All reagents and standards were of analytical reagent grade unless stated otherwise. Kaempferol (≥97%), luteolin (≥98%), quercetin (≥95%) and rutin (≥95%) were purchased from Sigma-Aldrich (St. Louis, MO, USA) and 200 mg/L stock solutions were prepared in ethanol (HPLC grade; Sigma). LC–MS grade acetonitrile (CH₃CN, 99%) (LabScan; Dublin, Ireland) and ultrapure water (Milli-Q Waters purification system; Millipore; Milford, MA, USA) were used for the HPLC-MS analyses. Folin–Ciocalteu's reagent and methanol were purchased from Merck (Darmstadt, Germany). 2,2-Diphenyl-1-picrylhydrazyl (DPPH), ABTS radical cation (2,2′-azino-bis(3-ethylbenzothiazoline)-6-sulphonic acid), DTNB (5,5-dithio-bis(2-nitrobenzoic) acid) and AChE (acetylcholinesterase (Electric ell acetylcholinesterase, Type-VI-S, EC 3.1.1.7), BChE (butyrylcholinesterase (horse serum butyrylcholinesterase, EC 3.1.1.8), L-DOPA (3,4-dihydroxy-L-phenylalanine), tyrosinase, acetylthiocholine iodide (ATCI) and butyrylthiocholine chloride (BTCl) were purchased from Sigma Chemical Co. All other chemicals and solvents were of analytical grade.

Lathyrus species were collected at flowering stage and corresponding information and localities are explained below. Taxonomic identification of the plant materials was confirmed by senior taxonomist Dr. Murad Aydın Sanda [Selcuk University, Science Faculty, Department of Biology (Botany)] based on Flora of Turkey (Davis, 1970) and A checklist of the Flora of Turkey (Vascular Plants) (Güner et al., 2012). Voucher specimens were deposited at the KNYA Herbarium of Department of Biology, Selcuk University, Konya-Turkey.

The plant materials were dried at room temperature. The dried aerial parts were ground to a fine powder using a laboratory mill. To obtain methanolic extracts, the air-dried aerial parts (10 g) were macerated with 200 mL of methanol at room temperature (25°C ± 1°C) for 24 h. The extracts were concentrated under vacuum at 40°C by using a rotary evaporator and stored at + 4°C in dark until use.

The HPLC system consisted of a vacuum degasser, an autosampler and a binary pump (Agilent Series 1100, Agilent Technologies, Santa Clara, CA, USA) equipped with a reversed phase Kinetex core-shell C₁₈ analytical column of 50 × 2.1 mm and 2.6 μm particle size (Phenomenex, Torrance, CA, USA). A C₁₈ Security Guard Ultra cartridge (Phenomenex) of 2.1 mm i.d. was placed before the analytical column. The best separation was achieved by using a mobile phase consisting of acetonitrile (A) and water-formic acid (100:0.1, v/v) (B). The following gradient program was used: 10% A (0 min), 25% A (10–20 min), 50% A (40 min), 100% A (42–47 min), and 10% A (49 min). The mobile phase flow rate was 0.4 mL min⁻¹. After filtration through 0.45 μm PTFE membrane filters, 10 μL of each extract was injected.

The HPLC system was connected to an ion trap mass spectrometer (Esquire 6,000, Bruker Daltonics, Billerica, MA, USA) equipped with an electrospray interface operating in negative ion mode. The scan range was set at m/z 100–1200 with a speed of 13,000 Da/s. The ESI conditions were as follows: drying gas (N₂) flow rate and temperature, 10 mL/min and 365°C; nebulizer gas (N₂) pressure, 50 psi; capillary voltage, 4,500 V; capillary exit voltage, −117.3 V. The acquisition of MSⁿ data was made in auto MSⁿ mode, with isolation width of 4.0 m/z, and fragmentation amplitude of 0.6 V (MSⁿ up to MS⁴). Esquire control software was used for the data acquisition and Data Analysis for processing.

