Synergistic Combination of NAPROC-13 and NMR 13C DFT Calculations: A Powerful Approach for Revising the Structure of Natural Products

This article describes the structure revision of nine triterpenoids that have been reported corresponding to the same 13C NMR data set. In addition, 13C NMR calculation shows that some chemical shift assignments must be swapped. Our analysis improves the fit between the experimental and calculated data. Correcting misassigned structures and correctly assigning each signal is essential for elucidating new structurally related compounds. Furthermore, the ambiguity of several compounds, the structure of which differs in the literature and the Sci-Finder database, has been eliminated. Misassigned structures were found by chemical shift searches in NAPROC-13, and the results provide two or more different compounds with the same 13C NMR data. The process to determine the correct, most likely structural proposal in agreement with the experimental 13C NMR data was carried out by DFT calculations.

−4 In other cases, the structure is probably correct, but one or more NMR signals are misassigned, which is seen relatively frequently in the literature.These errors usually occur when the sample contains impurities, signals are neglected due to their low intensity, or the multiplicity of some 13 C NMR signals is misinterpreted.In addition, an uncorrected typographical mistake could occur during the writing process.NAPROC-13 5 is a web-based application that contains 13 C NMR spectroscopic information for more than 26,000 natural products.More than 6% of the entries in NAPROC-13 are corrections to publications that contain structural errors.These corrections were incorporated after appearing in articles involving structural revisions or because an error or typo was noted as the entries were introduced into the database.
This article describes the structure revision of nine triterpenoids that were published with different structures corresponding to the same 13 C NMR data set.The misassigned structures were found by searches in NAPROC-13 for chemical shifts, and the results provided two or more different compounds with the same 13 C NMR data.DFT calculations were used to establish the correct or most likely structure proposal, in agreement with the 13 C NMR experimental data.The herein methodology used for structural revisions is different from the many and varied computer-assisted structure elucidation (CASE) methods available. 6To demonstrate the usefulness of our approach, a comparative study was performed for the structure revision of two misassigned straight-chain natural polyenols, A1 and B1 (Figure 1).In one case, the study was supported by a CASE method in combination with the DFT-parametric computational method DU8+. 7In the other case, the study was supported by the method described in this publication.For further details, see the Supporting Information.The misassigned proposals and the correct structures are structurally distant.Thus, despite the structural simplicity of both compounds A2 and B2, solving these structures is undoubtedly difficult (Figure 1).Furthermore, performing structural revision is even more difficult because the molecular ion deduced from the mass spectrum of A1 and the number of carbons deduced from the molecular formula of B1 are incorrect.
Ma et al. 8 and Siebatcheu et al. 9 described the structures of compounds A1 and B1 (Figure 1).According to Kutatelage et al., 7 both compounds contain different enol groups, which is very unlikely based on their stability; thus, the compounds may be misassigned substances.According to our approach for the revision of misassigned compounds, a search by 13 C NMR chemical shifts was carried out for putative misassigned structures A1 and B1 in NAPROC-13.The search results obtained for the A1 chemical shift provided, in addition to A1, uridine (A2) and three other compounds (SI: I.1 and I.2).The 13 C NMR chemical shifts of both substances A1 and A2 are almost identical.The structure of uridine is well established and solved by X-ray diffraction. 10The search of B1 chemical shifts yielded five additional compounds (SI: II.1 and II.2).All of them are either adenosine (B2) or compounds that have adenosine as a structural fragment.A combined search that also considered the molecular weight deduced from B1 mass spectra allows us to narrow down the result to only B1 and B2 (SI: II.3, II.4, and II.5).All of the carbons of both substances are color-coded according to the deviations between the searched chemical shifts and those of the values of the found compounds except for C-5 of compound B2, which resonates at 121 ppm and is not present in B1.The availability of the 13 C NMR spectrum in the SI in both publications that describe this compound allows us to observe the presence of a signal at 121 ppm, which has been neglected by the authors.Consequently, B1 and B2 are identical.The structure of adenosine was established by X-ray diffraction. 11Therefore, A1 and B1 structures could be revised in a matter of minutes using NAPROC-13.
The structure revision of nine triterpenoids found in NAPROC-13 is discussed below, for which the application of established CASE methods may not be productive due to the size of the molecule and multiple possibilities of functional location, which generates a considerable number of alternatives and makes density functional theory (DFT) calculation methods very expensive.In addition, the different conformational possibilities for each of the stereoisomers or regioisomers generated also must be considered.

