A biomimetic semisynthesis enables structural elucidation of selaginellin U: a tautomeric cyclic alkynylphenol from Selaginella tamariscina

Two new lactone-containing selaginellins T and U (1 and 2) together with eleven known selaginellin derivatives (3 and 7–16) were isolated from the whole plant of Selaginella tamariscina. The structure of tautomeric selaginellin U along with its biogenetic pathway was confirmed and proposed by a cross-validation of the semisynthesis of compound 4 from selaginellin (3) and derivation from 2 to 4. Additionally, compounds 3, 13 and 16 exhibited good inhibitory activities against β-site amyloid precursor protein cleaving enzyme 1 (BACE1) with IC50 values of 81.17, 51.13 and 48.89 µM, respectively.


Introduction
Selaginella tamariscina, a qualified species listed in the Chinese Pharmacopoeia, is abundant with bioactive biflavones, and has been used in traditional Chinese medicine for the treatment of amenorrhoea, dysmenorrhoea and traumatic injury [1]. Selaginellin derivatives, a rare family of natural pigments, are characterized by their acetylenic unit and p-quinone methide functionalities. However, these alkynylphenol components from this genus were still unknown until Cheng and colleagues first reported a tautomeric alkynylphenol from Selaginella sinensis [2]. Later, several selaginellins and their homologues with various bioactivities, including antibacterial activity [3], antifungal activity [4] and cytotoxicity [5][6][7], were identified. Recently, Yin and colleagues reported that 6/5/6 ring fused alkynylphenol selaginpulvilins A-D could strongly inhibit phosphodiesterase-4 (PDE4), which is a drug target for the treatment of asthma and chronic obstructive pulmonary disease [8]. Woo and colleagues found that selaginpulvilins A and D could act as protein tyrosine phosphatase 1B (PTP1B) inhibitor for the treatment of type 2 diabetes [9,10]. Inspired by the structural complexity and diverse bioactivities of selaginellin derivatives, herein, our chemical investigation of S. tamariscina led to the isolation and elucidation of two new lactone-containing selaginellins. (During the preparation of this manuscript, Yin and colleagues published the isolation and total synthesis of several lactone-containing selaginellins (see ref. [11]) including selaginpulvilin E, which was a new lactone before the publication), selaginellin T and U (1 and 2, figure 1), along with 11 known selaginellin derivatives. Meanwhile, we proposed a biosynthetic pathway for 1 and 2 (electronic supplementary material, scheme S1), as well as other selaginellins isolated from S. tamariscina in this study. As shown in scheme S1, oxidation cleavage of the alkynyl group of selaginellin A probably resulted in intermediate I, which underwent hydration followed by esterification to yield selaginellin T (1, figure 1). Selaginellin (3, figure 1) was oxidized to form the aldehyde (selaginellin O), which was further oxidized to give acid. Finally, F-ring closure between C-26 and C-28 and subsequent dehydration yielded selaginellin U (2, figure 1). In this work, the structure of compound 2 was determined by a crossvalidation of biomimetic semisynthesis from 3 and derivation from 2 to the same product 4. As a result, the biogenetic hypothesis was validated by this facile synthesis.

General experimental procedures
UV spectra were measured using a Shimadzu UV-2401A spectrophotometer (Shimadzu, Tokyo, Japan). IR spectra were recorded on a Bruker Tensor 27 spectrophotometer (Bruker Optics, Ettlingen, Germany) with KBr pellets. 1D and 2D NMR spectra were carried out on Bruker AM-400 and DRX-500 or AVANCE III-600 and AV 800 spectrometers with TMS as an internal standard (Bruker, Karlsruhe, Germany). ESI-MS was run on an Agilent 6540 Q TOF spectrometer (Agilent, Palo Alto, CA, USA). HR-ESI-MS was performed using an Agilent G6230 TOF MS system (Agilent, Palo Alto, CA, USA). Semipreparative HPLC was performed on an Agilent 1260 apparatus equipped with a diode-array detector and a Zorbax SB-C-18 (Agilent, 9.4

Semisynthesis
Aldehyde (5). To a stirred solution of 3 (50 mg) in acetone (20 ml), K 2 CO 3 (61 mg) and MeI (10 ml) were added. The mixture was sealed and stirred at 50°C for 15 h, and the solvent was evaporated under vacuum. The residue was purified by flash column chromatography on silica gel (CH 3 Cl/MeOH = 20 : 1, v/v) to afford trimethyl ether (50 mg). This compound (50 mg) was dissolved in DCM (25 ml) and to this solution was added activated MnO 2 (79 mg); the resulting mixture was stirred at 40°C for 24 h. The mixture was filtered and evaporated under vacuum, and the residue was purified by flash column chromatography on silica gel (CH 3 Cl/MeOH = 30 : 1, v/v) to give aldehyde 5 (48 mg, 90% yield over 2 steps). 1          baculovirus-expressed BACE1 and FRET peptide substrates. The first orally available non-peptidic β-secretase inhibitor LY2811376, which had an IC 50 value of 392 nM, was used as a positive control.   (table 1). Moreover, the 13 C-NMR data of compound 1 showed high similarity to those of selaginellin H in the literature [4] except for the absence of signals for the hydroxymethyl group, which indicated that the hydroxymethyl group was degraded in 1. This was supported by the 1 H-1 H COSY correlations (H15/H16 and H16/H17) of A-ring and detailed 2D analysis (figure 2).

