Background

In agriculture, productivity is affected by environmental conditions such as drought, salinity, high radiation and extreme temperatures faced by plants during their life cycle, that impose severe limitations to the growth and propagation, restricting their genetic potential and, ultimately, reflecting yield losses of agricultural crops. Although, advances have been achieved through classical breeding, further progress is needed to increase abiotic stress tolerance in cultivated plants. New gene targets need to be identified in order to reach these goals, requiring extensive studies concerning the biological processes related to abiotic stresses. Comparative analysis between primitive and related groups of cultivated species may shed some light on the understanding of these processes.

Microsatellites or SSRs (Simple Sequence Repeats) are sequences in which one or few bases are tandemly repeated, ranging from 1-6 base pair (bp) long units. They are ubiquitous in prokaryotes and eukaryotes, present even in the smallest bacterial genomes [1–3]. Variations in SSR regions originate mostly from errors during the replication process, frequently DNA Polymerase slippage. These errors generate base pair insertions or deletions, resulting, respectively, in larger or smaller regions [4]. SSR assessments in the human genome have shown that many diseases are caused by mutation in these sequences [5]. The genomic abundance of microsatellites, and their ability to associate with many phenotypes, make this class of molecular markers a powerful tool for diverse application in plant genetics. The identification of microsatellite markers derived from EST (or cDNAs), and described as functional markers, represents an even more useful possibility for these markers when compared to those based on assessing anonymous regions [6–8]. EST-SSRs offer some advantages over other genomic DNA-based markers, such as detecting the variation in the expressed portion of the genome, giving a ''perfect''marker-trait association; they can be developed from EST databases at no cost and unlike genomic SSRs, they may be used across a number of related species [9].

Many studies indicate UTRs as being more abundant in microsatellites than CDS regions [10]. In a study of micro- and minisatellite distribution in UTR and CDS regions using the Unigene database for several higher plants groups, higher occurrence of these elements in coding regions were found for all the studied species [11]. Disagreements between earlier reports and the later, reflect a deficiency in annotation when translated and non-translated fractions are separated in the Unigene transcript database. Dimer repeats were also frequent in CDS regions, which could be due to the fact that the Unigene database contains predominantly EST clusters. Therefore, there is a tendency for under-representing the UTR regions in the annotated sequences [11].

The characterization of tandem repeats and their variation within and between different plant families, could facilitate their use as genetic markers and consequently allow plant-breeding strategies that focus on the transfer of markers from model to orphan species to be applied. EST-SSR also have a higher probability of being in linkage disequilibrium with genes/QTLs controlling economic traits, making them more useful in studies involving marker-trait association, QTL mapping and genetic diversity analysis [9].

On model organisms, microsatellites have been reported to correspond to 0.85% of Arabidopsis thaliana (L.) Heynh, 0.37% of maize (Zea mays L.), 3.21% of tiger puffer (Takifugu rubripes Temminck & Schlegel), 0.21% of the nematode Caenorhabditis elegans Maupas and 0.30% of yeast (Saccharomyces cerevisiae Meyer ex. E.C. Hansen) genomes [10]. Moreover, they constitute 3.00% of the human genome [12]. All kinds of repeated element motifs, excluding trimers and hexamers, are significantly less frequent in the coding sequences when compared to intergenic DNA streches of A. thaliana, Z. mays, Oryza sativa subsp japonica S. Kato (rice), Glycine max (L.) Merr. (soybean) and Triticum aestivum L. (wheat) [10].

Close to 48.67% of repeat elements found in many species are formed by dimer motifs. In Picea abies (L.) H. Karst. (Norway spruce), for example, the dimer occurrence is 20 times more frequent in clones originating from intergenic regions vs. transcript regions [13]. Approximately 14% of protein translated sequences (CDS - coding sequences) contain repetitive DNA regions, and this phenomenon is 3 folds more frequent in eukaryotes than prokaryotes [14]. Clustering studies showing microsatellite occurrence in distinct protein families (non-homologous) from either prokaryotic or eukaryotic genomes, indicate that the origins of these loci occurred after eukaryotic evolution [14–16]. The highest and lowest repeat counts were found in rodents and C. elegans, respectively [3].

