The Arabidopsis flagellin receptor FLS2 mediates the perception of Xanthomonas Ax21 secreted peptides

Abstract

Detection of microbes by plants relies in part on an array of pattern-recognition receptors that recognize conserved microbial signatures, so-called “microbe-associated molecular patterns.” The Arabidopsis thaliana receptor-like kinase FLS2 is the pattern-recognition receptor for bacterial flagellin. Similarly to FLS2, the rice transmembrane protein XA21 is the receptor for the sulfated form of the Xanthomonas oryzae pv. oryzae secreted protein Ax21. Here we show that Ax21-derived peptides activate Arabidopsis immunity, triggering responses similar to those elicited by flagellin, including an oxidative burst, induction of defense-response genes, and enhanced resistance to bacterial pathogens. To identify Arabidopsis Xa21 functional homologs, we used a reverse genetics approach to screen T-DNA insertion mutants corresponding to all 47 of the Arabidopsis genes encoding non-RD kinases belonging to the interleukin-1 receptor-associated kinase (IRAK) family. Surprisingly, among all of these mutant lines, only fls2 mutants exhibited a significant loss of response to Ax21-derived peptides. Ax21 peptides also failed to activate defense-related responses in an fls2-24 mutant that does not bind Flg22. Moreover, a Flg22Δ2 variant of Flg22 that binds to FLS2 but does not activate FLS2-mediated signaling suppressed Ax21-derived peptide signaling, indicating mutually exclusive perception of Flg22 or Ax21 peptides by FLS2. The data indicate that FLS2 functions beyond flagellin perception to detect other microbe-associated molecular patterns.

Pattern-recognition receptors (PRRs) that recognize conserved microbial signatures, which are referred to as microbe-associated molecular patterns (MAMPs), are a key mechanism by which plants and other organisms detect microbes (1). Among several MAMPs detected by Arabidopsis thaliana, flagellin is the best studied. In Arabidopsis, the leucine-rich repeat (LRR) transmembrane receptor kinase FLAGELLIN SENSITIVE 2 (FLS2) is essential for flagellin perception (2). A 22-aa synthetic peptide (Flg22) corresponding to the recognized domain of flagellin activates FLS2-dependent signaling, triggering the same responses as the native flagellin protein from Pseudomonas syringae pv. tabaci (3). Flg22-triggered responses include activation of MAPK cascades, up-regulation of defense genes, transient production of an H₂O₂ oxidative burst, deposition of callose, and enhanced resistance against pathogens (2, 4, 5).

The Arabidopsis FLS2 receptor belongs to the IRAK family of receptor like kinases (RLKs), which includes two other well characterized MAMP receptors, Arabidopsis EFR (TU-elongation factor-receptor 1) and rice XA21 (Xanthomonas resistance protein 21) (6). These RLKs carry the non-RD domain, a motif that is found in many IRAK kinases that function in immune signaling pathways (6). The Arabidopsis genome encodes 47 non-RD IRAK kinases, of which 35 are RLKs and 12 are predicted to be cytoplasmic (6, 7).

XA21 recognizes the conserved Xanthomonas oryzae pv. oryzae (Xoo) Ax21 secreted protein (8). Rice plants carrying the Xa21 gene are fully resistant to Xoo carrying Ax21. A synthetic sulfated 17-aa peptide (axY^s22) derived from Ax21 (residues 17–33 of the Ax21 protein) binds XA21 and triggers enhanced resistance against Xoo. Replacing sulfated Tyr (at amino acid position 22 of Ax21) by Ala abolishes Ax21 perception, indicating that the sulfated Tyr residue is required for activity (9).

Here we present data showing that Ax21-derived peptides are also recognized by Arabidopsis, and surprisingly, that this recognition is mediated by FLS2, which was previously thought to be highly specific for flagellin. If individual MAMP receptors are capable of recognizing a variety of MAMPs, it increases the spectrum of microbe-derived molecules that can activate an immune response using a relatively limited number of PRRs.

Results

Ax21-Derived Peptides Are Perceived by Arabidopsis.

