FORCA, a promoter element that responds to crosstalk between defense and light signaling

Background Recognition of pathogenic microorganisms triggers in plants comprehensive transcriptional reprogramming. In order to identify transcriptome-level control elements required for plant immune responses we are examining several sets of genes found by microarray experiments to be co-activated in Arabidopsis thaliana (Arabidopsis) seedlings infected with the oomycete Hyaloperonospora parasitica. Promoter motifs conserved in clusters of co-expressed genes may be involved in mediating coordinated gene activity patterns. Although numerous studies identified such conserved promoter motifs in co-expressed gene sets, reports confirming their function as regulatory elements are rare. Results FORCA is a hexameric promoter motif that is conserved in clusters of Arabidopsis genes co-expressed in response to fungal or oomycete pathogens as well as defined light treatments. FORCA is generally more frequently present in Arabidopsis promoter regions than statistically expected. It constitutively interacts in a DNA-sequence specific manner with nuclear Arabidopsis proteins. These interactions are suppressed by defense-related stimuli and enhanced by prolonged exposure to constant light. Furthermore FORCA mediates constitutive reporter gene expression in transiently transformed Nicotiana benthamiana leaves as well as in stably transformed Arabidopsis plants. Its responsiveness to defense-stimuli is modulated by the duration of light exposure. In plants grown under normal light conditions or constant darkness defense-related stimuli result in suppression of FORCA-mediated reporter gene expression, while in plants grown under constant light exposure, defense-induction results in enhanced FORCA-mediated expression. In addition, we found plants subjected to constant light exposure to exhibit reduced susceptibility to virulent H. parasitica. Conclusion We propose that FORCA is a regulatory cis-element that is present in a wide variety of Arabidopsis promoters. It integrates light- and defense-related signals and participates in adjusting the transcriptome to changes in environmental conditions.


