NANA was purchased from Sigma-Aldrich (St. Louis, MO) as standard solution. Volume were measured accurately from the stock, diluted with fresh Milli-Q to produce a series of solution standards (1, 2, 5, 10, 15, 20 µM). Tyrosinase from mushroom (³1000 U/mg) and recombinant human tyrosinase (Activity ³ 95%) were purchased from Sigma-Aldrich. L-3, 4-dihydroxyphenyl-alanine (L-DOPA), vitamin C and tert- Butylhydroquinone (tBHQ) and alpha melanocyte stimulating hormone (α-MSH) were purchased from Sigma-Aldrich. All EBN samples and its adulterants were purchased in the market of Hong Kong.

For LC-MS/MS analysis, EBN samples were ground into powder (approximately 1 - 3 mm) and mixed thoroughly. Ten mg of each was weighed and extracted by 1 mL of fresh Milli-Q water under sonication for 10 min. Followed by centrifugation at 14,000 rpm for 5 min, the supernatants were filtered, and the filtrates were collected for LC-MS/MS analysis. For bioassay, one g of EBN sample, or its adulterants, was individually soaked in 100 folds of water for overnight at room temperature. The mixture was then stewed at 80˚C for 6 - 9 hours until completely molten and then left to cool down to room temperature. This method of preparation is a common cooking method for the consumers, as well as a high protein extraction efficiency [11] . The extracts were kept at 4˚C until further usage.

The liquid chromatograph was equipped with an Agilent 6410 Triple Quad MS/MS and a (2.1 × 100 mm) Eclipse XDB-C18 column (3.5 µm particle size). The injection volume was 2 µL. A 5-min linear gradient at flow rates of 0.4 mL/min between solvent A (Milli-Q water, 0.1% formic acid) and solvent B (Acetonitrile, 0.1% formic acid) was used. After reaching 80% B, the system returned to 100% A in 0.5 min. For column equilibration, a total cycle time of 10 min was needed. The MS was operated in negative electron spray ionization mode. A capillary voltage of 3.5 kV and a cone voltage of 10 V were applied. Source temperature was 100˚C, and desolvation temperature was 325˚C. Ultra-high purity nitrogen was used for cone gas (3.0 L/min), desolvation gas (10.0 L/min) and nebulising gas (35 psi). For collision induced dissociation (CID) a collision energy of 5 eV was used. Spectra from m/z 50 to m/z 1000 were recorded.

In calibration of NANA in EBN, a 5-point calibration curve having concentrations of 1, 2, 5, 10 and 20 µM was made. All calibrators were prepared in fresh Milli-Q water. Triplicate results were taken for each sample. Retention time of NANA was 0.65 - 0.69 min. A negatively single charged ion [M-H]⁻ of NANA (m/z 307.9) was selected as precursor ion for CID. The precursor ion was dissociated into two major product ions (m/z 87.0 and 170.0), and the product ion m/z 87.0 was the most abundant from NANA. Multiple reaction monitoring (MRM) was applied, the transitions m/z 307.9 à 87.0 and m/z 307.9 à 170.0 were chosen as the qualifiers, whilst the transitions m/z 307.9 à 87.0 was measured for quantification. The determination of citric acid was similar to that of NANA. The retention time of citric acid was at 1.50 min. The solution having known spiked amount of citric acid was defined as the quality control (QC) solution. Negatively single charged ion [M-H]⁺ of citric acid (m/z 191.0) was selected as precursor ion for CID. The precursor ion was dissociated into two major product ions (m/z 111.0 and 86.9), and the product ion m/z 111.0 was the most abundant from NANA. Multiple reaction monitoring was applied, the transitions m/z 191.0 à 111.0 and m/z 191.0 à 86.9 were chosen as the qualifiers, whilst the transitions m/z 191.0 à 111.0 was measured for quantification. The ethyl β-D-glucuronide (m/z 221.1) was used as the internal standard in the calibrators and QC samples, and the concentration was kept consistent at 100 ng/mL. The transitions m/z 221.1 à 84.9 and m/z 221.1 à 75.0 were chosen as the qualifiers, whilst the transitions m/z 221.1 à 84.9 was quantified as reference.

The enzyme activities of mushroom and human tyrosinase were monitored by dopachrome formation at 492 nm through the oxidation of substrate (L-DOPA). The reaction medium (200 mL) contained 0.5 mM L-DOPA in 50 mM sodium phosphate buffer (pH 6.8) and (pH 7.4). Five mM of vitamin C was used as an inhibitor control. The final concentration of mushroom tyrosinase was 0.2 mg/mL. In this method, 0.1 mL of different concentrations of effectors, including EBN extracts, adulterants of EBN and NANA, were added to the reaction medium. The reaction mixtures were loaded on a 96-well plate, and the formation of dopachrome was measured in optical density at 492 nm after 20 min of incubation under dark at room temperature. Absorption was recorded using micro-plate spectrophotometer.

