Stereostructure of luminamicin, an anaerobic antibiotic, via molecular dynamics, NMR spectroscopy, and the modified Mosher method

Results and Discussion

Distance and Dihedral Constraints Derived from NMR Experiments. Torsional constraints were obtained by applying the Karplus equation to vicinal proton–proton scalar coupling constants obtained by the 1D proton spectrum with high resolution (21). A proton–proton coupling constant >10 Hz was treated as indicating an anti H–H orientation and a dihedral-angle estimate of 180° ± 40°. A coupling constant <3 Hz was considered indicative of a gauche orientation, i.e., –90° ± 40° or 90° ± 40°. The resulting torsional constraints are given in Table 1.

To obtain distance constraints, 2D-TROESY spectra were measured with mixing times of 100, 150, 200, 250, and 300 ms. A plot of the volume of the cross-peak versus mixing time showed linearity up to 200 ms. Therefore, the proton–proton distance constraints were based on the integrated cross-peaks from the 200-ms spectrum. Fig. 7, which is published as supporting information on the PNAS web site, shows a part of the 2D-TROESY spectrum measured with a mixing time of 200 ms. Volumes of the five nonoverlapping geminal proton cross-peaks were averaged and used for calibrating measured ROE volumes. A distance of 1.8 Å was used as a distance reference for geminal proton cross-peaks. This calibration yielded the theoretically expected value, 2.4 Å, for the distance between H5 and H6. Distance constraints were classified into three categories corresponding to 1.8–2.7, 1.8–3.5, and 1.8–5.0 Å, corresponding to strong, medium, and weak ROEs, respectively. For the distance constraints related to methylene protons at C27 (H27_a,b), of which chemical shifts are overlapping, or methyl protons of 2-methoxy, 10-methyl, or 14-methyl groups, carbon atoms attached to these protons were used to estimate target distances. In such a case, 1.0 Å was added to the upper boundary of the constraints (22). For example, the strong ROE observed between H4 and the 14-methyl group is converted to a distance constraint between H4 and carbon atom of 14-methyl group (14MeC), the range of which is 1.8–3.7 Å. A total of 43 distance constraints were obtained as shown in Table 2.

Determination of Absolute Stereochemistry of Luminamicin. Table 3 lists the number of energetically possible conformers obtained from camdas calculation for 32 different configurations. About 2,500∼3500 distinct conformers were obtained for each configuration. Then, we calculated RMS deviations for all conformers generated by camdas to determine which configurations can adopt conformers accommodating the distance and dihedral constraints. All distance constraints were treated with equal weight, although they were classified as strong, medium, and weak. For constraints including nonoverlapping geminal protons (3-, 8-, 17-, 19-, or 20-methylene protons), careful analysis was done when estimating target dihedral angles and proton–proton distances. For example, H3_a (2.49 ppm) is included in a total of four constraints, i.e., two dihedral constraints and two distance constraints as shown in Tables 1 and 2. The proton corresponding to H3_a should simultaneously satisfy all of these four constraints. Because we cannot establish the stereospecific assignments on 3-methylene protons (H3_a and H3_b) in advance, we first need to consider all combinations when estimating target dihedral angles or target proton–proton distances. For example, we first estimate a total of four dihedral angles, i.e., two dihedral angles are combination between H2 and two 3-methylene protons, and other two are combination between H4 and two 3-methylene protons. Then, we determine which of two 3-methelyne protons can simultaneously satisfy two dihedral constraints included in Table 1. The identified proton can be considered to correspond to H3_a (2.49 ppm). Next, for the identified 3-methelyne proton, we estimate two target distances related to H3_a (2.49 ppm) in Table 2. The similar procedure was also performed for the constraints on 8-, 17-, 19-, or 20-methylene protons. Table 3 includes the smallest RMS distance deviations obtained for each configuration. The conformers satisfying all experimental constraints were found in only two configurations. For luminamicin with the R, S, S, S, S, R, and R configurations at C4, C7, C9, C10, C11, C12, and C13, respectively, C2, C16, C28, and C29 should have the R, R, S, and R configurations, respectively. On the other hand, for luminamicin with the S, R, R, R, R, S, and S configurations at C4, C7, C9, C10, C11, C12, and C13, respectively, the results embodied the S, S, R, and S configurations at C2, C16, C28, and C29, respectively. Obviously, the former configuration is an enantiomer of the latter one. This means that NMR constraints could not distinguish between enantiomers. This was an expected result, because distance and dihedral constraints can provide only relative structural orientation. The final determination of absolute stereostructure of luminamicin was done by using the result of the following modified Mosher method.