Background
Molecular recognition of pathogenic microorganisms triggers in plants comprehensive transcriptional reprogramming. A network of defense-regulators transduces information about the attacking microbe into appropriate transcriptional responses. Two functionally distinct classes of pathogen molecules are known to elicit plant immune responses [1][2][3]. Receptor-mediated recognition of pathogen/microbe associated molecular patterns (PAMPs/MAMPs) activates PAMP-triggered immunity (PTI), while recognition of pathogen effectors by plant disease resistance (R) proteins leads to effector-triggered immunity (ETI). ETI is a strong immune response that results in incompatible plant pathogen interactions where the plant is resistant and the pathogen is avirulent. Effectors are pathogen-derived molecules (proteins or small organic molecules) that are secreted into host tissues [4]. Some effectors have been shown to suppress PTI resulting in a weakened immune response called basal defense which is usually insufficient to halt growth and spread of the respective microbe. The typical outcome in this case is a compatible interaction, where the plant is susceptible and the pathogen is virulent.
A key player in the regulation of ETI as well as basal defense is the hormone salicylic acid (SA), which accumulates in infected tissues of several plant species and triggers downstream defense responses [5]. Several other small molecules and multiple proteins involved in defense regulation have been identified by genetic and biochemical studies [6]. In particular the immune system of Arabidopsis thaliana (Arabidopsis) has been extensively examined over the past two to three decades. While our knowledge about the Arabidopsis immune system is far from complete, it has become obvious in recent years, that its individual components are functionally interconnected in a complex manner and constitute a regulatory network rather than simple linear pathways. Crosstalk between distinct branches of this network has been found to be involved in fine-tuning defense outputs and maximizing the effectiveness of immune responses [7]. For example, sophisticated mechanisms mediating crosstalk between SA and jasmonic acid (JA)-dependent signaling processes can result in synergistic or antagonistic effects depending on the levels of each of these hormones [8][9][10].
In addition, there are extensive interactions of defense signaling with regulatory processes that are not primarily involved in plant immune responses. For example, crosstalk between defense mechanisms dependent on the defense hormones SA and JA on the one hand and each of the classical growth-controlling phytohormones, auxin, gibberellin, ABA, cytokinin or brassinosteroids on the other hand have been reported [11]. Details of these interactions are not well understood yet, but it seems that some of them benefit the host by enhancing immune responses, while others are beneficial for the pathogen by suppressing defense reactions. In addition, several labs reported interferences between light and defense signaling. Depending on the type of light signaling pathway involved, these interactions were found to be synergistic or antagonistic. For example, phytochrome-mediated light signaling can positively affect SA-dependent defense responses [12,13], while negative interferences were observed between defense and UV-light signaling [14]. Generally, these interactions seem to control resource allocation in the plant ensuring the maintenance of homeostasis and the adjustment of the host's metabolism to a state optimal to cope with the respective environmental challenges.
We are using interactions between the pathogenic oomycete Hyaloperonospora parasitica (Hp) and Arabidopsis to examine regulatory mechanisms operating at the interface between defense signaling and the regulation of the defense transcriptome [15][16][17]. Several transcriptional regulators have been implicated in these processes, such as NPR1, a nuclear transported transcriptional cofactor that acts downstream from SA and interacts with TGA-bZIP transcription factors [18]. Additional transcription factors, including WRKYs and ERFs, participate in the regulation of the plant defense transcriptome and disease resistance [19]. The DNA binding site preferences of these transcription factors have been well characterized. Typically TGA-bZIPs bind to TGA boxes (TGACG), WRKYs to W boxes (TTGACC/T) and ERFs to GCC boxes (AGCCGCC) [19]. These promoter sites have been shown in numerous studies to act as pathogen-responsive cis-elements mediating expression of individual defense-associated genes [19].
An enormous amount of global gene expression data from large-scale transcript profiling projects has accumulated during the past decade. These data allowed for the identification of gene clusters that are co-expressed under certain biological conditions and, hence, are likely to be subject to co-regulation by common mechanisms. Several studies revealed conserved sequence motifs in promoters of such co-regulated genes which may act as cis-elements mediating their coordinated activity [20,21]. In addition to already known cis-elements, some novel promoter motifs were found to be statistically enriched in coexpressed genes [15,22,23]. While in some cases, their function has been experimentally proven [22,23], reports confirming that such conserved promoter motifs are binding sites of nuclear proteins that can affect gene expression are still rare.
Here we report on the identification and functional characterization of a motif conserved in promoters of the FORC cluster, a set of Arabidopsis genes co-expressed in response to recognition of Hp as well as other oomycete and fungal pathogens. We found this motif, which we termed FORC A , to exhibit DNA sequence specific and differential interactions with nuclear Arabidopsis proteins. Furthermore, electrophoretic mobility shift assays as well as reporter gene assays showed that FORC A integrates stimuli related to defense-and light signaling. FORC A acts as a light-responsive enhancer, the activity of which is modulated by pathogen-perception. Under normal light conditions as well as constant darkness, FORC A activity is reduced by recognition of Hp as well as application of SA, while under constant light conditions its activity is enhanced by defense induction. Consistent with the observed functional connection between light and defense signaling, we observed that plants kept under constant light exposure exhibited reduced susceptibility to a virulent Hp isolate.

