Innate immunity: has poplar made its BED?


  • Hugo Germain,

    1. Natural Resources Canada, Canadian Forest Service, Laurentian Forestry Centre, 1055 du PEPS, PO Box 10380, Stn Sainte-Foy, Québec, QC, G1V 4C7, Canada
    Search for more papers by this author
  • Armand Séguin

    1. Natural Resources Canada, Canadian Forest Service, Laurentian Forestry Centre, 1055 du PEPS, PO Box 10380, Stn Sainte-Foy, Québec, QC, G1V 4C7, Canada
    Search for more papers by this author

Author for correspondence:
Hugo Germain
Tel: +1 418 648 4925



II.R-gene-mediated response in plants: an introductory overview679
III.R-protein pathways in poplar680
IV.What is the role of BED-NB-LRR?681
V.Downstream of R-proteins683
VI.The role of salicylic acid in biotrophic interaction683
VII.Concluding remarks684


The perennial plant model species Populus trichocarpa has received considerable attention in the last 5 yr because of its potential use as a bioenergy crop. The completion of its genome sequence revealed extensive homologies with the herbaceous annual species Arabidopsis thaliana. This review highlights the similarities and differences at the qualitative defence response components level, notably in putative NBS-LRR protein content and downstream defence regulators. With almost a twofold NBS-LRR gene complement compared with A. thaliana, P. trichocarpa also encodes some putative R-proteins with unusual architectures and possible DNA-binding capacity. P. trichocarpa also possesses all the known main components characteristic of TIR-NB-LRR and CC-NB-LRR signalling. However, very little has been done with regard to the components involved in the poplar qualitative response to pathogens. In addition, the relationship between plant-biotroph perception/signalling and the role of salicylic acid, an important defence compound, remains uncertain. This review aims to identify the genomic components present in poplar that could potentially participate in the qualitative response and highlights where efforts should be devoted to obtain a better understanding of the poplar qualitative defence response.

I. Introduction

The never-ending battle between plants and their pathogens is mediated by a two-layer immune system in which the first component, plasma membrane resident pathogen recognition receptors (PRRs), perceives microbe-associated molecular patterns (MAMPs). Upon MAMP recognition, PRRs will trigger MAMP-triggered immunity (MTI, formerly called basal resistance), a low-amplitude defence response usually sufficient to thwart the pathogen attack. However, pathogens use secreted MTI-inhibiting effectors that may then be recognized (or their resulting effect) by cognate resistance proteins (R-proteins), leading to a powerful defence response termed effector-triggered immunity (ETI, or race-specific resistance). The ETI response is characterized by the production of reactive oxygen species, the accumulation of nitric oxide, the stomatal closure and the activation of MAPK cascades, and it often culminates in the hypersensitive response (Delledonne et al., 1998; Bolwell, 1999; Zhang & Klessig, 2001; Jones & Dangl, 2006; Boller & Felix, 2009). Once an effector is recognized, it reduces the fitness of the carrier strain, and thus other effectors that were not under selective pressure may arise through population evolution as new virulence factors. Conversely, to avoid plant population extinction, one of the various polymorphic resistance loci must recognize these new effectors, which will confer heritable resistance (Jones & Dangl, 2006). This elegant model received broad support by the community and was coined the zigzag model.

