Author for correspondence: Francis Martin Tel: +33 3 83394080 Fax: +33 3 83394069 Email: firstname.lastname@example.org
• Regulated gene expression is an important mechanism for controlling ectomycorrhizal symbiosis development. This study aimed to elucidate the coordination between development of mycorrhiza and the differential gene expression in both partners.
• We analysed RNA levels from sequential samples of symbiotic tissues of Eucalyptus globulus bicostata and the basidiomycete Pisolithus microcarpus progressing through ectomycorrhiza development using cDNA arrays. We derived groups of coordinately expressed genes using hierarchical and nonhierarchical clustering algorithms.
• Five major distinct temporal patterns of induction/repression were observed with distinct groups of early, middle-, and late-transcriptionally responsive genes to symbiosis formation. At earliest stages, the differentially expressed fungal genes included cell wall symbiosis-regulated proteins, hydrophobins and mannoproteins, whereas transcripts coding for defense-related proteins were upregulated in plant tissues. Middle- and late-transcriptionally responsive genes coded enzymes of glycolysis, tricarboxylic acid cycle and amino acid biosynthesis, as well as protein synthesis, hormone metabolism and signal transduction components.
• This investigation confirms and extends earlier results which found that changes in morphology associated with mycorrhizal development were accompanied by changes in transcript patterns, but no ectomycorrhiza-specific genes were detected.
In addition to morphological and physiological changes, the interaction between the ectomycorrhizal fungus and its host root induces a cascade of changes in gene expression in both partners (Voiblet et al., 2001; Polidori et al., 2002; Johansson et al., 2004; Le Quéré, 2004). Detailed information on these molecular processes is essential for the understanding of symbiotic tissue development (Martin et al., 2001a; Wiemken & Boller, 2002). We therefore screened arrayed cDNAs to identify symbiosis-regulated (SR) genes in the ectomycorrhiza formed between Eucalyptus globulus bicostata (hereafter referred to as Eucalyptus) and Pisolithus microcarpus 441 (hereafter referred to as Pisolithus) during the fungal mantle formation (Voiblet et al., 2001). At this developmental stage, comparisons of gene expression in free-living partners and symbiotic tissues revealed significant differences in the expression levels for 17% of the genes analysed. This analysis has identified 65 SR genes, but no ectomycorrhiza-specific genes were detected. Recently, we obtained over 2200 expressed sequence tags (ESTs) from free-living Pisolithus and different symbiotic stages of the Eucalyptus–Pisolithus ectomycorrhiza (Peter et al., 2003). In the present study, we report on temporal patterns of gene regulation during the various developmental stages of the Eucalyptus–Pisolithus symbiosis using cDNA arrays containing the novel ESTs and we demonstrate that groups of coordinately and new differentially regulated genes are expressed at these different stages.
Materials and Methods
Biological materials and ectomycorrhizas synthesis
Hyphal plugs of 14-d-old vegetative mycelium of P. microcarpus isolate 441 (formerly P. tinctorius 441) (Homobasidiomycetes, Boletales, Sclerodermatineae, Pisolithaceae) were inoculated in Petri dishes (140 mm diameter) partly filled with 50 ml of low-sugar (5 mm) Pachlewski medium in 2.0% agar (Burgess et al., 1996). Fungal colonies were maintained in the dark at 25°C until a dense mycelial mat colonized the bottom of the plate. Seeds of E. globulus bicostata were germinated in a row 2 cm above the fungal mat. The Petri dishes were then slanted at 70° for 3 d in the dark at 25°C before being transferred in a controlled environment (16 h light, 150 µmol m−2 s−1, and 8 h dark; 23°C light and 20°C dark temperature). First contacts between the seedling primary roots and mycelium were observed 4 d later. Photographs of the in vitro system are available as Supplementary Material (Fig. S1) and at http://mycor.nancy.inra.fr/ectomycorrhizadb/pub.html (Fig. S2) (Burgess et al., 1996). Roots from control uninoculated plants, the actively growing edge of free-living mycelium colonies and ectomycorrhizal roots were harvested 4, 7, 12, and 21 d after contact, snap-frozen in liquid nitrogen and stored at −80°C. About 60 colonized root systems from five plates were pooled at each stage. Pooling samples before RNA extraction has the advantage of reducing the variation caused by biological replication and sample handling. Excess extraradical mycelium was removed using a razor blade. Two (days 7 and 12) to three (days 4 and 21) replicate experiments were carried out for each time-point.
