New Compounds from the Roots of Corsican Calicotome Villosa (Poir.) Link.: Two Pterocarpans and a Dihydrobenzofuran

Three new compounds, a dihydrobenzofuran (coumaran) derivative (compound 1) and two pterocarpans (compounds 2 and 3) were isolated from a root extract of Calicotome villosa growing wild in Corsica. Their structures were elucidated using 1D and 2D NMR spectroscopy and MS/MS as 2-(1-methylethenyl)-5-hydroxy-6-carbomethoxy-2,3-dihydro-benzofuran, 4,9-dihydroxy-3-methoxy-2-dimethylallylpterocarpan, and 4,9-dihydroxy-3′,3′-dimethyl-2,3-pyranopterocarpan.


Introduction
Calicotome villosa (Poir.) Link. (Fabaceae) is a shrub that can reach 2 m in high, with grey-tomentose stems and sharp terminations, villous pods, trifoliate and oval leaves, and yellow and grouped flowers during the spring season [1,2]. It is very common in the Mediterranean area and particularly in Corsica Island, where it grows near the sea, while the subspecies C. villosa subsp. intermedia is distributed especially in the North of Africa and Spain [3]. Solvent extracts, essential oil, and isolated compounds from C. villosa showed diverse biological activities including antioxydant, antimicrobial, anti-inflammatory, antidiabetic, hypotensive, diuretic, and vasodilatator [4][5][6][7]. Phytochemicals that belong to various families have been isolated and identified in C. villosa extracts from different parts of the plant, including alkaloids, flavonoids, steroids, anthraquinones, and phenol derivatives [2][3][4][7][8][9][10][11][12]. In addition, falcarinol (fatty alcohol), some oxygenated mono-and sesquiterpenes, furan derivatives, paraffins, and fatty acids have been identified in the essential oil from aerial parts of C. villosa [2,4,5]. Concerning root extracts, only one paper reported on the screening of total alkaloid, saponin, and flavonoid contents [13].
Identification of individual components of C. villosa root extracts has not yet been reported. In continuation of our investigations on the constituents of Corsican aromatic and medicinal plants, we report in this paper on the isolation and structural elucidation of three new compounds contained in C. villosa roots.

Results and Discussion
Air-died roots of C. villosa were extracted with methanol (soxhlet apparatus), yielding a methanolic extract (4.7%) after evaporation of the solvent. This extract was partitioned by column chromatography • Compound 1, the major component of fraction F2, was isolated in pure form by repetitive CC and its structure elucidated using a combination of spectroscopic techniques including 2D NMR "INADEQUATE" (Incredible Natural Abundance DoublE QUAntum Transfer Experiment) sequence.

•
As observed on their 13 C NMR spectra, the sub-fractions F3.B.3 and F3.B.7 (see experimental) contained only the unidentified compounds 2 and 3 with the relative ratios 3:1 and 1:4, respectively, according to the relative intensities of signals. Although we did not succeed in isolating 2 and 3 in a pure form, the full set of 1D and 2D NMR experiments was conducted on the sub-fractions and compounds 2 and 3 were unambiguously identified.

Determination of 4,9-dihydroxy-3-methoxy-2-dimethylallylpterocarpan (2)
Although compound 2 could not be obtained in pure form, its degree of purity allowed for its structural elucidation. Indeed, the 13 C NMR spectrum of sub-fraction F3.B.3 from EtOAc fraction (F3) displayed a series of 21 signals with strong intensities that belonged to compound 2. The main signals were accompanied by a series of much smaller signals, which belonged to compound 3, as demonstrated in the next paragraph.
Exact mass obtained from the MS/MS analyses ([M + Na] + , C21H22O5Na + , experimental mass = 377.1347, theoretical mass = 377.1359) for compound 2 led to the molecular formula C21H22O5 (monoisotopic neutral mass = 354.1467) in agreement with 13 C NMR and DEPT spectra ( Table 2). The formula of this compound displayed 11 degrees of unsaturation.
According to 13 C NMR chemical shifts, the molecule contained 14 sp 2 carbons (seven double bonds) and therefore four cycles.
Chemical shifts, multiplicities of signals, and coupling constant values are characteristic of the pterocarpan skeleton [15]. In the 13 C NMR spectrum, the occurrence of 12 out of 14 signals belonging

