Characterization of microsatellites and gene contents from genome shotgun sequences of mungbean (Vigna radiata (L.) Wilczek)

Background Mungbean is an important economical crop in Asia. However, genomic research has lagged behind other crop species due to the lack of polymorphic DNA markers found in this crop. The objective of this work is to develop and characterize microsatellite or simple sequence repeat (SSR) markers from genome shotgun sequencing of mungbean. Result We have generated and characterized a total of 470,024 genome shotgun sequences covering 100.5 Mb of the mungbean (Vigna radiata (L.) Wilczek) genome using 454 sequencing technology. We identified 1,493 SSR motifs that could be used as potential molecular markers. Among 192 tested primer pairs in 17 mungbean accessions, 60 loci revealed polymorphism with polymorphic information content (PIC) values ranging from 0.0555 to 0.6907 with an average of 0.2594. Majority of microsatellite markers were transferable in Vigna species, whereas transferability rates were only 22.90% and 24.43% in Phaseolus vulgaris and Glycine max, respectively. We also used 16 SSR loci to evaluate phylogenetic relationship of 35 genotypes of the Asian Vigna group. The genome survey sequences were further analyzed to search for gene content. The evidence suggested 1,542 gene fragments have been sequence tagged, that fell within intersected existing gene models and shared sequence homology with other proteins in the database. Furthermore, potential microRNAs that could regulate developmental stages and environmental responses were discovered from this dataset. Conclusion In this report, we provided evidence of generating remarkable levels of diverse microsatellite markers and gene content from high throughput genome shotgun sequencing of the mungbean genomic DNA. The markers could be used in germplasm analysis, accessing genetic diversity and linkage mapping of mungbean.


Background
Mungbean (Vigna radiata (L.) Wilczek) is an important food leguminous crop in Asia, with an annual production of around 3.5 -4.0 million tons [1]. The crop is grown principally for its protein-rich dry seeds (24% protein) which is a major protein source for people in Asian countries as part of a nutritionally balance diet [2]. It is popularly grown as a component in various cropping systems because of its ability to fix nitrogen in association with soil bacteria, early maturity (ca. 60  Mungbean is a self-pollinated diploid plant with 2n = 2x = 22 chromosomes and a genome size of 515 Mb/1C [3]. Genomic study in this crop is far behind other legume crops. Mungbean was among the primary crops that genetic linkage maps have been developed. However, the current linkage maps, based on RFLP and RAPD markers of mungbean, do not resolve 11 linkage groups [4]. Microsatellites or simple sequence repeats (SSRs) are markers of choice for crop improvement of many species because they are reliable and easy to score [5]. SSRs are clusters of short tandem repeated nucleotide bases distributed throughout the genome. SSR markers are co-dominant, multi-allelic and requiring small amount of DNA for scoring. The traditional method of SSR marker development involves construction of SSR-enriched library, cloning, and sequencing, which is costly and labor intensive. Nevertheless, significant efforts have been invested in development of SSR markers in recent years, but so far only 35 polymorphic SSR markers published for mungbean [6][7][8][9][10]. In a study by Somta et. al. (2008), more than 200 primer pairs amplifying SSRs were tested for polymorphism among 17 mungbean accessions, only 12 (5.7%) primer pairs were polymorphic. The authors suggested that the use of SSR markers has been limited due to the lack of polymorphism in this species [7].
Over the past few years, the introduction of a massivelyparallel pyrosequencing technology developed by 454 Life Sciences Technology has opened new possibilities for high-throughput genome analysis [11]. This new technology has been applied to the sequencing of microbial genomes, genotyping, genome resequencing, transcriptome profiling and methylation studies. Although, sequences generated by this technique are relatively short, there are evidences suggesting that this technique can be used to sequence plant genomes that are complex and large [12][13][14]. Wicker et al. (2006) suggested that 454 sequencing technology could reveal almost complete assembly of the entire gene sequences in 4 barley BAC clones at only 9-folds coverage and concluded that the method is a rapid and cost effective way of sequencing the gene-containing portions of the genome. Low coverage shotgun sequencing using 454 sequencing technology has also been used to study functional genomics in soybean [12], repetitive DNA in the pea genome [14] and transcriptome from a normalized cDNA library of Medicago truncatula [15]. Here, we report genome shotgun sequencing of the mungbean genomic DNA using 454 Life Sciences sequencing technology for isolation of SSR markers and characterization of gene content.

