A Comprehensive Toolkit for Inducible, Cell Type-Specific Gene Expression in Arabidopsis^,

A set of transgenic lines enables spatiotemporal control of gene expression in Arabidopsis.

Abstract

Understanding the context-specific role of gene function is a key objective of modern biology. To this end, we generated a resource for inducible cell type-specific transactivation in Arabidopsis (Arabidopsis thaliana) based on the well-established combination of the chimeric GR-LhG4 transcription factor and the synthetic pOp promoter. Harnessing the flexibility of the GreenGate cloning system, we produced a comprehensive set of transgenic lines termed GR-LhG4 driver lines targeting most tissues in the Arabidopsis shoot and root with a strong focus on the indeterminate meristems. When we combined these transgenic lines with effectors under the control of the pOp promoter, we observed tight temporal and spatial control of gene expression. In particular, inducible expression in F1 plants obtained from crosses of driver and effector lines allows for rapid assessment of the cell type-specific impact of an effector with high temporal resolution. Thus, our comprehensive and flexible method is suitable for overcoming the limitations of ubiquitous genetic approaches, the outputs of which often are difficult to interpret due to the widespread existence of compensatory mechanisms and the integration of diverging effects in different cell types.

The key to the evolutionary success of multicellularity, which arose independently in plants and animals, is the division of labor between highly specialized cell types. This requires the robust specification of cell fate through epigenetic and transcriptional programming, despite the identical genetic makeup of each cell. In plants, cell fate acquisition is based largely on positional information, which depends on cell-to-cell communication and medium- to long-distance morphogenetic signals that cooperate in organ patterning (Efroni, 2018). Conversely, individual genes, pathways, and metabolites can have diverse or even opposing roles depending on the tissue context. A prominent example for the context dependency of a fundamental patterning process is given by the interplay of the auxin and cytokinin phytohormones (Furuta et al., 2014; Greb and Lohmann, 2016; Truskina and Vernoux, 2018). In the shoot apical meristem, harboring the stem cell niche ultimately responsible for most aboveground plant organs, cytokinin signaling is associated with maintaining a pluripotent, undifferentiated state, whereas auxin signaling promotes differentiation. In marked contrast, auxin is required for stem cell maintenance in the root apical meristem (RAM; Pacifici et al., 2015; Weijers and Wagner, 2016). Therefore, the global effects of genetic lesions or of knockins can dilute and mask specific functions and often are difficult to interpret.

Routinely, stable genetic gain- and loss-of-function mutants remain the main pillar of the reductionist approach to biology, and the phenotypes of such mutants are assessed to deduce a function of the mutated locus in the wild type. However, the function of many gene products is context specific; thus, the phenotypes of mutants or transgenic lines can be complex. In addition, mutant organisms can undergo life-long adaptation, impeding the interpretation of their phenotype. Moreover, transgenic and mutational approaches can interfere with plant vitality, precluding an in-depth analysis.

Many of these problems can be overcome by inducible, cell type-specific expression mediated by two-component transcription activation systems (Moore et al., 2006). An expression cassette is constructed using a heterologous or synthetic promoter and, hence, is silent unless a cognate transcription factor is present. An efficient approach is to generate transgenic plants called driver lines that express the transcription factor in a spatially and temporally controlled manner and a responder line carrying the effector construct. After crossing of the two lines, expression can be induced and the phenotypic consequences of the effector can be studied. In the abstract, these expression systems are highly valuable because they ideally enable cell type-specific or stage-specific complementation or knockdown, facilitate time-resolved monitoring of the response to a given cue, can overcome the lethality of constitutive expression, and allow the study cell-autonomous and non-cell-autonomous effects with high temporal and spatial resolution. However, the considerable effort and time requirements for DNA cloning and the generation of stable transgenic plants are a major bottleneck curtailing their use to date. For the same reason and because distinct tissue-specific promoters were not always available in the past, attention is usually given to one tissue or cell type of interest at a time, and unbiased approaches targeting a larger spectrum of individual tissues are rarely followed.

Here, we report on the generation of a comprehensive set of Arabidopsis (Arabidopsis thaliana) driver lines suited for tissue-specific transactivation of an effector cassette in a wide range of cell types and with the possibility to monitor gene activation in space and time by a fluorescent promoter reporter. To ensure rapid, stable induction with minimal adverse effects on plant growth caused by the inducer, our system takes advantage of the widely used LhG4/pOp system (Moore et al., 1998; Craft et al., 2005; Samalova et al., 2005) combined with the ligand-binding domain of the rat glucocorticoid receptor (GR; Picard, 1993; Craft et al., 2005). LhG4 is a chimeric transcription factor consisting of a mutant version of the Escherichia coli lac repressor with high DNA-binding affinity (Lehming et al., 1987) and the transcription activation domain of yeast Gal4p (Moore et al., 1998). N-terminal fusion with the GR ligand-binding domain renders the transcription factor inactive in the cytosol through sequestration by HEAT SHOCK PROTEIN90 in the absence of the inducer. Nuclear import after treatment with the synthetic glucocorticoid dexamethasone (Dex; Picard, 1993) results in the transcriptional activation of expression cassettes that are under the control of the synthetic Op 5′ regulatory region consisting of a cauliflower mosaic virus (CaMV) 35S minimal promoter and two upstream lac operators (Moore et al., 1998; Craft et al., 2005). Combining multiple interspersed repeats of the operator elements (pOp4 and pOp6) and localized expression of LhG4 enables strong overexpression of a target gene in a cell type-specific manner (Craft et al., 2005).

