Introduction

Diatoms, which play a vital role in the biogeochemical cycle, are responsible for one-fifth of the primary productivity on Earth1,2,3. This phylogenetically young group includes two major classes, the pennates and the centrics and they are widespread in all kinds of aquatic environments4. The activity of diatoms dominates silicon cycling in the ocean5,6. For most diatoms, silicic acid greatly influences cell cycle progression and they will stop at the G1/S or G2/M transition if the silicon supply is deficient7,8. It has been suggested that biogenic silica is an effective pH buffer that facilitates the enzymatic conversion of bicarbonate to dissolved CO2, thus improving the efficiency of photosynthesis in diatoms9. Consequently, silicon plays a crucial role in the growth of diatoms.

Phaeodactylum tricornutum Bohlin is the first pennate diatom for which the complete genome is known10. It has three convertible morphotypes: oval, fusiform and triradiate11. Its outer shell is weakly silicified and silicification is restricted to one valve of the oval cells12,13. In a specific P. tricornutum accession, the most frequent morphotype is usually fusiform or triradiate, whereas the concentration of oval cells is very low. However, fusiform and triradiate cells can transform into oval forms under unfavorable growth conditions13. P. tricornutum is widespread in both coastal and inland waters, usually in unstable environments, including estuaries and rock pools13,14. Temperature, light quality and salinity change rapidly in these environments15 and the living environment in surface water greatly differs from that of bottom water. De Martino et al. (2011) proposed that different shapes of P. tricornutum are more prevalent under different culture conditions. For example, oval cells are better acclimated to benthic environments, as they have higher sedimentation rates and can adhere and glide on surfaces, whereas fusiform and triradiate cells are better acclimated to planktonic environments16. Oval cells also are better able to survive under stressed conditions (e.g., limited nutrients) and can convert to fusiform or triradiate cells under favorable conditions. Thus, it seems that conversion of the three morphotypes of P. tricornutum occurs when environmental factors change and the requirement of silicon for the growth of P. tricornutum is probably correlated with environment conditions.

In laboratory culture, the habitat is stable and conditions usually are optimal, which differs significantly from the situation in the natural environment. Under laboratory conditions, silicon is not required for the growth of P. tricornutum13, although all three morphotypes assimilate silicon11,17,18,19,20. Yang et al. (2013) reported that deprivation of silicon during P. tricornutum cultivation resulted in a better growth characteristics due to the omission of silicon induced cell breakage21. However, our previous study showed that miRNAs of silicon starved P. tricornutum were significantly different from those of the normal cultured cells22. As miRNAs are important post-transcriptional regulators of gene expression in eukaryotes23, these observed differences in miRNA demonstrate that silicon affects the physiological processes of P. tricornutum. However, silicon's role in P. tricornutum growth in nature is poorly understood.

In this study, we compared the growth rate of silicon starved P. tricornutum with that of normal cultured cells under different culture conditions. Pigment analysis, photosynthesis measurement, lipid analysis and proteomic analysis were performed to evaluate the role that silicon plays in P. tricornutum growth.

Results

Growth of P. tricornutum

To study the role that silicon plays in the growth of P. tricornutum, culturing experiments were conducted. Figure 1a shows little difference between the growth curves of P. tricornutum under normal and silicon starved culture conditions. Figures 1b–h show the growth curves of normal cultured and silicon starved P. tricornutum when cultured under low salinity (salinity 20‰), high light (2000 μmol m−2 s−1), different photoperiods and different nutritional deficiencies (iron starvation or nitrogen starvation). Silicon did not influence the growth rate of P. tricornutum under any of these conditions.

Figure 1
figure 1

Value of OD730 of P. tricornutum under different culture conditions.

(a) Normal (20°C, salinity 30‰, 24 μmol m−2 s−1 and 12:12 h light/dark cycle). (b) Low salinity (salinity 20‰). (c) High light (2000 μmol m−2 s−1). (d) 4:20 h light/dark cycle. (e) 20:4 h light/dark cycle. (f) 24:0 h light/dark cycle. (g) Iron starvation. (h) Nitrogen starvation. (Normal, normal cultured P. tricornutum. Si-, silicon starved cultured P. tricornutum.) The data are the mean of three independent experiments (±SD).

