INTRODUCTION
As photosynthesizing unicellular microorganisms, diatoms play an important role as primary producers in marine ecosystems and they are responsible for 20–25% of carbon recycling (Smetacek 1999, Scala and Bowler 2001). Another element that is recycled by diatoms on a global scale, affecting biogeochemistry, is silicon (Si), which is estimated to be recycled at 240 ± 40 Tmol per year (Nelson et al. 1995). Silicon constitutes the frustule which is silicified cell walls of diatoms. The intricately patterned nanostructures of this tiny shell have inspired many studies on the biomineralization process of Si, as well as applications in nanotechnology, such as bio-manufacturing, fabrication of nanostructures (Parkinson and Gordon 1999, Huang and Daboussi 2017, Roychoudhury et al. 2022), and biomedical applications (Sardo et al. 2021).
Observations using electron microscopy have greatly contributed to our understanding of the morphology and structure of frustules, and this knowledge has been applied to the classification of diatoms. Additionally, monitoring the exocytosis of the silica frustules has provided deeper insights into biosilicification and frustule formation. Fluorescent probes are highly effective tools for this investigation, because they can be used to study the incorporation of Si into newly synthesized frustules. LysoTracker HCK-123, introduced by Desclés et al. (2008), is an excellent tracer for newly synthesized biosilica shells. This probe offers several advantages over other fluorescence markers, including no need for ionophores, compatibility with common filter sets for visualization, an enhanced signal-to-noise ratio, and stability in acidic cellular compartments.
The absorption of Si into cells is a fundamental step in frustule formation. In aquatic environments, Si exists as dissolved silicates (DSi), such as silicic acid and Si(OH)4. Above certain concentrations, silicic acid can freely diffuse into the cells (Raven 1983, Thamatrakoln and Hildebrand 2008). However, the Si concentration in the ocean is less than 30 μM, and therefore, active silicon transporters are required (Tréguer et al. 1995, Thamatrakoln and Hildebrand 2008). In diatoms, silicon transporter proteins, known as SITs are reported to be responsible for silicon uptake. These integral membrane proteins are believed to function as symporters of Na+, demonstrating a relatively high affinity for Na+ ions (Knight et al. 2016).
In this study, we identified and analyzed SIT sequences and expression patterns of the centric diatom Thalassiosira eccentrica LIMS-PS-3165, isolated from the southern coastal region of South Korea. We also traced frustule synthesis via cell-cycle synchronization using a fluorescent probe under nutrient-limited conditions. The species in this study exhibited a unique shape and arrangement of areolae compared to other T. eccentrica reported earlier, which may be related to the frustule formation process. Although this study only examined some characteristics of the SIT proteins involved in Si absorption and transport, it provides important molecular biological insights into the local strain of T. eccentrica, a species that has been less studied. In particular, when compared with the well-studied model species T. pseudonana, this research could be valuable for future comparative studies.
MATERIALS AND METHODS
Algal strain and culture conditions
The T. eccentrica LIMS-PS-3165 strain was obtained from the Korea Institute of Ocean Science and Technology (KIOST) and was isolated from the coastal regions of Jeju, South Korea. Stock cells were batch cultured in L1 medium (Guillard and Hargraves 1993) at 20°C under a 12 : 12 light : dark regime at 55 μmol photons m−2 s−1 at 50% humidity. For pre-experimental culture, cells were harvested by centrifugation (1,500 rpm, 20°C, 5 min), and washed with L1 medium twice, and then inoculated into fresh medium at a density of 0.1 × 104 cells mL−1. When the cells reached to exponential growth phase (approximately 0.5 × 104 cells mL−1), cells were harvested by centrifugation, washed twice with fresh experimental medium, and inoculated into new media for experiments. The media used for the nutrient deprivation experiment lacked one of the three nutrients from the L1 medium: nitrogen (N), phosphorus (P), or silicon (Si). For the Si availability experiment, the washed cells were arrested by incubating in Si-lacking medium for 1 d, followed by transfer to either Si+ or Si− media.
Cell density was measured by counting under light microscopy using hemocytometer (Marienfeld, Gütersloh, Germany), eight times as technical replicates, and the counts were averaged after excluding the highest and lowest values. All growth experiments included a minimum of three biological replicates.