Perennial plants are different from annual herbaceous plants at several levels and the question arises whether the findings obtained in Arabidopsis can translate to trees. For example, the zigzag model entails that the plant population can ‘acquire’ genetic loci encoding R-proteins for every new virulence factor that will arise in the pathogen population. This model is conceivable for annual plants that have short generation time, but is it conceptually applicable to trees that have a life span of several decades or even centuries? How could these long-lived plants keep up with the short-generation, fast-evolving pathogens they face? Moreover, for Arabidopsis and most dicotyledonous species, salicylic acid (SA) is believed to be a very important component of the defence response to biotrophic and systemic-acquired resistance (SAR). It is also not clear whether the SA concentration can be modulated in the various clones of hybrid poplar, raising questions about the role of SA and SAR in poplar, and perhaps in trees in general. SAR may seem like a good strategy for plants of modest relative sizes but how could this systemic process in plants operate in plants that can be 5 m tall? In trees the poplar-rust pathogen has recently emerged as a model pathosystem and has been thoroughly studied at the transcriptome level thanks to the availability of microarray chips, leading to a better understanding of the quantitative and qualitative defence responses (Miranda et al., 2007; Rinaldi et al., 2007; Azaiez et al., 2009; Duplessis et al., 2009; Hacquard et al., 2010). This review will try to infer how poplar may respond to pathogens based on the knowledge we have of poplar defence and the knowledge gained from the poplar genome analysis (Tuskan et al., 2006).

II. R-gene-mediated response in plants: an introductory overview

Traditional plant breeders have relied on crosses between cultivars for the introgression of R-gene into susceptible crops. The response resulting from an incompatible interaction between a plant carrying a resistance protein and an avirulence gene product carried by the pathogen was described by Flor (1971) in his gene-for-gene model. In this model, Flor hypothesized that the interaction between a plant that possesses a resistance factor and a pathogen that has a cognate avirulence (Avr) factor would result in a resistance response. If the plant does not possess the R-gene or if the pathogens do not possess the Avr factor, the infection will prevail (Flor, 1971); this model fits with a receptor/ligand model. One important characteristic of poplar is that different poplar species can also be crossed to generate fertile F1 hybrids. This feature greatly increases gene flow in these obligate outcrossers, allowing new combinations of R-gene alleles and enhancing the potential of breeding for resistance (Bradshaw, 1996).

Although known Avrs are generally small molecules and R-proteins have a LRR domain believed to be involved in protein–protein interactions, only a few direct R/AVR interactions have been reported (Jia et al., 2000; Deslandes et al., 2003; Ueda et al., 2006), including Flor’s original flax-rust (or Melampsora lini/flax) model pathosystem (Dodds et al., 2006). A refinement to Flor’s hypothesis was suggested in which the role of the NB-LRR protein would be to guard or monitor the status of a host protein that is the target of an AVR. This refined model was initially used to describe the Prf/Pto/AvrPto interaction (Van der Biezen & Jones, 1998) and was later coined the guard model (Dangl & Jones, 2001). However, this model is also not perfect and another variant has emerged: the decoy model. In this new model, the plant protein targeted by the pathogen effector would have no function in host defence but would mimic a plant defence component (van der Hoorn & Kamoun, 2008). A good example of this would be Pto mimicking a defence component (such as the kinase domain of FLS2) to interact with AvrPto; and that this interaction is monitored by Prf (a NB-LRR), which subsequently triggers defence signalling (Xiang et al., 2008).

