A critical review on the use of DP4+ in the structural elucidation of natural products: the good, the bad and the ugly. A practical guide

Contents

• 1 Introduction

• 2 The good
  • 2.1 Simplicity
  • 2.2 Structural diversity
  • 2.3 Isomerism variety
  • 2.4 Variable number of candidate structures
  • 2.5 Flexibility in the level of theory
  • 2.6 Flexibility in the NMR data employed

• 3 The bad
  • 3.1 Energy miscalculations
  • 3.2 Similar chemical shifts
  • 3.3 Neglecting the real structure

• 4 The ugly
  • 4.1 Use of inappropriate distributions
  • 4.2 Incorrect chemical shifts calculations
  • 4.3 Large outliers
  • 4.4 General recommendations

• 5 Conclusion

• 6 Conflicts of interest

• Acknowledgements

• References

---

Abstract

Covering: 2015 up to the end of 2020

Even in the golden age of NMR, the number of natural products being incorrectly assigned is becoming larger every day. The use of quantum NMR calculations coupled with sophisticated data analysis provides ideal complementary tools to facilitate the elucidation process in challenging cases. Among the current computational methodologies to perform this task, the DP4+ probability is a popular and widely used method. This updated version of Goodman's DP4 synergistically combines NMR calculations at higher levels of theory with the Bayesian analysis of both scaled and unscaled data. Since its publication in late 2015, the use of DP4+ to solve controversial natural products has substantially grown, with several predictions being confirmed by total synthesis. To date, the structures of more than 200 natural products were determined with the aid of DP4+. However, all that glitters is not gold. Besides its intrinsic limitations, on many occasions it has been improperly used with potentially important consequences on the quality of the assignment. Herein we present a critical revision on how the scientific community has been using DP4+, exploring the strengths of the method and how to obtain optimal results from it. We also analyze the weaknesses of DP4+, and the paths to by-pass them to maximize the confidence in the structural elucidation.

---

Maribel O. Marcarino

Maribel Marcarino received her B.S. in chemistry from the National University of Rosario (UNR, Argentina) in 2019. That same year she joined Dr Sarotti's group at IQUIR-UNR as a PhD student, focusing on the development of new computational methods based on NMR calculations.

Soledad Cicetti

Soledad Cicetti received her B.S. in chemistry from the National University of Rosario (UNR, Argentina) in 2019. In 2020, she started her PhD studies at IQUIR-UNR in the Sarotti group, working in the ecofriendly synthesis of bioactive compounds and DFT calculations.

María M. Zanardi

Maria Marta Zanardi received her PhD from the National University of Rosario (UNR) in 2011 under the supervision of Prof. Alejandra G. Suárez. She was a postdoctoral researcher at the IQUIR-UNR with Dr Ariel Sarotti. Currently she is a researcher of CONICET at the INGEBIO-(UCA). Her research interests include development and optimization of new tools for structural elucidation based on NMR calculation, and the use of artificial intelligence for organic and medicinal chemistry research.

Ariel M. Sarotti

Ariel Sarotti received his PhD from the National University of Rosario (UNR, Argentina) in 2007. After postdoctoral training in computational chemistry, he became a researcher of the Argentine National Research Council (CONICET). Currently, he is an Associate Professor of Organic Chemistry at UNR and an Independent Researcher at CONICET. His scientific interests include green chemistry, asymmetric synthesis, computational chemistry, NMR spectroscopy, and medicinal chemistry.

---

1 Introduction

The exhaustive description of the molecular architecture of novel compounds, including connectivity, relative and absolute configurations, is of fundamental importance in the discovery of biologically active molecules, as their chemical and biological properties are strongly linked to their 3D structures. Several methods are currently used for structural elucidation, with X-ray crystallography analysis being the most unquestionable technique, though the need to generate diffraction quality crystals limits its scope. In contrast, due to its universality and effectiveness to unravel the structural mysteries of a vast range of organic molecules, NMR spectroscopy has become the leading methodology in the field. Over the last years the advances made in NMR have been noteworthy. Nevertheless, data misinterpretation is not uncommon, leading to a large number of erroneous structures published in the last decades.^1–5

Gauge-Including Atomic Orbitals (GIAO) NMR calculations at DFT levels emerge as an excellent complement to structural elucidation.^6–11 The discipline has experienced a tremendous growth over the last years, and nowadays the results of those calculations are often found as part of the routine in the structural elucidation of natural products. As a general trend, the procedure to determine the most likely structure among several candidates involves the following steps: (1) conformational search at a molecular mechanics level; (2) geometry optimization (in case DFT optimized structures are required); (3) NMR calculations (chemical shifts and/or J couplings); (4) energy calculations (can be done at the same or different levels employed in the previous steps); (5) calculation of the Boltzmann-averaged NMR chemical shifts and/or J couplings; (6) correlation of the calculated data with the experimental values. During the first decade of the XXI century, the agreement between calculated and experimental NMR data was determined with the aid of simple statistical descriptors, such as R², mean absolute error (MAE) or corrected mean absolute error (CMAE).¹⁰ The introduction of CP3 ¹² and DP4,¹³ both from the Goodman group, catalyzed the emergence of a new series of sophisticated approaches increasing in confidence. Among the Bayesian probability methods that were inspired by DP4 which are worth mentioning, we find DP4.2 ¹⁴ and DP4.AI ¹⁵ (Goodman group), DP4+ ¹⁶ (Sarotti group), J-DP4 ¹⁷ (Sarotti and Hernández Daranas groups), and DICE¹⁸ (Gonnella group). Among the non-probabilistic approaches, CASE-3D by Gil and Navarro-Vázquez¹⁹ and DU8+ by Kutateladze²⁰ stand out. While CASE-3D merges isotropic and anisotropic NMR measurements with DFT NMR calculations and conformational selection, DU8+ is based on fast NMR calculations obtained at low-cost DFT methods coupled with NBO-corrected J calculations.

Over the years, the scientific community has embraced computational NMR methods as an alternative and convenient strategy for the structural elucidation. The DP4+ probability is among the most popular and widely used methods today, due to its ease of use and the overall satisfactory and reliable performance. Its usefulness in structural elucidation is evidenced by the large amount of structures of natural products assigned with the aid of DP4+, and the trend suggests an increase of its use in the near future. This highlights the contribution of DP4+ to settle structural issues, preventing misassignments when experimental data is not conclusive, or providing additional support to newly assigned structures. However, all that glitters is not gold. Besides the intrinsic limitations of the method, we noticed that DP4+ was used improperly in many cases, which might affect the quality of the results. Considering the rising popularity of DP4+, we decided to make a critical review of the strengths and weaknesses of it, including a thorough analysis of how it has been applied. This review covers all the papers that cited DP4+ (source: Scopus) up to December 2020, and also includes a final section with descriptions and general recommendations to run DP4+ calculations.

2 The good

Since its introduction in late 2015, DP4+ has stood out as one of the leading toolboxes in structural elucidation with computational NMR methods and the structures of more than 200 natural and synthetic products were puzzled out with its aid.^21–217 This section is devoted to a thorough analysis of those studies highlighting the strengths and advantages of DP4+ during the elucidation stage.

Briefly, the DP4+ ¹⁶ probability is an improved version of DP4 ¹³ in which P(i) is the probability of candidate i (out of m isomers) to be correct, obtained through Bayes's theorem. It is based on the fact that the errors e between experimental, δ_exp, and calculated chemical shifts, δ_calc, (e = δ_calc − δ_exp) for a set of organic molecules obey a t distribution defined by three terms: μ (mean), σ (standard deviation) and ν (degrees of freedom). The original DP4 distributions have μ = 0 due to the scaling procedure to remove systematic errors, which is done according to δ_s = (δ_calc − b)/m, where b and m are the y-intercept and slope of a plot of δ_calc against δ_exp, respectively. Hence, the DP4 method is built with two sets of [σ, ν] values, corresponding to the errors in the carbon and proton series, respectively. However, in DP4+ we also included the errors due to unscaled chemical shifts (δ_calc) as we hypothesized that this would improve the stereochemical differentiation among closely related structures. Given the lack of scaling, the unscaled series are no longer centered in zero (μ ≠ 0). In addition, the data should be separated in terms of hybridization to furnish t distributed series (Fig. 1). Hence, six sets of [μ, σ, ν] parameters are needed to build DP4+ (one for scaled data, one for unscaled sp–sp² data, and one for unscaled sp³ data, both for carbon and proton chemical shifts). The corresponding sixteen parameters (note that the two scaled series have μ = 0) show dependence on the level of theory. In the original publication these were estimated at 24 different levels (combining B3LYP and mPW1PW91 functionals with six Pople's basis sets and two approaches to consider the solvent effect, using B3LYP/6-31G* optimized geometries in all cases). Recently, we reported a new customizable method to be employed with any desired level of theory (vide infra).²¹⁸ The use of higher levels of theory for the NMR calculation step resulted in another important improvement over the original DP4 protocol (which is based on chemical shifts computed at the fast but far than optimal B3LYP/6-31G**//MMFF level). As shown in Fig. 1, DP4+ was constructed as a function of the corresponding probabilities computed from scaled and unscaled chemical shifts, termed sDP4+ and uDP4+, respectively. Moreover, each DP4+ term can be calculated using only ¹H data, ¹³C data, or both. The great performance improvement due to the introduction of these additional parameters and upgrading the levels of theory is noteworthy. Other advantages of the method are:

Fig. 1 (Up) Error distribution plot of unscaled ¹³C chemical shifts. (Down) General equation of DP4+, in which P(i) gives the probability that candidate structure i (out of m possible candidates) is the correct one. T^ν is the cumulative T distribution function with ν degrees of freedom, whereas μ and σ are the center and standard deviation, respectively, of the error series.

2.1 Simplicity

In our opinion, besides its good overall performance (vide infra), the simplicity of conducting DP4+ calculations is surely one of the main reasons of its popularity. This can be easily done with a ready-to-use excel spreadsheet included by the authors as part of the ESI of the original paper,¹⁶ or also available at sarotti-nmr.weebly.com. Following the simple rules provided in the tutorial file, the DP4+ probabilities can be computed at 24 different levels of theory (original version)¹⁶ or any desired level of theory (updated version).²¹⁸ In the last case, the proper estimation of the [μ, ν, σ] terms should be carried out at the desired level. Despite DP4+ can be implemented in Python, C++, Matlab, or any other related platforms without difficulty, its distribution in excel format has certainly contributed to spreading the method among chemists or spectroscopists who are not experts in programming language.

2.2 Structural diversity

After a deep literature survey, we observed that the structures analyzed with DP4+ displayed a wide variety of arrangements, featuring diversity of shapes, functional groups and conformational freedom. The overall performance of DP4+ tends to be very good in general, even for challenging structural motifs (such as epoxides and spiroepoxides).²¹⁹ The range of molecular sizes goes from small structures with a few atoms (less than 20)¹⁰³ up to molecules with more than a hundred.^141,169 The average size of the studied systems was around 60 atoms, with a variable ratio of C, H, O, N as the prevalent atoms. We also found several cases of molecules containing other elements, such as S,^{29,32,55,92,140,164,212} P,²¹³ Cl,^{21,26,28,125,147} and Br.^29,212 In those cases of carbons attached to bromine (or other atom of the third row or greater) the errors are larger than the average due to the well-known heavy atom effect, consequence of neglecting the spin–orbit contributions from relativistic effects and minor contributions from electron correlation effects.¹¹ In those cases, it would be recommended to exclude those carbons of the DP4+ calculations in order to increase the confidence of the assignment.

About 70% of the compounds assigned in silico were further validated by additional studies such as more complex spectroscopic analysis, X-ray crystallography, total synthesis and electronic circular dichroism (ECD) among others, (for instance compounds 1–12, Fig. 2).^{32,46,57,78,82,103,126,135,138,149,151,173}

Fig. 2 Natural products assigned by DP4+ empirically validated by X-ray crystallography, circular dichroism and total synthesis. The number of compared candidates in each study is between parenthesis.

