Contribution to the Special Thematic Project “Advances in Chiroptical Methods”
Regular Article/Advances in Chiroptical Methods
Diastereoisomeric assignment in a pacifenol derivative using vibrational circular dichroism†
Version of Record online: 6 NOV 2009
Copyright © 2009 Wiley-Liss, Inc.
Special Issue: Advances in Chiroptical Methods
Volume 21, Issue 1E, pages E208–E214, 2009
How to Cite
Muñoz, M. A., Chamy, C., Carrasco, A., Rovirosa, J., San Martín, A. and Joseph-Nathan, P. (2009), Diastereoisomeric assignment in a pacifenol derivative using vibrational circular dichroism. Chirality, 21: E208–E214. doi: 10.1002/chir.20801
- Issue online: 26 FEB 2010
- Version of Record online: 6 NOV 2009
- Manuscript Accepted: 20 AUG 2009
- Manuscript Received: 8 JUL 2009
- Proyecto Anillo ACT-38
- diastereoisomeric assignment;
- relative configuration;
The configuration of a chiral center in semisynthetic (−)-(2R,5R,5aR,8ζ,9aS)- 2,8-dibromo-2,5,9,9a-tetrahydro-5-hydroxy-5,8,10,10-tetramethyl-6H-2,5a-methano-1-benzoxepin-7(8H)-one (3 or 4), prepared in two steps from (−)-(2R,5R,5aR,7S,8S,9aS)-2, 7-dibromo-8-chloro-2,5,7,8,9,9a-hexahydro-5,8,10,10-tetramethyl-6H-2,5a-methano-1-benzoxepin-5-ol, known as pacifenol 1, has been determined using vibrational circular dichroism (VCD) measurements. The vibrational spectra (IR and VCD) of diastereoisomers 3 and 4 were calculated using density functional theory (DFT) at the B3LYP/DGDZVP level of theory for the two conformers that in each case account for the total energetic distribution found in the first 10 kcal/mol range. The DFT conformational optimization of the 8R diastereoisomer 3 indicates the cyclohexanone exists almost exclusively in a boat conformation with a β-equatorial bromine atom and an α-axial methyl group at the chiral center alpha to the carbonyl group, while for the 8S diastereoisomer 4 a 5:1 conformational distribution in favor of a chair conformation with an α-axial bromine atom and a β-equatorial methyl group is calculated, suggesting due to well-known chair versus boat relative stabilities that the plausible diastereoisomer would be the 8S molecule. A comparison of the IR spectrum of the reaction product with the calculated spectra of 3 and 4 provided no means for the diastereoisomeric assignment, while from comparison of the VCD spectra it became immediately evident that the rearranged molecule possesses the 8R absolute configuration as shown in 3, in concordance with a single crystal X-ray diffraction study that could be refined to an R-factor of 2.9%. Chirality 21:E208–E214, 2009. © 2009 Wiley-Liss, Inc.
Pacifenol 1 is the first sesquiterpene of marine origin containing chlorine and bromine atoms that was originally isolated from the red alga Laurencia pacifica1 and has afterwards been isolated from many species of the genus Laurencia.2–6 The structure and absolute configuration of 1 followed1 from a very early single crystal X-ray analysis, performed using precession and Weissenberg cameras, which was later repeated to find a more accurate molecular representation.7 Detailed 1H- and 13C NMR studies8 of chamigrenes, which include pacifenol 1, are reported and several biological evaluations of these compounds are well-documented in the literature.9–11 The closely related dibrominated chamigrene 2, found as a constituent12 of Laurencia nipponica, could be prepared in quantitative yields after treatment of 1 with sodium hydride.13 In turn compound 2 is the first vinyl bromide chamigrene metabolite of marine origin. When diene 2 was reacted14, 15 with m-chloroperbenzoic acid it was cleanly transformed into a single alpha-bromoketone (3 or 4, Scheme 1) in which only one carbon–carbon double bond was oxidized and the vinyllic bromine atom underwent an unprecedented 1,2-bromine atom shift. Although the chirality of the newly formed stereogenic center, containing the migrated bromine atom, is not trivial to be ascertained mainly due to the lack of suitable NMR correlations, fortunately the reaction product provided solid material that could be evaluated by a single crystal X-ray diffraction study. This study allowed data refinement14 to a modest R-factor of 6%, no hydrogen atoms were depicted, and the results were not deposited at the Cambridge Crystallographic Data Centre. These limitations are amended in the present work during which the crystal structure was determined again and could nicely be refined to an R-factor of 2.9%.
