The Antarctic freshwater microalga Micractinium variabile KSF0031 was isolated from Ardley Island, South Shetland Islands, Antarctica (Chae et al. 2019). The culture was maintained in BG-11 medium (Stanier et al. 1971). The BG-11 medium contained the following (mM): NaNO₃, 17.60; K₂HPO₄, 0.23; MgSO₄·7H₂O, 0.30; CaCl₂·2H₂O, 0.24; citric acid, 0.031; ammonium ferric citrate green (AFCg), 0.021; EDTA·Na₂, 0.0027; Na₂CO₃, 0.19; trace metals, 0.058 (H₃BO₃, 0.046; MnCl₂·4H₂O, 0.009; ZnSO₄·7H₂O, 0.077; Na₂MoO₄·2H₂O, 0.0016; CuSO₄·5H₂O, 0.0003; Co(NO₃)₂·6H₂O, 0.00017).

Ten liters of seed with approximately 5 × 10⁵ cells mL⁻¹ was inoculated in 100 L of BG-11 medium at 18°C (Supplementary Fig. S1) and cultured for 24 d. The KSF0031 was irradiated with cool white fluorescent light (80 μmol photon m⁻² s⁻¹) on a 16 : 8 h light-dark cycle with filtered atmospheric air supplied at a rate of 5 L min⁻¹.

The essential components for the growth of KSF0031 were analyzed using the Plackett-Burman design (Plackett and Burman 1946). A classic way to screen for large variables is to use the Plackett-Burman design, which is programmed to identify critical parameters using n + 1 experiments (Ekpenyong et al. 2017). Based on these results, five elements of the BG-11 medium were selected and represented at two levels, low (−) and high (+) concentrations, producing a first-order model. The effect of each factor on cell numbers was calculated using Eq. (1):

, where Y denotes the response (number of cells), β₀ is the model intercept, β_i is the linear factor coefficient, and X_i is the level of each variable. The concentrations of the medium components were optimized using the Box-Behnken surface response method (Box and Behnken 1960). The Box-Behnken design was devised for analyses where each factor varies among the three levels (−1, 0, +1) (Czyrski and Sznura 2019). Accordingly, the individual factors in this study were coded into three disparate levels: (−) for low, (0) for intermediate, and (+) for high concentrations. A second-order polynomial model was designed to investigate the influence of individual components on the predicted response. A second-order polynomial model was calculated using Eq. (2):

, where Y denotes the predicted response (number of cells), β₀ is the constant coefficient, β_i, β_ij, and β_ii denote the regression coefficients of the model, and X_i and X_j represent the coded values of independent nutrient variables.

The growth of KSF0031 was analyzed by counting the cells using a hemocytometer and an optical microscope (Zeiss Axio Imager A2, Gottingen, Germany). Twenty milligrams of freeze-dried KSF0031 samples was used to extract fatty acids and phytol. Fatty acid methyl esters (FAMEs) were converted from crude lipids and analyzed as described previously (Kim et al. 2016). FAMEs and phytol were quantified using an external standard (1 mg of C22:0 in hexane); qualitative analysis was performed using a gas chromatography–mass spectrometer (TSQ 8000 EVO GC-MS; Thermo Fisher Scientific, Waltham, MA, USA) and reconfirmed using authentic standards (Supelco 37 Component FAME Mix and phytol; Sigma-Aldrich, St. Louis, MO, USA). The content (w w⁻¹, %) and yield (mg L⁻¹) were summed and expressed in terms of milligrams of FAMEs (or phytol) per gram of dry cell weight (DCW) and milligrams of FAMEs (or phytol) per liter of culture volume.

Results with a p-value < 0.05 were considered statistically significant. The optimization values of significant medium components were calculated using the Minitab software ver. 14 (Minitab Inc., State College, PA, USA) and linear multiple regression was used for the statistical analysis.

To eliminate unnecessary or negative elements of the microalgal medium, all nine components were analyzed individually (Fig. 1). In the elimination test to optimize the culture of KSF0031, we found five significant factors among the medium components, namely K₂HPO₄, MgSO₄·7H₂O, citric acid, AFCg, and Na₂CO₃, all of which had positive effects. In each medium, when any one of these five components were removed, the growth of KSF0031 was restricted compared to that observed upon using the original BG-11 medium. The non-significant factors were NaNO₃, CaCl₂·2H₂O, EDTA·Na₂, and trace metals. In each medium, when any of these four components were removed, the growth of KSF0031 was not significantly affected when compared to that observed upon using the original BG-11 medium (Fig. 1).

To select the component having optimal effects on KSF0031 cell numbers, the five components showing positive effects were tested using the Plackett-Burman design. Table 1 lists the test values, effects, t-statistics, and p-values. Results with a p-value lower than 0.05 were treated as showing a significant effect on the growth of KSF0031. NaNO₃, CaCl₂·2H₂O, EDTA·Na₂, and trace metals were not significant factors (p > 0.05) (Table 1). The remaining components (K₂HPO₄, MgSO₄·7H₂O, citric acid, AFCg, and Na₂CO₃) were significant (p < 0.05). These components were re-analyzed using a secondary Plackett-Burman design. Ten media, in each of which the concentrations were adjusted to -value (mM) and +value (mM), were prepared for the component analysis corresponding to the five significant variables (Table 2). Cell growth was measured in media adjusted to each concentration. Experiments were performed by varying the concentration range (0.2 times [−], 1.1 times [0], and 2 times [+] of the original concentration of BG-11) of the significant factors obtained through the elimination test, and the number of cells was measured. Three components showed significant effects, where two components, K₂HPO₄ and MgSO₄·7H₂O, exhibited positive effects; AFCg exhibited a negative effect (Table 2).

