A green separation mode of synephrine from Citrus aurantium L. (Rutaceae) by nanofiltration technology

Abstract Thermal breakage of alkaloid ingredients was a common problem to which attention should be paid in the application of fruit ingredients separation. In this study, the mathematical models were established to predict the rejection of synephrine from Citrus aurantium L. (Rutaceae). The experiment showed that there was a linear relationship between operation pressure and membrane flux. Meanwhile, under the influence of solution–diffusion effect and the charge effect, the mass transfer coefficient was power functioned with initial concentration. The mathematical model showed that the predicted rejections of synephrine from Citrus aurantium extract were well approximate to real ones, and the lipid‐lowering active ingredient had effectively enriched. The predicted model of nanofiltration separation has a preferable applicability to synephrine and provides references for nanofiltration separation, especially for raw food materials with synephrine.

NF is a potential technology for efficient reservation of foods and medicine without heat effect, but the NF separation behaviors of synephrine are unclear. Based on the solution-diffusion and Donnan effect (Yaroshchuk & Bruening, 2017), a commercial NF membrane was used to clarify the relationship between membrane transport mechanisms and molecular state, and construct mass transfer model to predict separation behaviors. Given today's green separation demand over the world, NF can effectively improve separation efficiency and reduce environmental pollution.

| Materials
Citrus aurantium were obtained from a local market, which were from Zhangshu (Jiangxi, China). Synephrine extract (>98.0%) was purchased from Nanjing Zelang Biological Technology Co., Ltd.
Synephrine reference substance (>98.0%) was obtained from the National Institute for the Control of Pharmaceutical and Biological Products (Beijing, China). HPLC grade methanol was purchased from Merck. All the other chemicals used in the experiments were of analytical grade, and deionized water was used.
NF membrane was carried out by using polyamide membrane with negative charge, molecular weight cutoff of 300 Da, a filtra-

| Sample preparation
Citrus aurantium extract solution: The Citrus aurantium were washed with purified water, and fruit branches were cut off. The Citrus aurantium was processed in a commercial juicer to yield the natural juice. The natural juice was pretreated by a 0.45-μm microfiltration membrane and kept at 4-7°C to prevent damage or degradation.
Synephrine reference solution: 0.0102 g synephrine reference was added to 50% methanol water, which was used as synephrine reference solution at a concentration of 1.02 mg/ml. Synephrine extract solution: Synephrine extract was dissolved in deionized water with the concentration of 200 μg/ml.

| pK a measurement
The pK a value of synephrine was determined to analyze the existence of solute so that it can adjust the existence of solute, and based on this, the effect on mass transfer process can be investigated.
Since synephrine molecular structure contains imide group, the pK a measurement of synephrine was according to the determination of monobasic weak alkali. The pH value was measured by a glass pH electrode at room temperature. First, the 20 ml precise amount of series of concentrations of synephrine extract should be placed in a clean dry beaker especially, according to hydrochloric acid standard titration. Then, the pH value was recorded once when 1.0 ml hydrochloric acid of the concentration of 0.1003 mol/L was added into the extract. After that, draw a curve with the value of pH and the data of hydrochloric acid volume (V). When the slope of this curve was close to 1.0, the amount of hydrochloric acid was changed to 0.2 ml each to record the pH value. Finally, the titration process can be ended until the slope of the curve went to infinity. By repeating the titration operation for three times, the pK a value can be calculated according to Equation (1).

