The experimental control of gene expression in specific tissues or cells at defined time points is a useful tool for the analysis of gene function. GAL4/VP16-UAS enhancer trap lines can be used to selectively express genes in specific tissues or cells, and an ethanol-inducible system can help to control the time of expression. In this study, the combination of the two methods allowed the successful regulation of gene expression in both time and space. For this purpose, a binary vector, 962-UAS::GUS, was constructed in which the ALCR activator and β-glucuronidase (GUS) reporter gene were placed under the control of upstream activator sequence (UAS) elements and the alcA response element, respectively. Three different GAL4/VP16-UAS enhancer trap lines of Arabidopsis were transformed, resulting in transgenic plants in which GUS activity was detected only on ethanol induction and exclusively in the predicted tissues of the enhancer trap lines. As a library of different enhancer trap lines with distinct green fluorescent protein (GFP) patterns exist, transformation with a similar vector, in which GUS is replaced by another gene, would enable the control of the time and place of transgene expression. We have constructed two vectors for easy cloning of the gene of interest, one with a polylinker site and one that is compatible with the GATEWAY™ vector conversion system. The method can be extended to other species when enhancer trap lines become available.
In current plant research, microarray analysis is often used to detect genes that are differentially expressed between tissues or in response to specific treatments. Once genes are identified and their expression patterns are known, their function is studied by defining the particular phenotypes in knock-out plants or in plants that show an over-expression of the genes of interest. The control of gene expression in both time and space would be an interesting and powerful tool to further refine the strategies for the study of gene function. This would certainly be the case for genes with expression only late in plant development. Constitutive expression from the embryonic stages onwards may render the interpretation of the roles of these genes more difficult. The same applies to genes for which inappropriate expression levels can be deleterious or even lethal.
To date, several inducible systems using ethanol, tetracycline, glucocorticoids and other chemicals have been developed to temporally regulate transgene expression in planta (Zuo and Chua, 2000; Padidam, 2003; Moore et al., 2006). The ethanol-inducible system consists of a tightly controlled ALCR activator and an alcA response element (Figure 1a). The inducer, ethanol, can be applied as ethanol vapour, by spraying the shoot or by watering the roots, and is a non-toxic, widely available and cheap chemical. Because of these particular features, the ALCR/alcA system has been widely used to control the time of gene expression (Caddick et al., 1998; Salter et al., 1998; Roslan et al., 2001; Sweetman et al., 2002; Junker et al., 2003; Laufs et al., 2003; Garoosi et al., 2005; Filichkin et al., 2006). In these studies, however, a constitutive promoter was used, making it impossible to control the exact location of gene expression. Recently, two groups have independently developed a new expression system that combines the advantages of temporal regulation by ethanol and spatial regulation by specific plant promoters. Deveaux et al. (2003) reported ethanol-inducible gene expression in defined subdomains of the shoot apical and floral meristems, and Maizel and Weigel (2004) combined the ethanol switch with a flower-specific promoter. This strategy, using a tissue-specific promoter, has also been combined with other inducible systems (Nguyen et al., 2004; Brand et al., 2006). Indeed, a variety of promoters provides an opportunity to express genes of interest in desired tissues or cell types (Canevascini et al., 1996; Potenza et al., 2004). It must be pointed out, however, that, to date, the number of well-documented plant promoters is rather small, and that many specific cells are not covered by specific promoters. This clearly restricts the use of these systems.
