In Arabidopsis thaliana, small peptides (AtPeps) encoded by PROPEP genes act as damage-associated molecular patterns (DAMPs) that are perceived by two leucine-rich repeat receptor kinases, PEPR1 and PEPR2, to amplify defense responses. In particular, expression of PROPEP2 and PROPEP3 is strongly and rapidly induced by AtPeps, in response to bacterial, oomycete, and fungal pathogens, and microbe-associated molecular patterns (MAMPs).
The cis-regulatory modules (CRMs) within the PROPEP2 and PROPEP3 promoters that mediate MAMP responsiveness were delineated, employing parsley (Petroselinum crispum) protoplasts and transgenic A. thaliana plants harboring promoter-reporter constructs. By chromatin immunoprecipitation in vivo, DNA interactions with a specific transcription factor were detected. Furthermore, the phastCons program was used to identify conserved regions of the PROPEP3 locus in different Brassicaceae species.
The major MAMP-responsive CRM within the PROPEP2 promoter is composed of several W boxes and an as1/OCS (activation sequence-1/octopine synthase) enhancer element, while in the PROPEP3 promoter the CRM is comprised of six W boxes. The WRKY33 transcription factor binds in vivo to these promoter regions in a MAMP-dependent manner. Both the position and orientation of the six W boxes are conserved within the PROPEP3 promoters of four other Brassicaceae family members.
WRKY factors are the major regulators of MAMP-induced PROPEP2 and PROPEP3 expression.
Plant innate immunity is based on a multilayer host defense response that is highly efficient at deterring infections by a broad range of diverse phytopathogens (Dodds & Rathjen, 2010). Conserved microbial components designated microbe-associated molecular patterns (MAMPs) are recognized at the plant cell surface by pattern recognition receptors (PRRs) that initiate a signaling cascade leading to MAMP-triggered immunity (MTI), one major branch of the immune system. Endogenous host molecules, termed damage-associated molecular patterns (DAMPs), also trigger a plant defense response via PRRs but are released only upon tissue damage mainly caused by pathogenic microbes (Boller & Felix, 2009). DAMP signaling has been proposed to intensify or prolong the stereotypical defense response triggered by MAMPs (Yamaguchi & Huffaker, 2011). MAMP/DAMP-triggered signaling results in massive transcriptional reprogramming of the host cell, which is a prerequisite for establishing and fine-tuning effective defenses (Moore et al., 2011; Boller & Flury, 2012).
In Arabidopsis thaliana, the active epitope of the bacterial flagellin, flg22, is perceived as a MAMP by the FLAGELLIN SENSING2 (FLS2) receptor, thereby initiating downstream signaling events (Gómez-Gómez & Boller, 2000). Similarly, the PEPR1 and PEPR2 receptors are involved in in eliciting DAMP signaling upon perception of their ligands (Krol et al., 2010; Yamaguchi et al., 2010). These host ligands are the products of a family of six genes, PROPEP-1 to -6, encoding precursor proteins that require subsequent processing (Pep1 to Pep6) and possibly export to the apoplast for receptor binding (Huffaker et al., 2006; Huffaker & Ryan, 2007). PROPEP2 and PROPEP3 expression was previously shown to be rapidly induced by pathogens and MAMPs including flg22, whereas the other genes were not significantly affected by such treatments (Huffaker et al., 2006). PROPEP1 expression, however, was induced by the plant hormone methyljasmonate (MeJA), by wounding, or by Pep1, the ligand product of PROPEP1 itself (Huffaker & Ryan, 2007).
Host transcriptional responses are a vital component of plant immunity, and several expression studies have revealed that transcript levels of numerous genes, including those encoding a multitude of transcriptional regulators, are strongly modulated upon pathogen perception (Eulgem, 2005). Major transcription factors that have been identified as important regulators of plant immunity include members of the basic leucine Zipper domain (bZIP), Myeloblastosis (MYB), Ethylene Responsive actor (ERF) and WRKY gene families along with co-regulators such as Nonexpressor of PR gene1 (NPR1), Calmodulin Binding Protein 60-like.g (CBP60g) and Systemic Acquired Resistance Deficient1 (SARD1) (Eulgem, 2005; Moore et al., 2011).
Transcriptional regulation of immune response genes requires the proper temporal and spatial binding of transcription factors to cis-regulatory DNA elements (CREs) often present in promoter regions. In particular, putative WRKY factor binding sites (W boxes) are often conserved in upstream regulatory regions of defense-related genes (Pandey & Somssich, 2009). Genome-wide expression profiling studies in combination with bioinformatics approaches are proving increasingly important in pinpointing potential relevant cis-acting DNA elements and their combinatorial role as cis-regulatory modules (CRMs), and in deciphering transcriptional networks (Priest et al., 2009; Ding et al., 2012). Still, despite the in silico identification of hundreds of putative stress-responsive CREs by such means (Zou et al., 2011), the number of experimentally verified cis-motifs and CRMs involved in the stimulus-dependent expression of specific plant defense genes remains very small. Thus, irrespective of the time- and labor-intensive work required for such functional analyses, they remain essential in testing and validating in silico models.
In this study we characterized the promoters of the two major MAMP/DAMP-responsive A. thaliana PROPEP genes and delineated functional CREs required for mediating gene expression. This work clearly demonstrated that WRKY-type transcription factors, via their action on W boxes, play a major role in regulating transcriptional outputs of PROPEP2 and PROPEP3. Our data showed that distinct CRMs are present in the respective gene promoters that are involved in conferring high MAMP inducibility.