Regardless of these hypotheses, R-proteins remain at the centre stage of how plants perceive the pathogen. Most identified plant R-proteins belong to the large group of NB-LRR proteins in which NB is a nucleotide-binding site that is required for ATP binding and hydrolysis (Tameling et al., 2002) and LRR stands for leucine-rich repeat. NB-LRR can be further separated into two distinct groups based on their N-termini (Martin et al., 2003; Belkhadir et al., 2004). Group one includes the toll-interleukin receptor domain (TIR) and the second group has an N-terminal coiled-coil domain (CC). The NB domain also contains motifs that are specific to TNL and CNL (Meyers et al., 2003). A total of 51 CC-NB-LRRs (CNL) and 93 TIR-NB-LRRs (TNL) were found in the genome of A. thaliana ecotype Columbia (Meyers et al., 2003) (Fig. 1). The TIR resistance pathway is mediated by the EDS1/PAD4/SAG101 (enhanced disease susceptibility 1/phytoalexin-deficient 4/senescence-associated genes 101) complexes. EDS1 forms distinct cytosolic and nuclear protein complexes with PAD4 and SAG101 (Feys et al., 2001, 2005). The CC pathway signals through NDR1 (nonrace-specific disease resistance 1), which localizes to the plasma membrane via a C-terminal glycosylphosphatidyl-inositol (GPI) anchor (Century et al., 1997; Aarts et al., 1998; Coppinger et al., 2004). The CC and TIR pathways converge at the synthesis of the defence hormone SA (Fig. 1). Following biotrophic pathogen detection by R-protein, SA accumulates in the infected plants. This pathogen-triggered accumulation is dependent on ISOCHORISMATE SYNTHASE 2 (Wildermuth et al., 2001). SA is a sufficient and necessary signal for SAR (Vernooij et al., 1994), a broad-spectrum and long-lasting systemic resistance (Durrant & Dong, 2004) mediated by the positive regulator NPR1 (nonexpressor of PR-1 genes) (Dong, 2004). The SA-dependent defence signalling pathway is associated with interactions with biotrophic pathogens, while the ethylene and jasmonic acid pathways (ET and JA), which are generally thought to be antagonistic to the SA pathway, are associated with necrotrophic pathogens (Glazebrook, 2005). Recent evidence shows that this antagonistic effect would be mediated, at least in part, by the transcription factor EIN3 (ethylene insensitive 3), which can directly bind to the SID2 promoter (SA synthesis) (Chen et al., 2009). Consistent with these observations, the ein3eil1 double mutant accumulates very high concentrations of SA and the double mutant displays enhanced resistance to virulent and avirulent strains of Pseudomonas syringae (Chen et al., 2009). Recent results also show that the JA pathway can be made insensitive to SA suppression if the ET pathway is induced (Leon-Reyes et al., 2010). Despite a large and old consensus among the community regarding the antagonistic relationship and the selectivity of the JA and SA pathways for necrotrophic or biotrophic pathogens, it was recently demonstrated that the SA pathway can positively contribute to the response to necrotrophic pathogens and that the ET and the JA pathways can also positively contribute to the response to biotrophic pathogens (Tsuda et al., 2009). Using infection and genetic interaction, the Katagiri group made quantitative measurements using combinatorial mutants of the ET, JA and SA pathways to identify the role of the wild-type genes rather than to analyse the effect of the mutant phenotypes (Tsuda et al., 2009). There is now accumulating evidence that, upon activation, R-proteins can go the nucleus themselves (Burch-Smith et al., 2007; Shen et al., 2007; Wirthmueller et al., 2007; Cheng et al., 2009) and the nucleocytoplasmic shuttling of the defence components is highly reminiscent of NF-κB nuclear import (Wiermer et al., 2010).

Figure 1.

 Main actors of the R-protein pathway in Arabidopsis and their homologues in poplar. NDR1, nonrace-specific disease resistance 1; NPR1, nonexpressor of PR-1 genes; EDS1, enhanced disease susceptibility 1; SA, salicylic acid; PAD4, phytoalexin-deficient 4; BNL, BED-NB-LRR; CNL, CC-NB-LRR; TNL, TIR-NB-LRR.

III. R-protein pathways in poplar

Although there has not been any R-AVR combination identified in poplar, there is mounting evidence to suggest that resistance to some pathogens would depend on R-proteins. One of the major threats faced by poplar is the foliar rust caused by the obligate biotrophic fungus Melampsora spp. Responses to Melampsora can be categorized into qualitative or quantitative defence responses. Different physiological races of rust can elicit incompatible or compatible (qualitative) reactions on a given clone of pure poplar species or on hybrid clones. For example, the hybrid Populus deltoides × Populus nigra clone ‘Ogy’ displays an incompatible reaction with Melampsora larici-populina isolates belonging to race E1 but a compatible reaction with race E2 M. larici-populina isolates. The incompatible reaction is characterized by highly localized early collapse and disorganization of the cytoplasm within 2 h after the appearance of the first haustoria (Laurans & Pilate, 1999). This reaction is indicative of R-protein-mediated hypersensitive response. Both types of response have been studied at the transcriptomics level. As in A. thaliana, the compatible and incompatible responses were found to trigger largely overlapping gene sets, including the well-known genes PR1, PR5, PR10 and NPR1 (Rinaldi et al., 2007). More recently, the quantitative defence response, which compares the plant response with the growth of two Melampsora species, was also analysed by microarray (Azaiez et al., 2009). The studied plant responses to a fully compatible Melampsora species with unrestricted growth, and another compatible species that elicits plant defence resulting in restricted growth, revealed gene sets partly overlapping but also specific to both responses (Azaiez et al., 2009).