The use of GAL4/VP16-UAS enhancer trap lines represents an attractive alternative to tissue-specific promoters, as there are hundreds of different expression patterns available. The vector is designed such that the expression of the GAL4/VP16 gene is driven by the fortuitous proximity of a promoter/enhancer, caused by the random integration of T-DNA into the plant genome (Figure 1b). The resulting GAL4 proteins bind to a responsive sequence (upstream activator sequence, UAS) that drives the expression of reporter genes, such as endoplasmic reticulum (ER)-targeted green fluorescent protein (GFP), as used in this study. In this case, the specific expression pattern of each trapped enhancer is indicated by green fluorescent cells. To obtain ectopic expression of target genes in specific plant tissues, it is necessary to cross a transgenic plant bearing the UAS gene of interest construct with a selected enhancer trap line, or to transform the enhancer trap lines. The GAL4 protein will activate the UAS sequence, resulting in the simultaneous expression of the reporter gene and the new gene of interest. Stocks of enhancer trap lines have been established for two well-characterized model plants, Arabidopsis thaliana and Oryza sativa (http://www.plantsci.cam.ac.uk/Haseloff/Home.html; Haseloff, 1999; Wu et al., 2003; Yang et al., 2004; Engineer et al. 2005; Johnson et al., 2005; Peng et al., 2005). They offer the possibility to express transgenes in cells or tissues that can be chosen from a still-growing library (Bougourd et al., 2000; Kiegle et al., 2000; Boisnard-Lorig et al., 2001; Sabatini et al., 2003; Johnson et al., 2005; Laplaze et al., 2005; Liang et al., 2006). This attractive system, however, lacks the temporal control of transgene expression.
To overcome all these limitations, we combined the ethanol-inducible system with the GAL4/VP16 enhancer trap lines, and achieved the controlled expression of β-glucuronidase (GUS) in both time and space in Arabidopsis.
Combination of enhancer trap lines and ethanol induction
In order to control transgene expression in the enhancer trap lines temporally by ethanol, the UAS element was cloned in front of the ALCR regulator (Figure 1c). The responder cassette used to screen for the effectiveness of the system, alcA::GUS, was placed in the opposite direction to UAS::ALCR. The modified alcA is composed of an ALCR-binding element and a downstream basal cauliflower mosaic virus (CaMV) 35S promoter. To ease insertion of the gene of interest, a 962-UAS::PL vector, harbouring an SnaBI-PacI-AgeI-EcoRI polylinker (PL) under the control of alcA, was constructed (Figure 1d), as well as a similar plasmid that can be used as a GATEWAY™ destination vector (962-UAS::GW; Figure 1e). Both plasmids can be provided on request and can be used after written permission has been obtained from Syngenta (Berkshire, UK) and VIB (Ghent, Belgium).
Ethanol does not affect the expression patterns of enhancer trap lines
Seeds of three enhancer trap lines, N9195, N9093 and N9113, were germinated in soil and treated with 1% ethanol after 5 days. Twenty-four hours later, the GFP pattern of the plantlets was examined using a confocal microscope. The patterns of control and ethanol-treated plants were identical, pointing to the absence of an ethanol effect on the enhancer trap lines (data not shown).
Ethanol-inducible expression of GUS in enhancer trap lines
Three enhancer trap lines with distinct GFP expression patterns, N9195, N9093 and N9113, were transformed with the binary vector 962-UAS::GUS by floral dip. Transformants were selected on the appropriate antibiotics and confirmed by polymerase chain reaction (PCR) of the GUS gene (results not shown). Seedlings from the T2 generation were grown in soil and assayed for GUS activity after 5 days. In the absence of ethanol, no GUS staining was detected in any line (data not shown). However, a clear GUS activity could be detected after 24 h of induction with 1% ethanol. In addition, the cells or tissues that were GUS-positive were identical to those in which GFP was detected. This confirms the temporal and spatial specificity of the system. In Figure 2, details of the GFP expression patterns and GUS staining of the transformants grown in soil are shown on confocal and digital camera photographs, respectively. The red colour in the photographs is caused by chlorophyll autofluorescence or propidium iodide staining of the cell walls.
Thirteen independent transformants of line N9113 were analysed. GFP was clearly present in all the tissues (Figure 2a); the same pattern was found after staining for GUS activity (Figure 2b). Intriguingly, two of the studied transformants from this transgenic line were completely silenced from the T1 generation onwards, as judged by the absence of GFP. All the other transformants expressed GUS in GFP-positive tissues, faithfully mimicking the original enhancer trap line. Suppression of gene silencing by treatment of the seedlings with 30 µm 5-aza-cytidine, known to work in silenced tobacco (Gälweiler et al., 2000), failed, but 30 µm 5-aza-2′-deoxycytidine (Engineer et al., 2005) recovered gene expression in silenced Arabidopsis enhancer trap lines. GFP was fully restored in both silenced lines at the correct locations (data not shown).
