Enhancement of protein production by microalgae Dunaliella salina under mixotrophic conditions using response surface methodology

Abstract

The present work was aimed at enhancing the protein production in Dunaliella salina by optimization of culture conditions. The interactive effects of medium composition on protein production were optimized using response surface methodology (RSM). A mathematical model was proposed to describe the kinetics of D. salina growth. Results showed that the mixotrophic effects were in conjunction with those of the mixture of NaCl and glucose. Based on the optimization study, protein production in D. salina could be markedly increased through glucose supplementation as low as 5 g L⁻¹. Optimal conditions were achieved at 288, 288 and 312 h with a doubling time of 1.13, 3.03 and 5.17 days for D₁, D₂ and D₃ strains, respectively. The overall protein production was enhanced by 3.4, 3.9 and 14 times in D₁, D₂ and D₃, respectively. The microplate-based approach enabled us to screen large numbers of experimental cultures in a time and cost effective manner. It also can be concluded that reducing the growth cycle through this cultivation system may expedite the analysis of transformants for production of recombinant proteins in microalgae.

---

Introduction

Dunaliella salina, unicellular halotolerant chlorophyta microalgae, is capable of producing valuable compounds such as β-carotene and glycerine.¹ The Lack of cell wall makes it easily digestible in both humans and animals. In addition, this microalga is highly efficient in the conversion of light and nutrients into biomass. Hence, it could accumulate more dry weight and glycerol under suitable cultivation conditions² as well as high protein content (57% of dry matter).³ D. salina is one of the most widely used microalgal species for mass culture with an annual production of 1200 tons.⁴ The products derived from D. salina could be categorized into three groups: β-carotene extracts, Dunaliella powder for human use and dried Dunaliella for feed use.⁵

One of the most recently used platforms, which is proposed for the purpose of molecular farming are microalgae as green “micro-bio-factories”. These microorganisms are the best model to this end, since they have the combination advantages of both bacteria and yeast (in terms of growth rate and low cost growth requirements) and animal and plants (having the ability of posttranslational modifications). In view of the high growth rate, the scalability and biosafety considerations such as the lab-suited microorganisms in controlled environments, possibility of using various microalgal species have been tested as the production platform for recombinant proteins and engineering of metabolic pathways in the production of increased levels of desirable compounds.^6,7 One of the main issues pertaining to the above is the low production yield. Accumulation of foreign proteins could be influenced by both upstream (genomics, transcriptomics, proteomics and metabolomics levels) and downstream (cultivation condition and physiological aspects of the platform organism) processes. In the upstream process many features such as utilization of strong promoters and untranslated regions (UTRs), codon adaptation, RNA secondary structure and instability, protein production and purification strategies should be taken into consideration. In downstream processing the optimum cultivation conditions with high biomass concentration is required to get high recombinant protein yield. Factors influencing microalgae growth and development including light (both intensity and photoperiod), temperature, nutrient requirements (carbon and nitrogen source), pH, aeration and contamination, may indirectly play important roles in protein accumulation.

Factors affecting the production of D. salina have been widely studied and reviewed. The effective factors on the growth and composition of D. salina are mainly light (quality and quantity), salinity, temperature and pH. Amongst these factors, salinity affects the biomass and pigment content of D. salina more than any other factors. There is also a synergistic influence between salinity and illumination of light quality and quantity in D. salina for production of carotenes.^1,8 However, the influence of salinity is strain-dependent and only a few Dunaliella strains have the high carotenogenesis potential.⁹

Two critical limiting factors, (1) labor costs and (2) low productivity, are involved in mass production of microalgal biomass. The use of efficient cultivation system such as the use of photobioreactor may be used to reduce the labor cost and to improve the production. High algal biomass (5 g L⁻¹) was obtained in cultivation using photobioreactor whereas very low algal biomass (0.5 g L⁻¹) was obtained in open pond system. Photobioreactors can effectively improve the labor costs and productivity (5 g L⁻¹ algal biomass compared to 0.5 g L⁻¹ in open ponds).¹⁰ In outdoor cultivation systems due to low light intensity, endogenous biomass consumption through respiration increases resulting lower productivity. The heterotrophic systems which rely only on the organic carbon source may be more costly than photoautotrophy. Furthermore, mixotrophic condition might reduce the cultivation costs compared to heterotrophic systems through shortening the growth cycle and thus increasing the biomass production.¹¹

Some microalgae are capable of transforming their growth habit from photoautotrophic to heterotrophic (utilizing organic materials as a source of energy for growth) or mixotrophic (combination of organic nutrition and light). This provides the feasibility of high growth rate and biomass production (up to 40 g L⁻¹).^12,13

Specific requirements with reference to the production of protein from algae (D. salina) have been reported. Protein production in algae can be modified by altering the cultivation conditions. Certain combinations of influencing factors need to be optimized for improvement of the cultivation process. With regards to the complexity of the influencing factors an in-depth knowledge of these factors needs to be substantiated for subsequent application in the optimization process. Most studies carried out to date claimed validation by statistical analysis and the combination of variables with their values and limits were arbitrarily chosen based primarily on personal experiences.^14,15

Conventional methods in optimizing fermentation/cultivation process require treating each factor separately which are laborious, incomplete and time consuming. If several factors are to be considered simultaneously their interactions are not discernible even for the dominant ones. These conventional approaches did not yield reliable results either. In this respect response surface methodology (RSM) and the experimental factorial design have been successfully applied for optimization of various biomanufacturing processes, which could also be used to investigate the interacting factors.^16,17

RSM, a non-conventional approach, is a collection of statistical and mathematical methods that can be used to quantify the interaction between different factors. This approach provides statistically reliable results with fewer numbers of experiments and is very useful for the development, improvement and optimization of the biomanufacturing processes.¹⁸ The statistical designs of RSM provide alternative methodologies to optimize a particular process by considering the mutual interactions among the various factors. This alternative approach is advantageous as it simplifies the optimization process and reduces experimental costs. Besides, the most significant factors that influenced the maximum response could also be determined. The RSM is based on response analysis, which is influenced by specific factors with the objective to define the optimum condition of the response by these factors. It is difficult to demonstrate the methodology in the form of space dimension but could only be visualized in the mind. RSM is useful in the identification of the direction in the next experiment towards an optimum point. From the optimum or near optimum point on the response surface, the equation can be determined.¹⁹

