The Multi-Template Molecularly Imprinted Polymer Based on SBA-15 for Selective Separation and Determination of Panax notoginseng Saponins Simultaneously in Biological Samples

The feasible, reliable and selective multi-template molecularly imprinted polymers (MT-MIPs) based on SBA-15 (SBA-15@MT-MIPs) for the selective separation and determination of the trace level of ginsenoside Rb1 (Rb1), ginsenoside Rg1 (Rg1) and notoginsenoside R1 (R1) simultaneously from biological samples were developed. The polymers were constructed by SBA-15 as support, Rb1, Rg1, R1 as multi-template, acrylamide (AM) as functional monomer and ethylene glycol dimethacrylate (EGDMA) as cross-linker. The new synthetic SBA-15@MT-MIPs were satisfactorily applied to solid-phase extraction (SPE) coupled with high performance liquid chromatography (HPLC) for the separation and determination of trace Rb1, Rg1 and R1 in plasma samples. Under the optimized conditions, the limits of detection (LODs) and quantitation (LOQs) of the proposed method for Rb1, Rg1 and R1 were in the range of 0.63–0.75 ng·mL−1 and 2.1–2.5 ng·mL−1, respectively. The recoveries of R1, Rb1 and Rg1 were obtained between 93.4% and 104.3% with relative standard deviations (RSDs) in the range of 3.3–4.2%. All results show that the obtained SBA-15@MT-MIPs could be a promising prospect for the practical application in the selective separation and enrichment of trace Panax notoginseng saponins (PNS) in the biological samples.


Introduction
Panax notoginseng (Burk.) F. H. Chen (P. notoginseng) is a species of the genus Panax, family Araliaceae, which has been officially recorded in USP Herbal Medicines Compendium and China Pharmacopeia [1,2]. The radix and rhizome of P. notoginseng-known as Sanqi or Tianqi in East Asian countries-is one of the primary herbs in natural products. Sanqi has wide-ranging pharmacological effect such as anti-inflammation [3], cardioprotective effect [4], anticarcinogenic effect [5], anti-atherosclerotic [6] as well as antioxidant [7], etc. The major bioactive constituents responsible for these pharmacological effects are extensively recognized as the Panax notoginseng saponins (PNS), among which ginsenoside Rb 1 (Rb 1 ), ginsenoside Rg 1 (Rg 1 ) and notoginsenoside absorb a group of structural analogues, are highly desirable for sustainable development. However, there is no publication to date that reports the simultaneous separation and enrichment of Rb 1 , Rg 1 and R 1 by MT-MIPs based on SBA-15 (SBA-15@MT-MIPs) in biological samples.
In this study, we prepared highly sensitive and selective polymers as a novel and inexpensive sorbent for the isolation and enrichment of Rb 1 , Rg 1 and R 1 in biological samples. The SBA-15@MT-MIPs were synthesized using R 1 , Rb 1 , Rg 1 as multi-template, acrylamide (AM) as functional monomer, ethylene glycol dimethacrylate (EGDMA) as cross-linker and ethanol as porogen. The resulting products were characterized by TEM, FT-IR, TGA, N 2 adsorption-desorption analysis. Adsorption properties of obtained SBA-15@MT-MIPs were investigated, which included adsorption isotherm, adsorption kinetics and selective recognition. In addition, a molecularly imprinted solid-phase extraction (MISPE) procedure coupled with HPLC was successfully developed to isolate and detect trace amounts of Rb 1 , Rg 1 and R 1 in rat plasma samples, which indicated that the novel SBA-15@MT-MIPs can provide a proper example of an analytical method for quality control, support pharmacological and clinical studies of PNS.
The powder of Panax notoginseng was donated by the Department of Pharmacognosy, Nanjing Medical University (Nanjing, China).

Apparatus
The FT-IR analysis was conducted in the form of KBr pellets and using a Fourier Transform Infrared (FT-IR) spectrometer (Bruker, Switzerland) in a range of 500-4000 cm −1 . Ultraviolet-Visible (UV-Vis) absorbance detection was achieved by a Shimadzu UV-2100 spectrophotometer (Shimadzu, Kyoto, Japan). Heating magnetic stirrer (C-MAG HS 7, IKA, Staufen, Germany) was applied during the preparation. The morphologies and structures of polymers were observed by a scanning electron microscope (SEM, FEI Quanta 200, FEI, Hillsboro, OR, USA) and a transmission electron microscope (TEM, JEM-2100, JEOL, Tokyo, Japan). A shake culture box (ZHLY-180, Shanghai, China) was applied during the binding experiment. N 2 adsorption-desorption analysis was carried out at 77 K using a Micromeritics ASAP 2010 system (Micromeritics, Norcross, GA, USA). Thermogravimetric analysis was conducted on a thermogravimetric analyzer (TGA, Pyris 1 DSC, Perkin-Elmer, Waltham, MA, USA) with a heating rate of 20 • C·min −1 from 25 • C to 700 • C under a stream of N 2 gas (200 mL·min −1 ). HPLC analysis was performed with a Shimadzu (Kyoto, Japan) system comprising LC-20AT pump, SPD-M 20A detector, CTO-20A column oven and HW-2000 chromatographic work station. The structural identification of target compounds was performed using an Agilent 1200 liquid chromatograph (Agilent, Santa Clara, CA, USA) coupled with an Agilent 6410B Triple Quad mass spectrometer. The SBA-15 was synthesized according to the previous literature [33]. In brief, in a 250 mL round-bottomed flask, 2.0 g P123 was mixed with 2.0 mmol·L −1 hydrochloric acid (60 g), 0.033 mol acetic acid and 0.83 mol water under vigorous stirring at 40 • C. Subsequently, 4.25 g TEOS was added into the abovementioned solution drop by drop under vigorous stirring for 24 h at 40 • C and the mixture was transferred into Teflon-lined autoclaves and heated at 100 • C for 48 h. Finally, the resulting precipitates were filtered and washed with distilled water and then dried at ambient temperature. The organic templates were removed by calcining in air for 6 h at a heating rate of 5 • C·min −1 up to 550 • C.

