Global and cell-type gene expression profiles in tomato plants colonized by an arbuscular mycorrhizal fungus


Author for correspondence:
Luisa Lanfranco
Tel: 039 011 6705969


  • Arbuscular mycorrhizal symbiosis develops in roots; extensive cellular reorganizations and specific metabolic changes occur, which are mirrored by local and systemic changes in the transcript profiles.
  • A TOM2 microarray (c. 12 000 probes) has been used to obtain an overview of the transcriptional changes that are triggered in Solanum lycopersicum roots and shoots, as a result of colonization by the arbuscular mycorrhizal fungus Glomus mosseae. The cell-type expression profile of a subset of genes was monitored, using laser microdissection, to identify possible plant determinants of arbuscule development,.
  • Microarrays revealed 362 up-regulated and 293 down-regulated genes in roots. Significant gene modulation was also observed in shoots: 85 up- and 337 down-regulated genes. The most responsive genes in both organs were ascribed to primary and secondary metabolism, defence and response to stimuli, cell organization and protein modification, and transcriptional regulation. Six genes, preferentially expressed in arbusculated cells, were identified.
  • A comparative analysis only showed a limited overlap with transcript profiles identified in mycorrhizal roots of Medicago truncatula, probably as a consequence of the largely nonoverlapping probe sets on the microarray tools used. The results suggest that auxin and abscisic acid metabolism are involved in arbuscule formation and/or functioning.


One of the most widespread mutualistic associations in nature is the arbuscular mycorrhizal (AM) symbiosis that is formed between soil fungi belonging to Glomeromycota and most land plants (Parniske, 2008). The ability to form this association is widely distributed throughout the plant kingdom, and involves most agricultural, horticultural and hardwood species (Bonfante & Genre, 2008; Smith & Read, 2008). The symbiosis develops in the plant roots where the colonization process occurs in sequential steps that involve both the epidermal and cortical cells. In the root cortex, the fungus develops intercellular hyphae and extensively branched intracellular hyphae called arbuscules (Bonfante et al., 2009).

The key functional benefit for both partners is the acquisition of nutrients: the fungus provides the plant with mineral nutrients (i.e. phosphate, nitrogen and sulphur) (Govindarajulu et al., 2005; Javot et al., 2007; Allen & Shachar-Hill, 2009) while, in return, it receives carbon compounds from the plant that are essential for the completion of its life cycle (Pfeffer et al., 1999). The symbiosis has a multifunctional nature because AM fungi perform other significant roles, including protection of the plant from biotic and abiotic stress (Pozo & Azcòn-Aguilar, 2007; Aroca et al., 2008). The exploitation of these plant-beneficial symbionts in agro-environments is therefore considered of high environmental relevance and economic value.

The establishment of the symbiosis requires complex developmental programmes whose genetic determinants, at least on the plant side, have been described in part through the characterization of mutant lines defective in the colonization process (Reinhardt, 2007; Parniske, 2008).

Transcriptomic studies, mainly based on microarray technology, have also been instrumental in deciphering the molecular mechanisms that accompany the formation of arbuscular mycorrhizas. In 2003, a 2268-probe cDNA array was used for the first time to monitor changes in transcript abundance along a time course of AM establishment in the model legume Medicago truncatula. The main result was the transient induction of defence-related genes during the early stages of the interaction (Liu et al., 2003). Later, new platforms, representative of a larger portion of the M. truncatula genome, became available (Hohnjec et al., 2005; Liu et al., 2007; Gomez et al., 2009). These studies revealed changes associated with metabolic pathways that control nutritional exchanges, cell wall modifications, secondary metabolism, signal transduction, protein turn-over and transcription (Hohnjec et al., 2005; Liu et al., 2007; Gomez et al., 2009) and allowed several mycorrhiza-responsive genes to be identified. Genome-wide transcript profiling in arbuscular mycorrhizas has been obtained for the monocotyledon Oryza sativa (Güimil et al., 2005) and, more recently, for the legume Lotus japonicus (Guether et al., 2009).

Spatial gene expression information is also particularly important in AM symbiosis because a mycorrhizal root is a heterogeneous cell environment that includes colonized and uncolonized cells. The expression patterns specifically associated with arbuscule-containing cells, the key structures of the symbiosis, were explored in tomato (Solanum lycopersicum) for the first time by Balestrini et al. (2007) using laser microdissection technology and, more recently, in the legumes L. japonicus and M. truncatula (Gomez et al., 2009; Guether et al., 2009).

To date, transcriptomics studies on arbuscular mycorrhizas have focused on a few plant species, two legumes (Medicago and Lotus) and rice (Oryza sativa), and most of these studies have dealt with changes in transcript levels in roots. Only Liu et al. (2007) monitored local (root) vs systemic (shoot) changes in expression patterns in mycorrhizal Medicago plants. Although tomato is an economically important crop and a model system, with a distinct phylogenetic position and with several resources for genomic research (Shibata, 2005; Barone et al., 2008), global gene expression analyses concerning AM symbiosis have never been undertaken.

In this work, the TOM2 microarray platform, which contains c. 12 000 genes (one-third of the whole tomato genome), was used to identify differentially expressed genes in the roots and shoots of S. lycopersicum plants inoculated with Glomus mosseae, an AM fungus widely distributed in agricultural and natural ecosystems. In addition, in order to identify any possible plant determinants of the arbuscule formation, a subset of genes induced in the mycorrhizal roots was selected and their cell-type expression profiles were monitored using laser microdissection (LMD) technology.

