Salvadora persica leaves: phytochemical profile and in vitro-inhibitory activity on inflammatory mediators implicated in periodontal disease

Abstract

Context

Virtually all parts of Salvadora persica L. (Salvadoraceae) are used in traditional medicine. The twigs and leaves are used for oral health, but leaves are far less investigated.

Objective

This study assesses the oral health-promoting potential of S. persica leaves with emphasis on anti-inflammatory and antiproliferative effects and provides an in depth-characterization of their metabolite profile.

Materials and methods

Hot-water and methanolic S. persica leaf extracts (1, 10, and 100 µg/mL) and their major constituents (5, 10, and 50 µM), were subjected to cellular assays on IL-8 and TNFα release in LPS-stimulated human neutrophils, NO-release in LPS/IFNγ stimulated mouse macrophages, and proliferation of HNO97 human tongue carcinoma cells. Metabolite profiling was performed by UHPLC-HRMS analysis. Major constituents were isolated and structurally elucidated.

Results and discussion

Both extracts showed pronounced anti-inflammatory activity in LPS-stimulated neutrophils. Major identified compound classes were flavonoid glycosides, the glucosinolate glucotropaeolin, phenyl- and benzylglycoside sulfates, and megastigmane glycosylsulfates, the latter ones identified for the first time in S. persica. Glucotropaeolin strongly inhibited the release of IL-8 and TNF-α (13.3 ± 2.0 and 22.7 ± 2.6% of the release of stimulated control cells at 50 µM), while some flavonoids and 3-(3′-O-sulfo-β-d-glucopyranosyloxy)-7,8-dihydro-β-ionone, a newly isolated megastigmane glycosylsulfate, were moderately active. Benzylisothiocyanate, which is likely formed from glucotropaeolin during traditional application of S. persica, showed considerable antiproliferative activity (IC₅₀ in HNO97 cells: 10.19 ± 0.72 µM) besides strongly inhibiting IL-8 and TNFα release.

Conclusions

Glucotropaeolin and benzylisothiocyanate are likely implicated in the oral health-promoting effects of S. persica leaves. The chemistry and pharmacology of the newly identified megastigmane glycosylsulfates should be further evaluated.

Keywords:

• Toothbrush tree
• anti-inflammatory
• antiproliferative
• UHPLC-HRMS metabolite profiling
• megastigmane

Introduction

Oral diseases constitute a major global public health issue that affects almost 3.5 billion people worldwide. Three out of four affected persons live in lower- and upper-middle income countries. Next to dental caries, severe periodontal disease as well as lip and oral cavity cancer are among the five main globally occurring oral diseases (World Health Organization 2022). Inflammatory mediators, such as the cytokines TNF-α and IL-8 (Nussbaum and Shapira 2011; Hoare et al. 2019; Saliem et al. 2022), and nitric oxide (NO) (Uğar-Cankal and Ozmeric 2006; Hussain et al. 2016) are known to be implicated in the pathogenesis of periodontal disease.

Salvadora persica L. (Salvadoraceae) is an evergreen shrub or tree occurring in Northern, Southwestern, and Central Africa as well as in the dry regions of India and Western Asia. It frequently grows on sandy, dry or saline soils (Khatak et al. 2010; Halawany 2012). The plant is known under the name ‘Arak’ in Arabic and ‘Peelu’ in Urdu. Its twigs and roots are the most common traditional sources of the first historically reported oral hygiene tool, a chewing stick called miswak, meswak or siwak (Sofrata et al. 2008; Ahmad and Rajagopal 2014). Until today, miswak use remains widespread in many parts of the world, and the World Health Organization recommends its application as an effective oral hygiene measure in regions where its use is customary (WHO, 1987; Sofrata et al. 2008).

S. persica twigs, stem bark and roots have been previously subjected to thorough phytochemical analysis. This revealed the presence of flavonoids, of phenolic acids, of fatty acids, of sulfur-containing compounds such as salvadoside and the sulfur-containing alkaloid persicaline, and of nitrogen-containing compounds such as benzylurea derivatives, 3-formylindole, butanediamides and the proline-betaines stachydrine and 4-hydroxystachydrine, as major secondary compound classes (Farag et al. 2018; Ayoub et al. 2021; Farag et al. 2021; Binsuwaidan et al. 2023).

The effectiveness of miswak application for dental plaque and gingivitis reduction has been shown in clinical studies (Ramli et al. 2022; Scott and Marshman 2022; Azizan et al. 2023). Miswak extracts and miswak-based preparations such as toothpastes or mouthwashes have been found to exert positive effects on several aspects of oral health in vitro, in vivo, or in clinical studies. They include antimicrobial and anticariogenic effects, reduction of dental plaque due to inhibition of oral microbes, as well as elevation of plaque pH caused by the high silica content. Furthermore, oral tissue healing effects as well as positive effects on gingival health and periodontal status have been reported (Ahmad and Rajagopal 2014; Kumar and Sharma 2021; Adam et al. 2023; Adilah Harun et al. 2023; Shaalan and El-Rashidy 2023).

Also the leaves of S. persica are traditionally used for oral health purposes, but not as commonly as the twigs and roots: S. persica leaves are considered to strengthen the teeth (Khatak et al. 2010), leaf decoctions are used as a mouthwash, and leaves are masticated for treating tooth and gum problems (Sher et al. 2010). The phytochemical profile of the leaves is less thoroughly investigated than that of the twigs. The main compound classes known so far are flavonoids (Ali et al. 1997; Owis et al. 2021), phenylpropanoids, gallic acid (Noumi et al. 2017; Ibrahim et al. 2022), the glucosinolate glucotropaeolin (Abd El-Wahab et al. 1990) the indole alkaloid salvadoricine (Malik et al. 1987), and volatile oil (Alali and Al-Lafi 2003). S. persica leaf extracts reportedly have shown in vitro antimicrobial activity against different bacterial strains (Aumeeruddy et al. 2018) as well as anti- quorum sensing and antibiofilm activity against oral Staphylococcus strains (Noumi et al. 2017) and antioxidant effects (Ibrahim et al. 2022).

During this study, the oral health- promoting effects of S. persica leaves were assessed. We emphasized on their potential anti-inflammatory effects in stimulated neutrophils and macrophages, and on their potential antiproliferative effects on a human tongue carcinoma cell line. Furthermore, the metabolite profile of S. persica leaf extracts was studied in depth. Identified major constituents were isolated or purchased, and their oral health-promoting potential was evaluated in addition.

Materials and methods

Chemicals and reagents

Water, acetonitrile, and formic acid for UHPLC-HRMS analyses were LCMS grade, the other solvents were of analytical grade. DMSO for pharmacological assays was from Carl Roth (Karlsruhe, Germany; 4720.2). TLC plates for monitoring fractionation and purification of compounds were coated with silica gel (Kieselgel 60 F254 aluminium plates, 20 × 20 cm, coat thickness 25 mm, Merck, Darmstadt, Germany; 1.05554.0001). Silica gel for VLC and silica gel CC (particle size 0.04-0.063 mm) were from VWR International (27,623.323). Cartridges for solid phase extraction (Isolute^® C18 (EC), 10 g) were from Biotage, Uppsala, Sweden (221-1000-F). Glucotropaeolin potassium salt was purchased from Phytolab, Vestenbergsgreuth, Germany (89216). Benzylisothiocyanate was from Alfa Aesar, Kandel, Germany (10752084).

Reagents for isolation and culturing of human neutrophils: Human Quantikine^® Pancoll human, density 1.077 g/mL was purchased from PAN-Biotech (Aidenbach, Germany; P04-601000). Fetal bovine serum (FBS) was obtained from Gibco (Grand Island NY, USA; A5256701) and penicillin–streptomycin from PAA Laboratories (Pasching, Austria; 15140122).

Reagents for assay on inflammatory functions in LPS-stimulated human neutrophils: Phosphate buffered saline (PBS) was purchased from Biomed (Lublin, Poland; 14190-250), RPMI 1640 Medium (R7509), f-MLP (formyl-Met-Leu-Phenylalanine) (F3506), LPS (from Escherichia coli 0111:B4) (2552117), cytochalasin A (C6637), HEPES solution (H0887), luminol (A8511), L-glutamine (G7513), propidium iodide (PI) (559763) and dexamethasone (>98%; D1756) were bought from Sigma–Aldrich (Steinheim, Germany). Human TNF-α (555212) and IL-8 (555244) Quantikine ELISA Kits were purchased from BD Biosciences (Franklin Lakes, NJ, USA).

