99th Dahlem Conference on Infection, Inflammation and Chronic Inflammatory Disorders: Innate immune responses in plants

Abstract

Plants rely exclusively upon mechanisms of innate immunity. Current concepts of the plant innate immune system are based largely on two forms of immunity that engage distinct classes of immune receptors. These receptors enable the recognition of non-self structures that are either conserved between members of a microbial class or specific to individual strains of a microbe. One type of receptor comprises membrane-resident pattern recognition receptors (PRRs) that detect widely conserved microbe-associated molecular patterns (MAMPs) on the cell surface. A second type of mainly intracellular immune sensors, designated resistance (R) proteins, recognizes either the structure or function of strain-specific pathogen effectors that are delivered inside host cells. Phytopathogenic microorganisms have evolved a repertoire of effectors, some of which are delivered into plant cells to sabotage MAMP-triggered immune responses. Plants appear to have also evolved receptors that sense cellular injury by the release and perception of endogenous damage-associated molecular patterns (DAMPs). It is possible that the integration of MAMP and DAMP responses is critical to mount robust MAMP-triggered immunity. This signal integration might help to explain why plants are colonized in nature by remarkably diverse and seemingly asymptomatic microbial communities.

Microbe-associated molecular pattern (MAMP)-triggered immunity

Exposure of plants to MAMPs leads to an increased ‘awareness’ for invading pathogens [1]. Different MAMPs, even those derived from distinct microbial classes, initiate stereotypic defence-associated responses. This physiological syndrome can be categorized broadly into rapid responses that are seen within 15 min, including an extracellular alkalinization, changes of calcium fluxes, an oxidative burst, the generation of ethylene and activation of mitogen-activated kinase (MAPK) cascades. Within 15–60 min a core set of approximately 100 defence-related genes is activated. Several hours later, anti-microbial compounds accumulate inside plant cells or are secreted, and callose, a (1,3) β-glucan, is laid down on the inner side of the plant cell wall.

The model plant Arabidopsis has the capacity to recognize at least three bacterial MAMPs derived from flagellin (flg22), elongation factor Tu (EF-Tu) (elf18) and lipopolysaccharide (LPS) (O-chain recognition of LPS). The former two MAMPs are detected by two leucine-rich repeat (LRR) domain kinases, flagellin sensitive 2 (FLS2) and elongation factor Tu receptor (EFR), which belong to the so-called LRR-XII class. The LPS receptor is still elusive. Fungal diseases are widespread in plant–microbe interactions. A classic fungal MAMP is chitin, a long-chain polymer of a N-acetylglucosamine and the main component of cell walls of higher fungi. A chitin binding protein, called chitin elicitor-binding protein (CEBiP), was purified from rice. CEBiP represents an extracellular glycoprotein with lysine motif (LysM) domains, a transmembrane domain, and a short cytoplasmic tail. LysM domains are important for chitin and peptidoglycan binding in animals. Arabidopsis insertion mutants disrupted in a gene encoding LysM-domain-containing proteins are non-responsive to chitin fragments [2]. The corresponding wild-type gene encodes a receptor-like kinase, designated CERK1 (or LysM RLK1), and consists of an extracellular part with three LysM domains, a transmembrane part and an intracellular protein kinase [2]. Because chitin binding to CERK1/LysM RLK has not been demonstrated, it is possible that Arabidopsis contains one or several CEBiP homologues that associate with CERK1/LysM RLK to form a functional receptor complex.

The FLS2 and EFR proteins, which detect, respectively, the presence of elicitor-active epitopes of bacterial flagellin and EF-Tu, serve as models for PRR-triggered immunity. For this reason, recent work on these receptors is described below in greater detail. In a landmark paper, Zipfel et al. (2004) showed that flg22 treatment of wild-type plants induced approximately 1000 genes within 30 min, but none in fls2 mutants. In addition, spray inoculation of fls2 plants with virulent Pseudomonas syringae permitted strongly enhanced bacterial growth in comparison to wild-type [3]. Thus, a single PRR can make a quantitative contribution to limit pathogen growth. The same work also underlines the importance of pathogen inoculation methods because infiltration of bacteria in leaves of wild-type and fls2 mutants resulted in indistinguishable bacterial titres. Only the former inoculation method requires bacterial motility for directional movement of P. syringae on the leaf surface towards stomata (natural openings of the leaf epidermis) and subsequent entry into the leaf interior for multiplication in the extracellular space. Of note, stomata close upon detection of flg22 and prevent infection of the leaf interior [4]. Although FLS2 is expressed in all organs and cell types of Arabidopsis, this points to the existence of a tissue- and/or cell-type specific immunity mechanism. This also explains why artificial pressure-infiltration of P. syringae into leaves of wild-type and fls2 mutants resulted in indistinguishable bacterial titres.