The total phenolic content was determined by Folin-Ciocâlteu method and expressed as gallic acid equivalents (GAEs/g extract), while total flavonoid content was determined by AlCl₃ method and expressed as rutin equivalents (REs/g extract) (Vlase et al., 2014; Zengin et al., 2016). The antioxidant capacity was investigated by different assays: radical scavenging assays (DPPH and ABTS), reducing power (CUPRAC and FRAP), phosphomolybdenum, and metal chelating (ferrozine method) (Dezsi et al., 2015; Toma et al., 2015; Mocan et al., 2017). Enzyme inhibitory activities were detected against cholinesterase, α-amylase and α-glucosidase in 96-well plates. The experimental procedures for antioxidant and enzyme inhibitory assays were previously described by Mocan et al. (2016b) and Zengin et al. (2016). The antioxidant and enzyme inhibitory effects were evaluated by IC₅₀ (%50 of free radical/ enzyme inhibition) and EC₅₀ (the effective concentration at which the absorbance was 0.5 in CUPRAC, FRAP and phosphomolybdenum assays) values. The extracts were evaluated at 0.5–5 mg/mL in the antioxidant and enzyme inhibitory assays. Trolox (0.05–0.5 mg/mL), EDTA (0.01–0.04 mg/mL), galantamine (1–5 μg/mL), kojic acid (0.1–0.4 mg/mL) and acarbose (1–4 mg/mL) were used as positive controls in these assays. The calibration curves were used to calculate IC₅₀ and EC₅₀ values for each extract and reference compounds.

Cells were plated at a density of 5 × 10⁵ cells/well in 6-well plate with 2 mL growth medium. Twenty-four hours after seeding, the medium was removed and replaced with a medium containing 500 μg/mL L. czeczottianus and 1,000 μg/mL L. nissolia methanolic extracts, which are close to IC₅₀ values. After 72 h, the cells were washed with 1X phosphate buffered saline (PBS) to remove cellular debris and photographed under an inverted microscope Leica DM IL LED using a DFC-290 camera (Leica, Wetzlar, Germany).

For all the experiments, the assays were carried out in triplicate. The results are expressed as mean values and standard deviation (SD). The differences between the different extracts were analyzed using one-way analysis of variance (ANOVA) followed by Tukey's honestly significant difference post-hoc test with α = 0.05. This treatment was carried out using SPSS v. 14.0 program.

The analysis of the phenolic composition in aerial parts of L. czeczottianus and L. nissolia was performed by HPLC-ESI-MSⁿ using the negative ionization mode. Two independent assays were carried out for each sample, obtaining similar data regarding the nature and relative intensities of the detected fragments. The base peak chromatograms of the methanolic extracts are shown in Figure 1.

The initial step for the characterization of the phenolic compounds consisted in the determination of the molecular weight of each compound. In the negative ionization mode (ESI⁻) MS¹ spectrum, the most intense peak usually corresponded to the deprotonated molecular ion [M-H]⁻. Compounds were numbered in both chromatograms by their order of elution. The tentative characterization of the detected compounds is shown in Table 1, and the discussion of the characterization is explained in the following sub-sections.

Phenolic compounds are known to serve as multifunctional bioactive components (anti-oxidant, anti-microbial, anti-inflammatory and anti-cancer agents). Several studies also concluded that these components are main contributors on the antioxidant effects of plant extracts (Shahidi and Ambigaipalan, 2015). In this context, they were selected as target compounds in the present study. Total phenolic and flavonoid contents were detected by colorimetric methods, namely Folin-Ciocβlteu and AlCl₃. The results are shown in Table 2. The total phenolic content in L. czeczottianus (63.16 mgGAE/g extract) was higher than in L. nissolia (26.47 mgRE/g extract). However, L. nissolia (20.97 mgRE/g extract) had higher concentrations of flavonoids than L. czeczottianus (14.16 mgRE/g extract). Apparently, the total flavonoids accounted for more than 50% of the total phenolics in L. nissolia extract. From these results, flavonoids were thus the major components in L. nissolia extract. This finding was confirmed also by HPLC-ESI-MSⁿ analysis.

Free radicals trigger a wide variety of chronic and degenerative diseases, such as Alzheimer's disease, diabetes mellitus and cancer. At this point, DPPH and ABTS radicals are widely used to assess free radical scavenging ability of plant extracts or synthetic compounds. In these assays, the initial absorbances (blue for ABTS and purple for DPPH) are decreased by antioxidants, and these changes are spectrophotometrically measured (517 nm for DPPH and 734 nm for ABTS). As can be seen in Table 3, L. czeczottianus was found to exhibit potent free radical scavenging activities in both DPPH (IC₅₀: 1.42 mg/mL) and ABTS (IC₅₀: 1.80 mg/mL) assays. The observed remarkable free radical scavenging activity may be explained by the high level of phenolics in the extracts. This is in accordance with the data published by other authors, indicating that the higher the total phenolic contents, the better the free radical scavenging activity (Bannour et al., 2016; Zhang et al., 2016). Also, some phenolics were reported as effective hydrogen donors by means of their hydroxyl groups (Sevgi et al., 2015).

