Exploring the anticancer and antioxidant properties of Vicia faba L. pods extracts, a promising source of nutraceuticals

Chemicals and reagents

Analytical grade acetone, methanol, ethanol and water (VWR International s.r.l., Milan, Italy) were used for the extraction of phenolic and flavonoid compounds. Trolox (6-hydroxy-2,5,7,8-tetramethilchroman-2-carboxylic acid, 97%), DPPH (2,2-Diphenyl-1-picrylhydrazyl, 95%), ABTS (2,2′-azino-bis-(3-ethylbenzothiazoline-6-sulphonic acid) diammonium salt, potassium persulphate, potassium acetate, aluminium chloride and Folin-Ciocâlteu reagent were obtained from Sigma-Aldrich (Milan, Italy).

Analytical standard grade vanillic acid (97%), ellagic acid (97%), p-coumaric acid (98%), ferulic acid (99%), quercetin (98%), gallic acid (98%) were also purchased from Sigma-Aldrich (Milan, Italy). HPLC gradient grade methanol and water (VWR International s.r.l., Milan, Italy) was used as chromatographic mobile phase. Formic acid (99%; Sigma Aldrich, Milan, Italy) was used to acidify the aqueous mobile phase, prepared with high-purity water. All samples were filtered through a syringe filters in PVDF, pore size 0.45 µm (Millipore, SER.DIA s.r.l., Reggio Calabria, Italy).

Plant material

The Vicia faba L. used were grown at a locality of Acri (39°29′38″04 Nord and 16°23′4″56 Est, Calabria, Southern Italy). The harvest was done in the period from April to May 2018 three times from the same soil at a time interval of 20 days from each other. The seeds were removed and the pods used were named according to the harvest period as VFI (first collection, 15–18 cm length, 4–5 small seeds inside), VFII (second collection, 20–25 cm length, 7–8 intermediate seeds inside), and VFIII (third collection, 25–30 cm length, 9–10 large seeds inside). The pods were cut into small pieces with a knife and freeze dried (Telstar freeze-dryer, mod. Cryodos), ground to a fine powder using a 60-mesh screen (particle size equal to 250 μm) and stored at −20 °C, before the extraction.

Preparation of the extracts

The flour was first extracted with acetone, then with methanol and finally with 70% aqueous ethanol. These solvents were used in the preliminary stage of the β-glucan isolation procedure in order to remove soluble compounds from the flour (Fazio et al., 2020). In particular, the use of these specific solvents was required as per protocol to remove lipids and soluble materials such as vitamins, polyphenols and others and then to facilitate the complete separation of fiber from other compounds that may be present. Dry matter (3 g) was weighed in 50 mL capped plastic tubes and 30 mL (g 10 mL⁻¹) of acetone were added. The suspension was left under stirring for 1 h at room temperature, then separated from the solid residue by centrifugation at 5,000 rpm for 15 min. The supernatant was filtered using sintered glass Buchner funnel (4 µm pore size) and collected in a previously weighed flask. The extraction was repeated three times on the resulting solid and filtrates combined. The solid residue from these extractions was extracted with methanol (3 × 30 mL) and centrifugated each time. The last extractions were performed on the solid residue using 70% aqueous ethanol (3 × 30 mL). The organic solvent from the three filtrates (acetone, methanol, and 70% aqueous ethanol) was separately removed under reduced pression. The crude extracts were conserved at −20 °C under N₂. The weight of each extract was expressed as the average value ± standard deviation.

Total phenolic content (Folin-Ciocâlteu assay)

The total phenolic content (TPC) was determined according to the modified Folin-Ciocâlteu method adapted by Plastina, Fazio & Gabriele (2012).

Briefly, 0.1 mL of the sample (1 mg mL⁻¹DMSO) were mixed with 0.5 mL of 1:10 diluted Folin–Ciocâlteu reagent and 2 mL of distilled water. The mixture was shaken for 4 min and then 0.4 mL of a 10% sodium carbonate solution (Na₂CO₃), were added and the mixture was stirred in the dark at room temperature. After 30 min under magnetic stirring the absorbance was measured at 765 nm using UV-Vis spectrophotometer (Ultrospec 2100 Pro; Amersham Biosciences/GE Healthcare, Amersham, UK), using an appropriate blank for background subtraction. Gallic acid standard solutions at five different concentrations in the range 1–600 µg mL⁻¹DMSO were prepared and used to construct a calibration curve (r² = 1; y = 3.394x − 0.0179).

