Cephalochromin (1) (Fig. 1), first characterized by Tertzakian et al.,1 is structurally related to ustilaginoidin A (2), isolated by Shibata et al.2 from Ustilaginoidea virens. Ustilaginoidin A (2) is the 9,9′ dimer of the commonly occurring metabolite 5,6,8-trihydroxy-2-methylbenzo[γ]chromen-4-one (3) (norrubrofusarin), isolated already in 1952 by Schmid and Seiler.3 The metabolite 1 was later found4 together with dihydroisoustilaginodin A and isoustilaginoidin A from the Verticillium Sp. K-113. Cephalochromin (1) was also identified5 as the first selective FabI (bacterial enoyl-acyl carrier protein reductase) inhibitor of Staphylococcus aureus and Escherichia coli. As can be seen from Figure 1, cephalochromin (1) may be viewed as the 2,3,2′,3′-tetrahydro derivative of the (aS)-enantiomer of ustilaginoidin A (2), although natural ustilaginoidin has (aR) axial chirality. Thus, cephalochromin (1) possesses central chirality elements as two tertiary stereogenic centers in addition to the axial chirality of the atropisomeric naphthopyranone systems. Although the axial chirality of some ustilaginoidins could be elucidated by Shibata and Ogihara6 and Koyama et al.7, the configuration of the stereogenic centers in cephalochromin (1) remained unknown. Chaetochromin A (4),7 which besides the (aS) axial chirality element has four stereogenic centers at C-2, C-3, C-2′, and C-3′, is even more closely related to cephalochromin (1). The absolute configuration (AC) of chaetochromin A was determined by X-ray crystallography of its 6-O-p-Br-benzoate.7 A positive exciton couplet in its electronic circular dichroism (ECD) sepctrum, derived from the coupling of the two naphthopyranone systems, was also correlated with its (aS) axial chirality. Recently, some of the authors isolated8 a large amount of cephalochromin (1) from the culture broth of the endophytic fungus Pseudoanguillospora sp. The spectral data were in perfect agreement with those published in the literature.1, 5 The AC of the atropisomeric naphthopyranone system of 1 could be established by the exciton-coupled circular dichroism method9, 10 supported by Time Dependent Density Functional Theory (TDDFT) calculations of the ECD spectra. However, the ECD calculated for 2-methylnaphtha-γ-pyrone moiety with the TDDFT method11 at different levels of calculation (B3LYP/TZVP//B3LYP/6-31G(d)) was found two orders of magnitude smaller than the exciton-coupled spectra and the AC of the C-2 and C-2′ stereogenic centers could not be determined from the ECD data. A similar predominance of the axial chirality element was previously observed in the CD spectrum of phomoxanthone A,12 and it seems to be of general importance in chiral biaryl systems.13 In this study, vibrational circular dichroism (VCD) is used to solve this fundamental stereochemical problem, the concurrent presence of two different chirality elements, waiting for a solution for more than 20 years.7
All calculations were undertaken with Gaussian program14 using B3LYP functional and 6-31G* and 6-311G* basis sets. The optimized geometry of isolated (aS,2R,2′R)-cephalochromin is shown in Figure 2. The stereogenic centers C-2 and C-2′ have the same configuration to maintain C2 symmetry of the homodimeric cephalochromin. There are four possible diastereomers for this molecule, namely (aS,2R,2′R), (aS,2S,2′S), (aR,2R,2′R), and (aR,2S,2′S). Earlier ECD studies7 reported the determination of axial chirality (+)-cephalochromin as (aS) on the basis of the positive exciton couplet centred around 280 nm. More recently, this assignment of axial chirality was confirmed independently by ECD calculation.8 Starting from the structure with (aS)-axial chirality, the geometries of two diastereomers (aS,2R,2′R) and (aS,2S,2′S) have been optimized, with methyl groups at 2 and 2′ positions in equatorial positions. The hydroxyl groups at C-8 and C-8′ can take three possible orientations15: (a) cis orientation for both HOC-8C-9 and HOC-8′C-9′ (labeled as cis–cis conformer); (b) trans orientation for both HOC-8C-9 and HOC-8′C-9′ (labeled as trans–trans conformer); and (c) cis orientation for HOC-8C-9 and trans orientation for HOC-8′C-9′ or vice versa (labeled as cis–trans conformer). Of these three conformers of isolated 1,1′-bi-2-naphthol molecule15 cis–trans and trans–trans conformers have ∼4.2 and 7.6 kcal/mol higher energy, respectively, over cis–cis conformer. Similar relative energy trend was found16 for the three conformers of isolated 6,6′-dibromo-1,1′-bi-2-naphthol. Furthermore, cis–cis conformer was found16 to be favored in non-hydrogen bonding CH2Cl2 solvent for 6,6′-dibromo-1,1′-bi-2-naphthol. The same relative energy trend is found for cephalochromin as well. For example, at B3LYP/6-311G* level, trans–trans conformer has 7.0 kcal/mol higher energy than cis–cis conformer for both (aS,2R,2′R) and (aS,2S,2′S) diastereomers. Therefore, the present calculations used cis–cis conformer. The hydroxyl groups at C-5, C-6, C-5′, and C-6′ are constrained to favor intramolecular hydrogen bonding such that the hydroxyl group at C-5/C-5′ is hydrogen bonded to the carbonyl oxygen at C-4/C-4′. In addition, the hydroxyl group at C-6/C-6′ is hydrogen bonded to the oxygen atom at C-5/C-5′. These intramolecular hydrogen-bonded structures are expected to be favored in non-hydrogen bonding solvents (such as CH2Cl2 used in the current experimental studies) and have lower energies over those that have either both free OH groups or one free OH group at C-6/C-6′.
