About Us | Help Videos | Contact Us | Subscriptions
 

The Plant Genome - Special Submissions

Inside Arbuscular Mycorrhizal Roots – Molecular Probes to Understand the Symbiosis

 

This article in TPG

  1. Vol. 6 No. 2
    unlockOPEN ACCESS
     
    Received: June 04, 2012
    Published: January 7, 2013


    * Corresponding author(s): bbarbazuk@ufl.edu
 View
 Download
 Alerts
 Permissions
Request Permissions
 Share

doi:10.3835/plantgenome2012.06.0007
  1. Daniel Ruzickaa,
  2. Srikar Chamalab,
  3. Felipe H. Barrios-Masiasc,
  4. Francis Martind,
  5. Sally Smithe,
  6. Louise E. Jacksonc,
  7. W. Brad Barbazuk b and
  8. Daniel P. Schachtmana
  1. a Donald Danforth Plant Science Center, 975 N. Warson Rd., St. Louis, MO 63132
    b Dep. of Biology and the UF Genetics Institute, Univ. of Florida, Cancer & Genetics Research Complex, Room 407, 2033 Mowry Rd., PO Box 103610, Gainesville, FL 32610
    c Dep. of Land, Air and Water Resources, Univ. of California-Davis, Plant and Environmental Sciences Bldg., One Shields Ave., Davis, CA 95616
    d INRA, UMR1136 INRA-Nancy Université ‘Interactions Arbres/Microorganismes,’ Centre de Nancy, 54280 Champenoux, France
    e Soil Group, School of Agriculture, Food and Wine, Waite Campus, The Univ. of Adelaide, Adelaide, South Australia 5005, Australia. Daniel Ruzicka and Daniel P. Schachtman, present address: Monsanto Company, 700 Chesterfield Pkwy., Chesterfield, MO 63017

Abstract

Associations between arbuscular mycorrhizal (AM) fungi and plants are an ancient and widespread plant microbe symbioses. Most land plants can associate with this specialized group of soil fungi (in the Glomeromycota), which enhance plant nutrient uptake in return for C derived from plant photosynthesis. Elucidating the mechanisms involved in the symbiosis between obligate symbionts such as AM fungi and plant roots is challenging because AM fungal transcripts in roots are in low abundance and reference genomes for the fungi have not been available. A deep sequencing metatranscriptomics approach was applied to a wild-type tomato and a tomato mutant (Solanum lycopersicum L. cultivar RioGrande 76R) incapable of supporting a functional AM symbiosis, revealing novel AM fungal and microbial transcripts expressed in colonized roots. We confirm transcripts known to be mycorrhiza associated and report the discovery of more than 500 AM fungal and novel plant transcripts associated with mycorrhizal tomato roots including putative Zn, Fe, aquaporin, and carbohydrate transporters as well as mycorrhizal-associated alternative gene splicing. This analysis provides a fundamental step toward identifying the molecular mechanisms of mineral and carbohydrate exchange during the symbiosis. The utility of this metatranscriptomic approach to explore an obligate biotrophic interaction is illustrated, especially as it relates to agriculturally relevant biological processes.


Abbreviations

    ABC, adenosine triphosphate-binding cassette; AM, arbuscular mycorrhizal; AS, alternative splicing; BLAST, basic local alignment search tool; cDNA, complementary DNA; CT, cycle threshold; EST, expressed sequence tag; GO, gene ontology; MFS, major facilitator superfamily; mRNA, messenger ribonucleic acid; myc-, without mycorrhiza; NCBI, National Center for Biotechnology Information; NM, nonmycorrhizal; NMD, nonsense-mediated messenger ribonucleic acid decay; nr, nonredundant; PCR, polymerase chain reaction; qRT-PCR, real-time reverse-transcription polymerase chain reaction; RNA, ribonucleic acid; RNA-seq, ribonucleic acid sequencing; RT-PCR, reverse transcription polymerase chain reaction; TC, tentative consensus; UTR, untranslated region

Arbuscular mycorrhizal (AM) fungi are important root symbionts that associate with the majority of land plants including most agricultural species (Smith and Read, 2008). They are obligate mutualistic biotrophs that provide an additional (fungal) pathway of mineral nutrient (mainly inorganic P, N, S, and Zn) uptake from the soil (Allen and Shachar-Hill, 2009; Govindarajulu et al., 2005; Javot et al., 2007), enhance drought tolerance (Aroca et al., 2008), and increased pathogen protection (Liu et al., 2007). In return for soil-derived nutrients, the plant supplies C to the fungus in the form of photosynthesis-derived sugars (Pfeffer et al., 1999). Establishment of the AM symbiosis involves major cellular and functional integration of plant and AM fungal development. Arbuscular mycorrhizal colonization of the roots results in formation of intracellular exchange interfaces in root cortical cells, which are composed of the membranes of both symbionts together with an interfacial apoplast between them. Outside the root a network of extraradical mycelium develops, extending beyond the rhizosphere and increasing the volume of soil from which essential nutrients are captured by an AM plant.

Development and function of the AM symbiosis requires the coordination of a wide range of cellular processes and results in changes to the plant transcriptome and metabolome. Examination of the transcriptome responses to AM fungal colonization have been conducted in a variety of plant species, mainly focusing on the identification of diverse sets of plant genes regulated by the symbiosis (Balestrini and Lanfranco, 2006; Breuillin et al., 2010; Fiorilli et al., 2009; Gomez et al., 2009; Grunwald et al., 2009; Liu et al., 2003). For example, studies in Oryza sativa L. (rice) and Medicago truncatula Gaertn. (barrel medic) have identified 239 and 377 genes regulated during AM symbiosis, respectively. The proteins encoded by these genes are involved in many different processes including primary and secondary metabolism, signal transduction, and transcriptional regulation (Güimil et al., 2005; Hohnjec et al., 2005). A recent study with sand-grown tomato inoculated with Glomus mosseae identified 655 plant genes differentially regulated in mycorrhizal roots (Fiorilli et al., 2009). None of these studies have provided comprehensive information on the fungal symbionts in mycorrhizal roots, largely because of the absence of fungal transcripts on the microarrays used.

