**1. Introduction**

Cancer is responsible for many people's deaths each year, and it represents the second most common cause of human death worldwide [1]. Even though this disease's incidence and mortality have declined over the past 20 years, there is still no reliable therapy for eradicating it. Recently, in addition to the traditional drug discovery approach, a novel strategy called drug repositioning has emerged. It lies in using the old for a novel purpose.

#### **Citation:** Rimpelová, S.;

Zimmermann, T.; Drašar, P.B.; Dolenský, B.; Bejˇcek, J.; Kmoníˇcková, E.; Cihláˇrová, P.; Gurská, S.; Kuklíková, L.; Hajd ˚uch, M.; et al. Steroid Glycosides Hyrcanoside and Deglucohyrcanoside: On Isolation, Structural Identification, and Anticancer Activity. *Foods* **2021**, *10*, 136. https://doi.org/10.3390/ foods10010136

Received: 14 December 2020 Accepted: 3 January 2021 Published: 11 January 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

This approach is economically much more feasible and faster than the traditional approach of drug approval [2].

Interestingly, drug repositioning has also been used for some cardiac glycosides (CGs) as a potential remedy for cancer. Several undergoing clinical trials on CG administration for cancer in mono or combination therapy can be found in [3]. Moreover, also promising is the fact that a lower incidence in some types of cancer in patients on CG therapy (cardiac conditions) has been reported [4]. While another study claims the opposite [5], others stand somewhere in between [6,7]. However, what is quite certain and makes CGs a new hope for cancer treatment are clinical data showing that CGs, mainly digoxin, significantly prolong the survival of cancer patients otherwise treated with traditional chemotherapeutics. What is also beneficial is the fact that CGs at multiple levels affect the immune response and trigger immunogenic cell death, which significantly contributes to their anticancer activity [8].

CGs are secondary plant metabolites found mainly in *Digitalis purpurea* L. and *Digitalis lanata* Ehrh. (digitoxin, Dg; digoxin), *Strophanthus gratus* (ouabain, Ob), and *Nerium oleander* L. (oleandrin). Their biological effect is associated with the interaction with Na+/K+-ATPase (NKA), the integral membrane protein of animal cells maintaining the balance of sodium and potassium ions. The CG pharmacophore is the structure of 5β,14β-androstane-3β,14-diol, which is substituted at the C-3 position by a saccharide moiety and the C-17 position by an unsaturated lactone [9]. According to the type of lactone, CGs are classified into cardenolides and bufadienolides; 5β,14β-androstane-3β,14-diol may be substituted in some other positions [10]. Biological activity has been shown to decrease after saturation of the lactone double bond [11,12]. Furthermore, biological activity is also affected by the type and number of carbohydrate units at the C-3 position, and it increases with the decreasing number of carbohydrate units [13,14]. The only exception is aglycone, the biological activity of which is lower than that of its glycosylated variants [13,15].

As aforementioned, CGs, as a unique group of metabolites, have been extensively utilized for the treatment of various heart conditions, and recently, they were also explored as possible anticancer agents (reviewed in [16]). Based on the positive ionotropic effect they induce, they are introduced as drugs in the treatment of heart failure and cardiac arrhythmias. The ionotropic effect is associated with NKA inhibition, which leads to an increase in intracellular Na+ concentration, resulting in an augmented influx of Ca2+ ions into the cell, followed by contraction of the heart muscle [17]. However, the interaction of CGs with NKA may not only be associated with disruption of Ca2+ homeostasis, as SERCA (Sarco-Endoplasmic Reticulum Calcium ATPase) inhibitors do [18]. It has also been found that at low (subclinical) concentrations of CGs, where there is little or no inhibition of NKA, this enzyme can serve as a receptor that activates non-receptor tyrosine kinases. This in turn leads to activation of mitogen-activated protein kinase (MAPK) and the triggering of the Ras/MAPK signaling pathway [19,20]. This further leads to stimulation or inhibition of cell proliferation, depending on the cell type: cancer cell proliferation is inhibited, while primary noncancerous cells are not [21–23].