Results and Discussion
Shotgun sequencing of Vigna radiata genome Sequencing of Vigna radiata genomic DNA was carried out using 454 Life Sciences technology on the Genome Sequencer (GS) FLX System. A total of 470,024 quality filtered sequence reads was generated with the average read length of 216 bases covering 100.5 Mb. All reads were deposited in NCBI Short Read Archive (ID = SRA003681) http://www.ncbi.nih.gov/Traces/. Assembly of the obtained nucleotide sequence reads was performed using the Newbler, de novo sequence assembly software [11]. Redundant reads were reduced to 46,646 contigs with the average contig length of 297 bases covering 13.85 Mb. The contig sequence data were reported in the DDBJ/EMBL/ GenBank nucleotide sequence databases with the accession number BABL01000001-BABL01046645. The contig length ranges from 89 bases to 44,462 bases. The average GC content of mungbean genomic DNA generated in this study is 34.69% which is consistent with the reports on GC contents in other plant genomes such as Arapbidopsis (36% [16]), grape (34.6% [17]), poplar (33.7% [18]), tomato (36.2% [19]) and potato (35.6% [19]). It is slightly higher than the mean of GC content for intergenic regions in the Arabidopsis genome (32.9%, Genome Indices 8/04: http://http//gi.kuicr.kyoto-u.ac.jp) [20]; but it is much lower than the average GC content of Arabidopsis coding sequences (44.5%) [21].