For a given R/AVR combination leading to qualitative resistance, crosses have shown that resistance was controlled by one single dominant gene or closely located genes. Mendelian segregation of qualitative resistance phenotypes in interhybrid crosses also suggests that resistance is controlled by one single dominant gene or closely located genes (Cervera et al., 1996; Villar et al., 1996; Tabor et al., 2000; Stirling et al., 2001; Zhang et al., 2001; Yin et al., 2004). The first putative R-gene-mediated resistance locus to be fine mapped was MER (Zhang et al., 2001). MER confers resistance to races E1, E2 and E3 of M. laricina-populina and it segregated in a Mendelian fashion when the resistant female parent P. deltoides was used in interhybrid crosses with male P. nigra or P. trichocarpa to generate a segregating mapping population (Zhang et al., 2001). The sequenced amplified fragment length polymorphism (AFLP) markers linked to the MER locus revealed the presence of three NB-LRRs in the region associated with resistance (Zhang et al., 2001); the locus was later mapped to chromosome XIX (Yin et al., 2004). The MXC3 locus conferring resistance Melampsora X columbiana pathotype 3 also segregated in a Mendelian manner in an F1 interspecific hybrid poplar pedigree (P. trichocarpa × P. deltoides), indicating again that the gene is a single dominant gene. Unfortunately, despite the saturation of genetic markers around the MXC3 locus, the gene responsible for resistance could not be identified because of a lack of recombination close to the marker (Stirling et al., 2001) as was previously observed for other R-gene loci (Ganal & Tanksley, 1996; Wei et al., 1999; Behura et al., 2004; Yang & Hua, 2004). MXC3 was later mapped to chromosome IV (Yin et al., 2004). Intriguingly, no NBS-LRR genes were found in the vicinity of MXC3. Two thaumatin-like pathogenesis-related proteins (PR5s), two receptor-like kinases (RLKs), one receptor-like protein (lacking a kinase domain) and one TIR-RLK were found. The four receptor-like genes (excluding the two thaumatins) were located c. 20 cM from the marker linked to the MXC3 locus, making them unlikely candidates. Although thaumatin-like pathogenesis-related (PR5) proteins have been shown to be involved in resistance, they are usually involved in downstream secondary responses and do not act as bona fide resistance genes per se. Gaps still exist in the poplar genome and the MXC3 gene could still be a NB-LRR located in one of those gaps (Yin et al., 2004).

IV. What is the role of BED-NB-LRR?

At the genomic level, poplar possesses a set of R-genes nearly twofold that of Arabidopsis (Kohler et al., 2008; Yang et al., 2008) (Fig. 1). Both grapevine and poplar display a higher degree of recent tandem duplication and gene conversion than Arabidopsis and rice (Yang et al., 2008). This feature could increase overall recombination events and thus lead to a larger number of disease-resistant alleles and curtail the longer generation time of perennial species.