PRR-triggered signalling

Little is understood about the precise signalling events that initiate with the perception of MAMPs. However, conclusive evidence is available that the extracellular LRR domain of FLS2 contains the binding site for flg22 and related peptides [5]. FLS2 associates upon flg22 binding instantaneously with BAK1 (BRI1-associated kinase 1), which appears to function as co-receptor of the brassinolide hormone receptor BRI1. EFR also engages BAK1 for MAMP signalling [6,7]. Thus, brassinolide hormone signalling and FLS2/EFR-dependent immunity share the same co-receptor for signal initiation [7], raising questions of how brassinolide hormone and FLS2/EFR-triggered signalling pathways diverge downstream of BAK1. Whether extracellular binding of flg22 to the LRRs is able to induce phosphorylation of the cytoplasmic domain of FLS2 and/or BAK1 is still unknown. Using transgenic plants expressing functional FLS2-GFP, evidence for ligand-induced receptor endocytosis has been obtained. It seems likely that this receptor endocytosis is necessary to stimulate an effective immune response, but could also have a role in receptor desensitization.

Using protoplasts from wild-type and fls2 mutant plants, Asai et al. provided evidence for the involvement of a MAPK cascade which culminated in the transcriptional induction of two WRKY transcription factors (TFs), WRKY22 and WRKY29 [8]. Similarly, rapid ligand-induced and FLS2-dependent protein phosphorylation has been demonstrated for a large number of host proteins. However, it remains to be shown which of the many plant MAPKs, WRKY TFs and phosphorylated host proteins are necessary to limit pathogen growth in a whole organism context. For example, there are 72 expressed WRKY genes in Arabidopsis. The presence of WRKY TF binding sites (C/TTGACC/T, designated W boxes) in several co-regulated Arabidopsis defence gene promoters provides strong evidence that zinc-finger-type WRKY factors play a broad role in regulating immune responses. Moreover, WRKY TFs are functionally connected, forming a transcriptional network composed of positive and negative feedback loops and feed-forward modules [9]. Whereas some WRKY factors appear to hold central positions mediating fast and efficient activation of defence programmes, there is an urgent need for the establishment of a global in vivo map of protein–DNA interactions associated with early MAMP-triggered immunity to obtain deeper insights into transcriptional re-programming events.

Pathogen effectors intercept PRR-triggered immune responses

Host-adapted bacteria, fungi, and oomycete pathogens evolved means to sabotage PRR-triggered immune responses [10,11]. This is perhaps best documented for bacterial effectors that are delivered directly into the plant cytoplasm by the type-three secretion system (TTSS). Unlike MAMPs that are conserved typically between members of a class of microbes, these effectors are normally specific for a pathogen strain. Indeed, plants have evolved a perception system to recognize the presence of these effectors (see below).

I will describe briefly the work on two well-studied bacterial effectors of P. syringae DC3000, called AvrPto and AvrPtoB. These two proteins are structurally unrelated and are injected into the plant cytoplasm through the TTSS. Tomato plants that are immune to P. syringae DC3000 carrying AvrPto harbour a resistance gene, called Pto. Unlike many other molecularly isolated resistance genes encoding NB-LRR-type proteins, Pto encodes a cytoplasmic protein kinase. However, the AvrPto-Pto-dependent resistance response also requires the presence of the NB-LRR immune sensor Prf. Compelling evidence shows that Pto and Prf associate in vivo in a co-regulatory interaction that requires Pto kinase activity for immune signalling [12]. Remarkably, there are other conserved host targets of AvrPto and AvrPtoB, and these are the cytoplasmic kinase domains of PRR complexes. For example, upon transient expression in Arabidopsis protoplasts, AvrPto can associate with FLS2 and inhibits in vitro the protein kinase activity of FLS2 and EFR. However, as discussed above, flagellin-induced immunity through FLS2 depends on the instantaneous association with another receptor-like protein kinase, BAK1 (brassinosteroid-receptor 1 associated kinase 1). Importantly, BAK1 probably plays a similar role in the transmission of signals initiated by the EF-Tu PRR [6,7]. Indeed, AvrPto and AvrPtoB appear to target the shared co-receptor BAK1 rather than the PRRs themselves [13]. This was based on the observation that Arabidopsis plants expressing AvrPto under constitutive or inducible promoters exhibit a phenotype resembling bak1 or weak bri1 (brassinosteroid insensitive 1) mutants. Upon simultaneous expression of tagged AvrPto or AvrPtoB and BAK1 in Arabidopsis protoplasts, BAK1-AvrPto or BAK1-AvrPtoB complexes could be co-immunoprecipiated. Perhaps most convincingly, both AvrPto and AvrPtoB inhibited the flagellin-induced FLS2-BAK1 association in the protoplast system, indicating that the interaction between the effectors and BAK1 leads to the dissociation of the FLS2-BAK1 signalling complex. AvrPto mutants unable to dampen PAMP signalling neither associated with BAK1 nor did these disrupt the FLS2-BAK1 complex. This inhibitory effect of AvrPto and AvrPtoB on BAK1 function was also observed in vivo using bacteria lacking avrPto and avrPtoB, or by a hrcC mutant lacking the entire TTSS. These data illustrate that effectors can have multiple and different targets in different hosts. Despite these seemingly convincing findings, I remain sceptical about claims in studies on effector targets, as many data are based on non-physiological expression levels of effectors.