Because Ax21 is conserved among Xanthomonas species, including Xanthomonas campestris campestris (Xcc), which is an Arabidopsis pathogen, we hypothesized that Arabidopsis may be able to respond to Ax21 similarly to other well-characterized MAMPs. Although supernatant extracts from Xcc do not normally elicit rice XA21-mediated immunity, they do when a plasmid containing the putative sulfotransferase RaxST from Xoo is introduced into the otherwise unrecognized Xcc bacterial strain (10), suggesting that the examined Xcc strains have the capability of normally secreting nonsulfated Ax21. Bacterial extracts from an isogenic Xcc ΔfliC mutant expressing a variant of flagellin that is not perceived by Arabidopsis, still triggered MAMP responses similar to wild-type Xcc, indicating that other MAMPs from Xcc besides flagellin are recognized by Arabidopsis (11), one of them potentially being Ax21.

To test whether Ax21-derived peptides are perceived by Arabidopsis, we infiltrated axY22, the nonsulfated version of the sulfated 17-aa peptide recognized by rice, into wild-type Col-0 plants carrying either WRKY11p::GUS or MYB51p::GUS reporter constructs. These two promoter-reporter constructs were previously shown to be activated by a variety of MAMPs, including Flg22 and Elf26 (a synthetic 26-aa peptide corresponding to elongation factor EF-Tu) (12). Indeed, infiltration of 1 μM axY22 triggered activation of WRKY11p::GUS and MYB51p::GUS similar to 1 μM Flg22 (Fig. 1). Surprisingly, the sulfated version of axY22 (axY^s22), as well as a derivative of axY^s22 that is unable to trigger XA21-mediated immunity, axY22A, which contains alanine instead of tyrosine at position 22 (9), also activated the Arabidopsis MAMP reporters (Fig. 1).

Fig. 1.

Ax21-derived peptides trigger MAMP-responsive gene expression in Arabidopsis. Leaves from 6-wk-old Col-0 transgenic plants carrying WRKY11::GUS or MYB51::GUS constructs were infiltrated with 1 μM peptides (Right) or water (Left) and stained with GUS, as described in SI Methods.

Treatment of plants with MAMPs has been shown to trigger a complex and multilayered defense response, including elicitation of a transient H₂O₂ burst (2). To facilitate the study of Ax21 perception in Arabidopsis, we developed a miniaturized and relatively high throughput assay to measure H₂O₂ bursts in 10-d-old seedlings germinated in liquid MS medium in 96-well assay plates. In agreement with the results shown in Fig. 1, peptides previously shown to be either active (axY^s22) or inactive (axY22, axY22A) in rice elicited an H₂O₂ burst in Arabidopsis (Fig. 2). However, in contrast to adult Arabidopsis plants, where relatively low concentrations of Ax21 peptides (1 μM) were sufficient to trigger reporter gene expression (Fig. 1), significantly higher concentrations (∼100 μM) of the Ax21 peptides were necessary to elicit H₂O₂ production in seedlings. The effective concentration Ax21 peptides that arise at sites of bacterial infection are not known, but concentrations of exogenously applied axY^s22 peptide in the range of 1–100 μM are also necessary to trigger immunity in rice (9).

Fig. 2.

Ax21-derived peptides trigger a hydrogen peroxide burst. Ten-day-old Col-0 wild-type seedlings were grown in 96-well plates and mock-treated or elicited with axY22, Flg22, or Elf26 (A) or with axY^s22 or axY22A (B) at the indicated concentrations: axY^s22 is the sulfated 17-mer that is active in rice; axY22 is the nonsulfated version of axY^s22; axY22A contains a Tyr→Ala substitution. Neither axY22 nor axY22A can trigger rice XA21-mediated immunity. Each point represents the mean of six seedlings. Error bars represent ± SD of the mean. Essentially identical results were obtained in at least three independent experiments.

To determine whether the recognition of Ax21 peptides by Arabidopsis results in a biologically significant response, we tested whether Ax21-derived peptides trigger enhanced resistance in seedlings to P. syringae, similarly to Flg22-elicited protection of seedlings against P. syringae (13). By using a P. syringae pv. maculicola (Psm) strain expressing the LUX operon from Photorhabdus luminescens, we were able to monitor bacterial growth by measuring light emission in a scintillation counter (adapted from refs. 13 and 14). For this MAMP-elicited protection assay, seedlings were grown in 12-well plates (20–30 seedlings per well) for 10 d and elicited with various MAMPs, including Ax21 peptides, for a period of 24 h before inoculation with Psm-LUX. To assess bacterial growth inside seedlings, washed seedlings were ground at different times after inoculation and luminescence was measured with a scintillation counter. Bacterial growth was estimated by converting light emission into CFUs (using an experimentally determined CPMs/CFUs conversion table). Although relatively high concentrations of the Ax21 peptides were necessary to detect protection activity (100 μM), as in the case of the oxidative burst (Fig. 2), peptides that are either active (axY^s22) or inactive (axY22 and axY22A) in rice were equally capable of triggering enhanced resistance against P. syringae in Arabidopsis (Fig. 3A). Similarly to the concentration of Flg22 required to elicit an oxidative burst in the seedling assay, 1 μM Flg22, but not lower concentrations, elicited protection against Psm-LUX. Importantly, both Flg22 and axY^s22 peptides also elicited enhanced resistance against X. campestris (Fig. 3B), which suggests that perception of Ax21-derived peptides could be part of a natural Arabidopsis defense mechanism against Xcc.

