipt-S line, it was retransformed with the activator construct pBIN-35S-LhG4 which expresses the LhG4 transcription factor (Moore et al., 1998), and 25 of 26 transgenic shoots recovered turned into teratomas unable to form roots (data not shown). In parallel, the three LhGR constructs were transformed into the same reporter line and seedlings of the T1 generation germinated on kanamycin were screened for dexamethasone-inducible ipt expression. After 4 weeks growth on medium containing 20 μm dexamethasone, cytokinin-related phenotypes were observed in five of the 36 pOp-ipt-S/LhGR lines (these were 1 of 14 LhGR-N lines, 3 of 12 LhGR-I lines, and one of 10 LhGR-C lines; these five lines were named S/N, S/I1, S/I2, S/I3, and S/C respectively; Figure 2a, pOp-ipt-S). The morphological alterations in these lines included inhibition of root growth, thickening of the hypocotyl, reduced cotyledon area and formation of shoots in axils of the cotyledons. No such phenotypes were observed in seedlings grown without dexamethasone or in control seedlings (pOp-ipt-S and SR1) with or without dexamethasone (Figure 2a, pOp-ipt-S). Northern analysis performed on seedlings of the five inducible lines (S/N, S/I1, S/I2, S/I3, S/C) showed that LhGR transcripts of the expected size could be detected in all cases (Figure 1c,d).
To monitor the subsequent development of the ipt phenotype, six plants from lines S/N, S/I3, and S/C lines were maintained on media with dexamethasone. After a further 2-week growth, the plants on dexamethasone exhibited reductions in stem elongation, leaf expansion, and apical dominance, while root development was arrested (Figure 2b,c,e). After a further 6 weeks growth most of the induced plants developed teratomas at the base of the stem (Figure 2b,d). To establish the ability of each line to develop roots, 6-week-old plants were cut at the base of the stem and transferred onto fresh media (with or without 20 μm dexamethasone). S/N, S/I3, and S/C plants grown on dexamethasone produced no roots even after 12 days whereas the average number of roots on control plants was at least 18.0 ± 5.0 and 27.3 ± 5.6 after 6 and 12 days respectively. When maintained in the absence of dexamethasone, the pOp-ipt-S/LhGR plants were indistinguishable from controls in all of the parameters described above, although one of the three S/C plants produced an axillary shoot at the base of the stem in the absence of inducer (Figure 2b,e and Table S1a).
Several plants of lines S/N, S/I1, S/I2, S/I3, and S/C germinated on dexamethasone-free medium were transferred to soil and their development in the greenhouse was monitored over the entire growth cycle to allow time for the cumulative effects of any leaky pOp activation to become apparent. Based on a number of morphological and temporal indicators of development, we were unable to identify any significant differences between the development of pOp-ipt-S/LhGR plants (Table S1b). In contrast, cytokinin-related phenotypes were observed in the inducible ipt lines after the application of dexamethasone to soil-grown plants at various stages of development (see below). To confirm these findings, activator lines derived by segregation from S/N, S/I3, and S/C (see Experimental procedures) were crossed with pOp-ipt-660 (Lexa et al., 2002), that elicits stronger induced phenotypes than pOp-ipt-S. As shown in Figure 2(a), induced phenotypes were indeed stronger with pOp-ipt-660, although uninduced seedlings again developed normally. Furthermore, when pOp-ipt-660/LhGR-N plants were allowed to complete a life cycle under growth room conditions no evidence of leaky ipt expression was obtained (Table S1c).
We concluded that all three LhGR fusions were capable of activating the pOp promoter in a dexamethasone-dependent fashion in tobacco though the frequency was apparently lower than with LhG4. T2 families of all five lines exhibited quantitatively and qualitatively similar phenotypes on dexamethasone containing medium indicating that the inducible pOp activation phenotype is inherited and stable over at least three generations.
Effect of different concentrations of dexamethasone
To investigate the dose–response characteristics of the pOp/LhGR system we chose to work with the LhGR-I activator line S/I1 which was generated with the weaker of the two pOp-ipt reporter lines described above. A double homozygous T2 line, S/I1-1, was identified and used for all subsequent experiments. Seeds were sown on MS media containing 0.625, 1.25, 2.5, 5, 10, 20, 30 μm dexamethasone to which ethanol was added to final concentration of 0.1% and on control medium containing only 0.1% ethanol. Figure 3(a) shows that even 0.625 μm dexamethasone almost completely induced the ipt phenotype.
