Increased biomass and lipid production by continuous cultivation of Nannochloropsis salina transformant overexpressing a bHLH transcription factor

Abstract Microalgae are promising feedstocks for sustainable and eco‐friendly production of biomaterials, which can be improved by genetic engineering. It is also necessary to optimize the processes to produce biomaterials from engineered microalgae. We previously reported that genetic improvements of an industrial microalga Nannochloropsis salina by overexpressing a basic helix‐loop‐helix transcription factor (NsbHLH2). These transformants showed an improved growth and lipid production particularly during the early phase of culture under batch culture. However, they had faster uptake of nutrients, resulting in earlier starvation and reduced growth during the later stages. We attempted to optimize the growth and lipid production by growing one of the transformants in continuous culture with variable dilution rate and feed nitrogen concentration. Relative to wild‐type, NsbHLH2 transformant consumed more nitrate at a high dilution rate (0.5 day −1), and had greater biomass production. Subsequently, nitrogen limitation at continuous cultivation led to an increased fatty acid methyl ester production by 83.6 mg l −1 day −1. To elucidate genetic mechanisms, we identified the genes containing E‐boxes, known as binding sites for bHLH transcription factors. Among these, we selected 18 genes involved in the growth and lipid metabolism, and revealed their positive contribution to the phenotypes via quantitative real‐time polymerase chain reaction. These results provide proof‐of‐concept that NsbHLH2 can be used to produce biomass and lipids.

Despite their advantages, biofuel production from microalgae has practical limitation about economic feasibility. Genetic engineering could resolve this problem by developing microalgae that have increased lipid and biomass productivity (Radakovits, Jinkerson, Darzins, & Posewitz, 2010). However, it is difficult to engineer microalgae with desirable phenotypes by the manipulation of individual enzymes. Thus, transcription factor (TF) engineering which regulates the expression of various genes simultaneously emerging in microalgae fields (Bajhaiya, Ziehe Moreira, & Pittman, 2017;Courchesne, Parisien, Wang, & Lan, 2009). It has been reported that nitrogen-responsive regulator and phosphorous starvation response 1 (PSR1) TFs play important roles in lipid accumulation (Bajhaiya, Dean, Zeef, Webster, & Pittman, 2016;Boyle et al., 2012). Other studies reported an increased lipid content in Chlamydomonas and Chlorella cells that overexpressed DNA binding with one finger (Dof)-type TFs (Ibanez-Salazar et al., 2014;Zhang et al., 2014).
In addition, we reported that the biomass and lipid production could be improved by the overexpression of stress-and/or lipid-related TFs, including Nannochloropsis salina by overexpressing a basic helix-loophelix transcription factor (NsbHLH2), AtWRI1, and NsbZIP1 (Kang et al., 2015a;Kang et al., 2017;Kwon et al., 2018).
Various cultivation methods are used for microalgal lipid production. Batch cultivation is widely used for the confirmation of phenotypes at the lab scale, and for mass production of biomass at the industrial scale. However, the culture conditions change as the cells get older in batch culture, so it can be difficult to determine optimal conditions for production of biomass and target products (Fernandes, Mota, Teixeira, & Vicente, 2015). On the other hand, continuous cultivation allows the maintenance of stable culture conditions. In particular, nutrient concentrations can be maintained by feeding fresh medium continuously using a chemostat, and cell density and photosynthetic efficiency can also be maintained using a turbidostat and luminostat. These various continuous modes allow to maintain optimal growth condition by providing nutrients continuously (Fernandes et al., 2015;Ho et al., 2014). As a result, the productivity of continuous cultivation is generally 2.3-to 5-times higher than batch cultivation systems (Fernandes et al., 2015;Lee et al., 2013). The additional advantages of continuous cultivation are reduced costs for harvesting, labor, cleaning, and sterilization. Continuous cultivation is thus a promising approach for industrial cultivation of microalgae (Fernandes et al., 2015;Mata, Martins, & Caetano, 2010).
