Biomass and RRR-α-tocopherol production in Stichococcus bacillaris strain siva2011 in a balloon bioreactor

Background Green microalgae represent a renewable natural source of vitamin E. Its most bioactive form is the naturally occurring RRR-α-tocopherol which is biosynthesized in photosynthetic organisms as a single stereoisomer. It is noteworthy that the natural and synthetic α-tocopherols are different biomolecular entities. This article focuses on RRR-α-tocopherol production in Stichococcus bacillaris strain siva2011 biomass in a bioreactor culture with methyl jasmonate (MeJa) elicitor. Additionally, a nonlinear mathematical model was used to quantitatively scale-up and predict the biomass production in a 20 L balloon bioreactor with dual variables such as time and volume. Results Approximately 0.6 mg/g dry weight (DW) of RRR-α-tocopherol was enhanced in S. bacillaris strain siva2011 biomass with the MeJa 50 μL/L for 24 hrs elicitations when compared to the control. The R2 value from the nonlinear model was enhanced up to 95% when compared to the linear model which significantly improved the accuracy for estimating S. bacillaris strain siva2011 biomass production in a balloon bioreactor. Conclusions S. bacillaris strain siva2011 is a new green microalga which biosynthesizes significant amounts of RRR-α-tocopherol. Systematically validated dual variable empirical data should provide key insights to multivariable or fourth order modeling for algal biomass scale-up. This bioprocess engineering should provide valuable information for industrial production of RRR-α-tocopherol from green cells.

Background RRR-α-tocopherol is a lipid soluble small molecule and the biologically active form of natural vitamin E. RRR-αtocopherol is exclusively biosynthesized by photosynthetic organisms or green cells including algae, plants, and cyanobacteria [1][2][3]. Plant-based products are a primary source of RRR-α-tocopherol in the human diet. For example, hazelnut is one of the richest sources of vitamin E [4]. It is known that vitamin E plays an important role in human nutrition as a natural antioxidant. It was recently proposed that vitamin E is active against oxidative stress-related diseases [5]. Reportedly, it suppresses telomerase activity in ovarian cancer cells [6]. Vitamin E enhances IL-2 production, gene expression, and is an effective therapeutic adjuvant [7]. Vitamin E deficiency affects both T and B immune cell functions [8], the α-tocopherol transfer protein (α-TTP) gene, and neurologic dysfunctions. In animal models, vitamin E mixtures inhibit colon, lung, mammary, and prostate carcinogenesis [9], as well as prevent diabetes [10]. Cosmetic industries also extensively use vitamin E in skin care products. In addition, RRR-α-tocopherol prolongs the shelf life of meat [11]. Thus, the use of RRR-α-tocopherol continues to increase in nutraceuticals [12].
Bioreactor technology is the key component for the industrial scale production of bioactive small molecules for pharmaceutical applications. For instance, bioreactor technology was successfully developed and scaled-up to 10,000 L for commercial-scale production of ginseng roots used for human health-related applications [13,14]. This reactor configuration has also been tested for RRR-αtocopherol production in lab-scale photosynthetic hazelnut root culture [15]. However, knowledge regarding the bioprocessing of green algal cells for production of RRR-α-tocopherol in balloon bioreactors is limited. Previously, Stichococcus bacillaris strain siva2011 biomass was scaledup in a lab-scale balloon bioreactor (4 L -8 L), and a linear fitting model for predicting scale-up was proposed [16]. The S. bacillaris strain siva2011 has unique lipid ( Figure 1) [16] and vitamin E metabolisms to lead to bioactive RRR-α-tocopherol production. A nonlinear model enables more accurate estimates and should provide insight for quantitatively predicating algal biomass accumulation in a large volume bioreactor. The objectives of this study were to: 1) evaluate RRR-α-tocopherol content from S. bacillaris strain siva2011 biomass with MeJa elicitation; 2) quantitatively predict S. bacillaris strain siva2011 DW in a 20 L bioreactor with simplified stepwise dual variables (time and volume) from 4 L and 8 L data using nonlinear regression. This systematic study could provide insight regarding stepwise nonlinear scale-up of S. bacillaris strain siva2011 biomass for RRR-α-tocopherol production in a balloon bioreactor.