Materials and Methods

Biological materials

Solanum lycopersicum (L.) (cv. Moneymaker) seeds were surface-sterilized by washing in 70% ethanol with a few drops of Tween 20 for 3 min and in sodium hypochlorite 5% for 13 min, and rinsed three times in distilled water for 10 min. The seeds were placed in agar:H2O (0.6%) in Petri dishes, incubated for 5 d in the dark (25°C) and then exposed to light for 4 d. The seedlings were then transferred to pots with sterile quartz sand. Inoculation of Glomus mosseae Gerd. & Trappe BEG12 (Biorize) was performed by mixing the inoculum with sterile quartz sand (30% v/v). The plants were grown in a growth chamber under a 14 h light (24°C)/10 h dark (20°C) regime, and watered at a rate of 125 ml per plant twice a week with water, and once a week with a modified Long–Ashton solution containing a low phosphorus concentration (3.2 μM Na2HPO4·12H2O) (Hewitt, 1966). The plants were harvested 42 d post-inoculation. Portions of the root system from each mycorrhizal plant were selected under a stereomicroscope on the basis of the presence of external mycelium. These root portions were mixed and pooled together and then divided into two samples, one to assess the level of mycorrhiza formation (done over 20 cm of root), and the other for RNA extraction. The mycorrhizal roots were stained with cotton blue and the level of mycorrhiza formation was assessed according to Trouvelot et al. (1986). Only roots showing high percentages for the four parameters considered [frequency of mycorrhiza formation (f %) > 50%, intensity of mycorrhiza formation (M %) > 10%, percentage of arbuscules within infected areas (a %) > 60% and percentage of arbuscules in the root system (A %) > 10%] were used for RNA extraction.

RNA extraction and microarray experiment

The total RNA was isolated from shoots and roots of single plants with Trizol reagent (Invitrogen, Carlsbad, CA, USA), according to the manufacturer’s instructions. RNA quantity and integrity were examined with Bioanalyzer 2100 (Agilent Technology, Santa Clara, CA, USA). The RNAs were pooled in three biological replicates, each pool containing RNAs from three or four plants (into each pool only one plant with an intensity of mycorrhiza formation < 20% was placed). Pools were prepared in the same way for the shoot and root samples.

The TOM2 microarrays were obtained from the Center for Gene Expression Profiles (CGEP; Cornell University, Ithaca, NY, USA). Each microarray contains 11769 oligonucleotide probes designed based on gene transcript sequences from the Lycopersicon Combined Built # 3 unigene database ( Three biological replicates were analysed and a ‘dye swap’ approach was adopted. Total RNA (500 ng) was used to generate direct fluorescently labelled cRNA using the Low RNA Input Linear Amp Kit (Agilent) according to the manufacturer’s instructions. Slides were treated following the pre-hybridization protocol provided by the manufacturer ( Microarray hybridization was performed using the Gene Expression Hybridization kit (Agilent). Post-hybridization was performed following the manufacturer’s instructions with slight modifications. An additional wash step in 0.05× SSC for 5 min and a dip in absolute ethanol were added before the final quick drying centrifugation. The slides were scanned using an Agilent microarray scanner (G2565BA) at a resolution of 10 μm and laser power set to 90%. The fluorescence data were processed using ImaGene software (version 5.6; BioDiscovery Inc.; Normalization and analysis of the microarray data were carried out using Limma (Bioconductor package) (Smyth, 2005). The values of all the spots on the arrays were per spot and per chip intensity-dependent (Lowess) normalized. Significant up- or down-regulated genes were filtered for a false discovery rate <0.05 and for normalized expression ratios greater or lower than 1.5- or 0.67-fold, respectively. Gene ontology (GO) term annotation was obtained using the software Blast2go (Conesa et al., 2005). The microarray data have been submitted to the ArrayExpress public database.

DNA extraction and PCR

The DNA extraction was performed on c. 100 mg of leaves using the DNA Plant Mini Kit (Qiagen) according to the manufacturer’s instructions. Genomic DNA was extracted from G. mosseae sporocarps. Approximately 50 sporocarps were added to 50 μl of 10× Red Taq (Sigma) buffer and crushed with a sterile pestle. The sample was heated at 95°C for 15 min and centrifuged at 12 000 g for 5 min. The supernatant was transferred to a new tube and stored at −20°C.

Primers were designed based on tomato sequences using the primer 3 software available at (Table S3). The presence of fungal DNA was evaluated using the G. mosseae 28S rDNA specific primers NDL22 and 5.21, described by Van Tuinen et al. (1998).

Polymerase chain reaction (PCR) assays were carried out in a final volume of 25 μl, containing 2.5 μl of 10× buffer, 1 μl of 2.5 mM dNTPs, 0.4 μl of each primer at 10 μM, 1 μl of Red Taq polymerase (Sigma), and 1.5 μl of a total DNA diluted 1 : 10. The PCR cycling programme consisted of: 95°C for 5 min, and 40 cycles of 94°C for 45 s, 65°C for 45 s and 72°C for 45 s.

Quantitative RT-PCR

The RNA samples used for the hybridization experiments and an additional pool from an independent sample were treated with Turbo DNase free (Ambion, Foster City, CA, USA) according to the manufacturer’s instructions. The DNA contaminations were evaluated by RT-PCR using 18S rRNA specific primers of tomato (Table S3) and the One Step RT-PCR kit (Qiagen).

