Abstract
This manuscript describes predicted NMR shifts for the limonoid natural product xylogranatin F. The 1H and 13C NMR shifts of four diastereomers were evaluated by GIAO and hybrid DFT/parametric DU8+ methods. The results of the 1H and 13C NMR calculations for both the GIAO method and the DU8+ calculations suggest the revised structure that was recently reassigned by chemical synthesis. Furthermore, we show that while DU8+ provides superior accuracy with less computation time, GIAO points to the correct structure with more distinguishable data in this case study.
Keywords: DU8+, GIAO NMR calculation, limonoid, natural product, structural reassignment
1. | INTRODUCTION
NMR calculations have become an increasingly useful and widely employed tool in both structural determination and revision of complicated molecules, especially in the context of natural products isolation and multistep synthesis (for reviews, see previous studies1–4; for several notable examples using NMR calculations for structural revision, see previous works5–11). Utilization of modern quantum chemical computations has the advantage of magnifying and quantifying subtle differences among similar structures (eg, diastereomers). Using a computational approach for structure determination instead of a purely spectroscopic one has the advantage of only requiring minimal quantities of material to obtain basic spectroscopic data. This exempts the need for a non-trivial amount of sample that is usually required for other methods, most notably 2D NMR techniques. Furthermore, a computationally powered approach has the advantage over authentic synthesis of being less time-intensive.
During the course of our total synthesis of granatumine A and related bislactone limonoid alkaloids (1-2), we determined by chemical synthesis that xylogranatin F (3) was structurally mischaracterized (Figure 1).12 Herein, we report the full analysis of the NMR shift calculations of the four most likely diastereomers by both GIAO and DU8+, which provides an approach for conducting NMR calculations within this class to furnish results consistent with authentic synthesis.
FIGURE 1.
Xylogranatin F and related limonoid natural products
In the original isolation paper, the authors proposed that xylogranatin F had the structure 3 wherein the relative configuration at positions C3, C5, and C10 was defined through two key NOESY contacts (Figure 1).13 Although the NOE signal between the protons at C5 and C19 (the substituent at C10) is unequivocal and establishes the cis fusion between rings A and F, the proposed NOE signals between the protons at C3 and C5, as well as between C3 and C19, were weak and ambiguous. It was unclear if what was interpreted as a signal was in fact an authentic NOE, or if instead it could be attributed to background noise. Suspecting a structural misassignment, we attempted to resolve this issue based on NMR calculations.
We first considered four different possible structures (3-6) by varying the stereochemistry at positions C3, C5, and C10 while maintaining the cis-fusion between rings A and F. The cis-fusion is supported by reliable NOESY data and furthermore by the presumed biosynthesis that involves a lactonization to form the left-most F-ring γ-lactone. Biosynthetically, the ring F lactone is proposed to form via an intramolecular cyclization of the tethered ester, which one might expect would proceed to provide the cis-fused relationship.14 Presumed biosynthesis aside, the spectral data were unambiguous on this matter.
We next conducted NMR calculations on the four diastereomers we considered to be the most likely candidates. In addition to inversion of the originally assigned C3 position—the one position for which biosynthetic considerations did not provide much insight—the pair of stereocenters at C5 and C10 were additionally inverted, maintaining the cis-relationship. These additional modifications were evaluated due to the possible inversion of the C5 and C10 stereocenters over the course of the biosynthesis. This is challenging to evaluate given the biosynthesis has not been experimentally characterized for the pyridine limonoids.
2. | MATERIALS AND METHODS
The general workflow for GIAO NMR calculations was performed according to the procedure by Hoye and co-workers.15,16 Six to 12 conformers were generated for each possible structure via molecular mechanics calculations using MacroModel before they were subjected to geometry optimization calculations using Gaussian '09. Geometry optimization calculations were performed in the gas phase using the B3LYP/6–31+G(d,p) level of theory. NMR single point calculations (GIAO) were performed on Gaussian '09 for all geometrically optimized structures using mPW1PW91/6–311+G(2d,p) in chloroform with the SMD solvation method. The NMR and free energy data were assembled and Boltzmann-averaged. All chemical shifts were scaled (http://cheshirenmr.info). The NMR data, as well as the maximum and average deviations between the scaled calculated and experimental chemical shifts, are shown in Tables 1 and 2. The scaled shifts were also subjected to statistical analysis using DP4.17
TABLE 1.