While these studies have improved our understanding of the changes in the plant transcriptome in response to the AM symbiosis, significantly less is known about the fungal transcriptome in mycorrhizal roots. Arbuscular mycorrhizal fungi have not been successfully cultured in vitro in the absence of a plant to produce structures that are similar to those found in plant roots. Therefore, few genome resources have been available until recently. A recent report detailing the Glomus intraradices transcriptome in germinated spores, extraradical mycelium, intraradical mycelium, and symbiosis roots identified fungal genes differentially expressed between the various tissues using inoculated and mock inoculated roots in Daucus carota L. (carrot) and M. truncatula (Tisserant et al., 2011). Efforts to sequence the Glomus intraradices genome have been slowed by challenges including a high content of transposable elements and heterokaryotic multinucleate hyphae leading to poor assemblies, presumably a result of duplications and polymorphism (Martin et al., 2008). While AM spore germination can be induced in the absence of a plant, the germinating spores are developmentally and functionally different from the fungus growing in symbiosis with the root. To more fully understand the function of the AM fungus in an active symbiotic context, we used a different approach, which included field sites where roots were colonized by a natural population of mycorrhizal fungi species and a tomato mutant with reduced colonization rates as a control to characterize the fungal transcriptome under field conditions in planta. These data provide a first step toward characterizing the many different functional aspects of the fungal side of the symbiosis based on the characterization of the genes that are expressed in fungal tissues inside the roots.

The reduced mycorrhiza colonization (rmc) mutant in tomato was a key tool used in this study to characterize the changes in the root transcriptome in mycorrhizal roots in the context of the soil environment. With the exception of one fungal species (Gao et al., 2001; Poulsen et al., 2005), AM fungi do not successfully colonize rmc mutant plants or support a functional mycorrhizal symbiosis (Barker et al., 1998; Larkan et al., 2007). Thus, the mutant can be used to provide close to a nonmycorrhizal (NM) treatment while keeping the roots in the same natural soil environment with an active microbial community including other bacterial and fungal species. It is important to note that rmc mutant plants are phenotypically identical to the wild-type tomato plants when grown in the absence of AM fungi (Cavagnaro et al., 2004). When the mutant is grown in the presence of AM fungi, rmc plant phenotypes include phosphate starvation, changes in mineral nutrient uptake, drought response, and lower yield, which has been attributed to the defects in the AM symbiosis (Cavagnaro et al., 2008; Ruzicka et al., 2012).

To delve deeper into the transcriptome changes in AM tomato roots, we used a de novo metatranscriptomics approach to identify plant and AM fungal genes whose expression is specific to or strongly induced in mycorrhizal roots. Arbuscular mycorrhizal and NM root transcripts from plants of wild-type or rmc tomato grown in organic farm field soil (Ruzicka et al., 2012) were sampled by a Roche-454 based ribonucleic acid sequencing (RNA-seq) strategy (Roche Applied Science). Subsequent computational analyses were used to assemble, annotate, and quantify the transcripts and determine the species of origin (tomato vs. AM fungal). By comparing transcript populations from AM compared to NM roots and benchmarking against public sequence data and transcript and genome data from a recent study (Tisserant et al., 2011), we identified putative mycorrhizal-specific plant and fungal transcripts, some of which were novel while others had been previously well characterized as mycorrhiza associated (Tisserant et al., 2011). This study demonstrates the use of a metatranscriptome approach (Marco, 2010) to identify AM fungal sequences related to an important symbiotic process and provides a powerful methodology to identify and sequence transcripts from this and other plant–microbe interactions without the need for artificial inoculation procedures or a priori reference genome sequence information.


Materials and Methods

Plant Material

Plant material, growth conditions, and mycorrhizal colonization analysis is summarized in Ruzicka et al. (2012). Analyses in the present study were among wild-type tomato plants grown in field soil and colonized by AM fungi (∼23% as determined by trypan blue staining and microscopy) (wild-type), rmc tomato mutants with reduced mycorrhizal colonization (<1%) grown in field soil (rmc), and wild-type plants grown in sterilized soil and not colonized with mycorrhiza (wild-type myc- [without mycorrhiza]) (Fig. 1) (Ruzicka et al., 2012). Molecular markers for Glomus intraradices alkaline phosphatase and AM β-tubulin were also used to confirm expression in mycorrhizal wild-type roots and absence in rmc root samples (Ruzicka et al., 2012).

Figure 1.
Figure 1.

Experimental design and arbuscular mycorrhizal colonization. Wild-type (A) and rmc mutant (B) tomato roots were stained with trypan blue to determine the number of mycorrhiza arbuscules (dark blue cells). A flow chart (C) shows the processing steps taken in this metatranscriptome analysis leading to the identification of putative fungal and tomato sequences specific to the mycorrhizal root. mRNA, messenger RNA; myc-, without mycorrhiza; cDNA, complementary DNA; BLAST, basic local alignment search tool; NCBI, National Center for Biotechnology Information; NR, nonredundant; WGS, whole genome sequence.

 

Sequencing (454 FLX) and Assembly

A flow chart summarizing the experimental design, sequencing, and assembly steps is presented in Fig. 1. Total ribonucleic acid (RNA) was extracted from root tissue from the three sample groups using the Qiagen RNeasy Plant Kit. Thirty micrograms total RNA consisting of 5 μg total RNA (deoxyribonuclease treated) from six pooled biological replicate samples was used in the Ambion MicroPoly(A) purist kit (Life Technologies Corporation) to enrich for polyadenylated messenger RNA (mRNA). The mRNA was concentrated and complementary DNA (cDNA) synthesis and polymerase chain reaction (PCR) amplification was performed on 100 ng enriched mRNA using the SMARTer PCR cDNA synthesis kit (Clontech Laboratories Inc.).