Besides this, CGs are also potent activators of the immune system response by induction of immunogenic cell death, which is a tremendous advantage over some other currently used chemotherapeutics, such as cisplatin, which lacks this effect. The immunogenic cell death is achieved by calreticulin exposure to the cell surface and the secretion of ATP and high-mobility group box 1 protein [24].

One of the widely studied CGs in terms of cancer is digoxin, currently included in a clinical trial for cancer combination therapy, in which it is co-administered with cisplatin (ClinicalTrials.gov). Another interesting and, in cancer treatment, possibly potent CG is Dg. This cardenolide type of CG also binds and inhibits NKA [25], and was shown to be potent in inhibiting cancer cell proliferation and cell cycle arrest [26–28] already at low nanomolar concentrations. Such concentrations are commonly found in the blood plasma of patients treated with Dg due to heart failure [13–15].

However, both Dg and digoxin lack cancer cell selectivity, resulting in high systemic toxicity often encountered in patients on CG therapy. Based on this, the seeking of novel CGs with improved properties and enhanced performance has not yet been finished. Therefore, we report on two very interesting CGs: hyrcanoside (Hyr, Figure 1) and deglucohyrcanoside (deHyr, Figure 1), about which there has not yet been much information. Hyr is a secondary plant metabolite of *Coronilla varia* L. Regarding cytotoxicity, the only information available is on Hyr-containing alcoholic extracts inhibiting the growth of KB cell [29] and human lymphocytic leukemia (P-388) and nasopharynx carcinomas (9KB) [30]. From earlier reports, it is obvious that deglucohyrcanoside acts as digoxin, but its cytotoxicity is several times lower [31–33], which indicates the potential of this CG as a therapeutic used in cancer treatment.

**Figure 1.** Structures of natural C-19-oxo cardiotonics and digitoxin. Carbon numbering of Hyr and deHyr is in grey. The ring designation of the steroid ring is highlighted by dots (i.e., the A ring is marked by a dot, B with , C with , and D with dot).

#### **2. Materials and Methods**

#### *2.1. Materials*

For thin-layer chromatography (TLC), aluminum silica gel sheets for detection in UV light (TLC Silica gel 60 F254, Merck, Prague, Czech Republic) were used. For TLC visualization, a diluted solution of H2SO4 in MeOH was used, and plates were heated. For column chromatography, 30–60 μm silica gel (ICN Biomedicals, Costa Mesa, CA, USA) was used. The NMR spectra were recorded by a 500 MHz instrument (JEOL, Tokyo, Japan) at 25 ◦C. The chemical shifts (*δ*) are presented in ppm, and the coupling constants (*J*) are

presented in Hz. The 1H and 13C chemical shifts are referenced to tetramethylsilane using the solvent signals CHD2SOCD3 2.50 ppm, CD3SOCD3 39.52 ppm, CHD2OD 3.31 ppm, and CD3OD 49.00 ppm. For the signal assignments and obtaining of the coupling constants, combinations of the following standard NMR sequences were used: 1D and selective homodecoupled 1H NMR spectra, 13C NMR spectra with or without 1H BB decoupling, 2D gCOSY, gTOCSY, dqf-COSY with 25% non-uniform sampling (NUS), 2D NOESY and ROESY (mix time 400 and 350 ms), 2D gHSQC and gHMBC (7 Hz) with adiabatic pulses and 25% NUS, 1D selective gNOESY1D and gROESY1D (various mix time), and gTOCSY1D (10, 15, 20, 40, 80, 160 ms). The content of Hyr and deHyr (dried extracts were dissolved in 50% MeOH) was verified using mass spectrometry, employing direct injection into the electrospray ionization source of QTRAP 6500+ (AB Sciex, Framingham, MA, USA) with a Turbo V Ion source. The mass spectrometer was operated in a positive scan mode with *m*/*z* ranging between 100 and 1000. The ion source settings were as follows: a temperature of 200 ◦C, a capillary voltage of 5500 V, curtain gas of 20 psi, nebulizer gas, and heater gas of 30 psi. The data were acquired and evaluated using Analyst 1.6.3 software (AB Sciex, Framingham, MA, USA). In further LC-MS analyses, the Quadrupole LC/MS (ESI ionization) with the Infinity III LC system (Agilent Technologies, Santa Clara, CA, USA) was used for LR-MS and HPLC analyses (C18 column: 100 mm, and UV detection).