Preparation of Vinyl Modified SBA-15
The surface of the as-prepared SBA-15 was modified by MPS according to our previous report [31]. Briefly, 500.0 mg SBA-15, 10 mL MPS and 50 mL anhydrous toluene were placed into a 100 mL three-necked flask and stirred. Then, the mixture solution was degassed with N 2 for about 10 min followed by heating the mixture to 55 • C for 24 h. Finally, the modified SBA-15 (SBA-15-MPS) containing vinyl terminal group was centrifuged, washed with toluene and methanol successively and dried under vacuum at 45 • C for 24 h.

Imprinting at the Surface of Modified SBA-15
The SBA-15@MT-MIPs were prepared by surface molecular imprinting technique, as shown in Figure 1. Here, a facile free-radical polymerization method was employed. Prior to polymerization, 0.2 mmol of the mixture of Rb 1 , Rg 1 and R 1 (the ratio of Rb 1 to Rg 1 to R 1 was 44:43:13 presented in P. notoginseng) [37] and functional monomer were dissolved in 30 mL of solvent, which was incubated for 4 h at ambient temperature for pre-polymerization to prepare the prearranged solution. Then 0.1 g SBA-15-MPS was dispersed in 20 mL of solvent by ultrasonic vibration. The prearranged solution, cross-linker (EGDMA) and initiator (AIBN) (the weight was about 5% of the total mass of templates, functional monomers and cross-linkers) were added to the above solution and mixed. The mixture was de-aerated by N 2 for 20 min and carried out in a shake culture box at 60 • C for 24 h. After the reaction, the SBA-15@MT-MIPs were centrifuged and washed successively with methanol to remove all absorbed oligomers and unreacted monomers. Finally, the SBA-15@MT-MIPs were washed repeatedly with methanol-acetic acid (9:1, v/v) to remove the templates until no templates in the supernatant were detected using a UV spectrophotometer. The resulting imprinted polymers were repeatedly washed with methanol until the supernatant was neutral and then dried in a vacuum at 40 • C for 24 h. The multi-template non-imprinted polymers based on SBA-15 (SBA-15@MT-NIPs) were also fabricated and purified under identical condition except that the templates were omitted. The SBA-15 was synthesized according to the previous literature [33]. In brief, in a 250 mL round-bottomed flask, 2.0 g P123 was mixed with 2.0 mmol·L −1 hydrochloric acid (60 g), 0.033 mol acetic acid and 0.83 mol water under vigorous stirring at 40 °C. Subsequently, 4.25 g TEOS was added into the abovementioned solution drop by drop under vigorous stirring for 24 h at 40 °C and the mixture was transferred into Teflon-lined autoclaves and heated at 100 °C for 48 h. Finally, the resulting precipitates were filtered and washed with distilled water and then dried at ambient temperature. The organic templates were removed by calcining in air for 6 h at a heating rate of 5 °C ·min −1 up to 550 °C.

Preparation of Vinyl Modified SBA-15
The surface of the as-prepared SBA-15 was modified by MPS according to our previous report [31]. Briefly, 500.0 mg SBA-15, 10 mL MPS and 50 mL anhydrous toluene were placed into a 100 mL three-necked flask and stirred. Then, the mixture solution was degassed with N2 for about 10 min followed by heating the mixture to 55 °C for 24 h. Finally, the modified SBA-15 (SBA-15-MPS) containing vinyl terminal group was centrifuged, washed with toluene and methanol successively and dried under vacuum at 45 °C for 24 h.

Imprinting at the Surface of Modified SBA-15
The SBA-15@MT-MIPs were prepared by surface molecular imprinting technique, as shown in Figure 1. Here, a facile free-radical polymerization method was employed. Prior to polymerization, 0.2 mmol of the mixture of Rb1, Rg1 and R1 (the ratio of Rb1 to Rg1 to R1 was 44:43:13 presented in P. notoginseng) [37] and functional monomer were dissolved in 30 mL of solvent, which was incubated for 4 h at ambient temperature for pre-polymerization to prepare the prearranged solution. Then 0.1 g SBA-15-MPS was dispersed in 20 mL of solvent by ultrasonic vibration. The prearranged solution, cross-linker (EGDMA) and initiator (AIBN) (the weight was about 5% of the total mass of templates, functional monomers and cross-linkers) were added to the above solution and mixed. The mixture was de-aerated by N2 for 20 min and carried out in a shake culture box at 60 °C for 24 h. After the reaction, the SBA-15@MT-MIPs were centrifuged and washed successively with methanol to remove all absorbed oligomers and unreacted monomers. Finally, the SBA-15@MT-MIPs were washed repeatedly with methanol-acetic acid (9:1, v/v) to remove the templates until no templates in the supernatant were detected using a UV spectrophotometer. The resulting imprinted polymers were repeatedly washed with methanol until the supernatant was neutral and then dried in a vacuum at 40 °C for 24 h. The multi-template non-imprinted polymers based on SBA-15 (SBA-15@MT-NIPs) were also fabricated and purified under identical condition except that the templates were omitted.

