Optimization of Ultrasonic-assisted Extraction of Fatty Acids in Seeds of Brucea Javanica (L.) Merr. from Different Sources and Simultaneous Analysis Using High-Performance Liquid Chromatography with Charged Aerosol Detection

Our research aimed to optimize the oil extraction process and determine the fatty acids in Brucea javanica (L.) Merr. seeds. The extraction technology was optimized using response surface methodology. A Box-Behnken design was employed to investigate the effects of three independent variables on an ultrasonic-assisted extraction technique, namely, sonication time (X1: 20–40 min), liquid–solid ratio (X2: 16:1 mL/g–24:1 mL/g), and ethanol concentration (X3: 90%–100%). The optimum conditions of sonication time, liquid–solid ratio, and ethanol concentration were 40 min, 24:1 mL/g, and 100%, respectively. The content of fatty acids and the oil yield were 14.64 mg/g and 16.87%, respectively, which match well with the predicted models. The optimum number of extraction times was eventually identified as two. A new rapid method for the qualitative and quantitative analysis of the fatty acids of B. javanica (L.) Merr. seed oil using HPLC with a charged aerosol detector was described. The fatty acid contents of 14 batches of B. javanica (L.) Merr. seed oil were determined, and the relevance and difference were analyzed by fingerprint analysis. The fingerprint has five common peaks, and the similarity was greater than 0.991. HPLC analysis represents a specialized and rational approach for the quality identification and comprehensive evaluation of B. javanica (L.) Merr. seed oils.


Introduction
Brucea javanica (L.) Merr. (Simaroubaceae plant), a traditional Chinese herb, is widely distributed in the southern provinces of China, such as Guangdong, Guangxi, Yunnan, and Hainan [1]. B. javanica (L.) Merr. seed oil (BJO) is an extract of the dried nucleoli of B. javanica (L.) Merr.. The primary component of BJO is fatty acid (FA), which includes oleic, linoleic, linolenic, palmitic, and stearic acids. Oleic and linoleic acids are the main active ingredients in BJO [2], which possesses various biological and pharmacological activities, and is commonly used to treat various cancers, such as lung, gastrointestinal, and liver cancers [3]. Its mechanisms of the tumor growth inhibition include inhibition of DNA synthesis, suppression of tumor multidrug resistance, and destruction of tumor cell membrane systems [4].
As is well known, the study of extraction methods is very important, and various techniques for the extraction of BJO have been developed, including supercritical fluid extraction, soakage extraction and ultrasonic-assisted extraction (UAE). It is reported that UAE can achieve better extractions of natural products [5,6]. The effect of ultrasonic waves is strong, as they can destroy the cells of plant, which is especially suitable for seeds [7,8]. UAE is seen as an ideal option for the edible oil industry because of improvements in efficiency and speed and because it can be performed at low operation temperatures which avoids thermal damage to the extracts and preserves the structural and molecular properties of bioactive compounds. Many factors affect the extraction efficiency of UAE. Some of these are ultrasonic power, extraction time, extraction temperature, and solvent to solid ratio [9].
The most commonly used methods for the analysis of FA include gas chromatography [15][16][17] and high-performance liquid chromatography (HPLC), typically connected with an ultraviolet (UV), evaporative light-scattering, diode array, and/or mass spectrometry detector [18][19][20]. Most FAs lack or have weak UV absorptions, and therefore, they must be derivatized before determination [21][22][23]. In 2003, a new alternative detector, the charged aerosol detector (CAD), was introduced [24]. CAD is a mass sensitive and universal detector suitable for the routine determination of many non-volatile or weak-volatile chemical species and other compounds that contain weak chromophores [25]. This detector can analyze triacylglycerols, lipids, and other substances [26][27][28] and it can also directly detect FA without derivatization [29,30].
The construction of a chromatographic fingerprint plays an important role in the quality control of complex herbal medicines [31]. This technique emphasizes the systemic characterization of sample compositions and focuses on identifying and assessing the stability of plants. The identity, stability, and consistency of Traditional Chinese Medicines (TCMs), as well as the identification of adulterants, can be determined by their chromatographic fingerprints [32].
The current study aimed to adequately extract BJO with RSM and establish a sensitive and selective HPLC-CAD method (for the simultaneous determination of five FAs and to develop a characteristic fingerprint to control the quality of the raw herb. This study focuses on the optimization of a complete set of extraction and analysis methods of BJO and the establishment of the chromatographic fingerprint of TCM to provide a reliable basis for the further study of B. javanica seeds.

