The air-dried roots of R. crispus were crushed into small grains and extracted with 90% aqueous MeOH. After removal of the organic solvent, the crude extract was suspended into water and fractionated with ethyl acetate. The ethyl acetate fraction was further applied to repeated column chromatography (CC) over Sephadex LH-20, macroporous resin D101, silica gel, and RP-18, followed with semipreparative HPLC, yielded 14 compounds. Among which, one seco-anthraquinone glucoside (1) and three naphtholones (2–4) are new compounds. Ten known compounds were identified as 3-acetyl-2-methyl-1,4,5-trihydroxy-2,3-epoxynaphthoquinol (5) , nepalensides A (6) and B (7) , polyanthraquinoside A (8) , emodin (9) , physcion (10) , chrysophanol (11) , emodin-1-O-β-D-glucopyranoside (12) , 6-methoxyl-10-hydroxyaloin B (13) , (10R)-3-methyl-1,8,10-trihydroxy-10-D-glucopyranosyl-9(10H)-anthracenone (14)  (Fig. 1), respectively, by comparison of their spectroscopic data with literature values. Seven compounds, 5–8 and 12–14, were isolated from R. crispus for the first time.
Structural identification of compounds
Compound 1 was obtained as yellowish amorphous powder. Its molecular formula, C21H22O11, was determined by negative HRESI-TOF–MS (m/z 449.1094 [M–H]−, calcd. for C21H21O11, 449.1089). In the 13C NMR spectrum of 1 (Table 1), 14 carbon signals due to one ketone (δC 203.3), one carboxyl (δC 170.8) and two benzene ring (δC 106.0–165.0, 12 × C) were observed, assignable to a carboxylated benzophenone. In addition, 6 carbon signals at δC 101.3 (C-1″), 74.7 (C-2″), 78.2 (C-3″), 71.7 (C-4″), 77.8 (C-5″), and 62.5 (C-6″) from a glucosyl moiety and a methyl signal at δC 21.4 were also observed. In the 1H NMR spectrum of 1, three characteristic proton resonances at δH 6.61 (1H, d, J = 8.3 Hz, H-4), 7.36 (1H, t, J = 8.3 Hz, H-5) and 6.56 (1H, d, J = 8.3 Hz, H-6) suggested the existence of a 1,2,3-trisubstituted benzene ring . Moreover, two aromatic singlet resonances at δH 6.90 (1H, s, H-3′) and 7.34 (1H, br s, H-5′)] and a 3-H singlet signal at δH 2.38 (3H, s) due to a methyl group were observed, together with one anomeric proton at δH 4.85 (1H, d, J = 7.8, H-1″) and a set of proton signals in the range between δH 2.3–3.8, disclosing the existence of a sugar moiety. The J1″−2″ coupling constant of anomeric proton (7.8 Hz) revealed the glucosyl anomeric center to be β configuration. The 1H and 13C NMR data of 1 were closely related to those of 6 . However, instead of a symmetric 1,2,3-trisubstituted benzene ring in 6, an un-symmetric 1,2,3-trisubstituted benzene ring appeared in 1. This was confirmed by the HMBC correlations of H-4 (δH 6.61) with C-1 (δC 114.9), C-2 (δC 165.0), C-3 (δC 160.0) and C-6 (δC 106.0), and H-6 (δH 6.56) with C-1 (δC 114.9), C-4 (δC 112.4) and C-7 (δC 203.3). Moreover, the HMBC correlations from methyl proton at δH 2.38 to C-3′ (δC 121.8), C-4′ (δC 140.8), and C-5′ (δC 122.8), from H-3′ (δH 6.90) to C-5′ and C-6′ (δC 132.3), and from H-5′ (δH 7.34) to C-3′, C-6′ and C-7′ (δC 170.8) (Fig. 2) confirmed the structure of 1. Therefore, the structure of compound 1 was established as shown in Fig. 1 and named as crispuside A.
Compound 1 is a new anthraquinone-related compound, whose formation mechanism might be similar to that of desmethylsulochrin, which was established by the ring-opening process of questin catalyzed by GedF (Geodin synthesis protein F) and GedK . In R. crispus, compound 1 maybe formed from reduction firstly and then ring-opening of ziganein-1-O-β-glucopyranoside catalyzed by GedF and GedK respectively (Fig. 3).
