|Home | About | Journals | Submit | Contact Us | Français|
We carried out transcriptional profiling analysis in 10 day-old Arabidopsis thaliana seedlings treated with oligogalacturonides (OGs), oligosaccharides derived from the plant cell wall, or the bacterial flagellin peptide Flg22, general elicitors of the basal defense response in plants. Although detected by different receptors, both OGs and Flg22 trigger a fast and transient response that is both similar and comprehensive, and characterized by activation of early stages of multiple defense signaling pathways, particularly JA-associated processes. However, the response to Flg22 is stronger in both the number of genes differentially expressed and the amplitude of change. The magnitude of induction of individual genes is in both cases dose dependent, but even at very high concentrations, OGs do not induce a response that is as comprehensive as that seen with Flg22. While high doses of either microbe-associated molecular pattern (MAMP) elicit a late response that includes activation of senescence processes, SA-dependent secretory pathway genes and PR1 expression are substantially induced only by Flg22. These results suggest a lower threshold for activation of early responses than for sustained or SA-mediated late defenses. Expression patterns of aminocyclopropane-carboxylate synthase genes also implicate ethylene biosynthesis in regulation of the late innate immune response.
Plant recognition of potential pathogens activates an intricate network of signal transduction pathways leading to metabolic reprogramming and production of an array of antimicrobial compounds. Not all pathways are activated by or effective against all pathogens, and the plant's response often displays some degree of specificity toward particular classes of pathogens (De Vos et al., 2005; Glazebrook, 2005). Defense modules mediated by the signaling molecule salicylic acid (SA), such as production of pathogenesis-related protein 1 (PR1) and activation of a defense-related senescence program, are associated with resistance to biotrophic pathogens, whereas necrotrophic pathogens are more effectively resisted by those components of the defensive arsenal that are regulated via jasmonic acid (JA) and ethylene (Et) (Glazebrook, 2005; Thatcher et al., 2005; Wiermer et al., 2005).
Early detection of potential pathogens occurs via recognition of Microbe Associated Molecular Patterns (MAMPs), such as bacterial flagellin, by transmembrane pattern recognition receptors (Jones and Dangl, 2006) that triggers a suite of immune responses that limit pathogen growth and damage to the host (Aziz et al., 2004; Zipfel et al., 2004; Ferrari et al., 2007). The relationship between MAMP-mediated defenses and SA-, JA- and Et-mediated responses has been unclear. Several lines of evidence suggest that MAMPs stimulate defense pathways that are independent of SA, JA, and Et. Resistance against Pseudomonas syringae or Botrytis cinerea can be induced by pre-treatment of Arabidopsis with Flg22 or OGs, respectively, independently of SA, JA, or Et (Zipfel et al., 2004; Ferrari et al., 2007). Consistent with these reports, induction by elicitors of specific defense-related genes has been demonstrated to be independent of SA, JA, or Et (Zhang et al., 2002; Ferrari et al., 2003; Ferrari et al., 2007). Conversely, MAMPs have also been reported to stimulate JA and ethylene production (Doares et al., 1995; Simpson et al., 1998; Kunze et al., 2004), as well as up-regulation of genes encoding proteins involved in the biosynthesis of JA and Et (Moscatiello et al., 2006) or pathogenesis-related proteins linked to SA-mediated responses (Gomez-Gomez et al., 1999). These data suggest that, in addition to innate immune responses that are activated independently of defense hormone signaling, MAMPs may also stimulate defense hormone-mediated effects.
From transcriptional profiles of elicitor-treated plant tissue or plants infected with effector-deficient pathogens, an understanding of the basis of MAMP-triggered immunity is beginning to emerge (de Torres et al., 2003; Navarro et al., 2004; Zipfel et al., 2004; Ramonell et al., 2005; Bae et al., 2006; Moscatiello et al., 2006; Qutob et al., 2006; Truman et al., 2006; Zipfel et al., 2006). Commonly, MAMP- induced early genes (within one hour) are functionally enriched for ones encoding enzymes for the synthesis of antimicrobial compounds and for proteins involved in signal perception and transduction, including receptor-like kinases, transcription regulatory factors, kinases, and phosphatases (Navarro et al., 2004; Zipfel et al., 2004; Moscatiello et al., 2006; Zipfel et al., 2006). Importantly, considerable overlap has been found in the responses to different elicitors (Qutob et al., 2006; Thilmony et al., 2006; Zipfel et al., 2006; Ferrari et al., 2007), suggesting that different elicitors activate conserved basal defense responses (Jones and Dangl, 2006). However, different experimental conditions (such as tissue type, environmental conditions, duration of treatment, and time of day at harvest) have made it difficult to accurately assess the degree of similarity between responses to diverse MAMPs.
In contrast, the particular suite of early signaling events, or associated kinetics and intensity, vary depending on the specific elicitor (Garcia-Brugger et al., 2006). This variability in response occurs despite the fact that many of the same signaling mechanisms are employed, such as activation of the same MAPK cascades, transient changes in concentrations of nuclear and cytoplasmic Ca2+, activation of kinases and phosphatases, accumulation of reactive oxygen species, and production of NO (Low and Merida, 1996; Yang et al., 1997; Chandra et al., 2000; Nuhse et al., 2000; Cessna and Low, 2001; Asai et al., 2002; Navazio et al., 2002; Zhang et al., 2002; Hu et al., 2004; Pedley and Martin, 2005; Garcia-Brugger et al., 2006; Qutob et al., 2006). For example, Lecourieux and colleagues compared eight different molecules that represent different classes of elicitors (six proteinaceous elicitors, including five that induce necrosis and one, Flg22, that is non-necrotic; and two oligosaccharide elicitors, including OGs) and reported that changes in Ca2+ concentration, while induced by all elicitors, varied in magnitude and timing depending on the stimulus (Lecourieux et al., 2005).
To ascertain the extent of similarity in the transcriptional changes that are induced following treatment with two elicitors which differ in source (endogenous versus exogenous) and in structural classification (carbohydrate vs proteinaceous), and which have been demonstrated to differentially stimulate early cellular changes, we have compared dynamic expression profiles of Arabidopsis seedlings treated with either OGs (which are produced upon degradation of plant cell wall pectin by pathogen-derived polygalacturonases) or Flg22 (a 22-amino acid peptide corresponding to a conserved domain of bacterial flagellin) at early and intermediate time points. The expression of several genes was assayed over an extended time-course to investigate the long-term patterns of transcriptional change resulting from exposure to elicitor. Our results indicate a highly correlated early response but differences in activation of late responses. At 1 hour, transcriptional changes imply activation of SA, JA, and Et signaling circuits, and of pathways for the biosynthesis of indolic antimicrobials, by both elicitors. At later times, genes with a role in SA-mediated secretory processes and senescence were strongly induced by Flg22. At high dose, OGs weakly induced these same responses, but failed to induce the canonical marker of SA signaling, PR1, which was up-regulated after Flg22 treatment. Differences between elicitors were also observed for induction of callose deposition.
