Atypical Lindenane-Type Sesquiterpenes from Lindera myrrha.

Two new lindenane sesquiterpenes were obtained from the roots of Lindera myrrha. These compounds were structurally elucidated by HRMS data, extensive NMR analyses, and comparison between experimental and theoretical 13C-NMR data. Myrrhalindenane A is the first monomeric seco-d lindenane displaying a non-rearranged, cyclohexanic C-ring. Myrrhalindenane B is the second occurrence of an angular lindenane-sesquiterpene related to a C6-C7 lactonization.


Introduction
Lindera is a core genus of the Litseeae tribe of the Lauraceae family [1]. Many Lindera plants are of salient economical interest for soap and lubricant manufacturing (especially Lindera communis and Lindera glauca) owing to their elevated fatty oil content, while others are used to produce fragrances, species, and even building timber. As to ethnopharmacological claims, Lindera aggregata is included in various preparations of the Chinese Pharmacopoeia for treating urinary system diseases and inflammatory-related health hazards [2]. Other plants are also used in folk medicine such as Lindera umbellate, which is endowed with antispasmodic properties and has beneficial effects on gastric ulcers, cholera, and beriberi [3]. Fueled by the diverse interests lying in these plants, a wealth of skeletons were reported to have occurred in this well-studied family, the most represented of which include sesquiterpenes (mainly lindenanes, eudesmanes, and germacranes), and aporphine alkaloids, Molecules 2020, 25, 1830; doi:10.3390/molecules25081830 www.mdpi.com/journal/molecules along with some typical α-methylene-γ-butyrolactones collectively known as butanolides, and a few emblematic polysubstituted cyclopentanediones designated as lucidones [2]. Within this thoroughly studied genus, Lindera myrrha (Lour.) Merr., a small shrub common in central Vietnam, long remained unstudied. Conducted in 1994, the first phytochemical investigation dedicated to this species led to the isolation of a suite of aporphine alkaloids, including a new noraporphine, oduocine; and a new oxaporphine, oxoduocine [4]. A novel dihydroisocoumarin, lindermyrrhin, was further described from L. myrrha [5], but as far as can be ascertained, its terpene content remained unstudied. With this in mind, our study focused on the sesquiterpenes of L. myrrha roots, leading to the isolation of two new structures: myrrhalindenanes A and B. The structures of the isolated compounds 1 and 2 were elucidated by the interpretation of their spectroscopic data and by comparison with those described in the literature.

Results and Discussion
Compounds 1 and 2 were isolated from the methanol extract of L. myrrha by repeated chromatographic fractionations, including column chromatography, size-exclusive column chromatography, and preparative TLC.
Compound 1 was isolated as a white, amorphous solid. Its molecular formula was determined to be C 15 13 C-NMR spectrum, along with HSQC data, exhibited 15 signals for carbons consisting of one carbonyl, one carboxyl, two olefinic quaternaries, an oxygenated tertiary carbon, an olefinic methine, an exo-methylene, three methines, and a quaternary carbon (Table 1). These functionalities accounted for 4 indices of H deficiency, defining the tricyclic scaffold of 1 ( Figure 1). The 1 H-1 H correlation spectroscopy spectrum of 1 showed a proton spin system of a 1,2-disubstituted cyclopropane ring (δ H 1.46 (H-1); δ H 0.71/1.52 (H 2 -2); and δ H 2.00 (H-3)) (Supplementary Materials). These structural features were evocative of a lindenane-type sesquiterpene [6]. The cautious analysis of the 2D-NMR spectra revealed a polycyclic framework embedded with a sterically congested cyclopentane, as deduced from the HMBC correlations from the angular methyl group at δ H 1.13 (H 3 -14) to the carbons resonating at δ C 31.1 (C-1), δ C 76.5 (C-5), and δ C 51.2 (C-9), and from the exo-methylene that was located at C-4 based on long-range heteronuclear crosspeaks between the olefinic protons at δ H 4.99/5.17 (H 2 -15) to C-3 (δ C 23.4) and C-5 (δ C 76.5). The chemical shift of the isolated diastereotopic methylene group at δ H 2.31/2.39 (each 1H, d, J = 15.5 Hz) hinted at it being vicinal to a carbonyl function. This tentative assignment was supported by the HMBC crosspeak from H 2 -9 to the carbon resonating at δ C 197.7 (C-8). Altogether, these spectroscopic data left no choice but to introduce a ∆ 6,7 moiety. The C-6 location of the olefinic