Fig. 3.

Ax21-derived peptides trigger enhanced resistance against bacteria. (A) Ten-d-old seedlings were grown in 12-well plates, elicited with 1 μM Flg22 or 1 μM, 10 μM, or 100 μM axY22, axY^s22, or axY22A for 24 h, and then infected with Psm-LUX. Bacterial titer was assessed as CFU/mg fresh weight seedling tissue 36 h after inoculation: axY^s22 is the sulfated 17-mer that is active in rice; axY22 is the nonsulfated version of axY^s22; axY22A contains a Tyr→Ala substitution. Neither axY22 nor axY22A can trigger rice XA21-mediated immunity. (B) Seedlings were elicited with 100 μM axY^s22 for 24 h and then inoculated with Xcc. Each column represents the bacterial titer 24 h after inoculation as CFU/mg fresh weight seedling tissue and is the mean of three wells containing 20 seedlings each. Error bars represent ± SD of the mean. Essentially identical results were obtained in at least three independent experiments. *P < 0.001 compared with mock (t test).

In an effort to determine which amino acids in the 17-aa synthetic axY^s22 peptide are important for perception by Arabidopsis, we tested a previously described collection of 17 peptides, each one carrying an alanine in place of the original amino acid, except for residue #1 (which normally contains an Ala at that position) that was replaced by Gly (9). Surprisingly, all of these modified peptides exhibited a similar level of activity as axY22 or axY^s22 in the P. syringae protection assay when tested at 100 μM (Fig. S1A). However, replacing Ala by Gly at position #1 (axY^s22-A1), or Glu by Ala at position #2 (axY^s22-A2), increased the activity of the peptide by at least 10-fold (Fig. S1 B and C). That is, the two peptides containing substitutions at one of the two N-terminal residues were as active at 10 μM as axY22 or axY^s22 were at 100 μM.

One explanation for the unexpected result that all of the substituted Ax21 peptides had activity at 100 μM is that all peptides at such a relatively high concentration would nonspecifically elicit a robust defense response or would directly inhibit the growth of P. syringae. To test this possibility, we pretreated seedlings with high concentrations of peptides that have been reported to be inactive in Arabidopsis. A Flg22-like peptide derived from Agrobacterium tumefaciens (Flg22^A.tum), or a Flg22-derived peptide Flg22Δ2, both previously shown to be inactive in adult Arabidopsis plants (15), were also inactive at 100 μM in the seedling protection assay (Fig. S2), suggesting that a high concentration of any peptide does not trigger nonspecific activation of MAMP-elicited responses. To further test the specificity of peptide perception in seedlings, we tested Flg22 and Elf26 perception in fls2 and efr receptor mutants, respectively. High concentrations of active Flg22 or active Elf26 did not trigger a measurable response in the corresponding receptor mutants (Fig. S3), suggesting that high concentrations of peptides do not bypass the requirement for a specific receptor. Finally, to further assess whether the Ax21 peptide response is receptor-specific, we performed a saturation experiment with the axY^s22 peptide. We found that at concentrations greater than 100 μM of peptide the response was saturated, indicating that Ax21 perception is most likely mediated by a specific receptor or receptors (Fig. S4).

Because the axY^s22-A1 peptide is 10-fold more active than the wild-type peptide in the Arabidopsis assays described in this section, we used 10 μM axY^s22-A1 for most of the experiments described below.

FLS2 Is Required for Perception of Ax21-Derived Peptides.