Characterization of polymorphic microsatellite markers in Vigna radiata
We isolated 1,493 microsatellite regions using the Troll software. There were 889 dinucleotide repeats (DNPs), 282 trinucleotide repeats (TNPs), 123 tetranucleotide repeats (TTNPs), 124 pentanucleotide repeats (PNPs) and 75 SSRs with hexanucleotide repeats or more. The distribution of the number of motif repeat ranged from 4 -30 repeats ( Table 1). The most common motif type of DNPs was TA/AT (89.3% of DNPs) followed by TC/AG (7.1% of DNPs) and AC/TG (3.6% of DNPs). The GC/CG motif was not found in the data set. TNPs were found at 282 SSR loci (18.9%), which was three times lower than that of DNPs. The TAA repeat was the most common motif type found at 184 loci (65.24% of TNPs). The least frequent TNP motif was GC-rich (GCG/CGC) found at only 2 loci. The genomic SSRs with GC-rich motif repeats are rare in most plants as previously reported in rice, corn, soybean [22], wheat [23], Arabidopsis thaliana, apricot, peach [24], coffee [25] and rubber tree [26]. In contrast, the GC-rich motifs have been reported as frequent motifs in studies on development of SSR from expressed sequence tags and genomes with methylation filtration [27][28][29][30]. Thus, GCrich SSR are most likely to be derived from the coding region of the genome. The frequency of identified SSR in mungbean was one SSR in every 67 kb (1,493 SSRs in 100.5 Mb) which is significantly lower than the SSR frequency in soybean (1/7.4 kb) [31]. Among plant species, the SSR frequencies range from 1/1.5 kb in coffee to 1/20 kb in cotton [25,31]. The observed low SSR frequency in this study is probably because a large proportion of reads from the low coverage sequencing (0.2x) of the mungbean genome were biased toward highly repetitive parts of the genome. From 1,493 identified SSRs, 192 SSRs were identified from contigs and 1,301 SSRs were from singletons. Among 192 contigs containing SSR motifs, majority of contigs were assembled from 2 reads (87 contigs) followed by 3 reads (48 contigs) and 4 reads (16 contigs) ( Table 2). By applying the Lander-Waterman model [32] to this dataset, there should be no contig assembled from more than 9 reads provided that all sequences were generated by chance from non repetitive DNA (Table 2). Therefore, 16 out of 192 contigs that were assembled from more than 9 reads are likely to represent repetitive sequences of the genome. It should be noted that loci present in multiple copies are not desirable for construction of genetic maps. Interestingly, there was a highly repetitive contig containing SSR (contig 44495) which was assembled from 3,174 raw reads. Sequence homology search revealed that contig 44495 is a fragment of the chloroplast genome. The number of chloroplast genome of higher plants can reach hundreds of copies per cell. Due to the deep sequencing nature of 454 technology, it is expected to obtain a large number of reads from sequences with multiple copies such as organellar genomes, transposons and ribosomal DNA [12]. The degree of sequencing over-representation in a repetitive  genome can be estimated from the difference between the observed read coverage and the predictions from the Lander-Waterman model ( Table 2) as suggested by [12]. It should be noted that the number of observed contigs with assembled reads = 2 was much lower than the prediction by the model. This was probably due to the effect of low sequencing coverage; thus it was not included in the calculation of the number of repetitive reads. In total, there were 241,410 reads (51%) present in multiple copies. We estimated that 51% of shotgun reads from 0.2× genome coverage represented repetitive DNA. This estimate is slightly more than the result from the DNA re-association kinetic study which estimated 46% of the total leaf DNA as repetitive sequences [33].
To evaluate these SSR loci in further detail, we designed 192 primer pairs to amplify all SSR loci identified from the contig data set. Among the 192 primer pairs evaluated in 17 mungbean accessions, 179 (93.23%) primer pairs were amplifiable and 127 (66.14%) primer pairs produced scorable bands. Of these, 58 primer pairs targeting 60 loci revealed polymorphism because 2 primer pairs, VR257 and VR400, were able to target 2 independent loci for each primer pair. Characteristics of all 60 loci are summarized in Additional File 1. These primer pairs were able to detect a range of 2 to 6 alleles with a mean of 2.6833 alleles per locus. Polymorphic information content (PIC) values ranged from 0.0555 to 0.6907 with an average of 0.2594 which is similar to the previous studies [7,34]. In this study, there were 33 pair-wise combinations that significantly deviated from linkage disequilibrium (LD). Genetic variation at a given locus in a population is meas-  [6,7,34]. The low level of heterozygosity and significant deviation from HWE are probably because mungbean is a highly self-pollinated species with an estimated outcrossing rate of only 1.1% [35].
We also tested the SSR locus in the highly repetitive contig 44495, which was a fragment on the chloroplast genome. The VR0453 locus, located in the non-coding region near the atpB gene in the chloroplast genome, had 2 alleles and showed relatively low PIC value of 0.1046 (see Additional File 1). Chloroplast microsatellites have been used in ecological and evolutionary studies, especially at the intraspecific level, because they are nonrecombinant, uniparentally inherited and effectively haploid [36]. However, the major barrier for utilization of chloroplast microsatellites is the low mutation rates associated with the chloroplast genome [37] leading to low polymorphism level of markers in the chloroplast genome.
Sequence homology search of other loci against the Genbank non-redundant protein database and the TIGR plant repeat databases [38] revealed that there were 5 loci (VR029, VR073, VR216, VR256 and VR323) matched  (Figure 1). The transferability rates of mungbean primers were between 45.80% (V. subterranean) and 91.60% (V. angularis). However, the amplification rate was reduced in Phaseolus vulgaris and Glycine max to 22.90% and 24.43%, respectively ( Figure 1). Transferability rate of mungbean genomic microsatellite markers to other Vigna species appeared to be more or less similar to previous studies. Somta et al. (2009) reported that amplification of genic microsatellite markers in 19 taxa of Vigna species was between 80% (V. aconitifolia) to 95.3% (V. reflex-pilosa) [39]. Whereas, Chaitieng et al. (2006) reported that amplification of azuki bean (V. angularis) microsatellite markers in V. mungo, V. radiata, V. aconitifolia and V. umbellata was between 68.8 to 90.2% [40]. The high amplification rates of both mungbean and azuki bean Cross-species amplification of 127 mungbean microsatellite markers in various species from genus Vigna, Phaseolus and Glycine   V a c V a n V a n ( w il d ) V s n V u n -U n g V u n -S e s P h a G ly % amplification microsatellite markers in Vigna species indicate high genome homology among species in this genus and are useful for genetics and genomics studies, especially genome mapping and comparative genomics.