The TPC in each extract was the mean of three assays expressed as mg of gallic acid equivalents per gram dry extract (mg GAE g⁻¹ DW).

Total flavonoid content

The total flavonoid content (TFC) was determined using a colorimetric method (Fazio et al., 2018). For each extract, 0.3 mL of the sample (1 mg mL⁻¹DMSO) were dissolved in 0.9 mL of MeOH. Then, 0.06 mL of 10% AlCl₃ solution, 0.06 mL of 1 M solution of potassium acetate and 1.68 mL of distilled water were added. After 30 min of incubation in the dark, under constant magnetic stirring at room temperature, the absorbance was measured at 420 nm using a UV-Vis spectrophotometer (Ultrospec 2100 Pro; Amersham Biosciences/GE Healthcare, Amersham, UK) against a blank. Three analyses were carried out for each sample. Quercetin standard solutions within the 12.5–100 µg mL⁻¹ DMSO range concentrations were prepared to construct a calibration curve (r² = 0.9993; y = 0.0073x + 0.001). The TFC was expressed as µg of quercetin equivalents per gram of dry extract (µg QE g⁻¹ DW).

Radical scavenging activity

Antioxidant activity of the ethanolic, methanolic and aqueous extracts from leaves was determined by DPPH and ABTS assays.

Qualitative and quantitative analysis of the phenolic and flavonoid compounds by HPLC-DAD

The quantitative analysis of the extracts was performed by high performance liquid chromatography (Shimadzu, Kyoto, Japan) equipped with Auto Sampler (SIL 20A), pumps (LC 20AD), oven (CTO-10AS vp), diode array-detector (SPD M20A) and a system controller CMB-20A. Mediterranea SEA (Terrassa, Spain) C-18 column (4.6 mm × 25 cm, 5 µm) was used for the analysis. The extracts were filtered through membrane filter (0.45 µm) into HPLC amber vials. The mobile phase was a mixture of water containing 1% of formic acid (A) and MeOH (B). All the analyses were performed in gradient elution mode as follows: 0 min, 10% B; 0–10 min, 20% B; 10–20 min, 30% B; 20–30 min, 40% B, 30–40 min, 50% B, 40–50 min, 60% B, 50–60 min, 70% B, 60–70 min, 80%, 70–80 min, 90% B, 80–90 min, 100% B, 90–120 min, 100% B, 120–125 min, 10% B. The flow rate and column temperature were 0.6 mL min⁻¹ and at 29 °C, respectively.

The qualitative analysis was performed compounds by comparing the chromatograms of the extracts with the reference compound ones. In addition, in order to assign the elution time of each compound, analyses were conducted on each extract that was enriched with the standards one by one. The individual polyphenolic compounds were quantified by the external standard method using the calibration curves of the following commercial standards: vanillic acid, ellagic acid, quercetin, p-coumaric acid, ferulic acid, chlorogenic acid and gallic acid which were detected at appropriate wavelengths (280, 254, 365, 271, 325, 327 and 276 nm, respectively).

Spectroscopic profiling via FTIR-ATR analysis

The infrared fingerprints of Vicia faba L. extracts were collected by using the Spectrum Two Fourier transform infrared (FTIR) spectrometer (Perkin Elmer Italia Spa, Milan, Italy), coupled with an attenuated total reflection (ATR) accessory consisting of a flat top-plate fitted with a ZnSe crystal. The ATR detector system was wiped before each analysis by using dry paper and methanol. The lab room air FTIR-ATR spectrum was used as background to verify the cleanliness, instrumental conditions, and interferences due to H₂O and CO₂. FTIR spectra of the samples, laid on the ATR top, were recorded in triplicate in the range 4,000–450 cm⁻¹. Scan numbers and resolution were optimized at 32 scans and 4 cm⁻¹, respectively (De Luca et al., 2016). A first selection of the range was necessary before data handling, variables matrix was cut keeping out the spectral region between 4,000 and 1,800 cm⁻¹. The range 4,000–2,500 cm⁻¹ contained mainly broad signals associated with the stretching vibrations of hydroxyl functions and the stretching vibrations of aromatic C-H that were poor in information about phenolic compounds. The range 2,500–1,800 cm⁻¹ did not contain spectral relevant information too, apart from the band corresponding to CO₂. Therefore, the remaining range 1,800–450 cm−¹ was arranged in the matrix and subdued to multivariate data modelling (Abbas et al., 2017).