Reagents for assay on NO release in LPS/IFNγ-stimulated mouse macrophages: LPS (E. coli 055:B5, L5418) and Griess Reagent (G4410) were purchased from Sigma-Aldrich, interferon-γ (mouse recombinant E. coli) was from Roche Diagnostics (Rotkreuz, Switzerland; 11276905001), and L-NMMA was purchased from Enzo Life Sciences (Farmingdale, NY, USA; ALX-106-001-M025). Medium and supplements for cell culture were purchased from Gibco and Thermo Fisher Scientific (Waltham, MA, USA). Sodium nitrite (purity > 98.7%) was from Carl Roth (1L4P.1).

Reagents for the HNO97 cell proliferation assay: FBS, antibiotics and DMEM high glucose medium were from Gibco/Thermo Fisher Scientific, Cell proliferation kit II was from Roche Diagnostics (11465015001, and vinblastine was from Sigma-Aldrich (V1377).

General experimental procedures

Semipreparative HPLC separations were performed on a Shimadzu semipreparative HPLC system (Shimadzu Corporation, Kyoto, Japan), using a Luna^® C18(2) (10 µm, 100 Å) column (250 × 10 mm, Phenomenex, Torrance, CA, USA, 00 G-4253-N0) as stationary phase.

NMR spectra for flavonoids were recorded in pyridine-d₅ (Eurisotop, Saclay France; D013FE) on a UnityInova 400 MHz instrument (Varian, Palo Alto, CA, USA), and NMR spectra for compound 41 were recorded in CH₃OH-d₄ (Eurisotop) on an Avance III 700 MHz instrument (Bruker, Bremen, Germany). Spectra were recorded at 25 °C and (CH₃)₄Si was used as internal standard. J is in Hz.

The absolute configuration of the glucose moiety present in compound 41 was determined as described previously: 0.3 mg of compound 41 were hydrolyzed and the released glucose converted to the respective butylglucoside with (2 R)-2-butanol (Sigma Aldrich, 236691) as described by Stambolov et al. (2023). The reaction mixture was evaporated to dryness and then dissolved in Sigma-Sil A (Sigma Aldrich, 1391) to prepare the respective trimethylsilyl ether. Reference sugars (d- and l- glucose, Sigma Aldrich, 158968 and G5500) were treated in the same manner, and the samples were analyzed by GC-MS as described in Alza et al. (2014).

Plant material and extract preparation

The plant name is in accordance with http://www.worldfloraonline.org/ (accessed 08.02.2023).

Leaves were collected on October 15, 2016 from shrub trees of S. persica cultivated in the Botanical Garden of the Botany Department of Sohag University, Sohag, Egypt. The material was authenticated by Ahmed Mazen. The leaves were dried for two days at 50 °C. A reference sample of the studied dried leaves is kept at the Institute of Pharmaceutical Sciences, Pharmacognosy, University of Graz, Austria (Reference code SPL2016-1).

Small- and large- scale extracts were prepared in February 2017. For the preparation of the hot water extract (SPI), 2.02 g of dried, cut S. persica leaves were mixed with 20 mL boiling water and kept at room temperature for 10 min. After that, the extract was filtered through a folded filter, frozen and lyophilized (yield: 195 mg dried hot water extract). For the preparation of the methanolic extract (SPM), 1.01 g of powdered S. persica leaves were mixed with 10 mL pure methanol (CH₃OH) and extracted for 5 min at room temperature in an ultrasonic bath. The mixture was centrifuged for 5 min at 4000 rpm, the supernatant was removed, and the extraction process was repeated twice with 10 mL of CH₃OH. The combined supernatants were concentrated by rotary evaporation and dried under a stream of nitrogen (yield: 42 mg dried CH₃OH extract). For compound isolation, 1.9 kg of powdered leaves were extracted by percolation with CH₃OH at room temperature (5 subsequent 24 h-cycles with 3 L CH₃OH each). The extract was concentrated on a rotary evaporator at 40 °C and dried under a stream of nitrogen (yield 300 g dried CH₃OH extract, 15.8%).

UHPLC-HRMS metabolite profiling of SPI and SPM

UHPLC-HRMS analyses were performed on a Dionex Ultimate 3000 HPLC system hyphenated with a QExactive Hybrid Quadrupole Orbitrap MS (Thermo Scientific), using an Acquity^™ Premier BEH C18 column (1.7 µm, 2.1 × 100 mm, Waters, 186003975) as stationary phase. The mobile phase consisted of H₂O + 0.1% HCOOH (A) and CH₃CN + 0.1% HCOOH (B). Column temperature was 40 °C, flow rate was 0.45 mL/min, and the gradient was as follows: 0–2 min, 2% B in A; 2–22.2 min, 2%–16% B in A; 22.2–31.2 min, 22%–70% B in A; 31.2–33.1 min, 70%–100% B in A; 33.1–34.6 min, 100% B in A; 34.6–35.3 min, 100%–2% B in A; 35.3–42 min, re-equilibration. The mass spectrometer was operated in the HESI negative ion mode, using the following parameters: probe heater temperature 350 °C; capillary temperature 330 °C; spray voltage 3.0 kV, sheath gas flow 50 arbitrary units; auxiliary gas flow 10 arbitrary units, S-lens RF level: 60 V; scan range: m/z 110–1650; resolution: 70,000 (FWHM at m/z 200) for full MS and 17,500 for data dependent MS/MS. Methanolic extract (SPM) and hot water extract (SPI) were dissolved at a concentration of 20 mg/mL with CH₃OH and CH₃OH:H₂O 1:1, respectively, centrifuged (10 min, 13,000 rpm), and 3 µL were injected. Peaks were annotated based on their m/z values, calculated molecular formula and MS/MS fragmentation patterns, using authentic reference compounds, literature or databases or theoretical fragmentation for annotation. Some of the compounds were also isolated and structurally elucidated.

Compound isolation

The dried CH₃OH extract (300 g) was suspended in 2 L water and subjected to subsequent liquid-liquid- extraction with n-hexane (5 × 600 mL), dichloromethane (CH₂Cl₂, 4 × 600 mL), ethyl acetate (EtOAc, 5 × 600 mL) and n-butanol (BuOH, 6 × 600 mL). The extracts were concentrated by rotary evaporation and dried under a stream of nitrogen. The EtOAc subfraction EA (2.65 g) was subjected to vacuum liquid chromatography (VLC) on silica gel (60 g, column diameter 4.5 cm, column height 12 cm), using a CH₂Cl₂− CH₃OH gradient (9:1–0:100; 38 fractions, each 50 mL). Fractions VLC1_11-13 (CH₂Cl₂:CH₃OH 8:2, 421 mg) and fractions VLC1_14-19 (CH₂Cl₂:CH₃OH 8:2, 396 mg) were combined and each subjected to solid phase extraction (SPE) using CH₃OH -H₂O mixtures with increasing CH₃OH proportion (20%–100% CH₃OH) for elution. Fractions 11–13_SPE3 (40% CH₃OH, 62 mg) and 14–19_SPE3 (40% CH₃OH, 62 mg) were purified by semipreparative HPLC with H₂O and acetonitrile (CH₃CN) as mobile phase. The column temperature was 25 °C, flow rate 4 mL/min and the gradient was 18%–20% CH₃CN in H₂O from 0 to 27 min. Purification of 11–13_SPE3 with this method afforded compound 34 (RT 24.6 min, 3.85 mg) and 39 (RT 25.9 min, 5.89 mg), and purification of 14–19_SPE3 led to the isolation of 32 (RT 18.5 min, 4.1 mg), 33 (RT 23.2 min, 7.26 mg) and 36 (RT 24.1 min, 9.85 mg) (numbering of compounds is as in Table 1).

Table 1. Annotation of peaks detectable in SPI and SPM in the HESI negative ion mode, numbered on basis of retention time.