B16 cells (CRL-6323™; American Type Culture Collection (ATCC), Manassas, VA) and A375 cells (CRL- 1619™, ATCC) were purchased from ATCC and cultured according to the recommendations provided. B16 and A375 cells were cultured in Dulbecco’s Modified Eagle’s Medium (DMEM) supplemented with 10% fetal bovine albumin and Penicillin-Streptomycin (100 U/mL). Cells were sub-cultured every alternate day.

For melanogenesis assay, two hundred thousands of B16 cells and A375 cells were seeded onto each well of 6-well culture plates and incubated for 2 days, or 4 days, with or without the extracts of EBN, respectively. Five mM of vitamin C, 50 µM of tBHQ and 2% kojic acid served as positive controls. For A375 cells, α-MSH was added on day 2 as to stimulate the formation of melanin. To extract melanin from cells, the samples were dissolved in 100 mL of 1 M NaOH after washing twice with PBS. Samples were incubated at 60˚C for 1 hour and mixed to solubilize the melanin. The optical density of mixed solution was detected at 405 nm, and the absorbance was converted to melanin concentration by a standard curve.

Twenty-four 3D human skin model constructed with keratinocytes and melanocytes in a ratio of originated from Asian skin tissue were purchased from MatTek Corporation (Lot # 21822). The model was specific for melanogenesis study (MelanoDerm^TM Part # MEL-300-A). All procedures for macroscopic and microscopic analysis followed the instructions provided by the manufacturer. In brief, the testing samples (20 mM NANA and 1 mg/mL EBN extract) and positive control (2% of kojic acid) were applied either to the inner chambers of trans-well (keratinocyte side) or culture medium directly (melanocyte side). Culture medium having same concentration of tested chemicals was replaced every 2 - 3 days. The inner chamber was rinsed twice with phosphate buffer provided by the manufacturer and replaced with fresh testing samples and positive control. Macroscopic and microscopic images were captured on day 18. Macroscopic images were captured by top view microscope (keratinocyte side), while microscopic images were captured by inverted microscope under 100× magnification (melanocyte side).

The concentration of protein was determined following the instructions of Bradford’s method with a kit from Bio-Rad Laboratories. The analysis was done on a 96-well microtiter plate. In brief, one part dye reagent concentrate was diluted with 4 parts of double distilled water before use. Six dilutions of BSA standard (0.05 - 0.5 mg/mL) were used for the test: Ten µL of each standard, or a sample solution, were added with 0.2 mL of diluted dye reagent into separate wells and mixed well. After 10 min incubation at room temperature, absorbance at 595 nm was taken. The concentration of protein was determined from the standard curve. Statistical tests were done by using student t test provided in GraphPad Prism 5.0. Statistically significant changes were classed as [*] where p < 0.05, [**] where p < 0.01 and [***] where p < 0.001.

EBNs are graded and priced according to their color, i.e. Red > White > Grass. These EBNs were collected from local market, and which were subjected to LC-MS/MS analysis. Three major fractions (named peak 1, 2 and 3) from EBN water extracts were notified and appeared at different retention time after the separation using the full scanning mode of LC-MS/MS chromatography (Figure 1(a)). Interestingly, the full scans of total ion chromatogram of different grades of EBN showed differentiable patterns. The full scan of Grass EBN showed a strong peak 1 and a barely recognizable peak 2. The full scan of White EBN showed both notable peaks 1 and 2. While the full scan of Red EBN showed all three distinguishable 3 peaks. After the analysis with molecular weight using MRM mode quantitation and the chemical standards, the identities of peak 2 and peak 3 were confirmed as NANA and citric acid, respectively (Figure 1(b)). Peak 1 consisted of compounds with different sizes, and the pattern of mass spectrum was similar to those spectrums of collagen-like proteins (data not shown). Having the identities of NANA and citric acid, the amounts of them in different grades of EBN were quantified by the established MS system. The contents of protein, free NANA and citric acid for different grades of EBN were summarized (Supplementary Table). The protein contents for different types of EBNs were rather similar; however, the contents of NANA and citric acid were significantly higher in Red EBN and lower in Grass EBN. Citric acid was not detected in Grass EBN.

