First published online March 30, 2007; 10.1104/pp.106.094987
Plant Physiology 144:347-366 (2007)
© 2007 American Society of Plant Biologists
PLANTS INTERACTING WITH OTHER ORGANISMS
Transcript Profiling of Poplar Leaves upon Infection with Compatible and Incompatible Strains of the Foliar Rust Melampsora larici-populina1,[W]
Cécile Rinaldi2,
Annegret Kohler2,
Pascal Frey,
Frédéric Duchaussoy,
Nathalie Ningre,
Arnaud Couloux,
Patrick Wincker,
Didier Le Thiec,
Silvia Fluch,
Francis Martin and
Sébastien Duplessis*
Unité Mixte de Recherche (UMR) 1136 Institut National de la Recherche Agronomique (INRA)/Nancy Université Interactions Arbres/Micro-organismes (C.R., A.K., P.F., F.D., F.M., S.D.), and UMR 1137 INRA/Nancy Université Ecophysiologie et Ecologie Forestières (N.N., D.L.T.), Centre INRA de Nancy, F–54280 Champenoux, France; Génoscope Centre National de la Recherche Scientifique UMR 8030, Centre National de Séquençage, F–91057 Evry cedex, France (A.C., P.W.); and Platform for Integrated Clone Management, Austrian Research Centers Seibersdorf Research Biogenetics, A–2444 Seibersdorf, Austria (S.F.)
 |
ABSTRACT
|
|---|
To understand key processes governing defense mechanisms in poplar (Populus spp.) upon infection with the rust fungus Melampsora larici-populina, we used combined histological and molecular techniques to describe the infection of Populus trichocarpa x Populus deltoides ‘Beaupré’ leaves by compatible and incompatible fungal strains. Striking differences in host-tissue infection were observed after 48-h postinoculation (hpi) between compatible and incompatible interactions. No reactive oxygen species production could be detected at infection sites, while a strong accumulation of monolignols occurred in the incompatible interaction after 48 hpi, indicating a late plant response once the fungus already penetrated host cells to form haustorial infection structures. P. trichocarpa whole-genome expression oligoarrays and sequencing of cDNAs were used to determine changes in gene expression in both interactions at 48 hpi. Temporal expression profiling of infection-regulated transcripts was further compared by cDNA arrays and reverse transcription-quantitative polymerase chain reaction. Among 1,730 significantly differentially expressed transcripts in the incompatible interaction, 150 showed an increase in concentration  3-fold, whereas 62 were decreased by 3-fold. Regulated transcripts corresponded to known genes targeted by R genes in plant pathosystems, such as inositol-3-P synthase, glutathione S-transferases, and pathogenesis-related proteins. However, the transcript showing the highest rust-induced up-regulation encodes a putative secreted protein with no known function. In contrast, only a few transcripts showed an altered expression in the compatible interaction, suggesting a delay in defense response between incompatible and compatible interactions in poplar. This comprehensive analysis of early molecular responses of poplar to M. larici-populina infection identified key genes that likely contain the fungus proliferation in planta.
Plants respond to microbial invasion by activating an array of inducible defense mechanisms (Nimchuk et al., 2003 ). After specific recognition of a pathogen, the hypersensitive response (HR) is a rapid and efficient plant resistance mechanism leading to cell death at the site of infection (Heath, 2000). Among the rapid defense mechanisms triggered in plant tissues are generation of reactive oxygen species (ROS) at the site of infection, cell wall thickening, and production of anti-microbial compounds and enzyme inhibitors (Glazebrook, 2005). Genes encoding pathogenesis-related (PR) proteins such as glucanases and chitinases are primary target genes triggered during the early response to pathogen attack (Van Loon and Van Strien, 1999) and are considered as a signature of the HR. Their expression could be directly targeted by pathogen-sensing systems through highly complex and interconnected networks of transduction pathways driving plant resistance (Katagiri, 2004). Several signal molecules, such as ethylene, salicylic acid (SA), or jasmonic acid, play an important role in these defense reaction-signaling networks (Shah, 2003), and recently a central role for the NONEXPRESSOR OF PR GENES1 (NPR1) protein has been highlighted (Dong, 2004). Considerable advances have been made in understanding plant resistance processes in the model plant Arabidopsis (Arabidopsis thaliana), particularly through extensive mutant analyses. It is now clear that the plant response to pathogen infection is associated with massive changes in gene expression (Schenk et al., 2000). Large-scale mRNA expression profilings have revealed that plants express similar sets of defense mechanisms in response to different pathogens (Tao et al., 2003; Eulgem, 2004). Dissection of the defense response at the molecular level has greatly helped in drawing a general model for pathogen resistance in plants (Nimchuk et al., 2003; Tao et al., 2003). However, it is not known whether such a model applies to long-term adaptation of resistance mechanisms in perennial plant species like trees. Such long-lived species are more prone to attacks by pathogens before reproduction, and their long generation time makes it impossible for them to match the evolutionary rates of a pathogen that goes through several generations every year. Annotation of the Populus trichocarpa genome has revealed an expansion of the NBS-Leu-rich repeat (LRR) resistance gene families (Tuskan et al., 2006), suggesting a possible adaptation to long-term exposure to pathogens.
The basidiomycete Melampsora larici-populina is responsible for the leaf rust disease in Populus species (Frey et al., 2005; Pinon and Frey, 2005). Urediniospore germlings of this obligate biotrophic fungus usually penetrate the host plant through stomatal openings, differentiate a series of infection structures in the intercellular space, and exhibit highly localized penetration of the host cell wall to establish a haustorium (Laurans and Pilate, 1999). Hyphae then proliferate in the leaf parenchyma and produce golden pustules filled with masses of urediniospores on the lower leaf surface. M. larici-populina causes severe economic losses in European poplar plantations and has been recently detected in North America (Newcombe and Chastagner, 1993). Selection for resistance to this biotrophic pathogen is thus an important challenge for poplar breeders (Dowkiw and Bastien, 2004). Severe damage occurs through decreased photosynthesis efficiency, early defoliation, and increased susceptibility to other pests and diseases (Gérard et al., 2006). To date, no resistant poplar cultivars are available, as new virulent strains of M. larici-populina are developing regularly. Sustainability of newly selected resistance requires a better understanding of the molecular mechanisms underlying Populus-Melampsora interactions. Most of the knowledge on lifestyle of fungal pathogens derives from studies in nonobligate biotrophs, and there are limited data about obligate biotrophs, such as rust fungi and powdery mildews. These fungi show a high level of compatibility with their host, but the reasons for the obligate biotroph status remains unknown (Mendgen and Hahn, 2002). To date, only sparse molecular data were obtained on defense mechanisms in poplar, and published data concern interactions with herbivores or viruses (Christopher et al., 2004; Smith et al., 2004; Major and Constabel, 2006; Ralph et al., 2006a). Molecular responses to Melampsora spp. invasion are unknown.