■ RESULTS AND DISCUSSION
The misassigned structures of nine triterpenoids with ursane, lupane, and oleane skeletons were computationally revised by calculating their 13 C NMR spectra.For this purpose, Spartan'20 12 was chosen, since it has implemented a protocol for calculating NMR chemical shifts in conformationally flexible molecules.These misassigned compounds were detected using NAPROC-13. 5In some of the examples presented, the structural revision involved a change in the triterpene skeleton.However, in the majority of cases, these compounds had already been published with the correct structure before.In addition, the differences between some   In a phytochemical study of the roots of Lantana camara L. (Verbenaceae), 13 a new triterpenoid with an ursane skeleton, 3β,19α-dihydroxyursan-28-oic acid (1P) (Figure 2), was isolated.Like many other triterpenoids of this group, it contains hydroxy groups at C-3 and C-19 and a carboxylic group at C-28.The distinct structural feature of this triterpene is the absence of a Δ 12 double bond, which is present in similar compounds of ursane group triterpenoids.Compound 1P is the only triterpene of the ursane group with the described structural features.These data, together with the value of the chemical shift assigned to Me-27 at δ 33.2 (in other related ursanes, Me-27 resonates at δ 18), caught our attention and prompted us to search for their chemical shifts in NAPROC-13. 5Furthermore, the values of the 13 C NMR shifts of the Ering carbons of this triterpene greatly differ from those of other ursanes.As a result of the search, in addition to 1P, two other triterpenoids were found, licanolide [3β-hydroxylupane-20(28)-olide] (2P) (Figure 2), a lupane isolated from Licania tomentosa (Chrysobalanaceae), 14 and 11α,12α-epoxy-3β-hydroxyolean-28(13β)-olide (1R) (Figure 1), an oleanane isolated from Paeonia japonica (Paeoniaceae). 15Compound 1R was previously isolated from Atractylis carduus L. 16 Compounds 1P, 2P, and 1R present very similar 13 C NMR data (see Table 1 and SI: II.6); however, as seen in Figure 2, their structures are significantly different.The compounds have different carbon skeletons of lupane, ursane, and oleanane types.Moreover, 1P contains a carboxylic acid and hydroxylfree groups, while 2P and 1R contain a γ-lactone.Compound 1R also possesses an epoxide group that is not present in the other two compounds.Despite this structural disparity, the reported mass spectra at 70 eV for the three triterpenoids showed identical m/z values of 470 [M] + for 1P and 1R and 471 [M] + for 2P.In its IR spectra, 2P shows an absorption band at 872 cm −1 characteristic of the epoxide group, similar to the band in the IR spectrum of 1R (868 cm −1 ).For 1P, no absorption bands below 1250 cm −1 are reported.Therefore, these three products, which have appeared in the literature under different structures, may correspond to the same compound.Determining the relationship between structurally different compounds is relatively straightforward using databases such as NAPROC-13, which offer the ability to search chemical shifts.In their 1 H NMR spectra, the three compounds show three signals between 3.0 and 3.4 ppm that are only compatible with the 1R structure.To determine the consistency between the proposed structures for 1P, 2P, and 1R and their corresponding 13 C NMR spectroscopic data, their 13 C NMR spectra were computationally calculated following the protocol described in the Experimental Section. 12he computational results were compared with the experimental data, and the results are summarized in Figure 2, in which the term max absolute expresses the largest deviations between the calculated and experimental chemical shifts; the statistical term rms is also reported underneath each structure.(11.5).Both proposed structures for 1P and 2P were assigned based on two-dimensional NMR analysis.The twodimensional spectra data for licanolide (2P) were reported later by the same authors. 17Comparison of the 13 C NMR signals of 2P with those of another triterpenoid with the same lupane skeleton, which differs only by free acid and hydroxyl groups instead of lactone, 3β,20-dihydroxylupane-28-oic acid, 18 shows significant differences between the signals assigned to the E-ring carbons for both substances.
