A Type I–Secreted, Sulfated Peptide Triggers XA21-Mediated Innate Immunity

See allHide authors and affiliations

Science  06 Nov 2009:
Vol. 326, Issue 5954, pp. 850-853
DOI: 10.1126/science.1173438

This article has been retracted. Please see:

Bacterial Trigger of Plant Protection

Innate immunity can be rapidly activated to defend a host plant against a microbial pathogen. The rice protein XA21, which is thought to be a cell surface–located receptor with a kinase domain, activates the plant's defenses in response to infection by certain strains of Xanthomonas bacteria. Lee et al. (p. 850) have now identified the bacterial gene that encodes the protein, AvrXA21, to which the plant receptor XA21 responds. The 194–amino acid protein needs to be secreted and sulfated to trigger the rice plant defense responses. Similarities exist between the receptor XA21 and other immune response receptors in both plants and animals.


The rice Xa21 gene confers immunity to most strains of the bacterium Xanthomonas oryzae pv. oryzae (Xoo). Liquid chromatography–tandem mass spectrometry analysis of biologically active fractions from Xoo supernatants led to the identification of a 194–amino acid protein designated Ax21 (activator of XA21-mediated immunity). A sulfated, 17–amino acid synthetic peptide (axYS22) derived from the N-terminal region of Ax21 is sufficient for activity, whereas peptides lacking tyrosine sulfation are biologically inactive. Using coimmunoprecipitation, we found that XA21 is required for axYS22 binding and recognition. axYS22 is 100% conserved in all analyzed Xanthomonas species, confirming that Ax21 is a pathogen-associated molecular pattern and that XA21 is a pattern recognition receptor.

In 1995 we showed that the rice Xa21 resistance gene, which encodes a protein with predicted leucine-rich repeat (LRR), transmembrane, juxtamembrane, and intracellular kinase domains, conferred immunity to diverse strains of the Gram-negative bacterium Xanthomonas oryzae pv. oryzae (Xoo) (1, 2). Subsequent discoveries in flies (3), humans (4), mice (5), and Arabidopsis (6, 7) revealed that animals and other plant species also carry membrane-anchored receptors [Toll in flies; Toll-like receptor 4 (TLR4) in mice and humans] with striking structural similarities to XA21 and that these receptors are also involved in microbial recognition and defense. Like XA21, these receptors typically associate with or carry non-RD (non–Arg-Asp) kinases to control early events of innate immunity signaling (8). Arabidopsis FLS2 (flagellin-sensitive 2) and EFR (elongation factor receptor) belong to the same class of plant receptor kinases (the LRRXII) as XA21 (8, 9).

Many of these cell surface receptors were later named pattern recognition receptors (PRRs) on the basis of their ability to directly recognize molecules that are conserved across a large class of microbes (10, 11). Such microbial molecules were called pathogen-associated molecular patterns [PAMPs, also known as microbe-associated molecular patterns (MAMPs)] (12).

Despite the similarity of the known PRRs to XA21, the classification of XA21 has been debated (13, 14). This is partly because XA21 was discovered before the terms PRR and PAMP were established (12) and partly because, under the classical definition of Flor (15), XA21 was called a “resistance” gene. Furthermore, because the molecule recognized by XA21 [previously called avrXa21 (avirulence Xa21) and here renamed Ax21 (activator of XA21-mediated immunity)] had not been identified, it was not known whether this molecule was conserved among a large class of microbes—a hallmark of PAMPs (12).

We previously identified six Xoo genes required for ax21 activity (the rax genes), which fall into two functional classes. The first class consists of three genes (raxA, raxB, and raxC) that encode components of a bacterial type I secretion system (TOSS) (16). The second class is involved in sulfation, including raxST, which encodes a protein with similarity to mammalian tyrosine sulfotransferases (16). Xoo strains carrying mutations in any of these rax genes no longer activate XA21-mediated immunity. None of the identified genes encodes an obvious activator of immunity.

