Arabidopsis EDS1 Connects Pathogen Effector Recognition to Cell Compartment–Specific Immune Responses

See allHide authors and affiliations

Science  09 Dec 2011:
Vol. 334, Issue 6061, pp. 1401-1404
DOI: 10.1126/science.1211641


Pathogen effectors are intercepted by plant intracellular nucleotide binding–leucine-rich repeat (NB-LRR) receptors. However, processes linking receptor activation to downstream defenses remain obscure. Nucleo-cytoplasmic basal resistance regulator EDS1 (ENHANCED DISEASE SUSCEPTIBILITY1) is indispensible for immunity mediated by TIR (Toll–interleukin-1 receptor)–NB-LRR receptors. We show that Arabidopsis EDS1 molecularly connects TIR-NB-LRR disease resistance protein RPS4 recognition of bacterial effector AvrRps4 to defense pathways. RPS4-EDS1 and AvrRps4-EDS1 complexes are detected inside nuclei of living tobacco cells after transient coexpression and in Arabidopsis soluble leaf extracts after resistance activation. Forced AvrRps4 localization to the host cytoplasm or nucleus reveals cell compartment–specific RPS4-EDS1 defense branches. Although nuclear processes restrict bacterial growth, programmed cell death and transcriptional resistance reinforcement require nucleo-cytoplasmic coordination. Thus, EDS1 behaves as an effector target and activated TIR-NB-LRR signal transducer for defenses across cell compartments.

Plant nucleotide binding–leucine-rich repeat (NB-LRR) proteins constitute a large family of intracellular receptors mediating strain-specific disease resistance (1). Recognition of pathogen effectors causes NB-LRR activation through adenosine triphosphate (ATP)–driven conformational changes that lead to induction of antimicrobial defenses and localized host programmed cell death (1, 2). Structural counterparts of NB-LRRs (called NACHT- or NOD-LRRs) regulate innate immune responses and apoptosis in mammalian cells (3), but in neither system are the mechanisms connecting receptor activation to defense reprogramming well understood. A major class of plant NB-LRR receptor with N-terminal TIR (Toll–interleukin-1–receptor domain) homology has evolved to intercept effectors from many different pathogen types (1). The nucleo-cytoplasmic lipaselike protein ENHANCED DISEASE SUSCEPTIBILITY1 (EDS1) controls basal immunity by restricting growth of virulent pathogens (46). EDS1 also signals downstream of activated TIR-NB-LRR receptors, to control host cell death and transcriptional mobilization of defense pathways (79). In Arabidopsis, TIR-NB-LRR receptor RPS4 recognizes a Pseudomonas syringae type III secreted effector AvrRps4 (10), and EDS1 nuclear accumulation is a prerequisite step for RPS4 resistance and associated transcriptional reprogramming (11). Coordination of the nucleo-cytoplasmic EDS1 pools through nuclear pore complexes is necessary for full TIR-NB-LRR immunity, which implies a need for EDS1 mobility inside cells (6, 11, 12).

We examined the molecular and subcellular relation between bacterial AvrRps4 protein and Arabidopsis RPS4 and EDS1. After delivery to plant cells, AvrRps4 is cleaved to release an 11-kD C-terminal fragment (AvrRps4C), which is necessary and sufficient for eliciting RPS4 immunity (13, 14). Although RPS4 associates mainly with endomembranes (9), RPS4 nuclear accumulation and genetic cooperativity with a nuclear WRKY transcription factor domain–containing TIR-NB-LRR receptor, RRS1, is required for AvrRps4-triggered immunity (9, 1517). Cytoplasmic membranes have emerged as an important cell compartment in which RPS4 and at least one other TIR-NB-LRR protein (SNC1) are constrained by the tetratricopeptide repeat (TPR) protein SRFR1 (SUPPRESSOR OF rps4-RLD1) and a SRFR1-interacting cochaperone SGT1 to prevent autoimmunity (18, 19).

