Metatranscriptomic analysis of small RNAs present in soybean deep sequencing libraries.

A large number of small RNAs unrelated to the soybean genome were identified after deep sequencing of soybean small RNA libraries. A metatranscriptomic analysis was carried out to identify the origin of these sequences. Comparative analyses of small interference RNAs (siRNAs) present in samples collected in open areas corresponding to soybean field plantations and samples from soybean cultivated in greenhouses under a controlled environment were made. Different pathogenic, symbiotic and free-living organisms were identified from samples of both growth systems. They included viruses, bacteria and different groups of fungi. This approach can be useful not only to identify potentially unknown pathogens and pests, but also to understand the relations that soybean plants establish with microorganisms that may affect, directly or indirectly, plant health and crop production.

Introduction

Until recently, analysis of the microbial diversity in environmental samples was conducted only after isolation, culture and identification of microorganisms and subsequent sequencing of cloned libraries (Cardenas and Tiedje, 2008). However, these conventional methods are limited to the minority of species that can be cultured (Chistoserdova, 2010). New culture-independent methods, such as metagenomics and metatranscriptomics, have been developed (Xu, 2006; Cardenas and Tiedje, 2008; Adams ; Warnecke and Hess, 2009; Chistoserdova, 2010). These methods refer to studies of the collective set of genomes and transcriptomes of mixed microbial communities and may be applied to the exploration of all microorganisms that reside in marine environments, soils, human and animal clinical samples, sludge, polluted environment, and plants (Kent ; Zoetendal ; Adams ; Al Rwahnih ; Poretsky ; Shi ; Desai ; Gifford ; Roossinck ). The metagenomic approaches, including metatranscriptomics, involve the sequencing of random DNA or RNA-derived complementary DNA (cDNA) profiles and subsequent determination of taxonomic diversity and prospective genes related to response to environmental conditions (Rosen ). With the advent of new sequencing technologies (high-throughput sequencing), more data can be generated in a relatively short time, in a practical and cost-effective way (Creer ). Moreover, this approach allows the direct sequencing of DNA or cDNA, avoiding any cloning bias and leading to large-scale studies (Adams ).

Metagenomic analysis of RNA deep-sequencing data has been used in plant disease diagnostics, such as in grapevine (Al Rwahnih ; Coetzee ), sweet potato (Kreuze ), tomato and Liatris spicata (Adams ). Coupling of metagenomics with pyrosequencing in these studies has also allowed the detection of bacterial and fungal RNA, suggesting that this approach can contribute to massive identification of crop-associated microbiota, including pathogenic, symbiotic, and free-living organisms.

In addition to other sequencing methodologies, deep sequencing libraries of small RNA (sRNA) can contribute to microbial identification studies as they may include non-coding RNAs (e.g. rRNA), small regulatory RNAs, such as microbial sRNAs (Shi ) and host small interfering RNAs (Kreuze ), as well as mRNA fragments that normally would not be represented in libraries enriched for polyA-tailed mRNA commonly used in metatranscriptomic studies. As an example of the application of high-throughput sequencing of plant sRNAs, Kreuze , using short read sequences of approximately 24 base pairs (bp), successfully identified viruses infecting sweet potato, even those present in extremely low titer symptomless infections.

With this in mind, the current study describes the use of high-throughput sequencing of sRNAs to identify potential pathogens and other microorganisms from samples of soybeans grown in the field and in controlled-environment conditions.

Material and Methods

Plant material

The sRNA sequences analyzed in this study were obtained from deep-sequencing libraries from different projects related to the soybean transcriptome (Genosoja and Genoprot). The libraries were constructed from root samples of soybean cultivars grown under greenhouse conditions, and from flower, seed and pod samples from soybean plants grown under field conditions.

Root samples were obtained from ‘Embrapa 48’ and ‘BR 16’ soybean cultivars grown in a greenhouse at Embrapa-Soja in Londrina, Brazil. Plants were cultivated using a hydroponic system composed of plastic containers (30 liters) and an aerated 6.6 pH-balanced nutrient solution. Seeds were pre-germinated on moist filter paper in the dark at 25 °C ± 1 °C and 65% ± 5% rhr. Plantlets were then placed in polystyrene supports with the roots of the seedlings completely immersed in the nutrient solution. Each tray containing the seedlings was maintained in a greenhouse at 25 °C ± 2 °C and 60% ± 5% rh under natural daylight (photosynthetic photon flux density (PPFD) = 1.5 x 103 μmoles m−2 s−1, equivalent to 8.93 x 104 lux) and a 12:12 h photoperiod. Roots of seedlings with the first trifoliate leaf fully developed (V2 developmental stage) were frozen in liquid nitrogen and stored at −80 °C until RNA extraction.