First single-strand cDNA was obtained from c. 1500 ng of total RNA using Oligo-dT (Invitrogen) primers and the StrataScript Reverse Transcriptase (Stratagene, La Jolla, CA, USA). The volume of RNA samples was brought to 40 μl and then 10 μl of Mix (composed of 0.6 μl of 500 ng μl−1 Oligo-dT and 9.4 μl of distilled water) was added. The samples were incubated for 5 min at 65°C and for 10 min at room temperature. A master mix (8.5 μl) containing 5 μl of StrataScript RT buffer, 1 μl of RNase inhibitor (40 U μl−1), 2 μl of dNTPs (10 mM) and 0.5 μl of RT StrataScript enzyme was then added. The samples were incubated at 42°C for 1 h.

Real-time PCR assays were carried out using Platinum Sybr Green qPCR SuperMix-UDG (Invitrogen) in an iCycler iQ apparatus (Bio-Rad). The reactions were conducted in a total volume of 25 μl, containing 12.5 μl of 2× Platinum PCR Supermix-UDG, 300 nM of each primer (Table S5) and 20 ng of cDNA template. The PCR cycling programme consisted of: 50°C for 3 min, 95°C for 3 min and 40 cycles each consisting of 95°C for 30 s and 60°C for 30 s.

A melting curve (55–95°C with a heating rate of 0.5°C per 10 s and a continuous fluorescence measurement) was recorded at the end of each run to assess amplification product specificity. All reactions were performed with three technical replicates and three biological replicates. PCR efficiency was determined from standard curves constructed of serial dilutions of tomato genomic DNA. The comparative threshold cycle method (Rasmussen, 2001) was used to calculate the relative expression level using ubiquitin (accession no. X58253) as the housekeeping gene.

Laser microdissection

The roots were dissected into 5–10-mm pieces in the home-made Methacarn fixative (absolute methanol : chloroform : glacial acetic acid 6 : 3 : 1) at 4°C overnight for paraffin embedding (Balestrini et al., 2007). Acetone was used as fixative for the second biological replicate (R. Balestrini, unpublished) to improve RNA integrity. The root pieces were placed in acetone under vacuum for 30 min, and then kept at 4°C overnight. The next day they were gradually dehydrated in a graded series of acetone: Neoclear (Merck, Darmstadt, Germany) (3 : 1, 1 : 1 and 1 : 3) followed by Neoclear 100% (twice) with each step being carried out on ice for 1 h. The Neoclear was gradually replaced with paraffin (Paraplast Plus; Sigma-Aldrich, St Louis, MO, USA). The embedding step was as described in Balestrini et al. (2007). Sections (14 μm) were cut using a rotary microtome and placed on Leica RNase-free PEN foil slides (Leica Microsystem, Inc., Bensheim, Germany) with diethyl pyrocarbonate-distilled water. The sections were dried at 40°C in a warming plate, stored at 4°C, and used within 2 d.

A Leica AS laser microdissection system (Leica Microsystem, Inc.) was used. The samples were deparaffinized in xylene for 10 min, dipped in 100% ethanol for 2 min, and then air-dried. After collection, an RNA extraction buffer from a Pico Pure kit (Arcturus Engineering, Montain View, CA, USA) was added. The samples were incubated at 42°C for 30 min, centrifugated at 800 g for 2 min, and stored at −80°C.

RNA was extracted with the Pico Pure kit (Arcturus Engineering), as described by Balestrini et al. (2007), and quantified using a NanoDrop 1000 (Thermo Scientific, Wilmington, DE, USA) spectrophotometer. DNA contaminations were evaluated by RT-PCR assays as previously described. RT-PCR assays were carried out using the One Step RT-PCR kit (Qiagen). The samples were incubated for 30 min at 50°C, followed by 15 min of incubation at 95°C. Amplification reactions were run for 37 cycles (42 for the bifunctional nuclease gene) at 94°C for 30 s, 60°C for 30 s, and 72°C for 40 s. The RT-PCR experiments were conducted on two different biological replicates. The PCR products were visualized by agarose gel electrophoresis.


Transcriptional responses in S. lycopersicum shoots and roots upon colonization by G. mosseae

A microarray analysis was performed on both the roots and shoots collected from plants 42 d post-inoculation with the AM fungus G. mosseae (BEG12) to provide an overview of S. lycopersicum transcriptional responses to AM fungal colonization. In these conditions, the mycorrhiza formation level was rather variable: only the mycorrhizal roots showing M % (intensity of mycorrhiza formation) values between 10 and 50% were therefore selected. The mycorrhizal plants displayed an increase in root and shoot biomasses compared with the nonmycorrhizal plants (Fig. S1).

The microarray analysis revealed 362 up-regulated and 293 down-regulated genes in the mycorrhizal roots (Table S1). Significant gene modulation was also observed in the shoots; in particular, 85 genes showed increased transcript levels while 337 genes were down-regulated (Table S2). The total number of mycorrhizal-responsive genes in the shoots and roots is shown in a Venn diagram in Fig. 1; it is worth noting that 104 genes are responsive in the two organs, but only a few genes (nine) showed an increase in transcript levels in both organs; most of the root up-regulated genes, in fact, showed an opposite regulation in the shoots.

Figure 1.

 Venn diagram showing the relationships between genes that show statistically significant differential expression in response to arbuscular mycorrhizal (AM) symbiosis in the shoots and roots. The number of genes responsive in both organs is shown in the overlapping portion. The circle area is a proportional representation of the gene number.

In order to confirm the microarray results, a subset of genes was subjected to validation by quantitative RT-PCR (qRT-PCR). Ubiquitin was used as a housekeeping gene for the normalization of the expression levels of the genes of interest. From the list of up- and down-regulated genes, 17 genes modulated in the roots and/or in the shoots, representing a range of biological functions, were selected (Table S3). qRT-PCR confirmed the microarray expression pattern for all of the genes with only one exception (Table 1).