Our work builds on these seminal studies by creating 19 well-characterized and stable driver lines targeting most cell types in Arabidopsis with a focus on the three main meristems of the plant, the RAM, the shoot apical meristem (SAM), and the cambium. Of note, for several cell types such as the pith in the inflorescence stem or the xylem pole pericycle cells in the root, inducible expression systems are not available so far. The driver lines were generated employing the fast and flexible GreenGate cloning system (Lampropoulos et al., 2013) but are compatible with any vector/transgenic line in which the expression of an effector is under the control of derivatives of the pOp promoter element (Moore et al., 1998). An important feature of our driver lines is the presence of a fluorescent reporter amenable to live imaging, which allows monitoring the spatiotemporal dynamics of gene induction and may serve as a readout for any effect on the respective tissue identity. Similarly, it allows us to assess whether the expression of the effector has an impact on the transcriptional circuitries targeting the promoter it is expressed from. The material described here allows testing the effect of genetic perturbations in a broad repertoire of individual tissues on a distinct developmental or physiological process. As transactivation efficiently occurs in the presence of the inducer in F1 plants derived from a cross between a driver and an effector line, the effect of a given expression cassette can be assessed relatively quickly in a wide range of cell types, demonstrating the usefulness of this resource for the broader research community.

RESULTS

Design of Driver Lines with Cell Type-Specific Expression of GR-LhG4

To generate a comprehensive set of driver lines expressing the chimeric GR-LhG4 transcription factor under the control of cell type-specific promoters, we made use of the Golden Gate-type GreenGate cloning system, which enables quick, modular, and scarless assembly of large constructs (Engler et al., 2008; Lampropoulos et al., 2013). Our design included, on the same T-DNA, the coding sequence for an mTurquoise2 fluorescent reporter (Goedhart et al., 2012) targeted to the endoplasmic reticulum (ER) through translational fusion with an N-terminal signal peptide from sweet potato (Ipomoea batatas) Sporamin A (SP; Lampropoulos et al., 2013) and the ER retention motif His-Asp-Glu-Leu (HDEL) under the control of pOp6 and a minimal cauliflower mosaic virus 35S promoter (pOp6:SP-mTurquoise2-HDEL; Fig. 1). In our setup, the GR-LhG4 transcription factor is expressed under the control of a tissue- or cell type-specific promoter. Consequently, GR-LhG4 activates the expression of the mTurquoise2 reporter and any other effector downstream of a pOp promoter after Dex treatment specifically in those tissues (Fig. 1). We anticipate that the most utility can be obtained from this system if lines harboring effector cassettes are crossed with driver lines and analyses are performed with F1 plants. However, other modes such as direct transformation of multiple driver lines or introgression into different (mutant) backgrounds also are conceivable. Notably, even though the mTurquoise2 reporter is expressed from the same T-DNA as GR-LhG4, there is no mechanistic difference from the activation of an effector in trans (Fig. 1).

Figure 1.

Overview of the Dex-inducible GR-LhG4/pOp system. In driver lines, expression of the synthetic transcription factor LhG4 is controlled by a tissue-specific promoter (pTS), whereas translational fusion with the ligand-binding domain of rat GR prevents nuclear translocation in the absence of the inducer (Dex). After crossing with an effector line harboring a transcriptional cassette under the control of a pOp element and a TATA box-containing minimal 35S promoter and the addition of Dex, GR-LhG4 drives the expression of the effector as well as the mTurquoise2 reporter encoded by the driver line.

To establish a rather comprehensive set of driver lines, we first selected respective tissue-specific promoters based on literature reports and our own expression data (Table 1). Subsequently, we generated stable transgenic driver lines in the Arabidopsis Columbia-0 (Col-0) background using 19 specific promoters that cover most cell types in the RAM, the SAM, and the cambium. Several of the promoters have been shown previously to work robustly in cell type-specific misexpression approaches (Nakajima et al., 2001; Weijers et al., 2006; Mustroph et al., 2009; Miyashima et al., 2011; Roppolo et al., 2011; Vatén et al., 2011; Naseer et al., 2012; Cruz-Ramírez et al., 2013; Ohashi-Ito et al., 2014; Wang et al., 2014; Chaiwanon and Wang, 2015; Serrano-Mislata et al., 2015; Vragović et al., 2015; Marquès-Bueno et al., 2016; Siligato et al., 2016; Doblas et al., 2017). Next, we generated T3 lines in which the resistance to the selective agent sulfadiazine appeared homozygous after segregating as a single locus in the T2 generation based on resistance or standard addition quantitative real-time PCR (SA-qPCR) analyses (Huang et al., 2013).

Table 1. Overview of promoters utilized in this study.