Inspection of molecular structures 3 and 4 reveals that their distinction seems to be an ideal case to explore the potential of vibrational circular dichroism (VCD) for a diastereoisomeric assignment since the molecule under study has a rigid scaffold with very limited conformational freedom, only one of five stereogenic centers requires assignment and this unknown center contains a heavy bromine atom.
VCD in combination with quantum mechanics calculations has proven to be very useful in the task to ascertain the absolute configuration of natural products16, 17 and to our knowledge no application for a typical diastereoisomeric assignment, in which epimers at a specific chiral center are involved, is described, although we reported a VCD study18 to distinguish diastereoisomers owing to the presence of a common chiral ester residue in molecules which in the absence of such ester residue would be enantiomers. Other chiroptical methods, like direct comparison of optical rotatory dispersion19 (ORD) or classical circular dichroism19 (CD) curves have in the present case disadvantages as compared to VCD. Although these methods are adequately recognized for cyclohexanones, due to the well established octant rule, in the present case its direct application is not easy due to the lack of comparison models since the six-member ring adopts a boat conformation.
In continuation of our studies of natural products20 and their derivatives21 using VCD, we herein present a detailed theoretical study of diastereoisomers 3 and 4 that allows the absolute configuration assignment of one newly formed chiral center using a combination of VCD spectroscopy, molecular mechanics and density functional theory (DFT) calculations at the B3LYP/DGDZVP level of theory. Our results will be of value for future configurational assignments of hemisynthetic derivatives obtained from syntons of known absolute configuration when the performed reaction yields either one or both possible diastereoisomers at a newly generated chiral center.
General Experimental Procedures
IR and VCD measurements were performed on a BioTools-BOMEM ChiralIR FT-VCD spectrophotometer equipped with a single photoelastic modulator. A sample of 8.5 mg of 3 was dissolved in 150 μL of 100% atom-D CDCl3, placed in a BaF2 cell with a pathlength of 100 μm and data were acquired at a resolution of 4 cm−1 during 10 h. The sample was available from the original study,14 which describes its preparation by treatment of 2 with m-chloroperbenzoic acid. Compound 3 was purified by fast chromatography through a gravity silica gel column using chloroform to afford prisms mp 168–169°C (Ref.14 mp 169°C). Its identity and purity was verified by 1H NMR measurement at 300 MHz on a Varian Mercury spectrometer using a 99.8% atom-D CDCl3 solution containing TMS as the internal standard immediately before VCD measurement.
Conformational searches were started using a Monte Carlo guided protocol considering an initial energy cutoff of 10 kcal/mol above the global minimum value. The searches were conducted independently for 3 and 4, and the conformations 3a-c and 4a-c found for each diastereoisomer were then submitted to geometry reoptimizations using the DFT B3LYP hybrid functional and the DGDZVP basis set. It followed that conformers 3b and 4c vanished in favor of conformers 3a and 4b, respectively, during this geometry reoptimization, and therefore only two conformers for each diastereoisomer account for the total conformational distribution. These four molecular arrangements were subjected to vibrational calculations using the same hybrid functional and basis set. The use of this B3LYP/DGDZVP combination of basis set and functional has shown to require less computing time than the 6-31G(d) basis set while producing very similar results, as is evident in figures of recently published work.20–22 This situation, seems to be associated with the fact that DGauss basis sets, such as DGDZVP, are optimized for DFT methods. Computer time could turn crucial when larger molecules are studied, as was the recent case of stypotriol triacetate,23 a C33H46O7 molecule with 300 electrons which required almost 1100 hours of computer time using the B3LYP/DGDZVP level of theory.