We examined the similarity in transcriptional reprogramming induced by OGs and Flg22 by comparing full-genome transcriptional changes in Arabidopsis seedlings after treatment with OGs (Ferrari et al., 2007) or Flg22 (Fig. 1). OGs were added to the medium to a final concentration of 50 μg/ml, a dose that had previously been shown to induce the maximal increase in cytosolic Ca2+ and H2O2 (Hu et al., 2004). Flg22 was given at a dose of 1 μM, a concentration that was reported to be saturating for medium alkalinization (Felix et al., 1999). For each treatment, three independent biological replicates were individually hybridized to Affymetrix ATH1 GeneChips, representing over 22,000 genes Raw data for each experiment are available at the Integrated Microarray Database System (http://ausubellab.mgh.harvard.edu/imds, under experiment name “Comparison of response to Flg22 and OGs elicitors). Gene expression data were analyzed with Rosetta Resolver (http://www.rosettabio.com/) (Sup. Table 1). As explained in Methods, genes that showed significant (P <= 0.01) change in expression in response to elicitor were selected by error-weighted ANOVA with multiple-testing correction. An additional arbitrary criterion of two-fold change was applied to select genes with robust up- or down-regulation. A combined total of 5289 genes were thus identified that showed altered expression after treatment with either elicitor at either time point (Sup. Table 2). At one hour, 1672 and 2586 genes had altered expression levels after OGs or Flg22 treatment, respectively, relative to water-treated controls (Fig. 1A). By three hours, the response to OGs had diminished, so that only 431 genes deviated from control levels, whereas the response to Flg22 was amplified to include 4413 genes with altered expression levels.
The set of 5289 significantly changing genes in response to OGs or Flg22 identified by Rosetta Resolver was used to determine correlation coefficients between responses to different treatments. Interestingly, the two elicitors induce highly correlated (0.957) responses 1 hour after the addition of elicitors to seedlings, despite the difference in the number of genes identified as differentially expressed after each treatment (Fig. 1B). This apparent discrepancy is a consequence of the fact that the use of threshold criteria to identify differentially expressed genes overestimates the difference between treatments. Of the genes that meet the criteria for Flg22 differentially expressed genes (DEG) but not OGs DEG at 1 hour, many do in fact show some perturbation after OGs, but not at a sufficient amplitude or probability (P <= 0.01) to be called DEGs (Sup. Fig. 1). Our analysis indicates that the calculation of correlation coefficients based on ratio values and measurement errors is a more accurate measure of similarity. By 3 hours, the responses to the two elicitors were more divergent (correlation coefficient = 0.788). Between 1 and 3 hours, the response to each elicitor also changed dynamically, resulting in similarity measures of only 0.572 and 0.803 for OGs and Flg22, respectively. In the case of OGs, the low correlation coefficient between the 1 and 3 hour time points indicates the transient nature of the early response.
To gain further insight into the similarities and differences between responses induced by different treatments, SOM clustering was used to group genes with similar expression patterns. Sixteen such groups were identified (Fig. 1C). In general, at 1 hour, most genes that were up-regulated by OGs (clusters 1 to 7) were also up-regulated by Flg22. However, the response to Flg22 was somewhat stronger in both the number of genes differentially expressed and the amplitude of change (for example, cluster 4). A small set of genes was induced at 1 hour by OGs, but not until 3 hours after Flg22 treatment (in cluster 2), suggesting that some responses are activated more quickly by OGs. At later time points, the response to each elicitor was markedly different: after treatment with OGs, expression of most genes returned to basal levels by 3 hours (clusters 3 to 7, 10, 12). In contrast, the response to Flg22 continued to be robust at 3 hours, with sustained induction or repression of many of the genes that were differentially regulated at 1 hour (clusters 1 to 6, 12). In addition, a large set of genes was regulated exclusively at the later time point by Flg22 (clusters 8, 11, 14).
The kinetic behavior of elicitor- induced genes was further examined by assaying expression of selected genes over a longer time course, from 0.5 hours to 48 hours. Genes that represent major expression clusters (Fig. 1C), or which are markers of specific defense pathways, were assayed by RT-Q-PCR (Fig. 2). Two overall conclusions are apparent: First, as observed from the array data, for many genes there is a higher amplitude response to Flg22 than to OGs. Secondly, as also concluded from the array data, the response to OGs diminishes more rapidly than after Flg22, even for genes with maximal induction at later time point (3h), so that most genes return to un-elicited levels by 6 hours. In addition, the data shown in Figure 2 confirm that Flg22 activates responses that are not activated by OGs in our assays.
Although not apparent in the array data, the RT-Q-PCR analysis showed that the response to Flg22 was also transient, diminishing slowly and returning to basal levels by 24 hours (Fig. 2). For some genes, especially early response genes such as CYP81F2 and the transcription factors WRKY29 and WRKY40, what appears to be a more durable induction after Flg22 treatment may be a consequence of a higher amplitude change with Flg22, as the kinetic pattern was the same with OGs or Flg22, with peak expression between 0.5 and 1 hour. But for other genes (ERF1, PGIP1 PILTP, and VSR), the kinetic pattern is different for the two elicitors, with peak expression in Flg22 treated samples occurring after levels in OGs treated samples are dropping. In addition to demonstrating the transient nature of the Flg22 response, the extended RT-Q-PCR time course shown in Figure 2 also identified WRKY40 and CYP81F2 as very rapidly induced genes, with peak expression at 30 minutes after addition of elicitor.
A calmodulin-like gene, CML41 (McCormack et al., 2005), was initially identified from the microarray hybridization as being potentially induced only by OGs but not by Flg22. However, when assayed over a longer time course, we observed that it was also induced by Flg22, but at a considerably later time, with peak expression at 12 hours compared to 3 hours for OGs (Fig. 2). Intriguingly, the peak expression for this gene coincides with times at which several of the early-response genes are returning to un-induced levels, and suggests that this gene has a role in dampening the immune response.
Of 113 Arabidopsis genes that have been linked with resistance to pathogens in previous functional (genetic) studies, 68% are regulated by exposure to either OGs or Flg22 (Table 1, Sup. Table 3). Interestingly, these defense-related genes exhibit many different patterns of expression. PAD4, EDS1, and NPR1 (Cao et al., 1994; Parker et al., 1996; Zhou et al., 1998), key regulators of resistance to biotrophic pathogens, are strongly but transiently upregulated at 1h by OGs and Flg22. AtPcb and EDS5/SID1, required for accumulation of reactive oxygen species and salicylic acid, respectively, (Nawrath and Métraux, 1999; Bindschedler et al., 2006) are also induced early by both elicitors, but their levels remain elevated at 3 hours. A large class of defense-related genes, including the respiratory burst homologs RbohD and RbohC (Torres et al., 2002); WRKY22, a transcription factor downstream of Flg22 recognition (Asai et al., 2002); and RIN4, a negative regulator of PAMP-triggered immunity (Mackey et al., 2002; Kim et al., 2005) are upregulated by OGs and Flg22 at 1 hour, but only sustained through 3 hours after Flg22 treatment. Some defense-related genes are only regulated significantly by Flg22 at 3 hours (i.e., not upregulated by OGs at 1 or 3 hr), including the resistance signaling protein PBS2/RAR1 (Tornero et al., 2002). Finally, another class of defense-related genes, which includes effector genes CYP83B1, CYP79B2, CYP79B3, SUR1, PAD3, and PAL1, that encode enzymes for the biosynthesis of the tryptophan secondary metabolites indole glucosinolates and camalexin (Bak et al., 2001; Mikkelsen et al., 2003; Mikkelsen et al., 2004) (Glazebrook and Ausubel, 1994), and of phenylpropanoids (Wanner et al., 1995), were up-regulated by both OGs and Flg22 most strongly at 3 hours.