cDNA array construction
The cDNAs used for array construction represent 1345 tentative consensus sequences (TCs) of P. microcarpus and 193 TCs of E. globulus (Peter et al., 2003). The expressed sequence tags (EST) and corresponding blastx files have been uploaded in a searchable MySQL database at EctomycorrhizaDB (http://mycor.nancy.inra.fr/pages/DNAdatabases/EctomycorrhizaDB/). The cDNA libraries have been constructed from RNA of free-living mycelium, 4-, 7-, 12-, and 21-d-old ectomycorrhizas (Peter et al., 2003) and from a subtractive cDNA library (Voiblet et al., 2001). Several cDNAs coding for known genes from the ectomycorrhizal basidiomycetes Laccaria bicolor and Hebeloma cylindrosporum were also arrayed on the membranes. These clones encompass a broad range of developmental and metabolic processes in different tissues and at different ectomycorrhizal developmental stages. Controls (human desmine cDNA and a sequence of the pTriplEx2 polylinker) were also spotted to assess the level of nonspecific hybridization. Three microliters of the bacterial stocks were used to polymerase chain reaction (PCR)-amplify cDNA inserts using primers present in vector arms as described by Kohler et al. (2003). The purity and length of all PCR products were checked by agarose gel electrophoresis. A total of 1831 PCR products (20–40 ng µl−1), which satisfied our quality controls, were arrayed from 384-well microtiter plates onto positively charged nylon membranes (Eurogentec, Seraing, Belgium) using the microGrid spotting device (BioRobotics, Cambridge, UK) with a 384-pin gadget. The 0.4 µm pins deposited 100 nl of each PCR product in duplicate with a spacing of 100 µm between spots on 7 × 10 cm arrays saturated with NaOH (0.1 m) at a final density of approximately 55 clones cm−2. The nylon arrays were then washed, blocked and baked according to the manufacturer's protocol (Eurogentec).
RNA isolation, probe synthesis and cDNA array hybridization
Total RNA was isolated from ectomycorrhizas at the different stages, free-living P. microcarpus 441 and uninoculated roots of E. globulus, as described by Logemann et al. (1987). Owing to the limited amount of symbiotic tissues, complex probes were then prepared by reverse transcription using SuperScript II reverse transcriptase (Life Technologies, Cergy Pontoise, France) and the SMART-PCR cDNA Synthesis Kit (Clontech, Palo Alto, CA, USA). A limited number (13–18) of PCR cycles were used to ensure that the cDNA synthesis was in the linear phase of amplification. Labelling of the cDNA probes was done in the presence of 30 µCi [33P]dCTP and 30 µCi [33P]dATP, and random hexamers using the Prime-a-Gene Kit (Promega, Madison, WI, USA) according to the manufacturer's instructions. The unincorporated labeled nucleotides were removed using QIAquick columns (QIAGEN, Hilden, Germany). The nylon arrays were preincubated in 30 ml of a hybridization solution (5 × standard saline citrate (SSC) solution, 5 × Denhardt's solution, 0.5% sodium dodecyl sulfate (SDS), 50 µg ml−1 shared salmon sperm DNA) for 4 h at 65°C in a HIR10M rotating hybridization incubator (Grant/Boekel, VWR International, Strasbourg, France). The arrays were then incubated in 10 ml of fresh hybridization solution containing the 33P-labeled probe at 65°C for 22 h. The hybridized arrays were washed successively for 3 × 5 min in 2× SSC at room temperature, 2 × 20 min in 2× SSC containing 0.05% SDS (65°C), 2 × 20 min in 1× SSC containing 0.1% SDS (65°C) and 2 × 20 min in 0.1× SSC containing 0.1% SDS (65°C). Arrays were then wrapped in plastic foils and exposed to a phosphorimaging screen (Eastman Kodak Company, Rochester, NY, USA) for varying periods (12 h to 3 d) and the target intensities were visualized by scanning at a resolution of 50 µm per pixel in a Personal Molecular Imager FX (Bio-Rad Laboratories, Hercules, CA, USA) that generated a 16-bit TIFF image. Two to three experiments with different arrays and independent cDNA probes derived from plant, fungal and ectomycorrhizal material corresponding to pools of at least 60 seedlings were performed for each time point, thus minimizing variation between individual samples, arrays or probes.
The 16-bit TIFF image obtained with the phosphorimager imaging system were imported into the x-dot reader program (version 2.0; Cose, Paris, France) and quantified. Detection and quantification of the 3700 signals representing hybridized DNA (each cDNA plus appropriate controls being present twice) were performed using the ‘volume quantification’ method. Each spot was defined by automatic grid positioning over the array image and average pixel intensity of each spot was determined. Net signal was determined by subtraction of the local surrounding background from the intensity for each spot. Spots deemed unsuitable for accurate quantification because of array artifact were flagged and excluded from further analysis. The data table generated by x-dot reader, containing the intensity of each spot, was then exported to the EXCEL X:mac worksheet programme (Microsoft Corporation, Redmond, WA, USA) for further manipulation. Spots that had an intensity of less than twofold that of the background were flagged as undetectable and had their intensity raised to a minimum threshold value of 0.1 to avoid spurious expression level ratios at the bottom of the spot intensity range. To take account of experimental variations in specific activity of the cDNA probe preparations or exposure time that might alter the signal intensity, the raw data obtained from different hybridizations were normalized (scaled) by the global normalization method (Baldi & Hatfield, 2002). Each (fungal or plant) measurements from each array (i.e. different time points) were divided by the sum of all (fungal or plant) measurements of that array. The expression profiles corresponding to individual time-points of the different treatments (mycelium, uninoculated, inoculated) were then scaled to a common reference profile based on these normalized centered values. This scales the total signal on all array to a value of one with the advantage that each individual measurement was expressed as a fraction of the total signal (i.e. as the fraction of total mRNA). It assumes that the total amount of mRNA and the large majority of genes remain relatively constant over the time-course experiment and that those genes whose expression levels are increased are countered by genes whose expression levels decrease. The Microsoft EXCEL software was used to calculate the expression level ratios (inoculated/uninoculated roots or fungal symbiotic tissues/free-living mycelium) at each time-point from replicate arrays.
