Molecular Disorder in (‒)-Encecanescin

(‒)-Encecanescin (1) has been isolated from the leaves of Eupatorium aschembornianum. Two conformers are present in the crystal structure as a result of molecular disorder. The structure of 1 was established by 1H- and 13C-NMR spectroscopy in CDCl3 solution using 2D NMR techniques (gHSQC, gHMBC and NOESY). A Monte Carlo random search using molecular mechanics followed by the geometry optimization of each minimum energy structure using density functional theory (DFT) calculations at the B3LYP/6–31G* level and a Boltzmann analysis of the total energies generated accurate molecular models describing the conformational behavior of 1. The three most stable conformers 2–4 of compound 1 were reoptimized at the B3LYP/6-311++G(d,p) level of theory using CHCl3 as a solvent. Correlations between the experimental 1H- and 13C-NMR chemical shifts (δexp) have been found, and the GIAO/B3LYP/6-311++G(d,p) calculated magnetic isotropic shielding tensors (σcalc) for conformers 2 and 3, δexp = a + b σcalc, are reported. A good linear relationship between the experimental and calculated NMR data has been obtained for protons and carbon atoms.

reported.A good linear relationship between the experimental and calculated NMR data has been obtained for protons and carbon atoms.Keywords: encecanescin; DFT calculations; X-ray diffraction; FTIR and NMR spectra

Introduction
Chromene (2H-1-benzopyran) ring derivatives are often found in natural heterocycles, and some have interesting biological activities [1].These compounds make up a new family of activators of potassium channels that are useful in the treatment of respiratory diseases as tracheal tissue relaxing agents [2].One such benzopyran derivative isolated from Ageratina asenii [3] is (+)-encecanescin, a dimeric chromene with a structure similar to a previously reported compound [4].Surprisingly, however, the authors found that the available encecanescin crystallized as a racemic mixture.
Later, crystals of (±)-encecanescin were analyzed by X-ray diffraction [5], revealing that it racemized during crystallization using 1:1 EtOAc/cyclohexane.In addition, it was possible to observe two molecules in the asymmetric unit that differ in geometry around C2.The rings attached to C2 exhibit a half-boat conformation, but differ in the orientation of the gem-dimethyl group present at C2.Our group recently obtained a white solid from Eupatorium aschembornianum by recrystallization using 95:5 hexane/EtOAc , which provided (-)-encecanescin (1) for the first time as a single crystal, with the structure shown in Figure 1 [6].However, the molecular crystal also exhibited disorder in the X-ray structure.In the present work, this disorder is explained in terms of two conformers in the solid state, and it is shown that this disorder can be deduced from quantum mechanical calculations.

Results and Discussion
(-)-Encecanescin (1) was isolated using a previously reported protocol [6].The dimeric structure of compound 1 was confirmed from the high-resolution mass spectra (MS-FAB+), which exhibited a peak at 450.2402 for C 28 H 34 O 5 .The compound was identified using NMR analysis by comparing the results with those previously reported [3,4], with the exception of the resonances for C-3, C-3'; C-4, C-4'; and C-5, C-5', which were reassigned in this study to 127.3, 122.7 and 124.0, respectively, based on gHMBC, gHSQC and NOESY spectra.

X-ray Crystallography
Crystals of 1 were grown from a 95:5 mixture of n-hexane with ethyl acetate.A crystal cut to the dimensions 0.24 × 0.20 × 0.18 mm was used for X-ray measurements at 293 K using an Enraf-Nonius KappaCCD diffractometer with graphite-monochromated λ Mo-Ka = 0.71073 Å.The structure was solved by direct methods and refined by full-matrix least-square calculations based on F 2 .Crystallographic calculations were performed using SHELXL-97 [7].The details of the crystal structure determinations and refinements are presented in Table 1.The best results were obtained for the disordered model in the crystal, and two conformers (1a, 1b) are found (Figure 1).The rings containing C2 in 1a and 1b show a half-boat conformation but differ in the orientation of the flag atom C2.