Most of the known MAMP receptors in plants, including FLS2, EFR, and rice XA21, belong to the IRAK family of non-RD kinases. In an effort to identify Arabidopsis Ax21 peptide receptor(s), we used the axYs22-A1 peptide in the seedling oxidative burst and protection assays to test a collection of 71 mutant lines consisting of one or two (when available) T-DNA insertion mutations corresponding to each of the 47 non-RD IRAK kinase genes. Surprisingly, only two independent T-DNA insertions in the fls2 gene compromised the H₂O₂ burst (Fig. 4A) or the protection against P. syringae (Fig. 4B) after axY^s22-A1 elicitation (Table S1). Similar results were obtained with these two fls2 mutants in larger scale assays in 12-well plates, where the two fls2 mutants also failed to respond to axY^s22-A1, axY^s22 or axY22 (Fig. 4C).

Fig. 4.

FLS2 is necessary for Ax21-derived peptide perception. (A) A hydrogen peroxide burst was elicited in 10-d-old Col-0 wild-type or fls2 mutant (SAIL_691_C04) seedlings in 96-well plates with axY^s22-A1 (10 μM) or Flg22 (1 μM). Each datapoint represents the mean of six seedlings (six wells). Error bars represent ± SD of the mean. (B) Ten-d-old Col-0 wild-type or fls2 (SAIL_691_C04) seedlings were grown in 96-well plates, elicited with axY^s22-A1 (10 μM) or Flg22 (1 μM) for 24 h and then infected with Psm-LUX. Bacterial titer was determined with a 96-well plate scintillation counter each hour between 20 and 24 h after inoculation. Each datapoint represents the mean of six seedlings (six wells). Error bars represent ± SD of the mean. (C) Ten-day-old Col-0 wild-type or fls2 mutant (SAIL_691_C04) seedlings were grown in 12-well plates and elicited with various Ax21 peptides and infected with Psm-LUX as in Fig. 3A. A second fls2 insertion mutant (Salk_121477) showed identical results. Each column represents the mean of three wells containing 20 seedlings each. Error bars represent ± SD of the mean. Essentially identical results were obtained in at least three independent experiments. *P < 0.001 compared with mock (t test).

Ax21 Peptides Elicit the Same Responses as Flagellin.

To further characterize FLS2-mediated perception of axY^s22-A1, we tested whether the NADPH oxidase RBOHD (respiratory burst oxidase homolog D), an enzyme that has been shown to be required for reactive oxygen species production after elicitation with Flg22 (16), is also necessary for the axY^s22-A1 triggered H₂O₂ burst. Indeed, an AtrbohD mutant did not exhibit an H₂O₂ burst following treatment with axY^s22-A1 (Fig. 5A). Moreover, we found that mutations in the FLS2 adaptor protein BAK1, which is required for Flg22-elicited signaling (17), partially compromised the enhanced resistance triggered by axY^s22-A1 (Fig. 5B). Finally, we found that in the seedling assay, 5 μM axY^s22-A1 activates the expression of a set of genes previously shown to be up-regulated by 1 μM Flg22 (18) in an FLS2-dependent manner (Fig. 5C). These data show that axY^s22-A1 triggers a similar cascade of downstream events as those triggered by Flg22, which is consistent with the hypothesis that FLS2 is the axY^s22-A1 receptor.

Fig. 5.

FLS2-mediated perception of Ax21-derived peptides mimics Flg22 perception. (A) A hydrogen peroxide burst was elicited in 10-d-old Col-0 wild-type or AtrbohD mutant seedlings with axY^s22-A1 (10 μM) or Flg22 (1 μM) in 96-well plates. Each datapoint represents the mean of six seedlings (six wells). Error bars represent ± SD of the mean. (B) Ten-d-old Col-0 wild-type, fls2 mutant (SAIL_691_C04), or bak1 mutant (SALK_116202) seedlings were grown in 12-well plates and elicited with axY^s22-A1 (10 μM) or Flg22 (1 μM) and then infected with Psm-LUX as in Fig. 3A. Columns represent the mean of three wells containing 20 seedlings each. Error bars represent ± SD of the mean. (C) Ten-day-old Col-0 wild-type or fls2 mutant seedlings were grown in 12-well plates and elicited with Flg22 (1 μM) or axY^s22-A1 (5 μM) for 3 h. RNA was extracted and qRT-PCR analysis was carried out as described in SI Methods. Gene expression is shown as fold-change compared with mock treatment. Columns represent the mean of three independent qRT-PCR reactions. Error bars represent ± SEM. Essentially identical results were obtained in at least three independent experiments. *P < 0.001 and **P < 0.01 compared with mock (t test).

