Production and characterization of a new glycolipid, mannosylerythritol lipid, from waste cooking oil biotransformation by Pseudozyma aphidis ZJUDM34

Abstract Mannosylerythritol lipids (MELs) are glycolipids possessing unique biosurfactant properties. However, the prices of substrates currently used for MEL formation caused its unsustainable commercial development. Waste cooking oil poses significant ecological and economical problems. Thus, the production of MELs from used waste cooking oil using the biotransformation route is one of the better alternatives to utilize it efficiently and economically. This work aims at the production of MELs using waste cooking oil instead of soybean oil and evaluating the major characteristics and compositions of MELs. The titers reached 61.50 g/L by the optimization of culture medium, higher than the counterpart (10.25 ± 0.32 g/L) of the nonoptimized medium. MELs exhibited good surface activity and better performance in contrast to MELs grown on soybean oil. The water phase behavior of MEL‐A was also evaluated. The process showed higher productivity of MELs with better surface activity and application stability than the conventional process using soybean oil. The findings of this study imply that the use of inexpensive fermentation substrates associated with straightforward downstream processing is expected to have a great impact on the economy of MEL production.

. In addition, almost all vegetable oils (except palm oil and coconut oil) have been found to serve as a good carbon source for MELs by different Pseudozyma spp. However, the sustainability of MEL production from the above-mentioned materials is doubtful due to high cost (Faria et al., 2014). Consequently, a positive strategy to reduce the production cost is to search for cheap substrates. So far, biosurfactant that has been shown could be produced using some renewable industrial residues such as olive oil mill effluent (Mercadé et al., 1993), oil refinery wastes (Bednarski, Adamczak, Tomasik, & Plaszczyk, 2004), distillery and whey wastes, potato process effluent (Dubey & Juwarkar, 2001), soapstock (Benincasa, Contiero, Manresa, & Moraes, 2002), and waste cooking oil (Raza, Khan, Khalid, & Rehman, 2006). Waste cooking oil is generated in large quantities during food preparation in both household and food processing industries (Hingu, Gogate, & Rathod, 2010;Zhang, Wang, & Mortimer, 2012). It has been considered a problematic waste product contributing to the environmental pollution. Currently, waste cooking oils are only allowed to produce biofuels. In general, the disposal of waste cooking oil is a growing problem that needs effective solution.
Biosurfactant production using the technique of microorganism transformation may be an effective way to utilize waste cooking oil.
There were some studies on glycolipid production using waste cooking oil. Waste frying oil and the methyl ester of coco/palm oil were used to produce sophorolipids by Candida bombicola in shaking flasks (Fleurackers, 2006). Haba, Espuny, Busquets, and Manresa (2000) reported the reuse of olive and sunflower cooking oil as substrate to produce rhamnolipids at equivalent productivity of 6.75-9.25 g/L (measured as 2.7 g rhamnose/L) in shaking flasks. Waste frying oil as the sole carbon source was used to synthesize rhamnolipids by Pseudomonas aeruginosa ZJU at a concentration of 12.47 g/L, and its mutant after treatment by UV light increased this productivity to 24.61 g/L cultured in shaking flasks. Future, the productivity of fermentation which conducted in a 50-L bioreactor reached over 20 g/L (Zhu et al., 2007).
There is no report on the MEL production from waste cooking oil, while Włodzimierz Bednarski, Adamczak, and Nawotka (2006) studied on the synthesis of MELs from glucose and products of enzymatically hydrolyzed lactose in milk permeate (after ultrafiltration) supplemented with the waste fats (fish, pork, and postrefining fatty acids), indicated that the fatty acids from the waste fats used seemed to be directly incorporated in the mannosylerythritol lipids.
To obtain higher MEL yield, the fermentation conditions have been widely studied, showing that the type and concentrations of nitrogen and inorganic salts, the amount of vegetable oil, inoculum size, and initial pH are potentially important impact factors (Luo et al., 2013). In this work, waste cooking oil was used as the sole carbon source to produce MELs by Pseudozyma aphidis ZJUDM34. The aim of this work was to use response surface methodology (RSM) to optimize the amount of waste cooking oil, inoculum size, initial pH, and medium volume for MEL production. Furthermore, to demonstrate the surface activity and stability, MELs from waste cooking oil were compared to that from soybean oil via the analyses of gas chromatography-mass spectrometry (GC-MS) and surface tension examinations.