Here, we investigated the interaction between an interamerican hybrid poplar, P. trichocarpa x Populus deltoides ‘Beaupré’ and M. larici-populina. This poplar hybrid has been largely used in commercial poplar cultivation in Europe and harbors a qualitative resistance to M. larici-populina (Pinon and Frey, 2005). It is derived from a cross between P. trichocarpa, a species for which the genome sequence is available (Tuskan et al., 2006), and P. deltoides, a species from which rust resistance loci are inherited (Jorge et al., 2005). Efforts had been made to localize rust resistance-related genes in poplar pedigrees through genetic approaches (Cervera et al., 2004). Different studies successfully mapped rust-resistance quantitative trait loci in P. trichocarpa and P. deltoides (Lescot et al., 2004; Yin et al., 2004). To characterize the specific host-response to either virulent or avirulent strains of M. larici-populina, we carried out a combined molecular and histological analysis of time-course infection through scanning electron microscopy and quantitative PCR (qPCR) to detect rust progression in planta. We also investigated lignin and ROS production in plant tissue to determine major differences in plant response between the compatible and incompatible interactions. Then, rust-responsive genes were identified using either a suppression subtractive hybridization (SSH)-cDNA library of poplar leaves infected by the avirulent M. larici-populina strain or NimbleGen whole-genome expression arrays. Finally, expression profiles of Populus defense-related genes during the time course of infection were confirmed with additional transcriptome-based approaches (reverse transcription [RT]-qPCR and cDNA arrays).
|
RESULTS
|
|---|
Time Course of Compatible and Incompatible Interactions
Rust development in P. trichocarpa x P. deltoides ‘Beaupré’ leaves was monitored at the macroscopic and microscopic levels over a period of 10 d postinoculation (dpi) with either compatible (98AG31; pathotype 3-4-7) or incompatible (93ID6; pathotype 3-4) strains of M. larici-populina. In the compatible interaction, uredinia formation was visible under the abaxial epidermis 5 dpi, and by 6 to 7 dpi, uredinia emerged through the epidermis and formed orange pustules of 1- to 2-mm diameter (Fig. 1
). Uredinia distribution was uniform on the leaf surface, and there were about 73 ± 6 pustules/cm2. In the incompatible interaction (93ID6), no lesion or pustule was observed on the leaf surface over a period of 10 d. The abaxial epidermis showed very localized necrotic zones, and dark dots inside mesophyll tissues were visible in transparency after 6 dpi (Fig. 1). Control leaves inoculated with water showed no pustules or necrotic lesions after 10 d.
View larger version (42K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Figure 1. Experimental design of time-course infection of detached leaves of P. trichocarpa x P. deltoides ‘Beaupré’ mock inoculated with water or inoculated with incompatible (avirulent, 93ID6) and compatible (virulent, 98AG31) strains of M. larici-populina. RNA was sampled in independent replicate experiments at 12, 24, and 48 hpi and were used for transcript profiling. Leaf tissues were also sampled at 2, 6, 12, 18, 24, 48, and 96 hpi and were then used for observation in the VPSEM, detection of ROS and lignin synthesis, and qPCR detection of pathogen growth in planta. Expression of disease (uredinia formation on abaxial epidermis) or resistance (localized hypersensitive reaction) was observed at 7 dpi. Necrotic lesions or uredinia on abaxial epidermis of foliar discs are shown at 7 dpi for the incompatible or the compatible interaction, respectively.
|
|
Infection structures developed by M. larici-populina at the leaf surface were monitored by aniline blue staining and light microscopy. Compatible and incompatible spores of M. larici-populina germinated within 2 h postinoculation (hpi) and germ tubes of different length (1 µm–1 mm) were observed on the leaf surface (Fig. 2A
). Most of the germ tubes ramified, formed appressoria, and had successfully penetrated plant tissues at 2 hpi. Low-temperature variable pressure scanning electron microscopy (VPSEM) was used to follow in planta colonization of M. larici-populina by direct observations of infected leaf sections. Fungal structures on the leaf surface were similar to those observed with aniline blue staining for both interactions. Penetration through stomata occurred for about one-half of the successful events of germination between 2 and 6 hpi (Fig. 2B). At 6 and 12 hpi, substomatal vesicles were observed in the substomatal chambers (Fig. 2C), and infection hyphae were developing in the mesophyll tissue from these vesicles. At 18 and 24 hpi, infection hyphae extended into the mesophyll and in some cases reached the palisadic mesophyll (Fig. 2, D and E). The infection hyphae terminated their growth on a mesophyll cell forming haustorial mother cells, while other infection hyphae continued their course into the mesophyll after branching (Fig. 2, D and E). Infection structures inside the cells (i.e. haustoria) cannot be observed using VPSEM. Most of the compatible strain hyphae observed by VPSEM invaded both the spongy and the palisadic mesophyll by 48 hpi. The number of infection hyphae dramatically increased for the compatible strain at 96 hpi (Fig. 2F). Observations were further made at 5 and 7 dpi. In the compatible interaction, hyphae totally invaded the plant tissue around primary infection sites at 5 dpi, domes were formed in the spongy mesophyll, and spore-forming cells were differentiating (data not shown). By 7 dpi, domes corresponding to uredinial pustules released spores on the leaf abaxial surface (Fig. 2G). In the case of the incompatible interaction, the number of hyphae observed in planta was similar to that of the compatible interaction until 24 hpi, and a limited number of hyphae was observed at later time points. At 96 hpi, a few hyphae were ramified or extended into the palissadic mesophyll (Fig. 2H).
View larger version (77K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Figure 2. Development of infection structures of compatible (A–G) strain 98AG31 and incompatible strain 93ID6 (H) of M. larici-populina during time-course infection of leaves of P. trichocarpa x P. deltoides ‘Beaupré’. A, Aniline blue-stained urediniospores producing several ramified germ tubes at vicinity of stomata and hyphal penetration through stomata without appressorium formation at 2 hpi. B, Urediniospores producing germ tubes and appressoria at vicinity of stomata and primary infection hyphae penetrating through stomata at 6 hpi. C, Substomatal vesicle formed under the abaxial epidermis and infection hyphae developing in the spongy mesophyll at 12 hpi. D, Infection hyphae colonizing the spongy mesophyll at 48 hpi. E, Infection hyphae developing from the spongy mesophyll to the palisade mesophyll at 48 hpi. F, Spongy and palisade mesophyll colonized by infection hyphae at 96 hpi. G, Uredinium formed on the abaxial epidermis releasing newly formed urediniospores at 196 hpi. H, Limited development of hyphae of the incompatible M. larici-populina strain 93ID6 in the mesophyll at 96 hpi compared to compatible strain in F. The hyphae of M. larici-populina were painted in red in F and H to help with visualization of fungal hyphae in plant mesophyll. Fungal infection structures are pinpointed by arrowheads. ap, Appressorium; gt, germ tube; ih, infection hyphae; inf, infection site; nf sp, newly formed spore; pih, primary infection hyphae; sp, spore; ssv, substomatal vesicle; ur, uredinium.
|
|
Fungal Growth in Leaves during Compatible and Incompatible Interactions
Assuming that the proportion of fungal and plant biomass present at any given time during an infection is equivalent to the proportion of fungal and plant DNA, quantitation of fungal nuclear ribosomal DNA (rDNA) internal transcribed spacer (ITS) can be used to estimate the extent of fungal growth in the plant (Boyle et al., 2005). Invasion of foliar tissues by compatible and incompatible strains was monitored in planta by qPCR quantification of M. larici-populina rDNA ITS. In planta growth of compatible and incompatible strains of M. larici-populina was similar during early stages of infection (2, 6, 12, and 24 hpi; Fig. 3
). A drastic increase in fungal DNA mass of about 12-fold was observed between 24 and 48 hpi for the compatible strain, while the incompatible strain showed a slower increase (4-fold). The amount of fungal rDNA ITS decreased for the latter strain between 48 and 96 hpi, indicating a possible hyphal decay (Fig. 3). In contrast, growth of the compatible strain dramatically increased at 96 hpi, and DNA mass was more than 500-fold higher than the amount measured for the incompatible strain.