DP4+ has been mainly employed in the assignment of natural products, but also it has stood out as a valuable tool in the structural elucidation of synthetic compounds (Fig. 3), including reaction intermediates (as compound 13 during Richnovsky's synthesis of kujonins A1 and A2)⁸⁵ and unexpected impurities (as the case of the demethylated intermediate of (+)-frondosin B which complicated the configuration assignment of the natural product).¹⁰⁵ In this case, the impurity of (+)-frondosin B was assigned as (R,R)-14 based on DP4+ calculations. Moreover, despite DP4+ is typically applied to evaluate stable molecules in their ground state, it can provide further insight into the identification of reaction intermediates. For instance, during the mechanistic study of the [Bi(OTf)₃] mediated mild electrophilic aromatic formylation with PhSCF₂H, three possible intermediates were studied (two conformational isomers 15a and 15b, and their geometric isomer 15c).⁵⁵

Fig. 3 Examples of non-traditional applications of DP4+.

2.3 Isomerism variety

Because of the comparison-based nature of DP4+, it can be used to discriminate among different types of isomerism, including diastereo-, regio- and constitutional isomers. In addition, it can handle both single and combined uncertainty.

2.3.1 Single uncertainty. Arises when only one type of isomerism is put into play. In the most typical example, the connectivity and regiochemistry of the molecule are irrefutably settled, but the relative configuration is uncertain. About 90% of the reported cases using DP4+ consists of this type of configurational analysis. The identification of relative configuration (such as compounds 1–14) has become particularly trendy, surely because of the deep-rooted difficulties associated with this task. Naturally, other types of isomerism uncertainty can be tackled for both chiral and achiral molecules (Fig. 4). For example, DP4+ allowed to establish the substitution pattern of the aromatic frames in compounds 18–21.^{29,104,111,214} Another interesting study was conducted with compound 16 as case study, in which the DP4+ calculations were done to examine the C1–C2′ and the C1–C3′ linkage in two regioisomers, further supporting the connection of the xanthone to the ribose via C1–C2′.¹⁴⁶ The elucidation of compounds 17,²¹²22,¹⁷² and 23 ⁴⁷ represent additional examples that outline the usefulness of DP4+ to establish molecular connectivity.

Fig. 4 DP4+ in the constitutional assignment for chiral and achiral compounds.

2.3.2 Combined uncertainty. When the connectivity or regiochemistry of chiral molecules is unsure, a lack of certainty in the configurational analysis might be expected as well (e.g. compounds 24–28, Fig. 5).^{40,128,169,215,216} For instance, Lee and co-workers were able to determine most of the structure of the diterpenoid 24,²¹⁵ but the experimental NMR data was inconclusive to establish the oxygenated motif (C₁₂–C₁₃–C₁₅). Hence, NMR calculations were carried out for ten possible structures, including three constitutional isomers with their corresponding diastereoisomers. Since the results indicated that the main difference laid upon C-15, the authors grouped δ_C-15 of the calculated compounds in order to establish a ranking for each constitutional isomer, and by direct comparison with the experimental values, the structure of the oxetane ring was settled. DP4+ analysis was then applied to the four resulting diastereoisomers, pointing with high confidence (>99.9%) towards the structure 24aa with an α-oriented cis-fused ring (Fig. 5a).

Fig. 5 DP4+ to tackle combined uncertainty problems.

Similarly, we found examples that simultaneously explore geometric and configurational isomers, such as the case of dictyospiromide (25), a diterpenoid with a novel structural scaffold isolated from a marine brown algae (Fig. 5b).²¹⁶ Its constitution and configuration were proposed after a thorough analysis of NMR and NOESY data, but further studies were needed to provide the full structure of the molecule. GIAO NMR calculations were carried out for four isomers arising from the difference in the configuration at C-2 (R or S) and the exocyclic alkene at C-1 (E or Z). On the basis of RMSD and MAE, isomer 1E,2R was identified as the most probable one, mainly on the basis of carbon data. However, the calculated values showed close similarity to those of the C-2 epimer (1E,2S). The DP4+ calculations revealed that the proton-based probability was ambiguous, but the characteristic compensation of the errors associated to DP4+ allowed the correct assignment by introducing the carbon data. Anisotropic NMR experiments validated the putative structure.

2.3.3 Separated stereoclusters. Current computational methodologies have been proved to be useful to differentiate among candidates bearing rigid structures and contiguous or near-by stereocenters.^10,13,19,20 However, the task becomes much more challenging when dealing with molecules featuring separated stereoclusters. Connecting the relative stereochemistry of non-interacting fragments is often difficult with standard NMR techniques, mainly when the molecule features flexibility. The effectiveness of quantum-based NMR methods to tackle separated stereoclusters has been previously covered.^{217,220–222}

Approximately 20% of the molecules assigned with the aid of DP4+ included stereoclusters separated through flexible spacers (such as ethylenes, non-stereogenic quaternary carbons, alkenes, heteroatoms, etc.). Some examples of this type of arrangement are present in compounds 29–36 (Fig. 6),^{79,88,89,92,154,164,168,217} and although we demonstrated that DP4+ can be applied to these types of compounds, the challenge of assessing the relative configuration becomes much more complicated than in other cases.¹⁶ It is fundamental to make a correct description of the chemical environment in order to avoid a wrong assignment, considering that some of the diastereoisomers might exhibit similar NMR chemical shifts.²²³ That is the reason for the keen remark in the inclusion of all the data available, especially during the study of this type of systems, to guarantee a confident result when using DP4+.

Fig. 6 Example of molecules containing separated stereoclusters assigned by DP4+.

The good capacity of DP4+ to connect the relative configuration of separated stereoclusters was exploited in a conceptually novel method to establish absolute configuration (AC) of organic molecules, and it was focused on chiral derivatization agents (CDAs) such as Mosher's reagent or its analogues (Fig. 7).²²⁰ Using an ambitious and large set of examples (114 systems) the absolute configuration of 96% of the cases could be determined using only a single derivatization experiment. The best results were achieved for secondary alcohols, along with secondary and primary amines as well. Primary and tertiary alcohols yielded more modest but still exciting predictions. A new DP4+ Integrated Probability (DIP) was then introduced to combine two independent DP4+ predictions into a single descriptor. This new probability proved to be useful in double derivatization approaches with remarkable results.

Fig. 7 Structural diversity of the compounds assigned by chiral derivatization and DP4+ analysis.

2.4 Variable number of candidate structures

The number of candidates is not a limitation for the method, though the typical applications found in the literature involves the comparison between few isomers. In fact, 70% of the reported DP4+ calculations were carried out considering only two candidates, 25% with four isomers, and the rest comparing more than eight structures. The highest number of isomers studied using DP4+ was 128 in the structure elucidation of pseudorubriflordilactone B (30).⁸⁹ Only considering a reduced number of isomers is beneficial regarding computational cost, but also because it minimizes the possibility of incorrect isomers fortuitously showing good fits with the experimental data, hence affecting the assignment. This reinforces the importance of conducting a careful analysis of the experimental data, including chemical shifts, J coupling constants and/or NOE interactions, in order to rule out those candidates a priori for further NMR calculations and DP4+ analysis.¹⁷

2.5 Flexibility in the level of theory

In order to provide flexibility when choosing the method to perform the NMR calculations, DP4+ was developed at 24 different levels of theory. These levels were generated combining two functionals (B3LYP and mPW1PW91), six basis sets (6-31G*, 6-31G**, 6-31+G**, 6-311G*, 6-311G**, and 6-311+G**) and two approaches to include the solvent effect (gas phase and PCM) for the GIAO NMR calculations, starting from B3LYP/6-31G* optimized geometries in all cases. Naturally, not all levels provide the same quality in the assignments, which is the reason we recommended PCM/mPW1PW91/6-31+G**//B3LYP/6-31G* for general applications.^16,219 We noticed that in about 30% of the publications this recommended level was used,^{40,55,78–117,214} whereas in 34% of the cases the authors decided to evaluate one of the other 23 levels available for DP4+.^{24,25,28,32,39,41,43,146–148,150–155,224} Noteworthy, in the remaining 36% papers the analysis was conducted using NMR shifts calculated at other levels for which DP4+ was not parameterized. This indicates a clear inclination of each research group towards conducting the geometry optimizations and GIAO calculations with pre-selected methods well-known to them. In this regard, it should be emphasized that using alternative levels does not represent a mistake by itself, but caution should be taken as the accuracy of the predictions may decrease, mainly when correlating NMR chemical shifts with [μ, σ, ν] values computed at different levels (vide infra). Recently we published the study exploring the sensitivity of the DP4+ method with the probability distribution terms. The results led us to develop a customizable DP4+ methodology, which allows calculations at any desired level of theory. The [μ, σ] terms can be fairly estimated from a small set of 8 rigid molecules for preliminary explorations. However, if more accurate DP4+ results are required, the [μ, σ, ν] parameters should be obtained from a more exhaustive number of molecules, such as those employed in the original publication.²¹⁸

2.6 Flexibility in the NMR data employed

2.6.1 Full or partial data. The successful performance of the DP4+ probability, compared with the original DP4, is the result of a constructive overall compensation of errors upon using both scaled and unscaled proton and carbon data, so including all types of data is beneficial for increasing the confidence in the assignment.^16,219,220 However, DP4+ analysis can be conducted using only partial data, which is the case in several publications.

2.6.2 Scaled and unscaled data. DP4+ synergistically combines the probabilities associated with scaled and unscaled chemical shifts, in such a way that a failure of sDP4+ is compensated by uDP4+, or vice versa, leading to satisfactory overall assignments.^16,218,219 In approximately 87% of the publications both sets of data were used, whereas in the remaining 13% only scaled data was employed.^{21–38,216,224} In some of these works, the exclusive use of scaled shifts was probably because the NMR shifts were carried out using levels of theory for which DP4+ was not parameterized. Since uDP4+ is more sensitive to the [μ, σ, ν] parameter set, it seems reasonable to consider only the sDP4+ term when the corresponding [μ, σ, ν] values are unknown at the selected level.²¹⁸

2.6.3 Proton and carbon data. A similar situation occurs when analysing the effect of using proton and carbon data in the DP4+ results. Even though reports suggest that proton data is more discriminating in terms of stereoassignment,^225,226 the overall DP4+ scores obtained with only proton or carbon data were equivalent.¹⁶ In addition, it was noticeable that the combination of both types of data were beneficial for the quality of the predictions. Here again, a synergistic compensation of errors often allows a positive compensation in mismatched cases. While a good assignment made by ¹³C often overrides an eventual bad assignment made by ¹H, or vice versa, a situation where a failure of one nucleus overcomes the success of the other is much less frequent.^16,222 The recommendation of using both types of data was taken into account in about 70% of the cases. In the remaining publications, the DP4+ calculations were conducted using only carbon data (28%)^{22–27,29–31,33–37,39–77,227} and to a much lesser extent, proton data (2%).^{96,132,228,229} Occasionally, a low quality ¹H NMR spectrum might complicate the analysis due to signal overlapping, and it might be tempting to fully exclude all the data in that spectrum. But we consider that, in those cases, including few diagnostic and well resolved ¹H NMR signals is more recommended than not using any signal at all. Albeit using only one source of data does not constitute a mistake by itself, it is clear that it reduces the confidence of the assignment. Hence, whenever possible, the DP4+ calculations should be carried out with as much experimental information as possible. In this regard, the possibility of including other nuclei to the DP4+ architecture would provide better classification capacity with molecules featuring those atoms. In DICE, a related Bayesian approach, Gonnella and co-workers demonstrated that the incorporation of ¹⁵N chemical shifts resulted in a superior performance of the method.¹⁸ We are currently working on an updated DP4+ version that allows adding chemical shifts of other nuclei of relevance in natural products (such as nitrogen and phosphorous), and will be published in due course.

