Arbutin, dimethyl sulfoxide, L-tyrosine, and mushroom tyrosinase were provided by Sigma-Aldrich (St. Louis, MO, United States). Tricaine, embryo culture medium, methylcellulose, and lytic buffer were supplied by Shandong Yixiyue Biotechnology Co., Ltd. (Weifang, China). Phosphate-buffered saline (PBS; pH 7.2) was obtained from Gibco (Waltham, MA, United States). Dipotassium hydrogen phosphate, potassium dihydrogen phosphate, and NaOH were acquired from Aladdin (Shanghai, China). All the organic solvents were supplied by Concord Technology (Tianjin, China). A Milli-Q water purification system (Millipore, Billerica, MA, United States) was used to obtain ultrapure water (18.2 MΩ cm), which was used for all the experiments.

Dryopteris crassirhizoma rhizome (voucher WLL-2018-03) was provided by Yiyuan Chinese Herb Medicine Co., Ltd. (Anguo, Hebei, China) in August 2018 and stored at the College of Public Health, Hebei University. Dried ground DCR (10 g) was extracted in triplicate with 100 ml of a 90% methanol aqueous solution over 24 h, and 1.59 g of solid extract powder was obtained after filtration, concentration, and lyophilization.

Mixtures of DCR extract samples with different concentrations, L-tyrosine (0.3 mM), and mushroom tyrosinase (50 units/ml) were incubated in PBS at 25°C, and their absorbances were recorded at 450 nm after 30 min. Arbutin, which was the positive control, was used as the reference compound. The tyrosinase inhibition values were calculated as follows:

where A is the absorbance of the reaction system comprising the DCR extract sample and tyrosinase, B is the absorbance of the reaction system comprising only the DCR extract sample, C is the absorbance of the reaction system comprising only tyrosinase, and D is the absorbance of the reaction mixture without DCR extract samples and tyrosinase.

A pair of healthy broodstock zebrafish was placed in an incubator equipped with a barrier, which was used to separate the male and female zebrafish overnight prior to mating. The barrier was removed the next morning, and light was used to stimulate the broodstock zebrafish to spawn. Then, 4 h post-fertilization (hpf), the zebrafish embryos were cultured with DCR (10, 50, and 100 μg/ml in embryo culture medium), arbutin (10, 50, and 100 μg/ml in embryo culture medium), or embryo culture medium in 96-well plates at 28°C under 14/10 h light/dark cycles. Each group comprised 60 embryos, and the corresponding culture media were changed daily. A stereomicroscope was used to record the melanin changes in the zebrafish embryos 72 hpf.

After treatment, 50 zebrafish embryos from each group were homogenized in the PBS using a tissue homogenizer, and the homogenates were centrifuged. The supernatant of each group was collected and used to determine the tyrosinase activity of the group. After protein concentration was normalized, the diluted supernatants containing tyrosinase (160 μl) and L-tyrosine solution (40 μl) were mixed and incubated at 37°C. After 1 h, 450 nm UV light was used to determine the absorbance of each mixture, and the tyrosinase inhibition values were calculated as follows:

where A is the absorbance of L-tyrosine incubated with the supernatant obtained from the zebrafish embryos treated with DCR or arbutin and A₀ is the absorbance of L-tyrosine incubated with the supernatant obtained from the zebrafish embryos treated with embryo culture medium.

To assess the melanin content of each group of zebrafish, the melanin precipitate from each group was collected and suspended in 400 μl NaOH (1 M). Next, 490 nm UV light was used to evaluate the dissolved melanin content, and the relative melanin content was calculated as follows:

where A is the absorbance of the melanin samples obtained from the zebrafish embryos treated with DCR or arbutin and A₀ is the absorbance of the melanin samples obtained from the zebrafish embryos treated with embryo culture medium only.

A TBE-300C HSCCC instrument (Tauto Biotechnique Company, Shanghai, China) equipped with an Isolera FLASH purification system (Biotage, Uppsala, Sweden) was used for separation. Before separation, 300 ml of the stationary phase was pumped into the coiled column and subsequently rotated at 400 rpm. After equilibrium was reached, 20 ml of the DCR extract solution was injected, and a series of mobile phases was delivered into the coil in the tail-to-head mode. A mixture of n-butanol (n-BuOH) and water [1:10 (v/v)] was used as the stationary phase. The injected DCR extract was dissolved in a mixture of ethyl acetate (EtOAc), n-BuOH, and water [1:1:10, (v/v/v)] to achieve a concentration of 25 mg/ml. A list of mobile phases is presented in Figure 6. Chromatograms were recorded at a wavelength of 254 nm.

