Novel Cytotoxic Sesquiterpene Coumarin Ethers and Sulfur-Containing Compounds from the Roots of Ferula turcica

Six new sesquiterpene coumarin ethers, namely turcicanol A (1), turcicanol A acetate (2), turcicanol B (3), turcica ketone (4), 11′-dehydrokaratavicinol (5), and galbanaldehyde (6), and one new sulfur-containing compound, namely turcicasulphide (7), along with thirty-two known secondary metabolites were isolated from the root of the endemic species Ferula turcica Akalın, Miski, & Tuncay through a bioassay-guided isolation approach. The structures of the new compounds were elucidated by spectroscopic analysis and comparison with the literature. Cell growth inhibition of colon cancer cell lines (COLO205 and HCT116) and kidney cancer cell lines (UO31 and A498) was used to guide isolation. Seventeen of the compounds showed significant activity against the cell lines.


Introduction
Cancers are rapidly increasing in incidence worldwide and, in total, are the second most important cause of death worldwide. According to research by the World Health Organization (WHO), cancer was the cause of death of 10 million people in 2020 [1]. Cancer is still an incurable disease; thus, there is a need to find new molecules in this field. Türkiye is one of the leading countries in its herbal richness, biodiversity, and ethnobotanical knowledge [2,3]. An important source that inspires researchers in the discovery of pharmaceuticals for the treatment of many diseases is the use of botanical resources. According to studies, 81% of cancer drugs approved between 1940 and 2014 are compounds of natural origin [4].
The genus Ferula is one of the largest genera of the Apiaceae family and ranks third in the world and first in Asia, with approximately 185 species [5]. About 26 species of Ferula grow in Türkiye (Iran-Turan region), 16 of which are endemic, and they are commonly referred to as "Çaksir" or "Çasir" [6,7]. Ferula turcica Akalın, Miski, & Tuncay is a new species defined as a member of the section Merwia in Türkiye [6]. The use of gum-like resins (oleo-gum-resin) obtained from Ferula species for the treatment of several diseases, including cancer, for thousands of years has been recorded in various sources, including Dioscorides' De Materia Medica and Avicenna's The Canon of Medicine [8][9][10][11]. The compounds identified in Ferula species and frequently encountered in gum-resin drugs obtained from Ferula species are mostly sesquiterpene esters [12][13][14], sesquiterpene coumarin ethers [15,16], and sulfur-containing substances [17,18].
Studies with sesquiterpene coumarin ethers have shown that these secondary metabolites have cytotoxic activity; they induce apoptosis in Jurkat-derived apoptotic cells and contribute to tumor suppression by inhibiting macrophage secretion and facilitating beneficial phenotypes [19][20][21]. Due to the high affinity of sesquiterpene coumarins such as conferone toward the -p-glycoprotein (Pgp) transporter, conferone has a synergistic effect on the cytotoxic activity of cancer drugs, such as vinblastine, whose effectiveness is reduced in the treatment of cancer [22]. Therefore, sesquiterpene coumarins constitute an important group for promising new drug discovery in the field of cancer.
In this study, the dichloromethane extract of Ferula turcica roots belonging to the Merwia section of the Ferula species in Türkiye was investigated for its cytotoxic secondary metabolites.

Results
The dichloromethane and methanol extracts of the roots of Ferula turcica were tested against COLO205 (colon), HCT116 (colon), UO31 (kidney), and A498 (kidney) cancer cell lines. Comparison of the cytotoxic activity of dichloromethane and methanol extracts of the roots of F. turcica showed that the cytotoxic constituents were mainly concentrated in the dichloromethane extract (Table 1). The dichloromethane extract of the roots of F. turcica was subjected to Sephadex LH-20 fractionation, followed by preparative HPLC with reverse-phase C18 columns to yield 7 novel ( Figure 1) and 30 known compounds.
Molecules 2023, 28, 5733 2 of 19 obtained from Ferula species are mostly sesquiterpene esters [12][13][14], sesquiterpene coumarin ethers [15,16], and sulfur-containing substances [17,18]. Studies with sesquiterpene coumarin ethers have shown that these secondary metabolites have cytotoxic activity; they induce apoptosis in Jurkat-derived apoptotic cells and contribute to tumor suppression by inhibiting macrophage secretion and facilitating beneficial phenotypes [19][20][21]. Due to the high affinity of sesquiterpene coumarins such as conferone toward the -p-glycoprotein (Pgp) transporter, conferone has a synergistic effect on the cytotoxic activity of cancer drugs, such as vinblastine, whose effectiveness is reduced in the treatment of cancer [22]. Therefore, sesquiterpene coumarins constitute an important group for promising new drug discovery in the field of cancer.
In this study, the dichloromethane extract of Ferula turcica roots belonging to the Merwia section of the Ferula species in Türkiye was investigated for its cytotoxic secondary metabolites.

