Chemical and Biological Characterization of Oleanane Triterpenoids from Soy

Soyasaponins are a group of complex and structural diverse oleanane triterpenoids found in soy (Glycine max) and other legumes. They are primarily classified into two main groups − group A and B − based on the attachment of sugar moieties at positions C-3 and C-22 of the ring structures. Group A soyasaponins are bidesmosidic, while group B soyasaponins are monodesmosidic. Group B soyasaponins are further classified into two subcategories known as 2,3-dihydro-2,5-dihydroxy-6 -methyl-4H-pyran-4-one (DDMP) and non-DDMP conjugated molecules. The preparation and purification of soyasaponin molecules is complicated by the presence of bioactive soy isoflavones, which often overlap with soyasaponin in polarity and must removed from extracts before biological assessment. Soyasaponin extracts, aglycones of group A and B and individual group B soyasaponins such as soyasaponin I have been reported to posses specific bioactive properties, such as in vitro anti-cancer properties by modulating the cell cycle and inducing apoptosis. The isolation, chemical characterization and detection strategies by HPLC and HPLC-MS are reviewed, along with the reported bioactive effects of soyasaponin extracts and individual molecules in cultured cancer cell experiments.


Soyasaponin classification
Soyasaponins are oleanane triterpenoid glycosides possessing complex and diverse structures. They are found in soy (Glycine max) and other legumes, such as green peas (Pisum sativum L) and lentils (Lens culinaris ) [1][2][3]. Soyasaponins are amphiphilic molecules, with polar water soluble sugar moieties attached to a nonpolar, water insoluble pentacyclic ring structure. Soyasaponins are categorized according to the individual aglycones (soyasapogenols), and there are two main aglycones, referred to as group A and group B, respectively. Group A soyasaponins are bidesmosidic saponins with two glycosylation sites at carbons 3 and 22 on the oleanane ring structure. Group A soyasaponins can be further divided into two groups, known as acetylated and deacetylated forms [4,5]. Group B soyasaponins have one glycosylation site on their aglycones (carbon 3) and are also classified into two groups, based on the conjugation at carbon 22 with a 2,3-dihydro-2,5-dihydroxy-6-methyl-4-pyrone (DDMP) moiety or non-DDMP conjugated soyasaponins. DDMP conjugated soyasaponins are known as αg, βa, βg, γa and γg, and non-DDMP conjugated soyasaponins are known as soyasaponins I, II, III, IV and V [5][6][7]; the chemical compositions are shown in Figure 1. DDMP group B soyasaponins are thought to be the more abundant group of soyasaponins in soy [5]. There are two different naming conventions currently utilized in the literature, which adds to the complexity of summarizing and interpreting the reported literature [3][4][5]8] and both conventions are shown in Figure 1. There is a third soyasaponin aglycone known as group E, which has a ketone at position C22 and has been reported to be formed during soy extraction [8][9][10][11]. Group E soyasaponins Bd and Be have been reported to be transformed into group B aglycone during acid hydrolysis [12], thus suggesting that the group E aglycone might be an artifact formed during alcoholic soyasaponin extraction [13] and not be naturally occurring. Generally, plants store saponins as glycosides, typically in the bidesmosidic form, that can be hydrolyzed to the monodesmosidic forms and this change from bidesmosidic to monodesmosidic has been reported to enhance the bioactivity of saponins [14].
Soyasaponins are thought to be bioactive molecules, and there are many reports relating the bioactive response to the soyasaponin structure [8,9,15,16]. In this review, the isolation, chemical characterization and detection strategies, focusing on HPLC and LC-MS, to analyze soyasaponins will be discussed, along with the reported bioactive effects of soyasaponins extracts and individual molecules assessed in cultured cancer cell experiments.