A cluster of Arabidopsis genes co-expressed during interactions with oomycete and fungal pathogens is enriched for the FORC A promoter motif
Using Affymetrix microarrays we previously identified clusters of genes that are co-expressed during Arabidopsis immune responses [15]. Two of these clusters, cluster I and the LURP cluster (designated as cluster II in [15]), were defined by distinct patterns of coordinated transcriptional up-regulation during incompatible and compatible interactions with the Hp isolates HpEmoy2, HpHiks1 and HpEmco5. While LURP transcripts exhibited a kinetic pattern of late and sustained up-regulation accumulating strongly between 12 and 48 hpi [16,24], cluster I transcript levels tended to increase earlier and predominantly accumulated within the first 12 h after infection [15]. Genetic disruption of SA-signaling and R protein-mediated Hp recognition strongly attenuated or delayed the response of LURP genes, but had only a minor effect on cluster I responses. Inspecting publicly accessible microarray data (Botany Array Resource) [25] we found that cluster I genes also exhibit a pronounced pattern of coactivation during interactions of Arabidopsis with the fungal pathogen Bortytis cinerea as well as a second oomycete, Phythophthora infestans (see Additional File 1). Therefore we renamed cluster I to FORC (Fungal and Oomycete Pathogen Response Cluster). Although LURP genes are also coordinately activated in response to the oomycete Phytohthora infestans, this set does not show an uniform response to fungal pathogens (not shown).
Tightly co-expressed subsets of both the FORC and LURP genes were found to be enriched for distinct promoter motifs, such as potential binding sites of WRKY transcription factors [15]. This suggested that members of the FORC cluster on the one hand and the LURP cluster on the other hand are controlled by common regulatory mechanisms. The fact that different motifs are conserved in the promoters of each of these two gene sets suggested that the FORC genes are controlled by mechanisms distinct from those controlling the LURP genes. Applying AlignACE [26] to the 1000 bp-upstream sequences of a strictly co-expressed subset of FORC genes (Pearson correlation of 0.95 to the average pattern of the FORC set; [15]) we found the hexameric motif T/ATGGGC to be significantly enriched compared to its statistically expected frequency (p = 3.0 E-4; see under "Methods"). We termed this motif FORC A .
FORC A specifically interacts with a nuclear DNA binding activity that is down-regulated by Hp and SA T/ATGGGC-containing promoter motifs have not been described as pathogen-response elements. Therefore we tested by electrophoretic mobility shift assays (EMSAs), if FORC A exhibits Hp-dependent differential interactions with nuclear Arabidopsis proteins ( Figure 1). Using the radioactively labeled FORC A -1 probe consisting of a single copy of the FORC A consensus sequence with the TTGGGC core and 8 bp of arbitrary flanking sequence we detected a constitutive DNA-binding activity producing a single band shift that is out-competed by unlabeled FORC A -1 probe ( Figures 1A-C). Surprisingly, this FORC A binding activity was down-regulated in response to infections with the avirulent Hp isolates HpHiks1 or HpEmoy2 as well as exogenously applied SA. Furthermore, we found FORC A / nuclear protein interactions to be light dependent. Nuclear extracts from plants kept for three days under constant light produced substantially more intense FORC A -shifts than those from plants kept for the same time in constant darkness (CL & CD in Figure 1B). Besides unlabeled FORC A -1 probe, unlabeled competitor probes containing wild-type FORC A permutations present in several FORC promoters (FORC A 2-6) clearly reduced the intensity of FORC A -1 shifts (Figures 1C &1D). Mutations in the sequences flanking the invariant TGGGC core of FORC A -3 did not result in reduced competition. However, the FORC A -3 mutA competitor probe, that lacks the TGGGC core, as well as competitor probes with sequences unrelated to FORC A (M5-2, M8-1) did not efficiently reduce the intensity of FORC A -1 band shifts. These results showed that the observed FORC A -binding activity is DNAsequence specific and that only the TGGGC sequence of FORC A and not sequences flanking this invariant core are critical for this interaction.