The theoretical infrared (IR) and VCD spectra were simulated using predicted frequencies as such without any scaling, with Lorentzian band shapes and 5 cm−1 half-width at half-peak height. The theoretical ECD spectra were simulated from the first 25 singlet → singlet electronic transitions using Lorentzian band shapes and 10-nm half-width at half-peak height. The predicted electronic transition wavelengths are used as such without any scaling. The UV–Vis spectra are presented as molar extinction coefficient (in L mol−1 cm−1), derived from dimensionless oscillator strength. The peak extinction coefficient of i-th band, ϵ, is related to oscillator strength, fi, as where Δi is the half-width at half-height of Lorentzian band. The predicted specific rotations at 633, 589, and 546 nm were obtained with Gaussian program.14 The UV–Vis, ECD, IR, and VCD spectra obtained at B3LYP/6-31G* and B3LYP/6-311G* levels are found to be similar. Those obtained at B3LYP/6-311G* level are shown in Figures 3–7.
The studied cephalochromin sample was isolated from an endophytic fungus, Pseudoanguillospora sp., isolated from the red algae Polyides rotundus and found on the shores of the Baltic Sea.
Optical rotation (OR) measurements were obtained at a concentration of 1.0 mg/mL (1.9 × 10−3 M) in CH2Cl2 solvent using a 0.5 dm cell and a Autopol IV polarimeter. Because of the strong absorption of visible light by cephalochromin, the ORs could only be measured at three wavelengths. The wavelengths and observed specific rotations (in deg cc/(g dm) units) are as follows: 633 nm, 303; 589 nm, 381; 546 nm, 507. The optical rotatory dispersion (ORD) spectra in this work are limited to these three wavelengths. The sample used for current experimental measurements is labeled (vide infra) as (+)589-[(-VCD)] where (+)589 refers to positive OR at 589 nm and [(-VCD)] refers to observed negative VCD at 1038 cm−1 in CH2Cl2 solvent. The UV–Vis and ECD spectra (Figs. 3 and 4) were recorded on a Jasco J720 spectrometer, using 0.01-cm path length quartz cell, at a concentration of 1.9 × 10−3 M in dichloromethane.
The IR and VCD spectra (Figs. 5 and 6) were recorded on a commercial Fourier Transform VCD spectrometer, Chiral IR, in the 2000–900 cm−1 region. The VCD spectra were recorded with 1-hr data collection time at 8 cm−1 resolution. Spectra were measured in CH2Cl2 solvent at 19 mg/mL (3.7 × 10−2 M) concentration. The sample was held in a variable path length cell with BaF2 windows. In the IR spectrum presented, the solvent absorption was subtracted. Similarly, the VCD spectrum of solvent measured under identical conditions was subtracted from that of the sample. The region ∼1295–1250 cm−1 has been removed from the experimental traces because of interference from strong solvent absorption. Because the region blocked from solvent absorption is very small, no attempts were made to measure the spectra in CD2Cl2 solvent.
RESULTS AND DISCUSSION
The optimized structure of (aS,2R,2′R), with cis–cis conformation for hydroxyl groups at C-8 and C-8′, obtained at B3LYP/6-311G* level is shown in Figure 2. The C-9C-9′ bond length in (aS,2R,2′R) diastereomer (1.50 Å) is slightly longer than that in (aS,2S,2′S) diastereomer (1.49 Å). The C-2C-11 bond length however is approximately the same (1.51 Å) in two diastereomers. There is a slightly larger difference in the dihedral angle around the C-9C-9′ bond among the two diastereomers (92° and 95° for (aS,2R,2′R) and (aS,2S,2′S), respectively). These observations indicate that there is some influence of stereogenic centers at C-2 and C-2′ on the orientation around C-9C-9′ bond and vice versa. Then, the optical activity arising from axial chirality may be expected to be influenced by the stereogenic centers at C-2 and C-2′, although the significance or magnitude of that influence cannot be quantified (vide infra) from structural parameters. The energy difference between the two diastereomers however is very small (∼0.1 kcal/mol).