Promoter	Gene	Expression	Reference
pSCR	SCARECROW	Endodermis, quiescent center (QC) in RAM, starch sheath in stem	Di Laurenzio et al. (1996); Wysocka-Diller et al. (2000)
pATHB-8	HOMEOBOX GENE8	Procambium, xylem precursors and columella in RAM	Baima et al. (1995)
pXPP	XYLEM POLE PERICYCLE	Xylem pole pericycle cells	Andersen et al. (2018)
pAHP6	HISTIDINE PHOSPHOTRANSFER PROTEIN6	Protoxylem precursors, pericycle, organ primordia in the SAM	Mähönen et al. (2006); Besnard et al. (2014)
pPXY	PHLOEM INTERCALATED WITH XYLEM	(Pro)cambium	Fisher and Turner (2007)
pTMO5	TARGET OF MONOPTEROS5	Xylem precursors	Schlereth et al. (2010); De Rybel et al. (2013)
pSMXL5	SMAX1-LIKE5	Phloem (precursors)	Wallner et al. (2017)
pCASP1	CASPARIAN STRIP MEMBRANE DOMAIN PROTEIN1	Endodermis	Roppolo et al. (2011)
pVND7	VASCULAR RELATED NAC-DOMAIN PROTEIN7	Protoxylem (differentiating) in root, vessels in stem	Kubo et al. (2005)
pAPL	ALTERED PHLOEM DEVELOPMENT	Phloem (differentiating)	Bonke et al. (2003)
pNST3	NAC SECONDARY WALL THICKENING PROMOTING3	Fibers	Mitsuda et al. (2007)
pWOX4	WUSCHEL RELATED HOMEOBOX4	(Pro)cambium	Hirakawa et al. (2010)
pLTP1	LIPID TRANSFER PROTEIN1	Epidermis in stem	Thoma et al. (1994)
pAT2G3830		Pith	Valério et al. (2004)
pML1	MERISTEM LAYER1	L1 layer, epidermis	Lu et al. (1996)
pCLV3	CLAVATA3	SAM stem cells	Fletcher et al. (1999)
pREV	REVOLUTA	SAM central zone	Otsuga et al. (2001)
pUFO	UNUSUAL FLOWER ORGANS	SAM peripheral zone	Levin and Meyerowitz (1995)
pCUC2	CUP-SHAPED COTYLEDON2	Boundaries in SAM and leaf	Aida et al. (1997)

Validation of the Specificity of Driver Lines

To confirm the expected expression patterns in the root, driver lines were germinated on medium containing 30 µm Dex or DMSO and analyzed with confocal laser scanning microscopy (CLSM) 5 d after germination (DAG). In each case, we recorded mTurquoise2-derived fluorescence in longitudinal optical sections of the root meristem (Fig. 2; Supplemental Fig. S1) and, where appropriate, in cross sections through the meristem or the differentiation zone (Supplemental Fig. S2). To visualize expression in the shoot, lines were grown on soil in long-day conditions and the aerial part of plants with 15-cm-tall inflorescence stems were dipped either in tap water containing 10 µm Dex (Fig. 3) or only the solvent DMSO (Supplemental Fig. S3). After 24 h, free-hand sections of the stem were stained with PI to highlight xylem elements and analyzed by confocal microscopy. To analyze expression in the SAM, inflorescence meristems of 15-cm-tall plants were treated with Dex 48 h before being dissected and imaged with CLSM, again using PI as a cell wall counterstain (Fig. 4). Reporter gene activities were consistent with the expected patterns and strictly dependent on the presence of Dex (Supplemental Figs. S1, S3, and S4). In addition, the complete absence of reporter activity in tissues adjacent to cells in which activity was expected suggested that the chimeric GR-LhG4 protein does not move between cells. We did not observe any negative effect of Dex treatment on plant growth (Supplemental Fig. S5).

Figure 2.

Analysis of induced driver lines in seedling roots. A, Schematic representation of root tissue layers. B to I, Induced driver line roots displaying fluorescence from propidium iodide (PI)-stained cell walls and the mTurquoise2 reporter (Fig. 1; Table 1). The indicated promoters mediate expression in the differentiating endodermis (B; pCASPARIAN STRIP MEMBRANE DOMAIN PROTEIN1 [pCASP1]), phloem precursor cells and adjacent pericycle cells (C; pHISTIDINE PHOSPHOTRANSFER PROTEIN6 [pAHP6]), xylem precursor cells (D; pTARGET OF MONOPTEROS5 [pTMO5]), xylem pole pericycle cells (E; pXYLEM POLE PERICYCLE [pXPP]), stele initials, cortex/endodermis initial (CEI), and columella initials (F; pHOMEOBOX GENE8 [pATHB-8]), endodermis, CEI, and QC ([G]; pSCARECROW [pSCR]), stele initials, phloem, and procambial cells (H; pSMAX1-LIKE5 [pSMXL5]), and procambial cells (I; pPHLOEM INTERCALATED WITH XYLEM [pPXY]). PI fluorescence is false colored in magenta, and mTurquoise2 fluorescence is false colored in green. Bars = 50 μm.

Figure 3.

