Perspective

To Nibble at Plant Resistance Proteins

See allHide authors and affiliations

Science  08 May 2009:
Vol. 324, Issue 5928, pp. 744-746
DOI: 10.1126/science.1171666

Abstract

To intercept invading microbes that threaten growth and reproduction, plants evolved a sophisticated innate immune system. Recognition of specialized pathogens is mediated by resistance proteins that function as molecular switches. Pathogen perception by these multidomain proteins seems to trigger a series of conformational changes dependent on nucleotide exchange. The activated resistance protein switches on host defenses, often culminating in the death of infected cells. Given their control over life and death, activity of these proteins requires tight regulation that involves intramolecular interactions between the various domains.

Discrimination between self and non-self is a fundamental ability of immune systems. Vertebrates rely on both an innate and an adaptive immune system of which the last is based on immunological memory. In contrast, plants primarily rely on their innate immune system in which each individual plant cell can autonomously mount a defense response (1). Two layers can be distinguished in the plant immune system. One is based on extracellular trans-membrane receptors that recognize conserved microbe-associated molecules and induce a relatively weak immune response that, nevertheless, effectively halts colonization by most microbes. The second layer is effective against specialized pathogens that can successfully break through the first layer and is based on highly polymorphic resistance (R) proteins. R proteins act mainly (but not exclusively) intracellularly and confer protection against (hemi-) biotrophic pathogens that need living host tissues for their proliferation. During infection, these pathogens (which include many viruses, bacteria, fungi, oomycetes, and nematodes) produce virulence factors (effectors), of which several suppress the first layer of the plant’s immune system (1), clearing the way for infection. Some effectors, or the perturbations they cause in the plant, are perceived by R proteins, which consequently set off strong defense responses in the plant that often leads to suicide of the infected cells (1).

Most R proteins are multidomain NB-LRRs (“nibblers”), named after their central nucleotide-binding (NB) and leucine-rich repeat (LRR) domains. The NB domain is part of a larger domain, the so-called NB-ARC domain, which consists of three subdomains: NB, ARC1, and ARC2 (2). The N termini of these R proteins are structurally diverse; some have homology to the Toll and human interleukin 1 receptor (TIR) and are called TIR-NB-LRRs. Others are commonly referred to as CC-NB-LRRs, because most carry predicted coiled-coil (CC) regions (1, 2). Plant NB-LRRs, together with the metazoan cell death regulators Apaf-1 and CED-4, form the NB-ARC family within the class of STAND [signal transduction adenosine triphosphatases (ATPases) with numerous domains] proteins (3). The NACHT sister family within this class encompasses the animal NLR (NACHT-LRR/NOD-LRR) innate immune receptors, where the NB domain is also fused to an LRR domain (4). STAND proteins are proposed to function as molecular switches, regulating cellular responses through nucleotide-dependent conformational changes (2, 3). Here, we discuss and evaluate the R protein–switch model in the context of other STAND proteins.

Because R proteins have the potential to trigger host cell death, their activity needs to be tightly regulated. They should be strongly inhibited in the absence of a pathogen, but rapidly activated upon attack. How is this process controlled? It appears that inappropriate activation is prevented by autoinhibition, which seems to be mainly accomplished by intramolecular interactions between the various domains. Interaction and mutagenesis studies with various NB-ARC and NLR proteins, including R proteins, identified both the N-terminal part of the repeat domain and the ARC2 subdomain to be essential for this autoinhibition (5, 6) (Fig. 1). Disturbance of the interaction between these two subdomains, by mutations or domain swaps, diminishes autoinhibition and results in constitutive R protein activation (5, 7).

Fig. 1

Model for R-protein activation. In the absence of a pathogen, NB-LRR R proteins reside in an autoinhibited, ADP-bound “OFF” state that is stabilized by the LRR domain. Effector-perception by the C-terminal part of the LRR domain changes the interface between its N-terminal part and the ARC2 subdomain, thereby creating a more open conformation of the R protein that is prone to nucleotide exchange. ADP/ATP exchange triggers a second conformational change, altering the interactions between the central NB-ARC, the N-terminal TIR/CC and C-terminal LRR domains resulting in the “ON” state. In the activated state, the NB subdomain becomes exposed to initiate defense signaling. ATP hydrolysis resets the protein into its ADP-bound autoinhibited “OFF” state. The model is a refined version of that presented in (2).

The LRR domain is not only involved in negative regulation, but provides positive control as well. Expression of truncated R proteins that lack the LRR domain and carry autoactivating mutations in the NB-ARC domain generally does not induce full host defenses unless the corresponding LRR domain is coexpressed (7). Furthermore, various studies have shown that the C-terminal part of the LRR domain provides pathogen recognition specificity (2, 7). Hence, the LRR domain has a dual function; it provides autoinhibition and it translates pathogen recognition into activation. How exactly the LRR recognizes a pathogen is unclear. Whereas some R proteins bind effectors directly, others require an intermediary host factor. This factor often interacts with the N-terminal domain of the R protein and could represent either the virulence target (thereby acting as a guardee) or a target mimic (thereby acting as a decoy) (8, 9). In this situation, the LRR is likely involved in sensing the effector-induced perturbations of the target. Either way, effector recognition evokes R-protein activation, a process that, as with other STAND proteins, requires the R protein to bind nucleotides [adenosine diphosphate or adenosine triphosphate (ADP/ATP)] (2, 1015).

