Food security has become an issue of global impor-
tance, and major price spikes for staples such as rice
and wheat have occurred in recent years. These price
spikes are partly due to the impact of plant diseases,
such as the spread of a new strain of the wheat stem
rust pathogen from East Africa into the Middle East
1
.
This has sparked an increased focus on improving
approaches to crop protection. The most effective and
environmentally sensitive approach to disease preven-
tion involves breeding crop plants for resistance. Indeed,
plant breeders have been using ‘resistance’ genes to con-
trol diseases in crop plants for almost 100 years, and the
effectiveness of this strategy sparked early genetic stud-
ies that defined ‘gene-for-gene’ relationships between
host resistance genes and pathogen virulence factors
2
.
However, only through recent molecular studies has
it become apparent that host resistance genes encode
components of the plant immune system that confer
the capacity to recognize and respond to specific path-
ogens. Plant immunity depends on cell-autonomous
events; these events are related to the innate immune
system in animals
3
but plants have a much bigger rec-
ognition repertoire to compensate for their lack of an
adaptive immune system. Ongoing research is revealing
the recognition capacity of the plant immune system,
and concurrent studies on pathogen biology are begin-
ning to unravel how these organisms manipulate host
immunity to cause disease. The recent convergence of
these two fields has dramatically changed our percep-
tion of plant–pathogen interactions and is providing
new approaches for crop protection.
Microbial plant pathogens almost always occupy
extracellular niches. Despite this, the nutrients that
enable pathogen growth are derived from host cells, and
the host cytoplasm and organelles are important sites of
molecular interaction. Plants have evolved two strategies
to detect pathogens
4,5
(FIG. 1). On the external face of the
host cell, conserved microbial elicitors called pathogen-
associated molecular patterns (PAMPs) are recognized
by receptor proteins called pattern recognition receptors
(PRRs)
6
. PAMPs are typically essential components of
whole classes of pathogens, such as bacterial flagellin or
fungal chitin. Plants also respond to endogenous mol-
ecules released by pathogen invasion, such as cell wall
or cuticular fragments called danger-associated molec-
ular patterns (DAMPs). Stimulation of PRRs leads to
PAMP-triggered immunity (PTI). The second class of per-
ception involves recognition by intracellular receptors of
pathogen virulence molecules called effectors; this recog-
nition induces effector-triggered immunity (ETI). This mode
of recognition leads to co-evolutionary dynamics between
the plant and pathogen that are quite different from PTI
as, in stark contrast to PAMPs, effectors are characteris-
tically variable and dispensable. Extreme diversification
of ETI receptors and pathogen effectors both within and
between species is the norm, whereas some PRR func-
tions are conserved widely across families. Generally, PTI
and ETI give rise to similar responses, although ETI is
qualitatively stronger and faster and often involves a form
of localized cell death called the hypersensitive response
(HR). PTI is generally effective against non-adapted
pathogens in a phenomenon called non-host resistance,
*Commonwealth Scientific
and Industrial Research
Organisation (CSIRO),
Division of Plant Industry,
GPO BOX 1600, Canberra,
Australian Capital Territory
2601, Australia.
‡
Research School of Biology,
Australian National
University, RN Robertson
Building, Biology Place,
Acton, Australian Capital
Territory 0200, Australia.
e-mails: peter.dodds@csiro.au;
john.rathjen@anu.edu.au
doi:10.1038/nrg2812
Published online 29 June 2010
Elicitors
Molecules that induce (‘elicit’)
an immune defence response.
In the context of this Review,
this term is used to refer to
both pathogen-associated
molecular patterns (PAMPs)
and effectors.
Pathogen-associated
molecular patterns
Any of a number of
conserved, usually structural,
molecules common to
pathogen organisms.
Plant immunity: towards an integrated
view of plant–pathogen interactions
Peter N. Dodds* and John P. Rathjen
‡
Abstract | Plants are engaged in a continuous co-evolutionary struggle for dominance
with their pathogens. The outcomes of these interactions are of particular importance
to human activities, as they can have dramatic effects on agricultural systems.