A similar experiment was conducted on soil-grown plants. Three weeks after transfer to soil, plants were watered with 50 ml of either 5, 10, 20, 40, or 60 μm dexamethasone (all with ethanol at 0.2% v/v) or with a control solution containing only ethanol. This treatment was repeated three times per week for 4 weeks and the resulting phenotypes are shown in Figure 3(b). The plants exhibited all the typical signs of cytokinin overproduction in a concentration-dependent manner. Their height was reduced, although the number of leaves on the main stem was not, indicating that the reduced height was attributable to shortening of the internodes. The plants showed early release of axillary buds from dormancy, delayed flowering, and a reduced number of fruits as flower buds fell off at an early stage. The senescence of the plants was delayed but leaves were wrinkled and chlorotic, with necrotic regions developing mainly at the margins, and the root system was reduced greatly (Figure 3c). In subsequent experiments we found that a single dexamethasone treatment was sufficient to induce the full phenotype (see below).
Different methods of dexamethasone application can result in different phenotypes
To determine whether dexamethasone would induce different phenotypes when applied generally or locally, plants of line S/I1-1 were either watered or painted with a 20 μm dexamethasone solution or the dexamethasone solution was applied to the axillary buds only (see Experimental procedures). The treatment was performed three times per week for 4 weeks. Painting leaves with dexamethasone induced the full range of phenotypes observed with soil watering suggesting that application via the roots and leaves resulted in similar levels and distribution of cytokinins (Figure 3d). In contrast, local application of dexamethasone to axillary buds resulted in a higher incidence of release from dormancy but with little influence on the height, flowering time, or fertility of the plants (Figure 3d). Nevertheless, the root system was still reduced and necrotic patches were observed on some leaves (not shown) suggesting that either dexamethasone or cytokinin was transported to other regions of the plant. S/I1-1 plants treated with the control solution were comparable to SR1 plants and displayed no sign of cytokinin overproduction (Figure 3d).
Testing for inhibitory influences of LhGR, dexamethasone, and ethanol on plant development
The inducing chemical or the chemically responsive transcription factor used in some other inducible expression systems can adversely affect plant function (Andersen et al., 2003; Corlett et al., 1996; Gatz, 1996; Kang et al., 1999; Ouwerkerk et al., 2001). To determine whether dexamethasone itself affects plant development, wild-type tobacco seeds were sown on media containing 5, 10, 15, or 20 μm dexamethasone to which ethanol was added to a final concentration of 0.1%. Control media contained either 0.1% v/v ethanol, 20 μm dexamethasone from a stock dissolved in DMSO (0.1% v/v final concentration), or no supplements. Seedling development was not influenced by even the highest concentrations of dexamethasone used, however 0.1% ethanol markedly inhibited seedling development (Figure 2f) as exemplified by shoot fresh weight of 2-week-old seedlings (Figure 2g). Monitoring seedling development over the first 2 weeks indicated that much of this reduction could be attributed to an approximately 2-day delay in germination (Figure 2h) followed by further delays in the subsequent rate of seedling development (not shown). To establish whether LhGR inhibits seedling development before or after activation with dexamethasone, untransformed seedlings and LhGR-N, -I, and -C activator lines derived by segregation from the five original pOp-ipt-S/LhGR transformants (see Experimental procedures) were sown on media containing either 20 μm dexamethasone and 0.1% v/v ethanol or ethanol alone. As before no effect of dexamethasone could be seen, although ethanol was again inhibitory and on each medium the LhGR activator lines were indistinguishable from the untransformed controls (Figure 2i).
Improved reporters pOp6 and pH-TOP
The data above show that the LhGR system is capable of inducing the pOp-promoter to generate strong cytokinin-overproduction phenotypes when used to express an ipt gene. However, when we tested LhGR-N, -I, or -C with pOp-GUS reporter lines (Moore et al., 1998), we found that induced GUS expression was barely detectable even in the best lines selected from a population of 52 primary transformants. Such low levels of induced activity would severely restrict the useful applications of the pOp/LhGR system, so we attempted to improve the transcriptional efficiency of the system. We have found that the efficiency of the pOp promoter in Arabidopsis can be improved by increasing the number of lac operator binding sites from two to six and spacing them to allow simultaneous occupancy by LhG4 or LhGR (Craft et al., 2005). Therefore, we tested the performance of plasmids carrying the improved operator array (Figure 1b) in conjunction with LhGR in tobacco. Plasmid pOp6-GUS was derived from pOp-GUS (Moore et al., 1998) by exchanging the two operators of pOp with the six-operator array. Plasmid pH-TOP carries two divergent minimal promoters flanking the array of six lac operators. One of these promoters drives expression of GUS while the other carries multiple cloning sites for insertion of genes of interest. In this way, GUS provides a simple screen for functional T-DNA insertions and for the efficiency of dexamethasone-induced gene expression. To test whether GUS activity also provides a quantitative indicator of the expression efficiency of a gene cloned in the multiple cloning site, we used a pH-TOP derivative, pH-Luc, carrying a luciferase reporter in this site (Craft et al., 2005).