In plant species, some reports have described the levels of occurrence of microsatellites associated to transcribed regions [7, 8, 10, 11, 17–22]. However, some comparative and/or descriptive approaches, still can offer new perspectives on the features of these markers. Furthermore, frequently new groups of plant species have their genome sequenced, enabling the reassessment of databases using new sequences, representing divergent evolutionary groups and/or with different genetic models.

The online platforms for nucleotide, protein and transcript (ESTs) databases available for the majority of species are relatively small when compared with model species, eg Physcomitrella patens (Hedw.) Bruch & Schimp., O. sativa and A. thaliana. Since the protocols for the isolation of repetitive element loci, such as microsatellites, require intensive labour and can be expensive, the exploitation of these elements in silico on databases of model plants and their respective transfer to orphan species, is a potentially fruitful strategy.

In this study we present our results on the SSR survey for the development of plant SSR markers. The survey was based on clustered non-redundant EST data, their classification, characterization and comparative analysis in eleven phylogenetically distant plant species including two green algae, a hepatic, two mosses, two fern, two gymnosperms, a monocot and a dicot.

Results and Discussion

We analysed 560,360 virtual transcripts with the SSRLocator software (Table 1). The species with most abundant records in Genbank was Arabidopsis thaliana with 224,496 virtual transcripts (40%), followed by Oryza sativa with 121,635 (21.7%), Physcomitrela patens with 79,537 (14.19%), Pinus taeda with 58,522 (10.44%) and Chlamydomonas reinhardtii with 40,525 (7.2%). The remaining species added up to 11.7% of virtual transcripts analysed. When total genome sizes are compared for the model plants included in this analysis, the virtual transcripts of P. patens (511 Mb) represent 0.01% of genome size. For O. sativa (389 Mb) and A. thaliana (109.2 Mb) the ESTs analysed represent 0.02% and 0.18%, respectively, of the genome. The highest average bp count per EST sequence was found for Selaginella spp. (924 bp) followed by M. polymorpha (777 bp), C. reinhardtii (775 bp) and P. taeda (760 bp). The lower average bp per sequence was found for G. gnemon (563 bp) and A. capillus-veneris (580 bp). For the model plants, A. thaliana showed the lowest average bp count (321 bp), with P. patens and O. sativa presenting similar bp counts (737 and 755 bp, respectively). Shorter observed sequences could be an indication of incomplete representation of genes, but one must keep in mind that average gene sizes could vary among species, i.e., rice fl-cDNAs (1,747 bp) are 14% longer than Arabidopsis fl-cDNAs (1,532 bp) (TAIR 9 and RIKEN, accessed in 12.2.2010). The overall bp counts are very similar to those found by other authors [23].

Table 1 EST database size and Overall occurrence of SSR, percentages and average length motifs per specie
Full size table

The frequency of SSR per EST database was higher (4.66%) in Selaginella spp virtual transcripts (Table 2). For model plants, 3.57% and 0.84% SSRs/EST were found for O. sativa and A. thaliana, respectively.

Table 2 EST database size and Overall occurrences of SSRs, percentages and average length motifs per species
Full size table