The recent convergence of molecular studies of plant immunity and pathogen
infection strategies is revealing an integrated picture of the plant–pathogen interaction
from the perspective of both organisms. Plants have an amazing capacity to recognize
pathogens through strategies involving both conserved and variable pathogen elicitors,
and pathogens manipulate the defence response through secretion of virulence
effector molecules. These insights suggest novel biotechnological approaches to
crop protection.
REVIEWS
NATURE REVIEWS
|
GENETICS ADVANCE ONLINE PUBLICATION
|
1
Nature Reviews Genetics
|
AOP, published online 29 June 2010; doi:10.1038/nrg2812
© 20 Macmillan Publishers Limited. All rights reserved10
Nature Reviews | Genetics
PAMPs
NB-LRR
Effector
Effector
Plant cell
PTI response
ETI response
Fungus/
oomycete
Extracellular space
Haustorium
Pilus
PRR BAK1
Bacterium
Pattern recognition
receptors
Plasma membrane-localized
receptors that recognize
the presence of pathogen-
associated molecular
patterns (PAMPs) in the
extracellular environment.
PAMP-triggered immunity
The plant defence response
elicited by pathogen-
associated molecular
pattern (PAMP) recognition.
Effectors
Proteins secreted by
pathogens into host cells
to enhance infection.
Many of these function to
suppress PAMP-triggered
immunity responses.
whereas ETI is active against adapted pathogens. However
these relationships are not exclusive and depend
on the elicitor molecules present in each infection.
Here, we provide an overview of the plant PTI and
ETI systems, highlighting recent advances and identify-
ing key gaps in our understanding of these processes. We
consider the roles of PRRs in initial pathogen perception,
our expanding knowledge of pathogen effectors and their
roles in suppressing PTI responses, the nature of effector
recognition and the downstream responses to pathogen
perception. Finally, we discuss briefly how this knowledge
is beginning to feed back into the agricultural context
that originally spawned the study of plant immunity.
Extracellular recognition by PRRs
PRRs have been reviewed recently
7
, so here we discuss
some important principles and recent findings relat-
ing to key proteins in the process of recognition of
extracellular pathogen molecules.
Pattern recognition receptors. Known PRRs fall into
one of two receptor classes: transmembrane receptor
kinases and transmembrane receptor-like proteins,
the latter of which lack any apparent internal signal-
ling domain
7
. Recent work has shown that endoplasmic
reticulum quality-control mechanisms are crucial for
PRR biogenesis (BOX 1). The receptor kinase gene family
has undergone huge expansion in plants: for exam-
ple, about 610 members are present in the Arabidopsis
thaliana genome, and many of these are responsive
to biotic stresses
8
. The receptor-like protein class has
57 members in A. thaliana
9
. The expansion of these
families is in contrast to the situation in animals, which
possess 12 Toll-like receptors that fulfil an equivalent
role to PRRs in plants
10
.
The PAMPs recognized by plants are multifarious
and include proteins, carbohydrates, lipids and small
molecules, such as ATP
6
. Recognition of PAMPs is
best understood in the case of the A. thaliana recep-
tor kinase FLAGELLIN SENSING 2 (FLS2), which
binds bacterial flagellin directly and then assembles
an active signalling complex. Although the PAMP
concept encompasses the idea that all PAMPs should
be recognized by all species, this has been found to
not always be the case, as perception of the bacterial
elongation factor EF-Tu is apparently restricted to the
Brassicaceae
11
. Similarly, the Xa21 receptor in rice pro-
vides race-specific resistance to the bacterial pathogen
Xanthomonas oryzae, and was recently shown to act as
a PRR for a novel sulphonated bacterial protein termed
Ax21 (REF. 12).