A mixture of DCR extract solution (0.1 mg/ml) and tyrosinase (100 units/ml) was incubated at 37°C for 30 min and subsequently centrifuged using a Microcon YM-10 centrifugal filter unit (Millipore, Billerica, MA, United States) at 10,000 × g for 30 min. A control group without tyrosinase was analyzed in parallel. The ultrafiltrates were collected for subsequent LC–MS/MS analysis.

An UltiMate 3000 high-performance liquid chromatography (HPLC) system (Thermo Fisher Scientific, Waltham, MA, United States) equipped with a Q Exactive Orbitrap mass spectrometer (Thermo Fisher Scientific, Waltham, MA, United States) was used for LC–MS/MS analysis. An Eclipse SB-C18 Rapid Resolution column (150 mm length, 4.6 mm ID, and 3.5 μm particle size; Agilent, Santa Clara, CA, United States) was used for separation. The sample’s injection volume and flow rate were 10 μl and 1 mg/ml, respectively. The elution gradient program was as follows: 0–8 min, 5–20% B; 8–25 min, 20–30% B; 25–60 min, 30–100% B; 60–67 min, 100% B; 67–70 min, 100–5% B; 70–80 min, 5% B; and the mobile phases comprised 0.1% formic acid in water (A) and methanol (B). The eluent was monitored at 254 nm. Peak identification was performed in the negative mode, and the electrospray ionization source conditions were set as follows: sheath gas flow rate: 45 arb, auxiliary gas flow rate: 15 arb, capillary temperature: 320°C, full mass resolution: 70,000, MS/MS resolution: 17,500, collision energy: 20/40/60 eV in the normalized collision energy model, and spray voltage: −3.1kV.

In silico docking was performed using the Surflex-Dock program version (Tripos, St. Louis, MO, United States), and the crystal structure of tyrosinase (2Y9X) was retrieved from the RCSB Protein Data Bank. Prior to docking, the macromolecules and small molecules were prepared, including structural file retrieval, water molecule removal, non-protein atom removal, structural defect correction, and molecular energy minimization. The “thresh” and “bloat” parameters of the docking protocol were set to be 0.5 and 1, respectively. The PyMOL (Schrödinger, New York, NY, United States) and LigPlot software (EMBL-EBI, Cambridge, United Kingdom) were used for data visualization.

All experiments were performed at least in triplicate, and the results are expressed as means ± standard deviations (SDs). Data analysis was performed using the SPSS software (IBM, Armonk, NY, United States). The mean values were compared using Student’s unpaired t-test or one-way analysis of variance (ANOVA), and statistical significance was set at p < 0.05.

The inhibitory effects of various concentrations of the DCR extracts on mushroom tyrosinase were tested in vitro. The DCR extracts considerably inhibited mushroom tyrosinase, and the inhibitions of the samples with DCR extract concentrations of 0.1, 0.5, and 1 mg/ml were 14.32, 53.61, and 75.47%, respectively (Figure 1). The half-maximal inhibitory concentration (IC₅₀) of the DCR extract for tyrosinase was 401 μg/ml. Experiments were performed using arbutin as the reference compound. The inhibitions of arbutin samples with concentrations of 0.1, 0.5, and 1 mg/ml were 56.03, 88.90, and 96.45%, respectively, and the IC₅₀ of arbutin was 86.9 μg/ml. The tyrosinase inhibitory activity of DCR was considerable yet lower than that of arbutin, as confirmed by the aforementioned experimental results and previously reported data (11, 12).