The Box-Behnken design was performed based on the statistical results of the Plackett-Burman design of KSF0031. It was applied to measure the interdependence of three component variables, K₂HPO₄ (X₂), MgSO₄·7H₂O (X₃), and AFCg (X₆), in their effects on cell biomass. For positive effect factors, (+) was set as twice the original concentration (mM) of BG-11 medium, and (−) was set as 1/10 of (+). For negative effect factors, (+) was set as the original concentration (mM) and (−) was set as 1/10 of (+). Other factors (NaNO₃, CaCl₂·2H₂O, citric acid, EDTA·Na₂, Na₂CO₃, and trace metals) were added at their original concentrations (mM) (Table 3).

The optimization values were calculated (Table 3). The estimated coefficients, t-statistics, and p-values were also calculated (Table 4). The effects of the other three variables at different levels are shown in Fig. 2. The expected number of KSF0031 cells was calculated using the Minitab software and determined to be 1.90 × 10⁷ cells mL⁻¹ with optimized concentrations of K₂HPO₄, MgSO₄·7H₂O, and AFCg as 0.4, 0.55, and 0.0021 mM, respectively. The final composition was as follows (mM): NaNO₃, 17.6; K₂HPO₄, 0.4; MgSO₄·7H₂O, 0.55; CaCl₂·2H₂O, 0.24; citric acid, 0.031; AFCg, 0.0021; EDTA·Na₂, 0.0027; Na₂CO₃, 0.19; and trace metals, 0.058 (Table 5).

The growth of KSF0031 and its production of lipids and phytol in the medium optimized via the Box-Behnken design were analyzed. Fig. 3B shows the number of KSF0031 cells mL⁻¹ in 100 L BG-11 culture media for 24 d. Maximal concentrations of (160 ± 8) × 10⁵ and (225 ± 11) × 10⁵ cells mL⁻¹ were obtained in the original and optimized BG-11 medium, respectively. The dry mass of harvested cells cultivated in the optimized BG-11 medium was 35.60 ± 1 g, which constituted a 37.5% improvement in biomass when compared with the dry mass of KSF0031 cells from the original BG-11 medium (Fig. 3C). Although the levels of the two components were elevated in the optimized medium, the production cost of KSF0031 in this medium could be significantly reduced (Supplementary Table S1). The biomass of KSF0031 was increased in the optimized medium where the concentration of the negative factor AFCg was reduced by 10 fold. This gain was due to relief from iron toxicity effects on the cells. Iron, which can be considered a heavy metal, plays an important role in metabolism at low concentrations but can be toxic to microalgal cells at high concentrations (Subramaniyam et al. 2016). Through statistical optimization, it is possible to determine the appropriate concentration of heavy metal ions (Jaafari and Yaghmaeian 2019). After cultivation for 24 d in the original and optimized media, lipids were extracted from the dried biomass to obtain FAMEs and phytol. The PUFA content in the total fatty acids of Micractinium reisseri was 66% (Abou-Shanab et al. 2014). The PUFA content of Micractinium singularis reported in 2020 was 55.5% (Jo et al. 2020). There was no information on PUFA production (mg g⁻¹ DCW) reported in either of these studies. In a study with M. reisseri reported in 2019, only 0.54 mg g⁻¹ DCW of PUFA was produced (Elshobary et al. 2019). Reports on PUFA and phytol production in Micractinium species are limited, and research on the applications of Micractinium sp. is in its infancy. To evaluate the production of PUFAs and phytol, gas chromatography analysis was performed (Fig. 4, Supplementary Fig. S2). The dominant fatty acids of KSF0031 were linolenic acid (C18:3) and hexadecatrienoic acid (C16:3). Sorokina et al. (2020) reported that the dominant fatty acids in Micractinium sp. IC-44 were C16:0 and C18:2 when cultured in Bold’s basal medium at 28°C. Although the two strains belong to the same genus, KSF0031 is a psychrotrophic microalga in Antarctica, which has the characteristic of producing more PUFAs (Teoh et al. 2004). Our data indicate that compared to the original BG-11, using the optimized BG-11 medium leads to higher FAME and phytol production values (13.03 and 10.98 mg g⁻¹ DCW, representing 12.04 and 48.78% yield increases, respectively) (Table 6). Phytol-producing microalgal strains have rarely been reported; the phytol contents of the ethanol extracts of Porphyridium cruentum, Phaeodactylum tricornutum, and Chlorella vulgaris were 3.1, 8.4, and 5.5%, respectively (Rodriguez-Garcia and Guil-Guerrero 2008). There have been no studies on phytol production using polar microalgae, and here, we performed an optimized medium design for the first time. The Antarctic microalga KSF0031 showed a high phytol content of 33.29% of total fatty acids, and we were able to improve the phytol content to 39.95% by using optimized BG-11 medium.