Lignans and a neolignan from an aqueous extract of Isatis indigotica roots

Abstract Four new lignans (1−4) and one new neolignan (5), along with two known lignan derivatives (6 and 7), were isolated from an aqueous extract of the Isatis indigotica root (ban lan gen). Their structures were determined by spectroscopic data analysis, chemical method, and theoretical calculation, for which 1 was proved by single-crystal X-ray diffraction. Compound 2 exhibited antiviral activity against influenza virus A/Hanfang/359/95 (H3N2) with an IC50 value of 11.1 µM and a selective index (SI) > 9, while 1 and 5 are the first examples of sulfonated lignan and neolignan from nature. Graphical Abstract

Based on the 1 H-1 H COSY and HSQC spectroscopic data, the proton and protonbearing carbon signals in the NMR spectra of 1 were assigned (Table 1). The presence of a 2-substituted 4-hydroxy-5-methoxyphenyl in 1 was deducible from the long-range two-and three-bond heteronuclear HMBC correlations of H-3/C-1, C-4, and C-5; H-6/C-2, C-4, and C-5; and 5-OCH 3 /C-5 ( Figure 2), together with their chemical shifts. Meanwhile, the presence of a 4 0 -hydroxy-3 0 -methoxyphenyl in 1 was determinable from the HMBC correlations of H-2 0 /C-3 0 , C-4 0 , and C-6 0 ; H-5 0 /C-1 0 , C-3 0 , and C-6 0 ; H-6 0 /C-2 0 , and C-4 0 ; and 3 0 -OCH 3 /C-3 0 , along with the 1 H-1 H COSY cross-peak of H-2 0 /H-6 0 /H-5 0 as well as the chemical shifts of these proton and carbon signals. In addition, the 1 H-1 H COSY cross-peak of H-8/H 2 -9 and the HMBC correlation of H-7/C-1, C-2, C-6, C-8, and C-9 demonstrated a linkage of C-1 through CH-7 and CH-8 to CH 2 -9. The 1 H-1 H COSY cross-peak of H-7 0 /H-8 0 /H 2 -9 0 and the HMBC correlation of H-7 0 /C-1 0 , C-2 0 , C-6 0 , C-8 0 , and C-9 0 revealed a connection of C-1 0 through CH-7 0 and CH-8 0 to CH 2 -9 0 . Moreover, the HMBC correlations of H-7/C-8 0 and H-7 0 /C-8 indicated the linkage of CH-8 and CH-8 0 , while the HMBC correlations of H-7 0 /C-1, C-2, and C-3 unraveled the connection between C-2 and CH-7 0 . A sulfonic acid and two hydroxyl groups have to be positioned at C-7, C-9, and C-9 0 to satisfy requirements of substitution and chemical shift of these carbons as well as of the molecular composition of 1. Accordingly, the planar structure of 1 was determined as shown in Figure 2, which is the first sulfonated lignan with the typical skeleton of 2,7 0 -cyclolignan [20]. The ROESY spectrum of 1 displayed the NOE crosspeaks H-7/H-9b, H-7 0 /H-9b and H-9a/H-9 0 a, indicating that these protons had the same orientation on the tetrahydronaphthalene ring, which was further confirmed by enhancements of H-7 and H-7 0 via irradiation of H-9b in the NOE difference spectrum. In addition, the NOE correlations of H-2 0 /H-8 0 and H-6 0 /H-8 0 , together with the enhancement of H-8 0 via irradiation of H-2 0 as well as the coupling constant J 7 0 ,8 0 (11.4 Hz), revealed that 4 0 -hydroxy-3 0 -methoxyphenyl and H-8 0 oriented on the other side of the tetrahydronaphthalene ring. After a CH 3 OH solution of 1 was storied under ambient conditions, a single crystal was obtained. Subsequent X-ray crystallographic analysis revealed that 1 was a calcium complex co-crystallizing with water to form a twin crystal (Supporting information, Figure S17). Although the absolute configuration could not be resolved due to formation of the twin crystal, the relative configuration of 1 was proved as shown in an ORTEP drawing of the crystal structure in Figure 4. The circular dichroism (CD) spectrum of 1 gave a typical positive coupling Cotton effect (Supporting information, Figure S19), originating from an exciton coupling of the p-p Ã transition of the two phenyl chromophores. An application of the exciton coupling rule [21] predicted that 1 had the 7 R,7 0 R,8R,8 0 S-configuration. The assignment was further supported by similarity between the experimental CD and calculation electronic circular dichroism (ECD) spectra (Supporting information, Figure  S5). Therefore, the structure of compound 1 was determined and named calcium isaticycloligsulfonate A, for which one hydroxyl group is retained at the calcium to maintain a charge equilibration of the molecular structure ( Figure 1).
