S. platensis strain C1 cultures were grown at 35°C under illumination by a 100 μEm^-2s^-1 fluorescent light with continuous stirring in 2 L of Zarrouk's medium [10]. The culture was grown until the optical density at 560 nm reached 0.4 (mid-log phase), and subsequently a cell sample was harvested by filtration before shifting the growth temperature (t = 0 min). The growth temperature was then immediately shifted from 35°C to 40°C and the culture was incubated for 45, 90, or 180 min before cell harvesting.

The harvested cells were washed and lysed as described previously [11]. The three subcellular fractions of Spirulina were separated according to the methods described by Murata and Omata, and Hongsthong et al. [11,12]. It should be noted that the soluble fraction contained cytoplasmic and periplasmic proteins. The purity of thylakoid (TM) and plasma membrane (PM) fractions were tested by scanning absorption spectra and western blot analysis as described previously [11]. The membrane pellet was resuspended in 500 μl of dissolving buffer, containing 2 M thiourea, 8 M urea, 20 mM Tris, 30 mM DTT, 1% (v/v) IPG buffer, 0.05% (w/v) β-dodecyl maltoside, and 4% (w/v) CHAPS, prior to protein precipitation using a 2D-clean up kit (GE Healthcare Biosciences, USA). The protein pellets were then dissolved in dissolving buffer without DTT before determining protein concentrations using a 2D-Quant kit protein assay (GE Healthcare Biosciences).

The pH of the protein samples was adjusted to 8.5 and 10 μg of each sample, prepared as described above, was labeled with fluorescent dyes, according to the manufacturer's instructions (GE Healthcare Biosciences). The proteins were separated by 2D-DIGE and statistically analyzed for differential expression as described previously [13]. Protein spot picking and in-gel digestion of the proteins of interest were carried out as described [13].

To study phosphorylated proteins, 2D-PAGE using 7 cm non-linear IPG strips, pH 3-10 and 4-7 (GE Healthcare Biosciences), in the first dimension were performed. Subsequently, the second dimension was conducted as described above, followed by western blot analysis. Three independent experiments were performed.

After protein digestion with trypsin, peptide samples were analyzed by MALDI-TOF mass spectrometry. For protein identification, the resulting peptide mass fingerprints (PMFs) were analyzed with our in-house software tool, using the unpublished S. platensis C1 database, which was generated from in silico digestion of the S. platensis C1 completed genome sequence. The search parameters and data filtering were carried out as described in detail previously [13]. PMF identification results of a protein spot were required to be reproducible in order to consider the spot as an identified protein.

Following 2D-PAGE, detection of phosphorylated proteins was performed by western blot analysis [14]. Phosphorylations on serine, threonine, and tyrosine residues were detected separately using monoclonal antibodies raised against the designated phosphorylated amino acid residues (Santa Cruz, USA, and Assay Designs, USA). Phosvidin and trypsin inhibitor were used as positive and negative controls, respectively. An equal amount of each protein sample was separated by 2D-PAGE, as described previously, and transferred onto a nitrocellulose membrane using a semi-dry electroblotter at a constant voltage of 20 V for 30 min at room temperature. A chemiluminescent western blot detection kit, with an HRP-based system and chemiluminescent molecular weight markers, was used (Pierce, USA) according to the manufacturer's instructions.

The resulting phosphoproteome spot-maps were then matched with the 2D-DIGE spot-maps over the same pH range. Finally, the analysis of differentially expressed proteins containing phosphorylated amino acid residues was performed.

The details of Spirulina RNA isolation were described previously [13]. The transcriptional expression levels of the genes of interest were analyzed by RT-PCR using an AccessQuick™ RT-PCR System (Promega, USA). The RT-PCR analysis was carried out according to the manufacturer's instructions. Details on primers are shown [see Additional file 1]. The densities of the RT-PCR product bands were quantified using the Image Quant TL program (GE Healthcare Biosciences). Normalization of the RT-PCR product levels was performed by comparing the density of the designated band to the density of the 16S rRNA band.

The protein expression dataset was validated for the input well-form of protein ratio values. The null values and those ratios that were extremely high or low, relative to the threshold value of 1e+-10, were filtered out. K-mean clustering was applied to obtain 23 profiles of the protein expression patterns. A good k-profile number was chosen by simulation, as described by Martin et al. [15].

A protein-protein interaction (PPI) network in Spirulina was constructed on the prototype PPI database of Synechocystis from CyanoBase [16]. Prototype construction is based on a graph in which nodes and edges represent proteins and interactions, respectively. Each interaction was experimentally identified by the yeast two-hybrid system. Source nodes represent the bait proteins, and prey proteins are represented by the target nodes. Next, homologous proteins, identified by BLAST similarity searches with significant values less than 1e-10, were mapped to their best-hit Synechocystis protein nodes. Finally, differentially expressed proteins in Spirulina were mapped to their corresponding nodes. The size of each node represents the level of differential expression.

RT-PCR was used to analyze the transcriptional expression levels of some differentially expressed proteins (Fig. 1, 2, 3, 4, 5, 6, 7, 8, 9 and 10) [see fig S5; Additional file 2]. The transcriptional expression patterns of DNA gyrase and chromosome segregation ATPase from the soluble fraction, and ABC transporter, S-adenosyl-L-homocysteine hydrolase and Δ⁹ desaturase from the thylakoid membrane were well correlated with the protein expression patterns. This correlation suggests that these proteins are likely regulated at the transcriptional level. Interestingly, the Δ⁹ desaturase gene, the first gene in the fatty acid desaturation process of Spirulina, was previously reported to be temperature-independent [9]. However, an earlier study by our group using Northern blot analysis [see fig S4; Additional file 2] demonstrates that this gene is indeed temperature-dependent, in agreement with the results obtained in the present study.

