Bond reactivity indices approach analysis of the [2+2] cycloaddition of jatrophane skeleton diterpenoids from Euphorbia gaditana Coss to tetracyclic gaditanone

The reaction mechanism of the intramolecular [2 + 2] cycloaddition from a jatrophane precursor to the gaditanane skeleton, an unprecedented 5/6/4/6-fused tetracyclic ring framework recently isolated from Euphorbia spp., was studied using the bond reactivity indices approach. Furthermore, six diterpenoids, including three undescribed jatrophanes isolated from E. gaditana Coss, were described. The structures of these compounds were deduced by a combination of 2D NMR spectroscopy and ECD data analysis.

Euphorbia gaditana Coss., belonging to the genus Euphorbia, is an endangered species from Southern Spain (western Andalusia) and Northern Africa (Algeria and Tunisia). In a previous chemical investigation of E. gaditana Coss, a previously undescribed jatrophane (1) together with an undescribed diterpenoid (2) with an unprecedented 5/ 6/4/6-fused tetracyclic ring skeleton, were isolated from the aerial parts of this plant. The name gaditanane was proposed for that carbon skeleton. A suggested biosynthetic relationship between both compounds was also established by chemical correlation (Flores-Giubi et al., 2017). Simultaneously, in a report from Mai et al. (2017), two other undescribed jatrophane-derived diterpenoid esters with a 5/6/4/6-fused tetracyclic ring skeleton, heliosterpenoids A and B, were isolated from E. helioscopia and were found to be potent inhibitors of β-glycoproteins and exhibited cytotoxic activity.
Therefore, as a part of our ongoing research on the identification and study of the reactivity of undescribed bioactive compounds potentially capable of modulating PKC isoforms (Daoubi et al., 2007;Flores-Giubi et al., 2019), a detailed study of the specialised metabolites in the roots and aerial parts of E. gaditana Coss has been conducted. A study of the cyclization to gaditanone from jathrophane metabolites isolated from E. gaditana Coss, has been modeled using bond reactivity indices approach.