Figure 2 shows the growth of cells cultured under different light qualities or low temperature (10°C). Silicon had no impact when P. tricornutum was cultured under red light (wavelength 647–700 nm) (Fig. 2a), whereas cells cultured under blue light (wavelength 470–475 nm) died after 3 days of culture (Fig. 2b). As 470–475 nm is located in the wave valley of the absorption spectra of chlorophyll a24. That is, the blue light cannot absorbed by chlorophyll a, that may be the reason for the death of the algae. Growth of normal cultured and silicon starved P. tricornutum differed when cultured under green light (wavelength 491–574 nm) and low temperature (Fig. 2c–e), as the cells grew more slowly when silicon was unavailable. These results show that silicon influences the growth of P. tricornutum under specific conditions, including low temperature and green light.

Figure 2
figure 2

Value of OD730 of P. tricornutum under different culture conditions.

(a) Red light (wavelength 647–700 nm). (b) Blue light (wavelength 470–475 nm). (c) Green light (wavelength 491–574 nm). (d) Low temperature (10°C). (e) Green light and low temperature. (Normal, normal cultured P. tricornutum. Si-, silicon starved cultured P. tricornutum.) The data are the mean of three independent experiments (±SD).

Pigment analysis

Under the normal culture condition, the concentrations of fucoxanthin and chlorophyll a were 11.88 ± 0.32 and 29.69 ± 1.47 (10−5 ng/cell) (Fig. 3a). When silicon was not available, the concentration of the two pigments decreased to 10.88 ± 0.13 and 24.71 ± 0.52 (10−5 ng/cell) after 48 h of cultivation. Thus, silicon influenced pigment biosynthesis or degradation in P. tricornutum.

Figure 3
figure 3

Pigment contents and photosynthesis data of P. tricornutum under different culture conditions.

(a) Fucoxanthin and chlorophyll a contents. (b) Electron transport rates of PSII (ETR(II)) and PSI (ETR(I)). (c) Effective quantum yields of PSII (Y(II)) and PSI (Y(I)) and maximum quantum yield of PSII (Fv/Fm). (Normal, normal cultured P. tricornutum. Si-, silicon starved cultured P. tricornutum.) The data are the mean of three independent experiments (±SD).

Chlorophyll fluorescence measurements

Figure 3c show the effective quantum yields of PSII (Y(II)) and PSI (Y(I)) and the maximum quantum yield of PSII (Fv/Fm). Y(II) and Y(I) values did not differ significantly between normal cultured and silicon starved P. tricornutum and the same was true for the electron transport rates of PSI (ETR(I))and PSII (ETR(II)) (Fig. 3b). The Fv/Fm value of normal cultured P. tricornutum was 0.69 ± 0.016, whereas the value was 0.66 ± 0.007 after 48 h of silicon starvation. Fv/Fm reflects the maximum quantum yield of PSII and is a sensitive indicator of the photosynthetic performance of plants. It provides important information about the effect of environmental stress on the plant25. The decline in the Fv/Fm value shows that the lack of silicon decreased the antireversion force of P. tricornutum, but the lack of differences in the other photosynthesis data showed that it did not influence the photosynthetic efficiency during the experiment.

Lipid analysis

The incipient relative fluorescence intensity of Nile red stained P. tricornutum was 45.08 ± 0.85 and the value dropped to 16.95 ± 1.21 and 11.47 ± 0.36, respectively, after re-culture in sterilized artificial seawater with normal or silicon-free f/2 media for 48 h. Figure 4 shows P. tricornutum cells stained with Nile red. Nile red stained lipid bodies showed characteristic yellow fluorescence under the microscope, with no significant difference between the two type of cultured cells.

Figure 4
figure 4

P. tricornutum cells stained by Nile red.

(a) Normal cultured P. tricornutum. (b) Silicon starved P. tricornutum.