Three stages of seed germination, pods, and mature seeds from the soybean cultivar ‘Conquista’ were also used in RNA extraction. In a growth chamber, seeds were incubated for 3, 5 and 7 days on rolls of moistened filter paper at 26 °C. Pods (R3-R5) were collected from field plants grown at the Federal University of Rio Grande do Sul (UFRGS), Porto Alegre, Brazil.

Flower samples (R2-R3) were collected from the soybean cultivar ‘Urano’ grown at the experimental field of the University of Passo Fundo (UPF), in Passo Fundo, Brazil. Collected flowers were immediately powdered in Trizol (Invitrogen, CA, USA) and stored until RNA extraction.

RNA extraction and sequencing

Total RNA was isolated from roots, seeds, seedlings, pods and flowers using Trizol (Invitrogen, CA, USA), and following the manufacturer’s instructions. RNA quality was evaluated by electrophoresis in 1.0% agarose gels, and the amount checked using a Quibit fluorometer and QuantiT RNA assay kit (Invitrogen, CA, USA) according to the manufacturer’s instructions. Approximately 10 μg of total RNA were sent to Fasteris Life Sciences SA (Planles-Ouates, Switzerland) for processing and sequencing using Solexa technology on an Illumina Genome Analyzer GAII. Briefly, the processing by Illumina consisted of the following successive steps: acrylamide gel purification of RNA bands corresponding to the size range 20–30 nucleotides (nt), ligation of 3′ and 5′ adapters to the RNA in two separate subsequent steps, each followed by acrylamide gel purification, cDNA synthesis, and a final step of PCR amplification to generate a DNA colony template library for Illumina sequencing. After removing vector sequences, reads of 19 to 24 nt were used for further analysis.

Sequence analysis

The detailed analysis of endogenous sRNAs, including microRNAs, obtained from the soybean libraries described above is the topic of a separate study and will not be discussed here. However, since in addition a large number of sRNAs unrelated to the soybean genome were identified, a metagenomic analysis was indicated to identify the origin of these sequences. To this end, all reads were assembled into contigs using the Velvet 0.7.31 de novo assembly algorithm (Zerbino and Birney, 2008) with the following parameters: hash length of 23, coverage cut-off of 50, expected coverage of 1,000, and a minimum scaffold length of 100.

Assembled contigs matching the soybean genome were eliminated from further analysis using BLAST (BLASTn) “stand alone” version 2.2.20 against the soybean genome database available at Phytozome with the following parameters: expectation value (−e) of 1e−10, cost to open a gap (−G) of −6, cost to extend a gap (−E) of −6, penalty for a nucleotide mismatch (−q) of −5. The remaining contigs were used to search the NCBI database using BLASTn (nucleotide blast) with default parameters and an expectation value of 10−5. The contigs were classified according to the sequence with the highest hit score found with BLASTn. The number of reads aligning to the contigs, coverage and average depth were determined with the SOAP tool (Li ). Default parameters were used and only filtered data (reads aligning to the references with a high confidence) are reported.

Results

The sRNA libraries analyzed in this study contained 5,627,802 reads (each consisting of 19–24 bp) from root, 8,610,347 from seeds and pods, and 9,314,206 from flower (Figure 1). The sRNA sequences were assembled into contigs ranging from 40 to 300 nucleotides, approximately. Sequence assembly produced 2,646, 15,521 and 28,382 contigs from root, seed and pod, and flower, respectively. After elimination of the soybean sequences, 253 (root), 2,574 (flower) and 1,959 (seed and pod) contigs remained for further analysis. These contigs were used in BLASTn searches against the NCBI database.

Figure 1

Chart flow of sequence analysis. The central column identifies the methods. The number of reads obtained at each step is indicated for samples obtained from a controlled environment (left) and from field samples (right).

After BLASTn annotation, a large number of sequences remained unidentified (Figure 1), accounting for 73.4% of the field samples (Figure 2A) and 37.2% of the controlled environmental samples (Figure 3A). Contigs that corresponded to soybean sequences, but could not be filtered by the local BLASTn, represented 17.4% of the total contigs from field samples (Figure 2A) and 5.1% from controlled environment samples (Figure 3A). There were also contigs corresponding to sequences from other plant species deposited in NCBI (Figures 2A and 3A).