BAK1, a central regulator of PAMP-triggered immu-
nity. Most known PRRs require the leucine-rich
repeat (LRR) receptor kinase BRASSINOSTEROID
INSENSITIVE 1-ASSOCIATED KINASE 1 (BAK1) for
function
13,14
(FIG. 2). An exception is the fungal chitin
receptor CHITIN ELICITOR RECEPTOR KINASE 1
(CERK1)
15,16
, which also responds to an unknown bac-
terial PAMP
17
. BAK1 is part of a family of five somatic
embryogenesis receptor kinase (SERK) members and
is also known as SERK3. It is not yet known whether
other SERK family members have redundant roles in
immune signalling. BAK1 does not have a direct role
in elicitor perception, but FLS2 rapidly forms a complex
with BAK1 after elicitation. This interaction results in
phosphorylation of both proteins, which peaks 30–60
seconds after elicitor treatment
18
. BAK1 also has a
role in the perception of other elicitors, probably
also through heterodimerization with PRRs in the
LRR-receptor kinase family.
As such, BAK1 is a central regulator of plant immu-
nity and consequently the target of several pathogen
virulence effector molecules
19
(see below). Despite
this, A. thaliana plants containing a null muta-
tion in the bak1 gene are actually marginally more
resistant to biotrophic pathogens, although they are
slightly more susceptible to necrotrophic pathogens
20
.
These phenotypes may be related to a deregulated
cell death phenotype that has been described in the
bak1 mutants
20,21
.
Figure 1 | The principles of plant immunity. Bacterial plant pathogens propagate
exclusively in the extracellular spaces of plant issues. Most fungal and oomycete
pathogens also extend their hyphae into this space, although many also form
specialized feeding structures, known as haustoria, that penetrate host cell walls but
not the plasma membrane. Other fungi extend invasive hyphae into plant cells, but
again do not breach the host membrane. Molecules released from the pathogens into
the extracellular spaces, such as lipopolysaccharides, flagellin and chitin (pathogen-
associated molecular patterns (PAMPs)) are recognized by cell surface pattern
recognition receptors (PRRs) and elicit PAMP-triggered immunity (PTI). PRRs
generally consist of an extracellular leucine-rich repeat (LRR) domain (mid-blue),
and an intracellular kinase domain (red). Many PRRs interact with the related protein
BRASSINOSTEROID INSENSITIVE 1-ASSOCIATED KINASE 1 (BAK1) to initiate the PTI
signalling pathway. Bacterial pathogens deliver effector proteins into the host cell
by a type-III secretion pilus, whereas fungi and oomycetes deliver effectors from
haustoria or other intracellular structures by an unknown mechanism. These
intracellular effectors often act to suppress PTI. However, many are recognized
by intracellular nucleotide-binding (NB)-LRR receptors, which induces effector-
triggered immunity (ETI). NB-LRR proteins consist of a carboxyl-terminal LRR domain
(light blue), a central NB domain (orange crescent) that binds ATP or ADP (yellow oval),
and an amino-terminal Toll, interleukin-1 receptor, resistance protein (TIR) or
coiled-coil (CC) domain (purple oval).
REVIEWS
2
|
ADVANCE ONLINE PUBLICATION www.nature.com/reviews/genetics
© 20 Macmillan Publishers Limited. All rights reserved10
Effector-triggered immunity
The plant defence response
elicited by effector recognition.
Biotrophic
Biotrophic pathogens
propagate in living plant tissue
and generally do not cause
necrosis as a result of infection.
They use various means, such
as haustoria production, to
extract nutrients from host cells.
Necrotrophic
Necrotrophic pathogens
actively induce necrosis in
infected tissues, often through
the production of toxins, and
obtain nutrients from the
dead host tissue.
Type-III secretion system
A syringe-like structure
produced by many plant
and animal pathogen bacteria
that allows direct secretion
of effector proteins from
the bacterial cytoplasm into
host cells.