Spectrophotometric measurement of mono silicic acid via molybdate blue method
Monosilicic acid, Si(OH)4 was determined based on the molybdate blue method using a scaled-down protocol provided with Hach silica testing kit (Hach Gmbh, Düsseldorf, Germany). One milliliter of the cell culture medium was centrifuged at 3,200 rpm to remove the cells. The remaining media samples were diluted by a dilution factor of 5. Forty microliters of molybdate 3 reagent was added and vortexed. The samples were incubated for 4 min. Forty microliters of citric acid reagent was added, vortexed, and incubated for 1 min. One packet of amino acid F reagent was dissolved in 1 mL dH2O and 40 μL of amino acid F reagent was added and vortexed. The sample was measured using an Ultrospec 2100 Pro (Amersham, GE Healthcare, Piscataway, NJ, USA) at OD803 using distilled H2O as a blank. A reagent blank sample was also prepared and measured but was omitted from the calculations owing to its negligible values. The optical density was measured for distilled H2O spiked with silicic acid at the range of 0–10 mg L−1. A standard curve was fitted to the standard samples to estimate silicic acid content.
Fluorescence microscopy observation
For fluorescence labeling of newly synthesized frustules, LysoTracker Yellow HCK-123 (Thermo Fisher Scientific, Waltham, MA, USA) was added to the experimental culture. HCK-123 stock solution (1 mM) was prepared in dimethyl sulfoxide, and added to the culture to final 1 μM. A 1 μL aliquot of cells was taken at specified time points for fluorescence microscopy using an Eclipse Ni-U microscope (Nikon, Tokyo, Japan) with a PhotoFluor LM-75 illumination system (89 North, Williston, VT, USA). Images were captured using a DS-Fi3 camera (Nikon) using NIS-Elements version 4.60, build 1171 (Nikon), as described in Won et al. (2023). Images of live cells, chlorophyll, and newly synthesized frustules were taken for each dataset. Live cells and chlorophyll were quantified to confirm cell viability. A minimum of 100 cells were analyzed in each experiment. To observe chlorophyll, a TxRed filter set was used, and a fluorescein isothiocyanate (FITC) filter was used to observe newly formed frustules stained with HCK-123.
Bioinformatic analyses of SIT proteins
Two SIT homologs, TeSIT1/2 and TeSIT3, were identified in the de novo assembled transcripts of T. eccentrica LIMS-PS-3165 (unpublished data). The coding sequences of the two SITs were deposited in GenBank with ID numbers PP526761 and PP526762, respectively (https://www.ncbi.nlm.nih.gov/gene). The predicted amino acid sequences of SITs listed in Table 1 were aligned and analyzed using the ClustalOmega web service from EMBL-EBI (Madeira et al. 2022). A phylogenetic tree of the SIT proteins was constructed using MEGA11 v11.0.10 (Tamura et al. 2021). The distances were computed using the Poisson correction method (Zuckerkandl and Pauling 1965), and the tree was evaluated through 500 bootstrap replicates (Kim et al. 2022). The phylogenetic tree was rooted with the SIT protein from Stephanoeca diplocostata, serving as a non-diatom SIT outgroup.
The structure of each SIT protein was predicted using Colabfold, which combines MMseqs2 and Alphafold2 to offer faster prediction of protein structures or complexes (Mirdita et al. 2022), and then visualized using UCSF ChimeraX (Pettersen et al. 2021). DeepLoc-2.0 web server (Thumuluri et al. 2022), Predotar v.1, and HECTAR (Gschloessl et al. 2008) were used to predict subcellular localization. Four web servers (Small et al. 2004) were used. Transmembrane regions were predicted using TMHMM 2.0, a web server (Krogh et al. 2001).
Real-time reverse transcriptase-polymerase chain reaction
Total RNA was isolated using a Hybrid-R RNA purification kit (Geneall Biotechnology, Seoul, Korea) and used as a template for cDNA synthesis using a High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems, Waltham, MA, USA), following the manufacturer’s instructions. For the transcription abundance, real-time PCR was performed using a PCR premix containing TB Green (TB Green Premix Ex Taq II; Takara Bio, Kusatsu, Japan). The thermal cycling profile comprised of 30 s pre-denaturing step at 95°C followed by 45 cycles of denaturing for 5 s at 95°C and annealing / extension for 30 s at 60°C. TATA-box binding protein was used as an internal control to normalize the data for relative quantification (Kang et al. 2012). Thermal cycling and quantification were performed using a Thermal Cycler Dice Real-Time System TP800 (Takara Bio), and each experiment was performed in triplicate. The primers used in these experiments are listed in Table 2.
