In this study, comparison of different lipid extraction methods from three green microalgal strains, Scenedesmus sp. ME02, H. tetrachotoma ME03 and Micractinium sp. ME05 were evaluated. All three strains were obtained from thermal water pools of Haymana, Ankara, Turkey and isolated in our laboratory [18] . For lipid extraction, 200 mL of microalgal cultures were grown in liquid Tris-Acetate-Phosphate (TAP, pH 6.8) or BG-11 [19] [20] media in one liter flasks with constant shaking at 24˚C ± 1˚C, 16 h light-8 h dark photoperiod with a light intensity of 54 µE∙m⁻²∙s⁻¹ until cultures reached stationary phase. All chemicals used in this study were technical grade and obtained from Sigma-Aldrich.

One part of extract was dissolved in forty parts of solvent in three different modified extraction methods: Bligh and Dyer, Folch and Soxhlet extraction method [11] -[15] . For all three methods, initially two flasks of 100 mL of the same microalgal suspension were separated and centrifuged at 3000 g for 10 min at 4^°C. Then, for dry weight calculation, one tube containing the pellet was taken and oven-dried. The second tube was used for lipid extraction according to the method of choice. All experiments were performed with three biological replicates (n = 3).

This method was originally described by Folch [11] . Pellet of 100 mL microalgal suspension was mixed with chloroform and methanol in a 2:1 v/v ratio and mixed thoroughly in a vortexer. Then, methanol and water were added to the mixture to a final ratio of 1:1:0.9 v/v for methanol:chloroform:water, respectively. The final ratio of pellet to methanol, chloroform and water mixture should be 1:40 w/v. The mixture was shaken for 10 min in a separating funnel and incubated until two distinct phases were visible. The lower phase containing the microal- gal lipids was separated and evaporated by a rotary evaporator (BUCHI Rotavapor R-200). Lipid content was quantified gravimetrically.

Bligh and Dyer method was carried out as described previously with some modifications [12] . Pellet of 100 mL microalgal suspension was mixed with methanol and chloroform in a 2:1 v/v ratio and mixed thoroughly in a vortexer. Then, chloroform and water were added to the mixture to a final ratio of 1:1:0.9 v/v for methanol: chloroform:water, respectively. The rest of the protocol is identical to that of Folch method as described above.

Soxhlet extraction method with n-hexane was performed with some modifications [14] [15] . One part of pellet was mixed with forty parts of n-hexane in the Soxhlet apparatus. The solvent was heated to reflux. Total oil of microalgae was dissolved in n-hexane. Pellet was left for 24 hours in Soxhlet apparatus and oil was taken from the distillation flask. Remaining n-hexane was evaporated by rotary evaporator (BUCHI Rotavapor R-200). Microalgae oil was weighed and lipid yield was calculated gravimetrically.

We carried out five different cell disruption methods, namely; homogenization assisted (H-A), microwave assisted (MW-A), ultrasonication assisted (US-A), bead assisted (B-A) and lyophilization assisted (L-A) and combinations of these together with the three solvent extraction procedures.

100 mL of microalgal suspension was homogenized in tubes on ice by a homogenizer (Heideoph Diax-900) with 20 s homogenization time and 5 s interval. We tried three different time points; 5, 10 or 20 min for the total time of homogenization.

100 mL of microalgal suspension was placed in a conventional microwave oven (Arcelik intellewave MD-599) and samples were disrupted at 900 W, 20 s microwave power and 5 s interval time for a total duration of 1, 5 or 10 min.

100 mL of microalgal suspension was extracted on ice via an ultrasonicator (Cole Palmer Ultrasonic Processor CPT) at 400 W, 20 s ultrasonication, 5 s interval time for a total duration of 5, 10 or 20 min.

100 mL of microalgal suspension was mixed with 2 g glass beads (0.4mm diameter in size) and shaken for 5, 10 and 20 min at 1000 rpm rotation speed.

Microalgal suspension culture was centrifuged at 3000 g for 10 min at 4˚C and pellet was freeze-dried (lyophilized) in a lyophilizator (Christ D-37520 Alpha 1-4 LD plus) for 24 hours. The lyophilized samples were weighed and 4 mL of solvent was added per 0.1 g lyophilized tissue in each of the three solvent extraction methods. For a double combination of L-A technique with other cell disruption methods, lyophilized samples were used in each subsequent assisted technique (i.e. MW-A, US-A and B-A) prior to solvent extraction.

