Portal de Periódicos da CAPES

Resistance to phytopathogens e tutti quanti : placing plant quantitative disease resistance on the map

2014; Wiley; Volume: 15; Issue: 5 Linguagem: Inglês

10.1111/mpp.12138

1464-6722

Fabrice Roux, Derry Voisin, Thomas Badet, Claudine Balagué, Xavier Barlet, Carine Huard‐Chauveau, Dominique Roby, Sylvain Raffaele,

Plant Pathogens and Resistance

Plant disease resistance can be seen as a process of a dual nature: both qualitative and quantitative. The nature of non-self molecules perceived by plants has led to the depiction of plant immunity as a two-layer defence system. The first layer is mediated by cell surface and intracellular pattern recognition receptors (PRRs) which perceive conserved microbial elicitors, termed pathogen-associated molecular patterns (PAMPs). The perception of these conserved elicitors initiates cascades of signalling and transcription events, known as PAMP-triggered immunity (PTI). Adapted pathogens secrete effector molecules able to suppress PTI, but which may also be recognized by plant intracellular resistance (R) proteins. This initiates effector-triggered immunity (ETI), the second layer of plant defence. ETI typically yields complete disease resistance phenotypes against pathogens containing the recognized effector, a process designated as qualitative resistance. By contrast, perception of a single PAMP typically has a weaker contribution to overall plant resistance. More generally, in the absence of qualitative resistance, an incomplete resistance phenomenon is often observed, leading to a reduction rather than absence of disease. This is usually referred to as quantitative disease resistance (QDR). The definition of QDR varies in the literature according to the interests of the authors, so that the QDR concept may be puzzling. We settled to define QDR based on the simultaneous occurrence of two observations: (i) the disease resistance phenotype follows a continuous distribution from susceptibility to resistance in host populations, as opposed to the 'susceptibility or resistance' binary phenotype observed in qualitative resistance (Fig. 1A); and (ii) the disease resistance phenotype is determined by multiple quantitative trait loci (QTLs) as revealed by crossing host parents with contrasted phenotypes. As a consequence, full disease resistance can be achieved by the introduction of a single R gene, but not by the introduction of a single QDR gene, in a susceptible genetic background. Based on this definition, the molecular architecture of QDR can be hypothesized as an intricate network integrating multiple response pathways to several pathogen molecular determinants and environmental cues (Fig. 1B). The qualitative–quantitative duality of plant disease resistance from the population biology and molecular biology points of view. The figure illustrates the qualitative (left panels) and quantitative (right panels) nature of resistance. From a population biology point of view (A), the disease resistance phenotype follows a binary 'susceptible or resistance' distribution for qualitative resistance, but a continuous distribution from susceptibility towards resistance for quantitative resistance. From a molecular biology point of view (B), qualitative resistance results from the perception of a single pathogen effector (Avr) by a plant resistance (R) gene, whereas quantitative resistance results from the integration of multiple perception pathways activated simultaneously, each having a relatively minor contribution to the overall resistance phenotype. QDR, quantitative disease resistance. Like other previous attempts, this definition tends to set QDR and qualitative resistance as mutually exclusive, based on apparently distinct phenotypic outcomes. The discrepancy between QDR and qualitative resistance is therefore dependent on our ability to read and interpret resistance phenotypes. Indeed, a binary 'susceptible or resistant' phenotype could appear to be continuous depending on the environment, or if the size of the host population analysed increases, or if the accuracy and resolution of the phenotyping methodology increases. Reciprocally, pathogen genotypes, environmental conditions and natural mutations in R genes, PRRs or modifier loci, altering the strength of resistance conferred by R genes and PRRs, are expected to introduce some degree of quantitative variation in PTI and ETI. R-mediated resistance can be seen as an extreme of the phenotypic spectrum, in which the switch from susceptibility to resistance in plant populations is reduced to a minimum of detectable transition states, probably relying on a reduced number of molecular differences between resistant and susceptible interactions (typically the presence/absence of an R gene and the recognized pathogen effector, Fig. 1B). Plant disease resistance is therefore intrinsically dual, with both a qualitative-like and a quantitative-like behaviour. This assumption is consistent with the observation that the genetic components underlying QDR and qualitative resistance can be partially overlapping (see next paragraph). It also suggests that the knowledge gained in recent years on the molecular determinants of qualitative resistance only represents the tip of the iceberg, with