Strain-specific immunity triggered by R proteins

Phytopathology research was transformed in the 1950s, when H. H. Flor proposed the ‘gene-for-gene’ hypothesis to explain his observations of strain-specific plant–parasite interactions, based on flax and flax rust. From these studies Flor inferred that plants can recognize pathogens through the action of disease resistance (R) genes, which confer immunity to host-adapted pathogens expressing unique corresponding avirulence (Avr) genes. We now know that Avr genes represent a subset of the effector repertoire of host-adapted parasites. Like PRR-triggered immune responses, R protein-conditioned immunity is also linked to an oxidative burst and defence gene activation, but differs both quantitatively and kinetically from the former, leading typically to host cell death at attempted invasion sites. This ‘hypersensitive response’ (HR) is thought to limit the spread of infection. Because PRR- and R protein-triggered output responses are similar, it is possible that the signalling pathways converge. Most known R genes encode intracellular proteins with a tripartite architecture consisting of N-terminal CC or Toll/interleukin-1 receptor (TIR) domains, a central NB site and C-terminal LRRs [14]. These plant R proteins are related structurally to a class of immune sensors of the vertebrate innate immune system known as nucleotide-binding oligomerization domain (NOD)-LRR, NOD-like receptors (NLRs), NACHT-LRR or caspase-recruitment domain (CARD), transcription enhancer, R (purine)-binding, pyrin, lots of leucine repeats (CATERPILLER) proteins.

Patterns of positive selection in R and Avr genes drive the co-evolution of plant and pathogen populations

Several studies in plants showed that residues of the LxxLxLxx β-strand motif in the LRR domain of R proteins are under stronger positive selection than other parts of the polypeptide. Diversifying selection in the LRR domain has been reported for intracellular CC/TIR-NB-LRR R proteins encoded by allelic variants of Arabidopsis RPP13, flax L and wheat Pm3, as non-synonymous substitutions are overrepresented in comparison to changes in synonymous sites. However, diversifying selection was observed both in the LRR and the TIR-encoding region of flax L, supporting previous findings that for some L proteins resistance specificity is determined solely by the TIR domain. Diversifying selection has also been detected in the flax rust AvrL567 and downy mildew ATR13 loci, whose products are recognized by a subset of flax L and by Arabidopsis RPP13 proteins, respectively. The 3D structures of two AvrL567 effectors are very similar, with polymorphic residues located on their surface [15]. Direct interactions were detected in yeast two-hybrid experiments between corresponding pairs of AvrL567 and L [16]. Together this led to the proposal that direct binding/recognition of non-self structures by intracellular CC/TIR-NB-LRR R proteins is one molecular mechanism that governs the co-evolution of plant–microbe interactions, resulting in diversifying selection at corresponding R and Avr loci [15].

Contrasting observations were made for three R proteins, where the presence of pathogen effectors is detected indirectly by monitoring their activity. The Arabidopsis R genes RPM1, RPS2 and RPS5, each conferring resistance to Pseudomonas syringae strains expressing cognate Avr genes, exhibit low levels of polymorphisms including simple presence/absence polymorphisms. Thus, for these R proteins there is no evidence for functional diversification through the generation of allelic resistance specificities. This might be a consequence of their direct binding to other host proteins, which serve as targets of the corresponding bacterial effectors [17,18].

