High-Throughput in Silico Structure Validation and ... - ACS Publications

Mar 24, 2017 - a large number of halogenated metabolites from algae, constitute a diverse and promising feedstock for molecular .... Our structural-ty...
4 downloads 11 Views 5MB Size
Featured Article pubs.acs.org/joc

High-Throughput in Silico Structure Validation and Revision of Halogenated Natural Products Is Enabled by Parametric Corrections to DFT-Computed 13C NMR Chemical Shifts and Spin−Spin Coupling Constants Andrei G. Kutateladze* and D. Sai Reddy Department of Chemistry and Biochemistry, University of Denver, Denver, Colorado 80208, United States S Supporting Information *

ABSTRACT: Halogenated natural products constitute diverse and promising feedstock for molecular pharmaceuticals. However, their solution-structure elucidation by NMR presents several challenges, including the lack of fast methods to compute 13C chemical shifts for carbons bearing heavy atoms. We show that parametric corrections to DFT-computed chemical shifts in conjunction with rf f-computed spin−spin coupling constants allow for fast and reliable screening of a large number of reported halogenated natural products, resulting in expedient structure validation or revision. In this paper, we examine more than 100 structures of halogenated terpenoids and other natural products with the new parametric approach and demonstrate that the accuracy of the combined method is sufficient to identify misassignments and suggest revisions in most cases (16 structures are revised). As the 1D 1H and 13C NMR data are ubiquitous and most routinely used in solution structure elucidation, this fast and efficient two-criterion method (nuclear spin−spin coupling and 13C chemical shifts) which we term DU8+ is recommended as the first essential step in structure assignment and validation.



INTRODUCTION

As accurate computations of NMR spectra become more practical, they nicely augment the experimental methods of structure elucidation. Computations of 13C chemical shifts have been employed most frequently,6 leading to many high profile structure revisions.7 Nuclear spin−spin coupling constants (SSCCs) contain more structural information but are difficult to compute. Recently, however, practical parametric methods have emerged for fast and accurate evaluation of SSCCs: Bally and Rablen’s work on linear scaling of the easily computed Fermi contacts8 and our own related method which we termed relativistic force f ield (rff).9,10 With these recent developments, practical computational structure elucidation tools are now available to practitioners in the field of organic chemistry to quickly solve structures of new organic compounds or revise misassigned structures. In the world of natural products, these computational tools are becoming a must for successful structure assignment. There is one prominent exception to this trend, having to do with the chemical shift calculations for carbon atoms bearing heavy elements, for example, halogens.11,12 These calculations are either not accurate at all or they are too expensive for large organic molecules because of significant (and difficult to calculate) spin−orbit effects of heavy atoms on magnetic

Halogenated natural products from marine organisms, including a large number of halogenated metabolites from algae, constitute a diverse and promising feedstock for molecular pharmaceuticals.1 Red algae from the genus Laurencia (family Rhodomelaceae, order Ceramiales) have emerged as one of the most prolific producers of hundreds of halogenated bioactive molecules, which are thought to be used by these organisms as defense chemicals (our recent search for “Laurencia” restricted to the titles of papers published by the ACS, Wiley, and Elsevier produced 477 hits). Many of these molecules exhibit useful pharmacological properties and remain in the focus of natural product chemists. The structures of crystalline compounds are unambiguously established by X-ray crystallography. More often than not, however, these natural products are oils, and their structures and stereochemistries are assigned on the basis of solution NMR experiments. Useful experimental NMR techniques are emerging to address some challenges, idiosyncratic to deciphering halogenated structures. For example, utilization of 35,37Cl isotope effects on 13C chemical shifts, first reported four decades ago,2 is becoming a practical tool for locating chlorine-bearing carbons in complex structures,3 especially in conjunction with sophisticated pulse sequences pioneered by Molinski,4 the Merck NMR group,5 and others. © 2017 American Chemical Society