Bioassay for BACE1 inhibitory activity
Selaginellin U (2) was obtained as a yellow oil, and showed ion peaks at m/z 525.13372 [M − H] − (calcd 525.1344) in HR-ESI-MS analysis corresponding to the molecular formula C 34 H 22 O 6 . The 1 H and 13 C-NMR spectra of 2 displayed an AB spin system (δ H 7.54 and 8.36, each 1H, d, J = 8.1 Hz) for the orthotetrasubstituted A-ring, two AA'BB' systems for the respective para-disubstituted B-and E-rings ((δ H 6.88 and 7.58, each 2H,d, J = 8.5 Hz), (δ H 6.69 and 6.97, each 2H,d, J = 8.4 Hz)), and an olefinic hydrogen (δ H 6.71, s). C26 was supposed to link to C-15 by HMBC correlation from H16 to C26 (figure 2). Strong HMBC correlations from H27 to C15 and C19 revealed that C27 linked to C14. The B-ring was determined to link to C28 by HMBC correlation from H33 to C28. The down-fielded C28 and up-fielded C15, as well as the single signal of H27 revealed that C26 and C28 probably were connected through an ester bond, resulting in the formation of a possibly new ring F fused with ring A at the C14 and C15 positions. However, the structural elucidation was stopped here because of the mismatching of the number of atoms present (C 22 H 13 O 4 ) with the molecular formula C 34 H 22 O 6, which resulted from HRMS. Selaginellins with pbenzoquinone-linked p-phenol motif always existed in two tautomers [2], resulting in delocalization between the C-and D-ring and rotation of the C7-C19 bond, and consequent broadening signals of CD rings, part of them even appearing to be obscured in the baseline ( figure 3). Subsequently, attempts to render well-defined NMR signals by acquiring the NMR spectra in a different solvent CD 3 OD (insoluble in other solvents) or with changing the test temperature (0°C and 40°C) failed. Therefore, a biomimetic semisynthesis of trimethyl ether 4 from selaginellin (3) was carried out for confirming the existence of CD rings to draw the whole structural picture of compound 2 (scheme 1).
As outlined in scheme 1, chemoselective methylation of three phenol hydroxyl groups of 3 under the condition of MeI/K 2 CO 3 and subsequent MnO 2 oxidation gave methylated aldehyde 5 in a 90% yield over two steps. Next, compound 5 was converted to acid 6 through pinnic oxidation in a 90% yield [12]. According to the hypothetical biogenetic pathway of 2 (electronic supplementary material, scheme S1), the end game was a biomimetic intramolecular cyclization of 6 in the presence of catalytic AgNO 3  2, compound 2 was subsequently fully methylated using MeI/K 2 CO 3 to give the desired product 4 in a quantitative yield. As expected, all overlapped signals were well resolved in the NMR spectra of compounds 4, 5 and 6. The tautomerization between phenol and benzoquinone at CD rings was blocked by methylation of three phenol hydroxyl groups successfully. The NMR data of compound 4 obtained from route 1 was coincident with those from route 2 (tables 1 and 2, δ ≤ 0.01 ppm in 1 H NMR and ≤ 0.1 ppm in 13 C NMR). Thus, selaginellin was probably regarded as the biogenetic precursor of 2.

Conclusion
In conclusion, two new lactone-containing selaginellins 1 and 2, together with eleven known ones, were isolated from S. tamariscina. Among them, selaginellin U (2), with a rare unsaturated δ-lactone ring fused with ring A, represents a new structure type for selaginellins. In addition, we first carried out biomimetic semisynthesis of compound 4, which not only enabled structural elucidation of 2 but also elucidated the plausible mechanisms of its biogenetic pathway. Furthermore, compounds 3, 13 and 16 exhibited promising application in the treatment of Alzheimer's disease owing to their good inhibitory activities against BACE1.
Data accessibility. The dataset supporting this article has been uploaded as the electronic supplementary material Competing interests. The authors declare no competing interests. Funding. This work was financially supported by the National Natural Science Foundation of China (U1502223, 81603000, 21602227).