The interaction of avirulence and resistance gene products in flax rust disease – providing advances in rust research^†

^†Contribution to the symposium “Signalling in Plant-Pathogen Interactions” held during the Canadian Phytopathological Society Annual Meeting, 22–25 June 2009, Winnipeg, MB.

Abstract

Pathogenic rust fungi constitute a major disease threat to agriculture, but their obligate parasitic lifestyle makes them difficult to study. Research on the model flax rust disease system has contributed greatly to our knowledge of rust infection and, in particular, the mechanisms of rust resistance and susceptibility controlled by resistance (R) genes in the host and avirulence (Avr) genes in the pathogen. Twenty flax R genes have been isolated and encode cytoplasmic proteins with nucleotide-binding domains and leucine-rich repeat regions. These proteins act as a surveillance system for the recognition of specific pathogen Avr proteins as signals of invasion and subsequently activate plant defences. Several of these rust Avr proteins have also been isolated and were found to be small secreted proteins that are expressed in haustoria (specialized structures that penetrate the host cell wall) and delivered into the host cytoplasm during infection. The identification of flax R genes and the corresponding fungal Avr genes has allowed more detailed analysis of the recognition events that trigger resistance, revealing a direct interaction between R and Avr proteins as the basis of resistance specificity in flax rust disease.

Résumé: L'infection par le champignon pathogène de la rouille constitue un risque majeur pour l'agriculture, mais son mode de vie parasitaire obligatoire le rend difficile à étudier. La recherche effectuée sur le système modèle de la rouille du lin a grandement contribué à notre compréhension de l'infection par la rouille et, particulièrement, des mécanismes de résistance et de réceptivité contrôlés par les gènes de résistance (R) chez l'hôte et les gènes d'avirulence (Avr) chez l'agent pathogène. Vingt gènes R du lin qui encodent des protéines cytoplasmiques avec domaines de liaison et régions répétées riches en leucine ont été isolés. Ces protéines agissent comme un système de surveillance affecté à la reconnaissance de protéines Avr d'agents pathogènes spécifiques. Elles déclenchent l'alerte au moindre signe d'invasion et activent par la suite les mécanismes de défense de la plante. Plusieurs de ces protéines Avr de la rouille ont également été isolées et nous avons remarqué qu'il s'agissait de petites protéines sécrétées qui sont exprimées dans les haustoriums (structures spécialisées qui pénètrent la paroi de la cellule hôte) et déposées dans le cytoplasme de l'hôte durant la phase d'infection. L'identification des gènes R du lin, et des gènes Avr correspondants, nous a permis de procéder à une analyse plus détaillée des phénomènes de reconnaissance qui provoquent la résistance, ce qui indique une interaction directe entre les protéines R et Avr en tant que fondement de la résistance à la rouille du lin.

Keywords:

• avirulence genes
• disease resistance genes
• effector proteins
• flax rust disease
• plant–pathogen interactions
• rust fungi

Mots clés:

• champignons de la rouille
• gènes d'avirulence
• gènes de résistance
• interactions plante-agent pathogène
• protéines effectrices
• rouille du lin

Introduction

Rust fungi are a large group of highly destructive plant pathogens with individual species generally infecting only a narrow range of host plants. Consisting of about 5000 species, rusts collectively cause disease on many important agricultural and silvicultural crops. These biotrophic fungi have evolved effective strategies to utilize the host's living cells as a nutritional source, a contrasting form of parasitism to that of necrotrophic fungi, which quickly kill their hosts to feed on the dead plant material. By maintaining a living environment, these biotrophic fungi are able to exploit a niche rich in nutrients. However, rusts are also obligate parasites making them completely reliant on living host tissue for all stages of their growth and reproduction, and are consequently notoriously difficult to study using in vitro methods.

