Absolute configuration of naturally occurring glabridin

Glabridin is a species-specific biomarker from the roots Glycyrrhiza glabra L. (European licorice, Fabaceae). Stereochemical analysis, including circular dichroism, NMR data and an X-ray diffraction data set with Bijvoet differences, confirms that glabridin, purified from its natural source, is found only in a C3 R configuration.

Abstract

The title compound {systematic name: 4-[(3R)-8,8-dimethyl-3,4-di­hydro-2H-pyrano[2,3-f]chromen-3-yl]benzene-1,3-diol, commonly named glabridin}, C₂₀H₂₀O₄, is a species-specific biomarker from the roots Glycyrrhiza glabra L. (European licorice, Fabaceae). In the present study, this prenylated isoflavan has been purified from an enriched CHCl₃ fraction of the extract of the root, using three steps of medium-pressure liquid chromatography (MPLC) by employing HW-40F, Sephadex LH-20 and LiChroCN as adsorbents. Pure glabridin was crystallized from an MeOH–H₂O mixture (95:5 v/v) to yield colorless crystals containing one mol­ecule per asymmetric unit (Z′ = 1) in the space group P2₁2₁2₁. Although the crystal structure has been reported before, the determination of the absolute configuration remained uncertain. Stereochemical analysis, including circular dichroism, NMR data and an X-ray diffraction data set with Bijvoet differences, confirms that glabridin, purified from its natural source, is found only in a C3 R configuration. These results can therefore be used as a reference for the assignment of the configuration and enantio­purity of any isolated or synthetic glabridin sample.

Introduction  

Glabridin, (I), is a prenylated isoflavan, which has previously been isolated from the roots of Glycyrrhiza glabra (Shibata & Saitoh, 1978 ▶; Zhang & Ye, 2009 ▶). Inter­estingly, glabridin is one of the most studied licorice flavonoids, and has been widely considered to be a phytoestrogen with numerous biological activities including anti­oxidant (Vaya et al., 1997 ▶), anti­bacterial (Fukai et al., 2002 ▶), neuroprotective (Cui et al., 2008 ▶) and potential antitumorogenic properties (Tsai et al., 2010 ▶; Hsu et al., 2011 ▶), as well as antinephritic, anti­bacterial and skin-whitening activities (Simmler et al., 2013 ▶ a). The structure of glabridin was first characterized in 1976 (Saitoh et al., 1976 ▶) and subsequently more fully investigated by various spectroscopic methods (Kinoshita et al., 1996 ▶; Kim et al., 2009 ▶). The chemical synthesis was reported in 2007 (Yoo & Nahm, 2007 ▶) and the X-ray crystal structure at 293 K was reported last year (Tanti­shaiyakul et al., 2012 ▶). However, the determination of the absolute stereochemistry and enantio­purity of natural glabridin remains uncertain.

Experimental  

Synthesis and crystallization  

The CHCl₃ extract (4.97 g) from G. glabra roots was fractionated on an HW-40F column (2.5 × 60 cm) with 100% MeOH elution at a flow rate of 2.5 ml min⁻¹, to yield seven fractions. The fourth fraction (1.92 g) was further separated on a Sephadex LH-20 column (1.1 × 9 m) eluted with 100% MeOH at a flow rate of 1 ml min⁻¹, to give a glabridin-enriched fraction (234 mg). The final purification step was carried out on a LiChroCN (1.1 × 30 cm) column using an isocratic elution with CHCl₃–MeOH (95:5 v/v). After drying under vacuum, the purified glabridin (10 mg), (I), was dissolved in MeOH–H₂O (2.5 ml, 95:5 v/v) and left in a round-bottomed flask at room temperature for 24 h in order to obtain clear colorless crystals.