Surprisingly, although it has been shown previously that the Flg22- and OGs- induced resistance to subsequent pathogen infection as well as the OGs- induced expression of AtPGIP1 are independent of SA, JA, and Et signaling (Ferrari et al., 2003; Zipfel et al., 2004; Ferrari et al., 2007), many components of these signaling networks were nonetheless upregulated by exposure of seedlings to elicitor (Table 1, Table S3). EDS5/SID1, SID2/ICS1, and NPR1 are critical genes for SA signaling, functioning either in SA biosynthesis or in SA signal transduction (Cao et al., 1997; Nawrath and Métraux, 1999; Wildermuth et al., 2001). All three were induced by OGs and Flg22, although the kinetic pattern varied for each gene. Several negative regulators of SA-mediated responses, WRKY7 and SIZ1 (Miura et al., 2005; Kim et al., 2006) were also induced, suggesting that complex modulation of downstream hormonal regulation of defense responses may be a characteristic of basal immunity.
As with SA, expression of a variety of JA pathway-associated genes, including the lipoxygenases LOX3 and LOX4, were activated by both OGs and Flg22 at 1 hour, but the response to OGs was less durable than the response to Flg22, with most genes returning to basal levels by 3 hours after treatment with OGs. Other JA-associated genes including ACX1 and OPR3 (Schaller et al., 2000; Castillo et al., 2004), were most strongly up-regulated following 3 hours treatment with Flg22. In addition, several members of the AP2/ERF transcription factor family that are induced by MeJA and the necrotrophic fungal pathogen Alternaria brassicicola (McGrath et al., 2005), including AtERF4, TDR1, and AtERF11, were regulated by OGs and Flg22. Another indication that OGs and Flg22 may activate JA signaling is the increase in expression of CYP79B2, CYP79B3, CYP83B1, and SUR1, involved in the biosynthesis of indole glucosinolates (Wittstock and Halkier, 2002; Mikkelsen et al., 2004) and of MYB51, a transcriptional regulator that promotes indole glucosinolate biosynthesis (Gigolashvili et al., 2007). Accumulation of indole glucosinolates triggered by elicitors of Erwinia carotovora is regulated by a JA-dependent, Et-independent pathway (Brader et al., 2001), and CYP79B2 and CYP79B3 can be induced by exogenous MeJA treatment (Mikkelsen et al., 2003; Devoto et al., 2005).
Biosynthetic genes for ethylene production.1-amino-cyclopropane-1-carboxylate synthases, (Argueso et al., 2007) were also up-regulated by both OGs and Flg22 but again the response was strongest with Flg22 and the patterns of expression varied for different ACS genes (Table 1; Sup. Table 3). Similarly, the ethylene receptors ETR1 & EIN4 (Bleecker et al., 1988; Hua et al., 1998) were significantly increased only after 3 hours treatment with Flg22, as was the negative regulator of ethylene responses CTR1 (Kieber et al., 1993). In contrast to genes with a role in JA signaling, major effects on ethylene biosynthetic and signaling genes were observed only after treatment with Flg22, and predominantly at a time when the response to OGs had largely diminished.
To gain more insight into the role of hormone-mediated responses in MAMP-triggered immunity, we also examined the behavior of several key transcription factors and kinases that have been shown to integrate signals from multiple hormone pathways. Ethylene and jasmonic acid signals are integrated by ERF1 and AtMYC2, which act conversely to activate appropriate responses. While ERF1 promotes expression of the Et -and JA-regulated defense response gene PDF1.2, AtMYC2 represses defense-associated JA-dependent gene induction but positively regulates wound responses (Lorenzo et al., 2003; Lorenzo et al., 2004; Lorenzo and Solano, 2005; McGrath et al., 2005). Although at 1 hour after the addition of OGs or Flg22, ERF1 expression was 3- to 3.5-fold higher than in control seedlings and continued to increase with longer exposure to Flg22, induction of the JA/Et downstream effector genes PDF1.2 and PR3 was not detected. AtMYC2 levels were not affected by OGs or Flg22.
WRKY40 and WRKY33 transcription factors are both modulators of SA and JA pathways, functioning as activators of JA-dependent defense pathways and repressors of SA signaling (Xu et al., 2006; Zheng et al., 2006). Both factors were strongly induced by OGs and Flg22 with expression peaking at 1 hour, and were still induced by Flg22 at 3 hours. The map kinase MPK4, which suppresses SA accumulation but promotes induction of the JA pathway (Brodersen et al., 2006), also had slightly increased levels after 3 hours Flg22.
Altered expression of genes involved in modulating levels of gibberellin and abscissic acid (ABA) was also detected, specifically strong up-regulation of the gene encoding the gibberellin catabolic enzyme Ga2ox6 (Thomas et al., 1999), and repression and up-regulation by Flg22 of ABA biosynthetic genes ABA2 (Leon-Kloosterziel et al., 1996) and ABA3 (Leon-Kloosterziel et al., 1996), respectively. Induction of a catabolic enzyme for growth-promoting gibberellins may be a mechanism for inhibiting growth when resources need to be diverted toward defense. ABA and JA/Et signaling pathways are mutually antagonistic (Anderson et al., 2004), and the decline in ABA2 expression occurs simultaneously with peak expression of the JA-pathway activators WRKY33 and WRKY40, while ABA3 induction occurs when WRKY33 and WRKY40 levels are dropping.
In summary, both OGs and Flg22 trigger rapid transcription of many genes encoding known components of Arabidopsis hormone and defense networks, including SA, JA, and Et biosynthetic genes and downstream signaling molecules.