The 13 C NMR spectra of the three substances (Table 1 and SI: II.6) show the same number of methyls and methines; however, 1R shows a quaternary carbon at 31.5 ppm, while 1P and 2P show a methylene, which also resonates at a similar chemical shift of 31 ppm (Table 1 and SI: II.6).The small differences observed in the remaining signals can be justified due to the use of different deuterated solvents in NMR experiments (CDCl 3 for 1P and 2P and C 5 D 5 N for 1R).Most likely, the higher purity of sample 1R prevented its solubilization in CDCl 3 and was decisive for the correct interpretation of the spectroscopic data and correct structure elucidation.When the experimental 13 C NMR chemical shifts of 1P and 2P (Table 1 and SI: II.6) are transferred to the 1R structure, assuming that the additional methylene described for 1P and 2P is actually a quaternary carbon and compared with the data obtained by computational calculation for 1R, a good fit is obtained.In addition, the rms values are close to those observed for the substance described as 1R (see SI: III.1, III.2, and III.3).Consequently, 1P with an ursane skeleton and 2P with a lupane skeleton are misassigned structures and actually correspond to 1R with an oleanane skeleton.The structure of 1R was corroborated because it was also reported as the compound obtained by the photochemical oxidation of oleanolic acid. 19As in many other reported examples of structural revisions, 1P and 2P with erroneous structures have been published with the correct structure as 1R several years before.
A phytochemical study carried out on Tripterygium doianum (Celastraceae) afforded a triterpenoid named 3β-acetoxyurs-11-en-30(13α)-olide (3P) 20 (Figure 3).The structural elucidation of this compound was based on the analysis of 1D and 2D NMR spectroscopic data.Additionally, by computational studies, a conformational search was carried out to establish the conformation of this compound and to rationalize the NOE effects observed.The authors state in the publication that 3P is the first 30,13α-olide ursene-type triterpenoid isolated from a natural source.This unique structural feature of 3P prompted us to perform a chemical shift search in NAPROC-13 to determine whether its chemical shift pattern is specific to this compound or if it is found in other structurally related substances.In addition to 3P acetate, the search on NAPROC-13 provided another triterpene named 3β-acetoxyurs-11-en-28(13β)-olide (3R) (Figure 3), isolated from Pieris japonica D. DON (Ericaceae). 21The 13 C NMR chemical shifts of 3R are practically identical to those of 3P (Table 1 and SI: II.6) with small differences that do not exceed 0.3 ppm in the most unfavorable case.This case is a new example of two triterpenoids with practically identical 13 C NMR spectra; both triterpenoids have an ursane skeleton and were assigned different structures.This prompted us to perform a computational calculation of the 13 C NMR spectra of both compounds 3P and 3R.The computational results were compared with the experimental data, and a summary of the results obtained is shown in Figure 3.In 3P, the maximum deviation of 17.9 ppm is that of C-20, which supports the carboxylic group of the lactone; moreover, other significant deviations affect the carbons in the vicinity.In contrast, when the 13 C NMR data of 3P are assigned to the 3R structure, they show a maximum difference of 3.8 and an rms value of 1.4 after several chemical shifts are swapped (SI: III.5).Therefore, the spectroscopic data for both substances show a good fit between the calculated and experimental values for 3R and do not seem compatible with the 3P structure (max abs = 17.9, rms = 5.6), as shown in Figure 3. Compound 3R was isolated from Euclea natalensis (Ebenaceae), 29 and its structure was confirmed by semisynthesis from acetylursolic acid. 30Therefore, compound 3β-acetoxyurs-11-en-30(13α)-olide (3P) is a misassigned compound, and its structure was revised to 3β-acetoxyurs-11en-28(13β)-olide (3R).This is another example in which a misassigned compound was published correctly more than 25 years ago.
The free carboxylic and hydroxyl groups present in 5P are lactonized in 6P and present an inversion at the C-20 stereocenter.The mass spectra of both compounds showed the same peak at m/z 455.When analyzing the IR spectrum of hydroxyacid 5P, we observed the presence of an absorption band at 1740 cm −1 compatible with a lactone group.Both substances, 5P and 6P, show practically identical 13 C NMR spectra (Table 1 and SI: II.7), suggesting that they are the same compound.The chemical shift in their 13 C NMR spectra assigned to C-20 of 83.9 ppm, which is characteristic of carbon that supports the non-carbonyl oxygen of lactone, together with the absorption band at 1740 cm −1 in the IR spectrum of 5P leads to the conclusion that the compound is 6P.To confirm this hypothesis, the 13 C NMR spectra of 5P and 6P were computationally calculated.As seen in Figure 5, the maximum deviation for 5P of −9.9 ppm corresponds to C-20, which supports the oxygenated function, and its rms is 3.5; in contrast, the deviation in C-20 is only −1.3 ppm for 6P.Despite a better approximation between the calculated and experimental data observed (max abs = 6.9 ppm; rms of 2.5), 6P presents important deviations in the carbons around C-18 and C-19.