Calibration Curves, Limits of Detection, and Quantification
The regression equations of the five FAs listed in Table 1 were linear with the correlation coefficients (R 2 ) and were between 0.9983 and 0.9999. The lowest LOD and LOQ were obtained with linoleic acid (0.368 µg/mL and 2.624 µg/mL, respectively), whereas the highest LOD and LOQ were obtained with stearic acid (1.958 µg/mL and 6.800 µg/mL, respectively). The chromatograms of mixed standard solution and sample solution are shown in Figure 1.    Table 2 shows the precision. Relative standard deviation (RSD) was used to express the precision. The RSD values of intra-day and inter-day precision for the peak area were lower than 3%, which indicates that this method is stable.   Table 2 shows the precision. Relative standard deviation (RSD) was used to express the precision. The RSD values of intra-day and inter-day precision for the peak area were lower than 3%, which indicates that this method is stable. Table 2. Intra-day and inter-day precision of standard solutions.

RSD (%) RSD (%)
Linolenic acid The results of the recovery experiment (Table 3) showed that the overall average recoveries were 94.88% to 105.00%, which indicated that the current method is robust and suitable for the determination of FAs in BJO.

Effect of Independent Variables on the Content of FAs and the Oil Yield
The content of FAs and the oil yield of BJO affected by different sonication time (20-60 min) are presented in Figure 2A, where two other factors, liquid-solid ratio and ethanol concentration, were fixed at 12:1 mL/g and 95%, respectively. The content of FAs and the oil yield of BJO increased during the initial 40 min and then slowed down until they reached an equilibrium. The content of FAs and the oil yield of BJO affected by the different liquid-solid ratio (8:1 mL/g-24:1 mL/g) are seen in Figure 2B, where two other factors, sonication time and ethanol concentration, were fixed at 30 min and 95%, respectively. The liquid-solid ratio significantly affected the content of FAs and the oil yield. The content of FAs and the oil yield of BJO increased rapidly with the liquid-solid ratio increased from 8:1 mL/g to 24:1 mL/g, and reached the maximum value at 24:1 mL/g. The content of FAs and the oil yield of BJO affected by different ethanol concentration (80-100%) are shown in Figure 2C, where two other factors, sonication time and liquid-solid ratio, were fixed at 30 min and 12:1 mL/g, respectively. The content of FAs and the oil yield of BJO were little when ethanol concentration was lower than 90%. However, the content of FAs and the oil yield increased rapidly when ethanol concentration changed from 90% to 100% and reached the peak value at 100% concentration. FAs and the oil yield of BJO affected by the different liquid-solid ratio (8:1 mL/g-24:1 mL/g) are seen in Figure 2B, where two other factors, sonication time and ethanol concentration, were fixed at 30 min and 95%, respectively. The liquid-solid ratio significantly affected the content of FAs and the oil yield. The content of FAs and the oil yield of BJO increased rapidly with the liquid-solid ratio increased from 8:1 mL/g to 24:1 mL/g, and reached the maximum value at 24:1 mL/g. The content of FAs and the oil yield of BJO affected by different ethanol concentration (80-100%) are shown in Figure 2C, where two other factors, sonication time and liquid-solid ratio, were fixed at 30 min and 12:1 mL/g, respectively. The content of FAs and the oil yield of BJO were little when ethanol concentration was lower than 90%. However, the content of FAs and the oil yield increased rapidly when ethanol concentration changed from 90% to 100% and reached the peak value at 100% concentration. Combining with the results above and considering the time and cost savings, this study took sonication times of 20, 30, and 40 min; liquid-solid ratios of 16:1, 20:1, and 24:1 mL/g; and ethanol concentrations of 90%, 95%, and 100% for further study objects in the BBD experiment. Combining with the results above and considering the time and cost savings, this study took sonication times of 20, 30, and 40 min; liquid-solid ratios of 16:1, 20:1, and 24:1 mL/g; and ethanol concentrations of 90%, 95%, and 100% for further study objects in the BBD experiment.