The ¹H NMR spectra (600 MHz in DMSO-d ₆ and in CDCl₃ at 10 mM) were acquired under quanti­tative conditions and recorded on a Bruker AVANCE 600 MHz spectrometer equipped with a 5 mm TXI cryoprobe. Off-line data processing was performed using the TOPSPIN 3.0.b.8 software package (Bruker BioSpin GmbH), and PERCH NMR software (Laatikainen et al., 1996 ▶) was employed for the full spin-system analysis. The NMR and circular dichroism (CD; 230–330 nm, in aceto­nitrile at 50 µM) data obtained for this secondary metabolite were in accordance with previously published data (Kinoshita et al., 1996 ▶; Kim et al., 2009 ▶; Yoo & Nahm, 2007 ▶). Chiral one-dimensional ¹H NMR was performed by adding a D₂O solution (20 mM) of a chiral reagent {sodium [(R)-1,2-di­amino­propane-N,N,N′,N′-tetra­acetato]­samarate(III) hydrate; TCI Co. Ltd} to the crystalline sample dissolved in DMSO-d ₆ (12 mM). Spectra were acquired at 600 MHz. Chiral ¹H NMR data processing was done following a Lorentzian-to-Gaussian modification (LB = −2.5 Hz, GF = 0.06), as reported previously (Jaki et al., 2004 ▶). If the sample were to contain two enantio­mers, a split in the proton resonance near the chiral center would be observed. The signal split should be all the more significant as the proportion of chiral reagent in the NMR sample increases. The chiral NMR method was previously assessed using a racemic sample of liquiritigenin (Part No. 00012290; Chromadex Inc.), from which the recorded CD spectrum (40 µM in aceto­nitrile) gave a flat line: 25 µl of samarate solution at 20 mM in D₂O added to 200 µl of a liquiritigenin solution at 10 mM in DMSO-d ₆ led to splitting of the signals from protons H2 (chiral center) and H3 (adjacent to the chiral center).

Chiral high-performance liquid chromatography (HPLC) was carried out on different types of stationary phases: permethyl­ated B cyclo­dextrins (Nucleodex β-OH, and α-, β- and γ-PM; Macherey Nagel, GmbH & Co.); sodium magnesium silicate particles and anion exchange with ruthenium (Ceramospher RU-2; Shiseido Co. Ltd); and derivatized cellulose or amylose tris­(3,5-dimenthylphenyl­carbamate) (Chiralpak 1B and 1A, respectively; Chiral Technolgies, Inc.). Different elution conditions using various solvent mixtures, such as MeOH–H₂O, ACN–H₂O and n-hexane–EtOH, and various flow rates from 0.7 to 1 ml min⁻¹, were also tested according to the type of stationary phase.

Several crystals were selected and screened, and complete X-ray data sets were collected on the LS-CAT beamline 21-ID-D (APS, Argonne, USA) at 0.7 Å to determine crystal quality. One small colorless crystal was selected for further data collection. An Ewald sphere of intensity data was collected and averaged to a quadrant in the Bravais lattice (Pmmm). The final unit-cell parameters were obtained from refinement using 9904 reflections to a maximum Bragg angle of 68.81°.

Refinement  

Crystal data, data collection and structure refinement details are summarized in Table 1 ▶. The space group was identified as P2₁2₁2₁ based upon systematic absences and intensity statistics, with a final total of 2879 unique observations. The positions of the H atoms were evident in a difference electron-density map, and were then freely refined, with U _iso(H) = 1.5U _eq(parent atom) for hy­droxy and methyl H atoms or 1.2U _eq(parent atom) otherwise. The structure has been deposited with the Cambridge Crystallographic Data Centre (deposition No. 918747).

Table 1. Experimental details.

Crystal data
Chemical formula	C20H20O4
M r	324.36
Crystal system, space group	Orthorhombic, P212121
Temperature (K)	92
a, b, c (Å)	6.3744 (10), 11.961 (2), 21.041 (3)
V (Å3)	1604.3 (4)
Z	4
Radiation type	Cu Kα
μ (mm−1)	0.76
Crystal size (mm)	0.15 × 0.05 × 0.03
Data collection
Diffractometer	Bruker Kappa APEXII DUO CCD area-detector diffractometer
Absorption correction	Multi-scan (SADABS; Sheldrick, 2004 ▶)
T min, T max	0.895, 0.978
No. of measured, independent and observed [I > 2σ(I)] reflections	47918, 2879, 2708
R int	0.046
(sin θ/λ)max (Å−1)	0.603
Refinement
R[F 2 > 2σ(F 2)], wR(F 2), S	0.026, 0.061, 1.03
No. of reflections	2879
No. of parameters	277
H-atom treatment	Only H-atom coordinates refined
Δρmax, Δρmin (e Å−3)	0.13, −0.13
Absolute structure	Flack x parameter determined using 1096 quotients [(I +) − (I −)]/[(I +) + (I −)] (Parsons et al., 2013 ▶)
Absolute structure parameter	−0.03 (6)