Figure 2

Comparative percentages of the contigs assembled from sRNA (19–24 nt) libraries of the field (crop soybean) for each sequence classification according to best hit with BLASTn. (A) All groups of organisms, (B) all groups excluding unidentified and plants and (C) best identified taxonomic groups. Bean Pod Mottle Virus (BPMV).

Figure 3

Comparative percentages of the contigs assembled from sRNA (19–24 nt) libraries of the soybean cultivated in a controlled environment for each sequence classification according to best hit with BLASTn. (A) All groups of organisms, (B) all groups excluding unidentified and plants and (C) best identified taxonomic groups.

Apart from the contigs mentioned above, 134 contigs from the controlled environment (root) and 335 from the field (flower, seed and pod) had hits to previously sequenced microorganisms and viruses (Figure 1) and provided the results shown in the following topics. These contigs were distributed in different taxonomic groups based on their best BLASTn hits (Figures 2B and 3B). Contigs showing similarities to sequences from different taxonomic groups (multiple affiliations) were classified at the taxonomic level immediately above.

Taxonomic origin of sRNAs from controlled environmental samples

The most represented taxon in the controlled environment samples belongs to the domain Eukaryota (49.3%), including unicellular eukaryotes (46.3%) and fungi (3.0%), followed by the domain Bacteria (32.8%). In addition, 17.9% of the microbial contigs could only be classified to the taxonomic level of the domain Eukaryota (Figure 3B).

Unicellular eukaryote organisms were found only in controlled environment samples. The amoeba genus Naegleria was well represented (12 contigs) within the kingdom Excavata, with one contig classified to species level, viz. Naegleria fowleri (Table 1, Figures 3C and 4C). Contigs corresponding to the phylum Kinetoplastida were distributed at different levels of taxonomic classification through the family Bodonidae, including Neobodo designis and Rhynchomonas nasuta. The family Trypanosomatidae was also identified within this phylum, including the genus Trypanosoma. Within the kingdom Chromalveolata (group Heterokontophyta), the class Chrysophyceae was identified through two contigs. The genera Chrysolepidomonas and Spumella (class Chrysophyceae) were represented by one and seven contigs, respectively. One contig was assigned and classified as pertaining to Spumella elongata. The genus Mallomonas was also identified within the kingdom Chromalveolata (group Heterokontophyta).

Table 1

Taxonomic affiliation of the 134 contigs assembled from libraries obtained from plants grown in a controlled environment (root samples).

