Polyphenolic Compounds from Lespedeza bicolor Protect Neuronal Cells from Oxidative Stress

Pterocarpans and related polyphenolics are known as promising neuroprotective agents. We used models of rotenone-, paraquat-, and 6-hydroxydopamine-induced neurotoxicity to study the neuroprotective activity of polyphenolic compounds from Lespedeza bicolor and their effects on mitochondrial membrane potential. We isolated 11 polyphenolic compounds: a novel coumestan lespebicoumestan A (10) and a novel stilbenoid 5’-isoprenylbicoloketon (11) as well as three previously known pterocarpans, two pterocarpens, one coumestan, one stilbenoid, and a dimeric flavonoid. Pterocarpans 3 and 6, stilbenoid 5, and dimeric flavonoid 8 significantly increased the percentage of living cells after treatment with paraquat (PQ), but only pterocarpan 6 slightly decreased the ROS level in PQ-treated cells. Pterocarpan 3 and stilbenoid 5 were shown to effectively increase mitochondrial membrane potential in PQ-treated cells. We showed that pterocarpans 2 and 3, containing a 3’-methyl-3’-isohexenylpyran ring; pterocarpens 4 and 9, with a double bond between C-6a and C-11a; and coumestan 10 significantly increased the percentage of living cells by decreasing ROS levels in 6-OHDA-treated cells, which is in accordance with their rather high activity in DPPH• and FRAP tests. Compounds 9 and 10 effectively increased the percentage of living cells after treatment with rotenone but did not significantly decrease ROS levels.


Introduction
Parkinson's disease (PD) is a common neurodegenerative disease of older age [1]. The pathogenesis of PD includes the death of neurons from oxidative damage as a result of an increase in the production of intracellular reactive oxygen species (ROS), which causes damage to lipids, proteins, and DNA. A combination of oxidative stress, mitochondrial dysfunction, protein misprocessing, and genetic factors plays a crucial role in the pathogenesis of age-related neurodegeneration [1,2]. The etiology of PD can be associated not only with aging but also with adverse environmental influences and neurotoxins. 6-hydroxydopamine (6-OHDA) and 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP) were shown to cause PD symptoms [3]. Several pesticides and herbicides (rotenone, paraquat (PQ), maneb (MB), and mancozeb (MZ)) are also neurotoxins and cause pathologies similar to PD [3]. The neurotoxic properties of these compounds are primarily due to their ability to generate toxic free radicals and reactive oxygen species in cells. For example, PQ affects the redox cycle and activity of the enzyme nitric oxide synthase in neuronal cells, which leads to increased production of ROS and increased levels of α-synuclein and

General Experimental Procedures
We recorded the UV spectra on a UV-1601 PC spectrophotometer (Shimadzu, Kyoto, Japan). The CD spectra were recorded using a Chirascan-Plus Quick Start CD Spectrometer (Applied Photophysics Limited, Leatherhead, UK) (acetonitrile, 20 • C). The 1 H, 13 C, and two-dimensional NMR spectra (HSQC, HMBC, COSY, ROESY) were recorded in CDCl 3 on a Bruker AVANCE III DRX-700 NMR spectrometer (Bruker, Karlsruhe, Germany). The chemical shift values (δ) and the coupling constants (J) are given in parts per million and Hz, respectively.

HR-ESI-MS
We performed HR-ESI-MS analysis on a Shimadzu hybrid ion-trap-time-of-flight mass spectrometer (Shimadzu, Kyoto, Japan). The following instrument settings were applied for analysis: drying gas (N 2 ) pressure-200 kPa; nebulizer gas (N 2 ) flow-1.5 L/min; electrospray ionization (ESI) source potential-3.8 kV for negative polarity ionization and 4.5 kV for positive polarity ionization; temperature for the curved desolvation line (CDL) and heat block-200 • C; detector voltage-1.5 kV, detection range-100-900 m/z. The mass accuracy was below 4 ppm. We acquired and processed the data using Shimadzu LCMS Solution software v3.60.361 (Shimadzu, Kyoto, Japan).

