Background

Due to the massive use of fossil fuels and deforestation, the carbon dioxide in the atmosphere is continuously increasing, causing global warming. Hence, the demand for sustainable alternative energy is promoting research on microalgae because of their potential for CO2 reduction and biofuel production. Microalgae are known to have much higher photosynthetic efficiency as compared to terrestrial plants [1, 2]. Indeed, around 27,200 tons of CO2 are commercially converted into microalgal biomass annually [3,4,5]. As such, microalgae are considered to be one of the best candidates for CO2 sequestration [6]. However, microalgae's low tolerance to high CO2 concentrations poses a significant challenge to its commercialization [7]. This is especially important when considering the utilization of industrial exhaust gas streams with high CO2 content (> 10%), which is the most economical and environmentally attractive option [8,9,10,11].

Chlorella sp. is one of the most studied microalgae for lipid and CO2 sequestration [12, 13]. A whole genome sequence has been unveiled in several Chlorella sp., and diverse genetic manipulation tools, including CRISPR/Cas9 and transformation methods, have been developed [14,15,16]. Recently, multiple transcriptomic analyses have been carried out under diverse conditions to understand the mechanisms of CO2 fixation, lipid accumulation, and stress resistance in Chlorella sp. (Table 1) [13, 17,18,19,20,21,22,23,24,25,26,27,28]. In particular, research has been conducted on the photosynthesis of these strains under high CO2 concentrations, as Chlorella is known to have strains that have a tolerance to high CO2 compared to other species [13, 22, 28,29,30].

Table 1 Transcriptomic studies conducted in Chlorella sp.
Full size table

Photosynthesis with CO2 fixation is one of the most crucial bioprocesses in microalgae. Photosynthetic efficiency is highly affected by the concentration of the dissolved CO2 in a media. Huang et al. estimated the Km value of RuBisCo (Ribulose-1,5-bisphosphate carboxylase) in Chlorella PY-ZU1 to be in the range of 80–192 μM, which exceeds the dissolved CO2 concentration (~ 15 μM) in water with air at an equilibrium state, indicating the necessity of CO2 concentrating mechanism (CCM) in microalgae [13, 31]. Carbonic anhydrase (CA) is the key enzyme for CCM, as it converts CO2 into more soluble HCO3−, allowing the dissipation of HCO3− into the cell. Then, CA helps CO2 fixation by reconverting the internalized HCO3− into CO2 at the thylakoid membrane under low CO2 concentrations. Hence, CA is highly expressed under CO2-limited conditions along with other CCM-related enzymes. Additionally, Jianhua et al. proposed several transporters related to CCM, which showed significant expression changes in the short term (0–24 h) by CO2 deprivation in Chlorella pyrenoidosa [21]. Besides, the previous transcriptomic profile of C. pyrenoidosa has also revealed 8 unigenes including ABC transporter, chlorophyll a/b binding protein, and photosystem I subunit V, which may play essential roles in CCM as well [20]. Although these findings have provided some information about CCM in Chlorella sp., there are still many unknown points underlying the CCM mechanisms.

The lipid accumulation mechanism in microalgae has been investigated over the years as lipids could be converted into biofuels or purified for human consumption (e.g. cooking oils, polyunsaturated fatty acids). In general, lipid production occurs primarily in the endoplasmic reticulum (ER) or chloroplast through the Kennedy pathway, storing energy in the form of triacylglycerol (TAG). Hence, many approaches for overexpression of Kennedy pathway genes including glycerol-3-phosphate dehydrogenase (GPDH), glycerol-3-phosphate acyltransferase (GPAT), lysophosphatidic acid acyltransferase (LPAT), phosphatidic acid phosphohydrolase (PAP), and diacylglycerol acyltransferase (DGAT) has been reported, especially in oleaginous algae species such as Phaeodactylum and Nannochloropsis [32,33,34]. The lipid accumulation pathway in Chlorella sp. has also been extensively studied at the transcriptome level under various stress conditions [18, 35,36,37].