One potential regulator of the FLS2–BAK1 com-
plex is the cytoplasmic protein kinase BOTRYTIS-
INDUCED KINASE 1 (BIK1). BIK1 was identified as
a potential regulator because bik1 is upregulated after
pathogen or elicitor treatment of A. thaliana leaves
22
.
BIK1 interacts with both FLS2 and BAK1 before elicita-
tion and seems to dissociate from the complex after elic-
itation. In vitro, BAK1 phosphorylates BIK1 and BIK1
phosphorylates both FLS2 and BAK1. In vivo,
BIK1 becomes phosphorylated 5–10 min after treat-
ment with flagellin
23
; this phosphorylation peaks
after the FLS2–BAK1 phosphorylation. Confusingly,
bik1 mutant A. thaliana plants are more resistant
to Pseudomonas syringae infection than wild-type
A. thaliana plants
22
as a result of them overproduc-
ing the defence hormone salicylic acid (SA), but they
are also more susceptible to infection with the necro-
trophic fungal pathogen Botrytis cinerea. Despite this,
deficiencies in FLS2-mediated immune responses
could be measured in these plants
23
. These contrast-
ing results make it difficult to ascribe a clear function
to BIK1 in plant immunity, and further studies will
be required.
Virulence activities of pathogen effectors
Successful pathogens are able to suppress PTI responses
and thereby multiply and cause disease. They achieve sup-
pression through the deployment of ‘effector’ proteins.
Studies of bacterial phytopathogens have provided
most of our understanding of effector strategies
and mechanisms. Individual phytopathogen strains
encode 20–30 effectors, which are highly regulated and
secreted directly into the host cytoplasm by a dedicated
needle structure, the type-III secretion system (TTSS)
24
.
The repertoire of individual effectors varies dra-
matically among closely related bacterial strains, and
effectors themselves act redundantly and are appar-
ently interchangeable
25
; examples of such effectors
are discussed below. Many effectors interfere directly
with PTI responses
26
, and bacterial mutants that lack
the TTSS system are non-pathogenic. Interestingly,
a number of examples show that transgenic overex-
pression of an individual type-III effector in the host
plant restores the ability of such bacterial mutants to
grow
27,28
, suggesting that bacterial pathogenicity only
requires suppression of PTI. However, contributions
of as yet undefined mechanisms to other processes,
such as nutrient acquisition, cannot be excluded.
Bacterial effector functions. Bacterial effectors have
molecular or enzymatic activities that specify both
their ability to modify host targets and their intracel-
lular recognition by ETI receptors
29
(see below). The
redundancy among effectors is illustrated by the unre-
lated P. syringae effectors AvrPto and AvrPtoB, which
both target the FLS2–BAK1 complex. Although the
models for how suppression works conflict in molec-
ular detail
19,30,31
, it is generally accepted that AvrPtoB
uses a dual strategy for kinase suppression: its amino-
terminal kinase-targeting domain is sufficient to sup-
press flagellin responses, and its carboxy-terminal
E3 ligase domain can tag interacting kinase proteins
with ubiquitin to direct them for degradation
32,33
.
AvrPtoB is known to target five host kinases of the
Pto/interleukin receptor-associated kinase (IRAK)
class
32
, but because this clade is hugely expanded in
plants
8
, there are probably many more such targets.
Likewise, AvrPto suppresses multiple PRR receptor
kinases, perhaps by acting as a kinase inhibitor
19,30,34
.
Overall, these effectors seem to be part of a bacterial
strategy that targets host kinases nonspecifically.