PeakNr	RT (min)	m/z	MS/MS fragments	monoisoto- pic mass	molecular formula	Δ (ppm)	ID^ID level	reference	compound class
1	0.79	317.0547	317.0552 [phone] (5); 96.9585 [HSO4]^− (75)	318.0621	C9H18O10S	3.14	hexosylsulfate derivative 1 isomer 1^3		glycosyl sulfate
2	1.23	317.0546	317.0551 (100); 241.0021 (5); 96.9586 [HSO4]^− (65)	318.0621	C9H18O10S	2.95	hexosylsulfate derivative 1 isomer 2^3		glycosyl sulfate
3	1.56	331.0704	331.0711 [phone] (5); 96.9585 [HSO4]^- (75)	332.0777	C10H20O10S	3.28	hexosylsulfate derivative 2 isomer 1^3		glycosyl sulfate
4	1.74	218.9968	218.9968 (30) 139.0400 [phone] (5) [HSO4]^− 96.9601 (5)	220.0042	C7H8O6S	1.9	methylpyrogallolsulfate isomer^2	(Pimpão et al. 2014)	hydroxyphenylsulfate
5	1.93	331.071	331.0710 [phone] (10) 96.0602 [HSO4]^− (75)	332.0777	C10H20O10S	3.47	hexosylsulfate derivative 2 isomer 2^3		glycosyl sulfate
6	2.27	331.0703	331.0710 [phone] (5) 96.9585 [HSO4]^− (65)	332.0777	C10H20O10S	3.16	hexosylsulfate derivative 2 isomer 3^3		glycosyl sulfate
7	2.64	365.055	365.0552 (65) 241.0026 (40) 96.9602 [HSO4]^− (100)	366.0621	C13H18O10S	1.97	hydroxybenzylalcohol hexosylsulfate isomer^2	(Xu et al. 2020)	benzylglycoside sulfate
8	2.79	345.0865	345.0869 [phone] (3) 96.9586 [HSO4]^− (65)	346.0934	C11H22O10S	2.75	hexosylsulfate derivative 3^3		glycosyl sulfate
9	3.52	395.0659	395.0660 (20) 241.0027 (40) 96.9602 [HSO4]^− (100)	396.0726	C14H20O11S	2.74	dimethoxyphenyl hexosylsulfate isomer 1^2	(Maldini et al. 2009)	phenylglycoside sulfate
10	4.05	408.0425	408.0423 (30); 259.0132 (5) 166.0320 (5) 96.0586 [HSO4]^− (100) 95.9524 [SO4]^− (70) 79.9576 (65) [SO3]^−74.9894 (30)	409.0501	C14H19O9NS2	1.87	glucotropaeolin^1	reference compound	glucosinolate
11	6.17	335.0448	335.0448 (55) 241.0027 (35) 96.9602 [HSO4]^− (100)	336.0515	C12H16O9S	3.35	phenyl hexosylsulfate isomer^3		phenylglycoside sulfate
12	6.76	327.0761	327.0760 (90) 256.9977 (10) 241.0027 (15) 96.9602 [HSO4]^− (100) 79.9576 [SO3]^− (20)	328.0828	C11H20O9S	3.19	hexosylsulfate derivative 4^3		glycosyl sulfate
13	7.12	395.0656	395.0655 [phone] (25) 96.9601 [HSO4]- (100)	396.0726	C14H20O11S	3.5	dimethoxyphenyl hexosylsulfate isomer 2^2	(Maldini et al. 2009)	phenylglycoside sulfate
14	7.36	349.0596	349.0598 (85); 241.0019 (25) 138.9707 (10) 96.9586 [HSO4]^− (100) 79.9576 [SO3]^- (10)	350.0672	C13H18O9S	2.35	salvadoside isomer 1^2	(Ayoub et al. 2021; Farag et al. 2021)	benzylglycoside sulfate
15	9.03	349.0596	349.0598 (95); 241.0019 (5) 96.9586 [HSO4]^− (100)	350.0672	C13H18O9S	2.27	salvadoside isomer 2^2	(Ayoub et al. 2021; Farag et al. 2021)	benzylglycoside sulfate
16	10.46	485.1696	485.1703 [phone] (10) 241.0031 (5) 96.9602 [HSO4]^− (60)	486.1771	C19H34O12S	1.46	megastigmane hexosylsulfate 1^3		megastigmane glycosylsulfate
17	11.33	469.1753	469.1742 [phone] (3) 96.9586 [HSO4]^− (45)	470.1822	C19H34O11S	3.09	megastigmane hexosylsulfate 2 isomer 1^3		megastigmane glycosylsulfate
18	11.39	393.0495	393.0494 (20) 241.0019 (60) 152.9863 (10) 96.9585 [HSO4]^- (100)	394.057	C14H18O11S	2.35	hexosylsulfate derivative 6 isomer 1^3		phenyl/benzyl-glycoside sulfate
19	11.74	374.0555	374.0557 (55) 241.0027 (20) 96.9602 [HSO4]^− (100)	375.0624	C14H17O9NS	2.42	hexosylsulfate derivative 5 (N-containing)^3		glycosyl sulfate
20	12.34	467.1587	467.1591 [phone] (5) 256.9976 (15) 96.9602 [HSO4]^- (70)	468.1665	C19H32O11S	3.04	megastigmane hexosylsulfate 4 isomer 1^3		megastigmane glycosylsulfate
21	12.38	393.0499	393.0498 (20) 241.0024 (65) 96.9601 [HSO4]^- (100)	394.057	C14H18O11S	3.36	hexosylsulfate derivative 6 isomer 2^3		glycoside sulfate
22	12.47	465.1439	465.1429 [phone] (5) 241.0028 (5) 96.9585 [HSO4]^- (85)	466.1509	C19H30O11S	2.91	megastigmane hexosylsulfate 3 isomer 1^3		megastigmane glycosylsulfate
23	12.84	465.1439	465.1440 [phone] (10) 241.0028 (5) 96.9602 [HSO4]^− (75)	466.1509	C19H30O11S	2.97	megastigmane hexosylsulfate 3 isomer 2^3		megastigmane glycosylsulfate
24	12.98	469.1744	469.1742 [phone] (5) 256.99856 (3) 241.0020 (3) 96.9586 [HSO4]^− (50)	470.1822	C19H34O11S	3.11	megastigmane hexosylsulfate 2 isomer 2^3		megastigmane glycosylsulfate
25	14.48	467.1595	467.1594 [phone] (3) 241.0036 (3) 96.9602 [HSO4]^- (50)	468.1665	C19H32O11S	3.94	megastigmane hexosylsulfate 4 isomer 2^3		megastigmane glycosylsulfate
26	16.68	467.16	467.1599 [phone] (3) 241.0029 (3) 96.9603 [HSO4]^− (60)	468.1665	C19H32O11S	3.88	megastigmane hexosylsulfate 4 isomer 3^3		megastigmane glycosylsulfate
27	16.99	465.1439	465.1436 [phone] (3) 241.0030 (5) 96.9602 [HSO4]^− (80)	466.1509	C19H30O11S	2.91	megastigmane hexosylsulfate 3 isomer 3^3		megastigmane glycosylsulfate
28	17.28	739.2119	739.2121 (30) 284.0330 [phone] (45) 227.0355 (20) 151.0032 (5)	740.2164	C33H40O19	5.2	mauritianin^0	isolated and elucidated	flavonol glycoside
29	17.72	739.2118	739.2119 (30) 284.0331 [phone] (40) 227.0353 (20) 178.9989 (3) 151.0040 (5)	740.2164	C33H40O19	5.12	clitorin^0	isolated and elucidated	flavonol glycoside
30	18.66	769.2228	769.2225 (45) 314.0438 (95) 299.0202 [phone] (10) 271.0251 (30) 243.0301 (20) 178.9987 (5) 151.0039 (5)	770.2269	C34H42O20	5.55	typhaneoside^0	isolated and elucidated	flavonol glycoside
31	18.95	769.2228	769.2230 (45) 314.0438 [phone] [phone] (10) 271.0251 (30) 243.0301 (20) 178.9982 (3) 151.0039 (5)	770.2269	C34H42O20	5.47	isorhamnetin-3-O-rha-(1->2)-[rha-(1->6)gal]^0	isolated and elucidated	flavonol glycoside
32	19.42	593.1523	593.1523 (40) 357.0510 (3) 285.0396 (35) 284.0329 [phone] (40) 227.0353 (15)	594.1585	C27H30O15	3.78	biorobin^0	isolated and elucidated	flavonol glycoside
33	21.12	593.1522	593.1525 (50) 451.1623 (3) 327.0512 (3) 285.0404 [phone] (85) 255.0302 (40) 227.0352 (20)	594.1585	C27H30O15	3.58	nicotiflorin^0	isolated and elucidated	flavonol glycoside
34	21.37	447.0938	447.0916 (45) 284.0332 [phone] (75) 227.0353 (50) 190.9962 (35) 174.9562 (90) 146.9614 (45)	448.1006	C21H20O11	3.63	astragalin^0	isolated and elucidated	flavonol glycoside
35	22.14	453.1804	453.1805 [phone] (3) 256.9978 (5) 241.0020 (3) 96.9602 [HSO4]^− (55)	454.1872	C19H34O10S	3.17	anisoposide A isomer^3		megastigmane glycosylsulfate
36	22.2	623.1632	623.1631 (35) 315.0512 [phone] (45) 300.0273 (55) 299.0201 (55) 271.0251 (30) 255.0303 (10) 243.0298 (15)	624.169	C28H32O16	4.09	narcissin^0	isolated and elucidated	flavonol glycoside
37	22.38	469.1753	469.1750 [phone] (1) 96.9602 [HSO4]^− (25)	470.1822	C19H34O11S	3.09	megastigmane hexosylsulfate 2 isomer 3^3		megastigmane glycosylsulfate
38	22.48	449.1492	449.1491 [phone] (3) 241.0029 (3) 176.9005 (3) 96.9602 [HSO4]^− (85)	450.156	C19H30O10S	3.58	tetraenoside C isomer 1^3		megastigmane glycosylsulfate
39	22.55	477.1042	477.1043 (70) 314.0435 [phone] (15) 299.0208 (20) 285.0411 (50) 271.0255 (70) 257.0454 (20) 243.0301 (10) 190.9967 (30) 151.0037 (15)	478.1111	C22H22O12	3.01	isorhamnetin-3-O-glc^0	isolated and elucidated	flavonol glycoside
40	22.83	469.1752	469.1751 [phone] (3) 96.9602 [HSO4]^− (25)	470.1822	C19H34O11S	2.97	megastigmane hexosylsulfate 2 isomer 4^3		megastigmane glycosylsulfate
41	23.15	451.1648	451.1648 [phone] (10) 285.0248 (3) 256.9971 (5) 241.0034 (1) 180.9128 (3) 96.9602 [HSO4]^- (65)	452.1716	C19H32O10S	3.52	3-(3′-O-sulfo-β-glucopyranosyloxy)-7,8-dihydro-β-ionone ^0/1	isolated and elucidated	megastigmane glycosylsulfate
42	23.69	449.1492	449.1489 [phone] (1) 256.9976 (5) 241.0028 (3) 96.9602 [HSO4]^- (95)	450.156	C19H30O10S	3.51	tetraenoside C isomer 2^3		megastigmane glycosylsulfate
43	24.18	451.1647	451.1646 [phone] (3) 256.9949 (5) 241.0020 (3) 96.9602 [HSO4]^− (60)	452.1716	C19H32O10S	3.16	3-(3′-O-sulfo-β-glucopyranosyloxy)-7,8-dihydro-β-ionone isomer 1^2	isomer of 41	megastigmane glycosylsulfate
44	24.62	451.1651	451.1648 [phone] (5) 241.0034 (3) 180.9126 (3) 96.9602 [HSO4]/ (60)	452.1716	C19H32O10S	4.04	3-(3′-O-sulfo-β-glucopyranosyloxy)-7,8-dihydro-β-ionone 2^2	isomer of 41	megastigmane glycosylsulfate
45	25.24	495.1913 [M + HCOO]^-	495.1911 (50) 256.9977 (15) 241.0034 (1) 225.0080 (1) 96.9602 [HSO4]^− (100)	450.1924	C20H34O9S	3.62	hexosylsulfate derivative 7^3		glycosyl sulfate
46	25.46	601.2913	601.2911 [phone] (3) 256.9977 (3) 241.0027 (5) 96.9602 [HSO4]^− (65)	602.2972	C26H50O13S	4.03	hexosylsulfate derivative 8^3
47	26.73	327.2181	327.2182 [phone] (10) 229.1446 (30) 221.1190 (10) 211.1344 (55) 183.1392 (15) 171.1029 (45) 97.0661 (15) 85.0297 (20)	328.225	C18H32O5	4.61	trihydroxyoctadecadienoic acid FA18:2 + 3O isomer^2	https://mona.fiehnlab.ucdavis.edu/spectra/display/VF-NPL-QTOF000050	oxidized fatty acid
48	27.18	329.2338	329.2339 [phone] (30) 211.1343 (45) 171.1028 (80) 139.1130 (35) 127.1130 (15) 99.0817 (15)	330.3406	C18H34O5	4.8	trihydroxyoctadecenoic acid FA18:1 + 3O isomer^2	https://mona.fiehnlab.ucdavis.edu/spectra/display/PR309108	oxidized fatty acid
49	28.27	307.1919	307.1923 (20) 289.1806 (15) 235.1343 (80) 211.1343 (40) 209.1186 (35) 185.1185 (80) 125.0973 (65) 121.0660 (100) 97.0660 (85) 71.0503 (555) 65.0398 (75)	308.1988	C18H28O4	4.8	phytoprostane FA 18:1; [9-13cy5:1];OH;oxo isomer^2	(Collado-González et al. 2015)	oxylipin
50	29.65	309.2076	309.2074 (60) 291.1971 (35) 273.1864 (15) 209.1548 (25) 197.1185 (60) 185.1185 (35) 171.1028 [phone] (10) 137.0973 (40) 125.0971 (15) 119.0865 (35) 111.0817 (25)	310.2144	C18H30O4	4.99	hydroperoxyoctatrienoic acid isomer^2	(Geng et al. 2015)	oxidized fatty acid