Data quality assessment was performed using analysis of variance (t-test) and a Bayesian statistical framework implemented in the Cyber-T web interface (http://www.igb.uci.edu/servers/cybert/) (Long et al., 2001; Baldi & Hatfield, 2002). Based on the statistical analysis, a gene was considered significantly up- or down-regulated if it met all four criteria: (1) t-test P-value < 0.001; (2) mycorrhiza vs control fold change = 2.5 or 0.4; (3) the trend (up- or down-regulation) was consistent in all data sets; and (4) there were significant fold changes in all of two or three data sets. For the final analysis, fold changes of genes significantly differentially expressed were averaged. Microsoft EXCEL spreadsheets with these data set are available in Supplementary Material Table S1 and at the EctomycorrhizaDB. Fold changes were supplied to various programmes for clustering. Hierarchical clustering to identify structure in the expression data was done using using EPCLUST (http://ep.ebi.ac.uk/EP/EPCLUST/) (Brazma & Vilo, 2000). Clustering tools based on hierarchical neighbor joining (Eisen et al., 1998), k-means, SOM (Tamayo et al., 1999), GPB (Rasmussen et al., 2003) and additional clustering algorithms were used to identify groups of coregulated genes using epclust, genecluster 2 (http://www.broad.mit.edu/cancer/software/genecluster2/gc2.html) and wcluto (http://cluto.ccgb.umn.edu/cgibin/wCluto/wCluto.cgi; Rasmussen et al. 2003) softwares. In genecluster, the number of SOM rows and SOM columns were set to 2 and 3, respectively, to explicitly build a SOM with six clusters (Fig. S4 in Supplementary Material). The number of iterations was set to 100 000. The default parameters were used for the other settings. epclust and wcluto default parameters were used for settings.
The procedures described in this paper complied with the MIAME standards for microarray data (Brazma et al., 2001).
Time sequence of the ectomycorrhizal development
The in vitro system used in this study for producing Eucalyptus–Pisolithus ectomycorrhizas (Supplementary Material, Fig. S1) was slightly different from the Petri-dish techniques (Malajczuk et al., 1990) previously used to study protein synthesis (Hilbert et al., 1991) and gene expression (Voiblet et al., 2001). This in vitro system was designed to mimic events that occur in the rhizosphere (Burgess et al., 1996). Seedlings germinated in the presence of fungal exudates and the two partners grew simultaneously with lateral tips being initiated in contact with mycelium. By day 4, the tap roots were c. 2 cm long and were in contact with the edge of the fungal colonies (Supplementary material Fig. S2 at EctomycorrhizaDB) (Burgess et al., 1996). Both seedlings and fungus were actively growing, so that within 7 d the whole root system was in contact with the mycelium. After 12 d, several root tips (first order lateral) (c. 2–4) were observed. These root tips continued to emerge until 21 d after contact between symbionts. A 7-cm tap root could carry up to a dozen of lateral roots. As shown previously (Burgess et al., 1996), stimulation of lateral tip emergence was significant from day 12 and by day 21 the number of lateral tips emerged on inoculated seedlings was double that of the uninoculated seedlings (data not shown). Over the time sequence this results in samples containing tips at different stages of development (Burgess et al., 1996). By doing this we are diminishing the chance of observing symbiosis-related changes in gene expression, but we avoid transfer-induced mechanical stresses. Here, we distinguished four main stages in mycorrhizal development: colonization of the root cap (4 d), mantle formation on tap and lateral roots (4–7 d), Hartig net formation on lateral roots (7–12 d) and mature mycorrhiza (outer and inner mantle and Hartig net) formed on lateral roots (21 d).
The relative amounts of RNA from Eucalyptus and Pisolithus in the RNA samples prepared from these roots were estimated by RNA gel blot analysis using species-specific rDNA internal transcribed spacer sequence probes (Carnero Diaz et al., 1997). As expected, the percentage of RNA from Pisolithus increased with increasing colonization of the root system, from 30% in 4-d-old mycorrhizal roots to 50–60% in 21-d-old samples (data not shown).
