Pigments of the Moss Paraleucobryum longifolium: Isolation and Structure Elucidation of Prenyl-Substituted 8,8′-Linked 9,10-Phenanthrenequinone Dimers

In a search for new secondary metabolites from mosses, leucobryns A–E, axially chiral 9,10-phenanthrenequinone dimers, were isolated from Paraleucobryum longifolium (1–5), together with diosmetin triglycoside. Leucobryns B (2) and C (3) were proved to be homodimeric atropodiastereomers containing both axial and central chirality elements, while leucobryns D (4) and E (5) were found to be heterodimeric atropodiastereomers containing central chirality in only one of the two monomeric units. Axial chirality of the compounds was determined by ECD measurements and sTDA ECD calculations, while the central chirality elements were assigned by TDDFT-SOR calculations. Leucobryns represent the first 9,10-phenanthrenequinone dimers, the monomers of which are linked through their C-8 atoms. Leucobryns B–E contain an uncommon C10 monoterpenoid side chain, in which isoprenoid units are joined by 3,4 linkages. Leucobryns A and B exhibited weak antiproliferative activity against several human cancer cell lines.

■ RESULTS AND DISCUSSION P. longifolium was extracted with MeOH, and the evaporated extract was subjected to column chromatography (CC) on polyamide. Fractions eluted with 80% MeOH displayed a strong violet color. Dark violet compounds 1−5 were isolated using vacuum liquid chromatography, gel filtration on Sephadex LH-20, and RP-HPLC.
Compound 1 was obtained as a dark violet amorphous powder with UV maxima at 542.5, 370, 305, 272, and 213.5 nm. The molecular formula was established as C 40  resonances, which indicated that compound 1 is a symmetric dimer ( Figure 1).

Journal of Natural Products
Article with O-functionalities at C-2 (δ C 155.84), C-5 (δ C 156.81), and C-7 (δ C 157.65) and C-functionalities at C-1 (δ C 133.91) and C-8 (δ C 121.19) (Figure 2). The alternative anthraquinone structure was ruled out because the strong HMBC crosspeak observed between H-4 and C-5a suggested that they are separated by three rather than four bonds. The other three-bond correlations of H-4 with C-2 and C-1a had similar intensities. On the other hand, a weak four-bond heteronuclear interaction was detected between H-4 and the carbonyl carbon at δ C-10 192.77. Furthermore, the proposed structure was confirmed by a weak cross-peak exhibited between H-6 and C-4a, which would not be expected in an anthraquinone, due to the long distance (5 bonds) between these atoms.
The presence of the methoxy group at C-7 was established by HMBC correlations between the OCH 3 protons and C-7 and C-6 and by a NOESY cross-peak between H-6 and the OCH 3 group. A 3-hydroxyisopentyl substituent was corroborated with the HMBC cross-peaks observed between the methyl protons at δ H 1.28 (H 3 -14, H 3 -15) and C-13 (δ C 71.98) and C-12 (δ C 44.21); this group was placed at C-1 on the basis of the H-11/C-1, H-11/C-1a, H-11/C-2, and H-12/C-1 HMBC correlations. A hydroxy group, concluded from the HRESIMS data, is attached at C-7, as confirmed by its chemical shift value of δ C 157.65. In accordance with all these data, the structure of compound 1, named leucobryn A, could be established as a dimer in which the monomers are connected via their C-8 atoms as depicted in Figure 1.
Compound 2 possessed a molecular formula of C 50 13 C JMOD spectra revealed a structure containing 25 carbon atoms only, suggesting a symmetric dimeric structure, similarly to compound 1. The 1 H NMR and 13 C JMOD spectra of 2 also resembled those of 1. In particular, the 1 H and 13 C chemical shifts of the 9,10-phenanthrenedione part were close to those of 1 (Tables 1 and 2). The main differences were found in the C 5 prenyl side chain, which was changed to a C 10 moiety in 2. Combined analysis of 1 H NMR, 13 C JMOD, and HSQC spectra proved that this C 10 substituent consists of a nonprotonated, oxygenated carbon (δ C 71. 46 28.88, 29.30). Based on the 1 H− 1 H COSY spectrum, two structural fragments could be unambiguously assigned for the C 10 part: −CH 2 −CHR−CH 2 −CH 2 − (A) and CH 2 (CH 3 )C− (B) (Figure 3).
Compound 3 (leucobryn C) was isolated the same way as 2, and in the final step of the purification procedure compounds 2 and 3 were separated by RP-HPLC. The HRESIMS data of 3 displayed a sodium adduct ion at m/z 869.3507 [M + Na] + , indicating the same molecular formula of C 50 H 54 O 12 as that of compound 2. 1D and 2D 1 H/ 13 C NMR characteristics of 2 and 3 were similar, with only minute differences observed in the

