Matrine (purity 99.9%) and oxymatrine (purity 94.3%) were purchased from ChromaDex (Irvine, California, USA). High performance liquid chromatography (HPLC) grade methyl alcohol (MeOH), acetonitrile, and ace- tone were purchased from Tedia (Ohio, USA). Hydrophilic lipophilic balanced (HLB) SPE (60 mg, 3 mL) from Waters (Milford, USA) were used, and six carbonaceous SPE cartridges were purchased; ENVI-Carb SPE car- tridge (0.25 g, 3 mL and 0.5 g, 6 mL) from Supelco (Philadelphia, USA), activated carbon (0.5 g, 6 mL) from Silycycle (Quebec, Canada), Carbograph SPE tubes (0.5 g, 8 mL and 1 g, 15 mL) from Chrom Tech (Minnesota, USA), and Bond Elut Carbon (0.5 g, 6 mL) from Agilent (California, USA). C₁₈ SPE (0.5 g, 6 mL) was from Phenomenex (California, USA). Twenty one commercial biopesticides containing S. flavescens extract were purchased from 14 Korean local companies.

The HLB clean-up procedure as reported by Lee et al. was followed [33] . The HLB cartridge was first condi- tioned with 2 mL MeOH and 2 mL distilled water (DW), then the sample was loaded and washed with 5% aqueous MeOH (2 mL). The cartridge was dried by vacuum, and eluted with 5 mL MeOH. The eluate was eva- porated and re-dissolved with 1 mL MeOH for UPLC analysis.

In order to compare the recovery rate of matrine and oxymatrine, the six carbonaceous SPE were tested. Each cartridge was placed onto female-luer port of the SPE vacuum manifold and conditioned with 3 mL of MeOH and 3 mL of DW. A portion of 1 mL aqueous solution of matrine (10 mg·L⁻¹), oxymatrine (10 mg·L⁻¹), and the biopesticides diluted 10 times with DW, were loaded each into conditioned cartridges, respectively, and were eluted with MeOH (3 Í 3 mL) after washing/non-washing with 12 mL of DW. Eluate evaporated slowly under reduced pressure, and the dried residues were re-dissolved into 1 mL of MeOH for analysis. UPLC system (Wa- ters, MA, USA) with UV detector was used for matrine and oxymatrine analysis.

ENVI-Carb SPE cartridge (0.5 g) was selected according to the result of Section 2.2. The second clean-up procedure was conducted with C₁₈ SPE cartridge (0.5 g) to remove unknown impurities for reduce the burden on UPLC. C₁₈ SPE cartridge was placed to the bottom of the ENVI-Carb cartridge after DW washing as men- tioned in Section 2.2. The eluate of ENVI-Carb was fully loaded to the C₁₈ cartridge and the eluate was col- lected. Then the C₁₈ cartridge was additionally eluted with 9 mL of MeOH. The second clean-up efficiency was confirmed with absorbance of eluate from 210 to 800 nm using UV-Vis spectrophotometer (Lambda 650, Perkin-Elmer, London, UK).

Matrine and oxymatrine were analyzed on a Waters UPLC system. One microliter of a sample was introduced to a Waters Acquity BEH Phenyl column (1.7 mm particle size, 3.0 × 100 mm, Waters, Boston, USA). The column was maintained at a temperature of 40˚C and a constant flow rate of 0.5 mL·min⁻¹. The UPLC was operated us- ing a gradient mobile phase of solvent A (0.05% formic acid in DW) and B (acetonitrile). The gradient was 0% - 10% B (0 - 10 min), 10% - 50% B (10 - 15 min), and 50% - 100% B (15 - 17 min). The absorbance of matrine and oxymatrine was measured at 217 nm.

