New Triterpenes from Maytenus robusta: Structural Elucidation Based on NMR Experimental Data and Theoretical Calculations

Leaves of Maytenus robusta (Celastraceae) were subjected to phytochemical investigation mainly directed at the isolation of pentacyclic triterpenes. The compounds friedelin (1), β-friedelinol (2), 3-oxo-21β-H-hop-22(29)-ene (7), 3,4-seco-friedelan-3,11β-olide (8), 3β-hydroxy-21β-H-hop-22(29)-ene (9), 3,4-seco-21β-H-hop-22(29)-en-3-oic acid (10), 3,4-seco-friedelan-3-oic acid (11), and sitosterol were identified in the hexane extract of M. robusta leaves. Compounds 8 and 9 are described herein for the first time. The structure and stereochemistry of both compounds were experimentally established by IR, HRLC-MS, and 1D (1H, 13C, and DEPT 135) and 2D (HSQC, HMBC and COSY) NMR data and supported by correlations with carbon chemical shifts calculated using the DFT method (BLYP/6-31G* level). Compounds 7 and 10 are also described for the first time, and their chemical structures were established by comparison with NMR data of similar structures described in the literature and correlations with BLYP/6-31G* calculated carbon chemical shifts. Compound 9, a mixture of 11 and sitosterol, and 3β,11β-dihydroxyfriedelane (4) were evaluated by the Ellman’s method and all these compounds showed acethylcholinesterase inhibitory properties.


Introduction
Secondary metabolites are isolated from plants and animals, and many of them have been used as sources of derivatives with a large spectrum of biological activities [1], including effects in the treatment of Alzheimer's disease (AD). AD is a progressive neurodegenerative disorder characterized by a decline in memory and cognitive abilities. About 34 million people around the World have AD, being the major cause of dementia in elderly people [2]. Acetylcholinesterase (AChE) inhibitors are a group of drugs frequently investigated for the symptomatic treatment of AD [3]. Alternatively, the literature also describes some relationships between pentacyclic triterpenes and treatments for AD [4][5][6][7].

Compounds 1 and 7
The hexane extract (HE) fractions eluted with 9:1 hexane-chloroform provided a white solid. The IR spectrum of this solid shows two intense absorptions at 1,714 and 1,701 cm −1 which are attributed to carbonyl groups. The 13 C-NMR spectrum shows two groups of 30 signals, each with significant differences in intensity. The 13 C-NMR data of the group of high-intensity signals (named triterpene 1) present a signal at δ C 213.2 which is characteristic of a carbonyl carbon. The NMR data are similar to the corresponding ones described in the literature for the triterpene friedelin [31]. In turn, the 13 C-NMR data of the group of low-intensity signals (named triterpene 7) present a signal at δ C 218.2 which is also characteristic of a carbonyl carbon. Two other signals at δ C 148.6 and 110.1 (non-hydrogenated and methylene carbon atoms, respectively) are characteristic of alkenyl carbon atoms. The 1 H-NMR data show a signal at δ H 4.78 (integrated for two hydrogen atoms) which is also characteristic of an alkenyl group. These signals are in agreement with a hopane-type skeleton containing a carbonyl carbon. The 13 C-NMR data of 7 were compared with the hopane-type skeleton data compiled in the literature for 12 and 13 (see Figure 2 and Table 1). Triterpene 7 only differs in relation to the substituent at C-3 of the ring A of 12 and the stereochemistry of the C-21 in the ring E of 13. In fact, the NMR data of C-1 to C-15 of 13 are very similar to the corresponding data of 7. On the other hand, the NMR data of C-14 to C-30 (except for C-23 and C-24) of 12 are very similar to the corresponding data of 7. As result, it can be proposed that the chemical structure of 7 is a combination of the rings A-C of 13 and rings C-E of 12. The chemical structure of 7 is thus in agreement with that of the compound 3-oxo-21β-H-hop-22(29)-ene, a triterpene which was not yet described in the literature. Moreover, the intensity and integration of the carbonyl carbon atom signals based on quantitative 13 C-NMR analysis indicates a mixture 2:1 of compounds 1 and 7, respectively. BLYP/6-31G* geometry optimization calculations were carried out for 7 with a starting geometry based on the stereochemistry proposed to 3-oxo-21β-H-hop-22(29)-ene (see Figure 2). The most stable optimized geometry of 7 (E = −1246.52767272 a.u.) presents rings A-D in the chair conformation and ring E in the envelope one. Moreover, the C-29 of the allyl group is positioned close to the methyl group at C-28. Carbon chemical shift calculations were carried out on the optimized geometry of 7 at the same level of theory. Correlations between calculated and experimental 13 C-NMR carbon chemical shift values of the data of 7 (Table 1) provided high correlation coefficient (R 2 = 0.99330) and slope of the R 2 curve (α = 0.91728). These theoretical results are also in agreement with the stereochemistry of 3-oxo-21β-H-hop-22(29)-ene proposed for 7. Table 1. 13 C-NMR data of triterpene 7, compared with the corresponding data described in the literature for 12 [26] and 13 [27], and 13 C-NMR data of triterpene 10, compared with the corresponding data described in the literature for 14 [28] and 15 [26].

