Antiausterity Activity of Secondary Metabolites from the Roots of Ferula hezarlalehzarica against the PANC-1 Human Pancreatic Cancer Cell Line.

Human pancreatic cancer is one of the most aggressive types of cancer, with a high mortality rate. Due to the high tolerance of such cancer cells to nutrient starvation conditions, they can survive in a hypovascular tumor microenvironment. In this study, the dichloromethane extract of the roots of Ferula hezarlalehzarica showed potent preferential cytotoxic activity with a PC50 value of 0.78 μg/mL. Phytochemical investigation of this extract led to the isolation of 18 compounds, including one new sesquiterpenoid (6) and one new monoterpenoid (18). All isolated compounds were evaluated for their preferential cytotoxicity against PANC-1 human pancreatic cancer cells by employing an antiausterity strategy. Among them, ferutinin (2) was identified as the most active compound, with a PC50 value of 0.72 μM. In addition, the real-time effect of ferutinin (2) and compound 6 against PANC-1 cells, exposed to a nutrient-deprived medium (NDM), showed cell shrinkage, leading to cancer cell death within a short period of exposure. Compounds 2 and 6 also inhibited colony formation of PANC-1 cells. The present study indicates that the dichloromethane extract of the roots of F. hezarlalehzarica is a rich source of bioactive compounds for targeting PANC-1 cells.

Human pancreatic cancer is one of the most aggressive and lethal types of cancer, with median survival rates of less than six months after diagnosis.[1] It frequently occurs in industrialized countries, and reports show that it is the seventh leading cause of cancer-related mortality worldwide, with a 5-year survival rate of only 9%.[2] Inefficient diagnosis and the remarkable drug resistance of pancreatic cancer cells have elevated the demand for developing new therapies. Recent studies have shown that the microenvironment of PANC-1 cells and the activation of stellate cells lead to dense stroma formation, providing an immunosuppressive environment and resistance to chemotherapy that accounts for the survival of these cells.[3] Tumor cells, therefore, grow rapidly and invasively. An antiausterity strategy is a method of targeting resistance of pancreatic cancer cells to nutrient starvation conditions.[4] This method was developed previously and applied successfully for the discovery of potent natural products such as arctigenin, formulated as an enriched extract (GBS-01), which has completed phase IIa clinical trials for targeting pancreatic cancer in gemcitabine refractory pancreatic cancer patients.[5]

In the framework of a recent screening program on the antiausterity activity of more than 66 plants from Iran, Ferula hezarlalehzarica Ajani (Apiaceae) showed promising activity on PANC-1 cells with a PC50 value of 0.78 μg/mL. The genus Ferula comprises more than 170 species worldwide, of which 30 grow in Iran, including 16 species endemic to the country.[6] These plants have been used traditionally for treating skin infections, hysteria, and stomach disorders.[6]F. hezarlalehzarica was identified for the first time in 2008,[7] and there have been no prior phytochemical investigations of this plant to date. In the present study, reported for the first time are the isolation, identification, and determination of the absolute configuration of secondary metabolites of F. hezarlalehzarica. Furthermore, the antiausterity activity of isolated compounds was evaluated, and the effects of two of the most active compounds, 2 and 6, were investigated on cell morphology and colony formation of PANC-1 cells.

