Novel Flavonol Glycosides from the Aerial Parts of Lentil (Lens culinaris)

While the phytochemical composition of lentil (Lens culinaris) seeds is well described in scientific literature, there is very little available data about secondary metabolites from lentil leaves and stems. Our research reveals that the aerial parts of lentil are a rich source of flavonoids. Six kaempferol and twelve quercetin glycosides were isolated, their structures were elucidated using NMR spectroscopy and chemical methods. This group includes 16 compounds which have not been previously described in the scientific literature: quercetin 3-O-β-D-glucopyranosyl(1→2)-β-D-galactopyranoside-7-O-β-D-glucuropyranoside (1), kaempferol 3-O-β-D-glucopyranosyl(1→2)-β-D-galacto-pyranoside-7-O-β-D-glucuropyranoside (3), their derivatives 4–10, 12–15, 17, 18 acylated with caffeic, p-coumaric, ferulic, or 3,4,5-trihydroxycinnamic acid and kaempferol 3-O-{[(6-O-E-p-coumaroyl)-β-D-glucopyranosyl(1→2)]-α-L-rhamnopyranosyl(1→6)}-β-D-galactopyranoside-7-O-α-L-rhamnopyranoside (11). Their DPPH scavenging activity was also evaluated. This is probably the first detailed description of flavonoids from the aerial parts of lentil.


Introduction
Lentil (Lens culinaris Medik) whose cultivation started in the Near East in the Neolithic period, is one of the earliest domesticated plants. Nowadays, lentil is a crop of high importance in many countries of Northern Africa, Western and Southern Asia, as well as in Canada, which is the world leader in its production [1]. The nutritional value of legumes is commonly known, especially their high content of good quality protein. Lentil grain contains, on average, about 28% of protein, and is rich in lysine and several other essential amino acids. It is also a good source of minerals (Ca, Fe, K, Mg, P, Zn), some B-group vitamins and pantothenic acid [2]. Lentil straw finds use as a valued fodder in many parts of the Near East [3].
There is a broad literature on basic nutrients and raffinose family oligosaccharides of lentil grain, but the number of publications about lentil secondary metabolites is much more limited. The seeds of this plant are known to contain phytosterols, phytic acid, saponins and phenolic compounds. The seed phenolics are represented by condensed tannins (present in significant amounts, especially in the seed coat), phenolic acids, lignans, stilbens, and flavonoids. The reported lentil flavonoids comprise mainly catechin and and glycosidic derivatives of kaempferol, quercetin, myricetin, apigenin and luteolin [2,[4][5][6][7][8][9]. The marked differences in flavonoid profiles among individual studies can be explained by the use of different cultivars and plant growth conditions. While secondary metabolites of lentil seeds are well characterized, it seems there are hardly any data available about secondary metabolites in other organs of this plant.
Flavonoids are a group of phenolic compounds very widely distributed in the plant kingdom [10]. They are common food constituents, extensively investigated due to their antioxidant properties, diverse biological activities and role in prevention of cardiovascular diseases and cancer [11,12]. Leaves and stems of numerous legumes are known to be a rich source of various types of flavonoids, but there is extremely little information about phenolic compounds from the aerial parts of lentil. A few articles about lentil sprouts provide some more detailed data about phenolics, though flavonoids were only preliminarily identified. However, sprouts of the lentil were reported to contain acylated glycosides of kaempferol and quercetin [13][14][15]. Because of the broad bioactive potential of flavonoids, finding new sources and new types of these compounds still remains an important task, and the aim of our study was to isolate and identify flavonoids from aerial parts of the lentil cultivar Tina.