Most microalgal CO2 conversion studies have been conducted under moderately high CO2 conditions of 5% or below, and microalgae that grow well under 5–20% CO2 are considered CO2-tolerant strains [40]. Chlorella sp. ABC-001 showed enhanced growth under 10–15% CO2 conditions as compared to ambient air conditions [38, 39]. Thus, it is necessary to understand insights into the response of CO2-tolerant strain under high CO2 conditions for the use of industrial flue gases. Additionally, most studies focusing on lipid synthesis and transcriptomic analysis manipulate media composition or impose artificial stress to observe short-term changes. However, in large-scale cultivation, applying artificial stress is complex. This highlights the significance of studying natural nutrient depletion, an aspect that has received comparatively less attention in the field. Accordingly, it is meaningful to study the relationships between phenotypes and transcriptomic analysis according to CO2 and nutrient concentrations.

In this study, we conducted a time-course transcriptomic analysis of Chlorella sp. ABC-001, the high-CO2 tolerant strain, under ambient air and 10% CO2 conditions. We analyzed transcriptomic changes related to CO2 concentrations, cell growth phases, lipid accumulation, and nutrient depletion with a macroscopic perspective. Furthermore, we identified critical genes related to each metabolism and discussed them in detail. These findings are expected to enhance our understanding of the photosynthesis and lipid accumulation metabolism in the green alga Chlorella sp., especially under high CO2 conditions. Our study will also provide valuable insights into the metabolic engineering and synthetic biology of microalgae.

Results

Cultivation of Chlorella sp. ABC-001 under ambient air and 10% CO2 conditions

In this study, triplicate biological cultivations of Chlorella sp. ABC-001 were carried out in different CO2 concentrations using photobioreactors for transcriptomic analysis. As described in Fig. 1, an apparent difference in growth and lipid accumulation was observed under the 10% CO2 condition compared to the ambient air condition. The maximum growth rate was faster with CO2 (μmaxCO2: 0.84 d−1, μmaxAir: 0.72 d−1), and the cells grew up to 2.5-fold higher cell density, accumulating almost twice more lipids (Fig. 1a, b). The nutrients were depleted faster in CO2 conditions, which was consistent with faster cell growth (Fig. 1c). Nitrogen and sulfur were almost depleted on day 3 and day 4 at 10% CO2, respectively, while it took 6 days to consume all these nutrients under ambient air conditions. The pH was maintained between 6 and 8 during the whole cultivation in both conditions, indicating that the effects of pH change may be negligible (Fig. 1d). Transcriptomic analysis through RNA sequencing was conducted to unravel the governing mechanisms and discover critical genes causing the phenotypic changes under different CO2 concentrations.

Fig. 1
figure 1

Cultivation of Chlorella sp. ABC-001 under different CO2 conditions. a Growth curve of cells cultivated with ambient air and 10% CO2. The line graph represents cellular density, and the bar graph represents dried cell weight. b The lipid contents of cells as measured after transesterification into FAME. c Changes in nitrate concentration (NO3−) and sulfate concentration (SO42−). d pH changes. Error bars stand for the standard error calculated from three independent experimental data sets. Asterisks indicate the significant difference between cells cultivated under ambient air and 10% CO2 determined by Student’s t-test (*P < 0.05, **P < 0.01, ***P < 0.001)

Full size image

RNA-sequencing and unigene assembly

The RNA samples were collected on days 1, 3, and 7 to represent the early-exponential phase, mid-exponential phase, and stationary phase under both ambient air and high CO2 conditions. For simplification, cells collected from each condition will be represented by ‘A1, A3, A7’ and ‘C1, C3, C7’ (A: ambient air, C: 10% CO2, 1: day1, 3: day3, 7: day7). The total number of 10–16 million (M) raw reads from the ambient air and high CO2 samples were sequenced, and 8–14 M clean reads were generated after trimming and sorting, as described in the material and method section. The clean read ratio of the raw sequence was about 87.75%, and the pairs plot analysis showed that the reproducibility of biological triplicate samples was higher than 0.9 in most cases, ensuring high reproducibility and reliability of the data (Additional file 1: Fig. S1). Clean reads were pooled for assembly, and a total of 10,246 annotated unigenes were generated, as summarized in Additional file 1: Table S1. The top 3 species represented in the National Center for Biotechnology Information (NCBI) non-redundant (NR) database were Auxenochlorella protothecoides, Chlorella variabilis, and Coccomyxa subellipsoidea, accounting for 52.7% of the transcripts (Additional file 1: Fig. S2). The whole transcriptomic data obtained in this study were deposited in the NCBI database under BioProject accession number PRJNA757763.