We determined in which subcellular compartment AvrRps4C activates host defenses by expressing AvrRps4 with a C-terminal yellow fluorescent protein (YFP) fusion alone or YFP attached to a eukaryotic nuclear localization (NLS), a nuclear export (NES), or respective mutated nls and nes sequences (20). In Agrobacterium-mediated transient expression assays of Nicotiana benthamiana leaves, AvrRps4C-YFP (as well as AvrRps4C-YFP-nls and AvrRps4C-YFP-nes) displayed a nucleo-cytoplasmic distribution, monitored by live-cell imaging (fig. S1). AvrRps4C-YFP-NLS was detected only in nuclei and NES-tagged AvrRps4C-YFP in the cytoplasm of N. benthamiana cells, which suggested that AvrRps4C can be forced into either compartment (fig. S1). Expression of the same constructs in Arabidopsis stable transgenic plants of a nonresponding eds1 mutant in Arabidopsis accession Columbia (Col eds1-2) conferred similar AvrRps4C distribution patterns (Fig. 1A), although YFP fluorescence was considerably lower than in the transient assays. None of the eds1 transgenic lines displayed stunting or necrosis, which are hallmarks of autoimmunity (7, 9, 21), consistent with eds1 failing to respond to AvrRps4C (Fig. 1B). Multiple independent transgenic eds1 lines for each construct were crossed with wild-type Col to introduce functional EDS1. All F1 progeny produced severely dwarf plants that died after 5 to 6 weeks, except those expressing AvrRps4C-YFP-NES, which grew normally, although slightly less well than Col wild type (Fig. 1B). AvrRps4C-YFP accumulation was similar in all lines (Fig. 1C). Therefore, active exclusion of AvrRps4C-YFP from host nuclei eliminates EDS1-dependent stunting of Arabidopsis.

Fig. 1

Consequences of AvrRps4C subcellular mislocalization in Arabidopsis. (A) Confocal live cell images of representative epidermal cells from leaves of 3-week-old stable transgenic Col eds1-2 plants expressing AvrRps4C-YFP under control of a cauliflower mosaic virus (CaMV) 35S constitutive promoter and fused at the C terminus to functional or mutated NLS/nls and NES/nes motifs, as indicated. Scale bars, 20 μm. (B) Growth of representative 4-week-old Col eds1-2 plants [described in (A) (top)] and hemizygous F1 progeny from a cross between the different Col eds1-2 transgenic lines and Col wild type (bottom). Scale bars, 2 cm. (C) Immunoblot analysis of total protein extracts separated by SDS-polyacrylamide gel electrophoresis (SDS-PAGE) from 4-week-old EDS1/eds1-2 F1 plants expressing AvrRps4C-YFP, AvrRps4C-YFP-NLS or -nls, or AvrRps4C-YFP-NES or -nes variants and probed with antibodies against green fluorescent protein (α-GFP). Ponceau S staining shows equal transfer of protein samples to the membrane.

We investigated whether nuclear targeting of AvrRps4C is required to trigger resistance to bacterial infection. After being tagged with the hemagglutinin (HA) epitope, AvrRps4-HA or the AvrRps4-HA-NLS or -nls or NES or -nes variants (for simplicity denoted AvrRps4C) were cloned into a bacterial effector delivery vector and expressed in P. syringae pv. tomato strain DC3000 (Pst) to allow secretion (fig. S2A) and release of HA-tagged AvrRps4 C termini after in planta processing (14, 22). We performed Pst infection assays on leaves of Arabidopsis Col and another RPS4-expressing resistant accession, Ws (10, 17). In Ws and Col eds1 mutants (eds1-1 and eds1-2, respectively), multiplication of all bacterial strains was as high as virulent Pst DC3000, which reflected their equal capacity to infect a nonresponding host (Fig. 2A and fig. S3A). Restriction of Pst/AvrRps4C-HA growth and disease symptoms was partially dependent on RPS4 in both accessions (Fig. 2A and fig. S3, A and B), as reported previously (9, 17). Pst/AvrRps4C-HA-NES grew to the same extent in wild-type Ws and Col as AvrRps4C-HA in the rps4 mutants (Fig. 2A and fig. S3A), which suggested that RPS4 resistance to AvrRps4C-HA-NES was compromised. Growth of all other strains was similar to AvrRps4C-HA on wild-type plants (Fig. 2A and fig. S3A). Thus, an AvrRps4C nuclear pool is required to trigger full resistance to Pst infection. Forced nuclear localization of AvrRps4C-HA-NLS did not circumvent the need for RPS4 or EDS1 in the immune response (Fig. 2A and fig. S3A) nor did it alleviate disease susceptibility of Col transgenic plants expressing a mutated RPS4 receptor that fails to accumulate in nuclei (RPS4-HSnls rps4-2) (fig. S3C) (9). Therefore, competence of AvrRps4C-HA-NLS in triggering resistance also requires nuclear RPS4.