Autorepression and a catalytic cycle for NB-LRR-type immune sensors

The central NB and C-terminal LRRs are common modules found in plant R and vertebrate NLRs. In contrast, a structurally diverse range of domains was apparently co-opted during evolution N-terminal to the NB domain, including CC or TIR domains in plants and in vertebrates a CARD, or pyrin domain (PYD), or baculovirus IAP repeats (BIRs). The central NB domain is part of a larger domain, called NB-ARC, due to its occurrence in plant R proteins and the apoptotic regulators human apoptotic protease-activating factor 1 (APAF-1) and its Caenorhabditis elegans homologue CED-4. NB-ARC domain-containing proteins belong to the family of STAND (signal transduction ATPases with numerous domains) NTPases that are found in archaea, bacteria, fungi, plants and animals. STAND ATPases are modular proteins and display a wide range of fusions to domains involved in protein–protein or protein–DNA interactions, small-molecule-binding domains, as well as catalytic domains involved in signal transduction. These proteins are considered to act as regulatory signal transduction switches. A critical aspect of this switching is reversible, NTP hydrolysis-powered, conformational changes that are relayed to effector domains. STAND NTPases are unusual because the regulatory switch, scaffolding and, occasionally, sensory as well as signal generating moieties are integrated into a single multi-domain protein. A highly conserved MHD-motif (hxhHD) of plant R proteins is located in the ARC2 subdomain [19]. Mutagenesis of either the histidine or aspartate in several R proteins as well as human NOD2 results in autoactivation, indicating that these residues are important to keep the receptors in an inactive conformation. Biochemical analysis of two autoactivating mutations of the tomato I-2 R protein, which confers resistance to the fungal pathogen Fusarium oxysporum, showed markedly reduced ATP hydrolysis in vitro but did not affect nucleotide binding. This suggests that the ATP bound form is the ‘on state’ while ATP hydrolysis switches the protein back to the ‘off state’. Autoactivating mutations in plant R proteins and human NOD2 also map in the linker region between ARC2 and the LRR region, as well as in the N-terminal part of the LRRs. This points to the existence of additional receptor regions that keep the protein in an autorepressed form in the absence of a cognate pathogen effector.

LRR domains of R proteins are thought to be major determinants of effector recognition. This is supported by studies on several LRR domain-containing protein families, which are involved in ligand binding or protein–protein interactions such as the mouse Toll-like receptor TLR-4 or Arabidopsis TIR1 ubiquitin ligase. A typical LRR consists of 20–29 residues that form a β-strand often followed by an α helix, which are arranged parallel to a common axis. The conserved segment of 11 amino acids (LxxLxLxxN/CxL, where x stands for any residue and the L position can be occupied by any hydrophobic amino acid) of the β-strand/β-turn motif lies on the concave face of the solenoid structure. The leucine or related residues of the LRR motif point inwards and build a hydrophobic core that gives a lateral stabilization to the repeat architecture, as deduced from solved crystal structures. The convex face of the LRR is often formed by an α-helix. The crystal structures of LRR proteins revealed that the concave site, consisting of the β-strand motif and the adjacent turns, are the main site of protein interaction and ligand binding.

Domain swap experiments between the highly sequence-related potato CC-NB-LRR-type Rx and GPA2 R proteins produced autoactive variants upon inappropriate pairings of ARC2 and LRR domains, suggesting that intramolecular interactions between these two domains regulate the receptor's transition from an autorepressed to an active state. Together, this has led to a model in which the direct or indirect recognition of pathogen effectors by the polymorphic LRR region initiates a first conformational change. This facilitates exchange of ADP by ATP, which in turn is thought to trigger a second conformational change that renders the respective N-terminal effector domain (CC, TIR, CARD, PYR, BIR) accessible for associations with downstream targets. Subsequent ATP hydrolysis switches the receptor back to its autorepressed form [19].