Domain	Kingdom	Organism	Environment	No. of contigs	Identity (%)	Sequence type
Eukaryota		Fungi/Unicellular eucaryotes	Soil, plants, water, humans and animals	25	81–100	18S rRNA, 26S rRNA, 28S rRNA and Genomic DNA
Eukaryota	Chromalveolata	Stramenopiles	Water and soil	5	98–100	28S rRNA and 18S rRNA
Eukaryota	Chromalveolata	Mallomonas sp. (algae)	Water	2	94–100	ITS1-5.8 rRNA-ITS2-28S rRNA
Eukaryota	Chromalveolata	Chrysophyceae (phytoplankton)	Marine, lake and river water	2	100	28S rRNA
Eukaryota	Chromalveolata	Spumella sp.(flagellate)	Water	6	98–100	18S rRNA
Eukaryota	Chromalveolata	Spumella elongate (flagellate)	Water	1	100	18S rRNA
Eukaryota	Chromalveolata	Chrysolepidomonas sp. (flagellate)	Water	1	98	Genomic DNA
Eukaryota	Excavata	Kinetoplastida	Water	2	100	28S rRNA
Eukaryota	Excavata	Trypanosomatidae	Marine, lake and river sediments	3	90–100	24S α rRNA and genomic DNA
Eukaryota	Excavata	Trypanosoma sp.	Marine, lake and river sediments	1	100	18S rRNA, ITS1, 5.8S
Eukaryota	Excavata	Bodonidae	Marine, lake and river sediments	3	100	18S rRNA
Eukaryota	Excavata	Rhynchomonas nasuta (flagellate)	Water and soil	1	100	18S rRNA
Eukaryota	Excavata	Neobodo sp. (flagellate)	Marine, lake and river sediments	2	95–98	18S rRNA
Eukaryota	Excavata	Neobodo designis (flagellate)	Marine, lake and river sediments	8	100	18S rRNA
Eukaryota	Excavata	Naegleria sp. (amoeba)	Water and soil	12	91–100	18S rRNA, 28S rRNA, extrachromosomal rRNA plasmid DNA
Eukaryota	Excavata	Naegleria fowleri (amoeba)	Water and soil	1	100	18S rRNA
Eukaryota	Rhizaria	Cercozoa (ameboflagellates)	Soil and water sediments	2	89–100	18S rRNA and 28S rRNA
Eukaryota	Rhizaria	Cercomonas sp. (ameboflagellates)	Soil and water sediments	4	100	28S rRNA
Eukaryota	Rhizaria	Gymnophrys sp. (ameboflagellates)	Soil and water sediments	1	100	28S rRNA
Eukaryota	Viridiplantae	Pyramimonas sp. (microalgae)	Marine, lake and river water	3	95–96	28S rRNA
Eukaryota	Viridiplantae	Chlorococcum sp. (microalgae)	Marine, lake and river water	1	100	28S rRNA
Eukaryota	Fungi	Fungi	Soil, plant, human, animal, insect	1	94	18S rRNA
Eukaryota	Fungi	Acaulospora sp. (arbuscular mycorrhiza)	Plant roots	1	87	18S rRNA
Eukaryota	Fungi	Rozella sp.	Endoparasite of other fungi	1	92	28S rRNA
Eukaryota	Fungi	Dikarya	Soil, plant, human, animal, insect	1	98	26S rRNA
Bacteria		Bacteria	Soil, plants, water, humans and animals	10	92–100	16S rRNA and 23S rRNA
Bacteria		Proteobacteria	Soil, plants, water	2	97–100	16S rRNA and 23S rRNA
Bacteria		Alphaproteobacteria	Soil, plants, water	2	100	16S rRNA
Bacteria		Rhizobiales	Soil and plant root nodules	1	96	Genomic DNA
Bacteria		Mesorhizobium sp.	Soil and plant root nodules	1	98	23S rRNA
Bacteria		Rhizobium sp.	Soil and plant root nodules	1	100	16S rRNA
Bacteria		Bradyrhizobiaceae	Soil and plant root nodules	1	100	16S rRNA
Bacteria		Bosea sp.	Soil and plants	1	100	16S rRNA
Bacteria		Sphingomonadaceae	Soil	1	100	16S rRNA
Bacteria		Sphingobium sp.	Soil	1	100	23S rRNA
Bacteria		Caulobacteriaceae	Water	1	100	23S rRNA
Bacteria		Phenylobacterium sp.	Water	1	98	23S rRNA
Bacteria		Betaproteobacteria	Soil, plants, water	4	100	16S rRNA and genomic DNA
Bacteria		Methylophilaceae	Soil	1	98	23S rRNA
Bacteria		Laribacter sp.	Water	1	95	Genomic DNA
Bacteria		Oceanospirillales	Lakes, rivers and sea hypersaline and highly alkaline.	1	97	Genomic DNA
Bacteria		Burkholderiaceae	Soil and plants	1	100	23S rRNA
Bacteria		Burkholderia sp.	Soil and plants	1	96	Genomic DNA
Bacteria		Ralstonia solanacearum	Soil and water	1	100	Genomic DNA
Bacteria		Comamonadaceae	Water	1	100	16S rRNA and 23S rRNA
Bacteria		Verminephrobacter sp.	Lumbricid earthworms (Lumbricidae nephridia)	1	96	Genomic DNA
Bacteria		Delftia acidovorans	Soil and water	1	100	16S rRNA
Bacteria		Leptothrix sp.	Aquatic environments with sufficient organic matter	1	90	23S rRNA
Bacteria		Pseudomonas sp.	Plant leaves	1	100	16S rRNA
Bacteria		Ignatzschineria larvae	Insect mid gut (adult flesh fly)	1	100	23S rRNA
Bacteria		Bacterioidetes	Soil, plants, water	1	100	Genomic DNA
Bacteria		Flectobacillus sp.	Water	1	98	16S rRNA
Bacteria		Cytophaga hutchinsoni	Soil and plant debris (cellulose degradation)	3	100	Genomic DNA
Total				134

Figure 4

Taxonomic levels of microorganisms distributed among the libraries of small RNAs in soybean cultivated in a controlled environment and in the field. (A) Bacteria (B) Fungi and (C) other Eukaryota.