Materials and Methods
Plants, cell culture and growth conditions
All experiments were performed with Arabidopsis thaliana Columbia-0 (wild type (WT)) plants, or stable transgenic lines in the Col-0 background. Plants were grown for 4–5 wk under short-day conditions in closed cabinets (Snijder Scientific, Tilburg, the Netherlands: 20°C, 16 h light : 8 h darkness, and 80% humidity) in soil or for pathogen treatments in 42-mm Jiffy-7 pots (Jiffy, Mölln, Germany). Before sewing the Jiffy pot, peat pellets were soaked in water containing 0.1% liquid fertilizer ‘Wuxal’ (Manna, Düsseldorf, Germany). Seedlings were grown in 0.5× Murashige and Skoog (MS) medium, pH 5.7, after selection on MS agar plates. Parsley (Petroselinum crispum) Pc5/3 cells were cultivated as previously described (Meier et al., 1991).
For flg22 or elongation factor18 (elf18) treatments, mature leaves were hand-infiltrated with 1 μM flg22 solution using a 1-ml syringe or by vacuum infiltration. For seedlings, plantlets were transferred after BASTA® (Bayer Crop Sciences) selection on MS plates into microtiter plates containing liquid MS medium, and 24 h thereafter flg22 was added to a final concentration of 1 μM. For infections, Botrytis cinerea isolate 2100 (CECT2100; Spanish Type Culture Collection, University of Valencia, Valencia, Spain) was used, which was cultivated on potato dextrose plates at 22°C for 10 d. Spores were collected, washed and frozen at −80°C in 0.8% NaCl at a concentration of 107 spores ml−1. For infection of A. thaliana plants, the spores were diluted in Vogelbuffer (in 1 l: 15 g sucrose, 3 g Na-citrate, 5 g K2HPO4, 0.2 g MgSO4 7H2O, 0.1 g CaCl2 2H2O and 2 g NH4NO3) to 5 × 105 spores ml−1. For droplet infections, 5 μl was applied to single leaves of 5-wk-old plants. For mock treatment, Vogelbuffer was used. One day before and during the entire infection process plants were incubated at high humidity under a hood.
Translational promoter/reporter-gene fusions were achieved by ligation of appropriate PCR-derived PROPEP gene fragments into a linearized pUC-GUS vector (van de Löcht et al., 1990). Constructs MS23-3C and MS23-3Cmut were generated by ligation of PCR fragments into the HindIII and SpeI sites of the pBT10-GUS vector (Sprenger-Haussels & Weisshaar, 2000).
Mutations of all described W boxes were achieved by altering the core sequence 5′-TGAC-3′ to 5′-TGAA-3′ or by the PCR-based megaprimer method (Landt et al., 1990). Similarly, the activation sequence-1/octopine synthase (as-1/OCS) enhancer sequence 5′-ACTACGTAA-3′ within the PROPEP2 promoter was mutated to 5′-ACTTCGAAA-3′, and the putative D box motif in the PROPEP3 promoter from 5′-TTCAAACAAA-3′ to 5′-TTCAAGAAAA-3′. The correctness of all DNA constructs was determined by the Max Planck Genome Center Cologne using BigDye terminator chemistry on Applied Biosystems 377 sequencers. Oligonucleotides were purchased from Gibco (Life Technologies).
Transient transfection experiments
Transient expression analyses with parsley cell culture protoplasts were performed as previously described (van de Löcht et al., 1990). In each assay, 2 × 106 protoplasts were transfected with 5 μg of GUS reporter plasmid linearized with ScaI. MAMP treatment was carried out with 333 ng ml−1 pep25 derived from the oomycete Phytophthora sojae (Nürnberger et al., 1994). Suspensions of transfected protoplasts were split, and one half was treated for 18 h with the MAMP while the other half remained untreated. Aliquots were subsequently harvested and treated as described previously (Rushton et al., 1996).
Construction of transgenes and generation of transgenic plants
For appropriate reporter constructions, the BamHI/SfoI GUS reporter fragment of pUC-GUS was ligated into the BamHI/StuI sites of the binary vector pGreenII0229, and for the GFP fusions the NotI/XhoI eGFP reporter fragment of pAM-Kan-MCS-eGFP (obtained from M. Kwaitaal, MPIPZ Köln, Germany) was ligated into NotI/XhoI sites of pGreenII0229.
GUS reporter fusions of all PROPEP promoter fragments were ligated into the HindIII/SnaBI sites of the pGreenII0229-GUS vector. For the PROPEP1-GFP fusion the GUS reporter gene in the PROPEP2B:GUS constructs was replaced by the PCR amplified coding sequence (cds) of PROPEP1 (At5 g64900). Subsequently, the PROPEP2B promoter and PROPEP1cds were ligated into the EcoRI/XhoI sites of the pGreenII0229-eGFP vector.
All constructs were transformed into the Agrobacteriumtumefaciens strain GV3101-pMP90 (Koncz & Schell, 1986), including the helper plasmid pSoup. Agrobacteriumtumefaciens-mediated transformation into pepr1 pepr2 and Col-0 plants was performed as described previously (Logemann et al., 2006). Transformants were selected by spraying BASTA. Subsequent selection was achieved on 0.5× MS medium, pH 5.7, 1% sucrose, 50 μM phosphinotrycine and 0.8% agar. For each construct, eight to 40 individual lines were analyzed.
Histochemical staining of plant tissue was performed by vacuum infiltration (three times for 1–2 min each) with X-Gluc solution (100 mM NaPO4, 2 mM K3Fe(CN)6, 0.1% Triton-X-100, and 0.5 mg ml−1 X-Gluc) followed by incubation at 37°C overnight, and subsequent destaining of the tissue in 70% ethanol.
Plant protein extracts
For enzyme activity assays, 100–200 mg of plant material was ground in 300 μl of GUS extraction buffer (50 mM NaPO4, pH 7.0, 1 mM EDTA, 0.1% Triton and 10 mM β-mercaptoethanol) using a TissueLyser (Qiagen, Hilden, Germany). Enzyme activity assays were performed as described previously (Rushton et al., 1996).