Analysis of differentially expressed genes across different growth phases and CO2 concentrations

Pairwise transcriptome comparison was performed to identify differentially expressed genes (DEGs). We compared expression levels under high CO2 conditions based on those under ambient air conditions on days 1, 3, and 7 (C1A: C1 vs A1, C3A: C3 vs A3, C7A: C7 vs A7). Basically, downregulated DEGs were more dominant than upregulated DEGs in all three comparisons under high CO2 conditions, and the number of downregulated DEGs also increased as the cultivation progressed (Fig. 2a).

Fig. 2
figure 2

Transcript profiling changes by CO2 concentration. a The numbers of upregulated and downregulated DEGs in each condition. b Venn diagram representing the number of non-redundant DEGs from different CO2 concentrations. The left and right Venn diagram each represents upregulated and downregulated DEGs. C1A, C3A, and C3A represent the DEGs comparison of C1 vs A1, C3 vs A3 and C7 vs A7, respectively

Full size image

Among all the identified DEGs, only one gene, carbamoyl phosphate synthetase B (Csorokiniana1SL000732t0001), showed a consistent upregulation in 10% CO2 during the whole cultivation period (Fig. 2b). Carbamoyl phosphate synthetase participates in the nitrogen metabolism pathway, catalyzing the first committed step in pyrimidine and arginine production [41]. Thus, we assume that the high expression level of carbamoyl phosphate synthetase was due to the faster nitrogen uptake in high CO2 conditions. The most notable upregulated gene was the cobalamin-independent methionine synthase (METE, Csorokiniana1SL000035t0006), which was observed in the C1A comparison. The expression level of METE was 20–32 fold higher in the C1 condition as compared to any other conditions including A1, A3, A7, C3, and C7 (Additional file 2). As METE is also involved in the synthesis of DNA, RNA, and protein, it seems to help with rapid growth in the early stage under high CO2 conditions.

There were 18 constitutively downregulated genes including carbonic anhydrase, ferredoxin hydrogenase, and starch synthase (Fig. 2b). The carbonic anhydrase and ferredoxin hydrogenase play essential roles in CCM [13]. Under the conditions of a sufficient external supply of carbon dioxide, the two genes might not need to be upregulated. Additionally, starch synthase was notably downregulated under high CO2 conditions as carbon storage is not preferred under favorable conditions. Indeed, the carbohydrate content of cells cultivated under ambient conditions was higher (40.7%) compared to those cultivated under 10% CO2 conditions (32.2%) on day 7 (Additional file 1: Fig. S3).

Gene ontology analysis across different growth phases and CO2 concentrations

To further understand the function of the DEGs in each phase, Gene Ontology (GO) enrichment analysis was performed. Only the top 7 terms in each category of biological process (BP), cellular component (CC), and molecular function (MF) are listed (Fig. 3).

Fig. 3
figure 3

GO analysis of upregulated and downregulated non-redundant DEGs during cultivation under different CO2 concentrations. The top 7 terms were listed in each analysis based on the cellular growth phase of day 1 (a, d), day 3 (b, e), and day 7 (c, f). The size of the circle represents the gene numbers, and the reference is on the left bottom side of the figure

Full size image

The early growth phase on day 1, the photosynthesis and energy-related genes in BP were mainly upregulated under high CO2 conditions (Fig. 3a). In CC, the genes of chloroplast-related organelles, such as the plastid, chloroplast, and thylakoid, were upregulated. In MF, the chlorophyll binding process is classified as the highly upregulated DEGs. On the other hand, there were no downregulated DEGs under high CO2 conditions on day 1, except for the glutamine metabolic process (Fig. 3d).

At the mid-exponential phase on day 3, more diverse changes were detected from both upregulated and downregulated DEGs (Fig. 3b and e). In MF, the cells primarily concentrated on protein-level regulation (including catalytic activity, oxidoreductase, chaperone, deacetylase, and phosphatase) under elevated CO2 conditions. Conversely, transcription-level regulation (such as RNA binding, small molecule binding, and mRNA binding) was downregulated. In addition, both upregulated and downregulated DEGs on day 3 were related to chloroplast and plastid stroma in the CC category. These findings suggest that the cells transitioned to the exponential growth phase on day 3. Indeed, Chlorella sp. ABC-001 exhibited accelerated growth under high CO2 conditions, and this might lead to a halt in the development of photosynthetic function while concentrating on a range of catalytic activities.