Antiradical Activity
The 2,2-diphenyl-1-picrylhydrazyl (DPPH • ) radical-scavenging effect of polyphenolic compounds 7-11 was determined as described in [24]. Polyphenolic compounds 7, 8, 9, 10, and 11 were added to DPPH • solution in MeOH (10 −4 M) at concentrations from 6 to 34 µM. The reacting mixture was kept in the dark for 30 min at room temperature. Then, we measured the absorbance at 517 nm using a Shimadzu UV 1240 spectrophotometer (Shimadzu, Kyoto, Japan). Equation (1) was used to calculate the DPPH • radical-scavenging effect (%): where A 0 is the absorbance of DPPH • solution of a blank sample (without polyphenolic compounds); A X is the absorbance of DPPH • solution in the presence of different concentrations of polyphenolic compounds. Quercetin was used as a reference compound. All experiments were performed in triplicate. We calculated the half-maximal scavenging concentration (SC 50 ) for polyphenolic compounds by plotting the DPPH • scavenging effect (%) against the concentrations of polyphenolic compounds. SC 50 values are given as the mean value ± SEM.

Ferric Reducing Antioxidant Power (FRAP) Assay
We performed the FRAP assay as described in [29]. We prepared the FRAP reagent by mixing 2.5 mL of TPTZ (2,4,6-tris(2-pyridyl)-s-triazine) solution (10 mM) in 40 mM HCl and 25 mL of FeCl 3 solution (20 mM) in acetate buffer solution (300 mM, pH 3.6). Polyphenolics 7-11 were added to 3 mL of FRAP reagent at concentrations from 6 to 34 µM. The mixture was kept in the dark at room temperature for 4 min. Then, we measured the absorbance at 595 nm using a Shimadzu UV 1240 spectrophotometer. Equation (2) was used to calculate the FRAP values for polyphenolic compounds 7-11: where C Fe is the concentration of Fe 2+ (µM) formed in the reaction; C X is the concentration of polyphenolic compounds in the reacting mixture. The concentration of Fe 2+ (µM) formed in the reaction was determined using the calibration curve obtained for different concentrations of FeSO 4 ·7H 2 O.

Quantum-Chemical Modeling
We applied density functional theory (DFT) with the nonlocal exchange-correlation functional B3LYP [30], the polarization continuum model (PCM) [31], and split-valence basis set 6-311G(d), implemented in the Gaussian 16 package of programs [32] to perform the quantum-chemical calculations for compounds 9 and 10 in acetonitrile solvent. The molecular cavity was modeled according to unified force field (radii = UFF). The detailed conformational analysis of compounds 9 and 10 preceded the following calculations of their chiroptical properties.
The statistical weights (g im ) of different conformations were calculated according to Equation (3): where the summation was performed over all found stable conformations of the stereoisomer under study, and ∆G im = G i − G m ; G = E el + G tr,T + G rot,T + G vib,T is the sum of electronic, translational, rotational, and vibrational contributions to the Gibbs free energy, respectively, calculated at temperature T = 298.15 K; the subscript "m" denotes the conformation, for which G is minimal. The excitation energies and the rotatory strengths were calculated using time-dependent density functional theory (TDDFT), cam-B3LYP functional theory [33], and a PCM model and basis set, used previously for conformational analysis. Each individual transition from the electronic ground state to the i-th calculated excited electronic state (1 ≤ i ≤ 115) was simulated as a Gaussian-type function. The bandwidths, taken at 1/e peak heights, were chosen to be σ = 0.24 eV. The total theoretical ECD spectrum was obtained after statistical averaging over all selected conformations using Equation (4): where i denotes different conformations of the stereoisomer under study. Conformations with Gibbs free energies in the region of ∆G im ≤ 5 kcal/mol were accounted for.