Nine different transformants of line N9093 were subjected to GFP and GUS analysis. In all transformants, GFP was expressed especially in the distal part of the root and at the transition between root and hypocotyl (Figure 2c); a similar pattern was found for GUS activity (Figure 2d). In the same line, the stomata were brightly fluorescent (Figure 2e), whereas GUS signals were only present in some stomata of plants grown in soil and induced with ethanol on the roots (Figure 2f). Plants that were germinated on Murashige and Skoog (MS) medium showed GUS expression in all stomata, even in the absence of ethanol (Figure 2g).
Four independent transformants of line N9195 were analysed. The T1 and T2 generations of line N9195 behaved differently. GFP was present in the leaves and some specific tissues of the root for T1 seedlings (Figure 2h,j), whereas, in T2 seedlings, a very faint GFP signal was confined to the tip of cotyledons (Figure 2l, see circle). The change in expression patterns could be ascribed to gene silencing, as treatment of the T2 seedlings with 30 µm 5-aza-2′-deoxycytidine restored the GFP and GUS expression. Again, 30 µm 5-aza-cytidine failed to do so (data not shown). It should be noted that GUS activity was consistent with the GFP signals for both T1 and T2 plantlets (Figure 2i,k and see circle in Figure 2m), whether they were silenced or not.
All the transformants from lines N9113, N9093 and N9195 were grown on MS medium in the absence of ethanol, and were subsequently histochemically stained for GUS activity. Leaky expression was observed in all the transformants, except in the silenced ones (Figure 2g and data not shown).
The GUS gene was used as the transgene in the new system, and it was demonstrated that, on ethanol induction, its expression was confined to the cells and tissues that were specifically chosen from the library. A leaky expression of the transgene was noticed when plantlets were grown on MS medium, as reported by Roslan et al. (2001). These authors suggested that Arabidopsis plants grown on agar-containing medium, in which oxygen diffusion is limited, change to an anaerobic respiration, which probably results in the formation of endogenous ethanol. This leaky expression was absent when plants were grown on soil. A second observation was that gene silencing can occur from the T1 or T2 generation onwards. It has been reported that silencing of the GAL4-UAS system can take place in transgenic plants at the level of the GAL4-binding sites (Gälweiler et al., 2000; Engineer et al., 2005) and/or at GAL4/VP16 transgene expression (Engineer et al., 2005). As reported previously (Engineer et al., 2005), treatment with 5-aza-2′-deoxycytidine suppressed gene silencing in the silenced transformants, whereas application of 5-aza-cytidine failed to rescue this silencing. Nevertheless, this substance was shown to restore GAL4-mediated GUS expression in tobacco (Gälweiler et al., 2000). The exact mechanism underlying this silencing/rescuing phenomenon is still not clear. As GFP is concomitantly expressed with the transgene, this can serve as a control. Gene silencing can be easily detected as the absence of GFP. The transformed plants that display this aberrant behaviour can thus easily be discarded, or treated with 5-aza-2′-deoxycytidine.
To date, more than 10 reports have successfully employed the Haseloff GAL4/VP16-UAS enhancer trap lines for the mis-expression of genes (for reviews, see, for example, Moore et al., 2006). It is somewhat surprising that silencing of gene expression under UAS control has only been described in one study (Boisnard-Lorig et al., 2001). It is possible that some transgenic lines from the Haseloff collection are not prone to the methylation of GAL4-binding sites. Although the possibility of gene silencing in some cases may affect our strategy to some extent, our study undoubtedly indicates that the combination of the ethanol switch and enhancer trap lines should be regarded as a very attractive complement to existing systems that use the ethanol switch in combination with cell or tissue type-specific promoters (Deveaux et al., 2003; Maizel and Weigel, 2004).