Flg22 Binding Domain of FLS2 Is also Required for Ax21 Peptide Activity.

The FLS2 receptor is a 1,173-aa protein that consists of an intracellular kinase domain, a hydrophobic membrane-spanning domain, and an extracellular domain composed of 25 LRRs (19). A single nucleotide change in LRR #10 (G → R 318, referred to as the fls2-24 allele) abolishes the binding of Flg22 without compromising the accumulation or stability of FLS2 protein (15, 20), as do other nearby mutations in the proposed binding domain for Flg22 (21). To initiate studies to determine whether Flg22 and axY^s22-A1 use the same binding domain to activate FLS2, we tested the fls2-24 mutant for H₂O₂ production after elicitation with axY^s22-A1 peptide. The fls2-24 mutant did not exhibit an H₂O₂ burst after axY^s22-A1 elicitation (Fig. 6A). To further test the hypothesis that Flg22 and axY^s22-A1 share a binding site on FLS2, we performed competition experiments between axY^s22-A1 and Flg22Δ2, a Flg22-derived peptide that has been shown to compete with Flg22 by binding to the FLS2 receptor but not triggering FLS2 activation (15). The addition of Flg22Δ2 at the same concentration as axY^s22-A1 strongly reduced the H₂O₂ burst triggered by axY^s22-A1 (Fig. 6B), whereas a 10-fold molar excess of Flg22Δ2 was necessary to partially inhibit the Flg22-triggered H₂O₂ burst (Fig. S5A). As a control, a 10-fold molar excess of Flg22Δ2 did not have any measurable effect on the Elf26-triggered H₂O₂ burst (Fig. S5B). Although competition through binding at separate sites is possible, the simplest explanation for Flg22Δ2 suppression of both Flg22- and axY^s22-A1-mediated signaling is competition for the same binding domain on FLS2.

Fig. 6.

The Flg22 binding domain of FLS2 is also required for Ax21-derived peptide perception. (A) A hydrogen peroxide burst was monitored in 10-d-old Ler wild-type or fls2-24 mutant seedlings in 96-well plates after elicitation with Flg22 (1 μM) or axY^s22-A1 (10 μM). (B) A hydrogen peroxide burst was monitored in Col-0 wild-type seedlings after elicitation with axY^s22-A1, Flg22Δ2, or a combination of both peptides. Each datapoint represents the mean of six seedlings (six wells). Error bars represent ± SD of the mean. Essentially identical results were obtained in at least three independent experiments (A) or two independent experiments (B).

Synthetic Peptide axY^s22-A1 Is Not Contaminated with Flg22.

To rule out the possibility of contamination of the axY^s22-A1 stock solution with Flg22, we analyzed axY^s22-A1 by mass spectrometry (Fig. S6). We did not detect any contaminants at the concentration used in our assays (10 μM). In particular, Flg22 with a detection limit of 10 nM (Fig. S6) was not present in the stock. Because concentrations of Flg22 below 1 μM are insufficient to trigger protection against Psm-LUX or to elicit an oxidative burst in Arabidopsis seedlings in our assay, these results make it extremely unlikely that Flg22 contamination of the axY^s22-A1 stock solution would explain FLS2-dependent axY^s22-A1 peptide perception.

Discussion

Plant defense against microbial attack uses a limited number of preformed receptors that recognize pathogen-related signature molecules. MAMP receptors, such as the Arabidopsis receptor kinases FLS2 and EFR, recognize highly conserved pattern molecules, such as flagellin and elongation factor EF-Tu, respectively. If plants had promiscuous MAMP receptors that were able to recognize multiple MAMPs, it would expand the variety of MAMPs that an individual plant could recognize without devoting a large fraction of the genome to PRRs. In addition, because MAMPs are likely to be present in low concentrations in a natural infection, the simultaneous recognition of multiple MAMPs may help to strengthen MAMP-mediated signaling. The promiscuity of FLS2 in recognizing both flagellin and the Ax21-derived peptides produced by Xanthomonas suggests that the Arabidopsis immune system has evolved to maximize the utility of a limited number of PRRs.