Sequencing (454-FLX) sequencing was performed on each of the three root cDNA preparations at the Florida Genomics Core Facility as described in Margulies et al. (2005) with slight modifications as specified by 454 Life Sciences. Each of the three samples was deposited on one-half plate and run on the 454-FLX sequencer, a discriminating suffix was appended to the read names from each sample, and the reads were pooled before assembly with the expectation that redundancy present between pools would result in longer sequences and improve annotation but still allow the contribution of each sample to be easily determined. The pooled reads were assembled with the Roche 454 Newbler assembler version 2.0 (Margulies et al., 2005).

To identify and remove transcript assemblies that were mitochondrial, chloroplast, insect, or bacterial in origin, all assemblies were aligned by the basic local alignment search tool (BLAST) (WUBLASTN version 2.0 [Gish, W., personal communication, 2012], B = 5; V = 5; E = 0.000001; links; topcombo N = 1) to bacterial (75,342 sequences), insect (75,439 sequences), chloroplast (14,293 sequences), and mitochondrial sequences (65,137 sequences) collected from GenBank. In addition, any 454 transcript assembly that aligned to insect or bacterial sequences annotated as “ribosome,” “ribosomal,” or “histone” was retained to avoid removing tomato or mycorrhizal representatives of highly conserved ribosomal or histone sequences. Approximately 17.5% (5388) of the transcript assemblies demonstrated significant sequence similarity to sequences within the collection of insect, organellar, and bacterial sequences and were removed from further analysis. The assembly.ace files produced by Newbler assembler were consulted to assign all reads incorporated into a given contig to a particular RNA source discriminating suffix (see above). This generated an approximate expression profile for each transcript assembly across the three RNA sources.

Transcript Assembly Annotation

The contig assembly sequences were initially annotated by sequence database BLAST searches and functional assignments were inferred by applying a locally installed Blast2Go pipeline (Conesa et al., 2005). This installation (October 2009) retrieves gene ontology (GO) data both locally and from the Blast2GO database from a variety of sources including the National Center for Biotechnology Information (NCBI) nucleotide database, GO mapping tables, and Protein Information Resource protein database. The set of transcript assemblies was compared against NCBI’s nonredundant (nr) amino acid database by BLASTX. We separated sequences that originated from the tomato plant from those of fungal origin. In addition, all sequences were aligned with GMAP (Wu and Watanabe, 2005) to the draft tomato whole genome sequence assembly version SL2.40 (The Tomato Genome Consortium, 2012; http://solgenomics.net/organism/Solanum_lycopersicum/genome [accessed 31 May 2012]). GMAP is a splice aware alignment program that rapidly enabled us to confirm those assemblies whose best hits were to tomato and plant genes as being tomato in origin and map some assemblies to tomato that had no hits in the sequence database. The GMAP alignment data also provided further evidence that those assemblies whose best hit was fungal in origin were not found in the tomato genome. Mycorrhizal associated sequences identified from wild-type roots grown in field soil were searched against the Glomus intraradices genome assembly “test 14” and the MIRA expressed sequence tag (EST) assembly version 2.0 available at the French National Institute for Agricultural Research Glomus intraradices genome database (http://mycor.nancy.inra.fr/IMGC/GlomusGenome/index3.html (accessed 30 June 2011)).

Polymerase Chain Reaction Analysis Methods

Quantitative real-time reverse transcription PCR (RT-PCR) was performed on the individual biological replicate samples (n = 3) as described in Ruzicka et al. (2010). Supplemental Table S1 contains sequences of gene-specific primer pairs. Supplemental Table S2 contains the summary relative expression information for each sequence assembly assayed by real-time PCR (qPCR). Tomato actin was used as the reference control gene to normalize for root tissue input. The relative expression of the various target genes was analyzed according to the ΔΔ cycle threshold (CT) method, and standard error was computed from the average of the ΔCT values for each biological sample (n = 6) (Livak and Schmittgen, 2001).


Results

To characterize the AM fungal transcriptome as well as the tomato transcriptome changes that are related to colonization of the root system by mycorrhiza, we compared RNA transcript profiles in field-grown wild-type tomato plants. The roots sampled were colonized by the AM fungi (wild-type) naturally occurring in an organically managed field and compared to rmc tomato plants with reduced mycorrhizal colonization (rmc) grown in the same location. In addition, root RNA isolated from wild-type tomato plants grown in sterile soil in a greenhouse (wild-type myc-) were sampled and used as a reference to identify tomato genes not expressed in rmc due to the mutation that limits colonization. Ribonucleic acid was isolated from the three root sample groups, converted to cDNA, and sequenced on the Roche 454-FLX sequencer. Over 1.15 million sequence reads were obtained from nearly 361 million bp across all three sample groups. These were assembled into 30,063 contig sequences after filtering contaminants (see Materials and Methods and Supplemental Table S3). The contig assemblies have an average length of 757 bp and a range of 91 to 5359 bp. After contig assembly, the sample origin of the input reads for each contig was deconvoluted and then analyzed. The read-count comparison among the three sample groups is illustrated in the Venn diagram in Fig. 2A. The comparison between wild-type and rmc root samples identified 2210 fungal and tomato sequence assemblies identified from mycorrhizal roots.

Figure 2.
Figure 2.