proton was validated based on the HMBC correlations from H-6 to C-5, the quaternary olefinic carbon resonating at δ C 163.1 (C-7), and to C-8. In the end, the C-9 location of the side chain was established owing to the HMBC correlations from the methine at δ H 3.50 (H-11) to both C-7 and C-8. This methine was deduced to have been substituted by a methyl and a carboxylic acid group based on (i) the COSY crosspeak between this and the methyl protons at δ H 1.13 (CH 3 -14), and (ii) the HMBC correlation from these methyl protons to both C-7 and the carbon resonating at δ C 174.0 (C-12). These spectroscopic features determined the planar structure of 1, namely myrrhalindenane A, as indicated in Figure 2. The NOESY correlations between H 2 -2 and H 3 -14 determined their synfacial orientation. Aside from the doubts regarding C-5 configuration, the absolute configuration assignment of C-11 represented a vexing problem in its achiral environment. These spectroscopic features led us to consider four different stereochemical arrangements, as indicated in Figure 3. DFT-NMR chemical shift calculations and the subsequent DP4 probability method [7] were performed on these different candidates. This DP4 application demonstrated the structural equivalence of 1 with diastereoisomer 1C with 88.8% probability ( Figure 3).              Compound 2 was obtained as a white, amorphous solid. Its molecular formula, C 15 H 18 O 5 , was established from the sodiated ion peak at m/z 301.1047 (calculated for C 15 H 18 O 5 Na), differing from compound 1 by an additional oxygen atom. Notwithstanding their common lindenane core, the NMR data revealed some salient structural differences between these compounds. The 13 C-NMR data revealed the lack of the exo-methylene moiety and the loss of the olefinic proton although a tetrasubstituted double bond could be identified. In line with this latter point, the downfield 1 H chemical shift of the signal related to the methyl CH 3 -13 (δ H 1.92 vs. 1.26), combined with the shielding of the corresponding carbon (δ C 9.8 vs. 16.6) were evocative of its location on a double bond [8]. Conversely, the 1 H-NMR spectrum displayed further signals corresponding to a tertiary methine at δ H 2.27 (1H, d, J = 12.5 Hz), coupled with an oxygenated methine at δ H 5.03 (1H, d, J = 12.5 Hz). Likewise, an additional set of oxygenated diastereotopic methylene at δ H 3.80/3.67 could be identified, as well as a new tertiary oxygenated methine at δ C 80.1. Along with the unchanged carbonyl moieties at δ C 197.7 and 173.9, these functionalities represented three indices of hydrogen deficiency, determining the tetracyclic appendage of 2. The oxygenated methylene could be located at C-4 based on the long-range heteronuclear correlations from these protons to C-3 (δ C 28.7), C-4 (δ C 80.1), and C-5 (δ C 63.8). The joint HMBC correlations from the methyl protons at δ H 1.11 and of the diastereotopic methylene signals at δ H 2.62/2.67 (each 1H, d, J = 16 Hz) to the carbon resonating at δ C 63.8, validated the occurrence of a methine at this specific position (C-5). The chemical shift of C-4 (δ C 80.1) defined the presence of a hydroxy group on it. Such B-ring structures are recurrent within lindenane sesquiterpenes, falling into the third subtype defined by Du [9]. The tetracyclic core of 2, and the unchanged chemical shifts of both C-8 and C-9 left no possibility but to introduce an additional α-methyl-∆ α,β -γ lactone fused ring at C-6/C-7. This assumption was validated by the correlations from the olefinic-located methyl at δ H 2.27 to the quaternary carbons C-7 (δ C 154.6) and C-11 (δ C 132.1), to the carbonyl-type carbon C-12 (δ C 173.9), and from the oxymethine proton H-6 to C-11. These spectroscopic data were fully consistent with those of formerly reported sesquiterpene lactones [10,11]. The antiperiplanar orientation of H-5 and H-6 could be determined from the magnitude of the coupling constant value (J = 11.5 Hz) [11]. Having in mind, i) the consensual trans arrangement of the hydrindane system in lindenane sesquiterpenes, and ii) the antifacial orientations of H-5 and H-6, only left the configuration of C-4 pending assignment [12,13]. A preferred configuration for C-4 prevails with a β-OH group and an α-oxygenated methylene moiety, so that Du's lindenane sesquiterpene subtypes define the absolute configuration of this stereocenter [9]. Nevertheless, exceptions were reported throughout literature [11,14,15], so assigning the