The average motif length, excluding compound SSRs, was 27.03 bp. Mesostigma EST database shows the longest SSR average size with 34.13 bp, and the shortest size was found for Marchantia polymorpha with 22.56 bp mean size. The SSR size for model plants was similar. For P. patens, O. sativa and A. thaliana, average sizes of 24.2, 23.4 and 26.5 bp were found, respectively. A total 1,106 EST sequences contained more than one SSR. Among the species, O. sativa and P. patens are on the extremes of the distribution with 37.34% and 3.46% of virtual transcripts containing one or more microsatellites. However, Adiantum capillus-veneris EST database contained the highest percentage of transcripts displaying more than one SSR (20.86%) based on the database size. Similar results were found in our group [11], using the Unigene database for grasses and other allies. In the same study, rice was shown to have the highest frequency of ESTs containing more than one SSR (11.28%). In the present study, a similar value was found for rice (10.20%). These small differences could be due to different redundancy reduction parameters used in Unigene species database and CAP3 default settings. Other reports for higher plants [19, 20, 24–26], showed different ranges, but never higher than 2-3 fold. The variations encountered in different reports are related to the strategy employed by investigators (software, repeat number and motif type) [11]. The results for each species, regarding the percentage of SSRs found per EST database size are shown on Table 2.

The microsatellite survey using SSRLocator showed that 13,133 SSRs were available as potential marker loci. From those, 12,585 loci were found in single formation and only 590 were found in compound formation. The fern A. capillus-veneris showed the highest percentage (20%) of compound SSR loci. When compared with other available SSR marker search tools, similar results were found. Using MISA software, a total of 13,861 SSRs were available as potential marker loci, being 13,172 SSRs single and 689 compound SSRs for all studied species. Adiantum EST database showed the highest percentage of SSR in compound formation (15.55%). This trend does not hold for the majority of lower plants. P. patens, for example, presented few EST-SSRs in compound formation (3.57%) and possibly the fern lower database size is masking the results. When it is compared with the majority of plant groups, P. taeda is the only species showing a high percentage of compound SSRs (5.81%), corroborating other studies which report that compound and imperfect tandem repeats are most common in pines [27–29].

A total of 3,723 EST-SSRs were found in P. patens database using the MISA software [23]. The SSRLocator analysis resulted in 2,839 SSR for this species. When the same non-redundant databases were run in other bioformatics tools, the results were similar to MISA. Using the SciKoco package [30] combined with MISA, Sputinik and Modified scripts, it was possible to narrow SSR results to a 2-fold range variation.

The search for repetitive elements in EST databases of the eleven taxa listed above enabled the comparison of patterns of occurrence of these elements in lower and higher plants (Figure 1). In some species such as C. reinhardtii, Mesostigma viride and bryophytes, we found that dimer (NN) microsatellites are more common when compared to higher plants (Figure 2). The trimer (NNN) microsatellites are predominant in higher plants (See additional files), in agreement with other SSR survey studies [6, 10, 11, 21] supporting the relative distribution of motifs in these plant groups. However, gymnosperm species showed the lowest SSR occurrence within the derived plant groups. Pinus and Gnetum results indicate low SSR frequencies as intrinsic characteristics of gymnosperms, such as suggested by other results obtained with distinct methods [10, 23, 28, 29]. The patterns of occurrence of dimers and trimers found in the EST databases of the selected species are shown on Additional files 1 and 2, respectively.

Figure 1
figure 1

SSR motifs occurrences by plant group studied. SSR motifs (%) in all plant groups studied (Chlorophyta+Mesostigmatophyceae = unicelullar green algae; Bryophyta l.s. = hornworts, liverworts and mosses; Filicophyta+Lycopodiophyta = ferns; Cycadophyta+Coniferophyta = Gimnosperms; Magnoliophyta = flowering plants)

Full size image
Figure 2
figure 2

Predominant loci containing dinucleotide microsatellites motifs per species.

Full size image

The average GC-content in the 11 datasets was 48.55%. Significantly increased GC-contents were detected for the green algae Chlamydomonas (57.22%) and Mesostigma (51.36%), for the moss Syntrichia ruralis (54.75%) and the fern moss Sellaginella spp. (51.38%). These results are in agreement with other genomic comparative analyses of a wide range of plant groups, where the lower groups presented the higher contents [23, 31, 32]. The remaining species showed similar results (Table 1).

