Transcriptomic changes arising during light-induced sporulation in Physarum polycephalum
© Barrantes et al. 2010
Received: 30 October 2009
Accepted: 17 February 2010
Published: 17 February 2010
Skip to main content
© Barrantes et al. 2010
Received: 30 October 2009
Accepted: 17 February 2010
Published: 17 February 2010
Physarum polycephalum is a free-living amoebozoan protist displaying a complex life cycle, including alternation between single- and multinucleate stages through sporulation, a simple form of cell differentiation. Sporulation in Physarum can be experimentally induced by several external factors, and Physarum displays many biochemical features typical for metazoan cells, including metazoan-type signaling pathways, which makes this organism a model to study cell cycle, cell differentiation and cellular reprogramming.
In order to identify the genes associated to the light-induced sporulation in Physarum, especially those related to signal transduction, we isolated RNA before and after photoinduction from sporulation- competent cells, and used these RNAs to synthesize cDNAs, which were then analyzed using the 454 sequencing technology. We obtained 16,669 cDNAs that were annotated at every computational level. 13,169 transcripts included hit count data, from which 2,772 displayed significant differential expression (upregulated: 1,623; downregulated: 1,149). Transcripts with valid annotations and significant differential expression were later integrated into putative networks using interaction information from orthologs.
Gene ontology analysis suggested that most significantly downregulated genes are linked to DNA repair, cell division, inhibition of cell migration, and calcium release, while highly upregulated genes were involved in cell death, cell polarization, maintenance of integrity, and differentiation. In addition, cell death- associated transcripts were overrepresented between the upregulated transcripts. These changes are associated to a network of actin-binding proteins encoded by genes that are differentially regulated before and after light induction.
Physarum polycephalum, commonly known as "slime mold", belongs to the mycetozoan group of Amoebozoa. Physarum follows a complex life cycle with haploid and diploid cell types, formed in temporal order as triggered by environmental stimuli . The plasmodium is a multinucleate single cell whose nuclei display natural synchrony with respect to cell cycle and differentiation status. During several days of starvation, a plasmodium grown to macroscopic size becomes competent to sporulate. Sporulation can then be triggered experimentally by exposing a competent plasmodium to a pulse of blue or far-red light, or to heat shock. As plasmodial nuclei are synchronous, and because the timing of the differentiation program in individual plasmodia can be reproduced experimentally, the stage-specific gene expression program that leads to sporulation can be analyzed at high resolution [2–4]. Recent large scale cDNA surveys [5, 6] provide a basis for studying the phenomenon at the transcriptomic level.
In order to identify the differentially expressed genes associated with the commitment to sporulation, we characterized and compared two cDNA libraries prepared from competent and light-induced plasmodia using massive parallel sequencing technology . We employed this method because it does not rely on reference transcripts for quantitation, previous cloning steps are not required, it does not have an upper limit for quantitation, and it is a relatively unbiased approach . From the comparison of annotations and transcript quantitations, we found that most differentially expressed genes encode proteins associated to a network of actin-binding proteins. Components of this putative interaction network are associated to development, DNA repair, cell division, calcium release, cell death, and maintenance of cell integrity.
Separate cDNA libraries were constructed from polyA+ RNA isolated from two sources: (i) competent plasmodia; and (ii) sporulation- induced plasmodia (competent plasmodia harvested six hours after exposure to far-red light). The cDNAs libraries were then analyzed using massive parallel sequencing [7, 8]. Transcripts were annotated at every bioinformatic level, and the annotation data was used to infer hypothetical interaction networks from differentially regulated genes. The whole approach is summarized in the [Additional File 1 : Figure S1].