Evaluation of the Binding Ability of SBA-15@MT-MIPs
To evaluate the adsorption performance of SBA-15@MT-MIPs, the isotherm, kinetics and selective adsorption experiments were carried out. All experiments were repeated three times in parallel. In all experiments, 20 mg polymers were dispersed in a 5 mL ethanol solution. The mixture was incubated on a shaker at 200 rpm for a period of time and then filtered using a 0.22 µm microporous membrane. The maximum absorbance of Rb 1 , Rg 1 and R 1 at about 204 nm before and after adsorption was measured by UV-Vis spectrophotometer and the corresponding concentrations were calculated through the calibration curve. The adsorption capacity (Q, µmol·g −1 ) of the template bound to the polymer was calculated by the following Equation (1): where Q e is the equilibrium adsorption amount of Rb 1 , Rg 1 and R 1 , C 0 and C e represent the initial concentration and the equilibrium concentration of Rb 1 , Rg 1 and R 1 , respectively (mmol·L −1 ), V is the volume of solution tested and m is the mass of the polymer. In order to investigate the recognition performance of adsorbents for target molecules, an equilibrium adsorption experiment was studied at various concentrations of the mixture solution of Rb 1 , Rg 1 and R 1 ranging from 0.5 mmol·L −1 to 4.0 mmol·L −1 . The mixture was kept shaking for 80 min at ambient temperature.
To investigate the adsorption rate of the obtained SBA-15@MT-MIPs, an adsorption kinetics experiment was conducted. The adsorption time was changed at regular intervals from 0 min to 80 min, while the mixture solution of Rb 1 , Rg 1 and R 1 was kept at a concentration of 3.0 mmol·L −1 , respectively.
To measure the specificity of SBA-15@MT-MIPs, UA and OA were adopted as structural analogues to compare with Rb 1 , Rg 1 and R 1 . A selective adsorption experiment was investigated using solutions of R 1 , Rg 1 , Rb 1 , UA and OA at a concentration of 3.0 mmol·L −1 and incubation time of 60 min. The imprinting factor (IF) was used to evaluate the specific recognition property of the polymers towards the targets and structural analogues, which was calculated by Equation (2): here Q MIP and Q NIP are the adsorption capacities of the template and/or structural analog on SBA-15@MT-MIPs and SBA-15@MT-NIPs, respectively. To investigate the stability and reusability of SBA-15@MT-MIPs, six adsorption-desorption cycles were conducted by using the same SBA-15@MT-MIPs. The adsorption procedure was the same as the binding experiment. The concentration of the mixture solution of Rb 1 , Rg 1 and R 1 and incubation time was 3.0 mmol·L −1 and 60 min, respectively. Elution of Rb 1 , Rg 1 and R 1 from SBA-15@MT-MIPs was performed using methanol-acetic acid (9:1, v/v) and methanol in turn.
2.6. Separation and Determination of Rb 1 , Rg 1 and R 1 in Real Samples 2.6.1. Preparation of Plasma Samples 50.0 g of P. notoginseng powder was weighted accurately and placed into a 500 mL round-bottom flask and 400 mL of 70% ethanol was added. The mixture was refluxed for 4 h at 80 • C. The extraction was repeated two additional times and the combined extracts were concentrated to about 10 mL under vacuum and then diluted to 50 mL with physiological saline. Ten male SD rats (250-280 g) were fed with standard laboratory food and water ad libitum except for fasting 12 h prior to experiment. The extract solution was orally administered to rats at the doses equivalent to 10 g·kg −1 of the powder of P. notoginseng and equal volume of normal saline for the blank group. At 50 min after drug dosing, orbital blood samples were collected into heparinized microfuge tubes and immediately centrifuged at 3000 rpm for 15 min to obtain plasma, which was precisely collected and stored at −20 • C until analysis. In order to reduce complexity of plasma matrix, the extraction was carried out by adding quadruple volume of methanol to the plasma and the mixture was vortexed for 30 s and then centrifuging at 12,000 rpm for 10 min to precipitate the denatured proteins. The supernatant was evaporated to dryness under a gentle N 2 stream at room temperature and then re-dissolved with 5 mL toluene-ethanol (7:3, v/v) and collected as analytical samples.
2.6.2. Enrichment of Rb 1 , Rg 1 and R 1 from Plasma Samples by MISPE Firstly, 500 mg of SBA-15@MT-MIPs and SBA-15@MT-NIPs were packed manually in an empty solid-phase extraction (SPE) column. Each cartridge was rinsed with methanol-acetic acid (9:1, v/v) and no template was detected in the obtained solution by HPLC. After being conditioned with 5 mL toluene-ethanol (7:3, v/v), 5 mL of analytical samples were loaded into the cartridge and sealed. Subsequently, it was washed with 5 mL of washing solution to remove the interfering substances and then was dried thoroughly by a vacuum pump. Finally, it was eluted with 5 mL of methanol-acetic acid (9:1, v/v) to get the desired Rb 1 , Rg 1 and R 1 . Both the washing and elution fractions were evaporated to dryness under a gentle N 2 stream at room temperature and the obtained residues were re-dissolved in 1 mL methanol for HPLC analysis.

Chromatographic Analysis
The amount of Rb 1 , Rg 1 and R 1 was analyzed by an HPLC system. A Shimadzu shim-pack C 18 column (250 mm × 4.6 mm, 5 µm) was used as the stationary phase. The mobile phase consisted of ultra-water (A) and acetonitrile (B) and was run in an isocratic mode at a flow rate of 1.0 mL·min −1 . The following gradient was performed: 0-23 min, 23% B; 23-40 min, 35% B. The column temperature was maintained at 35 • C. Aliquots of 20 µL were injected. Monitoring and quantitation of Rb 1 , Rg 1 and R 1 were performed at 204 nm.
Agilent 1200 liquid chromatograph combined with an Agilent 6410B Triple Quad mass spectrometer was used to identify Rb 1 , Rg 1 and R 1 in biological samples. An Agilent zorbax eclipse plus C 18 column (150 mm × 2.1 mm, 5 µm) was used as the stationary phase. The mobile phase for LC-MS analysis consisted of acetonitrile-water (5:5, v/v) at a flow rate of 0.5 mL·min −1 .