FORC A is a light-responsive promoter element whose activity is modulated by defense-related stimuli
To test effects of FORC A on transcriptional regulation, we fused a FORC A trimer consisting of three distinct FORC A permutations to the minimal CaMV35S promoter followed by GUS in pCambia1281X (3xFORC A -pCAMBIA). Untreated Nicotiana benthamiana leaves transiently transformed with 3xFORC A -pCAMBIA by Agrobacterium tumefaciens exhibited strong GUS activity after histochemical staining or quantitative GUS activity assays ( Figure 2A). Treatment of transformed N. benthamiana leaves with 1 mM SA clearly reduced this activity. With a construct containing a trimer of a motif unrelated to FORC A (3xM5-pCAMBIA) we observed no significant reporter gene activ-ity, while a trimer of a second unrelated motif, M6, mediated a strong increase of GUS activity upon SA treatment (see Additional File 2). Formally the SA-triggered reduction of 3xFORC A -pCAMBIA expression in our agro-transient assays may be due to effects of SA on the viability of Agrobacterium or other effects unrelated to FORC A -mediated gene expression. However, we believe that this is quite unlikely for the following reasons. (1) we also observed a SA-triggered reduction of 3xFORC A -pCAMBIA expression in stably transformed Arabidopsis lines (see below); (2) motif M6 mediated SA-inducibility of GUS expression in our agro-transient assays; (3) treatment with 1 mM SA did not reduce GUS expression mediated by the constitutive CaMV35S promoter in previously published agro-transient assays with tobacco leaves [27,28].
pCAMBIA1281X contains a copy of the full CaMV35S promoter close to its multiple cloning site. In order to rule out that this promoter influences the expression pattern observed with 3xFORC A -pCAMBIA, we constructed a second pair of reporter gene vectors using pBI101 (Clontech), which lacks any functional promoters in the vicinity of its multiple cloning site. The construct 3xFORC A -pBI101 contains a FORC A trimer fused to the minimal CaMV35S promoter followed by GUS. 3xFORC A -mut-pBI101 is identical to 3xFORC A -pBI101, but contains block mutations eliminating the TGGGC core motif of each FORC A copy. Consistent with the results observed with the pCAMBIA versions, 3xFORC A -pBI101 exhibited in agro-transient expression assays a strong constitutive activity that is suppressed by SA, while 3xFORC A -mut- Figure 1 In vitro-interactions of FORC A with nuclear Arabidopsis proteins. A, B: Interactions of FORC A with nuclear protein extracts from Arabidopsis Col-0 seedlings grown under normal light conditions and left untreated (UN) or treated with salicylic acid (SA), infected with the avirulent Hp isolates Hiks1 and Emoy2 or exposed for three days to continuous light (CL) or continuous darkness (CD). Panel (A) shows the whole gel for the 0 h untreated samples; panel (B) shows gel sections with the FORC A -specific band shift for all tested conditions. NE: nuclear extract, Comp: unlabeled competitor probe, SB: FORC A -specific band shift, FP: free probe.C: FORC A derived oligonucleotides used in EMSAs. The gene IDs specify the promoters from which each motif is derived. Positions representing the defining T/ATGGGC core sequences of the shown wild type FORC A permutations are highlighted by a grey box. Sequences mutated in the tested FORC A -3 permutations are underlined. D: EMSA competition analysis of FORC A -1 interactions using Arabidopsis nuclear protein extracts from Arabidopsis Col-0 seedlings exposed to CL.

In vitro-interactions of FORC A with nuclear Arabidopsis proteins
NE:

Sequence
Motif/Gene ID pBI101 did not show any detectable reporter gene activity ( Figure 2B).
We also generated transgenic Arabidopsis lines transformed with 3xFORC A -pCAMBIA. T2 progeny from two independent primary transformants were selected for the presence of the transgene and examined for GUS expression by histochemical staining with X-Gluc ( Figure 3A). Both tested lines exhibited identical GUS expression patterns. Four day-old 3xFORC A -pCAMBIA seedlings showed constitutive GUS activity in stems which is most pronounced in vascular tissues ( Figure 3A-a). Ten day-old seedlings exhibited 3xFORC A ::GUS expression in all plant tissues. In these plants, GUS staining is in particular strongly detectable in vascular tissues and the epidermal tissues surrounding the bases of trichomes ( Figure 3A normal light conditions as well as after 3 days exposure to constant darkness ( Figure 3B). However, after constant light treatment, SA triggered a strong increase of reporter gene expression in these lines. Consistent with the SAinduced suppression of FORC A -mediated GUS expression observed in 3xFORC A -pCAMBIA Arabidopsis lines under normal light conditions, GUS expression in these lines grown under normal light conditions is also down-regulated after infection with the virulent Hp isolate Noco2 ( Figures 4A, B &4E). In Col-0 plants, HpNoco2 is known to induce a basal defense response that is dependent on SA [29].
In our assays the high stability of the GUS enzyme in plant tissue [30] resulted already in strong X-Gluc-staining of 3xFORC A -pCAMBIA seedlings prior to infection with HpNoco2 (not shown). To be able to detect a reduction of this background activity, we had to incubate the seedlings for at least 7 days after spray-infection with HpNoco2 spores. At this timepoint a dense network of Hp hyphae had typically developed in cotyledon and leaf tissues. To eliminate the possibility that the reduction of GUS activity in infected tissues is due to disease associated necrosis, we stained HpNoco2-sprayed seedlings 7 dpi with trypan blue, which marks dead plant cells dark blue. As shown in Figure 4D dead plant cells are absent in HpNoco2 infected 3xFORC A -pCAMBIA seedlings 7 dpi and only trypan bluestained Hp structures are visible in these samples. This indicates that the observed reduction of GUS expression is a result of down-regulated 3xFORC A activity.
Taken together our reporter gene assays using stably transformed Arabidopsis lines, transiently transformed N. benthamiana leaves and EMSAs with nuclear Arabidopsis proteins showed that FORC A is a constitutive promoter element, the activity of which is modulated by defenseand light-associated signaling processes.

