Comparative MS- and NMR-Based Metabolome Mapping of Egyptian Red and White Squill Bulbs F. Liliaceae and in Relation to Their Cytotoxic Effect

Urginea maritima L. (squill) species is widely spread at the Mediterranean region as two main varieties, i.e., white squill (WS) and red squill (RS), that are recognized for several health potentials. The major secondary metabolite classes of the squill are cardiac glycosides, mainly, bufadienolides, flavonoids, and anthocyanins. Herein, a multiplex MS and NMR metabolomics approach targeting secondary and aroma compounds in WS and RS was employed for varieties classification. Solid-phase micro extraction-gas chromatography/mass spectroscopy (SPME-GC/MS), ultra-high-performance liquid chromatography/mass spectrometry (UPLC/MS), as well as nuclear magnetic resonance (NMR) provided fingerprinting and structural confirmation of the major metabolites for both types of the squill. For comparison of the different platforms’ classification potential, multivariate data analysis was employed. While Bufadienolides, viz. “hydroxy-scilliglaucosidin-O-rhamnoside, desacetylscillirosidin-O-rhamnoside and bufotalidin-O-hexoside” as well as oxylipids, were enriched in WS, flavonoids, i.e., dihydro-kaempferol-O-hexoside and its aglycon, taxifolin derivative, were predominant in RS. A cytotoxicity screening against three cancer cell lines, including breast adenocarcinoma (MCF-7), lung (A-549), and ovarian (SKOV-3) cell lines was conducted. Results revealed that WS was more effective on A-549 and SKOV-3 cell lines (WS IC50 0.11 and 0.4 µg/mL, respectively) owing to its abundance of bufadienolides, while RS recorded IC50 (MCF7 cell line) 0.17 µg/mL since is is rich inflavonoids.


Introduction
Urginea, closely related to the "Drimia" genera belonging to family Asparagaceaeis, is widely spread at the Mediterranean region, Africa and India [1]. In Arabic, it is known as Basal farion, Onsul, and Samm el-Far [2]. Other common names are termed squill and sea onion [3]. U. martima species comprises two varieties to include white and red, with both recognized for their bulb's medicinal value and abundance of cardiac glycosides [4,5].

Identification of Volatile Organic Compounds (VOCs)
The volatiles profiling of RS and WS was determined using headspace SPME coupled with GC-MS, as reported in the current study, for the first time. A total of 26 volatiles were identified belonging to monoterpenes (10), aliphatic hydrocarbons (6), oxygenated compounds (4), in addition to a heterocyclic, an aromatic, and a sesquiterpene, as listed in Table 1. Due to the higher percentiles of monoterpenes and aliphatic hydrocarbons in RS compared to WS (47.82% and 37.26% vs. 25.69% and 19.76%), respectively, RS had a stronger aroma profile than WS. Other classes found exclusively in WS included heterocyclics, oxygenated hydrocarbons, aromatics, sesquiterpenes, oxygenated monoterpenes, ca. 4.57%, 12.15%, 1.75%, 8.22% and 2.29%, respectively, as shown in Figure 1.   Figure 1. SPME-GC/MS chromatogram of WS and RS headspace volatiles. The corresponding volatile names for each peak followed that listed in Table 1.

Secondary Metabolites Profiling via UHPLC/MS
UPLC-MS profiling was employed for secondary metabolites, and profiling in WS is denoted in pink color versus RS in red color. Figure 2 highlights the major differences between the two varieties. Diversity of secondary metabolite classes were identified in both squill, including mostly cardiac glycosides "bufadienolides" (50) followed by flavonoids (40) including"flavonols, flavanones and dihydroflavonols". Minor classes of coumarins (1), anthocyanins (2), and phenolic acids (5) were detected, with full spectral data Figure 1. SPME-GC/MS chromatogram of WS and RS headspace volatiles. The corresponding volatile names for each peak followed that listed in Table 1.

