October 31, 2022
by Mikhail Elyashberg, Leading Researcher, ACD/Labs
This example application of ACD/Structure Elucidator (ACD/SE) comes from a publication by M. Elyashberg and coworkers  devoted to structure revisions of a set of natural products whose structures were already published in literature.
Kim and coworkers  isolated a new neolignan, piperkadsin C (1), and elucidated its structure from HRMS and NMR spectra (1H, 13C, HSQC, HMBC).
Compound 1 was obtained as a viscus gum. In the positive mode FABMS of 1, a quasimolecular ion peak at 340 [M]+ was observed, and the molecular formula of 1 was determined to be C20H20O5 by HR-FABMS: observed, 340.1315 (calculated for C20H20O5, 340.1311).
However, this structure was revised by the same group in their next article . In the positive mode FABMS, a molecular ion peak [M+H]+ at m/z 357 was observed, and the molecular formula of the compound was determined to be C20H20O6 by the HR-FABMS spectra and a molecular ion peak [M+H]+ at m/z 357.1335 was observed (calcd. for C20H20O6, 357.1338). The following revised structure was proposed:
As we see, the molecular formula of the revised structure contains one additional oxygen. It is interesting that in both cases calculated and measured molecular masses agreed with great accuracy.
To verify the revised structure 2, the NMR spectroscopic data presented in the article  were entered into ACD/Structure Elucidator (see Table 1).
Table 1. NMR spectroscopic data of the structure 2
|Label||δC||δC Calc (HOSE)||CHn||δH||C to H HMBC|
|C 2||108.800||106.580||CH||6.920||C 7, C 6, C 1, C 4, C 3|
|C 5||108.100||107.780||CH||6.760||C 6, C 1, C 4, C 3|
|C 6||122.700||123.980||CH||6.750||C 7, C 5, C 2, C 1, C 4|
|C 7||65.300||68.450||CH||3.640||C 9, C 8, C 14, C 2, C
6, C 1, C 13, C 15
|C 9||16.800||20.600||CH3||0.960||C 8, C 14, C 7, C 15|
|C 11||157.200||118.150||CH||7.000||C 16, C 12, C 10, C 13,
|C 14||58.600||56.020||CH||3.360||C 9, C 8, C 7, C 12, C
10, C 13, C 15
|C 16||33.000||35.700||CH2||2.910||C 18, C 10, C 17, C 11,
|C 17||135.500||134.430||CH||5.820||C 16, C 18, C 10|
|C 18||116.700||116.990||CH2||5.060||C 16, C 10, C 17|
|C 19||101.100||101.220||CH2||5.940||C 4, C 3|
|C 20||59.200||55.130||CH3||3.820||C 12|
The 13C chemical shifts assigned by authors  were set to carbon atoms of 2, and chemical shift prediction was performed using three algorithms implemented into the ACD/Structure Elucidator: HOSE code based, incremental and neural networks. Results are shown in Figure 1.We see that the average deviations are large, which hinted that the structure is problematic. Figure 1 shows that the proposed structure contains four nonstandard correlations (NSC) of 4-bonds length. To verify the structure using ACD/SE, a Molecular Connectivity Diagram (MCD) was created in which the “green fragment” (confirmed by 13C prediction) was pre-defined (Figure 2). Also, the caron atom at 197.2 ppm was pre-defined to be a carbonyl, as there isn’t any other option for that chemical shift value.
As structure 2 contains four NSCs, Fuzzy Structure Generation was run from the MCD (Figure 2) with the following options: the presence of four NSCs of unknown lengths was admitted; 13C chemical shifts were calculated during the generation; structural filtering was switched off to provide saving of the proposed structure 2. Results: k =667→(removal of duplicates)→606, tg = 44 s.
After 13C chemical shift prediction using all three methods, the output file was ranked in increasing order of dA average deviations. The nine top ranked structures are presented in Figure 3.
Figure 3 shows that all structures are characterized by large average and maximum deviations, while the proposed structure 2 was placed in 9th position by the ranking procedure. This means that all generated structures, including the revised one, are incorrect. We hypothesized that there is a probability to deduce a correct structure from the molecular formula and NMR data used for deducing structure 1.
With this in mind, we entered the indicated data (Table 2) into the ACD/SE and, as in the previous case, verified structure 1 (see Figure 4).
Table 2. NMR spectroscopic data used for deducing structure 1 
|Label||δC||δC calc (HOSE)||CHn||δH||H to C HMBC|
|C 2||108.800||107.290||CH||6.920||C 7, C 6, C 1, C 4, C 3|
|C 5||108.100||108.010||CH||6.760||C 6, C 1, C 4, C 3|
|C 6||122.700||122.240||CH||6.750||C 7, C 5, C 2, C 1, C 4|
|C 7||65.300||75.920||CH||3.640||C 9, C 8, C 14, C 2, C
6, C 12, C 1, C 15
|C 9||16.800||21.680||CH3||0.960||C 8, C 14, C 7, C 12, C
13, C 11
|C 11||157.200||132.440||CH||7.000||C 16, C 8, C 12, C 10,
C 13, C 15
|C 14||58.600||52.450||CH||3.360||C 9, C 8, C 7, C 12, C
10, C 13, C 15
|C 16||33.000||35.440||CH2||2.910||C 18, C 10, C 17, C 11,
|C 17||135.500||134.660||CH||5.820||C 16, C 18, C 10|
|C 18||116.700||116.910||CH2||5.060||C 16, C 10, C 17|
|C 19||101.100||101.270||CH2||5.940||C 4, C 3|
|C 20||59.200||59.030||CH3||3.820||C 13|
Figure 4 shows that structure 1 is wrong, as it has large average and maximum deviations. Moreover, this structure contradicts Bredt’s rule and cannot exist due to its instability .
The MCD created from the data presented in Table 2 is shown below in Figure 5
Fuzzy Structure Generation was initiated with options selected automatically. Structure filtering was switched off as in the previous case. Results: k = 6, tg = 6 m 50 s, 4 (from 44) connectivities have been extended during generation. The ranked output file is shown in Figure 6.
We see that the two best structures #1 and #2 are similar and differ only by the positions of the OH and O-CH3 groups. The original structure 1 was placed in 5th position. We subsequently used the methodology suggested by authors of ref. . On the basis of DFT-based 13C chemical shift prediction using the DU8+ approach, it was determined that structure #2 is the correct one with a Mean Average Error of 1.08 ppm. Structure #1 has a MAE value of 2.17 ppm.
Thus, the correct structure of piperkadsin C was determined as a result of the combined application of CASE and DFT-based chemical shift prediction for the top structures of the ranked output file.
- Elyashberg, M.; Novitskiy, I. M.; Bates, R. W.; Kutateladze, A. G.; Williams, C. M. (2022). Reassignment of Improbable Natural Products Identified through Chemical Principle Screening. European Journal of Organic Chemistry, e202200572. https://doi.org/10.1002/ejoc.202200572
- K.H. Kim, J. W.Choi, S. K. Ha, S.Y. Kim, K. R. Lee. (2010). Neolignans from Piper kadsura and their anti-neuroinflammatory activity. Bioorganic & Medicinal Chemistry Letters, 20, 409–412.
- K.H. Kim, J. W.Choi, S. K. Ha, S.Y. Kim, K. R. Lee. (2010). Corrigendum to ‘‘Neolignans from Piper kadsura and their anti-neuroinflammatory activity”. Bioorganic & Medicinal Chemistry Letters, 20, 3186–3187.
- A. V. Buevich, M. E. Elyashberg. (2016). J. Nat. Prod., 79 (12), 3105–3116.