Bio-guided isolation of androsta-1,4-dien-3,16-dione as a vasodilator active principle from the inflorescence of Ravenala madagascariensis Sonn. (Strelitziaceae)

Abstract Androsta-1,4-dien-3,16-dione was isolated for the first time from the plant kingdom of the ethanolic extract of the Ravenala madagascariensis’ inflorescence by the bio-guided method. Its structure was elucidated by NMR and MS spectroscopic data analysis. The vascular effects of ethanol extracts, fractions and androsta-1,4-dien-3,16-dione were assessed on the phenylephrine pre-contracted isolated rat aorta. The isolated compound exerted the most potent vaso-relaxing effect (EC50 = 109.32 ± 15.82 µM) than the ethanol extract and fractions. The pharmacological mechanism of its vaso-relaxing action was analysed on isolated rat aorta using free-endothelial vascular tissue, specific contracting reagents (CaCl2 and KCl), antagonist (propranolol), enzyme inhibitors (L-NAME, methylene blue) and channel blocker (glibenclamide). Its vaso-relaxing activity could be due, at least partly, to the non-specific inhibition of the calcic influx. Graphical Abstract


Introduction
Arterial hypertension is one of the most prevalent non-communicable diseases and constitutes the main risk factor for morbidity and mortality related to cardiovascular diseases (annually, 17.9 million dead patients worldwide) (WHO. 2017). Morbidities caused by this chronic disease are becoming heavier and heavier for the world public health and economy, in particular undeveloped countries, but also low and middleincome countries (Lekoubou et al. 2010). At the beginning of the XXI century, the number of the patients suffering from arterial hypertension in the world was estimated at 972 million of which 65.7% belonging to the economically developing countries and the projection for 2025 estimates that it could reach to 1.563 billion where 1.15 billion cases could be observed in the economically developing countries (Lekoubou et al. 2010). Recent statistics published by WHO indicate that the most important hypertension prevalence is observed in the African region (46%) (WHO 2014). During the last decades, its prevalence in the Sub-Saharan countries was alarmingly increased. Unfortunately, only (i) 27% of them were aware of their hypertensive status before the survey, (ii) 18% were subjected to medical treatments and (iii) 7% checked their blood pressure (Ataklte et al. 2015). In Madagascar, hypertension prevalence is 35.8% (Nulu et al. 2016). Some differences in the hypertension epidemiological profile are observed comparing rural and urban populations. Its prevalence is less important in the rural population (27.6%), even if only 0.5% of the hypertensive-diagnosed rural patients were treated with medicines (Ratovoson et al. 2014). Generally, Malagasy rural people use plants to treat specific diseases, in particular diseases requiring a long treatment period such as arterial hypertension. The most important reasons are due to sociocultural and financial issues, insufficiency of sanitary infrastructures, lack of health assistance systems, and its low efficacy. The most used traditional preparation is the herbal tea prepared as infusion and decoction. Several Malagasy plants are known for their antihypertensive action. Ravenala madagascariensis (local name: 'ravinala', meaning 'leaves of the forest') has been investigated in this research. Ethnomedicinal investigations, carried out by Razafindraibe et al. (2013) on the autochthon populations in the Malagasy Eastern coast, showed that local populations usually treat hypertension symptoms by daily drinking three glasses of ravinala fresh leaf decoction during a week. This preparation is also used to treat some digestive problems, liver diseases, and other diseases (Suroowan and Mahomoodally 2020).
Other medicinal properties were reported for different parts of ravinala as antidiarrheal (Razafindraibe et al. 2013), anti-inflammatory (Rakotoarivelo et al. 2014), and anti-fibril activities (activity against atrial fibrillation) (Rasoanaivo et al. 1992); it is also used for the treatment of the kidney stones (Priyadarshini et al. 2010a). Ravinala shows also antiseptic, antidiabetic (Jain and Srivastava 2005;Priyadarsini et al. 2010b), and larvicidal activities . Some chemical investigations (Ramiarantsoa et al. 2008(Ramiarantsoa et al. , 2014 allowed to isolate different bioactive compounds, as b-sitosteryl-D-glucoside, (2E, 7R, 11R) phytyl-3,7,11,15-tetramethylhexadec-2-enyl pentadecanoate, (24S, 31S)-cycloartan-31,32-diol, and cycloartanol. The latter is known for its antidiabetic activity (Tanaka et al. 2006;Ragasa et al. 2013). Vegetable oils extracted from seeds and arils of R. madagascariensis fruits mainly contain oleic acid, palmitic acid, and several sterols (Rabarisoa et al. 1981). From the stem an edible fat is obtained, similar to the oil extracted from the butter of the butter tree (Butyro spermumparkii, syn. Vitellaria paradoxa), which is also known as karit e oil and is widely used in cosmetics. Indeed, the sap of the plant or an extract from the young leaves has been reported as a hydrating active agent in a cosmetic composition, to restore, maintain or reinforce the hydration state of the skin (Reninel and Andre 2014).
R. madagascariensis is an endemic herbal plant belonging to the Malagasy flora, widely known as the traveller's palm or traveller's tree. It is the only member of the genus Ravelana, belonging to the Strelitziaceae family. It especially grows in the plant secondary formations of the eastern part of Madagascar. Its leaves are used as the main building material for the roof and wall in the local constructions due to their high resistance to the humidity. This plant is a symbol of Madagascar and therefore its photo appears in the Malagasy state seal. Nowadays, the plant is also cultivated as ornamental and therefore introduced in several countries.
A chemical and pharmacological investigation of different parts of this plant was carried out as a contribution to the scientific valorisation of Malagasy popular medicine. This study aimed to identify and isolate the bioactive compound responsible for the antihypertensive property of this species, and to elucidate the pharmacological mechanism of its action.