The transcriptional patterns of sensory box/GGDEF and RNA-directed DNA polymerase, both from the soluble fraction, were different than their protein expression patterns. The transcripts of these genes increased throughout the experimental time period (3 hours), while their protein levels initially increased, followed by reduction to steady state protein levels. Importantly, phosphorylation was detected on both of these proteins. This suggests that the post-translational modification might play a role in the function of these proteins in response to the high temperature stress.

In the thylakoid membrane fraction, the transcription patterns of DnaK and N-6 DNA methylase (adenine specific) were not well correlated with their protein expression patterns, and phosphorylation was not detected for these proteins. These results suggest that these proteins might be regulated at either the post-transcriptional level or the post-translational level (except phosphorylation), although further investigation will be required to confirm this hypothesis. There are some interesting facts related to these three proteins in the photosynthetic membrane of Spirulina. It has been reported that the N-6 DNA methylase uses S-adenosyl-methionine as a methyl-donor for its DNA-methylation reaction in plants [30]. It should be noted that the S-adenosyl-L-homocysteine hydrolase, the enzyme responsible for regeneration of S-adenosyl-methionine [31,32], was also up-regulated in the same fraction.

In Synechococcus sp. PCC7942, photosystem II (PSII) is drastically deactivated at 40°C [33]. The chaperone DnaK (Hsp70) is present in plant-chloroplast as a component of multi-chaperone complex [34] and plays a critical role in photoprotection and repair of PSII during and after photoinhibition [35,36]. Thus, it is expected that this chaperone was up-regulated in response to the high temperature stress in Spirulina, although its regulation should be further investigated.

Adenylate cyclase, which is known to localize in the thylakoid membrane of cyanobacteria, plays a key role in cAMP biosynthesis [37]. The level of cAMP is regulated by red/far red light and thus adenylate cyclase works in association with phytochrome [37], an up-regulated protein found in Spirulina-TM. These proteins are part of the cAMP-dependent light signaling cascade. Adenylate cyclase has also been reported to be regulated at the post-translational level by ligand binding, protein binding and phosphorylation [37]. Together, our results demonstrate the association between high temperature response and the light signaling cascade.

The proteins with significantly differential expression in each subcellular fraction were clustered, based on their expression patterns [see fig S6; Additional file 2]. According to Lacerda et al., the expression patterns in response to stress can be classified into three major groups: resistance, adaptation and sustained tolerance [38]. The results shown in Fig. 11, 12 and 13 demonstrate that the majority of proteins in every subcellular fraction belong to the sustained tolerance expression pattern. If all differentially expressed proteins are set as 100%, the percentages of the resistance, adaptation and sustained tolerance groups are: (i) 7%, 3% and 46% in the PM fraction, (ii) 9%, 10%, 46% in the soluble fraction and (iii) 18%, 12% and 58% in the TM fraction, respectively. It should be noted that some patterns do not fit into any categories, and these patterns were mostly found in the PM fraction. Moreover, the plasma membrane, where the environmental changes are first encountered, is the only site where the resistance proteins are present at a significantly higher level than the adaptation proteins.

Site-specific DNA methyltransferase (cytosine-specific) is the only resistance protein that was identified in this study. The level of this protein initially increased and subsequently decreased in the TM fraction (Table 4). DNA methylase is involved in the DNA repair system, and it shows the same expression pattern in response to cadmium stress, which is known to induce DNA-damage [38]. Most of the two component signal transduction systems, stress-related proteins and proteins involved in DNA-damage and DNA-repair are classified in the sustained tolerance group (Fig. 11). This suggests the critical role of these proteins in the tolerance to high temperature stress in Spirulina. It is noteworthy that the resistance proteins (short-term only response) were present at a significantly higher level (2-fold) in the thylakoid membrane than in the other two fractions (Fig. 13). Additionally, adaptation proteins (long-term only response) were found at a higher level in the soluble and the thylakoid membrane fractions (Fig. 12 and 13) than the plasma membrane fraction (Fig. 11).

Several differentially expressed proteins identified in this study can be mapped onto the PPI network available on Cyanobase (Fig. 14, 15 and 16). The potential PPIs shown in the three subcellular fractions represent interesting linkages or cross-talks among the three cellular compartments. For example, in the PM fraction, two component system sensory histidine kinase (spot#1564), ABC transporter (spot#2179), ferredoxin-glutamate synthase (spot#1388) and carboxypeptidase (spot#1284) show interactions with the photosynthetic system. In the soluble fraction, the phosphorylated form of multi-sensor signal transduction histidine kinase (spot#1883) interacts with several periplasmic proteins. However, in the TM fraction, the same protein was found in the non-phosphorylated form. The interactions found in the thylakoid membrane also show communication with the other two fractions.

Additionally, PPI networks clearly demonstrate the linkage between high temperature stress and nitrogen and ammonia assimilation in Spirulina. It is well established that photosynthesis and nitrate reduction are closely related in cyanobacteria and plants, via the nitrate reductase requirement of photoreduced ferredoxin [39,40]. In response to heat stress, inhibition of photosynthesis and nitrate reductase was observed. Moreover, it was reported by Rajaram and Apte [40] that a Hsp60 family protein, Cpn60, which is induced by heat stress and stabilized by nitrogen supplementation, either from nitrate or ammonia, is essential for the thermal stability of these vital metabolic processes.