Results
The dichloromethane and methanol extracts of the roots of Ferula turcica were tested against COLO205 (colon), HCT116 (colon), UO31 (kidney), and A498 (kidney) cancer cell lines. Comparison of the cytotoxic activity of dichloromethane and methanol extracts of the roots of F. turcica showed that the cytotoxic constituents were mainly concentrated in the dichloromethane extract (Table 1). The dichloromethane extract of the roots of F. turcica was subjected to Sephadex LH-20 fractionation, followed by preparative HPLC with reverse-phase C18 columns to yield 7 novel ( Figure 1) and 30 known compounds.

Characterization of Cytotoxic Compounds
Six new sesquiterpene coumarin ethers, namely turcicanol A (1), turcicanol A acetate (2), turcicanol B (3), turcica ketone (4), 11 -dehydrokaratavicinol (5), and galbanaldehyde (6), and a new sulfur-containing compound, namely turcicasulphide (7), were isolated from the dichloromethane extract of the roots of Ferula turcica (Figure 1). Turcicanol (1) was isolated as an amorphous white powder. The (+)-HRESIMS of 1 showed a [M + H] + molecular ion peak at m/z 383.2219, suggesting a molecular formula of C 24 H 30 O 4 for 1 with ten degrees of unsaturation. The 1 H-NMR spectrum of 1 was closely similar to that of conferol (8) (Supplementary Materials Figure S67); thus, this compound should be an unsaturated bi-cyclic drimane sesquiterpene ether of umbelliferone. The most significant difference between the 1 H-NMR spectra of conferol (8) and turcicanol A (1) was the lack of ABX signals of the H-11 a and H-11 b protons located at δ H 4.02 and 4.17 ppm (each 1H, dd). The 1 H-NMR spectrum of 1 displayed two AB-type doublets at δ H 4.40 and 4.55 ppm (each 1H) (see Table 2); such difference strongly suggests that the double bond of 1 was located between C-8 and C-9 , and the H-9 proton of conferol (8) was not present.  Figure S9 and Figure 2b) clearly confirmed the relative stereochemistry of turcicanol A as depicted in the formula 1 (Figure 1). Thus, turcicanol A (1) is a C-8 -C-9 double-bond isomer of conferol (8).
Turcicanol (1) was isolated as an amorphous white powder. The (+)-HRESIMS of 1 showed a [M + H] + molecular ion peak at m/z 383.2219, suggesting a molecular formula of C24H30O4 for 1 with ten degrees of unsaturation. The 1 H-NMR spectrum of 1 was closely similar to that of conferol (8) (Supplementary Materials Figure S67); thus, this compound should be an unsaturated bi-cyclic drimane sesquiterpene ether of umbelliferone. The most significant difference between the 1 H-NMR spectra of conferol (8) and turcicanol A (1) was the lack of ABX signals of the H-11′a and H-11′b protons located at δH 4.02 and 4.17 ppm (each 1H, dd). The 1 H-NMR spectrum of 1 displayed two AB-type doublets at δH 4.40 and 4.55 ppm (each 1H) (see Table 2); such difference strongly suggests that the double bond of 1 was located between C-8′ and C-9′, and the H-9′ proton of conferol (8) was not present. The 13     Turcicanol A acetate (2) was isolated as an amorphous white powder. The [M + H] + molecular ion peak observed at m/z 425.2325 indicated a C 26 H 32 O 5 molecular formula for 2 with 11 degrees of unsaturation. The 1 H-NMR spectrum of 2 was similar to that of turcicanol A (1) with the exception of the ca. 1.5 ppm downfield shift of the H-3 signal to δ 4.69 ppm, and the presence of a methyl singlet at δ H 2.07 ppm clearly suggested the presence of an acetoxy group in 2. The HMBC correlation from H-3 (δ H 4.96) to the ester carbonyl at δ C 171.1 established the acetyl group at position 3 of 2. The 13 C-NMR, 2D-COSY, HSQC, and HMBC spectra (Supplementary Materials, Figures S14-S17 and Figure 3a and Table 2) confirmed the proposed structure of 2 as turcicanol A acetate. In addition, the NOE correlations observed in the 2D-NOESY spectrum of 2 (Supplementary Materials, Figure S18 and Figure 3b) clearly confirmed the relative stereochemistry of turcicanol A acetate as depicted in Figure 1. molecular ion peak observed at m/z 425.2325 indicated a C26H32O5 molecular formula 2 with 11 degrees of unsaturation. The 1 H-NMR spectrum of 2 was similar to tha turcicanol A (1) with the exception of the ca. 1.5 ppm downfield shift of the H-3′ signa δ 4.69 ppm, and the