Multivariate data analysis

Principal component analysis (PCA) allows to get the information available in the spectral fingerprint by projecting the samples and spectral variables on a set of new orthogonal variables called principal components (PCs). PCA was performed in order to estimate the metabolic differences in terms of phenolic compounds from the different fava bean samples and extracts (Marrelli et al., 2020). Original data consisting of the Fava bean extracts were arranged in a matrix where 27 samples were described by their FTIR-ATR spectra. PCA analysis was performed by using the SVD (Singular Value Decomposition) algorithm and the full cross-validation procedure used in order to select the optimal number of PCs. The Unscrambler X software version 10.5 from CAMO (Computer Aided Modelling, Trondheim, Norway) was used for the chemometric treatment of the spectral data (Spatari et al., 2017).

Cell culture

All the cell lines used in this work were purchased from American Type Culture Collection (ATCC, Manassas, VA, USA). They were maintained at 37 °C in a humidified atmosphere of 95% air and 5% CO₂ and periodically screened for contamination as already reported (Barbarossa et al., 2021; Ceramella et al., 2019; Iacopetta et al., 2020a).

MTT assay

The in vitro antitumor activities of all the target compounds were determined by using the MTT (Sigma) assay, as already described (Sinicropi et al., 2018). Cells were exposed to the different extracts dissolved in DMSO at different concentrations (50, 100, 200, 400 and 500 µg/mL) for 72 h. Then MTT was added as already reported (Ceramella et al., 2021). The IC₅₀ values were obtained by using GraphPad Prism 9 (GraphPad Software, La Jolla, CA, USA). Data are representative of three independent experiments performed in triplicate; standard deviations (SD) have been shown.

TUNEL assay

Apoptosis was detected by the TUNEL assay, according to the guidelines of the manufacturer (CF^TM488A TUNEL Assay Apoptosis Detection Kit; Biotium, Hayward, CA, USA), with some modifications. The cells were grown on glass coverslips and, after treatment with the most active extracts at their IC₅₀ values, they were washed and methanol-fixed. The TUNEL reaction mixture containing the terminal deoxynucleotidyl transferase (TdT) were added, as already reported by Iacopetta et al. (2017). Samples were washed, incubated with DAPI (Sigma, 0.2 mg/mL) and then observed and imaged under a fluorescence microscope (Leica DM6000; 20× magnification, λ_ex/em maxima of 490/515 nm for CF^TM488A or 350/460 nm for DAPI). Images are representative of three independent experiments.

Immunofluorescence analysis

Cells were seeded in 48-well culture plates containing glass slides and then serum-deprived for 24 h and incubated with the most active extracts for 24 h (concentration equal to their IC₅₀ value). The cells were incubated with primary antibody, as previously described (Ceramella et al., 2019; Sinicropi et al., 2021). The rabbit anti β-Tubulin were purchased from Santa Cruz Biotechnology and diluted 1:100 before the use. The secondary antibody used was Alexa Fluor® 546 conjugate goat-anti-mouse (1:500; Thermo Fisher Scientific, MA, USA). Leica DM 6000, 40× magnification was used in order to detect fluorescence and the images were processed using LAS-X software. All the experiments were performed in triplicate.

Docking studies

The crystal structure Vinblastine in complex with a tetrameric assembly of Tubulin (Waight et al., 2016) (PDB code 5j2t), has been used as a protein target for our simulations. The atomic coordinates of the ligands we tested in silico, have been built and energy minimized using the program MarvinSketch (ChemAxon ltd, Budapest, Hungary). To better understand the binding modes to the protein target, the mechanisms of action of our molecules and to determine their binding energies, we used the program Autodock v.4.2.2. (Morris et al., 2009) and its graphical interface ADT (Sanner et al., 1999). We built a cubic searching grid composed by 126 nodes in each dimension, and we centred it on the protein centre. Nodes are spaced 0.375 Å each other. We run our simulations adopting the program default values. The protein target and the ligands were prepared using the ADT graphical interface: polar hydrogens were added, Kollman charged assigned to the ligands and eventually solvatation parameters added. Tubulin has been considered as a rigid object while all our ligands were taken as fully flexible. Our simulation was carried out as already reported (Iacopetta et al., 2020b; Lappano et al., 2015; Rosano et al., 2012; Viale et al., 2011).