Indirect recognition of effectors by R proteins

Two well-studied examples of indirect non-self recognition by R proteins will now be described. In Arabidopsis, the CC-NB-LRR-type protein RPM1 confers immunity against P. syringae expressing either of two sequence-unrelated type III effector proteins, AvrRpm1 or AvrB. Once Arabidopsis AvrRpm1 or AvrB are delivered into host cells, the plasma-membrane-associated RPM1-interacting protein (RIN4) is phosphorylated [18]. RPM1 is thought to detect this RIN4 modification and subsequently activates unknown signalling pathway(s) leading to defence gene expression and the onset of a HR [18]. Another Arabidopsis CC-NB-LRR-type R protein, RPS2, specifically detects and mounts an immune response to P. syringae isolates expressing the AvrRpt2 effector. This bacterial effector undergoes self-cleavage after delivery by the TTSS into host cells through the assistance of a plant encoded cyclophilin (a folding catalyst that facilitates cis/trans isomerization of prolyl bonds) and acts as cysteine protease [17]. While attempts to detect direct interactions between RPS2 and AvrRpt2 were unsuccessful, both were found to associate physically with Arabidopsis RIN4. A RPS2-RIN4 complex is constitutively present in healthy plants, but RIN4 disappears upon AvrRpt2 delivery into plant cells. Importantly, mutations in any of three C-terminal AvrRpt2 residues, predicted to be essential for catalytic activity of the Pseudomonas protease, disrupts its own processing in planta, the RPS2-dependent immune response, as well as RIN4 elimination. This led to the hypothesis that RPS2 might recognize the result of AvrRpt2's proteolytic activity, i.e. the removal of RIN4. Taken together, RIN4 appears to be a host target for multiple Pseudomonas effector proteins. Unfortunately, the role of RIN4 in cellular reprogramming during pathogenesis remains unclear. One idea is that Arabidopsis RIN4 and tomato Pto (see above) serve solely as ‘decoys’ in effector perception by R proteins. Indirect recognition of non-self (= recognition of modified self) by R proteins is an elegant alternative solution compared to direct non-self perception by the adaptive immune system in vertebrates. R protein-mediated surveillance of only those host protein assemblies that are critical for successful pathogenesis of parasites could have been an important step for plants to survive with a limited set of receptors per individual. Note that there are only ∼120 NB-LRR-type proteins in the Arabidopsis genome.

Nucleocytoplasmic trafficking and nuclear action of R proteins

How plant NB-LRR proteins activate immune responses following recognition of pathogen-derived effectors has been a major question since the molecular isolation of the founding family members. Recent findings point to nuclear actions of the CC- and TIR-type receptor families. Fractionation of cell extracts using transgenic plants that express native levels of the barley mildew-resistance locus A (MLA) CC-NB-LRR-type R protein as well as visualization of a fluorochrome-marked MLA variant localized the majority of the receptor to the soluble cytoplasmic fraction and approximately 5% to the nucleus [20]. Perturbation of nucleocytoplasmic MLA partitioning by expression of a receptor fusion protein containing a nuclear export signal (NES), which enhances nuclear export over import, abrogated MLA-specified disease resistance. Similarly, adding a NES to the tobacco TIR-type N receptor, which conditions immunity against the tobacco mosaic virus (TMV) upon recognition of the p50 TMV replicase, impaired both N nuclear accumulation and TMV disease resistance [21]. Nuclear action of MLA and N was unexpected, because both proteins lack a canonical nuclear localization signal (NLS). Unlike this, the Arabidopsis TIR-type RPS4 protein, conditioning immunity to P. syringae strains expressing avrRps4, contains a bipartite NLS and this targeting signal is required for both nuclear import and disease resistance. Similar to barley MLA, less than 10% of total cellular RPS4 was found in Arabidopsis nuclei preparations, while the bulk of the receptor associates with endosomes.

Transcriptional reprogramming of plant cells upon pathogen attack is extensive, affecting between 3 and 12% of the 24 000 Arabidopsis genes upon fungal or bacterial challenge, respectively. How the perception of non-self structures by PRRs and R proteins leads to transcriptional activation of defence response genes has been a long-standing question. In this context, nuclear activities of MLA, N, and RPS4 appear to be relevant. Quantitative fluorescence lifetime imaging of fluorochrome-tagged receptor was used to visualize in vivo in nuclei an effector-dependent physical association between the MLA receptor and two WRKY TFs (WRKY1 and WRKY2 TFs [20]), suggesting that the TFs serve as immediate downstream targets of the activated receptor. This protein–protein association is mediated by the invariant N-terminal CC domain of allelic MLA receptors. Because the polymorphic C-terminal LRR region of MLA has been shown to determine recognition specificity, it is possible that this region senses, directly or indirectly, the presence of fungal effectors, while the N-terminal CC of the activated receptor acts as a signal relay moiety to the WRKY TFs. Accordingly, different structural modules at opposite ends of the receptor might account for sensory and signal transmission subfunctions. While it remains to be seen whether MLA and RPS4 proteins detect the corresponding effectors in the cytoplasm and/or nucleus, the cytoplasmic pool of tobacco N appears to detect the TMV p50 viral effector. When the p50 effector was fused to the NES, thereby depleting the nuclear p50 pool and enforcing cytoplasmic localization, plant cells retained the ability to trigger N-mediated disease resistance [21]. Thus, sensory and signal transmission activities of N might take effect in different compartments.