We reported previously that the overexpression of NsbHLH2 TF could improve the production of biomass and lipids in N. salina (Kang et al., 2015a). However, lipid productivity of NsbHLH2 transformants became similar to that of wild-type (WT) at the end of the cultivation under the batch culture due to nutrient depletion. In the present study, we thus subjected one of the NsbHLH2 strains to continuous cultivation using a flat-panel photobioreactor (PBR) for maximized production of biomass and lipids by providing nutrients constantly and maintaining optimal conditions, which was not possible with the batch mode. We then attempted to identify the genes possibly regulated by NsbHLH2 TF that have roles in the growth and lipid synthesis (Anderson, Muff, Georgianna, & Mayfield, 2017), and verified the messenger RNA (mRNA) expression of these genes via quantitative real-time polymerase chain reaction (qRT-PCR). Taken together, this study provides proof-of-concept that TF engineering with NsbHLH2 can be used for production platform, which can be managed/optimized by continuous cultivation in PBRs.
A flat-panel PBR was used for the continuous cultivation of WT and transformant cells (Sung, Lee, Kim, Nam, & Chang, 2017).
This PBR has dimensions of 35 mm × 240 mm × 310, and 2.5 l working volume. LED irradiation (100 μmol photons m −2 s −1 ) was applied to each side of the PBR, so the total light intensity was 200 μmol photons m −2 s −1 . Microalgal cells were cultivated in the PBR with 0.1 vvm of 2% CO 2 at 25°C. F2N medium without Tris-HCl was used for continuous cultivation. For continuous cultivation, F2N medium was supplied using a peristaltic pump (EMS Tech, Yongin-si, Gyeonggi-do, South Korea). The working volume was maintained by an overflow channel in the PBR.

| Growth and nutrient concentration analysis
The cell growth was assessed by the measurement of the cell density and dry cell weight (DCW). The cell density was determined using a Cellometer Auto X4 Cell Counter (Nexcelom Bioscience, Lawrence, MA). DCW was determined by passing cells through a GF/C filter paper (Whatman, Maidstone, Kent, UK), washing with deionized water, drying at 105°C overnight, and then weighing. The specific growth rate was calculated by the following equation (Bailey & Ollis, 1977;Monod, 1949): where, X f is the biomass concentration of the feed stream, X is the biomass concentration in the PBR, D is the dilution rate, and µ is the specific growth rate. As the feed stream has no biomass, X f can be ignored. In addition, dX dt is zero when the culture is at steady state. Thus, the above equation can be simplified to: Therefore, the specific growth at a steady state is the same as the dilution rate. Biomass productivity at steady state was thus calculated by multiplying DCW and dilution rate. We calculated the maximum specific growth rate (μ m ) and the halfsaturation constant (Ks) using the Monod equation (Monod, 1949): where, μ m is the maximum specific growth rate, S is the concentration of the limiting substrate, and Ks is the half-saturation constant. By inverting Equation (3), we obtain the following: By plotting of 1/D and 1/S, we estimated μ m and Ks.

| Fatty acid methyl ester analysis
The fatty acid methyl ester (FAME) analysis was conducted as previously described (Kang et al., 2015b Figure S1). The Pearson chi-square test was used for the statistical analysis of the significance (α = 0.05). On the basis of GO annotations, 18 genes related to the growth and lipid synthesis were selected to investigate mRNA expression levels. Supporting Information Figure S2 shows the locations of the CDS and E-boxes in these 18 selected genes.
The expression of mRNA of the selected gene was determined by qRT-PCR, as previously described (Kang et al., 2017). The cells which F I G U R E 1 The growth and nutrient concentration analysis of NsbHLH2 transformant according to the dilution rate. (a) Cell density, (b) biomass productivity, and (c) concentration of NO 3 . Feed NaNO 3 concentration was fixed at 427.5 mg l −1 . The data points represent the average of samples and error bars indicate standard error (n = 3). Significant differences against WT for the same dilution rate conditions, as determined by Studentʼs t test, are indicated by asterisks (*p < 0.05, **p < 0.01, ***p < 0.001).  Table S1 shows the primers that were used.