Dimer and Trimer most frequent motifs

For algae species, the most frequent dimer motifs were AC/GT and CA/TG (Figure 2). For example, in C. reinhardtii, from 548 dimer occurrences, 199 AC/GT and 233 CA/TG motifs were found. The predominant trimer motifs found were GCA/TGC, CAG/CTG and GCC/GGC (Additional file 3) with 55, 46 and 39 occurrences in 263 trimers found for algae species. For nonvascular plants, the predominant dimer motifs were AG/CT (239/1,049), AT/AT (226/1,049) and GA/TC (340/1,049), as found for P. patens. For mosses, the most frequent trimers found within the studied species were GCA/TGC, AAG/CTT and AGC/GCT. For vascular plants, the most frequent motifs were AG/CT and GA/TC. In O. sativa, 246 (43%) and 191(33%) occurrences for these motifs were found, respectively, in a total of 578 dimer occurrences. The GC/GC was only detected in C. reinhardtii. There has been a report on the abundance of GC elements in Chlamydomonas genome libraries [33].

For the other species this motif has not been reported in high frequencies [10, 11, 23, 28, 34].

Among trimer motifs, there was a predominance of AAG/CTT, AGA/TCT, GGA/TCC and GAA/TTC in higher plants. In lower plants, the motifs GCA/TGC and CAG/CTG were predominant. The trimer motif CCG/CGG is predominant in the algae C. reinhardtii and the model moss P. patens, and could reflect the high GC content in these two species. However, this relationship does not hold for the other cryptogams analysed. The increased CCG/CGG frequency has been described earlier for grasses and has been related to a high GC-content [10]. In this context, the CCG/CGG increase in Chlamydomonas and P. patens was consistent, but, a previous study reported that it can not be taken as a rule, since higher GC values were found for other lower groups with low CCG/CGG contents [23]. For rice CCG/CGG is the predominant motif and its content appears to be high in the members of the grass family [11, 21].

Comparing all plant groups selected for this in silico study, the most frequent dimer motifs found were AG/CT and GA/TC, occurring for all plant species. The most frequent trimers were AAG/CTT and GCA/TGC occurring in the 11 studied species.

Tetramers, Pentamers and Hexamers

Tetramer and pentamer motifs were rare for all studied species except for M. viride. This algae showed the higher frequencies in loci formed by motifs longer than three nucleotides with 36.95% of tetramer and 19.56% of pentamer motifs. Although these results are in agreement with other study [23], it is difficult to state that this is a rule for this species, since the EST database size for Mesostigma is the smallest one available among the studied databases. In general, tetramer and pentamer motifs predominantly found for Oryza, Physcomitrella and Selaginela where CATC/GATG, CTCC/GGAG, GATC/GATC, TGCT/AGCA (Additional file 4) and CTTCT/AGAAG, GGAGA/TCTCC, GGCAG/CTGCC, TCTCG/CGAGA and TGCTG/CAGCA (Additional file 5) and these were the most frequent motifs, at least for two out of three of these species.

Hexamer motifs were predominant in novel taxa such as gymnosperms and flowering plants [3, 21, 35]. P. taeda and G. gnemom showed the highest frequency (26.95%) of these motifs, but none of the hexamer motifs found in Gnetum and Pinus were found in common with other plant EST databases. However, one can not state the absence of hexamer motif patterns in plant groups, since in Bryophytes there is a possibility of patterns occurring within closely related groups. For P. patens and M. polymorpha the AGCAGG/AGCAGG, AGCTGG/CCAGGT, CAGCAA/TTGCTG and TGGTGC/GCACCA motifs occur in both species (Additional file 6). Based on plastid molecular data, Marchantiophyta and Bryophyta originated about 450 Mya [36] and its possible that some repeats are conserved for recently formed groups, but it would be necessary to include others species in further analyses to confirm this hypothesis. For the other SSR types (7, 8, 9 and 10 repeats) frequencies were very low (less than 2 occurrences per motif) and were not further characterized.

