Preliminary experiments demonstrated that the pressure, temperature and modifier concentration were the main factors affecting the total amount of microbial lipids in the scCO₂ extraction. The range of pressures, temperatures and modifier concentrations were defined by considering the deterioration of bacterial and archaeal cell with increasing temperature [21,22], and the alteration of cell structure with increasing pressure [19]. Therefore, the effects of pressure (10–30 MPa), temperature (60–100 °C) and modifier concentration (5–15%, v/v) on the total amounts of RQ, PLFA, and PLEL extracted were investigated. The central composite design matrix of the SFE method and total amounts of the microbial lipids that were extracted from anaerobically digested sludge are reported in Table 1. The independent and dependent variables were analyzed to obtain regression equations that could estimate the total amounts of lipid biomarkers extracted within a given range.

Table 2 shows the regressions coefficients that were obtained by fitting the experimental data to the second-order response models for the total amounts of RQ, PLFA, and PLEL extracted. The goodness of the fit of models was verified by the coefficient of determination (R²). In this experiment, the R² values for RQ, PLFA, and PLEL were 0.86, 0.90 and 0.88, respectively. The experimental data were in agreement with the predicted values for the total amounts of RQ, PLFA, and PLEL extracted. The values of adjusted R² ranged from 0.76 to 0.83, suggesting that the total variation in the total amounts of RQ, PLFA, and PLEL was attributed to the independent variables. In addition, the portion of total variation that was not explained by the models ranged from 0.17% to 0.24%.

Analysis of variance (ANOVA) was applied to evaluate the significance of different model equations associated with models parameters. A summary of the ANOVA for each microbial lipid biomarker from anaerobically digested sludge is shown in Table 3. The ANOVA of the quadratic regression model from the experimental data of microbial lipid biomarkers demonstrated that the model was highly significant with Fisher’s F-test values ranged from 8.84 to 12.63 (p-value < 0.01). The lack of fit F-value ranged from 1.05 to 2.43, and implied that the lack of fit was not significant relative to the pure error. A visualization of the interactions between two experimental factors with the third factor, which was held constant at its central level, is plotted by the three-dimensional (3-D) response surface in Figures 1–3.

Figure 1a–c shows the interactions and effects of pressure, temperature, and modifier (methanol) concentration on the total amount of RQ extracted. The total amount of RQ extracted increased with increasing modifier concentration. However, increasing the extraction temperature resulted in a decrease in the total amount of RQ extracted in the tested conditions. All independent variables significantly affected the total amount of RQ extracted, with temperature exhibiting a negative influence. The interaction between the temperature and modifier concentration had a significant effect on the total amount of RQ extracted (p-value < 0.01, Table 2). Pressure increased the total amount of RQ extracted up to approximately 22 MPa, with no significantly further improvement thereafter. When considering the variables individually, the extraction conditions that maximize the total amount of RQ extracted were 22 MPa, 60 °C and a 15% (v/v) modifier concentration.

Increasing temperature is known to reduce the solvent density and consequently decrease the total amount of RQ extracted at constant pressure. At the same time, higher temperatures promote the solubility of the solute and increase the yield by high mass transfer of solute in the matrix and/or from the matrix to the solvent [28]. Therefore, the increase in temperature could have either a positive or a negative effect based on the balance between CO₂ density and solute vapor pressure. An increase in pressure results in an increase in CO₂ density by increasing the solvating power of the supercritical fluid [20]. An increase in pressure from 10 to 20 MPa induced a steady increase in the total amount of RQ extracted in this study.

The effects of pressure, temperature and modifier concentration on the total amount of PLFA extracted from anaerobically digested sludge are shown in Figure 2a–c. All independent variables significantly affected the total amount of PLFA extracted (p-value < 0.01, Table 2). The complex interaction between the temperature and modifier concentration had a negative effect. At a fixed extraction pressure of 20 MPa, the total amount of PLFA increased with increasing modifier concentration and temperature. It reached a maximum value at approximately 80 °C and 13% (v/v) of modifier concentration with no further improvement thereafter.

In this study, the total amount of PLFA extracted was increased with increasing pressure from 10 to 25 MPa. Above this range of pressure, a decrease in the total amount of PLFA extracted with increasing pressure was observed. The volatility and polarity of the extracted compounds might be responsible for these results and we can conclude that the composition of the extract can be controlled by pressure. When considering the variables individually, the extraction conditions that maximize the total amount of PLFA extracted were 25 MPa, 80 °C and a 13% (v/v) modifier concentration. These results indicate that the extraction of PLFA requires a higher temperature compared to that of RQ.