Transcriptome analysis of mycorrhizal wild-type, nonmycorrhizal (NM) rmc mutant, and NM wild-type myc- (without mycorrhiza) root samples. Venn diagram (A) displays the contributing source of sequence reads to each of the 30,063 total sequence assemblies. Pie charts display the annotated species of each of the 30,063 total sequence assemblies (B) and of each of the 2210 mycorrhizal specific assemblies (C). Scatter plots display the relationship between rmc and mycorrhizal wild-type reads for all of the plant (D) and fungal (E) sequence assemblies. The best-fit linear trend line, slope equation, and R2 value are displayed for each.

 

The sequence assemblies were annotated and functional categorization was inferred with Blast2Go (Conesa et al., 2005) (see Materials and Methods and Supplemental Table S3). Eighteen thousand four hundred seventy-one of the 30,063 total contigs had a significant hit to the NCBI nonredundant nucleotide database (61.4%). Of these 18,471 contigs, 17,650 aligned best to plant sequences while 762 aligned best to fungal sequences. Because approximately 39% or 11,592 of the total contigs did not have a significant match in the NCBI nr database, all contig sequences were also aligned to the draft tomato whole genome assembly (version SL2.40) (The Tomato Genome Consortium, 2012). Ten thousand two hundred twenty-five of these 11,592 contigs aligned to the tomato genome, suggesting a large number of novel tomato transcripts were identified and correctly assembled de novo. Of the remaining 1375 assembled contigs without significant alignments to available sequence collections, 982 of these contigs were identified and found only in mycorrhizal roots (zero reads in rmc and wild-type myc-). After aligning all sequences to NCBI nr and tomato whole genome databases, 92.7% aligned to tomato sequences, 2.5% to fungal sequences, and 4.6% remained unidentified (Fig. 2B).

Of the 2210 contigs that were specific to AM roots, 448 (20.3%) exhibited very high sequence similarity to fungal sequences in the NCBI nr database (Fig. 2C). These included genes categorized as transporters, mineral nutrient metabolism genes, cell wall metabolism genes, and general carbohydrate and lipid metabolism genes (Table 1). Table 2 highlights the read-count data that was used to quantify the expression of fungal sequences annotated as transporters and includes putative Fe, Zn, adenosine triphosphate-binding cassette (ABC), and diverse major facilitator superfamily (MFS) transporters. This list of 2210 contigs identified in mycorrhizal roots is significantly enriched in putative fungal sequences compared to the percentage of putative fungal sequences found in the complete contig sequence dataset (20.3 vs. 4.1%, χ2 P-value < 0.0001). While excluded from the list of fungal transcripts, the 983 AM associated transcripts with no hit in the NCBI nr database (Fig. 2C) could represent novel orphan genes from Glomeromycota species or perhaps other microbial species associated with plant roots that require the functional rmc allele. All 27,867 plant and 762 fungal sequence assemblies were plotted on scatter plots to visualize the relationship of reads from wild-type vs. rmc root samples (Fig. 2D and 2E). Sequence assemblies with an equal number of reads from each root sample would be expected to exhibit a ratio of 1 and be plotted on the diagonal while sequence assemblies specific to the mycorrhizal root sample would be plotted along the horizontal x axis (y = 0). As expected, the majority of plant sequences display an expression ratio close to 1 (slope of the best-fit line = 1.02, R2 value = 0.845, P < 0.0001) (Fig. 2D). The scatter plot of fungal sequences exhibits a high number of mycorrhizal root-specific sequence assemblies (slope = 0.027, R2 = 0.007, P = 0.02) (Fig. 2E). This graphical display of sequence assembly reads between the mycorrhizal and NM root samples provides a visualization further supporting the enrichment of mycorrhizal-specific sequences in the fungal dataset.


View Full Table | Close Full ViewTable 1.

Functional categorization based on gene ontology annotation of the 448 mycorrhizal root-specific fungal sequence assemblies.

 
Functional category No. contig sequences Percent of mycorrhizal specific contigs Selected gene annotations of interest†
Cell growth and division 2 0.43% Cell division cycle family member and cell polarity protein
Cell structure 24 5.17% Cytoskeleton proteins and histone family members
Disease defense 13 2.80% Ferrooxidoreductase and Cu-Zn superoxide dismutase
Energy 18 3.88% Malate dehydrogenase, NADP-dependent malic enzyme, and isocitrate dehydrogenase
Intracellular trafficking 5 1.08% Synaptobrevin vesicle-associated membrane protein and Rab GDP-dissociation inhibitor
Metabolism 140 30.17% Inorganic pyrophosphatase, chitin synthase export chaperone, ornithine aminotransferase, hexokinase, chitin synthase, and glycerol dehydrogenase
Protein storage and destination 50 10.78% 14-3-3 Family protein
Secondary metabolism 6 1.29% Cytochrome p450
Signal transduction 26 5.60% GTP-binding protein, RHO GTPase, and G-protein complex alpha subunit
Transcription 10 2.16% RNA polymerase, RNA binding, and zero DNA-binding transcription factors found
Translation 91 19.61% Ribosomal proteins, initiation and elongation factors
Transporter 31 6.68% See Table 2
Unknown 48 10.34%
NADP, nicotinamide adenine dinucleotide phosphate; GDP, guanosine diphosphate; GTP, guanosine triphosphate; RNA, ribonucleic acid.

View Full Table | Close Full ViewTable 2.

Fungal mycorrhizal root-specific transporter sequence assemblies tested by real-time reverse-transcription polymerase chain reaction (qRT-PCR) to validate results from 454 sequencing.