Microalgae as the lab-suited microorganisms in containment environments are among the best models for molecular farming. Most recently, these green “micro-bio-factories” are increasingly used as expression platforms. Most approaches for improvement of the recombinant protein production in molecular farming have been focused on upstream strategies. While the downstream strategies which also play critical roles in this regard, have less studied. In the present study, among the factors influencing productivity of the recombinant proteins in microalgae expression platforms we focused on downstream bioprocesses. The main objective of this study was to optimize the variables of microplate cultivation method of D. salina for enhancement of protein production using response surface methodology. The interactive effect of mixotrophy condition on growth rate, cell concentration and protein content were also analysed.

Experimental details

Microalgal species and cultivation procedures

D. salina strains, D₁ and D₂ were obtained from the Department of Biology, University of Isfahan, Iran and D₃ (UTEX 1644) was purchased from UTEX Collection (Texas University of Austin, USA). A preliminary study to determine the preferred medium for growth of D. salina was carried out by comparing three different media including modified Johnson,¹⁰ 2× Erdschrider (UTEX, the culture collection of algae) and Ben Amotz.⁹ The modified Johnson medium (M) demonstrated the ideal medium and was used for subsequent experiments. The strains were plate-streaked on solid medium containing ampicillin (100 μg mL⁻¹) and kanamycin (50 μg mL⁻¹). The stock culture preparation was carried out through inoculation of the liquid medium by a single colony obtained from the agar plate. This was performed for each strain separately from the plates that did not show any signs of contamination. The axenicity of stock cultures in terms of bacterial contamination was later confirmed through PCR amplification of 16S rDNA using primers as described by Weisburg et al. (1991).²⁰ To start the cultivation, three different strains of D. salina (D₁, D₂ and D₃) were cultured into four 96-well microplates (Nunc, Inc.). Two hundred microliters of aliquots were pipetted into each well. Four microplates were prepared with one as control (without microalgae) and two for mixotrophic cultures of 24 (M24) and 16 hours (M16) illumination, as well as one for heterotrophic condition in the absence of light (H). A light equipped incubator with controlled temperature at 25 °C and an irradiance of 40 μmol m⁻² s⁻¹ supplied by a cool-white fluorescent light was used to cultivate microalgae for a period of two weeks. The cultures were agitated at 150 RPM using rotary shaker (Heidolph Titramax 1000) for 5 minutes, daily. Cultivation using microplates was conducted in triplicates for subsequent experiments.

Taxonomic identification of the strains

Three microalgae strains were identified by optical microscopy and 18S rDNA sequencing. Total genomic DNA was extracted according to Kadkhodaei et al. (2011)²¹ and DNA quality and quantity were assessed through spectrophotometry and agarose gel electrophoresis. For molecular identification of the strains, four sets of primer pairs described by Gonzalez-Ballester et al. (2011),²² Olmos et al. (2009)²³ and Wan et al. (2011)²⁴ were selected to amplify 18s ribosomal RNA. All PCR amplifications were performed in a final volume of 50 μL containing 50 ng of DNA template, 1× buffer, 1.5 mM MgCl₂, 0.2 mM dNTP, 0.4 μM each primer, and 1 unit of Taq-polymerase (NEB, Ipswich, MA, USA). The temperature for primer annealing (T_a) was experimentally optimized using the gradient function of the thermocycler. The cycling program were performed with following conditions: 95 °C for 5 min, 35 cycles [94 °C for 30 s, T_a for 30 s, 72 °C for 90 s], followed by a final extension step of 72 °C for 10 min. The PCR products were visualize on 1.5% agarose (Amresco, Inc.) gel in TBE and staining with SYBR Gold™ (Invitrogen, Inc.) or GelRed™ (Biotium Inc., Hayward, CA). The fragments were then gel purified using Qiagen kit (Qiagen, Inc.). PCR products from each strain were sequenced with the primers used for amplification in an automatic DNA Genetic Analyzer (ABI Prism 3130 Genetic Analyzer) using the BigDye Terminator cycle sequencing kit v1.1 (Applied Biosystems). The resulting electropherograms were analyzed with CLC Main Workbench software v6.9.1 (CLCBio, Inc.) and subsequently aligned and phylogenetic dendrogram was created using Neighbor-Joining (NJ) algorithm. Bootstrap values based on the analysis of 1000 bootstrap replicates were calculated to assess the confidence degree assigned to the nodes in the phylogenetic trees. The identity of PCR products was confirmed through BLAST (Basic Local Alignment Search Tool) from the database of the NCBI. Similarly, the strains sequences herein obtained and those present in NCBI database stored in a local database to perform a local alignment by the tool present in CLC software.

Analytical procedures

Cell density was estimated by the mean number of cells which obtained from the direct cell counting. Three samples of each strain were used in cell counting using a hemocytometer and a light microscope (Nikon Eclipse TE200S). Maximum absorbance was inspected by scanning the culture sample from 400 to 700 nm with 10 nm increments (ESI Fig. 2†). The scanned data were normalized and plotted using Statistica software v.10 (StatSoft, Inc.). In order to determine the algal density, the highest absorbance value was used. The relationship between cell number and spectrophotometer absorbance were expressed using a general power equation (eqn (1))²⁵ (ESI Table 2 and ESI Fig. 3†):

Y = a(X)^b (1)

where; Y: absorbance, X: cell density (cell per mL), and, a and b: calibration coefficients.