Microalgae belonging to kingdom Plantae were represented by the genera Chlorococcum and Pyramimonas. The phylum Cercozoa belonging to kingdom Rhizaria was represented by six contigs, with four pertaining to the genus Cercomonas and one to the genus Gymnophrys.

The kingdom Fungi was poorly represented in the controlled environment libraries (Figures 3B, C). One contig was assigned to the genus Rozella and another one to the genus Acaulospora (Figure 4B). We also found a contig with 94% of identity to an uncultured soil fungus and one with multiple affiliations within the subkingdom Dikarya (Table 1).

The domain Bacteria was the most diverse taxon in controlled environment samples (Figures 3B,C). Contigs affiliated with multiple taxa could be classified only to the domain Bacteria, phyla Bacterioidetes and Proteobacteria, class Alphaproteobacteria and families Sphingomonadaceae, Bradyrhizobiaceae, Caulobacteriaceae, Burkholderiaceae and Comamonadaceae.

The classes Alpha- and Gammaproteobacteria were equally represented, followed by Betaproteobacteria. Within the class Alphaproteobacteria, contigs corresponded to the order Rhizobiales, including the genera Bosea, Rhizobium and Mesorhizobium. There were contigs assigned to families Caulobacteriaceae and Sphingomonadaceae, including the genera Phenylobacterium and Sphingobium, respectively (Figure 4A).

Within the class Gammaproteobacteria, contigs were derived from the order Burkholderiales, families Burkholderiaceae (including Burkholderia and Ralstonia solanacearum) and Comamonadaceae (including the genus Verminephrobacter and the species Delftia acidovorans) and genus Leptothrix. The group Oceanospirillales was represented by a single contig. Within the group Pseudomonadales, the genus Pseudomonas was the only one to be identified with a single contig. The group Xanthomonadales was represented by the species Ignatzschineria larvae (Figure 4), and the phylum Bacterioidetes by the species Cytophaga hutchinsoni and genus Flectobacillus, both members of the family Cytophagaceae (Figure 4).

Further contigs were classified at the genus leve and shared high identity (at least 87%) to nucleotide sequences of species deposited in the NCBI database (Supplementary Material, Table S1). These may belong to the same or a closely related species, viz. the unicellular eukaryotes Naegleria gruberi, Mallomonas asmundae and Chrysolepidomonas dendrolepidota, the bacteria Phenylobacterium lituiforme, Laribacter hongkongensis and Leptothrix cholodnii, and the fungus Acaulospora scrobiculata.

Taxonomic origin of sRNAs from field samples

The BLAST analysis revealed that 87.2% of the contigs assembled from field samples were related to BPMV (Bean Pod Mottle Virus) (Figure 2B). These contigs were assembled from seed and pod samples and showed identities ranging from 94 to 100% for both RNA1 and RNA2 sequences from the virus (Table 2). The remaining sequences were fungi (9.9%) and bacteria (3.0%).

Table 2

Taxonomic affiliation of the 335 contigs assembled from libraries obtained from soybean plants grown in the field (flowers, pods, mature seeds and samples of germinating seeds).