| Fermentation strain
The experimental strain was P. aphidis DSM 70725, which obtained from Deutsche Sammlungvon von Mikroorganismen und Zellkulturen GmbH (DSMZ), Braunschweig, Germany. The P. aphidis ZJUDM34 mutant was evolved from wild-type yeast using UV F I G U R E 1 General structure of mannosylerythritol lipids (MELs). MEL-A: R1/R2=acetyl-; MEL-B: R1=acetyl-, R2=H; MEL-C: R1=H, R2=acetyl-; MEL-D: R1/R2=H mutagenesis and low-energy N + implantation methods. It exhibited enhanced ability to produce MEL-A. This mutant was stored in our laboratory. Stock cultures were grown for 2 days at 28°C in a liquid medium containing (g/L) yeast extract 3.0, malt extract 3.0, peptone from soybeans 5.0, and glucose 10.0 at initial pH and then mixed with an equal volume of 50% glycerol. They were stored at −80°C and renewed every half-year.
Waste cooking oil was collected from major restaurant in Hangzhou with soybean oil in the preparation of cooked food. The crude waste oil was preliminarily prepared using the process as shown in Figure 2, prior to use as the carbon source in replace of soybean oil.

| Inoculum preparation
At first, the yeast was activated by inoculating 1-ml stock cultures into 50-ml YM medium composed of (g/L): yeast extract 3.0, malt extract 3.0, peptone from soybeans 5.0, and glucose 10.0 at the natural pH in 250-ml Erlenmeyer flasks, and was then incubated for 36 hr at 28°C and 180 rpm. Seed culture was prepared by inoculating 1 ml of the activated cells in 250-ml Erlenmeyer flasks containing 50-ml culture medium (g/L) (NaNO 3 3, KH 2 PO 4 0.3, MgSO 4 ·7H 2 O 0.3, yeast extract 1, and glucose 40) and incubated for 2 days at 28°C and 180 rpm. Then, the inoculum cells were prepared by the following procedure: Seed was centrifuged to obtain the cells, which were washed two times with 0.9% physiologic saline, dissolved in an appropriate amount of 0.9% physiologic saline to formulate into the inoculum with the amount of 0.12 g/ml cell wet weight.

| Fermentation medium and conditions
The liquid submerged fermentation (g/L) medium was comprised of NaNO 3 3.0, KH 2 PO 4 0.3, MgSO 4 ·7H 2 O 0.3, yeast extract 1.0, and certain amount of waste cooking oil (ml/L) or soybean oil (ml/L) depending on experimental design. Fermentation was inoculated with the above inoculum and conducted in 250-ml Erlenmeyer flasks for 10 days at 28°C and 180 rpm.

| Experimental design
Response surface methodology is a statistical method to find the optimal process parameters to solve multivariable problems with multiple regression analysis for the quantitative data obtained from properly designed experiments. In order to obtain optimum conditions for increasing MELs' yield, a central composite design with five coded levels was performed. In the experimental design, four factors (waste cooking oil amount, inoculum size, medium volume in 250-ml Erlenmeyer flasks, and initial pH) that having impact on the production of MELs were identified by the optimization strategies.
All the experiments were performed in triplicate. The test factors were coded according to the following equation (Zhang, Dong, Fan, Jiao, & Chen, 2015): where x i is the coded value of the ith independent variable, X i is the natural value of the ith independent variable, X 0 is the natural value of the ith independent variable at the center point, and ΔX i is the step change value. The range and the levels of the test factors with both coded values and natural values investigated in this study are given in Table 1. The production of MELs was considered as the response.
The quadratic model for predicting the optimal point was expressed based on the following equation: where Y was the response variable, b 0 , b i , b ii , and b ij were the regression coefficients variables, for intercept, linear, quadratic, and interaction coefficients, respectively, and x i and x j were independent variables. Results were analyzed using the response surface regression (RSREG) procedure (SAS Institute Inc, Cary, NC, USA).
Three-dimensional surface plots were drawn to illustrate the main and interactive effects of the independent variables on MEL production. Solving the regression equation and analyzing the response surface contour plots, the optimum values of the selected variables were obtained for the further experiments (Ghorbani, Karimi, Biria, Kariminia, & Jeihanipour, 2015).
To make clear the fermentation process, according to the optimal results of 2.3.1 experiments, design experiments to conduct the assay of biomass, pH, and MEL yield at different culture times.