A further example of overlapping effector func-
tions involves the host protein RPM1-INTERACTING
PROTEIN 4 (RIN4), which is targeted by the P. syringae
effectors AvrB, AvrRPM1 and AvrRpt2 through dif-
ferent molecular strategies
35,36
. Recently, it was shown
that the P. syringae effector HopF2 may also target
RIN4 (REF. 37). Overexpression of HopF2 prevented
degradation of RIN4 by the protease AvrRpt2 but
did not alter the interactions of RIN4 with AvrRPM1
or AvrB. Bacteria that lack HopF2 have increased
growth on lines that lack RIN4, suggesting that RIN4
could indeed be a target for virulence, but an indirect
cause for this observation was not ruled out. RIN4
is a negative regulator of both PTI and ETI
28,38
, and
also interacts with the plasma membrane H
+
-ATPases
AHA1 and AHA2 to enhance stomatal opening
39
, a
key event during bacterial pathogenicity on leaves.
Thus it is not clear how targeting of RIN4 by multiple
effectors would enhance bacterial virulence, as disrup-
tion of RIN4 should actually restrict pathogenicity.
However, the number of effectors involved in this
process is consistent with RIN4 being an important
virulence target.
Box 1 | Pattern recognition receptor biogenesis
Most eukaryotic membrane proteins undergo quality control during folding and
maturation in the endoplasmic reticulum (ER), a process termed ER‑QC
114
.
A number of recent studies show that the biogenesis of a pattern recognition
receptor (PRR), the EF‑Tu receptor (EFR), is regulated by this mechanism
115–119
.
After secretion into the ER, proteins are modified at glycosylable Asn residues by
an oligosaccharyltransferase complex, which covalently attaches a complex
polysaccharide containing three terminal glucose residues. The glucose moieties
are subsequently trimmed by glucosidases I and II. A single glucose residue is
added back by UDP‑glucose:glycoprotein glucosyltransferase (UGGT) near regions
of protein disorder. Monoglucosylated proteins interact with the lectins calnexin
(CNX) or calreticulin (CRT) to retain misfolded substrates in the ER. In this way,
UGGT acts as a folding sensor, and glycosylation is intimately related to protein
maturation. Terminally misfolded proteins are degraded.
Another ER folding pathway is based on the chaperone BiP (a form of heat shock
protein 70 (Hsp70)). Unfolded proteins undergo cycles of BiP binding and release,
which is regulated by Hsp40 co‑chaperones containing J domains (for example, the
ERdj protein). Forward genetic screens showed that Arabidopsis thaliana genes
encoding glucosidase II, UGGT, CRT3, ERdj3B and ERD2b are required for EFR
function and accumulation. In addition, STT3A, a subunit of the oligosaccharyl‑
transferase complex, was necessary for EFR biogenesis. Finally, STROMAL‑
DERIVED FACTOR 2 (SDF2) resides in a protein complex with ERdj3B and BiP, and
was also required for EFR maturation. Plants with mutations in these genes are
generally more susceptible to pathogens, indicating that EFR is not the only
immune protein that is governed by ER‑QC. However, neither FLAGELLIN SENSING 2
(FLS2) nor CHITIN ELICITOR RECEPTOR KINASE 1 (CERK1) function is significantly
affected in these mutants.
REVIEWS
NATURE REVIEWS
|
GENETICS ADVANCE ONLINE PUBLICATION
|
3
© 20 Macmillan Publishers Limited. All rights reserved10
Nature Reviews | Genetics
PRR BAK1 PRRBAK1
Bacterium Apoplast
PRRBAK1
PP
CDPKs
BIK1
a
bc
Plant cell MAPKs
Haustoria
(sing. haustorium.) Specialized
structures produced by
some fungal and oomycete
pathogens. Haustoria extend
through the plant cell wall
and expand in the host cell.
They remain surrounded by
a host-derived membrane
and hence are topologically
extracellular and separated
from the host cytoplasm.
Hemibiotrophic
Hemibiotrophic pathogens
incorporate aspects of both
biotrophic and necrotrophic
infection strategies. Often
this involves an initial
biotrophic infection phase
during which the pathogen
spreads in host tissue,
followed by a necrotrophic
phase during which host
cell death is induced.