The effects of pressure, temperature and modifier concentration on the total amount of PLEL extracted from anaerobically digested sludge are shown in Figure 3a–c. Although the total amount of RQ extracted decreased with temperature, the total amount of PLEL exhibited the opposite trend. The modifier concentration showed the most significant effect (p-value < 0.01, Table 2).

At a given modifier concentration, the total amount of PLEL extracted increased rapidly with increasing extraction temperature. The total amount of PLEL extracted increased with increasing modifier concentration and reached a maximum value at a given pressure. With further increase in modifier concentration, the total amount of PLEL extracted slightly decreased. The fluid density is highly sensitive to temperature, especially near the system’s critical pressure. Therefore, the total amounts of PLEL extracted changed over the range from 60 to 100 °C. A moderate increase in temperature leads to a large decrease in fluid density, with a consequent reduction in solubility [29]. However, an increase in temperature will also accelerate mass transfer and improve the total amount of extracted compounds. Extraction conditions that maximized the total amount of PLEL extracted were 27 MPa, 95 °C and a 15% (v/v) modifier concentration.

A multiple-response optimization simultaneously maximizes all of the applied responses as described elsewhere [30]. In this study, the multiple-response optimization was performed by maximizing all of the dependent variables (total amounts of RQ, PLFA and PLEL extracted) together. The optimized extraction conditions were a pressure of 23.6 MPa, temperature of 77.6 °C, and 10.6% (v/v) of modifier concentration. The total amounts of RQ, PLFA, and PLEL were calculated to be 21.27, 596.11, and 6.98 nmol/g-dry sludge, respectively, under these conditions.

The optimized extraction conditions from the previous experiments were applied to the scCO₂ extraction experiment. Verification experiments were performed five times under the optimized conditions (23.6 MPa, 77.6 °C, and 10.6% (v/v) methanol). The total amounts of RQ, PLFA, and PLEL obtained were 22.17, 604.61, and 6.78 nmol/g-dry sludge, respectively, with relative standard deviations (RSDs) of 9.02%, 11.04%, and 3.68%, respectively.

The ultra performance liquid chromatography (UPLC) and GC-MS chromatograms of the microbial lipid biomarkers extracted from the anaerobically digested sludge by SFE are shown in Figures 4 and 5. RQ were determined by UPLC, while PLFA and PLEL were simultaneously determined by GC-MS. Eight types of RQ, 30 types of PLFA, and 1 type of PLEL were identified in the samples, illustrating the sensitivity of SFE technique for the simultaneous extraction of both bacterial and archaeal lipids from anaerobically digested sludge. The same numbers and types of microbial lipid biomarkers were obtained using the organic solvent extraction method.

Figure 6 shows the mole concentrations of microbial RQ, PLFA, and PLEL extracted from anaerobically digested sludge by SFE and conventional organic solvent extraction. A mole percent basis was used for RQ, PLFA, and PLEL data. It indicated the SFE method did not alter the microbial community structure information. The mole concentrations of the individual biomarkers were nearly identical for the two methods. Microbial lipid extracts from the SFE method were in agreement with the conventional solvent extraction method. However, SFE method proved to be simple, fast, and with less consumption of organic solvents.

The nomenclature of the PLFA used in this study is designated by the total number of carbon atoms, the number of double bonds, and the position of the double bond closest to the methyl end (ω) of the molecule [31]. For example, the PLFA 18:2ω6c is 18 carbons long, with two double bonds positioned six carbons from the methyl end of PLFA, and exhibit a cis configuration. The prefixes i, a, and cy indicate iso branching, anteiso branching and cyclopropyl groups, respectively. For RQ nomenclature, ubiquinones and menaquinones with n isoprene units in their side chain were abbreviated as UQ-n and MK-n, respectively. For example, UQ-8 indicates a ubiquinone with 8 isoprene units and MK-8(H₂) denotes a menaquinone with 8 isoprene units and one double bond saturated with 2 hydrogen atoms [32]. The PLEL are classified into di- and bidiphytanylglycerol ether phospholipids. Di-O-phytanylglycerol ether or diether(DE)-lipid contains two 20 carbon saturated and tetra-O-dibiphytanylglycerol ether or tetraether(TE)-lipid contains two 40 carbon isoprenoid hydrocarbons in ether linkage to glycerols [33].