Reducing power is considered as an important antioxidant feature and it is reflected by the electron-donating ability of antioxidants. For this purpose, CUPRAC (Cu²⁺ →Cu⁺) and FRAP (Fe³⁺ →Fe²⁺) assays were performed. L. czeczottianus exhibited stronger reducing abilities in both assays as similar to the free radical scavenging abilities (Table 3). From these results, the high contents of phenolics in the extracts indicated that these compounds contribute to the higher reducing abilities. Similar results were obtained by several researchers, who reported a strong relationship between total phenolic content and reducing abilities (Beghlal et al., 2016; Smeriglio et al., 2016). Phosphomolybdenum assay is based on the reduction of Mo (VI) to Mo (V) by antioxidants at acidic pH; then, green phosphate-Mo (V) complex is formed, which has its maximum absorbance at 695 nm. The phosphomolybdenum activity was observed as lower for L. czeczottianus (EC₅₀: 1.29 mg/mL) than L. nissolia (EC₅₀: 0.86 mg/mL). (Table 3). Because phosphomolybdenum assay is considered as a total antioxidant capacity assay, our findings for this assay might be attributed also to non-phenolic antioxidants such as tocopherol or vitamin C. These results were consistent with other reports (Albayrak et al., 2010; Zengin et al., 2015). Also, some researchers observed a weak correlation between total phenolic content and phosphomolybdenum assay (Nićiforović et al., 2010; Sarikurkcu et al., 2015).

Transition metals (iron and copper) play an important role in Fenton and Haber-Weiss reactions, which are main reactions involved in the production of free radicals. In this sense, the chelating ability of these metals is considered to have a significant role among the antioxidant mechanisms. The metal chelating ability of Lathyrus extracts was determined by ferrozine method and tabulated in Table 3. In contrast to the free radical and reducing power assays, L. nissolia (IC₅₀: 1.14 mg/mL) exhibited stronger metal chelating activity compared to L. czeczottianus (IC₅₀: >5 mg/mL). Thus, the observed activity for L. nissolia may be caused from non-phenolic chelators, including peptides, polysaccharides or citric acid. In accordance with our results, some authors reported that the metal chelating activity of phenolics plays a minor role among antioxidant mechanisms, and is suggested by several researchers as a secondary antioxidant mechanism (Rice-Evans et al., 1996; Gursoy et al., 2009; Wang et al., 2009). To the best of our knowledge, this study is the first report on the antioxidant properties of these Lathyrus species. In this direction, the obtained results could be useful to provide new perspectives on the Lathyrus genus, and its species could be regarded as potential candidates for developing novel phyto-pharmaceuticals.

Alzheimer's disease (AD) and diabetes mellitus (DM) are major health problems due to the severe increase of their prevalence. For example, recent reports indicated DM prevalence to be 9% among adults, and it is supposed to be well above 15% until 2025 (Waltenberger et al., 2016). Today, more than 50 million people are also affected by AD, and this number is estimated to increase to 131.5 million people by 2,050 (Prince, 2015). Therefore, new effective strategies have vital importance in the treatment of these diseases (Rescigno et al., 2002; Mao et al., 2015; Zengin et al., 2016). Moreover, key enzyme inhibition theory is accepted as one of the most efficient strategies to counteract these diseases (Etxeberria et al., 2012; Murray et al., 2013; Mocan et al., 2016a,b). At this point, enzyme inhibitory assays are considered a very useful tool for investigating the biological properties of herbal extracts or isolated compounds (Bahadori et al., 2016; Dinparast et al., 2016). Considering the mentioned aspects, the enzyme inhibitory properties of L. czeczottianus and L. nissolia toward cholinesterases, α-amylase, α-glucosidase, and tyrosinase were investigated, and the results are gathered in Table 4.