Results and Discussion

Methanolic extracts of 66 plants from Iran were screened for their antiausterity activity on human pancreatic cancer cells (PANC-1) in nutrient starvation conditions using a previously established method.[4] The root extract of F. hezarlalehzarica showed the most promising activity, with a PC50 value of 0.78 μg/mL (Figure S1-A, Supporting Information). HPLC-DAD analysis of the extract indicated that the major compounds were rather lipophilic. Thus, extraction of the root material was performed with dichloromethane, followed by a reassessment of its activity (Figure S1-B, Supporting Information). Subsequent purification steps of the dichloromethane extract led to the isolation of a new daucane-type sesquiterpenoid (6) and a new monoterpenoid (18), along with 16 known mono- and sesquiterpene derivatives (Figure ). Jaeschkeanadiol (1),[8] ferutinin (2),[9] teferin (3),[9] 10α-hydroxyferutinin (4),[10] kuhistanicaol G (5),[11] felikiol (7),[12] felikiol 8-angelate (8),[12] two lancenotriol p-hydroxybenzoate isomers (9S = α-OH; 9R = β-OH) (9, 10),[13] lancerodiol p-hydroxybenzoate (11),[13] webiol 3-angelate (12),[12] 8,9-epoxyferutinin (13),[14] (±)-tschimgin (14),[15,16] (±)-tschimganin (15),[15,16] fenoferin (16),[17] and stylosin (17)[17] were isolated previously from different Ferula species in the past. Although their relative configurations were established, the absolute configuration of most of them was not determined except for compound 1, but with somewhat confusing literature values.[9−14] For this purpose, electronic circular dichroism spectra of selected compounds were simulated for the first time in the present study by quantum chemical calculation methods and compared with experimental spectra. Compounds 2, 4, 5, 9–11, and 13 share a p-hydroxybenzoic acid ester moiety or, in the case of 3, the corresponding 3-methoxy-4-hydroxybenzoic acid ester unit, which are suitable as ECD chromophores by giving two Cotton effects around 250 and 225 nm. Since all the above-mentioned compounds showed the same relative configuration at C-6 (according to the NOESY experiments performed) and displayed similar experimental ECD spectra, calculations were only performed for compound 13, and the results obtained were extended to compounds 2–5 and 9–11 (Figures S18–S20, Supporting Information). For compounds 6, 7, 8, 12, and 18, ECD calculations were performed individually. As the NOESY correlations were not conclusive in the case of compound 12, NMR chemical shift calculations along with DP4+ probability evaluation were applied for the determination of its relative configuration prior to ECD calculation. In contrast, compounds 14 to 17 were isolated as racemic mixtures, since no optical activity was observed during the measurement of the respective optical rotations. The structures of the compounds isolated with their absolute configuration are shown in Figure .

Figure 1

Structure of isolated molecules (1–18) from the roots of F. hezaelalehzarica.

Compound 6 was isolated as a colorless gum. Its molecular formula was deduced as C22H30O5 based on the HRESIMS data (found m/z 397.1959, calcd for C22H30O5Na [M + Na]+, m/z 397.1985). The UV spectrum of 6 showed an absorption maximum at 254 nm (log ε 2.68), pointing to a similar chromophore to that in compounds 2–5, 9–11, and 13–18. The 1H NMR spectrum revealed four methyl group resonances at δH 0.87 (H-12), 0.89 (H-13), 1.31 (H-14), and 1.55 (H-15) and those for six methylene protons at δH 2.81 (H-7a), 2.08 (H-7b), 1.97 (H-3a) 1.60 (H-3b), 1.68 (H-2a), and 1.39 (H-2b) and signals for five methine protons at δH 5.86 (H-6), 5.78 (H-9), 5.51 (H-10), 2.42 (H-5), and 1.65 (H-11), along with four aromatic methine groups at δH 7.92 (H-2′, H-6′), 6.87 (H-3′, H-5′), displaying a para-substituted aromatic ring pattern. The 13C NMR spectrum showed 22 carbon signals, including those for two quaternary carbons at δC 85.7 (C-4) and 84.2 (C-8) and for para-substituted phenyl ring carbons at δC 160.4 (C-4′), 132.2 (C-2′ and C-4′), 122.8 (C-1′), and 115.6 (C-3′ and C-5′), along with an ester carbonyl at δC 166.9 (C-7′). Analysis of the COSY and HMBC correlations revealed a daucane scaffold for this compound.[18] The NMR data of 6 were similar to those of the already isolated and reported compound kuhistanicaol G (5).[18] However, a detailed comparison of the NMR data revealed diagnostic differences in the chemical shifts of C-8 (δC 84.2) and C-6 (δC 70.6) of 6, in contrast to those of kuhistanicaol G [C-8 (δC 83.6) and C-6 (δC 72.8)]. Detailed analysis of the HMBC spectra did not allow the positioning of the p-hydroxybenzoic acid moiety due to the lack of a cross-peak to the corresponding ester carbonyl (δC 165.0), suggesting a connection to a tertiary alcohol group at another position. This assumption was confirmed by long-range experiments (NOESY and ROESY). NOESY data showed a correlation from δH 1.55 (H-15) to the protons of the p-hydroxybenzoic acid ester group at δH 7.92 (H-2′ and H-6′), confirming the connection of the ester group at C-8. The ROESY spectrum additionally displayed NOEs from H-14 to H-6 and from H-2′ and H-6′ to H-12 and H-13, which later confirmed the structure of the compound. As a result, the structure of compound 6 was deduced as (1R*,4R*,5S*,6S*,8R*)-4,6-dihydroxy-4-isopropyl-1,8-dimethyl-1,2,3,4,5,6,7,8-octahydroazulen-8-yl 4-hydroxybenzoate (Figure ). The obtained 3D structure of 6 was subjected to conformational analysis, resulting in five conformers within an energy window of 5 kcal mol–1 (Figure , Supporting Information). Further optimization and calculation of ECD spectra was done by B3LYP/6-31++G(d,p)//B3LYP/6-31++G(d,p)/CPCM in acetonitrile. As depicted in Figure , the calculated ECD spectrum (red line) showed two positive Cotton effects at 260 and 195 nm and two negative Cotton effects at 237 and 214 nm, which were in good agreement with the experimental spectrum, while the opposite enantiomer (blue line) showed a mirrored ECD spectrum. Therefore, the absolute configuration of 6 was deduced as 1R, 4R, 5S, 6S, and 8R.