Model Fitting
Analysis of variance (ANOVA) for the model is presented in Table 4. The results indicated that the model used to fit response variable was significant (p < 0.05, p < 0.01) and adequate to represent the relationship between the response and the independent variables. The coefficient (R 2 ) of the content of FAs and the oil yield were 0.8650 and 0.9456, respectively. The calculated models had no significant lack of fit at p > 0.05, which suggests a good fit. The predicted models reasonably represented the observed values. The content of FAs and the oil yield were affected most significantly by ethanol concentration (X 3 ) (p < 0.01, p < 0.0001), followed by liquid-solid ratio (X 2 ) (p < 0.05, p < 0.01), and sonication time (X 1 ) (p = 0.2669, p = 0.1212). The content of FAs shows that the quadratic parameters (X 1 2 , X 3 2 ) were significant (p < 0.05, p < 0.01), whereas the quadratic parameters (X 2 2 ) and all the interaction parameters were insignificant (p > 0.05). In the oil yield, all the interaction parameters (X 1 X 2 , X 1 X 3 , and X 2 X 3 ) and the quadratic parameters (X 1 2 , X 2 2 , and X 3 2 ) were insignificant (p > 0.05).
The predicted response could be expressed by the following second-order polynomial equations in terms of coded values: where Y 1 and Y 2 are the FA contents and the yield, respectively, and X 1 , X 2 , and X 3 are the coded variables for ultrasonic time, liquid-solid ratio, and ethanol concentration, respectively.

Analysis of Response Surface
The 3D response surface plots were obtained by varying two variables within the experimental range under investigation and holding another variable at its "0" level. Figures 3 and 4 show the 3D response surface plots and 2D contours for the optimization conditions of ultrasonic extraction of BJO, respectively. The effects of sonication time (X 1 ) and liquid-solid ratio (X 2 ) on the content of FAs and the oil yield are shown in Figure 3a,d, respectively. The response values were increased with increases in sonication time (X 1 ) and liquid-solid ratio (X 2 ). Figure 3b,e show the effect of sonication time (X 1 ) and ethanol concentration (X 3 ) on the content of FAs and the oil yield, respectively. The response values increased with the increases in sonication time (X 1 ). Further increase in ethanol concentration (X 3 ) resulted in the enhancement of this trend. Figure 3c,f show the effect of liquid-solid ratio (X 2 ) and ethanol concentration (X 3 ) on the content of FAs and the oil yield, respectively. Both increased with the increases in liquid-solid ratio (X 2 ) and ethanol concentration (X 3 ). the effect of liquid-solid ratio (X2) and ethanol concentration (X3) on the content of FAs and the oil yield, respectively. Both increased with the increases in liquid-solid ratio (X2) and ethanol concentration (X3).    g, a, b, and c) and oil yield (%, d, e, and f). Two variables were plotted against each other in each panel.

Optimization of Extraction Parameters and Validation of the Model
The optimized results showed that the maximum content of FAs and oil yield by UAE could achieve 13.94 mg/g and 17.86%, respectively. The software predicted the optimum sonication time,

Optimization of Extraction Parameters and Validation of the Model
The optimized results showed that the maximum content of FAs and oil yield by UAE could achieve 13.94 mg/g and 17.86%, respectively. The software predicted the optimum sonication time, liquid-solid ratio, and ethanol concentration to be 40 min, 24:1 mL/g, and 100%, respectively. Three parallel experiments were carried out under the optimal conditions. The actual values of the content of FAs and the oil yield were 14.64 mg/g and 16.87%, respectively, which were close to the predicted values. This result indicated that the optimization in the present study is reliable.