View larger version (12K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Figure 3. Time-course infection of leaves of P. trichocarpa x P. deltoides ‘Beaupré’ by compatible (98AG31) and incompatible (93ID6) strains of M. larici-populina. Development of the two rust strains was followed in planta by specific amplification of the rDNA intergenic ITS region from total DNA extracted from inoculated leaf tissues at 2, 6, 12, 24, 48, and 96 hpi. Pathogen growth curves correspond to  Ct of fungal ITS amplicons measured by quantitative PCR. Note the log scale. Estimates of fungal mass DNA are indicated on the graph for the two strains at 12, 24, 48, and 96 hpi. Amplifications were carried out on three biological replicates, and significant differences between the two M. larici-populina strains are indicated by a star (t test; P < 0.05).
|
|
Detection of ROS and Lignin Monomers
Leaf tissues inoculated with compatible and incompatible strains of M. larici-populina showed no endogenous ROS accumulation based on diaminobenzidine (DAB) staining (data not shown). In contrast, control leaves with hydrogen peroxide (H2O2) injection and wounded leaves exhibited DAB precipitates (data not shown). Phloroglucinol staining is considered to be specific for cinnamaldehyde end groups present in lignins (Nakano and Meshitsuka, 1992). At 2, 6, 12 (data not shown), 24, and 48 hpi, a light red coloration of vessels of major orders was observable on leaf inoculated with the incompatible strain, and no coloration was visible for the rest of the leaf tissues (Fig. 4A
). An intense red coloration of leaf tissues was detected at 96 hpi in leaves inoculated with the incompatible strain (Fig. 4, A and B), whereas staining was restricted to lower orders of leaf vessels in the case of the compatible interaction or in control leaves (Fig. 4A). Infection hyphae of the compatible strain could be observed as yellowish dots in the mesophyll tissue at 96 hpi (Fig. 4B). In control leaf tissues, a light red coloration was restricted to vessels.
View larger version (72K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Figure 4. Wiesner coloration revealing monolignol accumulation by a red coloration in leaf discs of P. trichocarpa x P. deltoides ‘Beaupré’ sampled on detached leaves inoculated with compatible (98AG31) or incompatible (93ID6) strains of M. larici-populina or mock inoculated with water. A, Cleared leaf discs at 24, 48, and 96 hpi in compatible and incompatible interactions and in the control treatment. B, A 10x close-up of leaf discs at 96 hpi. M. larici-populina infecting hyphae are visible by transparency in the case of the compatible interaction.
|
|
Defense-Related Genes in a SSH cDNA Library of Rust-Infected Leaves
SSH technology is a powerful approach to identify genes differentially expressed by cells or organisms in specific developmental stages (Diatchenko et al., 1996). A subtractive cDNA library from incompatible rust-infected leaves (12, 24, and 48 hpi), subtracted by RNA from noninoculated leaf tissues, was constructed. Assembly of 1,999 ESTs obtained from the 5' and 3' ends of 1,152 SSH cDNAs produced 967 nonredundant tentative consensus (TC) sequences corresponding to 486 different genes. These sequences have been deposited in the National Center for Biotechnology Information (NCBI) database (accession nos. CT027996–CT029994; CT033829). Among them, 357 ESTs (37%) corresponded to singletons, 610 ESTs (63%) were clustered in 129 TC, and approximately 90% of the ESTs were supported on both 5' and 3' ends. A BLASTN search showed that 467 of these ESTs have a homolog in the P. trichocarpa genome sequence in the Joint Genome Institute (JGI) database. The remaining TCs showed no significant similarity to any genes in the NCBI databases, suggesting that they are either absent from the current P. trichocarpa genome sequence assembly (version 1.1) or they might be specific to P. trichocarpa x P. deltoides ‘Beaupré’. The sequences showed no significant hits against M. larici-populina and other basidiomycetes genomes on the JGI portal (data not shown).
A summary of homology searches against GenBank using the BLAST algorithm is shown in Supplemental Table S1. ESTs from Rubisco and chlorophyll a/b-binding protein (CAB) genes were not identified in the SSH cDNA library, whereas they represent 24% of a ‘Beaupré’ cDNA leaf library (Kohler et al., 2003), indicating that the SSH library was efficiently depleted from constitutively expressed transcripts. This analysis indicated that the most prevalent ESTs represented genes directly connected with plant defense and nitrogen metabolism. Among these abundant ESTs, an EST matching an inositol-3-P synthase (I3PS) gene was the most frequently detected sequence (199 occurrences; 20% of ESTs; Table I
). Several transcripts corresponding to defense-related genes (PR-1, PR-2, PR-3, and PR-5) and pathogen perception (receptor-like kinase [RLK], LRR, and ankyrin protein) were also among the most abundant SSH ESTs. For example, PR-1 transcripts represented about 4% of the SSH transcripts, indicating a striking up-regulation of its expression. A least three distinct types of chitinase-like genes, including basic (PR-8) and acidic (PR-3) chitinases, were expressed in rust-infected leaves. Transcripts coding for Asp aminotransferase, Asn synthetase, and NADH-Glu synthase, with no obvious direct defensive roles, were also abundant in the SSH library.
View this table:
[in this window]
[in a new window]
|
Table I. Most prevalent rust-responsive transcripts in leaves of P. trichocarpa x P. deltoides ‘Beaupré’ inoculated with the incompatible strain 93ID6 of M. larici-populina measured by ESTs redundancy in a SSH-enriched cDNA library
nd, Not detected.
|
|
Identification of Rust-Responsive Genes at 48 hpi Using Whole-Genome Oligoarrays
Microscopy observations (Fig. 2) and qPCR measurement of fungal DNA (Fig. 3) showed a shift in fungal progression between compatible and incompatible interactions at 48 hpi. A strong difference in lignin deposition at infection sites was observed in the case of the incompatible interaction at 96 hpi (Fig. 4), suggesting that the host molecular response probably initiated after the fungus entered into the mesophyll (12 hpi; Fig. 2, D and E) and when haustorial infection structures were differentiating. We thus investigated changes in gene expression in ‘Beaupré’ leaves at 48 hpi in compatible and incompatible interactions (Fig. 1) using the NimbleGen Populus whole-genome expression oligoarray (Tuskan et al., 2006). We identified 280 (0.4%) and 1,730 (2.6%) transcripts differentially accumulated (2-fold) in the compatible and incompatible interactions, respectively, compared to control leaves mock inoculated with water (Table II
; Supplemental Table S2). Among these transcripts, 150 showed an increased concentration 3-fold in the incompatible interaction and only seven in the compatible interaction.
View this table:
[in this window]
[in a new window]
|
Table II. Expression ratios of rust-responsive transcripts from P. trichocarpa x P. deltoides ‘Beaupré’ measured with the NimbleGen P. trichocarpa whole-genome expression oligoarray at 48 hpi in ‘Beaupré’ leaves inoculated with incompatible (I48) or compatible (C48) strains of M. larici-populina
|
|
Incompatible Interaction
The transcript showing the highest rust-induced accumulation (32-fold) in the incompatible interaction corresponded to a P. trichocarpa gene model (protein identification [ID] no. 678883) with no sequence similarity in the nonredundant NCBI database or the Arabidopsis genome. The corresponding genomic sequence is located at the beginning of scaffold 5,059 of the P. trichocarpa genome assembly and is truncated in 5'. This transcript showed a strong homology with several Populus ESTs in NCBI dbEST. These ESTs are longer in their 5' and 3' ends of nucleotide sequences and encode an 82-amino acid polypeptide with no ortholog in databases. SignalP 3.0 (http://www.cbs.dtu.dk/services/SignalP/) analysis identified a signal peptide of 24-amino acid length, and Phobius (http://phobius.cgb.ki.se/) predicted a noncytoplasmic localization for the protein, suggesting a putative localization outside of the plant cell.