3 The bad

The main limitation of any comparison-based method arises when an incorrect isomer shows a fortuitously better agreement with the experimental data than the right candidate does, and DP4+ is not the exception. Indeed, in the original publication a few molecular systems could not be properly solved by DP4+ (for example, compounds 44–45, Fig. 8),¹⁶ even at the recommended level of theory (PCM/mPW1PW91/6-31+G**). In subsequent publications, other unsuccessful examples were provided (for example, compounds 46–49, Fig. 8).^220,230,231 The extensive use of DP4+ puts the method constantly under scrutiny, and new cases with inconclusive results arise continuously. This highlights the fact that all current methods may show flaws in some certain scenarios. Understanding the reasons behind such failures, even after proper computational work, provides an excellent opportunity to develop improved methodologies. In this section, we discuss the background that might lead to potential misassignments and the alternatives to overcome those problems.

Fig. 8 Compounds incorrectly assigned by DP4+.

3.1 Energy miscalculations

Its favorable cost/accuracy ratio makes DFT the preferred alternative for quantum-based NMR calculations. However, this approach has intrinsic simplifications, which affect the quality of the predictions (including the Born–Oppenheimer approximation, the still imperfect nature of the exchange–correlation terms of all functionals, the use of implicit solvation models and the alternatives to tackle the gauge origin problem when using truncated basis sets).¹¹ Hence, error-free calculations should not be expected, not even for the simplest molecules using state-of-the-art computational procedures. Although there are many factors potentially influencing the accuracy of the NMR predictions, they can be summarized in two main categories: tensorial and energetic. The first one is related to the inability to perfectly reproduce the experimental isotropic shielding constants, whereas the second is associated with the errors involved in the energy simulations (only important when dealing with conformationally flexible systems). It would be reasonable to guess that the tensorial source of error would be the most decisive factor, but in our opinion, the energy term could have a greater influence when discriminating among flexible diastereoisomers. This is evidenced by the often excellent results obtained with most methods when applied to rigid molecules with a very limited or null conformational freedom.^13,16,19,20 On the contrary, it has been well documented that erroneous energy calculations might affect the Boltzmann-averaging, leading to low quality NMR predictions, with a potentially negative effect in the DP4/DP4+ results.^13,16

We recently showed that under a low energetic uncertainty (<1 kcal mol⁻¹), DP4+ values are not affected significantly.²³¹ On the other hand, larger errors in the energy calculations might lead to substantial discrepancies between the computed and actual conformational landscape. This becomes manifest in flexible molecules properly functionalized to give rise to intramolecular hydrogen bonding interactions (IHB), mainly when the experimental NMR spectra are collected in polar solvents, such as D₂O or CD₃OD. The IHB is a long-standing and well-known problem for those in the field, and it can lead to wrong DP4+ results when the most contributing conformations of the right isomer are spurious as a consequence of such interactions. We've thoroughly studied the impact of IHB in the stereoassignment of 40 known compounds of the hyacinthacine family, polyhydroxylated pyrrolizidine alkaloids featuring a rich conformational freedom with the possibility of building multiple IHB arrangements.^230,232 Following the standard DP4+ procedure at the recommended level of theory (PCM/mPW1PW91/6-31+G**//B3LYP/6-31G*), we found that 11 isomers were incorrectly assigned as consequence of improper conformational descriptions for the corresponding right candidate, highlighting the importance of IHB to bias the conformational amplitudes. Naturally, not all IHB lead to wrong results, as most of the leading conformations of the 29 compounds correctly assigned also displayed such interactions.

Therefore, when dealing with systems featuring IHB, the main goal is to define if those interactions are harmless or not in terms of DP4+ results, which is obviously impossible without knowing a priori which candidate is the correct isomer of the molecule. There are, however, additional strategies that can be followed to anticipate a possible misassignment as a consequence of IHB. Perhaps the easiest way is to recompute the DP4+ values after removing those featuring suspicious IHB interactions. In a preliminary study on the simplest members of the hyacinthacine family, we proved that these counterintuitive approaches allowed to improve the DP4+ results not only in the isomers where the standard procedure failed, but also keeping unchanged the DP4+ trends of the remaining five examples being correctly assigned.²³² In all those cases, removing the biased IHB shapes affected positively the quality of the NMR predictions by lowering the MAE (mean absolute error) values, clearly indicating that the predominance of those structures in the conformational landscape pictured by DFT is wrong. This was evidenced during the assignment of sphaerialactonam (50)⁹¹ and peyronetide A (51) (Fig. 9),¹⁰⁰ as we noticed large outliers in the calculated ¹H and ¹³C NMR resonances, respectively. In both cases, the conformations with largest amplitudes showed IHB interactions (either between C6–OH and the carbonyl groups at C5′ or C20′, or between C8–OH and ketone oxygen at C2′, respectively). The recomputed NMR shifts after neglecting those conformations significantly improved the match between experimental and calculated values. Although this modification did not change the sense of the assignment (the most likely candidate remained after conformational removal), this procedure allowed to rule out the possibility that DP4+ results could have been biased by a wrong conformational description.

Fig. 9 Representative conformations of sphaerialactonam and peyronetide A with spurious IHB responsible for large errors in the DFT NMR calculations.

The removal of conflicting conformations is an easy shortcut to potentially improve the NMR results and to support the assignment made by the standard formalism. There are, however, other alternatives to analyze the impact of spurious conformations in the DP4+ results. One of them is refining the Boltzmann amplitudes using SMD as the solvation model, a version of PCM that decomposes the solvation energy into SCRF bulk electrostatic contributions and the short-range solvent–solute interactions in the first solvation shell.²³³ In a thorough exploration of the conformational description of the hyacinthacine family, we observed sharped differences between the conformational amplitudes computed with PCM (the recommended for DP4+ for broad applications) and SMD, with the later providing a more realistic description in certain IHB systems.²³⁰ Another and more drastic approach involves neglecting all together the relative energies given by DFT calculations, and alternatively creating and evaluating different ensembles generated by removing conformations followed by a random relative energy distribution of the remaining shapes. This is supported by the idea that in a large set of ensembles, the majority of them would point towards the right isomer and reflect the correct final assignment. The averaged overall performance of this approach was excellent with 100% of the compounds belonging to the hycintacine family being correctly classified by DP4/DP4+.²³⁰ In comparison with other methods previously developed, this new and exciting approach is very different in its nature. The random ensemble strategy is based on the fact that one should not rely on a single determination to decide whether a putative structure is correct or not (Fig. 10).

Fig. 10 Random strategies employed to avoid the IHB problem.

The inadequate estimation of the conformer populations could be also tackled using advanced hybrid or double-hybrid functionals for the energy calculations, which were shown to provide superior performance than B3LYP.²³⁴ For example, in DP4.2 Goodman and co-workers found that better results were obtained by calculating the NMR chemical shifts at the mPW1PW91/6-311G* level and the relative energies for Boltzmann analysis at the M06-2X/6-31G** level.¹⁴ In DP4+ the estimation of Boltzmann amplitudes using new levels can be done in different ways. One of them, previously discussed, involves correlating the NMR chemical shifts computed at one of the 24 levels of theory (for example, PCM/mPW1PW91/6-31+G**//B3LYP/6-31G*) with the Boltzmann amplitudes refined at a new level (for example, SMD/M06-2X/6-31+G**).²³⁰ The other option is based on running the NMR calculations and/or the geometry optimization steps at new levels, now allowed in the updated version of DP4+. However, it is important to point out that a proper estimation of the [μ, ν, σ] terms should be carried out first to obtain meaningful results.²¹⁸

In any case, regardless the level of theory used for the geometry and energy calculations; it would be always wise to incorporate as much experimental information as possible to corroborate the DP4+ findings. For example, whenever available, the homo- and heteronuclear ³J coupling constants and interatomic distances (estimated from NOE/ROE correlations) should be used to check consistency of the DP4+ analysis. In case of improper conformational description, the parallel application of these data would allow to detect the failure, hence unleashing a more detailed study (for example, refining the Boltzmann amplitudes with new levels).

3.2 Similar chemical shifts

The DP4+ results are usually robust and correct when the isomers under consideration show acceptable differences in at least a few chemical shifts, as is often the case. However, when two or more isomers show differences in their chemical shifts below the precision limit of the method, the results might become occasionally erratic. In this regard, it is only enough that the incorrect isomer shows a slightly better fit than the correct one to influence the DP4+ results. This could be the situation when dealing with isomers with separated stereoclusters featuring the same relative configuration in some of them but opposite configurations in others. For example, during the first total synthesis and structural elucidation of (+)-cryptoconcatone H⁹³ (Fig. 11) by Pilli and co-workers, DP4+ calculations were carried out over the eight possible diastereoisomeric candidates to validate the putative structure before facing the synthetic effort. The calculations strongly suggested that the putative structure (52a) was erroneous, whereas the two candidates featuring the cis/trans configuration at the tetrahydropyran ring were the most likely structures of the natural product (52b and 52c). The calculated NMR chemical shifts of these two isomers displayed excellent agreement with the experimental data reported for the natural product, with CMAE values of 1.0 and 1.3 ppm (carbon data) and 0.08 and 0.10 ppm (proton data), respectively. The better match observed with 52b, which impacted in the DP4+ probability values (99.9% vs. 0.1%), was mainly due to small errors computed for the conserved regions of the molecule. However, the real structure of the natural product was determined as 52c after total synthesis of the two candidates. Although recomputing the DP4+ by taking into account only the NMR data from the most relevant and differentiating region (in this case, the tetrahydropyran fragment) reduced the preference towards 52b, and this result revealed the potential problem arising whenever two isomers display similar NMR properties. This case study was subsequently analyzed in detail, but we could not find any evident proof of a bad representation of the conformational landscape that might have negatively influenced the NMR predictions of 52c. We concluded that the problem was not related to a poor prediction of the conformational landscape of the correct structure, but merely to a better match with a similar, yet different, shape. In later publications, the potential problems of setting the conformational amplitudes by fitting instead of relying on Boltzmann analysis using the corresponding relative energies computed at DFT levels were also discussed.^231,235 Forcing all isomers to have computed chemical shifts as close as possible to the experimental values jeopardizes the assignment as some incorrect isomer could end up having a better agreement, and therefore higher DP4+ values, than the correct structure.

Fig. 11 Putative structure of (+)-cryptoconcatone H, and most likely structures by DP4+ calculations.

3.3 Neglecting the real structure

Another inherent problem associated with most computational methods for structural elucidation is related to the inability to unequivocally assess the correctness of a given structure.^236,237 Instead, the most probable structure is selected among several candidates that are ranked depending on the degree of fit with the experimental values. Hence, if the real structure of the natural product is not included in the set of candidates, the methods will inevitably fail as they were not design to disprove all the provided options (mathematically, the sum of the probabilities of all candidates should be 100%). As stated above, the main application of DP4+ to date has been related to the determination of the relative configuration of a molecule whose planar structure is known. Therefore, the reliability of the final prediction will be linked to that assignment, which if incorrectly conducted would inevitably lead to a DP4+ failure, regardless the probability value computed for the most likely isomer. The elucidation of littordial F represents an interesting case study to exemplify the above. This natural product, isolated in 2019 from the leaves of Psidiumlittorale by Xu and Feng groups (Fig. 12),⁶² was proposed as a novel meroterpenoid with a unique 6/8/9/4-tetracyclic core structure. In order to establish the configuration at C-10′ the authors evaluated the two possible diastereoisomers with DP4+ using only ¹³C NMR data, suggesting that the real structure should be the 1R,4S,5R,9S,10′S isomer (53). However, in 2020 George et al. revised the structure to the corresponding 6/6/9/4-tetracyclic core (54) based on biosynthetic considerations, further confirmed by biomimetic total synthesis and NMR studies.

Fig. 12 The case of Littordial F reflects the problem of neglecting the real structure.

4 The ugly

DP4+ is a valuable tool for the in silico structural elucidation of natural products, affording overall reliable predictions (see “The good”), though in some challenging and specific cases the performance drops due to the intrinsic limitations of the method (see “The bad”). However, DP4+ results might be affected also by improper computational work and/or misuse of the excel toolbox. From our experience, and after a detailed analysis of the hundreds of cases reporting the use of DP4+, we've detected the most common pitfalls that might end up affecting the results, which will be discussed in this section. To avoid unnecessary speculation about the certainty of the results reported in those few papers where the following errors were detected, we decided not to cite them in this section (though they were cited in this review). After all, this part is dedicated to showing the nature of the errors and how to prevent them in future studies.

