Bioguided Identification of Active Antimicrobial Compounds from Asphodelus bento-rainhae and Asphodelus macrocarpus Root Tubers

Root tubers of Asphodelus bento-rainhae subsp. bento-rainhae (AbR), a vulnerable endemic species, and Asphodelus macrocarpus subsp. macrocarpus (AmR) have traditionally been used in Portugal to treat inflammatory and infectious skin disorders. The present study aims to evaluate the in vitro antimicrobial activity of crude 70% and 96% hydroethanolic extracts of both medicinal plants, specifically against multidrug-resistant skin-related pathogens, to identify the involved marker secondary metabolites and also to assess the pre-clinical toxicity of these medicinal plant extracts. Bioguided fractionation of the 70% hydroethanolic extracts of both species using solvents of increasing polarity, namely diethyl ether (DEE: AbR-1, AmR-1), ethyl acetate (AbR-2, AmR-2) and aqueous (AbR-3, AmR-3) fractions, enabled the identification of the DEE fractions as the most active against all the tested Gram-positive microorganisms (MIC: 16 to 1000 µg/mL). Furthermore, phytochemical analyses using TLC and LC-UV/DAD-ESI/MS techniques revealed the presence of anthracene derivatives as the main constituents of DEE fractions, and five known compounds, namely 7′-(chrysophanol-4-yl)-chrysophanol-10’-C-beta-D-xylopyranosyl-anthrone (p), 10,7′-bichrysophanol (q), chrysophanol (r), 10-(chrysophanol-7′-yl)-10-hydroxychrysophanol-9-anthrone (s) and asphodelin (t), were identified as the main marker compounds. All these compounds showed high antimicrobial activity, particularly against Staphylococcus epidermidis (MIC: 3.2 to 100 µg/mL). Importantly, no cytotoxicity against HepG2 and HaCaT cells (up to 125 µg/mL) for crude extracts of both species and genotoxicity (up to 5000 µg/mL, with and without metabolic activation) for AbR 96% hydroethanolic extract was detected using the MTT and Ames tests, respectively. Overall, the obtained results contribute to the concrete validation of the use of these medicinal plants as potential sources of antimicrobial agents in the treatment of skin diseases.


Introduction
Antimicrobial resistance (AMR) is a growing global healthcare problem due to the loss of efficacy of first-line antibiotics. Many pathogens are developing resistance to multiple drugs, making infections difficult or, in some cases, impossible to treat [1]. In response to the increasing demand for alternative medicines, the screening of natural products has emerged as one of the most successful methods for detecting/identifying antibacterial agents. Although in recent decades, the majority of new antibacterial drugs were from natural sources [2], only a small fraction of marine, fungal and plant resources have been investigated, and nature still offers a high potential for drug-lead discovery, notably among anti-infective compounds [3].
The Portuguese flora exhibits a considerable abundance of Asphodelus species, subspecies and varieties compared to the rest of Europe and the Mediterranean Basin. Besides the above-mentioned medical applications, root tubers are also used as daily food in the Iberian Peninsula, after being moistened and fried to eliminate the astringent compounds [6].
Asphodelus bento-rainhae subsp. bento-rainhae P. Silva is a vulnerable [23] endemic species from the Gardunha mountain range [24], located in the central region of Portugal, coexisting with Asphodelus macrocarpus subsp. macrocarpus Parlatore in the same geographical area. Both species are commonly known by the Portuguese name "abrotea", and their root tubers have traditionally been used for the treatment of skin diseases such as scabies, dermatophytosis and warts in Portugal.
The objective of this study is derived from the fact that although there are promising ethnomedical, phytochemical and biological data related to the Asphodelus species, to the best of our knowledge, no scientific studies on Asphodelus bento-rainhae and Asphodelus macrocarpus root tubers have been documented so far. Thus, the present study aims to establish the chemical profiles of the potential antimicrobial constituents together with the pre-clinical safety evaluation and validation of the use of the studied plants as herbal medicines.

Drug-Extract Ratio (DRE)
The drug-extract ratio was calculated as 1.9:1 and 5.5:1 for the A. bento-rainhae root tuber (AbR) and 2.7:1 and 6.7:1 for the A. macrocarpus root tuber (AmR) 70% and 96% hydroethanolic extracts, respectively. Considering these results, AbR exhibited a higher percentage of yield in both hydroethanolic extracts compared to the AmR extracts. Moreover, extraction with ethanol 96% noticeably reduced the percentage of yield in both species.