Results and Discussion
A preliminary UHPLC-MS/MS analysis of methanol extract from lentil aerial parts revealed the presence of numerous phenolic compounds ( Figure 1) having flavonoid-like UV spectra. They were tentatively identified as kaempferol and quercetin glycosides, most of them acylated with phenolic acids. Three-step chromatographic separation of the extract led to the isolation of 18 flavonoids, including all major and several minor compounds (   NMR spectroscopy data included various 1D [ 1 H, proton-decoupled 13 C and DEPT-135, selective excitation 1D-TOCSY and 1D-ROESY (mixing times of 120 and 250 ms respectively)] and 2D [ 1 H-1 H gCOSY (magnitude mode), 1 H-1 H TOCSY, 1 H-1 H ROESY, 1 H-13 C gHSQC, 1 H-13 C gHSQC-TOCSY (mixing time of 80 ms) and 1 H- 13 C gHMBC ( n JCH = 8 Hz)] spectra (Figures S1-S193). Hydrolysis of the purified flavonoids, followed by the determination of absolute configuration of monosaccharides showed that these were D-Glc, D-Gal, and D-GlcA for compounds 1, 3-10, 12-15, 17 and 18, while 11 contained D-Glc, D-Gal, and L-Rha (see Supplementary Figures S211-S226). Analyses of the remaining products of alkaline and acid hydrolyses of these flavonoids confirmed the identity of their aglycones and constituent phenolic acids (see Supplementary Figures S195-S209). In the case of the 3,4,5-trihydroxycinnamic acid, present in the compound 4, an analytical standard was not available and the compound was tentatively identified on the base of its molecular mass. We could not find any precise literature data about its UV maxima, but the UV spectrum we obtained (see Supplementary Figure S196) was similar to one presented in the work of Kopycki et al. [16]. , corresponding to the kaempferol ion Y0 − . A precise structure elucidation was possible after 1D and 2D-NMR analyses of these compounds. The 13 C-NMR spectrum of 1 showed 33 signals, sorted by 13 C and DEPT-135 experiments into 2 CH2, 20 CH and 11 quaternary carbon atoms. The aromatic region of the 1 H and COSY spectra of 1 exhibited the presence of two sets of aromatic protons, characteristic for the quercetin aglycone (Table 1). One set corresponded to a tetrasubstituted aromatic ring with two meta-coupling protons and appeared at δH 6.77 (d, J = 1.6 Hz, H-8) and 6.50 (d, J = 1.5 Hz, H-6), which were correlated in the HSQC spectrum with their aromatic carbon atoms at δC 95.8 and 100.7 ppm, respectively. The other set corresponded to 3,4-dihydroxyphenyl group at δH 7.80 (d, J = 1.9 Hz, H-2'), 7.62 (dd, J = 8.4, 1.8 Hz, H-6'), and 6.91 (d, J = 8.4 Hz, H-5'), in accordance with AMX system of ring B of the aglycone. The assignments of all carbons of the flavonol moiety were accomplished by interpretation of the HSQC and HMBC spectra. Observed heteronuclear multiple bond connectivity (HMBC) correlations from H-2' and H-6' to C-2, 4 J correlation from H-8 to C-4 and downfield shifted resonance at δC 159.4 for C-2, indicated that the compound 1 contained 3-O substituted quercetin. This was further supported by the result of UHPLC analysis of non-polar products of acid hydrolysis of 1. The carbohydrate region of 1 H NMR spectrum showed the presence of the oxymethine protons in the range δ 3.36-4.14. Moreover, three anomeric proton signals at δH 5.35 (d, J = 7.6 Hz, H-1Gal), 4.79 (d, J = 7.1 Hz, H-1Glc) and 5.20 (m, H-1GlcA) were also observed, indicating the presence of three sugar units. Based on the values of coupling constants (J > 7 Hz), and the analysis of 1 H-, 13 C-NMR spectra, and 1D TOCSY, COSY, TOCSY, HSQC, HSQC-TOCSY and HMBC data, the three sugar units were elucidated as β-galactopyranoside δH/C 5.35 (H-1Gal)/101.6 (C-1Gal), β-glucopyranoside δH/C 4.79 (H-1Glc)/105.1 (C-1Glc) and β-glucuropyranoside δH/C 5.20 (H-1GlcA)/101.4 (C-1GlcA) ( Table 2). It was observed that the glucuropyranosyl moiety demonstrated non-first order 1 H-NMR spectrum, which was not the result of impurities or other physical factors. In our opinion it was caused by the NMR phenomenon called virtual coupling [17].  Typically it occurs when the chemical shift difference between two J coupled nuclei is of the same order as the coupling constant. It is said that this type of coupling shows dependence on solvent and field strength. Other factors influencing the complexity of the spectrum are both steric and electronic contributions. As it is shown later in this paper, other isolated compounds, substituted in either C-2GlcA or C-6Glc (or both) do not show the phenomenon of virtual coupling, but it is still visible in the GlcA of compound 3. The long range correlations observed in the HMBC spectrum between the anomeric proton of the glucose (δH 4.79, H-1Glc) and C-2 of galactose (δC 80.8) indicated the presence of interglycosidic linkage between these hexosyl units (1→2). This was further supported by the NOE effect detected in the rotating frame nuclear Overhauser effect spectroscopy spectrum (ROESY) between H-1Glc and H-2Gal. The 3-O glycosidation site was determined mainly by NOE effect in ROESY spectrum between the anomeric proton of galactopyranoside (δH 5. 35, and ring B (δH 7.80, H-2' and δH 7.62, H-6') of the quercetin moiety. The other, indirect evidence, was the downfield shifted resonance at δC 159.4 for C-2, as mentioned earlier. The correlation observed in the HMBC spectrum from anomeric proton at δH 5.20 (H-1GlcA) to carbon C-7 (δC 164.3) and the NOE effect visible in the ROESY spectrum between H-1GlcA and H-6/8 indicated that the point of attachment of β-glucuropyranosyl unit to quercetin was at C-7 position. Therefore, the compound 1 was identified as quercetin 3-O-β-D-glucopyranosyl(1→2)-β-D-galactopyranoside-7-O-β-D-glucuropyranoside. Since the compound has not been reported before, we propose to name it lensoside A.