From the pyrosequencing, we obtained a total output of 61.9 Mb from two runs, corresponding to the starved (26.1 Mb) and light-induced (35.8 Mb) plasmodia libraries. Considering that Physarum possess a 300 Mb genome, and assuming that 10% is encoding genes, we estimate a 2.06× coverage of protein coding sequences. The assembled sequencing output consisted of 26,037 sequences, and large cDNAs from this assembly (>500 bp) were then joined to a previously available sequence dataset , to form a comprehensive set of representative transcripts. This analysis produced 16,669 sequences, 13,169 of these containing transcript abundance data: 125,456 reads from competent and 99,632 reads from light-induced plasmodia, respectively [Additional File 2]. We used this abundance data (number of reads for each assembled transcript) as a measure of expression, which we defined as "hit counts". The remaining contigs without hit counts consisted of previously sequenced clones from a normalized cDNA library prepared from competent plasmodia , indicating that the normalization produced a broader coverage of transcripts. From 11,399 cDNA contigs detected in the competent plasmodia library (10,689 in light-induced plasmodia), over 4,227 were represented with at least five hits (3,553 in light-induced plasmodia [Additional File 3 : Figure S2]). Conversely, 8,711 transcripts (52,3%) were found with 5 or less sequence hits in both samples. For statistical reasons, no statement on the differential expression from this fraction could be made. Between contigs with lowest hit counts, 2,437 cDNA species were represented by just one hit (competent plasmodia), and 2,621 from light-induced sample [Additional File 3 : Figure S2].
We compared the transcript hit counts between different libraries as a measure of differential gene expression. As most contig species were represented by low hit counts, we normalized the number of hits. To this end, we first obtained the relative frequency (number of hits divided by the total hits on a given condition), and later we calculated the relative frequencies for each contig in the two cDNA samples compared to each other. Given that each EST belongs to a single gene, the significance of its differential expression depends only on the number of hits, respect to the total number of hits on each library . Following these assumptions, we found 2,772 cDNAs that displayed significant differential expression (P -value < 0.05). All contig species, regardless of whether differentially expressed or not were submitted to the Sequence Read Archive subset of GenBank  (Accession numbers SRX012830 and SRX012831).
The newly assembled contigs were compared against sequence databases using BLASTX . This analysis revealed that 3,310 sequences have significant similarity (≤ 1 × 10-15) to existing sequences in SwissProt , 3,651 to the protozoa subset from RefSeq , and 3,345 to proteins of the related model organism Dictyostelium discoideum (dictyBase; ). From the 13,169 sequences with hit counts data, orthologs were identified for 5,544 transcripts (1,287 of these with significant differential expression; [Additional File 4]).
The Gene Ontology (GO) project  is an annotation framework that provides a standardized vocabulary that is used to assign function to uncharacterized sequences, based on three main categories: biological processes (BP), molecular functions (MF) and cellular components (CC). We employed BLAST2GO , a tool that associates GO terms to sequences based in several annotation evidences, to classify gene function in our dataset. Using the BLASTX hits (annotation e-value cutoff < 1 × 10-6), together with GO terms previously extracted from InterPro domain searches , we inferred 13,068 GO annotations for 3,304 (20%) cDNAs, with 11,446 annotations belonging to 2,459 sequences with hit counts data. Transcripts were associated to biological processes (n = 2,437; 15%), molecular functions (n = 2,801; 17%), and cellular components (n = 2,023; 12%). As many as 2,136 (13%), 1,663 (10%) and 1,645 (10%) sequences were annotated with a combination of MF and BP terms, MF and CC, and BP and CC terms respectively, and 1,487 cDNAs were annotated with MF, BP and CC terms altogether. Later, in order to analyze the differences between the two condition groups with respect to the GO annotations, we carried out Fisher exact tests using the Gossip module  from BLAST2GO. We found that the 'cell development' (GO:0048468), 'cell death' (GO:0008219) and 'death' (GO:0016265) GO terms were overrepresented in cDNAs with higher relative frequencies in light-induced plasmodia (false discovery rate < 0.01), as compared to competent plasmodia [Additional File 6 : Table S2].
Summary of the transcriptome sequencing and annotation.