The Optimization of Preparation Conditions of SBA-15@MT-MIPs
The successful preparation of MIPs largely depends on the judicious selection of the functional precursor, such as functional monomer, solvent/porogen and their relative molar ratio, etc. So, the preparation conditions of SBA-15@MT-MIPs should be optimized.
The selectivity of SBA-15@MT-MIPs is closely related to the number and shape of binding sites within the polymer network structure, as well as the rigidity of the complexes formed by template molecule and functional monomer [38]. Therefore, monomer selection is the first priority for the synthesis of SBA-15@MT-MIPs with good affinity and high selectivity. In this study, three potential functional monomers (4-VP, AM and MAA) were screened to prepare the polymers. In Table 1, SBA-15@MT-MIPs-3 showed the highest Q (108.9 µmol·g −1 ) and IF values (2.11). It was concluded that the monomer based on AM created SBA-15@MT-MIPs with higher affinity binding sites for Rb 1 , Rg 1 and R 1 compared to other study monomers. Although the nitrogen of pyridine group in the 4-VP and the oxygen of the carbonyl in the MAA and AM were likely to become the hydrogen bond acceptors for the protons of the hydroxyl groups of the Rb 1 , Rg 1 and R 1 , the monomer based on AM created SBA-15@MT-MIPs with higher affinity binding sites for Rb 1 , Rg 1 and R 1 compared to other study monomers. The reason may be that strong complexes were formed between AM and templates [39]. Therefore, AM was chosen as the functional monomer in the experiment. In addition, the ratio of functional monomer to the cross-linker has an influence on binding capacity and rigid polymer network. In this study, 1:4 was the optimum molar ratio of monomer to cross-linker according to previous paper reports [40,41]. Generally, the ratio of template to functional monomer has influence on binding affinity and imprinting effect of the polymers, so we examined the effect of different molar ratios of template to functional monomer in a range of 1:3-1:6. The molecular structures of Rb 1 , Rg 1 and R 1 can be seen in Figure 2. The results in Table 1 revealed that the SBA-15@MT-MIPs-3 had the highest adsorption capacity and selectivity with the molar ratios of 1:4, while other molar ratios showed lower adsorption capacity and selectivity. The reason may be that fewer templates induce fewer binding sites in polymers due to fewer template-monomer complexes but residual templates produce lower specific binding capacity and higher non-specific binding capacity [42]. So, 1:4:16 was selected as the optimum molar ratio of template to functional monomer to cross-linker and their exact molar amount was 0.2, 0.8 and 3.2 mmol, respectively. network. In this study, 1:4 was the optimum molar ratio of monomer to cross-linker according to previous paper reports [40,41]. Generally, the ratio of template to functional monomer has influence on binding affinity and imprinting effect of the polymers, so we examined the effect of different molar ratios of template to functional monomer in a range of 1:3-1:6. The molecular structures of Rb1, Rg1 and R1 can be seen in Figure 2. The results in Table 1 revealed that the SBA-15@MT-MIPs-3 had the highest adsorption capacity and selectivity with the molar ratios of 1:4, while other molar ratios showed lower adsorption capacity and selectivity. The reason may be that fewer templates induce fewer binding sites in polymers due to fewer template-monomer complexes but residual templates produce lower specific binding capacity and higher non-specific binding capacity [42]. So, 1:4:16 was selected as the optimum molar ratio of template to functional monomer to cross-linker and their exact molar amount was 0.2, 0.8 and 3.2 mmol, respectively. The selectivity and affinity of synthesized SBA-15@MT-MIPs through non-covalent interaction between active sites of SBA-15@MT-MIPs and imprinted molecules are closely related to the polarity of the porogens. In addition, porogens have a substantial impact on the self-assembly of templatefunctional monomer and the size of mesoporous and macropores in the particles. Rb1, Rg1 and R1 are valuable bioactive molecules but its high polarity [43]. So, the choice of solvents is very challenging. It is well known that high polarity porogenic solvents are likely to disturb the monomer-template interactions [44] but the low polar solvents could have failed to dissolve templates and functional monomer completely. Therefore, we chosen tetrahydrofuran, ethanol and ethanol-acetonitrile (2:3, v/v) as solvent. The results were shown in Table 1. When using tetrahydrofuran and ethanolacetonitrile (2:3, v/v) as the porogens, the Q and IF of the SBA-15@MT-MIPs were not the highest. It was indicated that the hydrogen bond interaction between templates (Rb1, Rg1 and R1) and AM in ethanol was stronger than the other choices mentioned [45]. Rb1, Rg1 and R1 were sparingly soluble in ethanol-acetonitrile (2:3, v/v), whereas slightly soluble in tetrahydrofuran. Ethanol was chosen to The selectivity and affinity of synthesized SBA-15@MT-MIPs through non-covalent interaction between active sites of SBA-15@MT-MIPs and imprinted molecules are closely related to the polarity of the porogens. In addition, porogens have a substantial impact on the self-assembly of template-functional monomer and the size of mesoporous and macropores in the particles. Rb 1 , Rg 1 and R 1 are valuable bioactive molecules but its high polarity [43]. So, the choice of solvents is very challenging. It is well known that high polarity porogenic solvents are likely to disturb the monomer-template interactions [44] but the low polar solvents could have failed to dissolve templates and functional monomer completely. Therefore, we chosen tetrahydrofuran, ethanol and ethanol-acetonitrile (2:3, v/v) as solvent. The results were shown in Table 1. When using tetrahydrofuran and ethanol-acetonitrile (2:3, v/v) as the porogens, the Q and IF of the SBA-15@MT-MIPs were not the highest. It was indicated that the hydrogen bond interaction between templates (Rb 1 , Rg 1 and R 1 ) and AM in ethanol was stronger than the other choices mentioned [45]. Rb 1 , Rg 1 and R 1 were sparingly soluble in ethanol-acetonitrile (2:3, v/v), whereas slightly soluble in tetrahydrofuran. Ethanol was chosen to be the polymerization solvent, in which Rb 1 , Rg 1 and R 1 can completely dissolve. So, SBA-15@MT-MIPs were prepared in ethanol in this study.
From all the above, the three parameters including the selection of the functional monomers, the molar ratios of template to functional monomer and the optimization of polymerization porogens have been optimized and the optimal material was SBA-15@MT-MIPs-3, which was employed for further study.

Characterization of SBA-15@MT-MIPs
The FT-IR spectra ascertained the successful synthesis and modification of SBA-15. Meanwhile, SBA-15@MT-MIPs had been prepared successfully. As shown in Figure 3A From all the above, the three parameters including the selection of the functional monomers, the molar ratios of template to functional monomer and the optimization of polymerization porogens have been optimized and the optimal material was SBA-15@MT-MIPs-3, which was employed for further study.

