Tetraselmis sp. strain M8 (Chlamydomonadaceae), was isolated from a rock pool at the Sunshine Coast (26°39′39′′S, 153°6′18′′E) and was cultured in the Algae Biotechnology Laboratory at the University of Queensland (Lim et al., 2012). It was grown in f/2 medium (Guillard and Ryther, 1962) in artificial sea water and then scaled up to a 1000 L outdoor open raceway pond equipped with an airlift mechanism for mixing. Microalgae were harvested by centrifugation after 5 days of nitrogen starvation to provide a high-lipid content biomass. The biomass for the initial baseline experiments was provided by a 10 L laboratory-grown culture (using clear 20 polyethylene bags and air bubbling under 16:8 h day: night illumination); while for the direct and indirect transesterification experiments, the outdoor raceway culture was used.

Five millimolar stock solution of SYTOX Green in DMSO (Invitrogen, Ltd., UK) was used for staining of damaged cells. A total of 0.5 μL of this solution was used for staining 0.5 mL of cell suspensions followed by 5 min incubation at room temperature in the dark. The observation of the stained cells was carried out through utilizing a fluorescence microscope (BX61, Olympus, Japan).

Nile red (Sigma-Aldrich) was prepared by dissolving 10 mg of Nile red powder in 1 mL acetone and stored in the dark at 4°C as a 31.41 mM stock solution. TAG observation was conducted using an Olympus BX60 fluorescent microscope with an Olympus U-RFL-T burner.

An Avanti centrifuge (HP-20 XPI) (Beckman Coulter) was used for centrifugation of volumes higher than 1 L and a Techno Spin R centrifuge (Sorvall Instruments) was used for small amounts (<100 mL). Lyophilization was carried out by using a DYNAVAC Freeze drying unit model FDI. For turning the dried biomass to fine powder, milling was accomplished by a Retsch NM 301 mixer mill machine.

The biomass preparation was decided as per the extraction experiments on both lyophilized dry biomass and wet concentrated algal paste. The best solvent composition for Soxhlet extraction was selected based on the results of the Soxhlet extraction through utilizing hexane or a solvent mixture of hexane–ethanol (3:1 ratio). The pre-treatments were carried on Tetraselmis sp. M8 followed by direct and indirect transesterification as summarized below. The procedures of the experiments are described in Figure 1.

Tetraselmis sp. M8 slurry from the same batch was used for comparisons within an experiment.

Lipid extraction from microalgae was carried out following either the procedure by Bligh and Dyer (1959) or by using a Soxhlet (Foss Soxtec 1043). For lipid extraction according to Bligh & Dyer, 1 g of lyophilized Tetraselmis sp. M8 or the equivalent DW of paste were placed in 40 mL glass vials with Polytetrafluoroethylene (PTFE) lined caps. Five milliliters chloroform and 10 mL methanol were then added and the mixture was vortexed for 5 min. Then an additional 5 mL chloroform was added and vortexing was conducted for another 5 min followed by addition of distilled water (the amount of distilled water addition was based as per the moisture content of the biomass to result in a ratio of 2:2:1.8 of chloroform/methanol/distilled water as detailed by Bligh and Dyer). Then centrifugation at 1000 × g for 7 min was conducted to form phase separation, and subsequently the organic layer (chloroform with extracted lipids) was transferred to pre-weighted glass tubes and the chloroform was evaporated using a vacuum desiccator.

Soxhlet extraction with hexane or hexane–ethanol was carried out using a Foss Soxtec 1043 equipped with six extraction chambers and valves for gathering and recycling the distilled solvent. Solvents used were hexane, methanol, chloroform, and ethanol; all were HPLC grade sourced from Merck KGaA. Each of the six cellulose thimbles was loaded with about 0.7 g of lyophilized Tetraselmis sp. M8 biomass, which were placed in pre-weighted aluminum cups previously filled with either 52 mL hexane or a hexane–ethanol (3:1; v/v) solvent mixture. The extraction (solvent circulation) was performed for 6 h followed by 30 min rinsing and 30 min evaporation. Weighing the samples and gravimetric measuring of the lipid extractions were carried out using a digital Mettler AM50 scale with 1 mg accuracy.

For direct transesterification, 4 mg of lyophilized algal pellet was hydrolyzed and methyl-esterified with 300 μL of 2% H₂SO₄ in methanol solution at 80°C by shaking (480 rpm) for 2 h on a thermal-mixer. Prior to the esterification, 50 μg of heneicosanoic acid (C21) was added to the pellet in each sample as an internal standard. After esterification, 300 μL of 0.9% (w/v) NaCl and 300 μL of HPLC grade hexane were added and vortexed for 20 s. Phase separation was performed by centrifugation at 16,000 × g for 3 min and the hexane layer was used for FAME profile analysis by GC-MS.

Indirect transesterification was conducted by applying two extraction techniques (Bligh & Dyer or Soxhlet) followed by transesterification of the recovered triglycerides for GC-MS. Bligh & Dyer extraction was performed on 1 g lyophilized Tetraselmis sp. M8 biomass. The Soxhlet extraction technique was conducted by utilizing a solvent mixture of hexane–ethanol with a ratio of 3:1 on 0.7 g lyophilized Tetraselmis sp. M8. Both extraction techniques were carried out as mentioned for the baseline experiments.

