Compound 1 was isolated as a white amorphous powder. The molecular formula was deduced to be C₂₁H₂₉N₂O⁺ from the molecular ion peak [M]⁺ at m/z 325.2282 (calculated for C₂₁H₂₉N₂O⁺, 325.2274) in the positive-ion mode of HRESIMS (Figure S1). The ¹H NMR spectrum of 1 (Table 1, Figure S2) displayed the presence of the characteristic signals of two aromatic proton sets of a monosubstituted aromatic ring at δ_H 7.64 (2H, t, J = 7.5 Hz), δ_H 7.62 (1H, t, J = 7.5 Hz), and δ_H 7.58 (2H, t, J = 7.5 Hz); 1,2,6-trisubstituted aromatic ring at δ_H 7.18 (1H, t, J = 7.0 Hz) and δ_H 7.17 (2H, d, J = 7.0 Hz); two pairs of relatively deshielded methylene groups at δ_H 4.94 (2H, s) and δ_H 4.16 (2H, s); two symmetric methyl groups at δ_H 2.30 (6H, s); and another two pairs of ethyl groups at δ_H 3.67 (4H, m) and δ_H 1.56 (6H, t, J = 7.5 Hz). The ¹³C NMR data of 1 (Table 1, Figure S3), assigned with the aid of the HSQC (Figure S5) and HMBC experiments (Figure S6)confirmed 21 carbon signals composed of four methyl groups at δ_C 8.4 (2 × C) and δ_C 18.7 (2 × C); four methylene carbons at δ_C 54.9 (2 × C), δ_C 55.7, and δ_C 63.4; 12 aromatic carbons (δ_C 128.7, 129.1, 129.5 (2 × C), 130.7 (2 × C), 132.3, 134.1 (2 × C), 134.2, 136.7 (2 × C)); and a carbonyl carbon at δ_C 164.1. The partial structures of 1 were determined by 2D NMR experiments (¹H-¹H COSY and HMBC). The gross structure of 1 was finally elucidated by the characteristic NMR signals, and its molecular formula (C₂₁H₂₉N₂O⁺) was confirmed by HRESIMS. The ¹H-¹H COSY correlations (Figure S4) between H-2/H-3/H-4/H-5/H-6 as well as the HMBC correlations of H-2(H-6)/C-7 (δ_C 63.4) and H₂-7/C-1, C-2, and C-3 verified the presence of benzyl functionality (Figure 3). Furthermore, the methylene of H₂-7 showed HMBC correlations with three other carbons: C-8 (δ_C 55.7), C-1′ (δ_C 54.9), and C-1″ (δ_C 54.9) (Figure 3), providing evidence that a quaternary atom linking C-7, C-8, C-1′, and C-1″ is present. Based on the ¹H-¹H COSY spectrum of H₂-1′/H₂-2′ and H₂-1″/H₂-2″ and the HMBC correlations of H₂-1′/H₂-1″ with C-2′/C-2″, along with their symmetric NMR signals, the two ethyl units were assigned and confirmed to be attached to the quaternary atom by the HMBC correlations of H₂-1′/H₂-1″ with C-7 and C-8. The relatively deshielded methylene carbon NMR signals of C-7, C-8, C-1′, and C-1″ and the markedly diminished intensity of the carbon NMR signals observed for C-1′, and C-1″ at δ_C 54.9 confirmed that the quaternary atom linking them could be a quaternary ammonium cation, which finally led to the partial structure of A (Figure 3). Another spin system was observed as a cross-peak between H-12/H-13/H-14 in the ¹H-¹H COSY spectrum, representing the 1,2,6-trisubstituted aromatic ring, which was assigned to the 2,6-dimethylated benzene as the partial structure of B by the HMBC correlations of H-12(H-14)/C-10, H-13/C-11(C-15), C-16(C-17)/C-10, C-11, and C-12 (Figure 3). Finally, the connectivity through the amide bond between the two partial structures of A and B was suggested by the C=O and NH moieties remaining from the molecular formula (C₂₁H₂₉N₂O⁺) of 1, the detected HMBC correlation of H₂-8/C-9 (δ_C 164.1), and the characteristic ¹³C chemical shifts of C-8 (δ_C 55.7) and C-10 (δ_C 134.2), although the key HMBC correlation between C-9 and C-10 was missing in 1 due to the absence of protons. Accordingly, the complete structure of 1 was established, as shown in Figure 1, and it was identified to be denatonium.