Secondary Metabolites Profiling via UHPLC/MS
UPLC-MS profiling was employed for secondary metabolites, and profiling in WS is denoted in pink color versus RS in red color. Figure 2 highlights the major differences between the two varieties. Diversity of secondary metabolite classes were identified in both squill, including mostly cardiac glycosides "bufadienolides" (50) followed by flavonoids (40) including"flavonols, flavanones and dihydroflavonols". Minor classes of coumarins (1), anthocyanins (2), and phenolic acids (5) were detected, with full spectral data of identified peaks presented in Table 2.
The UV spectra of the major classes of secondary metabolites were observed at λ 280-299 nm. For bufadienolides glycosides, they were observed at 292-299 nm [17], while phenolics showed UV max at 280-283 and 296-340 nm [18]. Different classes have been annotated in both white and red squill, viz. bufadienolides, flavonoids, phenolic, amino, and fatty acids, as detailed in the next subsections. All identified secondary metabolites were listed in Table 2          Bufadienolides are C-24 steroids with a pyranone ring at C-17β that are naturally present in plants and mammalian animals, specifically toads and snakes [19]. Bufadienolides were found more abundant in WS than RS as revealed from the inspection of the chromatograms of both WS and RS at retention time "R t ." from 5 to 13 min. A series of different classes of bufadienolides were identified in WS and RS and included hydroxyoxobufadienolide with a C-19 aldehyde group, i.e., bufotalidin, and hydroxybufaenolides with a C-19 methyl group, including scillarenin and scilliphosidin. Notably, identified bufadienolides structures were illustrated in Figure 3. present in plants and mammalian animals, specifically toads and snakes [19]. Bufadienolides were found more abundant in WS than RS as revealed from the inspection of the chromatograms of both WS and RS at retention time "Rt." from 5 to 13 min. A series of different classes of bufadienolides were identified in WS and RS and included hydroxyoxobufadienolide with a C-19 aldehyde group, i.e., bufotalidin, and hydroxybufaenolides with a C-19 methyl group, including scillarenin and scilliphosidin. Notably, identified bufadienolides structures were illustrated in Figure 3.  Figure S1C).
Identified hydroxybufaenolides are subdivided by the number of hydroxyl groups to mono, di, tri and tetra hydroxyl derivatives and were identified in both WS and RS as listed in Table 2.

Flavonoids
Compared to WS's abundance of bufadienolides, flavonoids were more abundant in RS, with several being reported for the first time in this current study. Beside a positive mode, improved detection of flavonoids was observed in a negative mode, with MS2 fragments found to be characteristic of flavonoids, i.e., amu 151, 125 and 153 as well as neutral loss of amu −28 "CO", −18 "H 2 O" and −44 "CO 2 " [38,39]. The next subsection shall summarize identification of the different flavonoid subclasses and distribution in Urginea species.

Identification of Coumarins
The hydroxycoumarin class has been annotated in both squill types, represented as L29 and N15 m/z 163 (C 9 H 7 O 3 ) + and 161 (C 9 H 5 O 3 ) − , respectively, and is in agreement with previous reports for coumarins in U. indica species [48,49].

Phenolic Acids
Phenolic acids are aromatic secondary plant metabolites found ubiquitously in plants and play a role in food quality and their organoleptic properties [50]. The primary detected phenolic acid was vanillic acid in L26 at m/z 169 (C 8

Amino Acids and Fatty Acids
Few amino acids were identified in both squills, including inL12 at m/z 205 (C 11 H 13 N 2 O 2 ) + , L8 at m/z 182 (C 9 H 12 NO 3 ) + , and L9 m/z 166 (C 9 H 12 NO 2 ) + for tryptophan, tyrosine, and phenyl alanine, respectively. Compared to amino acids showing earlier elution, fatty acids were observed at the end of the chromatogram, considering their lypophilic nature in both varieties. An example of major fatty acids includes L114 at m/z 279 (C 18 H 31 O 2 ) + [M+H] + octadecatrienoic acid and L120 at m/z 281 (C 18 H 33 O 2 ) + linoleic acid. No difference in amino acids was observed among squill varieties compared to flavonoids; however, they later appear as stronger marker for variety type.