 
Gene annotation† Contig identification Glomus intraradices genome hit (1 = yes and 0 = no) Total reads Wild-type reads rmc reads Putative substrate
MFS carboxylic acid transporter contig02160 0 124 123 1 Organic acid or carbohydrate
Membrane Zn transporter contig05640 0 11 10 1 Zn
Phosphate-repressible Na+ phosphate cotransporter contig05781 0 14 13 1 Phosphate
MFS nicotinic acid transporter contig01830 1 8 8 0 Coenzyme precursor
Transmembrane protein GPR1, FUN34, and yaaH family contig02702 1 26 26 0 Undefined or acetate
Mitochondrial carrier protein contig07972 1 2 2 0 Undefined or coenzyme A
Inorganic P-repressible Na+ phosphate cotransporter contig10052 1 10 10 0 Phosphate
Mitochondrial phosphate carrier protein 2 contig11495 1 2 2 0 Phosphate
MFS nicotinic acid transporter contig11875 1 10 10 0 Coenzyme precursor
GPR1, FUN34, and yaaH family transmembrane protein contig12669 1 7 7 0 Undefined or acetate
Mitochondrial phosphate carrier protein contig15671 1 5 5 0 Phosphate or mitochondrial
Voltage-dependent ion-selective channel contig15872 1 2 2 0 Ion or mitochondrial
Nitochondrial phosphate carrier protein contig24447 1 5 5 0 Phosphate
MFS glucose transporter contig03656 0 9 9 0 Carbohydrate
MFS d-galactonate or pantothenate transporter liz1 contig08298 0 4 4 0 Organic acid or carbohydrate
MFS high-affinity nicotinic acid transporter contig10741 0 6 6 0 Nicotinic acid
ABC metal ion transporter contig11511 0 2 2 0 Ion
Mitochondrial dicarboxylate carrier contig11578 0 3 3 0 Organic acid or carbohydrate
Choline transport protein contig12672 0 3 3 0 Ion, lipid, and amino acid
Amino acid permease contig12757 0 3 3 0 Amino acid
MFS quinate transporter contig16036 0 2 2 0 Amino acid precursor
MFS high-affinity nicotinic acid transporter contig16425 0 3 3 0 Coenzyme precursor
General amino acid permease gap1 contig16959 0 4 4 0 Amino acid
ABC transporter cdr4 contig17103 0 3 3 0 Undefined
High affinity Fe permease transporter contig26941 0 3 3 0 Fe
Plasma membrane Fe permease contig02193 0 11 11 0 Fe
MFS, major facilitator superfamily; ABC, adenosine triphosphate-binding cassette.

To further test whether these transcripts originated from AM fungal species, the 448 fungal sequences identified only in the mycorrhizal roots were subsequently aligned to the EST and whole genome sequence databases from the AM species Glomus intraradices (Tisserant et al., 2011). Fifty-one percent (228) of these align to Glomus intraradices sequences (BLASTN e-value < 10–5 and score > 25) and therefore were subsequently classified as bona fide AM fungal transcripts. The number of nonoverlapping transcripts is not unexpected given that the field soil where the tomato mycorrhizal roots were grown contained diverse AM fungi species whose gene sequences would not align to Glomus intraradices sequences and the physiology of field-grown roots could result in significantly different gene expression in the AM fungi as compared to previous studies that used pot-grown plants (Tisserant et al., 2011). The remaining 220 fungal mycorrhizal root-specific transcripts were conservatively classified as putative AM fungal transcripts. Two hundred ninety-eight fungal transcript sequence assemblies contained sequence reads obtained from both wild-type and rmc samples. These were subsequently aligned to the Glomus intraradices EST and whole genome sequence databases to test whether fungal sequences found both in wild-type and rmc roots may also have originated from AM fungi species. Seventy-seven of these contigs (30%) aligned to Glomus intraradices sequences, suggesting that some AM fungal transcripts were found in common to both wild-type and rmc root samples. Arbuscular mycorrhizal fungi transcripts from intraradical hyphae that have penetrated the root cortex or extraradical hyphae that surround the root itself would be present in rmc samples. Highly conserved gene sequences from non-AM fungi species are also likely to align to Glomus intraradices database sequences. Thirty-nine of these 77 sequences encode highly conserved sequences including 40S and 60S ribosomal protein translation factors.

This study also identified tomato genes only expressed in the tomato mycorrhizal root. There were 774 tomato contigs composed of reads that were found exclusively in the wild-type mycorrhizal root samples. The list of 774 tomato mycorrhizal root-specific contigs was subsequently filtered to only include sequence assemblies with at least eight reads in wild-type to reduce false positives (genes with low expression that falsely appear to be mycorrhizal root specific). Conversely, sequences with at least 10 reads in wild-type and with 10-fold induction in mycorrhizal roots representing tomato genes strongly induced by the AM symbiosis were added to the list of filtered mycorrhizal root-specific tomato genes to produce a final list of 100 tomato contigs referred to throughout as “mycorrhiza-induced” (Table 3; Supplemental Table S3). Among these 100 transcript assemblies, only five were previously identified in tomato mycorrhiza microarray studies (Fiorilli et al., 2009; Ruzicka et al., 2012), demonstrating the sensitivity and coverage benefits of deep sampling with next-generation sequence technologies. The five genes that overlapped with previous studies were phosphate transporter 3 (contig02966), cytochrome p450 CYP733A1 (contig02137), glucuronosyl transferase (contig28650), indole-3-acetic acid-amido synthetase GH3.3 (contig03103), and subtilisin-like protein (contig09522). To test for false positives in the 454 data set among the tomato and fungal mycorrhiza induced and specific gene lists, we used quantitative RT-PCR to assay the expression of 20 mycorrhiza-induced tomato transcripts and 54 mycorrhiza-specific fungal transcripts (Supplemental Tables S1 and S2). All 20 (100%) tomato sequences were confirmed to only be expressed in wild-type roots, suggesting that our filtering criteria for tomato genes were stringent and likely to have underestimated the number of mycorrhiza-induced tomato genes reported in this study. Forty-six of the 54 (85%) fungal sequences were confirmed to exhibit mycorrhizal root-specific expression indicating that perhaps as high as 15% of false positives remain in the list of 448 mycorrhizal root-specific fungal transcripts. This may also be due to the inclusion of sequence assemblies with very few total reads.