1 was treated with (R)-(+)- and (S)-(–)-2-methoxy-2-trifluoromethyl-2-phenylacetic acid (MTPA) in the presence of EDCI and DMAP to afford the (R)-(+)- and (S)-(–)-MTPA esters (2 and 3) (see Scheme 1, which is published as supporting information on the PNAS web site). Fig. 3 shows the Δδ values (δ _S –δ _R ) obtained from the ¹H-NMR data of 2 and 3, which are included in Table 4, which is published as supporting information on the PNAS web site. The Δδ values for H₃-14-Me, H-16, and H-29 were negative, whereas positive Δδ values were obtained for H-26 and H₂-27, thus indicating an R configuration. The absolute configurations therefore at C2, C4, C7, C9, C10, C11, C12, C13, C16, C28, and C29 of 1 were assigned as S, S, R, R, R, R, S, S, S, R, and S, respectively.

This configuration is the same as that found in nodusmicin (7) and nargenicin (8), which have a chemical structure quite similar with luminamicin (Fig. 4).

Description of the 3D Structure of Luminamicin. The process described above led to a set of two energy-minimized structures for luminamicin, which have Merck Molecular Force Field energies of 135.6 and 142.1 kcal/mol, respectively. Fig. 5 shows stereopairs of the best-fit superposition of the heavy atoms for these two structures. The structures only differ with respect to the orientation of the 2-methoxy group. As shown is Fig. 1, luminamicin consists of three different rings, i.e., a decalin, a 10-membered macrolactone, and a 14-membered macrolactone. The earlier NMR study described only the stereochemistry of the decalin ring (Fig. 2). Both C3 and C14 of the 10-membered macrolactone ring occupy equatorial positions relative to the decalin ring, as indicated in Fig. 2. The two strong ROEs between H4 and 14-methyl group and H12 and 14-methyl group conclusively established the spatial orientation of the 14-methyl group. It must be on the same side with H4 and H12 of the decalin ring. The configuration of C2 was well defined by the ROEs between H2 and H4, 2-methoxy group and H27_a,b, and 2-methoxy group and H25. These ROEs result in the 2-methoxy group oriented to the 14-membered macrolactone. The 10- and 14-membered macrolactones were clearly found to be cis-fused in this study, because critical strong ROEs were observed between H16 and H29, H16 and 14-methyl group, and H29 and 14-methyl group, indicating that these protons are on one side of the molecule. Both of H17_a and H17_b have gauche orientation relative to H16, because both correlations of 16H to H17_a and H17_b are strong in the TROESY spectrum. The small values of J _16,17a (<1.0 Hz) and J _16,17b (<1.0 Hz) are also consistent with these gauche orientations. The orientation of olefinic group (C25 and C26) was established by the strong correlations of H25 to H20_b and H27 in the TROESY spectrum. H25 is oriented inside the molecule and H26 is turned outside. The stereochemistry of C28 was determined by the strong ROE between H25 and H28, resulting in the 28-hydroxyl group oriented outside the molecule. The resulting 3D structure of luminamicin shows an interesting feature that the maleic anhydride functionality in conjugation with olefinic group of the 14-membered macrolactone is nearly perpendicular to the plane of the decalin and 14-membered macrolactone rings, as shown in Fig. 5.

Luminamicin has two hydroxyl groups at C11 and C28, which are expected to react with (R)-MTPA or (S)-MTPA. However, it appears that these reagents can react with only the 28-hydroxyl group. With respect to this phenomenon, we have obtained some relevant information from the 3D structure of luminamicin. As shown in Fig. 6, two hydroxyl groups at C11 and C28 are placed on one side of luminamicin. In addition, the 28-hydroxyl group is more exposed to the solvent, because its solvent-accessible surface area (36.9 Å) is larger than that of the hydroxyl group at C11 (29.7 Å). Therefore, we can consider that the reagents appear to more easily react at C28, and the (R)-MTPA or (S)-MTPA ester group attached at C28 most likely hinder the reaction occurring at C11.

In conclusion, the absolute stereostructure of luminamicin, an anaerobic antibiotic, has been determined by using the conformational analysis via high-temperature MD, NMR spectroscopy, and the modified Mosher method. Luminamicin (1) could be a new lead for medicines to compare with vancomycin, which is used clinically in pseudomembranous colitis therapy.