Cell Line and Culture Conditions
The murine neuroblastoma cell line  We incubated Neuro-2a cells (1 × 10 4 cells/well) at 37 • C in a CO 2 incubator for 24 h until they formed an adherent monolayer. Then, 20 µL of the tested solution was loaded onto the cells and incubated for 24 h. After incubation, the medium containing the polyphenolic test compounds was replaced by 100 µL of fresh medium. Then, we added 10 µL of MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide) (Sigma-Aldrich, St. Louis, MO, USA) stock solution (5 mg/mL) to each well and incubated the microplate for 4 h. Then, 100 µL of SDS-HCl solution (1 g SDS/10 mL dH 2 O/17 µL 6 M HCl) was added to each well, followed by incubation for 18 h. We measured the absorbance of the converted dye formazan on a Multiskan FC microplate photometer (Thermo Scientific, Waltham, MA, USA) at a wavelength of 570 nm [34]. We performed all experiments in triplicate and expressed the cytotoxic activity as percent of living cells.

Mitochondrial Membrane Potential (MMP) Detection
The cells were incubated for 1 h in a 96-well plate (1 × 10 4 cells/well) with polyphenolic compounds (1 and 10 µM). Then, PQ (500 µM) was added, and the cell suspension was incubated for 1 h. Cells incubated without PQ and compounds were used as positive control, and cells with PQ alone were used as negative control. We added the tetramethylrhodamine methyl (TMRM) (Sigma-Aldrich, St. Louis, MO, USA) solution (500 nM) to each well, and incubated cells for 30 min at 37 • C. After the incubation, we measured the intensity of fluorescence with a PHERAstar FSplate reader (BMG Labtech, Ortenberg, Germany) at λ ex = 540 nm and λ em = 590 nm. We processed the data using MARS Data Analysis v3.01R2 (BMG Labtech, Ortenberg, Germany) and presented results as percentages of the positive control value.

Statistical Analysis
We carried out all the experiments in triplicate and performed Student's t-test using SigmaPlot 14.0 (Systat Software Inc., San Jose, CA, USA) to determine statistical significance.

Mitochondrial Membrane Potential (MMP) Detection
The cells were incubated for 1 h in a 96-well plate (1 × 10 4 cells/well) with polyphenolic compounds (1 and 10 µM). Then, PQ (500 µM) was added, and the cell suspension was incubated for 1 h. Cells incubated without PQ and compounds were used as positive control, and cells with PQ alone were used as negative control. We added the tetramethylrhodamine methyl (TMRM) (Sigma-Aldrich, St. Louis, MO, USA) solution (500 nM) to each well, and incubated cells for 30 min at 37 °C. After the incubation, we measured the intensity of fluorescence with a PHERAstar FSplate reader (BMG Labtech, Ortenberg, Germany) at λex = 540 nm and λem = 590 nm. We processed the data using MARS Data Analysis v3.01R2 (BMG Labtech, Ortenberg, Germany) and presented results as percentages of the positive control value.