Journal of Natural Products
Article chemical shifts of H-14, H-16, C-10, C-11, and C-14 (Tables 1  and 2). These data suggest that the structural differences are explicable in terms of the stereochemistry of the compounds. Compound 4 (leucobryn D) was isolated as a dark violet amorphous solid. HRESIMS data revealed a molecular formula of C 45 (Tables 1 and 2), indicating that compound 4 is a heterodimer constructed by the monomeric moieties of 1 and 2. As in compounds 1 and 2, the monomeric constituents of 4 are connected via their C-8 atoms. 2D NMR studies provided further evidence to confirm the structure of this compound as depicted in structural formula 4.
Compound 5 (leucobryn E) was separated from 4 by RP-HPLC using a gradient system of MeCN−H 2 O mixtures. All HRESIMS and NMR characteristics of compound 5 were highly similar to those of 4, suggesting a molecular pair with the same 2D structure, but with different chirality.

Journal of Natural Products
Article For the configurational assignment of the axial chirality elements of 1−5, conformational analysis and the sTDA (simplified Tamm−Dancoff approximation) approach 9 were applied on the arbitrarily chosen (aS,12R,12′R)-2 and (aR,12R,12′R)-3 stereoisomers differing in the axial chirality. When both central and axial chirality elements are present in biaryl compounds, the ECD spectrum is governed by the axial chirality element, which can be elucidated by ECD calculations. 10 The experimental ECD spectrum of 1, containing only an axial chirality element, was found to be similar to those of 2 and 4, but with weaker intensity, and nearly mirror-images of the ECD spectra of 3 and 5 were recorded, indicating that the central chirality elements of 2−5 do not significantly contribute to the ECD spectra (Figure 4a).
In axially chiral biaryls, the absolute configuration of the central chirality elements was determined by ECD analysis of the monomers co-isolated from the same source 11,12 or, alternatively, by VCD calculations. 13 The configurational assignment of both central and axial chirality elements in axially chiral natural biaryls is considered a challenging task by even the combination of different chiroptical methods and their computation. 10a,11a, 14 Owing to the substitution pattern in the ortho-positions of the biaryl axis, biaryls 1−5 were expected to have a rotational energy barrier exceeding 93.6 kJ/ mol, which is considered the minimum rotational energy barrier for the presence of separable atropisomers. 15 Biaryls 1−5 are expected to have a large number of ECD transitions to compute, which required the use of the sTDA method 9c,d developed for large molecules and/or for compounds with a large number of ECD transitions. This method was developed for complex molecules, the TDDFT calculation of which would be quite demanding computationally or not feasible at an advanced DFT level. Despite the strengths of this method, there are only a few examples for application in the field of natural products. 16,17 The initial Merck Molecular Force Field (MMFF) conformational search of (12R,12′R)-2/3 resulted in 1326 conformers in a 21 kJ/mol energy window, including both the (aS,12R,12′R)-2 and (aR,12R,12′R)-3 stereoisomers, which were separated manually. Subsequently, they were reoptimized at the AM1 semiempirical level to reduce the number of conformers 18 and then at the CAM-B3LYP/TZVP 19 PCM/MeCN level.
sTDA-ECD calculations were performed for conformers above 1% population at the CAM-B3LYP, LC-BLYP, 20 and ωB97X 21 levels, for which the required single-point calculations were performed with the same functionals and the TZVP basis set. All of the Boltzmann-averaged ECD spectra of (aS,12R,12′R)-2 and (aR,12R,12′R)-3 consistently gave moderate to good agreement with the experimental ECD spectrum of 2 and 3, respectively, and all the low-energy conformers of the same axial chirality exhibited similar ECD spectra, permitting the elucidation of the absolute configuration of the axial chirality as (aS) in 2 and (aR) in 3 ( Figure  4b and Figure 5).
Thus, compounds 2 and 3 are atropodiastereomers with identical absolute configuration of their stereogenic centers. Comparing the ECD spectra of 1, 4, and 5 to those of 2 and 3 revealed the absolute configuration of 1 and 4 as (aS) and that of 5 as (aR) (Figure 4a).
