Chemical Constituents of the Leaves of Campanula takesimana (Korean Bellflower) and Their Inhibitory Effects on LPS-induced PGE2 Production

Campanula takesimana Nakai (Campanulaceae; Korean bellflower) is one of the endemic herbs of Korea. The plant has been used as traditional medicines for treating asthma, tonsillitis, and sore throat in Korea. A hot water extract of the leaves of C. takesimana exhibited a significant inhibitory effect on lipopolysaccharide (LPS)-stimulated prostaglandin E2 (PGE2) production. Repetitive chromatographic separation of the hot water extract led to the isolation of three new neolignan glucosides, campanulalignans A–C (1–3), with 15 known compounds (4–18). The structures of new compounds 1–3 were elucidated by analyzing nuclear magnetic resonance (NMR) spectroscopic data, along with high resolution quadrupole time of flight mass (HR-Q-TOF-MS) spectrometric data. Among the isolates, simplidin (7), 5-hydroxyconiferaldehyde (11), icariside F2 (12), benzyl-α-l-arabinopyranosyl-(1″→6′)-β-d-glucopyranoside (13), and kaempferol 3-O-β-d-apiosyl (1→2)-β-d-glucopyranoside (15) were isolated from the Campanulaceae family for the first time. The isolates (1, 2, and 4–18) were assessed for their anti-inflammatory effects on LPS-stimulated PGE2 production on RAW 264.7 cells. 7R,8S-Dihydrodehydrodiconiferyl alcohol (5), 3′,4-O-dimethylcedrusin 9-O-β-glucopyranoside (6), pinoresinol di-O-β-d-glucoside (8), ferulic acid (10), 5-hydroxyconiferaldehyde (11), and quercetin (18) showed significant inhibitory effects on LPS-stimulated PGE2 production.