Chiroptical Spectroscopic Studies
To assess the interaction between axial and central chirality elements (axial chirality around 9-9′ bond and stereogenic centers at 2 and 2′ positions), one can analyze the difference between the spectra predicted for diastereomers, (aS,2R,2′R) and (aS,2S,2′S). If there is no significant interaction between axial and central chirality elements then one-half of the difference, [(aS,2R,2′R) − (aS,2S,2′S)]/2, should represent the contribution of (2R,2′R) configuration in the ECD/ORD/VCD spectra. On the other hand, one-half of the sum spectrum, [(aS,2R,2′R) + (aS,2S,2′S)]/2, should represent the contribution of (aS) configuration in ECD/ORD/VCD spectra. Thus, this difference in spectral analysis provides information on contributions from individual chirality elements. In the corresponding absorption spectra, the difference [(aS,2R,2′R) − (aS,2S,2′S)] should be zero if there is no significant interaction between axial and central chirality elements.
UV-Vis and ECD Spectra
In Figure 3, the experimental UV–Vis spectrum is compared with the predicted UV–Vis spectra for diastereomers (aS,2R,2′R) and (aS,2S,2′S), along with the above-mentioned difference spectrum, [(aS,2R,2′R) − (aS,2S,2′S)]. From these spectra, it is apparent that the UV–Vis spectra of individual diastereomers are slightly different. There are shifts in band positions among the two diastereomers and also some bands are apparent in the absorption spectrum of one diastereomer and not the other. Nevertheless, the magnitudes in difference spectrum are not large.
In Figure 4, the experimental ECD spectrum is compared with the predicted ECD spectra for diastereomers (aS,2R,2′R) and (aS,2S,2′S), along with the difference spectrum, [(aS,2R,2′R) − (aS,2S,2′S)]. From these spectra, it is apparent that the ECD spectra of individual diastereomers are essentially the same. The magnitudes in the difference spectrum, 0.5 × [(aS,2R,2′R) − (aS,2S,2′S)], are so small that the ECD spectral features seen for individual diastereomers can be regarded as arising predominantly from axial chirality. The contributions from stereogenic centers C-2 and C-2′ to the observed ECD spectrum can be regarded as negligible. Then, the positive ECD couplet seen in the experimental ECD spectrum (positive at 292 nm and negative at 266 nm) can be attributed solely to axial chirality (aS), as has already been noted before.8
IR and VCD Spectra
The experimental IR spectrum is compared with the predicted IR spectra for diastereomers (aS,2R,2′R) and (aS,2S,2′S), along with the difference spectrum, [(aS,2R,2′R) − (aS,2S,2′S)], in Figure 5. From these spectra, it is apparent that the IR spectra of individual diastereomers are essentially the same. The magnitudes in the difference spectrum, [(aS,2R,2′R) − (aS,2S,2′S)], are so small indicating that the IR spectral features seen for individual diastereomers are little influenced by the stereogenic centers C-2 and C-2′. The experimental absorption bands have a reasonably good agreement with the absorption bands predicted for the diastereomers.
The experimental VCD spectrum is compared with the predicted VCD spectra for diastereomers (aS,2R,2′R) and (aS,2S,2′S), along with the difference spectrum, [(aS,2R,2′R) − (aS,2S,2′S)], in Figure 6. From these spectra, it is apparent that most of the features in the VCD spectra of individual diastereomers are similar but, importantly, mirror image VCD signs are seen for diastereomers at 1054, 949, and 929 cm−1. Among these three VCD bands, the first one at 1054 cm−1, negative for (aS,2R,2′R) and positive for (aS,2S,2′S), is of larger magnitude and the latter two, positive for (aS,2R,2′R) and negative for (aS,2S,2′S), are of smaller magnitudes, with the one at 949 cm−1 being the weakest. Because the corresponding experimental bands at 1038, 930, and 910 cm−1 have negative, positive, and positive VCD signs, respectively, one can assign the ACs at C-2 and C-2′ positions as (R) for the sample investigated here. It should be noted that although the predicted VCD bands under consideration are not among the strong bands, these bands are reproduced in both 6-31G* and 6-311G* calculations and therefore these predictions are considered to be reliable.