Since both the experimental and computed ECD spectra of (aS,12R,12′R)-2 and (aR,12R,12′R)-3 showed a mirror-image relationship, in principle the determination of the central chirality element was not feasible with the ECD spectra.  nm) falls in the higher-wavelength range of the highestwavelength ECD transition of 2 and 3, and they had the same signs as those of the corresponding ECD transition. 24 In order to check that the SR values of 2 and 3 recorded at 589.3 nm belong to the monotonously increasing wavelength range of the ORD spectrum without a change of sign (ORD anomaly) or they contain the effect of the ORD anomaly, SR values of (aS,12R,12′R)-2 and (aR,12R,12′R)-3 were calculated at 10 selected wavelengths in the range of 575−800 nm with four methods (Figures 6 and 7). These SR calculations consistently showed that there was no sign inversion of SR values in the 600−800 nm wavelength range, while anomalous ORD appears around the wavelength of the sodium D line. The ORD anomaly expected below 600 nm and the MMFF-AM1clustering-DFT reoptimization sequence for the conformational analysis to reproduce the biaryl dihedral angles and the side-chain conformation may expose some uncertainty in the determination of the central chirality. However, the computed SR values clearly support the conclusion, and characteristic differences in the magnitudes of the computed SR values for (aS,12R,12′R)-2 and (aR,12R,12′R)-3 appeared consistently at all the computed wavelengths. 25 The SR calculations of (aS,12R,12′R)-2 and (aR,12R,12′R)-3 represent the first example to determine the central chirality of a homodimeric biaryl natural product in the presence of axial chirality on the basis of the SR calculations. VCD measurement and calculations were reported earlier to determine both axial and central chirality for the biaryl natural product (aS,2R,2′R)-cephalochromin, the SR calculation of which could not decide on the central chirality. 13 However, the same VCD approach failed to determine the central chirality of flavomannin A in the presence of axial chirality. 10a SR calculations were also successfully utilized to determine the central chirality of quinolinophane derivatives containing both planar and central chirality elements. 26 The recent example of leucobryns B and C also demonstrated that the specific rotation values can contain stereochemical information, which are either missing from or supplementary to the ECD data, and it can be utilized to determine stereochemical properties, which would not be feasible by using only ECD calculations.

Journal of Natural Products
Article plant species. 41,42 9,10-Phenanthraquinones produced by microorganism have been reported to have strong antibiotic properties and antiviral activity. 35,36 The possible biogenetic origin of phenanthraquinones is rarely discussed in the literature. Regarding their oxygenation patterns, acetate-derived polyketide biogenesis was suggested for some phenanthraquinones based on isotope feeding experiments. 36,43,44 On the other hand, phenanthrenes can be formed by intramolecular oxidative ortho-coupling of dihydrostilbenes (bibenzyls). 45 In such cases the hydroxylation/ methoxylation pattern of at least one of the two aromatic rings in phenanthrene corresponds to the 3,5-disubstitution pattern of stilbenes. This structural feature is also characteristic of leucobryns. Considering that bibenzyls and bis-bibenzyls belong to the most important secondary metabolites of bryophytes, their biogenetic relationship with dimeric phenanthraquinones can be accepted. The occurrence of bibenzyl− phenanthrene and bibenzyl−dihydrophenanthrene compounds in some bryophytes (Buzzania and Cavicularia species) 24,25 provides further evidence for the presumption that leucobryns are biosynthesized by dimerization of bibenzyls with a C−C linkage, followed by oxidative steps yielding a 9,10-dioxo functionality. Similar to leucobryns 1−5, certain bibenzyls are also substituted with prenyl or geranyl groups. 1,2 The antiproliferative action of the isolated compounds (1− 5) was determined against four human cancer cell lines of gynecological origin, but only moderate (<10 μM) or negligible activities were measured (Table S6).