Figure 2

Key HMBC correlations of compounds 6 and 18.

Figure 3

Comparison of experimental ECD spectra of compounds 6 and 18 with their calculated spectra.

Compound 18 was obtained as a colorless gum. The HRESIMS displayed a [M – H]− ion at m/z 289.1479, indicating a molecular formula of C17H21O4 (calcd for 289.1445). The UV spectrum (CH3CN) of 18 showed an absorption maximum at 254 nm (log ε 3.67), pointing to a similar chromophore to that in compounds 2–6, 9–11, and 13–17. The 1H NMR spectrum revealed three methyl groups at δH 1.22 (H-9), 1.14 (H-8), and 0.82 (H-10); two methylene protons at δH 2.42 (H-6a), 1.11 (H-6b), 1.68 (H-7a), and 1.58 (H-7b); five methine protons at δH 5.86 (H-6), 5.78 (H-9), 5.51 (H-10), 2.42 (H-5), and 1.65 (H-11); and four aromatic methine groups at δH 7.92 (H-2′, H-6′) and 6.87 (H-3′, H-5′), corresponding to a para-substituted aromatic ring. The 13C NMR spectrum showed 17 carbon signals, from which two quaternary carbons at δC 85.7 (C-4) and 84.2 (C-8) and carbons of a para-substituted phenyl ring at δC 160.4 (C-20), 132.2 (C-2′ and C-6′), 122.8 (C-17), and 115.6 (C-19 and C-21) along with an ester carbonyl at δC 166.9 (C-16) were characteristic for a p-hydroxybenzoic acid ester unit. The 10 remaining carbons could be identified as those of a monoterpene unit. COSY correlations indicated a spin system of H-6/H-5/H-4/H-7. Analysis of the HMBC spectrum showed a connection from H-7a (δH 1.68) to C-6 (δC 40.0) over a quaternary carbon at δC 48.2 (C-1) in which the position of C-1 was confirmed by two 3J correlations from H-4 (δH 1.79) and H-5 (δH 4.29) to C-1 (δC 48.1). The H-4 (δH 1.79) resonance was also involved in further vicinal connections to C-9 (δC 19.7), C-2 (δC 85.0), C-5 (δC 71.6), and C-6 (δC 40.0). In turn, the H-2 signal at δH 4.55 showed HMBC correlations to signals of the methyl groups C-9 and C-10 (δC 19.7 and 29.9), C-6 (δC 40.0), C-7 (δC 37.2), C-8 (δC 19.0), C-4 (δC 56.1), and an ester carbon at δC 166.6, confirming the position of the p-hydroxybenzoic acid group. Therefore, the planar structure of 18 was deduced as 5-hydroxy-1,3,3-trimethylbicyclo[2.2.1]heptan-2-yl 4-hydroxybenzoate (Figure ).