Characterization of SBA-15@MT-MIPs
The FT-IR spectra ascertained the successful synthesis and modification of SBA-15. Meanwhile, SBA-15@MT-MIPs had been prepared successfully. As shown in Figure 3A  The FT-IR spectra of R1, Rg1, Rb1, SBA-15@MT-MIPs before eluting templates and SBA-15@MT-NIPs were explored and the corresponding results were presented in Figure 3B. The as-prepared SBA-15@MT-MIPs before eluting templates ( Figure 3B(d)) clearly displayed the characteristic peaks of R1, Rg1 and Rb1, including the peaks at 1078 cm −1 and 1045 cm −1 corresponding to [47], while the peaks were not shown in the spectrum of SBA-15@MT-NIPs ( Figure 3B(e)). It can be inferred that R1, Rg1 and Rb1 were existed in SBA-15@MT-MIPs before elution. The peak at 1726 cm −1 was ascribed to the stretching vibration absorption of the C=O, which demonstrated the existence of EGDMA and AM at SBA-15@MT-MIPs and SBA-15@MT-NIPs. In conclusion, the results indicated that MIP layers were grafted onto the surface of SBA-15-MPS through the free-radical polymerization method.
The representative morphology and microstructure of SBA-15 and SBA-15@MT-MIPs were characterized by SEM and TEM, respectively. The size-uniformity of SBA-15 with short hexagonal rod-like structures approximately 1.4 μm were observed in Figure 4A. Compared with SBA-15, significant differences in morphology and particle shape can be seen from Figure 4B, SBA-15@MT-MIPs with a honeycomb-like form exhibited high density of macropores, which were ascribed to the generated imprinted polymer thin layers. Besides, the results confirm that agglomerated particles The FT-IR spectra of R 1 , Rg 1 , Rb 1 , SBA-15@MT-MIPs before eluting templates and SBA-15@MT-NIPs were explored and the corresponding results were presented in Figure 3B. The asprepared SBA-15@MT-MIPs before eluting templates ( Figure 3B(d)) clearly displayed the characteristic peaks of R 1 , Rg 1 and Rb 1 , including the peaks at 1078 cm −1 and 1045 cm −1 corresponding to [47], while the peaks were not shown in the spectrum of SBA-15@MT-NIPs ( Figure 3B(e)). It can be inferred that R 1 , Rg 1 and Rb 1 were existed in SBA-15@MT-MIPs before elution. The peak at 1726 cm −1 was ascribed to the stretching vibration absorption of the C=O, which demonstrated the existence of EGDMA and AM at SBA-15@MT-MIPs and SBA-15@MT-NIPs. In conclusion, the results indicated that MIP layers were grafted onto the surface of SBA-15-MPS through the free-radical polymerization method.
The representative morphology and microstructure of SBA-15 and SBA-15@MT-MIPs were characterized by SEM and TEM, respectively. The size-uniformity of SBA-15 with short hexagonal rod-like structures approximately 1.4 µm were observed in Figure 4A. Compared with SBA-15, Polymers 2017, 9, 653 9 of 18 significant differences in morphology and particle shape can be seen from Figure 4B, SBA-15@MT-MIPs with a honeycomb-like form exhibited high density of macropores, which were ascribed to the generated imprinted polymer thin layers. Besides, the results confirm that agglomerated particles had nano-sized which provide suitable structure for SPE adsorbents. To investigate the morphology and microstructure thoroughly and intuitively, the TEM images of prepared materials were obtained. As can be seen in Figure 4C,D, all samples displayed the well-ordered hexagonal mesoporous structures, similarly indicating the structure of SBA-15 was maintained after modification and polymerization. However, noticeable differences between them can be distinguished in the images. As illustrated in Figure 4D, the imprinted polymer layers with uniform thickness were densely combined with the surface of mesoporous carriers. As expected, we can posit that the novel MT-MIPs based on mesoporous carriers were successfully prepared. Moreover, the thin imprinted layers could be beneficial for the mass transfer between templates and the surface of SBA-15@MT-MIPs.
Polymers 2017, 9, 653 9 of 18 had nano-sized which provide suitable structure for SPE adsorbents. To investigate the morphology and microstructure thoroughly and intuitively, the TEM images of prepared materials were obtained. As can be seen in Figure 4C,D, all samples displayed the well-ordered hexagonal mesoporous structures, similarly indicating the structure of SBA-15 was maintained after modification and polymerization. However, noticeable differences between them can be distinguished in the images. As illustrated in Figure 4D, the imprinted polymer layers with uniform thickness were densely combined with the surface of mesoporous carriers. As expected, we can posit that the novel MT-MIPs based on mesoporous carriers were successfully prepared. Moreover, the thin imprinted layers could be beneficial for the mass transfer between templates and the surface of SBA-15@MT-MIPs. To investigate the thermostability and solid content of the materials, Figure 5 shows the TGA curves of SBA-15, SBA-15-MPS and SBA-15@MT-MIPs in the temperature range from 25 °C to 700 °C. As shown in Figure 5(a), The TGA curve of SBA-15 had only one weight loss stage in the temperature below 100 °C, which was assigned to the release of physically and chemically adsorbed water in SBA-15. Figure 5(b) illustrates that SBA-15-MPS had two weight loss steps in the range of 25-700 °C. The first step below 100 °C was the release of physically adsorbed water and solvent residues in SBA-15. The second weight loss step showed around 14% weight loss in the temperature range of 100-550 °C, which corresponded to the degradation of the MPS silane coupling agent. Analogously, in the TGA curve of SBA-15@MT-MIPs ( Figure 5(c)), the first weight loss step corresponded to the surface adsorbed water. The next much higher weight loss (around 82%) was observed from 100 °C to 482 °C, which could be assigned to the removal of the organic content in the imprinted layers. Meanwhile, it was confirmed that imprinted polymer layers were successfully synthesized.
N2 adsorption-desorption analysis was used to evaluate the porosity changes of the mesoporous SBA-15 by the introduction of MPS and MIP layers. The surface areas were measured with Brunauer-Emmett-Teller (BET) theory. Total pore volumes and pore size distributions were obtained based on the adsorption branch of the isotherm with the Barrett-Joyner-Halenda (BJH) method. The results were shown in Table 2. After the attachment of MPS, the BET surface area, pore volume and pore size of SBA-15-MPS decreased slightly compared with SBA-15. However, the surface area (SBET), pore volume (VT) and pore size (DP) of SBA-15@MT-MIPs decreased drastically to 140.2 m 2 ·g −1 , 0.1018 cm 3 ·g −1 and 34.28 Å , respectively. These changes provided proof that the imprinted layers existed in the nanopores of SBA-15, which has been reported elsewhere [48]. To investigate the thermostability and solid content of the materials, Figure 5 shows the TGA curves of SBA-15, SBA-15-MPS and SBA-15@MT-MIPs in the temperature range from 25 • C to 700 • C. As shown in Figure 5(a), The TGA curve of SBA-15 had only one weight loss stage in the temperature below 100 • C, which was assigned to the release of physically and chemically adsorbed water in SBA-15. Figure 5(b) illustrates that SBA-15-MPS had two weight loss steps in the range of 25-700 • C. The first step below 100 • C was the release of physically adsorbed water and solvent residues in SBA-15. The second weight loss step showed around 14% weight loss in the temperature range of 100-550 • C, which corresponded to the degradation of the MPS silane coupling agent. Analogously, in the TGA curve of SBA-15@MT-MIPs ( Figure 5(c)), the first weight loss step corresponded to the surface adsorbed water. The next much higher weight loss (around 82%) was observed from 100 • C to 482 • C, which could be assigned to the removal of the organic content in the imprinted layers. Meanwhile, it was confirmed that imprinted polymer layers were successfully synthesized.

