Angesicolide A (1) was isolated as colorless oil. The HR-ESI–MS of 1 (m/z 397.2015 [M + H]+, calcd for 397.2010) indicated its molecular formula of C24H28O5, with 11 degrees of unsaturation. IR band at 1758 cm−1 indicated the presence of an α,β-unsaturated lactone. Compound 1 demonstrated 24 carbon signals in 13C NMR, which was classified into two carbonyl carbons (δC 168.4, 168.2), six olefinic carbons, two oxygenated carbons (δC 89.9, 61.6), four methines, eight methylenes, two methyls (δC 13.8, 13.6) by DEPT experiments. The 1H NMR (Table 1) demonstrated the presence of two butylidene side chains [δH 5.11 (1H), 2.33 (2H), 1.47 (2H), and 0.93 (3H); δH 3.27 (1H), 1.90, 1.78 (each 1H), 1.54 (2H), and 0.93 (3H)], four methine signals (δH 3.04, 3.58, 3.03, 3.57) and four methane signals [δH 2.35 (2H), 1.80 (2H), 2.17, 2.04 (each 1H), 1.86 (2H)]. These signals suggested 1 to be a phthalide dimer with the same basic skeleton as chaxiongnolide A [15]. Striking differences included an extra epoxy group based on same degree of unsaturation and the absence of a double bond. The crucial HMBC correlation between H-8ʹ (δH 3.27) with C-10ʹ (δC 19.2), C-9ʹ (δC 29.8), and C-3ʹa (δC 158.4) indicated that the epoxy group was situated at C-3ʹ and C-8ʹ. Further 1H–1H COSY, HSQC, and HMBC data verified the constructive planar structure (Fig. 2).

Key 1H–.1H COSY (bold lines) and HMBC (arrows) correlations of compounds 1–3

The relative stereochemistry of 1 was elucidated based on interpretation of ROESY interactions. The butylidene side chain took Z-form by the ROESY correlation of H-8 and H-4a. Furthermore, the ROESY correlations of H-7′/H-5a and H-7/H-5′a indicated β-orientations for H-6 and H-7, and α-orientations for H-6′ and H-7′. Consequently, compound 1 was determined to possess the relative configurations of 6R*,7R*,6′R*, and 7′R* (Fig. 3). Although H-4′a showed a strong ROESY correlation with H-8′, it was insufficient to determine the relative configuration of C-3′ and C-8′. Therefore, two possible stereoisomers have been proposed: (6R*,7R*,3′R*,6′R*,7′R*,8′S*)-1A or (6R*,7R*,3′S*,6′R*,7′R*,8′R*)-1B. These configurations were subsequently subjected to NMR calculations by employing the GIAO approach at the wB97XD/6–31+G(d,p) level of theory [16]. The results indicated the whole relative configuration of 1 to be 6R*,7R*,3′R*,6′R*,7′R*,8′S* with a DP4 + MM probability of approximately 75% (Fig. 4B).

Key ROESY (dashed lines) correlations of 1–3

The.13C NMR calculation results of two plausible stereoisomers and DP4 + MM probability analysis at wB97XD/6–31+G(d,p) level. A,B Linear correlation plots of calculated vs. experimental NMR chemical shift values and DP4 + MM probability analysis for 1A and 1B; C comparison of the experimental and calculated ECD spectra of 1. D,E Linear correlation plots of calculated vs. experimental NMR chemical shift values and DP4 + MM probability analysis for 2A and 2B; F comparison of the experimental and calculated ECD spectra of 2

Angesicolide B (2) was obtained as colorless gum. 2 had the molecular formula of C24H28O5 deduced by the HR-ESI–MS data (m/z 414.2271 [M + NH4]+, calcd for C24H28O5NH4 414.2275), with 11 degrees of unsaturation. The existence of hydroxy and α,β-unsaturated lactone moieties was evidenced by the IR absorption bands at 3437, 1772 and 1710 cm−1. 2 showed 24 carbon signals including eight quaternary carbons, seven methines, seven methylenes and two methyls. The high similarity NMR data with that of levulanolide A indicated both compounds shared the same carbon skeleton [17], except for the existence of an additional hydroxyl group in 2 (δH 4.64; δC 68.1). The HMBC correlations of H-9 (δH 4.64) with C-8 (δC 113.0) and C-10 (δC 29.9) finally established the hydroxyl group was attached to C-9 (Fig. 2).

The α-orientations of H-6 and H-7, and the β-orientation of H-6′ were confirmed by the ROESY correlations of H-4′a/H-7, H-4′a/H-8′, H-7/H-8′, H-5/H-6′, and H-6/H-5′. Consequently, the relative configurations of C-6, C-7, C-3′a, and C-6′ were assigned as 6R*,7R*,3′aR*, and 6′R*, respectively (Fig. 3). The stereochemistry of the hydroxyl group at C-9 could not be determined through ROESY correlations. Consequently, two possible relative configurations were proposed for compound 2: (6R*,7R*,9R*,3′aR*,6′R*)-2A and (6R*,7R*,9S*,3′aR*,6′R*)-2B. The subsequent experimental and the calculated NMR data established the relative configuration of 2 as 6R*,7R*,9R*,3′aR*,6′R*, with a probability of approximately 85% for DP4 + MM (Fig. 4E).

Angesicolide C (3) was purified as yellow oil. It had the same molecular formula of C12H12O3 with that of senkyunolide C [18], based on the HR-ESI–MS data at m/z 205.0861 [M + H]+ (calcd for 205.0859). Upon analysis of the 1D and 2D NMR data, 3 possessed the same planar structure as that of senkyunolide C. The most pronounced difference was the enolicdouble bond in 3 took E-configuration, as verified by the absence of ROESY correlation of H-8 and H-4 and the corresponding upfield chemical shifts at C-4 (ΔδC − 8.2 ppm) and C-8 (ΔδC − 3.7 ppm) in 3.

Based on the specific optical rotation and chiral HPLC data, compounds 1 and 2 were a mixture of enantiomers, which were further treated by chiral separation to give two pairs of enantiomers [(+)-1/(−)-1, (+)-2/(−)-2]. As shown in Fig. 4, the calculated ECD results finally established the absolute configurations of (+)-1 and (−)-1 as (6S,7R,3ʹS,6′S,7′R,8ʹR)-1 and (6R,7S,3ʹR,6′R,7′S,8ʹS)-1 (Fig. 4C), respectively. Similarly, the absolute configurations of (+)-2 and (−)-2 were assigned as (6R,7R,9R,3′aR,6′R)-2 and (6S,7S,9S,3′aS,6′S)-2 (Fig. 4F).

Compound 4 was firstly isolated from A. sinensis. Comparing its spectral data with the reported data, compound 4 was identified as (±)-lyocasuarolide A [19]. Compounds (+)-4 and (−)-4 were obtained by chiral HPLC analysis, see Supporting Information for details.

All the isolates (1–4) were evaluated in vitro for their inhibitory effects against NO production in LPS-induced RAW 246.7 mouse macrophages. As shown in Fig. 5 and Figure S1, compounds 2 and 4 and their enantiomers exhibit inhibitory activity against LPS-induced NO production without cytotoxicity. As shown in Table 2, compounds 2, (+)-2, (−)-2, 4, (+)-4, and (−)-4 exhibited significant NO inhibitory effects with IC50 values between 1.23 and 5.55 μM, which were more potential than the positive control l-NMMA.

The inhibition rates of compounds 1–4 against LPS-induced NO production in RAW 264.7 cells