Domain	Kingdom	Organism	Environment	No. of contigs	Identity (%)	Sequence type
Eukaryota	Fungi		Soil, dung, plant, decaying wood, fungal, insect, human	1	100	18S rRNA
Eukaryota	Fungi	Dikarya	Soil, dung, plant, rock, decaying wood, fungal, insect, human	3	100	18S, ITS1, 5.8S rRNA, ITS 2, 28S rRNA
Eukaryota	Fungi	Basidiomycota	Plant, humans, animals, water, soil	4	100	18S rRNA and 28S rRNA
Eukaryota	Fungi	Agaricomycetes	Plant, plant debris	1	100	28S rRNA
Eukaryota	Fungi	Xanthophyllomyces sp. (yeast)	Plant	1	94	28S rRNA
Eukaryota	Fungi	Tremellales	Human, parasites of other fungi	2	100	5.8S rRNA gene, ITS2, 26S rRNA
Eukaryota	Fungi	Trichosporon sp. (yeasts)	Soil, human	1	100	18S rRNA
Eukaryota	Fungi	Dioszegia sp. (yeasts)	Plant	2	100	26S rRNA, 18S rRNA
Eukaryota	Fungi	Pezizomycotina	Plants, soil, debris plant, rock	9	100	18S rRNA gene, ITS1, 5.8S rRNA, ITS 2, 28S rRNA
Eukaryota	Fungi	Sordariomycetes	Soil, dung, plant, decaying wood, fungal, insect, human	1	100	28S rRNA
Eukaryota	Fungi	Dothideomycetes	Plants, soil, debris plant, rock	1	100	28S rRNA
Eukaryota	Fungi	Capnodiales	Plants, soil, debris plant, rock	2	97–100	18S rRNA and 28S rRNA
Eukaryota	Fungi	Mycosphaerellaceae	Plants, soil, debris plant	3	100	18S rRNA and 28S rRNA
Eukaryota	Fungi	Cladosporium sp.	Plants, soil, debris plant	2	98–100	18S rRNA
Bacteria		Bacteria	Plant, humans, animals, water, soil	1	100	16S rRNA
Bacteria		Rhizobiales	Soil and root plant nodules	1	100	16S rRNA
Bacteria		Enterobacteriaceae	Plant, humans, animals, water, soil	4	100	16S rRNA, 23S rRNA and genomic DNA
Bacteria		Buchnera aphidicola	Insect	1	100	Genomic DNA
Bacteria		Burkholderia sp.	Soil, plant, human	1	100	Chromosome 2
Bacteria		Streptomyces sp.	Plant	1	100	16S rRNA
Bacteria		Bacillales	Plant, humans, animals, water, soil	1	100	23S rRNA
Aphanabionta		Bean pod mottle virus	Plant	292	94–100	RNA1, soy RNA1-b polyprotein, RNA2, putative defective coat protein, capsid polyprotein
Total				335

Within the kingdom Fungi, contigs were classified according to the subkingdom Dikarya and phylum Basidiomycota. One contig was assigned within the phylum Basidiomycota, subphylum Agaricomycetes, and two contigs were assigned to the order Tremellales. Within the order Tremellales, two contigs represented the genus Dioszegia and one contig the genus Trichosporon (Figure 4B). In addition, one contig was not identical but showed high identity (94%) to the species Xanthophyllomyces dendrorhous and could be classified only in the genus Xanthophyllomyces (Table 2) which belongs to the class Tremellomycetes.

The phylum Ascomycota subphylum Pezizomycotina was the most represented fungal taxon (Figure 2C). It was identified through nine contigs that had multiple affiliations within this taxon, these being with the classes Sordariomycetes and Dothideomycetes. Within this latter there were contigs affiliated to the order Capnodiales, including the family Mycosphaerellaceae and the genus Cladosporium (Figure 4B).

A single contig was classified only at the domain level Bacteria. There were contigs to the class Gammaproteobacteria, one to the genus Burkholderia and four to the family Enterobacteriaceae, including one to Buchnera aphidicola. There were also contigs associated with the class Alphaproteobacteria (order Rhizobiales) and the phyla Firmicutes (order Bacillales) and Actinobacteria from the genus Streptomyces (Figures 2 and 4A, Table 2).

Discussion

High-throughput sequencing allows the direct sequencing of DNA or cDNA from the environment, avoiding any cloning bias and also being less costly and faster than the Sanger method (Cardenas and Tiedje, 2008). The Illumina (Solexa) technology is capable of generating 36 million reads with average length of 35 bp within 4 days, which is several times higher than the output of traditional cloning libraries. The short read lengths are compensated by massive output, speed, simplicity and coverage, including regions recalcitrant to cloning.

To obtain ribosomal sequences (commonly used in the identification of species) from complex microbial communities, classical polymerase chain reaction (PCR) has been used with primers complementary to highly conserved regions of rRNA (Rosen ), representing the bias of being a targeted approach (Bailly ).

Because the new sequencing technologies have the potential to sequence technically difficult regions, such as those that form firm secondary structures, they are useful in the analysis of rRNA (Cardenas and Tiedje, 2008).

The metagenomic analysis of sRNA high-throughput libraries, as done in this study, is a non-targeted approach that avoids amplification steps and the design of the ‘universal primers’. In this way, the present study provided several rRNA sequences that could be used in taxonomic identification (Tables 1 and 2).

Here, we demonstrate the feasibility of metatranscriptomics/metagenomics to study the microbial communities present in soybean plants cultivated in field and controlled environment. In this work, high-throughput sequencing generated several records for bacteria, fungi, unicellular eukaryotes, and viruses at different taxonomic levels (species, genus, family, order, class, phylum, kingdom or domain).