4.1 Use of inappropriate distributions

DP4+ was built on the basis of the scaled (δ_s) and unscaled (δ_u) errors (δ_calc − δ_exp) being random variables that follow t-like distributions. In the case of δ_u series, the errors depend on the hybridization of the nuclei, leading to two series: one for sp–sp² carbons (or protons attached to them) and another one for sp³ carbons (or protons attached to them). Since each t distribution is characterized by a set of three parameters: mean (μ), standard deviation (σ) and degrees of freedom (ν), the DP4+ equation requires the knowledge of six [μ, σ, ν] parameter sets (three for ¹³C data composed of one distribution for scaled shifts and two distributions for unscaled shifts, and the corresponding three distributions for ¹H data). Hence, any mistake related to the use of improper distributions might impact on the DP4+ values leading to potentially wrong assignments. The sensitivity of DP4+ to each of the statistical parameters was recently explored.²¹⁸ This can take place in two different ways:

4.1.1 Improper level of theory. The original DP4+ was developed from the [μ, σ, ν] sets estimated for 72 known molecules at 24 different levels of theory in the GIAO NMR calculation stage (combining two functionals, six basis sets, and two ways to treat solvation) using B3LYP/6-31G* optimized geometries. Beyond the intrinsic differences on the results obtained at each level (we recommended PCM/mPW1PW91/6-31+G**//B3LYP/6-31G*), the potential problems arising when correlating NMR chemical shifts with [μ, σ, ν] parameters obtained at different levels should be emphasized. One involuntary situation takes place when the computational work is done at one of the allowed 24 levels of theory but the choice is not properly indicated in the user selection drop-down lists available in the excel spreadsheet, Fig. 13. For instance, DP4+ affords the right assignment for compound 55 at the recommended PCM/mPW1PW91/6-31+G** level, for which the isotropic shielding values were computed. However, if a different level is selected in the excel toolbox, the DP4+ values will change as the Bayesian analysis is done with improper [μ, σ, ν] parameters (Fig. 13). The severity of the discrepancy strongly depends on the system under study, and the differences between the proper and improper [μ, σ, ν] values. In this particular case, choosing a wrong functional or solvation method affords slightly different, yet qualitatively similar results (that is, the sense of the assignment remains). However, selecting an inadequate basis set considerably affects the results by inverting the probability ratios. Note that changing the [μ, σ, ν] values has major influence on the uDP4+ probabilities because the corresponding distributions are not centered in zero (μ ≠ 0). In addition, if the calculated NMR data is inserted as isotropic shielding values (σ_x), the unscaled chemical shifts will be automatically computed using the default reference standard (TMS) values corresponding to the selected level of theory, and not the level actually used in the NMR calculations.

Fig. 13 Changes in DP4+ by selecting improper levels in the excel spreadsheet. The isotropic shielding values of 55 and ent-55 were carried out at the PCM/mPW1PW91/6-31+G**//B3LYP/6-31G* level.

A similar and potentially problematic situation arises when the NMR chemical shifts are computed at different levels for which DP4+ has not been parameterized and tested yet. In this case, unless the [μ, σ, ν] values of the new level are known, and are similar to one of those reported 24 levels available in DP4+, any selection made in the excel spreadsheet might end up in a mistaken assignment. This triggers a complex crossroad: without previous knowledge, it is difficult to determine which level to choose in the excel spreadsheet, but the final results will almost certainly depend on that selection. In addition, it is difficult to estimate the exact impact of this mistake, because it largely depends on the system under study and the differences in the [μ, σ, ν] values of the two levels of theory (the employed and the indicated one). Fig. 14 shows how the cumulative probability for a given error (difference between experimental and calculated NMR shifts) changes by choosing improper distributions. To overcome those limitations, we introduced a new customizable DP4+ version, which allows the user to select any level of theory of their preference. The new excel spreadsheet handles the previously explored 24 levels (default settings), and any new level as well (custom settings). If this option is selected, the user must introduce the sixteen [μ, σ, ν] values corresponding to the desired level using a set of known molecules. For preliminary calculations, we showed that these terms can be estimated using a reduced set of 8 molecules with no significant change in the overall DP4+ values. The new excel spreadsheet allows the automatic calculation of the [μ, σ, ν] parameters once the calculated NMR isotropic shielding values for the test set is entered.²¹⁸

Fig. 14 Graphical representation of the change in the cumulative probability for a given error by choosing different distributions.

4.1.2 Improper hybridization. Since the probability distributions of the unscaled chemical shifts largely depend on the hybridization of the nuclei, it is important to take it into account to properly compute the uDP4+ probabilities. This can be easily done in the excel spreadsheet by typing a letter “x” in the corresponding column for those sp–sp² carbons, or protons attached to them. However, if this mark is accidentally placed in a sp³ nucleus, or if analogously a sp–sp² nucleus is not marked, the spreadsheet will carry out uDP4+ calculations with improper [μ, σ, ν] sets, impacting on the overall DP4+ results. Sometimes this mistake is evident (for example, carbons in the carbonyl region without any mark, or highly shielded carbons in the aliphatic zone with the “x” mark). However, in other cases special care must be taken, as is the case of carbons in the 80–120 region, which could be both sp³ or sp–sp² depending on their chemical environment.

4.2 Incorrect chemical shifts calculations

The excel toolbox allows three different ways of data entrance: isotropic shielding values, unscaled chemical shifts or scaled chemical shifts. If the first option is selected, the unscaled chemical shifts are computed according to δ_u = σ₀ − σ_x (where σ₀ is the isotropic shielding value of TMS at the level of theory selected in the drop-down menu, and then the scaled chemical shifts are obtained as discussed above. This is the recommended way because it rules out possible errors made by the user during the chemical shift calculation and/or the subsequent scaling procedure. In addition, any other improper computational work will surely affect the results, regardless the way NMR data is inserted in the excel toolbox. The number of situations that might end up in wrong calculations is immense, but the most common ones are briefly discussed below:

4.2.1 Improper conformational sampling. When dealing with conformationally flexible molecules, the appropriate exploration of the conformational landscape plays a fundamental role in subsequent results. Missing potentially relevant conformations by using a very low energy cutoff, few steps in the conformational sampling, erroneous selection of the dihedral angles to rotate, or ring systems to flip, among others, might lead to negative consequences in the NMR prediction.

4.2.2 Improper labeling. When extracting the isotropic shielding values from the output files it is necessary to know which signal of resonance belongs to which nuclei, and this is done using the labeling scheme. The labeling should be analyzed in detail, otherwise the experimental and calculated NMR chemical shifts of nuclei occupying different positions in the molecule will be correlated. The atom numbers of different candidate structures could be different without it affecting the calculations, as long as they remain the same within each isomer. It's a serious mistake if the labeling changes in different conformers of the same isomer since two (or more) nuclei from different parts of the molecule would be treated as if they had the same chemical environment. In some publications we noticed alarmingly high errors (>20 ppm for carbon data, >2 ppm for proton data) fully inconsistent with the candidate structures that surely arose from labeling errors.

4.2.3 Signal averaging. The chemical shifts of equivalent nuclei that show fast interconversion should be averaged (such as the case of methyl groups, or some methylene groups). Treating the signal of each individual proton independently is wrong (for example, computing three different chemical shifts for the same methyl group). Another problem arises when dealing with diastereotopic methylene protons, which are often arbitrarily correlated. Unless the discrimination of both signals as pro-R and pro-S can be made with additional NMR information (such as NOE or J coupling), the most convenient way to tackle this issue is to order the experimental and calculated values upside down (that is, matching the most deshielded experimental value with the most deshielded calculated one).

4.2.4 Other errors. A comprehensive analysis of all sources of improper computational work is beyond this review article. However, neglecting frequency analysis (the NMR calculations should be carried out with local minima geometries), and not removing duplicates (each conformer should be counted only once during the Boltzmann weighting step) are amongst the errors that should be avoided.

4.3 Large outliers

Although DP4+ is based on the analysis of the errors that arise from the NMR calculation process, when these errors are too large caution must be taken. The distributions of scaled chemical shifts are centered in zero, with standard deviations in range 1–2 ppm and 0.1–0.2 ppm for carbon and proton shifts, respectively, using DFT optimized geometries. Hence, when alarmingly large scaled errors (>1 ppm for proton data, >10 ppm for carbon data) are observed, this could indicate a potential problem to be further analyzed in detail. The most common situations which may be encountered are described below:

4.3.1 All isomers showing the same outliers. This is a common problem when dealing with nuclei whose chemical shifts values are not properly reproduced by the calculations (for instance, carbon attached to heavy atoms). In this case, it would be wise to recompute the DP4+ values after removing those potentially conflicting signals. However, if the outliers show up in systems for which such large discrepancies are not expected, it would indicate that the real connectivity might be different than the considered one.

4.3.2 Some isomers showing very large outliers. It is expected that diastereoisomers display differences in their computed NMR chemical shifts. Nevertheless, these discrepancies should not be exorbitantly large. We have seen reports with diastereoisomers differing in more than 40 ppm in the calculated ¹³C values for the same position, which could hardly be justified from configurational differences. On the contrary, the origins of such differences might probably rely on improper computational work (vide supra).

4.4 General recommendations

Despite each research group has its own preferences for conducting the NMR calculations, herein we present general recommendations to carry out the DP4+ analysis.

(a) When dealing with flexible molecules, a thorough conformational sampling ought to be done using a safe energy cutoff (5 kcal mol⁻¹).

(b) All conformations found should be optimized at the B3LYP/6-31G* level, and duplicates should be removed. A frequency analysis on the most stable structures should be done to verify the nature of the stationary point found.†

(c) The NMR calculations should be done with all significantly populated conformations found in the previous step. It is not recommended to keep only the global minimum for further analysis.

(d) The NMR calculations should be done with the GIAO method, using any of the 24 levels of theory available for DP4+. For general purposes, we recommend PCM/mPW1PW91/6-31+G**.†

(e) When inserting the experimental and calculated data in the excel spreadsheet, it is important to differentiate the sp–sp² and sp³ nuclei. In addition, the NMR data shall be fully assigned.

(f) The calculated NMR data should be thoroughly revised. Alarmingly large errors (mainly for the most likely candidate) need to be deeply analyzed in search for inconsistencies. If diastereoisomers are scrutinized, very large differences in the calculated NMR values for the same nucleus might be indicative of a mistake.

(g) When dealing with flexible molecules conveniently functionalized with groups that could afford intramolecular H-bonding interactions, whenever the conformational landscape is dominated by shapes featuring IHB interactions, it is recommended to recompute the DP4+ analysis by averaging the isotropic shielding values with SMD-derived Boltzmann amplitudes, or by evaluating the system with the Random DP4+ approach.

(h) It is always recommended to validate the DP4+ results with the experimental NMR information available (such as homo- and heteronuclear coupling values and/or interatomic distances obtained through NOE/ROE experiments).

5 Conclusion

It has been demonstrated that DP4+ is a powerful and easy-to-use toolbox for the structural elucidation of natural products. It can be applied either independently or in combination with other methods. However, to obtain meaningful results the computational work should be done properly, and the data should be manipulated following the suggestions. By understanding the scope and limitations of the method, and how to use it properly, the chances of arriving at a good determination maximize. We consider that with the details and recommendations provided in this review, DP4+ will continue to facilitate the structural determination of new and valuable natural products with high confidence.

6 Conflicts of interest

There are no conflicts to declare.

7. Acknowledgements

We are grateful to UNR (BIO 316 and BIO 500) and ANPCyT (PICT-2016-0116, PICT-2017-1524 and PICT-2019-4052) for financial support. M. O. M. and S. C. thank CONICET for the award of a fellowship.