Phytochemical Screening and Antimicrobial Activity
Hyphenated analytical techniques were applied for the phytochemical dereplication of the samples. Following our previous study's results [25], the chromatographic profiles of AbR and AmR extracts showed excellent qualitative similarity in their chemical composition, characterized by the presence of terpenoids, phenolic acids and anthracene derivatives. Therefore, in continuation of the above-mentioned study searching for potent antimicrobial metabolites from these Portuguese Asphodelus species, liquid-liquid frac- tionations of both plant extracts with increasingly polar solvents, namely diethyl ether (AbR-1, AmR-1), ethyl acetate (AbR-2, AmR-2) and water (AbR-3, AmR-3), were performed. Both species' crude extracts and their subsequent L-L fractions were then submitted to in vitro antimicrobial evaluation in order to select the most active fractions for further phytochemical identification of their lead secondary metabolites.
The antimicrobial activity of the crude extracts, their derived L-L fractions and isolated compounds were evaluated through determination of the MIC values, an in vitro quantitative method of susceptibility testing against both selected Gram-positive and Gram-negative resistant pathogens.
As shown in Table 1 Considering the previously reported results of the antimicrobial activity of Asphodelus spp. root tuber crude extracts and in agreement with the obtained results verified in our species, weak to moderate activities against a similar pathogen panel with MIC values higher than 2000 µg/mL were observed [8]. The methanolic root extracts of A. luteus and A. microcarpus showed antimicrobial potential against methicillin-resistant S. aureus (MRSA), with MIC values of 650 to 1250 and 1250 to 2500 µg/mL, respectively [7]. Screening A. microcarpus tuber methanolic extract using an agar well diffusion assay revealed moderate activity against S. aureus, with an inhibition diameter zone of 14 mm [26]. Furthermore, the 80% hydromethanolic whole plant extract of A. tenuifolius was also found to be significantly active against S. aureus, E. coli, P. aeruginosa and K. pneumonia, with inhibition diameter zones of 16, 29, 18 and 18 mm, respectively, determined using the disc diffusion method [11,13].
Since anthracene derivatives are considered the main secondary metabolites of Asphodelus species, the detected and identified compounds could effectively be used for the chemotaxonomic classification of both A. bento-rainhae and A. macrocarpus species [14,37].

Antimicrobial Activity of the Major Marker Compounds and 96% Hydroethanolic Extracts of Both Asphodelus Root Tubers
The results for the antimicrobial activity of the five isolated major marker compounds of both diethyl ether L-L fractions (AbR-1, AmR-1) are presented in Table 2. Additionally, considering the chemical class and polarity of these compounds and in order to verify whether the activity of the total extract is relevant to the major or minor constituents, a less polar hydroethanolic extract (96%) of both species was also prepared and tested.  The AbR 96% hydroethanolic extract was found to be the most active crude extract and showed higher contents of marker metabolites in comparison to AmR 96% and both species' 70% hydroethanolic extracts, which is in accordance with the fundamental role of these compounds in the antimicrobial activity exhibited by these medicinal plants.
All the tested compounds were found to be active against all the tested Gram-positive strains, particularly against Staphylococcus epidermidis, with MIC values between 3.2 and 100 µg/mL (Table 2). Among these strains, teicoplanin-and linezolid-resistant S. epidermidis INSA958 showed the highest susceptibility to all the tested compounds. Moreover, chrysophanol, the major marker compound of both species, showed remarkable activity (MIC: 3.2 µg/mL) against this strain, which is often found on the human skin and mucous membrane; however, according to hospital surveillance reports, the bacterium is a common cause of nosocomial wound infections. Similar to the obtained results of the tested crude extracts, no activity regarding these compounds was found against the tested Gram-negative microorganisms in the tested range of concentrations (up to 200 µg/mL). To the best of our knowledge, no data related to the resistant strains employed in this study have been reported; however, chrysophanol and its derivatives have been previously reported to have potential antibacterial activity against other S. aureus strains (MIC values of 90 to 190 µg/mL) [42].
So far, there has not been enough research to explain the antibacterial mechanism of these compounds; however, according to the existing studies, the cell walls of Gram-positive bacteria, compared to Gram-negative bacteria, are more sensitive to many antibiotics and antimicrobial chemical compounds/herbal drugs [43]. The lipopolysaccharide layer and periplasmic space of Gram-negative bacteria are the reasons for the relative resistance of Gram-negative bacteria [44].