The growth rates (day⁻¹), from the beginning to the end of the experiment as well as other growth parameters such as divisions per day and generation time were calculated according to Stein (1979).²⁶

Growth rate; K′ = Ln(N₂/N₁)/(t₂ − t₁) (2)

where; N₁ and N₂ are biomass at time1 (t₁) and time2 (t₂), respectively

Division per day; Div. day⁻¹ = K′/Ln2 (3)

Generation time; Gen' t = 1/Div. day⁻¹ (4)

Glucose uptake was determined using the YSI bioanalyzer (Illinoise, USA) and expressed as the eqn (5).

= ΔS_g/S_g (5)

where; ΔS_g: is the total weight of glucose consumption from the beginning to the end of experiment, S_g: is the weight of glucose supplemented to the medium at the beginning of cultivation.

Biomass (X) measured as dry cell weight was estimated using filtration and oven drying according to the method suggested by Zhu and Lee (2002).²⁷ Biomass yield was determined by the equation:

Y = (X_t − X₀)/(S₀ − S_t) (6)

where; X_t: is the final biomass concentration (g L⁻¹); X₀: is the initial biomass concentration (g L⁻¹); and S₀ and S_t: are the initial and final concentrations of the carbon source (g L⁻¹), respectively.

Total nitrogen content was estimated by the micro-Kjeldahl method^28,29 and then converted to protein using conversion factor of 5.95 which was suggested by López et al. (2010)³⁰ for microalgae and cyanobacteria undergoing rapid growth. To determine the protein content, three replicate samples of known volume and cell number were centrifuged in 3000 g and 4 °C for 15 min. The pellet was resuspended in 10 mL of sodium phosphate buffer containing 1% (w/v) sodium dodecyl sulphate (SDS) and the solution was sonicated on ice water (3 cycles) and then centrifuged to remove the pellet. The protein content was normalized to the cell numbers. Since the Kjeldahl nitrogen value has been shown to have better correlation with Lowry protein measurement for microalgae,³⁰ this indirect protein content measurement was chosen in opposite to a direct protein measurement.

Experimental design

Based on prior mixotrophic cultivations with modified Johnson medium, carbon source (glucose) and salinity (sodium chloride), which were the main factors affecting the growth performance of D. salina strains were selected as numeric experimental factors and photoperiod (periodic illumination times) and types of strains (D₁, D₂ and D₃) were selected as categorical experimental factors. The initial ranges were selected according to preliminary studies and previous works. A central composite design (CCD) with a full-factorial design consisting of a two-factor-five-level pattern with 13 design points (nine combinations with five replicates on the center point) using response surface methodology (RSM) was applied in this study. The design matrix of CCDs used is presented in Table 2.

Table 1 Growth performance for three strains of D. salina in M24^a

Strain	μmax	Divmax/day	Min td (h)	Nmax (10^6 cell per mL)	Dmax
a μmax = maximum growth rate, Nmax = maximum cell density, Dmax = the day with maximum Nmax, Min td = minimum generation time. μmax, Nmax, Dmax and Min td was calculated with the average values.
D1	0.25	0.37	65.24	2.27	12
D2	0.52	0.75	32.11	1.67	12
D3	0.4	0.58	41.09	0.50	13

---

Table 2 Growth parameters calculated for each treatment in three strains of D. salina in M24^a

Run	Glucose concentration (X1) (g L^−1)	NaCl concentration (X2) (M)	Growth rate		Protein production (g/100 g)		Glucose consumption (g L^−1)
			Observed	Predicted	Observed	Predicted
a ND*: not detected, ±: standard deviation of triplicate data.
D1
1	−1(5)	−1(1)	0.54 ± 0.04	0.54	0.27 ± 0.08	0.27	0.4 ± 0.07
2	1(15)	−1(1)	0.22 ± 0.07	0.23	ND*	ND*	2.4 ± 0.12
3	−1(5)	1(2)	0.64 ± 0.07	0.65	0.79 ± 0.08	0.79	1.52 ± 0.09
4	1(15)	1(2)	0.43 ± 0.09	0.41	0.71 ± 0.03	0.71	1.56 ± 0.08
5	−2(0)	0(1.5)	0.66 ± 0.05	0.68	ND*	ND*	0 ± 0.00
6	2(20)	0(1.5)	0.62 ± 0.07	0.61	0.83 ± 0.09	0.82	1.66 ± 0.08
7	0(10)	−2(0.5)	0.63 ± 0.07	0.66	0.27 ± 0.07	0.27	1.39 ± 0.15
8	0(10)	2(2.5)	0.60 ± 0.07	0.55	0.56 ± 0.05	0.56	1.89 ± 0.16
9	0(10)	0(1.5)	0.66 ± 0.08	0.63	0.89 ± 0.02	0.89	1.50 ± 0.10
10	0(10)	0(1.5)	0.54 ± 0.06	0.63	0.85 ± 0.09	0.89	1.55 ± 0.05
11	0(10)	0(1.5)	0.61 ± 0.03	0.63	0.89 ± 0.02	0.89	1.53 ± 0.04
12	0(10)	0(1.5)	0.65 ± 0.05	0.63	0.90 ± 0.07	0.89	1.52 ± 0.09
13	0(10)	0(1.5)	0.69 ± 0.03	0.63	0.91 ± 0.07	0.89	1.54 ± 0.08
D2
1	−1(5)	−1(1)	1.31 ± 0.09	1.14	0.39 ± 0.01	0.39	1.00 ± 0.15
2	1(15)	−1(1)	1.24 ± 0.08	1.15	ND*	ND*	4.17 ± 0.29
3	−1(5)	1(2)	0.78 ± 0.04	0.76	1.49 ± 0.08	1.49	1.76 ± 0.09
4	1(15)	1(2)	0.47 ± 0.09	0.53	1.53 ± 0.08	1.53	1.97 ± 0.15
5	−2(0)	0(1.5)	0.91 ± 0.07	0.97	0.90 ± 0.09	0.9	0 ± 0.00
6	2(20)	0(1.5)	0.76 ± 0.06	0.75	1.26 ± 0.09	1.26	3.96 ± 0.40
7	0(10)	−2(0.5)	1.38 ± 0.09	1.48	0.43 ± 0.07	0.42	4.54 ± 0.32
8	0(10)	2(2.5)	0.53 ± 0.04	0.48	ND*	ND*	3.61 ± 0.21
9	0(10)	0(1.5)	0.92 ± 0.03	0.87	0.89 ± 0.07	0.83	2.67 ± 0.10
10	0(10)	0(1.5)	0.72 ± 0.04	0.87	0.82 ± 0.09	0.83	2.35 ± 0.09
11	0(10)	0(1.5)	0.86 ± 0.06	0.87	0.78 ± 0.03	0.83	2.54 ± 0.10
12	0(10)	0(1.5)	0.83 ± 0.08	0.87	0.83 ± 0.06	0.83	2.61 ± 0.08
13	0(10)	0(1.5)	0.91 ± 0.07	0.87	0.83 ± 0.03	0.83	3.00 ± 0.10
D3
1	−1(5)	−1(1)	0.23 ± 0.06	0.22	0.49 ± 0.04	0.56	4.14 ± 0.55
2	1(15)	−1(1)	0.18 ± 0.07	0.19	0.50 ± 0.09	0.55	11.6 ± 0.87
3	−1(5)	1(2)	0.20 ± 0.01	0.2	0.80 ± 0.08	0.81	3.73 ± 0.27
4	1(15)	1(2)	0.17 ± 0.03	0.18	0.75 ± 0.06	0.73	10.92 ± 0.97
5	−2(0)	0(1.5)	0.27 ± 0.08	0.27	1.36 ± 0.08	1.33	0 ± 0.00
6	2(20)	0(1.5)	0.23 ± 0.03	0.22	0.10 ± 0.02	0.05	8.48 ± 0.73
7	0(10)	−2(0.5)	0.17 ± 0.06	0.17	0.46 ± 0.04	0.48	6.6 ± 0.71
8	0(10)	2(2.5)	0.15 ± 0.05	0.14	0.53 ± 0.02	0.56	6.62 ± 0.58
9	0(10)	0(1.5)	0.18 ± 0.08	0.2	0.51 ± 0.09	0.54	7.3 ± 0.63
10	0(10)	0(1.5)	0.22 ± 0.06	0.2	0.59 ± 0.06	0.54	6.33 ± 0.66
11	0(10)	0(1.5)	0.21 ± 0.03	0.2	0.57 ± 0.09	0.54	7.17 ± 0.70
12	0(10)	0(1.5)	0.20 ± 0.03	0.2	0.57 ± 0.04	0.54	5.67 ± 0.71
13	0(10)	0(1.5)	0.21 ± 0.04	0.2	0.49 ± 0.09	0.54	4.14 ± 0.69