Fig. 2

Bacterially secreted AvrRps4C-HA directed to Arabidopsis nuclei limits pathogen growth without inducing host cell death. (A) Four-week-old Ws, eds1-1, and rps4-21 plants were inoculated by being sprayed with virulent Pst or Pst strains expressing AvrRps4C-HA or AvrRps4C-HA-NLS or -nls or -NES or -nes variants. Bacterial numbers measured at 3 dpi are shown. All bacterial strains had similar entry rates measured at 3 hpi (data not shown). Standard errors (SE) were calculated from three biological samples per genotype. Asterisk (*) indicates significant differences (P < 0.05) compared with Pst AvrRps4C-HA infection of Ws (Student’s t test). Three independent experiments gave similar results. (B) Ion leakage measurements were made at the indicated time points in leaf discs of 4-week-old Ws, eds1-1, and rps4-21 plants after infiltration with Pfo-expressing AvrRps4C-HA. Error bars represent SE of four samples per genotype. Three independent experiments gave similar results. (C) Four-week-old Ws plants were infiltrated with Pfo alone or Pfo strains delivering AvrRps4C-HA or AvrRps4C-HA-NLS or -nls or NES or -nes variants, and ion leakage was measured at the indicated time points, as described in (B). Weak programmed cell death elicitation by AvrRps4C-HA-nls is likely because of incomplete loss of “nls” nuclear targeting (fig. S1). Error bars represent SE of four samples per bacterial strain. The experiment was performed three times with similar results.

Host cell death at infection sites is often used as an indicator of functional effector recognition, although restriction of pathogen growth does not always correlate with cell death (2326). We measured whether the different levels of resistance to bacteria expressing AvrRps4C-HA-NLS and AvrRps4C-HA-NES mirrored cell death in Col and Ws. In order not to confuse NB-LRR–triggered cell death with disease-associated necrosis, the different AvrRps4C-HA variants were delivered through the type III secretion system of a noninfectious bacterial strain (Pseudomonas fluorescens, Pfo) (fig. S2B) (20). Cell death, quantified by electrolyte leakage from leaves after Pfo/AvrRps4C-HA infiltration, was abolished in eds1 and reduced in rps4 mutants (Fig. 2B, fig. S4, A and B). Partial cell death elicitation by AvrRps4C-HA-NES correlated with intermediate Pst growth suppression (compare Fig. 2C, Fig. 2A, and fig. S3A). By contrast, AvrRps4C-HA-NLS failed to elicit cell death (Fig. 2C), although it induced full resistance (Fig. 2A, fig. S3A). These results show that host-programmed cell death depends on cytoplasmic AvrRps4C-HA and that suppression of bacterial growth can be uncoupled from cell death by mislocalizing AvrRps4C-HA to nuclei. Weak cell death elicitation by Pfo-delivered AvrRps4C-HA-NES (Fig. 2C) points to a need for cooperativity between nuclear and cytoplasmic AvrRps4 for this defense output. With its nucleo-cytoplasmic signaling partner PAD4 (Phytoalexin Deficient 4), EDS1 drives transcriptional amplification of defenses, ensuring “reinforcement” of resistance around local NB-LRR resistance foci and induction of systemic immunity (27, 28). We therefore measured the competence of Pst-delivered AvrRps4C-HA-NLS or AvrRps4C-HA-NES forms in triggering EDS1-dependent transcriptional reprogramming and systemic resistance (11, 28). Compared with Pst/AvrRps4C-HA, both mislocalized AvrRps4C-HA forms failed to induce gene expression changes (fig. S5) or systemic immunity (fig. S6), underscoring the importance of nucleo-cytoplasmic AvrRps4C for transcriptional reinforcement of resistance.