Chemical investigation
Leaves, petioles, and inflorescences were separately investigated to determine the most active aerial part of R. madagascariensis for the anti-hypertensive activity. The ethanol crude extract of each plant material has been tested on the phenylephrine pre-contracted isolated rat aorta. The inflorescence (EEI) extract showed the most potent vaso-relaxing effect with an EC 50 ¼ 1.27 ± 0.05 mg/ml so it was investigated for the bio-guided fractionation by liquid-liquid partition (water-ethyl acetate) giving two fractions where the ethyl acetate fraction was the most active (EC 50 ¼ 667.22 ± 75.10 mg/mL). Several sub-fractions were obtained using silica gel column chromatography and the purification on preparative TLC allowed for the isolation of a pure white amorphous powder (35 mg, purity at 99%) from the most active fraction. Its molecular formula was determined to be C 19 H 24 O 2 by HR-ESI-MS ([M þ H] þ 285.1849 (calculated mass 285.1855)) suggesting the presence of 8 degrees of unsaturation. The UV absorption maxima at 298 and 240 nm suggested the presence of the carbonylic group and/or olefinic group. The structure of the isolated compound was determined as 10, 13-dimethyl-7,8,9,11,12,14,15,17-octahydro-6H-cyclopenta[a]phenanthrene-3,16-dione or androsta-1,4-dien-3,16-dione.
Comparative TLC of different extracts from leaves, petioles, or inflorescences was performed and showed the presence of androsta-1,4-dien-3,16-dione in all the considered organs with the highest concentration in the inflorescence. It suggests that this steroid should also be the bioactive molecule in the leaves, the plant material used in traditional medicine. In any case, other molecules may synergistically act together with the isolated bioactive molecule. The role of these compounds on plant antihypertensive properties should be investigated in further studies.