Two to three mature leaves or nine to 12 seedlings were ground in liquid nitrogen using a TissueLyser (Qiagen), resuspended in 150–200 μl of 2X Laemmli loading buffer and subsequently incubated three times for 5 min at 95°C with cooling between on ice.
Fifteen microliters of total protein extracts was separated by 12% or 15% SDS-PAGE and transferred to PVDF membranes. Five per cent milk powder (Roth, Karlsruhe, Germany) was used for blocking. Mouse anti-GFP immunoglobulin G (IgG) (Roche) was used as the primary antibody (1 : 3000 dilution). Sheep anti-mouse Horseradish Peroxidase (HRP) (GE-Healthcare, Braunschweig, Germany) was used as the secondary antibody (1 : 5000 dilution). GFP proteins were detected using Thermo Scientific SuperSignal West Femto (part no. 34095; Thermo Scientific, Karlsruhe, Germany) 1 : 1. CL-XPosure films were exposed for 1–15 min.
Chromatin immunoprecipitation (ChIP)
Twelve-day-old seedlings, grown in 1× MS medium containing 1% sucrose, were treated with 1 μM flg22 for 2 h before processing following the protocol of Gendrel et al. (2005) with some modifications. Each 2-g sample was cross-linked by vacuum infiltration of 1% formaldehyde solution for three subsequent 7-min treatments. Sonication of the isolated nuclei was performed on a Bioruptor Next Gen sonicator (Diagenode, Liége, Belgium) in ice-cold water for 5 × 30 s with a 30-s pause between sonications. The sheared chromatin (25 μg) was diluted 10-fold in ChIP dilution buffer (Gendrel et al., 2005) containing protease inhibitors (Roche) and phosphatase inhibitor (Sigma) to lower the SDS concentration to 0.1%. After preclearing with proteinA agarose beads, the chromatin was incubated overnight with rabbit polyclonal antibodies to Human influenza hemaglutinin (HA) (ChIP grade, Abcam, Cambridge, UK) at 4°C on a nutator. Immune complexes were collected by incubation with proteinA agarose. After washing of the beads, elution of the immune complexes, reverse cross-linking and recovery of the DNA were performed as described previously (Gendrel et al., 2005). One microliter of DNA was used for qPCR. Primers are listed in Supporting Information Table S1.
Quantitative real-time PCR (qRT-PCR)
Total RNA was isolated from seedlings 0, 2, and 4 h post MAMP treatment and reverse-transcribed with oligo dT primer to produce cDNA using the SuperScript first-strand synthesis system for reverse-transcription PCR (Invitrogen), following the manufacturer's protocol. cDNA corresponding to 2.5 ng of total RNA was subjected to qPCR with gene-specific primers using the Brilliant SYBR Green qPCR core reagent kit (Agilent Technologies, Stratagene Products Division, La Jolla, CA, USA). The qPCRs were performed on an iQ5 multicolor real-time PCR detection system (Bio-Rad, München, Germany), with two technical replicates in the same run and three biological replicates in different runs. For normalization, the endogenous gene At4 g26410, showing highly constant expression under varying stress conditions, was used (Czechowski et al., 2005). The results were analyzed using the ΔΔCt method (Livak & Schmittgen, 2001). Data are shown as the mean ± standard deviation (SD) from three biological replicates.
Leaves of PROPEP2B:PROPEP1-GFP and PROPEP2Bmut5:PROPEP1-GFP transgenic plants were infiltrated with 1 mM flg22 for 16 h before analysis. Detection of expressed protein was by confocal laser scanning microscopy with a Zeiss LSM 700 (Carl Zeiss Jena, Jena, Germany) using a 40 × 1.3 numerical aperture NeoFluar oil-immersion objective to collect images. GFP was detected using the 488-nm laser line for excitation and two-channel imaging of emission, 505–550 nm (green/GFP) and 650–720 nm (red/chlorophyll) plus an additional channel for transmission. Images processing was with Zeiss zen 2011 software and Adobe Photoshop CS3.
Sequence conservation analysis and identification of putative CREs
Genome-wide orthologous gene relationships between the genomes of A. thaliana (http://www.arabidopsis.org/index.jsp), Arabidopsis lyrata (Hu et al., 2011), Brassica rapa (Wang et al., 2011) and Arabis alpina (Fig. S12) were analyzed using three independent approaches; reciprocal best BLAST hit (Moreno-Hagelsieb & Latimer, 2008), INPARANOID (Östlund et al., 2010), and Mercator (Dewey et al., 2006). PROPEP3 orthologous genes were unambiguously identified by all three methods. For PROPEP2, no reliable ortholog relationship between any of the other genomes and A. thaliana could be established. Multiple sequence alignment of the 500-bp PROPEP3 ortholog promoter sequences was performed using the MultiZ and tba alignment programs (Blanchette et al., 2004). Using the phast package (Siepel et al., 2005), nucleotide-level conservation scores (PhastCons scores) were generated using default parameters and A. thaliana as the base sequence. De novo motif prediction was carried out with meme software using the ‘any number of repeats’ model (Bailey et al., 2006).