Genes encoding enzymes known to be associated with the host defense response were highly induced at 48 hpi in the incompatible interaction and presented no significant regulation in the compatible interaction. These genes encode several types of PR proteins, such as PR-1 homologs, basic glucan-endo-1,3- -glucosidase (PR-2), thaumatin-like, and osmotin-like proteins (PR-5), which were also identified in the SSH cDNA library and PR-10-like proteins. All those PR transcripts showed an increase 3-fold compared to the mock-inoculated treatment.
Among other transcripts showing an important induction in the incompatible interaction (5-fold accumulation), we detected transcripts coding for components of the signaling pathways (calmodulins), an RLK, and the I3PS identified in the SSH library. Transcripts related to secondary metabolism and cell wall synthesis, such as dirigent-like proteins, chalcone-, flavonol-, and tropinone reductases, were also detected among transcripts significantly accumulated in the incompatible interaction supporting the observed accumulation of phloroglucinol-stained lignin monomers (Fig. 4A).
Several rust-induced genes corresponded to different members of the glutathione S-transferase (GST) gene family. Twelve different GST transcripts showed at least a 2-fold accumulation in the incompatible interaction compared to the mock-inoculated treatment. The most strongly accumulated GST transcripts consisted of two phylogenetic groups of sequences (group 1, protein ID nos. 276538, 277644, and 272826; and group 2, protein ID nos. 657351 and 820835; Table II) that shared a high sequence similarity (90%). Alignment of the oligonucleotide probe sequences matching these GST sequences revealed a significant overlap that may have resulted in possible cross hybridizations between transcript species. Nevertheless, all transcripts that were strongly accumulated belonged to the tau GST class and not to other classes (i.e. phi, theta, and zeta GSTs) described so far in plants (Dixon et al., 2002; Wagner et al., 2002).
There were a few transcripts (292) showing a decrease in concentration at 48 hpi in the incompatible interaction. These include several chloroplastic transcripts, as well as different transcripts coding transposable elements, including the magali Spm-like transposable elements (protein ID no. 418172) located within a rust-resistance locus in P. deltoides (Lescot et al., 2004). A transcript encoding a Lys decarboxylase also showed a dramatic decrease in concentration at 48 hpi (8-fold). Most of these transcripts also presented a decrease in concentration in the compatible interaction (Table II).
Compatible Interaction
Only a few transcripts were significantly up- or down-regulated in the compatible interaction (158 and 122, respectively). The highest increase (4.2-fold) corresponded to a transcript coding for an anthocyanin acyltransferase-like protein. Several transcripts corresponding to signaling components (Ser/Thr kinase, calcium-binding protein) and LRR-containing proteins, including an RLK, were significantly induced (2.5-fold). Interestingly, a transcript coding a 1-pyrroline-5-carboxylase (protein ID no. 421059) showed a 2.5-fold induction. This protein is involved in Pro biosynthesis, and a regulation of transcripts involved in the catabolism of Pro was previously described specifically in the flax (Linum usitatissimum)-Melampsora lini compatible interaction (Ayliffe et al., 2002).
Identification of Rust-Responsive Genes at 48 hpi Using cDNA Microarrays
Before the availability of the NimbleGen whole-genome expression oligoarrays, we carried out a series of transcript profilings of rust-infected (incompatible interaction) and mock-inoculated (control) P. trichocarpa x P. deltoides ‘Beaupré’ leaves at 48 hpi using 28 K Platform for Integrated Clone Management (PICME) cDNA microarrays. Thus, these datasets obtained on different plants (year 2004) than those for oligoarray-based expression profiling (year 2005) were compared to the oligoarray expression profiles.
We identified 1,614 transcripts corresponding to 1,055 P. trichocarpa gene models that were significantly accumulated or decreased (2-fold) at 48 hpi in the incompatible interaction compared to control leaves. Among these transcripts, 218 showed an increased concentration 3-fold and 136 a decreased concentration  3-fold. Transcripts accumulated in response to the rust infection are mostly identical to those described in the oligoarray expression profiling (Table III
; Supplemental Table S3). The highest levels of rust induction (8-fold) were observed for transcripts coding PR proteins, such as PR-1, PR-2, PR-5, PR-8, and PR-10, and the rust-induced secreted protein (RISP) number 678883. GST transcript corresponding to protein ID number 820835 showed a 5.9-fold induction. In addition, two transcripts coding GSTs not significantly regulated in the whole-genome oligoarray analysis and belonging to the same rust-induced tau GST class (see above) were detected as significantly induced on PICME microarrays (protein ID no. 670248, 9.4-fold; protein ID no. 658124, 6-fold).
View this table:
[in this window]
[in a new window]
|
Table III. Expression ratios of rust-responsive transcripts in leaves of P. trichocarpa x P. deltoides ‘Beaupré’ measured with the PICME 28 K cDNA Populus microarray at 48 hpi in ‘Beaupré’ leaves inoculated with the incompatible (I48) strain of M. larici-populina versus mock-inoculated leaves
The highest expression ratio is presented for gene models (P. trichocarpa protein ID no.) represented by different cDNA probes on the array.
|
|
Several Cyt-P450 transcripts showed contrasting expression profiles, some being strongly induced and others repressed as observed on the whole-genome oligoarray. Several other transcripts related to redox regulation also showed a slight decrease in concentration at 48 hpi in the incompatible interaction (e.g. peroxidases, thioredoxin), whereas other transcripts in the same cellular category were strongly induced (e.g. GSTs, protein disulfide isomerase, and peroxidases). Several transcripts involved in the photosynthetic machinery and carbon metabolism (e.g. light-harvesting complex CAB, PSI and PSII polypeptides, Rubisco, RuBisCO activase) and two different transcripts involved in thiamin biosynthesis showed a decreased concentration at 48 hpi in the incompatible interaction.
Validation of Rust-Regulated Genes Using cDNA Macroarrays and RT-qPCR
Expression data were further carried out at 12, 24, and 48 hpi in both compatible and incompatible interactions by either a RT-qPCR approach or reverse-northern on a Populus 4 K cDNA Nylon macroarray (Supplemental Table S4) with different sets of biological replicates than those used in the expression profiling experiments described above.
We measured transcripts coding for PR-1, PR-5, and PR-10 proteins as typical genes triggered by host defense reactions at 48 hpi in leaf tissues. The transcripts coding for I3PS (protein ID no. 832275), the dirigent-like protein (protein ID no.711753), NPR1 (protein ID no. 253241), and the RISP (protein ID no. 678883) that showed the highest transcripts accumulation based on the whole-genome oligoarrays, cDNA microarrays, or SSH library sequencing were also measured. Genes coding for a PSI center reaction subunit (protein ID no. 711610) and the small subunit of the Rubisco (protein ID no. 813777) that were slightly down-regulated during both types of interactions were included in the set of genes tested by RT-qPCR.
Strong accumulation of the selected rust-induced transcripts was confirmed by RT-qPCR amplification in leaf tissues challenged by the incompatible strain of M. larici-populina compared to mock-inoculated tissues. Maximum induction of PR genes was reached at 48 hpi (Fig. 5
), and, interestingly, NPR1 transcript showed a peak of expression at 24 hpi. The transcript coding the RISP (protein ID no. 678883) with the highest accumulation (32-fold) detected at 48 hpi with whole-genome oligoarray profiling showed a different profile with RT-qPCR. A strong induction (7-fold) was measured at earlier time points in the incompatible interaction, and a lower induction level was detected at 48 hpi. We thus measured the level of this latter transcript by RT-qPCR with the RNA samples used to perform the whole-genome oligoarray hybridizations, and we observed a 56 (±8)-fold accumulation at 48 hpi (Fig. 5). This observation confirmed that the rate of induction is influenced by the physiological status of the infected plants rather than strong technical biases in array measurement. In some cases, RT-qPCR revealed a late induction of transcripts coding PR proteins (e.g. PR-1 and PR-10; Fig. 5) at 48 hpi in the compatible interaction with lower levels than those reached in the incompatible interaction, whereas array analysis did not reveal such induction.