Determination of 4,9-dihydroxy-3-methoxy-2-dimethylallylpterocarpan (2)
Although compound 2 could not be obtained in pure form, its degree of purity allowed for its structural elucidation. Indeed, the 13 C NMR spectrum of sub-fraction F3.B.3 from EtOAc fraction (F3) displayed a series of 21 signals with strong intensities that belonged to compound 2. The main signals were accompanied by a series of much smaller signals, which belonged to compound 3, as demonstrated in the next paragraph.
Exact mass obtained from the MS/MS analyses ([M + Na] + , C 21 H 22 O 5 Na + , experimental mass = 377.1347, theoretical mass = 377.1359) for compound 2 led to the molecular formula C 21 H 22 O 5 (monoisotopic neutral mass = 354.1467) in agreement with 13 C NMR and DEPT spectra ( Table 2). The formula of this compound displayed 11 degrees of unsaturation.
According to 13 C NMR chemical shifts, the molecule contained 14 sp 2 carbons (seven double bonds) and therefore four cycles.
Chemical shifts, multiplicities of signals, and coupling constant values are characteristic of the pterocarpan skeleton [15]. In the 13 C NMR spectrum, the occurrence of 12 out of 14 signals belonging to sp 2 carbons confirmed the pterocarpan skeleton that includes two phenyl moieties (A and D cycles; Table 2). Moreover, the value of the 3 J coupling constant between H6a and H11a (6.6 Hz) confirmed the cis junction of the two aliphatic cycles, characteristic of a pterocarpan skeleton [16], in agreement with the correlation plots observed in the COSY and NOESY spectra (Table 2).
Lastly, the remaining signals are characteristic of the γ, γ'-dimethylallyl substructure, corroborated by appropriate correlations in the HMBC spectrum. In that spectrum, correlations between H1 and C1, C2, and C3 allowed the positioning of the γ, γ'-dimethylallyl group on carbon 2 of cycle A of the pterocarpan. 1 H and 13 C NMR spectra evidenced also a methoxy group (-O-CH 3 ) as well as two "phenol" functions. The positioning of the methoxy group and the two hydroxyl functions were confirmed by correlations observed in the HMBC spectrum (Table 2, Figure S12). Indeed, HMBC correlations between the proton at 4.95 ppm (singlet) with aromatic carbons C8, C9, and C10 on the one hand, and the proton at 5.47 ppm (singlet) with aromatic carbons C3, C4, and C4a on the other hand evidenced two hydroxyl groups linked at C9 and C4, respectively. In addition, the correlation between the signal at 3.86 ppm (3H), belonging to the methoxy group, with the quaternary carbon at 145.52 ppm indicated the positioning of this group linked at the C3 position of the ring. Thus, compound 2 is a new pterocarpan, named 4,9-dihydroxy-3-methoxy-2dimethylallylpterocarpan, whose full NMR data are reported in the Table 2 and Figures S12-S20 ( 1 H and 13 C NMR spectra, HSQC, HMBC, COSY, and NOESY). Isoprenyl pterocarpans have been reported as secondary metabolites from various plants belonging to the Fabaceae family, for instance Pueraria mirifica [17,18], as well as pterocarpans containing the dimethylbenzopyrane substructure in Lespedeza floribunda [19]. 2D NMR was efficient to locate the various substituents on the pterocarpan skeleton. to sp 2 carbons confirmed the pterocarpan skeleton that includes two phenyl moieties (A and D cycles; Table 2). Moreover, the value of the 3 J coupling constant between H6a and H11a (6.6 Hz) confirmed the cis junction of the two aliphatic cycles, characteristic of a pterocarpan skeleton [16], in agreement with the correlation plots observed in the COSY and NOESY spectra (Table 2). Lastly, the remaining signals are characteristic of the γ, γ'-dimethylallyl substructure, corroborated by appropriate correlations in the HMBC spectrum. In that spectrum, correlations between H1′ and C1, C2, and C3 allowed the positioning of the γ, γ'-dimethylallyl group on carbon 2 of cycle A of the pterocarpan. 1 H and 13 C NMR spectra evidenced also a methoxy group (-O-CH3) as well as two "phenol" functions. The positioning of the methoxy group and the two hydroxyl functions were confirmed by correlations observed in the HMBC spectrum (Table 2, Figure S12). Indeed, HMBC correlations between the proton at 4.95 ppm (singlet) with aromatic carbons C8, C9, and C10 on the one hand, and the proton at 5.47 ppm (singlet) with aromatic carbons C3, C4, and C4a on the other hand evidenced two hydroxyl groups linked at C9 and C4, respectively. In addition, the correlation between the signal at 3.86 ppm (3H), belonging to the methoxy group, with the quaternary carbon at 145.52 ppm indicated the positioning of this group linked at the C3 position of the ring.
Thus, compound 2 is a new pterocarpan, named 4,9-dihydroxy-3-methoxy-2dimethylallylpterocarpan, whose full NMR data are reported in the Table 2 and Figures S12-S20 ( 1 H and 13 C NMR spectra, HSQC, HMBC, COSY, and NOESY). Isoprenyl pterocarpans have been reported as secondary metabolites from various plants belonging to the Fabaceae family, for instance Pueraria mirifica [17,18], as well as pterocarpans containing the dimethylbenzopyrane substructure in Lespedeza floribunda [19]. 2D NMR was efficient to locate the various substituents on the pterocarpan skeleton.  Thus, compound 2 is a new pterocarpan, named 4,9-dihydroxy-3-methoxy-2dimethylallylpterocarpan, whose full NMR data are reported in the Table 2 and Figures S12-S20 ( 1 H and 13 C NMR spectra, HSQC, HMBC, COSY, and NOESY). Isoprenyl pterocarpans have been reported as secondary metabolites from various plants belonging to the Fabaceae family, for instance Pueraria mirifica [17,18], as well as pterocarpans containing the dimethylbenzopyrane substructure in Lespedeza floribunda [19]. 2D NMR was efficient to locate the various substituents on the pterocarpan skeleton.