13C NMR chemical shifts calculated with GIAOa
 | |||||||||
|---|---|---|---|---|---|---|---|---|---|
| Position | Wu et al 13C δ, ppm13 | 3 δ, ppm | Δ(3-reported), ppm | 4 δ, ppm | Δ(4-reported), ppm | 5 δ, ppm | Δ(5-reported), ppm | 6 δ, ppm | Δ(6-reported), ppm |
| 1 | 156.7 | 155.4 | −1.3 | 157.0 | 0.3 | 155.4 | −1.3 | 156.9 | 0.2 |
| 2 | 130.6 | 130.4 | −0.2 | 130.0 | −0.6 | 130.5 | −0.1 | 130.2 | −0.4 |
| 3 | 75.6 | 75.5 | −0.1 | 76.2 | 0.6 | 75.6 | 0.0 | 76.0 | 0.4 |
| 4 | 36.3 | 39.9 | 3.6 | 39.2 | 2.9 | 40.0 | 3.7 | 39.2 | 2.9 |
| 5 | 45.8 | 49.8 | 4.0 | 46.1 | 0.3 | 49.8 | 4.0 | 46.2 | 0.4 |
| 6 | 31.0 | 33.4 | 2.4 | 32.5 | 1.5 | 33.4 | 2.4 | 32.5 | 1.5 |
| 7 | 175.2 | 176.0 | 0.8 | 176.4 | 1.2 | 176.1 | 0.9 | 176.4 | 1.2 |
| 8 | 124.3 | 123.5 | −0.8 | 123.4 | −0.9 | 123.5 | −0.8 | 123.4 | −0.9 |
| 9 | 158.1 | 157.4 | −0.7 | 158.4 | 0.3 | 157.2 | −0.9 | 158.2 | 0.1 |
| 10 | 84.1 | 83.7 | −0.4 | 84.2 | 0.1 | 83.4 | −0.7 | 84.0 | −0.1 |
| 11 | 28.0 | 29.8 | 1.8 | 29.9 | 1.9 | 29.7 | 1.7 | 29.9 | 1.9 |
| 12 | 30.3 | 31.3 | 1.0 | 31.2 | 0.9 | 30.9 | 0.6 | 31.1 | 0.8 |
| 13 | 37.7 | 41.1 | 3.4 | 41.1 | 3.4 | 41.9 | 4.2 | 41.5 | 3.8 |
| 14 | 157.2 | 159.8 | 2.6 | 159.7 | 2.5 | 160.0 | 2.8 | 159.7 | 2.5 |
| 15 | 111.2 | 110.5 | −0.7 | 110.4 | −0.8 | 110.4 | −0.8 | 110.3 | −0.9 |
| 16 | 165.2 | 164.3 | −0.9 | 164.3 | −0.9 | 164.2 | −1.0 | 164.2 | −1.0 |
| 17 | 80.9 | 79.6 | −1.3 | 79.6 | −1.3 | 81.0 | 0.1 | 80.2 | −0.7 |
| 18 | 15.8 | 15.3 | −0.5 | 15.2 | −0.6 | 14.7 | −1.1 | 15.0 | −0.8 |
| 19 | 28.5 | 27.3 | −1.2 | 26.2 | −2.3 | 27.0 | −1.5 | 25.9 | −2.6 |
| 20 | 119.8 | 121.3 | 1.5 | 121.1 | 1.3 | 121.9 | 2.1 | 121.4 | 1.6 |
| 21 | 141.4 | 140.8 | −0.6 | 140.8 | −0.6 | 141.2 | −0.2 | 141.0 | −0.4 |
| 22 | 110.0 | 110.2 | 0.2 | 110.2 | 0.2 | 109.6 | −0.4 | 109.9 | −0.1 |
| 23 | 143.3 | 142.0 | −1.3 | 142.0 | −1.3 | 141.2 | −2.1 | 141.6 | −1.7 |
| 28 | 23.5 | 24.1 | 0.6 | 22.3 | −1.2 | 15.4 | −8.1 | 19.2 | −4.3 |
| 29 | 21.4 | 15.6 | −5.8 | 19.2 | −2.2 | 24.0 | 2.6 | 22.3 | 0.9 |
| 30 | 133.9 | 131.7 | −2.2 | 133.0 | −0.9 | 131.6 | −2.3 | 133.1 | −0.8 |
| MAE | 1.5 | 1.2 | 1.8 | 1.3 | |||||
| RMSD | 2.0 | 1.5 | 2.5 | 1.7 | |||||
| Max | 5.8 | 3.4 | 8.1 | 4.3 | |||||
| DP4 | 0.6% | 81.8% | 0.0% | 17.6% | |||||
Chemical shifts with an absolute deviation greater than 3.0 ppm are colored in red.