Table 1.   Microarray and qRT-PCR log ratio of genes regulated in roots and shoots of mycorrhizal plants
SGN IDPutative annotationRootsShoots
  1. 1The result was obtained from an additional independent sample. LePT1 and LePT3, tomato phosphate transporter; PR Protein, pathogenesis related protein.

SGN-U222064AP2 domain transcription factor-like4.417.07/6.72/8.75  
SGN-U216203C-repeat binding factor2.555.37/5.20/3.70  
SGN-U216204C-repeat binding factor2.543.02/2.95/3.88/4.271  
SGN-U216297AP2 domain transcription factor2.455.87/4.23/1.27/5.301  
SGN-U231884Myb transcription factor myb1171.532.33/2.73/0.07−1.49−3.43/−0.37/−3.63
SGN-U230270Receptor-like protein kinase ark10.961.33/0.67/1.93  
SGN-U217373Basic helix-loop-helixfamily protein−1.35−0.20/−0.67/−1.80  
SGN-U213812Serine decarboxylase1.32.22/1.68/1.951.622.90/3.62/1.05
SGN-U221489Transcription factor0.922.02/1.38/2.511.574.05/2.28/0.50
SGN-U213245WRKY type binding protein1.464.13/3.30/1.53  
SGN-U212927Osmotin-like protein  −3.15−5.03/−2.50/−5.70/−4.031
SGN-U212883Chitinase  −2.13−2.38/−0.93/−3.13/−5.431
SGN-U212922PR protein  −3.17−4.57/−1.13/−4.57−5.501
SGN-U214985Pathogenesis-related protein 4b  −1.3−1.73/1.12/−1.50/−1.601
SGN-U214651Wound-induced protein  −1.70−0.93/−0.67/−2.37

Because of their crucial role in the symbiosis, the transcription profile of the phosphate transporter (PT) genes was investigated. Among the three PT genes described in tomato as mycorrhiza-inducible (LePT3, LePT4 and LePT5; Nagy et al., 2005), LePT3, the only one present on the TOM2 array, was up-regulated in the mycorrhizal roots. Real-time RT-PCR experiments also showed the up-regulation of LePT5 (fold change 300) and LePT4 in mycorrhizal samples; LePT4 transcripts were not detected in nonmycorrhizal roots.

The gene ontology (GO) annotation was obtained to ascribe differentially expressed genes to specific functional categories, and to determine their clustering, using the Blast2go software (Conesa et al., 2005). The GO annotation was missing for 68 (16%) of the root-responsive genes and 109 (17%) of the shoot-responsive genes. The functional categories were organized in eight more comprehensive groups (Fig. 2). The cell organization and protein modification category was the predominant category in root and shoots, followed by primary metabolism and nucleus organization, and transcription regulation. In addition, the categories membrane and cell wall organization and defence and response to stimuli were also well represented (Fig. 2). In the roots all the categories showed similar numbers of up- and down-regulated genes (Fig. 3). A different situation was observed in the shoots, where down-regulation was prevalent in all categories. A more detailed description of genes involved in cellular organization and protein turnover, phytohormone metabolism and transcription is given in the following paragraphs.

Figure 2.

 Functional distribution of genes showing differential expression in the roots and shoots of Solanum lycopersicum colonized by Glomus mosseae; genes were grouped in eight more comprehensive groups on the basis of their gene ontology (GO) annotation. Shoots, white bars; roots, black bars; TOM2, grey bars.

Figure 3.

 Representation of the functional distribution of up- and down-regulated genes in the roots and shoots of mycorrhizal plants, according to the more comprehensive categories defined in Fig. 2. Data are expressed as a percentage of the total number of regulated genes belonging to each category.

Cellular organization and protein turnover

The microarray results revealed the differential expression in the roots of many genes involved in cellular organization, which may be required for fungal accommodation (Bonfante, 2001). In the list of up-regulated genes, we identified two putative kinesin-related proteins (SGN-U222911, up-regulated 6.40-fold; SGN-U224477, up-regulated 1.61-fold). An up-regulation was also observed for a gene (SGN-U215602; 4.06-fold) showing similarity to ATPases involved in cell division and vesicle and membrane fusion (Buaboocha et al., 2001). Considering cell wall-degrading enzymes, a beta-xylosidase alpha-l-arabinosidase (SGN U236747; up-regulated 7.78 fold) was up-regulated in mycorrhizal roots.

Consistent with data from the literature, protein turnover seems to be a major process in mycorrhizal roots (Liu et al., 2003; Hohnjec et al., 2005; Guether et al., 2009). A subtilisin-like protease (SGN-U227064; up-regulated 2.14-fold) and a serine carboxypeptidase (SGN-U224353; up-regulated 1.57-fold) were found to be up-regulated in the AM roots. In addition, six U-box proteins, two of which are members of the ubiquitin-protein ligases class (SGN-U214212; up-regulated 6.29-fold; SGN-U214213; up-regulated 9.04-fold), were also up-regulated. Five genes related to proteinase inhibitors, described as mycorrhiza-responsive by Hohnjec et al. (2005) and Guether et al. (2009), in contrast showed a decrease in transcript levels in the tomato roots.


Our transcriptome analysis has highlighted the regulation of many genes directly and/or indirectly involved in the biosynthesis and catabolism of, and responses to, phytohormones. In particular, several genes involved in auxin and abscisic acid (ABA) metabolism were found to be regulated in the roots and shoots of the mycorrhizal plants compared with the controls.