QDR-specific molecular components and mechanisms, and their connections to other signalling pathways, still to be discovered. Known plant receptors mediating resistance typically belong to two classes: PRRs frequently contain, or associate with, non-arginine–aspartate (non-RD) kinases, whereas R genes mediating ETI are nucleotide binding-leucine-rich repeat (NB-LRR) receptors lacking kinase domains. Although numerous disease resistance QTLs have been identified in plants, the genes and mechanisms underlying QDR remain largely unknown. The recent cloning of a limited number of QDR genes through QTL mapping approaches nonetheless suggests that some QDR genes may be members of the two well-described classes of plant immune receptors. Indeed, NB-LRR genes underlying resistance QTLs have been cloned in rare instances, such as RCG1, conferring QDR to anthracnose stalk rot in maize, and RLM1, conferring QDR to the blackleg disease in Arabidopsis. R-gene loci have also been frequently co-localized with resistance QTLs, but their roles in QDR have not been demonstrated. These findings have led to the hypothesis that weak alleles of R genes can cause QDR. However, most QDR genes identified to date do not correspond to typical immune receptors and encode a broad range of molecular functions (see discussion in Huard-Chauveau et al., 2013). Three kinases or kinase-like genes, presumably acting as defence signalling proteins, have been reported as QDR genes (WKS1/Yr36 in wheat, WAKL22/RFO1 and RKS1 in Arabidopsis); other QDR genes encode putative transporters (at the Lr34 locus in wheat and Rhg1 locus in soybean) and three QDR genes correspond to previously unidentified defence genes: the soybean wound-inducible domain protein WI12, which may be involved in the production of compounds toxic to nematodes, the soybean RHG4 serine hydroxymethyltransferase and the rice proline-containing protein Pi21. These molecular functions have not been associated previously with plant disease resistance, suggesting that the molecular mechanisms underlying QDR may be more diverse than anticipated. We should, however, acknowledge that, as for most quantitative traits, the identification of QDR genes has focused on large-effect QTLs (i.e. QTLs explaining more than 20% of phenotypic variance). It remains to be determined whether small-effect QDR genes will broaden the range of molecular functions associated with plant disease resistance. In addition, although cross-talk has been found between PTI and ETI pathways, the connections of these pathways with QDR remain scarce and the placement of the identified QDR components in known disease resistance pathways remains enigmatic. As ETI relies on the specific recognition of highly variable pathogen effectors by the corresponding plant R-gene product, it generally functions only against specific isolates of a pathogen species. In some rare cases, R genes of the NB-LRR class are effective against multiple isolates and species, such as RPW8, which confers resistance against multiple powdery mildew pathogens in Arabidopsis (Xiao et al., 2001). It has been proposed that broad-spectrum R genes do not recognize pathogen effectors directly or that they recognize effectors conserved across several pathogen lineages. By contrast, QDR genes typically provide broad-spectrum resistance. The ABC transporter Lr34 is a representative example, which confers QDR against both ascomycete and basidiomycete fungal pathogens in wheat (Krattinger et al., 2009). Again, broad-spectrum resistance conferred by QDR genes may relate to the perception and response pathways to well-conserved microbial signatures, such as conserved effectors and PAMPs. For instance, expression of the PRR gene EFR confers QDR to adapted and nonadapted pathogens in several plant families (Lacombe et al., 2010). Alternatively, like most QDR genes identified to date, they may correspond to downstream effector response components. These could either act at the intersection of several effector perception pathways or represent common host targets indirectly manipulated by several effectors. Indeed, we speculate that QDR results from the integration of multiple perception pathways activated simultaneously, each having a minor contribution to the overall resistance phenotype (Fig. 1B). The global analysis of Arabidopsis proteins interacting with 30 effectors of the bacterial pathogen Pseudomonas syringae and 53 candidate effectors of the oomycete pathogen Hyaloperanospora arabidopsidis revealed an intricate network of interactions with hubs associated with the response to multiple effectors from both pathogens (Mukhtar et al., 2011). Many QDR components identified to date may correspond to such hubs, or response pathway integrators, for signalling pathways initiated by multiple recognition events. Multiple signals from different pathogen races, strains, pathovars or species, possibly recognized by different receptors, converging for signalling on such hubs, could account for the broad-spectrum nature of QDR. In this context, R gene-mediated resistance could be an extreme form of activation of the same signalling network in which signalling through a single major pathogen effector is strongly predominant. A number of race-specific resistance QTLs (reviewed, for instance, in Poland et al., 2009) also suggest that network hubs are not the only components of the QDR response. Downstream effector targets more or less sensitive to manipulation by one or several effectors might lead to QDR. Future molecular and biochemical studies encompassing both plant and pathogen diversity should provide access to the full complexity of plant immune response signalling networks. Although the recognition of pathogen effectors by NB-LRR plant proteins is associated with resistance to biotrophic pathogens, it can specifically trigger susceptibility to some necrotrophs (Lorang et al., 2007). It is becoming apparent that components of the plant immune response to biotrophs are frequently exploited by necrotrophic fungi to facilitate infection. However, among the numerous resistance QTLs against various necrotrophic and hemibiotrophic pathogens that have been reported, some are also efficient against obligate biotrophic pathogens. This is the case, for instance, for Lr34 in wheat, which confers resistance to leaf, stripe and stem rust hemibiotrophic pathogens, as well as to the obligate biotrophic powdery mildew pathogen Blumeria graminis f.sp. tritici (Krattinger et al., 2009). Most pathogens against which QDR is the predominant form of plant resistance are generalists, infecting a broad range of host species. Sclerotinia sclerotiorum (Perchepied et al., 2010) and Botrytis cinerea (Denby et al., 2004), each infecting several hundreds of plant species, are paradigmatic examples. However, in a few cases, QDR has been observed against host-specific pathogens, such as B. graminis mentioned earlier. Furthermore, QDR has been observed against viral, bacterial and eukaryotic pathogens (notably, fungal, oomycete and nematode pathogens). Finally, plants exhibit QDR responses against root-infecting, foliar and vascular pathogens, suggesting that several plant cell types express components of the QDR machinery. Thus, QDR appears to be a rather ubiquitous process, raising the question of whether specific QDR mechanisms exist against each class of pathogen. The example of Lr34, efficient against rust and powdery mildew pathogens with different lifestyles, suggests that some common genetic bases exist for QDR against diverse pathogens. Because of the intense selection pressure imposed by qualitative resistance on avirulent pathogen populations, it is expected that a higher degree of host specialization will be observed in pathogens confronted with qualitative resistance than in those confronted with QDR. Qualitative resistance may therefore become a frequent form of resistance in plant–pathogen interactions with a long co-evolution history or in environments in which one pathogen population is largely predominant. Conversely, QDR may become predominant in complex environments in which diverse pathogens co-occur. Theoretical models of population evolution including the simultaneous contribution of multiple R genes should help in testing this hypothesis. Over the past two decades, an ever-increasing interest has been focused on understanding the evolutionary forces shaping the natural variation of R genes. Interestingly, contrasting evolutionary dynamics have been observed among genes associated with the different layers constituting the complex plant immune response. Although evidence of selection acting on signalling genes and response genes appears to be sparse, strong signatures of selection have been commonly observed for genes involved in the direct or indirect recognition of pathogens. In particular, many cases of balancing selection have been documented in the latter class of genes, with the observation of two major patterns of variation at R genes: i.e. the presence of a large number of diverse functional alleles or presence–absence polymorphisms with functional alleles and a null allele segregating within a species. Both patterns of variation are thought to result from frequency-dependent selection, leading to the maintenance of long-lived polymorphisms associated with R genes over evolutionary time. Until very recently, whether or not QDR genes showed hallmarks of selection similar to those for gene-for-gene resistance was an open question. A study on RKS1, conferring quantitative resistance in A. thaliana to the vascular bacterial pathogen Xanthomonas campestris, revealed that alternative resistance mechanisms may share common selective pressures (Huard-Chauveau et al., 2013). Indeed, similar to the RPM1 and RPS5 R genes, a strong signature of balancing selection acting on two highly divergent haplotypes has been detected in RKS1. This signature of selection is further supported by the presence of polymorphic populations at the RKS1 locus across the native range of A. thaliana. A striking difference between the RPM1 and RPS5 R genes and RKS1 can, however, be highlighted. Presence–absence polymorphisms are subject to balancing selection at the RPM1 and RPS5 genes. In contrast, balancing selection was found to act on two highly differentially expressed RKS1 haplotypes (Huard-Chauveau et al., 2013). Because the low-RKS1-expressed haplotype is associated with susceptibility to X. campestris, we may wonder why natural selection maintained this susceptible functional allele