Identification of Minor Benzoylated 4-Phenylcoumarins from a Mammea neurophylla Bark Extract

Through dereplication analysis, seven known Mammea coumarins were identified in a fraction obtained from Mammea neurophylla dichloromethane bark extract selected for its ability to prevent advanced glycation end-product (AGE) formation. Among them, a careful examination of the NMR dataset of pedilanthocoumarin B led to a structural revision. Inspection of LC-DAD-MSn chromatograms allowed us to predict the presence of four new compounds, which were further isolated. Using spectroscopic methods (1H-, 13C- and 2D-NMR, HRMS, UV), these compounds were identified as new benzoyl substituted 4-phenylcoumarins (iso-pedilanthocoumarin B and neurophyllol C) and 4-(1-acetoxypropyl)coumarins cyclo F (ochrocarpins H and I).


Introduction
The vascular endothelium is the innermost layer of cells in vessels. In diabetes, chronic hyperglycemia induces advanced glycation end-product (AGE) formation. Their accumulation in tissues and blood contributes to endothelial dysfunction [1,2] and, consequently, to disorders, such as atherosclerosis, cardiovascular events and graft rejection, through the expression of inflammatory and immune mediators on the endothelial cell surface [3].
An anti-AGE screening of different plant extracts obtained from Clusiaceae and Calophyllaceae species previously led to selecting Mammea neurophylla (Schltr.) Kosterm, an endemic neocaledonian shrub, for further phytochemical and biological investigations [4,5]. In the dichloromethane (DCM) bark extract, Mammea coumarins were demonstrated to be responsible for the anti-AGE, anti-inflammatory and vasorelaxing effects [5]. This paper deals with the study of minor compounds that were not investigated during the aforementioned preliminary study.

Results and Discussion
The M. neurophylla DCM bark extract was fractionated using flash chromatography to give 26 fractions, which were tested for their anti-AGE properties [5,6]. Among them, Fraction VIII exhibited even a stronger anti-AGE potential than the natural reference compound quercetin (IC50 0.04 and 0.06 mg/mL, respectively). Therefore, this fraction was first analyzed following a rapid, sensitive and simple dereplication method of Mammea coumarins using LC-DAD-MS n that we recently published ( Figure S1) [7].

Dereplication of Known Mammea Coumarins
By comparison of UV spectra, as well as mass spectra and fragmentations (Table 1) and by interlocking literature data via the Scifinder database [8], seven known coumarins were presumably detected in the fraction and associated with peaks 8.1 to 8.11 ( Figure S1). As far as peaks 8.1, 8.4 and 8.8 to 8.11 were concerned, mass fragmentation of the quasi-molecular peak in positive ionization mode suggested a 4-(1-acetoxypropyl)coumarin (Mammea E type coumarins), as a loss of both 42 and 60 u was observed to form a stable ion. In negative mode, the loss of 60 u confirmed this substitution. Their UV spectrum (λmax = 222, 295, sh 325-330 nm) indicated 4-alkyl-5,7-dihydroxycoumarins exhibiting an 8-acyl substituent [9][10][11]. For 8.10 and 8.11, mass fragmentations were typical of prenylated Mammea E with fragments at m/z 315 [M + H − 60 − 56] + . Thus, considering their hypothetical molecular mass (430 g·mol −1 ) and retention time, a 3-methyl-1-oxobutyl or a 2-methyl-1-oxobutyl could be proposed at C-8 to suggest Mammea E/BA (1) and Mammea E/BB (2) for peaks 8.11 and 8.10, respectively [7]. For peaks 8.8 and 8.9, losses of 42 and 60 u in positive ionization mode were preceded by a water loss, suggesting a substitution by a 2-hydroxy-3-methylbut-3-enyl group at C-6. Considering their molecular mass (446 g·mol −1 ), as well as the loss of 86 u of the base peak [M + H − H2O] + in positive mode, the acyl group at C-8 should be a 3-methyl-1-oxobutyl or a 2-methyl-1-oxobutyl. Peaks 8.9 and 8.8 were therefore hypothesized to be neurophyllol A (3) and neurophyllol B (4). The structures of these four coumarins were confirmed by NMR experiments after purification steps [4,12]. Finally, as fragmentations for peaks 8.1 and 8.4 were identical with those observed for Mammea cyclo F coumarins ( Figure S2) [7], they were identified as ochrocarpin F (5) and G (6), which was confirmed by NMR spectral analysis after isolation [13].
The UV spectrum of peak 8.6 (λmax = 255, 313 nm) showed a bathochromic shift in comparison to 4-phenylcoumarins, usually substituted by an acyl group at C-6 or C-8, suggesting a less common substitution at this position. Such a UV spectrum could correspond to a 4-phenylcoumarin substituted with a benzoyl group at position 8 [14]. For these Mammea A coumarins, the quasi-molecular ion ([M + H] + or [M − H] − ) usually appeared as the base peak in both ionization modes. Thus, 426 g·mol −1 was proposed as a hypothetical molecular mass. The loss of 56 u to form a stable ion in positive ionization mode corresponds to a non-cyclized 4-phenylcoumarins bearing a prenyl group. The revised structure of pedilanthocoumarin B (7) [14] (see below) was attributed to peak 8.7 (λmax = 255, 301 nm), whereas peak 8.6 was considered as associated with a new compound.