During infection, rust fungi penetrate the leaf surface of their host, either through stomata or by direct penetration of epidermal cells, and produce intercellular hyphae. Contact with mesophyll cells by growing intercellular hypha induces the formation of specialized feeding structures known as haustoria. These fungal cells begin as a narrow, hyphal-like projection that penetrates the mesophyll cell wall. Once the cell wall is breached, the distal end expands, invaginating the host plasma membrane to differentiate into a haustorium, and thus remains separated from the host protoplast. This host–haustorium interface provides intimate contact between the fungus and the plant and is believed to play an important role in the uptake of nutrients, the exchange of molecular signals, and ultimately in establishing biotrophy (Panstruga, 2003; Voegele & Mendgen, 2003). Indeed, haustoria are a feature of many biotrophic fungi and oomycetes, and are also produced during transient biotrophic stages of some hemibiotrophic species (Perfect & Green, 2001).

It is now widely accepted that haustoria secrete a suite of effector proteins to modulate plant innate immune responses and alter host cells to facilitate a successful infection in susceptible cultivars (Kamoun, 2006; Catanzariti et al., 2007; Birch et al., 2009; Panstruga & Dodds, 2009; Tyler, 2009). In a countermeasure to their actions, plants have evolved an efficient surveillance system to detect a subset of these effectors and mount a defence response. Thus, in resistant cultivars, host cells containing nascent haustoria are sites of the hypersensitive response (HR), a plant-induced cell death that restricts the pathogen's growth (Kobayashi et al., 1994; Heath, 1997). This response is the result of effector recognition by a corresponding plant resistance (R) protein and is associated with a rapid and very effective defence response that prevents disease. This form of plant resistance is known as effector-triggered immunity (ETI), but was first described genetically as ‘gene-for-gene’ resistance (Flor, 1971), due to its dependence on the presence of both an R protein in the host and matching effector in the pathogen with the absence of either resulting in disease. Since an effector that is recognized by an R protein prevents pathogen growth, they are also referred to as avirulence (Avr) proteins.

Given their importance as key components of ETI, numerous plant R proteins and many pathogen effector proteins from different pathosystems have been identified. R proteins share structural domains which have been used to categorize them into different classes (McHale et al., 2006). By far the most common class are the cytoplasmic NB-LRR proteins that contain a nucleotide-binding (NB) and C-terminal leucine-rich repeat (LRR) domain. These proteins can be further categorized based on the presence of either a coiled coil (CC) or TIR domain, which has homology to the Toll and interleukin-1 receptors, at the N-terminus. By contrast, Avr proteins show a great deal of sequence and functional diversity (Chisholm et al., 2006). With few exceptions, these proteins are secreted from the pathogen and in many cases targeted into the host cell. The Avr proteins from bacteria are injected directly into host cells by the type III secretion system and many have been shown to facilitate disease, particularly through the suppression of various basal plant defence responses (Abramovitch et al., 2003; Hauck et al., 2003; DebRoy et al., 2004; Kim et al., 2005; Janjusevic et al., 2006). The Avr proteins from biotrophic fungi and oomycetes are also delivered into host cells, however, the precise translocation mechanisms are not yet known (Catanzariti et al., 2007; Tyler, 2009). To date, several oomycete Avr proteins have also been found to promote virulence by suppressing basal defences (Sohn et al., 2007; Dou et al., 2008a).

An extremely useful model system to study ETI against rust pathogens is the interaction between the fungus Melampsora lini (Ehrenb.) Desm. and its host flax (Linum usitatissimum L.). This pathosystem was the basis of the seminal gene-for-gene concept first proposed by Harold Flor (1971), whose extensive work on the inheritance of rust resistance in the host and avirulence in the pathogen, as well as considerable research by several others (Shepherd & Mayo, 1972; Lawrence et al., 1981; Lawrence, 1988; Islam et al., 1989), resulted in a comprehensive understanding of the genetics of this plant–pathogen interaction. The identification of flax R genes for over half of the defined rust resistance specificities, and the more recent identification of several corresponding rust Avr genes, has allowed insights into host–pathogen specificity and the molecular basis of the R–Avr interaction. In this review, we focus on the discovery of rust effectors and their recognition by R proteins in this model plant–rust interaction.