Computer programs: APEX2 (Bruker, 2007 ▶), SAINT (Bruker, 2007 ▶), SHELXS97 (Sheldrick, 2008 ▶), SHELXL2013 (Sheldrick, 2008 ▶), ORTEP-3 for Windows (Farrugia, 2012 ▶) and publCIF (Westrip, 2010 ▶).

Results and discussion  

After the final step of the medium-pressure liquid chromatography (MPLC) purification process, natural glabridin was crystallized from a mixture of methanol and water to give clear colorless crystals of (I) containing one mol­ecule per asymmetric unit (Z′ = 1) in the space group P2₁2₁2₁ (Fig. 1 ▶). A comparison of bond lengths obtained for this and the previous determination is given in the Supplementary materials. An inter­molecular hydrogen bond is observed between atom O1B (O1′′ according to the biogenetic labeling) and atom H4A on atom O4A (O4′) of an adjacent mol­ecule, with an O—H⋯O angle of 178 (2)° (Table 2 ▶). In addition, it has been reported (Tantishaiyakul et al., 2012 ▶) that there is a putative O—H⋯π inter­action between the hy­droxy group on atom O2A and C1A–C6A aromatic ring of a neighboring mol­ecule (Table 2 ▶).

Figure 1.

The hydrogen-bonding scheme of glabridin in the unit cell. Dashed lines indicate hydrogen bonds.

Table 2. Hydrogen-bond geometry (Å, °).

CgA is the centroid of the C1A–C6A ring.

D—H⋯A	D—H	H⋯A	D⋯A	D—H⋯A
O4A—H4A⋯O1B i	0.86 (3)	1.96 (3)	2.8214 (19)	178 (2)
O2A—H2A⋯CgA ii	0.84 (3)	2.46 (2)	3.1505 (16)	139 (2)

Symmetry codes: (i) ; (ii) .

The one-dimensional ¹H NMR spectra (600 MHz, in DMSO-d ₆ and CDCl₃; see Supplementary materials) obtained for the present sample of glabridin are in accordance with previously published data (Kinoshita et al., 1996 ▶; Kim et al., 2009 ▶). The two-dimensional NMR spectra (HSQC and HMBC recorded at 600 MHz in DMSO-d ₆) further confirmed the structural assignments. Additionally, the enantio­purity of the crystalline sample has been evaluated by means of chiral HPLC, using different types of stationary phases and elution systems. Under all of these conditions, the glabridin sample eluted only as a single enantio­mer. Moreover, chiral ¹H NMR in DMSO-d ₆ was also performed using a chiral water-soluble reagent (samarate). Even after adding 50% molar concentration of the samarate, no splitting of glabridin ¹H NMR signals was observed, suggesting that only one enantio­mer was present in the NMR tube. Taken together, these observations suggest that the collected sample was enantio­pure.

The stereochemistry of glabridin was determined using Cu Kα radiation and by carefully measuring a large number of Bijvoet differences. We assert that the isolated glabridin crystallized as a pure C3 R enantio­mer (Fig. 2 ▶). The Flack parameter (Flack, 1983 ▶) refined to −0.14 (19) for the ‘hole-in-one’ fit, and to −0.03 (6) for the 1096 selected quotients from Parsons’ method (Parsons et al., 2013 ▶). PLATON (Spek, 2009 ▶) was also used to analyze the 1096 Bijvoet differences (Hooft et al., 2008 ▶) and suggested that the glabridin structure is enantio­pure: the probability P2 (true) = P3 (true) = 1.000 with P3(racemate-twin) = 5.0 × 10⁻²⁷, P3 (false) = 1.0 × 10⁻⁹⁷, G = 1.09 (10), and the Hooft parameter y = −0.05 (5) for the C3 R enantio­mer. CRYSTALS (Betteridge et al., 2003 ▶) was also used to refine the structure. It reported a Flack parameter of −0.13 (16), a G parameter of 1.03 (12) and a Hooft parameter of −0.01 (6).