The SOM clusters (Fig. 1C) were analyzed using the “Analyze Gene Sets” function of Genomica (Segal et al., 2004) (http://genomica.weizmann.ac.il/), which identifies functional classes of genes that are enriched (overrepresented) in a particular cluster compared to the gene set on the ATH1 array. For this analysis, gene functions were assigned according to the GOslim gene ontology annotations from TAIR (ftp://ftp.arabidopsis.org/home/tair/Ontologies/Gene_Ontology/) (Berardini et al., 2004). As shown in Table 2, clusters of predominantly up-regulated genes (clusters #1–8) are enriched for genes responding to abiotic or biotic stimulus or to stress, or that encode proteins involved in signal transduction, including kinases and receptor complexes. Cluster 7 genes, which are transiently upregulated at 1 hour, are especially enriched for receptor binding (selectively interacting with one or more specific sites on a receptor molecule) or receptor activity (combining with an extracellular or intracellular messenger to initiate a change in cell activity). Cluster 8 genes are strongly induced only by Flg22 at 3 hours, and this cluster is enriched for proteins localized to the endoplasmic reticulum (ER), the principal functions of which are synthesis and processing of proteins targeted to membranes, vacuoles, or the secretory pathway. The expression of secretory pathway proteins induced during the systemic acquired resistance defense response is controlled by the signaling molecule NPR1 (Wang et al., 2005). Significantly, of eighteen NPR1-dependent genes that are induced 1.6-fold or more during SAR, nine are grouped with cluster 8 genes, suggesting that Flg22 but not OGs induce the secretory processes that are regulated by NPR1. Clusters 10 through 16 are characterized by down-regulation, with cluster 10 genes being more responsive at 1 hour while clusters 11 to 16 are predominantly regulated at 3 hours by Flg22. Cluster 10 is enriched for transcription factors, many of which are involved in developmental processes. Clusters 11 through 16 are enriched for electron transport or energy pathways and for plastid components, suggesting activation of the senescence program that often accompanies pathogen infection.
Induction of a senescence program is a common response to multiple environmental stresses, including pathogen infection (Gan and Amasino, 1997). During senescence, nutrients are recycled to other organs of the plant. Functionally, activation of a senescence program may constitute an important component of defense by reducing nutrient availability to pathogens. One of the early changes in senescence is the breakdown of the chloroplast, where a large portion of the nitrogen in a leaf cell is stored (Gan and Amasino, 1997), and senescing leaves are characterized by reduced levels of photosynthesis-related genes (Morris et al., 2000) and chlorophyll.
To determine whether Flg22 or OGs treatment affects elements of the photosynthetic apparatus, we displayed expression data for clusters of genes that were down-regulated at 3 hours, which our functional analysis (Table 2) had indicated to be enriched in electron transport and plastid components, onto photosynthesis pathways using MapMan (Thimm et al., 2004). Several key enzymes of the Calvin cycle, including Rubisco large subunit, fructose-bisphosphate aldolase, and fructose-1,6-bisphosphatase, are down-regulated at 3 hours after addition of Flg22 but not after addition of OGs (Sup. Fig. 2). In addition, a number of molecules that function in the light reactions of photosynthesis were specifically down-regulated by Flg22 at 3 hours, including a photosystem I reaction center subunit, photosystem II light harvesting complex subunit, ferrodoxin, and light harvesting chlorophyll A/B binding protein. Clusters of genes specifically downregulated by Flg22 at 3 hours were also enriched for chloroplast protein synthesis.
Expression of two WRKY transcription factor genes, WRKY6 and WRKY53, has been associated with senescence. WRKY6 is most strongly expressed at early and intermediate stages of senescence in leaves (Robatzek and Somssich, 2001) and WRKY53 regulates a set of stress-, defense- and senescence-associated genes (Miao et al., 2004). Both transcription factors were induced by OGs and Flg22 at 1 hour but were still elevated at 3 hours only after Flg22 treatment. Expression of WRKY6 is highest at 1 hour, consistent with its having a role early in senescence. Expression of WRKY53 is most strongly induced in samples treated with Flg22, regardless of time. Consequently, both OGs and Flg22 appear to initiate a senescence program, but it is propagated only after Flg22 treatment.
We examined the possibility that the difference in pathway activation observed between OGs and Flg22 was due to differences in dose of the two elicitors. Seedlings were treated with concentrations of OGs ranging from 2 μg/ml to 1250 μg/ml and compared with treatments of 1.6 to 1000 nM Flg22. Three genes that are up-regulated early by both OGs and Flg22 but more highly induced by Flg22 were assayed at one hour after addition of elicitor. In each case, an increase in OGs concentration resulted in higher amplitude induction, but even at the highest dose of OGs, transcript levels did not match those observed after Flg22 treatment (Fig. 3A). One difficulty in interpreting these results is that, whereas 1 μM Flg22 appears to be a saturating dose, the highest feasible concentration of OGs tested failed to saturate the response. Similarly, no increase in the duration of the response was seen after treatment with high concentrations of OGs (Fig. 3B).
Another potential reason for the difference in response to OGs and Flg22 is lifetime of the elicitor in the medium. To ascertain whether there were differences between the two elicitors in their persistence during treatments, medium in which seedlings (“conditioning”) had been exposed to elicitor for 1 or 3 hours was removed and added to fresh (“naive”) seedlings, which were subsequently assayed for elicitor response. As shown in Fig. 3C, naive seedlings failed to respond to medium in which OGs at 50 μg/ml had been pre-incubated for 1 hour with conditioning seedlings, indicating that free OGs were no longer in the medium, in contrast to a comparable experiment with 1 μM Flg22 in which the conditioned medium was still able to elicit induction of CYP81F2, WRKY29 and WRKY40 (Fig. 3C). To compensate for the disappearance of OGs from the medium, OGs was added repeatedly to the medium, every 30 minutes, and seedlings assayed for induction of CYP81F2, WRKY29 and WRKY40 at 1 or 3 hours. Although repeated applications of Flg22 resulted in sustained induction of these genes, no significant increase in response to OGs was observed (Fig. 3B and 3D). Desensitization following prolonged or repetitive exposure to elicitors has been reported previously for OGs (Binet et al., 1998; Mathieu et al., 1998; Chandra et al., 2000; Navazio et al., 2002) and chitin oligomers (Felix et al., 1998). Although internalization of the elicitor receptor and consequent unavailability for subsequent binding has been reported for the Flg22 receptor FLS2 (Robatzek et al., 2006), in other systems desensitization appears to occur downstream of receptor binding, as the attenuation of response extends to heterologous elicitors (Felix et al., 1998; Chandra et al., 2000). Furthermore, Mathieu and colleagues have previously reported that OGs are degraded in medium containing suspension-cultured tobacco cells (Mathieu et al., 1998). Therefore, we cannot distinguish between desensitization, or different characteristics of the elicitors and their receptors, or different concentrations at the site of perception as the basis for the different responses to these two elicitors.
The similarity in OGs- and Flg22-induced enhanced resistance to Botrytis cinerea (Ferrari et al., 2007) suggests that comparable mechanisms may be employed for recognition and signal transduction for the two elicitors. FLS2, which has structural motifs characteristic of receptor kinases (LRR domain, transmembrane domain, kinase domain), has been identified as the receptor for Flg22, and Arabidopsis lines that lack functional FLS2 are attenuated in their response to Flg22 (Gomez-Gomez and Boller, 2000). To determine whether FLS2 is specific for Flg22-mediated signaling or has a role in transmitting the signal from stimulation with OGs, we tested the response of three genes, (the cytochrome P450-encoding gene CYP81F2, a FAD linked oxidase encoding gene, and the transcription factor-encoding gene WRKY40) that are strongly upregulated after 1 hour exposure of seedlings to OGs, in wild-type and fls2 mutant plants. OGs and Flg22 each induced high-level accumulation of CYP81F2, FAD linked oxidase, and WRKY40 transcripts in wild-type Col-0 and La-er (Fig. 4A). In the La-er fls2-24 mutant and in wild-type ecotype Ws-0, which carries a mutation in fls2, the response to Flg22 was either greatly diminished (La-er fls2-24) or completely abolished (Ws-0), as expected, while the response to OGs was unaltered (Fig. 4A). Differences in the activities of the Ws-0 fls2 and EMS-generated La-er fls2-24 alleles were also observed in assays of seedling growth inhibition by Flg22 (Gomez-Gomez and Boller, 2000).