Carbon
Compound/ C 7 12  The HE fractions eluted with 7:3 hexane-chloroform provided a white solid (named triterpene 2). The IR spectrum shows an absorption at 3471 cm −1 , which is attributed to a hydroxyl group. The absorptions at 1384 and 1172 cm −1 can be attributed to the asymmetric and symmetric C-O stretches, respectively. The 1 H-NMR and 13 C-NMR spectra shows a large signal at  H 3.74 and a signal at  C 72.8 which are characteristic of a carbinolic carbon. The NMR data are similar to the corresponding ones described in the literature for the triterpene β-friedelinol [32].
These theoretical results are in agreement with the stereochemistry of 3,4-seco-friedelan-3,11-olide proposed for 8, a triterpene not yet described in the literature.

Compound 9
The HE fractions eluted with 3:2 hexane-chloroform also provided a white solid with molecular formula C 30 Table 3 Table 3. NMR data of triterpene 9 and corresponding data described in the literature for 12 [26]. The NMR analyses of 9 are in agreement with the data of the triterpene 3-hydroxy-21-H-hop-22(29)-ene. In fact, the 13 C-NMR data of 9 were compared with the hopane-type skeleton data compiled in the literature for 12 [26], and seen to only present a significant difference in the stereochemistry at C-3. The NMR data of C-6 to C-8, C-10 to C12, C-14 to C-23, and C-27 to C-30 of 9 are very similar to the corresponding data of 12 (Table 3). BLYP/6-31G* geometry optimization calculations were carried out for 9 with a starting geometry based on the stereochemistry proposed to 3-hydroxy-21-Hhop-22(29)-ene ( Figure 2). The most stable optimized geometry (E = −1247.70633974 a.u.) presents the rings A, B, C, and D in the chair conformation and the ring E in the envelope one. Moreover, the C-29 of the allyl group is positioned close to the methyl group at C-28. Carbon chemical shift calculations were carried out to the optimized geometry of 9 at the same level of theory (BLYP/6-31G*). Correlations between values of calculated carbon chemical shifts and experimental 13 C-NMR data of 9 (Table 3) provided a high correlation coefficient (R 2 = 0.98817) and slope of the R 2 curve ( = 0.93702). These theoretical results are in agreement with the stereochemistry of 3-hydroxy-21-H-hop-22(29)-ene for 9, a triterpene not yet described in the literature.