#### *2.2. Extraction of C. varia Seeds and Isolation of Compounds*

Seeds of crown vetch (*Coronilla varia* L., Fabaceae, sometimes also *Securigera varia* (L.) Lassen, or *Coronilla pendula* Kit.) purchased from Agrostis Trávníky Ltd., Rousínov u Vyškova (CZ) were ground on an electric blade coffee grinder to a fine powder. Powdered seeds (500 g) were mixed with ethanol (500 mL) and kept at room temperature for seven days. Then, the mixture was filtered through frita, the filtrate was evaporated, re-dissolved in 200 mL of EtOH, and transferred onto a chromatography silicagel column (length of 24 cm, a diameter of 6 cm, 800 g of silica gel). Seed contents were eluted at first by ethanol and then by a gradient mixture of EtOH-H2O (10:1→ 3:1, *v*/*v*), respectively. Thus, two fractions were obtained: ethanolic one (H1) and fractions eluted by the mixture of aqueous EtOH (H2). MS and TLC evaluations of fraction H1 showed that mainly daphnoretin, scopoletin, and umbelliferone (structures are shown in Figure S1) were present (after evaporation, a dry residue of 85 g) [29]. On the other hand, fraction H2 (we obtained a brown-red syrupy evaporate of 70 g) contained solely traces of the aforementioned three compounds plus (−)-epicatechin (Figure S1), deHyr, and some unidentified polar compounds [29]. As a major component, Hyr was identified. The content of both fractions was subjected to column chromatography.

Chromatographic purification of H1 (84 g) over a silica gel column (length of 24 cm, a diameter of 6 cm, 800 g of silica gel, using DCM-MeOH, 20:1→ 3:2, (*v*/*v*) as eluent) afforded mostly triacylglycerols, as expected, and some minor products. Only about 120 mg of crude Hyr was obtained. This was in accordance with the preliminary LC-MS spectra recorded for fraction H1, which showed only a negligible amount of compounds of interest. On the contrary, chromatographic purification of 69 g of the second major fraction H2 (length of 30 cm, a diameter of 4 cm, 600 g of silica gel, using DCM-MeOH, 10:1→ 3:2, *v/v* as an eluent) provided four fractions (f1–f4), and LC-MS confirmed the presence of both deHyr (in fraction 2) and Hyr (in fraction 4).