A molecular link between R- and PRR-triggered immune responses?

The effector-dependent association between MLA and WRKY TFs appears to contribute to receptor-triggered disease resistance and host cell death at attempted fungal infection sites. However, the WRKY TFs interacting with MLA act as repressors of MAMP-triggered immune responses and might have a role in preventing ‘chronic inflammatory responses’ and/or to dampen immune responses below a threshold that is detrimental to attacked plant cells. It has been postulated that MLA receptors interfere with the WRKY repressor function, thereby de-repressing MAMP-triggered immune responses. The de-repression could amplify MAMP-triggered immune responses and, in principle, would be sufficient to drive plant cells into suicide. Thus, the effector-triggered MLA WRKY association could serve as nexus to integrate signals generated by PRRs and R proteins.

Unlike direct links between MLA and the transcriptional machinery, nuclear RPS4 activity requires EDS1, a protein of unknown biochemical function(s) that lacks known chromatin- or DNA-binding domains and resides in both cytoplasmic and nuclear compartments [22]. RPS4-triggered immunity, but not nucleocytoplasmic partitioning or receptor stability, is abolished in an eds1 null mutant background. Together with an almost complete breakdown of RPS4/EDS1-dependent activation/repression of approximately 130 defence-related genes in eds1 plants [22], this suggests that EDS1 acts as intermediary positive signal transducer between the receptor and defence gene expression.

Further evidence for transcription machinery-associated functions of plant immune sensors comes from a functional analysis of Arabidopsis RRS1, which conditions disease resistance to the bacterial pathogen Ralstonia solanacearum expressing the cognate effector PopP2. RRS1 is unusual because it encodes a TIR-NB-LRR R protein with a carboxy-terminal WRKY domain. The latter is shared by all WRKY transcription factor family members and is known to bind to cis-active DNA elements, termed W-boxes (see above). The TTSS-delivered effector PopP2 carries a bipartite nuclear localization signal and is targeted specifically to host cell nuclei. Transient gene expression experiments in Arabidopsis protoplasts using fluorochrome-tagged RRS1 and PopP2 demonstrated that nuclear visualization of RRS1 requires co-expression of PopP2, while expression of RRS1-GFP alone did not produce a fluorescence signal. As RRS1 and PopP2 were also shown to interact in yeast two-hybrid experiments, it is possible that the association with PopP2 either induces conformational changes in RRS1-GFP, thereby producing a detectable fluorescence signal, or that RRS1-GFP forms a heterocomplex, which is resistant to degradation. A 3 base pairs (bp) insertion mutation in RRS1 (synonym SHL1) that results in the addition of a single amino acid in the WRKY domain, thereby impairing its DNA-binding activity, leads to chronic expression of defence genes and occasional cell death in the absence of the parasite. One interpretation is that in healthy plants the wild-type protein must bind to DNA to repress plant defence gene expression. In this scenario, the association between RRS1 and PopP2 could serve as a trigger to sequester the R protein away from DNA, thereby allowing defence gene expression.

Because barley MLA, tobacco N, as well as Arabidopsis RPS4, each localize to the cytoplasm and nucleus in healthy plants, nucleocytoplasmic partitioning is an intrinsic effector-independent feature of these receptors. This partitioning is expected to engage the nuclear import and export machinery. Indirect evidence for this comes from genetic experiments in which a gain-of-function mutation in an Arabidopsis TIR-NB-LRR gene was used to study the requirements of an ‘autoimmunity’ phenotype. Mutant snc1 plants express the autoactive SNC1 protein carrying a single amino acid substitution between the NB LRR domains, leading to chronic activation of defence responses and disease resistance to bacterial and oomycete pathogens. Recessive mutations in two suppressor loci of the snc1 genotype, MOS3 and MOS6, each affect components required for protein passage through the nuclear pore. MOS3 is homologous to vertebrate nucleoporin 96 (Nup96) and resides at the nuclear rim. Vertebrate Nup96 and the yeast homologue, C-Nup145p, serve as components of the conserved Nup107-160 nuclear pore subcomplex, which is localized to both sides of the nuclear pore and regulates nuclear pore complex assembly and mRNA export. MOS6 encodes importin α3, a family of proteins known to function as adapters by binding to NLS containing cargo proteins and to importin β. The latter interacts with Nups to traverse nuclear pore complexes, thus implying mechanistically linked functions for MOS3 and MOS6 in nucleocytoplasmic trafficking. Functional specialization of importin family members is indicated by the fact that importin α3 represents one of eight importin homologues present in the Arabidopsis genome. The biological significance of snc1 suppressors comes from fully or partially restored susceptibility to virulent bacterial and oomycete pathogens in mos3 and mos6 plants, respectively. Finally, analysis of mos2 single mutants indicated that this suppressor of snc1‘autoimmunity’ is required for resistance specified by multiple R genes and MAMP-triggered basal resistance. MOS2 encodes a novel nuclear protein that contains one G-patch and two KOW (Kyprides, Ouzounis, Woese) domains and has homologues across the animal kingdom. The presence of both G-patch and KOW domains in the MOS2 protein suggests that it could function as an RNA binding protein critical for plant innate immunity.