Peaks printed in bold constitute structurally elucidated compounds, while peaks not printed in bold were only tentatively annotated on the basis of their mass spectrometry data. ID levels are according to the MSI standards (Sumner et al. 2007; Blaženović et al. 2018): 0, isolated pure compound, unambiguous 3D structure; 1, confident 2D structure, identified by reference standard match (retention time and MS/MS); 2, tentative annotation by comparison of structural formula and MS/MS fragmentation pattern with data from databases or literature. 3, no data available in literature or databases; compound or compound class tentatively annotated on the basis of the MS/MS fragmentation pattern (theoretical interpretation and/or comparison to related compounds).

The BuOH subextract (27.8 g) was fractionated by silica gel VLC (350 g, particle size 0.04–0.063 mm, column diameter 9.5 cm, column height 11 cm). The mobile phase consisted of a CH₂Cl₂–CH₃OH–H₂O-gradient, starting with CH₂Cl₂:CH₃OH:H₂O 95:5:0.5, and gradually increasing the CH₃OH–H₂O-proportion. In total, 36 fractions were collected.

The combined fraction VLC2_13–14 (1.24 g) as well as fractions VLC2_15 (0.89 g) and VLC2_16 (0.81 g) were individually further fractionated by SPE with CH₃OH and H₂O, both containing 0.1% trifluoroacetic acid (TFA), as eluent (5%–100% CH₃OH/0.1% TFA in H₂O/0.1% TFA). From each of these three SPE separations, fractions 3 and 4 (15 and 20% CH₃OH/0.1% TFA in H₂O/0.1% TFA), and in case of VLC2_15 also fraction 5 (30% CH₃OH/0.1% TFA in H₂O/0.1% TFA) were combined due to similarities in their TLC and LC-MS profiles, yielding subfraction SPE_345 (237 mg). This subfraction was further separated by column chromatography on silica gel (30 g, particle size 0.04–0.063 mm; column diameter 2 cm, column height 30 cm), using a gradient of CH₂Cl₂–CH₃OH with increasing proportions of CH₃OH (5%–50% CH₃OH), and 59 fractions (500 drops/fraction) were collected with an automatic fraction collector. A combined subfraction of this separation, CC_32_33 (CH₂Cl₂:CH₃OH 88:12, 12 mg) was purified on silica gel TLC plates, using n-BuOH:n-propanol:glacial acetic acid:H₂O (38:18:2:2) as mobile phase. This afforded compound 41 (Rf 0.67, 4.66 mg). Fractions VLC2_20–21 (3.27 g) were combined, and a part of this fraction (1.17 g) was further separated by SPE, eluting with CH₃OH–H₂O mixtures with increasing CH₃OH proportion (20%–50% CH₃OH). Fraction 20–21_3 (49 mg) was purified with semipreparative HPLC, using CH₃CN:H₂O 14:86 (isocratic) as mobile phase. This separation afforded four more compounds: 28 (RT 31.4 min, 3.89 mg), 29 (RT 34.0 min, 5.39 mg), 30 (RT 36.2 min, 3.42 mg) and 31 (RT 39.0 min, 3.11 mg).