The innate immune response elicited by Flg22 is mediated by the leucine rich repeat FLS2 receptor kinase, which, in cooperation with BAK1, promotes the activation of the mitogen-activated protein cascade (MEKK1, MKK4/5, and MPK3/6) that specifically regulates the transcription factor genes WRKY22 and WRKY29 (Asai et al., 2002). As shown in Figures 2 and and3,3, OG treatment also results in the transcriptional upregulation of WRKY29.
Although OGs and Flg2 are detected by different receptors (Figure 4A), both OGs and Flg22 elicit MPK3 activation in Arabidopsis protoplasts (Figure 4B). OG-dependent MPK3 activity is apparent after 3 minutes of exposure to elicitor, and returns to basal levels by 10 minutes. In contrast, MPK3 activity induced by Flg22, while strong at 3 minutes, is still robust at 10 minutes following elicitation. Our results are similar to those reported for by Nühse and colleagues for AtMPK6, which is maximally activated both more rapidly and more transiently by OGs than by Flg22 (Nuhse et al., 2000).
Although we did not detect any increase in expression of the SA-dependent effector genes WRKY70 and PR1 in the samples used for expression profiling at 1 and 3 hours, the fact that ICS1 and NPR1 were up-regulated at early time points (Table 1) led us to assay WRKY70 and PR1 expression at later times. To minimize concentration differences between OGs and Flg22, seedlings were treated with either a high dose of OGs, 1250 μg/ml, or 50 μg/ml OGs, or with high or low doses of Flg22, 1 μM or 10 nM. After Flg22 treatment, WRKY70 was consistently induced around 20-fold, with increased expression detectable by 6 hours, and maximal expression at 12 hours (Fig. 5). Up-regulation of WRKY70 following OGs elicitation was variable: in some but not all experiments, WRKY70 transcript levels increased after OGs elicitation. Induction was both earlier (3 hours) and lower magnitude compared to Flg22-treated samples. However, PR1 was significantly induced only in Flg22-treated samples, peaking at 12 hours (Fig. 5). The expression pattern of the gene encoding the secretory pathway protein VSR was similar to that of WRKY70, with reproducible and strong up-regulation only after Flg22 treatment, although the time of peak expression varied between experiments (Fig. 5; Sup. Fig. 3). In contrast, LHCA6, which encodes a chlorophyll a/b binding protein, was consistently down-regulated by high concentrations of OGs. However, this gene too was more responsive to Flg22, which elicited both greater amplitude and more durable repression of LHCA6 expression (Fig. 5).
To verify our results for PR1 expression, we assayed induction of PR1 in a PR1GUS reporter line. GUS expression was induced by Flg22 in seedlings and in leaves from adult plants that were submerged in or infiltrated with elicitor solution (Fig. 6 A–C). However, no induction of PR1GUS was observed after addition of OGs, in either seedlings or adult leaves, regardless of the method of treatment (Fig. 6 A–C). The failure of OGs to induce PR1 was not specific to the accession Col-0, as PR1 was also not induced in Ws-0 seedlings that had been exposed to OGs for 24 hours (Sup. Fig. 4). That PR1 was being induced via SA signaling was confirmed by the loss of PR1 induction in the SA biosynthetic mutant sid2-2 (Fig. 7).
Formation of cell wall appositions, comprised in part of callose, is an early response that is critical in the establishment of non-host and basal resistance (Hauck et al., 2003; DebRoy et al., 2004) and that is a target for suppression by bacterial virulence factors (DebRoy et al., 2004). Deposition of callose can be triggered by treatment of seedlings with the elicitor Flg22 (Gomez-Gomez et al., 1999). To determine if OGs also activate this basal defense response, seedlings were treated with OGs or Flg22 for 18 hours and stained with aniline blue for callose visualization. Only seedlings that had been exposed to Flg22 formed callose-containing papillae in the cotyledons (Fig. 6D). As a control for the negative results obtained with OGs in the seedling assay, callose induction was also assayed in leaves of 4-week old plants of Arabidopsis accession Col-0. At 18 hours after infiltration, leaves treated with OGs had formed multiple callose-containing papillae that were similar to those formed after Flg22 treatment (Fig. 6F). However, if the leaves were treated with OGs or Flg22 by submerging them in a solution containing the elicitor, callose deposition was observed only after treatment with Flg22 (Fig. 6E). These differences in the ability of OGs and Flg22 to elicit callose formation suggest that callose formation can only be induced by OGs if another stress such as mechanical perturbation is also perceived. Alternatively, the difference observed after submerging or infiltration of adult leaves could be a consequence of differential access of OGs to the site of their perception. However, the fact that OGs fail to induce PR1 expression under conditions where they elicit other defense responses suggests that additional triggers are needed for PR1 induction.
Transcriptional responses to OGs and Flg22 elicitors are highly correlated at early stages. This finding is consistent with reports from other investigators that transcriptional changes triggered by elicitors are similar regardless of the particular MAMP. Little difference in transcriptional reprogramming was observed between infections with flagellin-presenting or flagellin-deficient strains of P. syringae pv tomato DC3000 (Pst), or between Pst and E. coli, suggesting multiple MAMPs induce changes similar to those seen with Flg22 (Thilmony et al., 2006). Additionally, a high correlation between transcriptional responses has been reported for seedlings treated with Flg22 or Ef-Tu (Zipfel et al., 2006).
At least part of the early transcriptional response occurs independently of defense hormone signaling, as a number of rapid response genes are induced independently of SA, JA, and Et (Ferrari et al., 2007). However, among genes that are strongly induced at 1 hour by both Flg22 and OGs there are multiple components of known defense signaling pathways. Some of the most highly up-regulated genes function in oxylipin signaling, either as putative JA biosynthetic enzymes (LOX4, induced 100-fold) or as transcription factors that mediate the response to JA (WRKY40, induced 19- to 34-fold). Both elicitors also trigger a rapid increase in the expression of WRKY33, which, like WRKY40, activates JA-pathway responses and represses SA signaling; ERF1, which integrates JA and Et signals and activates JA/Et dependent defense responses including increased expression of PDF1.2 and B-Chi; and TDR1, a member of the AP2/ERF transcription factor family that is induced by MeJA. An increase in genes encoding enzymes for JA biosynthesis following treatment of Arabidopsis with OGs has also been reported by Moscatiello and colleagues, although the particular suite of genes expressed varied from those identified in this work, perhaps reflective of differences between seedlings, the tissue used in this study, and suspension-cultured hypocotyl cells (Moscatiello et al., 2006).