On day 7, a notable decrease in transmembrane transporter activity was observed in downregulated GO terms for MF under high CO2 conditions (Fig. 3f). This included diverse transporters such as metal transporters, ATP-binding cassette transporters, sugar transporters, and amino acid transporters (Additional file 2). As photosynthetic activity requires many membrane transporters [42], we expect the decreased photosynthetic activity may heavily cause the decreased membrane transporter activity. The ratio of OD680/OD750, which serves as an approximate indicator of photosynthetic efficiency and the physiological state of cells with their chlorophyll content [43], displayed a decreasing pattern under high CO2 conditions after day 3 (Additional file 1: Fig. S4). In contrast, cells grown under ambient air maintained a consistent OD680/OD750 ratio, further supporting our explanations.

Transcriptomic profile analysis of CCM and the Calvin cycle

To understand the mechanism behind CO2 fixation, growth, and lipid production under high CO2 conditions, we analyzed the associated transcriptomic profiles (Fig. 4). We first looked into the expression patterns of carbonic anhydrase (CA) genes and the Calvin cycle-related genes (Fig. 4). In Chlorella sp. ABC-001, 6 CAs were identified, and the CAs generally showed lower expression levels under CO2-supplemented conditions. Particularly, the CA3 and CA4 showed the most significant changes, 25–40-fold lower expression levels under high CO2 conditions than under ambient air conditions. This means that the CAs, especially CA3 and CA4, are not necessary under high CO2 conditions. Conversely, they are expected to play a critical role in the carbon concentrating mechanism under low CO2 conditions in Chlorella sp. ABC-001. The bicarbonate transporter 1, which is an essential transporter for carbon fixation, showed higher expression in high CO2 conditions on days 1 and 3.

Fig. 4
figure 4

Relative expression levels (log2 fold change) of key enzymes participating in carbon fixation, Calvin cycle, nitrogen uptake, and lipid biosynthesis. The Gene IDs of each enzyme are listed in Additional file 1: Table S3. (GAPDH Glyceraldehyde-3-phosphate dehydrogenase, GPDH glycerol-3-phosphate dehydrogenase, GPAT glycerol-3-phosphate acyltransferase, LPAT lysophosphatidic acid acyltransferase, PAP phosphatidic acid phosphohydrolase, DGAT diacylglycerol acyltransferase, PDAT phospholipid-diacylglycerol acyltransferase, bHLH basic helix-loop-helix, MYB myeloblastosis, bZIP basic leucine zipper, ACC1 Acetyl-CoA carboxylase 1, FAD Fatty acid desaturase)

Full size image

The expression patterns of two potential RuBisCo activases were also analyzed. The RuBisCo activase 1 and 2 showed higher expression levels on day 3 under high CO2 conditions. In particular, the RuBisCo activase 2 was significantly upregulated on day 1 at the early growth phase. On day 1, most of the genes in the Calvin cycle showed higher expression levels in the CO2-supplemented culture, whereas it became the opposite on day 7.

Transcriptomic profile analysis of lipid synthesis

In microalgae, nitrogen limitation is the most effective stress to trigger lipid accumulation in the cell [44]. According to Fig. 1c, nitrogen starvation was initiated after day 3 and day 6, under 10% CO2 and ambient air conditions, respectively. From the RNA-seq comparison of CO2 vs ambient air, a sudden increase of expression level in nitrate reductase (NR), nitrite reductase (NiR), and cofactor of nitrate reductase (CoNiR) were found on day 3, when nitrogen was almost depleted in 10% CO2 conditions (Fig. 4). The expression levels of NR, NiR, and CoNiR also consistently increased from day 1 to day 7 in both cultivation conditions (Additional file 1: Fig. S5). High expression levels of NR after nitrogen depletion at the mid-exponential phase have been reported by other researchers as well [13].

Microalgae generally accumulate lipids in the form of tri-acyl-glycerol (TAG), which is synthesized via the Kennedy pathway. Under high CO2 conditions, FAME contents were almost 40% on day 7, which was 3.1-fold higher than those on day 3 (Fig. 1b). The gene expression patterns of the Kennedy pathway were also related to the phenotypes. The genes encoding for glycerol-3-phosphate dehydrogenase (GPDH), lysophosphatidic acid acyltransferase (LPAT), diacylglycerol acyltransferase (DGAT), and phospholipid-diacylglycerol acyltransferase (PDAT) showed an increased expression level under high CO2 conditions on day 3 and 7. However, PAP seemed not to be affected much by CO2 concentrations. A distinct decrease in the expression level of GPDH, LPAT, PAP, DGAT, and PDAT was observed under 10% CO2 conditions on day 1. As C1 is the condition that exhibited a very active metabolism in photosynthesis and energy generation for growth compared to A1 (Figs. 1a and 3a), lipid synthesis seems to be strictly repressed at this point.