Determination of 4,9-dihydroxy-3 ,3 -dimethyl-2,3-pyranopterocarpan (3)
Compound 3 could not be obtained in pure form, but its degree of purity allowed for its structural elucidation. Indeed, in the sub-fraction F3.B.7 compound 3 was accompanied by compound 2, previously identified. The relative ratio evaluated by comparing the mean intensities of the signals of protonated carbons of every molecule in the spectrum was equal to 4:1. Therefore, the full set of 2D NMR experiments was conducted on fraction F3.B.7.
Chemical shifts, multiplicities of signals, and coupling constant values, as well as the number of sp 2 carbons belonging to two phenyl moieties, are characteristic of the pterocarpan skeleton [14] ( Table 3). Once again, the value of the 3 J coupling constant between H6a and H11a (6.5 Hz for compound 3) confirmed the cis junction of the two aliphatic cycles [16].

Plant Material
Roots of C. villosa were collected in September 2018 in Ajaccio, Corsica, France (41°54′45.1′′ N; 8°39′9.8′′ E), in collaboration with the Conservatoire Botanique de Corse. Our samples of C. villosa were identified as similar to the voucher specimen number 12057/1 (Jeanmonod D. and Roguet D.) deposited in the Conservatoire et Jardin Botanique de Genève.

Extraction and Isolation
Air-dried roots of C. villosa (192.6 g) were first extracted (soxhlet apparatus) with methanol (500 mL) for 24 h followed by concentration under reduced pressure to afford a dark brown extract (9.18 g). This extract was chromatographed over a silica gel column (200-500 µm; 150 g) and Concerning degrees of unsaturation, eight of these belonged to two phenyl moieties and two others are due to the B and C oxygenated cycles of the pterocarpan skeleton. According to the chemical shift values of signals, the two last degrees of unsaturation corresponded to an isolated double bond on the one hand, and a fifth cycle on the other hand.
We have already seen that the molecule contained the -O-CH 2 -CH-CH-O-sub-structure and two phenyl rings characteristics of the pterocarpan skeleton. According to carbon chemical shift values, compound 3 possesses an identical D ring to that of compound 2 and other pterocarpans such as desmocarpin, erysubin C, or edudiol [21] (Table 3). Therefore, the fifth cycle is linked to the aromatic A ring and includes an oxygen atom linked to a quaternary carbon at δC = 77.31 ppm and a double bond with two coupled hydrogens at δ = 5.57 ppm and δ = 6.31 ppm ( 3 J = 9.8 Hz, COSY spectrum), describing a pyrane sub-structure (or benzopyrane if cycle A is included in the sub-structure) ( Figure S21). According to the HMBC correlations, the proton at 6.31 ppm (doublet, linked to the carbon at 121.79 ppm, C1 ) correlates with the carbons at 118.41 ppm (methine), 116.12 ppm, and 140.30 ppm (quaternary carbons). These correlations indicated that the pyrano ring is connected with the A ring by carbons C2 and C3. Compound 3 is also a new pterocarpan, named 4,9-dihydroxy-3 ,3 -dimethyl-2,3-pyranopterocarpan (full 1D and 2D NMR data reported in the Table 3, Figures S21-S29). This compound may be seen as 4-hydroxyneorautenol. The biosynthetic pathway up to compounds 2 and 3 starting from phenylalanine may be hypothesized, similar to that mentioned by Goel et al. [22]. F3.A.13). DEPT spectra were recorded with the same parameters as 13 [23] for other pterocarpans, Figure S30).