Analysis of induced driver lines in the stem. A, Schematic representation of inflorescence stem tissue layers. B to I, Induced driver line stems displaying fluorescence from PI-stained cell walls and the mTurquoise2 reporter (Fig. 1; Table 1). The promoters mediate expression in differentiated phloem (B; pALTERED PHLOEM DEVELOPMENT [pAPL]), xylem fibers and interfascicular fibers (C; pNAC SECONDARY WALL THICKENING PROMOTING3 [pNST3]), starch sheath (D; pSCR), cambium (E; pWUSCHEL RELATED HOMEOBOX4 [pWOX4]), xylem vessels (F; pVASCULAR RELATED NAC DOMAIN PROTEIN7 [pVND7]), epidermal cells (G; pLIPID TRANSFER PROTEIN1 [pLTP1]), the incipient phloem (H; pSMXL5), and pith (I; pAT2G38380). PI fluorescence is false colored in magenta, and mTurquoise2 fluorescence is false colored in green. Bars = 50 μm.

Figure 4.

Analysis of induced driver lines in the SAM. A, Schematic representation of cell identity domains in the SAM. B to G, Induced driver line stems displaying fluorescence from PI-stained cell walls and the mTurquoise2 reporter (Fig. 1; Table 1). The left and middle images are maximum projections of confocal stack, and the right images consist of a single median confocal xy section and xz and yz views of the stack. The indicated promoters mediate expression in the L1 layer/epidermis (B; pMERISTEM LAYER1 [pML1]), the stem cell domain (C; pCLV3), the central zone (D; pREVOLUTA [pREV]), the peripheral zone (E; pUNUSUAL FLOWER ORGANS [pUFO]), the boundary domain (F; pCUP-SHAPED COTYLEDON [pCUC2]), and organ primordia (G; pAHP6). PI fluorescence is false colored in magenta, and mTurquoise2 fluorescence is false colored in green. Bars = 20 µm.

Characterization of Gene Activation

We next tested whether the dose-response and induction dynamics observed previously with the GR-LhG4 system (Craft et al., 2005) were recapitulated in our setup. To this end, we germinated the pSCR driver line mediating GR-LhG4 expression in the QC and the endodermis (Di Laurenzio et al., 1996; Wysocka-Diller et al., 2000) on plates containing solvent or 0.1, 1, 10, or 100 µm Dex. Visualizing reporter fluorescence 5 DAG indeed revealed increasing reporter activity with increasing Dex concentrations (Fig. 5A), arguing for the possibility to fine-tune gene expression by adjusting the levels of the inducer. We noticed that QC cells showed markedly stronger fluorescence compared with the endodermis, putatively reflecting higher promoter GR-LhG4/reporter stability in the QC, as this was not observed with previously published lines using the same promoter fragment (Gallagher et al., 2004; Heidstra et al., 2004; Cruz-Ramírez et al., 2013). Therefore, we quantified fluorescence separately in the QC cells and the endodermal initials (Fig. 5, C and D). Whereas the QC did not show a significant difference in fluorescence intensity between any of the treatments (Fig. 5C), the endodermis fluorescence intensity correlated with the concentration of the inducer until saturation appeared to be reached between 10 and 100 µm Dex (Fig. 5D). Consequently, we concluded that, to fine-tune gene expression by applying different Dex concentrations, the appropriate concentration range has to be determined for each promoter and cell type individually.

Figure 5.

Dose-response and time-course analyses of driver line seedling roots. A, The pSCR driver line was grown on 0, 0.1, 1, 10, and 100 µm Dex and imaged 5 DAG. B, Time course of pSCR driver line induction for 1, 6, and 24 h with 10 µm Dex. Bars = 50 µm. C, Quantification of the mTurquoise2 fluorescence intensity dose response in QC cells and CEI (cells outlined in white in A). D, Quantification of mTurquoise2 fluorescence intensity of the first three endodermal cells after the CEI (cells outlined in blue in A). E, Quantification of the induction time course (B) in QC cells, CEI, and the first three endodermal cells. Significant differences in C to E are based on the results of a two-tailed Student’s t test with P < 0.05 (a), P < 0.01 (b), and P < 0.001 (c), n = 3 to 6 roots each. AU, arbitrary units.

To further assess induction kinetics, the pSCR driver line was germinated on plates with control medium and transferred onto plates containing 50 µm Dex after 5 d. As expected, a time-dependent increase of reporter activity was observed over a period of 24 h (Fig. 5B). Combined quantification of fluorescence in the QC and the endodermis initials detected reporter activity 6 h after induction (Fig. 5E), and the activity values were close to the values of constitutive Dex treatment after 24 h (Fig. 5, D and E). These observations suggested that 6 h are sufficient to allow the nuclear import of GR-LhG4, the induction of gene transcription, and initial protein translation and that, within 24 h, protein levels reached a steady-state level. In addition, 5-d-old roots that were induced at 2, 3, or 4 DAG showed similar reporter activities, demonstrating that responsiveness to the inducer is sustainable (Supplemental Fig. S6). To assess the kinetics of reporter expression after removal of the inducer, we germinated the pSCR>GR>mTurquoise2 line on Dex-containing medium and transferred the seedlings to Dex-free medium 2 DAG. Quantifying reporter fluorescence revealed that, 1 d after transfer, fluorescence intensity was indistinguishable from that in control plants transferred to inducer-containing plates but declined over the course of the next 2 d to hardly detectable levels (Supplemental Fig. S7).