Rust resistance genes in flax

Thirty-one different rust resistant specificities have been identified in flax which map to just five different multiple-allele loci (K, L, M, N, and P) with different genetic arrangements (Islam & Mayo, 1990). R genes for 20 of these specificities have been cloned from four loci and each encodes an intracellular TIR-NB-LRR protein (Table 1; Lawrence et al., 1995, 2010; Anderson et al., 1997; Ellis et al., 1999; Dodds et al., 2001a, 2001b). The L locus contains a single gene with alternative alleles that represent different resistance specificities for which all 12 have been cloned. The nucleotide sequences of these genes share greater than 90% DNA sequence identity, with the greatest variation occurring within the LRR domain, mainly through point mutations, although in L1, L2 and L8 there are also duplications and deletions in this region. Furthermore, it appears that just single intragenic sequence exchange events separate L6 from both L11 and L7, since all of the sequence variation between these pairs occurs in the LRR and TIR domains, respectively.

Table 1.  Flax rust resistance genes

Loci	Locus complexity	Encoded specificities	Cloned	Reference
K	∗	2	–	–
L	single gene	12	L, L1–L11	Lawrence et al. (1995); Ellis et al. (1999)
M	∼15 genes	7	M, M1, M3	Anderson et al. (1997); Lawrence et al. (2010)
N	4 genes	3	N, N1, N2	Dodds et al. (2001a)
P	6–8 genes	6	P, P2	Dodds et al. (2001b)
Note: ∗No genetic studies for this locus have been reported.

The other loci from which R genes have been cloned are M, N and P which, unlike L, are complex loci containing a number of closely linked paralogues arranged in tandem with generally only one member conferring the designated resistance specificity. The M genes are closely related to those at the L locus with more than 80% DNA sequence identity and these are believed to be homoeologous loci from the ancient tetraploid flax genome. The L and M proteins have a hydrophobic N-terminal region that is predicted to be a membrane anchor, and expression of these sequences fused to a GFP reporter gene found the protein at the Golgi membrane or plasma membrane, respectively (P.N. Dodds, unpublished data), suggesting this region may facilitate membrane association. The N and P genes have low sequence identity to L and M genes, with the P genes having the greatest divergence among all the cloned flax rust genes. Genes from this locus also encode an additional 150 amino acids after the LRR region, referred to as a C-terminal non-LRR (CNL) domain.

The specificity of these flax R genes against different rust isolates has been explored extensively and, using sequence data and domain swap experiments, it was shown that the LRR domain is the major determinant of Avr recognition specificity (Ellis et al., 1999; Dodds et al., 2001a, 2001b). This is consistent with the LRR domain being the most variable region between allelic R genes. An exception is the L6 and L7 genes which differ only within the TIR domain. The L7 gene confers a weak resistance allowing a small amount of rust sporulation, compared with L6 which completely restricts pathogen growth (Islam & Mayo, 1990). The cloning of the corresponding Avr genes (see below) revealed these R genes have the same specificity, and differ only in the strength of their reaction. Thus, the TIR domain appears to have a signalling role rather than a role in pathogen recognition.

Avirulence genes in flax rust

Unlike their R gene counterparts, the flax rust Avr genes are dispersed throughout the M. lini genome, however there are several tightly linked genes that usually segregate as gene clusters. Currently, flax rust Avr genes have been cloned from four loci (Table 2; Dodds et al., 2004; Catanzariti et al., 2006). These genes were identified using two different approaches, but in each case, candidate genes were mapped by restriction fragment length polymorphisms (RFLPs) in an F₂ family segregating for multiple Avr specificities. Genes that co-segregated with avirulence were then confirmed to encode the Avr specificity by observing an R gene-dependent HR when transiently expressed in flax lines of the appropriate genotypes.

Table 2.  Cloned flax rust avirulence genes

Avr locus	Mature protein size∗	No. cloned genes and rust strains identified from	Protein features	References
AvrL567	127	12 (C, H, I, J,Bs1, Fi)	Directly interacts with L5/L6 in yeast	Dodds et al. (2004); Dodds et al. (2006)
AvrM	184–349	6 (CH5)	Directly interacts with M in yeast	Catanzariti et al. (2006, 2010)
AvrP/P123	88–94	6 (CH5, H, I, J,Bs1, Fi)	10 Cys; Kazal consensus seq	Catanzariti et al. (2006); Dodds & Thrall (2009)
AvrP4	67	3 (CH5, WA)	6 Cys; potential Cys-knot structure	Catanzariti et al. (2006)
Note: ∗Predicted amino acid size after signal peptide cleavage.