Statistical analysis

The experimental data regarding were expressed as means of three replications ± standard deviation. Statistical analyses were performed using GraphPad Prism 8.0.2 software and evaluated by two-way ANOVA followed by a multi-comparison Turkey test, and by multi-comparison Dunnett’s test to evaluate the significance of EC₅₀ values against DPPH and ABTS. Significance was established at p values (*) < 0.05, (**) < 0.01, (***) < 0.001, and (****) < 0.0001.

Recovery of the extracts from dry powder

The water content (%) and the yield of dry matter (g 100 g⁻¹) were shown in Fig. 1. The yield (%) of each extract for all collections was reported in Fig. 2.

Data reported in Fig. 2 showed that the acetone extracts have a low yield in all three stages of collection (2.2–3.3%), but with greater recovery from VFII (VFII vs. VFII, VFIII *** p < 0.001). The major quantitative recovery occurs for methanol extracts (23.7–31.8%), with significance ****p < 0.0001 compared to acetone and ethanolic extracts (VFa, VFe) from each collection and also to methanolic extracts from the three stages of collection (VFIm, VFIIm, VFIIIm), which means that methanol has higher extraction efficiency despite its dielectric constant, being lower than that of 70% EtOH. The latter has an intermediate extraction efficiency compared to the other two solvents (6.5–10.9%), but significantly different (****p < 0.0001) in the ethanolic extract from the three harvests. Likely, as suggested from the Folin-Ciocâlteu assay, it has affinity for compounds that are not necessarily phenolic.

The quantitative recoveries from acetone and methanol extracts are more abundant in the medium stage of maturation (33 mg and 318 mg respectively) than from the ethanolic aqueous extract, for which the highest recovery is observed in the third collection (108.8 mg).

Total phenolic content (TPC) by Folin-Ciocâlteu assay

The total phenolic content (TPC) in each extract was reported in Fig. 3 and was a function of the extraction solvent and collection time, decreasing with the maturation in ethanolic extract from 224.2 ± 16.6 (first collection) to 124.7 µg ± 0.7 GAE g⁻¹_extract (third collection, VFI vs.VFII and VFIII ****p < 0.0001, VFII vs VFIII *p < 0.05) while increasing in acetone extract from 92.7 ± 7.6 µg GAE g⁻¹_extract in the first collection to 238.6 ± 11.9 µg GAE g⁻¹_extract in the second collection (****p < 0.0001). In the third collection it was halved compared to the content of stage I (****p < 0.0001).

In methanol, the initial content (181.7 ± 4.7 µg GAE g⁻¹_extract) was reduced by about one third the second collection (65.9 ± 1.1 µg GAE g⁻¹_extract, ****p < 0.0001) and, although without significance, it increased by 7% compared to the content of stage II (70.9 ± 0.8 µg GAE g⁻¹_extract).

The higher phenolic content was found in the acetone extract of the second collection (238.6 ± 11.9 µg GAE g⁻¹_extract), followed by the ethanolic extract of the first collection (224.2 ± 16.6 µg GAE g⁻¹_extract). The minor content of polyphenols was found in the acetone extract from VFIII (41.2 ± 0.8 µg GAE g⁻¹_extract).

The samples showed a progressive increase in the phenolic content with increasing polarity of the extraction solvent, apart from TPC in the second collection that is maximum in the acetone extract (238.6 ± 11.9 µg GAE g⁻¹_extract) despite the lower polarity of the solvent (ε = 20.7).

Total flavonoid content (TFC)

Figure 4 showed the TFC in all the extracts. Similarly to TPC, the TFC varied as a function of the extraction solvent and collection time. Flavonoid content in acetone extracts decreased as the ripening progresses (VFI vs. VFII ***p < 0.001, VFI vs VFIII ****p < 0.000, VFII vs VFIII ns), while TFC in ethanolic extracts increased from 4.9 ± 1.8 µg QE g⁻¹_extract in the first collection to 5.3 ± 1.4 µg QE g⁻¹_extract in the third collection, without any significance. In methanolic extracts, the initial content (8.3 ± 0.6 µg QE g⁻¹_extract) remained unchanged in the second collection (8.4 ± 0.1µg QE g⁻¹_extract) and then decreased in the last collection time (5.3 ± 0.8 µg QE g⁻¹_extract).