Received: January 24, 2017 Published: March 24, 2017 3368

DOI: 10.1021/acs.joc.7b00188 J. Org. Chem. 2017, 82, 3368−3381

The Journal of Organic Chemistry

Featured Article

shielding of directly attached carbons. For halogenated marine natural products, this challenge was recognized four decades ago.13 While computational methods exist to explicitly account for spin−orbit (SO) contributions,14 they are impractically slow for large organic systems and, at times, unpredictable. Kaupp and Malkin have shown15 that, when SO contributions are not accounted for, the discrepancy between the GIAO calculated and experimental values for organic iodo compounds could be as large as 20−30 ppm. What made matters more confusing is their finding that the explicit evaluation of all two-electron SO integrals in fact degraded the accuracy of chemical shift calculations. For this reason, they suggested that a computationally much more expedient mean-field approximation of SO terms should be preferred. Braddock and Rzepa in their structural reassignment of obtusallenes7b went further and suggested that a simple fixed value correction could successfully substitute for the difficult SO calculations in their entirety. For example, an empirical correction of −3 ppm was applied to all sp3 carbons bearing chlorine atoms, and −14 ppm to sp3 carbons bearing Br in the DFT 13C chemical shift calculations with the aug-cc-pVDZ basis set. As we demonstrated in the past, spin−orbit coupling contributions could be understood in simple structure-hybridization terms and could be parametrized to accelerate computations in many instances, regardless of whether it is the electronic spin−orbit coupling in triplet organic diradicals16 or nuclear spin−orbit contributions to indirect SSCCs.9 This is why we were intrigued by observations made by Kaupp and coworkers17 that the spin polarization induced by heavy-atom SO coupling interacts with nuclear magnetic moments of the system primarily by the same Fermi contact mechanism which made the Bally−Rablen approach to fast evaluation of SSCCs successful in the first place. Kaupp maintains that the “rules governing the propagation of ‘heavy atom’ effects through the molecule are...closely analogous to the well-established mechanisms for indirect Fermi-contact nuclear spin−spin coupling.” Given that our fast parametric rff method works well on a vast test set of proton spin−spin coupling constants (>3000 JHH, rmsd = 0.29 Hz), we hypothesized that a similar scaling could be employed to simulate effects of SO coupling on chemicals shifts, i.e., that Braddock’s simple numerical offsets could be more accurately and more generally implemented in the form of binomial scaling of the DFT/GIAO-computed isotropic magnetic shielding values for halogenated molecules. As we show below, this indeed was the case. Another critical point is that, in conjunction with the rff method we now have a synergistic integrated approach which uses two primary criteria, (i) SSCCs and (ii) 13C NMR chemical shifts, and, as a secondary criterion, 1H NMR chemical shifts, all for more reliable structure elucidation. By design, these calculations are computationally inexpensive and fast. On a single 12-core node of an aging Linux cluster, the DFT runs for each conformer, including full geometry optimization and calculations of NMR parameters, are completed in less than 30 min of wall-clock time for the majority of structures reported in this work (see the histogram in Figure 1). This allows for fast virtual screening of massive collections of compounds in a highthroughput manner to quickly identify misassigned structures. There are a number of approaches to fast computer-aided structure elucidation (CASE) which utilize parametrized chemical shifts, offering practical tools for structure discovery,18 but these are often only as good as the quality of calculated chemical shifts. For this reason, better hybrid CASE approaches

Figure 1. Wall-clock time distribution histogram for DFT calculations of structures, reported in this study, on a 12-core node of a Linux cluster.

are suggested with the DFT-computed shifts augmenting empirical corrections.19 On the other end of this spectrum, the DP420 method is gaining popularity, especially in cases when experimental data is limited. Instructively, to improve the predictive performance, Sarotti has recently developed an upgraded version of it, DP4+,21 in which both scaled and unscaled shifts are computed at a higher level of theory to address inadequacies of DP4 in dealing with unusually hybridized carbons, for example in oxiranes. This fine balance between the expediency and the accuracy of computations is the grand challenge of computer-assisted structure elucidation protocols. It is evident that the majority of misassignments could be readily identified utilizing ubiquitous 1D proton and carbon NMR data, provided that there is an accurate and fast method for computing both chemical shifts and spin−spin coupling constants. We believe that with our hybrid DFT/parametric method we are one step closer to addressing this grand challenge. In this paper, we demonstrate how this approach could be applied to a particularly challenging structure and stereochemistry assignment in compounds containing heavy atoms. We re-examined here more than a hundred prominent halogenated natural products, revising some of the originally proposed structures. As there were no existing methods for such fast in silico NMR analysis of halogenated molecules, we are not particularly surprised to see a rather unsettling rate of structure/ stereochemistry misassignment (∼13%) in the structurally rich world of red algae metabolites.