TABLE 2.
1H NMR chemical shifts calculated with GIAOa
 | |||||||||
|---|---|---|---|---|---|---|---|---|---|
| Position | Wu et al 1H δ, ppm13 | 3 δ, ppm | Δ(3-reported), ppm | 4 δ, ppm | Δ(4-reported), ppm | 5 δ, ppm | Δ(5-reported), ppm | 6 δ, ppm | Δ(6-reported), ppm |
| 1 | |||||||||
| 2 | |||||||||
| 3 | 4.48 | 4.40 | −0.08 | 4.34 | −0.14 | 4.41 | −0.07 | 4.33 | −0.15 |
| 4 | |||||||||
| 5 | 2.97 | 2.43 | −0.54 | 2.90 | −0.07 | 2.42 | −0.55 | 2.90 | −0.07 |
| 6α | 2.58 | 2.91 | 0.33 | 2.44 | −0.14 | 2.79 | 0.21 | 2.88 | 0.30 |
| 6β | 2.98 | 2.77 | −0.21 | 2.86 | −0.12 | 2.91 | −0.07 | 2.44 | −0.54 |
| 7 | |||||||||
| 8 | |||||||||
| 9 | |||||||||
| 10 | |||||||||
| 11α | 3.10 | 2.96 | −0.14 | 2.98 | −0.12 | 2.98 | −0.12 | 2.97 | −0.13 |
| 11β | 3.20 | 2.98 | −0.22 | 2.99 | −0.21 | 3.05 | −0.15 | 3.06 | −0.14 |
| 12α | 1.87 | 1.63 | −0.24 | 1.63 | −0.24 | 1.92 | 0.05 | 1.76 | −0.11 |
| 12β | 1.75 | 1.69 | −0.06 | 1.68 | −0.07 | 1.86 | 0.11 | 1.77 | 0.02 |
| 13 | |||||||||
| 14 | |||||||||
| 15 | 6.57 | 6.37 | −0.20 | 6.34 | −0.23 | 6.39 | −0.18 | 6.34 | −0.23 |
| 16 | |||||||||
| 17 | 5.22 | 5.07 | −0.15 | 5.07 | −0.15 | 5.12 | −0.10 | 5.09 | −0.13 |
| 18b | 1.15 | 1.13 | −0.02 | 1.14 | −0.01 | 0.98 | −0.17 | 1.07 | −0.08 |
| 19b | 1.83 | 1.64 | −0.19 | 1.69 | −0.14 | 1.64 | −0.19 | 1.68 | −0.15 |
| 20 | |||||||||
| 21 | 7.56 | 7.23 | −0.33 | 7.24 | −0.32 | 7.41 | −0.15 | 7.32 | −0.24 |
| 22 | 6.52 | 6.52 | 0.00 | 6.52 | 0.00 | 6.22 | −0.30 | 6.39 | −0.13 |
| 23 | 7.48 | 7.27 | −0.21 | 7.27 | −0.21 | 7.26 | −0.22 | 7.28 | −0.20 |
| 28b | 1.16 | 1.00 | −0.16 | 1.12 | −0.04 | 0.80 | −0.36 | 0.68 | −0.48 |
| 29b | 0.83 | 0.80 | −0.03 | 0.66 | −0.17 | 1.00 | 0.17 | 1.11 | 0.28 |
| 30 | 8.05 | 8.15 | 0.10 | 7.83 | −0.22 | 8.16 | 0.11 | 7.83 | −0.22 |
| MAE | 0.18 | 0.14 | 0.18 | 0.20 | |||||
| RMSD | 0.22 | 0.17 | 0.22 | 0.24 | |||||
| Max | 0.54 | 0.32 | 0.55 | 0.54 | |||||
| DP4 | 0.0% | 100.0% | 0.0% | 0.0% | |||||
Chemical shifts with an absolute deviation greater than 0.20 ppm are colored in red.