Introduction
Campanula takesimana Nakai (Campanulaceae), known as Korean bellflower, is a native herb of Korea, growing on Ulleng island. The leaves and stems of C. takesimana are shiny, and the flowers have spots on a pale purple background. It is clearly distinguished from C. punctata, commonly distributed in Korea, China, and Japan, by its smaller overall size and lighter purple color of flower [1]. Because it has long flowering period and shape of flower, this plant has been commonly cultivated for a decorative plant. Besides the value of horticulture, roots of C. takesimana have been used for herbal remedies for asthma, tonsillitis, and sore throat in traditional Korean medicine [2].
It was reported that the n-butanol and ethyl acetate fractions of C. takesimana exhibit free radical-scavenging, tyrosinase inhibition, and superoxide dismutase (SOD)-like activities [2]. Although extensive research has been done on the pharmacological activities and/or phytochemical constituents of the plants belonging to family Campanulaceae, yet only few studies have been conducted on the genus Campanula, despite their frequent traditional use [3]. Moreover, chemical constituents of C. takesimana have never been reported from previous studies.
As part of our continuing project to search for bioactive compounds with anti-inflammatory activities and identify chemical constituents of higher plants, we observed that a hot water extract from the leaves of C. takesimana showed a significant inhibitory effect on lipopolysaccharide (LPS)-stimulated prostaglandin E2 (PGE2) production in RAW 264.7 macrophages. LPS is one of the most potent activators of macrophages, and LPS-stimulated macrophages and monocytes are known to produce inflammatory mediators. In this regard, inflammatory mediators, such as PGE2, nitric oxide (NO), and numerous pro-inflammatory cytokines released by the activated macrophages are important targets for the treatment of inflammation [4]. Thus, compounds that can regulate the production of pro-inflammatory mediators like PGE2 may be promising therapeutic agents for the treatment of various inflammatory diseases.
In the present research, we concentrated on isolating secondary metabolites from the hot water extract and identifying the active constituents. Three new neolignan glucosides (1-3) and 15 known compounds (4)(5)(6)(7)(8)(9)(10)(11)(12)(13)(14)(15)(16)(17)(18) were identified by repetitive chromatographic separation of the hot water extract. The structure of the new compounds (1)(2)(3) was elucidated by analysis of one-dimensional (1D) and two-dimensional (2D) nuclear magnetic resonance (NMR) spectroscopic and high resolution quadrupole time of flight mass (HR-Q-TOF-MS) spectrometric data. The structures of the known compounds were identified in comparison with previously published values. Thereafter, the isolates were assessed for their inhibitory effects on the production of the pro-inflammatory mediator, PGE2, in LPS-stimulated RAW264.7 macrophages. We describe in this paper the isolation of the secondary metabolites from the leaves of P. japonicus, structure elucidation of the three new neolignan glucosides (1)(2)(3), and inhibitory effects of the isolates on PGE2 production.
Compound 3 had a molecular formula of C32H42O17, as established by its molecular ion peak [M−H] − at m/z 697.2336 (calcd for C32H41O17, 697.2344) in negative mode of HR-Q-TOF-MS ( Figure  S13) and was also the same as 1 and 2. The 1 H and 13 C NMR spectra of 3 were almost identical with those of 1 and 2, except for the existence of a vinyl alcohol group in 3 instead of an allyl alcohol group in 1 and 2. 1     Compound 2 was obtained as a pale yellow powder. The molecular formula of 2 was assigned as C32H42O17 by its HR-Q-TOF-MS ion peak at m/z 697.2354 [M−H] − (calcd for C32H41O17, 697.2344) in negative mode ( Figure S7) and was the same as 1. The 1 H and 13 C NMR spectra of 2 strongly resembled those of 1. However, the carbon chemical shift of one methoxy group in aromatic ring moved to the downfield (δC 62.7) in 2 (Table 1, Figure S9). In the 1 H NMR, the difference between compounds 1 and 2 was shown on the aromatic singlet signal of 2 shifted to the downfield (δH 6.87 (1H, s, H-6)), whereas two olefinic protons of 2 shifted to the upfield (δH 6.81 (1H, d, J = 15.5 Hz, H-7) and δH 6.17 (1H, dt, J = 15.5, 6.0 Hz, H-8)) (Table 1, Figure S8). By analyzing the HMBC spectrum, the carbon that linked to the methoxy proton showed correlation with methylene proton peaks at δH 3.55 (2H, m, H-9′) ( Figure 2). For these reasons, the location of one methoxy group was suggested to be C-3 of aromatic ring, rather than C-5 of 1. Thus, the structure of the new compound 2 was elucidated as 5,3′-demethyl tangshenoside Ⅲ, and named as campanulalignan B.
Compound 3 had a molecular formula of C32H42O17, as established by its molecular ion peak [M−H] − at m/z 697.2336 (calcd for C32H41O17, 697.2344) in negative mode of HR-Q-TOF-MS ( Figure  S13) and was also the same as 1 and 2. The 1 H and 13 C NMR spectra of 3 were almost identical with those of 1 and 2, except for the existence of a vinyl alcohol group in 3 instead of an allyl alcohol group in 1 and 2. 1 Table 1). 13 ) and HMBC (  Compound 2 was obtained as a pale yellow powder. The molecular formula of 2 was assigned as C32H42O17 by its HR-Q-TOF-MS ion peak at m/z 697.2354 [M−H] − (calcd for C32H41O17, 697.2344) in negative mode ( Figure S7) and was the same as 1. The 1 H and 13 C NMR spectra of 2 strongly resembled those of 1. However, the carbon chemical shift of one methoxy group in aromatic ring moved to the downfield (δC 62.7) in 2 (Table 1, Figure S9). In the 1 H NMR, the difference between compounds 1 and 2 was shown on the aromatic singlet signal of 2 shifted to the downfield (δH 6.87 (1H, s, H-6)), whereas two olefinic protons of 2 shifted to the upfield (δH 6.81 (1H, d, J = 15.5 Hz, H-7) and δH 6.17 (1H, dt, J = 15.5, 6.0 Hz, H-8)) (Table 1, Figure S8). By analyzing the HMBC spectrum, the carbon that linked to the methoxy proton showed correlation with methylene proton peaks at δH 3.55 (2H, m, H-9′) ( Figure 2). For these reasons, the location of one methoxy group was suggested to be C-3 of aromatic ring, rather than C-5 of 1. Thus, the structure of the new compound 2 was elucidated as 5,3′-demethyl tangshenoside Ⅲ, and named as campanulalignan B.
Compound 3 had a molecular formula of C32H42O17, as established by its molecular ion peak [M−H] − at m/z 697.2336 (calcd for C32H41O17, 697.2344) in negative mode of HR-Q-TOF-MS ( Figure  S13) and was also the same as 1 and 2. The 1 H and 13 C NMR spectra of 3 were almost identical with those of 1 and 2, except for the existence of a vinyl alcohol group in 3 instead of an allyl alcohol group in 1 and 2. 1 Figure S7) and was the same as 1. The 1 H and 13 C NMR spectra of 2 strongly resembled those of 1. However, the carbon chemical shift of one methoxy group in aromatic ring moved to the downfield (δ C 62.7) in 2 (Table 1, Figure S9). In the 1 H NMR, the difference between compounds 1 and 2 was shown on the aromatic singlet signal of 2 shifted to the downfield (δ H 6.87 (1H, s, H-6)), whereas two olefinic protons of 2 shifted to the upfield (δ H 6.81 (1H, d, J = 15.5 Hz, H-7) and δ H 6.17 (1H, dt, J = 15.5, 6.0 Hz, H-8)) (Table 1, Figure S8). By analyzing the HMBC spectrum, the carbon that linked to the methoxy proton showed correlation with methylene proton peaks at δ H 3.55 (2H, m, H-9 ) (Figure 2). For these reasons, the location of one methoxy group was suggested to be C-3 of aromatic ring, rather than C-5 of 1. Thus, the structure of the new compound 2 was elucidated as 5,3 -demethyl tangshenoside III, and named as campanulalignan B.  Figure S13) and was also the same as 1 and 2. The 1 H and 13 C NMR spectra of 3 were almost identical with those of 1 and 2, except for the existence of a vinyl alcohol group in 3 instead of an allyl alcohol group in 1 and 2.  Table 1). The 13 C NMR and HSQC of 3 indicated the presence of a vinyl alcohol group, which includes one oxygenated carbon signal (δ C 72.1 (C-7)), one olefinic carbon signal (δ C 141.7 (C-8)), and one exomethylene signal (δ C 115.2 (C-9)), (Table 1, Figure S16). The presence of the vinyl alcohol group in 3 was also supported by 1 H-1 H COSY data, which showed the correlation at H-7/H-8 and H-8/H-9a, 9b (Figure 2, Figure S17).
The location of the vinyl alcohol group was unambiguously determined by the HMBC correlation from H-6 to C-7 to be at C-1. Two β-d-glucopyranosyl groups and two methoxy groups were located at C-4, C-4 , C-5, and C-5 , respectively, as 1, through HMBC spectral interpretation (Figure 2). The chemical structure of 3 was established by the results noted above and by comparison with tangshenosides II and III [6,7], as shown in Figure 2, and named as campanulalignan C. However, the absolute configuration at C-7 of 3 was not determined due to the limited amount of sample obtained.