Conformational searches were made using the Spartan'04 software package,24 while geometry reoptimizations and vibrational spectra were calculated using the Gaussian 03W software package.25 Typical calculations required between 30 and 35 hours of computational time per conformer when using a desktop personal computer (PC) with 2 Gb RAM operated at 3 GHz. Tabulated theoretical vibrational frequencies, rotational strengths and dipole strengths were obtained from the calculations using GaussView software. Since the measured IR and VCD frequencies derive from an anharmomic force field and the calculated frequencies derive from a harmonic force field, the intense bands of the calculated IR spectrum and those of the experimental spectrum were compared to obtain an anharmonicity factor of 0.97 which then allowed scaling the calculated VCD frequencies. Experimental vibration frequencies, rotational strengths and dipole strengths were converted to molecular absorptivities (M−1 cm−1). They were obtained from experimental IR and VCD spectra by Lorentzian fitting with half-widths of 6 cm−1 using the PeakFit software.26 To compare calculated and experimental data, the bands of the theoretical IR spectrum were numbered and assigned to the bands in the measured IR spectrum using described methodology.17 Construction of a plot of calculated versus measured frequencies gave an R2 correlation coefficient of 0.9266 in good agreement with published cases.27 We have also used these plots to compare experimental and calculated 1H NMR coupling constants during a VCD study of verticillanes.28 The calculated frequencies are 1–5% higher than experimental values, pointing to the use of an anharmonicity factor of 0.97, after which the bands differ between −2% and +2%. The same assignment was used for the VCD spectra in which the bands were positive and negative. After rotational strengths were obtained, they were successfully compared to the experimental VCD spectrum, fully substantiating the stereochemistry of 3.
Single Crystal X-Ray Analysis of 3
Slow evaporation of an acetonitrile solution gave crystals suitable for X-ray analysis. A prism measuring 0.38 × 0.06 × 0.04 mm was mounted on a Bruker-Nonius CAD4 diffractometer. The crystal was orthorhombic, space group P212121, with cell dimensions a = 6.557(2) Å, b = 12.307(1) Å, c = 19.318(2) Å, V = 1589.9(5) Å3, ρcalcd = 1.739 g/cm3 for Z = 4, C15H20O3Br2, Mw = 408.13, and F(000) = 816 e. Unit cell refinements were done using the CAD4 Express v 2.0 software package provided by the diffractometer manufacturer. A total of 1256 reflections were collected using Cu Kα radiation (λ = 1.54184 Å) within a θ range of 4.48–59.93° for 0 ≤ h ≤ 7, 0 ≤ k ≤ 13, 0 ≤ l ≤ 21. The data were corrected for background, Lorentz polarization, and absorption (μ = 6.639 mm−1), while crystal decay was negligible. The structure was solved by direct methods using the Sir2004 program.29 For the structural refinement, the nonhydrogen atoms were treated anisotropically, and the hydrogen atoms were refined isotropically. The unique reflections were 1188, the observed reflections were 1126, and final discrepancy indices, refining 198 parameters, were RF = 2.9% and RW = 7.5%. The final difference Fourier map was essentially featureless, the highest residual peak and hole having densities of 0.246 and −0.348 e/A3.
RESULTS AND DISCUSSION
The molecular scaffold of the tricyclic oxidation product (3 or 4) obtained from 2 imposes several restrictions to conformational freedom due to the spiro nature of the two six-membered rings and the cis-fused tetrahydrofuran and cyclohexanone. Construction of Büchi solid Dreiding models reveals that only two conformational arrangements of the cyclohexanone are feasible. One of them is a chair-like arrangement in which the equatorial H6β atom and the carbonyl group are coplanar, and the other is a boat-like conformation in which the carbonyl group and the equatorial H6α atom become coplanar. In both cases the methyl group and bromine atom substituents at C8 hold their special arrangement since only C6 and C7 are twisted during a chair to boat interconversion. A third molecular arrangement in which C8 and C9 of the chair conformation are twisted to afford the cyclohexanone in boat shape is completely ruled out due to severe steric interaction between one of the methyl groups of the gem-dimethyl group and either the bromine atom or the methyl group substituents located at C8. From these considerations one could assume, due to the very well established conformational preference of cyclohexanones to undertake chair conformations rather than boat conformations,30 that diastereoisomer 4 would be the reaction product of 2 with m-chloroperbenzoic acid.