1H shifts of homotopic protons were averaged.
DU8+ calculations were performed as previously described,18 except the GIAO computations were performed with the Gaussian '09 default PCM method (in chloroform). The structures were pre-optimized with the force field MMFF94 as implemented in OpenBabel before they were subjected to geometry optimization calculations using Gaussian '09.19 Geometry optimization calculations were performed using the B3LYP/6–31G(d) level of theory with the default PCM solvation method (chloroform). NMR single point calculations (GIAO) were performed on Gaussian '09 for all geometrically optimized structures using ωB97xD/6–31G(d) in chloroform with the default PCM solvation method. The DU8+ empirical corrections for 13C chemical shifts were then applied.18 The NMR data were Boltzmann-averaged.
3. | RESULTS AND DISCUSSION
3.1. | GIAO NMR calculation results
The results of the GIAO NMR calculations are consistent with the reassignment of xylogranatin F to the 3S-5S-10S structure (4), ie, the C3-epimer of the original structure. According to the 13C NMR calculation results, the chemical shifts of the originally proposed structure, 3R-5S-10S (3), have an average deviation of 1.5 ppm and RMSD of 2.0 (Table 1). Furthermore, the chemical shifts of several carbons on rings A and F around the C3 stereogenic center differ by more than 3.0 ppm. Specifically, calculated 13C chemical shifts of C4 and C5 on ring A are 3.6 ppm and 4.0 ppm more downfield than the reported values, respectively. On the other hand, those of C29 are 5.8 ppm more upfield than the experimentally observed shifts. In comparison, structure (4) has the lowest average chemical shift deviation of 1.2 ppm and RMSD of 1.5. The deviations between the calculated 13C NMR chemical shifts of C4, C5, and C29 for 4 and the reported 13C NMR shifts are all within 3.0 ppm. In addition, 4 is predicted to be the correct structure with 81.8% probability by DP4 analysis based on the 13C NMR shifts.17
The 1H NMR calculation results also suggest structural misassignment (Table 2). The chemical shifts of the originally proposed structure, 3R-5S-10S (3), have an average deviation of 0.18 ppm and RMSD of 0.22 (Table 2). In addition, the deviations between the calculated 1H NMR chemical shifts of the protons on C5 and C6 and the values reported by Wu et al are more than 0.20 ppm. In contrast, the 3S-5S-10S structure (4) has the lowest average chemical shift deviation of 0.14 ppm with RMSD of 0.17. The deviations between the calculated 1H NMR chemical shifts of the protons on C5 and C6 for 4 and the reported 13C NMR shifts are all within 0.20 ppm. Additionally, 4 is predicted to be the correct structure with 100.0% probability using the 1H NMR chemical shifts only, as well as using both proton and carbon data by DP4 analysis.17
3.2. | DU8+ calculation results
The results of the DU8+ calculations confirmed the structural misassignment of xylogranatin F (Tables 3 and 4). The most striking discrepancy between the experimental 13C NMR chemical shift data and the calculated data for the originally proposed structure 3 was the chemical shift of C29 (off by 7.6 ppm). The 13C chemical shift of C5 also deviated by 3.0 ppm. In contrast, the calculated chemical shifts of 3S-5S-10S structure (4) were in an excellent agreement with the experimental data: RMSD = 0.8 ppm, MAE = 0.6 ppm, and the maximum deviation of 2.0 ppm.
TABLE 3.