NOESY correlations revealed NOEs from H-5 to H-10 and from H-2 to H-9 and H-7b, resulting in a relative configuration of 1R*, 2S*, 4R*, and 5S*. Subsequently, the 3D structure was subjected to conformational analysis followed by geometrical optimization and ECD simulation, which led to the determination of the absolute configuration of compound 18 as 1R, 2S, 4R, and 5S (Figure ).

Preferential Cytotoxicity against PANC-1 Cells

Human PANC-1 cells show a remarkable tolerance to nutrient-deprived conditions and can survive under an extreme austere hypovascular tumor microenvironment[19] for prolonged time periods. An antiausterity strategy targets the tolerance of cancer cells to nutrition starvation, thereby killing the cancer cells selectively. Previous studies have shown the effectiveness of this approach for the discovery of potential anticancer agents.[20,21] All compounds isolated therefore were investigated for their preferential cytotoxic activity in the nutrient-deprived medium (NDM) along with the nutrient-rich medium (DMEM). The activity is expressed as the PC50 value, which is the concentration at which 50% of cancer cells were preferentially killed in the NDM without apparent toxicity in the DMEM. As shown in Table , most of the isolated compounds, in particular compounds 2, 3, and 10, exhibited promising antiausterity activity with PC50 values of 0.7, 1.2, and 1.3 μM, respectively. The potency of compound 2 was equivalent to that of the positive control, arctigenin (PC50 = 0.9 μM), an established antiausterity agent. As most of the active compounds bear a benzoic acid ester unit, the activities of these units (p-hydroxybenzoic acid and p-benzoic acid methyl ester) were additionally assessed as pure compounds. Interestingly, no preferential cytotoxic activity was observed at the investigated concentrations, suggesting that a combination of a terpene unit with an ester moiety is crucial for the activity of these compounds, which was supported also by the high PC50 value of 1 (90.5 μM), representing an unfunctionalized sesquiterpene alcohol. Subsequently, all isolated compounds were also tested in a nutrient-rich medium (DMEM), and, interestingly, no cytotoxicity was observed at the applied concentrations.

Table 2

Preferential Cytotoxicity Activity of Isolated Compounds from F. hezarlalehzarica Roots on Human PANC-1 Pancreatic Cancer Cells

compound	PC50a (μM)	compound	PC50a (μM)
1	90.5	11	3.0
2	0.7	12	57.4
3	1.2	13	35.9
4	3.8	14	1.5
5	3.5	15	3.8
6	1.5	16	5.9
7	15.3	17	3.2
8	73.1	18	16.4
9	8.9	PHBAb	>100
10	1.3	BAMEc	>100
arctigenin	0.9	CH2Cl2 extract	1.1 (μg/mL)

Concentration at which 50% of the cells were killed preferentially in nutrient-deprived medium (NDM).

p-Hydroxybenzoic acid.

Benzoic acid methyl ester.

Preferential cytotoxic activities of compound 2 (A) and compound 6 (B) against the PANC-1 human pancreatic cancer cell line in nutrient-deprived medium (NDM) and Dulbecco’s modified Eagle’s medium (DMEM).

In the next step, ferutinin (2) and its new and active analogue 6 were selected for further investigations on PANC-1 cells to determine their effects on cell death in real time. Cells were treated with 5 μM of the compounds selected in NDM and incubated in a stage-top CO2 incubator at 37 °C on a digital cell imaging station, and images were captured every 10 min under the phase-contrast mode for 24 h. As shown in Figure (Supporting Movies 1 and 2), the treatment of PANC-1 cells with compound 2 or 6 resulted in cell morphology alterations, which were already visible after 4 h of incubation time. Compounds 2 and 6 caused shrinkage of PANC-1 cells within 4 h of treatment and plasma blebbing from 10 h, leading to total cell death within 24 h.

Figure 5

Effects of compounds 2 and 6 on the morphology of PANC-1 cells. Cells were treated with 5 μM of compounds 2 and 6. Photographs were taken at different time points (0, 4, 8, 12, 16, and 20 h).