---

A polynomial model to predict the response and optimal levels were calculated using eqn (7).

Y = b₀ + b₁X₁ + b₂X₂ + b₁₁X₁² + b₂₂X₂² + b₁₂X₁X₂ (7)

where; Y: is the response variable, b₀: is interception coefficient, b₁ and b₂: coefficients of the linear effects, b₁₁ and b₂₂: coefficients of quadratic effects, b₁₂: coefficients of interaction effects for two variables and X₁ and X₂: two different independent variables (X₁: glucose concentration, X₂: NaCl concentration).

The obtained experimental values were fitted by different polynomial equations which were provided in Design Expert software. Statistical analysis of data and plots were constructed using Design of Expert software version 7 (Stat Ease, Minneapolis, MN, USA), Sigmaplot 12.3 (Systat software, Inc.) and Statistica 10 (Statsoft, Inc). The analysis of variance (ANOVA) was employed to determine the significance of the model parameters in RSM. R² and adjusted R² values were calculated to evaluate the performance of the regression model.

The R² and adjusted R² were calculated using eqn (8) and (9).

where; X: is predicted value of protein production, Y_i,exp: is the observed value of protein production and ȳ_i: is the average value of observed protein production.

The adjusted R² was calculated using eqn (9),

where; N: is the total number of observations and K: is the number of input variables.

Results and discussion

Molecular identification

The axenicity of cultures was confirmed by PCR amplification of bacterial 16S rDNA (Fig. 1). On the other hand, the 18S rDNA sequences obtained for the strains demonstrated similar results, reflecting the high conservation of the molecular markers used as shown in Fig. 1. All PCR amplified products demonstrated a size between 1600 and 1750 bp. The phylogenetic tree obtained by Neighbor-Joining (NJ) determined the genetic distance and relationship among the studied strains and those already deposited in NCBI GenBank which all belonged to Dunaliella species. According to the genetic distances, strain D₃ is far from the other two strains D₁ and D₂ which are more close to each other falling in the same subcluster (Fig. 2). Molecular identification of the microalgae strains revealed their relatedness to Dunaliella salina. The contaminations affect the growth parameters as well as interfere with protein content analysis of the microalgae cultures. Therefore, the axenicity tests were performed as the prerequisite to validate the purity of the samples for further experiments. Phylogenetic analysis also confirmed the identification of species used in the study.

Fig. 1 Molecular identification (Lanes 1–4) and axenicity assessment (Lanes 5–7) of Dunaliella strains used in the study by 18S and 16S rDNA PCR amplifications, respectively. Lane M: molecular size marker (100 bp DNA Ladder PLUS, 100–5000 bp; OZ Biosciences, France), Lane 1 and 5: D₁, Lane 2 and 6: D₂, Lane 3 and 7: D₃, Lane 4: negative control (without DNA).

Fig. 2 Neighbor-Joining phylogenetic tree showing the relationships among the studied Dunaliella strains performed from the alignment of 18s amplicons and those reference sequences deposited to NCBI GenBank. The phylogenetic tree was constructed using the Neighbour-Joining method (CLC Main Workbench 6). The scale bar represents the number of substitutions per nucleotide position, showing high similarity between the Dunaliella samples and GenBank sequences.