Physcomitrella patens SSR loci versusGene Ontology assignments

For the 4,909 SSR loci found for P. patens EST sequences, 1,750 had GO assignments. More than 25% of these hits were exclusive to P. patens. However, up to 70% of SSR loci were found as conserved across the moss and the higher plant species O. sativa, Vitis vinifera L. and A. thaliana. On Table 3, the distribution of the best Blast hits is presented.

Table 3 Distribution of Blast hits for Physcomitrella patens SSR loci sequences against several taxa with GO assignment
Full size table

Regarding biological processes, the majority of SSR loci found were involved with metabolic (32.17%) and cellular (31.02%) processes (Figure 3). Comparing all P. patens genome sequences with Gene Ontology assignment and those containing SSRs (Figure 4), there was a concentration of SSRs in metabolic process genes. Biological adhesion, rhythmic processes, growth and cell killing processes had the lowest SSR contents among the P. patens transcripts. Similar results were found comparing P. patens and A. thaliana EST libraries [37]. This author suggested that genes that are involved in protein metabolism and biosynthesis are well conserved between mosses and vascular plants. These patterns were confirmed for mosses using Syntrichia ruralis and P. patens transcript databases, respectively [38, 39]. For cellular components (Figure 5) the majority of SSRs found are related to intracellular component gene sequences (52.52%) and membrane elements (12.15%). This ontology levels were reported as the majority of GO assignments in for P. patens annotated sequences [39]. Currently, more than half of cellular component GO annotations for P. patens genome [32] are related with membrane structure (Figure 6). Our results show the enrichment of SSR occurrence mainly for genes related to this structural level. The whole genome molecular function assignment level in Gene Ontology revealed a predominance of binding genes (80.51%), suggesting these are representatively higher in P. patens genome (Figure 7). However, when EST sequences containing SSRs are assessed with the Gene Ontology assigned molecular function (Figure 8), a relative increase of other functions is revealed. Sequences associated with binding decrease (42.81%), and those related to catalytic activity (33.76%), and structural molecule activity (10.80%) increase. These findings agree to the expectations concerning the cellular function and are consistent with ratios observed for rice, Arabidopsis, and for the bryophytes Syntrichia ruralis and P. patens [32, 38–41]. The higher occurrence of SSR loci in this ontology level indicate a good potential for using these molecular markers to saturate pathways associated to those functions described above.

Figure 3
figure 3

Distribuition of Physcomitrella patens SSR loci within sequences of known biological processes in Gene Ontology.

Full size image
Figure 4
figure 4

Distribuition of Physcomitrella patens genome sequences with Gene Ontology assignments into biological processes. (Data: Rensing et al., 2008).

Full size image
Figure 5
figure 5

Distribuition of Physcomitrella patens SSR loci within sequences of known cellular component in Gene Ontology.

Full size image
Figure 6
figure 6

Distribuition of Physcomitrella patens genome sequences with Gene Ontology assignments into cellular component. (Data: Rensing et al., 2008).

Full size image
Figure 7
figure 7

Distribuition of Physcomitrella patens SSR loci within sequences of known molecular function in Gene Onthology.

Full size image
Figure 8
figure 8

Distribuition of Physcomitrella patens genome sequences with Gene Ontology assignments into molecular function. (Data: Rensing et al., 2008).