Effect of Extraction Times on Content of FAs and Oil Yield
The content of FAs and the oil yield of BJO became higher with the increase of extraction time. In the first extraction, the content of FAs and the oil yield of BJO were the highest (14.64 mg/g, 16.87%), and the content of FAs and the oil yield of the two instances that followed decreased quickly (The content of FAs and the oil yield obtained by second times were 2.81 mg/g and 5.7%, respectively, and the third times were 0.67 mg/g and 1.2%, respectively.). The relative cumulative value of two-time extraction has reached 95%, indicating that two-time extractions of BJO have been completed. The oil yield extracted twice was 22.57%, which was slightly higher than that of Ge et al. (21.35%) [33]. The two times were selected, taking into account the cost of extraction.
When considering extraction of oleic acid as the main active substance of BJO, the herbal sources from Guangxi and Guangdong are the better alternatives. However, in order to obtain higher total contents of FAs, the herbs from Guangxi and Guangdong should be chosen as the preferred sources instead of those from Fujian, Hainan and Yunnan. The experimental results also indicated that the varieties and contents of FAs in different sources were significantly different. Due to the planting area, harvest time, climate, and other factors, even if the herbs came from the same sources, their content of FAs of BJO were also somewhat different.

HPLC Fingerprint, Cluster, and Principal Component Analysis
The chromatogram of sample 1 (S1) was set as the reference map, and the fingerprint of the FA part of BJO of 14 batches was established. The control fingerprint was automatically generated by the system, and five common peaks were obtained. The similarity of all samples was higher than 0.991, which indicates that the FA composition of BJO of different regions had good similarity. The cluster analysis and principal component analysis were performed with the peak area of five common peaks of FA. Figure 5A,B show that the 14 samples could be clustered into two categories. Samples 1, 10, and 12 were combined into one class, and the rest of the samples were a another category. The contribution rate of the first and second principal components were 78.9% and 18.8%, respectively. The results were affected by linoleic, oleic, palmitic, and stearic acids ( Figure 5C).

Materials and Standards
Ethanol, acetonitrile, formic acid, and petroleum ether (60 • C to 90 • C) were obtained from Sinopharm Chemical Reagent Co., Ltd., Shanghai, China. Deionized water was purchased from Wahaha Co., Ltd., Hangzhou, China. Five FA standards (i.e., linolenic acid, linoleic acid, palmitic acid, oleic acid, and stearic acid) were purchased from Sigma (St. Louis, MO, USA). The raw B. javanica seeds were collected from different regions in China (Guangdong, Guangxi, Fujian, Hebei, Yunnan, and Hainan). All samples were authenticated by Dr. Hongmei Zhang of School of Pharmacy, Shanghai University of Traditional Chinese Medicine and stored at the Experiment Center for Teaching and Learning, Shanghai University of Traditional Chinese Medicine.

Extraction of Oil
The dried seeds were ground (60-mesh) with a mill (FWJ-03, Minye Industrial and Trading Co., Shanghai, China). The powder (5 g) was placed into a conical flask (250 mL). Ethanol (120 mL) was added and then the flask was placed in an ultrasonic cleaning bath (SB5200D, 40 kHz, Ningbo Scientz Biotechnology Co., Ltd., Zhejiang, China) for 40 min. The filtrate was collected, concentrated, and extracted with petroleum ether. BJO was obtained after solvent evaporation.

Preparation of Sample Solution
The BJO of different origins was extracted according to the optimal condition of the above extraction experiments. Appropriate BJO was weighed and dissolved in acetonitrile. The sample was swirled for 2 min, then sonicated for 2 min to disperse the contents evenly, and finally filtered through a 0.45 µm membrane.

HPLC-CAD Analysis
Measurements were carried out using HPLC U3000 with a Corona ® charged aerosol detector (Thermo Scientific, Idstein, Germany). A XB-C18 (150 mm × 4.6 mm, 5 µm) column (Welch, Shanghai, China) was used for sample separation at 30 • C. The injection volume was 10 µL. Isocratic elution (15:85) was applied using 0.05% (v/v) formic acid in water as mobile phase A and 0.05% (v/v) formic acid in acetonitrile as mobile phase B. The flow rate was set to 1 mL/min. The settings for the CAD were as follows: gas, nitrogen; pressure, 35 psi; filter, none; collection frequency, 2 Hz; and acquisition range, 100 pA.