View larger version (15K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Figure 5. RT-qPCR expression patterns for transcripts of proteins PR-1, PR-5, PR-10, I3PS, NPR1, dirigent-like protein, RISP number 678883, ribulose 1,5-bisphosphate carboxylase oxygenase, arabinogalactan protein, and PSI center reaction IV. Total RNA of mock-inoculated or inoculated leaves of P. trichocarpa x P. deltoides ‘Beaupré’ with either compatible (black bars) or incompatible (white bars) strains of M. larici-populina 12, 24, and 48 hpi was isolated, and aliquots of 1 µg were used for first-strand cDNA synthesis. PCR was performed with 2 µL of first-strand cDNA. A control with no RT in the first-strand cDNA synthesis reaction mix was included to control for the lack of genomic DNA. The Populus ubiquitin transcript (GenBank ID CA825222) was used as a control for transcript not regulated by rust. Inset, Data for the RISP number 678883; transcript concentration was measured by RT-qPCR on the RNA samples used for the whole-genome oligoarray analysis at 48 hpi (n = 3).
|
|
Comparison of the Different Transcript Profiling Approaches
Genes showing striking differences in transcript concentration during the incompatible interaction were detected by the different transcript profiling approaches, i.e. SSH cDNA sequencing, whole-genome oligoarrays, cDNA microarrays, cDNA macroarrays, and RT-qPCR, indicating consistency of the various approaches (Fig. 5; Table IV
), although the regulation ratio may vary. For example, the RISP transcript that showed the highest accumulation (32-fold) based on the whole-genome oligoarrays was represented by several ESTs on the cDNA microarrays and showed a level of accumulation over 10-fold (protein ID no. 678883; Table III). In contrast, the strongly induced transcript coding PR-5 protein (protein ID no. 669475) showed a 28.8-fold induction based on the cDNA microarray and was quite abundant in the SSH library (2% of the cDNA clones), whereas a lower level was detected on the whole-genome oligoarray. The I3PS transcript that was highly abundant in the SSH library (approximately 20.4% of the cDNA clones) only showed approximately 5-fold accumulation in both whole-genome oligoarray and cDNA microarray analyses. In addition to bias resulting from the different technologies (full-length cDNA versus 60-mer oligonucleotide probes), differences in mRNA accumulation detected between the various profiling approaches likely reflect the fact that RNA were extracted from different sets of biological replicates with delayed plant defense response due to the variable physiological status of cuttings grown in greenhouse. A high proportion of rust-induced genes were identified in the SSH library. This technique presents the potential of identifying rare transcripts or genes expressed locally that may be missed in microarray expression profiling (e.g. statistics stringency in array analysis). Thus, this approach is not redundant but complementary to array-based transcriptome profiling.
View this table:
[in this window]
[in a new window]
|
Table IV. List of significantly rust-induced transcripts detected using different transcriptome-based expression analyses in leaves of P. trichocarpa x P. deltoides ‘Beaupré’ inoculated by the incompatible M. larici-populina strain 93ID6 compared to mock-inoculated (water) control leaves at 48 hpi
Expression ratios (fold-induction) measured with NimbleGen whole-genome oligoarrays, PICME cDNA microarrays, or Nylon-based cDNA macroarrays are given. Transcripts were selected when identified through at least two different approaches and based on their expression ratios (at least 3-fold induction in one of the transcriptomic assay). Matches of a transcript sequence with an EST from a rust-responsive Populus leaves cDNA library (SSH) are indicated (x). –, Data are not available (not detected, not significant, or not present on array or in cDNA library).
|
|
|
DISCUSSION
|
|---|
Reactions that lead to programmed cell death in incompatible interactions between plant and hemi-biotrophic or necrotrophic pathogens preventing the pathogen to spread in plant tissues have been largely described in several plant pathosystems (Heath, 2000). In contrast, interactions involving biotrophic pathogens, with their sophisticated type of pathogenesis that keeps plant cells alive and minimizes tissue damage in susceptible hosts, are poorly known. The uredinial stage of rust fungi is generally taking place through stomatal penetration, and most studies suggest that host compatibility requires the ability for the fungus to avoid or negate prehaustorial defenses within the substomatal cavity of the host leaf, breach the mesophyll plant cell wall to form the first haustorium, and develop a biotrophic interaction with the living invaded cell to support further fungal growth (Schulze-Lefert and Panstruga, 2003; Glazebrook, 2005; Spanu, 2006). Recent work described haustorially expressed secreted proteins of the rust biotrophic pathogen M. lini (Catanzariti et al., 2006) that led to HR when expressed in planta, indicating a probable direct R-Avr recognition system in flax challenged by an avirulent strain of M. lini (Dodds et al., 2006).
In this study, we describe at the microscopic, histological, and transcriptomic levels a novel pathosystem involving P. trichocarpa x P. deltoides ‘Beaupré’ challenged by urediniospores of the leaf rust basidiomycete M. larici-populina. ‘Beaupré’ is resistant to M. larici-populina isolate 93ID6 (pathotype 3-4) and susceptible to isolate 98AG31 (pathotype 3-4-7; Barrès et al., 2006). Unexpectedly, there was no significant difference in fungal growth from spore germination to contact with mesophyll cell between these two isolates. Spore germlings of both isolates were able to penetrate the leaf through stomatas in the first hours after inoculation, forming primary infection structures (i.e. substomatal vesicles) in the spongy mesophyll and reaching mesophyll cells for infection. Appressoria were formed on the leaf surface but are not a prerequisite to penetrate inside plant tissues, as frequent direct hyphal penetration through stomatas were observed. Based on the amount of fungal rDNA, differences of growth in planta were only noticeable between compatible and incompatible isolates after 24 hpi (Fig. 3). The compatible isolate then spread in leaf tissues and colonized the whole mesophyll, while the incompatible strain remained in the spongy mesophyll (Fig. 2, F and H). Fungal growth increased until the compatible strain developed spore-forming cells and produced typical golden pustules filled with masses of urediniospores on the lower leaf surfaces between 6 and 7 dpi. At this stage, the resistant phenotype was generally characterized by the presence of scattered necrotic lesions and the absence of macroscopic symptoms. Similar responses have been reported in P. deltoides x Populus nigra ‘Ogy’ inoculated with virulent and avirulent isolates of M. larici-populina (Laurans and Pilate, 1999).
We were not able to detect H2O2 accumulation through DAB staining in leaf tissues challenged by the incompatible strain of M. larici-populina. H2O2 production possibly occurred transiently and only in plant cells challenged by M. larici-populina during infection. Supporting this assumption, several genes encoding enzymes of the redox regulation pathways such as GSTs, ascorbate peroxidases, and superoxide dismutase were highly up-regulated at 48 hpi. Studies conducted at the protein level confirmed the up-regulation of thioredoxin and peroxiredoxin during Populus-Melampsora interaction (Rouhier et al., 2004; Vieira Dos Santos et al., 2005). The lack of H2O2 accumulation has been previously reported in plants interacting with biotroph pathogens (Glazebrook, 2005). This may reflect the specific and complex biotrophic relationships between rust and living host cells. As observed by Laurans and Pilate (1999) and in this study, the necrotic tissues are highly localized to a limited area (Fig. 1), supporting the fact that H2O2 production related to HR was not spreading far from infection sites in leaf tissues.