View Full Table | Close Full ViewTable 3.

Functional categorization of the postfiltered mycorrhizal root-specific and 10-fold induced tomato sequence assemblies.

 
Functional category No. contig sequences Percent of mycorrhizal induced contigs Selected gene annotations of interest†
Cell growth and division 0 0.00%
Cell structure 1 1.01%
Disease defense 9 9.09% PRp27-like secretory protein, early nodulin 55-2, and glutathione-S-transferase
Energy 0 0.00%
Intracellular trafficking 0 0.00%
Metabolism 20 20.20% Class iii and class iv chitinases, photoassimilate responsive protein, acyl:CoA ligase, lipid transfer protein, and autoinhibitied H+ atpase
Protein storage and destination 13 13.13% Ubiquitin extension protein, ubiquitin conjugating enzyme, and serine peptidase
Secondary metabolism 6 6.06% Cytochrome p450 704A14, 716A1, and 733A1
Signal transduction 1 1.01% Nodulation receptor kinase
Transcription 0 0.00%
Translation 0 0.00%
Transporter 8 8.08% Phosphate transporters 3, 4, and 5; PIP aquaporin; ammonium transporters 4 and 5; and ABC transporters
Unknown 41 41.41%
CoA, coenzyme A; PIP, plasma membrane intrinsic protein; ABC, adenosine triphosphate-binding cassette.

To identify potential mycorrhizal root-specific splice variants, the tomato contig sequences were aligned to The Institute for Genome Research unigene sequence database (http://compbio.dfci.harvard.edu/cgi-bin/tgi/gimain.pl?gudb=tomato [accessed 25 Nov. 2012]) (Quackenbush et al., 2001) and draft tomato genome sequence (version SL1.03; http://solgenomics.net/organism/Solanum_lycopersicum/genome [accessed 31 May 2012]). Because sequence assembly results in contigs that are nonidentical by design, multiple contigs that align to the same tentative consensus (TC) unigene and a unique genome locus may represent splice variants. Identification of contigs that exhibit this behavior but that have different expression patterns may represent mycorrhizal-specific splice variants. A set of contigs that aligned to unigene TC243593 (annotated as ubiquitin conjugating enzyme and meloidogyne-induced giant cell protein) exhibited very different expression patterns from one another. Regions of unigene TC243593 spanned by contig29628 and contig29644 were detected in wild-type mycorrhizal, rmc, and wild-type myc- root tissue while regions of unigene TC243593 spanned by contig02787 were only detected in mycorrhizal wild-type root samples. Further investigation using the tomato whole genome sequence identified two additional tomato transcript contigs that are specific to mycorrhizal root that mapped to the genomic region adjacent to TC243593. By mapping these five contig sequences to the corresponding tomato genome sequence, we found that transcripts from all three tissue samples shared the 5′ untranslated region (UTR), coding exons, and stop codon sequence regions (Fig. 3A). However, the transcript specifically found in wild-type mycorrhizal roots contained a 1700 nucleotide extension of the 3′ UTR beyond the poly-A tail of the constitutive transcript region. When mapped to the genomic sequence, this long mycorrhizal-specific 3′ UTR sequence also spanned an intron. Primers were designed to amplify the transcripts found in AM or NM root samples (Fig. 3B) and RT-PCR confirmed the expression specificity of these alternative transcripts (Fig. 3C).

Figure 3.
Figure 3.

Alternative splicing specific to the mycorrhizal root identified by mapping sequence assemblies to the tomato whole genome sequence. A. Gene model of ubiquitin conjugating enzyme (TC243593) sequence assemblies mapped to the tomato genome demonstrating 3′ untranslated region (UTR) alternative splicing between mycorrhizal wild-type and rmc samples. The 5′ UTR, coding sequence (CDS), and 3′ UTR regions are displayed as colored blocks and introns as thin lines. Primers were designed to span the extended 3′ UTR of the mycorrhizal-specific transcript (product 1) as well as the constitutive region (product 2) to test for the different isoforms in all three ribonucleic acid populations using reverse transcription polymerase chain reaction (B).

 

Discussion

The rationale for transcriptome sequencing rather than using microarrays was based on the expectation that sequencing provides a sensitive assay and that currently available tomato microarrays provide only limited coverage of the tomato transcriptome. In addition, we expected that RNA sampling and sequencing of AM roots would identify AM fungal transcripts, including AM fungal genes necessary for establishing and maintaining the symbiotic association. Roche 454 sequencing was used to enable longer sequence read lengths whose de novo assembly produced over 30,000 contiguous sequences in this study that aligned to either the tomato draft whole genome or known plant or fungal sequences in the public domain. Only approximately 4.5% of all sequence assemblies from this study could not be aligned or annotated. Because the sequenced RNA populations originated from roots grown in active soil ecosystems, these unaligned sequences may represent novel transcripts from root colonizing microorganisms.