Table 1 shows the fatty acid composition of normal cultured and silicon starved P. tricornutum. Total saturated fatty acids and total unsaturated fatty acids of normal cultured P. tricornutum represented 13.95% and 61.89% of the fatty acid content, respectively. The amount of total saturated fatty acids increased to 15.07% and that of total unsaturated fatty acids dropped to 61.14% when the cells were starved of silicon for 48 h. These results indicate that silicon deficiency did not influence the lipid content but did influence the fatty acid composition of P. tricornutum.

Table 1 Fatty acid composition of P. tricornutum under different conditions. (Normal, normal cultured P. tricornutum. Si-, silicon starved P. tricornutum.)

Protein extraction

2-DE analysis of the extracted total protein was conducted before LC–MS/MS was carried out to ensure that the protein was properly prepared for further analysis. In Fig. 5, the protein spots are clearly displayed. Analysis using PDQuest 2-D Analysis Software (Bio-Rad) identified 1,293 and 1,559 protein spots in normal cultured and silicon starved P. tricornutum, respectively. The 2-DE results indicated that there were no problems with these proteins, thus they were suitable for subsequent LC–MS/MS analysis.

Figure 5
figure 5

Proteins separated (1.5 mg) by 2-DE (pI4-7).

(a) Normal cultured P. tricornutum. (b) Silicon starved P. tricornutum.

Protein expression and identification by LC–MS/MS analysis

LC–MS/MS analysis

For LC–MS/MS, each sample was analyzed three times. The whole analysis yielded 374 positive identifications with a protein score exceeding 15. Among them, 33 were up-regulated and 37 were down-regulated in silicon starved cells. The proteins of interest are shown in Table 2. The potential cellular functions of all of the differentially expressed proteins identified were searched for in UniProt (http://www.uniprot.org/) (Table 2, Fig. 6). More than one-third of the proteins had no notation in the database. Other proteins were classed into categories such as carbon metabolism, lipid metabolism, protein synthesis and photosynthesis. Below we describe our proteome results according to this classification.

Table 2 Functional categories of selected proteins. (Differential expression proteins involved in carbon metabolism, pigment synthesis, photosynthesis and lipid metabolism in 48 h silicon starved P. tricornutum compared with that normal cultured. UniProt, Accession numbers from http://www.ebi.uniprot.org/index.shtml. Up/Down, Silicon starved P. tricornutum compared with normal cultured cells.)
Figure 6
figure 6

Functional categories of differential expressed proteins separated by LC–MS/MS.

Central carbon metabolism

The abundance of two proteins, fructose-1,6-bisphosphate aldolase precursor (UniProt Accession number Q84XB5) and fructose-bisphosphate aldolase (B7GE67), which are related to glycolysis, changed after 48 h silicon starvation. Expression of two proteins, acetyl-CoA carboxylase (B7G7S4) and long chain acyl-CoA synthetase (B7FYK0), which are involved in lipid synthesis, was up-regulated when silicon was unavailable. Acetyl-CoA carboxylase catalyzes the first reaction of the fatty acid biosynthetic pathway, which involves the formation of malonyl-CoA from acetyl-CoA and CO226. This reaction is a committed step in fatty acid synthesis and the enzyme is key in regulating rates of fatty acid synthesis27. However, not all of the malonyl-CoA generated by acetyl-CoA carboxylase is a precursor for de novo fatty acid biosynthesis and elongation; some of it contributes to secondary metabolites, such as flavonoids, stilbenoids, malonic acid and malonyl derivatives28. It is difficult to conclude that the up-regulation of acetyl-CoA carboxylase can directly lead to the accumulation of fatty acids.

Expression of two subunits (H1A8C7 and A0T0E2) of the key enzyme of the Calvin cycle, ribulose-1,5-bisphosphate carboxylase/oxygenase (Rubisco), was down-regulated when the cells were lack of silicon. Rubisco catalyzes the rate-limiting step of CO2 fixation in photosynthesis. The large subunit of Rubisco contains the active site, whereas the small subunit is required for maximal catalysis and contributes to CO2/O2 specificity29. However, down-regulation of only one enzyme of the Calvin cycle may not influence CO2 fixation and algal growth rate.