Delineation of the PROPEP1 and PROPEP2 promoters mediating MAMP responsiveness
Of the six A. thaliana PROPEP genes, expression of PROPEP1, PROPEP2 and PROPEP3 was shown to be induced by Botrytis cinerea, Phytophthora infestans, MAMPs such as flg22, wounding, and MeJA treatment (Huffaker et al., 2006). To define the promoter regions of these three genes that mediate MAMP responsiveness, we initially employed the well-established parsley transient expression system and the pathogen-derived peptide MAMP pep25 (Rushton et al., 2002; Hahlbrock et al., 2003; Koschmann et al., 2012). We generated promoter-GUS reporter constructs encompassing the entire 5′ upstream regions of PROPEP1 (1.73 kb), PROPEP2 (0.56 kb) and PROPEP3 (1.52 kb). These sequences were derived from the annotated A. thaliana genome at The Arabidopsis Information Resource (TAIR), and encompassed the entire intergenic regions up to the next 5′ open reading frames. In addition, numerous promoter derivative-GUS constructs, as outlined below (Figs 1–3), were generated. DNA constructs were transfected into parsley protoplasts and the cells subsequently treated for 18 h with pep25 before harvesting. Transient expression of the reporter gene was monitored by measuring β-glucuronidase activity.
Searches in databases that identify CREs, such as AtcisDB (Yilmaz et al., 2011), PLACE (Higo et al., 1999), and TRANSFAC (Hehl & Wingender, 2001), revealed that the 1.73-kb PROPEP1 promoter contains several putative transcription factor binding sites (Fig. S1). For three PROPEP1 promoter deletion constructs, the pep25-induced and basal expression levels were assayed and the ratio between the two determined (Fig. 2). Both the −1737 and the −761 PROPEP1 promoters conferred pep25 inducibility of the reporter gene, whereas the shortest version, −236, did not. However, the overall measured GUS activity levels were low compared with the PcPR1-1A promoter-GUS construct (c. 18 000–22 000 pmol min−1 mg−1; data not shown) which served as a positive control for pep25 responsiveness (Tovar Torres et al., 1993). This weak response is consistent with reported microarray data obtained in A. thaliana (Huffaker et al., 2006). Thus, despite its responsiveness, this promoter was not further dissected.
For the 560-bp PROPEP2 promoter, databases searches also identified numerous putative CREs that may serve as potential binding sites for Auxin Response Factor (ARF), bZIP, DNA-binding with One Finger (Dof), MYB and Trihelix factors. However, the most conspicuous feature appeared to be the clustering of several W boxes that are bound by WRKY transcription factors (Rushton et al., 2010; Fig. S3). A promoter deletion series was generated for the PROPEP2 promoter, taking into account the distribution of these W boxes, and the appropriate promoter-GUS reporter constructs were functionally assayed in parsley protoplasts (Figs 1a, S3). Strong pep25-dependent β-glucuronidase activity was observed for all constructs bearing promoter sequences down to −294 relative to the translation start site (constructs 2A–C). Further deletion down to −163, however, completely abolished pep25-induced expression and drastically reduced overall GUS activity (construct 2D). The differences in the fold inductions of constructs 2A–2C were mainly attributable to differences in the basal levels measured for the constructs rather than in the overall GUS activity levels, although overall GUS activity was also slightly reduced for construct 2C. These results demonstrated that the major DNA elements required for pep25 inducibility reside in the promoter region −294 to −163. The region −348 to −163 harbors four W boxes in reverse orientation between −348 and −294, and two W boxes in head-to-head orientation along with an overlapping as-1/OCS enhancer sequence between −294 and −164 (Fig. 1b). The as-1/OCS enhancer is known to bind members of the bZIP transcription factor family (Jakoby et al., 2002).
To pinpoint critical elements mediating pep25 responsiveness, a set of mutations within these motifs were generated within the context of construct 2B. For the W boxes, a single base substitution within the core motif (TGAC to TGAA) was employed that abolishes WRKY factor binding and W box-dependent gene activation (Yu et al., 2001). For the overlapping as-1/OCS enhancer, two base substitutions disrupting the ACGT core to TCGA or CCGG were made. As shown in Fig. 1(b), simultaneous mutation of all four W boxes between −348 and −294 (mut1) reduced the overall levels of pep25-dependent GUS activity and slightly reduced fold-induction values. This is similar to the differences observed between the deletion constructs 2B and 2C (Fig. 1a) and indicated that these W boxes only partly mediate MAMP-dependent expression. By contrast, mutations of the two downstream W boxes had a more profound affect on pep25-dependent GUS activity levels and on fold inducibility (mut2). The weaker contribution of the four upstream W boxes was also evident when comparing the values obtained with constructs mut2 and mut3, in which all six W boxes were mutated. Mutation of the as-1/OCS enhancer region alone (mut4) also strongly reduced pep-25 dependent GUS activity and fold-induction values. However, as with construct mut3, MAMP-dependent activation of the reporter gene was still observed. Simultaneous mutations of all tested elements drastically reduced GUS activity and rendered the construct basically nonresponsive to pep25 (mut5).
We conclude that W boxes together with the as-1/OCS enhancer act as the major CRM to mediate MAMP responsiveness of the PROPEP2 gene.