Our analyses identified over 1400 nonplant mycorrhizal root sequences including those annotated as fungal (448) and those displaying no homology to any known sequence (982). The 448 fungal assemblies included 26 sequences similar to known transporter genes (Table 2). Of these, 10 were similar to Glomus intraradices transporter genes reported in Tisserant et al., suggesting a certain degree of overlap in the two studies (Table 2) (Tisserant et al., 2011). Mineral uptake from beyond the rhizosphere and transport to the plant host is an essential function of AM fungi; therefore, these additional sequences represent a new and deeper look into what molecules may play a role in the symbiosis. Of particular interest are the seven MFS transporter sequences. Major facilitator superfamily gene members constitute a diverse group of membrane transporters found across all kingdoms whose substrates include mono-, di-, and oligosaccharides as well as inorganic and organic cations and anions (Pao et al., 1998). The fungal sequences found only in mycorrhizal roots were most similar to MFS members that transport sugars and organic acids, and the transport of sugars and other C sources from the plant root to the fungus across the symbiotic interface is a key function of the mycorrhizal symbiosis. Previous work identified hexose as the major form of sugar transported from the plant root to the fungus in vivo, and a recent report identified a Glomus spp. monosaccharide transporter critical to the symbiosis (Helber et al., 2011; Pfeffer et al., 1999; Shachar-Hill et al., 1995). The MFS genes identified in this study did not exhibit sequence identity above 61% to those reported previously (Helber et al., 2011; Schussler et al., 2006), with the exception of one partial assembly (Contig13369, which is 282 bases in length) that exhibits 79% identity to a Glomus intraradices sugar transporter (MST3: gb|HQ848964.1), suggesting these may be new MFS sequences from different subfamilies. It remains to be seen whether organic molecules in addition to hexose such as dicarboxylates are also transported from the plant to the fungus; however, the number of diverse mycorrhizal root-specific fungal MFS genes reported here (Table 2) suggests the possibility of multiple forms of carbohydrates being transported.

There have been reports of the role AM symbiosis plays in plant tolerance to high Zn soils (Audet and Charest, 2006; Hildebrandt et al., 2007) as well as in improving plant nutrition under low Zn (Cavagnaro, 2008; Chen et al., 2003; Subramanian et al., 2008) and low Fe conditions (Caris et al., 1998; Wang et al., 2007, 2008). The sequence assembly contig05640 encodes a putative fungal Zn transporter that is 83% similar to the Zn transporter EAU29689 from Aspergillus terreus. Contig05640 is only 17% similar to the Zn-induced Glomus spp. Zn transporter (EB741033) proposed to function in mycorrhizal Zn detoxification (Hildebrandt et al., 2007). This serves to illustrate the high degree of sequence diversity observed within Zn transporters across species (Simm et al., 2011). This new putative fungal Zn transporter, along with two fungal Fe transporters (contig26941 and contig02193), may play roles in mineral nutrition in mycorrhizal plants in agreement with the increased Zn and Fe levels often observed in mycorrhizal plants (Cavagnaro et al., 2008).

The analysis identified eight mycorrhiza-induced tomato transporter genes including three phosphate transporters (contig02966, contig27416, and contig01354), two ammonium transporters (contig01385 and contig04044), an aquaporin (contig01860), and two ABC transporters (contig 04420 and contig 04135). The three phosphate transporter sequences encode previously identified mycorrhizal specific phosphate transporters PT3, PT4, and PT5, respectively (Gomez-Ariza et al., 2009; Nagy et al., 2005). The two ammonium transporter sequences encode mycorrhizal specific ammonium transporters AMT4 and AMT5 (Ruzicka et al., 2010) that share a high degree of similarity to M. truncatula and Lotus corniculatus L. var. japonicus Regel mycorrhiza-specific ammonium transporters (Gomez et al., 2009; Guether et al., 2009). Two tomato ABC transporters were also identified in the tomato mycorrhiza-induced dataset that have already been reported in the literature. Contig04420 is 76% similar to the M. truncatula ABC transporter STR, and contig04135 is 79% similar to the M. truncatula ABC transporter STR2. Both STR and STR2 are required for the AM symbiosis in M. truncatula and form a heterodimer that localizes to the periarbuscular membrane (Zhang et al., 2010). The mycorrhizal-specific aquaporin sequence identified here belongs to the major intrinsic protein (MIP) plasma membrane intrinsic protein (PIP) gene family and is similar to the constitutively expressed Gossypium hirsutum L. (cotton) X intrinsic protein (XIP) 1;1 (Park et al., 2010). Previous studies have suggested a potential role for aquaporins in the symbiosis (Uehlein et al., 2007) but had not identified mycorrhizal-specific aquaporin gene members. The beneficial effect of the AM symbiosis on plant drought tolerance has been well documented (Porcel et al., 2006), and this novel mycorrhizal-specific aquaporin may play an important role in this mechanism. In summary, it should be noted that seven of the eight mycorrhiza-induced tomato transporters identified in the present study have either been previously reported in tomato or are homologous to previously reported mycorrhizal-specific sequences in other plant species. Given the spectrum of molecules that have been hypothesized to be transported across the periarbusclar membrane and the lack of additional novel plant transporters identified in this study, it is reasonable to propose that some constitutively expressed plant transporters may also function in facilitating nutrient transfer at the periarbuscular membrane in mycorrhizal plants.

Of the 19 tomato mycorrhiza-induced metabolism genes identified in this study, 15 were related to carbohydrate, cell wall, or lipid metabolism. This list includes seven tomato chitinase and three endoglucanase genes. Previous studies have identified transitory increases in chitinase and glucanase activity in mycorrhizal roots (García-Garrido and Ocampo, 2002), and there is one report of a mycorrhizal specific class iii chitinase identified in Medicago truncatula (Salzer et al., 2000). Chitinases are important defense response genes in plants. Their gene products degrade cell walls of pathogenic fungi and may control intraradical fungal growth. The identification of multiple mycorrhizal-specific chitinases and glucanases suggests that fungal wall remodeling is an important process in the establishment or maintenance of the symbiosis. It has been suggested that these chitinases may function in suppressing the plant defense reactions at later stages of AM development by degrading chitin fragments from fungal cell walls that would otherwise elicit host defense responses (Salzer et al., 2000). The identification of multiple chitin and glucan biosynthesis genes supports the proposition that plant oligosaccharide remodeling enzymes function in coordination with fungal cell wall remodeling in the arbuscule.