Accumulation of the transcripts for a putative indole-3-acetic acid (IAA) amido synthetase (SGN-U224367; up-regulated 25-fold) was observed in the mycorrhizal roots. The translated protein shared 80% identity with the auxin- and ethylene-responsive GH3-like protein from Capsicum chinense (Liu et al., 2005) and 77% identity with AtGH3-3 from Arabidopsis thaliana (Staswick et al., 2002, 2005). These proteins were shown to catalyse the formation of amino acid conjugates with IAA and jasmonic acid (JA). In addition, five genes belonging to the category auxin-responsive genes were up-regulated in the roots.

De novo ABA biosynthesis involves cleavage of carotenoid precursors by 9-cis-epoxycarotenoid dioxygenase (NCED) (Nambara & Marion-Poll, 2005). An increase in NCED transcript levels was detected in the mycorrhizal roots, but the gene was not responsive in the shoots. In addition, two genes related to ABA catabolism, a putative cytochrome P450 (CYP707A3) (SGN-U223227; up-regulated 6.09-fold) and a putative CYP707A2 (SGN-U237910; up-regulated 1.94-fold), showed increased transcript levels in the roots, while a third gene also involved in ABA catabolism, a putative CYP707A4 (SGN-U224438; down-regulated 0.59-fold), was down-regulated. Interestingly, eight ABA-responsive genes (four transcription factors and four genes ascribed to the calcium ion binding category) showed increased transcript levels in the roots while they were not regulated in the shoots. A decrease in phytoene synthase (PSY) transcripts was also observed in the shoots of the mycorrhizal plants. As PSY mediates the first committed step in carotenoid biosynthesis from which ABA is also derived, its down-regulation may be related to a decrease in ABA biosynthesis.

It is well known that phytohormones can mediate plant responses to biotic and abiotic stresses. The ‘response to biotic and abiotic stimuli’ category showed a high number of responsive genes in both the roots and the shoots. Within the list of up-regulated genes in the roots, some are involved in ethylene and gibberellin metabolism.

Many genes in shoots belonging to this category (pathogenesis related protein (PR protein), osmotin-like protein, chitinase, glutathione S-transferase, and hypersensitive-induced response protein) showed down-regulation. The gene most strongly up-regulated in the shoots was an S-adenosyl-l-methionine synthetase (SGN-U232236; up-regulated 8.7-fold); this is a key enzyme in plant metabolism, which catalyses the biosynthesis of S-adenosyl-l-methionine (SAM) from metionine and ATP. SAM is a precursor for the biosynthesis of ethylene (Yang & Hoffman, 1984) and polyamines (Heby & Persson, 1990) and is involved in methylation reactions (Tabor & Tabor, 1984).

Transcription factors

Seventy-seven transcription factor (TF)-encoding genes were modulated in the roots and 34 in the shoots; the majority possess zinc finger domains. Eighteen out of 20 zinc finger proteins showed an up-regulation in the mycorrhizal roots. Two zinc finger TFs were up-regulated in the shoots, while six were down-regulated. The roots and shoots had just four genes in common, and only one was found to be up-regulated in both organs.

A high transcript accumulation was also detected in the roots for seven Myb TF genes; this family is involved in the control of several processes, such as secondary metabolism and plant organ development (Allan et al., 2008). Myb TFs have recently been described as mycorrhiza-responsive in different plant systems (Liu et al., 2003; Hohnjec et al., 2005; Guether et al., 2009). Only three Myb TF genes were identified in the shoots and all of them showed down-regulation.

We also identified six genes encoding ethylene-responsive element binding proteins (EREBPs) (SGN-U213917, SGN-U214815, SGN-U214425, SGN-U219120, SGN-U216050 and SGN-U226547), and five genes belonging to the AP2 superfamily. An AP2 domain transcription factor-like gene (SGN-U222064; up-regulated 21-fold) was the TF gene with the highest up-regulation in the mycorrhizal roots.

Our array analysis revealed two genes belonging to the SCARECROW (SCR) family, GRAS4 (SGN-U218577; up-regulated 1.87-fold) and GRAS9 (SGN-U225065; up-regulated 1.59-fold), to be slightly up-regulated in the mycorrhizal roots. There is evidence that members of the GRAS family are essential for nodule development (Udvardi et al., 2007) and they may be important for the regulation of gene expression in AM roots of M. truncatula and L. japonicus (Gomez et al., 2009; Guether et al., 2009). However, the GRAS genes that are regulated in the tomato roots do not show high similarity with the GRAS genes that are up-regulated in the mycorrhizal roots of M. truncatula and L. japonicus.

Most TFs regulated in the roots do not show modulation in shoots. We observed a larger number of genes in the shoots with decreased transcript levels. In particular, five genes belonging to the WRKY transcription factor family (SGN-U218605, SGN-U226247, SGN-U214599, SGN-U232567 and SGN-U214610) were down-regulated in the shoots and not modulated in the roots.

Comparative analyses with M. truncatula mycorrhizal roots

An analysis of previously published transcript profiles has highlighted the presence of a set of mycorrhiza-responsive genes in M. truncatula (Krajinski et al., 2000;Doll et al., 2003; Frenzel et al., 2005; Liu et al., 2007; Uehlein et al., 2007; Gomez et al., 2009). We compared these sequences with those present in the TOM2 platform using a tblastx analysis with an E value cut-off of 1e–5. Only 27 sequences out of 41 from Liu et al. (2007) had a corresponding sequence in TOM2. Among these, six were modulated in the tomato mycorrhizal roots. However, only two genes, both involved in abscisic acid catabolism (SGN-U223227 and SGN-U237910), were up-regulated, as described in M. truncatula (Table S4).

