Hyperisampsins H–M, Cytotoxic Polycyclic Polyprenylated Acylphloroglucinols from Hypericum sampsonii

Six new polycyclic polyprenylated acylphloroglucinols (PPAPs), named hyperisampsins H–M (1–6), were isolated from the aerial parts of Hypericum sampsonii, together with five known analogs (7–11). The structures of 1–6 were established by extensive spectroscopic analyses, including HRESIMS and NMR. In addition, the absolute configurations of these new compounds were determined by electronic circular dichroism (ECD) calculations. Compounds 1 and 2 represent the first examples of PPAPs possessing a unique γ-lactone ring at C-23, while 3–6 differed from normal PPAPs with an unprecedented 1,2-dioxane ring. Compounds 1–7 were evaluated for their cytotoxic activities against a panel of human cancer cell lines in vitro, of which 3, 4, and 6 exhibited significant cytotoxic activities with IC50 values ranging from 0.56 to 3.00 μM. Moreover, compound 3 induces leukemia cell apoptotic death, evidenced by activation of caspase-3, degradation of PARP, up-regulation of Bax, and down-regulation of Bcl-2 and Bcl-xl.

Scientific RepoRts | 5:14772 | DOi: 10.1038/srep14772 A side-by-side comparison of the 1 H and 13 C NMR data of 1 with those of 8 along with a tracing of the connectivities observed in the HMBC and 1 H-1 H COSY spectra (Fig. 2) revealed that both 1 and 8 have the same bicyclo[3.3.1]nonane core structure. However, compound 1 differs from 8 by the absence of two olefinic carbons and two methyl groups, which, together with the presence of an unexpected carboxyl group at δ C 175.5 on the side chain at C-23, imply that 1 is an oxidative degradation product of 8 10 . This conclusion was supported by the 1

H-1 H COSY correlations of H-25/H-27 and H-22/H-23 and by
the long-range HMBC interactions from H-23 to C-4, H-26 to C-23, C-24, and C-25, and from H-25 to C-23, C-27, and C-28. In addition, with 14 indices of hydrogen deficiency occupied by the tricyclic core, a phenyl group, three carbonyls, and three double bonds, the remaining index indicated the presence of an unprecedented γ-lactone ring between C-28 (δ C 175.5) and C-24 (δ C 85.7).
The relative configurations at C-1, C-5, and C-7 of the core skeleton of 1 were established through a NOESY experiment (Fig. 2) and was further confirmed by comparing its 1 H and 13 C NMR data (Tables 1  and 2) with those of sampsonione K (8) 10 . The NOESY correlation between Me-38 and H-6β revealed their cofacial and axial relationship. Therefore, the large coupling constant of H-6β/H-7 (J 6β/7 = 7.3 Hz) and the absence of coupling between H-6α and H-7 indicated that H-7 was equatorial and β oriented. These analyses were also supported by the literature 1 , which indicated that when the isoprenyl at C-7 is in an axial position, the deviation of the chemical shifts of the two protons attached to C-6 should be in the range of 0.0-0.2 ppm, while the chemical shift of C-7 should be in the range of 45-49 ppm 1 . With compound 1, the deviation of the chemical shifts of H-6α and H-6β (Δ δ H = 0.1 ppm), along with that of C-7 (δ C 47.6), corresponded well with the above conditions, confirming that the isoprenyl at C-7 was in the axial orientation. Furthermore, the α-configuration of H-23 was determined by the NOESY correlation of H-23/H-6α, which was identical to that of 8 10 . To determine the configuration of C-24, the rotation energy barrier of the single bond C-23/C-24 was calculated ( Fig. 3a) 26 . As the literature reported, rotational energy barrier of 20 kcal mol -1 was a threshold to distinguish between atropisomers and non-atropisomers 26 , therefore, theoretically, the carbon-carbon bond between C-23 and C-24 could rotate to certain extent. However, the calculated Boltzmann distribution of the most stable conformation of 1 (conformation 1a) could reach up to 96.09% (Fig. 3), which together with the observed NOESY correlations (Fig. 2b) of Me-26 with H-23 and H-22 and of H-23/H-25 indicated that this rotation was restricted 27 , and suggested a R* configuration for C-24. Thus, the structure of 1 was elucidated as shown, and it represents the first example of a PPAP with a unique γ-lactone ring. The molecular formula of hyperisampsin I (2) was identical to that of 1 as revealed by the HRESIMS spectrum (m/z 581.2798, [M + Na] + ). The structural elucidation of 2 was straightforward, as a comparison of its 1 H and 13 C NMR data (Tables 1 and 2) with those of 1, suggested that 1 and 2 shared great structural similarity. Careful analysis of the 2D NMR ( 1 H-1 H COSY and HMBC) spectra of 2 suggested that the planar structure of 2 was identical to that of 1. The relative configurtions at C-1, C-5, and C-7 were established to be identical to those of 1 in the same mannar as described for 1. In additon, the NOESY correlation of H-23/H-6α suggested that the H-23 of 2 was also α-oriented. Therefore, the only difference between 2 and 1 was the relative configuration of C-24, which was revealed by the NOESY cross-peaks (Fig. 2b)  Careful analysis of the HMBC and 1 H-1 H COSY spectra of 3 revealed that it possessed the same carbon connectivities as 8, albeit with significantly higher degrees of oxidation. Considering the chemical shifts of C-24 (δ C 80.9) and C-28 (δ C 87.0) in 3, along with the 14 indices of hydrogen deficiency required by its HRESIMS, it was reasonable to locate a 1,2-dioxane ring at C-23 via a peroxide linkage between C-24 and C-28 13 , which was also supported by the TLC detection colorated with KI-starch ( Figure S1). This deduction was further confirmed by a comparison of the 13 C NMR chemical shift values of this fragment (sequence from C-24 to C-28) in 3 with those of the 1,2-dioxane ring reported in the literature 28 . Thus, the planar structure of 3 was elucidated.