Isolation and chemical characterization
The petroleum ether and dichloromethane extracts from the aerial parts and roots of E. gaditana Coss were further purified leading to the isolation and structural elucidation of three undescribed jatrophanes (3-5), along with the three previously known compounds (1, 2 and 6) (Fig. 1). The structures of compounds 3-5 were established based on spectroscopy analysis including 1D and 2D NMR and HRESIMS experiments. Absolute configuration for compound 3 was established by comparison among its experimental and calculated ECD, where the starting point conformation used for computational calculations was deduced from NOESY data. This, in turn, supported the proposed absolute configurations given for compounds 4-5.
Compound 3 was isolated as an amorphous solid, and its molecular formula C 33 H 44 O 10 (12 unsaturations) was deduced on the basis of the observed sodium adduct ion in its HRESIMS (m/z 623.2834 [M+Na] + ). Both 1 H and 13 C NMR data analysis (Table 1) revealed four ester residues attributable to the presence of three acetoxy groups and one benzoyloxy group. In addition of these signals, 13 C NMR and DEPT spectra displayed signals for the remaining 20 skeletal carbons, encompassing five methyl groups, one methylene group, 11 methine groups (including five oxygenated sp 3 carbons and three olefinic carbons), and three nonprotonated carbons (including one oxygenated sp 3 carbon and one olefinic carbon).
The HMBC cross-peaks from H-8 (δ H 5.34 ppm) and H-14 (δ H 4.79 ppm) to δ C 169.2 and 170.4 ppm, respectively, indicated two acetoxy groups attached to C-8 and C-14 whereas the benzoate chain was located at C-9 due to HMBC cross-peaks from H-9 (δ H 4.60 ppm) to carbonyl carbon of the benzoyl group at δ C 167.4 ppm. The remaining acetoxy group could not be directly attached by HMBC experiments due to the lack of direct correlations. The comparison of the chemical shifts and coupling constants of H-3 and H-2 with those observed in the literature for other related jatrophanes pointed out that it must be located at C-3 (Hohmann et al., 2002(Hohmann et al., , 2003Valente et al., 2004;Flores-Giubi et al., 2017;Mai et al., 2017). Finally, the free hydroxyl groups were deduced to be attached to C-7 and C-15. The gCOSY correlations observed for protons H-7 (δ H 4.16 ppm) and OH-7 (δ H 4.85 ppm) indicated the presence of a hydroxyl group at C-7. The remaining one hydroxyl group must be located at C-15 due to the shielding of that nucleus in the 13 C NMR spectrum in comparison to other related-jatrophanes (Table S9) ( Fig. 2) (Hohmann et al., 2002(Hohmann et al., , 2003Valente et al., 2004;Flores-Giubi et al., 2017;Mai et al., 2017). A new signal at δ H 7.27 ppm in the 1 H NMR spectrum in C 6 D 6 , which partially exchanged with D 2 O, confirmed the hydroxyl group at C-15 (Table S8 and Fig. S14).
The relative configuration of 3 was deduced by the interpretation of the NOESY spectrum. The correlations of H-4/H 3 -17/H-7/H-8 and H-4/ H-13/H-11 indicated α-orientations for those protons, whereas the correlations of H-5/H-9 were used to place these protons on the opposite face of the macrocycle. The NOESY correlations of H-11/H 3 -17/H-13 and the coupling constant of J 11,12 = 15.5 Hz revealed the E-geometry of the C-11/C-12 double bond, while the E stereochemistry of the C-5/C-6 olefin linkage was deduced from NOESY correlations between H-4/H 3 -17 and H-5/H-9. Further NOESY correlations of H 3 -16/H-3/H-4 and H-12/H 3 -20/H-14 revealed their proximity in space (Fig. 3). Moreover, the coupling constant J 14,13 = 10.3 Hz (Table 1) was compatible with an anti-disposition of H-13 and H-14 and it was in agreement with the values described in literature for other related-jatrophanes (Valente et al., 2004;Mai et al., 2018b).
On the basis of the α-assumed configuration of H-4 and β-orientation of 15-OH, characteristic of all jatrophane-type diterpene derivatives isolated to date (Fakunle et al., 1989;Tian et al., 2014), and NOESY correlations the absolute configuration for 3 was suggested to be assigned as 2R,3S,4S,7S,8S,9S,13S,14R,15R. This absolute configuration of 3 was supported by comparison of the experimental ECD spectrum with the computed ECD spectrum calculated from quantum mechanical time-dependent density functional theory (TDDFT) calculations with B3LYP/6-31G++ (d,p) level of theory, using the Gaussian 16 program (Li et al., 2010). The theoretically calculated ECD spectrum of 3 was in good agreement with the experimental ECD spectrum (Fig. 4).
Compounds 4 and 5 showed the same molecular formula of C 35 H 46 O 11 based on a sodium adduct ion at m/z = 665.2946 [M+Na] + and m/z = 665.2944 [M+Na] + , respectively in their HRESIMS spectra. Both the 1 H and 13 C NMR data (Table 1) indicated that the structure of compounds 4 and 5 closely resembled jatrophane 3, the difference being the presence and position of an additional acetoxy group.
The spectroscopic data of 6 are in agreement with those reported in the literature (Flores-Giubi et al., 2017).
In a range of concentrations between 1 and 25 μg/mL, none of the tested compounds showed significant activity as compared to controls. These results agreed with previous studies that have shown that acetylation of the 15-hydroxyl and acetoxylation at C-2 are critical for the HIV-1 latency antagonizing activity (Bedoya et al., 2009).