MAMP recognition in mammals is carried out by Toll-like receptors (TLRs), transmembrane receptors with LRR external domains, and associated non-RD cytoplasmic kinases, which together are functionally equivalent to FLS2 or XA21. The mammalian receptor TLR2 was initially thought to bind MAMPs as different in structure as LPS, peptidoglycan, and lipoproteins from diverse bacteria and parasites. Later studies, however, demonstrated that LPS and peptidoglycan preparations were contaminated with lipoproteins or lipopeptides (22), which are now thought to be the actual ligands of TLR2. Because all of the peptides used in our study were synthetic (HPLC-purified and subjected to mass spectrometry to confirm mass and sequence and potential contamination with Flg22) (Fig. S6), the apparent promiscuity of FLS2 that we observe cannot be explained as an artifact caused by flagellin or Flg22 contamination.

Mammalian TLR receptors, such as TLR2 and TLR6, can form heterodimers, which broaden the range of specificity of these receptors. TLR2-TLR6 heterodimers recognize 2-acyl-lipoproteins, whereas TLR6 homodimers recognize lipoteicoic acid and zymosan and TLR2 homodimers seem to be inactive (23). A similar mechanism may operate in plants and is a viable hypothesis for the apparent promiscuity of FLS2, although partner MAMP receptors for FLS2 have not been identified. Alternatively, because Flg22 and Ax21 peptides do not show any obvious sequence similarity, it is possible that FLS2 may function as a coreceptor for another protein that is the actual receptor for the Ax21-derived peptides. Consistent with the idea that MAMP receptors function in complexes with transmembrane receptor partners, FLS2 and EFR function in concert with the coreceptor kinase BAK1, which is also required for BRI1-mediated perception of brassinolide (17, 24). However, there is no evidence that BAK1 interacts directly with brassinolide or MAMPs.

In contrast to the coreceptor model, our data suggest that the Flg22- and Ax21-dervived peptides directly bind to FLS2 because an FLS2 mutation that affects the binding of Flg22 to FLS2 blocked Ax21-derived peptide elicited responses (Fig. 6A). Furthermore, a deleted version of Flg22, Flg22Δ2, which functions as a suppressor of Flg22 responses (15), partially suppressed the axY^s22-A1–mediated oxidative burst (Fig. 6B), again suggesting that Flg22- and Ax21-derived peptides may compete for the same binding site on FLS2 (Fig. 6B). Whether FLS2 is the only receptor for Ax21-derived peptides or whether FLS2 is recruited in a receptor complex together with other receptors and adaptors that may modulate its specificity, remains to be determined.

If FLS2 is not the direct or the only receptor providing the binding site for Ax21 peptides, it is unlikely than any of the other 34 Arabidopsis non-RD RLKs function as an Ax21 peptide receptor, because T-DNA insertions in the corresponding genes do not affect the ability of Ax21 peptides to activate an Arabidopsis immune response, unless they have redundant functions in partnering with FLS2 for Ax21 perception. We limited our search for potential Ax21 receptors to those with the non-RD types of kinase domain (Table S1) because genomic analysis indicated that the presence of the non-RD motif is highly predictive of a function in the innate immune response in both plants and animals (6). For example, all plant non-RD kinases that have been assigned a physiological function serve a key role in innate immunity, including three of the best-studied plant PRRs, FLS2, EFR, and XA21. Because XA21 is closely related to Arabidopsis non-RD RLKs of the subfamily LRR-XII, which includes FLS2 and EFR (6, 25), the identification of AtFLS2 as an Xa21 functional homolog is perhaps somewhat predictable. Nevertheless, because the extracellular LRRs of FLS2 and XA21 have no compelling similarity and FLS2 was thought to be specific for flagellin perception, the result that perception of Ax21-derived peptides requires FLS2 is striking.

Given the significant level of identity between FLS2 and XA21, and in light of our data suggesting that FLS2 may be the receptor for Ax21-derived peptides, it is interesting to note that rice cultivars that lack Xa21 but that still encode functional OsFLS2 do not respond to Ax21-expressing Xanthomonas strains (8) and do not perceive Ax21 peptides (9). These latter data suggest that OsFLS2 does not have the broad ligand specificity exhibited by AtFLS2, or that the experimental protocols used in the rice experiments were not sensitive enough to detect this broad specificity.