Full size image

Predicted coding for SSR loci

The predicted amino acid content for the SSR loci detected in the eleven species studied is shown in Figure 9. The amino acids arginine (Arg), alanine (Ala) and Serine (Ser) were predominant for all species. Alanine was predominant for the majority of cryptogams, ranging from 14.85% to 29.7%. Exceptions were observed for Adiantum, Mesostigma and Physcomitrella, in which serine (Ser), glutamic acid (Glu) and leucine (Leu) were the predominant amino acid (up to 17%). Serine (up to 11%) was predominant for fern species and for Gnetum and Arabidopsis, Pinus and Oryza showed arginine as the predominant amino acid (10.46% and 23.31%, respectively). Tyrosine (Tyr), asparagine (Asp), aspartic acid (Asn) were the amino acids found at lower frequencies among SSR loci for all species and were practically absent in the algae species surveyed. In bryophytes, methionine was only found in Physcomitrella, but at a small frequency (1.7%). For all higher plant species databases used in this survey, arginine, alanine, serine, glutamic acid, proline (Pro) and leucine were among the predominant amino acids, agreeing with previous reports for flowering plants [11, 3, 22, 42–45]. No reports were found for amino acid distribution in SSR loci in lower plants.

Figure 9
figure 9

Predicted amino acid occurrences in SSR loci within plant groups studied.

Full size image

The small EST databases available for some species did not seem to have hampered the results, since the predicted loci distribution found were consistent within the taxonomic groups. The absence of a relationship between genome size and tandem repeat loci content were reported based in grass genome studies [11], where large genomes such as sugarcane (Saccharum officinarum L.), maize and wheat did not present higher frequencies of SSR loci.

Relationship of Codon-bias with EST-SSR motif occurrences

The high GC-content in some EST-SSR motifs found in the present study can be a result of a codon usage preference by plant species. When we compare the codon usage for the model species included in this study (Chlamydomonas reinhardtii, Physcomitrella patens, Oryza sativa and Arabidopsis thaliana) the occurrence of some repeat motifs are reflected in codon-bias known for each species. Higher frequencies of GC were found in the first and third codon position for all four species. However, for the basal plant (C. reinhardtii), the preference for GC3 was much higher than the other three species. The first (GC1) and the third (GC3) codon position reached 64.8% and 86.21% of the occurrences, respectively. For rice, GC1 and GC3 frequencies were 58.19% and 61.6%, respectively. For the other model plants, the occurrences at GC3 were lower than the occurrences in GC1, i.e., for Physcomitrella patens and Arabidopsis thaliana, GC1 (55.49% and 50.84%, respectively) and GC3 (54.6% and 42.4%, respectively) values were found. When one associates these codon usage values with the SSR motif frequencies found, a striking result is obtained for C. reinhardtii and rice. In the first, the most frequent motifs were GCA/TGC, CAG/CTG and GCC/GGC and could be explained by the GC1s and GC3s codon preference. In rice the CCG/CGG predominant motif could also be a reflection of GC3s codon preference. For Arabidopsis, the most frequent motif found in this study (GAA/TTC) is also the most preferred codon used by this species (GAA) with 34.3% of the occurrences. It also reflects the GC1 preference in the codon usage in this species. In the model moss species the most frequent motifs do not show a relationship with the GC codon usage (Figure 10). Despite the similarities in average codon bias between P. patens and Arabidopsis thaliana, the distribution pattern is different, with 15% of moss genes being unbiased [46]. An association between the frequency of microsatellite motifs and codon usage could explain the occurrences found in P. patens. For example, the most representative motifs GCA/TGC, AAG/CTT and AGC/GCT are also found among the most used codons GCA, AAG and AGC (20.7%, 33.6% and 15%, respectively).

Figure 10
figure 10

Correlation between synonymous codon usage bias and GC composition in Physcomitrella patens EST-SSR sequences.

Full size image

The width of the GC3 distribution in flowering plants was found to be a result of variation in the levels of directional mutation pressure or selection against mutational biases. Likewise, the low frequency of GC2 occurrences is a result of a strong selective pressure against peptide substitution. The balance between these forces could be shaping the distribution of EST-SSR by means of codon usage preference [47].