Biosynthetic study of the intramolecular [2 + 2] cycloaddition
The isolation of 6, the proposed biosynthetic precursor for gaditanone (2), from E. gaditana Coss encouraged us to gain insight on the nature of the intramolecular [2 + 2] cycloaddition leading to the gaditanane skeleton by computational analysis. Two possible biosynthetic routes were hypothesized (Fig. 7). On the one hand, the jatrophane ester 1 might undergo cyclization following the pathway A to obtain a tetra-fused derivative (7) that could be further oxidized to form gaditanone (2). On the other hand, a previous oxidation of 1 leading to 6 would be a required step, previous to its cyclization to yield gaditanone (2) (pathway B).
In a cycloaddition reaction there is an electron density transfer from the nucleophilic to the electrophilic species (Kumar et al., 2016;Qiu et al., 2018). In order to translate this concept into a computational analysis, a recent methodology has been used for obtaining bond reactivity indices (Δf i NBO ) (Sánchez- Márquez, 2016;Morales-Bayuelo et al., 2018) and philicities for every Natural Bond Orbital (NBO). These bond reactivity indices are obtained through projection of the function dual descriptor on the localized orbitals obtained in the NBO analysis allowing the reaction site of an electrophile or nucleophile to be determined within a much-localized molecular region. A negative small value of Δf i NBO together with a high partial occupancy indicates a nucleophilic behaviour whereas a positive value of Δf i NBO and a small partial occupancy corresponds to an electrophilic behaviour. For instance, this approach has been successfully applied to study previously the regioselectivity in Diels-Alder reactions (Morales-Bayuelo et al., 2018). In addition, in this study, a Second-Order Perturbational Theory Analysis (SOPTA) (Reed et al., 1988) has been also used to complement and corroborate the results. To the best of our knowledge, these quantum methodologies have not been previously applied in clarifying biosynthetic pathways of terpenes. Firstly, the pathway A proposed in Fig. 7 was studied considering a singlet state in a concerted mechanism. Therefore, in compound 1 the π bonding orbitals 167 (C11-C12), § 168 (C5-C6) and the π anti-bonding orbitals 181 (C11-C12) and 182 (C5-C6) would be responsible for the formation of the four-membered ring (Table S3, Supporting information; Jatrophane numbering has been used for describing the orbital interactions throughout the manuscript. For correspondence between jatrophane and gaussian numbering used in table S3, see Table S2 of Supporting Information). However, a [2 + 2] cycloaddition it does not seem to be very favoured in that model, since the virtual orbitals 181 and 182 have little electrophilic character (Δf i NBO 0.054 and 0.058, respectively). Furthermore, they have very high orbital energy and according to the geometry that the model predicts, they are a long way from the occupied orbitals to interact with them (Table S3, Supporting information). As in the previous case, in an alternative non-concerted mechanism through triplet state, the formation of the carbocycle would occur by the interaction of the π bonding orbitals 170 (C5-C6), § 171 (C11-C12) and the π anti-bonding orbitals 181 (C5-C6) and 182 (C11-C12) ( On the other hand, it was carried out orbital energy diagrams of the singlet and triplet states of compound 6 (Fig. 7, Pathway B). In the singlet state, NBO 182 (C5-C6) § showed a high electrophilic character (see Δf i NBO in Table S6, Supporting Information). However, NBO 181 (C11-C12) showed a much less electrophilic character, with Δf i NBO values below NBOs 172 and 176 respectively located on carbonyl group at C-7 and on carbonyl of benzoate substituent at C-9. Furthermore, NBOs 161 (C11-C12) and 162 (C5-C6) § are highly stabilized since they are located below the occupied NBOs 163-171 in the orbital diagram, disfavouring the orbital interactions (Table S6, Supporting Information; For correspondence between jatrophane and gaussian numbering used in table S6, see Table S5 of Supporting Information). If a non-concerted mechanism is considered, two new lone pairs type orbitals appeared in Fig. 7. Proposed biosynthetic routes.
C-5 (NBO 172) and in the oxygen of the carbonyl group at C-7 (NBO 163), and a new double bond C6-C7 (NBO 160) is formed (Table S7, supporting information). The proposed resonance structures of unpaired electrons in the triple state are shown in Fig. 8. The comparison of the resonance structures with the orbital diagram for this state showed a full correspondence of structure II with the new orbitals (see Fig. S7) described for the triplet-state (Table S7, Supporting Information; For  correspondence between jatrophane and gaussian numbering used in  table S7, see Table S5 of Supporting Information). The Δf i NBO values (Table S7, supporting information) indicated that the radical electron NBO 172 (C5) § (Δf i NBO − 0.386) has the highest nucleophilic character and the second radical electron is found in orbital 163 (C7-CO) § (Δf i NBO − 0.042) that has a slight nucleophilic character.
On the other hand, the π bonding NBO 160 (C6-C7) § showed the highest electrophilic character (Δf i NBO 0.295). This information indicates that C5 is very nucleophilic and C6 very electrophilic (Fig. S7, Supporting Information). For this reason, the interaction with the occupied orbital 162 (C11-C12) and the virtual orbital 182 (C11-C12) § would give rise to the C5-C11 and C12-C6 bonds in gaditanone (2). These interactions were corroborated in the SOPTA study, which showed two important interactions. The first one between the NBO 163 (C7-CO) § and the π antibonding NBO 173 (C6-C7) § with a value for energy of 8.19 kcal/mol, and a second interaction of NBO 163 (C7-CO) § with the anti-bonding NBO 194 (C7-C8) § with energy equal to 9.99 kcal/mol. These interactions would stabilize by delocalization the structures shown in Fig. 8. In contrast to pathway A, the excitation from singlet to triplet seems to be determinant in the change of reactivity of the orbitals involved in the pathway B.