This scheme of glycosidation, along with the type of sugar moieties, was observed in all isolated compounds, except for molecules 2, 11 and 16. Furthermore, on the basis of NMR spectra and the results of UHPLC analyses of non-polar products of acid hydrolysis, we established that the quercetin nucleus was also present in compounds 2, 4, 5, 7, 8,12,14,15,16,17 and 18, which will be discussed later.
The 13 C-NMR spectrum of 3, like 1, showed 33 signals, sorted by 13 C and DEPT-135 experiments into two CH2, 21 CH and 10 quaternary carbon atoms. The aromatic region of the 1 H and COSY spectra of 3 exhibited the presence of two sets of aromatic protons. One set corresponded to a tetrasubstituted aromatic ring with two meta-coupling protons and appeared at  Table 2). As in the case of the compound 1, the glucuropyranosyl moiety demonstrated non-first order 1 H NMR spectrum caused by virtual coupling. The long range correlations observed in the HMBC spectrum between the anomeric proton of the glucose (δH 4.78, H-1Glc) and C-2 of galactose (δC 80.5) indicated the interglycosidic linkage between these hexosyl units (1→2). This, along with the correlations observed in the HMBC spectrum from anomeric protons at δH 5.44 (H-1Gal) to carbon C-3 (δC 135.2), δH 5.20 (H-1GlcA) to carbon C-7 (δC 164.3) and the NOE effect visible in the ROESY spectrum between H-1Gal and H-2'/6' together with NOE effect between H-1GlcA and H-6/8 indicated that the points of glycosidation were identical as in 1. Therefore, 3 was identified as kaempferol 3-O-β-D-glucopyranosyl(1→2)-β-D-galactopyranoside-7-O-β-D-glucuropyranoside, named lensoside B. Furthermore, it was established on the basis of NMR spectra and the results of UHPLC analysis of non-polar products of acid hydrolysis that the kaempferol nucleus was also present in compounds 6, 9, 10, 11, and 13, which will be discussed later.
Compounds 4-10 form another group with a common structural scheme. Their UV spectra were characterized by a distinct (18- The UV and MS spectra, as well as the higher retention times of these flavonoids clearly suggested that they might be acylated forms of compounds 1-3. NMR analyses confirmed this conclusion. The 13 C-NMR spectrum of the compound 4 showed 42 signals, sorted by 13 (Table 1), but also a set of coupled doublets E-α-H and β-H at δH 7.13 and 5.85 corresponding to E-(Jα,β = 15.8 Hz) olefinic moiety and a downfield shifted aromatic singlet at δH 6.23 (2H) correlated in the HSQC spectrum with its aromatic carbon atom at δC 108.5. The long-range correlations observed in HMBC spectrum suggested that this is a E-3,4,5-trihydroxycinnamoyl moiety. For the Glc residue H-6 protons and C-6 carbon were downfield shifted to δH 4.41 (m, 2H) and δC 64. 8, espectively (Table 3). Additionally, protons H-6Glc exhibited 3 J correlation in the HMBC spectrum with a carbonyl group resonated at δC 167. 5, corresponding to C-9triOHCin (COO − ). Therefore, 4 was a monoacylated derivative of 1, established as quercetin Compounds 5, 7 and 8, similarly to 4, were monoacylated with phenolic acids. All of them shared the same basic skeleton, identical with 1, which was confirmed with COSY, ROESY, HSQC and HMBC spectra along with selective experiments (1D TOCSY and 1D ROESY) used for the determination of nature of sugar moieties. The aromatic region of the 1 H and COSY spectra of 5 contained sets of resonances characteristic for E-(Jα,β = 15.9 Hz) olefinic moiety and ABX system corresponding to a 3,4-dihydroxyphenyl group at δH 6.75 (d, J = 1.9 Hz, H-2Caf), 6.62 (d, J = 8.3 Hz, H-5Caf), and 6.59 (dd, J = 8.2, 1.9 Hz, H-6Caf) and it was identified as E-caffeoyl group. Likewise, the aromatic region of the 1 H and COSY spectra of 7 contained one pair of E-α-H and β-H doublets (δH 6.02 and 7.35 with Jα,β = 15.9 Hz), but it also exhibited a AA'XX' system corresponding to p-hydroxyphenyl group at δH 7.12 (d, J = 8.1 Hz, H-2/6Cou) and 6.67 (d, J = 8.1 Hz, H-3/5Cou), which was interpreted as E-p-coumaroyl moiety in turn.     