In addition, other known compounds contained in *C. varia* were detected by LC-MS in fraction f2, namely daphnoretin and umbelliferone in fraction f1 and (−)-epicatechin and scopoletin in f3 (structures are shown in Figure S1). DeHyr and Hyr fractions were further purified. Hyr fraction f4 of H2 was purified using a silica gel column (length of 30 cm, a diameter of 4 cm, 600 g of silicagel, using DCM-MeOH, 10:1→ 6:1, (*v*/*v*) as eluent) to provide fairly pure Hyr. Due to the fact that Hyr contains both hydrophilic disaccharide and lipophilic steroidal moiety, standard chromatographic purification was not efficient enough to provide pure Hyr (or deHyr). Thus, the final processing step was the crystallization of Hyr from the solution. Finally, the crude in absolute MeOH was dissolved, and the careful addition of Et2O showed signs of crystals being formed. Crude Hyr was thus dissolved in MeOH (100 mL), then, Et2O (50 mL) was slowly added and the solution left to stay at 4 ◦C overnight. The white solids formed were fritted and washed with ether. Hyr was obtained as white crystals (1.634 g, 2.40 mmol; 3.3%). Purified deHyr fraction H1 (length of 30 cm, a diameter of 4 cm, 600 g of silica gel, using DCM-MeOH, 10:1, *v*/*v* as an eluent) provided a notably lower amount of crude deHyr. This product was also crystallized by dissolving it in a mixture of MeOH (10 mL) and acetone (20 mL) with the subsequent addition of Et2O (20 mL), then it was left overnight at 4 ◦C. This process provided deHyr as white crystals (96 mg, 0.19 mmol; 0.2%). MS analyses confirmed the corresponding molecular composition (Figures S6 and S7). The purity and identity of the isolated compounds were checked by LC analysis (Figures S8 and S9) and NMR (Section 3.1) spectroscopy, respectively.

#### *2.3. In Silico Modeling*

The Maestro program 2019-3 (Schrödinger, LLC, New York, NY, USA) was used for all structural modifications and subsequent molecular docking of the tested CGs into NKA. The CG structures were obtained from the ChemSpider database (www.chemspider.com). Using the LigPrep module, the missing hydrogen atoms were added to the ligands. Then, the ligands' structure was converted to 3D, and their energy was minimized using the OPLS3e force field. The structure of NKA with the code name 4RET (organism: *Sus scrofa*) was obtained from the ProteinDataBank database (www.rcsb.org). Non-protein moieties (aspartyl phosphate, cholesterol, sucrose, digoxin, Mg2+, *N*-acetyl-D-glucosamine, phospholipids, and water) were removed from the NKA structure. This was followed by adding hydrogen atoms to the NKA structure. Then, the amino acids were assigned a protonation state corresponding to pH = 7 using the PROPKA function, and the energy of the molecule was minimized by the OPLS3e force field. Using the amino acids L-Thr114, L-Asp121, and L-Thr797, a ligand-binding site was defined, which consisted of two cubes with an edge length of 15 and 35 Å (a small and large cube, respectively). The center of the molecule should not leave the smaller cube and the molecule as a whole should not leave the larger cube. Subsequently, a subset of a spherical pocket with a radius of 4 Å was defined using the constraints function, into which the C and D cycles of the steroid skeleton were fixed. The CG ligands were docked into NKA with an extra precision mode.

#### *2.4. Cell Lines*

If not indicated otherwise, the cell lines were purchased from the American Type Culture Collection (ATCC). The CCRF-CEM line is derived from T-cell childhood acute lymphoblastic leukemia, which shows the highest chemosensitivity in our tumor cell lines panel. K562 is the erythroid-myeloid precursor cell line derived from chronic myeloid leukemia carrying the *BCR-ABL* hybrid gene. A549, MCF-7, PC-3, 5637, U-2 OS, and MiaPaCa-2 are cells derived from lung, breast, prostate, bladder carcinoma, osteosarcoma, and pancreatic adenocarcinoma, respectively. HEK 293T are transformed human embryonic kidney cells, and L929 (Sigma, St. Louis, MS, USA) are transformed mouse fibroblasts. HCT116, cells from colorectal carcinoma, and their *p53* gene knock-out variant (HCT116p53-/-) were purchased from Horizon Discovery. This cell line is a model of human tumors bearing *p53* loss-of-function mutations or biallelic deletion of the *p53* gene, frequently associated with poor prognosis. The MRC-5 and BJ cells were used as noncancerous cells; more specifically, they are human fibroblasts from the lungs and foreskin, respectively. According to the supplier's recommendations, all cells were cultured at 37 ◦C in a 5% CO2 atmosphere and 100% humidity. The culture media used were: DMEM, RPMI 1640, and MEM (according to a cell line); all were supplemented with 5 g L−<sup>1</sup> glucose, 10% fetal calf serum, 2 mM glutamine, 100 U mL−<sup>1</sup> penicillin, 100 μg mL−<sup>1</sup> streptomycin, and NaHCO3. Cells were passaged every two or three days using 0.25% trypsin plus 0.01% EDTA (ethylenediamine tetraacetic acid) in phosphate-buffered saline.