Because EDS1 is central to qualitatively and spatially different AvrRps4-elicited resistance outputs, we looked for molecular association between AvrRps4C and Arabidopsis EDS1 using fluorescence lifetime imaging microscopy (FLIM). In nuclei of N. benthamiana cells transiently coexpressing AvrRps4C-CFP (with cyan fluorescent protein, CFP) and EDS1-YFP, a significant reduction of CFP lifetime because of a transfer of energy between the two fluorophores was detected (Fig. 3A,B), which suggested that AvrRps4C and EDS1 interact in this subcellular compartment. EDS1 also immunoprecipitated with AvrRps4C-HA in Arabidopsis Col leaf extracts expressing dexamethasone (dex)-induced AvrRps4-HA (pDex:AvrRps4-HA) (13) but not with TIR1-HA-StrepII (TIR1-HS) (36) protein from a control transgenic line (Fig. 4A). AvrRps4C association with EDS1 could reflect effector targeting of EDS1 in order to disable a key basal resistance regulator, which might then be intercepted indirectly by RPS4 (2, 5). However, independent studies suggest that EDS1 signals downstream of activated TIR-NB-LRR receptors (5, 79). Notably, autoimmunity caused by RPS4 overexpression in Arabidopsis does not even marginally circumvent a requirement for EDS1 (9). We therefore examined whether EDS1 might serve both as a target or “bait” for AvrRps4C and RPS4 signal transducer by also interacting with RPS4. Whereas no transfer of energy was detected between AvrRps4C-CFP and YFP-RPS4, a significant reduction of the CFP lifetime was observed in N. benthamiana nuclei coexpressing EDS1-CFP and YFP-RPS4 (Fig. 3, A and B), which suggested that EDS1 and RPS4 interact. Further evidence for RPS4-EDS1 association within a complex was obtained from immunoprecipitation (IP) experiments in Arabidopsis. IPs were performed with α-HA or α-EDS1 antibodies on extracts of Col plants overexpressing RPS4-HA-StrepII (RPS4-HS). At 19° to 22°C, these plants are stunted because EDS1-dependent immunity is triggered through RPS4 activation (9). Plants can be propagated at 28°C because of suppression of the immune response and reduced EDS1 expression (29). After shifting the plants to 19°C, EDS1 immunoprecipitated with RPS4-HS but not with control TIR1-HS (Fig. 4B and fig. S7A), which suggested that some EDS1 resides in a complex with activated RPS4-HS. Neither the presence of EDS1 nor dex-induced AvrRps4-HA substantially changed the mainly microsomal distribution of RPS4 in fractionation experiments (9). We therefore tested whether a subpool of EDS1 interacts with AvrRps4C or RPS4 associated with membranes. Although dex-induced AvrRps4C-HA distributed equally between soluble and microsomal fractions, EDS1 was almost entirely soluble and an EDS1 AvrRps4C-HA complex was detected in the soluble fraction (Fig. 4C). After shifting the RPS4-HS line to 19°C, RPS4-HS protein was mainly microsomal, and an EDS1 IP signal was detected in the soluble phase (Fig. 4C and fig. S7B). No interaction of EDS1 in either fraction was observed with TIR1-HS (fig. S7C). These results suggest that in resistance-activated Arabidopsis extracts EDS1 resides in soluble, and thus potentially mobile, complexes with RPS4 and/or AvrRps4C.

Fig. 3

EDS1 interacts with AvrRps4C and RPS4 in nuclei of N. benthamiana leaf cells. (A) Subcellular localization of transiently expressed specific proteins in epidermal cells after Agrobacterium-mediated transformation. Expression of the respective proteins alone or together did not cause cell death. Confocal images were taken 48 hours after infiltration. Scale bars, 20 μm. (B) FRET-FLIM measurements show that EDS1 interacts with AvrRps4C and RPS4 in nuclei of N. benthamiana cells. No interaction was observed between AvrRps4C-CFP and YFP-RPS4 or EDS1-CFP and YFP. Interaction between EDS1-CFP and its known nuclear interactor SAG101-YFP was measured as a positive control. (τ, CFP mean lifetime in nanoseconds; sem, standard error of the mean; N, total number of measured nuclei; E, percentage of FRET efficiency).

Fig. 4

Arabidopsis EDS1 resides in complexes with AvrRps4C and RPS4. (A) EDS1 IP with dex-induced AvrRps4C-HA in Arabidopsis total leaf extracts. Input samples were taken from Col, a control line (pTIR1:TIR1-HS), 8 hours after spray infection with Pst/AvrRps4 or pDex:AvrRps4-HA plants 48 hours after dex treatment. IPs were performed with antibody against HA (α-HA). Antibodies used to detect proteins on immunoblots of SDS-PAGE separated total (input) and eluted (IP) samples are indicated. Asterisk (*) nonspecific signal cross-reacting with α-EDS1; arrowheads indicate detected proteins. (B) EDS1 IP with constitutively expressed RPS4-HS in Arabidopsis total leaf extracts. Input samples, pTIR1:TIR1-HS 8 hours after spray infection with Pst/AvrRps4 and p35S:RPS4-HS 8 hours after shifting plants from 28° to 19°C, were immunoblotted as in (A). (C) EDS1 IP with AvrRps4C and overexpressed RPS4 in the soluble fraction of Arabidopsis leaf cells. Total (T) extracts of pDex:AvrRps4-HA plants 48 hours after dex induction and p35S:RPS4-HS plants 8 hours after shifting to 19°C were separated into soluble (S) and microsomal (M) fractions. Fractions were immunoprecipitated with α-HA and processed as in (A). VAMP722 antibody was used as a marker for the microsomal fraction and an antibody against phosphoenolpyruvate carboxylase (PEPC) for the soluble fraction, as indicated. Asterisk (*) nonspecific cross-reacting signal. (D) Model for different subcellular immune outputs triggered by RPS4 with EDS1 in response to AvrRps4C (see text).