Sample preparation and analytical methods were validated in terms of linearity, limit of quantitation (LOQ) and recovery. The linearities of the calibration curve for matrine and oxymatrine were obtained from the concentra- tions 1, 5, 10, 25 and 50 mg·L⁻¹, respectively. LOQs for matrine and oxymatrine were calculated from the con- centration that produced a signal-to-noise ratio of 10. Recoveries of matrine and oxymatrine were tested by spiking the standard material at a concentration of 10 mg·L⁻¹ in the extract of neem and its biopesticide that were not containing the analytes. The precision of the method was expressed in percentage (%) of relative stan- dard deviation (RSD).

Recently, Lee et al. reported the analytical method using HLB SPE and UPLC for the quantitative analysis of bioactive limonoids in the biopesticides containing neem extract [33] . First, the HLB clean-up method was ap- plied to the purification method for quantitation of alkaloidal matrines, which were spiked at a concentration of 10 mg·L⁻¹ in the biopesticide containing neem extract. However, the recovery was below 60% on the standard spiked water sample. To find the reason for the low recovery on HLB experiment, the partition coefficients (log P) were compared with azadirachtin which was an indicator of neem extract, matrine, and oxymatrine. The values were calculated from ALOGPS v2.1 that was provided by the Virtual Computational Chemistry Laboratory [34] . The calculated log P values of azadirachtin, matrine and oxymatrine were 1.53, 1.44, and −0.33, respectively. However, matrine which has tertiary amine could be easily protonated and present as charged amine form in water solution due to its pKa [35] . Therefore, the log P would be compared with the protonated matrine, and it was calculated to −1.37 from ALOGPS. These values of oxymatrine and protonated matrine were lower by about 70 and 800 times, respectively, than that of azadirachtin. Therefore, the HLB SPE method would not be suitable for the purification of alkaloidal matrines on the mixture of S. flavescens.

Furthermore, the LC condition had to be optimized to get the best result particularly on good sensitivity for matrine and oxymatrine. Acquity UPLC BEH Phenyl column has been known to provide complementary selec- tivity to C₁₈ stationary phase, especially for aromatic compounds. However, C₁₈ column was used for HPLC analysis of matrine and oxymatrine in previous studies [29] . Acquity UPLC BEH Phenyl and C₁₈ column were tested to compare the sensitivity of matrine and oxymatrine. The sensitivity of BEH phenyl column was a little better than that of C₁₈ column. The optimized analytical condition was achieved using a mobile phase consisting of 0.05% formic acid and acetonitrile on an Acquity UPLC BEH Phenyl column. Figure 1 presents the chroma- togram of matrine and oxymatrine in biopesticide sample. The retention times of both materials under analysis condition from Section 2.4 were 5.75 and 6.84 min, respectively.

This study tried to find out the suitable SPE for the clean-up of biopesticides containing S. flavescens extract and various supplementary materials, and selected carbonaceous SPE. Carbonaceous SPE can be applied to the aqueous sample unlikely to silica gel or Florisil cartridge. Also, many reports have described the extraction of polar analytes and/or multi-residue extractions such as chloroaniline, chlorophenols, and polar pesticides with graphitized carbons [36] -[39] . The optimum dilution of sample was considered due to the viscosity and crudity of S. flavescens extract and its biopesticides. The dilution rate was applied from 1 to 100 folds with DW in this experiment, and the diluted samples were applied to carbonaceous SPE for the purification of matrines from S. flavescens extract and its biopesticides.

Six carbonaceous cartridges were tested to compare the recovery rate of matrine and oxymatrine with or without washing step. The recoveries of matrine and oxymatrine both with and without DW washing were 78.6% - 131.0% in ENVI-Carb (0.25 and 0.5 g) and Bond Elut Carbon tube (0.5 g), whereas below 48% in the other SPE cartridge (Figure 2). When the purification efficiency was compared with the absorbance at 217 nm on the eluate from ENVI-Carb SPE cartridge, the disturbance was reduced by around 25% after washing with 12 mL DW without recovery changes (Figure 3). In order to find out the optimal dilution rate of biopesticide, the sam- ples were diluted 1, 10, 20, 50, and 100 times with DW and investigated for the recoveries of matrine and oxymatrine after ENVI-Carb clean-up. The recoveries of matrine and oxymatrine of biopesticide without dilution were 68.0% and 73.0%, respectively, but the recoveries of 10, 20, 50, and 100 times diluted samples ranged from 96.3% - 106.5% (data not shown). Finally, ENVI-Carb 0.5 g and dilution rate of 10 times were selected for the clean-up method which showed the best recovery of both matrine and oxymatrine. However, the color of ENVI-Carb eluate of biopesticides was still evident, hence, second purification was considered to diminish the impurities and to reduce the column damages on the instrument.