In conclusion, this novel approach describes a system that enables the temporal control of transgene expression in any desired cell. The system could be extended to any species, given the existence and presence of enhancer trap lines, and will serve as a handy tool to study gene function and role.
The UAS promoter was amplified from pET15 with the primers pUAS1 (5′-AGCGGCCGCCAAGCTTGCATGCCTGCA-3′) and pUAS2 (5′-AGCGGCCGCGTGTCCTCTCCAAATGAAATGAA-3′). Purified PCR products were digested with NotI, extracted from the gel, and finally cloned into NotI-digested pLP962 to result in the plant transformation vector 962-UAS::GUS. This vector was introduced into the Agrobacterium tumefaciens strain GV3101 by the freeze–thaw method.
For the construction of the 962-UAS::PL plasmid, the primers pAlc1 (5′-AGAGCTCGACGGTATCGATAAGCTTGGCC-3′) and pAlc2 (5′-AAGCTTGAATTCACCGGTTAATTAATACGTACTGCAGGTCGTCCTCTCCAAATGA-3′) were used to introduce a PL composed of four unique enzyme sites (italic), downstream of the alcA fragment. PCR products were digested with EcoRI and HindIII, and a 326-bp fragment was purified from the gel. 962-UAS::GUS was cut with EcoRI and partially digested with HindIII; a larger fragment with a size of about 9.5 kb was extracted from the gel. The two fragments were ligated to form the 962-UAS::PL vector.
The GATEWAY™-compatible 962-UAS::GW vector was created by ligation of the 1711-bp fragment of EcoRV-digested pB2GW7 (Karimi et al., 2002) with SnaBI-digested 962-UAS::PL, which was dephosphorylated before the ligation reaction.
Plant growth and transformation
The enhancer trap lines were collected from NASC and grown in soil or on MS medium [4.4 g/L Murashige and Skoog medium including vitamins, 10 g/L sucrose at pH 5.7 solidified with 6 g/L Gelrite (Duchefa, the Netherlands)] in a growth chamber using a 16-h light/8-h dark cycle at 22 °C.
Agrobacterium cells harbouring 962-UAS::GUS were used to transform three different homozygous enhancer trap lines by floral dip (Clough and Bent, 1998). Transgenic plants were selected on MS medium supplemented with 20 mg/L hygromycin B.
To induce gene expression, plantlets were watered with a 1% ethanol solution. After 24 h, plants were observed under a confocal microscope or assayed for GUS activity.
Histochemical staining for GUS
The GUS assay was carried out as described previously (Jefferson et al., 1987) on 5-day-old whole seedlings subjected to 24-h induction with 1% ethanol. In a control experiment, plants that did not receive ethanol treatment were assayed. Plantlets were stained for 6 h at 37 °C, followed by a 70% ethanol treatment to extract chlorophyll.
The GFP pattern of Arabidopsis seedlings was visualized with a Nikon C1 confocal microscope (Nikon, Brussels, Belgium). Cell walls were stained by treating the seedlings with propidium iodide for 5 min, followed by three rinses with distilled water.
Photographs of the GUS-assayed plantlets were taken using a Zeiss Axioskope (Carl Zeiss, Jena, Germany) equipped with a Nikon DXM1200 digital camera.
The research was funded by a University of Antwerp-grant (UA-BOF), a Research Grant of the Research Foundation – Flanders (FWO) (grant G.0101.04) and by the Interuniversity Attraction Poles Programme – Belgian State – Belgian Science Policy (IUAP VI/33). B. Van Loock is funded by a PhD grant from the Institute for the Promotion of Innovation through Science and Technology in Flanders (IWT – Vlaanderen). K. Vissenberg is a postdoctoral fellow of FWO. The authors acknowledge Syngenta (Berkshire, UK) for the source of the Alc materials, VIB (Ghent, Belgium) for pB2GW7 and thank FWO for financial support. Dr P. Laufs, Dr J. Haseloff and Dr R.-X. Fang are thanked for the provision of pLP962, pET15 and Agrobacterium tumefaciens strain GV3101, respectively.