Photosynthesis and pigment biosynthesis

Several proteins, such as fucoxanthin chlorophyll a/c proteins (FCPs) (B7G871, etc.), were down-regulated after 48 h silicon starvation. FCPs are the antenna proteins of diatoms; they are homologs of the light-harvesting complexes (LHC) of higher plants. Down-regulation of FCPs might be accompanied by down-regulation of photosynthesis and photosynthetic pigment production.

The levels of a predicted protein (B7FP19) related to the chlorophyll biosynthetic process decreased when the silicon was unavailable. This indicates that the chlorophyll content may have decreased as well. Pigment analysis showed decreased levels of chlorophyll a and fucoxanthin, which was consistent with the proteome results. Photosynthesis analysis showed that the Fv/Fm value dropped, whereas other photosynthesis data did not change, after 48 h of silicon starvation. As previously mentioned, lack of silicon decreased the antireversion force of P. tricornutum, but it did not appear to influence the photosynthetic efficiency.

Protein metabolism and chromosome integration

Expression of two translation elongation factors (B5Y4J2 and B8LEI9) was down-regulated when silicon was unavailable. Three predicted proteins (B7G715, B7FQU7 and B7G9G2) that function as structural constituents of ribosomes were regulated in two ways: two were up-regulated and one was down-regulated. A ubiquitin extension protein (B7FY02) was down-regulated. The proteome results suggested that protein synthesis declined and protein degradation increased when the cells were silicon starved. Three predicted proteins (B5YMD6, B7GB44 and B5Y3W7) for histone H3, nucleotide binding protein and tubulin, respectively, were up-regulated when silicon was unavailable. The increased chromosome and cytoskeleton components suggested that chromatin tended to concentrate to the chromosome, prepared for a suitable condition for mitosis when the cells were lack of silicon.

Signal transduction

Expression of a polyubiquitin (Q1EFP9) and an annexin (B5Y5D4) was down-regulated when silicon was absent. Polyubiquitins are polymers formed through ubiquitin-ubiquitin conjugation, which usually occurs within cells and can be linked to target proteins30. Annexins are Ca2+ and phospholipid binding proteins and they are present in both eukaryotes and prokaryotes31. Mortimer et al. (2008)32 posited that annexins may be central regulators or effectors of plant growth and stress signaling. The decrease in stress tolerance associated proteins indicated that when the cells lacked silicon, their stress tolerance may have declined.

Expression of a predicted protein (B7G386) with peroxidase activity and expression of a precursor of mutase superoxide dismutase (B7G0L6) were up-regulated when the cells were silicon starved. Peroxidase can respond to oxidative stress and superoxide dismutase catalyzes the dismutation of superoxide radicals33. As the content of fucoxanthin, an effective antioxidant34, decreased when silicon was unavailable, the level of toxic reactive oxygen species (ROS) may have increased. Thus, the up-regulation of peroxidase and superoxide dismutase may have been a response to the increased level of ROS.

A predicted protein (B7G195) belonging to the small heat shock protein (HSP20) family was up-regulated after 48 h silicon starvation. HSPs can be induced by a wide variety of stressors, including elevated temperature. Expression of two predicted proteins (B7G4Y3 and B7G0Y4) that function as an intracellular protein transporter and an ammonium transmembrane transporter, respectively, was up-regulated when silicon was absent.

Analysis of mRNA expression

The mRNA expression levels of a randomly selected set of differentially expressed proteins in normal cultured and silicon starved P. tricornutum were measured by qRT-PCR using the RPS (ribosomal protein small subunit 30S) gene as an internal control. The RPS gene was stably expressed in P. tricornutum. The primer efficiencies of RPS, PFC (protein fucoxanthin chlorophyll a/c protein, B7G871), ANN (annexin, B5Y5D4), RBC (ribulose-1,5-bisphoaphate carboxylase/oxygenase small subunit, A0T0E2) and UBI (polyubiquitin, Q1EFP9) were 104.0, 105.0, 96.0, 100.9 and 103%, respectively. The level of ANN, RBC and UBI expression in the normal cultured P. tricornutum was higher than that in the silicon starved cells (Fig. 7). This finding agrees with the proteome results, which showed a tendency for expression to decline at both the transcriptional and translational level when silicon was absent. The lack of a significant difference in PFC expression suggests that this gene may be post-transcriptionally controlled.