B3LYP Calculations
The theoretical conformational distribution of 1 was obtained by a Monte Carlo random search.A total of 10 minimum energy structures were found within a molecular mechanics energy range of 10 kcal•mol −1 .All of these structures were subjected to geometry and energy optimization by density functional theory (DFT) calculations employing the B3LYP/6-31G* basis set.According to these calculations, the original group of 10 structures was reduced to a group of three (within a 0-3 kcal•mol −1 range), as seven conformers appeared as duplicates.These three structures were submitted to geometry reoptimization using DFT calculations at the B3LYP/6-311++G(d,p) level of theory in a CHCl 3 solution.Figure 2 shows the total DFT energy in solution, the relative energy and the conformational population of the three optimized conformers of 1 (2, 3 and 4), which account for 99.99% of the conformational population according to the DFT total energy values.Geometry optimizations included a frequency calculation to verify that an energy minimum had been reached.Given that conformer 4 has a relative energy of 2.263 kcal•mol −1 , its contribution to the equilibrium (ca.1.1%) can be neglected.

Figure 2. Conformational distribution of (-)-encecanescin (1).
The selected calculated B3LYP bond lengths, bond angles and torsion angles are given in Table 2. Most of the calculated bonds are slightly longer than the experimental ones, except C(10)-C(5), C(10)-C(9), C(2)-O(1) and C(9)-C (8), which are shorter.The calculated bond angles agree with the experimental values within 1.3°, excluding angles within the rings containing C2, ranging from 5.2° to 35.6°.The largest differences between the X-ray and B3LYP data are in the torsion angles, which vary from 2.5° to 69.6°.The B3LYP calculations accurately reproduce the signs of the torsion angles.

FTIR and Raman Spectra
The observed and calculated harmonic frequencies of the two conformers (2 and 3) of (-)-encecanescin (1) and their tentative assignments are presented in Table 3.A comparison of the calculated and experimental frequencies reveals important differences.Two factors may be responsible for the disagreements between the experimental and computed spectra of the studied structures.The first is that the experimental spectrum was recorded for the molecule in the solid state, while the computed spectra correspond to isolated molecules in CHCl 3 solution.The second is the fact that the experimental values correspond to anharmonic vibrations, while the calculated values correspond to harmonic vibrations.The overestimation of the computed wavenumbers is quite systematic, and a scaling procedure was used to obtain the predicted frequencies [8,9].

1 H-and 13 C-NMR Spectra
The 1 H-and 13 C-NMR signals of 1 were assigned based on the observed gHSQC, gHMBC and NOESY correlations in CDCl 3 [10] and are listed in Tables 4 and 5.The gHSQC spectrum (Figure 3) shows cross peaks between the resonances of 1 H and those of the 13 C atoms to which the protons are attached.The horizontal axis corresponds to the 1 H spectrum and the vertical one to the 13 C spectrum.On the other hand, in the HMBC spectrum, correlations between the protons or carbons through two and three bonds are observed (Figure 3).Starting from the characteristic resonance of the H-15 methoxyl proton (δ 3.67), it was possible to assign the resonance of the sp 2 carbon C-7 (δ 157.7) based on its gHMBC correlation with H-15.On the other hand, the signal at δ 6.32 was assigned to H-8 due to its cross peak with CH 3 O (δ 3.67) in the NOESY plot.Analogously, the methine proton resonating at δ 4.59 produces cross peaks with carbons C-5 (δ 124.0) and C-6 (δ 125.1), thus revealing its position on C-11.The resonance of methyl protons C( 12)H 3 at δ 1.46 is coupled through two bonds to the carbon C-11.Furthermore, C-7 correlates in the gHMBC through two and three bonds with the protons at δ 6.32 and δ 7.10; therefore, these signals must be assigned to the protons H-8 and H-5, respectively.Similarly, relevant cross peaks were observed for proton H-5 at δ 7.10 through three bonds with the carbons C-4 at δ 122.7 and C-11 at δ 68.6, confirming its assignment to H-5.In addition, the signal at δ 152.8 correlates in the gHMBC through three bonds with the protons H-5 at δ 7.10 and H-4 at δ 6.34; therefore, this signal must be assigned to carbon C-9.The resonances of the methyl protons C(13)H 3 and C( 14)H 3 at δ 1.46 and δ 1.43 are coupled through three bonds to carbon C-3 at δ 127.3.Thus, the signal at δ 5.46 was assigned to H-3 based on its cross peak with C(13)H 3 and C( 14)H 3 and H-4 at δ 6.34 in the NOESY plot.The signal at δ 113.8 correlates in the gHMBC through three bonds to the carbon with the protons H-8 and H-3; therefore, this signal must be assigned to carbon C-10.Accordingly, the remaining carbons C-8 and C-2, resonating at δ 99.2 and δ 78.3, respectively, were assigned based on the gHSQC correlations.
Table 5. Experimental chemical shifts (δ exp , CDCl 3 ) vs. the isotropic magnetic shielding tensors (σ calc ) from the GIAO/B3LYP/6-311++G(d,p) calculations for encecanescin (1); δ exp = a+b•σ calc : (a) 13 C (a = 173.81;b = −0.946;r 2 = 0.9986) and (b) 1 H (a = 32.101;b = −1.0124;r 2 = 0.9952).The relationship between the experimental 13 C and 1 H chemical shifts (δ exp ) and the GIAO (gauge-independent atomic orbitals) magnetic isotropic shielding constants (σ calc ) calculated for conformers 2 and 3 in CHCl 3 are generally linear and are well described by the equation δ exp = a + b•σ calc [11].The slope and intercept of the least-squares correlation line (Figure 4a,b, Tables 4 and 5) are utilized to scale the GIAO magnetic isotropic shielding constants, σ calc , and to predict the chemical shifts, δ pred = a + b•σ calc .The correlations between the experimental chemical shifts and calculated magnetic isotropic shielding constants are generally better for carbon-13 atoms than for protons; however, in this case, the correlations are good for both carbons and protons.This finding can be explained by the absence of hydrogen bonds and other strong interactions that mainly affect outer H atoms.The magnetic isotropic shielding constants confirm the correct assignments of the chemical shifts to the aforementioned atoms.