Total 454 reads
Total sequencing output (Mb)
Reads from the competent plasmodia library
Reads from the light-induced plasmodia library
Contigs with hit counts
Contigs with at least 5 hits in both libraries
More abundant in the library from competent plasmodia
More abundant in the library from light-induced plasmodia
Downregulated, significant differential expression
Upregulated, significant differential expression
Total contigs with blastx results (e-value < 1E-3)
Contigs with blastx results and hit counts
Contigs with blastx results and significant differential expression
Total contigs with GO annotations
Total contigs with KEGG orthologs
Total contigs with InterPro results
Contigs with GO annotations and hit counts
Contigs with KEGG orthologs and hit counts
Contigs with InterPro results and hit counts
Top 20 Annotated Transcripts Downregulated in Light-induced Plasmodia
Cell division control protein 16, CDC16
Phospholipase D (PLD1)
Poly ADP-ribose polymerase 1 (PARP-1)
Uncharacterized transporter YHL008C
Subtilisin inhibitor 1, ASI-I
Glutamate decarboxylase, GAD
Flap endonuclease 1, FEN-1
Putative ankyrin repeat protein, YL715
F-box/LRR-repeat protein 5, FBL5
Pumilio homolog 2
Kv channel-interacting protein 2, Kcnip2
Actin-binding protein 120, ABP-120
POLTERGEIST Protein phosphatase 2C 32, PP2C
Glyoxalase II, Glx II
Pyruvate decarboxylase isozyme 2, PDC2
Apoptosis inhibitor 1, IAP-1
Top 20 Annotated Transcripts Upregulated in Light-induced Plasmodia
ATP-dependent RNA helicase, DDX42
Regulator of G-protein signaling 2, RGS2
ADP-ribosylation factor GTPase-activating, AGD8
Serine/threonine-protein kinase, pakA
Myosin regulatory light chain 1, MLR1
Uncharacterized protein DDB_G0285917
Transmembrane protein 63A, TM63A
Dosage-dependent cell cycle regulator 2, DCR2
Non-histone chromosomal protein 6, NHP6
Calcium-transporting ATPase, PAT1
Zinc finger- containing protein 18, NHN1
Rab RING finger 7, RR7
Cell division control protein 31, CDC31
Putative FNIP repeat-containing protein, L281
Proteasome subunit alpha type-7, PSMA7
The development of plasmodia competent for sporulation includes growth arrest, condensation of cellular constituents, and mitosis . Sporulation of competent plasmodia can be triggered by a light pulse. Some proteins associated with the light-induced pathways that lead to sporulation have been described [2–4], suggesting that several signaling mechanisms are involved, but there are no studies that describe changes at the level of the whole transcriptome. In the present study we identified the most up- and downregulated transcripts, which are associated to a network of putative interactions (Figure 4). The network is hypothetical as interactions used for inference are based on data obtained from different organisms. For the sake of simplicity, the discussion will be focused on genes with predicted significant interactions (P > 0.9).
The actin cytoskeleton of Physarum is essential for locomotion, division, and other biological processes . Assembly and disassembly of actin filaments is controlled by a group of actin-binding proteins, whose activities in turn are regulated by specific signaling pathways. Physarum cell types differ in actin organization but express the same actin genes, suggesting that changes in actin-binding proteins are responsible for the differences in actin organization . Physarum possesses several classes of actin- binding proteins, and most of these proteins display cell type-specific patterns of expression, but their precise roles are not known [29, 30]. Nevertheless, expression changes in genes coding for actin-binding proteins correlate with modifications in cell organization and behavior . In the present study, we found that some actin-binding genes were linked specifically to stages before and after photoinduction in the starved Physarum plasmodium.