Multivariate Data Analysis of UPLC-MS Dataset
To aid in identifying further markers for each squill variety in an untargeted manner, unsupervised principle component analysis (PCA) and orthogonal projection to latent structures analysis (OPLS-DA) were attempted in both negative ( Figure 4) and positive ionization modes ( Figure 5). Complete segregation between the two squill varieties was observed, highlighting that RS was more rich in phenolics, whereas WS was abundant in bufadienolides. A score plot model derived from the negative ionization mode of UHPLC/MS prescribed by PC1 and PC2 accounted for 72 and 17%, respectively, of the total variance ( Figure 4A), with clear segregation of RS from WS. As for revealing the metabolites mediating RS and WS segregation, a PCA-loading plot ( Figure 4B) revealed an abundance of flavanols, identified as dihydrokaempferol-O-hexoside and taxifolin-O-hexoside derivatives, in RS, which are likely to serve as precursors for anthocyanins solely found in RS variety. The OPLS-DA model ( Figure 4C) further confirmed the PCA results from the S-plot ( Figure 4D) revealing that dihydrokaempferol-O-hexoside, its aglycon, and vanillin rhamnoglucoside, were abundant in RS compared to WS.
The results from the UHPLC/MS-derived model in the positive ionisation mode were comparable to those in the negative mode ( Figure 5A). The two squill varieties, RS and WS, were clearly separated in the PCA model, which was prescribed by PC1 66% and PC2 23%. Furthermore, according to Figure 5B, in accordance with the negative ionisation mode, RS was more abundant in dihydrokaempferol, which is a biosynthetic precursor for anthocyanins in RS [53]. On the other hand, WS had higher concentrations of bufadienolides, such as hydroxy-scilliglaucosidin-O-rhamnoside, desacetylscilliglaucosidin-O-rhamnoside, and bufotalidin-O-hexoside as well as oxylipids, such as linoleic acid and linoleyl alcohol. Supervised OPLS-DA model. Figure 5C,D confirmed linoleic acid enrichment in WS and identifying dihydro-kaempferol "aromandrin" and the bufadienolide "scillipheoside-Oglucoside" as markers for RS. mode, RS was more abundant in dihydrokaempferol, which is a biosynthetic precursor for anthocyanins in RS [53]. On the other hand, WS had higher concentrations of bufadienolides, such as hydroxy-scilliglaucosidin-O-rhamnoside, desacetylscilliglaucosidin-Orhamnoside, and bufotalidin-O-hexoside as well as oxylipids, such as linoleic acid and linoleyl alcohol. Supervised OPLS-DA model. Figure 5C,D confirmed linoleic acid enrichment in WS and identifying dihydro-kaempferol "aromandrin" and the bufadienolide "scillipheoside-O-glucoside" as markers for RS.  mode, RS was more abundant in dihydrokaempferol, which is a biosynthetic precursor for anthocyanins in RS [53]. On the other hand, WS had higher concentrations of bufadienolides, such as hydroxy-scilliglaucosidin-O-rhamnoside, desacetylscilliglaucosidin-Orhamnoside, and bufotalidin-O-hexoside as well as oxylipids, such as linoleic acid and linoleyl alcohol. Supervised OPLS-DA model. Figure 5C,D confirmed linoleic acid enrichment in WS and identifying dihydro-kaempferol "aromandrin" and the bufadienolide "scillipheoside-O-glucoside" as markers for RS.

NMR Metabolites Fingerprinting
To provide a broader coverage of squill metabolome, NMR was employed to provide insight on both secondary and primary metabolites, especially with the later class not detected using LCMS. NMR offers also improved structural elucidation tool aided by its extensive 2D NMR experiments and quantitative determination of the major metabolites [54] for quality control purposes. Major classes detected in squill using NMR included sugars, flavonoids, bufadienolides, phenolics, and amino and fatty acids ( Figure 6 and Table 3). Nevertheless, compared to MS, NMR suffers from low sensitivity and from signal overlap, especially in aliphatic regions. To overcome the problem of signal overlap, 2D NMR experiments were employed to allow for the resolution of overlapped signal along the second dimension, i.e., carbon in the case of HMBC [55,56]. 1 H NMR spectra from WS showed the signals relative richness in both varieties (Figure 7 and Table S1).
positive ionization mode of RS and WS squill varieties.