General
The infrared spectrum was recorded on a Varian FT-IR spectrometer (Palo Alto city, CA, USA).The Raman spectrum of a crystalline sample was measured using a Thermo Scientific DXR Raman microscope (Madison, WI, USA) equipped with a 532 nm laser with a power of 10 mW and an exposure time of 104 s.High-resolution mass spectroscopy was performed using a JEOL spectrometer (model 102 ASX, Jeol).
The 13 C-and 1 H-NMR spectra were recorded on an Agilent 400 MR DD2 spectrometer (Santa Clara, CA, USA) operating at 100 MHz for 13 C and 400 MHz for 1 H.The 13 C and 1 H chemical shifts were measured in CDCl 3 relative to TMS as an internal standard.Typical conditions for the proton spectra were as follows: pulse width of 45°, acquisition time of 2.5 s, FT size of 32 K and digital resolution of 0.3 Hz per point.Typical conditions for the carbon spectra were as follows: pulse width of 45°, FT size of 65 K and digital resolution of 0.5 Hz per point.The number of scans varied from 1200 to 10,000 per spectrum.All proton and carbon-13 resonances were assigned by 1 H (NOESY) and 13 C (gHSQC, gHMBC), respectively.All 2D NMR spectra were recorded at 298 K on the Agilent 400 MR DD2 spectrometer operating at 100 MHz ( 13 C) and 400 MHz ( 1 H), with a FT size of 2 × 2 K and a digital resolution of 0.3 Hz per point.

Materials
Eupatorium aschembornianum leaves were collected in San Juan Tlacotenco, Tepoztlan, Morelos State, México, during August 2007.A specimen from the original collection can be found in "Jorge Espinosa Herbarium-Hortorio" in the Biology Area of Chapingo Autonomous University, with voucher number 1835.

Conclusions
(−)-Encecanescin (1) has been isolated from leaves of Eupatorium aschembornianum.The structure of 1 was established by X-ray diffraction and characterized by FTIR, Raman and NMR spectroscopy and DFT calculations.The X-ray analysis showed that the molecule is non-planar and is present as a mixture of two conformers in the crystal (2 and 3).Molecular modeling of 1 using the Monte Carlo protocol followed by geometry optimization at the B3LYP 6-31G(d,p) level of theory and a Boltzmann analysis of the total energies confirmed that 2 and 3 are the two most stable conformers of 1. Good correlations between the experimental 1 H and 13 C chemical shifts in CHCl 3 and the GIAO/B3LYP/6-311++G(d,p) calculated magnetic isotropic shielding tensors for both conformers (δexp = a + b•σ calc ) confirmed the geometry of 1.

Table 4 .
(1)bon-13 chemical shifts (δ, ppm) in CDCl 3 and calculated GIAO nuclear magnetic shielding (σ cal ) for (−)-encecanescin(1).The predicted GIAO chemical shifts were computed from the linear equation δ exp = a + b•σ calc with a and b determined from the fit to the experimental data.