NB-LRR proteins
A class of intracellular
receptor proteins containing
nucleotide-binding (NB) and
leucine-rich repeat (LRR)
domains that recognize specific
pathogen effectors.
It is important to note that not all effectors target PTI.
One example of an alternative bacterial effector strategy
is given by the transcription activator-like (TAL) effec-
tors of Xanthomonas spp., which are transcription fac-
tors that induce the expression of specific host genes,
some of which contribute to symptom development
40
.
Unlike AvrPto and AvrPtoB in Pseudomonas spp., TAL
effectors do not seem to act redundantly because several
of them are essential for virulence. They interact specifi-
cally with a site in the target gene promoters through a
central tandem repeat region that forms a DNA-binding
domain
41–43
. Strikingly, two hypervariable amino acid
residues in each repeat specify interaction with a charac-
teristic nucleotide in the effector recognition site. Thus,
the nucleotide sequence of the target DNA can be pre-
dicted by the amino acid sequence of the tandem repeat
domain. Biotechnologically this is significant because it
enables precise modification of gene expression in vivo,
including turning this system against Xanthomonas spp.
by engineering AvrBs3-responsive elements (known as
UPA sites), upstream of active resistance genes
44
. In
nature, this strategy has been pre-empted in some plant
species: target sites for certain TAL effectors have been
incorporated upstream of the resistance genes Bs3 and
Xa27 in pepper and rice, respectively
45,46
.
Eukaryotic effectors. Data on eukaryotic effectors and
their functions are sparse in comparison with data on
bacterial effectors. Both fungal and oomycete patho-
gens produce effectors that are secreted through the
endomembrane system and are subsequently delivered
into host cells by unknown mechanisms
47,48
. Oomycete
effectors characteristically contain the internal motif
Arg-X-Leu-Arg (RXLR, in which X represents any
amino acid), which is required for delivery into plant
cells. Genome sequencing of Phytophthora infestans
49
,
the Irish potato famine pathogen, revealed 563 RXLR
effector genes. Seventy of these genes are under diver-
sifying selection and only 16 share orthologues in the
genomes of 2 other sequenced Phytophthora spp., which
indicates that very strong selection processes act on
these effectors. A further 196 effectors of a separate class
(known as Crinkler proteins) are encoded by P. infestans.
Such generalized identification of fungal effector genes
has been restricted by the lack of conserved motifs to
aid genome interrogation, but genome analysis of sev-
eral fungal pathogens predicts that they have complex
and diversified secretomes
50,51
. The massive expansion
in eukaryotic effector repertoires relative to bacterial
effector repertoires may suggest a requirement for more
diverse effector functions by eukaryotic pathogens,
possibly to support their more specialized nutrient
acquisition strategies.
Some data support roles of P. infestans effectors in
suppression of immunity
52
; for example, Avr3a sup-
presses elicitor-induced cell death through interac-
tion with the host CMPG1 E3 ligase
53
, but in general
very little is known about effector functions in fungi
or oomycetes. However, many other potential roles
remain, such as establishment of the pathogenic niche
through development of the haustoria feeding struc-
tures and manipulation of host cell death during the
hemibiotrophic lifestyle.
Sedentary nematode pathogens of plants form pro-
longed associations with roots, in which they induce the
formation of novel host structures, such as multinucle-
ate giant cells, from which they feed using a specialized
proboscis called a stylet. The stylet also delivers salivary
secretions into host cells; proteomic analysis of saliva
from one such species, Meloidogyne incognita, identi-
fied 486 potential effector proteins
54
. Ongoing genom-
ics analyses of such species will identify many more and
help in elucidating the pathogenic strategies of these fas-
cinating organisms. In addition, viral pathogens encode
specific suppressors of the small RNA pathway to pre-
vent degradation of their genomes and/or abrogation of
viral gene expression
55
.