The relative configurations of C-1, C-5, C-7, and C-23 in 3 were established to be the same as those of 1 by careful analyses of the 1 H and 13 C NMR spectra and the NOESY spectra (Fig. 2). Considering the distributions of 1 based on the former conformation analyses, the S* configuration of C- 24   , revealed that H-28 was axial and β-oriented. Therefore, the relative configuration of 3 was established, and it appears to be the first example of PPAPs possessing an unexpected 1,2-dioxane ring. Detailed inspection of the 1 H and 13 C NMR spectroscopic data (Tables 1 and 2) of compounds 4-6 indicated that their structures closely resembled that of 3. Comprehensive analyses of the HMBC, 1 H-1 H COSY, and HRESIMS spectra of 4-6 suggested that compound 6 has the same planar structure as that of 3, while 4 and 5 might have structures with 1,2-dioxane rings or 1,2-dioxepane rings (4a/4b and 5a/5b, Fig. 4). In the end, the complete structures of 4 and 5 were determined to be 4a and 5a by comparing the experimental 13 C NMR data with those calculated for 4a, 4b, 5a, and 5b (Fig. 4). The relative configurtions at C-1, C-5, C-7, and C-23 of 4-6 were identical to those of 3 as revealed by their NOESY experiments and coupling constants. Consequently, the NOESY correlations from Me-26 to H-23 and H-22 and from H-23 to H-25 in 5 suggested an R* configuration of C-24. As with 5, the relative configurations of C-24 in compounds 4 and 6 were also established by NOESY experiments. Similar to 3, the orientations of H-28 in 4-6 were all set in axial orientations, owing to the large coupling constants of H-28 (9.7-11.0 Hz). Thus, the structures of 4-6 were established and named as hyperisampsins K-M.
Comparison of the electronic circular dichroism (ECD) spectra (Fig. 5) of the closely related compounds 1-6 suggested that all of them had a strong positive Cotton effect (CE) at λ max 270 nm and two negative CEs at λ max 246 and 305 nm. To determine the absolute configurations of 1-6, the ECD spectra of two simplified models A and B (Fig. 5) were calculated using time-dependent density functional theory (TDDFT) with Gaussian 09. The ECD spectra of 1-6 were subsequently compared with the calculated ECD curves of models A (1S,5R,7S) and B (1R,5S,7R) (Fig. 5), which revealed a good agreement between the experimental curves and the calculated curve of model A (1S,5R,7S). In addition, the experimental ECD curves of 1-6 were closely similar to those of hyperattenins A and B 21 , suggesting the same absolute configurations of the core structures. Thus, the absolute configurations of 1-6 were determined as shown.
Cytotoxic Activities Evaluation. Compounds 1-7 were tested for their cytotoxic activities against five human tumor cell lines, including a myeloid leukemia line (HL-60 cells), a hepatocellular carcinoma  (Table 3) 19 . Among the compounds tested, compounds 3, 4, and 6 exhibited the most potent cytotoxicity, with IC 50 values ranging from 0.56 to 3.00 μM. Compounds 5 and 7 showed moderate cytotoxic activities, with IC 50 values over a range of 3.03-25.92 μM.
Flow cytometry analysis of cell apoptosis. To analyze the potential cell apoptosis induced by these cytotoxic PPAPs, two acute myeloid leukemia cell lines (HL60 and NB4) were treated with compound 3 for 48 hours. As show in Fig. 6(A,B), treated HL60 and NB4 cells underwent dramatic cellular apoptosis in a concentration-dependent manner. Compared with the control group, the 0.75 μM compound 3 resulted in about 64.2% and 92.0% apoptosis incidence in NB4 and HL60 cells, respectively.