Expression characteristics of pOp6-GUS/LhGR and pH-Luc/LhGR lines
To test the effectiveness of the new reporter constructs we transformed them into the activator lines that were generated by segregation from the original pOp-ipt-S/LhGR lines S/N, S/I3 and S/C described above. The majority of T0 plants exhibited dexamethasone-inducible GUS expression in infiltrated leaf tissue, so T1 seeds from 23 selected pOp6-GUS/LhGR lines and 16 pH-Luc/LhGR lines were grown on selective media with or without 20 μm dexamethasone. Strongly inducible GUS expression was detected by histochemical staining in 38 of these lines. After 3 weeks of growth in Petri dishes, seedlings were assayed histochemically for GUS activity. Extractable GUS activity was determined in seedlings of nine selected pOp6-GUS/LhGR-N lines grown for 3 weeks on inducing or control media and induced activities ranged from 2 to 72 pmol 4-MU min−1μg−1 protein (an average of 26 pmol 4-MU min−1μg−1 protein; Figure 4a). In many cases, the GUS activity reached or exceeded levels comparable to those obtained with the CaMV 35S promoter (typically 10–30 pmol 4-MU min−1μg−1 protein; Moore et al., 1998; Weinmann et al., 1994). The background activity (average 0.006 ± 0.006 pmol 4-MU min−1μg−1 protein) was comparable to the controls (average of 0.009 ± 0.001 pmol 4-MU min−1μg−1 protein) in all cases but one (line 15S/N, 0.34 ± 0.01 pmol 4-MU min−1μg−1 protein). This represents an average 4300-fold induction for the eight most inducible lines.
The GUS activity detected in 10 selected pH-Luc/LhGR-N lines (Figure 4b) was significantly lower than in pOp6-GUS/LhGR-N lines (average of four compared with 26pmol 4-MU min−1μg−1 protein) and the average background activity was 10-fold higher than untransformed controls. Consequently, the average induction of GUS activity in the pH-Luc/LhGR-N lines was 20-fold. These seedlings were also used for determining extractable luciferase activity (Figure 4c). The background activity was comparable to control SR1 plants in most cases and the average induction of luciferase activity by dexamethasone in the nine best lines was 500-fold. Surprisingly, plotting induced GUS and luciferase activities for 10 independent pH-Luc/LhGR-N lines (Figure 4d) showed that the correlation was poor. Nevertheless, the best five GUS-expressing lines included four of the five best luciferase expressers.
Time course of dexamethasone-induced gene expression
To establish the rate of pOp6-GUS/LhGR-N activation under optimal conditions seedlings were transferred into liquid medium and induced with 20 μm dexamethasone for various times before being assayed by Northern blot of total RNA. GUS transcripts could not be detected in the absence of dexamethasone but were detectable 1 h after its addition and increased in abundance reaching a steady-state during the subsequent 7 h (Figure 5a,b). When extractable GUS activity was determined, increases were detected 2 h after application of dexamethasone and activity had not reached a steady-state even 34 h later (Figure 5c). In similar experiments with pH-Luc/LhGR-N, both GUS and luciferase activities were induced rapidly after addition of dexamethasone and the proteins accumulated but did not reach their maximum even 36 h after induction (Figure 5d,e).
Dose–response characteristics of dexamethasone-induced gene expression
To establish the dose–response curve for activation of pOp6-GUS/LhGR-N under optimal conditions, dexamethasone was added to seedlings in liquid medium at final concentrations between 0.1 and 20 μm as these concentrations spanned the range of inducing concentrations reported for other dexamethasone-inducible systems. Surprisingly, analysis of total RNA extracted 24 h after induction showed that the lowest concentration tested (0.1 μm) was sufficient to induce GUS transcripts to their maximum abundance (Figure 5f,g). When extractable GUS activity was determined 48 h after application of dexamethasone at concentrations ranging from 0.1 nm to 10 μm maximum activity was reached at about 0.1 μm (Figure 5h and data not shown). Remarkably, 20% of maximum activity was measured even with a dexamethasone concentration of 0.1 nm. Furthermore, the non-induced activity was not significantly different from the control SR1 line alone. Similar experiments with a pH-Luc/LhGR-N line also showed that there was no significant difference in GUS and luciferase activities induced with 0.1 and 10 μm dexamethasone (Figure 5i,j).