41, experimental data: white amorphous solid; $\left[\alpha \right]\begin{array}{c}25\\ D\end{array}-17.6°\left(c\hspace{0.17em}0.27,\text{methanol}\right);$ ESIMS [M-H]⁻ 451.1646 (calcd. or C₁₉H₃₁O₁₀S⁻: 451.1643; ¹H NMR (700 MHz, CD₃OD) and ¹³C NMR (700 MHz, CD₃OD), see Table 2; experimental NMR and HRMS spectra, see Supplemental Material, Figures S1–S6. 28–34, 36, 39 (known compounds): ¹H and ¹³C data, see Supplemental Material, Tables S1–S9.

Figure 1. Effects of SPM and SPI on proinflammatory functions in human neutrophils stimulated with LPS. (A) Effect on IL-8 production (% of stimulated control sample) (B) Effect on TNF-α production (% of stimulated control cells). NST: non-stimulated control cells; ST: stimulated control cells. Statistical comparison between treatments was performed by ANOVA with Dunnett’s multiple comparison post hoc test. ^#, p < 0.001 versus non stimulated control; ***, p < 0.001 versus stimulated control.

Table 2. ¹³C (175 MHz) and ¹H (700 MHz) NMR resonance values of compound 41 in methanol-d₄ and comparison with carbon data of icariside B₆ recorded in pyridine-d₅ (Miyase et al. 1988).

Atom	compound 41 (methanol-d4)		icariside B6 (pyridine-d5)
	δC [ppm], type	δH [ppm], (J in Hz)	δC [ppm]
1	38.7, C	–	38.0
2	47.3, CH2	1.85 ddd (12.0, 3.5, 2.2) 1.49 t (12.0)	47.0
3	73.2, CH	4.07 m	71.9
4	39.7, CH2	2.02 m 2.35 m	39.6
5	126.2, C	–	125.3
6	137.5, C	–	136.9
7	22.9, CH2	2.30 m 2.18 m	22.4
8	45.0, CH2	2.53 m 2.57 m	44.4
9	211.6, C	–	208.2
10	29.8, CH3	2.13 s	29.9
11	28.7, CH3	1.04 s	28.4
12	30.0, CH3	1.04 s	29.8
13	19.9, CH3	1.61 s	19.9
1′	102.0, CH	4.53 d (7.3)	102.6
2′	73.9, CH	3.32	75.4
3′	85.8, CH	4.23 t (8.7)	78.6
4′	70.2, CH	3.52 t (8.8)	71.8
5′	77.5, CH	3.34	78.5
6′	62.5, CH2	3,86 dd (11.8, 2.4) 3,68 dd (11.8, 5.4)	63.0

Cellular in-vitro assays

Assay on IL-8 and TNF-α production in LPS- stimulated human neutrophils

All tested samples were dissolved in DMSO (stock solutions with 10 mg/mL for extracts or 10 mM for pure compounds) and further diluted with RPMI 1640 medium. Prepared dilutions were directly added to the cell cultures. Dexamethasone at 50 μM was used as a positive control. The final concentration of DMSO (<0.1%) did not influence the assays.

Peripheral venous blood taken from healthy human donors (18–35 years old) was purchased from Warsaw Blood Donation Centre. Donors did not smoke or take any medication. They were clinically recognized to be healthy and routine laboratory tests showed all values to be within the normal ranges. Neutrophils were isolated by dextran sedimentation and centrifugation in a Ficoll Hypaque gradient and then re-suspended in an RPMI 1640 medium supplemented with 10% Fetal Bovine Serum (FBS), 10 mM HEPES, 2 mM l-glutamine and Penicillin/Streptomycin (10,000 U/mL). The cell suspension was transferred to 24-well plates and incubated for 24 h at 37 °C under humidified 5% CO₂ air for cytokines analysis. Neutrophils were treated with extracts at final concentration of 100, 10 and 1 µg/mL or compounds at final concentrations 50 µM, in the case of active compounds also at 10 and 5 µM.

Cytotoxicity of the test samples was determined by a standard flow cytometric probe using propidium iodide (PI) staining: Neutrophils (2 × 10⁶/mL) were cultured in 4-well plates in RPMI 1640 medium containing 10% FBS, 10 mM HEPES, 2 mM l-glutamine and Penicillin/Streptomycin (10,000 U/mL) for 24 h at 37 °C with 5% CO₂ in presence of extracts or compounds. After 24 h the neutrophils were harvested and centrifuged (1500 rpm; 10 min; 4 °C), washed once with cold PBS and re-suspended in 400 µL of PBS. Five µL of PI (50 µg/mL) solution were added to the cell suspensions. After 15 min of incubation at room temperature cells were analyzed by BD FACSCalibur flow cytometer (BD Biosciences), 10,000 events were recorded per sample. Cells that displayed high permeability to PI were expressed as a percentage PI(+) (i.e., dead) cells.

For assay on IL-8 and TNF-α production, neutrophils (2 × 10⁶) were cultured under the same conditions as for cytotoxicity assay, in the absence or presence of LPS (100 ng/mL) and test samples (added 30 min prior to LPS addition). After 24 h the neutrophils were harvested and centrifuged (774 × g; 10 min; 4 °C). The amount of released cytokines in the supernatant was measured by enzyme-linked immunosorbent assay (ELISA) following the manufacturer’s instructions (BD Biosciences). The effects on IL-8 and TNF-α production were calculated relative to the respective cytokine levels in the stimulated control samples which lacked the test compounds.

The results were expressed as mean ± SD from at least three independent experiments, each performed using cells from a different donor. The Shapiro-Wilk test was applied to determine the normality of the results. The statistical significance of differences between means was assessed by ANOVA with Dunnett’s multiple comparison post hoc test. P values below 0.05 were considered statistically significant. All analyses were performed using Statistica 13.1. GraphPad Prism (version 5.01), which was also used to plot the data.

Assay on nitric oxide (NO) production in from LPS/IFNγ stimulated mouse macrophages

Extracts and compounds were dissolved in DMSO and further diluted with PBS buffer. Diluted samples were added to the cell suspensions at final concentrations of 50 µg/mL (extracts) and 50 µM (pure compounds). The final DMSO concentration of 0.1% did not affect cell viability. RAW 264.7 mouse macrophages (Mus musculus, CVCL_0493, Elabscience, EP-CL-0190) were routinely cultivated in house (Fischer 2006; Blunder 2010) in high glucose DMEM containing 10% FBS, 100 U/mL penicillin, 100 µg/mL streptomycin, 1.25 µg/mL amphotericin B, 4 mM glutamine, 4.5 g/L glucose and 25 mM HEPES. For the assay, cells were seeded in 24-well plates (1 mL/well) at a density of 1 × 10⁶ cells/mL and incubated for 24 h. Inflammation was induced by exchanging medium to DMEM without phenol red (supplements remain the same as in routine culture) and stimulating the cells with LPS (0.5 µg/mL) and interferon-γ (IFNγ, 5 ng/mL) as previously described (Nikzad-Langerodi et al. 2017), using the non-specific NOS-inhibitor L-NMMA at 100 µM as positive control. Afterwards, 100 µL of the supernatants were transferred to 96-well plates to evaluate NO production. 100 µL of Griess reagent (40 mg/mL) were added and optical density was photometrically determined at 535 nm using a Hidex Sense plate reader (HIDEX, Turku, Finland). NO-production was calculated by comparison with a NaN₃ standard curve and set in relation to a control sample stimulated with LPS and containing 0.1% DMSO instead of test samples. Tests were performed in two replicates on two consecutive days (n = 4). Statistical significance between solvent treated LPS/IFNγ-stimulated control samples and test samples was determined by Kruskal-Wallis One Way Analysis on Ranks followed by Dunn’s test using SigmaPlot 13.0 (Systat Software Inc., Chicago, IL, USA).

Assay on inhibition of HNO97 carcinoma cell growth

The ability to inhibit the growth of HNO97 human tongue carcinoma cells was examined by means of an XTT assay. Samples were dissolved in DMSO and diluted with PBS to reach final concentrations of 50 µg/mL (extracts) and 50 µM (pure compounds). The final DMSO concentration of 0.5% did not affect cell viability. The cell line (Homo sapiens, CVCL_D227) was purchased from Cytion (Eppelheim, Germany; 300129) and kindly provided to us by Core Facility Alternative Biomodels and Preclinical Imaging, Medical University of Graz (Graz, Austria). Cells were cultured in DMEM high glucose medium (4.5 g/L d-glucose, Gibco, Thermo Fisher) containing 10% FBS, 100 U/mL penicillin and 100 µg/mL streptomycin and 4 mM l-glutamine. The XTT viability assay was performed in accordance with the manufacturer’s protocol. In brief, 5,000 cells/well of HNO97 carcinoma cells were seeded into 96-well plates (100 μL, flat bottom) and grown overnight before the test compounds were added. After 72 h of treatment, a freshly prepared XTT solution (5 mL of XTT solution mixed with 100 μL of electron coupling reagent) was added (50 µL/well) and analyzed after 90 min of incubation at 37 °C using a HIDEX sense plate reader. Vinblastine served as positive control (0.01 μM). Results were expressed relative to solvent control incubated with 0.5% DMSO instead of test samples. Statistical significance between solvent-treated control samples and test samples was determined with One Way Analysis on Ranks followed by Dunn’s test using SigmaPlot 13.0, and IC₅₀ values were determined with the same software using the four-parameter logistic curve. Eight different concentrations and two different cell passages each tested in three independent wells were used.