Based on the previous results, the effects of 2 and 6 on colony formation of PANC-1 cells were evaluated. This assay is used for the investigation of the development and chemosensitivity of early tumors.[22] In order to examine the effects of compounds 2 and 6 on colony formation, PANC-1 cells were treated with three different concentrations (12.5, 25, and 50 μM). As shown in Figure , compounds 2 and 6 decreased colony formation of PANC-1 cells significantly in a concentration-dependent manner. Treatment of PANC-1 cells with a concentration of 12.5 μM of compound 2 could further inhibit the growth of the cells after 10 days by a factor of 50% when compared to a control. Although compound 2 was the most active in the preferential cytotoxicity assay, its analogue, compound 6, reduced colony formation by 80% at a concentration of 12.5 μM.

Figure 6

Effects of 2 and 6 on colony formation of PANC-1 cells. (A) PANC-1 cell colonies treated with different concentrations of 2 and 6. (B) Graph showing mean values of the area occupied by PANC-1 cell colonies (three replicates). ****p < 0.0001, ***p < 0.001, *p < 0.05 when compared with the untreated control group.

Ferutinin (2) was isolated previously from different Ferula species,[11,13,14,18,23] and previous investigations have indicated its cytotoxic activity against different cancer cell lines.[24,25] In a study performed by Arghiani et al.,[26] ferutinin (2) revealed a discernible antitumor activity on human colon cancer cells in vivo and was able to suppress tumor growth significantly. Ferutinin (2) and other isolated derivatives are also potential agonists (ERα) or antagonists (ERβ) of estrogen receptor subtypes,[27] among which ferutinin (2) acts as a nonsteroidal phytoestrogen at the ERα subtype with an EC50 value of 33.1 nM.[28] Among all these studies, there has been no prior work of the potential antineoplastic activities of ferutinin (2) and its derivatives targeting pancreatic cancer cells. In this study, ferutinin (2) and its derivatives exhibited promising antiausterity activity on PANC-1 cells. Further investigation revealed that ferutinin (2) and compound 6 affect cell membrane integrity and inhibited cell colony formation. Therefore, the roots of F. hezarlalehzarica may represent a valuable source of potential antitumor compounds that target pancreatic cancer cells.

Experimental Section

General Experimental Procedures

Reagents and solvents were of analytical grade and purchased from VWR International (Darmstadt, Germany) if not otherwise stated. Solvents used for HPLC analysis were obtained from Merck (Darmstadt, Germany). Ultrapure water for the HPLC analysis was produced by a Sartorius Arium 611 UV water purification system (Sartorius AG, Göttingen, Germany). Optical rotations were measured on a JASCO P-2000 polarimeter. UV/vis and ECD spectra measurements were carried out on a JASCO J-8000 spectropolarimeter (JASCO, Japan). IR spectra were recorded on a Bruker Alpha FT-IR apparatus equipped with a platinum ATR module. One- and two-dimensional NMR experiments were recorded on a Bruker Avance II 600 spectrometer (Bruker) operating at 600.19 MHz (1H) and 150.92 MHz (13C) at 300 K (chemical shifts δ in ppm, coupling constants J in Hz), with deuterated chloroform (chloroform-d) or acetone (acetone-d6) as solvents, containing TMS 0.03%. These solvents were purchased from Euriso-top SAS (Saint-Aubin Cedex, France). LC and LC-ESIMS parameters: stationary phase: YMC-Pack Pro C18 RS 5 μm, 150 × 4.6 mm; mobile phase: A = H2O, B = acetonitrile; gradient: 0 min: 2% B; 30 min: 98% B; 40 min: 98% B; 40.1: stop; temp: 35 °C, flow: 0.8 mL/min, injection volume: 5–15 μL, sample concentration = 2 mg/mL in CH3CN or THF. HPLC-ESIMS experiments were performed on an Agilent HP1100 system (Waldbronn, Germany) coupled with an Esquire 3000 Plus ion trap (Bruker Daltonics, Bremen, Germany), using electrospray ionization (ESI). High-resolution mass spectrometry was performed on a Bruker TOF-Q mass spectrometer using a direct injection of isolated compounds dissolved in MeOH in concentrations of 10–30 μg/mL. HPLC analysis was carried out on a Shimadzu LC-20AD XR system (Düsseldorf, Germany) equipped with a DAD detector, autosampler, and column thermostat. Preparative TLC was performed on precoated silica gel 60 F254 plates (Merck, Darmstadt, Germany, 1 mm thickness).