To identify Ax21, we fractionated the supernatant of Xoo strain PXO99 cultures on a C18 reversed-phase high-performance liquid chromatography (RP-HPLC) column (Fig. 1A) and carried out bioassays of seven HPLC peptide-enriched fractions (Fig. 1B) using our previously established methods (17). An active fraction that was able to trigger XA21-mediated immunity (Fig. 1) was subjected to liquid chromatography–tandem mass spectrometry (LC-MS/MS) (18). Fifteen peptides from the LC-MS/MS spectra matched eight Xoo proteins (18), including two peptides that corresponded to the N-terminal and C-terminal regions of a 194–amino acid protein encoded by PXO_03968 (boxes in Fig. 1C and fig. S1).

Fig. 1

Isolation of Ax21. (A) RP-HPLC elution profile of peptides secreted from Xoo strain PXO99 (carrying Ax21 activity). Peptide-enriched samples from the PXO99 supernatant were separated on a reversed-phase C18 column (1 × 250 mm, flow rate 0.05 ml/min) with a 10 to 90% acetonitrile gradient containing 0.1% trifluoroacetic acid. A280, absorbance at 280 nm. (B) Lesion length measurements of XA21 rice leaves pretreated with RP-HPLC fractions followed by inoculation with PXO99ΔraxST. Lesion lengths were measured 12 days after PXO99ΔraxST inoculation. Each value is the mean ± SD from nine inoculated leaves. (C) Deduced amino acid sequence of Ax21. The two peptides (boxed) identified from the biologically active fraction were sequenced using LC-MS/MS. Predicted sulfated tyrosines Y22 and Y144 are underlined. The dashed box indicates one of the peptide used in the Ax21 bioassay shown in Fig. 3. (D) Mass (LTQ) spectrum of the axY22 peptide corresponding to the N-terminal region [first box in (C)] of Ax21. The spectrum corresponding to the peptide derived from the C-terminal region [second box in (C)] of Ax21 is shown in fig. S1.

To identify which gene encodes Ax21, we generated Xoo strains carrying a mutation in each of the individual genes. Whereas a PXO_03968 knockout strain caused long lesions and grew to high levels on XA21 leaves (Fig. 2), none of the other strains did (fig. S2). These data indicate that the PXO_03968 gene encodes Ax21. We further showed that Ax21 secretion requires raxA and raxC (fig. S3) (18).

Fig. 2

A mutation in ax21 abolishes Ax21 activity. (A) Lesion lengths of rice leaves measured 12 days after inoculation with Xoo strains PXO99, PXO99ΔraxST, or PXO99Δax21. Suspensions of each strain [108 colony-forming units (CFU)/ml] were scissor-inoculated onto rice leaves (TP309-XA21, resistant to PXO99; TP309, susceptible to PXO99). Images are representative of five independent experiments. (B) Growth of PXO99, PXO99ΔraxST, and PXO99Δax21 populations in inoculated rice leaves. Bacteria were extracted from the leaves at 0, 3, 6, 9, and 12 days after inoculation, plated on selective media after serial dilution, and colonies counted after a 3-day incubation at 28°C. Each value is the mean ± SD from nine inoculated leaves.

To test the importance of the putative tyrosine sulfation sites on Ax21 (19), we synthesized seven peptides: two carrying sulfated tyrosines in the target residues [Tyr22 and Tyr144 (Y22 and Y144)], two carrying nonsulfated tyrosines, two carrying alanines in place of the tyrosines, and one corresponding to the C-terminal region of Ax21 (Fig. 3A) (19). XA21 rice leaves were pretreated with each peptide (100 μM in water). The 17–amino acid peptide carrying Y22 sulfation (axYS22) activated XA21-mediated immunity (Fig. 3B). To further quantify this response, we characterized the activity of the axYS22 synthetic peptide by growth curve analysis. Pretreatment of XA21 rice leaves with the axYS22 peptide triggered resistance to PXO99ΔraxST, as reflected in a reduction in PXO99ΔraxST population growth by three orders of magnitude. The nonsulfated peptide (axY22) was unable to trigger XA21-mediated immunity (Fig. 3C).