In particular, a putative PDAT (Csorokiniana1SL007740t0001) and GPDH genes showed the most noticeable changes, which increases the likelihood of its correlation with lipid accumulation. PDAT, similar to DGAT, delivers an acyl group to diacylglycerol (DAG) in the last step of TAG biosynthesis. However, unlike DGAT, which takes acyl-group from acyl-CoA, PDAT functions on diverse kinds of substrates such as phospholipids (phosphatidylcholine, phosphatidylenolamine), membrane lipids (monogalactosyl diacylglycerol, digalactosyl diacylglycerol), and DAG [45]. Accordingly, GPDH and LPAT are also expected to be important for lipid accumulation according to their increased expression pattern.

We have broadened our analysis to explore additional factors and pathways involved in lipid synthesis in microalgae, not limited to the Kennedy pathway (Fig. 4). Among the transcription factors (TFs) known to promote lipid synthesis in microalgae, basic helix-loop-helix (bHLH) [46] and myeloblastosis (MYB) [47] TFs displayed increased expression profiles under high CO2 conditions, which correlated with lipid accumulation. In contrast, basic leucine zipper (bZIP) TFs [48] did not exhibit a clear relationship with lipids. Acetyl-CoA carboxylase and β-oxidizing enzymes, responsible for the synthesis and cleavage of acyl-CoA, respectively, showed elevated expression levels under 10% CO2 conditions during the stationary phase. Additionally, we identified multiple potential fatty acid desaturase (FAD) genes; all 10 FADs consistently demonstrated lower expression levels under high CO2 conditions on day 7.

Validation of transcriptomic analysis via qRT-PCR

To validate the quality of the RNA-sequencing data obtained in this study, qRT-PCR was carried out by targeting several genes exhibiting different expression patterns (Fig. 5). Indeed, the qRT-PCR results were almost consistent with the RNA-sequencing data. CA3, CA4, and CA5 showed low expression levels under high CO2 conditions and high expression levels under ambient air conditions by qRT-PCR. Although the expression level of NR1 was lower in qRT-PCR compared to RNA-seq, the overall expression pattern was consistent in both analyses. The CoNiR1 also showed highly conserved expression patterns between qRT-PCR and RNA-sequencing results. Based on these results, we confirmed the credibility of the RNA-sequencing analysis carried out in this study.

Fig. 5
figure 5

Validation of RNA-sequencing data by qRT-PCR. Each data was normalized by the expression level of A1. Error bars stand for the standard error calculated from three independent experimental data sets. Specific gene IDs are listed in Additional file 1: Table S3

Full size image

Discussion

As the potential of Chlorella for CO2 fixation and production of lipids/high value-added substances has been highlighted, a number of transcriptomic approaches have been conducted to elucidate the underlying mechanisms in recent days [13, 17,18,19,20,21, 49, 50]. These researches covered a broad range of areas such as hydrogen photoproduction, TAG synthesis, stress tolerance, and starch synthesis. Due to the dynamic activity of cellular metabolism and the formidable amount of transcriptomic data, various interpretations are possible depending on sampling points, comparison groups, and analytic criteria.

In this study, we performed a transcriptomic analysis of Chlorella sp. ABC-001 under different CO2 concentrations (10% CO2 and ambient air) at three different time points (early exponential, mid-late exponential, and stationary growth phase). From transcriptomic analysis, a total of 10,246 putative genes were annotated, which is an improved result compared to previous studies (9590) on transcriptome profiling of C. pyrenoidosa in response to different CO2 concentrations [20]. Based on the transcriptome data, we illustrated the central carbon metabolism of Chlorella sp. ABC-001 with their expression levels at high CO2 conditions as compared to ambient air conditions according to the time point in Fig. 6.