Expression characteristics of pOp6-GUS/LhGR plants in soil
We next investigated the induction characteristics of the pOp6-GUS promoter with LhGR-N under greenhouse conditions. Three 8-week-old plants of line 17S/N were watered with 50 ml of 20 μm dexamethasone on two consecutive days and samples were taken from the third oldest leaf at the time of the first dexamethasone treatment and on several subsequent occasions over an 8-week period. Induced GUS activity was detectable within 24 h of induction and peaked after 3 weeks (Figure S1a). Histochemical staining of leaf samples indicated that after 1 week, GUS activity was predominantly confined to the vascular tissues. In contrast, when leaves were painted with a 20 μm solution of dexamethasone, GUS staining was observed in both the vascular tissues and the lamina (data not shown). To determine the distribution of GUS activity throughout the plant following induction via the soil, plants were induced as above and 10 days after the first dexamethasone treatment GUS activity was measured in each leaf. Highest GUS activities were recorded in the fifth oldest leaf with substantially less activity being found in the youngest and oldest leaves (Figure S1b). To establish whether localized induction could be achieved by painting individual leaves with a dexamethasone solution, young growing leaves or older fully expanded leaves were painted on one side of the mid-vein and GUS activity in the painted and non-painted halves was determined 1, 3, and 7 days later. There was apparently a small amount of transport of dexamethasone into the non-painted areas where the appearance of GUS activity paralleled that in the painted areas (Figure S1c,d). When a leaf in the middle of the plant was painted with dexamethasone and GUS activity was measured in this and the two adjacent leaves, there was no induction of GUS activity in the lower leaf over 7 days but there was some weak induction in the upper leaf in one of the three plants tested (Figure S1e). We conclude that the pOp6/LhGR system functions efficiently in soil-grown plants allowing systemic induction of the promoter in the shoot by application of dexamethasone to the soil and localized induction in individual leaves by leaf painting.
Expression characteristics of pOp6 with the ipt reporter gene
The data reported above indicate that the improved promoters in pOp6-GUS and pH-Luc provide high levels of inducible expression using GUS or luciferase reporters. However, we wanted to ensure that this improved efficiency had not been achieved at the expense of increased uninduced expression that might preclude the use of these promoters with deleterious genes. For this reason the ipt coding sequence (Heidekamp et al., 1983) used previously with pOp was used to replace GUS in pOp6-GUS and luciferase in pH-Luc to generate pOp6-ipt and pH-ipt. We then asked whether we could recover phenotypically normal plants that exhibited strong inducible ipt-related phenotypes. The homozygous activator line LhGR-N was retransformed with pOp6-ipt and pH-ipt. Although some primary transformants failed to root, we recovered six pOp6-ipt/LhGR-N and 23 pH-ipt/LhGR-N lines that formed roots and grew normally after transfer to the greenhouse. T1 seeds from these 29 lines were plated with and without dexamethasone to screen for induced and uninduced phenotypes.
Five of the six pOp6-ip/LhGR-N lines and all 23 of the pH-ipt/LhGR-N lines exhibited strong inducible phenotypes that were more severe than those observed with pOp-ipt. In most cases seedlings plated on dexamethasone-containing medium failed to develop beyond the cotyledon stage and seedlings of some lines were killed upon induction (Figure 6a,b). Three-week-old plants from representative pOp6-ipt/LhGR-N and pH-ipt/LhGR-N lines were transferred into soil along with control plants and pOp-ipt lines that carried the same LhGR-N activator locus. Either 10 or 21 days later, these plants were induced by watering once with 50 ml of 20 μm dexamethasone. In each case, within 3 days of induction, necrotic lesions were observed on the leaves of pOp6-ipt/LhGR-N and pH-ipt/LhGR-N plants but not on any of the pOp-ipt/LhGR-N or control plants. Photographs taken about 2 weeks after induction reveal that the pOp6-ipt/LhGR-N and pH-ipt/LhGR-N plants had ceased to develop further and, in the case of the older plants, severe wilting had occurred (Figure 6c,d). The pOp-ipt/LhGR-N lines developed typical cytokinin-related phenotypes as reported above but their response was weaker than that of the pOp6-ipt/LhGR-N and pH-ipt/LhGR-N plants. In contrast, we noted no abnormalities in the growth or appearance of uninduced lines either as seedlings on plates or in the growth room in soil. To verify that non-induced pOp6-ipt/LhGR-N and pH-ipt/LhGR-N plants exhibited no measurable alteration in development we measured a number of developmental parameters in 7- and 9-week-old plants (Table S1c). There was no significant difference between the height, leaf number, flowering time, or fruit set of the pOp6-ipt/LhGR-N or pH-ipt/LhGR-N plants and the control plants in this population.
We conclude that the improved promoter can be used to generate plants that exhibit extreme cytokinin-related phenotypes and that any uninduced expression in these plants is so low that it was unable to alter plant development during the entire growth cycle.