Poplar contains a small subset of atypical putative R-genes that seem to have arisen from domain fusion, such as TIR-NBS-LRR-TIR, TIR-NBS-LRR-NBS and NBS-LRR-TIR, and others also found in the CNL family or a mix of the two families (TCNL) (Kohler et al., 2008; Yang et al., 2008). Such chimeric putative R-proteins were also reported in Arabidopsis (Meyers et al., 2003). The presence of a putative R-gene family in poplar, the BED-NB-LRR family (henceforth called BNL) comprising 32 members, seems to be unique to poplar in the dicots family (Table 1). Blast search does not reveal any hits with a BNL architecture outside of the poplar species in the dicots except for a TIR-NB-LRR-BED-TIR in Vitis vinifera. In monocots, there are eight occurrences of BNL in the rice genome. One of these genes, Xa1, was shown to confer resistance to bacterial blight caused by Xanthomonas oryzae pv. oryzae (Yoshimura et al., 1998). We inferred a phylogenetic tree of all BNLs using the NB-ARC domain (Fig. 2). Poplar BNLs are supported on a different clade from V. vinifera and rice BNL. In addition, and as expected, BNLs on chromosome XIX are more alike than those located on other chromosomes (see grey-shaded box in Fig. 2). Some BNLs, presently positioned on the scaffold, also fall within the grey-shaded box. The discovery of the BED domain was first published in 2000 and was named BED finger after two Drosophila known proteins named BEAF and DREF containing this domain (Aravind, 2000). It is rather surprising that both rice and poplar seem to have independently acquired this gene architecture. The BED domain is a ubiquitous zinc finger DNA-binding domain and its DNA-binding sequence was only recently identified using CHIP sequencing in mouse cells (Markljung et al., 2009). Unfortunately, the DNA-binding sequence of the BED domain is relatively short (8 bp), making its random presence more frequent. We performed an in silico search of a poplar promoter repository using the BED nucleotide-binding sequence to assess whether its presence would be enriched in promoters located upstream of defence-related genes. Perhaps owing to the fact that the BED-binding domain is very short, and therefore not very specific, we observed no correlation between the occurrence of BED-binding sites and a defence function for the genes associated with promoters that contained the BED-binding domain (H. Germain & A. Séguin, pers. obs.). Most BNLs in poplar arose from common ancestors and 27 out of 32 have close homology with At4g27220, four show close homology with At4g10780 and one shows close homology with At4g26090; all are closely located genes on A. thaliana chromosome IV (Table 1 and Supporting Information Table S1) for sequence and alternate gene model). Intriguingly, most BNLs are found in the upper peritelomeric end of poplar chromosome XIX (Fig. 3), a genomic region rich in putative R-genes where MER is located and thought to be an incipient sex chromosome where segregation distortion and suppressed recombination are observed (Yin et al., 2004). Zhang et al. (2001) mapped the MER locus several years before the full genome sequence of poplar became available and yet they found, through cloning and sequencing, that some of their AFLP markers were NB-LRRs encoding genes. We now know that the AFLP marker AF393739 used by Zhang et al. (2001) is in fact Poptr0019s00510, a BNL (Fig. 3). A total of 20 BNLs (out of 32) are associated with the MER locus, making them good candidates for being the MER gene, if they are indeed R-genes (for the precise chromosomal location of these BNLs, see Table S2).