In root tissues of soybean plants cultivated in a controlled environment, three groups of organisms were detected, these being unicellular eukaryotes, bacteria and fungi. The unicellular eukaryotes were the most abundant group of organisms in root tissues of soybean cultivated in a controlled environment. Unicellular eukaryotes included some subgroups of eukaryote microorganisms, such as microalgae (genera Pyramimonas and Chlorococcum, these belonging to the division Chlorophyta, kingdom Plantae, and the genus Mallomonas belonging to the phylum Heterokontophyta or Stramenopiles, kingdom Chromalveolata), flagellates (including Neobodo designis, Rhynchomonas nasuta and the genus Trypanosoma, belonging to the class Kinetoplastida, kingdom Excavata, and the genus Spumella belonging to the phylum Heterokontophyta or Stramenopiles, kingdom Chromalveolata), amoeba (genus Naegleria, including N. loweri, belonging to the class Heterolobosea, kingdom Excavata), and ameboflagellates (genera Cercomonas and Gymnophrys belonging to the phylum Cercozoa, kingdom Rhizaria), which are free-living organisms in freshwater, soil and marine habitats (Bailly ). The fact that plants grown in a controlled environment were cultivated in a hydroponic system with nutritive solution could explain the high percentage of unicellular eukaryotes found in this habitat, since these organisms feed on the sediment of organic matter present in water.

The domain Bacteria was the second highest group in terms of occurrence in root samples, including endophytic and epiphytic bacteria. Species, genera, families, and orders related to growth promotion and nodulation through endosymbiosis with rhizobia (Juteau ; Kuklinsky-Sobral ; Delmotte ; Ikeda , 2010; Okubo ) were found in this work (Table S1). They include the genera Rhizobium, Mesorhizobium, Sphingobium, Burkholderia, Bosea and Pseudomonas, the families Bradyrhizobiaceae, Sphingomonadaceae and Rhizobiaceae, and the order Rhizobiales. We also found some species and genera involved in the degradation of cellulose in plant debris and that of organic matter in humid and freshwater habitats, such as the species/genera Cytophaga hutchinsoni, Delftia acidovorans, Ignatzschineria larvae, Laribacter (Woo ), Leptothrix and Phenylobacterium.

Ralstonia solanacearum is considered a bacterial pathogen. It causes wilt in important crops (including soybean) in other countries. Nevertheless, populations of Ralstonia solanacearum may occur as free-living microorganisms in watercourses or in a latent form in plants without causing disease (Grey and Steck, 2001; Mole ). The genus Pseudomonas identified in this study includes members that are pathogenic to soybean, such as P. syringae pv. glycinea, or non-pathogenic ones, surviving as saprophytes, epiphytes or endophytes (Kuklinsky-Sobral ).

Endophytic and epiphytic bacteria can contribute to the health, growth and development of plants. Promotion of plant growth by these bacteria may result from indirect effects, such as the biocontrol of soilborne diseases through competition for nutrients, siderophore-mediated competition for iron, antibiosis, or the induction of systemic resistance in the host plant. Direct effects, such as the production of phytohormones, providing the host plant with fixed nitrogen, or solubilization of phosphorus and iron present in soil, may also be of relevance (Kuklinsky-Sobral ).

The bacterial families Commamonadaceae and Caulobacteriaceae have members involved in sediment degradation in freshwater. Other groups of bacteria, such as members of the family Methylophilaceae and the order Oceanospirillales can survive in soil and humid habitats.

Fungi present in the roots of soybean plants were found to belong to the genus Rozella, these being involved in the biological control of other fungi in plants. The genus Acaulospora forms arbuscular mycorhiza in plants, thus increasing nutrient absorption from soil, mainly phosphorus.

In seeds, seedlings, pods and flowers from soybean plants cultivated in field (crop), three groups of organisms were identified: virus, fungi and bacteria. The virus group was represented by a single member only, the Bean Pod Mottle Virus (BPMV). This virus is widespread in the major soybean growing areas throughout Brazil and the world (Anjos ; Giesler ). 0Mottling originates at the hilum and is also referred to as “bleeding hilum” since the hilum color seems to bleed from its normal zone. From then on, the mottling of the seed has similar coloration as the hilum (Giesler ). BPMV was detected mainly in the mature seed library of sRNA, and seeds used in the RNA extraction procedure showed symptoms of mottling (data not shown), predicting the occurrence of BPMV infection.