Microbial lipids were grouped into saturated fatty acids, gram-positive and gram-negative bacteria, actinobacteria, fungi, and methanogenic archaea according to the guidelines that have been described elsewhere [34–36]. Saturated fatty acid group consisted of 12:0, 14:0, 15:0, 16:0, 17:0, 18:0, 20:0 and 22:0 PLFA. Gram-positive bacteria group consisted of MK-6, MK-7, and MK-8 RQ, and i14:0, i15:0, a15:0, i16:0, i17:0 and a17:0 PLFA. Gram-negative bacteria consisted of 16:1ω7c, 16:1ω8c, 16:1ω9c, 18:1ω7c, 18:1ω8c and cy17:0 PLFA. Actinobacteria group consisted of MK-9, MK-10, MK-8(H₂), MK-9(H₂), and MK-9(H₄) RQ and 10Me16:0 PLFA. Fungi group consisted of 16:1ω5c, 18:1ω9c, 18:1ω9t, 18:2ω6c, 20:1ω9c, 20:1ω9t, 22:1ω7c, 22:1ω9c and 22:1ω9t PLFA. Methanogenic archaea group consisted of diether(DE)-lipid derived from PLEL.

All solvents were HPLC grade, and all other chemicals were analytical grade. Methanol, acetone, chloroform, ethanol and hexane were purchased from Wako Co., Japan. The fatty acid standard (nonadecanoic acid methyl ester), diether lipid standard (1,2-di-O-hexadecyl-rac-glycerol), bis(trimethylsilyl)trifluoroacetamide (BSTFA), and 3 N methanolic-HCl were purchased from Sigma-Aldrich Co., Japan. The ubiquinone (UQ-10) and menaquinone (MK-7) standards were purchased from Nacalai Tesque Co., Japan. Standard stock solutions were stored at 4 °C. Sep-Pak^® Plus Silica cartridges (Cat. no. WAT020520) were purchased from Waters Co., Japan.

The anaerobically digested sludge used in this study was obtained from an 8-L lab-scale semi-automatic continuously stirred bioreactor (38 °C) treating a mixture of cow manure with food waste at the Department of Environmental and Life Sciences, Toyohashi University of Technology, Japan. Anaerobically digested sludge is a semi-solid material that is released during anaerobic digestion as a complex mixture of microbial fragments and organic colloids [37]. Prior to SFE extraction, the digested sludge samples were dried in a vacuum-freeze dryer for 24 h and homogenized by crushing. Freeze-dried digested sludge samples were stored at −20 °C until analysis.

Samples were extracted according to the modified Bligh and Dyer method [4] using a mixture of chloroform, methanol and phosphate buffer (50 mM, pH 7.4) at a ratio of 1:2:0.8 (v/v/v). After overnight extraction, chloroform and milli-pure water were added to the extract in equal amounts to yield a two-phase system with a final volume of 1:1:0.9 (v/v/v). The supernatant (upper phase) was transferred with a pipette into a test tube. The solvent was removed from the test tube under constant nitrogen flow with a water bath at a temperature that was not higher than 37 °C. The total lipid extract was fractionated into neutral lipids, glycolipids, and phospholipids, with chloroform (10 mL), acetone (10 mL) and methanol (10 mL), on acidic silicic columns. The neutral lipid fraction was cleaned with 0.2 μm spin filter. The extracts containing RQ were resolved in 200 μL acetone, and quantified by UPLC [16]. The phospholipid fraction was derivatized to form PLFA as theirs fatty acid methyl ester and PLEL as theirs etherlipid derivatives. PLFA and PLEL were simultaneously quantified using GC-MS [33,38]. The organic solvent extraction protocol for this study took 3 days to complete.

All supercritical fluid extractions were performed using an in-house constructed supercritical fluid extraction system. This system is equipped with a high-pressure pump (SCF-201, Jasco Co., Japan), a backpressure regulator (880-81, Jasco Co., Japan) and an oven (GC A353, GL Sciences Inc., Japan). In total, 100 mg of freeze-dried digested sludge was placed in a 1-mL stainless steel extraction cell. One end of the extraction cell was connected to a high-pressure pump through a 2-m preheating coil (0.5 mm i.d. stainless steel tubing), which was placed in an oven. The liquid CO₂ was compressed with a cooler and then allowed to flow into the extraction vessel at a flow rate of 2.7 mL/min. The liquid CO₂ and modifier (methanol) were continuously mixed in-line. The modifier is a polar or non-polar miscible solvent, and it is used to modify the polarity and solvent strength of the SFE. All operations were performed in dynamic extraction mode for 15 min. The dynamic mode means the mixture was passed over the activated sludge sample inside the extraction vessel. During the extraction, extracted lipids were collected in a dark glass tube to prevent photodecomposition and then dried under a stream of nitrogen gas.

The total lipid extracted obtained from SFE was fractionated into neutral lipids, glycolipids and phospholipids with chloroform (5 mL), acetone (5 mL) and methanol (5 mL) on SepPak^® Plus silica cartridges. The neutral lipid fraction was collected in a 15-mL glass test tube and evaporated to dryness under a stream of nitrogen. The neutral lipid fraction containing ubiquinones and menaquinones were resolved in 100 μL acetone and analyzed using UPLC without further purification [15]. The fraction containing phospholipids was also collected in a 15-mL glass test tube and evaporated to dryness under a stream of nitrogen.