#### *2.5. MTS Cytotoxic Assay*

To evaluate compound cytotoxicity, cells were seeded in 384-well microtiter plates in a volume of 30 mL. The next day, aliquots of the tested derivatives were transferred with Echo550 acoustic liquid handler (Labcyte) to obtain dose response curves with dilution factor 4. The experiments were performed in technical duplicates and three or more biological replicates. After 72 h of incubation in a humidified incubator, 4 mL of the MTS/PMS stock solution were pipetted into each well. After another 1–4 h of incubation, the absorbance at 490 nm was measured using an EnVision Multilabel Plate Reader (PerkinElmer). IC50 values were calculated from the appropriate dose response curves in Dotmatics software using the following equation: IC50 = (ODdrugexposed well/mean ODcontrol wells) × 100% [34].

#### *2.6. Mice and Peritoneal Primary Cells*

Female mice of the inbred strain C57BL/6, eight to ten weeks old, were purchased from Charles River Deutschland (Sulzfeld, Germany). They were kept in transparent plastic cages in groups of ten. The animals were housed with food and water ad libitum, lighting was set on 6–18 h, and the temperature was set at 22 ◦C. All protocols were approved by the institutional ethics committee (MSMT15894/2013-310). Animals, killed by cervical dislocation, were intraperitoneally injected with 8 mL of sterile saline. Pooled peritoneal cells collected from mice were washed in sterile saline, re-suspended in the culture medium, and seeded into 96-well microplates in final 100-μL volumes (Costar, Cambridge, MA, USA). The final density of the cells was 0.25 × <sup>10</sup><sup>6</sup> per mL−1. The cultures were maintained with or without compounds for 24 h at 37 ◦C with 5% CO2 in a humidified Heraeus incubator in complete RPMI-1640 (Merck-Sigma, St. Louis, MS, USA), which contained 10% heat-inactivated fetal bovine serum, 2 mM of L-glutamine, 50 <sup>μ</sup>g·mL−<sup>1</sup> of gentamicin, and <sup>5</sup> × <sup>10</sup>−<sup>5</sup> M of 2-mercaptoethanol (all Merck-Sigma, St. Louis, MS, USA). Compounds (Ob, Cy (cymarin), Hyr, deHyr) were prepared as 100 mM stock solutions in dimethyl sulfoxide (DMSO) with cell culture grade. The next dilution continued immediately before the experiment with the culture medium. To eliminate the influence of DMSO, equal levels of DMSO were added to the experimental groups.

#### *2.7. Viability of Mouse Primary Cells*

The viability of mouse peritoneal cells was determined using a colorimetric assay based on the cleavage of the tetrazolium salt WST-1 by mitochondrial dehydrogenases in viable cells (Merck-Sigma, St. Louis, MS, USA). The cells were cultured as described above. After 24 h of culture, the WST-1 was added, and the cells were kept in the Heraeus incubator at 37 ◦C for an additional 3 h. Optical density at 450/690 nm was determined. The cytotoxicity of the tested CGs was expressed as a percentage. The test samples were related to control samples consisting of untreated cells and samples with 100% dead cells evoked by 1% Triton, according to the formula: ((exp. value − control value)/(Triton value − control value)) × 100. All control and experimental variants were run in quadruplicates. The data were analyzed using GraphPad Prism software 6.05 (GraphPad, San Diego, CA, USA). Values were expressed as the mean ± standard error of the mean (SEM).