| Nanofiltration separation
In order to ensure that the separation performance of the membranes was not changed during filtration experiments, step 1, NF membrane was washed with deionized water to decrease membrane pollution and to maintain membrane flux. The flux was controlled by transmembrane pressure which was adjusted by the power and flow velocity valve of peristaltic pump.
Step 2, remaining water was pumped from the NF apparatus to decrease dilution effect.
Step 3, as the membrane flux dropped by 10%, repeat step 1.
During the experiment, the channel of feed solution, filtrate solution, and rejected solution were placed in the same container.
In order to improve NF membrane pollution, synephrine extract solution was pretreated by a polyether sulfone microfiltration membrane with a pore diameter of 0.45 μm and a filtration area of 0.3 m 2 .
Before sampling analysis, membrane module was pressurized at the test pressure for minimum 2 hr to reach the steady-state conditions and the adsorption-desorption equilibrium between solutes and membrane. The concentrations of the feed and permeate were analyzed with Agilent 1100. The rejection was calculated according to Equation (1).
where C o and C p are the solute concentrations in feed and permeate solution. Each measurement was performed in triplicate.
The initial concentration of synephrine extract solution was 200, 150, 100, 50, and 10 μg/ml for transmembrane pressure 0.2, 0.4, 0.6, 0.8, 1.0, and 1.2 MPa, respectively. The pH of synephrine extract solution was changed according to pK a , so that synephrine can be divided into three forms (ionic state, molecular state, and mixed state).

| Nanofiltration mass transfer model
Based on the solution-diffusion effect (Paul, 2004;Weng et al., 2016;Wijmans & Baker, 1995), it is assumed that in a NF process, the solvent and solute dissolve in the liquid-membrane interface after approaching to the film material, and then, the absorbed molecules transport through the membrane as a consequence of concentration gradient and pressure difference. The equations of the solution-diffusion model can be written as: is the osmotic pressure difference across the membrane; K is the solute partition coefficient; δ (cm) is the effective thickness of a film; DK/δ (cm/s) is the solute transport parameter; N A (mol/cm 2 s) is the solute flux; C m (mol/L) is the total concentration of solute in the liquid-membrane surface; and C p (mol/L) is the total concentration of solute in the permeate.
The rejections of solute can be divided into the observed rejection (R o ) and the real rejection (R r ), R o is experimentally measurable interception, which can be calculated by Equation (5). R r is the real rejection of components, which is difficult to calculate from experimental data due to the value of C m is hard to be determined directly.
R o is used for determining the membrane parameters.
Based on the solution-diffusion model and Equation (3)-Equation (6), the relationship between R o and k can be expressed as: Equation (7)

| HPLC conditions
Analysis for synephrine was carried out on a Hanbon C 18 column (250 × 4.6 mm, 5.0 μm). The binary mobile phase consisted of potassium phosphate monobasic (0.6 g potassium phosphate monobasic, 1.0 g sodium dodecyl benzene sulfonate, and 1 ml acetic acid were added to 1,000 ml deionized water) and methanol (50:50, v/v) at a rate of 1.0 ml/min. The detection wavelength was set to 275 nm.
The column temperature was sustained at 30°C, and the injection volume was 10 μl.

| Linear relationship
0.05, 0.10, 0.20, 0.50, 1.00, and 2.00 ml synephrine reference solution were, respectively, added to 10 ml methanol, and these samples were detected by Agilent 1100 HPLC. The calibration curve was operated by plotting the peak areas (Y) versus the concentration of synephrine (X), the linear regression equation of which was Y = 5.32 X + 5.57 (R 2 = 0.999 2). The calibration curve showed excellent linearity over the concentration ranges of 5.1-204.0 μg/ml for synephrine.

| Separation model verification
According to the pK a of synephrine, Citrus aurantium extract, of which the concentration of synephrine was detected by HPLC, was, respectively, adjusted to the completely ionic state, the totally molecular state, and the mixed state. The solutions were microfiltered through a 0.45-μm-pore size membrane. The R o value of synephrine at the different transmembrane pressure, respectively, equaled to 0.2, 0.4, 0.6, 0.8, 1.0, and 1.2 MPa, and the predicted one of model was compared with verify the practicability of the mass transfer model.