The inhibitory effects on cholinesterases of both species are modest, a higher effect being observed for L. nissolia (IC₅₀: 1.22 mg/mL for AChE). Additionally, L. czeczottianus extract exhibited a high anti-tyrosinase effect, whereas L. nissolia was not active toward this enzyme. High anti-tyrosinase effects and modest inhibitory effects on cholinesterases were already reported in our previous studies concerning L. aureus and L. pratensis (Llorent-Martínez et al., 2016). Moreover, higher inhibitory effects toward α-glucosidase in comparison with α-amylase were observed in this study, the results being in line with our previous findings (Llorent-Martínez et al., 2016; Mocan et al., 2016b). Nonetheless, previous reports indicated that phenolic compounds have lower α-amylase inhibitory activity and a stronger inhibition activity against yeast α-glucosidase (Apostolidis et al., 2011). According to our results, the Lathyrus species have similar α-amylase inhibition and but L. nissolia exerted stronger α-glucosidase inhibition as compared to L. czeczottianus. Also, additional computational studies (molecular docking) were performed to understand possible interactions between the target compounds [5-O-CQA (4), isoschaftoside (22) and vitexin (33)] and tested enzymes as well as to get further insights into the modulation of biological response.

Enzymatic assays performed on the extract of L. czeczottianus evidenced a modest inhibition activity to AchE, and strong tyrosinase, whereas the L. nissolia extracts have shown relevant inhibition activity for AchE and α-glucosidase. The best and most representative enzyme-ligand complexes are reported in Figures 2–8. Among the identified bioactive compounds, vitexin (33) and 5-O-CQA (4), which were found to be dominant in both extracts (especially vitexin), and isoschaftoside (22) which is prevalent in the extract of L. nissolia, were selected for docking experiments. These compounds have been docked to AchE, BChE, α-glucosidase, α-amylase and tyrosinase.

The best pose found for 5-O-CQA (4) docked to AchE (docking score −9.01 Kcal/mol) was stabilized by several interactions with the residues surrounding the enzymatic pocket involving Trp84 with one π–π stack and His440 with two hydrogen bonds. The best ranked pose of vitexin (33) to the same enzyme (score = −8.30 Kcal/mol) interacts with the binding pocket of the AchE by two hydrogen bonds with Arg289, and one hydrogen bond with Ser286, Tyr334, Asp72, and Tyr70. The best pose found for iscoschaftoside (22) docked to AchE was stabilized by two hydrogen bonds to Ser286, two hydrogen bonds to Asp285, one hydrogen bond to Tyr121 and a π–π stack to Trp279 (score = −11.05 Kcal/mol). The best pose found for 5-O-CQA (4) in the enzymatic pocket of BchE was stabilized by the formation of one hydrogen bond with His438, Gln197 and Leu286 with a docking score = −9.39 Kcal/mol. The best pose for isoschaftoside (22) was stabilized by the formation of hydrogen bond with Glu197, Ser198, Leu286, Ser 287 and a π–π stack with Tyr332 (score = −10.08). The best pose found for vitexin (33) in the interaction to the enzymatic pocket of BchE was stabilized by the formation of three hydrogen bonds, respectively, with Gln197, Asp70, Thr120, and a much lower docking score: −11.24 Kcal/mol. These data are in agreement with the literature data in which isoschaftoside (22) has shown a high inhibitory activity to AchE and BChE with IC₅₀ of 0.23 μM and 3.11 μM respectively (Hung et al., 2015).