Inhibitory Effects of Compounds on LPS-Induced PGE 2 Production in RAW 264.7 Macrophages
The isolates 1, 2, and 4-18 were evaluated for their inhibitory effects of LPS-stimulated PGE 2 production in RAW 264.7 macrophages. Compound 3 was not subjected to this experiment due to the limited amount obtained. Except for 2, 3, and 4, all other compounds did not exhibit cytotoxicity in cell viability tests at concentrations up to 100 µM. The effects were assessed using the inhibitory rate on PGE 2 , which are listed in Table 2. Quercetin (18) showed the most potent inhibitory effect on LPS-induced PGE 2 production in RAW 264.7 macrophages (97% inhibitory rate at 50 µM concentration), among isolated compounds from C. takesimana. 5-Hydroxyconiferaldehyde (11) also decreased remarkably the PGE 2 production with 85.17% inhibitory rate. Additionally, 7R,8S-dihydrodehydrodiconiferyl alcohol (5), simplidin (7), pinoresinol di-O-β-d-glucoside (8), and ferulic acid (10) showed inhibitory effects over 50% on the PGE 2 production at 50 µM concentration while the new compounds 1 and 2 did not exhibit a significant effect (<50% inhibitory rate). Among the active compounds, quercetin (18), 7R,8S-dihydrodehydrodiconiferyl alcohol (5), pinoresinol di-O-β-d-glucoside (8), and ferulic acid (10) have already been reported to have anti-inflammatory effects, whereas there is no report on biological activities related to anti-inflammatory effects for simplidin (7) and 5-hydroxyconiferaldehyde (11). Therefore, inhibitory effects of 7 and 11 on LPS-induced PGE 2 production at various concentrations were investigated. As shown in Figure 3A,B, simplidin (7) and 5-hydroxyconiferaldehyde (11) significantly and concentration-dependently suppressed LPS-stimulated PGE 2 production. Compounds 7 and 11 did not influence the RAW 264.7 cell viability in either the presence or absence of LPS at concentrations up to 50 µM for 24 h, suggesting that 7 and 11 may be major constituents of the leaves of C. takesimana showing anti-inflammatory activities. Accordingly, further pharmacological studies are demanded to investigate the mechanisms responsible for the inhibitory effects of 7 and 11 on inflammatory mediator in RAW 264.7 macrophages.
All the isolates from the leaves of C. takesimana except for compound 3 were estimated for their inhibitory effects on the production of PGE2 in the LPS-induced RAW 264.7 macrophages. Inflammatory mediators, such as NO and PGE2, and numerous pro-inflammatory cytokines released by the activated macrophages are important targets for the treatment of various inflammatory diseases, including multiple sclerosis, rheumatoid arthritis, and obesity [23,24]. iNOS and COX-2 protein expression are remarkably elevated in the inflammatory processes, resulting in increased production of NO and PGE2, respectively. These overexpressed pro-inflammatory mediators can lead to damage in living cells and tissues, resulting in various physiological disorders associated with inflammation [4,23]. Therefore, compounds that inhibit the production of pro-inflammatory mediators like NO and PGE2 can be candidates for anti-inflammatory drugs [4].
All the isolates from the leaves of C. takesimana except for compound 3 were estimated for their inhibitory effects on the production of PGE 2 in the LPS-induced RAW 264.7 macrophages. Inflammatory mediators, such as NO and PGE 2 , and numerous pro-inflammatory cytokines released by the activated macrophages are important targets for the treatment of various inflammatory diseases, including multiple sclerosis, rheumatoid arthritis, and obesity [23,24]. iNOS and COX-2 protein expression are remarkably elevated in the inflammatory processes, resulting in increased production of NO and PGE 2 , respectively. These overexpressed pro-inflammatory mediators can lead to damage in living cells and tissues, resulting in various physiological disorders associated with inflammation [4,23]. Therefore, compounds that inhibit the production of pro-inflammatory mediators like NO and PGE 2 can be candidates for anti-inflammatory drugs [4].