In this study, we did a comparative analysis of three different solvent extraction methods with or without assisted techniques for lipid extraction from three thermo-resistant microalgae strains, namely; Scenedesmus sp. ME02, Hindakia tetrachotoma ME03, and Micractinium sp. ME05 isolated from thermal water flora of Central Anatolia [18] . Initially, three solvent extraction methods; Soxhlet method, Bligh and Dyer, and Folch methods were used with no additional assisted technique. Results are summarized in Figure 1. For Scenedesmus sp. ME02, percent lipid contents were as follows: 9.0%, 9.3% and 10.3% total lipid.gdryweight^-1 with Bligh and Dyer, Soxhlet and Folch methods, respectively. For H. tetrachotoma ME03, final lipid yield was 4.7% total lipid∙gdryweight⁻¹ with the Folch method and 7.0% with Bligh and Dyer and Soxhlet methods. Finally, Micrac- tinium sp. ME05 had the highest lipid content among three species with all three methods tested. Percent lipid content for Micractinium sp. ME05 was 13.7%, 13.3%, and 11.7% with Soxhlet, Bligh and Dyer and Folch methods, respectively.

Next, we employed five different cell disruption techniques for assisted extraction of lipids from microalgal cells. These techniques are as follows: Homogenization assisted (H-A), microwave assisted (MW-A), ultrasonic- cation (US-A), glass bead (B-A) and lyophilization assisted (L-A) methods.

For four of these five methods, namely; H-A, MW-A, US-A and B-A, we tested three different time durations of cell disruption (Table 1). For the H-A method, increasing the time of homogenization from 5 min to 10 min slightly improved the lipid yield for all three species and solvent extraction methods. A further increase in homogenization time to 20 min, on the other hand, did not lead to further improvement in yield. Therefore, 10 min was determined as the total time of homogenization.

^*indicates durations that were used as data points in figures 2-4.

For the MW-A method, three different time durations of 1, 5 and 10 minutes were tested. 5 and 10 minutes of cell disruption in the microwave oven led to slightly higher percent lipid yield in most cases compared to one minute of disruption (Table 1). Between 5 and 10 minutes, we chose the latter as the total duration of microwave time due to the consistency of final percent yields between three biological replicates reflected by low standard error.

Finally, 20 minutes of incubation time was determined as the ideal duration of disruption for both US-A and B-A techniques.

Once the duration of disruption for each technique was determined, we compared all five extraction assisted techniques as well as double combinations of lyophilization assisted (L-A) method with MW-A, US-A and B-A in terms of percent lipid yield (Figures 2-4). For Scenedesmus sp. ME02, US-A together with the Soxhlet method resulted in an oil recovery of 14%. L-A and US-A combined further increased the lipid yield to 15.7%

(Figure 2). For H. tetrachotoma ME03, US-A and L-A in combination with the Soxhlet method (n-hexane) were equally effective leading to an increase in lipid yield from 7.0% to 8.7%. For the same species, using the

Soxhlet method together with L-A and US-A techniques combined further increased the lipid yield to 9.3%. Finally for Micractinium sp., Soxhlet extraction using the lyophilized samples (i.e. L-A technique) yielded in a nearly two fold increase in total lipid content (22.7%) compared to Soxhlet extraction alone (13.7%). However, the highest increase in percent lipid yield was achieved with L-A and US-A combined (27.3%) (Figure 2).

Next, we compared the effectiveness of each assisted or a combination of assisted technique(s) together with the Bligh and Dyer solvent extraction method for all three species used in our study. The results were similar to those of Soxhlet method with minor differences (Figure 3). A combination of L-A and US-A resulted in the highest increase in percent lipid yields in Scenedesmus sp. ME02 (12.7%), H. tetrachotoma ME03 (9.3%) and Micractinium sp. ME05 (24%) compared to Bligh and Dyer method alone.

For the Folch method, although some improvement was achieved particularly with L-A and US-A combination in H. tetrachotoma (7.3%), increase in lipid yield was not nearly as high as either the Soxhlet or the Bligh and Dyer methods in Scenedesmus sp. (11.3%) (Figure 4).

Taken together, extraction assisted techniques significantly improve the percent lipid yield when combined with a solvent extraction method. Lyophilization and ultrasonication and a combination of both proved to be particularly effective for three green microalgae species used in this study.