Results

SPM and SPI display pronounced inhibitory effect on proinflammatory functions in LPS- stimulated neutrophils

Addition of LPS to human neutrophils led to highly increased production of the inflammatory mediators IL-8 and TNF-α. Both, SPM and SPI strongly and dose-dependently inhibited the production of both, IL-8 and TNF-α (Figure 1), without exhibiting a negative impact on cell viability (only 12.7 ± 0.6% and 14.9 ± 1.1 of cells treated with 100 µg/mL SPI and SPM, respectively, displayed high permeability for propidium iodide). At contrast, the extracts had no effect on NO production in RAW 264.7 macrophages stimulated with LPS/IFN-γ (Figure 2A). In HNO97 human tongue carcinoma cells, SPI had no impact on cell viability, whilst SPM slightly, but insignificantly enhanced their proliferation (Figure 2B).

Figure 2. (A) Effect of SPM and SPI on NO production in RAW 254.7 macrophages stimulated with LPS/IFN-γ. ST: stimulated cells, NST: non-stimulated control cells; L-NMMA: positive control; ** p < 0.01. (B) effect of SPM and SPI on the viability of HNO97 human tongue carcinoma cells; 0.1% DMSO: control cells treated with solvent; VBN: positive control; * p < 0.05. Statistical comparison to ST (A) and to 0.1% DMSO (B) was performed by Kruskal–Wallis One way analysis of variance on ranks with Dunn’s multiple comparisons versus control group post hoc test.

Megastigmane glycosylsulfates, flavonoids and glucosinolate are major SPM and SPI constituents

UHPLC-HRMS analysis revealed a very similar metabolite profile for SPI and SPM in the HESI negative mode (Figure 3). A high number of constituents were detected in this ionization mode, while positive mode ionization was only observable for a limited number of analytes (data not shown). Therefore, negative mode data were used for compound annotation. Overall, 50 peaks could be annotated in SPI and SPM (Table 1).

Figure 3. UHPLC-HRMS analysis of SPI and SPM. HESI negative mode base peak chromatograms (m/z 110-1650) of SPI (A) and SPM (B). Peak numbers are as indicated in Table 1.

Sulfur-containing constituents

Several types of sulfur-containing constituents could be annotated in SPM and SPI: First, the major peak eluting at 4.05 min (10) was identified as the glucosinolate glucotropaeolin, which has been previously described as a constituent of S. persica (Ezmirly and Seif El-Nasr 1981). It was the only glucosinolate identified in SPM and SPI. Second, 32 compounds, that displayed a fragment with m/z 96.959 [HSO₄]^- in their MS/MS spectrum, were detected (Table 1). This fragment indicated the presence of a sulfate group in these compounds. Except for peak 4 which was tentatively annotated as methylpyrogallol sulfate, the other 31 peaks shared a further common fragment with m/z 241.002 in their MS/MS spectra. This fragment was annotated as [hexosylsulfate-H₂O]⁻, Therefore, it was concluded that 31 of the 32 detected sulfates possessed a hexosylsulfate part.

The calculated molecular formulas of 17 sulfates indicated that they all share a molecular formula with 19 carbons. One of them, compound 41, could be isolated from the BuOH subextract of S. persica methanolic leaf extract by subsequent purification using silica gel VLC, RP-18 SPE, silica gel CC and eventually silica gel TLC. The isolation was complicated by the fact that the compound displayed no UV absorption, therefore the use of preparative HPLC with UV detection was not possible.

HRMS analysis indicated a molecular formula of C₁₉H₃₂O₁₀S. From the NMR data (Table 2) it could be deduced that compound 41 consisted of three parts, an exocyclic butanone moiety, a substituted cyclohexene unit, and a hexose. According to the observed HMBC correlations, the methyl groups C-11 and C-13 were attached to the same quaternary carbon and exhibited HMBC correlations to the olefinic carbon C-6 (137.5 ppm). The protons H-7 and H-8 of the exocyclic butanone showed HMBC correlation to the same carbon (137.5 ppm) defining its point of attachment. Methyl group C-13 was found to be attached to the olefinic carbon C-5, while the anomeric proton of the hexose showed an HMBC correlation to C-3 (73.2 ppm). Within the hexose, the observed coupling pattern of H-1′, H-3′ and H-4′ indicated the axial orientation of protons H-1′ to H-5′, hence the sugar was identified as glucopyranose in the β-form. The high carbon shift value (85.8 ppm) of C-3′ clearly pointed to sulfation at this position (complete proton and carbon assignments, see Table 2).

The absolute configuration of the glucose moiety was determined to be d. Therefore, the isolated compound 41 was identified as the megastigmane glycosylsulfate 4-[2,6,6-trimethyl-4-(3-O-sulfo-β-d-glucopyranosyloxy)cyclohex-1-en-1-yl]butan-2-one, or 3-(3′-O-sulfo-β-d-glucopyranosyl-oxy)-7,8-dihydro-β-ionone, respectively (Figure 4). For determination of the absolute configuration in position 3 of the megastigmane part, it would be necessary to produce the deglycosylated compound in order prepare and analyze its Mosher esters (Hoye et al. 2007). This was not possible with the low available amount of compound 41 and needs to be done in future studies.

Figure 4. Structures of compounds isolated and structurally identified from S. persica methanolic leaf extract (compound numbers are as in Table 1).

Based on comparing their MS/MS fragmentation patterns with that of isolated compound 41, further 16 peaks were tentatively annotated as megastigmane glycosylsulfates (for more details on their annotation, please refer to the text in the Supplementary Material): Peaks 35, 38, 42, 43 and 44 contained 19 carbon atoms and 10 oxygen atoms. Peaks 43 and 44 displayed the same molecular formula as compound 41 (C₁₉H₃₂O₁₀S), while in peaks 38 and 42 (C₁₉H₃₀O₁₀S) two hydrogen atoms less, and in peak 35 (C₁₉H₃₄O₁₀S) two hydrogens more were present compared to compound 41. In peaks 17, 20, 22-27, 37 and 40, 11 instead of 10 oxygen atoms, and in peak 16, 12 oxygens were present. Three isomers were detected with a calculated molecular formula C₁₉H₃₀O₁₁S (peaks 22, 23 and 27), three with C₁₉H₃₂O₁₁S (peaks 20, 25, 26), four with C₁₉H₃₄O₁₁S (peaks 17, 24, 37, 40), and the molecular formula of peak 16 was C₁₉H₃₄O₁₁S. This is the first report on megastigmane glycosylsulfates and generally on the compound class of megastigmanes in S. persica. Due to a lack of indicative MS/MS fragments, mass spectrometry data only allowed tentative annotation, but did not enable us to conclude on the exact molecular constitution of the detected megastigmane glycosylsulfates, except for isolated compound 41.

Further hexosylsulfate peaks could be tentatively annotated as phenylglycoside sulfates (peaks 9, 11, 13) and benzylglycoside sulfates (peaks 14 and 15). The benzylglycoside sulfate salvadoside that has the same molecular formula as peaks 14 and 15, has been previously isolated from S. persica stems (Kamel et al. 1992), whereas phenylglycoside sulfates have hitherto not been detected in the species.