The VCD band at 1054 cm−1 in the predicted spectrum for (aS,2R,2′R)-cephalochromin originates from two vibrations, one at 1054.1 cm−1 of A symmetry and another at 1054.2 cm−1 of B symmetry. Both are predicted to have negative rotational strengths, although the later has larger magnitudes for both dipole and rotational strengths. The displacements for B symmetry vibration are displayed in Figure 7. Both A and B symmetry vibrations predicted with a frequency of ∼1054 cm−1 involve methine bending motion at C-2 (the methine hydrogen atom moves into, and away from, the pyranone ring) coupled with rocking motion of methyl group attached to C-2 and methylene twisting motion at the adjacent carbon, C-3. The predicted VCD band at 949 cm−1 arises from two vibrations, one at 953 cm−1 (A symmetry) and another at 949 cm−1 (B symmetry), but the associated normal modes have rotational strengths with opposite signs resulting in mutual cancellation and reduced intensity. As a result, the intensity associated with predicted VCD band at 949 cm−1 is quite weak. The predicted VCD band at 929 cm−1 also arises from two vibrations, one at 930 cm−1 (A symmetry) and another at 927 cm−1 (B symmetry), both with positive rotational strengths. The normal modes associated with 930 and 927 cm−1 vibrations (as well those at 953 and 949 cm−1), also involve methine bending motion at C-2, coupled to those of methyl group (attached to C-2) and methylene group at the adjacent carbon, C-3 (similar to those at 1054 cm−1 shown in Fig. 7). All these vibrations directly influence the stereogenic center C-2 and therefore associated VCD can be expected to reflect the AC at C-2.
The magnitudes in the difference VCD spectrum, [(aS,2R,2′R) − (aS,2S,2′S)], are generally small, so the dominant VCD spectral features seen for individual diastereomers mostly arise from the axial chirality (aS). The large VCD bands seen in the experimental spectrum at 1647(−), 1632(+), 1600(−), and 1585(+) originate from (aS) configuration and are not influenced significantly by the stereogenic centers at 2 and 2′, as the corresponding predicted VCD bands at ∼1687(−), 1678 (+), 1634(−), and 1620(+) are present in the spectra for both diastereomers with nearly equal intensities. The same is true for the experimental VCD bands at 1169 and 1126 cm−1, which correlate with those predicted at 1185 and 1148 cm−1 and are not influenced by the stereogenic centers 2 and 2′. The VCD bands seen in the experimental spectrum in the 1500–1200 cm−1 region however are somewhat congested and seem to have overlapping contributions from (aS) configuration and the stereogenic centers 2 and 2′, because the difference VCD spectrum shows some features in this region, although not of large magnitude.
The experimental specific rotations at three wavelengths, 633, 589, and 546 nm, are compared with the predicted specific rotations for two diastereomers in Figure 8. The predicted signs of specific rotations are positive for both diastereomers at all three wavelengths. Therefore, the predicted signs of ORs cannot distinguish the two diastereomers. Based on predicted magnitudes alone, the predicted specific rotations for (aS, 2S, 2′S) diastereomer are closer to the experimental values, so one may presume that the experimental data may belong to (aS, 2S, 2′S) diastereomer. This conclusion will contradict that noted earlier from VCD data. To predict reliable OR magnitudes,17, 18 however, very large basis sets consisting of polarization functions would be needed and, furthermore, vibrational contributions and solvent effects need to be incorporated into calculations. Therefore, in this work, we have placed emphasis only on the predicted signs, and not on the magnitudes, of OR.
To summarize the three chiroptical spectroscopic data, predicted ECD spectra and OR signs cannot distinguish the diastereomers of cephalochromin. However, VCD spectra, with three bands in the ∼1050–900 cm−1 region, provide a clear way to distinguish the two diastereomers of cephalochromin. Among the corresponding three experimental VCD bands, the one at 1038 cm−1 has relatively larger magnitude and therefore this VCD band is used to designate the diastereomers. Thus, the sample used in the current studies has been designated, following an earlier notation19 that used ECD bands, as (+)589-[(−VCD)]-(aS,2R,2′R), and its diastereomer as (+)589-[(+VCD)] − (aS,2S,2′S). The current results for cephalochromin emphasize the utility of VCD for other molecules where signs of ECD bands and OR cannot distinguish the diastereomers.
By exploiting the experimental VCD and corresponding theoretical VCD, in the ∼1050–900 cm−1 region associated with the vibrations of the stereogenic centers, ACs of cephalochromin have been assigned for the first time. This report demonstrates that when a molecule contains axial chirality along with stereogenic centers, each chirality element can be identified and characterized by means of VCD spectroscopy. Although, similarly to the ECD spectrum, the VCD spectrum of cephalochromin is also dominated by the axial chirality element, some vibrational bands in the ∼1050–900 cm−1 region are predicted to give opposite VCD signs for the two possible diastereomers allowing the assignment of the ACs of central chirality elements. On the other hand, ECD spectroscopy and signs of OR are not so useful in this regard.