We calculated lipid productivities of Scenedesmus sp. ME02, H. tetrachotoma ME03, and Micractinium sp. ME05 as a factor of gram per liter dry weight per day (Table 2). We previously determined and described growth characteristics of each three species [18] . Based on the results, Scenedesmus sp. ME02 had the lowest lipid productivity (0.006 g∙L⁻¹∙day⁻¹), whereas Micractinium sp. ME05 showed the highest lipid productivity rate (0.05 g∙L⁻¹∙day⁻¹). Lipid productivity of H. tetrachotoma ME03 was calculated to be 0.01 g∙L⁻¹∙day⁻¹.

The choice of an effective extraction method depends on both the type of microalgae used and the solvent of choice. Previous studies emphasized importance of selecting an optimal extraction method suitable for different microalgae with different cell sizes, cell wall structure, lipid and fatty acid characteristics [21] -[23] . Lee et al.

observed significant differences in lipid content from Botryococcus sp., Chlorella vulgaris and Scenedesmus sp. using different extraction methods [21] . All microalgal strains used in this study belong to different taxonomic groups (i.e. Scenedesmus, Hindakia and Micractinium genera). However, the extraction method of choice did not show a significant difference among any of the three species used (Figure 1).

An appropriate solvent extraction method may be selected on a number of criteria. For instance, in a comparative study, Guckert et al. found Soxhlet method with methylene chloride:methanol (2:1) to be the most selective method for extraction of neutral lipids among others [24] . However, Soxhlet method is not suitable for extraction of unsaturated lipids due to instability of these lipids at elevated temperatures during reflux. High temperature can also accelerate transesterification of lipids in the presence of methanol, changing their natural form. In this respect, Bligh and Dyer method with chloroform and methanol causes less artifacts, therefore; results obtained with this method are considerably more reliable and reproducible [24] . However, Bligh and Dyer method is practically not convenient for large scale extractions due to high amounts of toxic waste generated, which requires expensive recycling methods and compromises user safety [25] .

Hexane as an extraction solvent, on the other hand, is advantageous for large scale extractions due to being cheaper and less toxic compared to other solvents like chloroform and methanol [26] [27] . Additionally, it has no affinity for non-lipid contaminants during extraction. In this study, the solvent extraction method of choice alone did not differ significantly for any of the three species used.

In the present study, lipid extraction from dry material (i.e. lyophilized samples) proved to be more effective than using wet material. Among other benefits like easy handling and storage [28] , lyophilization may also assist in direct and rapid contact of samples with the extraction solvent by increasing the surface area [29] without altering the initial solvent to water ratio in contrast to using wet material.

Several studies tested the effectiveness of microwave-assisted disruption for lipid extraction from microalgae. In a previous study, Lee et al. found among other methods, microwave to be the most effective in oil extraction from Botryococcus sp., Chlorella vulgaris and Scenedesmus sp. [21] . In yet another study, Balasubramanian et al. showed a 30% increase in lipid yield using the microwave assisted extraction compared to control from Scenedesmus obliquus [30] .

In a recent study, Koberg et al. successfully employed a one-step biodiesel production method via microwave and ultrasonication directly from Nannochloropsis biomass [31] . In yet another study, Cui et al. used microwave irradiation for direct transesterification of Cryptococcus curvatus and effectively determined optimum conditions of biodiesel production [32] . In addition, Cheng et al. studied direct transesterification of wet Chlorella pyrenoidosa via microwave and compared it to two-step process of biodiesel production. They found that one step process displayed six-fold higher biomass than that of two-step process for wet Chlorella pyrenoidosa [33] . In our study, microwave assisted technique did not result in significant improvement of lipid yield. Increasing the microwave processing time to 20 minutes or higher and using a high-power industrial scale microwave oven could help in improving the results.

Shen et al. showed that various cell disruption techniques including bead-beating, French press, sonication and wet milling followed by solvent extraction had different effects on the lipid contents of Scenedesmus dimor- phous and Chlorella protothecoides [34] . For instance, in S. dimorphus, wet milling resulted in more than five fold increase in lipid yield compared to solvent extraction alone. In C. protothecoides, on the other hand, bead beating was the most effective technique in lipid recovery. In a previous study, Araujo et al. studied different lipid extraction methods and found the Bligh and Dyer method as suitable extraction method for Chlorella vulgaris [35] .

Lee et al. found bead-beating and microwave methods to be the most efficient methods for Botryococcus sp. compared to homogenization, French press and lyophilization while sonication was the least effective [26] . Cell wall structure, cell size, bead size and duration and speed of beating are important parameters affecting the efficiency of the bead-beating method. Smaller bead size can be tried for more effective extraction as 4 mm bead size used in our study may have been less effective for disruption of cells of approximately 2 - 3 µm in size [18] . Optimization of these parameters in future studies may improve the lipid yield.