In the same publication that describes 6P, 22 two other related compounds are described and their structures are also revised: 4P in the present article and (3β,18α,19α)-ursane-3,20,28-triol, which was previously revised. 31Given the  revision of these two compounds, we calculated the 13 C NMR data of the other three stereoisomers of 6P by inverting the C-18 and C19 carbons, the results of which are summarized in Sections III.14 to III.17 in the Supporting Information.
The analysis of these data indicates that 5R (Figure 5) is the stereoisomer with a better fit between the calculated and experimental data and presents an inversion in both the C-18 and C-19 stereocenters with respect to the 6P compound.Consequently, 5P and 6P were revised to 3β-hydroxytaraxast-21-en-28(20β)-olide (5R), which is a new natural product previously not described in the literature.
A study carried out with Kokoona zeylanica Thwaites (Celastraceae) 33 described the isolation of six new friedelanes, which were characterized as 27-hydroxy D:A-friedooleananes (Figure 6, series a).The 13 C NMR data of two of these compounds, kokoonol (7a) and kokondiol (8a), were reported in another paper published by the same research group. 261α,26-Dihydroxy-D:A-friedoolean-3-one (8b) was isolated from Salacia reticulata var.Diandra, which also belongs to the family Celastraceae. 34Its 13 C NMR spectroscopic data are very similar to those of 8a.Subsequently, Giner et al., isolated from Caloncoba glauca 25 several new D:A-friedooleananes with oxygenated functions at C-27, including 7a.In their work, Giner et al. state that the physicochemical properties, such as melting point, optical rotation, and EIMS, of the product they describe as kokoonol (7a) were in close agreement with the published data for kokoonol by Gunatilaka et al. 33,26 Because of the similarity of the spectroscopic data between 8a and 8b, Gunatilaka et al. 35 carried out a thorough study of the NMR data, including 2D experiments and NOE studies on 8a isolated from S. reticulata and concluded that this compound is actually 8b.As a consequence, they revise the structures of the six D:A-friedooleananes previously described as 27-hydroxy D:A-friedooleananes, 26 including kokoonol (7a) and kokondiol (8a), to the corresponding 26-hydroxy D:A-friedooleananes, 7b, and 8b (Figure 6, series b).Both the compound isolated from Kokoona zeylanica, 33 as kokoonol, and the compound isolated from Caloncoba glauca, 25 also known as kokoonol, are referenced in Sci-Finder as 7b with the same CAS number (2183-92-7).Giner et al. did not carry out a comparative study between the 13 C NMR data of 7a published by them and the data reported by Gunatilaka et al. 26 A detailed analysis of the 13 C NMR data of both compounds, which were described under the same name kokoonol (Table 1), enabled us to observe some differences affecting the quaternary carbons C-13 (42.4/45.4ppm) and C-14 (42.5/38.4ppm) bearing C-26 and C-27, in addition to small differences in the neighboring methylenes (Figure 6b).This creates an uncertainty that supports the results obtained both research groups, who describe NOE effects that support the localization of the hydroxyl group at C-26 35 and at C-27 25 for kokoonol.Therefore, we decided to computationally calculate the 13 C NMR spectra of 7a and 7b to determine whether the compounds described by both research groups are the same substance and whether the small differences observed resulted from some impurity, an incorrect interpretation of the spectra, or a mistake in the writing of the paper.
The result of this computational study (Figure 7; SI: III.18, III.19, and III.20) clearly indicates that the substance described by Gunatilaka et al. 33 is 7b and corresponds with the structural revision performed by the same authors in a later publication. 35he substance described by Giner et al. 25 is 7a, as established by these authors in the original publication; however, the compound is different from the one in Sci-Finder under the same name, kokoonol.The 13 C NMR data of the compound described by Ginet et al. 25 show a very good fit with those obtained by computational calculation for 7a (max abs = 2.9; rms = 1.1) and is superior to those obtained for 7b (max abs = 3.8; rms = 1.9) (Figure 7; SI: III.18 and III.19).After considering a large number of misassigned signal swaps according to the computational result (SI: III.20; signs marked with an asterisk), the data obtained for the compound described by Gunatilaka et al. 33 show a closer approximation for 7b (max abs = 2.9; rms = 1.5) than for 7a (max abs = 3.9; rms = 1.7).Therefore, these are two different compounds.The substance correctly described by Giner et al. 25 corresponds to the structure initially erroneously proposed by Gunatilaka et al. 33 as kokoonol, contrary to the information in Sci-Finder, in which both regioisomers are stored with the same CAS number as 26-hydroxy D:A-friedooleanane.In more recent publications, such as those by Somwong et al., 36 citing kokoonol isolated from Salacia verrucosa, the location of the hydroxyl group in this molecule remains ambiguous.