Figure 3. Key ¹H-¹H COSY ( ) and HMBC ( ) correlations of compound 1. A and B represent two partial structures.

Table 1. ¹H and ¹³C NMR data of compound 1 in CD₃OD (δ in ppm, 850 MHz and 212.5 MHz for ¹H and ¹³C, respectively) ^a.

Position	Denatonium (1)
	δH (J in Hz)	δC
1		128.7 C
2	7.64, d (7.5)	134.1 CH
3	7.58, t (7.5)	130.7 CH
4	7.62, t (7.5)	132.3 CH
5	7.58, t (7.5)	130.7 CH
6	7.64, d (7.5)	134.1 CH
7	4.94, s	63.4 CH2
8	4.16, s	55.7 CH2
9		164.1 C
10		134.2 C
11		136.7 C
12	7.17, d (7.0)	129.5 CH
13	7.18, t (7.0)	129.1 CH
14	7.17, d (7.0)	129.5 CH
15		136.7 C
16	2.30, s	18.7 CH3
17	2.30, s	18.7 CH3
1′	3.67, m	54.9 CH2
2′	1.56, t (7.5)	8.4 CH3
1″	3.67, m	54.9 CH2
2″	1.56, t (7.5)	8.4 CH3

^a Coupling constants (Hz) are given in parentheses. ¹³C NMR assignments were based on HSQC, ¹H-¹H COSY, and HMBC experiments.

The structures of the known compounds (Figure 1) were determined to be trans-ferulic acid ethyl ester (2) [20], eugenin (3) [21], and α-L-glutamyl-L-Leucine (4) [22] by comparing their NMR spectroscopic data with those previously reported in the literature and MS data obtained from the LC/MS analysis. To the best of our knowledge, denatonium (1) was identified from a natural source for the first time, and compounds 2–4 were reported for the first time from P. multiflorus var. albus in this study. In natural product chemistry, it is important to take the necessary precautions during the isolation work in order to minimize the possibility of unexpected artifact isolation [23]. To verify whether compounds 1–4 were genuine natural compounds or artifacts, P. multiflorus var. albus was extracted with 80% methanol (v/v) for 10 h, and the resultant methanolic extract was subjected to LC/MS analysis. As a result, there was no peak with a molecular ion corresponding to compound 2 in the methanolic extract, whereas compounds 1, 3, and 4 were detected, suggesting that compound 2 was an artifact produced by the extraction with ethanol. In addition, to confirm that the isolated denatonium (1) is a natural compound, the methanol used for extraction was analyzed using ultra-performance liquid chromatography (UPLC) quadrupole time-of-flight (Q-TOF) high-resolution (HR)-MS because methanol can possess denatonium as a component. As a result, there was no detected peak for denatonium in the methanol solvent that we used for extraction (Figure S7), suggesting that the methanol used does not contain denatonium and the isolated denatonium can be a genuine natural compound.

Intriguingly, denatonium (benzyl-[2-(2,6-dimethylanilino)-2-oxoethyl]-diethylazanium), which is odorless and chemically stable, was found during research on local anesthetics in 1958 [24]. Since then, it has been widely used in various industries, such as cosmetics, pharmaceuticals, and material industries [25]. Interestingly, its bitterness and aversive taste have served to prevent young children from swallowing small household items, including toys and game packs, which has allowed denatonium to be widely employed in many household items. Denatonium is known to be one of the most bitter chemical compounds; thus, it was nominated in the Guinness Book of World Records as one of the most bitter compounds that people can use [26].