| Evaluation of lipid-lowering activity
Before the experiment, 50 SD rats were given basic feed for 3 days in the experimental environment. Random grouping after adaptation: The blank group continued to feed on the standard food, and 40 SD rats in the high-fat model group were fed high-fat diet. The rats were weighed every day, and the high-fat model was considered to be successful when the weight of the rats in the high-fat model group was at least 20% higher than that in the blank group.
The high-fat model group was randomly divided into four groups, 10 SD rats for each group, namely the high-fat model group, the positive drug control group, and the Citrus aurantium extracts and concentrate group. Moreover, the positive drug control group: orlistat 60 mg/(kg d), the high-fat model group was administrated equivalent distilled water, Citrus aurantium extracts group: 0.5 ml/(kg d), and Citrus aurantium concentrate group: 0.5 ml/(kg d). (2) After modeling, continuously administrate drug by drenching for 4 weeks, then put them to anesthesia, and get the orbital venous blood to do the test of TC, TG, HDL-C, and LDL-C content (Wang et al., 2019).

| pK a of synephrine with different concentration
The pK a of synephrine is between 8.25 and 8.84 with the concentration of synephrine ranges from 10 μg/ml to 200 μg/ml, respectively.
In order to study the influence of pH of solution on mass transfer coefficient, the synephrine extract solution was changed to pH 4.0 (ionic state), pH 10 (molecular state), and pH 8.5 (mixed state).

| Flux of nanofiltration
The What's more, pH has some effects on permeate flux as shown in  (Table 1), which also can be used to determine the extent of the mass transfer process impact of the charge effect. Meanwhile, the higher concentration of synephrine leads to the higher k, which indicates that the NF process depended not only on solution-diffusion model, but also on the charge effect.

| Ionic state
Ionic synephrine was difficult to get through the appearance of NF membrane because there was charge repulsion between synephrine and NF membrane. The k values of ionic synephrine dramatically decline compared with the molecular state data (Table 1), in that case, the charge effect was the leading factor in the process of NF separation, while the degree of rejection improved obviously contrasted to other solution environment, and the rejection was over 90% when it was at low concentrations. Furthermore, the value of ln[DK/δ] is relevant to the state of synephrine rather than the initial concentration of it. In addition, the value of ln[DK/δ] is higher in molecular state than in ionic state (Figure 2c).

| Verifying mass transfer model
The results showed that k performed correlation function with con-  Table 2, and then, the data of J v and R o can be calculated by Equation (7).

| Effect of synephrine samples on serum lipids of obese rat
Compared with the control group (Table 3), the level of TC, TG, and LDL-C in serum of the high-fat model group was significantly increased, but HDL-C was markedly lower than that in the control group (p < .01).
Compared with the Citrus aurantium extracts group, the level of TC, TG, and LDL-C in serum of the Citrus aurantium concentrate group was significantly decreased (p < .01), and HDL-C content was significantly increased (p < .01). After NF concentrating, the concentration of synephrine was effectively increased and the lipid-lowering activity was also enhanced (Figure 4).

| CON CLUS ION
Synephrine was separated and refined by NF technology at room temperature. In the separation process, synephrine molecules dissolved in the membrane surface and then transferred through membrane pore. The NF membrane surface carries negative charges, and as the pH of solution changes, ionic synephrine was difficult to attach to the membrane surface and then make synephrine hard to pass through membrane pores. Under the same concentration, the mass transfer coefficient was directly related to the pH.  TA B L E 3 Effect of synephrine samples on serum lipids of obese rat (X ± S, n = 10) mmol/L F I G U R E 4 Separation synephrine from Citrus aurantium L. (Rutaceae) by nanofiltration technology without thermal damage In this paper, NF separation is an effective technique for the concentration of synephrine from Citrus aurantium and the lipid-lowering activity was also enhanced. In addition, the NF technology can not only improve the technical level and products quality of raw food materials, but also be beneficial to the rationalization of nature plants resources.

ACK N OWLED G M ENTS
This work was supported by the National Natural Science Foundation of China (Grant No. 81603307, 81503258), Natural Science Fund for Colleges and Universities in Jiangsu province (Grant No. 17KJB360010).

CO N FLI C T O F I NTE R E S T
The authors have declared no conflict of interest.

E TH I C A L A PPROVA L
This study involved animal testing and was approved by the Institutional Review Board of the Nanjing University of Chinese Medicine. The protocol and procedures employed were ethically reviewed and approved by the same organ.