Calibration Curves, Limits of Detection, and Quantification
Linolenic acid, linoleic acid, palmitic acid, oleic acid, and stearic acid were accurately weighed and dissolved in acetonitrile to make stock solutions. The stock solutions were diluted to five appropriate concentrations and analyzed in triplicate to construct the calibration curves. The limits of detection (LOD) and the limits of quantification (LOQ) of each analyte were calculated on the peak response at signal-to-noise of 3 and 10, respectively.

Precision
The precision of the method was evaluated by intra-day and inter-day precision. Three samples concentrations were analyzed six times daily to determine the intra-day precision. Three samples were analyzed three times on three consecutive days to test the inter-day precision.

Accuracy
The accuracy of this method was evaluated by the recovery test. Experiments were conducted by adding three different concentrations (50, 100, and 150 mg/mL) of five standard compounds. Each set of samples was repeated three times, and the average recovery rate of each compound was calculated. The equation for calculating the accuracy of HPLC method was as follows: Recovery (%) = (detected amount − original amount) / added amount × 100%

Content of FAs and Oil Yield Determination
The efficiency of the UAE was evaluated using the content of FAs and the oil yield as indexes. The content of FAs were equal to the sum of the content of linolenic, linoleic, palmitic, oleic, and stearic acids. The oil yield was calculated as follows: W 0 and W S are the weight of the oil extracted from the sample (g) and the weight of the sample (g), respectively.

Optimization of UAE
The extraction conditions were determined first by a single-factor test, and the effects of independent variables and their interactions on response variables were evaluated by BBD. The optimal extraction conditions that achieve the highest extraction rate and the highest FA contents were predicted by establishing a mathematical model. RSM optimized the extraction parameters. A three-factor, three-level BBD was applied to determine the optimal conditions for the UAE of BJO. Table 6 shows the range and center point values of the three independent variables based on the results of the single-factor test. The sonication time (min, X 1 ), liquid-solid ratio (mL/g, X 2 ), and ethanol concentration (%, X 3 ) were chosen as independent variables. The content of FAs and the oil yield were selected as the responses for the combination of the independent variables. The design of experiments are presented in Table 7. Experimental runs were randomized to minimize the effects of unexpected variability in the observed responses.
Experimental data were fitted to a quadratic polynomial model and the obtained regression coefficient. The nonlinear computer-generated quadratic model used in the response surface was as follows: where Y, β 0 , β i , β ii , and β ij indicated the predicted response, the intercept term, the linear coefficient, the squared coefficient, and the interaction coefficient, respectively.

Determination of Extraction Times
Extraction was performed under optimum conditions. Three samples were taken, and each sample was extracted three times. The content of FAs and the oil yield were calculated.

Conclusions
The present study developed and applied a highly efficient extraction method to extract BJO. The main variables affecting the content of FAs and oil yield were optimized by RSM based on a BBD. The proposed method could provide favorable extraction efficiency under optimal conditions. Ultrasonic-assisted extraction times were investigated in order to achieve the higher extraction rate of BJO. Moreover, a rapid, sensitive, and reliable method for the quantitative analysis of FAs in B. javanica (L.) Merr. seeds was developed. The analytical method used in this study avoided the complicated pretreatment process of samples and showed good sensitivity and low detection limit. The method was successfully applied to the simultaneous determination of five kinds of FAs in 14 batches of B. javanica (L.) Merr. seeds from different regions in China. The analysis results showed that B. javanica (L.) Merr. seeds is rich in oleic and linoleic acids, and the content of FAs varied significantly in B. javanica (L.) Merr. seeds from different regions. Moreover, the results indicated that the proposed method was sufficiently competent to determine FAs in various natural medicines and TCM and demonstrates great potential to analyze formulations and products containing BJO.