The purification process is complex and laborious, with long sample preparation and extraction times and yielding relatively low amounts of soyasaponins from soy [3]. The amount of soyasaponins found in soy and related products are generally between 1.8% and 4.4%, depending on the variety and cultivation conditions of the soybeans [3]. The lack of a full complement of available commercial soyasaponin standards [20] has slowed the pace of research. Many of the current studies on soyasaponins have focused on extraction and analysis, usually producing relatively low yields and in insufficient quantities for biological activity testing of individual compounds.
Conventional extractions and preparation of soyasaponins from defatted soy flour or soy products typically involves the use of organic solvents, usually either aqueous ethanol or methanol with agitation at room temperature. Room temperature extraction is needed to ensure extraction and to prevent the breakdown of the DDMP conjugated compounds and most of the sugar glycoside molecules [21]. The extraction efficiency is dependent on three main factors: time, temperature (both ambient and solvent temperature) and choice of solvents [18][19][20]22]. The optimum extraction time has been reported to be between 4 to 6 h for the maximum yield of soyasaponins from soy flour using absolute methanol at 60 o C under reflux [23]. We have recently compared four different aqueous methanol extractions techniques: room temperature for 24 and 48 h with constant agitation, refluxing in methanol for 4 h at 60 °C, ultrasonic extraction, and Soxhlet extraction [19]. All four extraction techniques were suitable for extracting soyasaponins, but to different degrees. The Soxhlet extraction in methanol yielded significantly lower amounts of soyasaponins, compared to all other extraction techniques, while room temperature stirring in methanol yielded significantly higher amounts and an extraction time of 24 h was sufficient. The amounts of soyasaponins extracted from lowest to highest were found to be Soxhlet < reflux < sonication < 24 h stirring room temperature = 48 h [19]. A possible explanation for the observed differences in extraction may lie in the addition of thermal energy during the extractions, which can cause deglycosylation of saponin glycosides compared to room temperature extraction. Thermal energy likely transformed the DDMP conjugated group B soyasaponins such as βg into the corresponding non-DDMP soyasaponin I [7]. A similar funding was also reported for the extraction of dammarane triterpenoids saponins from ginseng (Panax quinquefolius) in boiling water [24].

Soyasaponin extract preparation
The basic extraction methodology used to extract soyasaponins typically involves preparing crude extracts, then employing a column chromatography concentration followed by purification and separation of individual soyasaponins [18,20,22]. However, soyasaponins are not the only groups of biologically active molecules found in soy and related products. Soy's biological activity can be attributed to three main classes of compounds: soy protein, isoflavones and the soyasaponins. Soy protein can be easily and selectively removed by ammonium acetate precipitation, however the isoflavones and the soyasaponins share overlapping polarities, making separation, quantification, and assessment of bioactivity difficult [19]. Employing column chromatography has been reported to be successful in separately the isoflavones from the soyasaponins. Decroos et al. [20] utilized an ethanol conditioned XAD-2 column to concentrate soyasaponins and isoflavones in defatted soy hypocotyls followed by fractioning and purification using preparative scale reverse phase chromatography [20]. The preparative chromatographic approach was shown to be effective in separating the isoflavones from the soyasaponins, based on different retention times, as the isoflavones were reported to elute earlier than the soyasaponins [20]. However, the purities of individual soyasaponins Ab, αg and βg were not optimal and required a second chromatographic step [20]. Preparative HPLC is an effective procedure to separate individual soyasaponins, but it is hampered by the large amounts of solvent needed and the low recovery yield. Semi-preparative purifications have also been successfully utilized for soyasaponin purification [22]. Alternatively, utilization of solid phase extraction has been reported to be a simple and economical alternative to achieve relatively pure soyasaponins between 85-90% [23]. We have recently reported a robust method based on solid phase extraction (SPE) utilizing various concentrations of methanol to separate the isoflavones from the soyasaponins, separate group B soyasaponins from group A and to produce relatively pure extracts [19,25]. A concentration of 50% methanol was found to be the optimum to separate the group B soyasaponins from the isoflavones, using SPE, virtually removing all but less than 1% of the isoflavones [19].