Study of Binding Properties
The binding properties of polymers towards Rb1, Rg1 and R1 was investigated by the static adsorption equilibrium experiment and adsorption kinetic studies.
Binding isotherms are a measure of the concentration dependent recognition behavior of a system. As shown in Figure 6A, the adsorption capacities of SBA-15@MT-MIPs and SBA-15@MT-NIPs for Rb1, Rg1 and R1 increased rapidly in the beginning stages and later slowed down with the increasing of initial concentration. Then the adsorption tended to be stable gradually and become constant at 3 mmol L −1 , due to the saturation of recognition sites. The saturated adsorption capacities of SBA-15@MT-MIPs and SBA-15@MT-NIPs for Rb1, Rg1 and R1 were calculated to 123.11, 58.41 μmol·g −1 , respectively. The IF of the SBA-15@MT-MIPs for templates (Rb1, Rg1 and R1) was also found satisfactory (IF = 2.11). Evidently, the amount of Rb1, Rg1 and R1 bonded on the SBA-15@MT-MIPs cavities was larger than that on the corresponding SBA-15@MT-NIPs at all the concentrations studies. This obvious preferential adsorption was ascribed to the SBA-15@MT-MIPs containing special binding sites complementary in size and shape to the template molecules and the existence of hydrogen bonds between the functional groups and template molecules.
The adsorption kinetic study can provide valuable information about both the binding and the rate-controlling mechanism. As indicated in the Figure 6B, the adsorption rate of the SBA-15@MT-MIPs towards the mixture of Rb1, Rg1 and R1 increased sharply in the first 30 min and then slowed down gradually to reach equilibrium within 60 min. The maximum absorption capacity of SBA-15@MT-MIPs for Rb1, Rg1 and R1 was 123.11 μmol·g −1 and the value of IF was 2.12. In addition, the dynamic curve of SBA-15@MT-NIPs showed a similar trend with that of SBA-15@MT-MIPs but the lower adsorption amount. These could occur because the specific recognition sites of SBA-15@MT-MIPs were generated in the molecular imprinting process and most of the binding sites were located in the proximity of the interior surface. So, target species could easily diffuse into the recognition sites of SBA-15@MT-MIPs, leading to the higher binding capacity and faster mass transfer rate.  Table 2. After the attachment of MPS, the BET surface area, pore volume and pore size of SBA-15-MPS decreased slightly compared with SBA-15. However, the surface area (S BET ), pore volume (V T ) and pore size (D P ) of SBA-15@MT-MIPs decreased drastically to 140.2 m 2 ·g −1 , 0.1018 cm 3 ·g −1 and 34.28 Å, respectively. These changes provided proof that the imprinted layers existed in the nanopores of SBA-15, which has been reported elsewhere [48].

Study of Binding Properties
The binding properties of polymers towards Rb 1 , Rg 1 and R 1 was investigated by the static adsorption equilibrium experiment and adsorption kinetic studies.
Binding isotherms are a measure of the concentration dependent recognition behavior of a system. As shown in Figure 6A, the adsorption capacities of SBA-15@MT-MIPs and SBA-15@MT-NIPs for Rb 1 , Rg 1 and R 1 increased rapidly in the beginning stages and later slowed down with the increasing of initial concentration. Then the adsorption tended to be stable gradually and become constant at 3 mmol·L −1 , due to the saturation of recognition sites. The saturated adsorption capacities of SBA-15@MT-MIPs and SBA-15@MT-NIPs for Rb 1 , Rg 1 and R 1 were calculated to 123.11, 58.41 µmol·g −1 , respectively. The IF of the SBA-15@MT-MIPs for templates (Rb 1 , Rg 1 and R 1 ) was also found satisfactory (IF = 2.11). Evidently, the amount of Rb 1 , Rg 1 and R 1 bonded on the SBA-15@MT-MIPs cavities was larger than that on the corresponding SBA-15@MT-NIPs at all the concentrations studies. This obvious preferential adsorption was ascribed to the SBA-15@MT-MIPs containing special binding sites complementary in size and shape to the template molecules and the existence of hydrogen bonds between the functional groups and template molecules.

Molecular Selectivity of the SBA-15@MT-MIPs
To measure the specificity of SBA-15@MT-MIPs, UA and OA were chosen as the reference compounds. The molecular structures of Rb1, Rg1, R1, UA and OA can be seen in Figure 2. From Figure  7A, we can see that the Q and IF of the SBA-15@MT-MIPs towards the templates of Rb1, Rg1 and R1 were significantly higher than those of the structural analogs, UA and OA. However, the binding capacities of SBA-15@MT-NIPs for the five analytes were almost the same and all lower than those of SBA-15@MT-MIPs. These results confirmed that SBA-15@MT-MIPs possessed higher selectivity and affinity toward Rb1, Rg1 and R1 owing to the existence of the specific recognition sites between the monomers and template molecules. As the competitive structural analog UA and OA, although UA and OA are similar to Rb1, Rg1 and R1 in the molecular volume and structures, the recognition sites were not complementary to the structural analogs. In other words, the memory of specific functional groups also played an important part in conformation memory [49].

Reusability of SBA-15@MT-MIPs
The stability of SBA-15@MT-MIPs was evaluated by comparing the adsorption capacity towards Rb1, Rg1 and R1 at each cycle. As shown in Figure 7B, the adsorption capacity decreased slowly with an increase in cycle times. After six adsorption-desorption cycles, the adsorption capacity was about 9.09% less. The slight decrease in adsorption capacity mainly resulted from the partial destruction of the recognition sites during the elution procedure. The results indicated that recognition, interaction and adsorption processes occurred reversibly and the polymers had excellent stability and acceptable reusability. So, it could be used for practical applications. The adsorption kinetic study can provide valuable information about both the binding and the rate-controlling mechanism. As indicated in the Figure 6B, the adsorption rate of the SBA-15@MT-MIPs towards the mixture of Rb 1 , Rg 1 and R 1 increased sharply in the first 30 min and then slowed down gradually to reach equilibrium within 60 min. The maximum absorption capacity of SBA-15@MT-MIPs for Rb 1 , Rg 1 and R 1 was 123.11 µmol·g −1 and the value of IF was 2.12. In addition, the dynamic curve of SBA-15@MT-NIPs showed a similar trend with that of SBA-15@MT-MIPs but the lower adsorption amount. These could occur because the specific recognition sites of SBA-15@MT-MIPs were generated in the molecular imprinting process and most of the binding sites were located in the proximity of the interior surface. So, target species could easily diffuse into the recognition sites of SBA-15@MT-MIPs, leading to the higher binding capacity and faster mass transfer rate.