Compounds 10 and 11
The HE fractions eluted with 1:1 hexane-chloroform provided a white solid. The IR spectrum of the solid shows a large absorption at 3250-2700 cm −1 and an intense absorption at 1701 cm −1 which are characteristic of a carboxylic acid group. Moreover, the absorptions at 1284 and 1049 cm −1 can be attributed to the asymmetric and symmetric C-O stretches, respectively. The 1 H-NMR spectrum shows a broad signal at  H 4.78 (integrating for two hydrogen atoms) which is characteristic of an alkenyl group. The 13 C-NMR spectrum shows two groups of 30 signals, each with significant differences in intensity. The 13 C-NMR data of the group of low-intensity signals (named triterpene 10) present a signal at  C 180.1 which is characteristic of a carboxylic carbon. The signals at  C 148.7 and 110.1 (non-hydrogenated and methylenic carbon atoms, respectively) are characteristic of an alkenyl group. The 13 C-NMR data of 10 were compared with the corresponding data compiled in the literature for 14 and 15 (see Table 1). Triterpene 10 only differs in the position of the allyl group and the opening of the ring A in relation to 14 and 15, respectively (see Figure 2). The NMR data of C-1 to C-5, C-8, C-9, and C-11 of 14 are very similar to the corresponding data of 10. On the other hand, the NMR data of the C-12 to C-22 of 10 are very similar to the corresponding data of 15. As result, it can be proposed that the chemical structure of 10 is in agreement with the structure of the triterpene 3,4-seco-21-H-hop-22(29)-en-3-oic acid, a triterpene which was not yet described in the literature. BLYP/6-31G* geometry optimization calculations were carried out for 10 with starting geometry based on the stereochemistry proposed for 3,4-seco-21-H-hop-22(29)-en-3-oic acid (see Figure 2). The most stable optimized geometry (E = −1322.93835876 a.u.), which does not have the ring A, presents the rings B, C, and D in the chair conformation and ring E in the envelope one. Carbon chemical shift calculations were carried out to the optimized geometry of 10 at the same level of theory (BLYP/6-31G*).
Correlations between values of calculated carbon chemical shifts and experimental 13 C-NMR data of 10 (Table 1) provided high correlation coefficient (R 2 = 0.97833) and slope of the R 2 curve ( = 0.87424). These theoretical results are in agreement with the stereochemistry of 3,4-seco-21-Hhop-22(29)-en-3-oic acid proposed for 10, a triterpene not yet described in the literature. In turn, the 13 C-NMR data of the group of high intensity signals (named triterpene 11) also presents a signal characteristic of carboxyl carbon (at  C 178.2). The 13 C-NMR data of 11 are similar to the corresponding data described in the literature for 3,4-seco-friedelan-3-oic acid [33]. Moreover, the intensity and integration of the carbonyl carbon atom signals based on quantitative 13 C-NMR analysis indicates a mixture 2:3 of compounds 10 and 11, respectively.

In Vitro AChE Inhibitory Activity
The AChE activity was measured for the triterpenes 4, 9, and mixture of 11 and sitosterol which were previously obtained from the leaves of M. robusta. The calorimetric method of Ellman was adapted for 96-well microplates in the assays at 25 °C [30]. The triterpenes 4 and 9 showed (64 ± 3)% and (76 ± 1)% of inhibition, respectively. The mixture of triterpene 11 and sitosterol exhibited very significant results, i.e., (94 ± 1)% of inhibition.

General Procedures
Uncorrected melting points were determined using a Microquímica apparatus, model MQAPF-302. Optical rotations were measured on a Perkin-Elmer Model 341 polarimeter using a 100 mm, 1.0 mL capacity cell. The IR spectra were taken on a Perkin Elmer-Spectrum One (ATR) spectrometer. The 1 H and 13 C-NMR spectra at 400.129 and 100.613 MHz, respectively, as well as the COSY, HSQC, and HMBC experiments were performed on a Brüker DRX400 AVANCE spectrometer, using CDCl 3 or a mixture of CDCl 3 /pyridine-d 5 as solvent, with direct or inverse probes and a field gradient. The chemical shifts were registered in ppm () relative to TMS as the internal standard. The coupling constants (J) were registered in Hertz. HR-APCIMS spectra were acquired on a Shimadzu LCMS-IT-TOF system. Analyses were carried out using manual injection. The samples were dissolved in CHCl 3 and then diluted with MeOH. Column chromatography (CC) processes were carried out using silica gel 60 (70-230 Mesh). Thin layer chromatography (TLC) processes were carried out using precoated silica gel plates.