Statistical Analysis
We carried out all the experiments in triplicate and performed Student's t-test using SigmaPlot 14.0 (Systat Software Inc., San Jose, CA, USA) to determine statistical significance.
We completely assigned all signals in the 1 H and 13 C NMR spectra of 10 on the basis of COSY, HMBC, and ROESY spectral data. Thus, compound 10 was determined to be a  (Table 1). Fifteen carbon atoms belonged to the coumestan skeleton (rings A-D), and 10 atoms formed a 3 -methyl-3 -isohexenylpyran ring (E). An ester carbonyl carbon signal was observed in the 13 C NMR spectrum of 10 at δ C 159.0 and was assigned to C-6 of the coumestan skeleton [22]. The 1 H NMR spectra of 10 showed the presence of an ABX spin system: the signals at δ H 7.85 (d, J = 8.5), 6.95 (d, J = 8.5), and 7.13 (s) were attributed to protons H-1, H-2, and H-4 of ring A, respectively (Table 1). A singlet signal at δ H 7.44 was ascribed to the H-7 proton (ring D of the coumestan skeleton). We also observed two broad singlet signals at δ H 6.54 and 5.50 due to OH-3 and OH-8 protons, respectively, in the 1 H NMR spectra of 10 ( Table 1). The 1 H NMR spectra of 10 revealed signals at δ H 6.89 (1H, d, J = 9.9 Hz) and 5.78 (1H, d, J = 9.9 Hz) due to the H-1 and H-2 protons of the AX-type olefinic proton system, suggesting that 10 had an oxidatively cyclized geranyl side chain similar to the structures of compounds 2 and 9 (Table 1) [24]. We completely assigned all signals in the 1 H and 13 C NMR spectra of 10 on the basis of COSY, HMBC, and ROESY spectral data. Thus, compound 10 was determined to be a derivative of lespedezol A 6 (7), previously isolated from L. homoloba but containing a cyclized geranyl side chain [22]. Compound 10 was named lespebicoumestan A. Compounds 9 and 10 had a 3 -methyl-3 -isohexenylpyran ring (E) and, hence, an asymmetric center at C-3 . Although lespedezol A 3 (9) had previously been isolated from L. homoloba [21], the absolute configuration of the asymmetric center at C-3 had not yet been determined. In order to determine the absolute configuration of the asymmetric center at C-3 in compounds 9 and 10, we compared their calculated theoretical ECD spectra with corresponding experimental ECD data. We used the cam-B3LYP exchange-correlation functional set [33] along with the 6-311G(d) basis set and polarization continuum model (PCM) [31], implemented in the Gaussian 16 suit of programs [32], to calculate the energies and rotatory strengths of vertical electronic transitions. First, conformational analysis was performed for each compound at the B3LYP/6-311G(d)_PCM level of theoretical modeling; optimized geometries and relative Gibbs free energies as well as statistical weights were thus obtained.
The choice of the cam-B3LYP functional model for excited states is approved because it accounts for long-range interactions, to some extent better than the B3LYP functional model. Finally, statistically averaged ECD spectra were obtained as a weighted superposition of Gaussian-type functions and chosen for simulation of the ∆ε i (λ) function shapes for individual electronic transitions. The same value of σ = 0.24 eV for the bandwidths at 1/e peak heights was used. One hundred fifteen excited states were calculated for each conformation.
A comparison of the experimental and theoretical ECD spectra obtained for 9 and 10 is presented in Figure 2a,b. The positions and relative intensities of individual bands in the characteristic region 200 ≤ λ ≤ 300 nm are well-reproduced. The discrepancies in properties of ∆ε calc (λ) and ∆ε exp (λ) occur in the long-wave region λ ≥ 300 nm, caused to some extent by underestimation of the contribution to the ∆ε calc (λ) from the E + conformations (Schemes S60-S61, Figure S62, Supplementary data). Thus, we performed several simulations of the averaged ECD spectrum for 9, in which relative amounts of the E − and E + conformations were varied manually. We found that the shapes, positions, and relative intensities of individual bands in the characteristic region 200 ≤ λ ≤ 300 nm are refractory to these variations, while they change dramatically in the λ ≥ 300 nm region Schemes S60-S61, Figure S62 (S61-S62, Supplementary data). We previously observed analogous behavior for compound 2 [24].
The good qualitative coincidence between theoretical and experimental ECD spectra allowed us to determine the absolute configuration of the asymmetric center at C-3 in compounds 9 and 10 as C3 -S.
We obtained compound 11 as a yellow, amorphous powder. We elucidated the structure of the new compound using extensive spectroscopic analyses. The molecular formula C 29  .2377) in the HR-ESI-MS spectrum of 11. The 13 C NMR spectrum of 11 showed the presence of 12 carbon atoms of two aromatic rings, 10 carbon atoms of a geranyl side chain, 5 carbon atoms of an isoprenyl side chain, and 2 carbon atoms of two carbonyl groups. The 1 H NMR spectrum of 11 exhibited resonances at δ H 6.44 (1H, s) and 7.18 (1H, s) of protons H-3 and H-6 , respectively (ring B), and a singlet at δ H 6.84 (1H) was attributed to H-8 (ring A) ( Table 2). The downfield-shifted chemical shift values of hydroxyl groups OH-4 and OH-2 at δ H 11.92 (1H, s) and 11.59 (1H, s) confirmed that they formed hydrogen bonds with carbonyl groups at C-2 and C-1, respectively. The signals of the geranyl side chain protons were observed at δ  Table 2). The geranyl substituent was determined to be located at C-5 because in the HMBC spectrum of 11 we observed cross-peaks between the proton signal at δ H 3.51 (1H, d, J = 7.3 Hz) of H-1 and the C-4, C-5, and C-6 carbon signals at δ C 158.9, 115.3, and 152.5, respectively ( Table 2). The signals of the isoprenyl side chain protons were observed at δ H 3.23 (2H, d, J = 7.2 Hz, H-1 ), 5.20 (1H, t, J = 7.2 Hz, H-2 ), 1.73 (3H, s, H-4 ), and 1.71 (3H, s, H-5 ) ( Table 2). The isoprenyl substituent was determined to be located at C-5 because in the HMBC spectrum of 11 we observed cross-peaks between the proton signal at δ H 3.23 (1H, d, J = 7.2 Hz) of H-1 and the C-4 , C-5 , and C-6 carbon signals at δ C 163.7, 119.9, and 133.9, respectively (Table 2). Thus, compound 11 was shown to be a stilbenoid, and the structure of 11 differed from that of bicoloketone (5) only by the presence of an additional isoprenyl side chain at C-5 . Compound 11 was named 5 -isoprenylbicoloketone.