Fig. 3

The axYS22 peptide is sufficient to trigger XA21-mediated immunity. (A) Synthetic peptides, including three corresponding to the N-terminal region of AX21 (axYS22, axY22, and axY22A), three corresponding to the central region (axYS144, axY144, and axY144A), and one corresponding to the C-terminal region (axM178), were tested for activity. Abbreviations for amino acid residues: A, Ala; D, Asp; E, Glu; F, Phe; G, Gly; I, Ile; K, Lys; L, Leu; M, Met; N, Asn; P, Pro; R, Arg; S, Ser; T, Thr; V, Val; W, Trp; Y, Tyr. (B) Five hours after peptide pretreatment, leaves were inoculated with PXO99ΔraxST and the lesions measured 12 days later. Each value is the mean ± SD from six leaves. (C) Growth of PXO99ΔraxST populations over time. TP309-XA21 leaves were pretreated with PXO99 supernatant (PXO99sup), water, or the synthetic peptides (axYS22 and axY22, 100 μM each). Bacterial cells were extracted from the leaves at 0, 5, 10, and 15 days after inoculation, plated on selective media after serial dilution, and colonies counted after a 3-day incubation at 28°C. Each value is the mean ± SD from eight inoculated leaves.

Bioassays with 17 axYS22 peptide variants carrying alanine substitutions identified eight amino acids critical for XA21-mediated immunity (fig. S4A) (18). A concentration of 1 μM is sufficient for PAMP activity (fig. S4B) (18).

In coimmunoprecipitation experiments with hemagglutinin (HA)–tagged axYS22 and extracts from leaves carrying a Myc-tagged XA21 protein (18), we observed labeling of a band migrating at 140 kD by SDS–polyacrylamide gel electrophoresis (PAGE) with antibodies to both Myc and HA (Fig. 4). The presence of 5- to 10-fold excess untagged axYS22 peptide suppressed the labeling of this band, whereas flg22ave from the rice pathogen Acidovorax avenae had no effect on axYS22-XA21 binding (Fig. 4). These experiments demonstrate that XA21 is required for axYS22 binding and recognition.

Fig. 4

XA21 is required for axYS22 binding. HA-tagged axYS22 cross-links to a 140-kD polypeptide that is immunoprecipitated by an antibody to Myc (Myc-XA21). (A) Before immunoprecipitation, the loading of equal amounts of protein (50 μg) from Kitaake and Myc-XA21 leaf extracts was confirmed using an antibody to actin (input). (B) Leaf extracts were incubated with 1 mM HA-axYS22 in the presence (+, 5 mM; ++, 10 mM) or absence (–) of the competitors axYS22 lacking the HA tag or flg22ave. After binding, cross-linking was initiated by the addition of sulfo-Ethylene Glycol bis (Succinimidyl Succinate). Duplicate protein gels were analyzed after separation by SDS-PAGE using antibodies to Myc (top) and to HA (bottom). Myc-XA21 and a proteolytic cleavage product of Myc-XA21 were detected at 140 and 110 kD, respectively, as reported previously (32). Arrows indicate the XA21 and Ax21-XA21 complexes.

Sequence analysis indicates that Ax21 is highly conserved in Xoo strains (KACC 10331 and MAFF 311018, both 98% identity), X. campestris pv. campestris (90%), X. axonopodis pv. glycinea (92%), X. axonopodis pv. vesicatoria 85-10 (Xav) (92%), and X. oryzae pv. oryzicola (98%) (fig. S5). The 17–amino acid axYS22 sequence is 100% conserved in these strains. Xylella fastidiosa and the opportunistic human pathogen Stenotrophomonas maltophilia also carry putative Ax21 orthologs (48% and 61%, respectively) and show 77% and 65% similarity, respectively, to the axYS22 sequence (fig. S5).

Because Xav carries predicted orthologs for Ax21, raxST, raxA, and raxB, which we have previously shown to be required for Ax21 activity (16), we hypothesized that Xav would express Ax21 activity. Indeed, we found that pretreatment of XA21 rice leaves with supernatants from wild-type Xav, but not Xav strains carrying a deletion of ax21 (18), can activate XA21-mediated immunity (fig. S6). These results indicate that the ax21 ortholog in Xav possesses the predicted biological activity.