Compound 8 exhibited in the aromatic region of 1 H and COSY spectra sets of resonances very similar to 5. One set of resonances was characteristic for E-(Jα,β = 15.9 Hz) olefinic moiety and an ABX system corresponded to 3,4- Table 4). The 1 H-NMR spectrum of 12 contained resonances typical for the quercetin nucleus and the basic skeleton was identical with 1 ( Table 1). The aromatic region of the 1 H and COSY spectra of 12 contained two pairs of E-α-H and β-H doublets (δH 7.65 and 6.34 with Jα,β = 15.9 Hz; δH 7.23 and 5.90 with Jα,β = 15.9 Hz), and exhibited two separate AMX systems corresponded to 3,4-dihydroxyphenyl groups, which were interpreted as two E-caffeoyl moieties.       − , corresponding to the quercetin radical ion, created after the loss of a deoxyhexose. On the base of its NMR spectra the compound 16 was identified as a widely occurring flavonoid, quercitrin, the quercetin 3-O-α-L-rhamnoside [19].
The ability of the purified flavonoids to scavenge DPPH radicals was assessed using a rapid TLC-DPPH test. Their antiradical activities were compared with the activity of rutin, and expressed as a sample activity/rutin activity ratio (Table 6). Compound 5 turned out to be a better radical scavenger than rutin, and the antiradical activities of 6, 7, 8,12,14,15 and 16 were also high. It should be noted that the scavenging effect of quercetin derivatives was in most cases much stronger than that of kaempferol glycosides. Moreover, acylation of flavonoid glycosides with caffeic acid significantly increased their antiradical properties, which is particularly visible when activities of the compounds 3 and 6, or 1 and 5 are compared.  On the contrary, acylation with p-coumaric or ferulic acid had a small effect on the antiradical activity of the investigated flavonoids, which is especially visible in the case of the kaempferol derivatives (compounds 9, 10,11,18). These observations can be explained by the fact that quercetin and caffeic acid were reported to be more efficient DPPH scavengers (which can be attributed to the presence of catechol moiety in their molecules) than kaempferol, ferulic acid, and particularly p-coumaric acid, [20][21][22][23].
It seems that the majority of the isolated flavonoids, except for compounds 2 and 16 [18,19], have not been described before in the scientific literature. However, it is possible that the compound 11 was previously detected in lentil seeds by Zou et al., who found two flavonoid derivatives with similar molecular masses (like 11, they gave deprotonated ions at m/z 1047) and UV spectrum [6]. Compound 11 is an acylated form of kaempferol 3-O-[β-D-glucopyranosyl(1→2){α-Lrhamnopyranosyl(1→6)}-β-D-galactopyranoside]-7-O-α-L-rhamnopyranoside, a flavonoid occurring in lentil seeds, and found also in the legume Ateleia chicoasensis and the cactus Cephalocereus senilis [24][25][26]. Moreover, a non-acylated flavonoid (or more flavonoids with nearly identical retention times), giving the deprotonated ion at m/z 901 and showing the relevant fragmentation pattern, was found to be the main phenolic compound of the Tina lentil seeds (see Supplementary Figure S227)). 7-O-glucuronides of flavonols were rarely found. Cichorium intybus and some Epilobium sp. contain simple kaempferol and quercetin 7-O-glucuronides, while the more complex glycosides were found in tulip (Tulipa gesneriana) and onion (A. cepa) [18,[27][28][29]. In contrast, 7-O-glucuronides of flavones are more widespread among plants, they can be found, e.g. in different plants belonging to the Lamiales order [30]. Among legumes, such compounds can be found in aerial parts of Medicago sp., known to contain different acylated and non-acylated 7-O-glucuronides of apigenin, luteolin, chrysoeriol, and tricin [31,32]. It seems the presence of the described 7-O-glucuronides of kaempferol and quercetin in aerial parts of lentil is interesting from chemotaxonoimic point of view, and these compounds can be used as molecular markers.