In the last decade several chemically inducible gene expression systems have been reported for plants (reviewed in Gatz and Lenk, 1998; Padidam, 2003; Zuo and Chua, 2000). The utility of any such system is determined in the main by its uninduced expression levels and its fold induction. The ideal system will deliver undetectable expression in the uninduced state but induced levels should be comparable to strong constitutive promoters such as CaMV 35S. In practice this requires the dynamic range to be 1000-fold or higher (Böhner et al., 1999; Padidam et al., 2003; Roslan et al., 2001; Zuo et al., 2000). It is also important that the inducing chemical can be applied systemically and that the concentrations required for full induction are substantially lower than those that interfere with plant physiology or development. Furthermore, the chemically responsive transcription factor should not cause any undesirable effects in the induced or uninduced state and the system should operate in several plant species including the common experimental species Arabidopsis and tobacco. Each of the systems reported to date satisfies several of these criteria but only one system appears to satisfy all (Padidam et al., 2003), although the system has yet to be tested with a physiologically active transgene. This paper together with an accompanying report (Craft et al., 2005) describes a new system, derived from the previously published pOp/LhG4 system (Moore et al., 1998) that satisfies the criteria outlined above. In combination with improved pOp promoters this modified transcription factor, LhGR, facilitates highly inducible and tightly regulated transgene expression in tobacco in response to nanomolar concentrations of dexamethasone. The system has been tested with the reporters GUS and luciferase as well as the physiologically active ipt sequence from A. tumefaciens (Heidekamp et al., 1983).
In the native GR, the LBD is situated at the caroboxy terminus of the protein, but we found that it worked effectively at all three positions in LhG4. This contrasted with our experience of the identical constructs in Arabidopsis where LhGR-N provided the best expression characteristics while LhGR-I proved ineffective and LhGR-C was leaky (Craft et al., 2005). Work is underway to investigate why these constructs behave differently in these two species. LhGR-N proved to be stringently repressed and highly inducible in both species.
When ipt was introduced into an LhGR-N activator line under control of the improved promoters, we found that in most cases induced ipt expression was so high that seedlings failed even to expand their cotyledons and eventually died. Plants treated with dexamethasone via the soil developed necrotic lesions and severely wilted within a week of treatment and also eventually died. Despite the severity of the induced ipt phenotypes we did not observe or measure any difference between the development or appearance of uninduced plants and controls. The ipt gene has been chosen for this type of analysis in part because plants are sensitive to low levels of expression that can be difficult to measure with conventional reporters such as GUS (Schmülling, 2002). This gene has been effectively regulated by the GVG and pOp/LhGR systems in Arabidopsis (Åstot et al., 2000; Craft et al., 2005) and by GVG in tobacco and lettuce (Kunkel et al., 1999) but has been used most extensively in tobacco in conjunction with inducible systems based on the Tn10-encoded tetracycline repressor (Gatz et al., 1992). The tetracycline-derepression system reported by Faiss et al. (1997) generated weaker phenotypes than those reported here, yet even the uninduced plants exhibited marked stunting indicating that the system was leaky (Böhner and Gatz, 2001). This problem was overcome by adding the GR domain and a transcriptional activation domain to the tet-repressor to generate a dexamethasone-inducible factor (Böhner et al., 1999). However in this system the ipt phenotypes induced by application of dexamethasone were even weaker than those obtained with the leaky tetracycline-derepression system (Böhner and Gatz, 2001). In comparison, the LhGR system appears to offer advantages: the original pOp promoter delivers phenotypes that appear to be of similar severity to those reported with the tet-repressor-based systems while the newer promoters exceed these levels, yet neither produces detectable uninduced activity in conjunction with the ipt sequence.
When dexamethasone was applied via the growth medium to plants in sterile culture or in soil, GUS activities reached 70 pmol 4-MU min−1μg−1 protein. The average induced GUS activity measured in the population of plants generated with pOp6-GUS/LhGR-N was 26 pmol 4-MU min−1μg−1 protein while the average uninduced activity was 0.006 pmol 4-MU min−1μg−1 protein. This corresponds to an average fold induction of 4300 and compares favourably with the average fold induction of other inducible systems (Böhner et al., 1999; Padidam et al., 2003; Roslan et al., 2001; Zuo et al., 2000). In some reports the fold inductions measured for individual plants were reported to exceed these values (up to 63 000-fold in one case; Böhner and Gatz, 2001). Similarly, individual plants within our population also appeared to exhibit substantially more than 4300-fold induction (up to 12 000-fold). However, such estimates are of questionable significance as they are highly sensitive to the estimate of uninduced activity and this is very difficult to make for a non-leaky system. Of more significance is the demonstration that the average induced activity in the pOp6/LhGR-N plants exceeds that previously reported for the pOp/LhG4 system in tobacco (Moore et al., 1998) and is similar to that reported for 35S-GUS (Moore et al., 1998; Weinmann et al., 1994). These high levels of induced GUS expression are consistent with the extreme severity of the ipt phenotypes we observed. The best lines exceeded the 35S-GUS levels several fold. In this respect the LhGR system is similar to the ecdysone-receptor-based systems (Martinez et al., 1999; Padidam et al., 2003). The XVE oestrogen-based system was also reported to direct expression levels that exceed those of CaMV 35S but this observation is difficult to compare with other studies because the XVE protein was expressed from the synthetic G10-90 promoter that is itself several fold more active than the CaMV 35S promoter (Zuo et al., 2000).