Conclusions
The bond reactivity indices study suggests that the intramolecular [2 + 2] cycloaddition leading to gaditanone (2) proceeds through a photochemical non-concerted mechanism of the enone 6. The stereoselectivity of the cycloaddition could be understood by the delocalization of electron density over specific NBOs presumably originated from the steric effects of the substituents in the transition state avoiding the cis/trans isomerization (Corey et al., 1964). Furthermore, the formation of a cyclohexanone moiety in the 5/6/4/6-fused tetracyclic ring skeleton both in gaditanone (2) (Flores-Giubi et al., 2017) and heliosterpenoids A and B (Mai et al., 2017) seems to play a structural stabilization role, favouring the cyclization of the jatrophane precursors.

General experimental procedures
Unless otherwise noted, materials and reagents were obtained from commercial suppliers and were used without further purification. Optical rotations were determined with a digital polarimeter. Infrared spectra were recorded on a FT-IR spectrophotometer and reported as wave number (cm − 1 ). ECD and UV spectra were recorded on a JASCO J-1500 CD spectrometer. 1 H and 13 C NMR measurements were recorded on Agilent 500 MHz NMR spectrometer with SiMe 4 as the internal reference. Chemical shifts were referenced to CDCl 3 (δ H 7.25, δ C 77.0). NMR assignments were made using a combination of 1D and 2D techniques. High resolution mass spectroscopy (HRMS) was performed in a QTOF mass spectrometer in the positive ion ESI mode. Purification by semi-preparative and analytical HPLC was performed with a Hitachi/ Merck L-6270 apparatus equipped with a differential refractometer

Extraction and isolation
The air-dried plant (1.8 kg) was powdered and extracted with 2.5 L of MeOH at room temperature for 3 × 24 h. The MeOH extract from the aerial parts and roots were evaporated under reduced pressure to yield a crude extract of 92 g and 11 g, respectively. The extract of the aerial was suspended in water (1 L) and then partitioned with petroleum ether (2 L), dichloromethane (1 L), and ethyl acetate (1 L) sequentially to give the corresponding petroleum ether (10.9 g), dichloromethane (0.5 g) and ethyl acetate (27.2 g) fractions. The petroleum ether-soluble extract showed diterpene signals according to 1 H NMR spectra and a portion (1.36 g) was subjected to column chromatography over silica gel, eluting with an increasingly polar gradient of ethyl acetate (0-100%) in petroleum ether, to afford 16 fractions (A1-A16), according to TLC analysis. Fractions (A6-A10) presented similar TLC profiles and were combined and further purified by analytical HPLC, to yield 1 (54 mg), 2 (1.4 mg), 3 (13.6 mg), 4 (8.2 mg), 5 (6.8 mg) and 6 (14.6 mg).

Computational methods
Conformational analysis of 3 was performed by using the semiempirical method PM6. The energies, oscillator strengths, and rotational strengths of the first 20 electronic excitations were calculated using a TD-DFT-B3LYP/6-31G++ (d,p) level of theory. The ECD spectrum of the conformer was simulated using a Gaussian function with a half-bandwidth of 0.33 eV. The corresponding theoretical ECD spectrum of the enantiomer (ent-3) was depicted by inverting that of 3. All quantum computations were performed using Gaussian 16 program package (Frisch et al., 2016), on a high performance computer located at CICA (Centro Informático Científico de Andalucía). In the 200-400 nm region, the theoretically calculated ECD spectrum of 3 was in good agreement with the experimental ECD spectrum (Fig. 4). This supported the assignment made of the absolute configuration for 3.
A detailed description of the methodology implemented in this version of the program has been included in reference (Morales-Bayuelo et al., 2018) in terms of various examples. Molecular orbital images have been obtained with GaussView software (Dennington et al., 2008) and second order perturbational theory analysis has been performed with the code NBO 3.0 (Carpenter and Weinhold, 1988;Glendening et al., 2001;Polo et al., 2019).

HIV-1 reactivation assays
Jurkat-LAT-GFT cells were used in the HIV-1 reactivation assays following the procedure described in the literature (Daoubi et al., 2007).

Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.