There are several articles describing phenolic compounds of lentil sprouts. Phenolic constituents of the lentil variety Aldona were analyzed, using LC-MS, by Troszyńska et al. [13]. They detected eight acylated and non-acylated glycosides of kaempferol and quercetin (only generally described), including an acylated kaempferol derivative showing a deprotonated ion at m/z 931. It seems this flavonoid may be identical to the compound 9 due to their similar fragment ions and UV spectra. Other publications present levels of some phenolic acids and flavonoid aglycones (luteolin, kaempferol, quercetin, daidzein, genistein, naringenin, catechin) but no information about flavonoid glycosides is available [14,15].

Purification of Phenolic Compounds
Fractions F1-F3 were subjected to low pressure reversed phase liquid chromatography on a C18 column (30 × 3.4 cm, i.d.; Lichroprep RP-18 40-63 μm, Merck). For separation of fraction F1, the column was equilibrated with 1.5% aqueous methanol with 0.1% formic acid. A 1.400 g portion of fraction F1 was dissolved in the same solution and loaded onto the column. The column was washed with the starting eluent, and the constituent phenolics were eluted with a stepwise gradient of: 5%-40% aqueous methanol containing 0.1% formic acid; 10 mL fractions were collected.
A similar experimental scheme was also used for fractions F2 and F3. Fraction F2 (1.400 g) was dissolved in 5% aqueous methanol with 0.1% formic acid and applied on the chromatographic column equilibrated with the same solvent. The column was subsequently washed with the starting eluent, followed by the step gradient of 10%-45% methanol containing 0.1% formic acid. Fraction F3 (1.443 g) was dissolved in 33% aqueous methanol with 0.1% formic acid and loaded onto the chromatographic column equilibrated with 25% methanol containing 0.1% formic acid. The column was washed with the starting solution and the sample constituents were eluted with a stepwise gradient of 30%-55% methanol.

Mass Analyses of Purified Compounds
Exact masses of lentil flavonoids were determined by direct infusion electrospray high resolution (Q-TOF) mass spectrometry (HRESI-MS), using a SYNAPT G2-S HDMS mass spectrometer (Waters). Fragmentation analyses were performed by direct infusion electrospray mass spectrometry, using a ACQUITY TQD mass spectrometer (Waters).

Determining the Absolute Configuration of Sugars
The absolute configuration of sugars was determined according to the modified method of Tanaka,et al. [33] Samples of monosaccharides obtained after the acid hydrolysis of flavonoids were dissolved in anhydrous pyridine (100 μL) containing L-cysteine methyl ester hydrochloride (0.5 mg) and heated at 60 °C for 1 h. Then of solution of o-tolyl isothiocyanate (0.5 mg) in pyridine (100 μL) was added, and the mixture was heated for another hour, at 60 °C. After cooling, samples were analyzed by UPLC-ESI-MS/MS. Chromatographic separations were carried out on a Acquity BEH C18 column (100 × 2.1 mm, 1.7 μm; Waters). Mass spectrometry analyses were performed in positive ionization mode, using the SRM method. Details of the analysis can be found in the work of Pérez, et al. [34]. D-glucose (Glc), D-galactose (Gal), D-glucuronic acid (GlcA) and L-rhamnose (Rha) were identified on the base of retention time and m/z values of authentic standards, derivatized in the same way.

TLC-DPPH Test
The ability of lentil flavonoids to scavenge DPPH radicals was determined using a TLC rapid test [35]. Briefly, standard solutions (1 mg·mL −1 ) of the purified flavonoids and rutin (positive control) were prepared. Aliquots of the standard solutions (3 µL) were applied onto a silica TLC plates, and the plates were developed as described above. They were subsequently dried and immersed for 3 s in freshly prepared 0.2% (w/v) methanolic DPPH solution. The test was performed in triplicate. The developed plates were kept in the dark for 30 min and then scanned in a flat-bed scanner. The obtained scans were analyzed using ImageJ image processing program. The antiradical activity of lentil flavonoids was expressed in relation to activity of rutin.