Figure 7
figure 7

Real-time fluorescent quantitative PCR (qPCR) of differential expressed proteins of normal and silicon starved P. tricornutum.

Discussion

The ocean is usually thought to be made of three principal layers, the surface layer is known as the mixed layer as much of the time it is mixed by currents, wind and waves and it is usually around 100–200 m thick35. However, the surface layer is not always well mixed, the very uppermost part of the surface layer is heated by the sun, the warm water floats on top and there is a sharp transition to the cooler water below35. The temperature can change from about 25°C at the surface to below 10°C at the bottom of the layer, which differs in different latitudes. This phenomenon usually exists in the spring and summer in temperature and polar water and the layer is called seasonal thermocline35. In clear ocean water blue light penetrates the deepest, secondly green light, for about 40–50 m, red light the least. However, coastal water often contains a lot of materials brought in by rivers that absorb blue light so that green penetrates deepest35. P. tricornutum can move vertically in the ocean and it is widespread and occurs at different depths in the ocean. The oval cells are better acclimated to benthic environments, whereas fusiform and triradiate cells are better acclimated to planktonic environments16. The growth curves in the current study show that silicon influenced the growth rate of P. tricornutum only under conditions of low temperature and green light (Fig. 2); there was little difference under other conditions (Fig. 1). As temperature is lower in deep water and green and blue light are predominant, growth of P. tricornutum cells could be affected when they move to deep water. Lu et al. (2001) has reported that a high proportion of ovals was produced in fusiform trains under long-term stress of lower temperature36. Silicon starvation may interfere with the conversion from fusiform and triradiate cells to oval cells, which is the silicification valve-containing morphotype that is better acclimated to unfavorable growth conditions13,36. Thus, we suggest that silicon may affect the growth of P. tricornutum in deep water.

If the growth rate of P. tricornutum can be changed by silicon availability in deep water, the light-harvesting complex may be affected as well. As green and blue light are predominant in deep water, the corresponding photosynthetic pigment may be also influenced. Figure 3a and table 2 show that fucoxanthin and protein fucoxanthin chlorophyll a/c protein decreased as the cells were starved of silicon. The absorption maximum of fucoxanthin is near 490 nm. When it attaches to a protein in cells, the maximum is shifted by about 40 nm to longer wavelengths, thereby making available a larger fraction of the green light37. As fucoxanthin decreased when the cells were silicon starved, the absorption of green light decreased. The effective binding of fucoxanthin to the light-harvesting complex may also influence the absorption of light. If fucoxanthin cannot bind to the light-harvesting protein effectively when the cell is silicon starved, then photosynthesis efficiency as well as growth rate may decrease correspondingly. The growth curves under green light were consistent with the pigment and proteome results.

Fucoxanthin is an effective antioxidant34, thus decreased fucoxanthin led to increased ROS. In response, expression of peroxidase and superoxide dismutase increased. Results of this study indicate that the gene for protein fucoxanthin chlorophyll a/c protein may be post-transcriptionally controlled and our previous study of the specific expression of an miRNA targeting a protein fucoxanthin chl a/c protein under silicon starvation supports this premise22.

Finley et al. (1987)38 reported that the yeast polyubiquitin gene is essential for resistance to stresses such as high temperatures and starvation. One kind of polyubiquitin, called K63-linked chains, was reported to function in four pathways: the inflammatory response, DNA damage tolerance, ribosomal protein synthesis and protein trafficking30. Another stress tolerance associated protein, annexin, has multiple functions, such as secretion, possible enzyme activity or interaction with other cellular proteins, interaction with actin, nucleotide phosphodiesterase activity and acting as a substrate for protein phosphorylation, calcium channels and peroxidases39. Annexins also may be central regulators or effectors of stress signaling and plant growth32. The decreased expression of two stress tolerance associated proteins in silicon starved P. tricornutum indicates that when the cells were starved of silicon, their stress tolerance may have declined.