13C NMR chemical shifts calculated with DU8+a
 | |||||||||
|---|---|---|---|---|---|---|---|---|---|
| Position | Wu et al 13C δ, ppm13 | 3 δ, ppm | Δ(3-reported), ppm | 4 δ, ppm | Δ(4-reported), ppm | 5 δ, ppm | Δ(5-reported), ppm | 6 δ, ppm | Δ(6-reported), ppm |
| 1 | 156.7 | 156.0 | −0.7 | 157.7 | 1.0 | 155.8 | −0.9 | 157.6 | 0.9 |
| 2 | 130.6 | 131.6 | 1.0 | 131.9 | 1.3 | 131.6 | 1.0 | 131.9 | 1.3 |
| 3 | 75.6 | 75.6 | 0.0 | 76.2 | 0.6 | 75.6 | 0.0 | 76.2 | 0.6 |
| 4 | 36.3 | 37.1 | 0.8 | 35.8 | −0.5 | 37.0 | 0.7 | 35.7 | −0.6 |
| 5 | 45.8 | 48.8 | 3.0 | 46.8 | 1.0 | 48.8 | 3.0 | 46.6 | 0.8 |
| 6 | 31.0 | 31.7 | 0.7 | 31.4 | 0.4 | 31.8 | 0.8 | 31.4 | 0.4 |
| 7 | 175.2 | 174.5 | −0.7 | 174.4 | −0.8 | 174.5 | −0.7 | 174.5 | −0.7 |
| 8 | 124.3 | 124.9 | 0.6 | 124.8 | 0.5 | 125.0 | 0.7 | 124.8 | 0.5 |
| 9 | 158.1 | 158.3 | 0.2 | 159.4 | 1.3 | 158.4 | 0.3 | 159.5 | 1.4 |
| 10 | 84.1 | 85.2 | 1.1 | 86.1 | 2.0 | 85.0 | 0.9 | 85.9 | 1.8 |
| 11 | 28.0 | 28.5 | 0.5 | 28.6 | 0.6 | 28.5 | 0.5 | 28.6 | 0.6 |
| 12 | 30.3 | 30.3 | 0.0 | 30.2 | −0.1 | 30.3 | 0.0 | 30.3 | 0.0 |
| 13 | 37.7 | 37.8 | 0.1 | 37.9 | 0.2 | 37.9 | 0.2 | 37.8 | 0.1 |
| 14 | 157.2 | 157.8 | 0.6 | 157.6 | 0.4 | 157.7 | 0.5 | 157.4 | 0.2 |
| 15 | 111.2 | 111.8 | 0.6 | 111.7 | 0.5 | 111.8 | 0.6 | 111.9 | 0.7 |
| 16 | 165.2 | 165.2 | 0.0 | 165.1 | −0.1 | 165.2 | 0.0 | 165.1 | −0.1 |
| 17 | 80.9 | 82.0 | 1.1 | 81.9 | 1.0 | 81.8 | 0.9 | 81.9 | 1.0 |
| 18 | 15.8 | 16.1 | 0.3 | 16.1 | 0.3 | 16.2 | 0.4 | 16.1 | 0.3 |
| 19 | 28.5 | 28.2 | −0.3 | 28.5 | 0.0 | 28.0 | −0.5 | 28.3 | −0.2 |
| 20 | 119.8 | 121.0 | 1.2 | 120.8 | 1.0 | 120.9 | 1.1 | 120.7 | 0.9 |
| 21 | 141.4 | 141.4 | 0.0 | 141.2 | −0.2 | 141.2 | −0.2 | 141.3 | −0.1 |
| 22 | 110.0 | 109.8 | −0.2 | 109.7 | −0.3 | 109.7 | −0.3 | 109.7 | −0.3 |
| 23 | 143.3 | 142.4 | −0.9 | 142.3 | −1.0 | 142.3 | −1.0 | 142.3 | −1.0 |
| 28 | 23.5 | 24.2 | 0.7 | 23.2 | −0.3 | 24.1 | 0.6 | 23.2 | −0.3 |
| 29 | 21.4 | 13.8 | −7.6 | 20.9 | −0.5 | 13.7 | −7.7 | 20.7 | −0.7 |
| 30 | 133.9 | 132.6 | −1.3 | 134.0 | 0.1 | 132.8 | −1.1 | 133.8 | −0.1 |
| MAE | 0.9 | 0.6 | 0.9 | 0.6 | |||||
| RMSD | 1.7 | 0.8 | 1.7 | 0.7 | |||||
| Max | 7.6 | 2.0 | 7.7 | 1.8 | |||||
Chemical shifts with an absolute deviation greater than 3.0 ppm are colored in red.