Folding and stabilization of intracellular R proteins

Genetic and biochemical experiments revealed evolutionary conserved proteins in vertebrates and plants that appear to serve critical functions in maintaining ‘optimal’ preactivation receptor levels, possibly by coupling receptor folding and degradation pathways. The co-chaperone-like proteins RAR1 and SGT1, as well as the cytosolic HSP90 chaperone, were identified originally in plants by mutational screens as essential components of a subset of R protein-triggered immune responses to diverse plant pathogens. Loss of disease resistance function in rar1 or sgt1 or hsp90 mutant plants results typically in a severe depletion of R protein levels in the absence of pathogen. Specific mutations in HSP90, inactivating its intrinsic ATPase activity, and direct physical associations between the chaperone and R proteins suggest that the latter are HSP90 ‘clients’ and become stabilized by the chaperone. As SGT1 and (metazoan) RAR1 share structural similarities with co-chaperones and bind to each other as well as to HSP90, the former two are thought to act as co-chaperones, possibly by positively modulating HSP90 activity on its R protein clients. The SGT1–HSP90 molecular chaperone machine was resolved recently by nuclear magnetic resonance (NMR) spectroscopy. A link to protein degradation comes from the finding that RAR1 and SGT1 each interact with subunits of the COP9 signalosome, a multi-protein complex of the ubiquitin–proteasome pathway, and from an association of SGT1 with SCF ubiquitin ligase components. The receptors themselves could serve as degradation targets. Because R protein levels are typically decreased in recessive rar1 and sgt1 single mutants, the corresponding wild-type genes could antagonize a default receptor degradation pathway such that receptor folding and degradation processes are coupled. Human homologues of SGT1 and cytosolic HSP90 form complexes with the CARD domain-containing NOD1 and NOD2 immune sensors as well as with several other NLRs, including NALP3. Functional dependence on SGT1 and HSP90 of receptors carrying unrelated N-terminal domains (TIR, CC, CARD, PYR) as well as direct binding of SGT1-HSP90 complexes to their LRRs indicates that folding of the signalling competent form occurs primarily via the polymorphic C-terminal receptor region [23].

Damage-associated molecular patterns (DAMPS) in plants?

It has been known since the 1980s that plants can recognize ‘endogenous’ elicitors such as pectic fragments derived from the plant cell wall (oligogalacturonides). A second example of an endogenous elicitor is AtPep1, a 23-aa peptide isolated from Arabidopsis leaves that signals the activation of components of the innate immune response. This peptide is derived from a 92-aa precursor encoded by a gene that is inducible by wounding, methyl jasmonate and ethylene. Six paralogues of PROPEP1 are present in Arabidopsis, and orthologues have been found in other plant species. The AtPep1 receptor has been identified using a ¹²⁵I-labelled AtPep1 analogue and was shown to be a membrane-resident LRR-RLK, designated PEPR1 [24]. Indeed, the available data suggest that PEPR1 amplifies innate immune responses during cell damage inflicted by invading microbes. PEPR1 receptors of intact neighbouring cells appear to recognize the endogenous peptide as a danger signal and trigger the initiation of a defence response. If so, then PEPR1 acts analogously to the MAMP-induced defence responses described above. Note that the proposed mode of action of membrane-resident PEPR1 is conceptually similar to the detection of modified self by a subset of intracellular R proteins (see above). Thus, it is possible that effector, MAMP and ‘endogenous’ elicitor-triggered immune responses represent variations of a common ‘danger’ theme. Clearly, the existence of endogenous elicitors in plants is conceptually analogous to the perception of danger signals in vertebrates [25]. Accordingly, plant cell wall-derived oligogalacturonides and AtPep1 can be considered as DAMPs. How plants integrate, regulate and partition immune responses triggered by DAMPs, MAMPs and effectors is largely unclear at present. Undoubtedly, a deeper understanding of their interplay will be critical for a comprehensive understanding of the plant immune system.