Hazel (1995) reported that almost all poikilotherms increase the level of unsaturated fatty acids to maintain the appropriate fluidity of membrane lipids when the ambient temperature decreases40 and Graham and Patterson (1982) found that the production of unsaturated fatty acids increases at low temperature in many plants41. Furthermore, cyanobacteria can enhance cold tolerance by increasing the amount of unsaturated fatty acids in genetically engineered membrane lipids as the cold-induced expression of genes for fatty acid desaturases increases42. Previous studies have shown that unsaturated fatty acids are important in maintenance of the photosystem II complex42,43, particularly at low temperatures. Table 1 shows the observed changes in fatty acid composition and that total unsaturated fatty acids decreased when the cells were silicon starved. The decrease in total unsaturated fatty acids may influence the growth of P. tricornutum at low temperature by decreasing its tolerance to cold; that is, silicon may be beneficial for P. tricornutum living in deep water.

Methods

Strains and culture conditions

For the P. tricornutum culture, f/2 medium44 made with steam-sterilized local seawater with f/2 vitamins (filter sterilized) and inorganic nutrients added was used. Cultures were incubated under cool-white fluorescent lights at 24 μmol m−2 s−1 on a 12:12 h light/dark cycle at 20°C for about 8 d and periodically stirred by hand. Cells were harvested by centrifugation for 5 min at 4,000 g and washed with sterilized artificial seawater45. Cells were re-cultured in normal or silicon-free f/2 media made with sterilized artificial seawater for 48 h. Cells were harvested by a two-step centrifugation process, initially at 4,000 g for 5 min and, after transfer to 1.5-mL Eppendorf tubes, 10,000 g for 2 min. Cell pellets were frozen instantly in liquid nitrogen and stored at −80°C before pigment extraction and protein extraction.

Growth measurement

Normal cultured cells and silicon starved cells for growth measurement experiments were cultured under conditions including low salinity (salinity 20‰), low temperature (10°C), different light quality (high light, red light, green light and blue light), different photoperiod and nutrient starvation (iron starvation, nitrogen starvation). The cell number was determined at a wave length of 730 nm using a UV-spectrophotometer (UV-1800, SHIMADZU, Japan) connected to a hemocytometer under an optical microscope (Nikon Eclipse E100, Japan).

Pigment analysis

The pigment extraction and HPLC analysis were carried out according to the method of Thayer and Björkman (1990) and Enriquez et al. (2010)46,47 with some modifications. All procedures were carried out in dim light. The collected cells were suspended in acetone/methanol (1/1, v/v), placed in an ice bath for 30 min and then centrifuged at 3,000 g for 5 min. The supernatant containing the pigments was collected and filtered through a 0.22 μm syringe filter. The filtered pigments were injected into an Agilent 1200 HPLC equipped with an Rx-C18 analytical column (4.6 × 250 mm) (Agilent Technologies Inc., Santa Clara, CA, USA). Analysis of the mobile phase, which consisted of water, methanol, acetonitrile and ethyl acetate, was programmed as follows: 0–15 min, linear gradient from I (15% water, 30% methanol, 55% acetonitrile, 0% ethyl acetate) to II (0% water, 15% methanol, 85% acetonitrile, 0% ethyl acetate); 15–17 min, linear gradient from II to III (15% water, 15% methanol, 35% acetonitrile, 35% ethyl acetate); and 17–40 min, linear gradient from III to IV (0% water, 30% methanol, 0% acetonitrile, 70% ethyl acetate). The column temperature was 50°C and the flow rate was 0.75 mL/min. The pigments were detected by their absorbance at 443 nm.