Overall, our understanding of effector proteins and
their host targets is at a very early stage. Sophisticated
biochemical screens for host protein targets that interact
with the diverse suites of pathogen effectors are likely to
lead to the identification of important components of
host defence mechanisms and teach us more about host
immune pathways and pathogenicity strategies.
Intracellular effector recognition
ETI is the second pathogen-sensing mechanism in
plants and is based on intracellular recognition of effec-
tor proteins
4,5
. Recognition events are mostly mediated
by a class of receptor proteins that contain nucleotide-
binding (NB) domains and LRRs (FIG. 1). Plant NB-LRR
proteins confer resistance to diverse pathogens, includ-
ing fungi, oomycetes, bacteria, viruses and insects. NB
and LRR domains are also present in NOD-like immune
Figure 2 | Formation of active pattern recognition receptor complexes.
a | Infectious pathogens, such as bacteria, shed pathogen-associated molecular
patterns (PAMPs; pink, yellow and purple shapes) into the apoplast, where they are
recognized by specific pattern recognition receptors (PRRs). b | Immediately after
ligand binding, the PRR forms an active complex with BRASSINOSTEROID
INSENSITIVE 1-ASSOCIATED KINASE 1 (BAK1). c | This results in transphosphorylation
(indicated by P) of the respective kinase domains of the PRR and BAK1. Signalling via
this active complex can be mediated directly by BOTRYTIS-INDUCED KINASE 1
(BIK1), or by mitogen-activated protein kinases (MAPKs) or calcium-dependent
protein kinases (CDPKs). This is a generalized model that is based on FLAGELLIN
SENSING 2 (FLS2), the receptor for bacterial flagellin.
REVIEWS
4
|
ADVANCE ONLINE PUBLICATION www.nature.com/reviews/genetics
© 20 Macmillan Publishers Limited. All rights reserved10
Nature Reviews | Genetics
a Direct b Guard/decoy c Bait
receptors (NLRs), which are involved in PAMP induction
of innate immunity responses in animals
3,56
, and in the
animal apoptotic factors apoptotic protease-activating
factor 1 (APAF1) and cell death protein 4 (CED4). Many
plant NB-LRR proteins also contain an N-terminal TIR
(Toll, interleukin-1 receptor, resistance protein) domain
related to the intracellular signalling domain of animal
Toll-like receptors
10
. A second common class of NB-LRR
proteins contain an N-terminal domain with a coiled-
coil (CC) domain, whereas others have no conserved
N-terminal region.
Direct and indirect recognition. NB-LRR proteins can
recognize pathogen effectors either directly by physical
association or indirectly through an accessory protein
that is part of an NB-LRR protein complex (FIG. 3). In
general, direct recognition has been demonstrated by
yeast two-hybrid (Y2H) assays, in some cases supported
by in vitro protein interaction assays. For example, the
rice CC-NB-LRR Pi-ta protein binds to the Magnaporthe
grisea effector AvrPita both in vitro and in Y2H assays
57
.
The flax TIR-NB-LRR L and M proteins also interact
in Y2H assays with the Melampsora lini fungal effec-
tors AvrL567 and AvrM, respectively
58–61
. These pairs of
receptor and effector proteins show evidence of strong
diversifying selection and are characterized by high
levels of sequence polymorphism between alleles in
the host and pathogen populations, respectively, with
these variants showing different recognition specifici-
ties. This is likely to be the result of antagonistic co-
evolution between the interacting components in the
host and pathogen.
Indirect effector recognition has been observed in
a number of cases. In the best-described models, the
effector interaction is mediated by an accessory protein
that is a pathogen virulence target or a structural mimic
of one. The effector induces a change in the accessory
protein that enables the accessory to be recognized by
the NB-LRR protein
62
. This strategy neatly sidesteps the
evolutionary advantage of the faster evolving pathogen,
as the host takes advantage of the pathogen’s virulence
strategy to drive the recognition. Three conceptual
models have been proposed to describe these mecha-
nisms (FIG. 3). The ‘guard’ model postulates that NB-LRR
proteins guard an accessory protein (or guardee) that
is targeted and modified by pathogen effectors
63
. This
model is exemplified by the A. thaliana RIN4 protein.