Specifically, we identified protist orthologs for actin-binding proteins, including Dictyostelium coaA (Coactosin A ) and ABP-120 (actin-binding protein 120), and actobindin from Acanthamoeba, which binds actin monomers (Table 2) . Coactosin A interferes with the capping of F-actin filaments , and is differentially expressed after metal exposure in worms . ABP-120 organizes filamentous actin into networks of fibers, and Dictyostelium cells lacking ABP-120 have a severe phototaxis defect at the multicellular slug stage . In addition, we noticed that transcripts coding for Physarum plasmodia-specific actin-binding proteins, such as profilin P (PROP)  and fragmin P (FRGP) , are downregulated after photoinduction (Figure 3). FRGP enables actin phosphorylation by the actin-fragmin kinase (AFK), and binds phosphorylated actin [29, 36]. Therefore it is possible that during sporulation these proteins are involved in the reorganization of the subcellular compartments via interactions with the actin cytoskeleton.
After several days of starvation, cell processes must be limited in order to save energy. Coordination of several biological processes is then required, and thus regulation of these phenomena needs a pleiotropic transducer like the cAMP, which targets several signaling pathways, including those that limit cell proliferation . Cell differentiation pathways regulated by cAMP levels have been described in Dictyostelium, a closely related protist . For Physarum, we found that MEI2, a transcript controlled via cAMP levels, is downregulated in the light-induced plasmodium (Figure 3; [Additional File 5 : Table S1]). MEI2 is an RNA-binding protein that encodes a cAMP-regulated positive regulator of meiosis in the yeast S.pombe [39, 40]. This gene product is functionally related to the actin cytoskeleton via the cAMP-dependent protein kinase A (PKA) [37, 38]. Other transcripts downregulated in light-induced plasmodia associated to cell division and DNA repair comprised FEN1, CDC16 and PUM2. First, the Flap endonuclease 1 (FEN1) appears in several processes linked to the maintenance of the genome integrity, such as the UV-induced DNA repair , as well as in DNA replication and DNA recombination . Second, the yeast cell division control protein 16 (CDC16), constitutes the catalytic subunit of the spg1p GTPase-activating protein, that is involved in the signal transduction controlling septum formation. CDC16 is involved in cytokinesis  and is essential for proliferation, as spores lacking a functional CDC16 gene complete mitosis without undergoing cell cleavage . Finally, PUM2 (Pumilio 2) encodes a RNA-binding protein associated to the control of meiosis during development . Consequently, starvation seems to be the signal that regulates cell division while protecting the cells from oxidative stress, through cAMP-regulated pathways (Figure 3).
Other downregulated transcripts in the light-induced plasmodium comprised orthologs of transducers, such as FBL5, a leucine-repeat protein linked to phosphorylation-dependent ubiquitination , PARP1, an Oryza poly ADP-ribose polymerase, a phospholipase D from Phytophtora (PLD1), and the Arabidopsis phosphatase 2C (PP2C, also known as Poltergeist). In plants, G-proteins are involved in phospholipase D activation, and this also seems to be the case for Phytophtora ; on the other hand, PP2C operates in several signaling pathways that regulate stem cell differentiation . It is then reasonable to consider that the differential expression of these transducers is also associated with the control of signaling mechanisms for differentiation, but more profound studies are needed to establish precise causal relationships.
Transcripts identified as calcium-binding proteins displayed different patterns of expression regulation. These were either down- (LAV1-2, KCNIP2 and GAD) or upregulated (MLR1, TRHY, and PAT1) after light induction. LAV1-2 is a plasmodium-specific RNA of unknown function that encodes a protein containing an EF-hand type domain whose calcium-binding activity has been observed in vitro in Physarum . LAV1-2 seems to act as a sensor of cell damage, releasing Ca2+ that leads to the activation of a plasmodium-specific transglutaminase, which separates damaged areas of a plasmodium . Other transcripts encoding orthologs of calcium-binding proteins, such as KCNIP2 and GAD, were also downregulated in the photoinduced plasmodium and have not been previously described for in Physarum. KCNIP2 encodes a potassium channel-interacting protein that probably modulates channels density in a Ca2+- dependent manner. In turn, the activation of glutamate decarboxylase (GAD) by calcium-bound calmodulin (CaM) is required for normal growth in plants . Previous studies have shown that the intracellular increase of calcium levels is correlated with increased concentrations of cAMP and with sporulation and differentiation in both Physarum and Dictyostelium [52, 53]. Moreover, actin filament crosslinking is affected by changes in intracellular calcium levels, which ultimately influences the cell contractility . Therefore it seems possible that these calcium-binding proteins coordinate the Ca2+ release as a means to influence the cell contractility through the interaction with the actin cytoskeleton (Figure 4).