NMR Metabolites Fingerprinting
To provide a broader coverage of squill metabolome, NMR was employed to provide insight on both secondary and primary metabolites, especially with the later class not detected using LCMS. NMR offers also improved structural elucidation tool aided by its extensive 2D NMR experiments and quantitative determination of the major metabolites [54] for quality control purposes. Major classes detected in squill using NMR included sugars, flavonoids, bufadienolides, phenolics, and amino and fatty acids ( Figure 6 and Table 3). Nevertheless, compared to MS, NMR suffers from low sensitivity and from signal overlap, especially in aliphatic regions. To overcome the problem of signal overlap, 2D NMR experiments were employed to allow for the resolution of overlapped signal along the second dimension, i.e., carbon in the case of HMBC [55,56]. 1 H NMR spectra from WS showed the signals relative richness in both varieties (Figure 7 and Table S1).  Table 3 for metabolite identification using 1D and 2DNMR.   Table 3 for metabolite identification using 1D and 2DNMR.      Table 3.

Quantification of Major Metabolites via 1 H-NMR
To aid in standardization of squill extract, 1 H-NMR was further used to determine absolute levels of major metabolites in squill varieties via integration of their well-resolved signals in NMR spectra [57]. The concentration of metabolites was calculated as µg/mg dry powder, as shown in Supplementary Material Table S1.
With regard to primary metabolites, unsaturated fatty acids were detected at much higher levels in WS at 89.6 ± 25.3 versus 2.84 ± 0.5 µg/mL in RS. In contrast, a comparable amino acid level was detected in both varieties exemplified by aspartic acid as major form at 43.1 ± 7.2 and 38.3 ± 4.9 µg/mL in WS and RS, respectively. Other less abundant amino acids, including glycine, alanine, and tryptophan, were detected at 19.8 ± 3.5; 8.5 ± 1.6; and 2.2 ± 0.18 µg/mL in WS versus lower levels in RS at 7.2 ± 1.0; 0.3 ± 0.1; and 1.0 ± 0.1 µg/mL in RS, respectively. Sugars were found at comparable levels in both WS and RS at 42.5 ± 1.2 and 36.6 ± 0.5 µg/mL, respectively.
With regard to secondary metabolites to influence squill health effects, higher levels of total bufadeinolides distinguished by α and β-unsaturated ketone of pyranone ring (H-22) were measured in WS at 17.5 ± 7.5 µg/mL while in RS, dihydrokampferol was predominated at 43.6 ± 2.3 µg/mL. Finally, coumarins detected at almost equal levels in both WS and RS at ca. 5-6 µg/mg.

Cytotoxic Screening Activity
Squill is recognized for its anticancer effect against various cancer cells and for its antioxidant and cytotoxic properties [58]. Consequently, a comparative cytotoxic assay of both WS and RS was evaluated on the different cell lines to include breast adenocarcinoma (MCF-7), lung (A-549) and ovarian cancer (SKOV-3) cell lines using Sulforhodamine B assay (SRB). The results revealed the potential cytotoxic activity of both varieties, with WS found to be more active against both cell lines A-549 and SKOV-3 than RS, as evidenced by its lower IC 50 values. The recorded IC 50 values of WS against A-549 and SKOV-3 cell lines were at 0.108 ± 0.003 and 0.690 ± 0.018 µg/mL versus RS's IC 50 values at 0.271 ± 0.005 and 0.912 ± 0.021 µg/mL, respectively. In contrast, the RS extract showed a more potent effect than WS on the MCF-7 cell line, with IC 50 values of 0.165 ± 0.007 and 0.326 ± 0.005 µg/mL, respectively. The doxorubicin "positive control" IC 50 values recorded on MCF7, A-549, and SKOV-3 cancer cell lines were 0.2 ± 0.004, 0.56 ± 0.003 and 0.2 ± 0.01 µg/mL, respectively (Table 4 and Figure 8).   The potential cytotoxicity of WS compared to RS could be attributed to its abundance of bufadienolides, as revealed from both LC/MS and NMR reported for its cytotoxic action, and suggests that bufadieonolides are more determinant than flavonoids regarding cytotoxic action in squill, at least in case of A-549 and SKOV-3 cell lines. Previous reports of hellebrigenin and bufatalin isolated from squill showed potential cytotoxic activity against leukemia, human colon carcinoma, human glioblastoma melanoma, and human liver carcinoma cells with IC 50 values ranging from 0.0007 to 0.16 µM [59]. Moreover, bufatalin induced apoptosis in human leukemia cells [60]. Scillarenin exhibited stronger cytotoxic action in the nanomolar range in comparison with bufatalin [61].
lines were at 0.108 ± 0.003 and 0.690 ± 0.018 µg/mL versus RS's IC50 values at 0.271 ± 0.005 and 0.912 ± 0.021 µg/mL, respectively. In contrast, the RS extract showed a more potent effect than WS on the MCF-7 cell line, with IC50 values of 0.165 ± 0.007 and 0.326 ± 0.005 µg/mL, respectively. The doxorubicin "positive control" IC50 values recorded on MCF7, A-549, and SKOV-3 cancer cell lines were 0.2 ± 0.004, 0.56 ± 0.003 and 0.2 ± 0.01 µg/mL, respectively (Table 4 and Figure 8).    Proscillaridin A, a cardiac glycoside isolated from U. maritima, was reported to exert an cytotoxic and/or antiproliferative effect against human breast cancer. proscillaridin A anticancer properties are mediated via its ability to block Na+/K+ ATPase, leading to increase in Ca 2+ levels, activating the AMPK pathway. Interestingly, on the opposite, the Ca 2+ level was reduced by ca. 30% after an 18 h administration of proscillaridin A to the normal lung fibroblast cell line CCD19-LU, which suggests differential action mechanisms against normal and cancer cells [62].
In a previous report, bufadienolides recorded effective cytotoxic action on human cancer cells [63]. Although in WS, the most abundant bufadienolide glycoside "scilliroside" is suggested to mediate the observed toxic action of Urginea sp. The lethal dose (LD 50 ) of scilliroside was 0.7 and 0.43 mg/kg for male and female rats in vivo, respectively. Scilliroside and its aglycon, scillirosidin, exert more toxic effect than other bufadienolides, such as proscillaridin and desacetylscillirosidin. This would be attributed for the presence of an acetoxy group at the C-6 position of scilliroside [63].