Molecular Selectivity of the SBA-15@MT-MIPs
To measure the specificity of SBA-15@MT-MIPs, UA and OA were chosen as the reference compounds. The molecular structures of Rb 1 , Rg 1 , R 1 , UA and OA can be seen in Figure 2. From Figure 7A, we can see that the Q and IF of the SBA-15@MT-MIPs towards the templates of Rb 1 , Rg 1 and R 1 were significantly higher than those of the structural analogs, UA and OA. However, the binding capacities of SBA-15@MT-NIPs for the five analytes were almost the same and all lower than those of SBA-15@MT-MIPs. These results confirmed that SBA-15@MT-MIPs possessed higher selectivity and affinity toward Rb 1 , Rg 1 and R 1 owing to the existence of the specific recognition sites between the monomers and template molecules. As the competitive structural analog UA and OA, although UA and OA are similar to Rb 1 , Rg 1 and R 1 in the molecular volume and structures, the recognition sites were not complementary to the structural analogs. In other words, the memory of specific functional groups also played an important part in conformation memory [49].
SBA-15@MT-MIPs. These results confirmed that SBA-15@MT-MIPs possessed higher selectivity and affinity toward Rb1, Rg1 and R1 owing to the existence of the specific recognition sites between the monomers and template molecules. As the competitive structural analog UA and OA, although UA and OA are similar to Rb1, Rg1 and R1 in the molecular volume and structures, the recognition sites were not complementary to the structural analogs. In other words, the memory of specific functional groups also played an important part in conformation memory [49].

Reusability of SBA-15@MT-MIPs
The stability of SBA-15@MT-MIPs was evaluated by comparing the adsorption capacity towards Rb1, Rg1 and R1 at each cycle. As shown in Figure 7B, the adsorption capacity decreased slowly with an increase in cycle times. After six adsorption-desorption cycles, the adsorption capacity was about 9.09% less. The slight decrease in adsorption capacity mainly resulted from the partial destruction of the recognition sites during the elution procedure. The results indicated that recognition, interaction and adsorption processes occurred reversibly and the polymers had excellent stability and acceptable reusability. So, it could be used for practical applications.

Reusability of SBA-15@MT-MIPs
The stability of SBA-15@MT-MIPs was evaluated by comparing the adsorption capacity towards Rb 1 , Rg 1 and R 1 at each cycle. As shown in Figure 7B, the adsorption capacity decreased slowly with an increase in cycle times. After six adsorption-desorption cycles, the adsorption capacity was about 9.09% less. The slight decrease in adsorption capacity mainly resulted from the partial destruction of the recognition sites during the elution procedure. The results indicated that recognition, interaction and adsorption processes occurred reversibly and the polymers had excellent stability and acceptable reusability. So, it could be used for practical applications.

Optimization of the MISPE Protocols
The washing procedure is a necessary procedure which could minimize the interferences for the analysis step and activate the binding sites of SBA-15@MT-MIPs for maximizing their interaction with the target analytes. Generally, low-polar solvent was preferred as washing solvent such as chloroform and toluene to disrupt the nonspecific binding between the polymers and matrix components [50]. Initially SPE cartridges containing 500 mg of SBA-15@MT-MIPs and SBA-15@MT-NIPs were loaded with 5 mL of 3.0 mmol·L −1 the mixture solution of Rb 1 , Rg 1 and R 1 in blank rat plasmas and then 5 mL of each washing solvent was applied. For optimizing the condition of the washing step, the various washing solvents, such as toluene-ethanol (7:3, v/v), acetonitrile, ethanol, ethanol-water (5:5, v/v) and water were studied. The results were shown in Figure 8. It can be observed that at least 70% of the Rb 1 , Rg 1 and R 1 were washed off from the SBA-15@MT-NIPs cartridge after it was washed by above five washing solvents, because there was only physical adsorption for SBA-15@MT-NIPs. However, when using toluene-ethanol (7:3, v/v) as washing solvent, the impurities in samples could be mostly cleaned up and the recoveries of Rb 1 , Rg 1 and R 1 were more than 92.47% in MISPE. Following are the reasons. Firstly, toluene-ethanol (7:3, v/v) can induce the formation of hydrogen bonding between templates (Rb 1 , Rg 1 and R 1 ) and functional monomers and at the same time interfering compounds were easier washed. Secondly, the specific interaction between the templates and SBA-15@MT-MIPs was easily disrupted by highly polar solvents (acetonitrile, ethanol, ethanol-water (5:5, v/v) and water). Thus, toluene-ethanol (7:3, v/v) as the optimal washing solution was chosen for running the subsequent experiments.
reasons. Firstly, toluene-ethanol (7:3, v/v) can induce the formation of hydrogen bonding between templates (Rb1, Rg1 and R1) and functional monomers and at the same time interfering compounds were easier washed. Secondly, the specific interaction between the templates and SBA-15@MT-MIPs was easily disrupted by highly polar solvents (acetonitrile, ethanol, ethanol-water (5:5, v/v) and water). Thus, toluene-ethanol (7:3, v/v) as the optimal washing solution was chosen for running the subsequent experiments. The selection of eluent plays a key role in excellent sensitivity and precision. It was reported that methanol-acetic acid (9:1, v/v) exhibits the good recovery [51,52]. The reason may be that acetic acid can break the formation of hydrogen bonding between template and functional monomer and template desorption was easier [50]. Thus, we applied 5 mL of methanol-acetic acid (9:1, v/v) as eluent agent.