Changes in gene expression during development
To examine gene-activity changes associated with the development of Eucalyptus–Pisolithus symbiosis, we performed large-scale expression profilings using high-density cDNA nylon arrays. Radioactive [33P]cDNAs were prepared from replicate sets of Eucalyptus mycorrhizal roots harvested 4, 7, 12, and 21 d after contact with Pisolithus and the corresponding free-living mycelium and non-mycorrhizal root controls. Non-mycorrhizal and ectomycorrhizal roots from about 60 seedlings were pooled. Radiolabeled probes were incubated with 3700-element nylon microarrays spotted in duplicate with PCR products corresponding to 1345 Pisolithus and 193 Eucalyptus ESTs, 170 additional cDNA clones from L. bicolor and H. cylindrosporum, and controls These ESTs corresponded to 1170 plant and fungal TCs (i.e. unique transcripts) (Peter et al., 2003). The arrayed Pisolithus cDNA were cloned from libraries constructed from RNA of free-living mycelium, 4-, 7-, 12-, and 21-d-old ectomycorrhizas (Peter et al., 2003) and from a subtractive cDNA library (Voiblet et al., 2001). These clones encompass a broad array of developmental and metabolic processes in different tissues and at different ectomycorrhiza developmental stages. Genes shown previously to be regulated in response to the development of the symbiosis (e.g. the hydrophobin HydPt-2; SRAP32 encoding symbiosis-regulated acidic mannoproteins of 32 kDa) were spotted several times (i.e. 23 ESTs of HydPt-2 and seven ESTs of SRAP32). At each time-point, we performed two to three microarray hybridizations using [33P]cDNA complex probes prepared from different samples of free-living mycelium, uninoculated and ectomycorrhizal root material treated under identical conditions (i.e. biological repeats).
The sensitivity of the detection system was tested by hybridizing a complex [33P]cDNA population prepared from free-living Pisolithus mycelium to cDNA arrays. Each array included two duplicate spots of the same EST. A Pearson correlation coefficient (r) of 0.99 between duplicated spots indicated good agreement for signal intensity higher than twice the background signal (Fig. 1a). As an indication of reproducibility, spot intensities were also compared between pairs of arrays hybridized with two sets of radioactive cDNAs prepared from the same RNA extract (i.e. technical repeats). The results (Fig. 1b) showed good reproducibility between arrays; in all cases the Pearson correlation coefficient of replicate filter pairs was > 0.94. Significant deviations between the two arrays occurred only at low signal intensity, when the accuracy of the spot-finding algorithm is reduced and the influence of background noise increases considerably. In a typical experiment, 1076 of the 1345 ESTs corresponding to Pisolithus cDNAs (80%) gave signal intensities above twice the background levels of hybridization (human or bacterial cDNAs) (data not shown). As a representative example, spot intensities were also compared between pairs of cDNA arrays hybridized with two sets of radioactive cDNAs prepared from two different RNA extracts from free-living Pisolithus mycelium grown in identical conditions (i.e. biological repeats). The results (Fig. 1c) also showed good reproducibility between arrays; the Pearson correlation coefficient (r) of replicate filter pairs was > 0.95. However, arrays prepared from different RNA extracts of ectomycorrhizal roots showed a lower reproducibility (0.75 < r < 0.89; data not shown) as a result of the difficulty in controlling the root colonization and mycorrhiza formation between batches of Petri dish systems.
Hierarchical clustering and principal component analysis
After filtering, the final data set comprised 844 fungal ESTs and 132 plant ESTs. Hierarchical clustering (Fig. 2) showed that the data sets corresponding to expression ratios of free-living mycelium, uninoculated and ectomycorrhizal roots were different, and that the transcriptional profiles from the 4-, 7-, 12- and 21-d ectomycorrhizas differed; data sets of 4-, 7- and 12-d being clustered. Principal component analysis (PCA) was performed to analyse the extent to which the variation in expression seen among the gene signals can be attributed to a limited number of variable components. The first two principal components (time and gene expression) accounted for 88.8% of the total variability seen in the data set (data not shown).
During the symbiosis development, 3–17% of the studied transcripts showed significant changes in expression (Cyber-T t-test Ln P-value < 1.0E-4) depending on the developmental stages. In addition, inferences were only made from genes showing a ratio above 2.5 (below 0.4). Among the Pisolithus genes, 96, 73, 221 and 30 were either up- or down-regulated at 4, 7, 12 or 21 d after contact, respectively. In Eucalyptus, 62, 60, 74, and 27 transcripts showed an altered level at 4, 7, 12 or 21 d after contact, respectively. Plant ESTs have been exclusively cloned from ectomycorrhiza cDNA libraries and this likely led to the high percentage of symbiosis-regulated genes in this set. No ectomycorrhiza-specific genes were detected. A list of these differentially expressed transcripts is available (Supplementary Material, Table S2).
Cluster analysis of Pisolithus gene expression during symbiosis development
The data set was clustered using nonhierarchical k-means, self-organizing maps (SOMs) (Tamayo et al., 1999), graph partitioning-based (GPB) (Rasmussen et al., 2003) and agglomerative algorithms (Rasmussen et al., 2003) able to recognize and classify features in complex, multidimensional gene expression data. For example, GPB clustering identified seven distinct groups of early-, middle- and late-transcriptionally responsive genes to symbiosis formation (Fig. 3). Represented in the corresponding hierarchical tree (Supplementary Material Fig. S3) are groups of genes whose transcript levels either increased (red) or decreased (green) during ectomycorrhizal development. We found that most genes assigned to major GPB clusters (Supplementary Material, Fig. S3 and Table S2) were also found in SOM clusters (Supplementary Material, Fig. S4), demonstrating that various algorithms generated consistent overall patterns.