Evaluation of the Cytotoxicity Potential
The results of the in vitro cytotoxicity evaluations of the A. bento-rainhae (AbR) 70% and 96% hydroethanolic extracts are presented in Figure 3. The analysis of these data obtained through the cell viability assay clearly showed that none of the tested extracts induced cytotoxicity in HepG2 cells. However, since this medicinal plant is commonly used for the treatment of skin disorders, we further assessed cytotoxicity using a skin cell type. For this, the AbR 96% hydroethanolic extract, as the most active extract with the highest contents of marker secondary metabolites, and its major constituent, chrysophanol, were selected.

Evaluation of the Genotoxicity/Mutagenicity Potential
Although the negative results of the genotoxic/mutagenic potential of the root tuber 70% hydroethanolic extracts of both species were previously reported by the authors [25], as suggested by the guidelines [45], a genotoxicity assessment of different herbal preparations should be evaluated in order to reflect, as far as possible, the full spectrum of the As previously observed with HepG2, the AbR 96% extract did not reduce HaCaT viability, indicating its safe use through topical application at concentrations up to 125 µg/mL, but the same was not observed with chrysophanol, which reduced cell viability by up to 50% with concentrations higher than 25 µg/mL.

Evaluation of the Genotoxicity/Mutagenicity Potential
Although the negative results of the genotoxic/mutagenic potential of the root tuber 70% hydroethanolic extracts of both species were previously reported by the authors [25], as suggested by the guidelines [45], a genotoxicity assessment of different herbal preparations should be evaluated in order to reflect, as far as possible, the full spectrum of the extracted components. Additionally, since the AbR 96% hydroethanolic extract exhibited the highest antimicrobial activity and quantity of the active secondary metabolites, it was selected for further genotoxicity/mutagenicity evaluations.
The obtained results of the Ames test for the AbR 96% hydroethanolic extract are presented in Table 3. According to the genotoxicity guidelines [46,47], a mutagenic substance in the bacterial reverse mutation (Ames) test should exhibit a reproducible dose-related increase in the number of revertant colonies for at least one of the tester strains. Additionally, the number of revertant colonies must be more than twice the number of colonies produced on the negative (solvent) control plates. For cytotoxicity, a reduction in the number of revertants and/or clearing or diminution of the background lawn might be detected [48][49][50]. The analysis of the results showed that none of the tested concentrations of this extract (up to 5000 µg/plate) enhanced the number of revertant colonies in any tested strains with or without metabolic activation compared to the negative control. Moreover, toxicity did not occur, since none of the above-mentioned requirements occurred at any tested concentration. Therefore, under the conditions of this study, the mutagenic potential essential to ensure the safety of these extracts was not observed. Even though there are studies indicating the genotoxic potential of chrysophanol in the Ames test with metabolic activation (S9) [51], the obtained negative results of the tested AbR 96% hydroethanolic crude extract (with and without metabolic activation), show that the presence of chrysophanol does not influence the genotoxicity of the crude extract. Insufficient amounts of the mutagenic constituents and their interactions in the extracts/complex mixtures are among the various theories that could explain this phenomenon. Additionally, human exposure to chrysophanol and its derivatives through AbR 96% hydroethanolic extract is expected to be negligible, concerning the expected mode of administration (topical application), since they need to undergo bioactivation, mediated by different isoforms of cytochrome P 450, to become genotoxic [52].

Plant Materials
Root tubers of A. bento-rainhae (AbR) and A. macrocarpus (AmR) were collected from Serra da Gardunha, Portugal, during root dormancy in November 2019. The corresponding voucher specimens were deposited in the Laboratory of Pharmacognosy, Department of Pharmacy, Pharmacology and Health Technologies, Faculty of Pharmacy, Universidade de Lisboa (voucher specimens: OSilva_201901-A. bento rainhae and OSilva_201902-A. macrocarpus). The collected samples were dried in a well-ventilated dark space at room temperature. The authors' previous monographic study give a more detailed description of both species' botanical identification and sample selections [25].
High-performance liquid chromatography (HPLC) was carried out using a Waters Alliance 2690 Separations Module coupled with a Waters 996 photodiode array detector (UV/DAD) (Waters Corporation, Milford, MA, USA). Crude extracts (20 mg/mL) and L-L fractions (10 mg/mL) were initially solubilized in acetonitrile/water, and standard solutions (1 mg/mL) were prepared in acetonitrile and filtered through a polytetrafluoroethylene syringe filter (0.2 µm). An Atlantis RP-18 T3 column (5 µm, 150 × 4.6 mm) was used for the analysis of 25 µL of the injected samples with a flow rate of 1 mL/min. Water with 0.1% (v/v) formic acid (solvent A) and acetonitrile (solvent B) were used as the mobile phase, and gradients of 95% A: 5% B to 0% A: 100% B for a total run time of 75 min were used. Chromatograms were monitored and registered on Maxplot (wavelength 240-650 nm), and the obtained data were analyzed using Waters Millennium ® 32 Chromatography Manager Software (Waters Corporation, Milford, MA, USA).
Mass spectrometry (MS) analysis was conducted using the same HPLC equipment in tandem with a triple quadrupole mass spectrometer (Micromass ® Quatro Micro TM API, Waters ® , Drinagh, Ireland) using an electrospray ionization source (ESI) operating in both positive and negative mode. Data were acquired and analyzed using MassLynx™ V4.1 software (Waters ® , Drinagh, Ireland).