Biochemical studies on the tomato R protein I-2 revealed that it tightly binds ADP in vitro and that mutations reducing its ATP-hydrolysis rate result in constitutive defense activation. On the basis of these data, it was proposed that R proteins function as nucleotide-controlled molecular switches (10). In this model, the ADP-bound state represents the “OFF” state and the ATP-bound state the “ON” state of the protein (Fig. 1). Recognition of an effector triggers a conformational change that results in an “intermediate” open state, which enables ADP to be exchanged for ATP. Upon ATP-binding, the R protein adopts its active conformation (“ON” state) that subsequently unchains, in a still unknown way, host defenses. ATP hydrolysis eventually returns the protein to its autoinhibited “OFF” state.

Recently, this model gained support by the observation that related STAND proteins also tightly bind ADP in their autoinhibited state (1618). Furthermore, the hypothesis that effector-binding sets the stage for nucleotide-exchange was substantiated by data on two STAND proteins: the Escherichia coli transcriptional regulator MalT and the human NLR protein NALP1 (NLRP1). Binding of the inducer, maltotriose or muramyl dipeptide, respectively, to these STAND proteins is an absolute requirement for their nucleotide exchange, as exogenously applied ATP was not bound in absence of the inducer (12, 18). Moreover, addition of the inducer in the absence of ATP led to a conformational change that likely corresponds to the “intermediate” open state proposed for R proteins upon effector recognition (10, 12) (Fig. 1). Analogous to the R protein I-2 (10), MalT became constitutively active after mutation of its catalytic residue, which abolished ATP hydrolysis and locked the protein in its ATP-bound state (18) (fig. S1). Naturally occurring mutations of the corresponding putative catalytic residue in the NLR proteins NOD2 and PYPAF1 (NLRP3) also resulted in constitutive activity and, as a consequence, autoinflammation (19, 20) (fig. S1). Finally, recent biochemical analysis showed that the active, oligomeric conformation of Apaf-1 harbors either deoxyadenosine triphosphate (dATP) or ATP [for simplicity, both are referred to as (d)ATP]. However, it remains to be resolved whether the monomeric autoinhibited form of Apaf-1 is bound to (d)ADP or to (d)ATP, and whether nucleotide-exchange has to be preceded by hydrolysis of prebound (d)ATP (16, 21). Taken together, these data link the “ON” and “OFF” state of STAND proteins to an ATP- and ADP-bound state, respectively, and support the hypothesis that effector recognition induces nucleotide exchange in R proteins (Fig. 1).

To date, it remains unclear how activated R proteins trigger defense signaling. STAND proteins such as Apaf-1 and CED-4 oligomerize and form ring-like structures that provide an activation platform for signaling components (11). For plant R proteins, effector-induced oligomerization has been described for only one R protein so far (22). Although some R proteins depend on other NB-LRR proteins for their function (23, 24), it remains an open question whether (hetero-) oligomerization is a common feature for R proteins. For a number of NB-ARC and NLR proteins, the N-terminal domain has been shown to interact with downstream signaling partners (25). However, no such partners have been identified to interact with the N-terminal domains of R proteins, and evidence for a signaling function of this domain remains slim (9). It now seems that the N terminus, together with the LRR domain, is actually involved in recognition rather than signaling, as suggested by its interactions with putative effector targets or target mimics (8, 9, 26).

If both the N- and C-terminal domains of R proteins are indeed involved in recognition rather than signaling, that leaves the NB domain, perhaps surprisingly, as a candidate to serve as an interaction platform for downstream signaling components. The observation that expression of only the NB subdomain of the potato R protein Rx triggers constitutive defenses in the absence of the pathogen lends support to this idea (26). A mechanism in which the NB domain itself is responsible for downstream signaling would be a unique property of R proteins, not shared with other members of the NB-ARC and NLR families. Direct signaling by the NB domain instead would resemble the molecular switch function of regulatory guanosine triphosphatases (GTPases) of the Ras superfamily (27). In these GTPases, different nucleotide-dependent conformations of two effector loops within the core nucleotide binding fold regulate downstream events. Nucleotide binding and hydrolysis—and, thereby, the activation state of these GTPases—are regulated by accessory proteins. It is tempting to speculate that in R proteins these functions are embedded in its multidomain structure.

In recent years, nibbling at NB-LRR protein function has shed some light on the molecular mechanistic basis of their activation and the role of nucleotide binding. To gain a deeper understanding of how these proteins regulate plant defenses, we now need to investigate their biochemical properties in more detail and to further analyze the dynamics and subcellular localization of NB-LRR signaling complexes in vivo. Identification of partners interacting with the NB domain should eventually reveal how R proteins activate host defenses. Another major challenge will be the elucidation of the three-dimensional structure of R proteins, preferably in the different conformational states, as this is the key for a full understanding of the molecular mechanisms underlying the choice between life and death.

Supporting Online Material

References and Notes

  1. We thank M. Rep, C. Testerink, and M. Joosten for critical review. We apologize to the many colleagues whose work could not be discussed due to size restriction. W.I.L.T. is supported by a VENI grant from the Netherlands Organization for Scientific Research (NWO). Research in the Takken lab is supported by the Centre for BioSystems Genomics (Netherlands Genomics Initiative/NWO) and by the European Union Integrated Project BIOEXPLOIT CT-2005-513959.
View Abstract

Navigate This Article