Cluster 2 (Fig. 3) contained 47 Pisolithus ESTs whose transcript levels were significantly higher in mycorrhizal roots at 4 d after inoculation and then declined as the symbiosis developed (Supplementary Material, Fig. S3 and Table S2) (Pattern I in Fig. 5). Their highest expression is taking place when Pisolithus hyphae colonized root caps and aggregated on root tips. Then, they showed a gradual decrease over time and a slight down-regulation in matured mycorrhiza (21 d). The transcripts coding for the hydrophobin gene hydPt-2, hydPt-3 and the RGD-containing mannoprotein SRAP32, robust molecular markers for early symbiosis development (Tagu et al., 1996, 2001; Laurent et al., 1999), fell into this cluster (2.2E-6 < CyberT Ln P-value < 1.2E-10). It also contains additional transcripts (e.g. SRAP17, a symbiosis-regulated acidic polypeptide of 17 kDa and hypothetical proteins) which are candidate markers for symbiosis-related changes in cell wall.
The cluster 4 (Fig. 3; 40 Pisolithus ESTs) encompassed early responsive genes whose expression increased in mycorrhizal roots, but showed their maximum transcript levels at 7 d after contact (Supplementary Material, Fig. S3 and Table S2) (Pattern II in Fig. 5). This cluster contained several fungal transcripts involved in primary carbon metabolism (i.e. hexokinase, NAD-malate dehydrogenase, aspartate aminotransferase, isocitrate dehydrogenase, pyruvate kinase, F0 ATPase, ATP/ADP carrier protein, and NADH (ubiquinone) dehydrogenase) (CyberT Ln P-value < 1.0E-5) (Fig. 4) suggesting that carbon transfer between the host plant and the mycobiont were already taking place at 7 d and this increased carbon flux stimulated glycolysis, the TCA cycle and respiration.
The cluster 6, and related clusters 3 and 5 (Fig. 3) (160 Pisolithus ESTs), encompassed genes showing maximal levels at 12 d after inoculation (Pattern III in Fig. 5). Several cellular functions are induced, including protein synthesis (ribosomal proteins, translation elongation factors) and fate (proteasome/ubiquitin components, pirin), mitochondrial activity (citrate synthase, isocitrate dehydrogenase, α-ketoglutarate dehydrogenase, malate dehydrogenase, ubiquinone-oxidoreductases, ATP synthase subunits) (Fig. 4), signaling pathway components (calcineurin B, protein kinase C inhibitor, ras, serine/threonine protein kinase) and transcription factors (CyberT Ln P-value < 1.0E-4). Transcripts involved in amino acid metabolism (NAD- and NADP-glutamate dehydrogenases, high-affinity methionine transporter, histidine kinase, homoserine kinase, nitrilase, δ-1-pyrroline-5-carboxylate dehydrogenase) (CyberT Ln P-value < 1.0E-5) (Fig. 4) also showed their highest expression at this stage.
Fungal transcripts having an increased level during mycorrhiza formation and showing their highest expression at 21 d postcontact were rare (e.g. FUN34/Outward Ammonium Transporter, 40S ribosomal protein S5 (CyberT Ln P-value, 5.2E-5)) (Pattern V in Fig. 5).
Unlike previous clusters, clusters 0 and 1 (Fig. 3) mainly contained fungal genes (65 ESTs) whose transcript levels were significantly lower in symbiotic roots (Pattern IV in Fig. 5). They mainly encompassed genes cloned from the vegetative mycelium library, such as the hydrophobin-related cerato-platanin, SnodProt1, metallothioneins, actin and actin-related proteins, proteins of the secretory system, translation elongation factor gamma, and hypothetical proteins (4.1E-6 < ln P-value < 7.5E-5) likely involved in functions preferentially expressed in the free-living mycelium.
Cluster analysis of Eucalyptus gene expression during symbiosis development
The group of early responsive plant genes (clusters 2 and 4; Fig. 3) contained the transcripts coding for various homologs of stress and defense-related proteins, such as a metallothionein-like protein, an hypersensitive-induced response protein, a pathogenesis-related (PR) protein, a cysteine proteinase inhibitor and a Bet v I allergen-related (major latex) protein (6.6E-11 < ln P-value < 7.3E-14) (Supplementary Material, Fig. S3 and Table S2). An elicitor-induced O-methyltransferase, identified in our previous cDNA array analysis of 4-d-old Eucalyptus–Pisolithus ectomycorrhiza (Voiblet et al., 2001), and a jasmonic acid-induced DNA-binding protein was also upregulated, but showed their higher expression 12 d after contact (cluster 5). The increased levels of PR proteins suggest that colonized root cells mounted defense reactions to restrict the fungal invasion.