| RESULTS
3.1 | Effect of dilution rate on cell growth, biomass, and FAME productivities In the previous study, we reported the development of two NsbHLH2 overexpressing transformants, 3-6 and 3-11 (Kang et al., 2015a). We selected the NsbHLH2 3-6 transformant for continuous culture experiments because 3-6 behaved better for the biomass productivity than 3-11 in batch mode experiments. Continuous cultivation was performed using different dilution rates (D) with fixed feed NaNO 3 concentration (427.5 mg l −1 ). As the dilution rate increased, the cell density of the WT and the transformant decreased, but the cell density of the transformant was higher than WT at all dilution rates ( Figure 1a). In particular, at high dilution rates of 0.35 and 0.5 day −1 , the cell density of the transformant was about 38% and 32% higher than that of WT, respectively ( Figure 1a). We also assessed the biomass productivity (Figure 1b). At a high dilution rate of 0.5 day −1 , biomass productivity of the transformant was increased to 0.38 g l −1 day −1 , which was the highest biomass productivity and about 30% F I G U R E 2 Correlation of 1/D and 1/S for the estimation of the maximum specific growth rate (µ m ) and half-saturation constant (K s ) in the Monod equation [Equation (4)] in the (a) WT and (b) NsbHLH2 transformant. NsbHLH2: Nannochloropsis salina by overexpressing a basic helix-loop-helix transcription factor; WT: wild-type The FAME analysis of NsbHLH2 transformant according to the dilution rate. (a) FAME content and (b) FAME productivity. Feed NaNO 3 concentration was fixed at 427.5 mg l −1 . The data points represent the average of samples and error bars indicate standard error (n = 3). Significant differences against WT for the same dilution rate conditions, as determined by Studentʼs t test, are indicated by asterisks (*p < 0.05, **p < 0.01, ***p < 0.001). FAME: fatty acid methyl ester; NsbHLH2: nannochloropsis salina by overexpressing a basic helix-loop-helix transcription factor; WT: wild-type [Color figure can be viewed at wileyonlinelibrary.com] higher than that of the WT (Figure 1b). The transformant also consumed more nitrate than WT at high dilution rates of 0.35 and 0.5 day −1 (Figure 1c).
We determined the maximum specific growth rate (μ m ) and the half-saturation constant (Ks) using the Monod equation (Monod, 1949). A plot of 1/D versus 1/S with Equation (3) allows calculation of the maximum specific growth rate of the WT (0.43 day −1 ) and transformant (0.47 day −1 ) and of the half-saturation constant of the WT (5.05 mg l −1 ) and the transformant (5.95 mg l −1 ; Figure 2). Thus, it was confirmed that the transformant had a higher growth rate by consuming more nitrate.
We then examined the effect of the dilution rate on FAME production ( Figure 3). The total FAME content of WT and the transformant was similar at all dilution rates ( Figure 3a). The FAME content of each strain was higher at a lower dilution rate (0.15 day −1 ) than at other dilution rate due to nitrate limitation ( Figure 1c).
Consequently, FAME productivity of the transformant was increased by 62.2 mg l −1 day −1 at a dilution rate of 0.15 day −1 (Figure 3b).
We also analyzed the effect of the dilution rate on FAME composition ( Table 1). The major FAMEs in N. salina were palmitic acid (PA; C16:0), palmitoleic acid (POA; 16:1), and eicosapentaenoic acid (EPA; C20:5). The transformant contained lower levels of PA and POA, but a higher level of EPA. As the dilution rate increased, the EPA content increased, and the transformant had the highest EPA content about 33% higher than WT at the dilution rate of 0.5 day −1 , revealing the potential of NsbHLH2 3-6 as a producer of EPA which is a high-value product.
3.2 | Effect of feed nitrate concentration on cell growth, biomass, and FAME productivities Once we determined the optimum dilution rate of 0.5 day −1 for biomass production, we varied another parameter, the feed NaNO 3 concentration, and tested its effect on the growth and lipid production ( Figure 4). Cell density and biomass productivity were higher in the transformant than WT at all tested feed NaNO 3 concentrations ( Figure 4a and b). Cell density and biomass productivity of the transformant and WT were lower at a 75 mg l −1 feed NaNO 3 concentration than at 225 mg l −1 or more feed NaNO 3 concentrations, probably due to the limited nitrate supply. Nitrate uptake was higher in NsbHLH2 than in WT at the feed NaNO 3 concentration of 225 mg l −1 or higher (Figure 4c), consistent with  Note. FAME: fatty acid methyl ester, NsbHLH2: nannochloropsis salina by overexpressing a basic helix-loop-helix transcription factor; WT: wild-type.