Pharmacological investigations
The vaso-relaxing activity of the isolated compound was comparable to that of 4androsten-3,17-dione (EC 50 ¼ 109.32 ± 15.82 vs 119.15 ± 19.22 mM; n ¼ 12; p > 0.05) in the same experimental conditions, which is a commercialized aromatase inhibitor (Abul-Hajj et al. 1995). This observation showed that the vaso-relaxing activity of these steroids did not depend either on the number of the unsaturated bonds or the position of the ketone functions (C-16 or C-17) on the steroid structure but may depend on the base structure of the steroid. Further studies are necessary to confirm this hypothesis. The presence of a carbonyl function in the steroid structure in the 16 position, instead of the usual 17 wherein it is a consequence of the oxidative loss of the side chain, is very unusual. The antihypertensive value of this plant could be due, at least, to its vaso-relaxing effect confirming its empirical use. The importance of vasoactive molecules as a direct vasodilator (e.g. organic nitrates, hydralazine, or openers of the ATP-dependent K þ channels) and compounds against vasoconstriction (e.g. calcium channel a 1 -adrenergic, AT 1 or ET A receptor competitive antagonists) in the management of essential hypertension is well established in previous studies (Taylor and Abdel-Rahman 2009). They may reduce peripheral arterial resistance and blood pressure.
The absence of endothelium did not affect the vaso-relaxing activity of the isolated compound (EC 50 ¼ 110.49 ± 15.48 lM). Furthermore, L-NAME which is an endothelial NO-synthase (eNOS) inhibitor, did not affect significantly its vaso-dilating effect (EC 50 ¼ 107.68 ± 17.95 lM). These results could indicate that its vaso-relaxing activity didn't involve the endothelial-derived relaxing factors (EDRFs) including nitric oxide (NO), prostacyclin (PGI2), and endothelium-derived hyperpolarizing factors (EDHFs). Certain vasodilator molecules, like acetylcholine, histamine or bradykinin, cause vasodilatation by activating eNOS, which catabolizes arginine to citrulline and NO. Nitric oxide, then, activates soluble guanylate cyclase in the muscular layer to generate cGMP, an intracellular second messenger for vascular smooth muscle relaxation (VSMC) (Ignarro 2019).
Comparatively, its vaso-relaxing effect was statistically less potent on the KCl precontracted aorta (EC 50 ¼ 166.16 ± 9.85 lM and Emax ¼ 542.95 ± 12.40 mM) rather than on PE pre-contracted organ (EC 50 ¼ 109.44 ± 15.83 lM and Emax ¼ 246.33 ± 5.98 mM) (p ˂ 0.02 and p < 0.05, respectively). PE induces vaso-contraction by stimulating the a 1 -adrenergic receptor subtype which is coupled to the Gq /11 subunit of the G protein and activates the enzyme phospholipase-C ß generating the two intracellular second messengers, diacylglycerol (DAG) and inositol triphosphate (IP 3 ). This latter binds to its receptor on the sarcoplasmic reticulum (IP 3 -Rs) and induces the release of Ca 2þ at the origin of the cellular contraction (Narayanan et al. 2012). Instead, KCl induces depolarization of VSMC and opens voltage-gated Ca 2þ channels (CaV1.2). In this case, vasoconstriction is mediated by Ca 2þ influx (Hosey et al.1996, Yamanaga et al. 1997). The Ca 2þ influx, which, by the calcium-induced calcium-release pathway, stimulates the Ca 2þ release from the sarcoplasmic endothelium via the ryanodine receptor (RYR) hence the contraction (Collier et al. 2000). In these two pathways, androsta-1,4-dien-3,16-dione can induce vasodilation with a more potent effect when the vasoconstriction is caused only by the calcium released from the sarcoplasmic reticulum.
In the presence of 10 À6 M propranolol, a competitive b 2 -adrenergic receptor antagonist, a reduction of the androsta-1,4-diene-3,16-dione EC 50 value on the PE pre-contracted rat aorta was observed (EC 50 ¼ 102.05 ± 10.56 lM), but the difference was statistically no significant (p > 0,05). It indicates that the vaso-relaxing activity may not be mediated by the stimulation of b 2 -adrenergic receptors.
Neither its EC 50 (95.29 ± 6.62 lM, p > 0.05) nor its maximal effect is significantly modified by 10 À5 M glibenclamide, which is a blocking agent of the K þ -ATP dependent channel. It means that 1,4-androstadien-3,16-dione can't act on this K þ channel subtype. Some antihypertensive molecules like diazoxide or minoxidil provoke vasodilatation by opening this channel.
In the presence of 10 mM methylene blue, the EC 50 of the androsta-1,4-dien-3,16dione was decreased to 99.23 ± 8.45 lM but not statistically different (p > 0.05). Methylene blue is a rather unspecific inhibitor of guanylate cyclase as reported by several studies (Salaris et al. 1991, Marczin et al. 1992, Mayer et al. 1993, Ashraf et al. 2004). In any case, its maximal effect was significantly reduced by 29.1 ± 3.40% in the presence of methylene blue (p < 0.02). It suggests that cGMP is probably involved, at least partly, in the vaso-relaxing activity of 1,4-androstadien-3,16 dione. In this case, androsta-1,4-dien-3,16-dione could activate soluble guanylate cyclase. It is a steroid compound which may passively pass through the cytoplasm membrane of the VSMC and could interact with this enzyme and/or sarcoplasmic/endoplasmic reticulum Ca 2þ -ATPases (SERCA).
In the presence of androsta-1,4-dien-3,16-dione, the EC 50 of the external Ca 2þ presented a significant increase, but not in a concentration-dependent manner. However, its E max decreased in concentration-dependent manner in the presence of the androsta-1,4-dien-3,16-dione (p < 0.001 with ANOVA). On the contrary, it is not the case with the verapamil which competitively inhibits the effect of CaCl 2 . It could indicate that this compound didn't act on the L-calcium channel subtype (Ca v 1.2). Cellular relaxation is due to the return of the calcium concentration in the cytoplasm to its resting value thanks to the Ca 2þ -ATPases pumps like the plasma membrane calcium ATPase (PMCA) and especially the SERCA. The androsta-1,4-diene-3,16dione has a vaso-relaxing effect on a pre-contracted preparation. This effect is certainly secondary to the decrease in the cytoplasmic calcium concentration. However, the mechanism of this drop is not yet determined. It may activate the recovery of Ca 2þ to its intracellular storage site.
The results of this research constitute potential scientific evidence of the antihypertensive value of the R. madagascariensis inflorescence. Further pharmacological studies are necessary to highlight the molecular mechanism of the vaso-relaxing activity of androsta-1,4-dien-3,16-dione and investigate acute and chronic toxicity of the used form in popular medicine.