Exposure to constant light enhances basal defense to Hp
The existence of crosstalk between defense and light signaling suggested that continuous exposure of Arabidopsis seedlings to light affects the efficiency of their defense responses. Indeed, we observed that Col-0 seedlings, when subject to three days of constant light exposure, appear to exhibit reduced susceptibility to HpNoco2 (Figure 5). This effect is more pronounced and highly significant in nahG seedlings, which are compromised in defense-associated SA accumulation and, hence, are hyper-susceptible to HpNoco2 [31]. In both Col-0 and nahG seedlings constant light pre-treatment resulted in the same level of protection to HpNoco2, indicating that this effect does not require SA.
Infections with virulent Hyaloperonospora parasitica suppress FORC A -mediated transcriptional activity

Discussion
Both PTI and ETI are associated with massive transcriptional reprogramming [19,32]. Although multiple types of cis-elements and transcription factors controlling defense-related genes have been identified, the important regulatory step of activation of the plant defense transcriptome is still poorly understood. Our current knowledge about defense-associated transcriptional regulation has mainly resulted from the analysis of individual "model genes" such as the Arabidopsis PR1 and Parsley PR10 genes [33,34]. The vast amount of global transcript profiling data that has accumulated for Arabidopsis and other model plants now allows us to address the identification of such regulatory mechanisms from a new perspective enabling researchers to directly pursue the discovery of transcriptome-level control elements. Multiple studies reported the existence of statistically conserved promoter motifs in clusters of co-expressed genes [20,21,35], but only a few reports provided evidence for their roles as regulatory promoter elements as well as binding sites of nuclear proteins [22,23].
Here we provide another successful demonstration supporting the feasibility of this approach. FORC A -related motifs have not been implicated in the control of defenseassociated gene expression before. We found FORC A to DNA-sequence specifically interact with nuclear Arabidopsis proteins. These interactions are differential and suppressed by defense-related stimuli. Furthermore, our data showed FORC A to function as a constitutive promoter element, whose activity can be modulated by the defense hormone SA or recognition of Hp.
The FORC cluster constitutes a set of genes that are coordinately up-regulated in response to infections by oomycete and fungal pathogens. In addition, 19 of the 43 FORC genes represented on the Affymetrix ATH1 array are significantly up-regulated by the SA analog BTH [36]. As in Arabidopsis plants grown under a normal light regime the constitutive FORC A activity is suppressed by defense-stimuli, including SA, this element cannot be sufficient to control the defense-associated expression of FORC genes. The FORC-type expression pattern can only be the result of the combined activity of FORC A and additional cis-elements that act as positive pathogen response elements. We previously reported the enrichment of a W box-like motif and an unknown motif in the promoters of FORC genes, which may mediate the pathogen-induced up-regulation of this cluster [15]. Indeed, we found these motifs to exhibit Hp-inducible band shifts in EMSAs with nuclear Arabidopsis proteins (not shown), suggesting a positive contribution to transcriptional activity. The role of FORC A may be to limit the amplitude of FORC gene induction and/or to mediate the down-regulation of FORCs after an early peak of their  Figure 6). The bean stem element 1 (SE1), that appears to act as a strong unspecific enhancer, and the oat light-repressor element 1 (RE1), which suppresses light-driven reporter gene expression, contain sequences that perfectly match the FORC A consensus [37,38]. The positive light response element PE1 from oat contains a sequence that matches 5 of the 6 conserved positions of FORC A [38]. RE1 and PE1 are conserved in the promoters of multiple phyA genes that are regulated in a light-dependent manner. Inspecting a data set on