Delineation of the PROPEP3 promoter mediating MAMP responsiveness
In silico analysis indicated that the 1.52-kb 5′ region of PROPEP3 contains numerous binding sites for various transcription factors (Fig. S4). We first generated a series of promoter deletions to uncover regions containing major elements mediating MAMP responsiveness. As this promoter also contains several, partly clustered, W boxes we designed the deletions such as to sequentially delete sets of W boxes (Fig. S4). Transient expression analyses revealed that comparable high levels of pep25-dependent GUS activity were sustained with all constructs still harboring the −442 PROPEP3 promoter region (Fig. 2a; constructs 3A–3C). These constructs enabled 125- to 155-fold increases in GUS activity upon pep25 application. Further deletion down to −321 significantly reduced GUS activity levels, and to a lesser extent, fold inducibility (construct 3D). A substantial decrease was further observed by removing additional sequences down to −215, although this short promoter still showed 24-fold induction upon MAMP treatment (construct 3E). To investigate the contribution of W boxes in mediating MAMP responsiveness of PROPEP3, single base pair mutations of the W boxes were again generated in the context of constructs 3C, 3D and 3E (Fig. 2b). Simultaneous mutations of the two most distal W boxes (construct 3Cmut1) had no significant effects on pep25 inducibility, although they increased basal reporter gene expression, thereby reducing fold-induction values. Somewhat surprisingly, simultaneous mutation of the four most proximal W boxes (3Cmut2) or of all six W boxes in construct 3C (3Cmut3), while reducing the absolute levels of pep25-dependent GUS activity to around 26% of those of the WT construct 3C, still conferred strong inducibility with high fold-induction values as a result of strongly reduced basal expression levels. Scanning for additional known DNA elements within this promoter region revealed one element, termed the D box, that has been shown to confer strong pep25 inducibility (Rushton et al., 2002). Unexpectedly, mutation of the D element (5′-TTCAAACAAA-3′ to 5′-TTCAAGAAAA-3′) in combination with all six W boxes actually boosted pep25-dependent GUS activity to high levels while basal expression levels remained very low (3Cmut4).
Simultaneous mutations of the two upstream or two downstream W boxes within the −321 deletion construct reduced pep25-dependent GUS activity levels to about 33% of that of construct 3D, while still allowing > 50-fold induction levels (3Dmut 1 and 3Dmut2). Mutation of all four W boxes in this construct, however, significantly reduced both GUS activity levels and fold inducibilty (3Dmut3). Similarly, mutations of the two remaining W boxes in construct 3E also drastically reduced pep25-dependent GUS activity levels to about 8% of those of the corresponding construct 3E, and strongly reduced fold-induction levels (3Emut1). These analyses showed that W boxes contribute significantly to MAMP-triggered expression of the PROPEP3p:GUS reporter gene in parsley protoplasts but that additional elements are also involved. In particular, the region from −442 to −321 appears to contain such elements. We fused this promoter fragment to the 35S CaMV (Cauliflower Mosaic Virus) TATA-box region and the GUS reporter (construct MS23-3C) and tested whether it was capable of mediating pep25-dependent expression. As shown in Fig. 2(c), this was indeed the case, with overall induced GUS activity levels similar to that of construct 3Cmut2 (Fig. 2b), in which the additional downstream W boxes were mutated. Surprisingly, however, mutation of the two W boxes within this fragment (MS23-3Cmut) strongly reduced both overall GUS activity levels and fold-induction values. This differs from our results with construct 3Cmut3 (Fig. 2b), indicating that these two W boxes have a stronger contribution when this region is taken out of its original promoter context.
In a final attempt to identify additional elements contributing to the MAMP-induced PROPEP3 expression, we generated as series of deletion constructs using as a template construct 3Cmut3, in which all six W boxes were mutated (Fig. 2b). Starting at position −442, we successively deleted 10 bp from the 5′-end down to position −351 (Fig. 3). Whereas deletion of the first 10 bp (construct Del1) did not significantly affect pep25-dependent activation of the reporter gene, further deletion of 10 bp (construct Del2) reduced overall MAMP-triggered GUS activity levels to about 30% of those of 3Cmut3 (Fig. 3). Unexpectedly, additional 5′ deletions of 10 and 20 bp (Del3 and Del4) resulted in enhancement of pep25-induced GUS activity levels, indicating a potential negative element(s) within this region. Further deletions down to and including the putative D box (Del5 to Del9) still enabled low levels of MAMP-dependent GUS activity but also elevated basal expression levels (Fig. 3). In summary, although discrete elements could not be identified, these data suggest that the region between −442 and −351 contains motifs that partly contribute negatively or positively to pep25-dependent expression in the transient protoplast assay.
Functionality of the PROPEP2 and PROPEP3 promoters in planta
Although the parsley protoplast system has proved extremely useful in identifying DNA elements responding to external stimuli (Rushton et al., 2002; Koschmann et al., 2012), in planta functional validation is required. For this, we selected a set of PROPEP2 and PROPEP3 promoter reporter constructs that were most informative in the transient expression assay and stably transformed these constructs into A. thaliana Col-0 plants. The transgenic lines were tested for their ability to activate the reporter construct following infiltration of the leaves with the MAMP flg22. As a wounding control, plants were infiltrated with MgCl2 buffer. For each construct, four independent transgenic lines were selected and subsequently tested for GUS activity in whole leaf extracts by a quantitative fluorometric assay. Although the relative GUS activity values varied between individual lines, probably as a result of positional effects and copy number, the behavior of these lines in their response to flg22, measured as fold induction, was very reproducible.
Lines carrying the entire 5′ region of PROPEP2 (PROPEP2A) strongly induced the expression of the reporter gene upon flg22 treatment (Fig. 4a). A similar induction level was observed with the −348 deletion construct (PROPEP2B). Both constructs also responded to wounding but to a much lesser extent. Mutations of all W boxes within the PROPEP2B promoter (PROPEP2Bmut3) did not affect the wound response but significantly reduced flg22 inducibility without completely abolishing it (Fig. 4a). However, both flg22 and wound responsiveness were completely abolished when the as-1/OCS enhancer element was simultaneously mutated along with the W boxes (PROPEP2Bmut5). The contribution of the as-1/OCS enhancer and of the two downstream W box elements for MAMP responsiveness was also supported by the −294 deletion construct lines (PROPEP2C) and by the failure of the −163 deletion lines (PROPEP2D) to respond to this stimulus. The differential response of these transgenic lines was also apparent after GUS activity staining of the leaves (Fig. S5). The importance of the W boxes and the as-1/OCS enhancer for in vivo MAMP-dependent gene activation was further provided by transgenic plants expressing a PROPEP1-GFP fusion protein driven by either the PROPEP2B or the PROPEP2Bmut5 promoter. Flg22 treatment resulted in clearly induced GFP signal being detected in leaf epidermal cells 16–18 h post treatment of PROPEP2B:PROPEP1-GFP plants, whereas no signal above background was observed in plants harboring the PROPEP2Bmut5:PROPEP1-GFP construct (Fig. 4b). Immunoblot detection confirmed that PROPEP1-GFP protein was virtually undetectable in flg22-treated transgenic lines harboring the PROPEP2Bmut5:PROPEP1-GFP construct (Fig. S6). Flg22-induced expression of PROPEP2B:PROPEP1-GFP, but not PROPEP2Bmut5:PROPEP1-GFP, was also observed in pepr1 pepr2 receptor double mutant plants (Figs 4c, S6), revealing that transcriptional activation was dependent on the W box/as-1/OCS CRM, but not on a functional DAMP perception system.