Flavonoids

Nine of the peaks (28–34, 36, 39) were annotated as flavonoid glycosides. All nine compounds were isolated from EtOAc or BuOH sub-extracts of S. persica methanolic leaf extract by means of VLC, RP-18 SPE and semipreparative HPLC, and their structures were elucidated by HRMS and 1D- and 2D- NMR spectroscopy. For all flavonoids, although they are known compounds, de novo structural assignment was performed with pyridine-d₅ as solvent. The complete proton and carbon assignments are listed in the Supplementary Material (Tables S1–S9). Compounds 28, 29 and 32–34 were identified as kaempferol glycosides, while compounds 30, 31, 36, and 39 were assigned as isorhamnetin glycosides (Figure 4). Compounds 34 and 39 were elucidated as 3-O-monoglycosides of kaempferol and isorhamnetin, respectively; 32 and 33 were kaempferol-, and 36 isorhamnetin- 3-O-diglycosides. 28–31 were assigned as branched triglycosides. 28–29, 30–31 and 32–33 constituted pairs of epimers, 28, 31 and 33 having a galactopyranose, and 29, 30 and 32 having a glucopyranose moiety directly attached to the flavonoid 3-O-position. The isolated flavonoids have already been isolated previously from S. persica (Ali et al. 1997; Owis et al. 2020)

Lipids

Finally, four peaks that could be detected in SPM and partially also in SPI were annotated as lipids: Three oxidized fatty acids (peaks 47, 48 and 50), as well as one phytoprostane (49).

Glucotropaeolin and its metabolite benzylisothiocyanate display most pronounced oral health-promoting effects in vitro

The constituents isolated from S. persica methanolic leaf extract and the commercially available major SPM and SPI constituent glucotropaeolin (GT) were subjected to the same pharmacological tests as the extracts. In addition, benzylisothiocyanate (BIT), a known myrosinase product of GT, was subjected to pharmacological testing.

At a concentration of 50 µM, potent anti-inflammatory activity in LPS-stimulated neutrophils was observed for GT and its myrosinase product BIT (Table 3). They dose-dependently inhibited LPS- induced TNF-α- and IL8-release in human neutrophils without impairing cell viability (Figure 5A,B and Table 3). BIT also moderately inhibited NO formation in LPS/IFNγ-stimulated RAW 264.7 mouse macrophages. Moreover, it was the only compound that showed a strong dose- dependent inhibition of HNO97 cancer cell growth, with an IC₅₀ value of 10.19 ± 0.72 µM (Figure 5C). The other tested SPM and SPI constituents displayed only moderate or no effects at the screening concentration of 50 µM (Table 3).

Figure 5. A, B: Effects of different concentrations of GT (10) and BIT on proinflammatory functions in human neutrophils stimulated with LPS. (A) Effect on IL-8 production. (B) Effect on TNF-α production. Statistical test was same as indicated in Figure 1. ^#, p < 0.001 vs. unstimulated control; ***, p < 0.001 vs. stimulated control; (C) Effect of different concentrations of BIT on cell viability of HNO97 tongue carcinoma cells. Dashed lines: 95% confidence band; dotted lines: 95% prediction band.

Table 3. Effects of pure compounds present in SPI and SPM on human neutrophils stimulated with LPS, on murine RAW 264.7 macrophages stimulated with LPS/IFNγ and on the viability of HNO97 cells.

		neutrophils stimulated with LPS			RAW 264.7 cells stimulated with LPS/IFNγ	HNO97 cells
		TNF-α [%± SD]^a	IL-8 [% ± SD]^a	PI (+)cells^b [% ±SD]	NO [% ± SD]^a	Viability [%± SD]^c
	unstimulated	6.0 ± 1.2	12.3 ± 3.0	13.2 ± 2.0	−2.2 ± 0.7	100 ± 0.02
	stimulated	100 ± 5.5^#	100 ± 5.9^#	16.3 ± 5.2	100 ± 02.7^#
Nr	stimulated + 50 µM
10	glucotropaeolin (GT)	22.7 ± 2.6***	13.3 ± 2.0***	9.3 ± 0.1	90.1 ± 5.5	126.0 ± 12.7^++
ND	benzylisothiocyanate (BT)	14.4 ± 0.4***	9.4 ± 1.8***	9.5 ± 0.1	62.5 ± 10.5*	6.6 ± 0.1^++
28	mauritianin	61.0 ± 0.3*	82.6 ± 9.5	12.5 ± 0.5	94.0 ± 2.5	119.9 ± 2.7
29	clitorin	77.7 ± 8.3	107.2 ± 0.9	14.0 ± 1.5	89.2 ± 1.7	124.4 ± 6.2^+
30	typhaneoside	69.6 ± 0.8*	85.9 ± 2.5	12.9 ± 0.8	95.5 ± 0.8	123.3 ± 4.48^+
31	isorhamnetin-3-rhamnosyl-(1->2)-[rhamnosyl-(1->6)-galactoside]	64.2 ± 1.5*	103.1 ± 15.9	12.1 ± 1.3	90.4 ± 4.2	126.6 ± 3.8^++
32	biorobin	83.2 ± 2.6	94.1 ± 6.2	15.1 ± 1.5	99.4 ± 5.6	124.6 ± 6.5^+
33	nicotiflorin	91.5 ± 3.8	107.4 ± 11.4	11.8 ± 1.4	93.1 ± 1.1	124.3 ± 3.5^+
34	astragalin	61.9 ± 2.7*	71.9 ± 1.8*	14.7 ± 1.9	94.1 ± 2.1	123.9 ± 5.7^+
36	narcissin	64.7 ± 3.8*	72.4 ± 6.3*	12.5 ± 0.8	100.2 ± 6.6	122.9 ± 3.7
39	isorhamnetin-3-O-β-D-glucopyranoside	54.6 ± 1.1*	87.2 ± 2.3	14.5 ± 2.2	95.0 ± 3.45	125.3 ± 6.2^++
41	3-(3′-O-sulfo-β-glucopyranosyloxy)-7,8-dihydro-β-ionone	36.3 ± 10.6**	61.1 ± 5.8**	16.3 ± 2.3	97.1 ± 5.6	102.0 ± 3.3
	positive control
	dexamethasone 50 µM	24.5 ± 3.7**	20.4 ± 2.0**
	L-NMMA 100 µM				58.1 ± 7.0**
	vinblastine 50 ng/ml					14.5 ± 0.8++

Compound numbering is as in Figure 1 and Table 1. Benzylisothiocyanate was not detected in the extract but is a common degradation product of glucotropaeolin (10).

^aGiven as % of release of stimulated control cells treated with vehicle.

^bExpressed as percentage of cells displaying high permeability for propidium iodide.

^cGiven as % of the viability of control cells treated with vehicle. Compounds showing >50 % inhibition printed in bold. The same statistical tests were used as indicated in Figures 1 and 2.

^#p < 0.001 versus un-stimulated control; *p < 0.05; **p < 0.01; ***p < 0.001 versus stimulated control; ⁺p < 0.05, ⁺⁺p < 0.01 versus solvent control.

Discussion

In our study, the oral health-promoting effects of S. persica leaves were assessed in different cellular models of inflammation and cancer, and their chemical composition was investigated by UHPLC-HRMS metabolite profiling. The investigated preparations were a hot water extract (SPI) and a methanolic extract (SPM) of the dried leaves. These preparations were used in order to extract both, the polar compounds which can be expected in traditionally used hot water preparations, and more non-polar compounds that can be expected to be released when leaves are masticated for the traditional treatment of tooth and gum problems (Sher et al. 2010).

By means of UHPLC-HRMS metabolite profiling, in addition to flavonoid glycosides, sulfur-containing constituents, including the glucosinolate GT, phenyl- and benzylglycoside sulfates, as well as megastigmane glycosylsulfates, were identified as predominantly occurring compound classes. Our study is the first report on the presence of megastigmanes in S. persica, and UHPLC-HRMS metabolite profiling indicated that the detected megastigmane glycosylsulfates occur in the studied leaf extracts at a high structural variety.

The megastigmane glycoside skeleton of the isolated and newly identified 3-(3′-O-sulfo-β-d-glucopyranosyloxy)-7,8-dihydro-β-ionone (41) is known. The 3 R-enantiomer, icariside B₆, has been isolated and structurally identified from Epimedium grandiflorum MORR. var. thunbergianum (MIQ.) NAKAI. III for the first time (Miyase et al. 1988). From the 3S-enantiomer, only the aglycone is described in the literature. It has been isolated from Chamaecyparis formosensis Matsumara (Cupressaceae) for the first time (Lin 1999). No sulfated derivatives of these compounds have been described to date.