Growth profile of D₁, D₂ and D₃

All three strains exhibited classical sigmoidal growth patterns comprising an initial lag phase with slow growth, log phase with exponential growth, and final stationary phase (Fig. 3). The growth of all three strains increased and reached stationary phase on day 12. The growth rate of strain D₃ was demonstrated to be less than D₁ and D₂, respectively. In addition, the maximum cell density (N_max) of strains D₁ (2.27 × 10⁶ cell per mL) and D₂ (1.67 × 10⁶ cell per mL) was higher than D₃ (0.50 × 10⁶ cell per mL) which showed a significantly lower growth rate at the end of day 14. The μ_max and t_d for D₂ is 2.1 and 1.3 fold higher than D₁ and D₃, respectively (Table 1, ESI Fig. 1†).

Fig. 3 Growth curves for three strains of D. salina based on cell density in M24. The simulated and experimental data represented by solid lines and symbols, respectively.

Optimization of culture conditions using RSM

Table 3 provides ANOVA and statistical parameters for significance and intercorrelations among the factors for growth responses. In D₁, significant differences existed for both salt and glucose concentrations. In this case, A (glucose), B (salt) and B² are significant model terms. The model F-value of 37.23 implies that the model is significant with insignificant lack of fit. The predicted R² of 0.7889 is in reasonable agreement with the adjusted R² of 0.9379. The model also showed adequate precision by measurement of signal to noise ratio. The ratio greater than 4 is desirable. Thus, a ratio of 20.887 indicated an adequate signal.³¹

Table 3 Analysis of variance (ANOVA) and coefficient estimate (CE) by regression model for the optimization of biomass by three strains of D. salina (D₁, D₂ and D₃) after 12 days of cultivation

Source	Coefficient estimate	Standard error	Sum of Square	Df	Mean square	F value	Prob > F
D1
Model			0.2	5	0.039	37.23	0.0001
Intercept	0.62	0.013
A-glucose	−0.026	9.35 × 10^−3	7.91 × 10^−3	1	7.91 × 10^−3	7.54	0.0286
B-NaCl	−0.11	9.35 × 10^−3	0.16	1	0.16	151.01	0.0001
AB	−0.037	0.016	5.53 × 10^−3	1	5.53 × 10^−3	5.27	0.0553
A^2	−8.16 × 10^−3	6.76 × 10^−3	1.52 × 10^−3	1	1.52 × 10^−3	1.45	0.267
B^2	−0.032	6.76 × 10^−3	0.023	1	0.023	22.32	0.0021
Residual			7.34 × 10^−3	7	1.05 × 10^−3
Cor total			0.2	12
D2
Model	0.87	0.046	0.82	5	0.16	13.23	0.0019
A-glucose	−0.055	0.032	0.036	1	0.036	2.89	0.133
B-NaCl	−0.25	0.032	0.75	1	0.75	60.46	0.0001
AB	−0.06	0.056	0.015	1	0.015	1.18	0.3138
A^2	−2.17 × 10^−3	0.023	1.08 × 10^−4	1	1.08 × 10^−4	8.72 × 10^−3	0.9282
B^2	0.028	0.023	0.017	1	0.017	1.41	0.2732
Residual			0.087	7	0.012
Cor total			0.9	12
D3
Model	0.2	6.24 × 10^−3	0.01	5	2.07 × 10^−3	9.17	0.0056
A-glucose	−0.013	4.33 × 10^−3	1.88 × 10^−3	1	1.88 × 10^−3	8.35	0.0233
B-NaCl	−6.58 × 10^−3	4.33 × 10^−3	5.20 × 10^−4	1	5.20 × 10^−4	2.31	0.1726
AB	2.92 × 10^−3	7.51 × 10^−3	3.40 × 10^−5	1	3.40 × 10^−5	0.15	0.7092
A^2	0.012	3.14 × 10^−3	3.13 × 10^−3	1	3.13 × 10^−3	13.89	0.0074
B^2	−0.011	3.14 × 10^−3	2.53 × 10^−3	1	2.53 × 10^−3	11.24	0.0122
Residual			1.58 × 10^−3	7	2.25 × 10^−4
Cor total			0.012	12

---

RSM models sum of squares demonstrated that the quadratic type is the highest order polynomial regression that highly suitable to explain the relationship between the input variables and responses. The reduced polynomial equations for the biomass and protein production by strains D₁, D₂ and D₃ were constructed in terms of coded values and empirical equations after replacement of coded values with actual values are given in Table 2. Regression coefficients for the related equations were generated by nonlinear estimation and analysed in each culture as shown in eqn (10)–(15).

Y₁: biomass

D₁: Y₁ = 0.62 − 0.026(A) − 0.11(B) − 0.037(A)(B) − 8.16 × 10⁻³(A²) − 0.032(B²) (10)

D₂: Y₁ = 0.87 − 0.055(A) − 0.25(B) − 0.06(A)(B) − 2.17 × 10⁻³(A²) + 0.028(B²) (11)

D₃: Y₁ = 0.2 − 0.013(A) − 6.58 × 10⁻³(B) + 2.92 × 10⁻³(A)(B) + 0.012(A²) − 0.011(B²) (12)

Y₂: protein production

D₁: Y₂ = 0.89 + 0.15(A) + 0.07(B) − 0.2(A)(B) − 0.092(A²) − 0.12(B²) (13)

D₂: Y₂ = 0.93 + 0.13(A) + 0.12(B) − 0.18(A)(B) + 0.07(A²) − 0.1(B²) (14)

D₃: Y₂ = 0.58 − 0.17(A) + 0.11(B) − 0.017(A)(B) + 0.074(A²) − 0.068(B²) (15)

where; Y₁ and Y₂ are biomass and protein production, respectively. A and B are the coded values of the independent variables, glucose and NaCl, in the media, respectively. Statistical parameters of the models developed through RSM for D. salina strains have been provided in ESI Table 1.†