Phloroglucinol staining confirmed that a massive production of monolignols was induced upon inoculation of plant tissues by the incompatible rust strain (Fig. 4). Such compounds are believed to play a role in plant defense (Dixon, 2001). Several genes of the phenylpropanoid pathways and dirigent-like proteins encoding genes were induced at 48 hpi in inoculated leaves prior to observation of the maximum level of phloroglucinol staining. Strong production of secondary metabolites in colonized leaves likely led to the synthesis of phytoalexins and lignin deposition in secondary cell walls restricting the fungal proliferation, as shown in the cowpea (Vigna unguiculata)-Uromyces vignae interaction and in other pathosystems involving biotrophic fungal pathogens (Heath, 1997, 2000). Callose synthase and Phe ammonia-lyase genes are often reported as marker genes of lignin and callose deposition, although the levels of induction may strongly vary from one interaction to another. In this study, the homologs of Populus Phe ammonia-lyase and callose synthase were not significantly induced during the incompatible interaction at the time points tested. Induction of dirigent proteins encoding transcripts has been reported in conifers submitted to wounding or insect attacks (Ralph et al., 2006b), and their products may participate in a general stress response in trees submitted to biotic or abiotic stress.
Studying the transcriptome of rust-infected leaves with whole-genome oligoarray harboring more than 45,000 putative gene models from the P. trichocarpa genome sequence (Tuskan et al., 2006) identified 2,397 rust-responsive genes (approximately 3% of the total set of arrayed genes) that may play a role in defense reactions in the incompatible interaction or, conversely, in supporting fungal growth in susceptible plants. As expected from previous studies in model species (Schenk et al., 2000; Mahalingam et al., 2003), many functional groups of genes were found to be involved in the defense response, including signal transduction pathways components, genes stimulated during biotic or abiotic stress responses, and genes of primary and secondary metabolisms. A striking alteration in steady-state RNA populations took place at 48 hpi, whereas few genes were regulated at earlier stages of interaction when tested by RT-qPCR or cDNA macroarrays. This suggests that gene expression in response to rust infection is activated later after the fungal cells entered into contact with the leaf surface and colonized the substomatal cavity. Recognition of the pathogen is likely taking place upon triggering of specific recognition mechanisms when fungal hyphae are elongating within the mesophyll and attempt to penetrate the plant cell wall barrier to form haustorial structures (Fig. 2). Presence of strongly inducible genes in the incompatible interaction indicates that the rust infection has had a significant impact on the leaf transcriptome despite the limited amount of fungal mycelium (Fig. 3) during the early stages of invasion. The observed differences between the transcriptomes of leaves infected by either the compatible or incompatible isolates were striking. The incompatible strain elicited up-regulation of a wide range of defense-related genes (e.g. PR proteins, GSTs), whereas these genes were not induced by the compatible strain.
Several genes that may participate in the perception of the rust pathogen by sensing avirulence products released by invading hyphae were differentially accumulated in the incompatible interaction between P. trichocarpa x P. deltoides and M. larici-populina. We identified transcripts coding for putative LRR disease resistance proteins (protein ID nos. 645750 and 826060) that showed an induction of their expression (Table IV). Transcripts coding for LRR receptor protein kinases (protein ID nos. 417599 and 171587), showing homology with RLK5 and PERK1, respectively, were also induced during the incompatible interaction. PERK1 encodes a putative receptor kinase with an extracellular domain with sequence identity to cell wall-associated extensin-like proteins. PERK1 transcripts are rapidly expressed upon mechanical wounding in response to Sclerotinia sclerotiorum and SA and methyl jasmonate application (Silva and Goring, 2002). RLK5, also named HAESA, is involved in the control of floral abscission in Arabidopsis (Haffani et al., 2004). These receptor kinases belong to a large family that had been shown to play a role in microbial sensing, and they may likely participate in pathogen perception in poplar. In flax, the M. lini products of the avirulence gene Avr567 are expressed in haustoria and recognized inside plant cells (Dodds et al., 2004). This recognition occurs through a direct interaction with the L resistance genes products of flax (Dodds et al., 2006). L genes are members of the NBS-LRR resistance gene family (Ellis et al., 1999). We did not identify any members of this family among rust-induced genes. These plant proteins are responsible for the detection of pathogens in natural conditions and should be constitutively expressed; thus, their expression may not necessarily be under transcriptional control during the infection process. None of the rust-induced putative receptors and LRR proteins is located in the superclusters of NBS-LRR R genes containing the MER locus (Tuskan et al., 2006) or close to the rust-resistance loci identified in poplar by genetic mapping (Cervera et al., 2004; Lescot et al., 2004; Jorge et al., 2005). Strikingly, a significant induction of RLK transcripts was also detected in the compatible interaction at 48 hpi (2.7-fold).
Upon recognition of the pathogen aggression, a complex network of signaling enzymes and molecules relays the information in the plant cell to the nucleus, where specific defense-related gene expression is triggered. As shown in several other plant-fungus interactions, components of the signaling pathways were induced in the incompatible interaction between P. trichocarpa x P. deltoides ‘Beaupré’ and M. larici-populina. These include transcripts from the calcium- and ethylene-related pathways, calmodulin, calreticulin, 1-aminocyclopropane-1-carboxylate oxidase, 14-3-3 proteins, and ethylene-responsive element-binding protein (EREBP) and MYB family transcription factors. NPR1 is an important regulator of PR gene expression through binding to transcriptional regulators TGA elements and a low accumulation (approximately 2-fold) of its transcript has been reported in different pathosystems (Glazebrook, 2005). P. trichocarpa x P. deltoides ‘Beaupré’ NPR1 transcript (protein ID no. 253241) was accumulated at 24 and 48 hpi in poplar leaves during the incompatible interaction (Fig. 5) prior the strong induction observed for targeted PR genes. Interestingly, we also detected the induction of a transcript coding for NPR1/NIM1-interacting protein (NIMIN-1) at 48 hpi (Table IV). NIMIN-1 directly interacts with NPR1 and can modulate its activity and expression of PR genes in Arabidopsis (Weigel et al., 2005). A gene that showed among the highest levels of induction, using the different transcript profiling approaches, encoded an I3PS. The protein is involved in the production of inositol-P, a metabolite that could lead to the production of various products such as phosphatidylinositides, cell wall components, or oligosaccharides of the raffinose series (Loewus and Murthy, 2000). Phosphatidylinositols, like inositol 1,4,5-trisphosphate, are important secondary messengers of the cell transduction pathways that play a crucial role in calcium homeostasis in plant cells (Munnik et al., 1998). Involvement of calcium-related signaling in plant-rust interaction has been previously reported (Xu and Heath, 1998). Several calcium-binding proteins encoding genes are induced in poplar leaves during the incompatible interaction with M. larici-populina, and I3PS may contribute to the production of phosphatidylinositols involved in calcium regulation. Considering the strong induction of I3PS transcript in poplar leaves during the incompatible interaction with M. larici-populina, addressing the exact role of inositol-P in response to either biotic or abiotic stress requires further investigation.
Within transcripts with the highest rust-induced accumulation (>10-fold) in the incompatible interaction, several encoded PR proteins, such as PR-1, PR-2 (1,3--glucanase), PR-3 (acidic chitinase), PR-5 (thaumatin-like protein), PR-8 (basic chitinase), and PR-10 (ribonuclease). All these enzymes are known for their antifungal properties (Van Loon and Van Strien, 1999) and are typical SA-induced marker genes of plant response to bacterial and fungal attacks (Van Loon and Van Strien, 1999; Schenk et al., 2000). There is evidence that Uromyces rust species are susceptible to apoplastic PR proteins, including PR-1 (Rauscher et al., 1999). Interestingly, a gene encoding a PR-10 protein was also activated in epidermal cells of resistant cowpea challenged by the cowpea rust fungus and prior of the fungus entering the cell lumen (Mould et al., 2003). We identified many expressed hypothetical proteins among rust-responsive genes. Of interest is the RISP whose transcript showed the strongest (32-fold) induction in the incompatible interaction in the whole-genome oligoarray analysis. The RISP may play a role in early defense against M. larici-populina, and it remains to address its exact role to determine whether it is a novel PR protein in Populus.