In all three harvesting, the TFC decreased with the increasing polarity and as the ripening progressed. The solvent with the best extraction efficacy against flavonoids was found to be acetone.

Radical scavenging activity

Two different in vitro assays, DPPH and ABTS, were used to evaluate the changes in free-radical scavenging abilities of all extracts from VFI, VFII and VFIII.

DPPH assay

DPPH radical was scavenged by aqueous ethanol, methanol and acetone extracts from the three collections in a concentration-dependent manner (Fig. 5).

All the extracts from VFI exhibited the greatest antioxidant activity, exceeding 90% of DPPH inhibition (from 93.83% to 91.59%, that is 228.0 ± 2.0 µg TE g⁻¹_extract to 220.8 ± 1.0 µg TE g⁻¹_extract) at the highest concentration tested (33.3 µg mL⁻¹). The results obtained from the VFI sample can be correlated to the content of phenols and flavonoids. At the same concentration, only the ethanolic extract from VFII inhibited DPPH radical by 90.6 ± 0.4% (218.2 ± 1.2 µg TE g⁻¹_extract). The extracts from VFIII exhibited the lowest activity at all concentrations: at 33.3 µg mL⁻¹, aqueous ethanolic, methanolic and acetone extracts inhibited DPPH radical by 63.8 ± 1.3% (142.7 ± 3.7 µg TE g⁻¹_extract), 46.2 ± 0.7% (124.6 ± 4.5 µg TE g⁻¹_extract) and 45.7 ± 0.7% (173.4 ± 1.5 µg TE g⁻¹_extract), respectively.

The EC₅₀ values of aqueous ethanolic, methanolic and acetone extracts from VFI were found to be 3.1 ± 0.5, 2.5 ± 0.4 and 7.9 ± 0. 8 µg mL⁻¹, respectively, indicating that the inhibition capacity of methanol against DPPH is comparable with the Trolox one, whereas that of acetone and ethanol is significantly lower than Trolox one (****p < 0.0001 and *p < 0.05, respectively)(Table 1). The EC₅₀ values of all extracts from VFII and VFII are significantly higher than the Trolox one (****p < 0.0001 and ***p < 0.001 for VFIIe).

Table 1:

EC₅₀ values. EC₅₀ values of the extracts against DPPH radical (expressed as µg mL⁻¹).

Samples	VFI	VFII	VFIII
Acetone	7.9 ± 0.8****	10.9 ± 1.3****	39.6 ± 1.6****
MeOH	2.5 ± 0.4	18.2 ± 1.0****	35.1 ± 1.5****
70% EtOH	3.1 ± 0.5*	4.1 ± 0.6***	21.5 ± 1.2****

DOI: 10.7717/peerj.13683/table-1

Notes:

The asterisks on the numbers indicate that means values were statistically different from the Trolox (EC₅₀ = 0.84 ± 0.3).

**** p < 0.0001.

*** p < 0.001.

* p < 0.05.

ABTS assay

ABTS radical was scavenged by aqueous ethanol, methanol and acetone extracts from the three collections in a concentration-dependent manner (Fig. 6). However, all the extracts exhibited a lower scavenging activity toward the radical cation ABTS•⁺ than toward the radical DPPH.

Aqueous ethanolic and methanolic extracts from VFI inhibited the radical cation ABTS•⁺ by 89.3 ± 1.3 (6.7 ± 0.3 µg TE g⁻¹_extract) and 72.7 ± 0.2% (5.45 ± 0.1 µg TE g⁻¹_extract), respectively, at the highest concentration tested (10 µg mL⁻¹). At the same concentration, acetone extracts from VFI showed the lowest percentage of radical inhibition (29.6 ± 0.6%, 2.1 ± 0.1 µg TE g⁻¹_extract). The EC₅₀ values of aqueous ethanolic, methanolic and acetone extracts from VFI were found to be 3.8 ± 0.6, 6.8 ± 0.8, and 24.4 ± 1.4 µg mL⁻¹, respectively, indicating that the inhibition capacity of 70% ethanol against ABTS•⁺ is comparable with the Trolox one, whereas that of acetone and ethanol is significantly lower than Trolox one (****p < 0.0001 and *** p< 0.001, respectively). The EC₅₀ values of all extracts from VFII and VFII are significantly higher than the Trolox one (****p < 0.0001) (Table 2).