The conformational distributions of 3 and 4 where asserted using Monte Carlo-MMFF94 conformational searches followed by geometry optimization at the B3LYP/DGDZVP level of theory (Fig. 1 and Table 1). In both molecules the MMFF94 calculations showed three low energy conformations dependent on two conformational features that were the hydroxyl group orientation and the chair-boat disposition of the α-brominated cyclohexanone. In the case of 3 the molecular mechanics calculations predicted two conformations with the cyclohexanone in a boat disposition and different hydroxyl group orientations, one intersecting the C5a-C5-Me bond angle with a C5a-C5-OH dihedral angle of 88.9°, and oriented towards the carbonyl oxygen atom (global minimum 3a), and the other intersecting the C4-C5-Me bond angle with a C5a-C5-OH dihedral angle of 177.0° (3b), being 0.11 kcal/mol more energetic than the first conformer. These two molecular arrangements where followed by a third conformer (3c) with an energy 2.75 kcal/mol above the global minimum, and arranged in a chair cyclohexanone disposition with the hydroxyl group orientation identical to that found in 3b. For isomer 4, also three equivalent conformations were found using the same search methodology, but with differences in their relative stability order. The global minimum 4a showed the same chair cyclohexanone conformational characteristics as conformer 3c with an energy difference of 0.65 kcal/mol above the global minimum 3a, while 4b being 1.07 kcal/mol above 4a showed to be conformationally equivalent to 3a, and 4c showed to be equivalent to 3b, as observed in Figure 1. Further geometry optimizations followed by vibrational calculations gave Gibbs free energy values that permitted a more accurate assessment of the conformational distributions. In both isomers conformations with the boat cyclohexanone disposition 3b and 4c changed their hydroxyl group orientations during the geometry optimization procedure towards the carbonyl oxygen atom, becoming 3a and 4b, respectively, and thus showing that conformations 3b and 4c are no longer local minima on the DFT potential energy surface. In addition, the relative order of stability of both distributions obtained from the DFT calculations remained unchanged as compared to the MMX calculations, predicting 3a to be 3.62 kcal/mol more stable that 3c, and 4a to be 0.99 kcal/mol more stable that 4b. This conformational preference of isomer 3 for the boat disposition can be explained based on an energetically favorable “anti” disposition of the equatorial bromine atom towards the carbonyl group (OCC8Br dihedral angle of −82.9°), which also leads to energy stabilization produced by the hydrogen bonding interaction between the hydroxyl group and the carbonyl oxygen atom of the cyclohexanone ring showing an OHOC interatomic distance of 2.5 Å. In contrast the hydrogen bond interaction in the boat configuration of 4b, in which the bromine atom is axial, leads to a more energetic “syn” orientation of the bromine atom towards the carbonyl oxygen with an OCC8Br dihedral angle of 49.9°, explaining the preference of 4 to the chair conformation that avoids this unstable disposition. Table 1 illustrates the conformational behavior on going from MMFF94 to DFT calculations.
Further comments to the molecular behavior are related to the presence of the hydroxyl and carbonyl functionalities, since some alcohols can associate in solution, as we have shown27 for a secondary nonintramolecularly bonded alcohol which provides good agreement of observed and calculated spectra only after acetylation. In contrast, for tertiary alcohols, due to steric hindrance, good spectral fits are easily obtained28 without derivatization. In the current case we are dealing with a tertiary alcohol which in addition is intramolecularly hydrogen-bonded to a carbonyl group and therefore no significant intermolecular association of the hydroxyl group might be expected.
The vibrational calculations performed for the obtained conformations, along with their conformational distributions, permitted to generate Boltzmann weighted plots of calculated IR and VCD spectra of diastereoisomers 3 and 4, as shown in Figures 2 and 3, respectively. Since the main objective of the present study is to evaluate the possibility and limitations to differentiate epimers like 3 and 4 in the infrared region of the electromagnetic spectrum, the calculated results need to be compared to find differences that can later be helpful in the configurational assignment of the reaction product. In the case of the IR spectra the changes in intensity and frequency are rather small and no band in particular stands out as useful for the differentiation purpose. The IR spectral presentation shown in Figure 2 is in full agreement with published cases18, 31 and allows a visual comparison in a similar way as organic chemists are used to seeing spectra, although in the present case we are dealing with absorption plots. Accordingly, a comparison between the theoretical IR spectra of both possible diastereoisomers and the observed IR spectrum of the reaction product showed a good agreement in both cases (Fig. 2), precluding any stereochemical assignation by these means. Nevertheless the opposite scenario is observed for the weighted VCD spectra of 3 and 4, in which several differences between them are evident at first glance (Fig. 3). Among them are the negative carbonyl band at 1728 cm−1, the most intense positive band located at 1360 cm−1, the positive band at 1168 cm−1 and the negative band at 1112 cm−1 in the VCD spectra of 3, which are not observed or differ considerably in sign and intensity in the calculated VCD spectrum of 4. As expected, these bands allow unambiguous assignment of the absolute stereochemistry of the C8 chiral center of the synthetic sample as the one depicted for 3, since the corresponding experimental spectrum shows the presence of the negative band at 1724 cm−1, the positive band at 1370 cm−1, the positive band at 1180 cm−1 and the negative band at 1129 cm−1, along with an overall spectral similarity when compared with the corresponding calculated VCD spectrum. In the present case it is also of relevance to note that due to the previously discussed intramolecular hydrogen-bonded hydroxyl group, the well studied carbonyl absorption artefacts32 seem to be significantly decreased making the sign and VCD band intensity more reliable as can be seen in the spectral comparison shown in Figure 3.