presence of a methyl singlet at δH 2.07 ppm clearly suggested presence of an acetoxy group in 2. The HMBC correlation from H-3′ (δH 4.96) to the e carbonyl at δC 171.1 established the acetyl group at position 3′ of 2. The 13 C-NMR, COSY, HSQC, and HMBC spectra (Supplementary Materials, Figures S14-S17 and 3a a Table 2) confirmed the proposed structure of 2 as turcicanol A acetate. In addition, NOE correlations observed in the 2D-NOESY spectrum of 2 (Supplementary Materi Figures S18 and 3b) clearly confirmed the relative stereochemistry of turcicanol A ace as depicted in Figure 1. Turcicanol B (3) was isolated as an amorphous white powder. The (+)-HRESIMS turcicanol B (3) showed an [M + H] + molecular ion peak at m/z 383.2225, which indica a C24H30O4 molecular formula for 3 with ten degrees of unsaturation. The 1 H-N spectrum of 3 was similar to that of turcicanol A (1). The only difference was the shif the H-3′ proton signal to δH 3.29 ppm (see Table 2) from δH 3.49 ppm. The 1 H-N spectrum of turcicanol B (3) showed a hydroxyl geminal H-3′ proton as a dd (J = 4.5, 1 Hz), suggesting an axial orientation; thus, 3′-OH of turcicanol B (3) should be equato (i.e., α-OH) ( Table 2). The 13 Table 2) from δ H 3.49 ppm. The 1 H-NMR spectrum of turcicanol B (3) showed a hydroxyl geminal H-3 proton as a dd (J = 4.5, 11.7 Hz), suggesting an axial orientation; thus, 3 -OH of turcicanol B (3) should be equatorial (i.e., α-OH) ( Table 2). The 13 Table 2) further corroborated the proposed structure of 4 as turcica ketone. Also, the strong anisotropic shift of the de-shielded H-2 protons (0.6-0.8 ppm) strongly suggested the presence of a keto group at the C-3 position (Table 2). Furthermore, the NOE correlations observed in the 2D-NOESY spectrum of 4 (Supplementary Materials, Figure S36 and   Table 2) further corroborated the proposed structure of 4 as turcica ketone. Also, the strong anisotropic shift of the de-shielded H-2′ protons (0.6-0.8 ppm) strongly suggested the presence of a keto group at the C-3′ position (Table 2). Furthermore, the NOE correlations observed in the 2D-NOESY spectrum of 4 (Supplementary Materials, Figures S36 and 5b) clearly confirmed the relative configuration of turcica ketone as shown in formula 4 ( Figure 1).  Figure S91) and compound 5 are similar except for the presence of two methylene proton singlets at δH 4.83 and δH 4.93 ppm in the 1 H NMR of 5 (brs 1H for each) and the lack of hydroxyl adjacent to the C12′ and C13′ methyl signals of karatavicinol in 5 suggested that a double bond between C-11′ and C-12′ was present in 5. Furthermore, due to the allylic positioning of the C-10′ hydroxyl group, the chemical shift of the oxygenated methine proton at C-10′ in the 1 H-NMR spectrum of 5 was shifted downfield ca. 0.7 ppm to δ 4.05 ppm ( Table 2). The 13 C-NMR, 2D-COSY, HSQC, and HMBC spectra (Supplementary Materials, Figures S41-S44 and 6a and Table 2) confirmed the proposed structure of 5 as 11′-dehydrokaratavicinol. The key NOE correlations observed in the 2D-NOESY spectrum of 5 (Supplementary Materials,   Figure S91) and compound 5 are similar except for the presence of two methylene proton singlets at δ H 4.83 and δ H 4.93 ppm in the 1 H NMR of 5 (brs 1H for each) and the lack of hydroxyl adjacent to the C12 and C13 methyl signals of karatavicinol in 5 suggested that a double bond between C-11 and C-12 was present in 5. Furthermore, due to the allylic positioning of the C-10 hydroxyl group, the chemical shift of the oxygenated methine proton at C-10 in the 1 H-NMR spectrum of 5 was shifted downfield ca. 0.7 ppm to δ 4.05 ppm ( Table 2). The 13 C-NMR, 2D-COSY, HSQC, and HMBC spectra (Supplementary Materials, Figures S41-S44 and Figure 6a and Table 2) confirmed the proposed structure of 5 as 11 -dehydrokaratavicinol. The key NOE correlations observed in the 2D-NOESY spectrum of 5 (Supplementary Materials, Figure S45 Table 2) confirmed the proposed structure of compound 6 as galbanaldehyde. The NOE correlations observed in the 2D-NOESY spectrum of 6 (Supplementary Materials, Figure S54 and Figure 7b) confirmed the relative configuration of galbanaldehyde as depicted in the formula 6 (Figure 1), which is identical to that of galbanic acid (27) [23][24][25].