#### *2.8. Analysis of Cell Cycle Arrest and Cell Death*

CCRF-CEM cells were seeded at a density of 1 × 106 cells per one mL in six-well plates (TPP) and treated the next day with CGs at concentrations corresponding to 1× or 5× the IC50 value. Together with the CG-treated cells, a vehicle-treated sample was harvested at the same time point. After 24 h, the cells were washed with cold phosphate-buffered saline, fixed dropwise in 70% ethanol, and stored overnight at −20 ◦C. The cells were then washed with hypotonic citrate buffer, treated with RNAse (50 μg mL<sup>−</sup>1), and stained with propidium iodide. Flow cytometry using a 488 nm laser (FACS-Calibur, Becton Dickinson, NJ, USA) was used for measurement. The cell cycle was analyzed by the ModFitLT program (Verity), and apoptosis was measured in a logarithmic model expressing the percentage

of particles with a lower propidium content than cells in the G0/G1 phase (<G1) of the cell cycle in the CellQuest program (Becton Dickinson). Half of the sample was used to label cells with pH3Ser10-FITC antibody (Exbio) for subsequent flow cytometry analysis of mitotic cells [35].

#### *2.9. BrDU Incorporation Analysis*

The cells were cultured in the same way as the cell cycle analysis method. Just before harvesting, 5-bromo-2 -deoxyuridine (BrDU) of 10 μM concentration was added to the cells for pulse-labeling for 30 min. Then, the cells were fixed with −20 ◦C cold 70% ethanol and stored in a freezer for 16 h. Before antibody staining, the samples were incubated on ice for 30 min, washed once with phosphate-buffered saline (PBS), and re-suspended in 2 M of HCl for 30 min at room temperature to hydrolyze their DNA. After neutralization with 0.1 M of Na2B4O7 (borax) solution, the cells were washed with PBS containing 0.5% Tween-20 and 1% BSA. This was followed by staining with the primary anti-BrDU antibody (Exbio) for 30 min at room temperature. Then, the cells were washed with PBS and stained with a secondary anti-mouse antibody conjugated to fluorescein isothiocyanate (Merck-Sigma, St. Louis, MS, USA) at room temperature in the dark. After another washing with PBS and incubation with propidium iodide (0.1 mg·mL−1) and RNAse A (0.5 mg·mL−1) for 1 h at room temperature in the dark, the cells were analyzed by flow cytometry using a 488 nm laser (FACSCalibur, Becton Dickinson, Franklin Lakes, NJ, USA) [35].

#### *2.10. BrU Incorporation Analysis*

The cells were cultured and processed as described above. Before harvesting, the cells were pulse-labeled with 1 mM of 5-bromouridine (BrU) for 30 min. The cells were then fixed in 1% buffered paraformaldehyde with 0.05% NP-40 at room temperature for 15 min, and then stored at 4 ◦C overnight. Before measurement, they were washed with 1% glycine in PBS, washed again with PBS, and stained with primary anti-BrdU antibody cross-reactive to BrU (Exbio) for 30 min at room temperature in the dark. From this point on, the experiment was performed exactly as in the method described above [35].

## **3. Results and Discussion**

#### *3.1. Isolation and Identification of Hyr and deHyr*

Isolation of the desired substances from the plant seeds consisted of several steps (Figure 2). First, it was necessary to grind the seeds and extract them. This procedure is described in detail in Section 2.1. Two extraction solvent systems were used: EtOH and aqueous EtOH, providing two extracts designated as H1 and H2, respectively. According to MS and TLC analyses, the desired substances (Hyr and deHyr) were mainly present in the H2 fraction. This was followed by purification of the extracts by chromatographic separation on silica gel. Separation of the components from extract H2 yielded pure Hyr (3.3%) and a very small amount of deHyr (0.0002%). Hembree et al. [30] described the isolation of Hyr and deHyr from 14.8 kg of powdered seeds with the yield of semi-pure Hyr of 2.57 g (0.00017%) and 70 mg (0.000005%) of deHyr. We hypothesize that the low yield of deHyr can be explained by its absence in the plant material, and the trace amount is formed during the processing. Taken together, our procedure has improved the yield of pure Hyr and deHyr by ca. 20,000 and 40 times, respectively.