Plant–microbe communities

I have added below a modified version of an Editorial that I have written together with Jefferey L. Dangl and Ton Bisseling for publication in Science to promote future research activities on this topic.

Parasitic and symbiotic associations between plants and microbes are merely the two extreme outcomes of a continuum of inter-organismal interactions affecting plant productivity. Remarkably little is understood about plant–microbe interactions that are, at first glance, ‘symptomless’. Complex communities of poorly studied plant-associated microbes are an untapped reservoir that can promote plant health and productivity.

One gram of soil typically contains ∼10¹⁰ bacteria. Microbial DNA fingerprints from plant roots (rhizosphere) or aerial organs (phyllosphere) have uncovered specific microbial communities thriving on the surface of, or within, healthy plant tissue. For instance, rhizosphere microbiomes attached to, and within the first few millimetres away from, the root surface are distinct from those in bulk soil, suggesting specific colonization events. Moreover, rhizosphere microbiomes (of bacteria and fungi) typically differ between plant species. Adding to this complexity, there are also differences in microbial populations along the longitudinal axis of roots, and between major developmental stages of the plant life cycle. The organic carbon flux from roots promotes the growth of microbial decomposers that, in turn, recycle plant nutrients for root uptake by transpiration-driven water fluxes. Seedlings exude 30–40%, and adult plants 20%, of photosynthetically fixed carbon into the rhizosphere in the form of poorly characterized rhizodeposits. This extrusion of nutrients outside the plant raises fundamental questions.

• Do plants feed and structure microbial rhizosphere communities to their advantage, and if so, how?

• Is the taxonomic diversity of microbiota related to functional diversity of food webs?

• Can the notoriously low heritability of plant growth be accounted for by large environmental interactions with microbial assemblies in the rhizosphere which are, in turn, influenced by soil types?

Purified isolates of rhizosphere-derived bacteria or fungi can promote plant growth, and a subset of rhizobacteria can suppress the growth of other soil-borne pathogens. Thus, the rhizosphere microbiome is likely to tune both maximal plant growth-promoting and protective functions.

Profiling techniques used currently to assess microbial population structure discriminate genetic fingerprints only at the species level. Advanced DNA sequencing technologies applied to rhizosphere and phyllosphere samples could overcome this limitation to define interspecies community structures. Ironically, very few studies have investigated the microbial populations inhabiting either Arabidopsis thaliana or model legumes cultivated under long-term sustainable agricultural practices. These model plants, and selected major crops such as corn and rice, provide genetic and genomics platforms for dissecting the organization and functions of rhizosphere and phyllosphere communities, and for identifying the plant loci that contribute to their formation. Recently, eight tested Arabidopsis ecotypes were found to exert a selective influence on bacteria associated with their roots, as determined by terminal-restriction fragment length polymorphism (T-RFLP) and ribosomal intergenic spacer analysis [26]. These accession-dependent community profiles also differed from control bulk soil. If this finding can be validated by others, then the plant genotype structures rhizobacterial communities. The availability of recombinant inbred lines (RILs) in Arabidopsis should provide the basis to dissect the genetic basis of these accession-dependent community profiles.

Problems and open questions

1. How do plants discriminate between pathogenic and beneficial (symbiotic) microbes? Why do symbionts escape defence?

2. Do plants feed and structure microbial rhizosphere communities to their advantage, and if so, how?

3. Is the taxonomic diversity of microbiota related to functional diversity of food webs?

4. Can the low heritability of plant growth be accounted for by large environmental interactions with microbial assemblies in the rhizosphere which are, in turn, influenced by soil types?

5. What are the nodes and mechanisms enabling plants to integrate and partition signals from DAMPs, MAMPs and effectors?

6. Post-activation signalling events of R proteins (and PRRs) are not understood. How do R proteins act in the nucleus?

7. A global in vivo map of protein–DNA interactions associated with early MAMP- and R-triggered immunity is needed for a deeper understanding of transcriptional reprogramming.

8. How exactly do plants integrate immune responses with phytohormone signalling pathways (jasmonate and salicylic acid pathways)? The receptor(s) for salicylic acid (and related metabolites) remains elusive.

Disclosure

None.