TABLE 4.
1H NMR chemical shifts calculated with DU8+a
 | |||||||||
|---|---|---|---|---|---|---|---|---|---|
| Position | Wu et al 1H δ, ppm13 | 3 δ, ppm | Δ(3-reported), ppm | 4 δ, ppm | Δ(4-reported), ppm | 5 δ, ppm | Δ(5-reported), ppm | 6 δ, ppm | Δ(6-reported), ppm |
| 1 | |||||||||
| 2 | |||||||||
| 3 | 4.48 | 4.74 | 0.26 | 4.73 | 0.25 | 4.77 | 0.29 | 4.70 | 0.22 |
| 4 | |||||||||
| 5 | 2.97 | 2.51 | −0.46 | 3.15 | 0.18 | 2.51 | −0.46 | 3.15 | 0.18 |
| 6α | 2.58 | 2.71 | 0.13 | 2.56 | −0.02 | 2.72 | 0.14 | 2.57 | −0.01 |
| 6β | 2.98 | 3.03 | 0.05 | 2.93 | −0.05 | 3.03 | 0.05 | 2.95 | −0.03 |
| 7 | |||||||||
| 8 | |||||||||
| 9 | |||||||||
| 10 | |||||||||
| 11α | 3.10 | 3.23 | 0.13 | 3.23 | 0.13 | 3.18 | 0.08 | 3.19 | 0.09 |
| 11β | 3.20 | 3.17 | −0.03 | 3.19 | −0.01 | 3.24 | 0.04 | 3.24 | 0.04 |
| 12α | 1.87 | 1.83 | −0.04 | 1.84 | −0.03 | 1.82 | −0.05 | 1.80 | −0.07 |
| 12β | 1.75 | 1.87 | 0.12 | 1.86 | 0.11 | 1.90 | 0.15 | 1.88 | 0.13 |
| 13 | |||||||||
| 14 | |||||||||
| 15 | 6.57 | 6.41 | −0.16 | 6.34 | −0.23 | 6.46 | −0.11 | 6.38 | −0.19 |
| 16 | |||||||||
| 17 | 5.22 | 5.31 | 0.09 | 5.30 | 0.08 | 5.29 | 0.07 | 5.29 | 0.07 |
| 18b | 1.15 | 1.37 | 0.22 | 1.35 | 0.20 | 1.37 | 0.22 | 1.37 | 0.22 |
| 19b | 1.83 | 1.78 | −0.05 | 1.92 | 0.09 | 1.77 | −0.06 | 1.91 | 0.08 |
| 20 | |||||||||
| 21 | 7.56 | 7.62 | 0.06 | 7.62 | 0.06 | 7.61 | 0.05 | 7.61 | 0.05 |
| 22 | 6.52 | 6.63 | 0.11 | 6.63 | 0.11 | 6.64 | 0.12 | 6.64 | 0.12 |
| 23 | 7.48 | 7.62 | 0.14 | 7.63 | 0.15 | 7.62 | 0.14 | 7.61 | 0.13 |
| 28b | 1.16 | 1.26 | 0.10 | 1.36 | 0.20 | 1.25 | 0.09 | 1.37 | 0.21 |
| 29b | 0.83 | 0.95 | 0.12 | 0.98 | 0.15 | 0.96 | 0.13 | 0.98 | 0.15 |
| 30 | 8.05 | 8.62 | 0.57 | 8.19 | 0.14 | 8.63 | 0.58 | 8.20 | 0.15 |
| MAE | 0.16 | 0.12 | 0.16 | 0.12 | |||||
| RMSD | 0.21 | 0.14 | 0.21 | 0.14 | |||||
| Max | 0.57 | 0.25 | 0.58 | 0.22 | |||||
Chemical shifts with an absolute deviation greater than 0.20 ppm are colored in red.
1H shifts of homotopic protons were averaged.