In this study, CCD experiments revealed that high level of NaCl concentration (Run 8) has a negative effect on growth of D. salina, compared to the addition of a low NaCl concentration (Run 5) with glucose concentration in the middle point. As a comparison between runs 5 and 6, the response of specific growth rate was higher in run 5. The highest specific growth rate was observed in run 7 where the maximum specific growth rates (μ_max) were 0.49, 0.21 and 0.09 for strains D₁, D₂ and D₃, respectively. The principal parameters such as glucose (X₁) and NaCl (X₂) enhanced biomass when their concentrations increased from low to high. It was also observed that variations in the NaCl concentrations caused a greater influence in biomass production as compared to glucose. The response surface curves (Fig. 4 and 5) showed the interactions of variables on microalgal growth during cultivation period of 14 days. Optimum results in biomass concentration were obtained when glucose was added to the central point with minimum NaCl concentration. In the optimization procedure, maximum predicted biomass was obtained when the concentrations of the factors were adjusted to 11.23, 2.76 and 0.01 g L⁻¹ glucose and 0.55, 0.5 and 1.42 M NaCl, for D₁, D₂ and D₃, respectively (Table 4).

Fig. 4 Three dimension response surface and contour line plots for the impact of mixotrophic condition on D. salina cultures biomass (cell concentration). NaCl: M, glucose: g L⁻¹.

Fig. 5 The impact of glucose and NaCl on the maximum protein production by three D. salina strains showed as 3D response surface and contour line plots. Protein (P): g/100 g, NaCl: M, glucose: g L⁻¹.

Table 4 Predicted optimal levels for cell concentration and protein content by RSM predicted response on these levels

Cell concentration
Strain	Glucose (g L^−1)	NaCl (M)	Cell conc. (×10^6 mL^−1)
D1	11.23	0.55	0.731
D2	2.76	0.5	1.380
D3	0.01	1.42	0.273

---

Protein content
Strain	Glucose (g L^−1)	NaCl (M)	Protein (g/100 g)
D1	14.93	1.33	0.982
D2	20.00	0.96	1.607
D3	0.00	2.02	1.291

---

In strain D₃, maximum total protein content was observed in the media containing low glucose concentration (5 g L⁻¹). While, in two other strains (D₁ and D₂) this was followed by two more peaks at 15 g L⁻¹ in both 1 and 2 M NaCl treatments. The significant effect of glucose supplementation in different NaCl concentration on biomass and protein content is shown in Fig. 4 and 5. Glucose supplementation increased the chlorophyll content, cell density and protein content in all strains. Comparison of maximum and minimum values in each strain revealed an increase of 2.61, 2.92 and 1.98 folds in optical density of strains D₁, D₂ and D₃, respectively. This increase in cell density was 9.3, 9.38 and 6.87 folds, respectively. For media without glucose addition, the protein content demonstrated 3.7, 3.94 and 1.01 folds increase in D₁, D₂ and D₃, respectively. When protein content was monitored among zero glucose treatments, a substantial variation was also observed at different NaCl concentrations. D. salina strains D₁, D₂ and D₃ showed 0.68, 1.68 and 1.84 folds increase (P < 0.01), respectively, as NaCl concentration increased from 0.5 to 2.5 M. In the optimization procedure where the protein content is important in D. salina cultivation in media without glucose, NaCl concentration of 2.48, 2.22 and 2.02 M resulted in the best response for the strains D₁, D₂ and D₃, respectively.

Composition of media for the cultivation of microalgae is a critical variable in defining cell density and growth rate. Organic carbon source supplementation supports rapid growth with high final cell concentration. Held (2011)³³ suggested that medium constituents play a more important role in providing energy rather than light. Carbon content showed to be a significant constituent of D. salina mixotrophic growth media. D. salina cultures grown in medium free from organic carbon source grew at considerably slower rates and much lower final densities. As a comparison, the cells E1 (D₂-10-0.5), D₁₂ (D₂-15-1), E2 (D₂-5-1), F7 (D₂-0-2) and D₅ (D₂-0-1.5) can be visually distinguished among the others (Fig. 6).

Fig. 6 A comparison between three mixotrophic (M24 and M16) and heterotrophic (H) conditions on the day eighth of cultivation for green algae D. salina. The replicates were considered to be five for center points and three for the other points.

This study produced results which corroborate the findings of a great deal of the previous work in this field. The findings of the current study are consistent with those of Wan et al. (2011)¹¹ who found that biomass and lipid production decreased at the highest glucose concentrations, but the content of protein and lipid were significantly augmented for mixotrophic conditions at least for some species. For the mixotrophic cultures in D₁ and D₂ at high glucose levels, much of the glucose was not consumed and remained in the medium. This is in agreement with the findings on the cultivation of microalgae C. vulgaris³⁴ and C. sorokiniana.¹¹ High glucose concentration in run 6 caused osmotic shock or osmotic stress. This might change the solute concentration around the microalgae cells, causing a rapid change in the movement of water across the cell membrane. In media containing high concentrations of any solutes (salts, substrates, etc.), water is drawn out of the cells through osmosis. In addition, such condition inhibits the transport of substrates and cofactors into the cell resulting in cell shocking. Cheirsilp and Torpee (2012)³⁵ reported that the chlorophyll content decreased with increasing initial glucose concentration for marine microalgae strains of Nannochloropsis and Chlorella. This is largely due to the increase of heterotrophic metabolism of glucose at higher concentrations of glucose.