Furthermore, the upregulated subset of calcium-binding proteins included MLR1, which inhibits cytokinesis in yeasts; trychohyalin (TRHY), which is involved in its own calcium-dependent processing during differentiation; and the Dictyostelium PAT1 ATPase. PAT1 is localized in the membrane of contractile vacuoles, and is a component of a calcium sequestration and excretion pathway, which functions to help maintain homeostasis, especially under conditions of Ca2+ stress . Thus these are candidates to control the intracellular calcium levels after light induction of starved plasmodia.
After photoinduction, a group of actin-binding proteins is upregulated including the elongation factor 1α (EF1A), Spire, and actophorin (Figures 3 and 4; Table 3; [Additional File 5 : Table S1]). Spire is a Drosophila gene involved in development through actin assembly. This gene is also widely distributed across the metazoan genomes. Spire mammalian isoforms are MAP kinase substrates, and data suggest that Spire evolved as an alternative independent mechanism of actin polymerization, necessary for cell polarization in multicellular organisms . Actophorin, in turn, binds actin monomers and separates actin filaments in a dose-dependent manner. Phosphorylation of actophorin blocks actin binding . In turn, EF1A, aside from its role in the protein synthesis, has a separate conserved actin-binding activity in eukaryota, initially observed in Dictyostelium , where it is predominantly found in actin-bound form . EF1A regulates the stoichiometry of cytoskeletal components, and the conservation of the EF1A-actin interaction across eukaryotes suggests its importance for cytoskeletal maintenance . Overexpression of EF1A in yeast results in effects on cell growth, and influences the actin distribution, morphology and budding in a dosage-dependent manner, although this increase of EF1A has no effect over the protein synthesis . In addition, changes in cytoskeletal redistribution of EF1A seem to be linked to the differentiation status, where the association between EF1A and microtubules gradually increases in differentiating cultures . Furthermore, EF1A stimulates actin remodeling and induces the formation of filopodia , and possibly connects these processes with signaling pathways .
We noticed that two coexpressed transcripts (the cysteine proteinase CYSP2 and the developmentally regulated gene CudA) are related to EF1A. First, cysteine proteinases are believed to participate in protein cleavage during the differentiation of Dictyostelium as a response to starvation , and these peptidases were copurified with EF1A in yeasts . CudA, on the other hand, is associated to the transition from slug migration to culmination in Dictyostelium. CudA expression levels depend on local cAMP concentration . Recent evidences show that CudA contains a novel DNA-binding site that is distantly related to the metazoan STAT domains, which participate in the regulation of developmentally controlled genes , and whose orthologs coexpress with EF1A . Yamada et al.  also proved a relationship between Dictyostelium CudA and a cDNA from Physarum, which corresponds to the contig reported here as a CudA ortholog. For these reasons, EF1A could work as a link between regulation of the protein synthesis, cytoskeletal maintenance, and signal transduction in slime molds (Figure 3).
Other developmentally regulated transcripts associated to the actin cytoskeleton included the cell wall integrity and stress response component (WSC1), which is a yeast membrane protein that acts as a sensor of cell wall damage , and CDC31, a constituent of the nuclear pore complex that is also involved in the maintenance of cell morphology (Table 3 and Figure 4). WSC1 is essential to keep the cell integrity, behaving like a stress-specific signal transducer that is involved in the reorganization of the actin cytoskeleton in response to osmotic shock [71, 72]. WSC1 is involved in the depolarization of the actin cytoskeleton , and, like CDC16 (downregulated in light-induced plasmodia), is entailed in cytokinesis .