To verify the reliability of our method in calculating the 13 C NMR spectroscopic data for 7a and 7b, which present small differences in 13 C NMR chemical shifts, we carried out a literature search for closely related D:A-friedooleananes to 7a, yielding trichadenic acid B (9a), 37 a D:A-friedooleanane isolated from Phyllanthus f lexuosus (Euphorbiaceae).Its structure was initially misassigned as D:A-friedooleanane 26- carboxylic acid (9b) (Figure 6a). 38Subsequently, its structure was unambiguously revised to D:A-friedooleanane 27-carboxylic acid (9a) by X-ray diffraction analysis (Figure 6a). 37The results obtained by computational calculation faithfully reproduce the experimental 13 C NMR data of compound 9a (SI: III.21).
Two regioisomers, pristimeronol (10a) 39 and salasone A (10b) 40 (Figure 6), are also described in the literature, and the 13 C NMR spectra of these compounds show great similarities.Additionally, computational calculations of the 13 C NMR spectra are in agreement with the calculated and experimental 13 C NMR data for 10a and 10b; these computational results perfectly reproduce the small differences observed in their experimental 13 C NMR data (SI: III.22 and III.23).Consequently, the computational data obtained allow us to correctly assign structures 7a and 7b as well as to validate the structures correctly published as 9a, 10a, and 10b.
In NAPROC-13, we located another pair of triterpenoids, 3β-hydroxyolean-28(19β)-olide (11P) and 28-oxyallobetulin (11R) (Figure 8), that were assigned the opposite configuration at C-18, although their 13 C NMR data were almost identical (Table 1 and SI: II.7).Compound 11P is an oleanane triterpenoid isolated from Diospyros angustifolia (Ebinaceae), 27 whereas 11R is a germanicane triterpenoid obtained from betulinic acid. 28To resolve this inconsistency, computational calculations of both epimers were performed to obtain their theoretical 13 C NMR chemical shifts and compare them with those published for these compounds.As shown in Figure 8, for 11P, the maximum deviation between the calculated and its experimental 13 C NMR chemical shift (max absolute) is 5.1, showing a value of rms = 1.8; for the C-18 epimer 11R, the values of the experimental and calculated chemical shifts present a closer approximation (max absolute 4.3 and rms 1.3) (Figure 8).The maximum deviation of 5.1 ppm in 11P corresponds to C-17 bearing the carbonyl carbon of the lactone, while the largest deviation in 11R corresponds to C-25 methyl, a value that seems to be a misprint considering the value of this chemical shift in other closely related compounds.DP4, with a value of 100%, clearly indicates that the correct structure is 11R.Unfortunately, 1 H NMR data, which could be definitive in clarifying the structure of this pair of epimers, including the H 13 −H 18 coupling constant, are not reported in these articles.However, the structure of 11R is well established by the published X-ray diffraction data for the corresponding acetate, 41 which is identical to the one obtained by acetylating 11R. 28Accordingly, 3β-hydroxyolean-28(19β)-olide (11P) was revised to 28-oxyallobetulin (11R).
In the same paper that described 11P, another triterpenoid, diospyrosooleanolide (12P), a 2α-hydroxy-3β-trans-p-coumaroyl derivative of 11P, was also described.The spectroscopic data of the terpenoid part are very similar to those of compound 11P; hence, the structure of diospyrosooleanolide (12P) was revised to the corresponding C-19 epimer 12R (Figure 9).