Isolation and Identification of the Main Marker Compounds
One gram of the active extract (AbR-1) was applied to the Sephadex LH-20 column. Several fractions were collected and concentrated through evaporation of the solvent. Then, the TLC control of the fractions was performed on silica gel 60 RP-18 plates using an H 2 O: MeOH (0.5:19.5, v/v) solvent system and screened under UV 254 and UV 366 . Fractions with similar profiles were mixed, and the collected fractions were bulked into six main fractions: AbR-1a (665 mg), AbR-1b (60 mg), AbR-1c (255 mg), AbR-1d (117 mg), AbR-1e (71 mg) and AbR-1f (29 mg). Compounds p, q, r, s and t were purified using a C18 reversed-phase silica gel column eluted with MeOH: H 2 O (90:10). The identification of compounds was based on co-chromatographic techniques and the obtained data related to the retention times, ultraviolet absorption and mass spectral characteristics recorded using LC-UV/DAD-ESI/MS, together with their TLC characteristics in comparison to those of standards and published data.

In Vitro Antimicrobial Activity
The broth microdilution method was used for an in vitro evaluation of the antibacterial potential [54], using 96-well tissue culture plates (VWR ® , Radnor, PA, USA) to determine the minimum inhibitory concentrations (MIC) of the tested samples against twelve reference (ATCC, LGC Standards S.L.U., Barcelona, Spain) and clinical strains (INSA clinical strains collection) of both Gram-positive (Staphylococcus aureus, S. epidermidis, S. saprophyticus, S. haemolyticus) ( Table 4) and Gram-negative (Escherichia coli, Klebsiella pneumoniae, Pseudomonas aeruginosa, Acinetobacter baumannii) multidrug-resistant bacteria.
The samples to be tested were initially prepared in water or DMSO 10%, and serial dilutions (2-2000 µg/mL for crude extracts/fractions and 0.2-200 for pure compound) were performed in a Mueller-Hinton medium and were distributed (50 µL) in each of the microplate wells using a microplate liquid handler (Precision TM BioTek, Winooski, VT, USA). Inoculums were prepared from a pure bacterial culture on agar, and suspensions with a turbidity of 0.5 for Gram-negative and 0.25 for Gram-positive bacteria on the McFarland scale (Grant Bio™ DEN-1B, Cambridge, UK) were prepared in Mueller-Hinton medium and stored at 4 • C until use. For MIC determination, the prepared suspensions were diluted at a ratio of 1:10, and 50 µL of this dilution was added to all the wells. To verify the absence of contamination and to check the viability of the inoculum, two controls were included for each tested sample, one plate in the absence of the extract solution and the other in the presence of the solvent (DMSO). As previously described, all experiments were carried out in triplicate to obtain consistent values.