in A. thaliana over a long time period. Investigations of the evolutionary dynamics of other QDR genes are clearly needed to establish: (i) whether this observation is commonplace among QDR genes; and/or (ii) whether the signature of selection depends on the position of the QDR genes in the defence pathways. Because the intensity of selection imposed by enemies clearly varies among natural populations and the phenotypic effect of an allele may depend strongly on the genetic background, we propose to explore the natural genetic variation of the cost–benefit trade-offs associated with QDR genes under increasing pathogen load. Estimating the cost–benefit trade-offs in different genetic backgrounds along a range of infection intensities will permit the elucidation of the evolutionary forces acting on QDR genes. In the case of the maintenance of long-lived polymorphism, the detection of a large fitness cost associated with the presence of the RPM1 gene in A. thaliana provided evidence that costs can contribute to the maintenance of an ancient R-gene polymorphism. The presence of a cost of resistance also seems to be an attractive hypothesis to explain the long-lived polymorphism at the RKS1 gene, but still needs to be tested. The efficiency of qualitative resistance rests on R genes and PRRs for which multiple functional alleles, differentiated by single nucleotide polymorphisms (SNPs), and presence–absence polymorphisms are the major patterns of variation. SNPs at R-gene and PRR loci can also introduce quantitative differences in the resistance phenotype. Indeed, some amino acid substitutions alter affinity in protein–protein and protein–ligand interactions. The nature and position of SNPs at a given locus may therefore drive a complete loss of resistance (qualitative resistance) or quantitative variations in the level of resistance (QDR). Presence–absence polymorphism may also condition QDR, such as at the Yr36 locus. Therefore, SNPs and presence–absence are polymorphisms that may govern qualitative or quantitative resistance. Other patterns of genetic variation seem to be typically associated with QDR based on current knowledge. Gene copy number variation (CNV) at the Rhg1 locus, involving multiple genes, conditions QDR against soybean cyst nematode (Cook et al., 2012). Similarly, QDR to rice blast and sheath blight diseases is correlated with the number of germin-like protein genes being silenced in a cluster located at a resistance QTL (Manosalva et al., 2009). Quantitative resistance variation may be associated with CNV in the case of resistance conditioned by limiting stoichiometric reactions, such as enzymatic conversions, where increasing protein amounts impact significantly on the efficiency of the pathway. In addition, gene duplications and cis and trans regulatory mechanisms may introduce quantitative variations in the output of resistance signalling pathways. The relationship between transcript abundance and QDR phenotype has been established recently through genome-wide analyses of transcript abundance and expression QTL (eQTL) mapping (Chen et al., 2010). Correlation between RKS1 expression and QDR phenotype in natural A. thaliana accessions and transgenic lines suggested an important role for RKS1 transcriptional regulation in the mediation of QDR to X. campestris (Huard-Chauveau et al., 2013). Analyses of natural variation in flagellin perception by FLS2 pointed towards regulatory differences underlying the response QTL. Genes underlying flagellin response QTLs may control FLS2 transcript abundance or post-translational modifications of FLS2, such as phosphorylation events (Vetter et al., 2012). The mapping of quantitative resistance loci typically yields multiple QTLs of small to moderate effect, explaining between 5% and 30% of the phenotypic variance. This complexity poses significant challenges for the functional characterization of resistance QTLs. Whether the phenotypic contribution of single QDR genes, underlying QTLs with a limited contribution to phenotype, can be robustly measured is the first part of the challenge. Whether multiple resistance QTLs can be robustly and accurately mapped onto a plant genome is another part. Recent technological progress offers promising perspectives towards accurate phenotypic assessment and mapping of QDR genes, pushing the limits of the smallest QDR contribution that can be measured. The assessment of the phenotypic contribution of genes underlying small-effect QTLs requires robust and accurate methods. Multiple quantitative traits can be measured to characterize QDR in a host–pathogen interaction, such as disease index, lesion size, latency period (filamentous pathogens), number of colony-forming units in planta (bacterial pathogens), enzyme-linked immunosorbent assays (viral pathogens), etc. The choice of a truly quantitative readout, which can be precisely and reliably measured, may be essential for both the accuracy of QTL mapping and the validation of QDR gene function. In addition, the more we dissect the QDR trait, the more direct access to its genetic bases we can expect. Defining the shortest path between the genetic bases of QDR and the trait measured, using molecular or subcellular markers, should magnify the phenotypic variance