Results and discussions
Natural vitamin E occurs in two general forms, namely the tocopherols and tocotrienols, which are collectively called tocochromanols. Each has four distinct isoforms having α, β, γ, or δ substitution on the chromanol. The natural α-tocopherol contains three methyl groups on the chromanol moiety at positions 5, 7, and 8. The phytyl or saturated side chain is attached to the C-2 position of the chromanol ring which has three chiral centers with a single RRR stereoisomer ( Figure 2). The chromanol groups have two fused rings, a phenol and a tetrahydropyran, sharing a 2 carbon bridge [17]. These rings are moderately polar, giving them an affinity for the cellular membrane surface while the phytyl tail is hydrophobic and normally associated with membrane lipids [18]. These structural features of RRR-α-tocopherol are efficiently acted on by the human hepatic α-TTP which is responsible for maintaining plasma α-tocopherol concentrations [19]. The synthetic α-tocopherol called all-racemic-α-tocopherol is not identical to RRR-α-tocopherol. It is an equimolar mixture of eight stereoisomers which possess three chiral centers at positions 2' , 4' , and 8' , giving rise to four diastereoisomeric pairs of enantiomers such as RRR, RSR, RRS, RSS, SRR, SSR, SRS, and SSS [20]. Moreover, α-TTP has a high affinity to RRR-α-tocopherol and has a 3-fold greater binding half-life when compared to synthetic α-tocopherol [21]. Thus, the bioactivity and the relative safety are different. Human proteins such as enzymes and receptors usually exhibit high stereospecificity [20]. Therefore, the natural RRR-α-tocopherol is more bioactive than the synthetic form.
RRR-α-tocopherol plays a major role as an antioxidant which prevents lipid peroxidation. In the photosynthetic cells, it may protect photosystem II during photoinhibition and repair chloroplast mechanisms [18]. This is due to the hydroxyl group on the C-6 position which is the active site that donates a hydrogen atom. The phenolic hydrogen atom is capable of scavenging lipid peroxy radicals and quenching singlet oxygen [22]. RRR-α-tocopherol is recycled in the photosynthetic cell by cytosolic ascorbate which oxidizes one-electron from the tocopheroxyl radical thus regenerating vitamin E [23]. This mechanism might protect the cell membranes.
RRR-α-tocopherol is biosynthesized in photosynthetic cells via two different pathways [24]. The phytyl domain precursor comes from an isoprenoid pathway, and the chromanol domain precursor comes from an alternative shikimate pathway homogentisic acid via complex enzymatic reactions [25]. RRR-α-tocopherol is found in the chloroplast envelope, thylakoids, and plastoglobuli of the plastid. The vitamin E biosynthetic pathway has been elucidated in Arabidopsis [26]. The genes associated with vitamin E biosynthesis in photosynthetic organisms have been well described in literature [3,27]. A significant metabolic engineering effort has been made to improve vitamin E content both in plants [28] and in cyanobacteria [29,30]. Moreover, tocopherol production in plant green callus [31], cell [32], and root cultures [15] have also been reported.
Tocopherol production can be enhanced by molecular elicitation which is non-transgenic. MeJa is a plant stress volatile signaling molecule which up-regulates several defense-related genes [37]. In plants, jasmonates are biosynthesized via the octadecanoid pathway; exogenous MeJa treatment is up-regulating secondary metabolic pathway genes especially those encoding for stress protection [38]. Therefore, it is used as one of the potential molecular elicitors in plant root culture to enhance several pharmaceutical molecule productions [15]. For instance, MeJa elicitation increases the activity of tyrosine aminotransferase in plant green cell culture which is one of the initial step enzymes involved in tocopherol biosynthesis [31,39]. To increase RRR-α-tocopherol content in S. bacillaris strain siva2011, MeJa elicitation was also used. The S. bacillaris strain siva2011 characteristics, culture conditions, and bioreactor experimental designs were previously reported for lipid production [16]. Chromanol nucleus Phytyl chain Figure 2 Chemical structure of RRR-α-tocopherol.
accumulated 0.7 mg/g DW of RRR-α-tocopherol which was detected starting during the early exponential growth phase. The lower concentration of MeJa, 50 μL/L, after 24 hrs elicitation, enhanced the production of RRR-αtocopherol to the highest concentration, 1.3 mg/g DW. The higher MeJa concentrations or longer elicitation periods were inhibiting both to the biomass growth and the resultant RRR-α-tocopherol production. MeJa can diffuse to cells either by intercellular migration or while in the vapor phase [40]. In plants, MeJa can transport from leaves to roots [41]. The vapor signaling can be transported to distal plants via air, and the intercellular signaling can be transported via vascular process [42]. For instance, in an in vitro root culture the MeJa elicitation could trigger the defensive molecules accumulation via intercellular transport [13][14][15], whereas in in vivo plants during herbivore attack it can act as a volatile signal [42]. In the algal culture, MeJa elicitation could trigger the RRR-α-tocopherol production by either intercellular signaling or both. The optimum concentration can upregulate the tocopherol biosynthetic pathway enzymes in S. bacillaris strain siva2011 which could increase the antioxidant. MeJa elicitation rapidly activates the defensive genes which also down regulates the photosynthetic system genes [38]. In addition, MeJa induces reactive oxygen species (ROS) which alter the mitochondrial and chloroplast dynamics [43]. Thus, the higher MeJa concentrations or longer elicitation periods can produce uncontrolled ROS which can precede chloroplast or photosynthetic dysfunctions which could be inhibiting the tocopherol biosynthesis and biomass accumulation in S. bacillaris strain siva2011. In green cells, chloroplasts are an essential organelle for energy capture and transduction; a decline in photosynthetic activity is closely related to the decrease in the biomass. The typical MeJa elicited cells' symptoms were loss of chlorophyll, which causes the decline in the net photosynthetic rate, and degradation of ribulose bisphosphate carboxylase, etc. [44]. For instance, 100 μL MeJa at 9 hrs elicitation altered chloroplast morphology and function which is associated with cell death [43]. Even though RRR-α-tocopherol indirectly regulates the amounts of jasmonic acid [18], the decline in chloroplast efficiency could down regulate the RRR-α-tocopherol metabolism. This suggests that higher MeJa concentrations or longer elicitation could be cytotoxic beyond what was studied.