The data points represent the average of samples and error bars that indicate standard error (n = 3). Significant differences against WT for the same dilution rate conditions, as determined by Studentʼs t test, are indicated by asterisks (*p < 0.05, **p < 0.01, ***p < 0.001).
NsbHLH2 3-6 showed a reduced composition of medium chain FAMEs, while increased polyunsaturated fatty acids (PUFAs) in most nitrate concentrations.

| Molecular analysis of NsbHLH2-regualted candidate genes involved in growth and lipid synthesis
We identified the genes possibly regulated by NsbHLH2 by screening promoter regions of N. salina for the presence of the E-box (CAnnTG), the binding site of bHLH TFs (Anderson et al., 2017). GO analyses of the identified genes revealed a wide variety of cellular components, molecular functions, and biological processes, including many with unknown functions (Supporting Information Figure S1). Among these, we selected 18 genes that have been known to be related to carbon utilization, glucose and cellulose metabolic processes, photosynthesis, and fatty acid metabolisms (Table 3). These genes and NsbHLH2 were subjected to qRT-PCR to determine if their expression is affected by transgenic NsbHLH2 TF, as summarized in Figure 6.
Pyruvate kinase (PK), which are involved in glycolysis (Bar-Even, Overall, NsbHLH2 appears to positively regulate genes involved in cellulose synthesis and degradation and lipid anabolic processes, while negatively regulated genes were involved in photosynthesis.
On the other hand, NsbHLH2 appeared not to consistently regulate the genes involved in glucose metabolism that lie in-between the carbohydrate and lipid pathways. This may represent activation of carbohydrate and lipid metabolism, resulting in the improvement of the growth and lipid production.

| Comparison of cell growth, biomass, and FAME productivities of NsbHLH2 transformant between batch and continuous cultivation
Since performance of NsbHLH2 transformants in batch culture has been reported (Kang, 2015), we compared our current data of continuous cultivation at a dilution rate of 0.5 day −1 to the previous ones (Table 4). Overall, biomass productivity was increased under continuous cultivation, compared with the batch culture. Under normal condition, biomass productivity of WT and NsbHLH2 3-6 was 71% and 114% greater under continuous cultivation than under batch cultivation, respectively. Under N limitation condition, FAME productivity of WT and NsbHLH2 3-6 was 52% and 89% higher under continuous cultivation than under batch cultivation, respectively. Furthermore, under continuous cultivation with normal condition, EPA composition was higher than any other conditions, resulting in the highest EPA productivity in NsbHLH2 transformant.
Taken together, continuous cultivation significantly improved biomass, FAME, and EPA productivities of NsbHLH2 transformant, which can be readily applied to scale-up cultivation for value-added products and biofuel production.

| DISCUSSION
Nannochloropsis are important industrial microalgae due to their robust growth and high lipid contents, and extensive efforts are dedicated to genetic improvements (Ma, Chen, Yang, Liu, & Chen, 2016;Poliner, Takeuchi, Du, Benning, & Farre, 2018;Radakovits et al., 2012). In particular, TF engineering has been used in Nannochloropsis for improving biomass and/or lipid production (Ajjawi et al., 2017;Kang et al., 2017;Kwon et al., 2018); however, these have been achieved in the batch culture in a laboratory scale, and lack assessment of potential for industrial production. We also reported a similar study of NsbHLH2 overexpression in Nannochloropsis showing limited improvement in biomass and lipid production under the batch culture (Kang et al., 2015a). We continued improvement of this strain for optimized biomass and lipid F I G U R E 5 The FAME analysis of NsbHLH2 transformant according to feed NaNO 3 concentration. (a) FAME content and (b) FAME productivity. The dilution rate was fixed at 0.5 day −1 . The data at 427.5 mg l −1 of feed NaNO 3 concentration are same as for the dilution rate of 0.5 day −1 in Figure  3. The data points represent the average of samples and error bars indicate standard error (n = 3). Significant differences against WT for the same feed NaNO 3 concentration, as determined by Studentʼs t test, are indicated by asterisks (*p < 0.05, **p < 0.01, ***p < 0.001). FAME: fatty acid methyl ester; NsbHLH2: nannochloropsis salina by overexpressing a basic helix-loop-helix transcription factor; WT: wild-type [Color figure can be viewed at wileyonlinelibrary.com] production through continuous cultivation by adjusting dilution rate and nitrate concentration in flat-panel PBR.