On the economic side, carbon source supply exhibited to be one of the main input costs among all costs related to microalgae (Spirulina) cultivations,³⁶ which holds true for Dunaliella cultures as well. Through optimization procedures in this study it was demonstrated that when there are limitations for the use of glucose (e.g. in economic side or contamination considerations), the level of glucose as low as 11.55 g L⁻¹ has equal influence to higher levels (17.74 g L⁻¹) in terms of desirability. Therefore, the growth of D. salina can be promoted with the lower level of glucose comparable with the higher concentrations resulting in a reasonable response for the further scale-up. In scaling up cultivations, production costs should be taken into account in economic side. However, the values of recombinant proteins will compensate these extra costs.⁸

Protein is known as a critical factor in algal cell division and chlorophyll as the main pigment in photosynthetic system.² Quantitation of total protein has been used as the means to normalize cellular reactions for several decades; based on the premise that on average, each cell has the same amount of protein.³² It has been demonstrated that the mixotrophic conditions (light and organic carbon supplementation) affect algal lipid, protein, carbohydrates and pigments biosynthesis differently. On the other hand, the level of protein content in microalgae (C. vulgaris) showed to be negatively correlated with carbohydrate and lipid content.^34,37 For example, increasing the initial glucose to 5 or 10 g L⁻¹ improved the lipid yield in C. sorokiniana and Chlorella, respectively.¹¹ In the same study an increase of protein content in D. salina reported under mixotrophic condition with a maximum at 15 g L⁻¹ of glucose, although it was minimal comparing the other studied microalgae species. This was confirmed in our study with a broader range of glucose combinations to various salt concentrations. In order to improve algal protein content through nutrient elements, nitrogen seems to be necessary for continuous protein biosynthesis which indirectly affects pigment formation as well. Our findings confirmed the former studies that supplementation of organic carbon source and energy (light and glucose) could convert the metabolic pathways in algal cells.³⁴ These results suggested that changes in the cellular biochemical composition were influenced by the trophic conditions and salt concentration in the medium.

In this study all experiments were done using microplate based system. Microplates not only offer the ability to measure many samples with very low volumes, but also the flexible array of wells density (6, 12, 24, 48, 96, 384, and 1536) which provide various options for the researcher to design the experiments based on the most appropriate sample number and reaction volume combination to fit their needs. In addition, automation can be easily performed for many repetitive and routine measurements since microplates have industry defined.³² Therefore, utilizing the cost-effective microplate-based approach more samples can be analyzed and screened with less need to lab facilities and incubation room, saving costs particularly in case of application of expensive chemicals.³⁸

The standard protocols to estimate algal density include direct cell counting, measurement of chlorophyll content and absorbance correlations. In spectrophotometric methods, different reading wavelengths of 750 nm,²⁵ 680 nm,³⁹ 600 nm (ref. 32) and 540 nm (ref. 13) have been suggested to monitor algal growth. These values are correlated to the light absorbance of chlorophyll. A peaks could be observed (680 nm), representing the wavelength of maximum sensitivity to quantify D. salina samples. Therefore, all further analyzed samples were read in this wavelength. However, Held (2011)³³ used 600 nm as the wavelength of choice to monitor growth in order to avoid influence from absorbing material.

As the other important factors on growth of microalgae D. salina which should be taken care particularly for production of recombinant proteins include pH, temperature and light intensity. The optimum pH for growth of D. salina was reported to be 7.5–8.0.¹⁰ Microalgae are sensitive to temperature changes, thus maintaining constant temperature is important for stable long-term cultivation. Kim et al. (2012)⁴⁰ revealed that the optimum temperature conditions for growth of Dunaliella strains was 27 °C. Light intensity is one of the most important factors in photoautotrophic conditions for photosynthesis in microalgae and it affects biomass productivity. The optimum light intensity for maximum productivity was 80 μE m⁻² s⁻¹ for Dunaliella strains. Light limitation will result in increasing pigment content of most species and shifts in fatty acid composition. In order to avoid photoinhibition, reduction in light intensity to some extent is preferable for many microalgae species. On the other hand, very low illumination might help green microalgae and cyanobacteria to survive in the vegetative state (not as cysts) for at least 6–12 months. However, the longer the stationary phase will result in longer lag phase when subculturing the cells in the fresh medium due to the shutdown of many biochemical pathways.

Conclusions

In summary, results from this study have demonstrated that D. salina can utilize glucose as the organic carbon substrate, and that the mixotrophic effects were in conjunction with that the mixture of NaCl and glucose for the production of biomass and protein. This alternative cultivation system could enhance the D. salina growth and biomass. It was observed that with an increase of glucose concentration from 0 to 15 g L⁻¹, biomass and protein content were increased. However, in the concentrations of glucose and NaCl higher than 15 g L⁻¹ and 2.5 M, respectively, these responses were dropped dramatically. Based on the optimization procedure, protein production in D. salina was markedly enhanced by glucose supplementation as low as 5 g L⁻¹. The optimal response obtained through nonlinear estimation was for 288, 288 and 312 h, and a doubling time of 1.13, 3.03 and 5.17 days for D₁, D₂ and D₃ strains, respectively. The runs with unsupplemented glucose presented the lowest doubling times. In terms of using this technique for recombinant protein production, glucose could be utilized to enhance both the growth rate and protein content. On the other hand, reducing the growth cycle expedites the analysis of positive transformants. Therefore, since the production level of recombinant proteins in an expression host is related to the total protein content, the authors conceptualized it as an application of the study to enhance the recombinant protein in response to cultivation parameters in the genetically modified microalgae. To be cost-effective, many samples could be screened using microplate based system for various applications such as optimization of the cultivation condition and downstream bioprocesses as well as screening of the transformants in molecular farming and recombinant protein production in microalgae. Also, the use of this technique may help in enhancement of biomass and protein content of microalgae to be used in human and animal food and feed.

Acknowledgements

We thank Prof. Dr Rasedee Abdullah, Prof. Dr Abdul Rahman Bin Omar, Abby Salleh and other members of LIVEs (laboratory of immunotherapeutic and vaccines). This work was supported by the following generous contributions: Grants from the UPM Institute of Biosciences to ABA, and UPM Graduate Research Fellowship to SK.