RIN4 forms exclusive complexes with the NB-LRR pro-
teins RPM1 and RESISTANCE TO PSEUDOMONAS
SYRINGAE 2 (RPS2)
36,64
. Degradation of RIN4 by the
protease effector AvrRpt2 de-represses RPS2, whereas
AvrB or AvrRPM1-mediated phosphorylation of
RIN4 activates RPM1 (REFS 35,36). Thus, modification
of RIN4 by the effectors explains how an individual
NB-LRR (in this case, RPM1) can recognize more than
one effector.
However, the guard model postulates that RIN4 is a
virulence target of the effectors, which is as yet unproven
(see also above). Also, this model creates an evolutionary
problem: RIN4 should evolve to avoid binding to the
effector proteins in the absence of RPS2 and RPM1, but
in their presence, selection will favour effector binding to
promote recognition
5
. To solve this problem, the ‘decoy’
model was proposed
62
, in which duplication of the
effector
target gene or independent evolution of a target
mimic could
relax evolutionary constraints and allow the
accessory protein to participate solely in effector percep-
tion. This situation is exemplified by the tomato NB-LRR
protein Prf, which forms a complex with the accessory
protein Pto kinase
65
. Pto kinase is closely related to the
kinase domains of FLS2 and CERK1, which are targets of
AvrPto and AvrPtoB
32,66
. Thus, Pto provides the recogni-
tion capability for Prf, and this drives diversification of
the Pto family to broaden the spectrum of recognition
capability
67
. In the decoy model, the accessory protein
specializes in perception of the effector by the NB-LRR
protein
but has no other function. This fails to explain
the requirement for Pto kinase activity in Prf activation
68
and the clear role of RIN4 in defence responses. A fur-
ther modification of the decoy concept is the bait-and-
switch model
69
, which envisages a two-step recognition
event. First, an effector interacts with the accessory
‘bait’ protein associated with an NB-LRR, and then a
subsequent recognition event occurs between the effec-
tor and NB-LRR protein to trigger signalling. That
is, the NB-LRR protein interacts with an effector target
(the bait) to facilitate direct recognition of the pathogen
effector, rather than recognizing the modified target as
envisaged in the guard model.
It is important to remember that these models are
generalizations based on limited specific examples, none
of which is yet fully understood. Thus, although they are
useful conceptual tools, they are unlikely to adequately
describe all effector recognition events and can be
restrictive. For instance, in addition to providing effec-
tor recognition, Pto seems to participate actively with
Prf in a highly evolved co-regulatory relationship
65,68
.
The massive diversity in effector and receptor biol-
ogy suggests that many variations on these themes,
and probably other novel recognition events, are likely
to occur. For example, the Pto kinase phosphorylates
the effector AvrPtoB, leading to inactivation of its
intrinsic E3 ligase activity
70
; this is an intriguing and
Figure 3 | Models of direct and indirect recognition. Plant nucleotide-binding
(NB)-leucine-rich repeat (LRR) receptors can recognize pathogen effectors by
either direct or indirect mechanisms. a | In direct recognition, the effector (green)
triggers immune signalling by physically binding to the receptor (purple, orange,
yellow and blue; see FIG. 1 for a description of the receptor). b | In the guard and
decoy models, the effector modifies an accessory protein (red), which may be its
virulence target (guard model) or a structural mimic of such a target (decoy model).
The modified accessory protein is recognized by the NB-LRR receptor. c | Under the
bait model, interaction of an effector with an accessory protein facilitates direct
recognition by the NB-LRR receptor.
REVIEWS
NATURE REVIEWS
|
GENETICS ADVANCE ONLINE PUBLICATION
|
5
© 20 Macmillan Publishers Limited. All rights reserved10