light triggered transcriptome changes in Arabidopsis [39] we also found FORC A to be conserved in the promoters of phytochrome A&B-dependently light-induced genes encoding proteins related to photosynthesis or localized to the chloroplast. In these 26 promoters the FORC A hexamer is significantly enriched compared to its statistically expected frequency (p = 5.82 E-9). If FORC A contributes to the co-regulation of these genes remains to be shown. In addition, a motif that perfectly matches the FORC A consensus was found to be conserved in Arabidopsis core promoters [40]. A function has not been assigned to this motif, but its preferential location at the distal periphery of core promoter regions may suggest a role distinct from that of conventional core promoter elements. FORC A does also perfectly match the recently identified "protein box" (PBX) which confers phase-specific diurnal and circadian reporter gene expression in Arabidopsis [41]. Promoters that contain PBX are enriched for gene ontology annotations related to protein synthesis.
Taken together, these reports suggest that FORC A /PBXrelated cis-elements are conserved throughout the plant kingdom and are generally involved in adjusting the transcriptome to daily changes in light-related (and possibly other) environmental conditions. Furthermore, these ciselements appear to participate in the regulation of a diverse set of genes involved in crucial metabolic processes such as photosynthesis and protein biosynthesis, but also defense. Consistent with such a broad role we observed a 2.9-fold higher frequency of the FORC A hexamer in Arabidopsis intergenic sequences than statistically expected. On average, FORC A is present 0.78 times per 1 kb in all 1 kb sequences upstream from the 33,282 loci of the TAIR8 Arabidopsis genome annotation http:// www.Arabidopsis.org, while only 0.27 occurrences of this hexamer per 1 kb are statistically expected in Arabidopsis intergenic regions.
It is presently unclear, what biological purpose it serves that defense-related stimuli reduce FORC A activity under normal light conditions or constant darkness, but enhance its activity under constant light exposure. This seemingly unorthodox and contradictory feature of FORC A may reflect that in natural environments the probability of plant/pathogen encounters can follow complex diurnal patterns which are further influenced by transient environmental factors, such as local light conditions, temperature, humidity and the presence of leaf surface water [42]. Hence, FORC A may act in concert with other regulatory elements to generate fine-tuned gene expression outputs appropriate for a changing array of environmental conditions.
Crosstalk between light and defense signaling has been described in several earlier reports and high light fluence rates or long durations of light exposure have been positively correlated with plant defense reactions [12,13]. For example, Genoud et al. reported interactions between phytochrome-mediated light responses and SA-triggered responses in Arabidopsis [12]. SA-dependent responses such as expression of the defense marker PR1 and resistance to avirulent Pseudomonas syringae bacteria were found to be triggered by exposure to high fluence rates of white light and to require a functional phytochrome signaling apparatus. It is unclear, however, if the functional characteristics of FORC A that we observed are related to the effects described by Gernoud et al., as different lightrelated parameters were tested in each study. While Gernound et al. varied the fluence rate of white light, we varied the length of the photoperiod. Multiple lesion mimic mutants also point to interactions between light and defense signaling, as they are known to exhibit spontaneous defense responses, such as elevated PR gene expression, upon exposure to high light [43,44]. As a possible connection between light and defense signaling FORC A is related to multiple described promoter elements Figure 6 FORC A is related to multiple described promoter elements. Known or hypothetical promoter elements with sequence similarities to FORC A are aligned. Positions matching the FORC A consensus motif are highlighted in grey.