Antioxidant Activity of Polyphenolic Compounds
The data on antioxidant activity (DPPH • -scavenging effect and FRAP assay) of compounds 1-6 have been previously published in [24]. Here, we evaluated the DPPH •scavenging effect and FRAP of polyphenolic compounds 7-11 isolated from L. bicolor. In the FRAP assay, polyphenolic antioxidants reduced the light blue Fe 3+ -TPTZ complex to the dark blue Fe 2+ -TPTZ complex. The change in color resulted in an increase in absorbance at 595 nm. The results of both tests are shown in Table 3. 19.4 ± 2.2 ** 1.45 ± 0.08 ** 1 Data are presented as the mean ± SEM, n = 3. *** p < 0.001, ** p < 0.005, and * p < 0.05 compared to quercetin. 2 Data for compounds 1-6 have been previously published in [24]. Compounds 1-11 exhibited a moderate DPPH-scavenging effect, which was comparable to the effect of ascorbic acid but smaller than that of quercetin (Table 3). Lespedezol A 2 (4) and lespedezol A 3 (9) possessed the most-significant DPPH-scavenging effect and FRAP among compounds 1-11 (Table 3) [24]. In the FRAP assay, all tested polyphenolics also showed moderate effects but were less active than quercetin and ascorbic acid. Lespedezol A 2 (4) was the most effective in the FRAP assay.

Influence on Viability and ROS Level in PQ-Treated Neuro-2a Cells
The percentage of living cells after treatment with neurotoxin was assessed by the MTT test. The percentage of living cells after treatment with PQ significantly increased when polyphenolic compounds 3, 5, 6, and 8 were added (Figure 3a-c). The neuroprotective effect of compound 6, measured by the MTT assay, was detected in the concentration range of 0.01-1 µM, and at a concentration of 1 µM this compound increased the percentage of living cells by 17%. Compound 3 increased the percentage of living cells after treatment with PQ by 8% at a concentration of 10 µM in the MTT assay. Compounds 5 and 8 increased the percentage of living cells after PQ treatment by 8-10% at concentrations of 1 and 10 µM in the MTT assay. Compounds 1, 2, 4, and 9-11 did not exhibit any effect on PQ-treated Neuro-2a cell viability in this test. Polyphenolics from L. bicolor did not significantly reduce ROS levels in PQ-treated cells (Figure 3d).