Surprisingly, although JA/Et signaling pathway genes are upregulated by OGs and Flg22, no significant, reproducible increase in transcript levels of the JA/Et dependent defense-related genes PDF1.2 or β-Chi (array and qRT-PCR data) was detected, and only a minimal increase in the JA/Et-induced HEL (PR4) gene (array data only) was observed (data not shown). In contrast to JA biosynthetic and signaling genes, increased expression of biosynthetic genes for Et was modest at 1h in our experiments. Under some growth conditions, endogenous Et is insufficient for enhanced MeJA levels to induce expression of PDF1.2 (Penninckx et al., 1998), which may account for the lack of PDF1.2 up-regulation in our experiments. Similarly to the JA/Et-regulated genes, a set of wound-induced JA-regulated genes that are repressed by Et, including VSP1/2 and THI2.1 (Tuominen et al., 2004; Lorenzo and Solano, 2005), were not activated by OGs or Flg22 in either the early time points assayed by microarray or in the later time points assayed by qRT-PCR (data not shown). VSP1/2 was actually slightly repressed after 3 hours Flg22 treatment. LOX4 is annotated as a putative JA biosynthetic gene, but experimental data demonstrating its activity in that pathway has not been reported. It is possible that the signaling compound produced by LOX4 activity is oxo-phytodienoic acid, which has activity overlapping with but distinct from that of JA (Taki et al., 2005).
Jasmonates have also been reported to regulate the indole glucosinolate biosynthetic genes CYP79B2 and CYP79B3 (Brader et al., 2001; Mikkelsen et al., 2003; Devoto and Turner, 2005). Both of these genes, as well as genes encoding cytochrome P450s involved in camalexin biosynthesis, were up-regulated by both OGs and Flg22, suggesting that PAMP/MAMP-induced resistance may be in part conferred by the production of low molecular weight antimicrobial compounds. Consistent with this hypothesis, Ferrari and colleagues have found that resistance to B. cinerea infection induced by pre-treatment with OGs or Flg22 is dependent on the camalexin biosynthetic gene PAD3 (Ferrari et al., 2007). The protective effects of glucosinolates have been most widely demonstrated in anti-herbivore defense, but glucosinolate-derived products have also been reported to possess anti-microbial properties (Tierens et al., 2001; Kliebenstein et al., 2005).
Biosynthetic genes for SA were also induced by both elicitors, but with peak expression after Flg22 treatment occurring later (3 hours) than for JA biosynthetic genes (1 hour). Increased expression of WRKY70, which positively regulates SA-dependent gene expression and negatively regulates JA-dependent gene expression, was variably observed at early time points (1 to 3 hours) after addition of OGs, but was reproducibly detected at significant levels at later times (6 to 12 hours) after Flg22 treatment. The differential regulation of this transcription factor by the two elicitors suggests that either some SA-mediated responses are induced by Flg22 but not OGs, or that a higher signal strength is needed for activation of this class of late defense responses. The fact that SA-dependent PR1 expression was up-regulated only after Flg22 treatment is consistent with either model. Secretory pathway genes were also induced specifically by Flg22, while the photosynthetic gene LHCA6 was down-regulated both by Flg22 and by OGs at high concentration. One function of the secretory pathway during defense activation is the proper folding, modification, or secretion of PR proteins, as shown by a reduction in extracellular PR1 in mutants deficient in these activities (Wang et al 2005). However, ER domains may also contribute to the formation of senescence-associated vacuoles (SAVs), vacuoles that develop in senescing leaf tissue and are lytic compartments containing senescence-associated proteases (Otegui et al 2005). In addition to a decrease in photosynthesis and an increase in protein breakdown during senescence, additional macromolecules including nucleic acids and starches, and lipids, are degraded for nutrient re-allocation (Otegui et al 2005). Several senescence-induced catabolic enzymes, including an RNase, a nuclease, and an α-amylase, are targeted to a secretory pathway (Taylor et. al 1993; Pérez-Amador et al 2000; Doyle et al 2007).
Several isoforms of the ethylene biosynthetic enzyme ACS were highly up-regulated by Flg22 but not by OGs. ACS catalyzes the rate-limiting step of ethylene biosynthesis (Chae and Kieber, 2005). While both elicitors trigger increased expression of ACS7 at 1 hour, ACS2 and ACS8 are induced most highly by Flg22 at 3 hours. Ethylene has been linked to the timing of the onset of senescence (Bleecker and Patterson, 1997), a process that was also most strongly induced by Flg22 at later time points. The possibility that the role of these ACS isoforms might be to induce a senescence program, in which photosynthesis-associated genes are down-regulated, is supported by our finding that genes whose expression is anti-correlated with ACS2 in the pathogen data set of AtGenExpress are highly enriched for chloroplast proteins (Expression Angler, Botany Array Resource (Toufighi et al., 2005)). ACS2 is one of two ACS isoforms that are phosphorylated by MPK6 (Liu and Zhang, 2004), which is a constituent of the MAP kinase cascade activated by Flg22 (Asai et al., 2002).
SA also has been shown to have a role in the regulation of expression of some senescence-associated genes during developmentally-regulated senescence (Morris et al., 2000). Consistent with this, senescing leaves are characterized by high levels of the SA-regulated PR gene PR1 (Robatzek and Somssich, 2001). SA and ethylene have also been reported to play a role in promoting pathogen-induced chlorosis, a senescence-like process (Bent et al., 1992; O'Donnell et al., 2001), and in a recent paper, SA signaling was shown to contribute to Flg22-induced repression of the chlorophyll a/b-binding protein gene (Tsuda et al 2008) Other studies as well have demonstrated that Et and SA both contribute to regulation of specific defense responses. Clarke and colleagues reported that in cpr6 mutants, constitutive PR1 expression is dependent on SA, but mediated through both SA:NPR1 and SA:Et/JA pathways (Clarke et al., 2000). Similarly, Et and SA act in concert to promote propagative cell death in the lesion mimic mutant vad1 (Bouchez et al., 2007). And enhanced ethylene signaling after grazing on Arabisopsis leaves by P. rapae primes subsequent TCV induction of SA-dependent PR1 expression (De Vos et al., 2006).
SA and Et pathways may interact through SA-induced transcriptional up-regulation of ACS2. In Arabidopsis seedling responses to exogenous SA, ACS2 is transcriptionally up-regulated (Genevestigator; Zimmermann et al 2005). We are currently investigating a role for ACS2 and additional ACS isoforms in Flg22-induced senescence and PR1 expression, and possible crosstalk between these pathways..