5-O-CQA (4) binds to the enzymatic pocket of the α-amylase by forming a series of hydrogen bonds with Gln63, Val163, Asp197 residues (docking score = −8.50 Kcal/mol), whereas vitexin (33) interacted with the enzymatic pocket by establishing hydrogen bonds with Tyr62, His201, Glu233, and Asp300 (docking score = −8.27 Kcal/mol). Isoschaftoside (22) best pose established several hydrogen bonds to Asp356, Asp330, Glu233, Asp197, and Trp59, and also two π–π stacks to Trp58 and Trp59. Among the three tested substances, it presented the most favorable docking score (score = −9.69 Kcal/mol) to this enzyme. Several papers were found in scientific literature reporting the antidiabetic properties of some medicinal plants containing isoschaftoside (Song et al., 2012; Khazneh et al., 2016), so supporting the hypothesis that isoschaftoside may be a good inhibitor of α-amylase. In a previous study, vitexin (33) proved to be a potent α-glucosidase inhibitor with a reported EC50 = 0.4 mg/mL, close to that of Acarbose (Yao et al., 2011) and it was found that its best pose interacted with the enzymatic pocket of α-glucosidase by forming three hydrogen bonds with Arg315, Asp215, Arg213, and a π–π stack with the aromatic ring of Phe178 (docking score = −7.29 Kcal/mol). 5-O-CQA (4) interacted only by forming hydrogen bonds with Asp69, Trp158, Glu343, and Asp307 (Docking score = −8.83 Kcal/mol); isoschaftoside (22) interact by forming hydrogen bonds to Glu411, Asp307, Thr310, Asp242 (score = −7.73 Kcal/mol). For tyrosinase, the docking of vitexin (33), isoschaftoside (22) and 5-O-CQA (4) were significantly different. Vitexin (33) and isoschaftoside (22) were both unable to bind directly to the copper atoms present in the enzymatic pocket of tyrosinase also the interactions to the enzymatic pocket are marginal (see Figure 1S in the supporting material); on the contrary, the best pose found for 5-O-CQA (4) evidenced the ability of the molecule to coordinate one Cu atom present at the binding pocket of the enzyme, as depicted in Figure 8. This finding is in full agreement with our enzymatic assays, and also with the reported inhibitory activity found for some caffeoylquininic acid derivatives toward the tyrosinase (Iwai et al., 2004). In conclusion, the docking experiments performed in this work can explain the literature data regarding the inhibitory activity of vitexin (33), isoshaftoside (22) and 5-O-QCA (4), being capable to interact with the residues present in the binding pocket of the considered enzymes. In particular, vitexin (33), which has been reported to be efficacious to control blood glucose in diabetic rats (Choo et al., 2012), is present in the extracts of both plants and can be addressed as one of the compounds responsible for the α-glucosidase inhibitory activity found for both extracts together with isoschaftoside (22) which has shown a remarkable docking score on alpha-amylase, whereas 5-O-CQA (4), as demonstrated in the docking experiments, may efficaciously inhibit the tyrosinase activity by binding to one Cu atom present in the enzymatic pocket.

Toxicity is a major issue to consider in drug screening or drug development from natural product derivatives. Some of the active compounds obtained from these natural products have biomedical uses such as anti-neoplastic and anti-inflammatory agents, and the toxicity of these compounds needs to be clarified by cytotoxicity assays in biological systems. In this study, we have tested the cytotoxicity of two different Lathyrus species (L. czeczottianus and L. nissolia) on human embryonic kidney cells, HEK-293, since these cells are widely used in drug toxicity studies as kidney may be one of the major target organ of several toxins (Mantle et al., 2015; Sang et al., 2016). Previous studies reported the use of HEK-293 cell line for neurobiological research as well, based on the findings that HEK-293 cells have many properties in common with neuronal precursors(Shaw et al., 2002; Thomas and Smart, 2005; Lin et al., 2014). For monitoring cytotoxicity, we used iCELLigence system, which enables to get real-time, label-free and non-invasive measurements from living cells (Bird and Kirstein, 2009; Martinez-Serra et al., 2014). In recent years, this system has been one of the gold standards in cell biology research since it measures the effect of the drugs without any additional treatment on the cells. In addition to this, RTCA systems have own data analysis software which calculates IC₅₀ values automatically. In this study, IC₅₀ values of L. czeczottianus and L. nissolia were determined as 413.19 ± 16.3 μg/mL and 1040.83 ± 48.8 μg/mL, respectively, from three independent experiments in a time and dose dependent manner after treating cells with six different concentrations of these two extracts (Figure 9). Furthermore, very high doses were tested, close to IC₅₀ concentrations of the extracts, on HEK-293 cell morphology and almost 50% cellular viability were observed as seen in middle and right panels compared to control cells on the left (Figure 10). From these results, the studied extracts had no cytotoxic activities against HEK-293 cell lines. However, L. czeczottianus had higher cytotoxic effect with lower value of IC₅₀ as compared to L. nissolia. In this direction, the observed remarkable activity for L. nissolia may be explained by the higher concentration of flavonoids in the extract. In fact, several authors reported that some flavonoids exhibited no cytotoxic effects on various cell lines (Kumar and Pandey, 2013; Sak, 2014). To the extent of our knowledge, only one previous study has been conducted on the cytotoxic evaluation of the Lathyrus species (L. laxiflorus) (Spanou et al., 2010).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.