For safety reason, EBN needs to be proper processed before the consumption by human [11] . As NANA and citric acid are the most abundant soluble chemicals and therefore considered to be the targeted inhibitors. Here, we aimed to reveal the amount of NANA and citric acid in EBN after standard processing procedures. Different sample preparation procedures were applied onto EBN preparation to mimic the standard cooking processes. The content of free NANA increased during the cooking processes, and an increase of ~4 folds after stewing in different types of EBN (Supplementary Figure 1(a)). The increase could be an outcome of the release of conjugated NANA as the free form after extensive cooking. In contrast, the content of citric acid dropped by 90% after the processing (Supplementary Figure 1(b)), suggesting the free salt form of citric acid within EBN. Thus, the cooking EBN could result in different amounts of NANA and citric acid.

Tyrosinase inhibition assay is a commonly used method in screening skin lightening agents [12] - [15] . The activity of tyrosinase (mushroom) was calculated on the formation of dopachrome from L-DOPA against time: the concentration of dopachrome was obtained by absorbance at 492 nm (Figure 2(a)). The inhibition of mushroom tyrosinase was determined in different EBN water extracts. Vitamin C was used as a positive control. White and Red EBNs inhibited the activity of tyrosinase by about 50% (Figure 2(b)). In contrast, Grass EBN reduced the tyrosinase activity of only ~10%. No inhibitory functions on tyrosinase for all common EBN adulterants, e.g. agar, fungi and pig skin. In order to search for ingredient corresponding for tyrosinase inhibition, the protein- depleted fraction of EBN water extract was tested. In different types of EBN, the inhibiting activity was enriched in the protein-depleted fraction (Figure 2(c)), which suggested the role of water soluble small chemicals in such function.

The skin lightening effect by NANA becomes the focus of study here. Having tyrosinase assay, NANA inhibited both mushroom and human tyrosinases in a dose-dependent manner (Figure 3(a)). The IC₅₀ of NANA on mushroom tyrosinase and human tyrosinase was 16.93 mM, and 0.10 mM, respectively (Figure 3(a)). The original Km for mushroom and human tyrosinase on L-DOPA were 1.491 mM and 0.545 mM, respectively; and Vmax were 0.394 µM/min and 0.308 µM/min, respectively. Observed from Lineweaver-Burk plot, the values of Km and Vmax of tyrosinase were altered in the presence of NANA: the changes were in a dose-dependent manner (Figure 3(b)). By adding 10 mM NANA to the mushroom enzyme, the Km increased from 1.491 to 1.818 mM, while the Vmax decreased from 0.394 to 0.287 µM/min. In human tyrosinase, the effect of NANA did not

change the Km value, while the Vmax decreased from 0.308 to 0.006 µM/min (Figure 3(c)). Referring to the Lineweaver-Burk plot, the inhibition mechanism of NANA on mushroom tyrosinase was classified as mixed type I inhibition, whilst the inhibition mechanism of NANA on human tyrosinase was classified as non-compet- itive inhibition. Enzymatic parameters of other known tyrosinase inhibitors were compared (Table 1). The enzymatic parameters of inhibition by NANA were comparable to those existing known inhibitors.

In cultured B16 mouse melanoma cells, the color of culture medium was turned into dark brown after 48 hours of culture (Figure 4(a)). This phenomenon was due to formation and release of melanin-rich melanosome from B16 melanoma [16] . For A375 human melanoma, after stimulated by α-MSH, culture medium was also turned into dark brown after 48 hours of culture (Figure 4(c)) Vitamin C, tBHQ and 2% kojic acid, served as positive control, successfully abolished the formation of melanin, and hence the medium color remained unchanged, i.e. inhibition of melanin formation (Figure 4(a) and Figure 4(c)). The melanin concentration extracted from the treated B16 mouse melanoma and A375 human melanoma also agreed with these observations. The melanin concentration of cultured B16 and A375 were significantly lower than the blank control after incubation with NANA and White or Red EBN extract; however, the abolishing effect was not found by the extract of Grass EBN (Figure 4(b) and Figure 4(d)).

In 3D human skin model, treatment of 2% kojic acid successfully lightened the apparent intensity of skin color (Figure 5(a)) and reduced the density of melanocytes (Figure 5(b)). Similar inhibitory results were obtained by NANA and the extract of White EBN (Figure 5(a), Figure 5(b)). No significant differences in inhibition effects between treatments on the sides of keratinocyte and melanocyte. Thus, kojic acid, NANA and EBN extracts were all suggested to be permeable through the 3D human skin model (Figure 5(a), Figure 5(b)).