Plant Material

The aerial parts and the roots of Ferula hezarlalehzarica were collected at Hezar Mountain in Kerman Province, Iran, in June 2016. Identification was performed by Mr. Mojtaba Khodami, and a voucher specimen (number: kf-1147) was deposited at the herbarium of the Kerman University on being identified. The plant material was shade-dried after collection and subsequently stored in the dark until further investigations.

Extraction and Isolation

Air-dried root material (250 g) was milled and extracted with 700 mL of dichloromethane utilizing an ultrasonic bath (10 min) followed by filtration. The solvent of the filtrate was removed by vacuum rotary evaporation at 30 °C. The procedure was repeated 10 times using fresh solvent. The combined extracts finally yielded 75 g of dry dichloromethane extract. A part of the dichloromethane extract (60 g) was fractionated by silica gel column (Ø = 10 cm, l = 40 cm) chromatography using gradient elution from 0% to 10% EtOAc using four steps (1000 mL, 2.5% increase in each step) followed by elution with 15, 20, 50, and 100% EtOAc (each 500 mL). Elution was continued with 98:2, 95:5, 90:10, and 80:20 EtOAc–MeOH gradients (each 500 mL). The fractions obtained were monitored by TLC and subsequently combined to afford 20 fractions (F1–F20). All combined fractions were also analyzed using HPLC-DAD. F5 (1.7 g) was chromatographed over a silica gel column (Ø = 2 cm, l = 100 cm) using isocratic elution with dichloromethane, which resulted in six fractions (F5-S1 to F5-S6). F5-S4 (50 mg) was purified by preparative TLC (PLC silica gel 60 F254, 1 mm layer, Merck, Darmstadt, Germany) with EtOAc–toluene (5:95) as mobile phase. The plate was developed twice, and the corresponding areas were removed and suspended using sonication in CHCl3–MeOH (50:50), which resulted in purification of compounds 16 (4.5 mg, R = 0.8) and 17 (10 mg, R = 0.9). F7 (3.5 g) was divided into three parts, and each part was separated by Sephadex LH-20 column chromatography (Ø = 2 cm, l = 100 cm), with CH2Cl2–acetone (85:15, v/v) as mobile phase. The fractions obtained from each column were monitored by TLC and combined, affording in total 16 fractions (F7-S1 to F7-S16). Fractions F7-S4 (100 mg) and F7-S5 (123 mg) were combined and chromatographed over a silica gel column (Ø = 1 cm, l = 40 cm) with a gradient elution of CH2Cl2–acetone (98:3 to 85:15) in five steps, resulting in eight subfractions (P1 to P8). F7-S4/5-P5 (18.8 mg) was recrystallized from n-hexane to afford compound 8 (10 mg). F7-S16 (50 mg) was subjected to preparative TLC (silica gel) using EtOAc–toluene (5:95) as mobile phase. The plate was developed twice, and the corresponding zones were removed and suspended using sonication in CHCl3–MeOH (50:50), which resulted in the purification of compounds 14 (8 mg, R = 0.52) and 15 (9 mg, R = 0.67). F10 (4.4 g) was chromatographed over a silica gel column (Ø = 2 cm, l = 100 cm) using a first isocratic elution with 100% CH2Cl2 followed by gradient elution of 6.25%, 20%, 50%, and 100% EtOAc. In total, 45 fractions were obtained (S1 to S45). Among them, the last fraction, F10-S45 (700 mg), contained pure compound 2. F10-S29 (56 mg) was subjected to Sephadex LH-20 CC (Ø = 1 cm, l = 40 cm) using isocratic elution with CH2Cl2–acetone (85:15, v/v) to afford 21 fractions, among which F10-S29-P4 (12, 12 mg) and F10-S29-P10 (3, 200 mg) were pure compounds. Further chromatography of F10-S29-P12 (131 mg) using a silica gel column (Ø = 2 cm, l = 100 cm) eluted with petroleum ether–tert-butyl methyl ether (25:75) led to the isolation of F10-S29-P12-N8 (1, 40 mg). F15 (600 mg) was loaded on a Sephadex LH-20 column (Ø = 2 cm, l = 100 cm) and eluted with CH2Cl2–acetone (85:15, v/v), which resulted in 32 fractions (F15-S1 to F15-S32) being produced. Of these, F15-S5 (7, 9 mg), F15-S17 (5, 3 mg), F15-S21 (4, 4 mg), F15-S24 (9, 7 mg), and F15-S32 (6, 5 mg) were pure compounds. Further chromatography of F15-S13 (100 mg) over a silica gel column (Ø = 1 cm, l = 40 cm), by an isocratic elution with CH2Cl2–acetone (98:2), led to six fractions (F15-S13-P1 to F15-S13-P6) being prepared. Chromatography of subfraction F15-S13-P3 (10 mg) using silica gel column chromatography (Ø = 0.5 cm, l = 20 cm) with petroleum ether–tert-butyl methyl ether (45:55, v/v, isocratic elution) as mobile phase resulted in the isolation of F15-S13-P3-N8/9 (13, 4.5 mg). F15-S29 (10 mg) was subjected to a silica gel CC (Ø = 0.5 cm, l = 20 cm) eluted with petroleum ether–tert-butyl methyl ether (30:70, v/v, isocratic elution), which resulted in the isolation of F15-S29-P4 (10, 2 mg) and F15-S29-P5 (11, 2 mg).