Ectomycorrhiza formation is under hormonal control (Burgess et al., 1996; Laurans et al., 2001) and these changes in hormone levels are accompanied by alteration in transcript abundance for genes involved in hormone metabolism, such as ethylene-forming enzyme-like dioxygenase and an auxin-induced aldo/keto reductase (1.5E-8 < ln P-value < 6.3E-10). This increased expression was taking place when root cortical cells expanded and lateral root tips proliferated (days 4 and 7) (Supplemental Material, Fig. S2B). Cellular functions upregulated by mycorrhiza formation 12 d after contact (clusters 3, 5 and 6) includes: protein synthesis and fate (ubiquitin/proteasome pathway, serine carboxypeptidase, translation initiation factors and ribosomal proteins), and nitrogen and carbon metabolism (alanine aminotransferase, choline-P-cytidylyltransferase and ubiquinol-cytochrome C reductase). Plant genes downregulated by symbiosis development contained abscisic stress-related proteins (cluster 0), a glycosyl transferase (cluster 1) and hypothetical proteins (9.3E-4 < ln P-value < 1.5E-7).
Early transcriptionally responsive genes to symbiosis formation Among the most abundantly expressed genes in differentiating ectomycorrhiza are many genes expected to be involved in the synthesis of the fungal cell wall and symbiosis interfacial matrix (Martin et al., 1999). Genes accumulating in Pisolithus cell walls, including genes encoding the different hydrophobins and SRAP32 mannoproteins (Tagu et al., 1996; Laurent et al., 1999; Voiblet et al., 2001; Peter et al., 2003), are upregulated in symbiotic tissues. These are candidate markers for symbiosis-related changes in cell wall and the encoded proteins likely function during symbiosis development. Activation of HydPt-2, coding for hydrophobin-2, is a well-known molecular marker for early symbiosis development (Tagu et al., 1996). In addition to HydPt-2, other members of the hydrophobin and SRAP32 mannoprotein multigene families (e.g. SRAP17) showed a similar expression profile. These transcripts were represented on the arrays by multiple spots and their nearly invariant clustering in the HydPt-2 group (GPB cluster 2 in Fig. 3) demonstrates the internal consistency of our cDNA array analysis. Members of the HydPt-2 cluster showed a unique overall expression profile (pattern I in Fig. 5). They were strongly activated in the early stages of mycorrhiza development (4 d after contact) when the root tips were colonized and the mantle formation was taking place. They were then going back to their constitutive level in the maturing mycorrhiza (12 d and 21 d). This expression pattern has been documented in our previous studies and was observed at both the transcript (using Northern blotting, RT-PCR and digital EST Northern) (see also Supplementary Material, Fig. S5) and protein levels for both hydrophobins and SRAP32 (Tagu et al., 1996; Laurent et al., 1999; Martin et al., 1999; Peter et al., 2003). For example, HydPt-2 transcripts were upregulated 3.2-fold (1.8–7.1 in the different biological replicates for the various HydPt-2 ESTs) in 4-d-old ectomycorrhizas based on cDNA array data (Supplementary Material, Table S2), whereas they were four times more abundant in mycorrhiza than in free-living hyphae, as determined by RNA blot analysis (Tagu et al., 1996) and reverse transcriptase polymerase chain reaction (RT-PCR) (Supplementary Material Fig. S5). In silico profiling based on EST abundance in cDNA libraries indicated that HydPt-2 ESTs represented 2.6% and 0% of the mycorrhiza and free-living mycelium cDNA libraries, respectively (Peter et al., 2003). Increased expression of hydrophobin and SRAP mannoprotein transcripts during the early stages of ectomycorrhiza development is suggestive of a direct participation of corresponding proteins in morphogenetic events related to the fungus adhesion to root surfaces. Interestingly, the expression of hydrophobins increases when the expression of the hydrophobin-like ceratoplatanin SnodProt (Pazzagli et al., 1999), highly expressed in the free-living mycelium, decreases (GPB Cluster 0, Pattern V in Fig. 5). Gene disruption should provide further insights into the function of these abundant and multiple mannoproteins and hydrophobins of Pisolithus.
Middle- and late-transcriptionally responsive genes to symbiosis formation There is evidence that ectomycorrhizal symbiosis brings about considerable modification of carbon metabolism in the host roots and in the mycobiont forming the association (Nehls et al., 2001a,b). An important question in relation to the physiology of ectomycorrhizal associations concerns the extent to which each partner contributes to the metabolism of carbohydrates. The utilization patterns of [1–13C]glucose by Eucalyptus seedlings and Pisolithus mycelium was dramatically influenced by mycorrhizal colonization, with a greater allocation of carbon to short chain polyols, arabitol and erythritol and to trehalose in the mycelium and a suppression of sucrose synthesis in the roots (see Fig. 3 in Martin et al., 1998). It appears that fungal metabolism dominates the assimilation of exogenous carbohydrates into symbiotic tissues. Here, several fungal transcripts coding for enzymes in the glycolysis (e.g. hexokinase, pyruvate kinase), tricarboxylic acid (TCA) cycle (e.g. citrate synthase, isocitrate dehydrogenase) and the mitochondrial electron transport chain (NADH dehydrogenase, F0 F1 ATPase) were upregulated in symbiotic tissues 7–12 d after contact (Fig. 4; patterns II and III in Fig. 5) confirming a general stimulation of the glucose respiration pathways. The hyphal tip produces the force to break the root surface and penetrates between epidermal cells to initiate the Hartig net (Massicotte et al., 1987a,b). The internal turgor pressure is believed to be generated by an influx of water caused by the osmotic gradient produced in the fungal cell. Arabitol and erythritol, accumulated by Pisolithus, may be the compatible solutes responsible for generating the needed hydrostatic pressure. This growth in the host apoplastic space, together with the proliferation of the hyphae forming the coenocytic Hartig net, may explain the highest demand in carbon metabolites and the accompanying upregulation of transcripts encoding enzymes of the glucose assimilation pathways. Microarray analysis also indicated a shift in the plant and fungal carbon metabolism in the Paxillus involutus–Betula pendula ectomycorrhiza (Johansson et al., 2004; Le Quéré, 2004). Transcripts coding for the NADP-glutamate dehydrogenase, aspartate aminotransferase and threonine synthesis were upregulated in symbiotic tissue, in agreement with the increased amino acid synthesis observed using 13C NMR (Martin et al., 1998). Whether gene expression of these metabolic genes is controlled by sugar- and amino acid-dependent regulation or by symbiosis-related developmental signals is not known (for a discussion see Nehls et al., 2001a).