8 References

1. B. K. Chhetri, S. Lavoie, A. M. Sweeney-Jones and J. Kubanek, Nat. Prod. Rep., 2018, 35, 514–531.
2. K. C. Nicolaou and S. A. Snyder, Angew. Chem., Int. Ed., 2005, 44, 1012–1044.
3. M. E. Maier, Nat. Prod. Rep., 2009, 26, 1105–1124.
4. T. L. Suyama, W. H. Gerwick and K. L. McPhail, Bioorg. Med. Chem., 2011, 19, 6675–6701.
5. H. D. Yoo, S. J. Nam, Y. W. Chin and M. S. Kim, Arch. Pharmacal Res., 2016, 39, 143–153.
6. M. O. Marcarino, M. M. Zanardi, S. Cicetti and A. M. Sarotti, Acc. Chem. Res., 2020, 53, 1922–1932.
7. A. Bagno and G. Saielli, Wiley Interdiscip. Rev.: Comput. Mol. Sci., 2015, 5, 228–240.
8. A. G. Kutateladze and T. Holt, J. Org. Chem., 2019, 84, 8297–8299.
9. G. Lauro and G. Bifulco, Eur. J. Org. Chem., 2020, 2020, 3929–3941.
10. N. Grimblat and A. M. Sarotti, Chem.–Eur. J., 2016, 22, 12246–12261.
11. M. W. Lodewyk, M. R. Siebert and D. J. Tantillo, Chem. Rev., 2012, 112, 1839–1862.
12. S. G. Smith and J. M. Goodman, J. Org. Chem., 2009, 74, 4597–4607.
13. S. G. Smith and J. M. Goodman, J. Am. Chem. Soc., 2010, 132, 12946–12959.
14. K. Ermanis, K. E. B. Parkes, T. Agback and J. M. Goodman, Org. Biomol. Chem., 2019, 17, 5886–5890.
15. A. Howarth, K. Ermanis and J. M. Goodman, Chem. Sci., 2020, 11, 4351–4359.
16. N. Grimblat, M. M. Zanardi and A. M. Sarotti, J. Org. Chem., 2015, 80, 12526–12534.
17. N. Grimblat, J. A. Gavín, A. Hernández Daranas and A. M. Sarotti, Org. Lett., 2019, 21, 4003–4007.
18. D. Xin, P. J. Jones and N. C. Gonnella, J. Org. Chem., 2018, 83, 5035–5043.
19. E. Troche-Pesqueira, C. Anklin, R. R. Gil and A. Navarro-Vázquez, Angew. Chem., Int. Ed., 2017, 56, 3660–3664.
20. A. G. Kutateladze and D. S. Reddy, J. Org. Chem., 2017, 82, 3368–3381.
21. D. Dardić, G. Lauro, G. Bifulco, P. Laboudie, P. Sakhaii, A. Bauer, A. Vilcinskas, P. E. Hammann and A. Plaza, J. Org. Chem., 2017, 82, 6032–6043.
22. J.-Q. Li, H.-W. Zhao and Z.-J. Ma, Tetrahedron Lett., 2020, 151874.
23. H. Sun, J. Tan, W. Lv, J. Li, J. Wu, J. Xu, H. Zhu, Z. Yang, W. Wang, Z. Ye, T. Xuan, Z. Zou, Z. Chen and K. Xu, Bioorg. Chem., 2020, 95, 103493.
24. Y. N. Shi, S. Pusch, Y. M. Shi, C. Richter, J. G. Maciá-Vicente, H. Schwalbe, M. Kaiser, T. Opatz and H. B. Bode, J. Org. Chem., 2019, 84, 11203–11209.
25. Y. N. Wang, G. Y. Xia, L. Y. Wang, G. B. Ge, H. W. Zhang, J. F. Zhang, Y. Z. Wu and S. Lin, Org. Lett., 2018, 20, 7341–7344.
26. W. X. Wang, X. Lei, Y. L. Yang, Z. H. Li, H. L. Ai, J. Li, T. Feng and J. K. Liu, Org. Lett., 2019, 21, 6957–6960.
27. M. Ye, W. Xu, D. Q. He, X. Wu, W. F. Lai, X. Q. Zhang, Y. Lin, W. Xu and X. W. Li, J. Nat. Prod., 2020, 83, 362–373.
28. J. Bhandari Neupane, R. P. Neupane, Y. Luo, W. Y. Yoshida, R. Sun and P. G. Williams, Org. Lett., 2019, 21, 8449–8453.
29. S. Sala, G. L. Nealon, A. N. Sobolev, J. Fromont, O. Gomez and G. R. Flematti, J. Nat. Prod., 2020, 83, 105–110.
30. W. X. Wang, G. G. Cheng, Z. H. Li, H. L. Ai, J. He, J. Li, T. Feng and J. K. Liu, Org. Biomol. Chem., 2019, 17, 7985–7994.
31. D. Zhang, W. Yi, H. Ge, Z. Zhang and B. Wu, J. Nat. Prod., 2019, 82, 2800–2808.
32. B. Khatri Chhetri, S. Lavoie, A. M. Sweeney-Jones, N. Mojib, V. Raghavan, K. Gagaring, B. Dale, C. W. McNamara, K. Soapi, C. L. Quave, P. L. Polavarapu and J. Kubanek, J. Org. Chem., 2019, 84, 8531–8541.
33. W. X. Wang, Z. H. Li, J. He, T. Feng, J. Li and J. K. Liu, Fitoterapia, 2019, 137, 104278.
34. L. H. Martorano, A. L. Valverde, C. M. R. Ribeiro, A. C. F. De Albuquerque, J. W. D. M. Carneiro, R. G. Fiorot and F. M. Dos Santos Junior, New J. Chem., 2020, 44, 8055–8060.
35. H. H. Sun, W. Y. Lv, J. Tan, Y. C. Tang, H. Zhu, J. B. Qu, J. Li, J. P. Wu, X. W. Chang, Z. C. Yang, W. X. Wang, Z. H. Chen and K. P. Xu, Nat. Prod. Res., 2020, 1–7.
36. A. L. Macedo, L. H. Martorano, A. C. F. de Albuquerque, R. G. Fiorot, J. W. M. Carneiro, V. R. Campos, T. R. A. Vasconcelos, A. L. Valverde, D. L. Moreira and F. M. dos Santos, J. Braz. Chem. Soc., 2020, 31, 2030–2037.
37. W. Zhou, F. Kang, L. Huang, J. Li, W. Wang, L. Xiao, Q. Wen, X. Yu, Y. Xu, Z. Zou, H. Zhou, H. Zang, S. Chen and K. Xu, Bioorg. Chem., 2020, 101, 103959.
38. J. Yuan, X. Wen, C. Q. Ke, T. Zhang, L. Lin, S. Yao, J. D. Goodpaster, C. Tang and Y. Ye, Org. Chem. Front., 2020, 7, 1374–1382.
39. R. S. Phansalkar, J. W. Nam, A. A. Leme-Kraus, L. S. Gan, B. Zhou, J. B. McAlpine, S. N. Chen, A. K. Bedran-Russo and G. F. Pauli, J. Nat. Prod., 2019, 82, 2387–2399.
40. Y. Y. Fan, L. S. Gan, H. C. Liu, H. Li, C. H. Xu, J. P. Zuo, J. Ding and J. M. Yue, Org. Lett., 2017, 19, 4580–4583.
41. D. Pan, X. Zhang, H. Zheng, Z. Zheng, X. Nong, X. Liang, X. Ma and S. Qi, Org. Chem. Front., 2019, 6, 3252–3258.
42. K. S. Salome and C. F. Tormena, J. Org. Chem., 2018, 83, 10501–10504.
43. K. Tanaka, H. Manabe, R. Irie and M. Oikawa, Bull. Chem. Soc. Jpn., 2019, 92, 1314–1323.
44. Y. Feng, S. Khokhar and R. A. Davis, Nat. Prod. Rep., 2017, 34, 571–584.
45. X. H. Liu, X. L. Hou, Y. P. Song, B. G. Wang and N. Y. Ji, Fitoterapia, 2020, 141, 104469.
46. Y. F. Liu, Y. H. Zhang, C. L. Shao, F. Cao and C. Y. Wang, J. Nat. Prod., 2020, 83, 1300–1304.
47. J. W. Tang, H. C. Xu, W. G. Wang, K. Hu, Y. F. Zhou, R. Chen, X. N. Li, X. Du, H. D. Sun and P. T. Puno, J. Nat. Prod., 2019, 82, 735–740.
48. D. H. Liu, Y. Z. Sun, T. Kurtán, A. Mándi, H. Tang, J. Li, L. Su, C. L. Zhuang, Z. Y. Liu and W. Zhang, J. Nat. Prod., 2019, 82, 1274–1282.
49. H. S. Magalhães, A. B. da Silva, N. R. F. Nascimento, L. G. F. de Sousa, M. J. S. da Fonseca, M. I. B. Loiola, N. K. V. Monteiro, F. W. Q. Almeida Neto, K. M. Canuto and O. D. L. Pessoa, Fitoterapia, 2020, 143, 104545.
50. A. N. L. Batista, F. M. Dos Santos, A. L. Valverde and J. M. Batista, Org. Biomol. Chem., 2019, 17, 9772–9777.
51. W. P. Ding, K. Hu, M. Liu, X. R. Li, R. Chen, X. N. Li, X. Du, P. T. Puno and H. D. Sun, Fitoterapia, 2018, 127, 193–200.
52. Z. hui Huang, X. hua Nong, X. Liang and S. hua Qi, Tetrahedron, 2018, 74, 2620–2626.
53. F. Zeng, C. Chen, A. A. Al Chnani, Q. Zhou, Q. Tong, W. Wang, Y. Zang, J. Gong, Z. Wu, J. Liu, J. Wang, H. Zhu and Y. Zhang, Bioorg. Chem., 2019, 86, 176–182.
54. A. Ledoux, A. St-Gelais, E. Cieckiewicz, O. Jansen, A. Bordignon, B. Illien, N. Di Giovanni, A. Marvilliers, F. Hoareau, H. Pendeville, J. Quetin-Leclercq and M. Frédérich, J. Nat. Prod., 2017, 80, 1750–1757.
55. N. M. Betterley, S. Kerdphon, S. Chaturonrutsamee, S. Kongsriprapan, P. Surawatanawong, D. Soorukram, M. Pohmakotr, P. G. Andersson, V. Reutrakul and C. Kuhakarn, Asian J. Org. Chem., 2018, 7, 1642–1647.
56. S. R. Lee, E. Choi, S. H. Jeon, X. Y. Zhi, J. S. Yu, S. H. Kim, J. Lee, K. M. Park and K. H. Kim, Molecules, 2018, 23, 2732.
57. Z.-Y. Yan, T.-M. Lv, Y.-X. Wang, S.-C. Shi, J.-J. Chen, Bin-Lin, Q.-B. Liu, X.-X. Huang and S.-J. Song, Phytochemistry, 2020, 175, 112361.
58. L. Shao, P. Wu, L. Xu, J. Xue, H. Li and X. Wei, Fitoterapia, 2020, 141, 104465.
59. L. Zhou, F. Y. Han, L. W. Lu, G. D. Yao, Y. Y. Zhang, X. B. Wang, B. Lin, X. X. Huang and S. J. Song, Phytochemistry, 2019, 164, 122–129.
60. T. Huang, S. H. Ying, J. Y. Li, H. W. Chen, Y. Zang, W. X. Wang, J. Li, J. Xiong and J. F. Hu, Phytochemistry, 2020, 169, 112184.
61. C. Huo, X. Lu, Z. Zheng, Y. Li, Y. Xu, H. Zheng and Y. Niu, Phytochemistry, 2020, 170, 112224.
62. H. L. Zhu, Y. W. Hu, W. Qu, J. Zhang, E. Y. Guo, X. Y. Jiang, W. Y. Liu, F. Feng and J. Xu, Tetrahedron Lett., 2019, 60, 1868–1870.