Cytotoxic Activity
The pure compounds of Ferula turcica were tested against colon cancer cell lines (COLO 205 and HCT 116) and kidney cell lines (UO31 and A498). The results are given in Table 3. According to the cytotoxicity studies, the acetylation of the hydroxyl C-10 position in the sulfur-bearing compounds (compounds 38, 39) preserved the cytotoxic activity in the UO31 and COLO205 cell lines and slightly increased the activity in HCT116; however, the loss of hydroxyl at the C-10 and substitution of a benzene ring led to the loss of cytotoxic activity (see compound 7) in all cell lines. The cytotoxic activity was observed in colladonin (9) but not in badrakemin (10), whose hydroxyl substitution at the C-3 was in the β position. In addition, the cytotoxic activity in gummosin (20), which is the C-9 epimer of badrakemin (10), and colladonin (9) increased the cytotoxic activity with axial stereochemistry. The acetate derivatives of badrakemin (10) and gummosin (20) as well as the cytotoxic activity of badrakemin acetate (11) increased slightly in HCT116 cell lines, and the activity of gummosin acetate (21) decreased in COLO205 cell lines, decreased in HCT116, increased in A498, and decreased in UO31 cell lines. As for the ketone derivatives, the oxidation products at C-3 of badrakemin (10) and gummosin (20) as well as oxidation did not cause any increase in badrakemone (12); it caused a decrease in cytotoxic activity in mogoltadone (22). Conferol (8) showed cytotoxic activity in colon cancer cell lines. We noted that even with the double-bond shift to C-8 and C-9 positions, cytotoxic activity in turcicanol A (1) is still significant in HCT116 colon cancer cells, just as in its isomers conferol (8) and gummosin (20). It was determined that the acetylation of the compound (turcicanol A acetate 2) causes activity to be lost in these four cell lines. In turcicanol B (3), an increase in cytotoxic activity was observed in the UO31 kidney cell line. When the cytotoxic activity results of the pure compound turcica ketone (4), an isomer of badrakemone (12) and mogoltadone (22), were examined, the endocyclic double bond was more cytotoxic than the exocyclic double bond, as in badrakemone (12) and mogoltadone (22). The cytotoxic activity results of galbanic acid (27) and its aldehyde derivative (6) showed that the aldehyde form of C-3 increased the cytotoxic activity on HCT116 colon cancer cell lines.
According to the IC 50 values, seventeen compounds (1, 3, 4, 6, 15, 17, 19-23, 25, 28-30, 32, 38, and 39) showed cytotoxic activity in this study. The most effective compounds against cancer cell lines were determined as conferol (8), gummosin (20), and persicasulphide A (38) and C (39), which is in agreement with the literature data. While it is true that no clear structure-activity patterns emerged from the testing, the diversity of structures may have limited any such conclusions.