FORC
A consensus: TTGGGC A in these mutants, light-driven accumulation of reactive oxygen intermediates (ROI), such as superoxide and hydrogenperoxide has been discussed [42,45]. ROI act as an early defense signal and participate in feedback mechanisms modulating SA-dependent defense responses [46,47]. Alternatively, such crosstalk may not be mediated by discrete signaling steps at all and may be merely a consequence of general disturbances of a highly complex plant signaling network [48]. At this point our knowledge about the architecture and dynamics of regulatory networks in plants is only rudimentary. Emerging new methodologies of computational biology, functional genomics and systems biology will allow us to shift the focus from understanding individual signaling circuits to a more holistic view of regulatory processes. Having promoter elements at hand that integrate signals from multiple input channels will aid in the dissection of such complex regulatory interactions.

Conclusion
FORC A (T/A T G G G C) is a hexameric promoter motif that is conserved in clusters of Arabidopsis genes coexpressed in response to fungal or oomycete pathogens as well as defined light treatments. It constitutively interacts in a DNA-sequence specific manner with nuclear Arabidopsis proteins. These interactions are suppressed by defense-related stimuli and enhanced by prolonged exposure to constant light. Furthermore FORC A mediates constitutive reporter gene expression in transiently transformed N. benthamiana leaves as well as in stably transformed Arabidopsis plants. Its responsiveness to defense-stimuli is modulated by the duration of light exposure. We propose that FORC A integrates light and defense related signals and participates in adjusting the transcriptome to daily changes in light-related environmental conditions.

Plants material and treatments
Arabidopsis ecotype Col-0, transgenic Col-0 nahG plants [31] and Nicotiana benthamiana plants were grown on soil under fluorescent lights (21°C, 100 E m -2 sec -1 ). Normal light cycles were 14 h day/10 h night. For some experiments, seedlings grown under normal conditions were exposed for additional three days to continuous light or dark treatment. Hyaloperonospora parasitica (Hp) was grown, propagated and applied to Arabidopsis as previously described [49]. Two week-old seedlings were sprayinoculated with Hp spore suspensions (6 × 10 4 to 1 × 10 5 spores ml -1 of water for HpHiks1 and HpEmoy2 and 2 × 10 4 spores ml -1 for HpNoco2) with Preval paint sprayers. HpHiks1, HpEmoy2 and HpNoco2 growth was determined 7 dpi by trypan blue tissue staining or visual sporangiophore counts [49]. SA was applied to Arabidopsis and N. benthamiana at a concentration of 1 mM in 0.05% EtOH using Preval sprayers.

Preparation of nuclear proteins and electrophoretic mobility shift assays
Nuclear protein extractions were performed as previously described [50] from 15 g of Arabidopsis whole seedlings that were untreated or pre-treated as described in the legend of Figure 1. Bio-Rad Bradford assays were used to determine protein concentrations. EMSAs were performed using the synthetic oligonucleotides (invitrogen) listed in Figure 1C. Fifty picomoles of double-stranded oligonucleotides were radio-labeled using T4 polynucleotide kinase (NEB) and γ 32 P-ATP. 2.5 picomole G50-purified radioactively labeled double stranded probe (10000 to 30000 cpm) was incubated with 15 μg of nuclear protein in a total volume of 30 μl of binding buffer (20 mM Hepes-KOH pH 7.9; 1.5 mM MgCl 2; 0.2 mM EDTA; 200 mM NaCl; 1 mM DTT) for 20 min at room temperature. For competition experiments, 100-fold molar excess of unlabeled competitor was added to the binding reaction. The EMSA reactions were subjected to electrophoresis on 5.4% non-denaturing polyacrylamide gels in 0.5 × tris borate EDTA buffer (TBE) at room temperature. The gels were vacuum dried and autoradiographed for 5 days at -80°C on HyBlot CL autoradiography film (Denville Scientific).