63. J. Xu, Y. W. Hu, W. Qu, M. H. Chen, L. S. Zhou, Q. R. Bi, J. G. Luo, W. Y. Liu, F. Feng and J. Zhang, Bioorg. Chem., 2019, 90, 103046.
64. Y. P. Song, F. P. Miao, X. H. Liu, X. L. Yin and N. Y. Ji, Mar. Drugs, 2019, 17, 252.
65. M. J. Garson, W. Hehre, G. K. Pierens and Suciati, Molecules, 2017, 22, 521.
66. X. L. Cheng, H. X. Li, J. Chen, P. Wu, J. H. Xue, Z. Y. Zhou, N. H. Xia and X. Y. Wei, Nat. Prod. Bioprospect., 2021, 11, 63–72.
67. J. Ma, G. Xia, Y. Zang, C. Li, J. Yang, J. Huang, J. Zhang, Y. Su, A. Wang and D. Zhang, Chin. Chem. Lett., 2020, 32, 1173–1176.
68. X. Zhen, M. J. Mao, R. Z. Wang, S. S. Chang, T. M. Xiao, Y. X. Wu, L. Y. Yu, Y. L. Song, M. H. Chen and S. Y. Si, J. Asian Nat. Prod. Res., 2020, 1–8.
69. Y. Gao, F. Stuhldreier, L. Schmitt, S. Wesselborg, Z. Guo, K. Zou, A. Mándi, T. Kurtán, Z. Liu and P. Proksch, Front. Microbiol., 2020, 11, 600983.
70. X. Zheng, A. Kadir, G. Zheng, P. Jin, D. Qin, M. Maiwulanjiang, H. A. Aisa and G. Yao, Bioorg. Chem., 2020, 104, 104261.
71. G. Zheng, A. Kadir, X. Zheng, P. Jin, J. Liu, M. Maiwulanjiang, G. Yao and H. A. Aisa, Org. Chem. Front., 2020, 7, 3137–3145.
72. L. Qin, W. Yi, X. Y. Lian, N. Wang and Z. Zhang, Tetrahedron, 2020, 76, 131516.
73. A. Mándi, J. Wu and T. Kurtán, RSC Adv., 2020, 10, 32216–32224.
74. W. Li, R. Yan, Y. Yu, Z. Shi, A. Mándi, L. Shen, T. Kurtán and J. Wu, Angew. Chem., 2020, 132, 13128–13136.
75. W. Yi, L. Qin, X. Y. Lian and Z. Zhang, Mar. Drugs, 2020, 18, 385.
76. L. J. Zhu, F. Cao, X. X. Su, C. Y. Li, B. Lin, H. F. Wang, X. S. Yao, X. Zhang, J. M. Jia and H. W. Liu, J. Org. Chem., 2020, 85, 8580–8587.
77. F. Li, S. Lin, S. Zhang, L. Pan, C. Chai, J. C. Su, B. Yang, J. Liu, J. Wang, Z. Hu and Y. Zhang, J. Nat. Prod., 2020, 83, 1931–1938.
78. A. Cerulli, G. Lauro, M. Masullo, V. Cantone, B. Olas, B. Kontek, F. Nazzaro, G. Bifulco and S. Piacente, J. Nat. Prod., 2017, 80, 1703–1713.
79. L. C. Forster, G. K. Pierens, A. M. White, K. L. Cheney, P. Dewapriya, R. J. Capon and M. J. Garson, ACS Omega, 2017, 2, 2672–2677.
80. P.-E. Campos, E. Pichon, B. Illien, P. Clerc, C. Moriou, N. de Voogd, C. Hellio, R. Trepos, M. Frederich, A. Al-Mourabit and A. Gauvin-Bialecki, Nat. Prod. Chem. Res., 2018, 06, 19–22.
81. Z. Feng, S. Chen, W. Wang, L. Feng, Y. Dong, Y. Zou, C. Ke, C. Tang, S. Yao, H. Zhang, L. Gan, Y. Ye and L. Lin, Fitoterapia, 2019, 139, 104378.
82. W. Wang, J. Yang, Y. Y. Liao, G. Cheng, J. Chen, X. D. Cheng, J. J. Qin and Z. Shao, J. Nat. Prod., 2020, 83, 1157–1166.
83. K. Fukaya, D. Urabe, M. Hiraizumi, K. Noguchi, T. Matsumoto and K. Sakurai, J. Org. Chem., 2020, 85, 339–344.
84. W. Li, Y. Q. Tang, J. Sang, R. Z. Fan, G. H. Tang and S. Yin, Org. Lett., 2020, 22, 106–109.
85. A. Burtea and S. D. Rychnovsky, Org. Lett., 2018, 20, 5849–5852.
86. Q. Y. Zhang, B. C. Yan, K. Hu, X. N. Li, H. D. Sun and P. T. Puno, Fitoterapia, 2020, 142, 104529.
87. K. L. Ji, Y. Y. Fan, Z. P. Ge, L. Sheng, Y. K. Xu, L. S. Gan, J. Y. Li and J. M. Yue, J. Org. Chem., 2019, 84, 282–288.
88. H. C. Xu, K. Hu, X. H. Shi, J. W. Tang, X. N. Li, H. D. Sun and P. T. Puno, Org. Chem. Front., 2019, 6, 1619–1626.
89. N. Grimblat, T. S. Kaufman and A. M. Sarotti, Org. Lett., 2016, 18, 6420–6423.
90. C. Li, A. M. Sarotti, J. Turkson and S. Cao, Tetrahedron Lett., 2017, 58, 2290–2293.
91. P. Huang, C. Li, A. M. Sarotti, J. Turkson and S. Cao, Tetrahedron Lett., 2017, 58, 1330–1333.
92. C. S. Li, A. M. Sarotti, P. Huang, U. T. Dang, J. G. Hurdle, T. P. Kondratyuk, J. M. Pezzuto, J. Turkson and S. Cao, Sci. Rep., 2017, 7, 10424.
93. F. Della-Felice, A. M. Sarotti and R. A. Pilli, J. Org. Chem., 2017, 82, 9191–9197.
94. Y. S. Cai, A. M. Sarotti, D. Gündisch, T. P. Kondratyuk, J. M. Pezzuto, J. Turkson and S. Cao, Bioorg. Med. Chem. Lett., 2017, 27, 4630–4634.
95. A. M. Sarotti, Org. Biomol. Chem., 2018, 16, 944–950.
96. D. A. Heredia, A. M. Durantini, A. M. Sarotti, N. S. Gsponer, D. D. Ferreyra, S. G. Bertolotti, M. E. Milanesio and E. N. Durantini, Chem.–Eur. J., 2018, 24, 5950–5961.
97. Y. H. Tsai, C. M. Borini Etichetti, C. Di Benedetto, J. E. Girardini, F. T. Martins, R. A. Spanevello, A. G. Suárez and A. M. Sarotti, J. Org. Chem., 2018, 83, 3516–3528.
98. Y. S. Cai, A. M. Sarotti, T. L. Zhou, R. Huang, G. Qiu, C. Tian, Z. H. Miao, A. Mándi, T. Kurtán, S. Cao and S. P. Yang, J. Nat. Prod., 2018, 81, 1976–1983.
99. F. Della-Felice, F. F. De Assis, A. M. Sarotti and R. A. Pilli, Synth, 2019, 51, 1545–1560.
100. C. Li, A. M. Sarotti, X. Wu, B. Yang, J. Turkson, Y. Chen, Q. Liu and S. Cao, Molecules, 2019, 24, 196.
101. F. Della-Felice, A. M. Sarotti, M. J. Krische and R. A. Pilli, J. Am. Chem. Soc., 2019, 141, 13778–13782.
102. J. H. Ke, L. S. Zhang, S. X. Chen, S. N. Shen, T. Zhang, C. X. Zhou, J. X. Mo, L. G. Lin and L. S. Gan, Fitoterapia, 2019, 134, 346–354.
103. V. A. Cosenza, D. A. Navarro and C. A. Stortz, Carbohydr. Polym., 2017, 157, 156–166.
104. M. T. Sibero, T. Zhou, K. Fukaya, D. Urabe, O. K. Karna Radjasa, A. Sabdono, A. Trianto and Y. Igarashi, Beilstein J. Org. Chem., 2019, 15, 2941–2947.
105. L. A. Joyce, C. C. Nawrat, E. C. Sherer, M. Biba, A. Brunskill, G. E. Martin, R. D. Cohen and I. W. Davies, Chem. Sci., 2018, 9, 415–424.
106. Y. E. Sim, O. Nwajiobi, S. Mahesh, R. D. Cohen, M. Y. Reibarkh and M. Raj, Chem. Sci., 2020, 11, 53–61.
107. B. Lipp, L. M. Kammer, M. Kücükdisli, A. Luque, J. Kühlborn, S. Pusch, G. Matulevičiūtė, D. Schollmeyer, A. Šačkus and T. Opatz, Chem.–Eur. J., 2019, 25, 8965–8969.
108. H. C. Xu, K. Hu, H. D. Sun and P. T. Puno, Nat. Products Bioprospect., 2019, 9, 165–173.
109. S. M. Isyaka, M. K. Langat, E. Mas-Claret, B. M. Mbala, B. K. Mvingu and D. A. Mulholland, Phytochemistry, 2020, 170, 112217.
110. M. M. Rob, A. Iwasaki, R. Suzuki, K. Suenaga and H. Kato-Noguchi, Plants, 2019, 8, 301.
111. P. Klein, P. Johe, A. Wagner, S. Jung, J. Kühlborn, F. Barthels, S. Tenzer, U. Distler, W. Waigel, B. Engels, U. A. Hellmich, T. Opatz and T. Schirmeister, Molecules, 2020, 25, 1451.
112. S. Saito, K. Atsumi, T. Zhou, K. Fukaya, D. Urabe, N. Oku, M. R. Ul Karim, H. Komaki and Y. Igarashi, Beilstein J. Org. Chem., 2020, 16, 1100–1110.
113. F. R. Jiao, B. Bin Gu, H. R. Zhu, Y. Zhang, K. C. Liu, W. Zhang, H. Han, S. H. Xu and H. W. Lin, J. Org. Chem., 2020 DOI:10.1021/acs.joc.0c02049.
114. S. W. Li, C. Cuadrado, X. J. Huan, L. G. Yao, Z. H. Miao, A. Hernandez Daranas and Y. W. Guo, Bioorg. Chem., 2020, 103, 104223.
115. M. Alilou, S. Marzocco, D. Hofer, S. F. Rapa, R. Asadpour, S. Schwaiger, J. Troppmair and H. Stuppner, J. Nat. Prod., 2020, 83, 2456–2468.
116. S. W. Li, C. Cuadrado, L. G. Yao, A. H. Daranas and Y. W. Guo, Org. Lett., 2020, 22, 4093–4096.
117. F. Wang, A. M. Sarotti, G. Jiang, J. C. Huguet-Tapia, S. L. Zheng, X. Wu, C. Li, Y. Ding and S. Cao, Org. Lett., 2020, 22, 4408–4412.
118. Y. L. Li, R. X. Zhu, G. Li, N. N. Wang, C. Y. Liu, Z. T. Zhao and H. X. Lou, RSC Adv., 2019, 9, 4140–4149.
119. M. K. Langat, A. Helfenstein, C. Horner, P. Tammela, H. Hokkanen, D. Izotov and D. A. Mulholland, Chem. Biodiversity, 2018, 15, e1800056.
120. M. F. Elsebai, H. A. Ghabbour, N. Legrave, F. Fontaine-Vive and M. Mehiri, Med. Chem. Res., 2018, 27, 1885–1892.
121. X. C. Guo, L. L. Xu, R. Y. Yang, M. Y. Yang, L. D. Hu, H. J. Zhu and F. Cao, Front. Chem., 2019, 7, 80.
122. H. Wakamatsu, S. Tanaka, Y. Matsuo, Y. Saito, K. Nishida and T. Tanaka, Molecules, 2019, 24, 4279.
123. Ł. Pecio, M. Alilou, S. Kozachok, I. E. Orhan, G. Eren, F. S. S. Deniz, H. Stuppner and W. Oleszek, Molecules, 2019, 24, 4162.