One of the key aspects of the definition of PAMPs is that they “are conserved within a class of microbes” (10). As a result of the explosion of studies on PRRs and PAMPs in both plant and animal systems, it has now become clear that PAMPs can be conserved quite widely across genera (e.g., flagellin) or more narrowly within a genus (e.g., Pep13) (20) and, further, that sequence variation and posttranslational modifications can modulate PRR-dependent pathogen recognition (18, 21).

In the XA21-Ax21 system, the axYS22 peptide sequence is invariant in all sequenced Xanthomonas species. Sulfation provides specificity to the system, just as flagellin or lipopolysaccharide recognition in some hosts is modulated by glycosylation or acylation, respectively (21, 22). Thus, Ax21 is a PAMP that satisfies the genetic definition of an avirulence factor because the presence or absence of sulfation on the conserved 17–amino acid epitope is decisive for its ability to trigger XA21-mediated immunity. Similarly, Xa21 is a disease resistance gene because it is the single polymorphic determinant in rice that confers resistance to strains of bacteria expressing sulfated Ax21, and it is also a PRR because it is required for recognition of a particular modified peptide epitope that is conserved across a microbial genus.

Thus, our data provide another example of bacteria-host interactions that can be attributed to the presence of genes encoding proteins (e.g., sulfotransferases, glycosylases, and acetylases) that modify conserved peptide epitopes (2125). Such examples indicate that successful pathogens of plants and animals have evolved methods of altering the PAMP to avoid detection by the host PRR. Conversely, the presence or absence of a particular PRR can have a marked effect on the resistance of the host to infection. Just as plants deficient in XA21 or FLS2 exhibit reduced resistance to phytopathogens, mice deficient for TLR4 or TLR2 are altered in their response to Mycobacterium tuberculosis infection (22). These studies have led to a convergence in our understanding of the molecular mechanisms governing the specificity of host-microbe interactions in plants and animals.

Our results thus demonstrate that the definitions of PAMPs and avr genes and those of disease resistance genes and PRRs cannot be strictly separated. Future usage of these terms should reflect the concepts presented by Medzhitov (12) and Flor (15), and must also take into account subsequent discoveries of the effects of PAMP polymorphism and posttranslational modifications. In plants, a disease resistance gene is merely an allele in the host genotype that confers resistance in a particular interaction. Such genes can encode diverse proteins (although, to date, the great majority encode intracellular nucleotide-binding LRR proteins) (2, 2628). Hence, the term “R gene” is merely operational and mechanistically agnostic.

Likewise, the term “avirulence gene” remains useful as a broad term that indicates a gene that encodes any determinant of the specificity of the interaction with the host. Thus, this term can encompass some PAMPs and pathogen effectors (e.g., bacterial type III effectors and oomycete effectors) as well as any genes that control variation in the activity of those molecules (2931)].

In the future, a diverse array of PAMPs from plant pathogens will likely be discovered. Many of these will almost certainly serve as ligands for the large class of predicted orphan PRRs present in the genomes of plant species (371 in rice; 47 in Arabidopsis) (8).

Supporting Online Material

Materials and Methods

Figs. S1 to S6

Tables S1 and S2


  • * These authors contributed equally to this work.

References and Notes

  1. See supporting material on Science Online.
  2. We thank B. Macek, M. Mann, C. Bertozzi, and D. Kliebenstein for technical assistance and discussions; R. Eigenheer and B. Phinney for LC-MS/MS service and HPLC equipment; and M. S. Chern, X. Chen, L. Bartley, A. Bent, J. Dangl, J. Dubcovsky, and B. Tyler for helpful comments and critical reading of the manuscript. S.-W. Lee was partially supported by KOSEF through PMRC of Kyung-Hee Univ., Korea. Supported by the USDA Cooperative State Research, Education and Extension Service (National Research Initiative grants 006-01888 and 2007-35319-18397) and by NIH grant GM55962.
View Abstract

Navigate This Article