Genes regulated in the host plant
Some nonspecific, broad-spectrum defenses (e.g. chitinases and peroxidases) are clearly mounted in plant hosts when ectomycorrhizal fungi penetrates directly into the root and digests its way through the apoplastic space (for a review see Martin et al., 2001a). These induced defense responses may limit the fungal invasion of root tissues. In Eucalyptus–Pisolithus ectomycorrhiza, the group of early responsive genes (GPB Cluster 2; pattern I in Fig. 5) contained transcripts coding homologs of stress- and defense-related proteins, such as an hypersensitive-induced response protein and PR-proteins, confirming that colonized root cells mounted defense reactions. Upregulation of an elicitor-induced O-methyltransferase and a jasmonic acid-induced DNA-binding protein at mid-stage supports this contention. Cellular functions upregulated by mycorrhiza formation also includes protein turnover (e.g. ubiquitin/proteasome pathway) as previously shown at the polypeptide level (Hilbert et al., 1991).
Laczko et al. (2004) showed that Pinus sylvestris produces increased amounts of neutral lipids (e.g. fatty acids) in response to Pisolithus tinctorius colonization, that the mycobiont takes up these fatty acids at the earliest contact sites, and that it transports these metabolites as neutral lipids to the extraradical mycelium. Expression of the choline-P cytidylyltransferase, a key enzyme in phosphatidyl choline biosynthesis, was increased in Eucalyptus root tissues 12 d after contact, suggesting that synthesis of membrane phospholipids was also increased in the Eucalyptus–Pisolithus ectomycorrhiza. By contrast, fungal ergosterol in Pinus–Pisolithus ectomycorrhiza only showed a slight increase in their concentration in the later stages of the symbiosis (Laczko et al., 2004). In our study, the transcript level of 3-hydroxy-3-methylglutaryl-coenzyme A reductase, lanosterol 14 α-demethylase, oxysterol-binding nuclear receptor, C-8,7 sterol isomerase, δ(24)-sterol c-methyltransferase, component in the synthesis of sterols were not regulated (Supplementary Material, Fig. S3).
A few root transcripts were strongly repressed during mycorrhiza formation (GPB Cluster 0). Among these transcripts, water stress-inducible abscisic stress proteins were represented, suggesting that, as expected, mycorrhiza formation facilitates water transfer to the plant and thus relieves stresses experienced by the growing roots.
The global gene expression analyses presented here add new information to existing models of ectomycorrhiza development (Fig. 5). To our knowledge, this is the first transcriptome study describing the time-course of ectomycorrhiza development. Expression profiling showed that developmental reprogramming takes place in roots and hyphae. A marked change in the gene expression in Eucalyptus–Pisolithus was observed at multiple levels: (1) a general activation of the fungal protein synthesis machinery and primary carbon metabolism probably supporting an intense cell division/proliferation; (2) increased accumulation of transcripts coding for cell surface proteins in fungal hyphae (hydrophobins and SRAPs) probably involved in the mantle and symbiotic interface formation; and (3) the upregulation of defense reactions and hormone metabolism in colonized roots. Changes in transcript levels for highly regulated SR-genes found in the current cDNA array study were confirmed by digital Northern (Peter et al., 2003), Northern RNA blotting (Tagu et al., 1996; Laurent et al., 1999), and RT-PCR (Supplementary Material, Fig. S5). Several of these cellular functions are also regulated in the Betula–Paxillus symbiosis (Johansson et al., 2004; Le Quéré, 2004) suggesting the induction of common genetic programmes in various ectomycorrhizal systems. At the different developmental stages studied, development of Eucalyptus–Pisolithus and Betula–Paxillus symbioses does not induce the expression of ectomycorrhiza-specific genes (Voiblet et al., 2001; Johansson et al., 2004; Le Quéré, 2004). So far, symbiosis-specific genes were not detected in the Pisolithus–Populus and Laccaria–Pseudotsuga ectomycorrhizas (A. Kohler et al., unpublished). The apparent lack of ectomycorrhiza-specific gene is striking and suggests that ontogenic and metabolic programmes leading to the symbiosis development and functioning are driven by changes in the organization of gene networks (e.g. differential arrays of pre-existing transcriptional factors and/or transduction pathways), rather than the specific expression of symbiosis-specific transcriptional factors or signalling components. A more complete analysis of this key question will await the completion of larger sets of ectomycorrhiza expression profiles on a wider range of associations.