In summary, both the in planta behavior of the selected PROPEP2p:GUS and the PROPEP2B:PROPEP1-GFP lines highlight a key role of specific W boxes in conjunction with the as-1/OCS enhancer in mediating MAMP-triggered expression of PROPEP2. Moreover, the results obtained with these lines closely mirrored the findings that were obtained in the parsley protoplast system (Fig. 1).
The importance of the PROPEP3 promoter regions downstream of −442 was also analyzed by generating stably transformed A. thaliana plants. As shown in Fig. 5, transgenic lines carrying the −442 PROPEP3:GUS construct (PROPEP3C) were highly responsive to flg22 treatment but only weakly induced upon wounding. Simultaneous mutation of all W boxes in the promoter region (PROPEP3Cmut3) strongly reduced MAMP-triggered expression. These findings partly differ from those of the transient expression studies (Fig. 2) where overall levels of GUS activity were also substantially reduced but where fold inducibility upon MAMP treatment remained high. Expression of the PROPEP3Cmut3-Del4 construct was similar to that of PROPEP3Cmut3, while expression of the PROPEP3Cmut3-Del7 transgenic lines was no longer responsive to flg22 (Fig. 5). Thus, flg22 responsiveness in 4-wk-old mature A. thaliana leaves appears to be mainly mediated via W boxes, whereas the region −442 to −372, which still allowed MAMP inducibility in the protoplast system (Fig. 3), appears not to be essential. The importance of the W boxes for flg22 responsiveness was further substantiated by results for the −321 PROPEP3 deletion lines that carried the construct with intact W boxes (PROPEP3D), and the nonresponsive lines PROPEP3Dmut3, in which all W boxes were mutated (Fig. 5). The region −442 to −321, which still showed MAMP responsiveness in the protoplast system when taken out of its promoter context and fused to the 35S CaMV TATA-box region and the GUS reporter (construct MS23-3C in Fig. 3c), was not significantly induced by flg22 in adult leaves of the transgenic lines (Figs 5, S7). Interestingly, however, these MS23-3C transgenic lines showed 6- to 9-fold flg22 responsiveness and relatively strong GUS activity levels in 10–14-d-old A. thaliana seedlings (Fig. S8). Moreover, this MAMP responsiveness in seedlings was mainly mediated by the two W boxes, as mutations within these sites (MS23-3Cmut) completely abolished inducibility (Fig. S8).
PROPEP2 and PROPEP3 are direct targets of WRKY33
Our promoter analyses revealed that W boxes are mediators of MAMP-induced expression of the PROPEP2 and PROPEP3 genes. This strongly suggests that WRKY transcription factors are key regulators of these genes. Microarray studies showed that expression of PROPEP2, and in particular PROPEP3 and WRKY33, is strongly induced by MAMP treatment, upon activation of DAMP signaling, and upon infection with B. cinerea (Zipfel et al., 2006; Yamaguchi et al., 2010; Birkenbihl et al., 2012). WRKY33 was also shown to be required for resistance to this fungal pathogen (Zheng et al., 2006). Moreover, as revealed by the differential response of the PROPEP2 and PROPEP3 promoter GUS reporter lines, the same W boxes appear to at least partly mediate the response to this necrotrophic fungus (Figs S9, S10). We therefore tested whether the promoters of PROPEP2 and PROPEP3 are direct in vivo targets for WRKY33. We performed chromatin immunoprecipitation (ChIP qPCR) experiments on wrky33 mutants (w33) and on a wrky33 line complemented with a WRKY33-HA construct under control of its native promoter (33HA; Birkenbihl et al., 2012). For ChIP, seedlings of the respective lines were treated for 2 h with flg22 before formaldehyde cross-linking. Using an antibody against the HA epitope we detected clear enrichment of WRKY33 at both PROPEP gene promoters (Figs 6a,b, S11). The positions of the enriched qPCR amplicons within the gene loci, indicative of WRKY33 binding, are marked in Fig. 6. Binding of WRKY33 at both promoters was dependent on flg22 treatment. Most significant WRKY33 binding was observed at the promoter regions of PROPEP2 and PROPEP3 containing the identified functionally important W boxes (Fig. 6a,b). The higher enrichment values for WRKY33 binding sites at the PROPEP3 locus correlate well with the stronger induced expression of PROPEP3 upon MAMP treatment (Zipfel et al., 2006). To assess the consequences of WRKY33 binding to these promoters, we analyzed expression of PROPEP2 and PROPEP3 in a previously described wrky33 mutant (Birkenbihl et al., 2012). As illustrated in Fig. 6(c,d), expression of these two genes was significantly reduced to about 50% of the levels observed in MAMP-treated WT plants. These data demonstrate that WRKY33 directly targets both genes in a MAMP-dependent manner and thereby positively contributes to the enhanced expression of these genes during MTI.