The first Avr genes to be isolated were three members of the AvrL567 locus. These genes were identified from a cDNA library created using suppressive subtractive hybridization to enrich for rust genes expressed during infection, and encode small secreted proteins with overlapping recognition specificities with the L5, L6 and L7 resistance proteins. The F₁ parent strain CH5, from which these genes were isolated, has two genes at the avirulence allele (A and B) and a single gene at the virulence allele (C). AvrL567-A is recognized by all three L proteins, while AvrL567-B is recognized only by L5 and L6, and AvrL567-C, which is derived from the virulent allele, is not recognized by any of the L proteins. Subsequent analysis in seven different rust strains identified a total of 12 variants designated AvrL567-A to -L, which occur at the locus in various arrangements of one to four genes (Dodds et al., 2006). Of these variants, seven encode avirulence specificities and are recognized by the L5, L6 or L7 proteins, while the other five are derived from virulence alleles and therefore are not recognized. The AvrL567 variants recognized by L6 are also weakly recognized by L7, as evident from a weaker HR during transient in planta expression. It is interesting that most of the AvrL567 proteins are recognized by both L5 and L6, as these proteins represent the two most divergent L variants (Ellis et al., 1999).

Consistent with intracellular localization of the flax rust R proteins, this R-Avr recognition occurs inside the plant cell, as in planta expression of AvrL567 proteins as cytoplasmic proteins (lacking the signal peptide) induces an R gene-dependent HR. The AvrL567 genes are highly polymorphic with 35 variable residues within the 127 amino acids of the predicted mature protein. The structure of this protein family was determined by X-ray crystallography, revealing that all but one of the side chains of these variable residues are surface exposed (Wang et al., 2007). The AvrL567 proteins have no structural or sequence similarity to any other known proteins and, beside the signal peptide sequence, do not contain any functional motifs. During infection, the AvrL567 proteins are expressed and secreted from the haustoria, however, it is not yet known how these proteins cross the invaginated host membrane to enter the host cell, or what virulence function they may carry out.

Based on AvrL567 characteristics, the second approach to identify flax rust genes corresponding to different Avr specificities was to generate a cDNA library from isolated haustoria, then search for ESTs containing a predicted secretion signal peptide. This strategy proved to be extremely successful, isolating AvrM, AvrP/P123 and AvrP4 genes from 20 putative haustorially expressed secreted proteins. Like AvrL567, these Avr elicitors are recognized by the corresponding R proteins inside the host cell (Catanzariti et al., 2006; Barrett et al., 2009), and they have no sequence similarity to currently known proteins or an assigned function. However, as these genes are maintained across different rust stains it is assumed they encode proteins with roles in pathogenicity. The AvrM locus in rust strain CH5 contains at least five variant genes, AvrM-A to -E, and a single gene designated avrM at the alternative allele derived from the virulent parent. The predicted mature proteins encoded by the six AvrM genes have 15 polymorphic residues and vary considerably in size (184–349 amino acids) due to truncations, deletions and a short duplication. Except for AvrM-E, which is a significantly truncated protein lacking the C-terminal region, AvrM proteins derived from the avirulence allele are all recognized by the M resistance protein, while the avrM protein is not. AvrM transcripts can also be detected in the germ tubes of spores that have been germinated in vitro, which is a feature not shared by genes cloned from other Avr loci.

Genes at the AvrP/P123 locus, like those of AvrL567, are highly polymorphic and encode overlapping avirulence specificities. The two genes identified from this locus in rust strain CH5 were designated AvrP and AvrP123, and occur as alternative alleles. AvrP is recognized by the P resistance protein and AvrP123 is recognized by P1, P2 and P3. Analysis of this locus in bs25, a recombinant strain derived from CH5, identified a recombinant gene containing an N-terminal AvrP sequence and C-terminal AvrP123 sequence (Dodds & Thrall, 2009). Expression of this recombinant gene revealed recognition only by the P2 resistance gene, suggesting avirulence specificity is conferred by different regions of the Avr protein, with polymorphisms in the N-terminus involved in recognition by P, P1 and P3, and C-terminal polymorphisms influencing recognition by P2. Analysis in other rust strains identified additional sequence variants, including one with avirulence specificity to P2 and P3 and a virulence variant that is not recognized by any of the P genes (Dodds & Thrall, 2009). The cloned AvrP/P123 genes encode predicted mature proteins of 88 or 94 amino acids with 38 variable residues and 10 conserved cysteine residues. Also conserved among the variants is a sequence signature found in the Kazal family of serine protease inhibitors, however, such a function has not been determined.