Prior to GC-MS analysis, the PLFA in phospholipid fraction were derivatized to produce their fatty acid methyl esters (FAMEs), as described elsewhere [38]. Phospholipid extracts were incubated at 80 °C in a heated block for 1 h after the addition of a methanolic HCl solution. The tubes were vortex-mixed every 10 min during the incubation and subsequently cooled for approximately 20 min. The phospholipid extracts were vortex-mixed for 15 s in 3 mL of hexane and centrifuged for 5 min at 2000 rpm. The hexane phase containing the FAMEs was evaporated to dryness under nitrogen. Diether lipids were converted to their corresponding O-trimethylsilylethers by treatment with bis(trimetylsilyl)trifluoroacetamide (BSTFA) [33]. An internal standard mixture of nonadecanoic acid methyl ester and 1,2-di-O-hexadecyl-rac-glycerol was added to the final phase containing FAMEs and etherlipids. The final phase was then transferred to an amber vial and stored at −20 °C.

Chromatographic separation of RQ was performed using the Waters Acquity UPLC™ system (Milford, MA, USA), including a binary solvent delivery manager, a sample manager, a column compartment and a photo-diode array detector (PDA-2996, Waters Co., USA), connected with Waters Empower Software. The analytical column was a Water Acquity UPLC™ BEH C₁₈ column (50 mm × 2.1 mm, 1.7 μm particle size). The mobile phase consisted of 100% methanol and pumped at a flow rate of 0.5 mL/min. Chromatography was performed at 35 ± 1 °C with a run time of 40 min. The autosampler temperature was set at 4.0 ± 1 °C, and the sample injection volume was 10 μL. The detector wavelengths were set at 275 nm and 270 nm for ubiquinones and menaquinones, respectively [21].

PLFA and PLEL concentrations were quantified on a Hewlett-Packard 6890 GC Series system with a 5973 N Mass Selective Detector (Wilmington, DE, USA). Standard GC analysis for the simultaneous determination of PLFA and PLEL has been described elsewhere [31]. In contrast, we used a cross-linked methyl silicone fused-silica HP-5MS capillary column (30 m × 0.25 mm i.d. × 0.25 μm film thickness), and a total run time of 138 min. Helium was used as the carrier gas at a flow rate of 0.9 mL/min. NIST 98 MS software library was used for PLFA and PLEL compounds identification. The initial oven temperature was 50 °C for 1 min and was increased at a rate of 30 °C/min (to reach 110 °C), 1 °C/min (to reach 220 °C), and 10 °C/min (to reach 320 °C). The final temperature was maintained for 15 min. One microliter of samples was injected splitless for 0.75 min at injector temperature of 320 °C. The temperature of the transfer line between GC and MS was 320 °C. Nonadecanoic acid methyl ester and 1,2-di-O-hexadecyl-rac-glycerol were used as internal standards.

The selection of the optimum combination of input variables can be solved by the development of mathematical model. In this work, the experimental design consisted of a response surface methodology with a central composite design (CCD) with three variables and five settings. Preliminary experiments indicated that the variables, such as pressure, temperature and methanol concentration, were the main factors that affected the total amount of lipid biomarkers extracted. Thus, these parameters were selected as the independent variables, and output data (total amounts of RQ, PLFA and PLEL) as the dependent variables. The selected experimental factors in coded and actual settings for the SFE are shown in Table 1. The upper limit of a factor was coded as +1.68 and lower limit as −1.68. The intermediate value was coded as 0. Factorial points were coded as +1 and −1. In the CCD test, 14 factorial points and 9 central points were employed to fit the full quadratic model. These runs and their combination sets are called experimental design matrix [39]. Nine replicates at the center of the design were used to estimate a pure error sum of the square. The experiments were conducted in a random order to maximize the effects of unexplained variability in the observed responses. Thus, 23 runs were carried out to estimate the effect of SFE parameters.

The following second-order polynomial equation was used:

This equation used the following variables: Y is the predicted response; k is the number of factor variables; β₀ is the model constant; β_j is the linear coefficient; β_jj is the quadratic coefficient; β_ij is the interaction coefficient; and x_i and x_j are the independent coded variables. Design Expert^® software (v8 trial, Stat-Ease Inc., Minneapolis, MN, USA) was used for the regression analysis and visualization of the response surface data. The fit of the regression model was checked by the adjusted coefficient of determination (R²). The statistical significance of the adjusted model was determined by the application of Fisher’s F-test. Analysis of variance (ANOVA) was performed to evaluate significant differences between independent variables. The adequacy of the model was determined by evaluating the pure error, the lack of fit, the coefficient of determination (p-value) and the Fisher test value (F-value).