(1R,4R,5S,6S,8R)-4,6-Dihydroxy-4-isopropyl-1,8-dimethyl-1,2,3,4,5,6,7,8-octahydroazulen-8-yl 4-hydroxybenzoate (6):

colorless gum; [α]D20 −30.0 (c 0.2, MeOH); UVmax (CH3CN) λmax (log ε) 254 (2.68), 195 (3.01) nm; ECD (c 100 × 10–6 M, CH3CN) [θ]260 +6592, [θ]237 −10 692, [θ]214 −36 042, [θ]195 +50 000; IR νmax 3320, 2961, 2935, 1685, 1607, 1592, 1514, 1273, 1164, 1028, 753 cm–1; 1H NMR and 13C NMR, see Table ; HRESIMS m/z 397.1959 [M + Na]+ (calcd for C22H30O5, m/z 397.1985).

Table 1

1H and 13C NMR Spectroscopic Data of Compounds 6 and 18 in CDCl3 (1H 600.19 MHz, 13C 150.92 MHz)a

	6		18
position	δC, type	δH (ppm, J in Hz)	δC, type	δH (ppm, J in Hz)
1	46.0, C		48.1, C
2	41.3, CH2	1.68, 1.39, m	85.0, CH	4.55, d (1.8)
3	31.1, CH2	1.97, dd (9.0, 14.0), 1.60, m	39.6, C
4	85.7, C		56.1, CH	1.79, brs
5	53.0, CH	2.42, d (10.7)	71.6, CH	4.29, d (6.5)
6	70.6, CH	5.86, ddd (3.0, 5.1, 10.7)	40.0, CH2	2.42, ddd (2.6, 6.6, 13.8)
				1.11, dq (13.6, 1.1)
7	40.9, CH2	2.81, dd (5.2, 15.0)	37.2, CH2	1.68, dd (10.9, 1.7)
		2.08, dt (15.01, 2.4)		1.58, brd (10.9)
8	84.2, C		19.0, CH3	1.14
9	130.6, CH	5.51, dd (1.8, 11.7)	29.8, CH3	1.22
10	141.6, CH	5.78, dd (11.7)	19.7, CH3	0.82
11	36.8, CH	1.65, m
12	18.6, CH3	0.87, d (2.2)
13	17.5, CH3	0.87, d (2.2)
14	21.5, CH3	1.31, s
15	24.7, CH3	1.55, s
1′	122.8, C		123.2, C
2′, 6′	132.2, CH	7.92, d (8.7)	132.0, CH	7.95, d (8.6)
3′, 5′	115.6, CH	6.87, d (8.7)	115.4	6.88, d (8.7)
4′	160.4, C		159.9, C
7′	166.9, C		166.6, C

For carbon numbering, see Figure .