| Determination of biomass, isolation, and purification of MELs
After fermentation, pH was measured with pH meter. Then, thoroughly mixed culture suspension with equal volume of ethyl acetate was extracted three times. Aqueous and organic phases were separated by

| Thin-layer chromatography
The purified MEL extracts were analyzed by thin-layer chromatography (TLC). As a solvent system, a chloroform-methanol-water (65:15:2) mixture was used. Location of the components on the plates was carried out by fumigating with iodine or spraying the plates with a solution of 0.1% orcinol in a 5% sulfuric acid solution and heating the plates for 5 min at 110°C.

| Methyl ester of samples
In this study, the methyl ester derivatives were prepared by the followed steps. The purified MEL fraction or oil sample (2.0 g) was mixed with 20 ml methanol and 0.5 ml KOH (1 M in methanol) in round bottom flask, then added a few zeolite, and begun to heat reflux with shaking the beaker, until the solution was clear. Afterward, the solution was cooled in cold water, transferred to a separating funnel, washed the flask with 10 ml n-heptane and transferred to the separating funnel, added to 20 ml distilled water, shook and still layered; the upper is ester layer, and the lower is aqueous layer. The lower layer was treated with 10 ml n-heptane extraction and mixed with the upper. Subsequently, the n-heptane solution was washed several times with distilled water until the water was neutral; acetate layer was obtained, dried with anhydrous sodium sulfate, filtered, and evaporated to 10 ml n-heptane solution for analysis.

| Surface-activity determination
Equilibrium surface (water/air) tension measurements of MEL solution were performed at 25°C with a QBZY-1 tensiometer (Fangrui Instrument Co., Shanghai, China) using the Wilhelmy plate method (Benincasa et al., 2002). To increase the accuracy of the surface tension measurements, an average of triplicates was determined. The platinum plate and all glassware used were cleaned in ultrapure water.

| Statistical analysis
The regression analysis of the obtained experimental data was per-

| Optimization under RSM-based experimental design
Previous investigations showed that waste cooking oil, inoculum size, culture medium volume, and initial pH all affected the production of MELs by P. aphidis ZJUDM34. Thus, to further improve the fermentation conditions for MEL production, RSM was employed to determine the optimal levels of four factors (waste cooking oil amount, inoculum size, medium volume in 250-ml Erlenmeyer flasks, and initial pH) that affected MEL production ( Table 2). As a result, the following regression equation was obtained: With Y, MEL production (the response); x 1 , waste cooking oil amount; x 2 , inoculum size; x 3 , medium volume in 250-ml Erlenmeyer flasks; and x 4 , initial pH (coded values). The significance of each coefficient was determined by p-values, which are listed in Table 3.
The regression equation indicated the R 2 value of 0.9780. This value ensured a satisfactory adjustment of the quadratic model to the experimental data and indicated that the model could explain 97.80% of the variability in the response. As can be seen from Table 3, the p-value and lack of fit check indicated that the model was significant for MEL production. As well, the production was The predicted maximum production of MELs is 61.51 g/L, which is much higher than the yield from the control fermentation (10.25 ± 0.32 g/L).

| Experimental validation of the optimized culture conditions
The predicted optimal fermentation conditions for MEL production composed of waste cooking oil amount 95 ml/L, 24.95 inoculum size (ml/L), 48.65-ml medium volume per 250 ml, and initial pH 5.76. The response surface plots are presented in Figure 3. In order to validate the adequacy of this optimized medium formula for MEL production, verification experiments were carried out at the predicted optimal soybean oil and waste cooking oil. The mean output of MELs from waste cooking oil was 55.00 ± 0.80, which was little lower than the predicted value (61.51 g/L) and slightly lower than 61.00 ± 0.80 g/L from soybean oil after 10 days (Figure 4a). In comparison, waste cooking oil showed lower MEL production and biomass than those of soybean oil (Figure 4b), whereas there was no obvious change of pH during the fermentation process (Figure 4c). These results suggested that the model obtained from RSM was adequate and waste cooking oil was a suitable carbon source for MEL production by P. aphidis ZJUDM34.

| Fatty acid profiles in substrates and MELs
Furthermore, by the use of TLC, MELs contain type-A, type-B, and type-C grown on soybean oil and type-A, type-B, type-C, and type-D from waste cooking oil ( Figure 5(1)). Soybean oil and waste cooking oil as the substrates were composed of a variety of fatty acids and likely influenced the properties of MELs; as a result, GC-MS was used for detecting fatty acid contents of substrate and MEL product. To compare the fatty acid profiles of soybean oil, waste cooking oil and a mixture of MEL derived from soybean oil or waste cooking oil are presented in Table 4 and Figure 5(2).