Extraction and Isolation of Constituents
Leaves of M. robusta were dried at room temperature until a constant weight was achieved (about one week) and finally powdered. A sample of this material (864.4 g) was submitted to extraction with hexane (3 L, 5 days, room temperature). A solid material (SM; 4.51 g) precipitated during solvent evaporation, being separated by filtration under reduced pressure. The SM was submitted to column chromatography using silica gel as the stationary phase (CCS) eluted with hexane, chloroform, ethyl acetate, and methanol in increasing polarity order. The triterpenes 1-6 ( Figure 1) were obtained, as previously reported [18].
The rest of the hexane extract provided a viscous crude oil (HE; 32.0 g) after complete solvent evaporation. A part of HE (31.43 g) was submitted to CCS eluted with hexane, chloroform, ethyl acetate, and methanol in increasing polarity order. The HE fractions eluted with hexane-chloroform (9:1) were again submitted to CCS eluted with hexane and chloroform in increasing polarity order. The fractions eluted with hexane-chloroform (1:1) provided a white solid (13.5 mg) which was identified as a mixture of the triterpenes 1 and 7. The HE fractions eluted with hexane-chloroform (4:1) provided a white solid (624.0 mg) which was identified as triterpene 1. The HE fractions eluted with hexane-chloroform (7:3) provided a solid (566.1 mg) which was identified as triterpene 2.
The HE fractions eluted with hexane-chloroform (3:2) were again submitted to CCS eluted with hexane, chloroform, ethyl acetate, and methanol in increasing polarity order. The fractions hexane-chloroform (3:7) provided a white solid (14.1 mg) which was identified as triterpene 8. The fractions eluted with chloroform (289.0 mg) were submitted to CCS eluted with chloroform, providing a white solid (103.0 mg) which was identified as triterpene 9.

Theoretical Methodology
Theoretical studies were carried out using the Gaussian 03 software package [34]. The geometries obtained from PM3 semi-empirical calculations were used as initial models in geometry optimizations employing DFT calculations with the Pople's split valence basis set 6-31G*. BLYP exchange-correlation functional was used in DFT calculations. The optimized geometries were characterized as true minima on the potential energy surface (PES) when all harmonic frequencies were real. The electronic-nuclear energy (E) of the optimized geometries was given in atomic unit (Hartree). This theoretical methodology has been efficiently employed in the study of different organic compounds, including terpenes [35][36][37][38].
The optimized geometries were used to calculate carbon chemical shifts at the same levels of theory. Values of calculated carbon chemical shift ( C ) were determined in relation to the corresponding calculated value for tetramethylsilane ( C 187.97). Correlations between  C values and experimental carbon chemical shifts ( C ) were obtained using software package Origin™ Standard 7.5.
The  C and  C values were plotted on the x and y axes, respectively. The  C / C correlation curves were given as linear fits with correlation coefficients (R 2 ) and slope of the R 2 curve () furnished by the program. The BLYP/6-31G* calculations usually give satisfactory results of carbon chemical shifts, as have been obtained in previous works [39][40][41].
Volumes of acetylthiocholine iodide (25 μL, 15 mM in water), DTNB (125 L, 3 mM in buffer A), buffer B (50 μL), and sample (25 μL, 10 mg/mL in MeOH diluted 10-fold with buffer C, resulting in a concentration of 1 mg/mL) were added into each well of a 96-well microplate. Instead of adding the sample solution, a volume of 25 μL of buffer C was employed to prepare the blank sample. The positive control was prepared under the same conditions, using physostigmine (eserine) as standard. Tests were carried out in quintuplicate. The absorbance was measured at 405 nm every 60 s by eight times using a Elisa Thermoplate microplate reader. After addition of 25 μL of acetylcholinesterase solution (0.226 U/mL in buffer B), the absorbance was again read every 60 s for ten times. The increase in absorbance relative to substrate spontaneous hydrolysis was corrected by reaction rate variation before and after addition of the enzyme. The inhibition percentage was calculated by comparing the rates of the sample with the blank.

Conclusions
The hexane extract of the leaves of M. robusta provided seven triterpenes. The triterpenes 1, 2, and 11 were also isolated in a previous phytochemical investigation. The triterpenes 8 and 9 are described for the first time in the literature. The triterpenes 7 and 10 are also new compounds, but both compounds were obtained as a mixture. Hopane and seco-hopane triterpenoids are not usual in species of the family Celastraceae. The combination of experimental NMR analyses with carbon chemical shift calculations was a useful procedure for the structural determination of these hopane and friedelane triterpenes. Compounds 4, 9, and the mixture of 11 and sitosterol showed acetylcholinesterase inhibitory properties. These compounds present hopane-and friedelane-type skeletons, suggesting biological potential of their derivatives for Alzheimer's desease.

Supplementary Material
Figures with liner fit curves obtained from correlations between experimental and calculated carbon chemical shifts are shown as Supplementary Material, which can be accessed at: http://www.mdpi.com/1420-3049/17/11/13439/s1. Tables with the geometric parameters and other results of all the optimized structures considered in this work are available from the authors upon request.