The fourth cloned flax rust Avr locus contains a single gene at alternative virulence (avrP4) and avirulence (AvrP4) alleles with specificity corresponding to the P4 resistance gene. Both genes encode a cysteine-rich protein with a predicted mature protein size of 67 amino acids. Sequence comparison of these proteins and an additional sequence variant isolated from another flax rust strain revealed these proteins to have seven polymorphic residues that cluster around the six conserved cysteines present at the C-terminus. Alanine mutations of these cysteines abolished R-mediated HR and are therefore likely to form important structural disulfide bonds (Catanzariti et al., 2006). AvrP4 homologues from a range of Melampsora spp. show an enormous level of amino acid divergence within the C-terminal cysteine-rich domain, with some variants differing at every residue in this region except for the conserved cysteines (Van der Merwe et al., 2009). This region of the protein has clearly been under strong diversifying selection during speciation within this genus, while the flanking regions have remained well conserved, which may be a response to host resistance or an adaptation to divergent host target proteins. It is also interesting that the cysteine residues fit the spacing found in the family of inhibitor cysteine knot structures, members of which include toxins and inhibitors of receptors or proteases (Pallaghy et al., 1994). However, the cysteine knot motif has been observed in many unrelated proteins with a variety of biological activities (Koomann-Gersmann et al., 1997).

Translocation of avirulence proteins into host cells

The discovery that flax rust Avr proteins are recognized by R proteins inside the plant cell supports the conclusion that rust pathogens deliver effectors into the host cell to carry out various virulence functions. The signal peptide of these Avr proteins presumably directs their secretion into the extrahaustorial space between the haustorial cell wall and the host plasma membrane, but the mechanism facilitating their translocation across the membrane into the host cytoplasm has not been determined. Further evidence for the movement of rust proteins across the haustorium–host interface exists with the secreted bean rust protein Uf-RTP1 from Uromyces fabae (Pers.) de Bary. Immunolocalization studies detected this protein initially within the extrahaustorial matrix, and as infection progressed, it accumulated within the cytoplasm of the infected host cell (Kemen et al., 2005). During later stages of infection, this protein was also found in the host nucleus, which is consistent with the presence of a predicted nuclear localization signal, and suggests a possible role in altering host gene expression during infection. As several other bean rust proteins expressed and secreted from haustoria were only detected within the extrahaustorial matrix, the localization of Uf-RTP1 implies a selective protein transport mechanism (Kemen et al., 2005).

The deployment of different sets of pathogen effectors to either the host cell cytoplasm or to the extracellular environment (either the extrahaustorial matrix or apoplast) is particularly evident among the large repertoire of secreted effector proteins from oomycete pathogens. Studies of several oomycete phytopathogens have identified conserved peptide motifs present in proteins known to enter host cells during infection (Rehmany et al., 2005; Tyler, 2009). These motifs are absent from proteins that are known to function in the apoplast, which are by contrast often cysteine-rich (Kamoun, 2006). Two families of host-cell targeted proteins have been identified in oomycetes, the RXLR effectors and the less characterized CRN (Crinkler) effectors. Sequence data have revealed RXLR proteins to be ubiquitous among this class of pathogens, and are classified based on a signal peptide sequence followed by an RXLR motif. In many of these effectors the RXLR motif is also followed by a stretch of acidic amino acids ending in an EER motif. Avr proteins have been identified from both biotrophic and hemibiotrophic oomycetes that also form haustoria during infection (Allen et al., 2004; Shan et al., 2004; Armstrong et al., 2005; Rehmany et al., 2005; van Poppel et al., 2008). These effectors are all recognized by cognate R genes inside the host cell and have an RXLR motif, which, along with the downstream EER sequence, was recently shown to be required for uptake into the plant cells (Whisson et al., 2007; Dou et al., 2008b). Moreover, one of these effectors was shown by microscopy to be located in host cells containing a haustorium (Whisson et al., 2007).