Mushroom tyrosinase has been commonly used to assess the tyrosinase inhibitory activity related to the melanogenesis inhibition of traditional medicinal plants and food items (19). However, the effect of DCR on melanogenesis in vivo remains unclear. Zebrafish embryos present significant physiological and genetic similarities with mammals. Moreover, pigmentation experiments using zebrafish require simple protocols (20). Therefore, zebrafish embryos have been increasingly used as in vivo test models, replacing mice and other animal models for phenotype-based melanogenesis inhibition studies. Consequently, the effect of DCR on melanogenesis inhibition was investigated using zebrafish embryos. The pigmentation of zebrafish larvae exposed to DCR or arbutin at 72 hpf decreased in a dose-dependent manner and was lighter than that of the control larvae. The melanin contents of the zebrafish larvae treated with DCR extracts with concentrations of 10, 50, and 100 μg/ml were 92.95, 64.34, and 42.85%, respectively, of the melanin content of the control group, whereas those of the zebrafish larvae treated with the arbutin concentrations of 10, 50, and 100 μg/ml were 77.73, 53.47, and 29.63%, respectively, of the melanin content of the control group (Figure 2). Moreover, no abnormalities were observed in the morphologies of the DCR- and arbutin-treated zebrafish larvae. These results suggested that the DCR extract inhibited melanogenesis in vivo.

To further determine whether melanogenesis inhibition in zebrafish embryos correlated with tyrosinase inhibition, the tyrosinase activity in zebrafish embryos was investigated. Treatment with DCR and arbutin considerably decreased the tyrosinase activity in zebrafish embryos at 72 hpf. The tyrosinase activities of the zebrafish embryos treated with DCR extracts at the concentrations of 10, 50, and 100 μg/ml were 90.34, 69.08, and 4.51%, respectively, of the tyrosinase activity of the control group, whereas those of the zebrafish treated with the arbutin concentrations of 10, 50, and 100 μg/ml were 72.14, 67.79, and 61.84%, respectively, of the tyrosinase activity of the control group (Figure 3). These results suggested that the DCR extract inhibited melanin synthesis via tyrosinase inhibition, thereby lowering pigmentation; moreover, these findings indicated that the DCR extract contained tyrosinase inhibitors.

To characterize the composition of the biologically active components of plant extracts, isolation is typically performed prior to evaluation or application. However, isolating biologically active components from DCR extracts is challenging because of the complex chemical composition of DCR (Figure 4). Alternatively, to characterize the tyrosinase inhibitors in DCR, an off-line hyphenation method combining HSCCC, affinity-based ultrafiltration, and LC–MS/MS was used in this study. A stepwise HSCCC was initially performed to enrich the minor components in DCR, and then, affinity-based ultrafiltration was conducted to identify potential tyrosinase inhibitors in DCR. Lastly, the structures of the potential tyrosinase inhibitors in DCR were determined using LC–MS/MS (Figure 5).

An appropriate liquid phase comprising two immiscible liquids that satisfy the golden rules proposed by Ito is typically required for HSCCC separation (21). The HSCCC procedure was adopted because assessing the numerous solvent systems and comparing the partition coefficients of the components using HPLC or thin-layer chromatography is time-consuming. Moreover, the separation of compounds with similar partition coefficients is restricted by the low resolution of HSCCC. Stepwise HSCCC separation was performed to maximize the enrichment of the minor components in the primary components of the complex extracts and simplify solvent system selection. In this study, nine mobile phases of the n-hexane–EtOAc–n-BuOH–water solvent system with a wide polarity range were used without calculating the partition coefficients of the components (Figure 6). After mobile phase elution, the residual stationary phase in the coil was washed with methanol. Lastly, seven fractions were obtained, and their HPLC profiles are presented in Figure 7. The experimental data indicated that the primary components in DCR were collected in fractions 1 and 2, whereas the minor components were enriched in fractions 3–7.