Plant Material
The leaves of Campanula takesimana Nakai (Campanulaceae; Korean bellflower) used in present study were provided by COSMAX BIO, Gyeonggi-do, Republic of Korea, in July 2019. The origin of the material was identified by D.S.J., one of authors, and a voucher specimen (CATA-2019) has been deposited in the Laboratory of Natural Product Medicine, College of Pharmacy, Kyung Hee University, Seoul, Republic of Korea.

General Experimental Procedures
General experimental procedures are in the Supplementary Materials.

Extraction and Isolation
Dried leaves of C. takesimana (2.5 kg) were extracted with distilled water (60 L) at 100 • C for 5 h, and the solvent was evaporated by rotary evaporator at 45 • C. The hot water extract (1.0 kg) was separated over column chromatography (CC) using Diaion HP-20 as the stationary phase with a gradient system of acetone-H 2 O (0:100 to 100:0, v/v) to give 17 fractions (F1~F17).

Measurement of PGE 2 Production
RAW 264.7 cells were treated with compounds (50 µM) 1 h prior to LPS (100 ng/mL) stimulation for 24 h. Cells were pretreated with positive controls (NS398) for 1 h, and then stimulated with LPS (100 ng/mL) for 24 h. The supernatant was collected and PGE 2 production detected by using PGE 2 ELISA Kits (R&D Systems, MN, USA).