Table 1.   List of all members of the BED-NB-LRR family (BNLs) and the presence of NLS or NES in their amino acid sequence
New annotationArabidopsis thaliana orthologuesNLS prediction PsortNES (number and position of amino acid)
POPTR_0001s41540At4g27220Negative (0.00)1 (686)
POPTR_0001s41570At4g27220Negative (0.00)2 (569, 571)
POPTR_0001s41680At4g10780Negative (0.00)2 (522, 597)
POPTR_0011s12620At4g27220Negative (0.00)5 (430, 433-6)
POPTR_0011s12500At4g27220notclr (0.40)3 (430, 673, 675)
POPTR_0011s12630At4g27220Negative (0.00)2 (969, 971)
POPTR_0019s00410At4g27220notclr (0.30)12 (670-8, 964, 969, 972)
POPTR_0019s00430At4g27220notclr (0.60)5 (661-2, 664, 896, 898)
POPTR_0019s00510At4g27220Positive (0.70)8 (578-584, 940)
POPTR_0019s00540At4g10780Positive (0.70)1 (854)
POPTR_0019s00570At4g27220Negative (0.00)6 (543, 937, 940, 942, 945, 948)
POPTR_0019s00620At4g26090notclr (0.30)4 (498, 500, 599, 601)
POPTR_0019s00700At4g27220Negative (0.00)2 (802, 1367)
POPTR_0019s01010At4g27220notclr (0.60)3 (688, 1282, 1366)
POPTR_0019s01020At4g27220notclr (0.30)4 (1028, 1385, 1387, 1470)
POPTR_0019s01080At4g27220Negative (0.00)9 (488-96)
POPTR_0019s01670At4g27220Negative (0.00)7 (234, 564, 566, 610, 892, 895, 897)
POPTR_0019s02040At4g27220Negative (0.00)1 (499)
POPTR_0019s02060At4g27220Negative (0.00)1 (402)
POPTR_0019s02150At4g10780Positive (0.70)2 (918, 926)
POPTR_0019s02170At4g27220notclr (0.40)2 (568, 840)
POPTR_0019s02180At4g27220Negative (0.00)2 (735, 1006)
POPTR_0019s02200At4g27220notclr (0.60)2 (676, 942)
POPTR_0019s03720At4g27220notclr (0.30)7 (443, 445, 818, 1041, 1043-4, 1090)
POPTR_0031s00350At4g10780Negative (0.00)9 (743-51)
POPTR_0031s00430At4g27220notclr (0.30)1 (333)
POPTR_0060s00250At4g27220notclr (0.30)3 (537-8, 540)
POPTR_0123s00220At4g27220Negative (0.20)2 (827, 850)
POPTR_0190s00200At4g27220notclr (0.30)9 (603-11)
POPTR_0190s00220At4g27220Negative (0.15)3 (567, 572, 1055)
POPTR_0287s00220At4g27220notclr (0.30)2 (806,1049)
fgenesh4_pg.C_LG_XIX000056, new annotation not foundAt4g27220Positive (0.80)1 (631)
Figure 2.

 Phylogenetic tree of all BNLs (BED-NB-LRR) from poplar, rice and one from Vitis vinifera. The tree was inferred using the method of neighbour-joining and used the full NB-ARC domain.

Figure 3.

 Location of BNL (BED-NB-LRR) on the peritelomeric region of chromosome 19 where MER is located.

From a signalling standpoint, the fusion of a DNA-binding domain with a NB-LRR domain is of great interest. Although some R-proteins have recently been shown to localize to the nucleus (Burch-Smith et al., 2007; Shen et al., 2007; Wirthmueller et al., 2007; Cheng et al., 2009), none have been shown to bind DNA or regulate transcription directly. Another alternative is that the BED domain of the putative resistance protein could act as a decoy (van der Hoorn & Kamoun, 2008) for another BED-containing transcriptional regulator, which would be the true target of the virulence factor. Supporting their possible role in the nucleus is the presence of a nuclear localization signal and a nuclear export signal in some of the BNL proteins (Table 1, Fig. S1).