To verify whether other compounds structurally related to 11P with the same configuration at C-18 occur in the literature, a Sci-Finder search was carried out.As a result, we found two other oleanane triterpenoids, 13a 42 and 14a 43 (Figure 10), which have, according to this database, the same configuration at C-13 and C-18 as the previously mentioned compounds 11P and 12P.In the articles that describe these compounds, the given structures are not coincident with those found in Sci-Finder, which were 3β-hydroxy-12-oxo-13Hα-   olean-28,19β-olide (13b) 42 and 3β,6β-dihydroxy-12-oxo-13Hα-olean-28,19β-olide (14b) 43 (Figure 10).These two compounds present an inverse trans-diaxial arrangement for H-13/18 in relation to those of 11R and 12R.Due to the discrepancy between these publications and the Sci-Finder database and because the C-13 configuration is inverted in these substances compared to the oleananes, we decided to perform 13 C NMR spectra computational calculations on 13a/ b and 14a/b to confirm their structures.Based on the results (SI: III.26−III.29), the above-mentioned publications described the correct representation, and, consequently, the structures reported in Sci-Finder are erroneous.The max absolute and rms terms for 13b (1.6 and 0.8, respectively) (SI: III.27) versus those obtained for 13a (2.0 and 5.8, respectively; SI: III.26) clearly indicate that 13b is the correct structure, which was confirmed by X-ray diffraction. 42For the calculation of the maximum absolute and rms values of 13a and 13b, the chemical shift of δ 39.4 assigned to C-24 was excluded, as it must be a misprint.This value deviated by more than 20 ppm from the value obtained by calculation and the value shown for other compounds with a similar environment published in the same article as 13b.Similar results were obtained when comparing the statistical results obtained for 14b versus 14a since 14b showed a max ab of 1.7 and an rms term of 0.9 versus the same values for 14a (max abs = 5.6, rms = 2.0) after swapping several chemical shifts, which significantly improved the adjustment (see SI: III.28 and III.29).

■ CONCLUSIONS
Searching chemical shifts for suspected misassigned compounds in NAPROC-13 can provide structurally different compounds that share the same 13 C NMR spectroscopic data.Subsequent calculation of the 13 C NMR spectra indicated the compound that most closely matched the experimental data in each case.It is relatively common that nowadays misassigned structures were correctly published in the past.In addition, if one of the structures found has been verified by X-ray diffraction, remaining misassigned structures can be checked very quickly.The examples herein provided reinforce the value of the method applied to correct natural product structures.

■ EXPERIMENTAL SECTION
Computational calculations were performed with Spartan'20 (Wavefunction Inc., Irvine, CA, USA).To obtain the 13 C NMR data by computational calculation, the automated protocol implemented in Spartan'20 was followed. 12This protocol consists of five steps: (1) Systematic conformational search using MMFF molecular mechanics, eliminating duplicate conformers and those with energy 40 kJ/mol above the global minimum; (2) geometric calculation using HF/3-21G, also eliminating duplicate conformers and those with energy higher than 40 kJ/mol above the global minimum; (3) energy calculation with the ωB97X-D/6-31G* model and removal of conformers above 15 kJ/mol with respect to the global minimum; (4) geometric calculation with the ωB97X-D/6-31G* model and removal of conformers with energies higher than 10 kJ/mol from that of the global minimum; (5) energy calculation with the ωB97X-V/6-311+G(2df,2p)[6-311G*]; and finally (6) the NMR calculations (following calculation of Boltzmann weights for conformationally flexible molecules) using the ωB97X-D/6-31G* method that has been corrected empirically based on the comparison of calculated and experimental 13 C shifts for ∼2000 rigid molecules.These corrections are on the order of 1−3 ppm.
■ ASSOCIATED CONTENT * sı Supporting Information

Figure 6 .
Figure 6.(a) C-26-or C-27-Functionalized friedelanes.(b) Experimental values of the 13 C NMR chemical shifts of the C and D rings of kokoonol with a hydroxyl group at C-27 (7a) or C-26 (7b).

Figure 7 .
Figure 7. Structures of kokoonol [26-OH] (7b) and kokoonol [27-OH] (7a).The 13 C NMR chemical shifts assumed for the two compounds are identical and correspond to those published by Giner et al. 25 Negative and positive deviations of δ C (calculated value − experimental value) are indicated for each carbon on the structures.Max y rms values are underneath each structure.

Figure 8 .
Figure 8. Structures of misassigned 3β-hydroxyolean-28(19β)-olide 11P and the correct structure of 28-oxyallobetulin 11R.Negative and positive deviations of δ C (calculated value − experimental value) are indicated for each carbon in the structures.Max y rms values are underneath each structure.

Figure 9 .
Figure 9. Structure of misassigned compound 12P and that of revised 12R.This review was performed by considering the spectroscopic data from 11P.

Table 1 .
13 C NMR Data for Compounds Revised 1−11 Recorded in CDCl