Identification of putative CREs by phylogenetic shadowing
The possible evolutionary conservation of the identified A. thaliana promoter regions containing important MAMP-responsive DNA elements in mediating PROPEP gene expression was also addressed by phylogenetic shadowing (Boffelli et al., 2003). This method takes into account the assumption that, during evolution, noncoding DNA sequences of orthologous genes rapidly diverge except for regions containing functionally important CRMs. We assessed the degree of conservation through a comparison of putative PROPEP orthologous regions between the related plant species A. thaliana, A. lyrata, A. alpina and B. rapa. Whereas it was not possible to clearly identify a PROPEP2 locus in any of the species other than A. thaliana, we did obtain orthologs for PROPEP3 from all three A. thaliana relatives. We then assessed the degree of conservation within the 500-bp upstream regions (relative to the translation start sites) of these orthologous loci using PhastCons (Siepel et al., 2005; Fig. 7). Although the 500-bp upstream region may not contain the full set of CREs impacting on the expression of PROPEP3 and its orthologs, it focuses on the regions that we found to have the largest impact on PROPEP3 expression. The degree of conservation in the promoter was found to be highly variable, with multiple short regions having a high degree of sequence conservation among the four species. To identify functionally important CREs within these conserved regions, we performed a search for functional elements using meme (Bailey et al., 2006). meme identified three highly significant motifs. The motif representing the W box was the most significant CRE, with an e-value of 3.1e-007. Annotation of the W box and the other elements in the promoter sequences revealed conserved domains that encompass the DNA elements that were functionally required for MAMP-induced gene expression in our transient and transgenic studies. Three important pairs of W boxes within the −168 to −103, −299 to −240, and −369 to −327 bp regions of the A. thaliana PROPEP3 promoter became apparent. Interestingly, the approximate distance and opposite directionality of the W boxes are maintained in the other plants (Fig. 7).
In addition, we identified two unknown motifs. The consensus sequence TTC/TAACAT appears three times in the promoter sequence of A. thaliana (blue element in Fig. 7). All three instances of this element are conserved in location and orientation in A. lyrata, and one such motif is conserved in all four species. This novel putative CRE resides in a short region of local conservation, as highlighted by the PhastCons scores. The other more degenerate motif (CG/ATC/TAC/AC/AAAC/A/G; green in Fig. 7) is found several times in all four species at similar locations.
In summary, de novo motif identification supports the hypothesis that W boxes, very probably in combination with other CREs, play a major role also in regulating the orthologous PROPEP3 genes in other Brassicaceae plants.
As in animals, host peptides derived from tissue damage act in plants as danger signals to activate a DAMP signaling cascade via dedicated plasma membrane-localized receptors. Rapid initiation of DAMP signaling is potentially important in triggering spatial and temporal enhancement of plant immunity (Yamaguchi & Huffaker, 2011). In A. thaliana, two key genes, PROPEP2 and PROPEP3, coding for such danger signals are transcriptionally activated upon treatment of cells with various MAMPS, including flg22 and elf18. By functionally dissecting the promoters of these two A. thaliana PROPEP genes, we identified critical CRMs that mediate MAMP-dependent gene expression. Our analyses showed that WRKY transcription factor binding sites (W box elements) are key components of the CRMs within the regulatory regions of both genes. Moreover, we could confirm direct in vivo MAMP-inducible binding of WRKY33 to such regulatory regions.
The role of WRKY factors in controlling A. thaliana PROPEP gene expression has been suggested based on the presence of several W boxes within their promoters (Yamaguchi et al., 2010). Although such in silico predictions are informative, they cannot address the specific contributions of the individual elements to a particular response and cannot predict with which other elements they will act in combination to mediate specific outputs. Moreover, efficient binding of certain WRKY factors to native W boxes requires additional specific nucleotide bases 5′ of the element (Ciolkowski et al., 2008).
In the case of PROPEP2, W boxes and an as-1/OCS enhancer constitute the key CRM required for high-level MAMP-induced expression. Based on these observations one can conclude that PROPEP2 promoter activity is mediated by a concerted action of WRKY and TGA factors upon MAMP perception. Both W box elements upstream of the as-1/OCS enhancer and the two W boxes downstream thereof strongly affected induced expression levels. However, analysis of the transgenic plants harboring the various PROPEP2:GUS reporter constructs suggests that W boxes within the −348 to −294 promoter region contribute to a greater extent, as deletion of this promoter region reduced flg22 inducibility by > 60% (Fig. 4a). Interestingly, the opposite was observed in the transient assays employing parsley protoplasts. Here, the two most proximal W boxes and the as-1/OCS enhancer were mainly responsible for high MAMP-induced expression, whereas the upstream W boxes contributed to a lesser extent (Fig. 1). Although employment of the transient expression system proved highly successful in delineating the important CRM within this gene promoter, such differences may be attributable to the dissimilarities of the assays, the plant cell material used, or differences in the magnitude of responses triggered by pep25 and flg22. In parsley cells, pep25 triggers strong and robust MAMP responses while flagellin (flg13) induces weaker early responses (Blume et al., 2000). By contrast, pep25 does not function in A. thaliana.