The extracted lipids were re-dissolved in 2 mL chloroform and a 100 μL aliquot was taken and dried down. Then the acid catalyzed derivatization was performed (similarly to what has been described in detail for the direct transesterification). Using 1 μL on a splitless injection system, GC-MS analyses were carried out on an Agilent 6890 GC coupled to a 5975 MSD. A DB-Wax column (Agilent, 122-7032) was used with running conditions as described in Agilent’s RTL DB-Wax method (Application note: 5988-5871EN). Identification of FAMEs was based on mass spectral profiles and retention times in the Agilent’s RTL DB-Wax method.

To quantify lipids potentially released to the medium after pre-treatments of algal biomass, 500 mL of slurry from each of the pre-treatments (including the untreated control), were placed in 500 mL Borosilicate glass volumetric flasks and were kept at 4°C overnight so that algal biomass completely settled and cooled down, allowing the accumulation of any released floating lipids on top of the supernatant. Then from each flask, 50 mL of the very top layer of the supernatant was transferred as 10 mL portions to glass vials for subsequent Bligh & Dyer extraction. Then, the settled slurries were shaken again to form homogenous slurries and from each, 50 mL were centrifuged at a high speed of 4000 × g for 15 min and the supernatant was transferred in 10 mL portions to glass vials for subsequent Bligh & Dyer extraction. A modified Bligh & Dyer extraction was conducted by adding 5 mL chloroform followed by 5 min vortexing, then 11 mL methanol were added followed by 1 min vortexing and lastly 6 mL chloroform were added and a final step of 5 min vortexing was carried out. The rate of solvents was selected in a way to form the ratio of 2:2:1.8 of chloroform: methanol: water in the end as has been suggested by Bligh and Dyer (1959). Then from the readily separated phases, the bottom layer of chloroform with any dissolved lipids was transferred to pre-weighted soda glasses for gravimetric analyses. All extraction experiments were carried out in triplicates.

The extracted lipid and FAMEs were compared according to different samples (algal paste and lyophilized biomass), different solvent systems, various pre-treatment systems, or different extraction methods using t test, one-way, and two-way ANOVA methods. The level of significant difference was at P < 0.05. Statistical analyses were conducted through GraphPad Prism 6.

Lipid extraction is currently considered a bottleneck in commercial algae cultivation for biodiesel production. One of the first questions is whether drying of algal biomass is required for efficient lipid extraction or whether extraction from wet biomass is possible. A comparison of the lipid extraction efficiency using the Bligh & Dyer method showed that algal paste (with a moisture content of 80.8%) was still suitable for lipid extraction, but yielded 32% less lipids than lyophilized dry biomass of Tetraselmis sp. M8 (Figure 2). The total lipid recovery from lyophilized biomass was 12.35 ± 0.24% of the DW, whereas, the same extraction method on paste of the same biomass showed a lower extraction efficiency (8.36 ± 0.49% of DW; P < 0.05). Fatty acid profiling by GC-MS showed that for almost all of the individual fatty acids, there was a considerable difference between the paste and lyophilized biomass.

Interestingly, although the amounts of fatty acids were different (lyophilized biomass showed considerable higher values compared to algal paste), they both resulted in similar fatty acid profiles. Nonetheless, the FAME recovery from the total extracted lipid of the lyophilized algal biomass was 77.31 ± 2.94%, while the FAME recovery from total extracted lipid from algal paste was only 65.59 ± 5.69% (P > 0.05) (Figure 2). The percentage of total saturated fatty acids from the total extracted lipid of the two samples were almost the same (around 16%), but lyophilized biomass showed higher amounts of total un-saturated fatty acids per biomass DW in comparison with algal paste (7.51 ± 0.231%, 4.116 ± 0.271, respectively; P < 0.05) (Figure 2). Individual FAME profiles are shown in Table 1.

Hexane is a common solvent for larger scale lipid extraction from biomass that is often used for vegetable oil extraction. However, Chen et al. (2012) tested different ratios of hexane to ethanol and reported that a mixture of hexane:ethanol with the ratio of (3:1) gave the best results. To test whether a solvent mixture may also lead to increased lipid extraction efficiencies for Tetraselmis sp. M8, a Soxhlet extraction was used. Lipid extraction using hexane led to 7.74 ± 0.68% extraction of total lipids from lyophilized algal biomass while there was a much higher lipid recovery for the Soxhlet utilizing a mixture of hexane–ethanol with ratio of 3:1 (14.31 ± 0.25%; P < 0.05; Figure 3). Interestingly, FAME profiling by GC-MS showed that not only the total lipid recovery increased by applying the hexane–ethanol (3:1) solvent mixture, but it also contained a higher percentage of fatty acids. The total fatty acids extraction rate by the solvent mixture as per the biomass dried weight was significantly higher than when the plain hexane extraction was used (7.372 ± 0.372% and 3.949 ± 0.238%, respectively; P < 0.05) (Figure 3). On the other hand, the C16 content comprised about 21% of the total fatty acids when using the solvent mixture, whereas, when plain hexane was used total fatty acids just included 14.6% of C16 (Table 2). Interestingly, from the total extracted lipid, the percentages of the total fatty acids were near identical for the two solvent categories (51.54 ± 3.25 and 51.12 ± 1.71% for hexane–ethanol (3:1) and plain hexane, respectively) (Figure 3). The proportion of total un-saturated FAMEs compared to total FAMEs was higher in the plain hexane Soxhlet extraction compared to when the solvent mixture was used (85 and 78.7%, respectively; P < 0.05) (Figure 3). Individual FAME profiles are shown in Table 2.