To characterize the tyrosinase inhibitors in DCR, the obtained HSCCC fractions were further analyzed via ultrafiltration combined with LC–MS/MS (22). The HPLC peak areas of the 22 compounds separated from DCR decreased, indicating their tyrosinase-inhibiting potentials. The structures of the 22 compounds, namely, 4-methyl-2-oxovaleric acid (1), 4-aminosalicylic acid (2), harmane (3), 2,3-dihydroxybenzoic acid (4), catechol (5), 2-hydroxyhippurate (6), 5,7-dihydroxy-2-(4-methoxyphenyl)-6-(3-methylbut-2-enyl)-2,3-dihydrochromen-4-one (7), salicylic acid (8), isobiflorin (9), biflorin, (10), (E)-1-(2-azido-3-nitrophenyl)-N-[(E)-(2-azido-3-nitrophenyl)methylideneamino]methanimine (11), 2-oxo-7-[(2S,3R,4S,5S,6R)-3,4,5-trihydroxy-6-(hydroxymethyl) oxan-2-yl]oxychromene-3-carboxylic acid (12), 7-hydroxy-6-methoxychromen-2-one (13), asterric acid (14), oryzalin (15), 3-(2,3-dihydroxy-5-methylphenoxy)-5-methylbenzene-1,2-diol (16), 4-methylumbelliferone (17), quercetin-3-O-glucoside (18), kaempferol 7-O-glucoside (19), kaempferol 3-O-glucoside (20), kaempferol 3-O-glucuronide (21), and dryopteroside (22), which are illustrated in Figure 7, were assigned by matching the precursors and MS/MS m/z values of the compounds with those compiled in libraries and comparing the data with references (Table 1 and Figure 8).

The isolation of minor and trace compounds from complex extracts is laborious (23). Therefore, molecular docking analysis was performed to rapidly confirm the binding mechanisms of the identified compounds to tyrosinase, and the results are presented in Table 2 and Supplementary Figures 1–23. Each of the binuclear copper atoms of tyrosinase is bonded to three histidine residues and plays a critical role in catalytic reactions (24). We hypothesized that the interactions between 3, 5, 6, 7, and 8 and the amino acid residues of tyrosinase were hydrophobic, whereas the other 17 compounds interacted with tyrosinase via hydrogen bonding and hydrophobic interactions. In particular, 4, 11, 15, and 16 formed hydrogen bonds with the amino acid residues of tyrosinase and did not interact with the copper atoms; 1, 2, 13, 17, and 21 formed hydrogen bonds with one copper atom and several amino acid residues; and 9, 10, 12, 14, 18–20, and 22 formed hydrogen bonds with two copper atoms and several amino acid residues. The binding stabilities toward and affinities for tyrosinase of the 22 compounds were assessed using crash, polar, chem, G-, D-, potential mean force, and C-scores (Table 2). Specifically, the crash scores reflected the incorrect penetration of the ligand in the active site pocket of tyrosinase; the polar scores reflected the ligand region; the D-scores were calculated using the charges and van der Waals interactions between proteins and ligands; the potential mean force scores indicated the Helmholtz free energies for the protein–ligand atom interactions; the G-scores were derived by evaluating the internal energies of hydrogen-bonding, protein–ligand, and ligand–ligand interactions; and the chem scores described the points where hydrogen-bonding, lipophilic contact, and rotational entropy changed. Each scoring method was used for different purposes, and the individual scores could not comprehensively evaluate the ligand–tyrosinase interactions. Therefore, the C-scores were calculated by combining the crash, polar, chem, G-, D-, and potential mean force scores, and the results were used to comprehensively assess the ligand affinity (25). The C-scores decreased as follows: 22 (6.88) > 19 (6.79) > 10 (6.72) > 16 (6.54) > 6 (5.87) > 9 (5.05) > 21 (5.05) > 12 (4.89) > 14 (4.76) > 13 (4.72) > 1 (4.64) > 2 (4.51) > 20 (4.49) > 17 (4.46) > 11 (4.32) > 18 (4.19) > 5 (4.11) > 7 (4.08) > 15 (3.86) > 3 (3.82) > 8 (3.25) > 4 (3.06). Arbutin, which is a known tyrosinase inhibitor, forms hydrogen bonds with one of the copper atoms and the His61, His94, Glu256, Asn260, and His296 residues of tyrosinase, and its C-score was 4.83. These results indicate that the 22 compounds formed stable conformations in the active site pocket of tyrosinase. Among the 22 compounds, 8 and 21 have been reported as tyrosinase inhibitors (26, 27), whereas 4 and 5 have been reported as tyrosinase substrates (28, 29). The rest of the 22 compounds have not been reported to interact with tyrosinase. Kubo et al. (30) reported that 18 and 20 were not tyrosinase inhibitors or substrates. However, Ohguchi et al. (31) demonstrated that 18 considerably inhibited the production of melanin in mouse B16 melanoma cells by suppressing tyrosinase expression. The inhibitory effects of the 22 compounds will be investigated in a future study.