To estimate the level of transcription mediated by the GR-LhG4/pOp system, we employed a line expressing PECTIN METHYLESTERASE INHIBITOR5 (PMEI5; Wolf et al., 2012) under the control of the strong and nearly ubiquitous 35S promoter (p35S:PMEI5). When comparing roots from the p35S:PMEI5 line with roots from a Dex-treated GR-LhG4/pOp line conferring expression of the same PMEI5 coding sequence in xylem pole pericycle (XPP) cells (designated as pXPP>GR>PMEI5; Craft et al., 2005), we observed PMEI5 transcript levels similar to or slightly exceeding those in the p35S:PMEI5 line (Supplemental Fig. S8). This was despite the fact that the XPP expression domain contains only approximately six cell files in the young root (Supplemental Fig. S2). Thus, we concluded that, although activating transcription in a very local manner, the GR-LhG4/pOp system can lead to strong expression in the respective cell types.

The ER-localized mTurquoise2 reporter present in our driver lines is transcribed from the same T-DNA that harbors the GR-LhG4 module (Fig. 1). To analyze the response of an independent T-DNA insertion carrying the pOp6 element in trans, we generated a transgenic line carrying an ER-targeted mVenus reporter under the control of the pOp6 promoter (pOp6:SP-mVenus-HDEL) and crossed it with the pSCR driver line. The resulting F1 plants did not show any reporter activity when grown on plates without Dex (Fig. 6), again confirming that the GR-LhG4/pOp system is fully Dex dependent. After Dex induction, we visualized both mTurquoise2 and mVenus fluorescence in the root and the stem and observed a complete congruence of both reporter activities (Fig. 6). Likewise, transgenic lines expressing a nucleus-targeted triple GFP fusion protein under the control of the pOp6 promoter were generated and crossed with the pCLAVATA3 (CLV3) driver line mediating expression in stem cells of the SAM (Fletcher et al., 1999). As expected, upon Dex induction, the 3xGFP-NLS signal was observed in a narrow domain at the tip of the SAM, which also expressed the mTurquoise2 marker (Fig. 6). Together, these observations confirmed the robust and specific transactivation of transgenes in F1 plants.

Figure 6.

Induction of mTurquoise2 and mVenus/3xGFP fluorescence in the root, stem, and SAM of F1 plants from a driver line-effector line cross. Cells are counterstained with PI (which, in the stem, highlights lignified vessel elements and fibers). Fluorescence channels are false colored. Bars = 50 µm for the root and the stem and 40 µm for the SAM.

Cell Type-Specific Induction of VND7 Demonstrates the Efficacy of Transactivation

To explore the potential of our lines to mediate the expression of a biologically active effector, we crossed the pSCR driver line with a line harboring the VND7 effector fused to the VP16 activation domain able to induce the formation of xylem vessels in a broad range of cell types (Kubo et al., 2005; Yamaguchi et al., 2010). F1 plants were grown on control medium for 5 d and then transferred to medium containing either 10 µm Dex or solvent. Five days later, fully differentiated vessel-like elements could be observed in the endodermis of the root and hypocotyl (Fig. 7), whereas in DMSO-treated controls, xylem elements were clearly restricted to the stele. These results demonstrate that this resource for cell type-specific and inducible transactivation can be used to study gene function with high spatiotemporal resolution.

Figure 7.

Cell type-specific induction demonstrates the efficacy of transactivation. Plants expressing VND7-VP16 as an effector in the endodermal cells (pSCR>GR>VND7-VP16) show ectopic vessel formation (white arrows) after 5 d of Dex induction in both root and hypocotyl endodermis, in contrast to DMSO-treated plants. The spiral secondary cell wall thickening was observed after fixing and clearing the samples and visualized by differential interference contrast microscopy. E, Endodermis; P, pericycle; X, xylem. Bars = 20 μm.

DISCUSSION

In this study, we combined the proven efficacy of the well-established GR-LhG4/pOp expression system (Craft et al., 2005; Rutherford et al., 2005; Samalova et al., 2005) with the ease of cloning enabled by the GreenGate system (Lampropoulos et al., 2013) to provide a comprehensive toolbox for inducible, cell type-specific expression in Arabidopsis. The driver lines described here cover a large proportion of the known cell types in the three main meristems of the plant, the RAM, the SAM, and the cambium. Our analysis demonstrates that this system achieves nonleaky, adjustable, and robust transactivation of effectors in the F1 generation after crossing with effector-carrying plants. Therefore, generating a line harboring an effector cassette under the control of the pOp6 promoter should enable users to rapidly assess a battery of different expression regimes for a wide range of applications. In most cases, the effector might be the coding region of a gene one may want to misexpress in a spatially and temporally controlled manner, but other uses are conceivable, such as adjustable (pulsed) expression of reporters, domain-specific knockdown through artificial microRNAs, cell type-specific complementation studies, the acquisition of cell type-specific transcriptomes/translatomes/proteomes/epigenomes, or the local induction of genome editing, for example through the expression of Cre recombinase or CRISPR/Cas9 modules (Birnbaum et al., 2003; Brady et al., 2007; Dinneny et al., 2008; Gifford et al., 2008; Mustroph et al., 2009; Deal and Henikoff, 2011; Hacham et al., 2011; Iyer-Pascuzzi et al., 2011; Petricka et al., 2012; Fridman et al., 2014; Adrian et al., 2015; Vragović et al., 2015; Efroni et al., 2016; Kang et al., 2017). Thus, this system should be a valuable tool for the generation of inducible genetic perturbations to overcome the limitations of end-point genetics and to study genetic activities in specific tissue contexts.