RESULTS AND DISCUSSION
Thalassiosira eccentrica LIMS-PS-3165, henceforth referred to as T. eccentrica, exhibited characteristics typical of centric diatoms, including circular valves with radial symmetry. Colony formation was not observed under normal or stress conditions. The cells generally displayed a golden-brown color (Fig. 1A). The diameter of the valve was approximately 10 μm, while the height of a single cell ranged from approximately 5 to 10 μm, suggesting various stages of cell growth. Scanning electron micrographic images of the valve revealed irregular patterning of the areolae (Fig. 1B). T. eccentrica was reported to be found globally except for Arctic and Antarctic areas, and also frequently found in Korean coastal area (Park et al. 2016). It was reported that this species has hexagonal areolae, distributed regularly over the valve face with one row of marginal spines (Park et al. 2009, 2016). However, the cells observed in this study had areolae in irregular shape and size which were randomly distributed over the valve face, and no spines was observed (Fig. 1B). In diatoms, cell maturity can generally influence the shape and arrangement of areolae, which tend to become more regular as cells mature (Sato 2010). However, based on repeated observations, these characteristics in this study do not appear to be due to cell maturity.
Nutrient deprivation and cell cycle arrest in diatoms observed in newly synthesized silica frustules
After inoculating cultures with HCK-123, we stained newly synthesized frustules and observed them using a fluorescence microscope equipped with a FITC filter set. The frustules were categorized based on three observable states of fluorescence staining, as illustrated in Fig. 2B. Middle staining indicates the initiation of the G2 + M phase, during which valve fragments begin to be synthesized at the center of the mother cell. Tip-stained cells were observed when the newly synthesized frustule fragments fully expanded across the cell, forming the hypotheca of the upper daughter cells and the epitheca of the lower daughter cells. This tip fluorescence indicated that the cells had completed one division cycle and were at the end of G1 phase. Fully stained cells suggested that the cells had undergone another cycle of division, as only one of the two valves was synthesized per division cycle (Fattorini and Maier 2021, Bulankova et al. 2022).
Cell cycle progression, as indicated by HCK-123 staining, varies under different nutrient limitations. Under Si-deprived conditions, frustule staining ceased at the tip stage. However, under phosphorus and nitrogen deprivation, full staining of frustules occurred on days 3 and 4, respectively, implying the resumption of the cell division cycle (Fig. 2, Supplementary Fig. S1).
At the 24-h mark, the most representative staining stage for all three treatments was middle-stained cells. However, at 72 and 96 h, tip-stained frustules were most common. Notably, in phosphorus-depleted P(−) media, a small portion of cells showing fully stained frustules was observed starting on day 3 (Fig. 2, Supplementary Fig. S1). When cells are supplied with Si after 1 d of arrest, around 90% of cells were labeled with fluorescence in one of the theca within 24 h, implying that 90% of cells had gone through one cycle of cell division (Fig. 3A). In Si-deplete medium, staining of one theca was observed in about 10% of cells within 24 h, while approximately 80% remained with middle staining. This suggests that, although the remaining Si in the medium was absorbed into the cells, cell division did not occur during the first 24 h of Si depletion. Nevertheless, as shown in Fig. 2B, the gradual increase in tip staining over 4 d in Si-deplete medium suggests that cell division is proceeding slowly using the absorbed Si.
Silicon (Si) is a critical element in diatoms and constitutes the silica cell wall. Its absence significantly affects cell division. Phosphorus (P) and nitrogen (N) are essential for cell growth. Nitrogen, a key component of proteins and nucleic acids, is vital for maintaining cell structure. Its deficiency not only reduces photosynthesis and chlorophyll synthesis, but also leads to the remodeling of carbon metabolism, affecting overall energy processes (Alipanah et al. 2015). Phosphorus is another major component of nucleic acids and phospholipids. Phosphorus limitation also affects photosynthesis, because the photosynthetic machinery relies on the phospholipid bilayer in the thylakoid membrane. Phosphate deprivation also affects N uptake and utilization, leading to growth retardation (Alipanah et al. 2018).
We found that deprivation of these three elements affected cell division. Especially, N and Si deprivation resulted in complete arrest of cell division (Supplementary Fig. S1), and fluorescent signal did not increase after 24 h (data not shown). However, cell density under P deprivation showed slight increase after the third day of deprivation. This modest increase is likely due to the recycling of these essential elements from reserves and metabolic remodeling in response to nutrient shortages (Kim et al. 2017). Unlike nitrogen and phosphorus, Si, which is available only as a soluble pool, cannot be accessed from storage, making it a more effective inhibitor of cell division (Parslow et al. 1984, Harrison et al. 1990).