Validation Assay
Under optimized conditions of MISPE coupled with HPLC, the linearity, the detection range (LODs) and limit (LOQs) were investigated. The results were shown in Table 3. Calibration curves were established for R1, Rg1 and Rb1 in rat spiked plasmas over a concentration range of 2.3-47.3 ng·mL −1 with benign regression coefficients R 2 > 0.995. The LODs and LOQs were determined based on the signal to noise (S/N) ratio of 3 and 10, respectively. The LODs and LOQs of the proposed The selection of eluent plays a key role in excellent sensitivity and precision. It was reported that methanol-acetic acid (9:1, v/v) exhibits the good recovery [51,52]. The reason may be that acetic acid can break the formation of hydrogen bonding between template and functional monomer and template desorption was easier [50]. Thus, we applied 5 mL of methanol-acetic acid (9:1, v/v) as eluent agent.

Validation Assay
Under optimized conditions of MISPE coupled with HPLC, the linearity, the detection range (LODs) and limit (LOQs) were investigated. The results were shown in Table 3. Calibration curves were established for R 1 , Rg 1 and Rb 1 in rat spiked plasmas over a concentration range of 2.3-47.3 ng·mL −1 with benign regression coefficients R 2 > 0.995. The LODs and LOQs were determined based on the signal to noise (S/N) ratio of 3 and 10, respectively. The LODs and LOQs of the proposed method for R 1 , Rg 1 and Rb 1 were in the range of 0.63-0.75 ng·mL −1 and 2.1-2.5 ng·mL −1 , respectively. The relative standard deviations (RSDs) for R 1 , Rg 1 and Rb 1 were 2.5%, 3.0% and 2.2%, respectively, which was repeated measurement for five times with 10 ng·mL −1 of R 1 and 20 ng·mL −1 of Rg 1 and Rb 1 , respectively. It was demonstrated the developed method had a satisfactory reproducibility. The accuracy and practicability of the method were verified by spiked recovery and summarized in Table 4. Since no Rb 1 , Rg 1 and R 1 in the collected the rat plasma samples of was detectable by the proposed method, spiked recoveries were carried out of the samples spiked with 2.5-16 ng·mL −1 R 1 , Rg 1 and Rb 1 to evaluate the developed method. As can be seen, the average recoveries at the middle concentration of the developed method for R 1 , Rg 1 and Rb 1 were 101.4%, 93.4% and 94.3%, respectively, with RSD < 4.2%. The results indicated a suitable improvement in the recovery and accuracy even at the low concentration was found and suggested that SBA-15@MT-MIPs can directly be used for selective adsorption and detection of R 1 , Rg 1 and Rb 1 in biological samples. In addition, we had compared the prepared MIPs as SPE sorbents with commercial C 18 column and polydimethylsiloxane stir bar (PDMS-SB) to validate the absorptive recovery of the prepared MIPs in our initial studies [53,54], which showed that the recoveries of the targets on MIPs were higher than that of commercial column. Therefore, SBA-15@MT-MIPs can be directly used for selective separation and determination of trace saponins in biological samples.

in Real Samples
To evaluate the enrichment and impurity-removing property of the MISPE column and whether the method is suitable for low concentration detection, the proposed method was applied to the analysis of Rb 1 , Rg 1 and R 1 in rat plasma samples and the results were shown in Figure 9. As shown in Figure 9(a), because of extremely low concentration, the analytes were hardly detectable without performing the separation and enrichment of process. Most of the matrix components only with few analytes were eliminated efficiently after being separated by MISPE column (Figure 9(b)). As shown in Figure 9(c), R 1 , Rb 1 and Rg 1 were remarkably concentrated and the interfering peaks arising from complex biological matrices were suppressed, indicating the remarkable preconcentration ability and desired specific recognition of the novel MISPE sorbents to R 1 , Rb 1 and Rg 1 . However, Figure 9(d) showed that most of the analytes and matrix components were washed out from the non-imprinted polymer solid-phase extraction (NISPE) column and few analytes were detected in the elution solution obtained from NISPE (Figure 9(e)). In addition, the amounts of R 1 , Rb 1 and Rg 1 that were detected from the rat plasma samples were 2.5 ng·mL −1 , 4.0 ng·mL −1 and 4.6 ng·mL −1 , respectively. Finally, the target analytes in the elution solution obtained from MISPE sorbents have been structurally identified based on HPLC-MS and the experimental results were listed in Table 5.  [57] As could be observed from the chromatograms, the sensitivities of Rb 1 , Rg 1 and R 1 in plasma samples were greatly enhanced with the MISPE coupled with HPLC-UV analysis. This process did not require special instrumentation, consumes much less toxic organic solvent and has a good clean-up and concentration effect for the analytes. These results demonstrated that the MISPE coupled with HPLC offers a suitable method for selective separation and determination of trace saponins in biological samples.   As could be observed from the chromatograms, the sensitivities of Rb1, Rg1 and R1 in plasma samples were greatly enhanced with the MISPE coupled with HPLC-UV analysis. This process did not require special instrumentation, consumes much less toxic organic solvent and has a good cleanup and concentration effect for the analytes. These results demonstrated that the MISPE coupled with HPLC offers a suitable method for selective separation and determination of trace saponins in biological samples.

Conclusions
In the present study, the innovative multi-template imprinted polymers based on SBA-15 have been prepared by combining the merits of surface imprinting technique and mesoporous material. The obtained SBA-15@MT-MIPs exhibited good characteristics such as excellent adsorption capacity, fast equilibrium kinetics and favorable selectivity towards Rb1, Rg1 and R1. Moreover, low LOD and good recovery were obtained, indicating the great applicability for selective separation of specific analytes from complex samples. Finally, the reliable and simple approach has been successfully employed for the efficient remove and quantitative enrichment of the trace level of PNS from plasma samples simultaneously in a single run. In conclusion, the developed method will find wider applications in diverse clinical and toxicological laboratories for the routine analysis of PNS in biological samples.

Conclusions
In the present study, the innovative multi-template imprinted polymers based on SBA-15 have been prepared by combining the merits of surface imprinting technique and mesoporous material. The obtained SBA-15@MT-MIPs exhibited good characteristics such as excellent adsorption capacity, fast equilibrium kinetics and favorable selectivity towards Rb 1 , Rg 1 and R 1 . Moreover, low LOD and good recovery were obtained, indicating the great applicability for selective separation of specific analytes from complex samples. Finally, the reliable and simple approach has been successfully employed for the efficient remove and quantitative enrichment of the trace level of PNS from plasma samples simultaneously in a single run. In conclusion, the developed method will find wider applications in diverse clinical and toxicological laboratories for the routine analysis of PNS in biological samples.