Megastigmanes are C13-norterpenoids structurally related to sesquiterpenes that are generated by oxidative cleavage of carotenoids (González-Delgado et al. 2017; Kornpointner et al. 2022). Until the early 2010s, around 150 megastigmanes have been reported (Breitmaier 2012), and although no more recent systematic review on this compound class is available in the literature, it can be assumed that the number has risen considerably since then (Kornpointner et al. 2022). Compared to that, reports on the detection of sulfated megastigmane glycosides in plants are very scarce to date. A megastigmane glycosylsulfate with the same molecular formula as in peaks 38 and 42, namely tetraenoside C, has been previously isolated from Tetraena mongolica Maxim. (Zygophyllaceae) (Wu et al. 2019), and a megastigmane glycosylsulfate with a molecular composition as in peak 35, anisoposide A, has been previously isolated from Mallotus anisopodus (Gagnep.) (Euphorbiaceae) (van Minh et al. 2009). Furthermore, corchoionoside C 6′-O-sulfate (Ozgen et al. 2010; Macahig et al. 2011) and tetraenoside A (Wu et al. 2019) with molecular formula C₁₉H₃₀O₁₁S, like in peaks 22, 23 and 37, as well as tetraenoside B (molecular formula C₁₉H₃₂O₁₁S, like in peaks 20, 25 and 26) (Wu et al. 2019) have been previously isolated from herbal materials. Due to a lack of indicative MS/MS fragments, we were unable to conclude whether the megastigmane glycosylsulfates detected in SPM and SPI correspond to the ones already described in the literature based on UHPLC-HRMS metabolite profiling. Their isolation and full structural elucidation are envisaged in a follow-up project.

(Kornpointner et al. 2022) recently isolated blumenol C sulfate, a non-glycosylated compound with sulfate directly attached to the megastigmane moiety, from the leaves of a Palicourea luxurians (Rusby) Borhidi specimen collected in a sulfate-rich environment. They suggested that the formation of this compound may be a defense mechanism against SO₄²⁻ - induced salt stress in the plant tissue of this species, however they did not assess whether the compound was also formed in specimens grown under lower SO₄²−- exposure. As far as S. persica is concerned, (Sawidis 2013) has demonstrated sulfur accumulation in the parenchyma cells of its stems, however it is not known whether sulfur is also accumulated in its leaves. Therefore, broader investigations on the occurrence of the newly detected megastigmane glycosylsulfates in different S. persica accessions grown on different soils, as well of different parts of the plant will be conducted in the future.

Pharmacological testing revealed that both S. persica extracts as well as their major constituent GT showed pronounced anti-inflammatory activities in LPS-stimulated human neutrophils. The inhibitory activity of the S. persica leaf extracts and GT on IL-8 and TNFα secretion is described herein for the first time.

BIT is known to be formed by the degradation of GT catalyzed by the enzyme myrosinase. In intact plant cells, this enzyme is kept separately from GT, but as soon as the cells are disrupted, the enzymatic cleavage takes place and BIT is formed (Nastruzzi et al. 2000; Kleinwächter et al. 2008). Although BIT was not detected in SPI and SPM in this study, it can be expected to be formed in certain traditional applications, e.g., when fresh leaves are masticated to relieve tooth and gum problems (Sher et al. 2010).

At contrast to the S. persica extracts and GT, BIT not only showed pronounced anti-inflammatory activities in LPS-stimulated human neutrophils, but also moderately inhibited NO release in LPS/IFNγ-stimulated RAW 264.7 macrophages, and it potently inhibited the proliferation of HNO97 tongue carcinoma cells. Similar observations have been made in a study assessing the anti-inflammatory and anticancer effects of Lepidium meyenii: in contrast to the intact glucosinolates, their myrosinase product BIT inhibited NO formation in LPS-stimulated mouse macrophages and the growth of five different tumor cell lines (Yan et al. 2022). Furthermore, numerous effects related to anti-inflammatory activity have been previously reported for BIT, among them inhibition of LPS-induced secretion of TNFα and several pro-inflammatory interleukins, but not of IL-8 (Ohta et al. 2004; Lee et al. 2009).

The major flavonoids present in SPM and SPI only showed moderate effects on LPS-induced TNFα secretion. Among them, isorhamnetin-3-O-β-d-glucopyranoside displayed the highest activity. The compound has previously shown moderate anti-inflammatory effects, e.g., reduction of TNF-α levels in human peripheral blood mononuclear cells stimulated with phytohemagglutinin (Abdallah and Esmat 2017) and in LPS/IFN-γ-stimulated mouse macrophages (Rao et al. 2008).

The newly isolated 3-(3′-O-sulfo-β-D-glucopyranosyloxy)-7,8-dihydro-β-ionone (41) moderately inhibited TNFα- and IL-8 secretion in LPS-stimulated neutrophils. While some unsubstituted megastigmanes and megastigmane glycosides have shown pharmacological effects, such as in vitro anti-inflammatory (Pan et al. 2019; Huang et al. 2023) and anticancer (Panza et al. 2011; Shu et al. 2013) activities, the pharmacological potential of megastigmane glucosylsulfates is hitherto more or less uninvestigated.

The onset and progression of periodontitis is attributed to the release of virulence factors by key bacteria in the dental biofilm that stimulate the inflammatory immune response in the host tissue, leading to increased interleukin production. IL-8 particularly acts as chemotactic agent, recruiting inflammatory cells such as neutrophils and macrophages to the periodontal lesion (Saliem et al. 2022) Arrival of these cells at an inflammation site has been shown to initiate the secretion of inflammatory chemokines, matrix metalloproteinases and cytokines, including TNF-α in circulating and wound neutrophils (Theilgaard-Mönch et al. 2004; Nussbaum and Shapira 2011). As periodontal disease proceeds, the growing populations of immune cells such as lymphocytes and T helper cells contribute to increasing levels of cytokines, such as TNF-α, IL-1, IL-4, IL-10, interferon-γ, and TGF-β (Graves 2008; Hoare et al. 2019; Saliem et al. 2022).

Despite being part of the nonspecific natural defense mechanisms of the oral cavity against pathogenic bacteria, NO, when produced at excessive levels, can also contribute to tissue damage in periodontal tissues and is thought to be involved in the pathogenesis of periodontal diseases (Uğar-Cankal and Ozmeric 2006). Endotoxins from Gram-negative bacteria occurring in periodontal disease and proinflammatory cytokines trigger resident as well as migrant cells to produce NO. The enzyme inducible NO synthase (iNOS) has been found to be upregulated during periodontitis and inflammation associated with peri-implantitis, leading to elevated NO levels (Hussain et al. 2016).

Conclusions

By means of UHPLC-HRMS metabolite profiling and subsequent isolation of major constituents, flavonoid glycosides, the glucosinolate GT, phenyl- and benzyl-glycoside sulfates and megastigmane glycosylsulfates were identified as major compound classes occurring in S. persica leaves. S. persica hot water and methanolic leaf extracts as well as their major constituent GT and the GT degradation product BIT were found to strongly inhibit the release of TNF-α and IL-8, two proinflammatory mediators implicated in periodontal disease, in LPS-stimulated neutrophils. BIT, which is likely formed by myrosinase-catalyzed degradation of GT during traditional application of S. persica leaves, additionally inhibited NO formation in LPS/IFNγ-stimulated RAW 264.7 macrophages as well as the proliferation of HNO97 tongue carcinoma cells, thereby further contributing to the oral health- promoting effects of S. persica leaves. Megastigmane glycosylsulfates were identified for the first time in S. persica and deserve deeper chemical and pharmacological investigation in the future.

Authors contributions

Conceptualization, EMPW, AMA, AKK,; methodology, EMPW, OK, AKK, NK; software; N/A; validation: EMPW, OK, AKK, TP; formal analysis, SK, BG, OK, NK, AMA, SK, CD, PR, TP, AKK; investigation, SK, BG, OK, AMA, PR, NK, CD, TP, EMPW; resources, AMA, RB, AKK; data curation, OK, EMPW; writing – original draft, EMPW, OK, BG; writing – review & editing, SK, BG, OK, AMA, PR, NK, CD, TP, EMPW, RB, AKK; visualization, OK, BG, AKK, NK, PR, TP, EMPW; supervision, EMPW, AMA, AKK, RB; project administration, EMPW, AMA, AKK; funding acquisition: RB, AKK, AMA.

Acknowledgements

We thank Ass.-Prof. Dr. Beate Rinner, Core Facility Alternative Biomodels and Preclinical Imaging, Medical University of Graz (Graz, Austria) for providing the HNO97 cell line. Robert Weis (University of Graz) is acknowledged for his support with the nomenclature of the new compound. NAWI Graz is thanked for supporting Central Lab Environmental, Plant & Microbial Metabolomics.

Disclosure statement

The authors declare no conflicts of interest.

Additional information

Funding

The research stay of Prof. A.M.A. Mazen at University of Graz (Austria) in September- October 2016 was financially supported by Erasmus Mundus (EU-Metalic Fellowship) and by Sohag University (Sohag, Egypt).