General Experimental Procedures
LC-MS analysis was performed with Agilent Technologies ® 6130 Quadrupole LC/MS (Santa Clara, CA, USA). UV-vis spectra were obtained using Shimadzu ® UV-1700 Phar-maSpec (Kyoto, Japan). IR spectra were determined using Bruker ® Alpha FT-IR (Billerica, MA, USA). NMR spectra of the compounds were acquired on a Bruker ® Avance III spectrometer operating at 600 MHz for 1 H and 150 MHz for 13 C in deuterated chloroform (Billerica, MA, USA). HRESIMS analysis of compounds 1-6 were performed using Agilent ® 6530 Accurate Mass Q-TOF (Santa Clara, CA, USA), while the HRESIMS data of turcicasulphide (7) were acquired on a Thermo Scientific-Q Exactive ® (Waltham, MA, USA). Optical rotation data were acquired using a Rudolph Analytical Autopol V Plus ® in dichloromethane (Hackettstown, NJ, USA). A Buchi rotary evaporator was used to evaporate the solvent of the extract (Buchi, Flawil, Switzerland). A Sephadex LH-20 (Sigma Chem. Co. 25-100 µm) (GE Healthcare, Chicago, IL, USA) column (5 × 100 cm) was used for the initial fractionation. A Gilson ® PLC 2050 was used for the further purification of the compounds (Saint-Avé, France). Hexane, dichloromethane, methanol, and acetonitrile (Merck, Darmstadt, Germany) were used during the chromatographic analyses.

Plant Material
The plant root materials used in this study were collected from the shores of Tuz Lake in Konya (Yavşan Tuzlası) on 16 June 2015, while the plant was fruiting, and the voucher specimen was archived in ISTE (Istanbul University Faculty of Pharmacy Herbarium) with the number 116,464. The species was identified by Prof. Emine Akalın and Hüseyin Onur Tuncay [6].

2DAY (Colon 2) XTT Cytotoxic Activity Assay
The two-XTT bioactivity test is an in vitro colorimetric cytotoxic activity test developed by the NCI MTP Assay Development and Screening Section [50] and used for this study. Colon (COLO205, HCT116) and kidney (A498, UO31) cancer cell lines were used during the tests. RPMI-1640 (Roswell Park Memorial Institute, Buffalo, NY, USA) medium, 10% FBS (fetal bovine serum), 1% glutamine, and 1% penicillin/streptomycin solutions were used for cell growth and treatment. Transfers were performed under laminar air flow in a sterile environment. The suspension containing the cells was seeded into 96-well plates with a volume of 45 µL with 3.5 × 10 5 cells per well. Then, the plate was incubated at 37 • C and 5% CO 2 for 24 h. The extract and pure compounds prepared in DMSO were added and incubated for another 48 h. After incubation, 10 µL of the tetrazolium salt XTT (2,3-bis[2-methoxy-4-nitro-5-sulfophenyl]-2H-tetrazolium-5-carboxanolide) was applied to the cells. After 4 h of incubation, dead cells were not stained with formazan dye, while viable cells could be counted in the EnVision plate reader under UV light (450 nm and 650 nm). Sanguinarine chloride hydrate was used as a positive control in the experiment.

Conclusions
A dichloromethane extract of Ferula turcica root was studied for the first time. Seven new and thirty-two known compounds (1-39) were isolated from the dichloromethane extract using bioactivity-directed fractionation, and their cytotoxic activities were investigated against COLO205, HCT116, A498, and UO31 cancer cell lines. The structures of the new compounds were determined by spectroscopic techniques, and the spectral data of the compounds are presented for the first time. Some structure-activity relationships of the compounds for cytotoxic activities illuminate the effects of substitution, oxidation, acetylation, and double bond positions.