Compounds 2 and 3, obtained as colorless powder, are a pair of enantiomers possessing the same molecular formula C13H16O4, as deduced by the negative HRESIMS (m/z 235.0979 [M−H]− calcd C13H15O4, 235.0976). The 1H, 13C NMR, and HSQC spectral data of 2 and 3 revealed the existence of two methyls [δH 2.48 (3H, s, H-11), δC 21.1; δH 1.52 (3H, d, J = 6.7 Hz, H-10), δC 21.1], two methylenes [δH 2.88 (1H, m, H-2a), 2.67 (1H, m, H-2b), δC 35.9; δH 2.27 (1H, m, H-3a), 2.08 (1H, m, H-3b), δC 32.6], one aromatic methine [δH 6.89 (1H, s, H-5), δC 121.9], two oxymethines [δH (1H, 4.78, dd, J = 8.0, 3.8 Hz, H-4), δC 68.1; δH 5.34 (1H, q, J = 6.7 Hz, H-9), δC 65.6], one carbonyl group (δC 206.1, C-1), and a set of quaternary aromatic carbons (δC 146.4, 147.2, 130.9, 161.5, 114.6). These spectroscopic features were similar to those of (4S,9S)-9-hydroxy-O-methylasparvenone (4S,9S-HM) , whose molecular weight was 252 Da, 16 Da more than those of 2 and 3. However, the chemical shifts of C-6 and C-11 in 2 and 3 were obviously different with those of 4S,9S-HM, indicating that the substituent at C-6 in 2 and 3 was methyl group, instead of a methoxyl group in 4S,9S-HM. Moreover, the HMBC correlations of H-11 (δH 2.48) with C-5 (δC 121.9), C-6 (δC 147.2) and C-7 (δC 130.9) confirmed the substitution of methyl at C-6 position. The 1H–1H COSY correlations between H-2 and H-3, and the key HMBC correlations from H-2 to C-1/C-3, from H-3 to C-1/C-2, from H-4 to C-5/C-8a, from H-11 to C-5/C-6/C-7, from H-5 to C-8a, from H-10 to C-7/C-9, from H-9 to C-6/C-7/C-8 determined the planar structures of 2 and 3 as 4,8-dihydroxy-7-(1-hydroxyethyl)-6-methyl-3,4-dihydronaphthalen-1(2H)-one. Comparison of the experimental ECD spectrum of 2 with the calculated ECD (Fig. 4) of the four stereoisomers, (4S,9S)-, (4R,9R)-, (4S,9R)- and (4R,9S)- HM  showed the ECD of 2 was more comparable with the computationally derived data for (4S,9S)- and (4S,9R)- with positive Cotton effects (CEs) at 248 and 214 nm and negative CE at 280 nm. However, the CE amplitudes of 2 were notably closer to those of the (4S,9S)- diastereomer, thus favoring the (4S,9S) absolute configuration for 2. Therefore, compound 2 was proposed as (4S,9S)-4,8-dihydroxy-7-(1-hydroxyethyl)-6-methyl-3,4-dihydronaphthalen-1(2H)-one. Similarly, the absolute configuration of 3 was deduced as (4R,9R)-4,8-dihydroxy-7-(1-hydroxyethyl)-6-methyl-3,4-dihydro-naphthalen-1(2H)-one. The structures of 2 and 3 were established as shown and named as naphthalenones A (2) and B (3), respectively.
Compound 4 was isolated as white powder. Its molecular formula was assigned to be C13H14O4, as deduced from the HRESIMS (m/z 233.0814 [M−H]−, calcd. C13H13O4, 233.0814). The UV spectrum showed absorption bands at λmax 286 nm. The 1D and 2D NMR data of 4 (Table 2) indicated it shared the same 3,4-dihydronaphthalen-1(2H)-one skeleton with 2. The molecular weight of 4 is 2 Da less than that of 2. Comparing the DEPT and 13C NMR data of 4 with those of 2 indicated that, a ketone group (δC 206.5) appeared in 4, instead of an oxymethine C-9 (δC 65.6) in 2. This was further verified by the HMBC correlation of H-10 with C-9 in the HMBC spectrum of 4. Comparing with the ECD curve of the known 7-acetyl-4R,8-dihydroxy-6-methyl-1-tetralone (ADMT) , compound 4 showed a negative CE at 260–270 nm in ECD spectrum, which is in contrast to ADMT (Fig. 4). Therefore, the structure of 4 was determined to be 7-acetyl-4S,8-dihydroxy-6-methyl-1-tetralone and named as naphthalenone C.