Phylogenetic relationship
To determine the genetic diversity structure and relationships between 35 genotypes of 20 taxa of Asian Vigna, polymorphism scores at 16 microsatellite loci without missing data were used (see Additional File 2). UPGMA cluster analysis was conducted using software NTSYSpc 2.2 [41]. Results from the cluster analysis revealed that all the genotypes of Asian Vigna could be clearly differentiated and classified into two groups; mungbean group and azuki bean group (Figure 2). The results were in agreement with previous studies using non-coding sequences of trnT-F [42,43]. In contrast, studies using AFLP [44], rDNA-ITS and atpB-rbcL sequences [45] recognized three groups within the Asian Vigna. In addition, it is worth noting that V. nepalensis, which has similar morphology [46] and close genetic relationship with V. angularis [43,45], was found to be highly distinct in our study. V. grandiflora previously shown to have high morphological and genetic similarity to V. radiata [46,47] was found to have closer genetic relationship with V. trilobata and V. stipulacea than V. radiate in this study. Also, V. subramaniana that was reported to be closely related to mungbean [45] appeared to be more distant from mungbean but more closely related to V. aridicola in our study. It should be noted that V. subramaniana has a complex taxonomic history, controversy in the literature and classification concerning the taxonomy of this species still remains [48]. The differences in the phylogenetic relationship of Asian Vigna may be explained by the differences in the methods used in the previous studies. Morphological traits [46], rDNA and cpDNA sequences [43,45] were used in previous studies to demonstrate phylogenetic relationship, while our study used SSR markers for demonstration. The use of PCRbased SSR markers may possibly result in size homoplasy of PCR products between/among species [49]. The same A dendrogram depicting genetic diversity and relationships among 35 genotypes from 20 taxa of Asian Vigna as revealed by the polymorphism of 16 mungbean microsatellite markers allele size of an SSR locus may contain different sequence variants; thus species sharing the same SSR allelic size include species that are identical by descent and species that have originated from convergent evolution.

Sequence annotation and gene ontology
The contigs were analyzed by GeneMark.hmm eukaryotic version 3.3 [50] to predict Open Reading Frame (ORF) using Medicago trunculata as a model organism and default parameter conditions. Results from GeneMark predicted a total of 44,112 ORFs. For functional annotation, the potential coding regions were analyzed by BLAST2GO [51] leading to consistent gene annotations, assigning gene names, gene products, EC numbers and Gene Ontology (GO) numbers. Gene Ontology provides a system to categorize description of gene products according to three ontologies: molecular function, biological process and cellular component. Sequence homology search revealed that there were 1,542 ORFs matches with non-redundant protein database with an E-value cut-off at E-6. Nine hundreds and fifty sequences were mapped to one or more ontologies with multiple assignments possible for a given protein within a single ontology. There were 647 assignments made to the molecular function ontology, with a large proportion of these in catalytic (42.72%) and binding activities (44.17%) categories (Figure 3a). Under the biological process ontology, 555 assignments were made with a large proportion of assignments fell into metabolic process and cellular process (such as secretory pathway, transcription and translation) categories (Figure 3b).

Similarity of mungbean predicted ORFs with other plant ESTs
To identify gene functions, the mungbean contig set was blasted (TBLASTX) to identify ESTs encoding similar proteins, at an e-value cutoff at E-6, against other plant gene indices collected in The Gene Index Databases, Dana Farber Cancer Institute, such as soybean (GMGI, 13.0), Arabidopsis (AGI, 13.0), rice (OGI, 17.0), M. truncatula (MTGI, 9.0) and Vitis vinifera (VVGI, 6.0) [52]. The number of sequences that showed similarity to encoding sequences is shown in Figure 4. Comparison between the mungbean dataset and the Glycine max gene index gave the highest number of matched sequences (7,940 sequences). V. radiata and G. max are grouped together as tropical season legumes or Phaseoloid exhibiting extensive genome conservation based on previous comparative genetic mapping [53,54]. The other Papilionoideae legume, M. truncatula, which is a cold season legume, also shares a large number of homologous sequences (5,759 sequences) with the mungbean dataset. A. thaliana and V. vinifera gave lower number of matched sequences to the mungbean dataset; 4872 and 4,949 sequences respectively. The lowest number of matched sequences (1,971 sequences) was observed when the mungbean dataset was blasted against the Oryza sativa gene index, the only monocot plant used in the comparison.