■ EXPERIMENTAL SECTION
General Experimental Procedures. UV spectra were measured on a Shimadzu UV-1800 UV−vis spectrophotometer in MeOH. The NMR spectra were recorded in methanol-d 4 on a Bruker Avance DRX 500 spectrometer operating at 499.9 MHz for 1 H and 125.7 MHz for 13 C and a Bruker Avance NEO 700 spectrometer operating at 700.25 MHz ( 1 H) and 176.08 MHz ( 13 C), respectively. All measurements were performed at 298 K. The signals of the methanol-d 4 ( 1 H: 3.35 ppm, 13 C: 49.3 ppm) were taken as reference. 2D NMR data were acquired and processed with Bruker TopSpin 4.0.5 software. Standard pulse programs available in the Bruker library were used for 2D 1 H− 1 H magnitude COSY (cosygpqf), gradient-and sensitivityenhanced HSQC (hsqcetgpsisp2.2), magnitude HMBC (hmbcgplpndqf), and EASY-ROESY (roesyadjsph) experiments. The high-resolution MS spectra were acquired on an FTHRMS-Orbitrap (Thermo-Finnigan) mass spectrometer equipped with an ESI ion source in positive ionization mode. HPLC was performed on a Waters 600 HPLC system (Waters Corporation, Milford, USA), equipped with a 2998 photodiode array detector and online degasser. ECD spectra were recorded on a JASCO J-810 spectropolarimeter. Solvents were distilled prior to use, and spectroscopic grade solvents were applied for spectroscopic measurements.
Plant Material. Paraleucobryum longifolium (Dicranaceae) was collected in Matrafured (Hungary) in June 2016. A voucher specimen (No. 883) has been deposited in the Herbarium of the Department of Pharmacognosy of University of Szeged, Szeged, Hungary. The plant material was dried at room temperature and chopped in a Retsch SM 100 chopper.
Extraction and Isolation. The plant material (4.0 kg) was extracted in a percolator with 75 L of MeOH at room temperature. The extract was evaporated (331.1 g) and the residue was subjected to column chromatography on polyamide using MeOH−H 2 O mixtures (20%, 40%, 60%, 80%, 100%). Fractions obtained with 80% MeOH contained compounds with a violet-black color. The 80% MeOH fraction was further separated by vacuum liquid chromatography (VLC) on normal-phase silica gel (particle size 45−63 μm) with a gradient system of CHCl 3 −MeOH (10:0, 9.6:0.4, 9.3:0.7, 9:1, 8:2, 7:3, and 0:1). This separation yielded 12 subfractions; among them subfractions 9−12 contained a MeOH-soluble, but a CHCl 3insoluble violet-black material, and subfraction 9 was chosen for further purification on NP-CC (silica gel, particle size 45−63 μm) using toluene−CHCl 3 −MeOH (4:12:3) as eluent. This chromatography afforded subfractions A−I; the violet-black material was concentrated in subfractions D, E, and F. Subfraction D was separated by reversed-phase flash chromatography (FC) using a gradient system of MeCN−H 2 O mixtures (from 50% to 50% to 100%  Computational Section. Mixed torsional/low-frequency mode conformational searches were carried out by means of the MacroModel 10.8.011 software by using the MMFF with an implicit solvent model for CHCl 3 . 47 Geometry reoptimizations were carried out first at the semiempirical AM1 level and then at the CAM-B3LYP/TZVP level with the PCM solvent model for both MeCN and acetone. Specific rotation data and single-point calculations for the sTDA ECD calculations were run with various functionals (B3LYP, BH&HLYP, CAM-B3LYP, and PBE0 for the SR and CAM-B3LYP, LC-BLYP, and ωB97X for the sTDA calculations) and the TZVP basis set as implemented in the Gaussian 09 package. 48 The sTDA calculations were performed with the sTDA 1.6 package. 49 Electronic circular dichroism spectra were generated as sums of Gaussians with 3000 cm −1 width at half-height, using dipole-velocity-computed rotational strength values. 50