N. benthamiana transient expression assays
Transient expression of 3xFORC A -pCAMBIA, 3xM5-pCAMBIA in Nicotiana benthamiana was based on the protocol established by Popescu et al. [52]. Plasmids 3xFORC A -pCAMBIA, 3xM5-pCAMBIA, 3xM6-pCAMBIA, 3xFORC A -pBi101, 3xFORC A -mut-pBi101 and p19 (a viral RNA silencing suppressor from the tomato bushy stunt virus) were first transformed into A. tumefaciens GV3101 by electroporation. One transformed colony (verified by PCR) for each plasmid was cultured at 28°C until the A 600 reached 1 to 1.5. The bacteria were centrifuged for 15 min at 2,000 g, resuspended in infiltration media (10 mM MES, 10 mM MgCl 2 , 200 μM acetosyringone) to a final A 600 of 1.5, and incubated for 3 to 4 h at room temperature. Bacteria were then washed with 1 volume of fresh infiltration media, and resuspended as described above. A mix of a 1:1 ratio (final OD of 0.75) between Agrobacteria containing 3xFORC A -pCAMBIA, 3xM5-pCAMBIA, 3xM6-pCAMBIA, 3xFORC A -pBi101, 3xFORC A -mut-pBi101 and those containing p19 were used for infiltration. The infiltration of bacterial mixtures was performed with a 1-ml syringe on the abaxial side of leaves of 3-week-old N. benthamiana plants. After 48 h, mock (EtOH 0.05%) or SA treatments were applied and leaves either shock frozen in liquid nitrogen or xgluc stained (see below).

Generation of transgenic Arabidopsis lines
35Smin-pCambia and 3xFORC A -pCAMBIA was introduced in the A. tumefaciens strain AGLO2 by electroporation [53]. Agrobacterium-mediated transformation of Col-0 (T 0 ) was performed by the floral-dip method [54]. 35Smin-pCambia and 3xFORC A -pCAMBIA transgenic plants were selected on 0.5 MS/.0.8% agar media containing 50 mg l -1 hygromycinin in the dark for 1 week and then transferred to soil. T2 lines were used for GUS staining and fluorometric analysis.

Fluorometric analysis of GUS activity
Protein was extracted from 15-20 Arabidopsis seedlings or 5-6 N. benthamiana leaves in MUG extraction buffer (50 mm sodium phosphate (pH 7.0), 10 mm EDTA (pH 8.0), 0.1% SDS and 0.1% Triton X-100). Ten microliters of crude extract was added to 290 μl of reaction mix containing 10 mM 4-MUG (Gold Biotechnology, St. Louis, MO) in MUG extraction buffer. Reactions were stopped at 15-min intervals by adding 50 μl of MUG reaction mixture to 200 μl 1 M sodium carbonate. Reactions were performed in 96-well plates at 37°C. Fluorescence was measured at an excitation wavelength of 365 nm and emission wavelength 455 nm using Wallac 1420-012 Multilabel Counter (PerkinElmer Life Science). Protein concentrations in extracts were determined using Bio-Rad protein assays following the manufacturer's instructions (Bio-Rad Laboratories). The GUS activity was expressed as picomoles of 4-methyl umbelliferone per milligram of protein per minute. Final values were adjusted to eliminate background fluorescence by subtracting the average GUS activity measured from non-βglucuronidase (uidA, GUS) expressing Arabidopsis seedlings. For N. benthamiana, GUS activity values are expressed as means of 5-10 samples each containing 5-6 leaves for three experimental repetitions. For GUS expressing transgenic Arabidopsis lines GUS activity values are means of 5-10 samples each containing 15-20 seedlings from two independent lines for three independent experimental repetitions.

Statistical analyses of conserved promoter motifs
P-values describing the enrichment of defined sequence motifs in promoter sets were calculated using the Poisson distribution as described previously [15]. However, due to the general conservation of FORC A in a large number of Arabidopsis promoters, the expected frequency of this motif was calculated based on theoretical assumptions and not based on its actual occurrence in all Arabidopsis promoters as previously [15]. Arabidopsis intergenic regions contain approximately 16% "Cs", 16% "Gs", 32%"As" and 32% "Ts". Based on this the expected frequency of the FORC A hexamer (A/T T G G G C) approximately equals 0.64 × 0.32 × 0.16 4 × 2 × 1000 = 0.27 occurrences on both strands of 1 kb of random intergenic DNA.