Chlorophyll fluorescence and P700 measurement

Assessment of photosynthesis was carried out as described48 with minor modification using a chlorophyll fluorometer Dual-PAM-100 (Heinz Walz GmbH, Germany). Briefly,samples of normal cultured cells and silicon starved cells were dark-adapted for 15 min before the measurement. The induction curve was measured using the dual channel mode (Fluo + P700) with actinic light set at 24 μmol m−2 s−1. The effective quantum yield and electron transport rate of both PS II and PS I (known as Y(II), Y(I), ETR(I) and ETR(II) respectively) and the maximum quantum yield of PS II (Fv/Fm) were calculated based on the data acquired during induction curve measurement.

Lipid analysis

Lipid content was determined by Nile red fluorescence staining following Liu et al. (2008)49, the total lipids were extracted with chloroform/methanol (2/1, v/v) and the composition of fatty acids was quantified by gas chromatography50. That is, normal cultured cells and silicon starved cells were mixed with Nile red solution (0.1 mg mL−1 in acetone) (100/1, v/v), respectively and settled for 7 min. Then the cells stained with Nile Red were analyzed on a fluorescence spectrophotometer (HITACHI F-4500, Tokyo, Japan) with 480 nm as the excitation wavelength and observed by Laser Confocal Microscopy, using a Carl Zeiss microscope with blue light as the excitation light.

Preparation of total protein

Cell pellets were ground in liquid nitrogen using a mortar and pestle with silica sand and polyvinyl-polypyrrolidone added. Pulverized cells were resuspended in 15 mL extraction buffer (5% w/v sodium dodecyl sulfate (SDS), 10% v/v glycerol, 5% v/v β-mercaptoethanol, 1% v/v complete protease inhibitor cocktail, 65 mM Tris-HCl pH 6.8) at 4°C for 1 h. The cell lysate was centrifuged at 8,000 g for 30 min at 4°C and the cell debris was removed. Proteins were precipitated overnight from the supernatant in 10% w/v trichloroacetic acid in 100% ice-cold acetone, centrifuged at 8,000 g for 30 min, rinsed three times in 100% ice-cold acetone and rinsed once in 80% ice-cold acetone51. The pellet was air dried at 4°C.

Two-dimensional gel electrophoresis (2-DE) and image analysis

The protein was resolubilized in 100 μL of 0.2 M NaOH52 and 2 mL of protein lysis buffer (7 M urea, 2 M thiourea, 4% w/v 3- [(3-cholamidopropy) dimethyl-ammonia]-1-propanesulfonate (CHAPS), 65 mM dithioethreitol (DTT)) overnight and then ultrafiltrated for 2 h. It was analyzed for total protein using bovine serum albumin as the standard53. A protein sample containing 1.5 mg of protein was mixed with 400 μL of solubilization buffer (7 M urea, 2 M thiourea, 4% w/v CHAPS, 65 mM DTT, 0.001% w/v bromochlorophenol blue, 0.2% w/v carrier ampholytes (Bio-lytes 3–10)). Immobilized pH gradient (IPG) strips (pI 4–7; 17 cm; Bio-Rad, Hercules, CA, USA) were rehydrated with the mixture at room temperature for 1 h then actively rehydrated at 50 V at 20°C for 18 h.

Isoelectric focusing (IEF) was performed at 20°C using the following parameters: linear 100 V for 1 h, 250 V for 1 h, 500 V for 1 h, a rapid increase from 500 V to 1,000 V for 1 h, a linear increase from 1,000 V to 8,000 V for 5 h and then a hold at 8,000 V for a total of 60 kVh. After focusing, the IPG strips were equilibrated in 10 mL of equilibration buffer I (6 M urea, 2% w/v SDS, 20% v/v glycerol, 0.375 M Tris-HCl pH8.8, 65 mM DTT) followed by equilibration buffer II (6 M urea, 2% w/v SDS, 20% v/v glycerol, 0.375 M Tris-HCl pH8.8, 260 mM iodoacetamide) for 15 min each at room temperature. The gel strips were then washed in MilliQ water and equilibrated for 2 min in SDS-PAGE running buffer. The gel strip was positioned on top of a vertical 12% polyacrylamide gel, held in place by molten agarose (0.5% w/v agarose, 25 mM Tris, 192 mM glycine, 0.1% w/v SDS, 0.001% w/v bromochlorophenol blue) and electrophoresed at a constant voltage of 80 V for 30 min and then 200 V for 8 h per gel. The gels were stained with blue silver54, then scanned with Image Scanner (Bio-Rad GS-800). The images were analyzed using PDQuest 2-D Analysis Software (Bio-Rad).