However, the five-bond separation between the closest stereogenic centers C3 and C13 in the two pyridine-separated cyclohexene moieties makes it unlikely to differentiate between diastereomers 4 and 6 based solely on the calculated NMR data. The same is true for the pair of diastereomers 3 and 5, which are also indistinguishable based on the calculated NMR data. It is understandable that with separation of stereogenic centers, the difference in 13C NMR chemical shifts for the two diastereomers becomes less pronounced such that it may drop below the accuracy of the computational method. The fact that the calculated 13C NMR shifts with DU8+ method for diastereomers 4 and 6 are so similar points to the obvious limitation of NMR in differentiating between such diastereomers based on shifts alone. The local environment in the fragment containing rings A and F allows for differentiation between the originally proposed structure (3) and its C3-epimer (4) based on calculated NMR shifts, but fails to relate stereogenic centers in this fragment and fragment containing rings C and D. What aggravates the situation is that there are no useful NOE enhancements across the pyridine spacer. In cases like this, one should take into consideration other criteria, for example, the biosynthetic origins of the natural products. Although the putative biosynthesis of the xylogranatins has not yet been experimentally corroborated, if this class of limonoid alkaloids is directly related to the mexicanolide-type limonoids,13,14 the configuration of the stereocenter at C5 should be S as in diastereomer 4 and not R as in 5 and 6.20
4. | CONCLUSION
In conclusion, NMR calculations were utilized in substantiating the structural misassignment of xylogranatin F as well as predicting the correct structure (4), which is a demonstration of the utility of NMR calculations for stereochemical assignment. The hybrid DFT-parametric DU8+ method has demonstrated superior accuracy using a fraction of computation cost. The MAE (δ 13C) of 0.6 ppm for the revised structure obtained with DU8+ is significantly better than MAE of 1.2 ppm achieved with a more standard GIAO method at the mPW1PW91/6–311+G(2d,p)//B3LYP/6–31+G(d,p) level of theory. Calculations of chemical shifts per conformer took under 7 minutes on a 16-core node of a Linux cluster for DU8+, while GIAO at the aforementioned level of theory, on the same node, took 10 times longer (70+ min). However, although the accuracy was lower for GIAO than DU8+, GIAO correctly predicted the right structure with larger differences predicted among multiple isomers. It is expected that the approach taken herein will help assist natural product structural assignment in other instances where clusterings of stereocenters are separated spatially.
Supplementary Material
ACKNOWLEDGEMENTS
Financial support for this work was provided by Yale University, Amgen, the Dreyfus Foundation, the Sloan Foundation, the NSF (CHE-1665342), and the NIH (GM118614). We gratefully acknowledge the National Science Foundation for financial support in the establishment of the Yale University High Performance Computing (HPC) Center (CNS 08-21132).
Funding information Alfred P. Sloan Foundation; Amgen; Camille and Henry Dreyfus Foundation; Division of Chemistry, Grant/Award Number: CHE-1665342; Division of Computer and Network Systems, Grant/ Award Number: CNS 08-21132; National Institutes of Health, Grant/Award Number: GM118614; Yale University; Yale University High Performance Computing (HPC) Center, Grant/Award Number: CNS 08-21132; NIH, Grant/Award Number: GM118614; NSF, Grant/Award Number: CHE-1665342
Footnotes
SUPPORTING INFORMATION
Additional supporting information may be found online in the Supporting Information section at the end of this article.