Plant Material
Samples of U. maritima (Linn) Baker "Sea Squill" (RS and WS) were collected in summer 2018-2019 from the El-Arish desert, Sinai, Egypt. RS and WS were authenticated by Prof. Zaki Turki (Department of Botany, Faculty of Science, El-Menoufia University, Shebin El-Kom, Egypt) as U. maritima (L.) Baker, and voucher specimens Sp. No HRE139 have been deposited at the Herbarium of the Department of Botany, Faculty of Science, Menoufia University, Egypt.

Secondary Metabolites Extraction and Preparation of NMR and MS Analysis Sample
The extraction protocol for NMR and MS analysis followed that by Farag et al. [14]. Briefly, a freeze-dried squill sample was mixed with 5 mL methanol with umbelliferone "internal standard (10 µg/mL) for the quantification of metabolites using UPLC/MS". The squill extract was vortexed and then centrifuged (3000× g) for half an hour to remove any plant wastes. NMR analysis was performed by 3 mL aliquot, then concentrated under "N 2 " stream. The dried squill extract was resuspended with CD 3 OD (700 µL) containing 0.94 mM HMDS, then centrifugation at 13,000× g for 1 min.

SPME/GC-MS
Solid phase micro extraction (SPME) technique was adopted for volatiles extraction as in Farag et al. [64]. Squill (5 g) was incubated at 50 • C for half an hour in a screw cap glass vial, through which the SPME fibers "stableflex fibers covered with divinylbenzene/carboxen/polydimethylsiloxane (DVB/CAR/PDMS, 50/30 µm), Supelco (Oakville, ON, Canada)" was placed for 15 min with the squill sample, then injected into the GC-MS injection port. The GC-MS specifications and the analysis method were previously explained in detail in a previous study by Farag et al. [64].

UHPLC/MS
The identification of secondary metabolites in the squill sample was following the specifications of the UPLC/MS as well as the procedure that was previously mentioned in Farag et al. [65]. Characterization of the secondary metabolites was carried by their UV-VIS spectra from 200-600 nm, the retention time (Rt.) relative to authentic, exact mass and upon comparing the mass spectra of those authentic, the natural products database dictionary (CRC), and the published literature [66].