V. Downstream of R-proteins

R-proteins are quite numerous and they converge to major regulatory nodes depending on their respective classifications. TNL signals through the EDS1/PAD4 complex while CNL signals via the NDR1 pathway. According to the new Poplar 2.0 genome annotation on Phytozome (, three copies of EDS1 would be found in poplar with 42.0, 42.1 and 27.2% identity, along with two copies of PAD4 having 48.4 and 46.4% identity with their Arabidopsis orthologues. In the CNL pathway, two copies of NDR1 are found, having 45.7 and 46.9% identity with their Arabidopsis counterparts. Although poplar seems to possess EDS1/PAD4 and NDR1, a role in defence has not yet been demonstrated. Whether the BNLs are functional R-proteins in poplar and whether they signal through the EDS1/PAD4 node or via the NDR1 node has not yet been demonstrated. Based on our observations, a hypothesis with regard to BED-NB-LRR signalling can be made. First is the presence and positive R-gene-like activity of a BNL in rice, even though rice lacks TNL. Second, all BNLs relate strongly to only three Arabidopsis NB-LRRs and all three are CNLs. Additionally, previous work by Meyers et al. (2003) identified different patterns in the NBS domain, namely the RNBS-A, RNBS-C and RNBS-D motifs that were different between the CNLs and TNLs (Meyers et al., 2003). Kohler et al. (2008) did a similar analysis using the poplar NB-LRR proteins and we find there is slightly more homology between BNL and CNL than with TNL, particularly in the RNBS-D motif (see the CAI/LFPxD section) and the RNBS-A motif (in the WxxVSQDFxxxxxEEL section) (Fig. S2). Based on these observations, BNL signalling via the NDR1 node or via a novel pathway would be more likely than via the EDS1/PAD4 node.

VI. The role of salicylic acid in biotrophic interaction

In Arabidopsis, race-specific pathogen recognition usually leads to localized HR and results in SAR in uninfected parts of the plants. The SAR response is dependent on the protein NPR1 and on the accumulation of SA, and leads to an increase in PR gene expression. In addition, NPR1 is also involved in the negative regulation of the JA pathway (for a review on NPR1, see Dong, 2004). In rice, which has a very high endogenous SA concentration, initial work has shown that SA-depleted plants were not more sensitive to pathogen infection than wild-type plants (Yang et al., 2004). Recent results showed that WRKY gene OsWRKY33 overexpression can trigger PR gene expression and increase SA accumulation, which leads to HR-like cell death (Koo et al., 2009), indicating that SA may have a role in the rice defence mechanism. Additionally, following probenazole treatment (a SAR inducer) in rice, free SA concentration increase is observed as well as OsPR1 transcript abundance, resulting in plants that are more resistant to infection by Magnaporthe grisea (Iwai et al., 2007).

The role of SA in poplar, or more generally in tree defence, has not been clarified, and whether the concentration of SA can be modulated in poplar remains controversial. Results by Morse et al. (2007) demonstrate that transgenic poplars overexpressing the widely used bacterial salicylate hydrolase NahG had unchanged free SA and catechol concentrations. However, glycosyl-conjugated SA was reduced by > 90% in two transgenic lines and there was a fivefold increase in glycosyl-conjugated catechol (downstream metabolite of SA degradation), indicating that free SA concentration is tightly regulated (Morse et al., 2007). Diara et al. (2005) used the ozone (O3)-induced hypersensitive response to assess whether SA, 1-aminocylcyclopropane-1-carboxylic acid (ACC, precursor of ethylene synthesis), ethylene and peroxide had similar kinetics and scale of production in the ozone-sensitive clone Eridano (P. deltoides×P. maximowizii) and the O3-resistant clone I-214 (P. deltoides×P. euramericana). In their treatment, a modest but significant increase in SA was observed in the sensitive clone (Eridano) while the I-214 clone, which had a much higher basal SA concentration, remained unchanged (Diara et al., 2005). The varying kinetics and magnitudes observed in specific hybrid to HR-inducing condition argues against a general and conserved role of SA in poplar defence response. The rapid up-regulation of poplar genes involved in JA and ET biosynthesis, such as allene oxide synthase (AOS) and ACC synthase, following Melampsora sp. infection, supports the positive contribution of JA and ET in response to biotrophic pathogens in trees (Azaiez et al., 2009). In glasshouse trials we observed that the exogenous application of isonicotinic acid (a SAR inducer) could trigger PR-gene expression and restrict the growth of the fully compatible fungal pathogen M. larici-populina in hybrid clones of P. tricocarpa×P. deltoides 3225 juveniles. These results suggest that young hybrids rely on SA for pathogen resistance. Whether this role is maintained as the tree ages remains to be clarified.