The kinetics of transcript accumulation in the pOp6/LhGR-N system are similar to those reported for other systems (Aoyama and Chua, 1997; Böhner et al., 1999; Roslan et al., 2001; Salter et al., 1998; Zuo et al., 2000). It is also potentially of value to be able to modulate gene expression by varying the dose of inducer and this requires a shallow dose–response curve. In the pOp6/LhGR-N system we found that induced GUS activity in seedlings reached its maximum with 0.1 μm dexamethasone and was induced to 50% by 1 nm dexamethasone. The pOp6/LhGR-N system in tobacco responds more sensitively than the dexamethasone-inducible GVG system which requires 10 μm for full induction in tobacco seedlings and 3 μm for induction to 50% (Aoyama and Chua, 1997). In fact, pOp6/LhGR-N in tobacco exhibits the most sensitive induction kinetics of any published system.
In agreement with other workers (Aoyama and Chua, 1997; Böhner et al., 1999), we found that seedlings developed normally at up to 30 μm dexamethasone which is 300-fold higher than the concentration required to saturate the response with pOp6/LhGR-N. Tobacco seedling development was sensitive to ethanol used to dissolve dexamethasone, so the use of alternative solvents or water-soluble dexamethasone formulations is advisable. Significantly, the concentration of ethanol that was sufficient to impair seedling development in vitro (0.1% v/v) was similar to that reported to be necessary for full induction of the AlcR-based ethanol-inducible system in seedlings (Salter et al., 1998). We did not observe any alterations in the development or appearance of tobacco plants carrying the LhGR fusions either with or without induction suggesting that, like LhG4, LhGR is not intrinsically inhibitory. This is also the case with the dexamethasone-inducible GVG and TGV systems in tobacco, although the GVG system induces growth abnormalities following induction by dexamethasone in Arabidopsis (Kang et al., 1999) and other species (Andersen et al., 2003; Ouwerkerk et al., 2001).
Dexamethasone can be applied to plant tissues at many stages using a variety of application procedures. We found that seedlings germinated on or transferred onto dexamethasone containing medium in vitro exhibited activation of pOp6/LhGR throughout the seedling, although transpiration is likely to be minimal under these conditions. Dexamethasone delivered through the roots in older plants in soil resulted in activation of pOp6/LhGR throughout the plant but principally in the vascular tissues whereas painting with dexamethasone solution resulted in induction throughout the leaf. Transport of dexamethasone from leaf to leaf is minimal allowing localized induction of individual leaves. In this respect dexamethasone is perhaps more versatile than ethanol, which is volatile and efficiently transported even between plants (Roslan et al., 2001). Indeed with pOp-ipt/LhGR we found that localized application of dexamethasone to leaf axils produced a different growth phenotype than widespread application via the soil or leaf painting. Similar observations have been made with the TGV system and with tetracycline (Böhner and Gatz, 2001; Faiss et al., 1997). The distribution of induced expression in tissues of soil-grown plants carrying the recent ecdysone-receptor-based system (Padidam et al., 2003) has not been reported but foliar application is apparently inefficient for this and another ecdysone-receptor-based system (Padidam, 2003). Finally there is no evidence of endogenous ligands for GR in plants which makes GR-based systems more attractive and versatile than ethanol and oestrogen-based systems for which endogenous ligands can exist in certain species or growth conditions (Salter et al., 1998; Zuo and Chua, 2000).
The strategy of using an increased number of operators when designing inducible systems has been successfully used several times (Aoyama and Chua, 1997; Böhner et al., 1999; Martinez et al., 1999; Padidam et al., 2003; Weinmann et al., 1994; Zuo et al., 2000). Although the pOp promoter was suitable for inducing cytokinin-related phenotypes in tobacco, LhGR directed very little dexamethasone-inducible GUS expression from pOp-GUS lines that were competent to respond efficiently to LhG4 (data not shown). The high sensitivity of the pOp6 promoter to LhGR-N may account for its efficient induction using a variety of dexamethasone application procedures with seedlings and mature plants. The proximity of the regulatory GR domain to the DNA-binding helices of LhGR may contribute to the efficient repression of transcriptional activity in the absence of inducer, providing tight regulation of even the pOp6 promoter in conjunction with the ipt reporter.