Positive and negative selection sites in EST-SSR across species

SSRs represent hyper mutable loci subject to reversible changes in their length [8]. Significant differences in SSR representations exist even among closely related species, suggesting that SSR abundance may change relatively rapidly during evolution [48]. To infer about the selection pressures (dN/dS ratio) on EST-SSR found for the 11 species chosen for this work, we used the common most frequent motif in all species (AAG/CTT and GCA/TGC). The dN-dS test revealed few negatively selected sites in the triplets for each EST-SSR (Additional file 7). The positive selection in SSR based sequence was reported in other studies [8, 49–51]. More than 50% of sites for both motifs analyzed across species were under a positive selection (dN/dS > 1), suggesting a weak selection pressure on these EST-SSR motifs, as was reported for other species [52, 53]. The occurrence of selective sweeps or background selection in ancestral lineages [54] cannot be discarded, however it could not be tested with the present data.

In silico transferability of EST-SSR across species

Across-species transferability of EST-SSRs is greater than genomic SSRs, as they originate from expressed regions and therefore they are more conserved across a number of related species [6].

The virtual PCR shows a lower transferability of Chlamydomonas reinhardtii EST-SSR for most of the plant species tested. The best results were found for Adiantum and Arabidopsis, where successful rates of positive EST-SSR amplicons derived from algae were 26% and 9%, respectively. When EST-SSR primers designed from Arabidopsis were used against other species, again low transferability rates were found, being the best positive cases found in Physcomitrella, Pinus and rice with amplification rates of 1.04%, 1.20% and 1.90%. The summary of in silico PCR results can be accessed in the Additional files section of this article. Some reports suggest that SSR markers have higher transferability rates when used between closely related species [6, 22, 55]. In this work virtual PCR amplification did follow the same trend.

For the positive EST-SSRs found for the in silico transfer, ten sets of Physcomitrella EST-SSR primers were used to illustrate the transferability results using an electronic tool [56] to simulate gel electrophoresis (Figure 11). For the three tested EST-databases only two primers amplified a single locus in each species (SSR9 and SSR10). In the other sets 2, 3 and even 4 virtual amplicons were observed (Additional file 8). For Chlamydomonas, 70% of the tested primers resulted in one amplicon and 10% each resulted in 2, 3 or 4 amplifications. However, only 20% of amplicons obtained in this algae species are related to the EST-SSR sequence, suggesting that the majority of designed EST-SSR primers act as degenerate when applied to Chlamydomonas. For rice, 30%, 40% and 10% of tested primers resulted in one, two or three amplifications, respectively. In Arabidopsis 40%, 40% and 20% of tested primers results in one, two or three amplifications, respectively. For both flowering plants, 50% of tested primers amplified moss EST-SSR homologue sequences, showing a high rate of success for transferability across species. These results agree with other studies where the transfer success rates decrease with the increasing evolutionary distance [55, 57–60]. The use of this molecular marker across distant taxonomical groups are not impossible, however our findings confirm that only a few retain their EST-SSR homologue sequences, making this effort hardly worthwhile [61].

Figure 11
figure 11

Eletronical eletrophoresis gel for 10 primers set design for Physcomitrella patens EST-SSR (SSRn) across Chlamydomonas reinhardtii (Chml) Oryza sativa (Os) and Arabidopsis thaliana (At) EST databases.

Full size image

Conclusions

These results make it possible to create strategies for transferring molecular markers based on microsatellites from model to orphan species.

Microsatellites were found in all species studied and variable transfer rates were found as a function of genetic distance among taxa. The motifs found are influenced by species codon usage preference. The two most common motifs among the eleven species are under a positive selection pressure. Primers generating one amplicon in the genome of origin may generate multiple amplicons in other taxa and only a few retain their original targeting sequence. The similarities between the results here presented and other initiatives using similar bioinformatics Perl scripts, such as MISA [23], support SSRLocator as a useful tool for SSR survey analyses.