Nine more genes were considered from Gomez et al. (2009): only three putative homologues have been found on the TOM2: one (triacylglycerol/steril ester lipase-like protein) was not responsive and two were down-regulated: SGN-U225376_copper transport protein (0.57-fold) and SGN-U214682_transfactor-like protein (0.58-fold).

The germin-like protein-encoding gene MtGLP1, described as up-regulated in Medicago mycorrhizal roots by Doll et al. (2003), has a corresponding sequence SGN-U216173 on TOM2; however, the gene was down-regulated in tomato mycorrhizal roots (0.28-fold).

Two aquaporin-encoding genes were also analysed from Krajinski et al. (2000) and Uehlein et al. (2007): the corresponding tomato sequences were down-regulated in mycorrhizal roots. The two lectin-encoding genes identified by Frenzel et al. (2005) do not have a corresponding sequence on TOM2.

Overall, the limited overlap with mycorrhiza-responsive genes in M. truncatula roots is probably attributable to largely nonoverlapping probe sets on the microarray tools used. It should also be noted that TOM2 lacks expressed sequence tags (ESTs) derived from mycorrhizal tissues.

Identification of novel genes expressed in arbuscule-containing cells

With the aim of identifying plant genetic markers of arbuscule formation and/or functioning, we selected a subset of genes induced in mycorrhizal roots and analysed their cell-type expression profile using LMD.

Approximately 1500 cortical cells were collected for three different populations: cells from control roots (C), noncolonized cells from mycorrhizal roots (MnM) and arbusculated cells (Myc). Two biological samples were considered for each cell population. In order to calibrate the amount of RNAs in the three different samples, RT-PCR assays using tomato 18S rRNA primers (Table S3) were performed (Fig. 4). As a positive control the expression profile of the phosphate transporter gene LePT4, which was previously shown to be expressed in arbusculated cells (Balestrini et al., 2007), was monitored. As expected, the corresponding PCR product was detected in the arbuscule-containing cells. RT-PCR assays using a G. mosseae-specific primer were also carried out to verify the presence of the fungus. A product of the expected size (380 bp) was detected in the arbuscule-containing cells as well as in the noncolonized cortical cells from the mycorrhizal roots (Fig. 4). This could be explained by the presence of intercellular hyphae, as already reported in Balestrini et al. (2007). Overall these results confirmed the quality of the microdissected samples.

Figure 4.

 Gel electrophoresis of RT-PCR products obtained from microdissected RNA samples using specific primers. C, cortical cells from nonmycorrhizal roots; MnM, noncolonized cortical cells from mycorrhizal roots; Myc, arbusculated cortical cells.

Specific primers were designed for 20 genes (Table S5) which were selected on the basis of fold change values obtained from the microarray analysis. Each primer pair was first tested on tomato plant DNA as a positive control and also on G. mosseae genomic DNA to exclude cross-hybridization. All the primers amplified a DNA fragment of the expected size and no signal from G. mosseae DNA was obtained. These primer pairs were then used in RT-PCR experiments on the calibrated RNAs from the three microdissected samples. Five genes showing similarity to a β-xylosidase α-l-arabinosidase (SGN-U236747), a putative kinesin-like protein (SGN-U222911), a bifunctional nuclease (SGN-U223510), a UDP-glucoronosyl UDP-glucosyl transferase protein (SGN-U214669) and a cytochrome P450 (putative CYP707 A3; SGN-U223227) out of 20 that were exclusively expressed in arbuscule-containing cells (Fig. 4). In addition, transcripts for a gene encoding a putative indole-3-acetic acid amido synthetase (SGN-U235006) were observed in arbuscule-containing cells as well as in the MnM sample; however, the signal intensity was higher in the arbusculated cells.

A gene related to ABA synthesis (NCED; SGN-U214605), which was found to be up-regulated in the microarray analysis, was also tested. Surprisingly, NCED transcripts were detected only in control cortical cells. This suggests that the up-regulation observed in the microarray was not associated with cortical cells but with other cell types (epidermis and central cylinder).


In this study we have used microarrays to identify differentially expressed genes in roots and shoots of tomato plants challenged by an AM fungus. Although the TOM2 chip only contains about one-third of the whole tomato genome, to our knowledge this is the first report of a large-scale gene expression analysis carried out in tomato mycorrhizal plants. The description of shoot responses is also of interest (Toussaint, 2007), but rather novel, with only a few reports found in the literature (Ruiz-Lozano, 2003; Taylor & Harrier, 2003; Caravaca et al., 2005) and only one based on microarrays (Liu et al., 2007). Microarray analysis, coupled to a qRT-PCR validation step, revealed local changes in gene expression in the roots and systemic alterations in the shoot transcriptome of the mycorrhizal plants.

Previous studies demonstrated that a small percentage (5–8%) of genes identified as differentially expressed during AM colonization were actually responsive to mycorrhiza-improved phosphorus nutrition (Liu et al., 2003; Güimil et al., 2005; Hohnjec et al., 2005). As high-phosphorus conditions were not considered in this study, we might assume that, similarly, a number of genes identified in the microarray analysis are indeed modulated primarily by the phosphorus nutritional changes occurring in mycorrhizal plants as a consequence of fungal colonization.