We achieved maximal biomass productivity of 377.8 mg l −1 day −1 in the transformant, 30% higher than WT, at a high dilution rate of 0.5 day −1 with a fixed feed of nitrate at 427.5 mg l −1 . This result is consistent with previous batch culture experiments (Figure 1b). As the dilution rate represents a growth rate at a steady state, as described by Equation (2) (μ = D; Bailey & Ollis, 1977;Monod, 1949), the exponential growth phase under batch culture is similar to continuous cultivation at a high dilution rate (0.5 day −1 ). Indeed, the transformant showed high biomass productivity by consuming more nitrate during the exponential phase of batch culture and at a high dilution rate of continuous cultivation (Kang et al., 2015a). In addition, continuous supply of nutrients with the high dilution rate provided a further increased biomass productivity of the transformant in continuous cultivation, 2.1-fold higher than in batch culture (Table 4).
Nitrogen starvation is the best method for the induction of lipid accumulation (Sharma, Schuhmann, & Schenk, 2012;Zienkiewicz, Du, Ma, Vollheyde, & Benning, 2016). We also improved lipid productivity in NsbHLH2 transformant by providing low feed NaNO 3 concentration of 75 mg l −1 at the fixed high dilution rate (0.5 day −1 ), mimicking the nitrate limitation conditions that we have used in the batch culture (Kang et al., 2015a). As a result, FAME productivity was increased up to 83.6 mg l −1 day −1 , which was about a 90% increase compared with the batch culture. Taken together, we verified that the transformant grew better than WT by consuming more nitrate using continuous cultivation. We could also overcome the limitations of batch culture, including exhaustion of certain nutrients toward the end of the cultivation and accompanied changes of culture conditions. Such nutrient depletion and changes of culture conditions can be overcome by feeding nutrient continuously and maintaining optimal culture condition constantly. This particularly benefited our transformant that consumes nutrient faster than WT, resulting in an increased biomass and lipid production compared with the batch culture.
To understand the mechanism behind the improved biomass and lipid productivities, we identified possible target genes based on the presence of the E-box that are known to be the binding site for bHLH TFs. Analyses of expression patterns of these genes and their assignment in the central carbon metabolic map revealed interesting patterns as shown in Figure 6 and Supporting Information Figure S3.
We confirmed mRNA expression levels of the genes under normal and N limitation conditions, and found that the expression patterns were different according to nitrogen concentration in the medium.
a These data are same as the data at a dilution rate of 0.5 day -1 in Table 1.
The data points represent the average of samples and error bars indicating standard error (n = 3). Significant differences against WT for the same feed NaNO 3 concentration, as determined by Studentʼs t test, are indicated by asterisks (*p < 0.05, **p < 0.01, ***p < 0.001).
T A B L E 3 Growth and lipid synthesis related genes with promoters containing E-box in N. salina CS-2 C: membrane; F: cellulose synthase (UDP-forming) activity; P: cellulose biosynthetic process P: starch metabolic process; P: sucrose metabolic process; P: UDP-glucose metabolic process Cellulase 2

| 563
First, positively regulated genes included CS-1, CS-2, CL2-1, and CL2-2 that are known to be involved in cellulose biosynthesis and degradation. These contrasting groups of genes may contribute to the growth via enhanced cellulose metabolism. Supporting this possibility, it has been reported that both CS and CL were highly expressed during the cell division cycle in dinoflagellate (Shi et al., 2017). In addition, overexpression of CS genes enhanced cell growth and biomass yield in plants (Hu et al., 2018). It has also been reported that cellulose synthesis and degradation are required for growth and stress response in plants (Kesten, Menna, & Sanchez-Rodriguez, 2017). Therefore, NsbHLH2 might be able to improve Other carbon metabolic genes that could be regulated by NsbHLH2 included PK, PPKD, and PEPCK, which are involved in the production of pyruvate, phosphoenolpyruvate, and OAA, respectively. Located at the metabolic hub of gluconeogenesis, glycolysis, and TCA cycle, these molecules and enzymes play important roles in carbon concentrating and partitioning depending on cellular needs (Johnson & Alric, 2013;Polle et al., 2014;Radakovits et al., 2012).