Identification of Major Metabolites via NMR Analysis
All spectra were analyzed using VNMRS 600 NMR spectrometer. All the specifications were described in detail as mentioned in Farag et al. [69]. The 2D-NMR spectra were reported at 599.83 MHz frequency using CHEMPACK 6.2 pulse sequences as COSY, HSQC, and HMBC. The optimization of HSQC and HMBC experiments were documented previously in [68]. Similar to the previous work [15], WS was used as a reference to demonstrate the identification of U. maritima metabolites. Interpretation was achieved by chemical shifts of standards using 2D-NMR and 1 H-1 H-correlation spectroscopy COSY and TOCSY, 1 H-13 C-HSQC, and HMBC.
Quantification of Major Metabolites via 1 H-NMR 16 metabolites were quantified by NMR spectroscopy Figure S1. The peak area of both target compounds and the internal standard (HMDS) specific protons were interpreted manually for both squill samples as described in [69].

Cytotoxic Screening Assay
The cell viability was conducted by SRB assay. A 100 µL of cell suspension aliquots (5 × 10 3 cells) was placed in 96-well plates, then incubated for 1 day. 100 µL media with squill extracts (0.01, 0.1, 1, 10, 100 µg/mL) were added to the cell suspension. After 3 days of treatment, the cells were fixed by changing the media with 150 µL (10% Trichloroacetic acid (TCA)) and incubated at 4 • C for 1 h. The cells were then washed with distilled water. SRB (70 µL) aliquots (0.4% w/v) were added and incubated in a dark place for 10 min. The plates were washed with acetic acid (1%) and allowed to dry overnight. Then, 150 µL of tris (hydroxymethyl) aminomethane (TRIS) (10 mM) was added, and absorbance was measured at 540 nm using a BMG LABTECH ® -FLUOstar Omega microplate reader (Biotechnology company in Ortenberg, Germany).

Statistical Analysis
The cytotoxic screening results were represented as averages of 3 independent experiments with their standard deviation (mean ± SD). For statistical significance determination, results were analyzed using one-way analysis of variance (ANOVA) followed by Dunnett's post hoc test to compare doxorubicin positive control with treatment groups of WS and RS extracts on different cancer cell lines tested in vitro using SRB assay. Graph pad prism version 5 was used where p was ≤ 0.05.

Conclusions
In this current study, the two squill varieties (red and white) were investigated using a metabolomic approach (PCA and OPLS-DA) coupled with different chromatographic (SPME-GC/MS and UPLC/MS) and spectroscopic techniques (1D and 2D-NMR) for the first time, where metabolites diversity was identified in the two varieties.
Volatiles assessment (SPME-GC/MS) resulted in identifying 27 volatiles in both red and white squills. Red squill was enriched with monoterpenes hydrocarbons (47.82%) than white one (25.90%), loading to more aroma profiling. The 1D and 2D-NMR spectroscopic technique was utilized to identify 16 major metabolites. The phenolic compounds were abundant in the red squill, whereas bufadienolides and fatty acids showed more intense peaks in the white one.
Secondary metabolites identification (UHPLC/MS) revealed 130 metabolites representing a myriad of classes. Bufadienolides class was the major one in white squill, whereas flavonoids were the major one in the red variety. The 1D and 2D-NMR spectroscopic technique was utilized to further identify 16 major metabolites. The phenolic compounds were abundant in the red squill, whereas bufadienolides and fatty acids showed more intense peaks in the white one.
Multivariate data analysis differentiated between both varieties and confirmed the abundance of flavonoids in red squill exemplified in dihydrokaempferol-O-hexoside, its aglycon, and taxifolin derivative, whereas fatty acids (oleic and linoleic acids) in addition to bufadienolides, viz. hydroxyscilliglaucosidin-O-rhamnoside, desacetylscillirosidin-Orhamnoside, and bufotalidin-O-hexoside are more abundant in the white one. A cytotoxicity screening was implemented on both squills against different cell lines revealed the effectiveness of white squill over red one due to its enrichment with bufadienolides class, which has yet to be confirmed using isolated bufadienolides to be conclusive.  Figure S8: Signal assignment of the proton markers for dihydro kaempferol (M15) observed in the 1 H-NMR spectrum of RS methanol extract; Table S1: 1 H-NMR quantification of most common primary and secondary metabolites detected in Urginea species, white squill (WS) and red squill (RS). Values are expressed as µg/mg dry powder ± S.D (n = 3). Chemical shifts used for metabolite quantification were determined in methanol-d 6