explained by QDR genes. The simultaneous recording of multiple QDR traits should further increase our ability to accurately quantify the phenotypic contribution of QDR genes. There is therefore a need for a detailed understanding of plant–microbe interactions at the subcellular and molecular level. Understanding how pathogens manipulate host gene expression and host cell architecture would, for instance, provide accurate readouts for QDR assessment. The finding of a reliable causal relationship between loci and QDR may be hindered by the availability of appropriate genetic resources. Alternatives to traditional linkage mapping, generally performed on the progeny of a cross between a susceptible and resistant parent, have emerged recently, including genome-wide association (GWA) mapping, which uses natural plant populations as mapping populations (see Bergelson and Roux, 2010 for details on comparative advantages of traditional QTL mapping and GWA mapping). An illustration of the power of GWA mapping to finely map QDR genes has been shown recently in Arabidopsis. The genetic markers most associated with natural variation of QDR to X. campestris were located in the vicinity or within RKS1, a gene previously cloned through a traditional QTL mapping approach based on a recombinant inbred line population (Huard-Chauveau et al., 2013). Because next-generation sequencing (NGS) technologies have recently provided access to a deluge of genomic data, enabling linkage mapping and GWA in diverse plant species, this NGS revolution should broaden our choice of host species (if the pathogen of interest infects several hosts), mapping population(s) and data analysis strategies available for accurate and reliable QDR gene identification. The use of reverse genetics approaches for testing the function of a candidate gene in QDR may be masked by the genetic background in which mutations and transgenes are introduced. To circumvent this issue, the CRE-Lox recombination system has been used to generate homozygous plant lines with identical genetic backgrounds, differing only with respect to the presence or absence of the RPM1 R gene (Tian et al., 2003). The recent development of genome editing technologies, including zinc-finger, TAL effector and Cas9 RNA-guided nucleases, offers new powerful methods for in situ genome modification, minimizing the effect of genetic background and alterations induced by the insertion of foreign DNA at QDR loci. The growing global demand for a sustainable food supply and the need to reduce reliance on pesticides have driven the search for complementary alternatives to plant disease control. Pathosystems relying on gene-for-gene interactions typically show boom-and-burst evolution cycles, with the emergence of a single major R gene (boom) rapidly broken down by the occurrence of virulent local pathogen populations (burst) (McDonald and Linde, 2002). As a consequence, the need to find alternative sustainable means to reduce crop losses caused by pests and diseases remains pressing. Although the pyramiding of either R or QDR genes has been shown to successfully slow down disease in many crop species, recent studies have demonstrated that pyramiding both R and QDR genes may be even more efficient in the control of disease (Brun et al., 2010). Although R genes may control most avirulent strains present in local pathogen populations, QDR genes may limit selection for virulent isolates by decreasing the effective pathogen population size and therefore the occurrence of mutations that may overcome R genes. However, it is important to remember that defence responses are often costly for plants. Although pyramiding both R and QDR genes appears to be a promising strategy for durable resistance, healthy plants need to keep defence and growth in balance with a minimum impact on other agronomically important crop traits and public acceptance. We note three nonexclusive ways to efficiently decrease the effect of the cost of multi-resistance whilst maintaining sustainable resistance: (i) the pyramiding of R and QDR genes that are expressed in different tissues; (ii) the identification of genes associated with a compensatory improvement of the multi-resistance cost; and (iii) a mosaic strategy spatially alternating R genes and QDR genes within crop fields. The current context of global changes probably favours pathogen emergence and higher/faster migration and adaptation potential in plant pathogens. Native natural plant populations and crops will probably face new aggressive pathogen strains. Because QDR genes often confer broad-spectrum resistance, we predict that they will be critical for the efficient control of epidemics. Determination of the diversity of QDR genes in plant populations thus appears to be an essential prerequisite to estimate the adaptive potential of natural plant populations and crops under potential changes in pathogen pressure as a result of global changes. SR is supported by a Marie Curie CIG grant ('SEPAraTE', contract 334036) and a starting grant of the European Research Council ('VariWhim', contract 336808). The Laboratoire des Interactions Plantes–Microorganismes (LIPM) is supported by the French Laboratory of Excellence project 'TULIP' (ANR-10-LABX-41; ANR-11-IDEX-0002-02).