Design of the Transactivation System

Two-component transactivation and chemically induced gene expression systems have been used widely by plant biologist in the past. For example, a large collection of enhancer-trapping lines based on the yeast Gal4 transcription factor (Haseloff, 1999; Engineer et al., 2005) are an invaluable tool for constitutive, tissue-specific transactivation in Arabidopsis (Aoyama and Chua, 1997; Sabatini et al., 2003; Weijers et al., 2003, 2005; Swarup et al., 2005). In addition, an inducible system based on Gal and cognate upstream activation sequence has been devised (Aoyama and Chua, 1997) but appears to induce unspecific growth defects (Kang et al., 1999). Transactivation based on LhG4 (Moore et al., 1998) shows only minimal detrimental effects on plant development, is thoroughly characterized and optimized (Moore et al., 1998, 2006; Baroux et al., 2005; Craft et al., 2005; Rutherford et al., 2005; Samalova et al., 2005), and has been used by the plant community in a number of studies (Schoof et al., 2000; Baroux et al., 2001; Eshed et al., 2001; Hay and Tsiantis, 2006; Nodine and Bartel, 2012; Sauret-Güeto et al., 2013; Hazak et al., 2014; Serrano-Mislata et al., 2015; Jiang and Berger, 2017). Parallel to the development of these tools for cell type-specific expression, a number of inducible systems have been conceived to enable temporal control of gene expression (Gatz et al., 1992; Weinmann et al., 1994; Caddick et al., 1998; Zuo et al., 2000). Subsequently, combining and optimizing the available technology has succeeded in generating tools to mediate inducible expression in a cell type-specific manner (Deveaux et al., 2003; Laufs et al., 2003; Maizel and Weigel, 2004; Craft et al., 2005).

For the generation of this resource, we built on groundbreaking previous work establishing the LhG4 system in combination with the GR ligand-binding domain (Craft et al., 2005), which has since been proven to be a valuable resource (Reddy and Meyerowitz, 2005; Ongaro et al., 2008; Ongaro and Leyser, 2008; Heisler et al., 2010; Jiang et al., 2011; Dello Ioio et al., 2012; Merelo et al., 2016; Caggiano et al., 2017; Tao et al., 2017). For the generation of our driver lines, we exploited the power of the GreenGate cloning system (Lampropoulos et al., 2013). We were able to rapidly assemble a large number of constructs efficiently, circumventing the bottleneck imposed previously by the challenging generation of large DNA constructs with varying promoter elements, coding regions, and terminators. Thus, the limiting factor in generating this resource was plant transformation and obtaining single-insertion, homozygous transgenic lines. As a general workflow, we aimed to generate at least 40 T1 transformants, then scored segregation ratios of antibiotic/herbicide resistance in the T2 generation and maintained lines in which the resistance segregated as a single locus. These lines usually showed similar characteristics concerning the response to the inducer and the expression levels achieved through transactivation (based on fluorescence intensity). Nevertheless, reporter expression in any set of newly generated driver lines should be assessed carefully and compared with the literature and within lines, as genome integration in the vicinity of endogenous promoter and/or enhancer elements might influence the expression pattern. As expected, we occasionally observed widespread silencing in the T2 generation of the driver lines, which did not correlate with any particular DNA element present in multiple constructs

An important feature of our driver lines is the incorporation of a reporter amenable to live imaging, which can be used to monitor the induction and visualize the spatial expression domain. In addition, it allows us to assess whether the expression of the effector has an impact on the transcriptional circuitries of the cell type it is expressed from. For some applications, the internal reporter of the driver lines also might serve as an inducible marker even in the absence of any further effector expression. We chose mTurquoise2 as a fluorescent reporter, since its spectral characteristics make it compatible with more widely used green and red fluorophores and it displays high photostability, fast maturation, and high quantum yield (Goedhart et al., 2012). The fluorescent protein was N-terminally fused with a signal peptide and modified with a C-terminal HDEL motif to mediate retention in the ER, which in our experience is the preferable subcellular localization for a fluorescent reporter when cross sections through the highly differentiated cells of the stem are required.

Transactivation Characteristics

Our system allows stringent temporal control of gene expression, as indicated by the lack of reporter expression in the absence of the inducer Dex. Moreover, the transactivated reporter faithfully reproduced previously described expression patterns associated with the respective 5′ regulatory regions, suggesting that the chimeric GR-LhG4 transcription factor is not cell-to-cell mobile. However, we noticed that, in some cases, transactivation led to slightly different expression patterns as compared with fusions of the same 5′ regulatory region with a reporter gene in cis. For example, expression driven from the CLV3 promoter seemed broader than what was described in pCLV3:XFP lines but consistent with a similarly designed pCLV3-driven transactivation (Serrano-Mislata et al., 2015), possibly because the multiple binding sites of the pOp6 promoter increase expression in cells where the CLV3 promoter is only weakly active. Alternatively, the high protein stability of the chimeric transcription factor, the reporter, or both might cause prolonged activity of these proteins in cells that are already displaced from the stem cell region. This potential issue is less relevant for organs such as the root, where cells of one cell type also largely have the same clonal identity (Kidner et al., 2000; Costa, 2016).