Phylogeny and conserved domains of SIT proteins: predicting functional differences
In our transcriptome analysis of T. eccentrica (unpublished laboratory data), we identified two SIT-coding sequences. One sequence was similar to SIT1 and SIT2 found in other diatoms, whereas the other was closer to SIT3; therefore, these two SITs were designated as TeSIT1/2 and TeSIT3, respectively. Sequence similarity was analyzed, and the protein structure was predicted based on the deduced amino acid sequences of the two SITs.
A phylogenetic tree constructed from deduced amino acid sequences from various diatoms including that of T. eccentrica is shown in Fig. 4. Fourteen SIT sequences were chosen from databases based on previous reports (Thamatrakoln and Hildebrand 2008, Durkin et al. 2016), and compared with TeSIT1/2 and TeSIT3.
The SIT1 and SIT2 sequences from T. pseudonana clustered with TeSIT1/2, with 100% bootstrap support, confirming a close relationship. The separation between the SIT1/2 and SIT3 sequences within Thalassiosira was well-supported with a 96% bootstrap value. The split between centric diatom SITs and most pennate diatom SITs was supported by 52% of the bootstrap values, with the SIT1 sequences of Fragilariopsis cylindrus and Phaeodactylum tricornutum diverging earlier than the separation between pennate and centric diatom SITs. Notably, SIT1 and SIT2 grouped separately from SIT3, indicating distinct evolutionary paths.
Previous studies on SIT sequences across various diatom species have shown that SITs from the Thalassiosira genus are clearly grouped within clade E (Hendry et al. 2018). Our findings suggest that within Thalassiosira, there are two distinct SIT lineages with different evolutionary histories. Evolutionary research on SITs indicates that SIT3 likely arose from a genome duplication event of SIT1-2, following a unique evolutionary path, particularly within Thalassiosirales (Bryłka et al. 2023).
In the alignments of the deduced amino acid sequences of TeSIT1/2 with those of SIT1 and SIT2 of the closely related species T. pseudonana, the pairwise identities were 83% and 79%, respectively. A comparison between the SIT3 of both species revealed 74% amino acid identity. In contrast, TeSIT1/2 and TeSIT3 showed only 50% similarity.
Despite relatively low similarity, 10 conserved transmembrane domains were detected in both TeSITs (Fig. 5). The 10 transmembrane domains predicted based on the hydropathy analysis by Thamatrakoln et al. (2006) was identified. However, when we attempted prediction using TMHMM, one of the 10 TM domains TeSIT1/2 had a low score, leading to an uncertain 3D model of the protein (Supplementary Fig. S3). Knight et al. (2023) found eight authentic transmembrane helices using the AlphaFold model, and the other two helices were partially submerged parallel to the membrane plane. Despite the small uncertainty due to differences in the algorithms used in the prediction, the combination of bioinformatics data and low-resolution experimental results collectively indicated the presence of 10 transmembrane domains. There were slight variations between the two TeSITs, which may contribute to functional differences between them.
Four GXQ motifs were conserved in all the aligned sequences, along with the positions of the motifs relative to the transmembrane domains. The GXQ motifs are reported to be located in pairs at the ends of helices 2 and 3 and helices 7 and 8, making them pairs that facilitate translocation if Si crosses the membrane (Thamatrakoln 2006, Knight et al. 2016), and are also found at the same position in T. eccentrica. It is presumed that the C-terminus of SIT1 interacts with other proteins and is less conserved (Hildebrand et al. 1997, 1998, Thamatrakoln et al. 2006, Durkin et al. 2016), which is also less conserved in SIT3 (Fig. 5).
Functional differences presumed by the transcriptional pattern and cellular localization of TeSITs
Although no significant differences in functional motifs were found between the two TeSITs, the transcription patterns in response to silicon availability suggest that they play different roles in silicon-related metabolic processes.
The relative mRNA expression levels of two silicon transporter genes, TeSIT1/2 and TeSIT3, were measured in conjunction with cell growth and Si concentration in the culture media at various time points.
The cell density of the experimental cultures remained relatively stable up to 9 h in both Si(+) and Si(−) conditions (Fig. 6A & B). The cell densities did not increase in the Si(−) condition as the Si concentration in the culture media slightly decreased. Conversely, under Si(+) conditions, the cell density showed a sharp increase between 9 and 12 h, followed by a steady increase until 48 h, albeit at a slower rate. The Si concentration in the Si(+) cultures decreased sharply in the first 3 h after Si was replenished and then continued to decrease at a stable rate until 48 h.