Compound 2 was isolated as a white amorphous powder with [a] 20 D þ85.3 (c 0.6, MeOH). The spectroscopic data (Table 1 and Experimental section) were similar to those of (À)-isolariciresinol (6) from the same extract (Supporting information, Figure S1), indicating that 2 was a stereoisomer of 6. The 2 D NMR analysis ( Figure  2) confirmed that 2 possessed the planar structure identical to that of 6. The ROESY cross-peaks of H-7b/H-8 0 and H-7b/H 2 -9, along with the coupling constants J 7a,8 and J 7b,8 (4.8 and 9.6 Hz), demonstrated that H-7b and H-8 0 had a cis orientation while trans-oriented with H-8. In addition, the ROESY cross-peaks of H-2 0 /H-9 0 b and H-6 0 / H-9 0 b, together with the small coupling constant J 7 0 ,8 0 (3.6 Hz), revealed H-7 0 and H-8 0 had a cis orientation. Accordingly, the relative configuration of 2 was assigned as shown in Figure 3, which was supported by comparing the coupling constants of 2 with those of 1 and 6. The CD spectrum of 2 gave a typical negative coupling Cotton effect (Supporting information, Figure S31), arising from an exciton coupling of the p-p Ã transition of the two phenyl chromophores. Based on the exciton coupling rule [21], the absolute configuration of 1 was assigned as 7 0 S,8S,8 0 S, which was further supported by ECD calculation (supporting information, Figure S7). Therefore, the structure of 2 determined and named isaticyclolignan A.
Compound 3, a white amorphous powder, has the molecule formula C 26 H 34 O 11 as determined by HR-ESI-MS and NMR spectroscopic data ( Table 2 and Experimental section). The NMR spectroscopic features were similar to that of 2, except for the appearance of resonances attributable to a b-glucopyranosyl unit. This suggests that 3 is a b-glucopyranoside of 2, which was supported by 2 D NMR spectroscopic data analysis of 3 ( Figure 3). In particular, the HMBC correlation of H-1 00 /C-4 and the   Figures S52 and S53) was the same as that of 2, while the glucose was identified as D-glucose by comparison of its specific rotation and 1 H NMR spectrum with those of authentic D-glucose (Experimental section and Supporting information, Figure S54). Moreover, the similarity of CD spectra between 3 and 2 supports that glycosylation at C-4 of 2 does not significantly change the exciton coupling Cotton effects of the two phenyl chromophores. Thus, the structure of 3 was determined and named isaticyclolignoside A.
Compound 4, a white amorphous powder, is a stereoisomer of 3 as demonstrated by spectroscopic data (Table 2 and Experimental section). Especially the NMR spectroscopic data of 4 were almost identical to that of (À)-isolariciresinol-4-O-b-D-glucopyranoside (7) from the same extract (Supporting information). However, the specific rotation [a] 20 This suggests that the aglycone in 4 is the enantiomer of that in 7, which was further  , unambiguously support an enantiomeric relationship between the two compounds. Therefore, the structure of 4 was determined and designated as isaticyclolignoside B. Although the structure of 4 was reported in the literature [22], there was no data to support the reported structure after retrieving the related references.
Compound 5 was isolated as a yellowish amorphous powder. The spectroscopic data (Table 1 and Experimental) revealed that 5 is an isomer of 1. However, comparison of the NMR spectroscopic data between the two isomers indicated replacement of the aliphatic unit (CH-7 0 ) in 1 by a methylene unit [d H 3.12 (dd, J ¼ 11.4 and 5.4 Hz, H-7 0 a) and 2.80 (m, H-7 0 b) and d C 33.0] in 5, while the chemical shifts of other proton and carbon resonances were significantly changed. The differences suggest that 5 is a 4-hydroxy-3,5 0 -dimethoxy-4 0 ,7-epoxy-8,3 0 -neolignan-9,9 0 -diol [20] containing the sulfonic acid group to match the requirements of the molecular composition and substitutions. The parent structure and location of the hydroxyl and methoxyl groups were verified by explanation of the 2 D NMR spectroscopic data of 5 (Figure 2), particularly by the HMBC correlations of H-2/C-1, C-3, C-4, and C-6; H-7/C-1, C-2, C-3 0 , C-4 0 , C-6, C-8, and C-9; H-8/C-1, C-2 0 , C-3 0 ; H-2 0 /C-4 0 and C-6 0 ; H-6 0 /C-4 0 and C-5 0 ; H 2 -7 0 /C-1 0 , C-2 0 , and C-6 0 ; 3-OCH 3 /C-3; and 5 0 -OCH 3 /C-5 0 ; along with the 1 H-1 H COSY cross-peaks of H-7/H-8/H 2 -9 and H 2 -7 0 /H-8 0 /H 2 -9 0 as well as their chemical shifts (Table 1). The sulfonic acid group has to be located at C-8 0 to satisfy the chemical shift requirements of H-8 0 and C-8 0 . The NOE correlations of H- 