configuration of these positions based solely on biosynthetic considerations is not a relevant approach to reliably establish the configuration of such compounds. To remedy this, DFT-NMR calculations and subsequent 13 C-NMR data comparison of the two possible epimers against the experimental data set, resulted in the prediction of diastereoisomer 2A with 100% probability (Figure 4). Compounds 1 and 2 were found to be unstable on storage. After three days at room temperature, both had undergone ca. 70% decomposition to yield a mixture of products. The minute amounts of compound precluded any further repurification attempt. In these conditions, the recorded ECD spectra provided no clear-cut match, irrespective of the absolute configuration used in TDDFT. This observation is in line with precedents having outlined the inherent instability of lindenane ring system [16,17], which was occasionally reported in the course of former phytochemical investigations [18]. Despite the lack of spectroscopic evidence, the consensual β-orientation of both the methyl and cyclopropyl functions not only in Lindera species [2], but also within the Chloranthaceae plants that produce a much higher number of these sesquiterpenes [19,20], gave strong support to the preferred absolute configuration depicted in Figure 1.
The determined A/B ring substitution pattern of myrrhalindenane A is common among lindenane sesquiterpenes, falling into the lindenane sesquiterpenoid subtype I, as defined by Du and co-workers [9]. Conversely, the occurrence of oxygenated substituents at C-5 is rather uncommon among lindenane sesquiterpenes, since this position is often substituted by an α-disposed hydrogen atom, or is unsaturated due to either a Δ 4,5 or a Δ 5,6 function [13]. A few structures were however reported to contain an oxygenated substituent at C-5 such as sarcandralactone A, which revealed a 5β-OH group [21] or the dimeric sarcandrolide F that exhibits a 5β-OOH group [22]. The side chain located at C-7 can be assumed to arise from the hydrolysis of a 2-methyl-2-butyrolactone or a 2methylbutyrolactone D ring related to the canonical lindenane skeleton. Only a few seco-D lindenanes have been reported to date. Some such compounds were formerly described in the Lindera species as strychnilactone [23], lindenanolide G [24], and linderagalactones B and C [25]. Nevertheless, all these structures undergo later rearrangement to afford a α-pyrone C-ring, therefore differing from the currently reported carbon skeleton. These compounds also differ from 1 by the constant occurrence of a Δ 7,11 moiety. Remarkably, a wealth of seco D-lindenanes were reported within lindenane sesquiterpenoid [4 + 2] dimers, especially from the Sarcandra species, e.g., sarcandrolides [22,26]; and various Chloranthus plant species such as shizukaol species [27][28][29], chlorahololides [30,31], spicachloranthins E and F [32], and chlorajaponilides [33], among many others. The biosynthesis of dimeric lindenane sesquiterpenes is deemed to proceed via a Diels-Alder reaction with Δ 4,15 and Δ 5,6 representing the diene reactive unit [34]. Furyldiene lindenanes and, more generally speaking, molecules displaying these structural features rendering them prone to undergoing Diels-Alder addition, seem to be too unstable to be isolable [9]. This inherent reactivity towards dimerization most likely accounts for 1 being the first reported seco D-lindenane monomer, which can be readily related to its Δ 6,7 function that prevents it from dimerizing. Lindermyrrhin B (2) is the second example Compounds 1 and 2 were found to be unstable on storage. After three days at room temperature, both had undergone ca. 70% decomposition to yield a mixture of products. The minute amounts of compound precluded any further repurification attempt. In these conditions, the recorded ECD spectra provided no clear-cut match, irrespective of the absolute configuration used in TDDFT. This observation is in line with precedents having outlined the inherent instability of lindenane ring system [16,17], which was occasionally reported in the course of former phytochemical investigations [18]. Despite the lack of spectroscopic evidence, the consensual β-orientation of both the methyl and cyclopropyl functions not only in Lindera species [2], but also within the Chloranthaceae plants that produce a much higher number of these sesquiterpenes [19,20], gave strong support to the preferred absolute configuration depicted in Figure 1.