In Vitro Cytotoxicity Evaluation using MTT Assay
In vitro cytotoxicity evaluation was performed using the methylthiazolyldiphenyltetrazolium bromide (MTT) reduction assay [55] on the human liver cell line HepG2 (ATCC, Cat. No. HB-8065, Manassas, VA, USA) and the human spontaneously immortalized keratinocyte cell line HaCaT (CLS, Cat. No. 300493, Eppelheim, Germany). HepG2 and HaCaT were inoculated at a density of 8.5 × 10 4 cells/cm 2 in α-MEM (Sigma-Aldrich ® , St. Louis, MO, USA) with 1 mM sodium pyruvate (PAN Biotech, Aidenbach, Germany), 1% non-essential amino acids (NEAA; PAN Biotech) and 10% fetal bovine serum (FBS, Gibco ® Thermo Fisher Scientific TM (Waltham, MA, USA), and of 4.0 × 10 4 cells/cm 2 in DMEM (Sigma-Aldrich ® ) with 4 g/L D-(+)-glucose (AppliChem, Darmstadt, Germany) and 10% fetal bovine serum (FBS, Gibco ® Thermo Fisher Scientific TM (Waltham, MA, USA), respectively. Both cell lines were maintained in a humidified chamber at 37 • C in a 5% CO 2 atmosphere. After 48 h, the medium was replaced with fresh medium with AbR 70% and 96% extracts and chrysophanol (9:1) at final concentrations of 25, 50, 75, 100 and 125 µg/mL for 48 h. Complete cell culture medium, DMSO 1% and DMSO 20% in α-MEM or DMEM were used as a positive, solvent and negative control, respectively. After cell washing with PBS, 200 µL 0.5 mg/mL MTT (Sigma-Aldrich ® ) was added to the cell culture medium. HepG2 and HaCaT were incubated for 3 h and 2 h, respectively, in a humidified chamber at 37 • C in a 5% CO 2 atmosphere. Next, 200 µL DMSO was used for solubilizing the purple crystals formed prior to measuring absorbance at 570 nm using a microplate spectrophotometer (SPECTROstar Omega; BMG LabTech, Ortengerg, Germany). The results are expressed as a percentage relative to the solvent control. Four wells were used for each sample, and at least two independent experiments were performed. Data analysis and graphs were plotted using GraphPad Prism ® software (version 9.0.0.121, GraphPad Software, San Diego, CA, USA). The results are presented as mean ± standard deviation. p < 0.05 was considered significant.

In Vitro Genotoxicity/Mutagenicity Evaluation using Ames Test
The screening of the genotoxicity potential was performed using a bacterial reverse mutation test (the Ames test) for the detection of genotoxic carcinogens and relevant genetic changes. The technique was conducted following the OECD No. 471 [46] and ICH S2 (R1) [47] guidelines as well as the published reference protocols [56,57].
Salmonella enterica serovar Typhimurium tester strains (TA98, TA100, TA102, TA1535 and TA1537) were used in this study (with and without metabolic activation) in a direct plate incorporation method. TA100, TA98, TA102 and TA1535 were kindly provided by the Genetic Department of the Nova Medical School of the Universidade NOVA de Lisboa (Portugal), having received them from Professor B.N. Ames (Berkeley, CA, USA). TA1537 was obtained from ATCC, NUMBER: 29630™, LOT: 7405375.
The AbR 96% hydroethanolic extract (50 mg/mL) was initially dissolved in DMSO, and 100 µL of the extract dilutions was mixed with 500 µL sodium phosphate buffer (0.1 M, pH 7.4) (in the assay without metabolic activation) or S9 mix (in the assay with metabolic activation). Then, 100 µL of the bacterial culture and 2 mL of melted top-agar, supplemented with 0.05 mM biotin and histidine, were added to the mixture. After a 48 h incubation at 37 • C, manual counting of His+ revertant colonies for each concentration was performed. The results are expressed as the mean number of revertant colonies with the standard deviation (mean ± SD). The positive controls were sodium azide (SA, 1.5 µg/plate for TA100 and TA1535), 2-nitrofluorene (2-NF, 5 µg/plate for TA98), 9-aminoacridine (9-AA, 100 µg/plate for TA1537) and tert-butyl hydroperoxide (tBHP, 50 µg/plate for TA102) in the assay without metabolic activation, and 2-aminoathracene (2-AA, 2 µg/plate for TA98 and 10 µg/plate for TA102, TA1535 and TA1537) and benzo(a)pyrene (BaP, 5 µg/plate for TA100) in the assay with metabolic activation. All assays were performed in triplicate to obtain consistent values.

Conclusions
Overall, the observed antimicrobial activity of both the A. bento-rainhae and A. macrocarpus root tuber 70% hydroethanolic extracts were similar to those obtained and reported from the other Asphodelus spp. tested against a similar panel of pathogens. However, the fractionation of these extracts and an enriched 96% hydroethanolic extract certainly enhanced their significant antimicrobial activity, as they contain the highest amounts of 1,8-dihydroxy anthracene derivatives, a known chemical class of secondary metabolites with potential antimicrobial activity.
Moreover, the isolated and identified chrysophanol derivatives could be considered important chemotaxonomic markers of both studied Asphodelus species.