Table 2:

EC₅₀values of the extracts against ABTS radical. EC₅₀values of the extracts against ABTS radical (expressed as µg mL⁻¹).

Samples	VFI	VFII	VFIII
Acetone	24.4 ± 1.4****	40.7 ± 1.6****	70.8 ± 1.8****
MeOH	6.8 ± 0.8***	47.1 ± 1.7****	38.6 ± 1.6****
70%EtOH	3.8 ± 0.6	12.3 ± 1.1****	12.9 ± 1.1****

DOI: 10.7717/peerj.13683/table-2

Notes:

The asterisks on the numbers indicate that means values were statistically different from the Trolox (EC₅₀= 1.98 ± 0.3).

**** p < 0.0001.

*** p < 0.001.

Qualitative and quantitative analysis of the phenolic and flavonoid compounds by HPLC-DAD

Seven compounds were identified and quantified by HPLC analyses in all samples (Fig. 7). The best quantitative and qualitative profile of polyphenols was recorded for VFI extracts, while the poorest extracts were from VFIII, except for the methanolic extracts from all three collections that contain the compounds identified in unchanged amount. In the methanolic and acetone extracts from VFI, seven compounds were identified among the standards, and five phenolics in the aqueous ethanolic extract (Table 3). The main compound in all the extracts was p-coumaric acid (21.9 ± 0.8–27.4 ± 5.1 µg g⁻¹). Ellagic acid was contained in small amount, in all tested extracts (0.8 ± 0.1–1.3 ± 0.1 µg g⁻¹). Chlorogenic acid was present only in methanolic extracts from all collections and in acetone extract from VFI. Quercetin and ferulic acid remained constant (about 4 and 19 µg g⁻¹, respectively) in all the extracts except for the EtOH 70% extract from VFII and for acetone and EtOH 70% from VFIII, in which they were lacking. A heat-map was shown (Fig. 8) to visualize the contents of phenolic compounds and to highlight their distribution in the acetone, methanol and ethanolic aqueous extracts from the three different collections.

Principal Component Analysis (PCA) to infrared fingerprint characterization

Figure 9 shows the Vicia faba L. extracts spectra recorded from extraction procedures, where the common spectral bands associated with aromatic rings and phenol moieties are visible. The methoxy group present in some phenolic compounds showed signals between 1,450 and 950 cm⁻¹. The phenolic acids were characterized by bands of unsaturated carboxylic acids (range 1,775–1,630 cm⁻¹), while flavonoids moiety mainly by bands associated with benzopyrylium and benzo-γ-pyrone vibrations in the regions 1,650–1,400 and 1,200–450 cm⁻¹. Phenolic compounds have a large and overlapped infrared fingerprint. The use of single peaks or limited wavenumber ranges to distinguish the different extraction procedures or ripening times seemed very difficult, making it necessary to use chemometric methodologies (Abbas et al., 2017). PCA elaboration was applied to the spectral data matrix, with the aim to discriminate the different metabolic contribution of Vicia faba L. bean samples. Figure 10 describes the plot scores by considering the first and second principal components with a total % explained variance of 95%. The sample grouping was evident and it was possible to distinguish the samples according to the extraction procedure and in some cases also the harvest time of the VFs. By comparing the X-loading values calculated for the elaborated waves with respect to the first two principal components (PCs), some characteristic peaks for each extraction technique were highlighted (Fig. 10). The correspondence between PCA and spectral fingerprint is shown in Fig. 9. PCA showed a relative abundance of information for the acetone extract in the spectral range 1,740–1,690 cm⁻¹, and these characteristic bands can be explained when hydroxybenzoic and hydroxycinnamic acids are present in the samples. The MeOH extract was characterized by the bands in the range 1,100–950 cm⁻¹, and the EtOH 70% extract in the ranges 1,590–1,550 and 530–510 cm⁻¹, confirming the differences in phenols composition according to the chromatographic study.