The precise uptake mechanism of these effectors has not yet been identified, although experimental evidence suggests that it may be independent of the pathogen. The RXLR effector Avr1b from the soybean pathogen Phytophthora sojae Kaufm. & Gerd., when expressed from Pichia pastoris (Guillierm.) Phaff and infiltrated into the leaf apoplast was able to trigger a cell death in leaves containing the cognate cytoplasmic Rps1b R gene (Shan et al., 2004). Furthermore, it was shown that the N-terminal leader sequence of Avr1b can direct the uptake of a GFP fusion protein, expressed and purified from Escherichia coli (Mig.) Castel. & Chalm., into the root cells of soybeans in the absence of the pathogen (Dou et al., 2008b). No common sequence motifs have been identified among the cloned flax rust effectors, but these proteins also appear to have intrinsic membrane translocation properties, suggesting they utilize a plant-derived transport system. For example, an M-dependent cell death response is triggered when AvrM is expressed in the plant either with or without the signal peptide. This necrosis was abolished by the addition of an ER retention signal only when AvrM is expressed with the signal peptide, suggesting the secreted AvrM protein re-enters the host cytoplasm in the absence of the rust. Furthermore, this translocation is also mediated by the N-terminal region, which is sufficient to direct re-uptake of a secreted GFP fusion protein when transiently expressed in planta (P.N. Dodds, unpublished data). Having independently evolved similar infection processes, it will be interesting to determine whether biotrophic fungi and oomycetes also have similar protein delivery mechanisms.

The molecular basis of Avr recognition in R-mediated flax rust resistance

The investigation of ETI in various plant pathogen systems has revealed that R proteins recognize corresponding effectors either through a direct interaction (receptor-ligand model) or indirectly via a third host protein (guard model). Under the latter scenario, R proteins have been shown to associate with (guard) a host protein that is targeted for modification by a pathogen effector, with alterations to the host protein resulting in the activation of resistance signalling pathways. Several well established examples of indirect R–Avr interactions include RPM1, RPS2 and RPS5 from Arabidopsis thaliana (L.) Heynh., which provide resistance to the bacteria Pseudomonas syringae van Hall (Axtell et al., 2003; Mackey et al., 2003; Shao et al., 2003). While a direct recognition mechanism involving the physical interaction between corresponding R–Avr protein pairs has been demonstrated for Pi-ta from rice with Avr-Pita from Magnaporthe grisea (Hebert) Barr (Jia et al., 2000); RRS1 from A. thaliana with the effector PopP2 from the bacteria Ralstonia solanacearum (Smith) Yabuuchi (Deslandes et al., 2003) and the N protein from tobacco with the p50 elicitor from tobacco mosaic virus (Ueda et al., 2006).

In the flax rust system, a direct physical association has been shown between the AvrL567 effectors and the corresponding L proteins (Dodds et al., 2006). Using a yeast two-hybrid assay, a pair-wise analysis of the 12 AvrL567 variants (AvrL567-A to AvrL567-L) against the L5 and L6 alleles found a strict correlation between the R–Avr interaction and the induction of HR in planta, indicating that direct R–Avr protein interaction is the basis for recognition specificity. Single amino acid changes made to the AvrL567 proteins and their position on the solved three-dimensional structure revealed that this interaction is contributed to by multiple opposing amino-acid contact points on the surface of the effector that have a cumulative effect on the strength of the interaction (Wang et al., 2007). Also tested in this yeast assay was a chimeric R protein that is identical to L6 except for 11 amino acids derived from L11 at the C-terminal end of the LRR domain. This chimera confers a new specificity, interacting only with AvrL567-J both in yeast and in planta, and implies that binding occurs between the Avr protein and the LRR domain, supporting previous observations that this domain controls recognition specificity. However, experimentation suggests that a functional NB domain is also required in Avr binding. A mutation in the P-loop motif of the NB domain that causes a loss of ATP binding eliminates both yeast interactions and the HR response in plants (Dodds et al., 2006), while deletion analysis revealed the smallest interacting protein includes both the NB and LRR domains (P.N. Dodds, unpublished data). It is thought that although the Avr protein binds to the LRR domain, the NB domain may be required to establish the correct protein conformation necessary for Avr binding. A similar situation has also been reported for the N protein from tobacco. This TIR-NB-LRR resistance protein requires both the NB and LRR domain to interact with the p50 effector, and this interaction is also dependent on ATP binding ability of the NB domain (Ueda et al., 2006). On the other hand, the CC-NB-LRR resistance protein Pi-ta directly interacts with Avr–Pi-ta solely through the LRR domain (Jia et al., 2000).