124. J. Li, C. Li, R. Riccio, G. Lauro, G. Bifulco, T. J. Li, H. Tang, C. L. Zhuang, H. Ma, P. Sun and W. Zhang, Mar. Drugs, 2017, 15, 129.
125. R. P. Neupane, S. M. Parrish, J. B. Neupane, W. Y. Yoshida, M. L. Richard Yip, J. Turkson, M. K. Harper, J. D. Head and P. G. Williams, Mar. Drugs, 2019, 17, 423.
126. D. B. Pu, B. W. Du, W. Chen, J. B. Gao, K. Hu, N. Shi, Y. M. Li, X. J. Zhang, R. H. Zhang, X. N. Li, H. Bin Zhang, F. Wang and W. L. Xiao, Org. Lett., 2018, 20, 6314–6317.
127. D. Pu, X. Li, J. Lin, R. Zhang, T. Luo, Y. Wang, J. Gao, M. A. Zeb, X. Zhang, X. Li, R. Wang and W. Xiao, J. Nat. Prod., 2019, 82, 2067–2077.
128. Q. Yang, K. Hu, B. C. Yan, M. Liu, X. N. Li, H. D. Sun and P. T. Puno, Org. Chem. Front., 2019, 6, 45–53.
129. J. Wang, Q. Ren, Y. Y. Zhang, R. Guo, B. Lin, X. X. Huang and S. J. Song, Fitoterapia, 2019, 138, 104352.
130. C. Ma, C. W. Meng, Q. M. Zhou, C. Peng, F. Liu, J. W. Zhang, F. Zhou and L. Xiong, Fitoterapia, 2019, 138, 104351.
131. P. Zhao, R. Guo, Y. Y. Zhang, H. Zhang, G. D. Yao, B. Lin, X. B. Wang, X. X. Huang and S. J. Song, Bioorg. Chem., 2019, 93, 103354.
132. P. Zhou, J. Hu, B. Wen, J. Ding, B. Lou, J. Xiong, G. Yang and J. Hu, Tetrahedron, 2020, 76, 131026.
133. H. M. So, J. S. Yu, Z. Khan, L. Subedi, Y. J. Ko, I. K. Lee, W. S. Park, S. J. Chung, M. J. Ahn, S. Y. Kim and K. H. Kim, Bioorg. Chem., 2019, 91, 103145.
134. J. Gao, X. Zhang, K. Shang, W. Zhong, R. Zhang, X. Dai, X. Li, Q. Wang, Y. Zou and W. Xiao, Chinese Chem. Lett., 2020, 31, 427–430.
135. R. Guo, T. Lv, F. Han, Z. Hou, G. Yao, B. Lin, X. Wang, X. Huang and S. Song, Chinese Chem. Lett., 2020, 31, 1254–1258.
136. D. Liu, Q. Yin, Q. Zhang, J. Xiang, C. Ruan, H. Liu, B. Li, W. Zhu, C. ping Yin and J. Fang, Phytochem. Lett., 2019, 34, 91–95.
137. T. Zhao, X. H. Nong, B. Zhang, M. M. Tang, D. Y. Huang, J. L. Wang, J. L. Xiao and G. Y. Chen, Phytochem. Lett., 2020, 36, 115–119.
138. Y. G. Cao, Y. L. Zhang, M. N. Zeng, M. Qi, Y. J. Ren, Y. L. Liu, X. Zhao, X. K. Zheng and W. S. Feng, J. Nat. Prod., 2020, 83, 1118–1130.
139. H. Cui, Y. Liu, J. Li, X. Huang, T. Yan, W. Cao, H. Liu, Y. Long and Z. She, J. Org. Chem., 2018, 83, 11804–11813.
140. F. Wang, W. Zhao, C. Zhang, S. Chang, R. Shao, J. Xing, M. Chen, Y. Zhang and S. Si, RSC Adv., 2019, 9, 16035–16039.
141. S. G. Li, X. J. Huang, Y. L. Zhong, M. M. Li, Y. L. Li, Y. Wang and W. C. Ye, Chem. Biodiversity, 2019, 16, e1900192.
142. S. X. Yang, W. T. Zhao, H. Y. Chen, L. Zhang, T. K. Liu, H. P. Chen, J. Yang and X. L. Yang, Chem. Biodiversity, 2019, 16, e1900364.
143. X. Y. Han, Y. X. Xie, C. Q. Wu, H. L. Ai, X. X. Lei and X. J. Wang, Chem. Biodiversity, 2019, 16, e1900471.
144. J. S. Wu, X. H. Shi, Y. H. Zhang, J. Y. Yu, X. M. Fu, X. Li, K. X. Chen, Y. W. Guo, C. L. Shao and C. Y. Wang, Front. Chem., 2019, 7, 763.
145. T. T. Fan, H. H. Zhang, Y. H. Tang, F. Z. Zhang and B. N. Han, Mar. Drugs, 2019, 17, 652.
146. D. A. Adpressa, K. J. Stalheim, P. J. Proteau and S. Loesgen, ACS Chem. Biol., 2017, 12, 1842–1847.
147. M. Chen, R. Wang, W. Zhao, L. Yu, C. Zhang, S. Chang, Y. Li, T. Zhang, J. Xing, M. Gan, F. Feng and S. Si, Org. Lett., 2019, 21, 1530–1533.
148. G. Chianese, H. B. Yu, F. Yang, C. Sirignano, P. Luciano, B. N. Han, S. Khan, H. W. Lin and O. Taglialatela-Scafati, J. Org. Chem., 2016, 81, 5135–5143.
149. S. R. Lee, H. B. Park and K. H. Kim, Anal. Chem., 2018, 90, 13212–13216.
150. J. Li, Y. Hu, X. Hao, J. Tan, F. Li, X. Qiao, S. Chen, C. Xiao, M. Chen, Z. Peng and M. Gan, J. Nat. Prod., 2019, 82, 1391–1395.
151. X. Luo, C. Chen, H. Tao, X. Lin, B. Yang, X. Zhou and Y. Liu, Org. Chem. Front., 2019, 6, 736–740.
152. Y. Nalli, S. Jan, G. Lauro, J. Ur Rasool, W. I. Lone, A. R. Sarkar, J. Banday, G. Bifulco, H. Laatsch, S. H. Syed and A. Ali, Nat. Prod. Res., 2021, 35, 471–480.
153. K. J. Park, C. S. Kim, Z. Khan, J. Oh, S. Y. Kim, S. U. Choi and K. R. Lee, J. Nat. Prod., 2019, 82, 1345–1353.
154. P. Sun, Q. Yu, J. Li, R. Riccio, G. Lauro, G. Bifulco, T. Kurtán, A. Mándi, H. Tang, T. J. Li, C. L. Zhuang, W. H. Gerwick and W. Zhang, J. Nat. Prod., 2016, 79, 2552–2558.
155. Y. Tang, Z. Z. Zhao, K. Hu, T. Feng, Z. H. Li, H. P. Chen and J. K. Liu, J. Org. Chem., 2019, 84, 1845–1852.
156. K. Hu, X. R. Li, J. W. Tang, X. N. Li and P. T. Puno, Chin. J. Nat. Med., 2019, 17, 970–981.
157. H. Nguyen Ngoc, M. Alilou, M. Stonig, D. T. Nghiem, L. T. Kim, J. M. Gostner, H. Stuppner and M. Ganzera, J. Nat. Prod., 2019, 82, 2941–2952.
158. M. Alilou, D. F. Dibwe, S. Schwaiger, M. Khodami, J. Troppmair, S. Awale and H. Stuppner, J. Nat. Prod., 2020, 83, 1099–1106.
159. M. S. Jo, S. Lee, J. S. Yu, S. C. Baek, Y. C. Cho and K. H. Kim, J. Nat. Prod., 2020, 83, 684–692.
160. J. Li, M. Chen, X. Hao, S. Li, F. Li, L. Yu, C. Xiao and M. Gan, Org. Lett., 2020, 22, 98–101.
161. W. J. Lu, W. J. Xu, Y. Q. Zhang, Y. R. Li, X. Zhou, Q. J. Li, H. Zhang, J. Luo and L. Y. Kong, Org. Chem. Front., 2020, 7, 1070–1076.
162. C. S. Kim, J. Oh, L. Subedi, S. Y. Kim, S. U. Choi and K. R. Lee, Chem. Pharm. Bull., 2018, 66, 839–842.
163. F. Cao, T. T. Sun, J. K. Yang, G. Z. Zhao, Q. A. Liu, L. D. Hu, Z. Y. Ma and H. J. Zhu, Nat. Prod. Res., 2019, 33, 2192–2199.
164. T. Seitz, R. E. Millán, D. Lentz, C. Jiménez, J. Rodríguez and M. Christmann, Org. Lett., 2018, 20, 594–597.
165. L. Dong, D. P. Qin, Q. Q. Di, Y. Liu, W. L. Chen, Y. X. Cheng and S. M. Wang, Org. Chem. Front., 2019, 6, 3825–3833.
166. N. Grimblat, T. S. Kaufman and A. M. Sarotti, Org. Lett., 2016, 18, 6420–6423.
167. C. Li, A. M. Sarotti, J. Turkson and S. Cao, Tetrahedron Lett., 2017, 58, 2290–2293.
168. N. Q. Tuan, J. Oh, H. B. Park, D. Ferreira, S. Choe, J. Lee and M. K. Na, Phytochemistry, 2017, 133, 45–50.
169. T. Tsujita, Y. Matsuo, Y. Saito and T. Tanaka, Tetrahedron, 2017, 73, 500–507.
170. L. L. Gou, K. Hu, Q. Yang, X. N. Li, H. D. Sun, C. L. Xiang and P. T. Puno, Tetrahedron, 2019, 75, 2797–2806.
171. L. Chen, J. N. Yao, H. P. Chen, Z. Z. Zhao, Z. H. Li, T. Feng and J. K. Liu, Phytochem. Lett., 2018, 27, 94–100.
172. S. Lavoie, A. M. Sweeney-Jones, N. Mojib, B. Dale, K. Gagaring, C. W. McNamara, C. L. Quave, K. Soapi and J. Kubanek, J. Org. Chem., 2019, 84, 5035–5045.
173. X. Wang, J. Liu, P. Pandey, F. R. Fronczek, R. J. Doerksen, J. Chen, X. Qi, P. Zhang, D. Ferreira, F. A. Valeriote, H. Sun, S. Li and M. T. Hamann, Org. Lett., 2018, 20, 5559–5563.
174. S. Zhang, Y. Huang, S. He, H. Chen, Z. Li, B. Wu, J. Zuo, T. Feng and J. Liu, RSC Adv., 2018, 8, 23914–23918.
175. J. J. Orejola, M. Era, Y. Matsuo, Y. Saito and T. Tanaka, Tetrahedron, 2020, 76, 131204.
176. I. Park, W. Lee, Y. Yoo, H. Shin, J. Oh, H. Kim, M. A. Kim, J. S. Hwang, J. S. Bae and M. Na, Int. J. Mol. Sci., 2020, 21, 3406.
177. S. S. Zhu, J. W. Liu, Y. M. Yan, Y. Liu, Z. Mao and Y. X. Cheng, Org. Lett., 2020, 22, 3428–3432.
178. C. T. Sun, J. P. Wang, Y. Shu, X. Y. Cai, J. T. Hu, S. Q. Zhang, L. Cai and Z. T. Ding, Nat. Prod. Res., 2020 DOI:10.1080/14786419.2020.1806272.
179. Y. Zhang, H. Hu and J. Luo, Nat. Prod. Res., 2020 DOI:10.1080/14786419.2020.1830397.
180. Y. A. Rincón, G. E. Siless, L. D. Amado, M. V. Dansey, E. Grassi, N. Schenone and G. M. Cabrera, Nat. Prod. Res., 2020 DOI:10.1080/14786419.2020.1752205.
181. N. K. T. Pham, T. T. L. Tran, T. H. Duong, N. T. Trung, D. C. T. Phan, D. T. Mai, V. K. Nguyen, B. L. C. Huynh, T. A. T. Nguyen, T. D. Tran, T. N. M. Tran and T. P. Nguyen, Nat. Prod. Res., 2020 DOI:10.1080/14786419.2020.1839456.