The determined A/B ring substitution pattern of myrrhalindenane A is common among lindenane sesquiterpenes, falling into the lindenane sesquiterpenoid subtype I, as defined by Du and co-workers [9]. Conversely, the occurrence of oxygenated substituents at C-5 is rather uncommon among lindenane sesquiterpenes, since this position is often substituted by an α-disposed hydrogen atom, or is unsaturated due to either a ∆ 4,5 or a ∆ 5,6 function [13]. A few structures were however reported to contain an oxygenated substituent at C-5 such as sarcandralactone A, which revealed a 5β-OH group [21] or the dimeric sarcandrolide F that exhibits a 5β-OOH group [22]. The side chain located at C-7 can be assumed to arise from the hydrolysis of a 2-methyl-2-butyrolactone or a 2-methylbutyrolactone D ring related to the canonical lindenane skeleton. Only a few seco-d lindenanes have been reported to date. Some such compounds were formerly described in the Lindera species as strychnilactone [23], lindenanolide G [24], and linderagalactones B and C [25]. Nevertheless, all these structures undergo later rearrangement to afford a α-pyrone C-ring, therefore differing from the currently reported carbon skeleton. These compounds also differ from 1 by the constant occurrence of a ∆ 7,11 moiety. Remarkably, a wealth of seco d-lindenanes were reported within lindenane sesquiterpenoid [4 + 2] dimers, especially from the Sarcandra species, e.g., sarcandrolides [22,26]; and various Chloranthus plant species such as shizukaol species [27][28][29], chlorahololides [30,31], spicachloranthins E and F [32], and chlorajaponilides [33], among many others. The biosynthesis of dimeric lindenane sesquiterpenes is deemed to proceed via a Diels-Alder reaction with ∆ 4,15 and ∆ 5,6 representing the diene reactive unit [34]. Furyldiene lindenanes and, more generally speaking, molecules displaying these structural features rendering them prone to undergoing Diels-Alder addition, seem to be too unstable to be isolable [9]. This inherent reactivity towards dimerization most likely accounts for 1 being the first reported seco d-lindenane monomer, which can be readily related to its ∆ 6,7 function that prevents it from dimerizing. Lindermyrrhin B (2) is the second example of a 3/5/6/5 tetracyclic lindenane-type sesquiterpene lactone formed at C-6 and C-7, with the first such occurrence being reported from Xanthium sibericum (Asteraceae) [11].

General
The NMR spectra were measured on a Bruker Avance III (500 MHz for 1 H-NMR and 125 MHz for 13 C-NMR, Bruker, Bremen, Germany) spectrometer with TMS as internal standard. Chemical shifts are expressed in ppm with reference to the residual protonated solvent signals (acetone-d6 with δ H 2.05, δ C 206.26, and 29.84) or the internal TMS (0.00). The HR-ESI-MS were recorded on a HR-ESI-MS Bruker microOTOF Q-II (Bremen, Germany). TLC was carried out on precoated silica gel 60 F254 or silica gel 60 RP-18 F254S (Merck, Darmstadt, Germany), and spots were visualized by spraying with 10% H 2 SO 4 solution followed by heating. Gravity column chromatography was performed with silica gel 60 (0.040-0.063 mm, Himedia, Mumbai, India).

Plant Material
The roots of Lindera myrrha were collected from Cu Chi District, Ho Chi Minh City, in July 2016. The botanical sample was authenticated by Dr. Pham Van Ngot, Department of Botany, Faculty of Biology, Ho Chi Minh University of Pedagogy. A voucher specimen (No UP-B05) was deposited in the herbarium of the Department of Organic Chemistry, Faculty of Chemistry, Ho Chi Minh University of Education.

Conclusions
The investigation of the so-far unstudied terpenic content of Lindera myrrha afforded two novel monomeric lindenanes. Despite the elevated number of such metabolites formerly reported to occur in Lauraceae and Chloranthaceae, these two compounds display unusual structural features. Among these, the combination of a native cyclohexanic C ring and of a seco-d cycle, unprecedented within monomeric lindenanes reported so far, is particularly worth being stressed out.
Supplementary Materials: The following are available online. 1 H-and 13 C-NMR spectra, HMBC spectra, and HRMS spectra for 1 and 2; atomic coordinates of the lowest-energy conformers of the four candidate diastereoisomers of 1 and 2.