Transcription factors, translation factors, cell structure, and cell growth and division factors were all underrepresented or entirely absent from this set of tomato mycorrhiza-induced genes, suggesting that these cellular functions may not require novel mycorrhizal-specific gene members. These processes may not be stimulated by the symbiosis or these classes of genes could be posttranscriptionally regulated in response to the symbiosis. Considering that transcription factors typically display relatively low expression levels, it is also possible that the conservative expression threshold filters imposed on the tomato mycorrhiza-induced dataset missed genes with overall lower expression levels. To address this, we expanded our analysis to include tomato contigs specific to wild-type mycorrhizal roots with at least four sequence reads (instead of 8) as well as tomato contigs five- to ninefold induced in wild-type mycorrhizal roots compared to rmc and wild-type myc- (instead of 10-fold). The 175 additional tomato genes identified by this search included additional transporter, carbohydrate and lipid metabolism, and transcription factor genes. Given these results, a more comprehensive survey using the Illumina HiSeq platform to provide deeper sequencing coverage and additional biological replicates would aid in quantifying more mycorrhiza-induced and mycorrhizal-repressed genes beyond the conservative thresholds imposed in this study.

The functional complexity of the transcriptomes of higher eukaryotes, including plants, is dramatically increased via the posttranscriptional regulatory mechanism of alternative splicing (AS) (Barbazuk et al., 2008; Reddy, 2007). Alternative splicing can generate multiple transcripts from a single gene, thereby increasing protein diversity (Kemal, 2003), and can alternatively reduce gene expression by generating aberrant transcripts that are degraded by nonsense-mediated mRNA decay (NMD) mechanisms (McGlincy and Smith, 2008). Earlier studies have identified AS in plants in response to abiotic and biotic stresses (Dinesh-Kumar and Baker, 2000) as well as a single report of AS involved in legume root nodulation (Combier et al., 2008). The identification of an alternatively spliced transcript between mycorrhizal and NM tomato root samples suggests AS likely plays a role in either the establishment and/or maintenance of the AM symbiosis or in the response to physiological differences between mycorrhizal and NM roots (e.g., phosphate starvation response). The presence of an extra-long 3′ UTR and a 3′ UTR intron is suggestive of posttranscriptional regulation via an NMD mechanism (Kebaara and Atkin, 2009; Kertész et al., 2006). Another possibility is that the alternative long 3′ UTR serves as a recognition site for microRNA-mediated translational regulation (Gandikota et al., 2007) although a search of the alternative splice sequence at miRBASE did not identify any significant hits (Kozomara and Griffiths-Jones, 2011). The core gene sequence expressed in both mycorrhizal and NM roots appears to be conserved across plant species including the nonmycorhizal model plant Arabidopsis thaliana (L.) Heynh. However, the splice variant region is only found in Solanaceae family species, suggesting its function may be specific to this plant family. Moreover, the functional annotation of this tomato gene sequence as a root-knot nematode (Meloidogyne incognita)-elicited giant cell protein suggests it may play a key role in diverse plant–microbe interactions (Bird and Wilson, 1994).


Conclusions

Despite the advent of metagenomic analysis of soil microbial communities in the last decade, challenges remain to link these metagenomic datasets with relevant ecological processes (Simon and Daniel, 2011). In this study we applied a metatranscriptomic approach to unravel functional aspects of the AM fungal–plant root symbiosis through identification of gene transcripts in field-grown AM roots. Gene-centric approaches to ecogenomic studies have in past successfully identified gene functions that are enriched in certain microbial environments and provide useful insights into the microbial community interaction with the environment (Larsen et al., 2011; Tringe et al., 2005). Our results demonstrate that deep sequencing and de novo assembly can be used to conduct metatranscriptome analysis of a complex sample such as mycorrhizal roots grown in an ecologically relevant context. Using the well-studied rmc mutant and wild-type tomato plants provided an opportunity to identify new aspects of the transcriptome in an obligate biotroph in its natural symbiotic environment through the identification of hundreds of putative fungal and plant mycorrhizal root-specific genes involved in diverse processes including transport and cell wall remodeling. Alignment of tomato sequences to the tomato genome map also identified a mycorrhizal-specific transcript splice variant. Future metatranscriptome and plant–fungus interaction studies will reveal more of the biological processes surrounding this symbiosis including the transcriptional responses of the AM fungal and plant symbionts to environmental change or sustainable agricultural management. This approach will improve our understanding of the role of these new genes in the symbiosis and the role of fungal microbes in the soil community.

Supplemental Information Available

Supplemental material is available at http://www.crops.org/publications/tpg.

Supplemental Table S1. The mycorrhizal root-specific tomato contig statistics and annotations.

Supplemental Table S2. Primer sequences for real-time reverse-transcription PCR (qRT-PCR).

Supplemental Table S3. The mycorrhizal root-specific fungal contig statistics.

Author Contributions

Daniel P. Schachtman, Louise E. Jackson, Daniel Ruzicka, and W. Brad Barbazuk conceived of the study and its design and drafted the primary manuscript. Additional text and discussion of the research was provided by Felipe H. Barrios-Masias, Louise E. Jackson, Sally Smith, and Francis Martin. Tissue samples, RNA isolations, PCR, and library preparation sequencing were done by Daniel Ruzicka, Felipe H. Barrios-Masias, and University of Florida Core Facilities. Sequence management, data assemblies, and other analyses were done by Daniel Ruzicka, Srikar Chamala, Francis Martin, and W. Brad Barbazuk. All authors contributed to and approved the final manuscript.

Acknowledgments

This work was supported by funds from the National Science Foundation grant #0723775 to D.P.S, W.B.B, and L.J. and by funds from USDA-NRI grant # 2007-35300-19739 to W.B.B and by the University of Florida to W.B.B. We are very grateful to Susan Barker for providing rmc seed as well as insightful discussion on the analysis.

 

References

Footnotes


Comments
Be the first to comment.



Please log in to post a comment.
*Society members, certified professionals, and authors are permitted to comment.