We used the LDM technology, a very powerful approach to study cell complexity in the AM symbiosis (Balestrini et al., 2007; Balestrini & Bonfante, 2008), to carry out additional experiments, in order to identify plant markers of arbuscule development. Six genes were found to be preferentially expressed in arbuscule-containing cells.

A limited overlap with a set of mycorrhiza-responsive genes up-regulated in M. truncatula roots (Doll et al., 2003; Frenzel et al., 2005; Liu et al., 2007; Uehlein et al., 2007; Gomez et al., 2009) was observed: the two plants share the up-regulation of two genes involved in ABA catabolism. This might be partly attributable to the fact that the TOM2 array is only partially representative of the tomato genome and it does not contain ESTs derived from mycorrhizal tissues; only 33 out of 55 M. truncatula sequences considered had a putative homologue on TOM2.

A second problem may be related to the limitations of tblastx analyses in the identification of orthologues. Tomato diverged from Medicago (and Arabidopsis) as much as 150 million years ago (Yang et al., 1999). A computational comparison among tomato, Arabidopsis and Medicago suggests that, excluding a group of genes that are highly conserved in plant species, Medicago and tomato sequences are probably too divergent to easily find true orthologues (Van der Hoeven et al., 2002).

We could also argue that the limited overlap may be attributable to the fact that different plant–fungus combinations were considered. Contrasting results for the two plants (M. truncatula and tomato) have already been reported concerning the expression in root tissues of aquaporin-encoding genes: one tonoplast (Krajinski et al., 2000) and two plasma membrane (Uehlein et al., 2007) aquaporins were found to be induced in mycorrhizal roots; by contrast, down-regulation was observed for LePIP1 and LeTIP in tomato mycorrhizal roots (Ouziad et al., 2006). Based on our similarity criteria (tblastx with an E value cut-off < 1e–5) the LeTIP gene, which is present on the TOM2 as SGN-U214295 and is also down-regulated in our samples, seems to be the orthologue of the aquaporin gene described as up-regulated by Krajinski et al. (2000). A more comprehensive comparative analysis carried out on a larger number of orthologues would better clarify the extent of common transcriptional changes shared by different, taxonomically unrelated plants.

Alterations of transcript levels in the shoots were more modest than in the roots and no mycorrhiza-specific gene was identified, in agreement with what has been described in a similar investigation on M. truncatula by Liu et al. (2007).

Interestingly, down-regulation of many defence-related genes was detected in the shoots. This is in contrast to what has been found in M. truncatula, where a number of defence-related genes were up-regulated in the shoots of mycorrhizal plants. In Medicago, this expression pattern was accompanied by increased resistance to a bacterial pathogen (Liu et al., 2007). Our data may suggest greater susceptibility of tomato mycorrhizal plants to foliar pathogens. Preliminary pathogenicity tests with the fungal leaf pathogen Botrytis cinerea showed a decrease of disease severity in mycorrhizal plants (data not shown). We speculate that the down-regulated genes may not play a prominent role in defence against this pathogen. A decreased ABA content in leaves might contribute to the increased tolerance to B. cinerea, as described by Audenaert et al. (2002). This hypothesis is supported by the microarray results which showed up-regulation of genes involved in ABA catabolism in shoots of mycorrhizal plants. Further investigation is needed to fully test this hypothesis.

The modulation of a number of transcription factors was evident in the roots, in agreement with previous studies (Hohnjec et al., 2005; Gomez et al., 2009; Guether et al., 2009) and also in the shoots, as also reported by Liu et al. (2007). In addition, same major differences concerning protein turnover and hormone metabolism have been noted.

Novel genes specifically expressed in arbuscule-containing cells

The microarray results have shown that the establishment of the AM symbiosis significantly alters the primary and secondary metabolism in both organs. Using the LMD technology, transcripts for a UDP-glucoronosyl/UDP-glucosyl transferase were exclusively detected in the arbusculated cells. UDP-glucoronosyl UDP-glucosyl transferases are members of a protein family which is involved in carbohydrate transport and metabolism. Arbuscule formation is accompanied by modifications in the carbon metabolism of the plant cells, probably in order to sustain the carbon flux towards the fungus, and this gene may mediate these changes. A gene belonging to the UDP-glucoronosyl/UDP-glucosyl transferase family was also found to be up-regulated in M. truncatula mycorrhizal roots by Hohnjec et al. (2005), although no information about mRNA spatial localization was given.

Intracellular accommodation of AM fungi dramatically changes the morphological organization of the host cells (Bonfante, 2001). A new interface compartment is created. Investigations on genes related to cell wall metabolism and recent functional genomic studies have led to new ideas on the genesis of the interface compartment (Balestrini & Bonfante, 2005; Balestrini & Lanfranco, 2006 and references therein). The presence of the transcripts for a β-xylosidase-α-l-arabinosidase (which contributes to the turnover of cell wall xylose and arabinose) in arbusculated cells suggests that this gene may exert a crucial role during fungal penetration, in particular during the formation of the periarbuscular matrix.

The cytoskeleton of invaded cortical cells also undergoes massive transient rearrangement (Genre & Bonfante, 2005). The activation of a α-tubulin (Bonfante et al., 1996) and a β-tubulin (Manthey et al., 2004) promoter in arbuscule-containing cells has clearly been demonstrated. Transcripts of the SGN-U222911 clone showing similarity to kinesin-related proteins were exclusively associated with arbusculated cells. Kinesin motor proteins play crucial roles in microtubule reorganization and vesicle transport (Guo et al., 2009); this finding points to a role of this gene in the extensive cytoskeletal reorganization that occurs in cortical cells that host arbuscules.