2/H-8 and H-6/H-8 in the ROESY spectrum, together with the chemical shift of C-8 (d C 53.4) [23], revealed the trans configuration of H-7 and H-8. Although the small specific rotation [a] 20 D À1.3 (c 0.31, CH 3 OH) suggested that 5 was obtained as a mixture of stereoisomers, the CD spectrum gave an obvious negative Cotton effect around 270 nm, predicting a 7 R,8S-configuration according to the helicity rule of the 1 L b band CD for the 7-methoxy-2,3-dihydrobenzo[b]furan chromophore [24]. Due to limitation of the obtained sample amount, the absolute configuration at C-8 0 was not further assigned, meanwhile, it was still not completely excluded that the sample was obtained as the mixture of C-8 0 epimers or scalemic stereoisomers since HPLC analysis using an ion exchange chiral column showed unresolved peaks in the chromatogram (Supporting information, Figure S80). Therefore, the structure of 5, at least as the main component of the isolated sample, was determined and designated as isatineoligsulfonic acid A.
It is worth noting that 5 should be obtained also as the salt form like 1. Because the possible cation was not assigned for 5, the acid structure is given and named. Additionally, in the 13 C NMR spectra of 1 and 5, the broadening (C-2 0 , C-3 0 , and C-6 0 for 1; C-8, C-9, and C-9 0 for 5) and pairing (C-2 0 of, C-5 0 , C-6 0 , C-7, and C-7 0 for 5; d C 0.05) carbon resonances, suggested that, in solution state there might be an equilibrium between the acid and salt forms or an transformation between their preferable conformational isomers due to the presence of the ionizable and bulk sulfonate group in the structures. Furthermore, although the calculated ECD spectra of 4 and 7 gave the reversed coupling Cotton effects, the calculated Cotton effects in the short wavelength region were different from that in the experimental CD spectra (Supporting information, Figures S56 and S94). This suggests that the theoretically calculated conformers originating from a rotation of the glycosylation bond differ from the real ones in the CH 3 OH solutions of 4 and 7. In addition, comparison of the calculated ECD spectra between 2 and 3, 3 and (8S,7 0 S,8 0 S)-3, and 6 and 7 further indicated that the b-D-glucopyranosyl indeed had significant influences on the Cotton effects [14].
The two known compounds were identified by comparison of their spectroscopic data with the reported data in literatures as (-)-isolariciresinol (6) [21]. Interestingly, burselignan and (þ)-isolariciresinol, which are enantiomers of 2 and 6, respectively, were reported from "ban lan gen" [2], though their specific rotation and/or CD spectroscopic data were not presented to support the assigned structures in the original literature [25]. Analysis of 4a and 6 by HPLC using different chiral columns and mobile phases excluded that these two compounds were obtained as scalemic mixtures, while 4a and 6 were proved as the enantiomeric pair (Supporting information, Figure S92).
The new isolates (1À7) were preliminarily evaluated by cell-based assays (Supporting information). However, only 2 showed antiviral activity against influenza virus A/Hanfang/359/95 (H3N2) with an IC 50 value of 11.1 mM and a selective index (SI) > 9, respectively, while the positive controls ribavirin and tamiflu gave IC 50 values of 1.43 and 0.44 mM as well as SI's of 814 and 2646.
In summary, six 2,7 0 -cyclolignans (1À4, 6, and 7) and one 4 0 ,7-epoxy-8,3 0 -neolignan (5) were isolated from the aqueous extract of "ban lan gen," of which 1 and 5 represent the novel sulfonated forms of lignan and neolignan metabolizing and occurring in nature. Especially single crystal X-ray analysis indicated that 1 was obtained as the calcium complex, for which the calcium could not be removed by trifluoroacetic acid (TFA) that was used as an additive agent of mobile phase during HPLC separation. This suggests that sulfonated lignan and/or neolignan, at least 1, is a calcium carrier in this plant, which is different from the sulfonated indole alkaloid, disodium isatindosulfonate I [19]. Moreover, in the preliminary assay, 2 showed antiviral activity against influenza virus A/Hanfang/359/95. This, together with other antiviral constituents previously discovered from the same extract [4, 5, 8-10, 12, 13, 15, 16, 18, 19], supports that multiple components play roles in a clinic application of the herbal medicines.