Mitochondrial Membrane Potential (MMP) Detection
We studied the effect of polyphenolic compounds from L. bicolor on PQ-induced mitochondrial dysfunction. The 23% decrease in tetramethylrhodamine methyl (TMRM) fluorescence after a 1 h exposure of Neuro-2a cells with PQ indicates that PQ causes depolarization of the mitochondrial membrane (Figure 3e). Among the tested compounds, pterocarpan 3 and stilbenoid 5 at a concentration of 10 µM were the most effective in this assay and increased the value of mitochondrial membrane potential by 16% and 23%, respectively.

Influence on Viability and ROS Levels in 6-OHDA-Treated Neuro-2a Cells
We evaluated the effects of the polyphenolic-compound set on cell viability in a 6-OHDA-induced neurotoxicity model. The percentage of living Neuro-2a cells after 6-OHDA treatment increased from 45% to 65% in the presence of polyphenolic compounds, compared with the control (Figure 4a-c). Pre-treatment of cells with the test compounds for 1 h before 6-OHDA addition provided an increase in the percentage of living cells with varying levels of statistical confidence.

Mitochondrial Membrane Potential (MMP) Detection
We studied the effect of polyphenolic compounds from L. bicolor on PQ-induced mitochondrial dysfunction. The 23% decrease in tetramethylrhodamine methyl (TMRM) fluorescence after a 1 h exposure of Neuro-2a cells with PQ indicates that PQ causes depolarization of the mitochondrial membrane (Figure 3e). Among the tested compounds, pterocarpan 3 and stilbenoid 5 at a concentration of 10 µM were the most effective in this assay and increased the value of mitochondrial membrane potential by 16% and 23%, respectively.

Influence on Viability and ROS Levels in 6-OHDA-Treated Neuro-2a Cells
We evaluated the effects of the polyphenolic-compound set on cell viability in a 6-OHDA-induced neurotoxicity model. The percentage of living Neuro-2a cells after 6-OHDA treatment increased from 45% to 65% in the presence of polyphenolic compounds, compared with the control (Figure 4a-c). Pre-treatment of cells with the test compounds for 1 h before 6-OHDA addition provided an increase in the percentage of living cells with varying levels of statistical confidence.

Influence on Viability and ROS Levels in Rotenone-Treated Neuro-2a Cells
We examined the effect of polyphenolic compounds from L. bicolor on cell viability in a rotenone-induced neurotoxicity model. The percentage of living Nuero-2a cells treated with rotenone was 68% compared to the control (Figure 5a-c). Pre-treatment of cells with polyphenolic compounds from L. bicolor for 1 h before rotenone addition provided an increase in the percentage of living Neuro-2a cells with different levels of statistical confidence.
Compounds 9 and 10 increased the percentage of living Neuro-2a cells after treatment with rotenone by 10.4% and 13.2%, respectively, compared to rotenone-treated cells (p < 0.05). The other compounds did not show significant improvement in this assay (Figure 3a-d).

Influence on Viability and ROS Levels in Rotenone-Treated Neuro-2a Cells
We examined the effect of polyphenolic compounds from L. bicolor on cell viability in a rotenone-induced neurotoxicity model. The percentage of living Nuero-2a cells treated with rotenone was 68% compared to the control (Figure 5a-c). Pre-treatment of cells with polyphenolic compounds from L. bicolor for 1 h before rotenone addition provided an increase in the percentage of living Neuro-2a cells with different levels of statistical confidence.
Compounds 9 and 10 increased the percentage of living Neuro-2a cells after treatment with rotenone by 10.4% and 13.2%, respectively, compared to rotenone-treated cells (p < 0.05). The other compounds did not show significant improvement in this assay (Figure 3a-d). (c) (d) Figure 5. The influence of polyphenolic compounds from L. bicolor on cell viability (a-c) and ROS levels (d) in Neuro-2a cells treated with rotenone (10 µM). The percentage of living cells treated with compounds and rotenone was measured by the MTT assay. Each bar represents the mean ± SEM of three independent replicates. (*), (**), and (***) indicate, respectively, p < 0.05, p < 0.005, and p < 0.001 versus rotenone-treated cells. The difference between control and rotenone-treated cells was considered significant.