We observed upregulation of PC, and downregulation of PK, PPDK, and PEPCK, which can result in the accumulation of OAA. OAA and associated metabolites may contribute to lipid biosynthesis, in which a cyclic reaction of OAA → malate → pyruvate (catalyzed by malate dehydrogenase, malic enzyme, and PC, respectively) is known to provide nicotinamide adenine dinucleotide phosphate, the critical reducing power for FA synthesis (Ratledge, 2014). It should also be noted that ATP: citrate lyase coverts citrate to OAA by producing acetyl-CoA that can be used for fatty acid synthesis (Liang & Jiang, 2013;Tan, Lin, Shen, & Lee, 2016). Overlapping with the so-called pyruvate hub (Dolch et al., 2017), these reactions may contribute to the increased lipid production.
We also observed an increased expression of FAE and MACS in the NsbHLH2 transformant that can increase long-chain FAs, while decreased the expression of membrane lipid catabolic genes (PL and LPL;Legeret et al., 2016;Trentacoste et al., 2013). The outcome of these metabolic regulations might increase accumulation of PUFAs including eicosapentaenoic acid (EPA) (Barker et al., 2007;Kaye et al., 2015), consistent with our FAME profile (  Figure 3a, 5a, and Supporting Information Figure S5). Therefore, the aforementioned changed carbon metabolism by NsbHLH2 TF might activate gluconeogenesis, glycolysis, and TCA cycle simultaneously without shifting the flow of carbon biased either way, resulting in an increased growth rate and biomass productivity. In addition, there were negative aspects for the growth such as downregulated photosynthesis genes (RLSMT and LHP), which can be compensated by positive impacts of metabolic pathways, including activated cellulose and chrysolaminarin metabolism, downstream carbon metabolism, and increased EPA contents.
Taken together, our NsbHLH2 overexpression line could be optimized for maximal production of biomass, lipids, and EPA via F I G U R E 6 The expression profiles of NsbHLH2 and NsbHLH2-regulated genes involved in growth and lipid synthesis in NsbHLH2 transformant. Each mRNA expression was measured at Day 0, 1, and 3 under normal (N) and nitrogen limitation (NL) conditions. The expression of all genes was determined by qRT-PCR and normalized to β-actin. Table 3 provides the full names of all the genes. The data points represent the average of samples and error bars indicate standard error (n = 3). Significant differences against WT for the same conditions and same time points, as determined by Studentʼs t test, are indicated by asterisks (*p < 0.05, **p < 0.01, ***p < 0.001). mRNA: messenger RNA; NsbHLH2: nannochloropsis salina by overexpressing a basic helix-loop-helix transcription factor; qRT-PCR: quantitative real-time polymerase chain reaction; WT: wild-type [Color figure can be viewed at wileyonlinelibrary.com] continuous cultivation, supported by the regulation of possible target genes that are involved in carbon and lipid metabolism. We previously reported that overexpression of NsbHLH2 can improve biomass and lipid production in Nannochloropsis (Kang et al., 2015a), and now provided a solid foundation that this strain can be used in lipid production via continuous cultivation.

| CONCLUSIONS
We examined the use of continuous cultivation in flat-panel PBR to grow the NsbHLH2 overexpressing transformant in an effort to ultimately develop a system for the industrial production of biofuels and biomaterials. We confirmed that the transformant showed enhanced biomass, FAME, and EPA production by consuming more nutrients using a continuous cultivation, compared with batch cultivation. In addition, N limitation with continuous cultivation led to the greatest FAME productivity in this transformant. Our examination of the mechanism of the enhanced growth and lipid production in the transformant indicated altered expression of numerous genes, especially those with roles in carbon metabolism.
Taken together, NsbHLH2 is expected to be used in mass production of biofuels and EPA from Nannochloropsis and other industrial microalgae.
T A B L E 4 Comparison of cell growth and FAME production of the WT and NsbHLH2 transformant under batch culture and continuous culture The batch culture of "N limitation" condition was conducted in Erlenmeyer baffled flasks (200 mL working volume) for 8 days (Kang et al., 2017).