**Figure 2.** Diagram presenting the process leading to the isolation of Hyr and deHyr from the seeds of *Coronilla varia*.

A thorough literature search has revealed that the name hyrcanoside has also been used for other chemical structures. Two different compounds were named hyrcanoside (Figure S2). A cardenolide, (3β)-3-[(4-*O*-β-D-glucopyranosyl-β-D-xylopyranosyl)oxy]-14-hydroxy-19 oxocarda-4,20(22)-dienolide (Hyr) (CAS Registry No. 15001-93-1; Figure 2 and Figure S2 compound Hyr), mostly named hyrcanoside (*Coronilla*) [30,31], which is the substance isolated from *Coronilla varia* (or from *C. varia* Prilipko, a synonym of *Securigera cretica* (L.) Lassen, *Securigera securidaca* (L.) Degen et Dörfler, and some other plants) [36], and a phenolglycoside that was isolated from *Dorema hyrcanum* Koso-Pol. or *Dorema glabrum* Fisch. & C.A. Mey., 1-[2-[(6-*O*-α-D-glucopyranosyl-β-D-glucopyranosyl)oxy]-6-hydroxy-4-methoxyphenyl]ethanone (CAS Registry No. 60197-59-3; Figure S2—compound s1) [37,38]. Hereby, we suggest naming this compound hyrcanoside (*Dorema*).

To make things even more complicated, Abubakirov [39] lists the structure of Hyr (Figure 1 and Figure S2) named as securidaside, and declares it identical with steroidal hyrcanoside. However, Zatula et al., in a series of earlier research articles [40–42], presents securidaside as C4(5) saturated, 5α,11β-hydroxy derivative s2 (CAS Registry No. 18309-58-5; for structure, see Figure S2). Zatula et al. in 1969 corrected [43] the structure to be identical with Hyr, from which it was cited by Abubakirov [39].

We also attempted to confirm the molecular structures of Hyr and deHyr by 1H and 13C NMR spectra (Table S1). Unfortunately, our values of the 13C chemical shifts were not in full accordance with the ones recorded in DMSO-*d*<sup>6</sup> [30], nor in CD3OD [44] (Hyr is named as securigenin glycoside s3; see Figures S4 and S5). Thus, we performed the signal assignment of all 13C and 1H signals of both Hyr and deHyr in both solvents. By the combination of 2D HSQC, HMBC, NOESY, TOCSY, and dqf-COSY spectra, followed by the series of the selective 1D TOCSY, NOESY, ROESY, and homodecoupled 1H spectra, we were able to assign all of the signals, including stereo positions (Table S1). We concluded that our assignments and characteristics of Hyr are fully consistent with its known crystal structure and molecular models [44]. Comparison of 1H and 13C chemical shifts of Hyr in CD3OD revealed that the chemical shifts of C3 and H3 signals are interchanged with C5 and H5, respectively [44]. After this correction, the 13C chemical shifts differed from −0.15 to 0.03 ppm, and 1H chemical shifts differed from −0.01 to −0.08 ppm; thus, we concluded that our Hyr isolated from *C. varia* and the securigenin glycoside s3 (= Hyr, Figure S3; the designation is different for data comparison) [44] isolated from *S. securidaca* are identical compounds. In contrast, the values of 13C chemical shifts of Hyr differed more significantly in DMSO-*d*<sup>6</sup> [30]. Even when the obvious interchange of the signals for C4 and C22 were corrected, there were still several deviations (1–11 ppm), which cannot be removed by an interchange. Since the compound was isolated from the same herb and its structure was confirmed by comparison with the synthetic standard, we concluded that the 13C chemical shifts are confused in [30].