Fig. 6
figure 6

Overall metabolic changes in Chlorella sp. ABC-001 at different CO2 environments. The color of each enzyme represents either upregulation (red) or downregulation (blue) of genes on day 1. Small boxes stand for the log2 fold changes of representative enzymes of each metabolic reaction. The asterisk indicates adj.p < 0.05. Specific gene IDs are listed in Additional file 1: Table S3. (FeH ferredoxin hydrogenase, SBPase sedoheptulose-1,7-bisphosphate, FBPase fructose 1,6-bisphosphatase, GAPDH glyceraldehyde-3-phosphate dehydrogenase, CA carbonic anhydrase 4, NR nitrate reductase1, NiR nitrite reductase1, METE cobalamin independent methionine synthase, Met methionine, Lipd syn represented by PDAT1, Starch syn represented by SS)

Full size image

During the early growth stage under high CO2 and nutrient conditions, we observed that a high CO2 level boosted the construction of photosynthetic organelles at the early stage and upregulated the genes related to the Calvin cycle and bicarbonate transporter to incorporate the inorganic carbon (Figs. 3, 4, and 6). At the same time, nitrate/nitrite reductases were also upregulated as nitrogen is an essential component for chlorophylls under high CO2 conditions. At the early growth stage (day 1), cells prioritize proliferation while inactivating energy-storage processes, including lipid/starch synthesis and the carbon concentrating mechanism. The intensive expression level of a cobalamin-independent methionine synthase (METE) at the early stage under high CO2 conditions also supports this idea. METE is the only enzyme that can synthesize methionine in an environment without additional vitamin B12 (cobalamin) [51]. The synthesized methionine is used for protein synthesis and also plays an important role in proliferation by methylation of DNA and RNAs [52, 53]. Fast nitrogen uptake may also be related to methionine synthesis because nitrogen is known to enhance the METE activity by 30–40 folds (Fig. 1c) [54].

From the middle of the exponential phase, cells started downregulating photosynthetic activity under high CO2 conditions. Particularly, the cells deactivated CCM and decreased the expression level of carbonic anhydrases because they do not need to use tremendous energy for CO2 transport under high CO2 conditions. We also found the low expression level of ferredoxin hydrogenase 4 (FeH) located in the chloroplast under high CO2 conditions (Additional file 2). The FeH might catalyze the production of hydrogen ions which is necessary for the reduction of bicarbonate and CO2 supplement. We also found the upregulation of NR, NiR, and CoNiR according to nitrogen limitation under 10% CO2 conditions, indicating that the cells subsequently would focus on lipid synthesis.

Chlorella sp. ABC-001 changed gene expression patterns for energy storage in lipids along with nutrient depletion and CO2 concentrations at the mid-exponential and stationary phases. Indeed, the variations in lipid composition observed in our study cannot be solely attributed to the direct influence of CO2 concentration. Instead, these results likely represent a complex array of interacting factors. The early onset of nitrogen starvation was primarily instigated by the high CO2 concentration, contributing to the change in the expression of lipid-related genes. Furthermore, shading effects resulting from high cell concentrations may also play a part in interacting with photosynthetic activity. According to the gene expression patterns under high CO2 conditions, GPDH and PDAT seemed to play important roles in lipid synthesis. GPDH1, GPDH2, and PDAT were downregulated at the early growth phase and highly upregulated at the stationary phase, where lipid was significantly accumulated (Fig. 4). Indeed, Yao et al. reported that the overexpression of endogenous GPDH in Phaeodactylum tricornutum resulted in a 60% increase in neutral lipids [55]. PDAT usually exists as a single-copy gene, even in oleaginous strains [56] and it has been suggested that PDAT contributes to TAG synthesis in Chlamydomonas reinhardtii, Myrmecia incisa and Nannochloropsis [45, 57, 58]. Recently, engineering of PDAT was actually proven to be effective for increasing the TAG contents in Nannochloropsis [58, 59]. The fatty acid composition of lipids also supported the transcriptomic and phenotypic data (Additional file 1: Fig. S6). On day 7, α-linolenic acid (C18:3) decreased, while oleic acid (C18:1) increased under 10% CO2 conditions. The change in fatty acid composition might be primarily attributed to the consistently lower expression levels of multiple FADs (Fig. 4). Additionally, the degradation of chloroplasts may contribute to the fatty acid composition, as polyunsaturated fatty acids are the primary components of membrane lipids [60, 61]. Consequently, fatty acids and acyl-groups from degraded membrane lipids, which correspond well with increased β-oxidase expression, can be used for de novo TAG synthesis by PDAT. Thus, the putative PDAT gene (Csorokiniana1SL007740t0001) showing a high expression level at the lipid accumulation stage in Chlorella sp. ABC-001 can be a promising genetic target for lipid production in microalgae.