Analysis and determination of soyasaponins
There are various reported methods for the determination of soyasaponins from soy and soy products. Separation by thin layer chromatography and quantification using a densitometer has been reported as an economical and effective way to separate and quantify soyasaponins [26], but highperformance liquid chromatography (HPLC) utilizing a reversed-phase column is the most prevalent analytical technique for soyasaponins analysis. Various detectors have been used, such as ultraviolet or photo diode array (PDA) [6,18,27], and evaporative light scattering detection (ELSD) [12,28]. The maximum absorption wavelength of most of soyasaponins is a about 205 nm [22], while some of the DDMP soyasaponins can reach 295 nm. Due to the large number of soyasaponin glycosides found in soy, the development of an all encompassing UV detection method is difficult to achieve [27] and often two different gradient programs are utilized to achieve full soyasaponin detection. An analytical method using UV was reported for the detection of all the known group B soyasaponins at a wavelength of 205 nm, provided standards have been prepared in advance [18]. ELSD, which is based on mass detection by light scattering after evaporation of the mobile phase, has been successfully used for the detection of soyasapogenols, soyasaponins and ginseng saponins [29]. Rupasinghe et al. [12] reported ELSD could detect the authentic soyasapogenols. However, ELSD like UV, has some disadvantages such as an extensive sample preparation and potential interference when detecting low quantities in serum [30]. Often one solvent gradient program is optimized to separate the more abundant soyasaponin glycosides and one is optimized for the separation of the soyasapogenols leading to a time consuming analysis [16,19]. Tables 1 and 2 lists the various HPLC analysis strategies employed in the recent literature to detect the soyasaponin glycosides and the soyasapogenols respectively.  Identification of the soyasaponins of interest typically requires MS analysis, either on a HPLC-MS system or an individual MS to confirm the molecular weights. HPLC-MS detection seems to be most relevant and effective method for the identification of soyasaponins [32], other triterpenoids [36] and specifically group B soyasaponins [32,34]. Decroos et al. [29] developed an HPLC-ELSD-ESI-MS method for analysis all groups of soyasaponins, including acetyl soyasaponins group A and DDMP group B soyasaponins. MS detection of oleanane triterpenoids is complex, and requires experienced personnel and expensive equipment, which is usually not available for daily routine analysis in all laboratories [37]. Complicated fragmentation patterns are produced during ionization, resulting in molecular weight confirmation issues. Heftmann et al. [38] reported oleanane triterpenoid ring structures of the soyasapogenols are themselves prone to fragmentation due to a reverse Diels-Alder reaction [38]. For all these reasons MS analysis of the soyasapogenols can be challenging. For example, soyasapogenol A have been reported to produced a fragmentation pattern with the most abundant ion in positive mode corresponding to molecular weight of 250, while the corresponding soyasapogenol B ion was observed at 234 [38]. These two fragment are caused by the reverse Diels-Alder reaction and, correspond to molecular weights of 474 and 458 of the respective soyasapogenols [38]. In our laboratory we have optimized the HPLC-ESI procedure on a Thermo Finnigan LCQ-ESI quadrapole ion trap LC-MS (Thermo Fisher Scientific, USA) system and have been successful in determining the molecular ions for a number of group B soyasaponins such as I, III, βa, βg; the fragmentation patterns are shown in Table 4 [19]. Soyasapogenol fragmentation and breakdown during ESI analysis of group B soyasaponins was avoided by carefully adjusting the fragmentation temperature and generally keeping the internal temperature below 250 °C. Table 3 summarizes the recent literature on the detection and use of HPLC-MS technique for soyasaponin analysis.