Our experiments, in agreement with previous results, suggested that GR-LhG4/pOp-mediated transactivation can achieve tissue-specific overexpression of the target gene, dependent on the concentration of the inducer. However, the possibility of squelching, the sequestration of general transcription factors required for other processes by the LhG4 activation domain, must be taken into account at very high expression levels. Consistent with previous reports (Craft et al., 2005), our analysis of the pSCR driver line revealed a linear dose response over at least 2 orders of magnitude, but the induction kinetics might be affected by the genomic location of the transgene and, thus, should be determined empirically for each line. It should be noted that the expression of effectors using LhG4/pOp systems can be quenched by adding isopropyl β-d-1-thiogalactopyranoside (Craft et al., 2005), which would allow pulsing experiments. However, we did not test the effect of isopropyl β-d-1-thiogalactopyranoside in our lines.

Distribution of Driver Lines and DNA Constructs

The lines described here, as well as DNA constructs, are available to the community upon request. While GR-LhG4 and the sulfadiazine resistance gene are expressed constitutively, care should be taken to amplify seeds only from noninduced plants to minimize the chance of inducing posttranscriptional gene silencing through the high expression levels of the reporter (Schubert et al., 2004; Abranches et al., 2005).

MATERIALS AND METHODS

Cloning

All constructs were produced by GreenGate cloning (Lampropoulos et al., 2013) using the modules described in Supplemental Table S1. The Eco31I (BsaI) sites of the SCR, PXY, and WOX4 promoters were removed by the QuikChange XL Site-Directed Mutagenesis Kit (Agilent Technologies) using the primers listed in Supplemental Table S1 following the manufacturer’s instructions. The Eco31I site of the ATHB-8 promoter was removed by amplifying the 5′ part of the promoter up to the endogenous Eco31I restriction site, which was mutated by a single-base exchange in the primer. This primer contained an Eco31I restriction site in the 5′ overhang. The 3′ fragment of the promoter was amplified with a forward primer directed against the region immediately 3′ of the endogenous Eco31I site (containing an Eco31I site in the 5′ overhang) and the reverse primer binding to the region immediately upstream of the ATG. The two fragments were amplified separately, digested with Eco31I, and ligated afterward. As Eco31I is a type II restriction enzyme, the recognition site in the primer overhangs were removed by digestion.

The repetitive sequences of the pOp promoter increase the likelihood of recombination events while amplifying the plasmids. To discriminate against clones with shorter pOp sequences, we designed primers that bind in the short flanking sequences at the beginning and end of pOp6 (pOp6_F, 5′-TGCATATGTCGAGCTCAAGAA-3′; and pOp6_R, 5′-CTTATATAGAGGAAGGGTCTT-3′) for PCR amplification and size assessment through gel electrophoresis. Final constructs were always confirmed by sequencing in Escherichia coli and Agrobacterium tumefaciens. The occasional recombination events were detected only in E. coli.

Plant Material and Growth Conditions

All constructs were transformed by the floral dip method (Clough and Bent, 1998) as modified by Zhang et al. (2006) into Arabidopsis (Arabidopsis thaliana) Col-0. Transformed seeds were selected on half-strength Murashige and Skoog plates containing 1.875 to 3.75 µg mL⁻¹ sulfadiazine or 7.5 µg mL⁻¹ glufosinate ammonium. Only single integration lines based on T2 segregation ratios were propagated to T3, in which plants homozygous for the resistance were selected. All plants were grown in long-day conditions (16 h of light/8 h of dark) at 22°C. For root analysis, plants were grown vertically on half-strength Murashige and Skoog plates containing 1% (w/v) Suc and 0.9% (w/v) plant agar (Duchefa; P1001). For the induction treatments on plates, the seeds were sown on plates containing Dex (Sigma-Aldrich; D4903) at the indicated concentrations while the same volume of DMSO (D139-1; Fisher Scientific) was added for the mock control. For the transactivation experiment, seeds were sown on plates without Dex and seedlings were transferred to Dex-containing plates at 1, 6, and 24 h before imaging 5 DAG. For analysis of the stem, the aerial parts of 15-cm-tall plants were dipped for 30 s in either tap water containing 10 µm Dex with 0.02% (v/v) Silwet L-77 (Kurt Obermeier) or water with the same volume of DMSO with 0.02% (v/v) Silwet. After 24 h, free-hand sections of the stem were performed with a razor blade. Sections were transferred to a small petri dish (35/10 mm; Greiner Bio-One) with 0.25 mg mL⁻¹ PI for 5 min and mounted on microscope slides to be visualized by CSLM. For SAM imaging, the inflorescence meristems of 25- to 30-DAG plants were sprayed with 10 µm Dex, whereas an equal volume of DMSO was added to water sprayed onto the mock controls. At 48 h after the treatment, the inflorescence meristems were dissected by cutting of the stem, flowers, and buds. The SAM was stained in 0.25 mg mL⁻¹ PI (Sigma-Aldrich; P4170) for 5 min, mounted in a 3% (w/v) agarose small petri dish (35/10 mm; Greiner Bio-One), and visualized by CLSM.