Western blot analysis of apoptosis related proteins. Caspase 3 activation is responsible for the proteolytic degradation of PARP, a hallmark of cells undergoing apoptosis. Bcl-2 family also plays a central regulatory role in the mitochondrial pathway of apoptosis. The balance between Bak verses Bcl-2 and Bcl-xl is important for apoptotic induction 29 . As shown in Fig. 6(C,D), compound 3 treatments activatedthe expression levels of Caspase 3 and PARP.It also up-regulated Bak, but down-regulated Bcl-2 and Bcl-xl. In conclusion, these data suggest that compound 3-induced apoptosis was mediated by the activation of caspase-3, upregulation of Bax, downregulation of Bcl-2/Bcl-xl, and degradation of PARP.
PPAPs are a special class of phloroglucinol derivatives, which have attracted great interest from both chemistry and pharmacology communities, since the report of the first natural occurring adamantyl derivative (plukenetione A) in 1996 30 . Recently, many bioactive PPAPs with complex and intriguing skeletons were reported, such as hyperuralones A and B 31 , hypersubones A and B 32 , and hyperisampsins A-D 19 . In this study, six new PPAPs (1-6), possessing novel unique γ-lactone or 1,2-dioxane rings, were isolated from the aerial parts of Hypericum sampsonii. Compounds 3, 4, and 6 exhibited significant cytotoxic activities with IC 50 values ranging from 0.56 to 3.00 μM. Moreover, we have demonstrated that compound 3 has the capacity to induce cell apoptotic death in leukemia cells, revealing, for the first  time, the mechanism of PPAP-mediated cytotoxicity, which may attract more attentions from synthesis chemistry and pharmacology communities. In conclusion, the novel structure of 3 combined with its significant cytotoxic activities reported in this study may greatly promote the anti-tumor studies of PPAPs, and further investigations on the mechanism and structure-function relationship for developing more excellent agent are necessary.
Scientific  nm; IR ν max = 3419, 1734, and 1700 cm −1 ; for 1 H NMR (400 MHz) and 13  Computational details. The theoretical calculations of compound 1 and the simplified models (A and B) of compounds 1-6 were performed using Gaussian 09. Conformational analysis was initially performed using Maestro in Schrödinger 2010 conformational searching together with the OPLS_2005 molecular mechanics methods. The optimized conformation geometries and thermodynamic parameters of all conformations were provided. The OPLS_2005 conformers were optimized at the B3LYP/6-31G(d, p) level. The theoretical calculation of ECD was performed using time-dependent density functional theory (TDDFT) at the B3LYP/6-31G(d, p) level in methanol with a PCM model. The calculated ECD curve was generated using SpecDis 1.51 33 . R vel was used in this work. The 3D structures of 4a/4b and 5a/5b, generated by Chem3D, were optimized in chloroform by using Gaussian 09 at the B3LYP/6-31G* level. Both optimized structures were then further used as the input structures for NMR calculations. For each conformation, the NMR calculation was performed using Gaussian 09 at the B3LYP/6-31G* level. Finally, the relative errors between the computed and recorded 13 C NMR spectra were calculated 34 .
Cytotoxic assay. Five human cancer cell lines (HL-60, SMMC-7721, A-549, MCF-7, and SW-480), together with one noncancerous cell line, the Beas-2B human bronchial epithelial cell line, were used in the cytotoxic activity assay. Cytotoxic activity was measured as described in our previous report 19 .
Flow cytometry analysis of cell apoptosis. Apoptosis analysis was carried out using an apoptosis detection kit (Keygen, Nanjing, China) according to the manufacturer's instructions. Briefly, HL60 and NB4 cells were exposed to vehicle (DMSO < 0.01%) and compound 3 (0.5 and 0.75 μM) for 48 h, then cells were collected and washed with cold PBS, and then resuspended in 500 μL binding buffer. After that, 5 μL of AnnexinV-FITC and 10 μL of PI were added. After supravital staining, cell apoptosis was analyzed by flow cytometry (Becton Dickinson, San Jose, CA, USA).

Western blot analysis.
Western blot analysis was conducted as described previously (citation).
Briefly, cells were treated with DMSO and compound 3 (0.5 and 0.75 μM) for 48 h, and then lysed in a radio immune-precipitation assay buffer. Protein concentrations were determined using a BCA protein assay kit (Byontime, Beijing, China). Samples were subjected to electrophoresis in 10% SDS-PAGE gels followed by transfer to PVDF membrane and probed with specific antibodies, including PARP, Bcl-2, BCL-XL, Bak, Cleaved Caspase 3, and β-Actin (Cell Signaling Technology, Inc.). Blots bands were visualized using the horseradish peroxidase conjugated secondary antibodies and chemiluminescent substrate.