General experiment procedures
The UV spectrum was recorded using HPLC (Alliance e2695, Waters) coupled to a PDA detector 2998 (Waters). The NMR spectra were recorded using a Bruker AMX-500 500 MHz NMR Spectrometer (Bruket Bruker BioSpin GmbH, Rheinstetten, Germany) at 298 K in DMSO-d6. HRESIMS/MS spectra were recorded with a mass spectrometer (Bruker Daltonik GmbH, Germany) in positive mode.
The powdered plant material (500 g) was macerated and stirred for 24 h in 90% ethanol at room temperature. The extraction process was repeated three times. Macerates were mixed, depigmented by passing through a charcoal layer, filtered, and evaporated at 40 C to dryness under reduced pressure using a B€ uchi rotavapor.

Animals
Adult male or female Wistar rats (150 ± 20 g, aged to 4 months) were used for the in vitro pharmacological tests. The research was carried out following the internationally accepted principles for laboratory animal use and care. All the experiments were approved by the local animal ethic committee (09/EA-IMRA/2018).

Isolated rat aorta preparation
The thoracic aorta was isolated from rats after purging the animal anaesthetized with chloroform. The aorta strip cleared from connective tissue was cut to 2 mm rings. Each ring preparation was suspended in the organ bath containing Krebs solution maintained at 37 C and continuously gassed with carbogen (5% CO 2 and 95% O 2 ) to maintain pH to 7.4. The tissue was mounted under an initial tension of 2 g. Isometric tension changes were detected. After 1 h of equilibration, its sensitivity was tested with 10 À5 M phenylephrine (PE).

Vaso-relaxing effect assessment
Then, the isolated rat aorta was pre-contracted with 10 À7 M PE. If the plateau of the contraction was reached, increasing concentrations of the studied sample were cumulatively added into the organ bath. The percentage of relaxation caused by each tested concentration was calculated considering 100% the maximal amplitude of the contraction due to phenylephrine. Concentration-response curves were applied. The concentration which exhibits 50% of relaxation (EC 50 ) was calculated by linear regression of the exponential part of the curve. The alcohol crude extracts of leaves, petioles, or inflorescences were separately tested. Samples were solubilized in dimethyl-sulfoxide (DMSO) and then diluted with distilled water to reduce the final rate of DMSO lower than 0.5%. The 0.5% DMSO was only used for the sample preparation. Extracts were diluted 20 times before injection in the organ bath reducing the DMSO concentration to 0.025%.

Bio-guided fractionation and isolation
The alcohol extract of the inflorescence was partitioned between distilled water and ethyl acetate. The vaso-relaxing effect of the organic fraction was the most powerful. For this reason, the fractionation continued on this fraction. Some chromatographic methods (glass column on silica gel, TLC, or a preparative TLC) were used to isolate and purify the bioactive compound. The bio-active constituent responsible for the vaso-relaxing activity was obtained in purified and powdered form. The purity of the isolated compound was verified using an HPLC (High-Performance Liquid Chromatography analysis (Alliance e2695, Waters) coupled to a photodiode array detector 2998 (Waters).
Its vaso-relaxing activity was assessed on the KCl pre-contracted isolated rat aorta or the denuded endothelial isolated rat aorta.

Data analysis
Each test was repeated at least in 4 independent experiments. All the EC 50 were expressed as mean ± standard error of the mean (s.e.m). For statistical evaluation of significant paired or unpaired differences, Student's t-test and one-way ANOVA followed by Tukey post-hoc were used (p ˂ 0.05). All statistical calculations were performed with IBM SPSS Statistics 22.0 (IBM, Armonk, NY, USA).

Conclusion
In this preliminary research, R. madagascariensis aerial parts, used by Malagasy people in the Eastern coast to treat high blood pressure, were studied. They cause a concentration-dependent vaso-relaxing effect on the PE pre-contracted isolated rat aorta, but the inflorescence effect is the most powerful. According to these results, its vaso-relaxation activity may contribute to its antihypertensive properties. Applying a bio-guided fractionation, a bioactive molecule (1,4-androstadien-3,16-dione), responsible for its vaso-relaxation effect, was identified and isolated in this study. Bioactive molecules isolated from plants are considered lead compounds at the activity level detected in this study, and after structure-activity relationship studies, their chemical structures are usually modified to improve biological activity. Further identification of other bioactive compounds could facilitate chemical and biological quality control of the raw material used to produce future phytomedicines to be used by the local population.