Effects on cancer cells viability of V.faba L. pods extracts

The cytotoxic effects of the V.faba L. pods extracts, at three different harvest periods, were examined against two human melanoma cell lines, A2058 and Sk-Mel-28, two human breast carcinoma cells, MCF-7 and MDA-MB-231, two uterine carcinoma cell lines, HeLa and Ishikawa, and the human colon carcinoma RKO cells using MTT assay. As normal cell lines, we adopted the human mammary epithelial cells, MCF-10A, and the human embryonic kidney epithelial cells, Hek-293. The IC₅₀ values were determined after 72 h of treatment at different concentrations (50, 100, 200, 400 and 500 μg mL⁻¹) of the acetone, methanol and ethanol VFs extracts and are reported in Table 4, together with those from Vinblastine, a well-known Vinca alkaloid used as reference molecule. The acetone extracts (VFIa, VFIIa and VFIIIa) resulted cytotoxic for all the cell lines used in these experiments, with few exceptions (Table 4). However, the lack of selectivity between cancer and normal cells make these extracts useless for further investigations. The three methanol and ethanol extracts exhibited, instead, a better cytotoxic profile and a selectivity between the cancer and normal cells. The VFIm extract was found to be particularly active toward the Sk-Mel-28 cells viability (IC₅₀ of 120.8 ± 1.2 μg mL⁻¹), with a lesser cytotoxicity against MCF-7 (IC₅₀ of 392.8 ± 0.6 μg mL⁻¹) and MDA-MB-231 (IC₅₀ of 467.3 ± 0.9 μg mL⁻¹) cells and lacked any effects on the other cell lines used, included the normal cells, MCF-10a and Hek-293. The VFIIm extract maintained a good anticancer activity against the Sk-Mel-28 cells, with an IC₅₀ value of 170.9 ± 1.0 μg mL⁻¹, but lost the activity against the breast cancer cells and resulted inactive toward all the other cells, included the normal ones. Finally, the VFIIIm extract exhibited a lesser activity against the Sk-Mel-28 cells (IC₅₀ of 256.1 ± 0.8 μg mL⁻¹) with respect to VFIIm and VFIm and gained a discrete activity against the MCF-7 cells (IC₅₀ of 330.8 ± 1.3 μg mL⁻¹); no cytotoxic activity was recorded against the other cell lines. The VFIe extract had the best activity against the Sk-Mel-28 cells, with an IC₅₀ value of 118.7 ± 0.5 μg mL⁻¹, a very low activity against the breast cancer cells (444.6 ± 1.2 and 403.0 ± 0.7 μg mL⁻¹ for MCF-7 and MDA-MB-231 cells, respectively) and a discrete activity against the HeLa cells, being the only one active against this cancer cells line and showing no effects on the normal cells viability and on the other cancer cells lines used. The VFIIe showed an IC₅₀ value of 175.1 ± 0.7 μg mL⁻¹ against the Sk-Mel-28 cells, with a cytotoxic profile comparable to that of the VFIIm extract. Finally, the VFIIIe extract was found active only against the Sk-Mel-28 and MCF-7 cells, with IC₅₀ values of 287.9 ± 1.3 and 326.7 ± 0.9 μg mL⁻¹, respectively, similarly to VFIIIm. Summing up, the VF acetone extracts are not selective between cancer and normal cells, whereas the VF methanol and ethanol extracts affect particularly the Sk-Mel-28 cells viability, with very close IC₅₀ values and following this trend in terms of activity: VFIe > VFIm > VFIIm > VFIIe > VFIIIm > VFIIIe. When compared to our extracts, Vinblastine exerted a stronger antitumor activity against all the used cancer cell lines, but accompanied by a dramatic cytotoxicity towards the normal cell lines (IC₅₀= 17.2 × 10⁻³ ± 0.6 and 33.5 × 10⁻³ ± 1.0 μg mL⁻¹ on MCF-10A and Hek-293, respectively). The VFIe extract is, overall, the most active of the series and is the only one that exhibited a discrete activity as well against the HeLa cells. Moreover, both the ethanol and methanol extracts did not affect the viability of the MCF-10A and Hek-293 normal cells.