4. Evaluation of the Biological Activities of Compounds 1–4

MSCs are pluripotent cells in bone marrow that are known to differentiate into osteocytes and adipocytes. As microenvironmental changes cause alterations in the regulation of gene expression in MSC differentiation, the alterations of related gene expression might disturb the balance between adipocyte progenitor and osteoprogenitor cells in patients with osteoporosis ^[27][28][29][27,28,29]. Thus, a therapy that can regulate gene expression in MSCs would be promising for the management of postmenopausal osteoporosis. To determine the regulatory effects of compounds 1–4 on MSC differentiation between adipogenesis and osteogenesis, all the compounds were examined for their effects on the differentiation of murine MSCs into adipocytes or osteoblasts. The murine MSC line C3H10T1/2 was treated with 20 µM of the compounds during adipogenesis, and the differentiated cells were stained with Oil Red O (ORO). Compound 4 slightly reduced the formation of lipid droplets, resulting in fewer ORO-stained cells, compared to the normally differentiated adipocytes (Figure 4A,B). In addition, C3H10T1/2 cells were cultured in osteogenesis-inducing media in the presence of compounds 1–4. The cells were then stained for alkaline phosphatase (ALP), which is considered a distinctive marker of osteoblast differentiation [30]. Cells treated with compounds 1 and 4 showed slightly higher staining intensity and ALP enzyme activity than the negative control group (Figure 4C,D).

Figure 4. The effects of compounds 1–4 on the differentiation of mesenchymal stem cells (MSCs) toward adipocytes and osteoblasts. (A) Suppressive effects of compounds 1–4 on the adipogenic differentiation of MSCs. C3H10T1/2 cells were treated with 20 μM of compounds 1–4. After adipogenic differentiation, the cells were stained with Oil Red O (ORO). (B) The intensity of stained lipid droplets was quantitatively examined. (C) Stimulatory effects of compounds 1–4 on osteogenic differentiation of MSCs. Fully differentiated C3H10T1/2 cells were stained with alkaline phosphatase (ALP) on day 9 post osteogenic differentiation with 20 µM compounds 1–4. (D) ALP enzyme activity was determined in osteogenically differentiated C3H10T1/2 cells treated with compounds 1–4. The values were calculated relatively by setting the untreated negative control (NC) to 1. Resveratrol (20 μM) and oryzativol A (5 μM) were added to the experimental set for adipogenesis and osteogenesis, respectively, as a positive control (PC). * denotes p < 0.05 and *** denotes p < 0.001.

Compounds 1 and 4 showed the regulatory effects on the differentiation between osteogenesis and adipogenesis of MSCs (Figure 4). Among the active compounds, compound 4 was not sufficient for further experiments to examine its effects. To test the effects of compound 1 on osteogenic differentiation, C3H10T1/2 cells were stained with ALP (Figure 5A), and ALP enzyme activity was measured (Figure 5B). Our results indicated that increased concentrations of compound 1 led to the formation of darker-colored cells (Figure 5A), which indicated that the treated cells exhibited greater promotion of bone differentiation than the control group (Figure 5B). Moreover, compound 1 slightly enhanced the gene expression of ALP (Figure 5C) and osteopontin (OPN) (Figure 5D), which are osteogenesis-related factors, during osteogenic differentiation in a dose-dependent manner.

Figure 5. The effects of compound 1 on osteogenic differentiation. C3H10T1/2 cells were treated with sequential concentrations (5, 10, 20, and 40 μM) of compound 1 during osteogenic differentiation. The effects of compound 1 were evaluated through ALP staining (A). The cells were evaluated by measuring the ALP activity (B). The mRNA expression of ALP (C) and OPN (D) was measured by real-time PCR. Oryzativol A (OryA) at a concentration of 5 μM was used as a positive control (PC). * denotes 0.01 < p < 0.05 and *** denotes p < 0.001.