In a compatible biotrophic interaction, the invading hyphae are able to alter the host-plant metabolism in such a way that increasing amounts of nitrogen and carbon metabolites are mobilized and translocated to fungal cells (Mendgen and Hahn, 2002; Panstruga, 2003; Both et al., 2005). In the compatible interaction between P. trichocarpa x P. deltoides ‘Beaupré’ and M. larici-populina, we did not detect a striking induction of metabolism-related elements, and only a few genes encoding transporters and enzymes of the carbon metabolism were slightly induced with the cDNA-macroarray experiment. We observed a 2.5-fold induction of a Populus transcript encoding a 1-pyrroline-5-carboxylate synthetase (protein ID no. 421059; Table II) in the compatible interaction, whereas no induction of this gene was observed in the incompatible interaction. This enzyme catalyzes the two first steps of Pro synthesis. The fis1 transcript encodes a 1-pyrroline-5-carboxylate dehydrogenase, which is specifically expressed in the flax-M. lini compatible interaction (Ayliffe et al., 2002). FIS1 is involved in the catabolism of Pro to Glu, and a possible link between such catabolic activity and the fungal metabolism remains unclear.
In conclusion, the rust-responsive genes from P. trichocarpa x P. deltoides ‘Beaupré’ presented here are a valuable resource for further functional genomics studies addressing mechanisms of durable resistance in a perennial species, Populus, and other Salicaceae. Comparative analysis of compatible and incompatible interactions showed the stimulation of several known genes involved in plant defense reactions to biotrophic pathogens like PR proteins targeted by plant recognition systems (i.e. R genes). New candidate genes, such as I3PS and RISP, which may participate to a specific Populus response to rust, were also detected. The accumulation of most rust-responsive transcripts occurred at a late stage (48 hpi) when fungal hyphae penetrate the mesophyll cells, although a few transcripts were induced at earlier time points. It appears that a perennial species, such as Populus, does not use specific arrays of defense proteins. Expansion of NBS-LRR genes as well as PR protein gene families in the Populus genome (Tuskan et al., 2006) may underlie specific recognition systems for an efficient targeting of certain PR proteins within expanded families. Further analyses, including gene inactivation, will address whether the observed quantitative differences in gene expression between the compatible and incompatible infections are sufficient to explain the drastic difference in the interaction of both M. larici-populina strains with ‘Beaupré’. Rust-responsive genes will be used in genetic studies, including ecotilling, comparing qualitative and quantitative resistances to rust in various poplar cultivars to help in marker-assisted breeding of new genotypes for durable resistance against M. larici-populina.
|
MATERIALS AND METHODS
|
|---|
Plant Materials and Growth Conditions
All experiments were performed on rooted cuttings of the hybrid poplar (Populus trichocarpa x Populus deltoides) ‘Beaupré’. For the analysis of compatible and incompatible poplar-rust interactions, P. trichocarpa x P. deltoides ‘Beaupré’ plants were grown for 12 weeks in a greenhouse from dormant cuttings in 5-L pots containing a sand-peat (50:50, v/v) mixture, with an initial fertilization of 1.45 g L–1 CaO and 6 g L–1 of slow release 13:13:13 N:P:K fertilizer (Nutricote T 100; Fertil). ‘Beaupré’ plants were watered daily with deionized water under 16-h/8-h photoperiod in greenhouse conditions with supplemental artificial light to complement to a minimum illumination of 200 µmol s–1 m–2 during the winter season. After 12 weeks, young trees were >1 m high and presented 10 to 14 fully expanded leaves.
Inoculation Procedures
Two isolates of Melampsora larici-populina were used in this study: the virulent 98AG31 (pathotype 3-4-7) and avirulent 93ID6 (pathotype 3-4) isolates (Barrès et al., 2006). The urediniospores of the two M. larici-populina isolates were multiplicated on detached leaves of P. deltoides x Populus nigra ‘Robusta’, which is susceptible to all M. larici-populina strains and collected at 10 dpi. The spores were dried and were kept in a dry atmosphere at 1°C. Fully expanded leaves from leaf plastochrony index (LPI) 5 to 9 were detached from several ‘Beaupré’ plants and spray inoculated on their abaxial surface with an urediniospore suspension in water-agar (0.1 g L–1) adjusted to 100,000 urediniospores mL–1 or with water-agar as a control (mock-inoculated leaves). Inoculations were done by pooling three leaves of different LPI from different plants (i.e. LPI 5 from plant 1, LPI 8 from plant 2, and LPI 9 from plant 3) for each treatment and each time point. The inoculated leaves (i.e. 27 leaves in total) were then incubated with the abaxial surface uppermost, floating on deionized water in 22.5- x 22.5-cm petri dishes, at 19 ± 1°C under continuous artificial illumination (fluorescent light, 25 µmol s–1 m–2), for various durations. The material harvested at different time points in the different treatments consisted of 7 cm2 (30-mm diameter) leaf discs randomly sampled on the overall leaf surface. The leaf discs were immediately snap-frozen in liquid nitrogen and transferred to –80°C until further analysis.
Microscopy Analysis and Scanning Electron Microscopy Analysis
Fungal infection structures at leaf surface were observed by clearing inoculated leaf discs (control, compatible, and incompatible) in boiling ethanol (70% v/v) for 10 min in a water bath, followed by 10 min incubation in an aniline blue solution (10 mg/mL; Sigma-Aldrich) and washes in distilled water. Observations by light microscopy were carried out at a magnification of 400x on a OPTIPHOT system (Nikon).
To obtain a high density of M. larici-populina urediniospores on the abaxial surface of poplar leaves for scanning electron microscopy observation, dry inoculations were performed with an air pistol (DIANA model 3; Mayer and Grammelspacher) with about 1 mg of spores (approximately 4 x 105 spores) inoculated for each shot. For each combination of time point (2, 6, 12, 24, 48, 96, 120, and 192 hpi) x treatment (compatible and incompatible interactions), two leaves were inoculated, and observations were carried out on three leaf fragments of 1 cm2, snap-frozen immediately after harvesting. Samples were fractured in liquid nitrogen and were attached to aluminum stubs on a Peltier stage (–50°C). They were then examined under a variable pressure scanning electron microscope (model 1450VP; Leo). Backscattered secondary electron images were observed at an accelerating voltage of 15 kV, a working distance of 10 mm, and at a pressure chamber of about 30 Pa. Digital images of samples (abaxial side and transversal section) were captured using the microscope software and edited with Adobe Photoshop CS2 (Adobe Systems France SAS) to adjust brightness and contrast or to artificially paint fungal hyphae colonizing the leaf tissues.
ROS and Lignin Monomers Detection
Production of ROS was investigated in inoculated leaf discs (ø 30 mm) using the DAB technique according to Thordal-Christensen et al. (1997). Leaf discs were incubated in a solution containing 0.2% (w/v) of 3,3-diamino-benzidine tetrahydrochloride (Sigma-Aldrich) in a shaking bath for 24 h at 25°C in the dark. Other infiltration procedures (e.g. vacuum-forced infiltration) were also tested. An evapotranspiration system was set with complete leaves where petioles were standing in DAB solution, as described by Orozco-Cardenas and Ryan (1999), for up to 24 h. Then, leaf discs were cleared in boiling ethanol (70%, v/v) in a water bath for 10 min and washed in water. Positive controls for DAB coloration were performed by injection of several H2O2 solutions of various concentrations (0.5%–20%; Sigma-Aldrich) through syringes on uninfected leaves and by wounding tissues of both noninoculated and inoculated leaves. Magnification for observation was of 100x. All observations were carried out on three replicates of each treatment (control, compatible, and incompatible) at 2, 6, 12, 24, 48, and 96 hpi. Light microscopy was carried on an OPTIPHOT system (Nikon).