Methods

An exploratory in silico analysis of SSRs was made in ESTs databases of 11 taxa, as follows: two unicellular green algae (Chlamydomonas reinhardtii Dang, Mesostigma viride Lauterborn.), three bryophytes s. l. [Marchantia polymorpha L., Physcomitrella patens and Syntricha ruralis (Hedw.) Weber & Mohr], two ferns (Selaginella spp. and Adiantum capillus-veneris L.), two gymnosperms (Gnetum gnemon L. and Pinus taeda L.) and two flowering plants, a monocot (Oryza sativa) and a dicot (Arabidopsis thaliana). These species were chosen because the amount of available ESTs data in Genbank (NCBI). As these databases may have redundancy, we used the program CAP3 [62] for MacOX, to construct contigs with the sequences and get non-redundant sequences for each database following the default settings.

Taxa data were loaded into the software SSRLocator [63], to investigate the presence of tandem repetitive elements (SSRs). The analysis was performed following the search parameters for repetitive elements in class I (≥ 20 bp) described as more efficient molecular markers [17]. Data resulting from in silico analyses were assessed for occurrence patterns in chosen taxa databases. The same analysis was performed using MISA script http://pgrc.ipk-gatersleben.de/misa/ software to search for SSR occurrences per contig. Several instructions in the algorithm used in SSRLocator resemble those from MISA [19] and SSRIT [17]. However, additional instructions have been inserted in SSRLocator's code. Instead of allowing the overlap of a few nucleotides when two SSRs are adjacent to each other and one of them is shorter than the minimum size for a given class as found in MISA and SSRIT, a module written in Delphi language records the data and eliminates such overlaps. For GC content, Perl scripts were used and the results were stored in text files (.txt) for later comparative analyses.

For the predicted amino acid contents in the SSR loci, an additional routine script was written in the SSRLocator software. This script determined which amino acids were coded by trimer, hexamer and nonamer motifs found in the EST database analysed [63].

To validate the frequencies obtained using the SSRLocator software, the Physcomitrella patens EST database was chosen.

This database was run with other SSR search scripts and softwares, such as MISA [19] and SPUTINIK [64], running in SCIROKO package [30], MINE SSR http://www.genome.clemson.edu/resources/online_tools/ssr, SSRIT following the SSR categories defined above [17]. The results were exported into Microsoft Excel spreadsheets (MacOSX-Oficce 2008) and respectively grouped by taxon.

A codon-bias for the model plants included in this research (Chlamydomonas reinhardtii, Physcomitrella patens, Oryza sativa and Arabidopsis thaliana) was made comparing with the preferencial codon table for each species available at http://www.kazusa.or.jp/codon/. The sequences containing EST-SSR for Physcomitrella patens was submitted to CodonO server [65] to confirm the preferencial codon usage compared with the know codon table for this species. To investigate the selective pressure on the triplets on the EST-SSR which occurs in all studied species a dN-dS statistics [66] was used to verify the synonymous and noun-synonymous substitutions in the preferential codons nearby the repeats chosen using the molecular phylogenetics package MEGA4 [67].

The Physcomitrella patens SSR results were run through a Gene Ontology (GO) assignment database in order to assess associations between SSR loci and biological processes, cellular components and molecular function of known genes. A fasta file with all EST-SSRs found in P. patens was subjected to Blast2GO software and ran against the GO annotated sequences, and the obtained hits were compiled.

To verify the potential transferability of this molecular markers we have tested in silico all EST-SSR found for the plant ancestral lineage, and for the derivative plant group, represented here by the green algae Chlamydomonas reinhardtii and Arabidopsis thaliana, across the others species EST database used for the present SSR survey. Electronic PCR [68] was used to verify the transferability of EST-SSRs across studied species. The positive results found were used to simulate a gel electrophoresis with aid of SIMGEL.exe included in the SPCR package [56] using the Physcomitrella patens EST-SSR sequences to design primers and Chlamydomonas, rice and Arabidopsis as templates. The virtual amplicons resulted for each primer set tested across species were aligned to verify the homology between the amplicons.