In an attempt to provide an easily scored marker of functional T-DNA integrations and of induction efficiency, we tested pH-TOP which drives transcription of a GUS reporter and a second gene-of-interest from the same pOp6 operator array (Craft et al., 2005). Transcription from minimal promoters flanking an enhancer element are correlated in individual transformants, although absolute expression levels can vary substantially between individuals (Ott et al., 1990). Surprisingly we found that GUS and luciferase activities driven from the same operator array in pH-Luc/LhGR-N transformants were poorly correlated. We have made similar observations with pH-TOP derivatives in Arabidopsis (Craft et al., 2005). This suggests that the principal determinants of position-dependent expression efficiency in pOp6 reporter constructs probably lie in regions outside the lac operator array. Modifying the minimal promoter sequence or length can significantly alter the efficiency of inducible gene expression (Martinez et al., 1999; Weinmann et al., 1994), so the two minimal promoters might be a source of variability in expression level between genes cloned in the pH-TOP vector. Alternatively, sequence-specific PTGS thresholds may independently affect the relative levels of each marker as described recently for Arabidopsis (Schubert et al., 2004). Nevertheless, pH-TOP may still be of some practical value because pre-screening pH-TOP transformants for high inducible GUS activities is likely to identify most of the lines that exhibit efficient inducible expression of the gene-of-interest. Furthermore, pH-TOP-driven expression in tobacco is mitotically stable, so it can be used to monitor the efficiency with which the reporter locus has been activated in any particular experiment.
In conclusion, we show that the pOp6/LhGR system provides a highly sensitive, efficient, and tightly regulated chemically inducible transgene expression system for tobacco. We have made similar observations with these constructs in Arabidopsis suggesting that this system might be applicable to a variety of plant species. Furthermore, the system described here is compatible with the previously reported LhG4 system for which a large number of tissue-specific activator constructs are available. Genes of interest introduced into tobacco under control of the pOp6 promoter could thus be activated in precise temporal or spatial domains after crossing with the appropriate LhGR or LhG4 activator line.
Standard procedures were used for DNA cloning and analysis (Ausubel et al., 1993). Kanamycin-resistant pBINPLUS (van Engelen et al., 1995) and pGreen (Hellens et al., 2000) and hygromycin-resistant pVKH18 (a derivative of pOp-GUS in which the pOp-GUS cassette was replaced by a polylinker; Moore et al., 1998) binary vectors for Agrobacterium-mediated plant transformation were used. pBIN-LhGR-N, -I, -C and pBIN-LhG4 have been described (Craft et al., 2005). The DNA sequence encoding the ipt gene was amplified from a cloned sequence kindly provided by Dr Csaba Koncz, Max-Planck-Institut fuer Zuechtungsforschung, Cologne, Germany, using forward 5′-AAAAAAGTCGACATGGACCTGCATCTAATTTTCGGTCCAAC-3′ and reverse 5′-AAAAACCCGGGC-TAATACATTCCGAACGGATGACCTTCGAATC-3′ primers and cloned as an SalI/SmaI fragment into multiple cloning site of pU-BOP (see below) downstream of a modified pOp promoter. An SacI/HindIII fragment containing the two lac operators, the minimal CaMV 35S promoter, ipt coding sequence and polyadenylation signal was isolated from this clone and inserted into pVKH18 vector to generate pOpBK-ipt. Second, a fragment containing the pOp6 promoter (Craft et al., 2005) was isolated as an SacI/SalI fragment and cloned into the SacI and SalI sites of pOpBK-ipt thereby replacing the pOp promoter with pOp6. pH-ipt was prepared by cloning the DNA sequence encoding the ipt coding sequence as a SalI/SmaI fragment into the SalI and SmaI sites of pH-TOP (Craft et al., 2005).
Construction of pU-BOP pOpBK a modified pOp promoter from which several restriction sites had been removed (Baroux et al., 2005), was isolated from pOpBK-GUS as an Ecl136II- HindIII fragment and inserted into the corresponding sites of pBluescript SKII (Stratagene, La Jolla, CA, USA). The HindIII site was digested and converted to an NheI site by treatment with Klenow fragment of DNA polymerase I and religation. An Ecl136IIKpnI fragment was then isolated and inserted into the corresponding sites of the pK18 polylinker followed by the polyA signal from pRT101 (Töpfer et al., 1987) isolated and inserted using BamHI and HindIII to generate pK-BOP. pU-BOP was constructed by transferring the pOpBK-polylinker-polyA cassette from pK-BOP into pUC19 using Ecl136II and HindIII.
Plant transformation and maintenance
The reporter lines pOp-ipt-S and pOp-ipt-660 were generated in Nicotiana tabacum cv. SR1 Petit Havana as described previously (Lexa et al., 2002). The binary vectors were transformed into A. tumefaciens strain GV3101::pMP90 (Koncz and Schell, 1986). Leaf disc transformation of activator lines and reporter lines was used to generate primary transformants that were selected either on kanamycin or hygromycin at 50 and 20 μg ml−1 respectively. In tissue culture, plants were grown on MS medium (Murashige and Skoog, 1962) supplemented with 3% (w/v) sucrose and 0.8% agar. Plants were grown with 16 h light/8 h dark at a daytime temperature of 25°C.