Discussion
We continued to study the chemical composition of polyphenolic compounds from L. bicolor root bark and isolated a new coumestan, lespebicoumestan A (10), and a stilbenoid, 5′-isoprenylbicoloketon (11), as well as previously known pterocarpans 1-3, and 6; pterocarpens 4 and 9; coumestan 7; stilbenoid 5; and dimeric flavonoid 8 (Figure 1). There was a considerable difference between the chemical compositions of polyphenolic compounds of L. bicolor growing in the Primorskiy region (Russian Far East) and in the Republic of Korea. Lee P.J. and coworkers isolated from L. bicolor 11 pterocarpans, 2 coumestans, and 2 arylbenzofurans with isoprenyl and geranyl substituents in their structures [25,27,28]. These compounds contained a methoxy group at C-1, whereas the C-1 position in pterocarpans and coumestans from stem bark and root bark of L. bicolor collected in the Primorskiy region was unsubstituted [24,26]. Besides, some pterorarpans and coumestans from L. bicolor growing in the Russian Far East had a 3′-methyl-3′-isohexenylpyran ring (E), presumably formed by oxidative cyclization of a geranyl side chain. However, pterocarpans and coumestans from L. bicolor growing in the Republic of Korea

Discussion
We continued to study the chemical composition of polyphenolic compounds from L. bicolor root bark and isolated a new coumestan, lespebicoumestan A (10), and a stilbenoid, 5 -isoprenylbicoloketon (11), as well as previously known pterocarpans 1-3, and 6; pterocarpens 4 and 9; coumestan 7; stilbenoid 5; and dimeric flavonoid 8 (Figure 1). There was a considerable difference between the chemical compositions of polyphenolic compounds of L. bicolor growing in the Primorskiy region (Russian Far East) and in the Republic of Korea. Lee P.J. and coworkers isolated from L. bicolor 11 pterocarpans, 2 coumestans, and 2 arylbenzofurans with isoprenyl and geranyl substituents in their structures [25,27,28]. These compounds contained a methoxy group at C-1, whereas the C-1 position in pterocarpans and coumestans from stem bark and root bark of L. bicolor collected in the Primorskiy region was unsubstituted [24,26]. Besides, some pterorarpans and coumestans from L. bicolor growing in the Russian Far East had a 3 -methyl-3 -isohexenylpyran ring (E), presumably formed by oxidative cyclization of a geranyl side chain. However, pterocarpans and coumestans from L. bicolor growing in the Republic of Korea contained a 3 ,3 -dimethylpyran ring (E) produced by oxidative cyclization of an isoprenyl side chain. In contrast to Far-Eastern L. bicolor, Korean L. bicolor did not contain stilbenoids and dimeric flavonoids.
Being natural antioxidants, pterocarpans and coumestans are promising candidates for the study of neuroprotective properties [10,11,25]. Nerve cells treated with various inducers of oxidative stress (PQ, 6-OHDA, rotenone) are often used as one of the generally accepted models for studying neurotoxic disorders, including PD [35,36].
In our study, pterocarpans 3 and 6, stilbenoid 5, and dimeric flavonoid 8 significantly increased the percentage of living Neuro-2a cells after treatment with PQ, but only pterocarpan 6 slightly decreased the ROS level in PQ-treated cells, which is in accordance with its rather high activity in the DPPH • and FRAP tests [24]. It is known that the effect of PQ on neurons is accompanied by impaired functioning of mitochondria due to changes in mitochondrial membrane permeability, membrane potential, and depolarization of mitochondrial membranes [37]. We presumed that these compounds could also increase cell viability by preventing depolarization of the mitochondrial membrane. In fact, pterocarpan 3 and stilbenoid 5 were shown to effectively increase the mitochondrial membrane potential of PQ-treated neuronal cells.