To improve data reproducibility, the second purification step was taken into consideration, and C₁₈ SPE was adopted to remove non-polar impurities. The C₁₈ SPE and ENVI-Carb SPE were connected sequentially after washing step of the method 2.3, and MeOH and acetonitrile were tested as elution solvents to find the suitable one because the two solvents were shown to have the good recoveries above 95%. The ENVI-Carb and C₁₈ SPE were eluted with 3 × 3 mL MeOH or acetonitrile, then 9 mL MeOH or acetonitrile was additionally eluted to C₁₈ SPE cartridge. The spectra for the eluates of two solvents were measured with UV-visible spectrophotometer, and MeOH showed better impurity removing efficiency than acetonitrile on the comparison of Abs₂₁₇ (Figure 3). And finally, MeOH was accepted as an elution solvent for the SPE clean-up. Second clean-up efficiency by C₁₈ SPE cartridge is presented in Figure 3. The greatest decrease of impurities was shown in clean-up with the combination of ENVI-Carb and C₁₈ SPE including DW washing, and there was no adverse effect to the recovery. Therefore, these results suggest that the clean-up method was accepted for matrine and oxymatrine analysis in biopesticides containing S. flavescens extract (Figure 4).

Figure 2. Recoveries of matrine (a) and oxymatrine (b) in the selected six cartridges with and without DW washing (I: ENVI-Carb 0.25 g, II: ENVI-Carb 0.5 g, III: Activated Carbon 0.5 g, IV: Carbograph 0.5 g, V: Carbograph 1 g, VI: Bond Elut 0.5 g). Vertical bars represent ±SD of the means.

^†The data represent the mean values ± SD of three replicates.

lations from 1 to 50 mg·L⁻¹ for both matrine and oxymatrine were detected with coefficient correlations of 0.9986 and 0.9996, respectively. The method LOQs of matrine and oxymatrine were 0.226 and 0.684 mg·L⁻¹, respectively. Validation of the analytical method was conducted in terms of recovery of the spiked sample that did not contain S. flavescens. Recoveries of 10 mg·L⁻¹ matrine and oxymatrine were found to be 105.0 and 103.6% using the established method. The inter- and intra-day precision of the method were determined from recovery of both materials on three separate days. The method was reliable since the RSD percentage (2.9% - 5.9%) below 15, which was the normal percent value. These results suggest that the experimental procedure including clean-up and instrumental analysis was suitable for use in the analysis of targeted materials in biopesticide samples.

The developed method for matrine and oxymatrine analysis was applied to the extract of S. flavescens and its biopesticides. Matrine and oxymatrine in all samples were detected at 0.00 - 0.66 and Table 2). Recently, Yang and Zhao [18] and Mao and Henderson [20] reported the antifungal and insecticid- al activities of matrines as EC ₅₀ and LD ₅₀ values; the values were 26 - 1133 mg·L ⁻¹ and 10.1 - 13.3 mg

per insect, respectively. If we only consider the concentration of alkaloidal matrines as bioactive materials, the contents of matrine and oxymatrine in some products among the tested biopesticides were assumed to be sufficient as crop protection agents. However, if we consider the practical use guideline of biopesticide, the concentration of the active ingredient would be expected to be lower than EC₅₀ and LD₅₀ level, because the general guideline for field application was suggested be 100 - 1000 folds diluted with water. In a survey of matrines present in biopesticides from S. flavescens, the concentration of bioactive materials showed wide ranges and these results would be related to the crop protection efficiency. Thus, the quality control of the commercial biopesticide should be performed.