Hydrolysis
Much of the early work on soyasaponins and other triterpenes such as the dammarane saponins from ginseng utilized acid or alkaline hydrolysis to deglycosolate the sugar moieties of the triterpene, leaving for the most part an intact aglycone [43]. The aglycone and related glycosides were then separated and visualized on thin-layer chromatography (TLC) before and after hydrolysis to confirm the presence of different aglycones and to assign new compounds to their respective groups based on TLC migratory patterns.
Acid hydrolysis is generally the preferred method for preparing soyasapogenols from crude soyasaponin extracts [13,16,26,37]. Acid hydrolysis in anhydrous methanol has been reported to enable the highest recovery of soyasapogenols A and B without producing artifacts [12,13]. Moreover, Ireland and Dziedzic [13] showed that hydrolysis for 3 h with 3% sulfuric acid in anhydrous methanol produced the greatest yield of soyasapogenols and anhydrous methanol has also been shown to increase the yield during acid hydrolysis [12].
Alkaline hydrolysis on the other hand, tends to produce partial hydrolysis of soyasaponins. Partial alkaline hydrolysis has been reported useful for preparing non-acetylated group A soyasaponins and non-DDMP group B soyasaponins [22,23]. Acetyl group A soyasaponins were reported to converted to non-acetyl soyasaponins A1 and A2 during saponification using alkaline treatment. DDMP conjugated group B soyasaponins are easily cleaved in alkaline conditions [6,19], [16] resulting in their corresponding non-DDMP molecule. Gurfinkel et al. [26] found the relative proportion of saponified soyasaponins significantly increased after alkaline treatment except for soyasaponin III. When compared to acid hydrolysis, alkaline treatment can assist with the cleavage of the DDMP conjugation without affecting the glycoside bond at position C-3 of the ring structure [19]. The DDMP moiety of group B soyasaponins has been found to be cleaved from soyasaponins easier than compared to the acetyl groups of group A soyasaponins [22]. In our laboratory we have utilized a characteristic partial alkaline hydrolysis (5% NaOH) in anhydrous methanol to produce an extract of approximately 65% soyasaponin I and 29% soyasaponin III [19] in sufficient amounts for cellular studies.

Measured bioactivities of soyasaponins in cell culture
Much of the reported bioactivity attributed to soyasaponins has utilized crude alcoholic extracts derived from defatted soy. Crude or total soyasaponin extracts are typically mixtures of group A and B soyasaponins with small amounts of soyasapogenols [16]. The biological activity of each individual saponin is currently unknown, although progress on the separation and isolation of sufficient quantities and bioactive testing is being made in our laboratory [19] and elsewhere [44,45]. Production of sufficient quantities of active extracts utilizing a bioassay guided fractionated approach extracts is a complex process [25] and the bioactivity of soyasaponins largely depends on their respective chemical structures. The presence of water soluble sugar moieties, DDMP and acetyl groups in the soyasaponin structures affects the polarity, which may mitigate changes in bioactivity [46]. Many studies have shown that mixtures of soyasaponins have measurable bioactivity in cell culture studies. Gurfinkel and Rao [9] reported that there was a relationship between structure and bioactivities with soyasapogenols A and B generally being more bioactive compared to the glycosides [9]. Research on the structureactivity relationships involving soyasaponins is ongoing. There is some evidence, as with many other saponins, that bioactivity of soyasaponins increases as sugars moieties are eliminated from the saponin structure, thereby reducing the polarity [14]. The position of the glycosides is also likely to determine certain bioactive functions such as membrane permeability. This has been demonstrated using dammarane saponins of ginseng [47]. Two molecules with similar chemical makeup, except for the location of a glucose moiety either at carbon 3 (Rh2) or carbon 6 (Rh1) had vastly different effects in cultured leukemia cells. Ginsenoside Rh2 was found induce apoptosis and concurrently increasing membrane permeation compared to Rh1 [47]. We are currently pursing bioactive classification of soyasaponins using cultured hepatocarcinoma cells and have begun with group B soyasaponins with the aim of providing evidence of structure function relationship for these molecules.