Lignin monomer production was examined using the Wiesner coloration procedure with standard protocol according to Nakano and Meshitsuka (1992). Control and inoculated leaf discs were incubated in a phloroglucinol solution (2% w/v) for 5 min in the dark. Disks were then incubated in a 6 N HCl solution for 5 min, rinsed, and kept in water for further observations. Observations were carried out on replicates at 2, 6, 12, 24, 48, and 96 hpi and captured with a digital camera Sony F707 (Sony).
DNA and RNA Extraction
Total DNA was extracted from leaf tissues with the DNeasy Plant Mini kit (Qiagen) from 100 mg of frozen (–80°C) material. RNA was removed by the addition of ribonuclease A during extraction. DNA quality was verified by electrophoresis on agarose gel, and DNA quantity was measured by spectrophotometry (Sambrook et al., 2001).
Total RNA extraction was performed with the RNeasy Plant Mini kit (Qiagen) from 100 mg of pooled (–80°C) foliar discs harvested from leaves of various LPI and various individual poplar plants for each treatment considered. Pooling of samples from different trees and LPI helped in minimizing the variations between individual RNA samples. Extraction from leaf tissue was modified as described in Kohler et al. (2004), and DNase I (Qiagen) treatment was included in the RNA extraction procedure according to the manufacturer's instructions to eliminate traces of genomic DNA. Quality and quantity of RNA samples were checked by electrophoresis on agarose denaturing gel and by spectrophotometry for cDNA library construction and cDNA array experiments (Sambrook et al., 2001), while quality and quantity of RNA samples used for hybridization with NimbleGen Populus arrays were assessed on the RNA analyzer Experion (Bio-Rad) following the manufacturer's recommendation.
Amplification of rDNA ITS by qPCR
Development of the compatible and incompatible rust strains was followed in planta by specific amplification of the nuclear rDNA ITS on total DNA extracted from inoculated leaf tissues (Boyle et al., 2005). The same amount of 100 ng DNA was used for each qPCR amplification. Amplifications were performed in 1x iQ SYBR Green Supermix (Bio-Rad) with 0.3 µM of specific 5'- and 3'-primers for M. larici-populina (GenBank accession no. AY375268; M. larici-populina primers ITS-Mlp-F, GAGCGCACTTTAATGTGACTC; ITS-Mlp-R, ACTTAATTAAGTTGATAGGG; P. Frey, C. Husson, and J. Pinon, unpublished data) with a MJ-opticon2 DNA engine (Bio-Rad). Assuming a signal intensity proportional to amplified ITS sequences, we considered the pathogen growth as the relative difference (2–Ct) of fungal ITS amplicons calculated for compatible or incompatible interactions compared to mock-inoculated tissues at 2, 6, 12, 24, 48, and 96 hpi. Amplifications were carried out in triplicates, and a statistical analysis (t test) based on Ct curves was performed with the software Statview 5.0 (SAS Institute) for Mac OS 9. A standard curve was drawn for conversion of Ct values to pathogen DNA mass (Boyle et al., 2005).
Sequencing of cDNAs from SSH Library
Double-stranded cDNAs corresponding to mRNAs expressed in P. trichocarpa x P. deltoides ‘Beaupré’ leaves upon infection with the incompatible strain of M. larici-populina at 12, 24, and 48 hpi (tester probe; 1/3, 1/3, 1/3) and cDNAs from ‘Beaupré’ mock-inoculated leaves at 12, 24, and 48 hpi (driver probe; 1/3, 1/3, 1/3) were separately obtained by using the SMART-PCR cDNA Synthesis kit (BD Biosciences). The mixed-tester cDNA pool was subtracted by the mixed-driver probe (SSH) following the manufacturer's instructions (PCR-Select cDNA Subtraction kit; BD Biosciences; Diatchenko et al., 1996). These subtracted cDNAs were subcloned in pGEM-T plasmids (Promega) and were used to transform Escherichia coli DH10b bacteria (Invitrogen).
Purification by rolling circle amplification and Dye Terminator sequencing of plasmid DNA from SSH clones were performed at the GENOSCOPE (Centre National de Séquençage) on ABI3730xl DNA analyzers (Applied Biosystems). Raw sequence data was edited using SEQUENCHER 4.2 (Gene Codes) for Mac OS X. Leading and trailing vector and polylinker sequences were removed by SEQUENCHER filters. Groups of sequences were assembled into clusters using the contig routine of SEQUENCHER and parsed using dedicated Perl scripts.
NimbleGen Populus Expression Oligoarrays
The P. trichocarpa whole-genome expression oligoarray version 2.0 (NimbleGen Systems) consisted of 65,965 probe sets corresponding to 55,970 gene models predicted on the P. trichocarpa genome sequence version 1.0 and 9,995 aspen cDNA sequences (Populus tremula, Populus tremuloides, and P. tremula x P. tremuloides). The Populus version 2.0 oligoarray (S. DiFazio, A. Brunner, P. Dharmawardhana, and K. Munn, unpublished data) is fully described in the platform GPL2699 stored in the Gene Expression Omnibus (GEO) at NCBI (http://www.ncbi.nlm.nih.gov/geo).
For hybridization with whole-genome oligoarray, a series of three replicates was obtained from mock-inoculated P. trichocarpa x P. deltoides ‘Beaupré’ leaves (control) and leaves infected with either compatible or incompatible strains of M. larici-populina at 48 hpi. Preparation of samples, hybridization procedures, and data acquisition and normalization were performed at the NimbleGen facility (NimbleGen Systems) following the manufacturer's procedures. Average expression levels were calculated for each gene from the independent probes on array and were used for further analysis. Log2-transformed data were calculated and were subjected to the CyberT statistical framework (http://www.igb.uci.edu/servers/cybert/; Long et al., 2001; Baldi and Hatfield, 2002) by using the ‘bayesreg’ script in the R statistical package (http://www.r-project.org/) with the following parameters: bayesT(aData,3,3,TRUE,1,TRUE,101,9). Statistical analyses were conducted separately for compatible and incompatible interactions versus mock-inoculated tissues in a simple paired data comparison model with all gene probe duplicates considered independently. Gene probes that showed a mean intensity close to the background level (i.e. lower than 200) in all experimental conditions were removed. Transcripts for which probe duplicates both showed a P value lower than 0.05 were considered as significantly regulated. A subset of transcripts that showed a 3-fold increase or 3-fold decrease in abundance in one treatment compared to the control was then selected. All materials and procedures descriptions comply MIAME standards set for array data (Brazma et al., 2001). The complete expression dataset is available as series accession number GSE7098 in the GEO at NCBI.
cDNA Microarrays
The PICME (http://www.picme.at/) Populus microarray, composed of 28,000 elements, including 23,500 cDNAs, is described in the platform GPL4874 stored in the GEO at NCBI. This set of cDNAs corresponds to approximately 10,000 different predicted gene models in the P. trichocarpa genome sequence (Tuskan et al., 2006). For hybridization with PICME microarrays, a series of three replicates was obtained from mock-inoculated P. trichocarpa x P. deltoides ‘Beaupré’ leaves (control) and leaves infected with the incompatible strain of M. larici-populina at 48 hpi. Sample details and hybridization procedures are fully described in sample series GSM162978 and GSM162979 |
TOP
ABSTRACT
RESULTS