Figure 2.

The mol­ecular structure of glabridin, showing the atom-labeling scheme according to the crystallographic coordinates. Displacement ellipsoids are drawn at the 50% probability level. The biogenetic labeling scheme relates to the crystallographic atom labels as follows: ‘A’ corresponds to primed (′) labels and ‘B’ corresponds to double primed (′′) labels, e.g. C3A is C3′ and C3B is C3′′.

The determination of absolute structure, and the relative uncertainties in the measurement of the intensities of the Friedel pairs, has been revisited (Flack, 2013 ▶; Parsons et al., 2012 ▶; Flack et al., 2011 ▶). In particular, it has been noted that the average of the observed intensities between Friedel pairs, 2A _o = |F _o(hkl)²| + |F _o()²|, can be in good agreement with the corresponding average of the calculated intensities of the model, 2A _c = |F _c(hkl)²| + |F _c()²|, but the differences between the observed intensities of the Friedel pairs, D _o = |F _o(hkl)²| − |F _o()²|, can be overwhelmed by random and systematic errors in the measurements when compared with the corresponding differences between the calculated intensities of the model, D _c = |F _c(hkl)²| − |F _c()²| (Flack, 2013 ▶). Ideally, the plots of 2A _o versus 2A _c and of D _o versus D _c would each yield linear fits with slopes of 1 and inter­cepts at zero. From Fig. 3 ▶(a), we see that 2A _o is in good agreement with 2A _c, with a slope near 1, but this is clearly not true for Fig. 3 ▶(b) with D _o plotted against D _c. However, the slope of the linear fit in Fig. 3 ▶(b) is near 0.2, suggesting a weak resonant-scattering contribution and the correct absolute structure assignment. Indeed, a plot of D _o versus D _c for D _o > 4σ results in a slope near 0.2 with a correlation coefficient of 0.77 (Fig. 3 ▶ c). Likewise, various assessment factors employing A and D can be defined to ascertain the magnitude of the resonant-scattering contributions to the Friedel pair differences: R_A = Σ|2A _o − 2A _c|/Σ|2A _o| = 3.5%, R_D = Σ|D _o − D _c|/Σ|D _o| = 92.0%, R_AD = Σ|D _o|/Σ|A _o| = 1.5% and 〈D _o/σ〉 = (1/n)Σ|D _o|/σ = 1.9. These assessment factors suggest that the resonant-scattering contributions to the observed intensities are weak, but still significant enough to assign the absolute structure correctly.

Figure 3.

2A and D plots of (a) 2A _o versus 2A _c, (b) D _o versus D _c for all data and (c) D _o versus D _c for all Friedel pairs with D _o > 4σ.

Circular dichroism (CD) of flavonoids can indicate the configuration of a sample regardless of its enantio­purity (Slade et al., 2005 ▶). If a sample is defined by an enantio­meric excess, its CD spectrum will only indicate the configuration of the enantio­mer present in excess (Simmler et al., 2013 ▶ b). Given that the enantio­purity and stereochemistry of the present glabridin sample have been determined by means of complementary techniques (chiral HPLC, NMR and X-ray diffraction), the CD spectrum recorded from the crystalline sample (50 µM in aceto­nitrile) displays the mol­ecular ellipticity of an enantio­pure glabridin in a C3 R configuration (Kim et al., 2009 ▶) (Fig. 4 ▶).

Figure 4.

The CD spectrum of C3 R enantio­pure glabridin at 50 µM in aceto­nitrile.

Conclusions  

The present description of the crystal structure of glabridin complements existing spectroscopic data, and confirms that glabridin naturally occurs in the C3 R configuration. From this stereochemical point of view, it is important to consider that all the biological activities reported for natural glabridin are likely obtained from the R enantio­mer. Therefore, the reported CD spectrum can be used as a reference for further characterization and elucidation of the enantio­purity of an isolated or synthetic glabridin sample. The results presented herein provide an unequivocal foundation for the structural characterization of this phytochemical and biological marker of European licorice, G. glabra L.