The picture emerging from these results is that the common, early response to OGs and Flg22 is activation of elements of JA and Et and SA defense signaling, but the late response is activation of SA-regulated processes. Similar kinetics have been reported for JA and SA levels in Arabidopsis following inoculation with P. syringae pv. tomato DC3000(avrRpt2): JA levels increase significantly within 3 hours, but comparable increases in SA do not occur until 12 hours post inoculation (De Vos et al., 2005). The loss of PR1 induction by Flg22 when administered at a sub-saturating dose suggests that activation of the late response may be dependent on signal strength reaching some threshold. Dose/response experiments indicate that even for genes that are induced by OGs, the activation is not as strong as with Flg22, which is consistent with the conclusion from microarray analysis that many genes identified as “Flg22 specific” at 1 hour are nonetheless somewhat responsive to OGs. We cannot rule out that the differences we see in response to OGs and Flg22 are attributable to lower concentrations at the site of elicitor perception, in which case the sets of responses to different treatments may define stages of defense with different activation thresholds.
Alternatively, plants may utilize characteristics of MAMPs to distinguish between different types of threat. Early signaling events differ for the two elicitors, as shown by both the difference in receptor for OGs and Flg22, and in the kinetics of activation of MPK3 (this work) and MPK6 (Nuhse et al., 2000), which function in a signaling cascade downstream from the Flg22 receptor FLS2 (Asai et al., 2002). Interestingly, the late responses reported here are those that have been shown by a substantial body of work to be effective against biotrophic pathogens, whereas JA/Et-mediated defenses are efficacious against necrotrophic pathogens. The establishment of biotrophy represents a more sophisticated interaction between host and pathogen than necrotrophy, and therefore presumably evolved at a later time. This evolutionary relationship between pathways may be reflected in the kinetics of their activation.
The Arabidopsis thaliana Columbia ecotype was used in this study unless otherwise indicated. For aseptic growth of seedlings, seeds were sterilized by treating them for 5 minutes in 70% ethanol followed by 5 minutes in 50% bleach (2.63 % sodium hypochlorite solution). They were then extensively washed with sterile water (5–6 times) and kept in darkness at 4C in 0.12% phytagar (GIBCO) for 2–5 days. 11 to 15 seeds were dispensed into each well of a 12-well tissue culture plate with 1 mL Murashige and Skoog Basal medium with vitamins (PhytoThechnologiy Laboratories) supplemented with 0.5% sucrose and 0.5 g/L MES, pH 5.7. Plates were sealed with Parafilm to prevent evaporation of the medium. Seedlings were grown at 22°C with a 16 hour photoperiod at a light intensity of 100 μE m−2 s−1 for 10 days before treatment. On the 8th day, the media was replaced with 1 mL of fresh media.
Seedlings were treated with elicitor by adding to the medium either oligogalacturonides (OGs) degree of polymerization 9–16, kindly provided by G. Salvi (University of Rome “La Sapienza”, Italy), or Flg22, a synthetic peptide of 22 amino acids (Felix et al., 1999) to final concentrations of 50 μg/ml or 1 μM, respectively, unless noted otherwise.
Total RNA was extracted from homogenized frozen seedlings using Qiagen RNeasy plant mini kits with on-column DNA digestion using Rnase-Free DNase (Qiagen). cDNA was synthesized using 1 μg of total RNA and the Iscript™cDNA Synthesis Kit (BIORAD) according to the manufacturer's instructions. Real-Time PCR reactions were performed in an iCycler iQ multicolor Real-Time PCR detection system (Bio-Rad) using iQ™ SYBR Green supermix reagent (Bio-Rad), 25 – 100 ng cDNA template, and 250 nM of each gene-specific primer in a final reaction volume of 20 μL. AGI number and primer sequences are as follows: CYP81f2 (At5g57220), 5'-GTGAAAGCACTAGGCGAAGC-3' and 5'-ATCCGTTCCAGCTAGCATCA-3'; PR1 (At2g14610), 5'-CGGAGCTACGCAGAACAACT-3' and 5'-CTCGCTAACCCACATGTTCA-3'; WRKY29 (At4g23550), 5'-ATCCAACGGATCAAGAGCTG-3' and 5'-GCGTCCGACAACAGATTCTC-3'; WRKY40 (At1g80840), 5'-GATCCACCgaCAAGTGCTTT-3' and 5'-AGGGCTGATTTGATCCCTCT-3'; WRKY70 (At3g56400), 5'-CCCAAGAAGTTACTTTAGATGCAC-3' and 5'-TTGCTCTTGGGAGTTTCTGC-3'; VSR (At1g30900), 5'-GAGAAGCGGATCAAGAATCG-3' and 5'-CAGTCATTGGATCGGTTGTG-3'; PI-LTP (At4g12490), 5'-CTGGTTCATCCGGAAACTGT-3' and 5'-CCTTCTGTTGCAAACGTTGA-3'; AtPGIP1 (At5g06860), 5'-GACGAATCTGACAGGTCCAA-3' and 5'-ATAGGCGAAGGTCAGGGACT-3'; PAD3 (At3g26830), 5'-ACGAGCATCTTAAGCCTGGA-3' and 5'-TCGGTCATTCCCCATAGTGT-3'; CML41 (At3g50770), 5'-CCGACGAAGATCACCAAAAT-3' and 5'-TGTCTGAGCTCAAAGGCTGA-3'; FAD linked oxidase (At1g26380), 5'-GACGACACGTAAGAAAGTCC-3' and 5'-CGAACCCTAACAACAAAAAC-3'; ERF1(At3g23240), 5'-TCGGCGATTCTCAATTTTTC-3' and 5'-ACAACCGGAGAACAACCATC-3'; LHCA6 (At1g19150)5'-TCCTCTCGGTTTAGGGTCTG-3' and 5'-TTGAGCCACAAACAGCGTAG-3'. The following PCR program was used for all PCR reactions: 95°C for 3 min, followed by 45 cycles of 95°C for 15 s, 55°C for 30 s and 95°C for 1min. Data were analyzed using Optical System Software, version 3.1 (Bio-Rad). Each sample reaction was run in triplicate for each gene assay. CT values were normalized to the CT value of UBQ10 (At4g05320), 5'-CACACTCCACTTGGTCTTGCGT-3' and 5'-TGGTCTTTCCGGTGAGAGTCTTCA-3', or (Figure 5) to the CT value of EIF4A1 (At3g13920), 5'-TCTGCACCAGAAGGCACA-3' and 5'-TCATAGGATGTGAAG AACTC-3'. Normalized transcript levels of each gene in elicitor-treated samples were compared with levels in mock-treated samples and the fold changes in expression level were calculated. Relative expression of the RT-PCR products was determined using Relative Standard curves method (User bulletin #2 ABI prism 7700 sequence detection system) or the ΔΔCt method (Livak and Schmittgen, 2001).
Three independent biological replicates for each treatment were analyzed. Total RNA was extracted from each sample using the Qiagen RNeasy Plant RNA Miniprep kit (Qiagen, Valencia, CA); samples were split in two before homogenization and re-pooled before loading on the RNA-binding column. RNA quality was assessed by determining the A260/280 ratio of RNA in Tris buffer and by checking the integrity of RNA on an Agilent 2100 Bioanalyzer (Agilent Technologies, www.agilent.com). Target labeling and microarray hybridizations were performed according to the protocol given in the Affymetrix GeneChip Expression Analysis Technical Manual 701025 rev 1 (for details, see Supplemental Methods). Arrays were scanned using an Affymetrix GeneArray® 2500 Scanner and Affymetrix MicroArray Suite v5.0 software.