Generally, DGAT genes have been studied well in microalgae because DGAT is in charge of the last step for lipid synthesis, transferring the acyl group to DAG for TAG synthesis. DGAT genes often exist with multiple copies in microalgae, and only some of them were proven to be effective in enhancing lipid accumulation [62, 63]. We also identified 5 DGAT genes in Chlorella sp. ABC-001. However, their expression patterns were not consistent with the lipid accumulation phenotype. Although the expression patterns of DGAT1, DGAT2, and DGAT5 are similar to the lipid accumulation pattern, the expression levels were not as high as that of GDPH and PDAT. This pattern was similar to GPAT and LPAT. Thus, The DGAT, GPAT, and LPAT from Chlorella sp. ABC-001 might not be suitable target genes for lipid synthesis.

Taken together, our analysis successfully elucidated the relationship between the phenotypic and transcriptomic changes of Chlorella sp. ABC-001 in response to CO2 conditions across various growth phases. Moreover, we identified key genes involved in carbon concentration mechanism (CCM), cell development, and lipid accumulation under high CO2 conditions. This study provides high-quality transcriptomic data that can be applied in microalgal synthetic biology and metabolic engineering for industrial purposes, particularly in utilizing industrial exhaust gas containing high CO2 concentrations. Our time-course transcriptomic analysis (days 1, 3, and 7) addresses a gap in the current research landscape, shedding light on the molecular mechanisms underlying long-term adaptations. While many studies focus on shorter time frames (24 or 48 h), we deliver a more comprehensive understanding of metabolic changes and adaptations over extended periods, better representing industrial microalgal cultivation conditions.

Conclusions

To address climate change issues, it is crucial to research the production of high-value materials utilizing industrial flue gases. However, conventional microalgae typically do not have high CO2 tolerance. In this study, we analyzed the transcriptome of Chlorella ABC-001, a high CO2 tolerant strain, over an extended period (7 days) during which considerable lipid accumulation occurred under high CO2 conditions. We confirmed that Chlorella sp. ABC-001 focuses on fast growth at the early stage under high CO2 conditions by upregulating the Calvin cycle, METE, and bicarbonate transporters. However, CA expression was decreased because CCM did not need to be significantly activated in the high CO2 supply. In addition, our transcriptomic analysis revealed that PDAT and GPDH, the first and final genes for TAG synthesis, are expected to play a crucial role in lipid synthesis under high CO2 conditions. This in-depth knowledge can contribute to developing more efficient microalgal cultivation strategies and innovative genetic engineering approaches for enhancing growth and lipid production capabilities.

Materials and methods

Strains and cultivation

Chlorella sp. ABC-001, a marine microalgae isolated from a coastal region in Jeju island (South Korea) was used in this study [39]. This species was chosen for its high CO2 tolerance, and the transcriptomic differences were analyzed under both aerobic and 10% CO2-supplied conditions [38]. The cells were maintained and cultivated in N8 media (1 g L−1 KNO3, 0.74 g L−1 Na2HPO4·2H2O, 13 mg L−1 CaCl2·2H2O, 10 mg L−1 FeEDTA, 50 mg L−1 MgSO4·7H2O, 12.98 mg L−1 MnCl2·4H2O, 3.2 mg L−1 ZnSO4·7H2O, 1.83 mg L−1 CuSO4·5H2O, 3.58 mg L−1 Al2(SO4)3·18H2O) through the whole experiments. The cultivation was performed in 500 mL column PBRs (photobioreactors) with working volumes of 400 mL under 200 μmol photons m−2 s−1 of light at 30 ℃. The initial concentrations in each experiment were fixed at 5 \(\times\) 106 cells mL−1. For aerobic conditions, ambient air was supplied at 0.6 vvm (volume gas per volume medium per minute), whereas 0.6 vvm of 10% CO2 was supplied for CO2-sufficient conditions.