One explanation for the current divergent specificities of OsFLS2 and AtFLS2 is to postulate the existence of an ancient progenitor of OsFLS2 and AtFLS2 that had the broad ligand specificity of AtFLS2. In the Arabidopsis lineage, a broad ligand-specificity of the ancient receptor may have been preserved, but in the rice lineage, a gene duplication may have allowed one paralog (FLS2) to lose its capacity to perceive Ax21 peptides as the other paralog (Xa21) evolved a high level of specificity for sulfated Ax21 peptides in the wild species Oryza longistaminata. One line of evidence in support of this model is that AtFLS2 recognizes many different variants of the axY^s22 17-mer in addition to Flg22 (Figs. 1 and 2, and Fig. S1). This proposed evolutionary course of events makes sense in light of the vast expansion of predicted PRRs in rice compared with dicots. With nearly 10-fold fewer predicted PRRs, [328 non-RD RLKs in rice versus 35 in Arabidopsis (1, 7)], Arabidopsis PRRs need to have broad specificity to detect as many MAMPs as rice.

Although microbes can potentially produce a very large number of conserved signatures, only a few MAMPs have been identified, and most of these trigger a similar set of responses. MAMP responses include alkalinization of the apoplast, transcriptional up-regulation of defense genes, activation of MAP kinases, deposition of callose, and the transient production of reactive oxygen species (oxidative burst). It is not known, however, if these responses are universal. We therefore suggest that the assessment of MAMP-enhanced resistance, which is likely to be the ultimate consequence of MAMP recognition, may be the most appropriate way to assess MAMP perception. In this article, we describe the development of a high throughput assay for detecting enhanced resistance against bacterial infection (Fig. 4B), which may assist in the identification of novel MAMPs.

The lack of physical information to conclusively define the ligand binding sites of FLS2 makes it difficult to understand how a single receptor mediates the perception of MAMPs with no apparent structural similarity. However, in light of our functional data showing that FLS2 mediates perception not only of flagellin but also of Ax21-derived peptides, it is possible that FLS2 (as well as XA21 and other PRRs) may mediate recognition of other ligands as well. Hence priorities for future research include not only identification of the binding sites of known ligands for well-characterized PRRs, but also discovery of additional putative ligands and potential partner proteins that may alter plant PRR extracellular domain configuration and ligand specificity.

Methods

Plant Growth.

Adult plants were grown in climate-controlled growth rooms (Conviron MTPS144) on Metro-Mix 360 soil (Sun Agro) at 22 °C, 75% humidity, and a 16 h photoperiod at 100 μE·m⁻²·s⁻¹ of light (for genotyping of T-DNA lines and seed propagation) or a 12-h photoperiod (for GUS staining). SALK_ and SAIL_ T-DNA insertion lines were obtained from the Arabidopsis Biological Resource Center (http://abrc.osu.edu/). GABI_Kat lines were obtained from the Arabidopsis Biological Resource Center or from NASC (http://arabidopsis.info/). An Arabidopsis atrbohD mutant was obtained from M. A. Torres (Universidad Politécnica de Madrid, Spain). The fls2-24 mutant was obtained from T. Boller (Botanical Institute, University of Basel, Switzerland). Genotyping was carried out by PCR of genomic DNA using PCR primers designed by SIGnal T-DNA Express and following the recommended combination of primers (http://signal.salk.edu/tdnaprimers.2.html). A complete list of PCR primers is shown in Table S2. For detailed information on genotyping see SI Methods.

Seedling Liquid Culture.

For growing seedlings in liquid medium in 12-well assay plates (BD Falcon; 353043), seeds were sterilized in 20% bleach (2 min), washed three times with sterile water, and 20 to 30 seeds were dispensed into wells containing 1 mL MS 1× medium (Murashige and Skoog basal medium with vitamins from Phytotechnology Laboratories supplemented with 0.5 g/L Mes hydrate and 0.5% sucrose at pH 5.7). Seedlings were grown for 10 d (replacing medium at day 8) at 22 °C, 95% humidity (to prevent medium evaporation) in a plant growth chamber (Conviron; E7/2) under 100 μE·m⁻²·s⁻¹ and a 16-h photoperiod.

For growing seedlings in 96-well plates (Greiner Bio-One; 655083), seeds were sterilized as above and several seeds were dispensed into each well containing 125 μL MS 1× medium (see above). At day 8, the medium was replaced using an eight-well vacuum manifold connected to a vacuum line and seedlings were thinned to leave a single seedling per well. Seedlings were grown for 2 more days at 22 °C, 95% humidity in a plant growth chamber (Conviron; E7/2) under 100 μE·m⁻²·s⁻¹ and a 16-h photoperiod.