General experimental procedures
Optical rotations were recorded on a Rudolph Research Autopol V polarimeter (Rudolph, NJ, USA). The UV spectra were measured on a V-650 spectrometer (JASCO, Tokyo, Japan). The CD spectra were recorded on a JASCO J-815 CD spectrometer (JASCO, Tokyo, Japan). The IR spectra were obtained on a Thermo Scientific Nicolet Is50 FT-IR ATR (Thermo Fisher Scientific, Madison, WI, USA). The HR-ESI-MS were measured on a Q Exactive Focus mass spectrometer (Thermo Fisher Scientific, MA, USA). The NMR spectra were recorded at 700 MHz, 600 MHz, or 500 MHz for 1 H and 175 MHz, 150 MHz, or 125 MHz for 13  ZWIX(þ) chiral column (150 Â 3 mm, i.d. 3 lm), or ZWIX(þ) chiral column (250 Â 10 mm, i.d. 5 lm) (Daicel Chemical Industries, Ltd., Tokyo, Japan). TLC was conducted on precoated silica gel GF254 plates. Spots were visualized under UV light (254 and 365 nm) and by spraying with 7% H 2 SO 4 in 95% EtOH followed by heating or by spraying with Dragendorff's reagent.

Plant material
For the plant material see Ref. [8].

X-Ray crystallography of 1
Triclinic, space group P -1 , a ¼ 16.5426 (5)  The data were collected on a Gemini E diffractometer with Cu Ka radiation by using the x scan technique. The crystal was kept at 100.15 K during data collection. Using Olex2 [S2], the structure was solved with the SIR2004 [S3] structure solution program using direct methods and refined with the SHELXL [S4] refinement package using CGLS minimization. Crystallographic data for the structure of 1 have been deposited with the Cambridge Crystallographic Data Centre as supplementary publication (CCDC 2163507). Copies of these data can be obtained free of charge via www.ccdc.cam.ac.uk/conts/retrieving.html (or from the Cambridge Crystallographic Data Centre, 12 Union Road, Cambridge CB21EZ, UK; fax: (þ44) 1223-336-033; or deposit@ccdc.cam.ac.uk).

Enzymatic hydrolysis of 3 and 4
Compounds 3 and 4 ($3.0 mg) were separately hydrolyzed in H 2 O (2 ml) with snailase (5.0 mg, CODE DE0274-10g, Beijing Biodee Biotech Co., Ltd., Beijing, China) at 37 C for 96 h. The hydrolysate was concentrated under reduced pressure, and separated by CC over Sephadex LH-20 (H 2 O) to obtain the aglycone 3a or 4a and sugar. The CD and 1 H NMR spectra of 3a were identical to those of 2 (Supporting information, Figures S31 and S53). The 1 H NMR spectrum of 4a was the same as that of 6, whereas the CD curve of 4a mirrored to that of 6 (Supporting information Figures  S67 and S82

Bioassays
See Supporting information.

Disclosure statement
No potential conflict of interest was reported by the authors.

Funding
Financial support from the National Natural Sciences Foundation of China (grant nos. 81630094 and 81730093) and CAMS Innovation Fund for Medical Science (2021-I2M-1-028) are acknowledged.