Nonlinear regression
Compared to green adventitious roots, photosynthetic algae have a higher ability to biosynthesize RRR-α-tocopherol in bioreactor cultures. The balloon bioreactor is a liquid-phase reactor with enhanced geometry and efficient fluid flow dynamics which could help provide higher mass transfer efficiency [14]. When compared to other photobioreactors, the balloon bioreactor had a larger headspace which efficiently captured light and enhanced photosynthesis [16]. Thus, S. bacillaris strain siva2011 was scaled-up in a balloon bioreactor (Figure 4) to investigate enhanced biomass accumulation. Of the four concentrations of CO 2 tested, 0.2% yielded the highest biomass of 3.45 g/L in 4 L and 3.79 g/L (DW) in 8 L on day 6 [16]. The RRR-α-tocopherol production was unchanged by the 0.2% CO 2 (data not shown).
To quantitatively predict larger-scale algal biomass production based on lab-scale test data, the following stepwise structured approach was proposed [45,46]: 1) to set up the simplest model to linearize with dual variables; 2) to model a nonlinear regression with dual variables; and 3) to demonstrate multiple variables based on a nonlinear regression. Although scale-up predication requires multiple variables, the 6 days algal culture does not significantly utilize all the media components. Therefore, selection of important dual variables can give insight on the efficiency of parameter selection for lab-scale validation and initial scale-up prediction. In addition, the multiple linear regression models might not incorporate the underlying nonlinear relationships [47]. So, in this study the second systematic approach was conducted to evaluate nonlinear modeling for scale-up prediction in a 20 L balloon reactor using dual variables. Nonlinear regression models are more important tools than linear models because they provide better parsimony, interpretability, and prediction [48]. Figure 5 illustrates nonlinear modeling of predicted S. bacillaris strain siva2011 biomass accumulation in a 20 L balloon bioreactor with 0.2% CO 2 . This model was generated using 4 L to 8 L data of S. bacillaris strain siva2011 and shows only small discrepancies between measured and predicted data. The  nondimensional DW* can be converted into dimensional DW in g/L by multiplying by the maximum DW experimentally obtained from the baseline test which in this case is 4 L. The nonlinear modeling agrees with measured data both qualitatively and quantitatively, where the modeling has enhanced the R 2 value up to 95% compared to the linear model value 87.4% [16]. This suggests that the nonlinear regression approach enhances accuracy of modeling which provides key scale-up informations of S. bacillaris strain siva2011 biomass for RRR-α-tocopherol production. In this study, the significance of this empirical approach provides insights into the application of a nonlinear regression model that increases the R 2 value and enhances the quantitative predication of S. bacillaris strain siva2011 scaled-up in a 20 L bioreactor. This allows for the systematic understanding and design of a multi-variable nonlinear regression experiment for significant biomass production.