Plant hormones are assumed to participate in the communication between AM fungi and plants (Ludwig-Müller, 2000), but their precise role in the interaction is still unclear (Hause et al., 2007 and references therein). In this study, genes involved in auxin and ABA metabolism were found to be responsive to AM colonization. Mycorrhizal roots exhibit morphological characteristics such as an increase in the number of lateral/fine roots during early growth phases, in a similar way to auxin-treated roots. Therefore, a role of auxins in the AM interaction was proposed (Ludwig-Müller, 2000). However, the determination of auxinic compound concentrations in control and AM roots from different plants has often produced contrasting results (Danneberg et al., 1992; Shaul-Keinan et al., 2002; Meixner et al., 2005). Our experiments have revealed the accumulation of transcripts for a putative IAA-amido synthetase in tomato mycorrhizal roots: the mRNAs were detected predominantly in arbuscule-containing cells. IAA-amido synthetases have been described as enzymes that are able to maintain auxin homeostasis by conjugating excess IAA to amino acids (Staswick et al., 2005). This expression pattern is in agreement with the observation that IAA and IBA amide conjugates increase during the later stages of AM colonization (Fitze et al., 2005). These results may reflect the existance of a complex mechanism that balances the concentrations of free and conjugated auxins, which can play a role in the various phases of AM symbiosis (Hause et al., 2007). Interestingly, it has recently been demonstrated that free IAA in rice induces the expression of expansins which control cell wall plasticity and, as a consequence, may render the plant more suscepitble to colonization (Ding et al., 2008). It is also worth noting that auxins, in particular high concentrations of IAA, were shown to inhibit AM fungi hyphal growth (Gryndler et al., 1998), thus also suggesting a role in the control of fungal morphogenesis.

Auxin and abscisic acid are thought to mediate tolerance to several abiotic stresses, which is often observed in mycorrhizal plants (Estrada-Luna & Davies, 2003; Aroca et al., 2008). There have been contrasting reports on the endogenous concentrations of ABA in mycorrhizal plants. A marked increase in ABA concentrations has been observed in Zea mays and Glycine max mycorrhizal roots (Danneberg et al., 1992;Bothe et al., 1994; Meixner et al., 2005). However, contradictory results have been obtained for ABA content in leaves of mycorrhizal plants (Allen et al., 1982; Aroca et al., 2008). It is important to underline that ABA has also been detected in fungal hyphae at higher concentrations than in the roots (Esch et al., 1994). These results indicate that the increase in ABA detected in mycorrhizal roots may be, at least in part, the result of the synthesis of this hormone by the AM fungus.

The endogenous ABA concentration is modulated by the balance between biosynthesis and catabolism. As far as ABA biosynthesis is concerned, the NCED gene has been proposed as the key regulatory enzyme while hydroxylation at the C-8′ position, which involves a small gene family (CYP707A genes), is considered the predominant ABA catabolism pathway (Nambara & Marion-Poll, 2005).

Our microarray data indicated an up-regulation of an NCED gene in the mycorrhizal roots (but not in the shoots); the genes related to ABA catabolism (CYP707A genes) instead showed an increase in transcript levels in both organs. In addition, ABA-responsive genes were up-regulated in the roots and down-regulated in the shoots of mycorrhizal plants. These data suggest that active synthesis and catabolism of ABA occur in mycorrhizal roots and the ABA content in the leaves of the mycorrhizal plants might be lower than in the controls. The LMD approach provided evidence that CYP707A3, a gene involved in ABA catabolism, is specifically expressed in arbuscule-containing cells, while NCED transcripts were detected only in control cortical cells. This result suggests that, in a mycorrhizal root, synthesis and catabolism of ABA may be confined to distinct cell types. A recent study describing the mycorrhizal phenotypes of tomato mutant lines impaired in ABA synthesis suggests a role of ABA in the arbuscule formation process (Herrera-Medina et al., 2007). All these data support the hypothesis that a balance between biosynthesis and catabolism of ABA is crucial for the differentiation of arbuscules.

An interesting case is that of the bifunctional nuclease whose transcripts localize in arbusculated cells. This sequence shows high similarity (∼86%) with a Lotus japonicus gene (TM0840.7.2), which has been found to be up-regulated (∼16 FC) in mycorrhizal roots (Guether et al., 2009) and with a M. truncatula gene (TC123731) found as ESTs within cDNA libraries from nodules, arbuscular mycorrhizas and virus-infected leaves. Bifunctional nucleases are enzymes that are induced during several developmental processes, such as germination, xylem differentiation, the hypersensitive and stress responses, senescence (Bariola & Green, 1997; Perez-Amador et al., 2000) and programmed cell death (Thelen & Northcote, 1989; Aoyagi et al., 1998). Although its exact role remains obscure, these data clearly suggest a role of the gene in AM symbiosis and in general in plant–microbe interactions.

In conclusion, this work provides evidence that AM symbiosis triggers considerable reprogramming in the root and shoot transcriptomes of S. lycopersicum and induces local and systemic alterations in the transcript profiles. Novel genes, which are candidates to participate in aspects of arbuscule development and/or functioning, have been identified on the basis of their expression pattern associated with arbuscule-containing cells. Our data indicate that auxin and ABA homeostasis in mycorrhizal plants deserves further investigation at both the organ and the cell-type level.


We are grateful to Raffaella Balestrini (Istituto per la Protezione delle Piante, CNR) for her help and excellent advice on laser microdissection and to Francesca Cardinale for critical discussions. The research was supported by a grant from the Regione Piemonte CIPE (Project B74) to LL and GPA and a University grant (60%) to LL.