Orthologs of certain genes highly upregulated in light-induced plasmodia are involved in signal transduction. These include transcripts linked to the GTP signaling (AGD8, YPTC6, RGS2), kinases (pakA) and phosphatases (DCR2). The serine/threonine-kinase pakA is a regulator of the myosin component of the cytoskeleton, required for cytokinesis and the regulation of the cytoskeleton during chemotaxis in Dictyostelium . In turn, the yeast dosage-dependent cell cycle regulator 2 (DCR2), is a phosphatase whose increased dosage alters cell cycle progression, while its loss delays the progression in the G1 phase . In addition, upregulated GTP signaling transducers included a putative GTPase- activating protein from Arabidopsis (AGD8); a Chlamydomonas GTP-binding protein (YPTC6); and RGS2, which acts as a negative regulator of G-protein signaling, a function that is evolutionarily conserved in yeast, C. elegans and mammals. Increased RGS2 expression is primarily mediated by the cAMP/PKA pathway , therefore it is possible that RGS2 is carrying out similar tasks in slime molds, where it could work in coordination with the other transducers, as hypothesized in Figure 4.
Comparison of GO terms between up- and downregulated groups showed that transcripts annotated for 'cell development' (GO:0048468), 'cell death' (GO:0008219) and 'death' (GO:0016265) were overrepresented in the upregulated group [Additional File 6 : Table S2]. However, all these ontologies belong to the same hierarchy, meaning that 'cell death' can be the product of either development or organismal death, and hence 'cell death' is the only difference between both expression groups. One of these cDNAs annotated for 'cell death' is Sequestosome 1, which is also included on the list of upregulated transcripts (Table 3). Sequestosome 1, also known as p62, is a multifunctional protein that targets polyubiquitinated proteins to degradation by proteasomes and autophagy . p62 knockouts significantly increased cell death , and this is probably linked to the interaction with atypical protein kinase C isoforms that are involved in pathways that control differentiation and apoptosis . Therefore it is likely that this gene product regulates cell death pathways linked to the commitment for sporulation.
Furthermore, other highly upregulated genes are also functionally linked to the protein turnover. These include the FKBP70 rotamase, which accelerates the folding of proteins during synthesis; the PSMA7 proteasome subunit, which together with the other subunits, suffer changes during the meiotic cell cycle ; and the endosome-lysosome vesicle traffic-related RR7 . It is likely then that these gene products, together with Sequestosome 1, are linked to the control of differentiation through post-transcriptional regulation.
The gain of sporulation-competence of Physarum plasmodia involves growth arrest, condensation of constituents, and mitosis and is a prerequisite before sporulation can be induced by light . Physarum gene expression has been shown to be cell type-specific, but existing studies have been focused only on individual genes [2–4]. Previously, we reported a library of 5,856 sequences obtained from plasmodia competent for the induction of sporulation . In the present study we used the massive parallel sequencing technology at the level of the whole transcriptome [7, 8] in order to identify global changes in expression that occur during light-induced sporulation of Physarum. We integrated the differentially expressed cDNAs into networks using interaction information from orthologs and the literature. We found that after light induction of a plasmodium the expression of transcripts linked to cell division and DNA repair is downregulated. In contrast, light-induction stimulated the expression of genes associated with protein turnover (proteases and proteasome transcripts), genes related to cell cycle progression, and genes involved in the maintenance of cell integrity and cytokinesis. These latter gene products might protect the cell against osmotic shock. Additionally, different groups of calcium-binding proteins are either down- or upregulated after light exposure. These gene products are candidates to control the intracellular calcium levels during sporulation. We postulate that these changes are associated with a network of actin-binding proteins (Figures 3 and 4), the components of which are differentially regulated upon plasmodial photoinduction. It seems that these gene products accomplish different tasks in each stage: the reorganization of the subcellular compartments in order to inhibit migration during starvation on one hand, and cell polarization and cytoskeletal redistribution after photoinduction mediated by a group of actin-binding proteins on the other. We expect that the precise representation of the proposed interaction networks may become available as gene knockout experiments, proteomic data, and comparative interactomics are integrated in future studies of this organism.