Conclusions
Photosynthetic microalgae are rich in RRR-α-tocopherol and a potential source for this natural antioxidant which is an essential human micronutrient. A significant advantage of this natural source is the maintenance of the specific bioactive form needed for nutrition and the elimination of possible issues with potential unknown or unexpected toxicities from the synthetic conformations. S. bacillaris strain siva2011 has a unique vitamin E biosynthetic mechanism capable of sustaining high levels of production, including inducible enhanced production which could provide a possible production platform of RRR-α-tocopherol for pharmaceutical industries. A nonlinear mathematical model was developed to model scale-up to production in 20 L reactors using an approach with a higher accuracy based on the dual variables tested in 4 L and 8 L reactors. The R 2 value from this study demonstrates that this nonlinear approach significantly improves estimation of S. bacillaris strain siva2011 biomass production in the bioreactor than does the linear model. Additional studies with progressively larger reactors (and models) will be needed to bridge the gap between laboratory and industrial scale. Nevertheless, this data provides enhanced bioprocess engineering information in the progression towards large-scale pharmaceutical RRR-α-tocopherol production from S. bacillaris strain siva2011 biomass.

Methods
Bioreactor culture, elicitation and analytics S. bacillaris strain siva2011 cells were cultured in a balloon type bioreactor (4 L and 8 L). The bioreactor experimental design and the biomass harvesting were performed as described by Sivakumar et al. [16]. For elicitation, three concentrations of filter sterilized MeJa 50, 100, and 200 μL/L were added to the S. bacillaris strain siva2011 culture on the 3 rd day. MeJa dissolved in ethanol and MeJa not dissolved in ethanol were tested, and both had a similar effect. Elicitations were carried out for 24, 48, or 72 hrs. Algal cells were harvested and freeze-dried according to the Sivakumar et al. [16] method. One gram of MeJa elicited and unelicited freeze-dried algal cells were used for analysis of RRR-αtocopherol. RRR-α-tocopherol was processed according to the Sivakumar et al. [4] method. The reversed phasehigh performance liquid chromatography chromatogram was acquired using a Dionex 3000 (LPG3400A) system (Thermo Scientific, Sunnyvale, CA) equipped with a thermostatted UltiMate 3000 autosampler and a Dionex RF 2000 fluorescence detector. The system was monitored by the software's chromeleon (6.80) for instrument control and data acquisition, data reprocessing, and solute quantification, respectively. An Agilent Zorbax Eclipse XDB-C8 (250 mm × 4.6 mm, mean particle size 5 μm) column or C18 (250 mm × 4.6 mm, mean particle size 5 μm) was used to separate RRR-αtocopherol. The mobile phase consists of a linear gradient of 90% methanol in water. The flow rate was 1 ml/ min. The total acquisition time was 35 min. The wavelength was set at 290 nm for excitation and 330 nm for emission. The authenticated RRR-α-tocopherol fluorescence spectra and retention time was used for HPLC confirmation of RRR-α-tocopherol in the samples. The RRR-α-tocopherol liquid chromatography mass spectrometry spectrum confirmation was acquired using a Shimadzu 8050 mass spectrometer.

Nonlinear regression
For a stepwise approach, the first phase of the nonlinear regression method begins with the simplification from five variables to two variables, time (t) and reactor volume (V), as shown in equation 1. The regression will be extended to multivariables when the two variables method is validated.
The associated variables DW, t, and V in equation 1 are nondimensionalized into DW*, t*, and V* as shown in equation (2) to (4).
Where DW max is the maximum DW [g/L] produced from 4 L test under 0.02% CO 2 fraction, and t max is the maximum time to reach the maximum DW. In this study, DW max and t max were 3.45 g/L and 6 days, respectively. The volume was standardized by the baseline: 4 L in this modeling.
A nonlinear model was assumed by combining 4 th order polynomials and the power form of the equation as shown in equation (5) in consideration of measured data of 4 L and 8 L.
All associated constants a, b, c, and d in equation (5) were determined from a nonlinear regression method as shown in equation (6). Equation (6) allows the predicting of DW in reactor volume 20 L at 0.02% CO 2 fraction from the correlation obtained from 4 L and 8 L measured data.

Statistical analysis
Bioreactor culture, elicitations, and analytical experiments were repeated at least three times, each with three replications. Results were expressed as the mean with standard errors. Stepwise dual nonlinear regressions were used to investigate the relationship between 4 L and 8 L in order to predict 20 L bioreactor scale-up.