References

1. A. Hosseini Tafreshi and M. Shariati, J. Appl. Microbiol., 2009, 107, 14–35.
2. W. Huang, B. Dong, Z. Cai and S. Duan, Afr. J. Biotechnol., 2011, 10, 13164–13173.
3. E. W. Becker, Biotechnol. Adv., 2007, 25, 207–210.
4. O. Pulz and W. Gross, Appl. Microbiol. Biotechnol., 2004, 65, 635–648.
5. P. Spolaore, C. Joannis-Cassan, E. Duran and A. Isambert, J. Biosci. Bioeng., 2006, 101, 87–96.
6. T. L. Walker, S. Purton, D. K. Becker and C. Collet, Plant Cell Rep., 2005, 24, 629–641.
7. A. Barzegari, M. A. S. Hejazi, N. Hosseinzadeh, S. Eslami and E. Mehdizadeh Aghdam, Mol. Biol. Rep., 2010, 37, 3427–3430.
8. J. Costa, L. Colla and P. D. Filho, World J. Microbiol. Biotechnol., 2002, 18, 603–607.
9. A. Ben-Amotz and M. Avron, Plant Physiol., 1983, 72, 593–597.
10. A. Richmond and Q. Hu, Handbook of Microalgal Culture: Applied Phycology and Biotechnology, John Wiley & Sons, Ltd, Oxford, UK, 2013.
11. M. Wan, P. Liu, J. Xia, J. N. Rosenberg, G. A. Oyler, M. J. Betenbaugh, Z. Nie and G. Qiu, Appl. Microbiol. Biotechnol., 2011, 91, 835–844.
12. A. Richmond, Handbook of Microalgal Culture: Biotechnology and Applied Phycology, Blackwell Science Ltd, 2004.
13. C. A. Santos, M. E. Ferreira, T. L. da Silva, L. Gouveia, J. M. Novais and A. Reis, J. Ind. Microbiol. Biotechnol., 2011, 38, 909–917.
14. A. Delgado, D. Brito, C. Peres, F. Noé-Arroyo and A. Garrido-Fernández, Food Microbiol., 2005, 22, 521–528.
15. S. Abbasiliasi, J. S. Tan, T. A. T. Ibrahim, S. Kadkhodaei, H. S. Ng, F. Vakhshiteh, Z. Ajdari, S. Mustafa, T. C. Ling, R. A. Rahim and A. B. Ariff, Food Chem., 2014, 151, 93–100.
16. S.-H. Ho, C.-N. N. Chen, Y.-Y. Lai, W.-B. Lu and J.-S. Chang, Bioresour. Technol., 2014, 163, 128–135.
17. P. Kittisuban, P. Ritthiruangdej and M. Suphantharika, LWT--Food Sci. Technol., 2014, 57, 738–748.
18. M. Manikandan, L. Pasić and V. Kannan, Bioresour. Technol., 2009, 100, 3107–3112.
19. J. Liu, J. Appl. Phycol., 2013, 26, 163–171.
20. W. G. Weisburg, S. M. Barns, D. A. Pelletier and D. J. Lane, J. Bacteriol., 1991, 173, 697–703.
21. S. Kadkhodaei, M. K. Nekouei, M. Shahnazari, H. Etminani, A. Imani, M. Ghaderi-Zefrehei, M. Elahy and A. B. Ariff, Aust. J. Crop Sci., 2011, 5, 1199–1209.
22. D. Gonzalez-Ballester, W. Pootakham, F. Mus, W. Yang, C. Catalanotti, L. Magneschi, A. de Montaigu, J. J. Higuera, M. Prior, A. Galván, E. Fernandez and A. R. Grossman, Plant Methods, 2011, 7, 24.
23. J. Olmos, L. Ochoa, J. Paniagua-Michel and R. Contreras, Saline Syst., 2009, 5, 5.
24. M. Wan, J. N. Rosenberg, J. Faruq, M. J. Betenbaugh and J. Xia, Biotechnol. Lett., 2011, 33, 1615–1619.
25. L. H. R. Rodrigues, A. Arenzon, M. T. Raya-rodriguez and N. F. Fontoura, J. Environ. Chem. Ecotoxicol., 2011, 3, 225–228.
26. J. R. Stein, Handbook of Phycological Methods: Culture Methods and Growth Measurements, Cambridge University Press, 1979.
27. M. Zhu, P. Zhou and L. Yu, Bioresour. Technol., 2002, 84, 510–516.
28. E. J. Havilah, D. M. Wallis, R. Morris and J. A. Woolnough, Lab. Pract., 1977, 26, 545–547.
29. J. Persson, M. Wennerholm, and S. O'Halloran, Handbook for Kjeldahl digestion, FOSS, Hilleroed, 4th edn, 2008.
30. C. V. G. López, M. D. C. C. García, F. G. A. Fernández, C. S. Bustos, Y. Chisti and J. M. F. Sevilla, Bioresour. Technol., 2010, 101, 7587–7591.
31. H. R. F. Masoumi, A. Kassim, M. Basri and D. K. Abdullah, Molecules, 2011, 16, 4672–4680.
32. P. Held, Biofuel Research, 2011, 1–5.
33. P. Held and K. Raymond, Biofuel Research, 2011, 1–5.
34. W. Kong, H. Yang, Y. Cao, H. Song and S. Hua, Food Technol. Biotechnol., 2013, 51, 62–69.
35. A. Eisentraeger, W. Dott, J. Klein and S. Hahn, Ecotoxicol. Environ. Saf., 2003, 54, 346–354.
36. R. Rojíčková and B. Maršálek, Chemosphere, 1999, 6535, 3329–3338.
37. B. Cheirsilp and S. Torpee, Bioresour. Technol., 2012, 110, 510–516.
38. Y. Liang, N. Sarkany and Y. Cui, Biotechnol. Lett., 2009, 31, 1043–1049.
39. S. Mandal and N. Mallick, Appl. Microbiol. Biotechnol., 2009, 84, 281–291.
40. W. Kim, J. M. Park, G. H. Gim, S.-H. Jeong, C. M. Kang, D.-J. Kim and S. W. Kim, Bioprocess Biosyst. Eng., 2012, 35, 19–27.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c5ra04546k

---