Physarum plasmodia of the white strain (LU897 × LU898) were hatched from spherules, and grown as microplasmodial suspensions for four days. The plasmodial mass was then applied to starvation agar plates. Microplasmodia spontaneously fused to give a single plasmodium on each plate. Plasmodia were then starved for six days in the dark at 22°C to obtain maximal competence for sporulation. To verify the sporulation-competent state, plasmodia were cut into two halves. One half was immediately frozen in liquid nitrogen for RNA extraction, and the other half was returned to the dark and incubated until the next day to verify that the plasmodium had not been induced to sporulation. To obtain light-induced plasmodia, competent plasmodia were irradiated for 30 min with far red light and then returned to the dark. Six hours after the start of irradiation, plasmodia were cut into two halves. One half was frozen in liquid nitrogen for RNA extraction. The other half was returned to the dark and incubated until the next day to verify the sporulation status [3, 4].
Transcript poly(A)+ RNAs were isolated by oligo-dT chromatography. cDNAs were prepared from these RNAs by the full-length enriched synthesis method (vertis Biotechnologie, Freising- Weihenstephan, Germany). First strand cDNA was synthesized using oligo(dT) adapter primers and MMLV H-reverse transcriptase. Following RNA hydrolysis, an adapter primer was annealed to the 3' end, and the produced fragments were PCR-amplified for 22 cycles with a proofreading enzyme. The cDNA libraries were then directly sequenced using the 454 GS FLX system (Roche Diagnostics, Mannheim, Germany) . Chromatograms were scored for quality, and the produced sequences were trimmed of adapter sequences, and coassembled into contigs using previously available transcriptomic data . For expression comparisons we obtained for each contig: (i) the number of reads (which we define as "hit counts") in both libraries; (ii), their relative frequencies (reads of a given contig divided by the total number of reads); and (iii) their relative frequencies. Statistical significance between the two hit counts for each contig species was then assessed .
Similarity searches against protein databases were performed using BLASTX implemented in the MIGenAS tool  (e-value 1 × 10-3). We utilized nine protein databases in this comparison: Swiss-Prot and TrEMBL (versions 56.3 and 39.3) , dictyBase  and RefSeq database subsets: mammalian, other vertebrates, invertebrate, protozoa, plant and microbial (release 31) . Functional annotation was carried out using BLAST2GO (version 2.2.3) . This procedure consisted of a similarity search against the non-redundant GenBank database , using BLASTX (e-value 1 × 10-3), followed by Gene Ontology (GO)  mappings extracted from similarity results and InterPro domain matches (InterPro release 18.0) . Annotation of sequences (cutoff value 1 × 10-6) was followed by their validation, and these annotations were extended using ANNEX . Statistical analysis of GO annotations between differentially expressed cDNAs was carried out using the Fisher exact test, as implemented in the GOSSIP module  of BLAST2GO. Sequences were also categorized in metabolic and signaling pathways, via similarity search against orthologs present in the KEGG database  using the KAAS server . In this case, we employed default parameters for ESTs. KEGG orthologs (KOs) were then plotted into the whole metabolic atlas, utilizing the KEGG mapping tool . Putative networks of correlated genetic interactions were generated from annotation information, using the MLE algorithm , as implemented in the Cytoprophet plugin  of Cytoscape .
Website for Affiliation 5: http://www.ovgu.de/ag-marwan/
elongation factor 1- alpha
cell wall integrity and stress response component
yeast meiosis- related protein
cAMP- dependent protein kinase A
Signal transducer and activator of transcription
false discovery rate
Kyoto Encyclopedia of Genes and Genomes
KEGG Automatic Annotation Server
We thank Markus Rampp (Rechenzentrum Garching of the Max Planck Society), for his support during the computational analysis. WIB is an International Max Planck Research School - Otto von Guericke University fellow. The experimental part of the study was supported by a cooperation contract with the Max Planck Society.
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.