Recent studies have shown that a total soyasaponin extract can inhibit the growth of Hela (cervical tumor) cells [44], Hep-G2 (hepatocarcinoma) cells [16], and in colon adenocarcinoma cells (HCT-15) [45,48] by inducing programmed cell death, either apoptosis or microautophagy. Apoptotic processes remove damaged or mutated cells and recycle the cellular components [49]. Total soyasaponin extract prepared from defatted soy flour was reported to reduced the growth of cultured Hep-G2 cells after 72 hours of treatment measure by the MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide) assay [16]. The LC 50 value was determined to be 0.6 mg/mL. In Hela cells after four days of treatment with a total soyasaponin extract the LC 50 was estimated to be 0.4 mg/mL [44]. In both studies a total soyasaponin extract reduced cell growth through the induction of apoptosis. In Hela cells after four days of treatment, total soyasaponins (0.4 mg/mL) showed an increase in sub-G1 cells (apoptotic cellular fragments) of 10% during cell cycle analysis and 9% in Hep-G2 cells [16,44]. Xiao et al. [44] found that soyasaponin treatment reduced mitochondrial transmembrane potential and increased the intracellular Ca 2+ concentration leading to apoptosis [44]. Total soyasaponin extracts have also been reported to reduce colon carcinoma cell growth in a number of studies [50][51][52][53]. Soyasaponins were reported to modify the cell membrane of cultured cells [54], potentially increasing membrane permeability. Increased membrane permeability has also been reported for the dammarane triterpenes derived from ginseng in both cultured leukemia and intestinal cells [47,55]. Soyasapogenols prepared by acid hydrolysis were reported to inhibit the growth of Hep-G2 cells in a dose-dependent manner. The LC 50 concentrations were determined to be 0.05 ± 0.01 mg/mL for soyasapogenol A and 0.13 ± 0.01 mg/mL for soyasapogenol B (Figure 2) [16]. Soyasapogenol A inhibited the growth of estrogen-insensitive human breast cancer cells (MDA-MB-231) at a concentration of 10 μM but stimulated the proliferation of estrogen sensitive cells (MCF-7) 2.5 fold [56]. Additionally, the ER-ERE DNA complex, a marker of estrogen activation, was induced by soyasapogenol A. Soyasapogenol B reduced the growth of MDA cells without a significant effect on MCF-7 cells at all concentrations tested [56]. The composition of the extracts were as follows: SG A contained soyasapogenol A (69.3%), SG B contained soyasapogenol B (46.2%), I and III contained 65% soyasaponin I and 29% soyasaponin III, Gr B reflux refers to a group B reflux prepared extract that contained I (32.9%), III (14.2%), Be (4.2%), βg (28.3%), βa (9.7%), γg (1.7%), <1% γa and 3% soyasapogenol B, Gr B 25°C refers to a room temperature methanolic extract consisting of I (10.4%), III (4.6%), Be (6.2%), βg (46.6%), βa (16.9%), γg (4%), γa (1.7%) and < 1% soyasapogenol B. The TS extract refers to a total soyasaponin extract with an unknown saponin composition [16,19,25].