(1R,2S,4R,5S)-5-Hydroxy-1,3,3-trimethylbicyclo[2.2.1]heptan-2-yl-4-hydroxybenzoate (18):

colorless gum; [α]D20 +2.3 (c 0.07, CHCl3); UVmax (MeOH) λmax (log ε) 275 (3.67) nm; ECD (c 2.3 × 10–3 M, CH3CN) [θ]241 +1068, [θ]216 +3690; IR νmax 3261, 2962, 1724, 1590, 1341, 1227, 1167, 1118, 1009 cm–1; 1H NMR and 13C NMR, see Table ; HRESIMS m/z 289.1479 [M – H]− (calcd for C17H22O4, m/z 289.1317).

1H and 13C NMR and DP4+ Calculations

Compound 12 was subjected to conformational analysis and geometrical optimization of using DFT/6-31G(d) in the gas phase in Gaussian 09. Subsequently, NMR chemical shift calculation was conducted using a gauge-independent atomic orbital (GIAO) method in rmpw1pw91/6-31+G(d,P)/CPCM/chloroform. The shift tensors obtained were adjusted further to chemical shifts by using TMS proton and carbon chemical shifts, which were calculated with the same method. All chemical shifts were Boltzmann-averaged, and unscaled chemical shifts were used for DP4+ probability calculation based on the method published by Grimblat et al.[29]

Calculation and Measurement of ECD Spectra

3D structures of isolated compounds were drawn in Maestro (Schrödinger LLC) and subjected to conformational analysis using MacroMoldel 9.1 (Schrödinger LLC) and OPLS-3 as a force field in the gas phase. Geometrical optimization and energy calculation of conformers occurring in an energy window of 5 kcal mol–1 were done by implementation of DFT/B3LYP/6-31G++(d,P) in the gas phase using Gaussian 09.[30] Subsequently, ECD spectra of optimized compounds were simulated using time-dependent density functional theory and a B3LYP and 6-31G++(d,p), basis set and level of theory and utilizing a continuum polarizable solvent model (CPCM) for modeling the effect of used solvent (acetonitrile). The obtained ECD spectra extracted with SpecDis 1.7 with a half-band of 0.2–0.3 eV were Boltzmann-averaged, and a UV correction of −30 to +30 nm was applied to compare them with experimental spectra obtained in acetonitrile.

Preferential Cytotoxicity Assay against PANC-1 Cells

The plant dichloromethane extract and all isolated compounds (1–18) were evaluated for their preferential cytotoxicity activity against human pancreatic cancer cells (PANC-1), based on a previously established method.[4]

Assessment of Morphology and Apoptosis of PANC-1 Cancer Cells

Compounds 2 and 6, as the most active and as a new compound, respectively, were selected to investigate their real-time effect on cell morphology and apoptosis of PANC-1 cells. For this study, PANC-1 cells were seeded in 12-well plates (1 × 106) and incubated in a humidified CO2 incubator for 24 h for cell attachment. The cells were then washed twice with PBS and treated with vehicle control or test compounds at a concentration of 5 μM in NDM and subjected to time-lapse microscopy using an EVOS digital microscope fitted with stage-top incubator in an interval of 10 min for 24 h.

Colony Formation Assay

PANC-1 cells were plated in 12-well plates at a density of 500 cells/well in DMEM (1 mL/well) and incubated at 37 °C under humidified 5% CO2 for 24 h for cell attachment. The cells were treated with compound 2 or compound 6 at concentrations of 12.5, 25, and 50 μM in DMEM and allowed to grow for 10 days. After 10 days, cells were washed with PBS, fixed with 4% formaldehyde, and stained with crystal violet for 10 min before measurement. Each experiment was repeated three times. The colony area measurement was carried out by the ImageJ plugin “Colony Area”. Statistical differences were analyzed with GraphPad Prism using the Student’s t test. p < 0.05 was considered significant (*).