Discovery of microRNA
To predict functional non-coding RNA, such as micro-RNA, in the mungbean dataset, we made computational prediction of potential microRNA using MiRFinder to search for the potential hairpin-loop structure in their sequences [55]. Next we calculated the minimal folding free energy (MFE) using Sfold [56]. There were 2,247 microRNA candidates with MFE < -25 kcal/mol which were selected for further analysis. Then we blasted the mungbean microRNA candidates against previously known microRNAs from Arabidopsis, rice, and other plant species to search for potentially conserved microR-NAs. A total of 4 miRNA candidates had sequence homology with miR171, miR408, miR1171 and miR414, which have been shown to target genes coding for SCARE-CROW-like proteins implicated in radial root pattern [57], plantacyanin [58], copper chaperone [59] and translation initiation factor [60], respectively (Table 3).

Conclusion
The results provided by the present study highlight a reliable and efficient way in obtaining polymorphic microsatellite markers and characterization of putative genes using shotgun genome sequences of Vigna radiata. A significance of the results from this study is that highthroughput shotgun sequences of mungbean can be useful not only for marker development, construction of linkage map, mungbean genetic improvement, phylogenetic relationship, but also for gene discovery as the paucity of DNA markers in cultivated mungbean has precluded detailed genetic research on this crop. Gene Ontology classification of the predicted mungbean ORFs according to molecular function (a) and biological process (b) using BLAST2GO [51] with E-6 cutoff Figure 3 Gene Ontology classification of the predicted mungbean ORFs according to molecular function (a) and biological process (b) using BLAST2GO [51]with E-6 cutoff.

a b
Prediction of sequencing coverage in contigs from a completely non repetitive genome was calculated according to the Lander and Waterman model [32]. The number of contigs expected containing a number of reads j is given by equation 1.
Where N is the number of reads, L is the read length, G is the haploid genome size in base pairs, and T is the base pair overlap required for contig formation (in this case T = 40).

Isolation, amplification and transferability of SSR markers
In order to identify microsatellite markers, non-redundant sequences were screened for SSRs using TROLL software http://wsmartins.net/webtroll/troll.html. For the searches, we defined SSRs as being DNP ≥ 14 bases; TNP ≥ 15 bases; TTNP ≥ 16 bases; HNP (and more) ≥ 16 bases [31]. For comparison of SSRs in plant genomic sequences, we used the criteria of SSR motif of ≥ 20 bases [31,62,63]. Primer pairs were designed to amplify microsatellite regions using PRIMER3 [64]. PCR was carried out in a total volume of 10 μL containing 2 ng of DNA template, 1× Taq buffer, 2 mM MgCl 2 , 0.2 mM dNTPs, 1 U Taq DNA polymerase (Fermentas) and 0.5 μM each of forward and reverse primers. Amplification was performed in a Gene-Amp PCR 9700 System thermocycler (Applied Biosystems) programmed as follow: 94°C for 2 min followed by 35 cycles of 94°C for 30 s, 50-65°C for 30 s, 72°C for 1 min, and a final extension step at 72°C for 10 min. Amplified products were separated on 5% denaturing polyacrylamide gels and visualized by silver-staining.

Analysis of polymorphic loci
Seventeen mungbean genotypes as listed in Additional File 3 were used for polymorphism analysis of SSR markers. Details of primer pairs for SSR markers are listed in Additional File 4. Scoring data from polymorphic loci were used to calculate Polymorphism Information Content (PIC) [65], Hardy-Weinberg equilibrium (HWE) [66], pairwise linkage disequilibrium (LD) using chisquare test, and observed heterozygosity and expected heterozygosity using the PowerMarker 3.25 software [67].  [41].

Analysis of gene content and annotation
The mungbean contig set was analyzed in two parts which are 1) gene prediction/Gene Ontology (GO) term annotation and 2) functional gene identification. Gene-  Mark.hmm eukaryotic version 3.3 [50] based on Hidden Markov Models was used to predict coding sequence (cds) of the contig set using Medicago trunculata as a model organism and default parameter conditions. For the functional annotation, the potential coding sequences were analyzed by BLAST2GO [51]. To identify gene functions, sequence similarity search program-BLAST was used to identify ESTs encoding similar proteins of the mungbean contig set. All