Soyasaponins and apoptosis
Both soyasapogenol A and B containing extracts have been reported to be able to induce apoptosis. Soyasapogenol A extract treated Hep-G2 cells induced 47 ± 3.5% of the cells to undergo apoptosis while soyasapogenol B extracted induced 15 ± 4.2% after 72 h treatment. Apoptotic fragments were confirmed by confocal laser scanning images showing evidence nuclear condensing (pyknosis) and fragmentation (karyorrhexis) consistent with the apoptotic program cell death [16] and representative sample images is shown in Figure 3. Yanamandra et al. [57] demonstrated that group B soyasaponins had pro-apoptotic and anti-invasive activities in human glioblastoma cells (SNB 19). A well characterized group B extract, containing mainly soyasaponins I, II, III, and IV reduced cell invasion 45% compared to untreated cells measured by an in vitro invasion assay. Furthermore, a loss of mitochondrial trans-membrane potential was measured along with increase release of cytochrome C and increased caspase activity [57]. Five different soyasaponin extracts were tested and the accumulation of sub-G1 apoptotic cells measured by flow cytometry [16,25]. Generally, soyasapogenol A containing extracted showed the greatest propensity to affect the cell cycle compared to soyasapogenol B containing extract tested at the LC 50 concentration compared to a fractionated extract or a total saponin mixture [16,25]. Extract preparation can influence the bioactive response of soyasaponins. Two different group B extracts were prepared, one prepared by ethanol reflux of soy flour, which is rich in non-DDMP group B soyasaponins, and one prepared by room temperature extraction containing an abundance of DDMP conjugated soyasaponins. The major different between the two extracts was the DDMP conjugated soyasaponin βg [15,25]. The non DDMP soyasaponins reflux extract inhibited the proliferation of Hep-G2 cells to a greater extent than the room temperature DDMP soyasaponin extraction. The LC 50 of the room temperature extract was found to be 0.9 ± 0.1 mg/mL and 0.5 ± 0.1 mg/mL for the reflux extract [15,25]. The reflux extract was found to induce apoptosis, as measured by the TUNEL assay, and affected the cell cycle progression whereas the room temperature extract induced differentiation of Hep-G2 treated measured by flow cytometry forward side scatter [15,25].
Soyasaponins have also been reported to induce macroautophagy, which is reported to be a type of programmed cell death [58]. Human colon cancer cells treated with soyasaponins suppressed proliferation, induced differentiation and inhibited protein kinase C activity [54]. Ellington et al. [45] reported that treatment of colon cancer cells (HCT-15) with an extract containing five different group B soyasaponins reduced cell growth after 24 and 48 h of treatment. Furthermore, treatment increased the percentage of cells in the S phase of the cell cycle while reducing cyclin-dependant kinase-2 (CDK-2) activity and a marker of macroautophagy (light chain 3) increased compared to non-treated cells. The induction of macroautophagy by group B soyasaponins was reported to be modulated by two important signaling pathway, group B treated cells were found to reduce Akt activity 50% affecting the phosphorylation of the ser 473 phosphorylation increasing activity of ERK1/2 (MAPK) by 60% [48].

Sialytransferase activity
Sialytransferase activity is associated with tumor metastasis and invasion [59,60]. Inhibition of sialytransferase activity is a useful target to delay the transformation of cells or slow the spread of metastasis. Wu et al. [61] showed that soyasaponin I was a highly specific inhibitor of in vitro sialytransferase activity [61]. Hsu et al. [62] confirmed that soyasaponin I was an in vitro sialytransferease inhibitor and was found to decrease α2,3-sialylations and ST3Gal IV expression which are important factors of the invasive behavior of tumor cells. Specifically, α2,3-linked sialic acids were suggested to play a role in the potential metastasis of murine melanoma cancer cell line B16F10 [63]. Soyasaponin I was found to specifically inhibit expression of α 2,3-linked sialic acids on the cell surface and to decrease the migration and cell adhesion to extracellular matrix proteins [63].

Conclusions
Soyasaponins are a group of structurally complex bioactive molecules. The extraction, isolation, and purification processes are challenging and many attempts have been made to characterize the chemical and biological activity of soyasaponins from soy. Generally, soyasaponins can be extracted in native form by room temperature methanolic extraction and can be adequate separated and identified by HPLC-MS provided sufficient chemical standards or molecular weight data are available. In terms of biological activity, classification of these molecules suggests that soyasapogenols have greater in vitro cellular anticancer activity such as inducing apoptosis compared to the corresponding glycosides. However, soyasaponin glycosides such as soyasaponin I and III do posses biological activity and may mitigate changes in cancer cell properties by inducing cellular differentiation or inhibiting enzymes involved in metastasis.