Quantitative protein analysis by liquid chromatography/tandem mass spectrometry (LC-MS/MS)

Protein solubilization was carried out as described55. Briefly, the protein was resolubilized in 8 M urea (125 mM NH4HCO3, pH 8) at 90°C for 20 min, cooled on ice and quantified53. Protein was in-solution tryptic digested as previously described. The lysate was reduced by 10 mM DTT at 37°C for 1 h and then alkylated with 50 mM iodoacetamide for 30 min in the dark. Sequencing grade modified trypsin was used for in-solution digestion with trypsin/protein at 1/30. The peptides were acidified with 1% formic acid and stored at −80°C for LC-MS/MS analysis. The tryptic digested peptides were loaded onto a 2.1 mm × 150 mm reverse-phase column (Zorbax SB-C18, Agilent) connected to an Agilent 1200 HPLC system.

Peptides were eluted with gradient mobile phase solution A (0.1% formic acid in distilled water) and B (0.1% formic acid in acetonitrile) following Blonder et al. (2007)56 with minor modifications. An acetonitrile gradient (3% B for 5 min, linear gradient from 3 to 50% B in 120 min; 95% B during 120 to 150 min) at a flow rate of 0.2 mL/min was used to elute the peptides into the Agilent 6520b Q-TOF mass spectrometer. The electrospray voltage used was 3.5 kV and the drying gas temperature was set at 350°C. The fragmentor was set at 175 for ion transfer. Nitrogen gas was used during MS/MS analysis. Data acquisition was carried out in auto MS/MS mode using MassHunter software (version B03.01, Agilent).

All MS/MS spectra were processed using the Spectrum Mill MS Proteomics Workbench (version A.03.03, Agilent) and filtered MS/MS spectra data were searched against the NCBI diatom database (4/30/13 download) for protein identification. Precursor mass tolerance was set at ±20 ppm. One missed cleavage site was allowed during sequence matching. All protein identifications with a protein score ≥ 15 and a peptide scored peak intensity (SPI) ≥ 60% were considered as positive identifications. One shared peptides grouping method were used for protein rolling up. Finally, NCBI accession numbers were changed to UniProt numbers and searched through UniProt for molecular function and cell location was summarized for the proteins in each sample.

Quantitative real time PCR (qRT-PCR) analysis

Total RNA was isolated using TRIZOL reagent (Gibco BRL, Grand Island, NY, USA), analyzed by agarose electrophoresis and reverse-transcribed to cDNA. All procedures were carried out as described57. The transcript level of several differentially expressed proteins in normal cultured and silicon starved P. tricornutum was measured by qPCR using an iQ5 multicolor real time PCR detection system (Bio-Rad) in a total volume of 25 μL containing 12.5 μL of 2 × SYBR green master mix (Tiangen Biotech, Beijing, China), 2.5 μL (2 mM) of each primer, 2.5 μL of the diluted cDNA mix and 5 μL of RNase-free water. The thermal profile for real time PCR was 95°C for 3 min followed by 35 cycles of 95°C for 15 s and 62°C for 30 s, followed by 72°C for 30 s. Dissociation curve analysis of the products was conducted at the end of each PCR reaction to confirm the presence of only one specific PCR amplification product. Triplicate qPCRs were performed for each sample and the data were analyzed using the 2−ΔΔCt method58. The primers used are shown in Table 3 and the RPS (ribosomal protein small subunit 30S) gene was used as the internal control59.

Table 3 Primers used in this study

Statistical analysis

All data shown are the mean ± standard deviation of results of three independent experiments. All statistical analyses were conducted using SPSS. One-way analysis of variance (ANOVA) was used to identify significant differences among groups. Differences were considered to be statistically significant at P < 0.05.