In conclusion, this investigation confirms and extends earlier results which found that changes in morphology associated with mycorrhizal development were accompanied by changes in transcript patterns (Voiblet et al., 2001; Johansson et al., 2004; Le Quéré, 2004) and that these changes commenced at the time of contact between the two partners long before the formation of functional ectomycorrhiza. Alteration in the concentration of transcripts coding for hydrophobins and mannoproteins SRAP32 was already documented (Tagu et al., 1996; Laurent et al., 1999; Voiblet et al., 2001), but the time frame of the drastic changes talking place in the carbohydrate, amino acid and hormone metabolism, and defence reactions in the Eucalyptus–Pisolithus symbiosis was unknown. The present experiment has elucidated the importance of coordination between development of mycorrhiza and the differential gene expression in both partners. Understanding the synchronization of these events is essential for understanding the determinants of symbiosis compatibility and mycorrhiza ontogenesis. Further studies are now needed to investigate the expression of the SR-genes identified in environmental samples.
We thank Drs Annegret Kohler, Martina Peter and Anne Jambois (UMR IaM) for valuable discussions during the course of this study. We are grateful to Dr Catherine Voiblet for her technical advices on cDNA arrays at the beginning of this project. Laccaria bicolor and P. involutus cDNAs were kindly provided by Martina Peter, Annick Brun and Michel Chalot (UMR IaM). The assistance of Marine Wasniewski and Natacha Steinwich in cDNA amplification is acknowledged. We also greatly appreciate the technical support of Christine Delaruelle and Sofia Kotowski (UMR IaM) and their dedicated assistance in sequencing and ectomycorrhiza production, respectively. Thanks to Christian Herbé (INRA-Nancy) for his help in Visual Basic macro programing. Thanks to Dr Anne Jambois and Dr Treena Burgess for the photos of ectomycorrhizas. This work was supported by INRA grants (programs Microbiology, Sequencing Symbiont & Pathogen Genomes, and Lignome) and the Région de Lorraine. The DNA Sequencing and Functional Genomics Facilities at INRA-Nancy was financed by grants from INRA and Région Lorraine through the Institut Fédérateur de Recherche no. 110.
Table S1 Raw and normalized data sets, and a list of regulated genes.
Table S2 Symbiosis-regulated genes found in the seven different wcluto graph partitioning-based (GPB) clusters. Clustering analysis was carried out on the 395 symbiosis-regulated Pisolithus microcarpus and Eucalyptus globulus genes (ratios > 2.5 and < 0.4).
Fig. S1 The Eucalyptus globulus–Pisolithus microcarpus in vitro Petri-dish system.
Fig. S2 (a) The development of Eucalyptus–Pisolithus ectomycorrhiza synthesized in vitro. (b) Longitudinal sections of the developing Eucalyptus–Pisolithus ectomycorrhiza synthesized in vitro (from Burgess et al., 1996 with permission).
Fig. S3 Hierarchical clustering tree view of 395 symbiosis-regulated Pisolithus microcarpus and Eucalyptus globulus genes. Clustering analysis was carried out using the graph partitioning-based (GPB) paradigm in wcluto. Each horizontal line displays the expression ratio for one gene in symbiotic tissues vs free-living partners. The GPB clustering allowed to define subset of genes sharing similar expression profiles (clusters 0–6). Each gene is represented by a row of colored boxes and each stage is represented by a single column. Regulation levels range from pale to saturated colors (red for induction; green for repression). Black indicates no change in gene expression. Eucalyptus globulus genes are prefixed by ‘Eg’.
Fig. S4 Self-organizing map (SOM) clusters of expression profiles during Eucalyptus–Pisolithus ectomycorrhiza development. Each graph displays the mean pattern of expression of the transcripts in that cluster (blue lines) and the standard deviation of average expression (red lines). The number of transcripts (expressed sequence tags) in each cluster is at the top center of each SOM. The y-axis represents normalized gene expression levels. A list of genes in each cluster is also provided.
Fig. S5 Reverse transcriptase-polymerase chain reaction (RT-PCR) expression patterns for representative Eucalyptus globulus and Pisolithus microcarpus SR-genes. Total RNA of free-living mycelium, nonmycorrhizal and mycorrhizal roots was isolated and aliquots of 1 µg were used for first-strand cDNA synthesis. A PCR was performed with 2 µl of first-strand cDNA and 20, 25, and 30 cycles. A control with no RT in the first strand cDNA synthesis reaction mix was included to control for the lack of genomic DNA. Eucalyptus- or Pisolithus-specific primers were used to amplify rDNA internal transcribed spacer sequence (ITS/5.8S rRNA) and to check for equal loading of plant or fungal RNA. Pm and Eg, P. microcarpus and E. globulus, respectively; M4, M7, M12, and M21 are 4-, 7-, 12-, and 21-d-old ectomycorrhiza, respectively.