VII. Concluding remarks

Poplar and Arabidopsis both belong to the Eurosid clade of dicots and are thus relatively close cousins in comparison with species such as cereal crops (monocotyledon) or conifer trees (gymnosperm). This phylogenetic relationship should sustain assumptions made from knowledge acquired in Arabidopsis pathosystems and thus circumvent the lack of data for poplar. Little is known about the active players involved in the poplar qualitative defence response, but solely based on genomic comparison with Arabidopsis, it seems that poplar has all the proper tools to launch an R-gene-mediated response. Poplar also has unique features, such as the BNL family, for which a role in tree defence is yet to be demonstrated. To account for the disadvantage of its long juvenile stage, which gives poplar (and trees in general) less generation per pathogen generation to evolve new R alleles, poplar has a higher degree of tandem duplication and gene conversion than Arabidopsis, and is an obligate outcrosser, which promotes genetic exchange and heterozygosity (Ingvarson, 2010), in contrast to A. thaliana, which is almost an obligate inbreeder and a highly homozygous species. The divergence with Arabidopsis probably occurs downstream of proteinaceous defence regulators at the hormonal level. There is a well-established consensus among plant pathologists that the plant response to biotrophic pathogens is mediated by SA, while the response to necrotrophic pathogens is mediated by JA and that both pathways are antagonistic (Koornneef & Pieterse, 2008). Only recently have we seen evidence that JA could have a positive output on biotrophic response and SA on necrotroph response (Tsuda et al., 2009). The extent of this contribution may be different in trees and Arabidopsis. For instance, auxin, gibberellic acid and abscisic acid also influence plant–pathogen interactions (Grant & Jones, 2009), but their contribution to the poplar defence response has not been investigated. Nonetheless, the fact that the poplar SA concentration appears to respond so differently to stimulus between hybrid clones and the fact that some hybrid clones appear insensitive to SA-stimulating conditions points to the fact that SA may not be the only hormone leading the response to biotrophic pathogens or that it is mediated by SA derivatives to keep the free SA pool constant for developmental purposes or other reasons.

The lack of a method to transiently assess gene functions in poplar (technical limitation) seriously impedes the investigation of gene functions and defence regulators. Transient gene silencing and gene overexpression technologies must be adapted to poplar to quickly move forward the functional analysis of genes in poplar. The successful complete sequencing of the poplar mosaic virus genome (PopMV) by Dr Malcolm Campbell’s group (Smith & Campbell, 2004) could potentially lead to the use of virus-induced gene silencing in poplar. However, since PopMV virulence varies greatly between hybrids, other less species-specific viral vectors should also be assessed. Because poplar is an obligate outcrosser, the T-DNA insertion strategy used to generate homozygous knock-out, such as the one so widely used in Arabidopsis, is irrelevant in this species. However, the gain of function could be investigated in activation-tagging poplar lines such as the 1800 independent activation-tagged lines produced and are available through collaboration with the Regan laboratory at Queen’s University (Harrison et al., 2007).

The availability of transcriptome analysis for poplar is of limited help when it comes to finding defence regulators. Unlike the downstream components of signalling cascades, the regulation of R-genes at the transcriptional level is usually relatively limited. Despite time and labour requirements, generation of stable transgenic lines remains the best approach to perform functional analyses in poplar. Complementation of Arabidopsis knock-out mutants with poplar genes is another strategy that needs to be implemented to confirm poplar protein function. Although this method has its limitations and does not allow the discovery of new processes or poplar-specific processes, it could help to confirm protein activity or function. Another aspect essential to a better understanding of poplar defence would be the availability of cloned Avr and a delivery system that could be used to dissect the poplar ETI pathway and analyse whether all the downstream regulators of R-gene signalling are redundant. At the present stage, functional research on poplar qualitative defence response still faces many hurdles. Overcoming these technical limitations would greatly help poplar to become a better model organism.