Microscopy

Root samples were imaged using a Leica TCS SP5 laser scanning confocal microscope with an HCX PL APO lambda blue 63× water-immersion objective. The mTurquoise2 fluorophore was excited by an argon laser at 458 nm, and emission was collected between 460 and 516 nm. The mVenus fluorophore was excited by 514 nm, and emission was collected between 520 and 580 nm. Cells were counterstained by PI (Sigma-Aldrich; P4170) and imaged with 488 nm for excitation, and emission was collected between 590 and 660 nm.

For stem and SAM samples, we used a Nikon (Minato) A1 confocal microscope with a CFI Apo LWD 25× water-immersion objective. The PI-counterstained cells were imaged with 561 nm for excitation and 570 to 620 nm for emission. mTurquoise2 fluorescence was acquired using excitation at 405 nm, and emission was collected between 425 and 475 nm. For the transactivation experiments, the 3xGFP-NLS signal in the SAM was imaged with 488 nm for excitation and 500 to 550 nm for emission. In the root, mVenus was excited with 514 nm, and the emission was collected between 500 and 550 nm.

For visualization of the xylem, plants were germinated on half-strength Murashige and Skoog plates and transferred 5 DAG to either 10 µm Dex- or mock solution-containing half-strength Murashige and Skoog plates. To visualize ectopic xylem formation, plants were collected 5 d after induction and fixed overnight in a 1:3 acetic acid:ethanol solution. Then, they were cleared in a 8:1:2 chloral hydrate:glycerol:water solution for at least 3 h. Samples were mounted on microscope slides containing 50% (v/v) glycerol solution, and bright-field images were obtained using an Axioimager M1 microscope equipped with an AxioCamHRc (Carl Zeiss).

qPCR and SA-qPCR Analyses

Analysis of PMEI5 expression by qPCR was performed as described (Wolf et al., 2012). For SA-qPCR, plant DNA extraction was performed as described by Allen et al. (2006) and SA-qPCR was performed as described (Huang et al., 2013). Quadruplicate qPCR was performed in a final volume of 12.5 µL, including 6.25 µL of ABsolute qPCR SYBR Green Mix (Thermo Scientific), 0.25 µL of each primer (10 µm), and 2 µL of genomic DNA (1.6 ng µL⁻¹) with different amounts (0, 1, or 3 µL) of plasmid (0.1 pg µL⁻¹) as a reference. The SulfR resistance gene was amplified with primers SulfR_Fwd (5′-GCATGATCTAACCCTCTGTCTC-3′) and SulfR_Rvs (5′-GAAGTCACTCGTTCCCACTAG-3′), and the plasmid target sequence was amplified with PL_Fwd (5′-GCCGTACTAAACCTCTCATCG-3′) and PL_Rvs (5′-CTGACCGGAAAGTTTGTTATTCG-3′).

Accession Numbers

The Arabidopsis Genome Initiative numbers of genes used in this study are as follows: SCR (AT3G54220), ATHB-8 (AT4G32880), XPP (At4g30450), AHP6 (AT1G80100), PXY (AT5G61480), TMO5 (AT3G25710), SMXL5 (AT5G57130), CASP1 (AT2G36100), VND7 (AT1G71930), APL (AT1G79430), NST3 (AT1G32770), WOX4 (AT1G46480), PMEI5 (AT2G31430), LTP1 (AT2G38540), ML1 (AT4G21750), CLV3 (AT2G27250), REV (AT5G60690), UFO (AT1G30950), and CUC2 (AT5G53950).

Supplemental Data

The following supplemental materials are available.

• Supplemental Figure S1. Analysis of DMSO-treated mock controls for driver line seedling root induction 5 DAG.

• Supplemental Figure S2. Analysis of induced driver lines in 5-DAG seedling root.

• Supplemental Figure S3. Analysis of DMSO-treated driver lines in the stem.

• Supplemental Figure S4. Analysis of DMSO-treated driver lines in the SAM.

• Supplemental Figure S5. Growth on 50 µm Dex does not impair root growth of Col-0.

• Supplemental Figure S6. Reporter activation in the pSCR>GR>mTurquoise2 line is sustainable.

• Supplemental Figure S7. Kinetics of pSCR>GR>mTurquoise2 reporter activity after removal of inducer.

• Supplemental Figure S8. Quantification of GR-LhG4-mediated transactivation.

• Supplemental Table S1. List of primers used and DNA constructs generated in this study.

Dive Curated Terms

The following phenotypic, genotypic, and functional terms are of significance to the work described in this paper:

• ATHB-8 Gramene: AT4G32880

• ATHB-8 Araport: AT4G32880

• PERICYCLE AmiGo: PO:0006203

• root AmiGo: PO:0009005