To assess the quality of each hybridization, we used Affymetrix MicroArray Suite v5.0 analysis software for reports of background intensity, signal to noise ratio, scaling factor for global normalization, and ratios of intensity between 3' and 5' probe sets for selected genes. Further data analysis was performed with Rosetta Resolver v3.2 Gene Expression Data Analysis System (Rosetta Inpharmatics, Kirkland, WA, USA), using Affymetrix. CEL files of array feature intensities and standard deviations as input. Determination of absolute intensity values, propagation of error and P-values, and normalization for comparing arrays in the Resolver system have been described in Waring et al (Waring et al., 2001) and are summarized below. For each probe set, comprised of multiple perfect match (PM) and mismatch (MM) probe pairs, an intensity difference between each PM and corresponding MM was calculated. Probe pairs that differed by more than 3 standard deviations from the mean PM-MM difference for the probe set were considered outliers and were not included in the final calculation of the mean PM-MM intensity difference. Calculation of the probability that a gene is present in the set of transcripts being analyzed was based on the intensities of negative control genes. To increase detection sensitivity, data from three biological replicates were combined. For each array, average intensities, associated intensity errors, and P-values were calculated for each probe set. For calculating average intensity from replicate samples, arrays were scaled to mean intensity, intensity values were transformed for homogenous variance, non-linear error correction was performed, and probe set average intensities computed taking into account measurement error calculations. P-values were calculated and intensity transformed back to the original scale. Ratios of treated versus control intensities were computed by calculating baseline mean background and signal, calculating ratio P-values, and building simple ratios. One-way error-weighted ANOVA was used to identify differentially expressed genes for each time point, using a threshold of P ≤ 0.01.
The intensity-based error model used in Resolver to estimate signal variability stabilizes variance estimation, thus improving the reliability of statistical hypothesis tests when the number of replicated samples is small (two to three) and variance estimations unreliable (Rajagopalan, 2003; Weng et al., 2006). By incorporating associated measurement error as well as expression level, error-weighted ANOVA yields more reliable prediction of differentially expressed genes. Multiple testing correction was performed using q-value. Only genes for which the absolute fold-change between treated and control samples was greater than or equal to 2 were considered to be up- or down-regulated.
To assess the degree of similarity between treatments, error-weighted correlation coefficients were calculated. Correlation coefficients are a more accurate measure of similarity than overlap between sets of differentially expressed genes because for any gene whose expression values under two different conditions are close to but on opposite sides of an applied threshold, the correlation coefficient would consider them to be similar, whereas the lists of differentially expressed genes would be in one case inclusive and in the other case exclusive. To evaluate the differences between these two methods for comparison, genes considered to change significantly after Flg22 treatment but not after OGs treatment were examined further to determine why they had not been selected as differentially expressed following OGs treatment. Approximately half of these genes failed to meet the two-fold change cutoff in OGs samples, but still met the probability threshold for significant change (Fig. S3). Therefore a comparison of lists comprised of genes which meet threshold criteria overestimates the difference between treatments, and the correlation coefficient is a more accurate measure of similarity. Clustering was performed with a Self-organizing Maps (SOM)-based algorithm using cosine correlation as the similarity measure. Gene annotation and assignment to functional categories were based on TAIR (http://www.arabidopsis.org/) gene onotology release 20070317 (ftp://ftp.arabidopsis.org/home/tair/Ontologies/). Functional enrichment of SOM clusters was evaluated with the analysis tool Genomica (http://genomica.weizmann.ac.il/; (Segal et al., 2004)) according to the author's instructions.
Arabidopsis mesophyll protoplast transient kinase expression assay was carried out as previously described (Kovtun et al., 2000; Sheen, 2001) using HA-tagged MPK3 and MPK6. Cells were incubated for 6 hours following transfection. Elicitations were done with 1 uM flg22 or 200 ug/ml OGs.
Plants were grown on Metro Mix 360 medium in a Conviron growth chamber with 60% relative humidity, 22°C/18°C day/night temperatures, and 12-hour photoperiod with light intensity of 75 μE m−2 s−1. One day before treatments the adult plants were covered with a transparent plastic dome to increase humidity, and after infiltration, plants were again covered and placed back into the growth chamber.
Leaf halves were infiltrated with 200 μg ml−1 OGs or 1 μM flg22 (approximately 100–150 μl for each leaf half) or with water as a control. A minimum of three leaf halves from each of three plants received one treatment. The leaves were treated and stained as described in Adam and Somerville (Adam and Somerville, 1996). Briefly, leaves were dehydrated and fixed in a solution of acetic acid/ethanol 1:3 (v/v) for at least two hours. The cleared tissues were washed in 50% and then 30% EtOH before staining for at least 1h in a solution of 0.01% aniline blue in 150 mM K2HPO4 (pH 9.5). Stained leaves were mounted in 50% glycerol. Microscopic analysis was carried out using ultraviolet epifluorescent illumination with DAPI_BFP filter (excitation wavelength 390nm, emission wavelength 461nm). Images were taken through an Apotome scope (ZEISS) using AxioVision software.
β-Glucuronidase (GUS) enzyme activity of PR-1GUS Arabidopsis, seedlings from three wells or leaves from three plants, was determined histochemically after treatment with OGs, Flg22 or water. The tissues were placed in 2ml of 50 mM sodium phosphate, pH 7, 0.5 mM potassium ferrocyanide, 0.5 mM potassium ferricyanide, 10 mM EDTA, pH 8, 0.1%tritonX-100 and 0.5mM 5-bromo-4-chloro-3-indolyl-β-D-glucuronide (X-gluc, Rose Scientific L.t.d). After vacuum-infiltration for 5 min the tissues were incubated overnight at 37°C and washed with sodium phosphate buffer. The samples were fixed with acetic acid/ethanol 1:3 (v/v), the chlorophyll was entirely removed by several wash in 70% ethanol, and the leaves or seedlings were mounted in 100% lactic acid.
We thank Jennifer Couget, Paul Grosu, and Reddy Gali (Bauer Center for Genomics Research, Harvard University, Cambridge, MA, USA) for assistance with microarray hybridization and data analysis and Christian Danna for help in identifying MAMP-induced transcription factor genes.
FUNDING This work was supported by the National Science Foundation (grant DBI-0114783 to F.M.A.); the National Institutes of Health (grant GM48707 to F.M.A.); the European Union (grant n. 23044 “Nutra-Snacks” to S.F.), the Ministero dell‚Università e della Ricerca (MIUR) Programma di ricerca di Rilevante Interesse Nazionale 2006 (S.F.); Fondo per gli Investimenti della Ricerca di Base 2001 and Cofinanziamento 2002 (G.D.L.), the Giovanni Armenise - Harvard Foundation (G.D.L.) and the Institute Pasteur - Fondazione Cenci Bolognetti, (G.D.L.).