Soybean oil
Waste cooking

| Critical micelle concentration
The surface (water/air) tensions of crude MEL from different substrates were comparatively measured. As shown in

| The stability of MELs to pH, temperature, and sodium chloride solution
To evaluate the pH effect on stability of MELs from different substrates, the surface tension of several MEL samples prepared with a concentration of 20.00 mg/ml at different pH values (2.00-12.00) was determined. As can be seen from Figure 6b, the surface tension values were between 30.00 and 35.00 mN/m as pH less than 10.00. However, if pH is higher than 10.00, the surface tension rapidly rises up to about 60.00 mN/m, which indicated that MELs from soybean oil or waste cooking oil were unsuitable at high alkali environmental condition. As presented in Figure 6c, both MELs produced from soybean oil and waste cooking oil remain the unaltered activity after incubation for 2 hr at −20°C, 0°C, Further, the salt tolerance of MELs was explored, as the results presented in Figure 6d, MELs from soybean oil and waste cooking oil showed constant surface tensions and even a downward trend as the salt concentration increases, which indicated that both of MELs retained good surface activity at high salt environment.

| DISCUSS IONS
The aim of this work was to investigate the use of waste cooking oil for MEL production by P. aphidis ZJUDM34. The technical feasibility of the transforming waste cooking oil into MELs was demonstrated.
The novel fermentation strategy for MEL production was developed by optimization using RSM. The novel MELs were further characterized through the comparative analyses of its surface property to that produced from soybean oil. In term of the fermentation capability and the major properties, it is obvious that the replacement of soybean oil with waste cooking oil (WCO) is feasible for MEL production. The choice of cheap raw materials is vital to overall economics of the fermentation process since they account for nearly 50% of final production cost (Makkar & Cameotra, 1999). Moreover, in contrast to waste cooking oil, the average price of soybean oil was up to about three times (Yaakob, Mohammad, Alherbawi, Alam, & Sopian, 2013). The finding of this study showed the complete replacement of fermentation substrate with waste oil for MEL production is feasible and economic.
In consideration of the reuse of waste used oil, the industrial significance is rather high. Presently, it was reported that world production of oils and fats is above 200 million tonnes, 81% of which are derived from plants (Smith, 2004). Most of the oils and fats are used in the food industry and other consumed factories, which generates great quantities of waste oils. Naturally, the disposal of waste cooking oil is a growing problem, which explains the increasing interest in the use of waste oils for microbial transformation. Fleurackers (2006) reported the use of waste frying oils for the production of sophorolipids at 50 g/L medium by Candida bombicola, which was comparable to the yield of more commonly used carbon sources such as oleic acid. After then, rhamnolipids were successfully produced by Pseudomonas aeruginosa grown on various waste frying oils, which indicated soybean waste frying oil was the best substrate, produced 9.3 g/L under batch-fed cultivation with the addition of rhamnolipid precursor (Raza et al., 2006). In this work, we found that P. aphidis ZJUDM34 can transform waste cooking oil into MELs, and the yield of MELs is close to that grown on soybean oil substrate at the optimized culture conditions and higher than the yields of sophorolipids and rhamnolipids.
Generally, the fatty acid compositions in MEL structure determine the characterization of MELs in actual application. The total fatty acid profile of purified MELs from WCO showed that fatty acid groups mainly composed of C18:n, followed by C16:0 (Table 4). Such profile was different from that found in MELs grown on soybean oil, preferentially C8:0 and C10:n, as also previously reported (Rau, Nguyen, Schulz et al., 2005). Meanwhile, the CMC values of MEL produced from soybean oil and WCO were both 20.00 mg/L. It implies that the use of WCO substrate represents the same characteristics as soybean oil in considering the surface property. Naturally, the replacement of major fermentation substrate with waste cooking oil remains the same chemical bioactivity.
In summary, this work shows that waste cooking oil can be used as an alternative substrate for MEL production from P. aphidis ZJUDM34, which helps to lower the production cost of MELs, facilitating its production and application in the commercial use.

ACK N OWLED G M ENTS
This study was financially supported by Public Projects of Zhejiang Province (LGF18C200003) and Nature Science Foundation of Zhejiang Province (LR13C200002), China.

E TH I C A L S TATEM ENT
This study does not involve any human or animal testing.

CO N FLI C T O F I NTE R E S T
The authors declare that they do not have any conflict of interest.