We also examined the effects of polyphenolic compounds on cell viability in a 6-OHDA-induced neurotoxicity model. We showed that pterocarpans 2 and 3, containing a 3 -methyl-3 -isohexenylpyran ring (E); pterocarpens 4 and 9, with a double bond between C-6a and C-11a; and lespebicoumestan A (10) significantly increased the percentage of living Neuro-2a cells and decreased ROS levels after treatment with 6-OHDA. Notably, compounds 9 and 10 both contain a 3 -methyl-3 -isohexenylpyran ring (E) and a double bond between C-6a and C-11a. Pterocarpans 2 and 3 were the most active in this assay and decreased the ROS level 4.5 times compared to 6-OHDA-treated cells. Notably, polyphenolic compounds from L. bicolor decreased the level of intracellular ROS much more effectively than quercetin (Figure 4d). Thus, pterocarpans and coumestans with an additional 3 -methyl-3 -isohexenylpyran ring demonstrated the most-significant activity.
We also studied the effect of polyphenolic compounds from L. bicolor on cell viability in a rotenone-induced neurotoxicity model. Pre-treatment of cells with polyphenolic compounds from L. bicolor before rotenone addition resulted in an increase in the percentage of living Neuro-2a cells, with compounds 9 and 10 being the most active. The presence of a 3 -methyl-3 -isohexenyl pyran ring (E) and a double bond between C-6a and C-11a in 9 and 10 may be responsible for the increase in cell viability. The other compounds did not show significant improvement of cell viability in this assay.
Previously, researchers from the College of Pharmacy, the Research Institute of Pharmaceutical Science and Technology (The Republic of Korea), and the Korea Research Institute of Standards and Science demonstrated that pterocarpan-type compounds (1methoxylespeflorin G11, bicolosin A, 1-methoxyerythrabyssin II, 8-methoxybicolosin C, 2-geranyl-1-methoxylespeflorin G11, and 2-geranylbicolosin A) exhibited significant neuroprotective effects against glutamate neurotoxicity in neuronal HT22 hippocampal cells [25]. The compound 2-geranyl-1-methoxylespeflorin G11-which has a geranyl group at C-2, a prenyl group at C-10, and a methyl group at C-8-was shown to attenuate apoptosis in HT22 cells by inhibiting intracellular ROS generation and mitochondrial dysfunction. Although in our study coumestan 10 effectively increased the percentage of living Neuro-2a cells after treatment with 6-OHDA-and rotenone, arylbenzofurans and coumestan isolated from Korean L. bicolor exhibited no protective effects [25].
Considering that isoflavonoids can quickly penetrate the blood-brain barrier [38], such compounds may be prospective agents in the treatment of PD.

Conclusions
Thus, pterocarpans 2, 3, and 6 and pterocarpens 4 and 9, as well as coumestan 10 from L. bicolor effectively protected PQ-and 6-OHDA-treated Neuro-2a cells from oxidative stress. The effect of polyphenolic compounds 3 and 5 from L. bicolor is mainly due to their ability to impair PQ-induced depolarization of the mitochondrial membrane, whereas compounds 2-4, 9, and 10 decreased ROS levels in 6-OHDA-treated Neuro-2a cells more effectively than quercetin. The effect of polyphenolic compounds on the viability of rotenone-treated cells was less dramatic.  Figure S58: The geometrical structure and atom numeration for compounds 9 and 10; Scheme S59: Abbreviations for conformations of compounds 9 and 10; Scheme S60: The influence of inversion of ring B on the geometrical structure of compound 9; Scheme S61: Equation for recalculation of the relative contributions of the E − and E + conformations to the ECD spectra of compounds 9 and 10; Figure S62: The influence of variation in the relative amounts of E − and E + conformations on the shape of the statistically averaged scaled ECD spectrum of compound 9.