182. X. Y. Hu, X. M. Li, S. Q. Yang, H. Liu, L. H. Meng and B. G. Wang, Mar. Drugs, 2020, 18, 194.
183. T. H. Duong, T. T. Nguyen, C. T. D. Phan, V. D. Nguyen, H. C. Nguyen, T. B. N. Dao, D. T. Mai, N. Niamnont, T. N. M. Tran and J. Sichaem, Nat. Prod. Res., 2020 DOI:10.1080/14786419.2020.1789980.
184. K. Calabro, L. K. Jennings, P. Lasserre, R. Doohan, D. Rodrigues, F. Reyes, C. Ramos and O. P. Thomas, J. Org. Chem., 2020, 85, 14026–14041.
185. H. Liu, X. Wang, Q. Shi, L. Li, Q. Zhang, Z. L. Wu, X. J. Huang, Q. W. Zhang, W. C. Ye, Y. Wang and L. Shi, ACS Omega, 2020, 5, 10167–10175.
186. W. Y. Zhang, Y. Zhong, Y. Yu, D. F. Shi, H. Y. Huang, X. L. Tang, Y. H. Wang, G. D. Chen, H. P. Zhang, C. L. Liu, D. Hu, H. Gao and X. S. Yao, J. Nat. Prod., 2020, 83, 3338–3346.
187. R. Kawazoe, Y. Matsuo, Y. Saito and T. Tanaka, J. Nat. Prod., 2020, 83, 3347–3353.
188. K. H. Lee, J. S. Yu, J. H. Choi, S. H. Kim, Y. J. Ko, C. Pang and K. H. Kim, Bioorg. Med. Chem. Lett., 2020, 30, 127641.
189. Q. Q. Shi, X. J. Zhang, Y. Zhang, Q. Wang, M. Amin, Q. Li, X. W. Wu, X. L. Li, R. H. Zhang, X. C. Dai and W. L. Xiao, Bioorg. Chem., 2020, 105, 104363.
190. J. S. Yu, C. Li, M. Kwon, T. Oh, T. H. Lee, D. H. Kim, J. S. Ahn, S. K. Ko, C. S. Kim, S. Cao and K. H. Kim, Bioorg. Chem., 2020, 105, 104397.
191. M. K. Langat, E. F. K. Djuidje, B. M. Ndunda, S. M. Isyaka, N. S. Dolan, G. D. Ettridge, H. Whitmore, I. Lopez, A. M. Alqahtani, I. Atiku, J. S. Lobe, E. Mas-Claret, N. R. Crouch, J. O. Midiwo, D. A. Mulholland and A. F. W. Kamdem, Phytochem. Lett., 2020, 40, 148–155.
192. N. Nath, J. C. Fuentes-Monteverde, D. Pech-Puch, J. Rodríguez, C. Jiménez, M. Noll, A. Kreiter, M. Reggelin, A. Navarro-Vázquez and C. Griesinger, Nat. Commun., 2020, 11, 4372.
193. Z. Hu, Z. Wu, Q. Su, M. Li, S. Wu, R. Meng, W. Ding and C. Li, Bioorg. Chem., 2020, 104, 104300.
194. Q. Q. Shi, X. J. Zhang, T. T. Wang, Q. Wang, T. T. Sun, M. Amin, R. H. Zhang, X. L. Li and W. L. Xiao, Org. Lett., 2020, 22, 7820–7824.
195. Y. Shu, J. P. Wang, X. Y. Cai, X. L. Li, J. T. Hu, C. T. Sun, L. Cai and Z. T. Ding, Tetrahedron, 2020, 76, 131520.
196. S. G. Li, Y. T. Wang, Q. Zhang, K. B. Wang, J. J. Xue, D. H. Li, Y. K. Jing, B. Lin and H. M. Hua, Org. Lett., 2020, 22, 7522–7525.
197. H. T. Li, R. T. Duan, T. Liu, R. N. Yang, J. P. Wang, S. X. Liu, Y. Bin Yang, H. Zhou and Z. T. Ding, Fitoterapia, 2020, 146, 104711.
198. Q. Li, W. Xu, R. Fan, J. Zhang, Y. Li, X. Wang, S. Han, W. Liu, M. Pan and Z. Cheng, J. Nat. Prod., 2020, 83, 2679–2685.
199. J. Wei Tang, K. Hu, X. Zheng Su, X. Nian Li, B. Chao Yan, H. Dong Sun and P. Tenzin Puno, Tetrahedron, 2020, 76, 131475.
200. S. Lee, D. Lee, R. Ryoo, J. C. Kim, H. B. Park, K. S. Kang and K. H. Kim, J. Nat. Prod., 2020, 83, 2737–2742.
201. Y. Kudo, C. T. Hanifin, Y. Kotaki and M. Yotsu-Yamashita, J. Nat. Prod., 2020, 83, 2706–2717.
202. B. Y. Hu, S. X. Wang, Y. M. Yan, J. W. Liu, D. P. Qin and Y. X. Cheng, Org. Chem. Front., 2020, 7, 2710–2718.
203. X. C. Guo, Y. H. Zhang, W. Bin Gao, L. Pan, H. J. Zhu and F. Cao, Mar. Drugs, 2020, 18, 479.
204. Q. Shi, S. Lu, D. Li, J. Lu, L. Zhou and M. Qiu, Fitoterapia, 2020, 145, 104635.
205. J. S. Yu, M. Park, C. Pang, L. Rashan, W. H. Jung and K. H. Kim, J. Nat. Prod., 2020, 83, 2261–2268.
206. J. P. Wang, Y. Shu, J. T. Hu, R. Liu, X. Y. Cai, C. T. Sun, D. Gan, D. J. Zhou, R. F. Mei, H. Ding, X. R. Zhang, L. Cai and Z. T. Ding, Org. Chem. Front., 2020, 7, 1463–1468.
207. X. R. Peng, Q. Q. Shi, J. Yang, H. G. Su, H. G. Su, L. Zhou, M. H. Qiu and M. H. Qiu, J. Org. Chem., 2020, 85, 7446–7451.
208. A. L. Duddupudi, P. Pandey, H. Vo, C. L. Welsh, R. J. Doerksen and G. D. Cuny, J. Org. Chem., 2020, 85, 7549–7557.
209. B. Fan, P. Dewapriya, F. Li, L. Grauso, M. Blümel, A. Mangoni and D. Tasdemir, Mar. Drugs, 2020, 18, 281.
210. P. Pel, H. S. Chae, P. Nhoek, Y. M. Kim, P. Khiev, G. J. Kim, J. W. Nam, H. Choi, Y. H. Choi and Y. W. Chin, Bioorg. Chem., 2020, 99, 103869.
211. C. X. Zou, Z. L. Hou, M. Bai, R. Guo, B. Lin, X. B. Wang, X. X. Huang and S. J. Song, Org. Biomol. Chem., 2020, 18, 3908–3916.
212. Y. Zou, X. Wang, J. Sims, B. Wang, P. Pandey, C. L. Welsh, R. P. Stone, M. A. Avery, R. J. Doerksen, D. Ferreira, C. Anklin, F. A. Valeriote, M. Kelly and M. T. Hamann, J. Am. Chem. Soc., 2019, 141, 4338–4344.
213. F. Della-Felice, A. M. Sarotti, M. J. Krische and R. A. Pilli, J. Am. Chem. Soc., 2019, 141, 13778–13782.
214. L. Andernach, L. P. Sandjo, J. C. Liermann, R. Schlämann, C. Richter, J. P. Ferner, H. Schwalbe, A. Schüffler, E. Thines and T. Opatz, J. Nat. Prod., 2016, 79, 2718–2725.
215. C. S. Kim, J. Oh, L. Subedi, S. Y. Kim, S. U. Choi and K. R. Lee, J. Nat. Prod., 2018, 81, 1795–1802.
216. P. Yan, G. Li, C. Wang, J. Wu, Z. Sun, G. E. Martin, X. Wang, M. Reibarkh, J. Saurí and K. R. Gustafson, Org. Lett., 2019, 21, 7577–7581.
217. G. Tarazona, G. Benedit, R. Fernández, M. Pérez, J. Rodríguez, C. Jiménez and C. Cuevas, J. Nat. Prod., 2018, 81, 343–348.
218. M. M. Zanardi and A. M. Sarotti, J. Org. Chem., 2021, 86, 8544–8548.
219. M. M. Zanardi, A. G. Suárez and A. M. Sarotti, J. Org. Chem., 2017, 82, 1873–1879.
220. M. M. Zanardi, F. A. Biglione, M. A. Sortino and A. M. Sarotti, J. Org. Chem., 2018, 83, 11839–11849.
221. F. Cen-Pacheco, J. Rodríguez, M. Norte, J. J. Fernández and A. Hernández Daranas, Chem.–Eur. J., 2013, 19, 8525–8532.
222. G. Zuber, M.-R. Goldsmith, T. D. Hopkins, D. N. Beratan and P. Wipf, Org. Lett., 2005, 7, 5269–5272.
223. C.-X. Wang, G.-D. Chen, C.-C. Feng, R.-R. He, S.-Y. Qin, D. Hu, H.-R. Chen, X.-Z. Liu, X.-S. Yao and H. Gao, Chem. Commun., 2016, 52, 1250–1253.
224. A. M. White, K. Dao, D. Vrubliauskas, Z. A. Könst, G. K. Pierens, A. Mándi, K. T. Andrews, T. S. Skinner-Adams, M. E. Clarke, P. T. Narbutas, D. C. M. Sim, K. L. Cheney, T. Kurtán, M. J. Garson and C. D. Vanderwal, J. Org. Chem., 2017, 82, 13313–13323.
225. M. G. Chini, R. Riccio and G. Bifulco, Eur. J. Org. Chem., 2015, 2015, 1320–1324.
226. D. J. Marell, S. J. Emond, A. Kulshrestha and T. R. Hoye, J. Org. Chem., 2014, 79, 752–758.
227. W. F. Xu, X. J. Xue, Y. X. Qi, N. N. Wu, C. Y. Wang and C. L. Shao, Nat. Prod. Res., 2021, 35, 490–493.
228. C. Jiménez, M. Blanco, C. Cuevas, R. Fernández, J. Rodríguez and G. Tarazona, Org. Lett., 2016, 18, 5832–5835.
229. S. C. Baek, K. H. Nam, S. A. Yi, M. S. Jo, K. H. Lee, Y. H. Lee, J. Lee and K. H. Kim, Foods, 2019, 8, 673.
230. M. M. Zanardi, M. O. Marcarino and A. M. Sarotti, Org. Lett., 2020, 22, 52–56.
231. M. O. Marcarino, M. M. Zanardi and A. M. Sarotti, Org. Lett., 2020, 22, 3561–3565.
232. M. M. Zanardi, M. A. Sortino and A. M. Sarotti, Carbohydr. Res., 2019, 474, 72–79.
233. A. V. Marenich, C. J. Cramer and D. G. Truhlar, J. Phys. Chem. B, 2009, 113, 6378–6396.
234. É. Brémond, M. Savarese, N. Q. Su, Á. J. Pérez-Jiménez, X. Xu, J. C. Sancho-García and C. Adamo, J. Chem. Theory Comput., 2016, 12, 459–465.
235. A. M. Sarotti, J. Org. Chem., 2020, 85, 11566–11570.
236. M. M. Zanardi and A. M. Sarotti, J. Org. Chem., 2015, 80, 9371–9378.
237. A. M. Sarotti, Org. Biomol. Chem., 2013, 11, 4847–4859.

---

Footnote

† In the case of the new updated version of DP4+, the geometry optimization and NMR calculations can be carried out at any desired level of theory. However, to obtain meaningful analysis a proper estimation of the [μ, ν, σ] set should be carried out.²¹⁸

---