The identified fungi were classified at species, genus, family, order, class, subphylum, and phylum levels. The yeast genera Xanthophyllomyces, Dioszegia and Thichosporon may occur in flower, seed, stem and leaf surfaces, and are essential for biological control of pathogens (Kucsera ; Wang ; Weber ). The genus Cladosporium found in this study in flower and seed tissues of soybean is known to contain entomopathogenic species with potential for biocontrol of pest insects (Pimentel ). The family Mycosphaerellaceae was detected in this work. It includes various genera, especially Cercospora, Pseudocercospora, Mycosphaerella, Septoria, Ramularia, etc. that represent more than 10,000 species. The genera Cercospora, Mycosphaerella and Septoria contain some species considered as pathogenic to soybean (Crous ). The subkingdom Dikarya, phylum Basidiomycota, subphylum Pezizomycotina, classes Dothideomycetes, Agaricomycetes and Sordariomycetes, and order Tremellales contain both phytopathogenic and nonpathogenic fungi species related to soybean, including species with potential for biocontrol of pest insects and diseases.

Moreover, in the field environment many bacteria were found associated with soybean tissue. The bacterium Buchnera aphidicola colonizes insects and may be present on plant surfaces. The genus Burkholderia includes endophytic species present in soil and plant (Kuklinsky-Sobral ). The family Enterobacteriaceae includes many endophytes, such as the genus Pantoea, which occur mainly in leguminous plants involved in biological control of phytopathogens (Delmotte ; Ikeda ). The orders Rhizobiales and Bacillales contain endophytes and species that colonize the rhizosphere and phyllosphere, and are related to soybean nodulation and biological control of pest insects, fungi and bacteria.

Several contigs could not be classified deeper into the eukaryote domain due to multiple affiliations or hitting to sequences from environmental samples (many from uncultured freshwater eukaryotes) (Table S1).

The high percentage of unknown sequences in this study could correspond to artifacts or to transcript fragments from poorly known taxa. Similarly, in many samples of a metagenomic study of plant viruses (Roossinck ), sequences without similarity to GenBank sequences represented more than half of the contigs. Furthermore, in a metatranscriptomic analysis of microbial communities from watercourse, half of the possible protein-encoding sequences from pyrosequencing had no significant hits to previously sequenced genes. The length of the contigs influenced this frequency, as the analysis of larger (> 200 bp) sequences resulted in twice the frequency of annotated sequences when compared to shorter (< 100 bp) reads (Poretsky ).

The sequencing of small RNA molecules in the size range of 19 to 25 was chosen as it corresponds to the sizes of small interfering RNAs (siRNAs) and virus derived interfering RNAs (vsiRNAs). These are normally produced when RNA interference mechanisms are activated in order to degrade endogenous or pathogen-derived RNAs. The size range of the sequenced small RNAs could be enlarged up to 100 nucleotides, in order to include other small RNA molecules originated by other processes of degradation. Sizes under 19 nucleotides should be avoided since they would have ambiguous matches in different genomes and would not be useful in species prediction.

The challenges that remain in metagenomics and metatranscriptomics include DNA and RNA extraction, low stability, abundance and proportion of mRNAs in total RNA extracts (Cardenas and Tiedje, 2008). In addition, many of the difficulties encountered in microbial diversity studies are due to the complexity of the microbial community and its unevenness (few populations are of high frequency and many populations of low abundance). These difficulties can be reduced in metatranscriptomics by focusing on the active populations in a sample (Morales and Holben, 2010). When analyzing a soil microbial community, Urich demonstrated that the deep-sequencing data from total RNA is naturally enriched in both functionally (such as mRNA libraries) and taxonomically relevant molecules, i.e. mRNA and rRNA, respectively.

Nevertheless, upon using deep-sequencing data, significant differences have been found in the taxonomic distribution of cDNAs derived from total RNA compared to DNA libraries from soil samples (Bailly ). Differences were also found when comparing cDNA derived from mRNA-enriched libraries with DNA libraries from marine microbial communities (Gilbert ). This suggests that both DNA and RNA sequences should be analyzed complementarily when investigating the diversity of species from environmental samples. In this sense, sRNA sequencing libraries can contribute to microbial identification with other types of sequences, viz. non-coding RNAs and mRNA fragments that would not normally be present in libraries enriched for polyA-tailed mRNA.