A direct interaction between M and AvrM has also been demonstrated by yeast two-hybrid analysis and shown to correlate with the in planta recognition specificity observed for each of the different AvrM variants (Catanzariti et al., 2010). Furthermore, this correlation extends to a series of AvrM truncations, whereby proteins that had lost their ability to induce an M-dependent necrotic response, also lost their ability to interact with M in the yeast assay (Catanzariti et al., 2010). Together, these data suggest that a direct interaction underlies the recognition event between these proteins. Work to solve the structure of AvrM is currently underway and may help determine the nature of this interaction and the role that variable residues play in specificity.

The molecular basis of effector recognition in R-mediated resistance may influence the co-evolution of R and Avr genes. R loci that encode proteins whose strategy is to guard host proteins targeted by effectors have generally been found to be conserved, and not under diversifying selection (Stahl et al., 1999; Mauricio et al., 2003). Pathogen effectors that are recognized indirectly by their virulence function (by modifying the guarded host protein) can only escape detection by becoming non-functional. This is reflected in the observation of such effectors as either present or absent within a pathogen species, and as such, the guarding R proteins that associate with the target host protein do not require further diversification. On the other hand, effector proteins that are recognized by a direct physical interaction by the R protein, can avoid detection by altering binding sites while maintaining their function. In response, selective pressure drives the evolution of R genes to new recognition specificities, with this ‘arms race’ giving rise to diversifying selection on both the R locus in the host and Avr locus in the pathogen. This situation is exemplified in the flax rust pathosystem, where sequence analysis at R and Avr loci show evidence for strong diversifying selection (Dodds et al., 2000, 2001a, 2001b, 2004, 2006; Catanzariti et al., 2006). In the flax R genes, this occurs largely in the region encoding the LRR domain, the major determinant of recognition specificity, and at the rust loci there is a propensity for overcoming resistance by sequence diversity rather than deletion or inactivation of the gene. It will be interesting to discover the basis of recognition for other flax R proteins to determine if a direct R–Avr interaction is common in the flax rust disease system, and perhaps extends to other biotrophic pathogens.

Summary

The comprehensive historical genetic analyses and more recent molecular studies on the model flax rust disease system have significantly advanced our understanding of host–rust interactions. The isolation of flax rust avirulence proteins, and other proteins secreted from haustoria, provides an exciting opportunity to investigate how rusts deliver effectors into plant cells, and to address what pathogenicity function they might carry out. Elucidating the processes used by rust fungi to establish an infection may uncover novel approaches to control these devastating plant pathogens. Further assistance in disease control will come from the considerable progress that has been made to understand rust resistance specificity. Experimental examination has revealed that the LRR domain of flax R proteins plays a pivotal role in the recognition of pathogen effectors, which in at least two cases is via a direct interaction. The emerging picture is that the LRR domain, coupled with the NB domain, has the capacity to interact with diverse pathogen effectors, and could therefore be engineered for new recognition specificities to be deployed in agriculture. It is anticipated that future work on the flax rust disease system will continue to provide valuable knowledge on plant–pathogen interactions and will remain an important model system for studying rust disease.

Notes

^†Contribution to the symposium “Signalling in Plant-Pathogen Interactions” held during the Canadian Phytopathological Society Annual Meeting, 22–25 June 2009, Winnipeg, MB.