Cell walls and cellular structures can potentially hinder solvent penetration leading to suboptimal lipid extraction efficiencies. To test this, we evaluated the effect of different cell-disrupting pre-treatments on the efficiency of total lipids/FAMEs recovery of lyophilized algal biomass through either direct or indirect transesterification techniques. Sixty liters of nitrogen-starved Tetraselmis sp. M8 slurry were harvested from an open raceway pond at the same time to provide enough biomass grown under the same conditions for all subsequent experiments to make a direct comparison of the results possible. As has been shown by baseline experiments, lyophilized Tetraselmis sp. M8 biomass showed better oil yield compared to wet algal paste (Figure 2). Hence, for all subsequent experiments, lyophilized biomass was used. Moreover, as the Soxhlet extraction with a solvent mixture using hexane–ethanol (3:1) showed more promising results than the Soxhlet extraction using plain hexane (Figure 3), the solvent mixture was used. Figure 1 shows an overview of the process of these experiments. Pre-treatments of algal biomass were chosen based on their potential scalability and ability to disrupt cells. Microscopic analyses showed that thermal and microwave pre-treatments led to effective cell disruption and were hence further investigated for their potential to improve lipid extraction efficiency. Figure 4 shows the total lipid recovery of lyophilized Tetraselmis sp. M8 biomass by applying indirect transesterification through Soxhlet or Bligh & Dyer extraction methods using different pre-treatments. The Bligh & Dyer method showed better oil recoveries than the Soxhlet method in all of the experiments. Considering just Bligh & Dyer extractions, the total lipid recovery percentage of the untreated M8 control samples was 18.46 ± 0.43%, but significantly higher (P < 0.05) lipid yields were gained from the microwave or thermal pre-treated biomass (23.36 ± 0.23 and 22.29 ± 0.23%, respectively).

Going through the Soxhlet results, the same order of extraction efficiency can be found, with the microwave and thermal pre-treatments giving the best oil yields of 20.26 ± 0.08 and 20.04 ± 0.41%, respectively, while the untreated control sample showed only minimal amount of 14.38 ± 0.93% lipid recovery.

Fatty acid profiling by GC-MS was performed for both extraction methods by indirect transesterification or direct transesterification for each of the pre-treatment systems; the obtained data are shown in Table 3. All of the methods resulted in similar fatty acid profiles, with C16, C16:4, and C18:3 contributing to more than half of the FAMEs in total.

Comparisons of fatty acids recovery for the different pre-treatment systems and extraction methods are shown in Figure 5. The untreated control samples of both direct and indirect transesterification experiments showed similar total FAME contents per DW. It is interesting to note that although Soxhlet extraction showed lower amounts of total lipid recovery than the Bligh & Dyer-based extraction, it was significantly more efficient for the total yield of FAMEs for samples that underwent thermal pre-treatment (P < 0.05). For both indirect transesterification procedures (following lipid extraction by Bligh & Dyer or Soxhlet), thermal and microwave pre-treatment gave significantly higher FAME yields. The highest yield of un-saturated fatty acids per DW was obtained when using the Soxhlet extraction after thermal or microwave pre-treatment (Table 3).

Within the indirect transesterification experiments, considering the total FAME recovery from the total extracted lipid, the best result was obtained by Soxhlet extraction of both thermal and microwave-pre-treated biomass with 86.06 ± 6.29 and 74.25 ± 11.1% recovery rates, respectively.

The main reason for applying the pre-treatments was to facilitate the extraction techniques by weakening microalgal cell walls so that the solvents can more readily permeate the cells to subsequently dissolve and extract the lipids. Hence, there was a concern that whether due to the pre-treatments, microalgal cells had collapsed and some of the intracellular lipids were released to the medium (seawater) and therefore were removed from the extraction system during centrifugation, which could result in under-estimation of the lipid recovery. However, microscopic analyses, including SYTOX Green staining of pre-treated algal slurries showed intact cells with no visible lipid bodies in the medium (Figure 6).

Microscopic visualizing the Nile Red stained samples of the supernatants of the pre-treatment systems under the UV light showed nothing more than a plain black background. To test larger volumes of the medium surrounding pre-treated algal cells, 50 mL of the supernatant from settled algal biomass was extracted via Bligh & Dyer. The results confirmed that there was no detectable lipid recovery from the supernatants of the pre-treatment systems. Hence, the aforementioned results can be considered as the precise total lipid extractions without significant losses.