Complementary evidence to establish stereostructure 3 for the oxidation product of 2 with m-chloroperbenzoic acid follows from a comparison33 of experimental rotational strengths of the reaction product with those of both possible diastereoisomers (3 and 4). The plots included in Figure 4 show an overall agreement only in the case of 3.
To compare the calculated molecular shape with experimental data, we turned our attention to the originally published14 solid state structure of 3. Since this structure was refined only to a modest R-factor of 6.0%, no hydrogen atoms were depicted, and the corresponding data are not deposited at the Cambridge Crystallographic Data Centre, we decided to repeat these measurements. The pertinent crystal data, data collection, structure determination and refinement procedures are detailed in the “Experimental” section and a representation of the final refined molecular model is given in Figure 5. The solid state structure is in excellent agreement with the conformation calculated by DFT as evidenced by the data summarized in Table 2. The solid state structure of pacifenol 1 also shows1, 7 the cyclohexane in a boat conformation.
The conformational distributions of both possible diastereoisomers (3 or 4) generated by reaction of pacifenol derivative 2 with m-chloroperbenzoic acid where asserted using Monte Carlo-MMFF94 conformational searches and geometry optimization at the B3LYP/DGDZVP level of theory. Comparing 3 and 4 revealed the presence of three similar conformations on the molecular mechanics potential energy surface, but with differences in their stability order. Subsequent geometry optimization calculations at the B3LYP/DGDZVP level of theory showed the presence of only two conformers in each case, and that these conformations are mainly determined by a hydrogen bond interaction between the hydroxyl group and the carbonyl oxygen atom of the cyclohexanone and by the syn or anti orientations of the bromine atom towards the carbonyl oxygen atom. These two conformational features caused the preference of the cyclohexanone in 3 and 4 for boat and chair conformations, respectively.
Furthermore, these conformational distributions, together with the calculation of vibrational frequencies and intensities, allowed obtaining weighed theoretical plots for IR and VCD spectra, which in turn permitted the configurational assignment for the reaction product of the oxidation reaction of 2 as the one depicted for 3. In the case of the calculated IR intensities and frequencies, these showed only small differences between isomers which avoided the use of these spectra in the task of diastereoisomeric differentiation. In contrast, the theoretical VCD weighed plots showed several differences between diastereoisomers, such as the negative carbonyl band at 1728 cm−1, the positive band at 1360 cm−1, the positive band at 1168 cm−1 and the negative band at 1112 cm−1 present in the theoretical VCD spectra of 3 that are absent in the spectrum of 4. These results allowed obtaining unambiguous stereochemical conclusions by comparison with the experimental VCD spectrum of the reaction product that showed these bands and overall VCD spectral similarity. In addition, the conformation of 3 in the solid state, as deduced from single crystal X-ray measurements, is also in good agreement with that calculated by DFT as evidenced by dihedral angles comparison of the cyclohexanone.
- 15Semisíntesis y actividad biológica de derivados del sesquiterpeno pacifenol. Bol Soc Chil Quim 1996; 41: 403–408., , , .
- 17Vibrational circular dichroism a new tool for the solution-state determination of the structure and absolute configuration of chiral natural product molecules. Nat Prod Comm 2008; 3: 451–466..
- 19Optical rotatory dispersion and circular dichroism in organic chemistry, Vol. 13. San Francisco: Holdey-Day; 1965. p 72–157..
- 23VCD determination of the absolute configuration of stypotriol. Joseph-Nathan, Nat Prod Comm 2009; 4: 1037–1040., , , , .
- 24Spartan '04, Irvine, CA: Wavefunction; 2004.
- 25Gaussian 03W, Pittsburgh, PE: Gaussian Inc.; 2003.
- 26PeakFit v4.12 for Windows, San Jose, CA: SeaSolve Software Inc.; 2003.
- 30Stereochemistry of carbon compounds. New York: McGraw Hill; 1962. p 204–269..