Structure Prediction of Structures of Original Mammea Coumarins
In addition to seven known coumarins (Figure 1), five previously unreported or revised compounds (7 to 11) were detected ( Figure 2).   Compound 8 (peak 8.6) exhibited similar mass spectra and fragmentation patterns as pedilanthocoumarin B (7), suggesting a 4-phenylcoumarin with the same substituents, i.e., one prenyl and one benzoyl moiety. As a bathochromic shift was observed on the UV spectrum (λmax = 255, 313 nm), 8 was hypothesized as iso-pedilanthocoumarin B, a 4-phenylcoumarin bearing a prenyl substituent at C-6 and a benzoyl group at C-8. Compound 9 (peak 8.3) showed a similar UV spectrum as pedilanthocoumarin B 7, suggesting a 4-phenylcoumarins with a benzoyl group at C-6. As observed for neurophyllols 3 to 4, the base peak in positive ionization mode [M + H − H2O] + corresponded to a water loss and was followed by either an additional water loss (−18 u) or the fragmentation of the pyran ring (−60 u). This suggested that the Mammea A coumarin was substituted by a 2-hydroxy-3-methylbut-3-enyl moiety at C-8 [7]. To our knowledge, neither 8 nor 9 have been previously identified.
The structures of Compounds 10 and 11 (peaks 8.2 and 8.5) seemed very close to those of ochrocarpin F (5) and G (6). Indeed, no differences were noticed on UV spectra, as well as mass spectra and fragmentations. As it was previously described that Mammea cyclo E and cyclo F show similar fragmentation pathways [7], 10 and 11 were hypothesized as 4-(1-acetoxypropyl)coumarins cyclo E or cyclo F substituted by a 3-methyl-1-oxobutyl or a 2-methyl-1-oxobutyl.

Purification and Structural Analysis of Coumarins 7 to 12
To verify our hypothesis, Fraction VIII was fractionated using flash chromatography, and Compounds 7 to 12 were purified by preparative HPLC.

Structure Revision of Pedilanthocoumarin B (7)
1 H-and 13 C-NMR data of 7 were identical to those previously reported by Sandjo et al., for pedilanthocoumarin B [14]. Indeed, pedilanthocoumarin B was identified as a 4-phenylcoumarin substituted by a prenyl group at C-6 and a benzoyl group at C-8 based on HMBC correlations between the prenyl methylene at δH 3.56 and the carbons at δc 109.2, 157.4 and 160.9, respectively assigned by the authors to C-6, C-5 and C-7. However, since C-8a also has a chemical shift at 157.4, carbons at δc 109.2, 157.4 and 160.9 could also be assigned to C-8, C-8a and C-7 from the same correlations. In our study, the 1 H-NMR spectrum of Compound 7, recorded in CDCl3 between 0 and 20 ppm, showed two hydroxyl signals at δH 8.60 and 9.05 ppm. They were respectively attributed to 5-OH and 7-OH, as indicated by a correlation between δH 9.05 and δc 109.2 on the HMBC spectrum. No signal corresponding to a strongly-chelated hydroxyl was observed, as expected for a 7-OH/8-benzoyl pattern [12,15]. These key features were not discussed in the first publication on pedilanthocoumarin B (7) [14]. Moreover, as explained below, NMR data for Compound 8 unambiguously led to a 6-prenyl-8-benzoylcoumarin. Consequently, Compound 7 was identified as pedilanthocoumarin B, but with a structure now revised as a 4-phenylcoumarin substituted by a benzoyl group at C-6 and a prenyl group at C-8.

Plant Material
As previously described, Mammea neurophylla bark was collected in November 1998 in the dry forest of "Conservatoire botanique de la forêt de Tiéa", North Province (New Caledonia). A voucher specimen (LIT-0660) was deposited at the Herbarium of the Botanical and Tropical Ecology, Department of the Institut de Recherche pour le Développement (IRD) Center, Noumea (New Caledonia) [4].

Extraction and Fractionation
As described by Dang et al. [5], air-dried and powdered bark (695 g) was extracted with DCM (2.5 L, 72 h) in a Soxhlet apparatus. The DCM crude extract was concentrated under vacuum at 40 °C to yield 29 g of dry extract. DCM bark extract (18 g) was subjected to flash chromatography using a PuriFlash PF-50SiHC/300G cartridge (Interchim, Clichy, France) eluted with C6H12/EtOAc (95:5 v/v to 70:30 v/v for 120 min, 50 mL/min) to afford 26 fractions, namely FI to FXXVI, in elution order.

Fraction VIII Profiling
LC-PDA-ESIMS analyses were performed on a Waters 2795 apparatus (Waters, Guyancourt, France) equipped with a Waters 2487 UV detector and coupled with an Esquire 3000 PLUS ESI ion trap mass spectrometer equipped with an electrospray source (Bruker, Wissembourg, France) assisted by the HyStar ® software (Bruker Daltonics, Wissembourg, France). A 10-µL sample was directly injected onto a Lichrospher 100 RP18 column (150 × 4.6 mm, 5 µm, Merck, Darmstadt, Germany). The column was set at 20 °C. Separation was achieved using an acidic water/MeOH system. The mobile phase was as follows: water (HCOOH-0.1%)/MeOH from 30:70 to 20:80 v/v in 45 min. The flow rate was 1 mL/min with a split of 90% before mass analysis. For the study, the wavelength was set at λ = 290 nm. For each peak of interest, the UV spectrum was obtained by scanning the sample in the range of 210 to 400 nm. The mass analyses were performed in both positive and negative modes using the conditions described by Dang et al. [7].