Selection of activator lines LhGR-N, -I and -C
LhGR-N, -I and -C activator lines were selected from T1 populations of lines S/N, S/I3 and S/C generated with the weak reporter line pOp-ipt-S. These lines were heterozygous for the pOp-ipt T-DNA at a single Mendelian locus. Segregating T1 progeny were selected on kanamycin (to select for LhGR) and 20 μm dexamethasone. Plants that exhibited no cytokinin-related phenotypes were transferred to the greenhouse and their pollen was used for pollination of reporter line pOp-ipt-660. The activator plants were also used for re-transformation with pOp6-GUS, pOp6-ipt and pH-Luc, and pH-ipt all of which encode hygromycin resistance in tobacco.
Methods of dexamethasone application
Dexamethasone (Sigma, Dorset, UK) was dissolved in ethanol and kept as 20 mm stock at −20°C. Unless otherwise stated 20 μm dexamethasone was used. Dexamethasone was added to MS media to achieve induction during seed germination or after seedling transfer in sterile conditions. Induction in liquid media was performed in conical flasks (250 ml) containing 100 ml MS media. Typically 2–3-week-old seedlings (20–40 individuals) were transferred into the flasks and kept on an orbital shaker (approximately 80 rpm) in a plant growth chamber. Two days later dexamethasone at the appropriate concentration was added. The seedlings were harvested after the required period, blotted dry, and freezed in liquid nitrogen. Tobacco plants grown in soil were watered with 50 ml of a solution containing dexamethasone. The treatment was performed either once or repeatedly as stated; in the meantime the plants were watered as required without any supplements. Tobacco leaves were painted using a paintbrush with a dexamethasone solution containing 0.02% Silwet L-77. Approximately 5–10 ml of the solution was used for a single treatment of all leaves, depending on the size of the plant. Approximately 1 ml was used for a single treatment of a single leaf. Approximately 0.5 ml of the solution was applied to the leaf axils temporarily covered with parafilm to prevent the solution from running down the stem.
Analysis of β-glucuronidase and luciferase reporter activity
Histochemical GUS staining and fluorometric GUS assay was performed according to Jefferson (1987). Extraction of luciferase and assays for relative light units (lu) were performed as described by Luehrsen and Walbot (1993). We found that CCLR extraction buffer used in this assay inhibited GUS activity by five- to 15-fold, so separate extracts were made according to Jefferson (1987) for comparison with GUS activities in other samples. The protein content of all extracts was determined spectrophotometrically using Bio-Rad Protein Assay Reagent (Bio-Rad laboratories, Hemel Hempsted, UK).
RNA gel blot analysis
Total RNA was extracted from plant tissue, dissolved in formamide, and 10 μg of each RNA sample was electrophoresed, blotted onto Hybond N+ membranes (Amersham, Cardiff, UK), hybridized with 32P-labelled probes, and analysed by autoradiography using Kodak X-AR film (Rochester, NY, USA), all as described in Moore et al. (1997). To detect the LhGR transcripts, the LhG4 sequence was isolated from pLhG4 (Moore et al., 1998) as an XbaI fragment. To detect the GUS transcript, the sequence was isolated from pRT103GUS (Töpfer et al., 1987) as an XhoI/SspI fragment. The film was developed using an AGFA Curix 60 X-ray film developer (AGFA, Herts., UK). In some cases the hybridizing radioactivity was quantified using a Bio-Rad Molecular Imager FX and Bio-Rad Multi-AnalystTM/PC Version 1.1 software. Hybridized probes were removed from filters by subjecting them to three washes with boiling strip solution (0.1 × SSC, 0.1% w/v SDS) and were hybridized with a sequence encoding a 571-bp fragment of a tobacco actin gene that had been amplified from genomic DNA using primers Nt-Act-3′a (5′-ATCCAGACACTRTACTTYCTCTC-3′) and Nt-Act-3′b (5′-TCCARACRCTGTAYTTCCTCTC-3′).
We are indebted to Prof. Peter Meyer, University of Leeds, for use of plant cultivation facilities during part of this work and to Rudolf Kalab for financial support. We are grateful to Csaba Koncz and Debora Grosskopf for providing the cloned ipt sequence, to Andrew Millar for providing the pLuc plasmid, to Helen Townley, Judith Craft, and Hazel Betts for the pOp6 promoter and pH-TOP. We thank John Baker for the photographic work. This work was supported by a Royal Society Joint Project Grant to IM and BB; by GA AS CR Grant No. IAA5004001; Ministry of Education of the Czech Republic Grant No. MSM143100008; AS CR Grant No. KSK5052113 to BB; and a Marie Curie PhD fellowship to MS.