The transcription patterns of the two TeSIT genes, TeSIT1/2 and TeSIT3, differed. The transcription of the TeSIT1/2 decreased during the first 12 h following Si resupply after 1 d of Si starvation (Supplementary Fig. S2A). This is in line with a previous study by Shrestha and Hildebrand (2015), which reported a decrease in the transcription of TpSIT1 and TpSIT2 during the first 9 h following Si resupply. However, the transcription of TeSIT1/2 gradually increased over the next 36 h. This is considered a response to the exhaustion of Si in the medium (Fig. 6A). In contrast, the transcription ratio of TeSIT3 remained relatively constant, and was not affected by the presence of Si (Fig. 6C, Supplementary Fig. S2B). This is in line with a previous study by Shrestha and Hildebrand (Shrestha and Hildebrand 2015), who reported differences between TpSIT1, TpSIT2, and TpSIT3. In their study, the sequences of TpSIT1 and TpSIT2 proteins showed higher similarity (88%) than those of TpSIT3 (46 and 74% with TpSIT1 and TpSIT2, respectively), and the transcription levels of TpSIT1 and TpSIT2 changed according to the cell cycle, in contrast to that of TpSIT3 which showed a relatively low and constant transcription level.
The transcription patterns in response to the absence or presence of Si also differed in previous studies. The transcription of TpSTI1 and TpSIT2 is upregulated under Si depletion and is downregulated upon Si replenishment (Shrestha and Hildebrand 2015, Hildebrand et al. 2018). In contrast, SIT3 expression is unaffected by Si levels (Hildebrand et al. 2018). Since silicic acid can easily diffuse into the cell, it is presumed that some SIT proteins do not actively contribute to the uptake of Si (Thamatrakoln et al. 2006), but rather sense internal Si levels and regulate intracellular Si metabolism. In addition, most diatoms possess multiple SIT genes, the expression levels of which change in response to Si availability; however, the abundance of transcripts and proteins does not always correspond, suggesting multiple levels of regulation (Thamatrakoln and Hildebrand 2008, Shrestha and Hildebrand 2015). In this study, although the differences were slight, the transcription patterns of TeSIT1/2 and TeSIT3 were nearly opposite (Fig. 6C, Supplementary Fig. S2), suggesting a difference in the functionality of the two proteins. This divergence may also reflect slight variations in the predicted protein structure and localization (Supplementary Fig. S3). Notably, TeSIT3 was predicted to localize to the endoplasmic reticulum (ER) (Supplementary Table S1). When predicted using HECTAR with the heterokont range, SIT3 is predicted to localize to the mitochondria and other locations. Although the predictions for subcellular localization vary depending on the methods, both DeepLoc and Predotar indicate a high probability of ER localization for SIT3. Previous studies have reported that ER localization predictions can vary depending on the analytical method used (Anand et al. 2018, Darabi and Seddigh 2018). Kumar et al. (2017) reported that only 50% of the proteins in an ER protein dataset had an ER signal, suggesting that these signals are not essential for their presence in the ER. Based on the predicted localization and transcription patterns, TeSIT3, unlike TeSIT1/2, is presumed to play an important role in concentrating silicon in silicon-deposited vesicles and in frustule formation, rather than in silicon uptake.
Further investigation is necessary to fully elucidate the functions of TeSIT1/2 and TeSIT3. However, based on their distinctive transcription patterns, one may be involved in the internal regulation of Si transport, whereas the other may regulate the uptake of Si. This is critical because the timing of Si uptake and incorporation must be closely coordinated, particularly because silica deposition in the cell wall is tightly linked to Si transport and influenced by intracellular components (Thamatrakoln 2006).
CONCLUSION
Thalassiosira eccentrica is prevalent worldwide and is especially common in the coastal areas of South Korea. However, it has been relatively understudied in biomolecular studies. Our experiments, which involved labeling newly synthesized frustules with a fluorescent probe under nutrient-deprived conditions, revealed that Si was the most potent element for arresting the cell cycle. Furthermore, the two TeSITs appeared to have different functions in Si uptake and utilization, as suggested by their sequence similarities and markedly different transcription levels. Employing genetic manipulation techniques to target these SITs will be the key to defining the specific functions of each TeSIT protein.