Plant NB-LRR receptors have evolved to recognize host cell interference by pathogen effectors in successive rounds of defense and counterdefense (1). Mechanisms underlying receptor activation, often with host cofactors that serve as effector baits or decoys, have begun to emerge (13). We provide evidence that Arabidopsis TIR-NB-LRR receptor RPS4 engages EDS1 to intercept a pathogen effector and transduce receptor activation to downstream defenses. All tested TIR-NB-LRR receptors converge genetically on EDS1 to trigger resistance (5). This may reflect collective TIR-NB-LRR guarding of a basal resistance regulatory hub against diverse pathogen effectors. An intrinsic basal resistance signaling activity of EDS1 might then have been co-opted by TIR-NB-LRRs for effector-triggered immunity. This notion is supported by Bhattacharjee et al. (30) who find that targeting of EDS1 by two sequence-unrelated bacterial effectors, AvrRps4 and HopA1, is intercepted by TIR-NB-LRR receptors RPS4 and RPS6, respectively.

Our work and other studies (2326) show that plant immunity can be effective without host programmed cell death, and therefore death per se does not define resistance. This has important implications for determining functional receptor-effector interactions. In resistance triggered by AvrRps4C, we establish that nuclear processes are essential for bacterial growth restriction, which reinforces earlier findings that nuclear accumulation of RPS4 (9) and EDS1 (11) is necessary for AvrRps4-triggered immunity. Nucleo-cytoplasmic AvrRps4C is, however, required to elicit host cell death and transcriptional defense amplification (figs. S5 and S6). We propose that bacterial AvrRps4C triggers distinct, but coordinated, subcellular defense branches through an RPS4-EDS1 receptor signaling complex that can accumulate in the cytoplasm and nucleus (Fig. 4D). At least one nuclear step is needed for resistance. Nuclear targeting of AvrRps4C may skew the host response toward nuclear antimicrobial defenses sufficient for local bacterial containment, thereby circumventing host cell death. A cytoplasmic component (possibly a modification or cofactor generated by an initial cell-death stimulus) and, crucially, nucleo-cytoplasmic coordination are required for transcriptional amplification of defenses leading to systemic immunity (28). This scheme is supported by results of Bhattacharjee et al. who detect RPS4-EDS1 and AvrRps4-EDS1 complexes associated with microsomes and inside nuclei of N. benthamiana cells (30). In their analysis, coexpression of AvrRps4 alters RPS4-EDS1 association at the microsomes (30). In Arabidopsis, presence of “activated” RPS4-EDS1 complexes in the soluble phase may underlie effective triggering of defenses in different cell compartments. Importance of subcellular dynamics was shown in activation of the TIR-NB-LRR receptor N by a viral helicase effector in tobacco cells (31). Cytoplasmic recognition of potato virus X viral coat protein by the nucleo-cytoplasmic NB-LRR receptor Rx is also critical for eliciting cell death and resistance (32, 33), which underscores the importance of cytoplasmic processes in programmed cell death. Plants may therefore be in a position to tune subcellular defense pathways for best effect against a particular mode of pathogen attack. Association between RPS4 and EDS1 as an effector-triggered receptor signaling module offers a new mechanism for innate immunity in which intracellular receptors employ a single key basal defense regulator to coordinate diverse immune outputs.

Supporting Online Material

Materials and Methods

Figs. S1 to S7

References (3455)

References and Notes

  1. Materials and methods are available as supporting material on Science Online.
  2. Acknowledgments: We thank J. Bautor for excellent technical assistance; J. Jones and K. H. Sohn (Sainsbury Laboratory, Norwich, UK) for the pEDV6 vector and advice on bacterial delivery assays; A. Jauneau for advice and technical support in fluorescence resonance energy transfer (FRET)–FLIM assays (FR3450); and C. Brière for statistical analysis of FRET-FLIM data (UMR CNRS-UPS 5546). This work was funded by The Max-Planck Society, an IMPRS PhD fellowship (K.H.), Deutsche Forschungsgemeinschaft SFB 670 grant (J.E.P.) and is part of the Laboratoire d`Excellence (LABEX) TULIP (ANR-10-LABX-41) program (L.D.).

Stay Connected to Science

Navigate This Article