The samples of LS, OJ, and LA used in the composite were harvested and purchased in Namwon City (Korea). After drying at 40℃ for 72 h with a dryer (LD9013; L'EQUIP Co., Korea), each herb was pulverized with a blender (HMF-3600TG; Hanil Electric, Korea). A total of 2 L of 50% ethanol per 100 g was used for extraction at 80℃ for 4 h, and the extract was filtered twice with nonwoven fabric and filter paper. The filtered extract used as a sample first was concentrated at 50℃ with a vacuum concentrator (R-100; BUCHI, Switzerland), then freeze-dried with a freeze dryer (MCFD8508; Ilshin Biobase Co., Korea), which resulting in a 17% extraction yield. The three herbs used in the experiment were authenticated by Professor Dae Keun Kim of the Woosuk University's Department of Pharmacy.

Standard CGA, ECG, and RMA (Biopurify Phytochemicals, China) with more than 98% purity were utilized for this study. In addition, the methanol and acetonitrile (J.T. Baker, Phillipsburg, USA) used in this study are HPLC-grade organic solvents, and the acid used was formic acid (Fluka, Germany).

The DPPH radical scavenging activity was measured by using a modified Blois method (Blois, 1958). After adding 247.5 μL of a 0.2 mM DPPH (Sigma-Aldrich, USA) solution to 2.5 μL of the sample, a reaction occurs in the dark at room temperature for 20 min. Measurements were taken at 517 nm using an ELISA reader (Bio-Rad, USA) after the reaction occurs. The sample's concentration when the absorbance of DPPH decreased by 50% (RC₅₀) was determined, and each sample was tested three times. L-ascorbic acid (Sigma-Aldrich) was the positive control (Ham et al., 2018).

For HPLC analysis, 20 mg of an equal composite of the three extracts was taken and ultrasonically extracted with 1 mL of 70% methanol for 30 min before being filtered through a 0.2 µm membrane filter (PALL Co., USA). The marker component of the standard product was also obtained similarly.

For the marker components' qualitative and quantitative analyses, a Hitachi Chromaster HPLC CM5000 system (Hitachi, Japan) was used, and detection was performed at multiple wavelengths using a diode array detector (DAD). YMC-Pack ODS-AM, C18 column (250×4.6 mm i.d; YMC, Japan) was used at 40℃. Its mobile phase was composed of water (A) containing 0.1% formic acid and acetonitrile (B) containing 0.1% formic acid with a flow rate of 1.0 mL/min. The slope of the mobile phase was 0-10 min, (A): 100%, 10-55 min, (A): 60%, 55-60 min, (A): 0%, 60-65 min, (A): 0%, 65-70 min, (A): 100%. Detection wavelengths were set at 213 nm, 326 nm, and 329 nm, and HPLC data analysis was performed using Agilent Open Lab Software (Agilent, CA).

The methodology was veridated by determining the linearity, range, LOD, LOQ, accuracy, precision, and content evaluation based on the Guidelines for Validation of Pharmaceutical Drugs from the Ministry of Food and Drug Safety (NIFDS, 2015).

Linearity was evaluated by preparing a calibration curve and calculating a regression equation based on the results obtained after diluting the marker compounds a three different concentrations (1, 50, and 100 µg/mL).

After preparing a composite of the three plant extracts and a standard mixture of each indicator component, the chromatogram was visually evaluated. The ultraviolet (UV) spectrum of the corresponding indicator component was assessed by extracting the standard mixture's chromatogram for comparison.

Consequently, both accuracy and precision were measured by comparing the integral values of the marker compounds' peaks. The accuracy was evaluated by adding three different concentrations of the standard product to the sample before measuring the recovery rate. In comparison, precision was determined using relative standard deviation (RSD) through intraday and interday tests. Each experiment was evaluated by calculating the results' RSDs obtained through three repeated experiments with three concentrations of mixed standard solutions.

The LOD and LOQ were calculated using the following formula (Chanchal & Swarnlata, 2008) based on the prepared calibration curve, where σ is the standard deviation of the response and S is the slope of the calibration curve.

LOD=3.3 (σ/S)

LOQ=10 (σ/S)

Based on the calibration curve obtained from the standard product analysis of the three marker compounds, each marker compound's area value was substituted with the linearity equation to calculate the marker compound's content.