Synthetic Peptides.

Flg22, Flg22^A.tum, Flg22Δ2 and Elf26 were synthesized by the Massachusetts General Hospital-Peptide/Protein Core Facility. Ax21-derived peptides were synthesized by Pacific Immunology Corporation. For a complete list of Ax21-derived peptides, see ref. 9. Lyophilized peptides were resuspended in sterile water. Mass spectrometry was used to determine the purity of the axY^s22-A1 peptide as described in SI Methods and Fig. S6.

Luminol Chemiluminescence Assay for H₂O₂ Detection in 96-Well Plates.

Ten-day-old seedlings were removed from the growth chamber 4 h after the beginning of the light period and kept in the dark for 30 min before elicitation. For the rest of the assay, plates were kept in the dark. Every plate contained 12 wells containing Col-0 wild-type seedlings in row A. Each plate also contained seven different T-DNA insertion lines in rows B to H, (12 seedlings per row; 1 seedling per well). In columns 1 to 6, seedlings were treated with water. In columns 7 to 12, seedlings were treated with peptides (10 μL, in water). A second plate containing the same distribution of wild-type and T-DNA lines was elicited with peptides (10 μL) in columns 1 to 6 and water in columns 7 to 12. After the addition of 10 μL of water or peptides, plates were centrifuged briefly at 30 × g (Beckman Coulter Allegra ×22 swinging arms centrifuge) to ensure that the added peptides were distributed into the medium and that seedlings were exposed to peptide. Immediately after centrifugation, 10 μL of a Luminol-HRP solution in 100 mM K₂/KPO₄ buffer pH 7.9 [0.5 μg/mL Luminol (A4685) plus 0.5 μg/mL Type VI-A HRP from Sigma (P6782)] was added to each well and the plates were briefly centrifuged again. Plates were placed into the 96-well scintillation reader and light emission was monitored using a 96-well scintillation counter (1450 Microbeta Wallac TriLux Scintillation/Luminescence counter). Every well was read for a total of 5 s in noncoincidental mode. Every plate was read in full 20 to 25 times (every 2.5 min) for a total of 40 to 50 min. Kinetics of H₂O₂ production were determined by integration of data for every well over the reading period. Every time point is the mean value of six seedlings (either mock or peptide elicited).

MAMP-Triggered Enhanced Resistance Assay in 96-Well Plates.

Ten-day-old seedlings in 96-well assay plates were grown and arranged in the assay plates and either mock-treated or elicited with peptides as described above for the oxidative burst assays. Seedlings were grown for an additional 24 h after the peptide or mock treatment and then inoculated with 10 μL of P. syringae pv. maculicola strain ES4326 (OD₆₀₀ = 0.0002) carrying the LUX operon from P. luminescens (Psm-LUX) (14). Bacterial growth was carried out as indicated in SI Methods. Inoculated seedlings were grown for 20 h and then transferred hourly to the 96-well scintillation counter for light quantification, as described above. Reads were repeated every hour for a total of 6 h. Kinetics of bacterial growth was determined by integration of data for each well. Each time point is the mean value of six seedlings (either mock or peptide elicited).

MAMP-Triggered Enhanced Resistance Assay in 12-Well Plates.

Ten-day-old seedlings, germinated and grown in 12-well plates, as described above, were mock-treated with water or treated with peptides for 24 h, inoculated with 100 μL of Psm-LUX (OD₆₀₀ = 0.002), and incubated an additional 36 h, after which seedlings from each well were removed, quickly dried on paper towels, and transferred to a sterile 2-mL Eppendorf tube. Samples were weighed (to calculate fresh weight) and 400 μL of sterile water plus one 5-mm stainless steel bead was added to each tube. Seedlings were ground with a TissueLyser at 25 shakes per second for 3 min. Aliquots of 100 μL from the ground seedlings were transferred to 96-well plates for light quantification, as desribed above. Enhanced resistance against X. campestris pv. campestris (Xcc strain 33919) was assessed by eliciting seedlings with Flg22 (1 μM) or axY^s22-A1 (10 μM) for 24 h and then inoculating with 10 μL of Xcc (OD₆₀₀ = 0.002). Bacterial growth was carried out as indicated in SI Methods. Seedlings were blotted dry and ground 36 h after inoculation as described above. Serial dilutions were plated on LB agar to determine CFUs.