REFERENCES
- 1.Bifulco G, Dambruoso P, Gomez-Paloma L, Riccio R. Determination of relative configuration in organic compounds by NMR spectroscopy and computational methods. Chem Rev. 2007;107 (9):3744–3779. [DOI] [PubMed] [Google Scholar]
- 2.Lodewyk MW, Siebert MR, Tantillo DJ. Computational prediction of 1H and 13C chemical shifts: a useful tool for natural product, mechanistic and synthetic organic chemistry. Chem Rev. 2012;112(3):1839–1862. [DOI] [PubMed] [Google Scholar]
- 3.Tantillo DJ. Walking in the woods with quantum chemistry-applications of quantum chemical calculations in natural products research. Nat Prod Rep. 2013;30(8):1079–1086. [DOI] [PubMed] [Google Scholar]
- 4.Grimblat N, Sarotti AM. Computational chemistry to the rescue: modern toolboxes for the assignment of complex molecules by GIAO NMR calculations. Chem A Eur J. 2016;22(35): 12246–12261. [DOI] [PubMed] [Google Scholar]
- 5.Rychnovsky SD. Predicting NMR spectra by computational methods: structure revision of hexacyclinol. Org Lett. 2006;8 (13):2895–2898. [DOI] [PubMed] [Google Scholar]
- 6.Lodewyk MW, Soldi C, Jones PB, et al. The correct structure of aquatolide–experimental validation of a theoretically-predicted structural revision. J Am Chem Soc. 2012;134(45):18550–18553. [DOI] [PubMed] [Google Scholar]
- 7.Novaes LFT, Sarotti AM, Pilli RA. Total synthesis and tentative structural elucidation of cryptomoscatone E3: interplay of experimental and computational studies. J Org Chem. 2015;80 (24):12027–12037. [DOI] [PubMed] [Google Scholar]
- 8.Grimblat N, Kaufman TS, Sarotti AM. Computational chemistry driven solution to rubriflordilactone B. Org Lett. 2016;18 (24):6420–6423. [DOI] [PubMed] [Google Scholar]
- 9.Wu J, Lorenzo P, Zhong S, et al. Synergy of synthesis, computation and NMR reveals correct baulamycin structures. Nature. 2017;547(7664):436–440. [DOI] [PubMed] [Google Scholar]
- 10.Shepherd ED, Dyson BS, Hak WE, et al. Structure determination of a chloroenyne from laurencia majuscule using computational methods and total synthesis. J Org Chem. 2019;84(9): 4971–4991. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Kutateladze A, Holt T. Structure validation of complex natural products: time to change the paradigm. What did synthesis of alstofolinine A prove? J Org Chem. 2019;84(12): 8297–8299. [DOI] [PubMed] [Google Scholar]
- 12.Schuppe AW, Zhao Y, Liu Y, Newhouse TR. Total synthesis of (+)-granatumine A and related bislactone limonoid alkaloids via a pyran to pyridine interconversion. J Am Chem Soc. 2019; 141(23):9191–9196. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Wu J, Zhang S, Bruhn T, Xiao Q, Ding H, Bringmann G. Xylogranatins F-R: antifeedants from the Chinese mangrove, Xylocarpus granatum, a new biogenetic pathway to tetranotriterpenoids. Chem A Eur J. 2008;14(4):1129–1144. [DOI] [PubMed] [Google Scholar]
- 14.Zhou ZF, Liu HL, Zhang W, et al. Bioactive rearranged limonloids from the Chinese mangrove Xylocarpus granatum Koenig. Tetrahedron. 2014;70(37):6444–6449. [Google Scholar]
- 15.Willoughby PH, Jansma MJ, Hoye TR. A guide to small-molecule structure assignment through computation of (1H and 13C) NMR chemical shifts. Nat Protoc. 2014;9(3):643–660. [DOI] [PubMed] [Google Scholar]
- 16.Neupane JB, Neupane RP, Luo YH, Yoshida WY, Sun R, Williams PG. Characterization of leptazolines A-D, polar oxazolines from the cyanobacterium leptolyngbya sp., reveals a glitch with the “Willoughby-Hoye” scripts for calculating NMR chemical shifts. Org Lett. 2019;21(20):8449–8453. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Smith SG, Goodman JM. Assigning stereochemistry to single diastereoisomers by GIAO NMR calculation: the DP4 probability. J Am Chem Soc. 2010;132(37):12946–12959. [DOI] [PubMed] [Google Scholar]
- 18.Kutateladze AG, Reddy DS. High-throughput in silico structure validation and revision of halogenated natural products is enabled by parametric corrections to DFT-computed 13C NMR chemical shifts and spin–spin coupling constants. J Org Chem. 2017;82(7):3368–3381. [DOI] [PubMed] [Google Scholar]
- 19.O'Boyle NM, Banck M, James CA, Morley C, Vandermeersch T, Hutchison GR. Open Babel: an open chemical toolbox. J Cheminformatics. 2011;3(33):1–14. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Tan QG, Luo XD. Meliaceous limonoids: chemistry and biological activities. Chem Rev. 2011;111(11):7437–7522. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.