Table 1 shows the sample's concentration required for 50% DPPH radical scavenging, which identifies the measured DPPH scavenging activities of the EtOH extract of LS, OJ, LA, and the composite. The OJ showed the highest DPPH scavenging activity of 48.21±4.0 µg/mL and its combination with LA showed RC₅₀ at 58.06±1.3 µg/mL. Meanwhile, the LOL composite showed RC₅₀ at 67.73±2.6 µg/mL, and the positive control L-ascorbic acid showed RC₅₀ at 19.81 µg/mL. Choi et al. (2008) reported that the MeOH extract of OJ could donate electrons, act as a reducing agent, and demonstrate the superoxide dismutase-like activity, suggesting that harvesting around August to October contributed to high anti-oxidative activityies. Im et al. (2017) reported that EtOH extracts of OJ inhibited the enzyme activities of collagenase, elastase, and tyrosinase by 85%, 45%, and 70%, respectively, at a concentration of 500 µg/mL. Meanwhile, Kim et al. (2012a) reported that the ethyl acetate fraction of LS had RC₅₀ values of 280 µg/mL for DPPH radical scavenging, suggesting that LS can act as a preservative for food, medicine, or cosmetics because it had excellent antibacterial properties against Bacillus cereus. Meanwhile, LA is used as a cosmetic ingredient because it possesses anti-aging, antioxidative, and inhibitory properties on tyrosinase activity (Hsu et al., 2007). The three plants' phenolic compounds had a high radical scavenging activity (Kadoma & Fujisawa, 2008), and the composite, as a cosmetic formulation, could prevent the overproduction of melanin through its anti-oxidative properties.

The study was conducted to predict several components to determine the marker compounds of EtOH extracts of LS, OJ, and LA. CGA, ECG, and RMA were selected, which were detected with higher content from the standard calibration curves obtained through liquid chromatography/mass spectrometry (LC/MS) analysis (Ham et al., 2018). The structures of these compounds are shown in Figure 1. Three marker compounds were analyzed with water containing 0.1% formic acid and acetonitrile made up of 0.1% formic acid through HPLC-DAD.

The marker compounds demonstrated showed high absorbance at 213 nm, 326 nm, and 329 nm on the HPLC-DAD analysis's UV absorbance spectrum (Figure 3). In Figure 2A, a mixture of the standards' chromatogram showed that the CGA, ECG, and RMA peaks showed different retention times (27.78 min, 34.50 min, and 39.23 min, respectively) without interference. These results demonstrated that the conditions of simultaneous analysis of three marker compounds were established.

As shown in Figure 2B, the LOL composite's chromatogram showed that the retention times of the three marker components' peaks from LOL were identical to the retention times of the standard materials. This result determined that CGA, ECG, and RMA were present in the LOL composite because its peaks on the UV absorbance spectrum are identical to the UV spectrum of standard products (Figure 3). Meanwhile, the chromatograms at 213 nm showed an increase in peak intensity with increasing absorbance of ECG, and the chromatograms at 326 nm and 329 nm revealed an increase in peak intensity with increasing CGA and RMA absorbance. The results confirmed that each marker component's peak increased on the LOL chromatogram when each standard material was injected with LOL as an internal standard (data not shown). Figure 2B also shows several peaks beside the marker components.

Linear regression analysis of each marker compound was evaluated using R², range, LOD, and LOQ (Table 2). All three marker compounds exhibited linearity with a correlation coefficient of R²>0.9997 in a 1-100 µg/mL concentration range. The LOD of CGA, ECG, and RMA were determined to be 0.83 µg/mL, 0.33 µg/mL, and 0.20 µg/mL, respectively, and the LOQ range of CGA, ECG, and RMA were 2.51 µg/mL, 0.99 µg/mL, and 0.61 µg/mL, respectively.

As shown in Table 3, the recovery test results measured after adding three concentrations (low, medium, and high) of each marker standard to the extracts, showed that the recovery of CGA was 102.00%-117.78%, and the relative standard deviation ranged 0.21%-1.38%. The ECG recovery rate was 94.40%-108.06%, and its relative standard deviation was around 0.84%-1.95%. Meanwhile, RMA showed a recovery rate between 93.79%-107.05% and a relative standard deviation of 0.66%-1.41%. The recovery data of all three compounds were between 93.79% and 117.78% and had RSD values less than 1.95%, demonstrating this method's reliability under established conditions.

The method's accuracy and precision were evaluated using triplicate analysis once a day over three days, and intraday accuracy of CGA, ECG, and RMA was 107.06%, 102.25%, and 92.24%, respectively. Conversely, inter-day accuracy was 110.39%, 104.03%, and 89.94% for CGA, ECG, and RMA, respectively. The recovery rates of all three marker compounds were between 89.94% and 110.39%, which are within 20% of the reference values of the Ministry of Food and Drug Safety's guidelines. All three markers had a range of 0.80%-1.77% for intraday RSD, and their inter-day RSD had a range of 0.72%-1.42%. These results reflect the guidelines' RSD of less than 5% (Table 4). Table 5 outlines the results of the quantitative analysis of the LOL composite's marker components, where CGA content was 13.99 µg/mg (1.4%), ECG was 8.28 µg/mg (0.8%), and RMA was 8.40 µg/mg (0.8%) with 0.76%-1.54% RSD. This validation method can control the quality of the cosmetic ingredients, such as LS, OJ, and LA, registered on INCI.