Similar to our observations for the A. thaliana PROPEP2, a distinct WRKY binding site and a closely adjacent as-1 site also form a CRM that positively modulates expression of Pathogenesis Related-1a (PR-1a) in tobacco (Nicotiana tabacum) upon treatment with salicylic acid and bacterial elicitors by binding NtWRKY12 and TGA1a (van Verk et al., 2008). In the case of A. thaliana PR-1, salicylic acid-induced expression appears to be achieved by the action of TGA factors binding to two as-1 sites in combination with both negatively and positively acting WRKY factors binding to distinct W boxes (Pape et al., 2010). Two W boxes and two as-1-like motifs within the rice (Oryza sativa) OsNPR1 gene promoter were also demonstrated to act as a CRM in mediating salicylic acid-responsive expression, again implicating WRKY and TGA factors in its regulation (Hwang & Hwang, 2010). The interdependence of distinct cis-acing DNA elements in mediating a unique promoter response was nicely demonstrated for the A. thaliana Cytochrome P-450 gene 81D11 (CYP81D11) by Köster et al. (2012). They showed that jasmonic acid induction of CYP81D11 requires both the presence of the TGA factor binding site as-1 and a G box that is bound by the basic Helix-Loop-Helix (bHLH) transcription factor MYELOCYTOMATOSE ONCOGENE-like2 (MYC2), and that these motifs act as one nonseparable unit.
We were also able to demonstrate the importance of functional W boxes within the PROPEP3 promoter in mediating MAMP-induced gene expression. In this case, however, we were unable to define additional CREs that may constitute the CRM. Our transient expression studies clearly showed that sequences upstream of −442 are dispensable for strong MAMP-induced expression. The six W boxes downstream of −442 all appear to contribute to the observed level of MAMP-induced expression (Fig. 2). Simultaneous mutations of all these six W boxes significantly reduced MAMP-triggered expression but did not completely abolish it (Fig. 2; PROPEP3Cmut3). These findings were mostly consistent with those obtained in the A. thaliana transgenic lines harboring the same reporter constructs (Fig. 5). However, in leaves of adult 4–5-wk-old plants, the six W boxes appear to constitute the major CRM as they contributed to > 90% of the MAMP-induced levels of the reporter gene (compare construct PROPEP3C with PROPEP3Cmut3 in Fig. 5). These findings are similar to those obtained for the A. thaliana resistance gene Resistance to Peronospora Parasitica8 (RPP8) whose induced expression upon downy mildew challenge and subsequent pathogen resistance was mainly dependent on three W boxes (Mohr et al., 2010). Moreover, the presence of the W boxes within conserved regions of the PROPEP3 orthologs from other Brassicaceae plants lends additional support to a vital role of WRKY factors in modulating the transcriptional activity of this key gene associated with DAMP signaling.
Whether the −442 to −351 PROPEP3 region actually contains additional novel elements that are functional in A. thaliana remains to be further investigated. Nevertheless, it is noteworthy that partial deletion of the conserved motif TTCAACAT, revealed by phylogenetic shadowing (between positions −393 and −386; Fig. 7), had a clear impact on MAMP-induced expression of PROPEP3 (compare PROPEP3Cmut3-Del4 and PROPEP3Cmut3-Del5 in Fig. 3).
As is the case for PROPEP2 and PROPEP3, closely adjacent W boxes, often in reverse orientation, are frequently found in stress-responsive gene promoters and in some instances have been shown to synergistically affect gene expression (Du & Chen, 2000; Mohr et al., 2010; Rushton et al., 2010). Close proximity of CRMs alone, however, is insufficient to explain synergistic transcription activation by bound factors, as this also appears to require stereo-specific positioning of the activators (Huang et al., 2012). Thus, it remains to be tested whether the close proximity of the individual W boxes to each other, or the W boxes to the as-1/OCS enhancer within the promoters, can actually activate PROPEP3 and PROPEP2 transcription more effectively.
We would like to note that all functionally relevant constructs tested in plants responded not only to flg22 but also in a similar manner to the MAMP elf18 (data not shown). Moreover, flg22 and elf18 were both able to activate PROPEP2B-driven expression of PROPEP1-GFP in the pepr1 pepr2 receptor double mutant, demonstrating that MAMP activation of the PROPEP genes does not require initial activation of the DAMP signaling pathway via receptor binding. In addition, most of the tested PROPEP promoter variants tested impacted wound-induced reporter gene expression in a qualitatively similar manner to MAMP treatment (Figs 4, 5). This is consistent with the previously observed wound-induced expression of PROPEP1, PEPR1 and PEPR2, and with the postulated role of DAMP signaling in perceiving and responding to host tissue damage (Huffaker et al., 2006; Yamaguchi et al., 2010). Although a direct role of the W boxes in the wound response was not addressed in this current study, their ability to mediate wound responsiveness has been shown previously (Rushton et al., 2002).
ChIP assays and analysis of wrky33 mutant plants clearly revealed that WRKY33 targets both the PROPEP2 and PROPEP3 promoters in a MAMP-dependent manner and positively contributes to their expression. However, other members of the WRKY family also appear to be involved in regulating these genes, as wrky33 plants still show MAMP responsiveness. Microarray studies have shown that many WRKY genes are induced by flg22/elf18 treatment, and Pep1 has been demonstrated to activate, next to WRKY33, the expression of WRKY22, WRKY29, WRKY53, and WRKY55 (Zipfel et al., 2006; Yamaguchi et al., 2010). Thus, it is conceivable that various W boxes will be bound by different WRKY factors and/or that the same W box will be associated with distinct WRKY members in a temporal manner. Moreover, despite the stereotypic binding preference of WRKY factors to their DNA, differences in binding site selectivity for some WRKY members have been observed (Ciolkowski et al., 2008). Nevertheless, such studies will remain challenging in the near future because of the current lack of highly selective antibodies against diverse A. thaliana WRKY proteins, and because of the time-consuming effort required to generate functional epitope-tagged derivatives in appropriate genetic backgrounds.
We are grateful to Dr Thorsten Nürnberger (ZMBP Tübingen) for providing seeds of the pepr1 pepr2 mutant. We thank Lydia Bollenbach for the maintenance and propagation of the cultured parsley suspension cells. This work was partly funded by an IMPRS PhD fellowship (V.R.) and by Deutsche Forschungsgemeinschaft grants SFB 670 and SO235/7-1 (I.E.S).