Compound 5 was isolated as colorless needle crystal with a molecular formula of C13H14O5, as determined by the ESI–MS (negative ion mode) m/z 249 [M−H]−, and 13C NMR and DEPT spectroscopic data. The 13C NMR spectrum of 5 showed one carbonyl (δC 208.6), six aromatic carbon signals (δC 157.4, 116.0, 129.9, 119.1, 137.8, 120.2) with three methines and one oxygenated quarternary carbon, and six sp3 carbon signals (δC 70.2, 65.6, 72.6, 68.1, 29.1, 15.9) assignable to two methyls, two oxymethines and two oxy quarternary carbons. The 1H NMR spectrum showed three characteristic proton resonances at δH 6.72 (1H, d, J = 7.9 Hz, H-6), 7.17 (1H, t, J = 7.9 Hz, H-7), 7.10 (1H, d, J = 7.9 Hz, H-8) suggesting the existence of a 5,9,10-trisubstituted benzene ring. In addition, two methyls (δH 2.36, s, H-12; 1.47, s, CH3-2) and two oxymethines (6.72, d, J = 7.9 Hz, H-6; 5.57, s, H-4) signals were observed. The above data indicated that 5 was a naphthoquinol derivative, whose epoxide ring was inferred by chemical shifts (δC 65.6, C-2; δC 72.6, C-3) and seven degrees of unsaturation. The detailed analyses of the 2D NMR (1H-1H COSY, HMQC, HMBC) spectra (Fig. 5) deduced the planar structure of 5, which is the same as the reported 3-acetyl-2-methyl-1,4,5-trihydroxy-2,3-epoxynaphthoquinol without determination of the absolute configuration . In the present study, fine crystal from acetone was obtained and the absolute configuration of 5 was determined by single crystal X-ray diffraction (CDCC Number: 2158643) (Fig. 5). The result confirmed the planar structure of 5, and revealed unambiguously the absolute configuration as 1R, 2R, 3S, 4S.
Anti-fungal and anti-inflammatory inhibitory activity
Compounds 1 and 6–14 were evaluated for their inhibitory effects against three skin fungi (Epidermophyton floccosum, Trichophyton rubrum, Microsporum gypseum) at a concentration of 100 μM (Table S1), as previously described , with terbinafine as positive control. Most of them displayed only weak antifungal activity against the three skin fungi, while compound 9 showed obvious antifungal activities against E. floccosum and M. gypseum with MIC50 values of 2.467 ± 0.03 μM and 4.673 ± 0.077 μM, respectively. In case of the antifungal effects against E. floccosum and M. gypseum, simple emodin type anthraquinone 9 showed the strongest inhibition, followed by oxyglucoside anthraquinone (12) or C-glucoside oxanthrones (13, 14), and finally glycosylated seco analogues (1, 6–8). These compounds (1, 6–14) were also evaluated for their anti-inflammatory activity by using LPS to induce the production of iNOS from mouse monocyte macrophage RAW264.7, with L-NMMA as a positive control. The inhibition rate was shown in Table S2, all 10 anthraquinones showed NO inhibitory activity at a concentration of 50 μM, and the order of their inhibition rates is as follows: 9 > 11 > 12 > 13 > 14 > 8 > 10 > 1 > 6 > 7. Compound 9 had the strongest anti-inflammatory effect, followed by oxyglucoside anthraquinone, C-glucoside oxanthrones and finally seco-anthraquinone glucosides. It is noted that the anti-inflammatory and anti-fungal potential of emodin (9) decreased when it become glycosides or seco-anthraquinone cleavaged between C-10 and C-4a.