Cells morphology changes and death by apoptosis

Sk-Mel-28 cancer cells were used for the next investigations, being particularly sensitive to the VFIe and VFIm. During the viability assays, we noticed a dramatic cell morphology change and an elevated number of floating cells, suggesting that they were dying (Fig. 11, VFIm and VFIe), whereas the vehicle-treated cells were adherent and possessed a normal polygonal morphology (Fig. 11, CTRL).

In order to characterize the cell death at the molecular level, we performed a TUNEL assay after 24 h of cells exposure to VFIe and VFIm extracts, used at their IC₅₀ values. Figure 12 shows that in the VFIe and VFIm extracts-treated cells is visible a green fluorescence localized in the cell nuclei (VFIm and VFIe, Fig. 12B) and not detectable in the vehicle-treated cells (Fig. 12B, CTRL), indicating DNA fragmentation and apoptosis.

Effects of methanol and ethanol V.faba L. pods extracts on cell microtubules

The morphological cells changes observed and the detected cell death by apoptosis pushed us to further investigate the mechanism(s) by which our extracts could triggers these effects. Considered the earlier observation, we carried out immunofluorescence assays on tubulin of the Sk-Mel-28 cells exposed to VFIe and VFIm extracts or only vehicle (negative control) for 24 h. As positive control we adopted one of the most know drugs targeting the tubulin, i.e. the vinblastine, able to inhibit the polymerization reaction. The obtained results are provided in Fig. 13, where the vehicle treated cells showed a normal tubulin organization, the microtubules are fairly distributed throughout the cell cytoplasm and sharply defined (CTRL, panel A). Conversely, under the vinblastine treatment (Fig. 13B, V) the microtubules organization seemed completely altered: indeed, there is a brighter fluorescence unevenly gathered around the cells nuclei, with a dot-like aspect due to the crystals and para-crystals formation. As well for the VFIe and VFIm treated cells, a similar result has been obtained and the microtubules resulted packed and dotted near the nuclei (Fig. 13A, VFIm and VFIe). Finally, we can deduce that one of the intracellular targets of the two above-mentioned extracts are the microtubules, which structural alteration results in a dysfunctional cell cytoskeleton responsible of the observed apoptotic death.

Docking simulations

We used molecular docking simulations to evaluate the possible binding modes of our phenolic compounds identified in Vicia faba L. extracts by HPLC and to calculate their binding affinities to a tetrameric assembly of tubulin. We initially performed a blind-docking using the program Vina (Trott & Olson, 2010) in order to identify the binding sites and then, through the program Autodock 4.2.2 (Morris et al., 2009) we calculated the affinities of the compounds to the protein target, in our case we used the three-dimensional structure of tubulin in complex with Vinblastine (Waight et al., 2016). The protocol we adopted for the simulations is the one we discussed in several other previous works (Abbassi et al., 2014; Rosano et al., 2016; Sinicropi et al., 2018) that allows us to identify the best binding mode basing on the clusterization of the results of our simulations together with the visual inspection of the binding area.

In this case, we identified three main binding zones, as shown in Fig. 14A. A first binding site (shown as an orange transparent surface) is occupied by ellagic acid and it is superposed to the binding site of a GDP molecule in the three-dimensional structure of tubulin in complex with vinblastine. A second area (purple transparent surface) is correspondent to the vinblastine binding cleft and, in our simulations, is occupied by the chlorogenic acid moiety. A third binding zone (green-violet surface) is the binding zone of the other compounds we tested (quercetin, and vanillic, gallic, ferulic and p-Coumaric Acids). Figure 14B–14H and Table 5 show the interactions between all the above-mentioned ligands and the protein amino acids in deeper details. Table 5 also reports the binding energies and the calculated affinity constant for each compound.

Taking into consideration the analysis of the clusterization of the results and the visual inspection of the molecular binding modes, the best among the ligand tested seems to be quercetin. Its binding to the beta subunit should prevent the stacking of a subsequent alpha subunit inhibiting therefore the correct polymerization of the assembly. Vanillic and gallic acids should act the same way of quercetin although with less efficiency being these ligands shorter. Ellagic Acid should interfere with the correct functioning of the tubulin assembly being its binding site superposed to the to the site dedicated to bind the GDP moiety. Ferulic, chlorogenic and p-Coumaric acids are positioned between the two tubulin monomers and they act as a link between the subunits α and β and thus they should prevent the correct depolymerization of the microtubular assembly.