Cultivational analysis

The growth of cells was measured daily in terms of cell density and dried cell weight (DCW) using a Cellometer (Auto X4, Nexcelom, USA) and filter paper (Whatman, USA) as described in previous publications [64]. The optical densities at wavelengths of 680 nm and 750 nm were measured using a UV–VIS spectrophotometer (UV-1800, Shimadzu, Japan). Accumulated lipids were determined by gas-chromatography (HP 6890; Agilent Technologies, CA, USA) after extraction and transesterification through the Bligh-Dyer method similar to our previous research [65]. The remaining nutrients in the media, particularly nitrate and sulfate, were measured with an elemental analyzer (FLASH 2000 series; Thermo Scientific, MA, USA) at the KAIST Analysis Center for Research Advancement [66]. The total carbohydrate content was quantified using the phenol–sulfuric acid method described in a previous study [38]. In brief, approximately 3–5 mg of dried biomass powder was weighed, thoroughly mixed in 10 mL of distilled water, and then 1 mL of the mixture was transferred to clean glass tubes. Next, 1 mL of 5% wt. phenol solution and 5 mL of concentrated sulfuric acid were added, and the tubes were left in the dark for 30 min to facilitate the reaction. The tubes were inverted several times to mix the solution well, and the absorbance at 470 nm was measured using a UV/Vis spectrophotometer (Shimadzu, Japan).

Transcriptomic analysis

RNA extraction, RNA sequencing, and annotations

The cultivated cells were harvested on days 1, 3, and 7 in each condition and were homogenized with a bead beater (Percellys 24, Bertin Technologies, Paris, France). Using NucleoZOL (Macherey–Nagel, Düren, Germany) according to the provided protocol, RNA was extracted from the homogenized cells. gDNA was then removed from the collected RNAs using DNA-freeTM DNase kits (Ambion, TX, USA). Finally, cDNA was synthesized using an oligo (dT) 20 primer (Invitrogen, CA, USA) and Superscript III™ Reverse Transcriptase (Invitrogen, CA, USA).

RNA-seq was conducted using an Illumina HiSeq2500, and 18 FASTQ files of sequences were yielded (A1, A3, A7, C1, C3, C7, three biological replicates per condition). Then, the transcriptome short reads were pre-treated using DynamicTrim and LengthSort provided by the Solexa QA package to sort out low-quality reads. The resulting clean-reads were assembled by Velvet (ver. 1.2.08) and Oases (ver.0.2.08) according to the provided protocols [67, 68]. The quality of RNA was assured by statistics of dynamic trim (Additional file 3), length sort (Additional file 3), final assembled transcripts (Additional file 3), and RNA integrity electropherograms (Additional file 1: Fig. S7). Annotation was conducted using BLASTX against amino acids from a diverse database including Phytozome v9, NCBI NR, Uniprotkb, KOG, and KEGG (e ≤ 1e−10). InterProscan was carried out using tools provided by EMBL as the default option.

Analysis of DEGs

For the selected clean reads, the read counts were calculated through mapping using Bowtie2 (v2.1.0) software (mismatch < 2 bp, penalty) [69]. Samples with high deviations were normalized by DEseq using R (Additional file 1: Fig. S8) [70]. Differentially expressed genes (DEGs) of Chlorella sp. ABC-001 during cultivation were selected according to the twofold change method and binomial method, where the criteria were set as ‘adjusted P-value (FDR) < 0.01’. When the log2 value of the fold change was higher than 1, it was annotated as an ‘up-regulation’, and when the value was lower than -1, it was annotated as a ‘down-regulation’. For GO (gene ontology) analysis, the selected DEGs were aligned to the sequences from the GO database and processed through Blast2GO. The significance level was set at 0.05 and categorized into three functional categories of BP (biological process), CC (Cellular component), and MF (molecular function). DEGs were annotated by BLASTX (e ≤ 1e−10, Best hits) against amino acid sequences from KEGG DB. The selected DEGs are described by the MA plot in Additional file 1: Fig. S9.

Quantitative real-time PCR (qRT-PCR) validation

The cDNA prepared for RNA sequencing was also used as a qRT-PCR template for validation. The housekeeping gene for TATA-box binding protein (TATA) was used as a control. All the primers used for qRT-PCR are listed in Additional file 1: Table S2. qRT-PCR was performed with 20 μL reaction volumes containing 2 μL of cDNAs (20 ng), 0.5 μL of each primer (10 μM), and 10 μL of Universal SYBR supermix (Bio-Rad, USA). The mixture was loaded on 96-well Hard Shell PCR plates (Bio-Rad, USA) and analyzed using a CFX96 Real-Time system (Bio-Rad, USA).

Statistic analysis

Three biological replicates were performed for all experiments, with results presented as mean ± standard error (SE). Statistical analysis was performed using an in-house script for DEG selection and metaboanalyst 5.0 (t-test, FDR) for other experimental data [71]. Detailed explanations for each case are described in the “Analysis of DEGs” section and legends of each figure.