RESULTS AND DISCUSSION The training set of reliable experimental 13C NMR chemical shifts of halogenated compounds was compiled from open sources, such as the Spectral Database for Organic Compounds (SDBS),22 NMRShiftDB,23 NMR spectra obtained in our laboratory, and NMR data from the primary literature, such as Roberts’ work.24 We limited the scope of this study to four of the most commonly encountered “heavy” elements in natural products: S, Cl, Br, and I. As there were a significant number of training set experimental chemical shifts compiled, we employed quadratic, not linear, corrections for added accuracy. The DFT (GIAO) calculations of isotropic magnetic shielding values were carried out at a light ωB97xD/6-31G(d) level of theory.25 For carbon atoms bearing heavy elements, these values were scaled with a binomial formula: Icorr = aI2 + bI + c. We considered such nonlinear correction “safe” given the large number of points in the experimental training set (more than 3369

DOI: 10.1021/acs.joc.7b00188 J. Org. Chem. 2017, 82, 3368−3381

The Journal of Organic Chemistry

Featured Article

400; S, 53; Cl, 119; Br, 214; I, 48) with resulting chemical shifts spanning a large range of values from negative to +170 ppm. In all cases, the Student’s t-test indicated a clear significance of the quadratic term (i.e., |T| > 1 for the quadratic fitting terms in all four C−X sets). Table 1 summarizes the quadratic scaling parameters for carbons singly substituted with heavy elements.26

Our structural-type corrections were developed on a massive data set of >3600 experimental 13C chemical shifts (this is in addition to more than 400 C−S/C−Hal experimental values) allowing for considerable improvements in the overall accuracy: rmsd of 1.49 ppm for the general “nonheavy” set, and rmsd of 1.69 ppm for C−S, 1.37 ppm for C−Cl, 1.52 ppm for C−Br, and 2.28 ppm for C−I. All empirical scaling parameters are based on the ωB97xD/6-31G(d)//B3LYP/6-31G(d) level of theory, which produced sufficiently accurate results in short calculation times. We will refer to this combined J/δ method as DU8+.30 As an example, Figure 2 illustrates in a graphical form a typical fit for 13 C chemical shifts of brominated carbons in the training set.

Table 1. Quadratic Scaling Parameters for Isotropic Magnetic Shielding Values Calculated for C−X Carbons26 a (I2) C−S C−Cl C−Br C−I

−4

−1.571 × 10 −3.661 × 10−4 −2.370 × 10−4 8.764 × 10−4

b (I1)

c (I0)

0.9959 1.0477 0.9810 0.6830

8.28 7.69 16.62 50.34

The corrected magnetic shielding values were then converted into chemical shifts using linear scaling: δ(ppm) = 201.8 − 1.044Icorr (the same linear scaling is applied to all carbons in the molecule, not only the ones bearing heavy atoms). Statistical analysis of our data sets shows that, on average, at this level of DFT theory the C−X chemical shifts calculated without the prescribed parametric corrections are significantly overestimated: by 9.8 ± 1.8 ppm for C−Cl, 13.3 ± 3.4 ppm for C−Br and 33.1 ± 7.2 ppm for C−I (standard deviations are used instead of errors here). To accelerate computations, inexpensive B3LYP/6-31G(d)optimized structures were used, the same light level of DFT theory successfully employed in our rff computations of spin− spin coupling constants. As we explored other options for improving cost-efficiency of the method, we have entertained the following two considerations. First, we explored the role of the solvent, which is normally assessed nonexplicitly with polarized continuum model (PCM) approximations. Any self-consistent reaction field (SCRF) computations bear additional cost. The binomial scaling coefficients developed in this work were intentionally trained on experimental data obtained for most common CDCl3 solutions. This allowed for faster gas phase computations which, with our additional scaling, adequately described spectra recorded in chloroform-d, as we show below. Second, to further improve the accuracy of these simplified DFT computations, the initial pass on the computed and scaled chemical shift values was analyzed for systematic deviations in the training set. As a result, we identified several structural types causing such systematic errors and introduced additional empirical corrections. Computationally, these structural types, encoded as smarts strings,27 are readily searched for and identified with the help of OpenBabel;28 for details, see the Supporting Information. Such corrections are important for sp2 fragments and particularly for extended polarized conjugated systems. One simple hand-waving rationale for this is that the porbitals of sp2 carbons in conjugated systems, for example, α,βunsaturated ketones, are directly exposed to solvent, whereas sp3 carbons are more “protected” from the media. Generally 13C NMR chemical shifts are much less sensitive to solvent effects than proton chemical shifts, which makes them superior for the purposes of structure elucidation. However, exceptions to this rule are well documented. In some cases, a pronounced solvent effect is mediated by a heavy atom. One relevant example reported by Goroff is the 13C chemical shifts in iodoacetylenes exhibiting a marked dependence on specific interactions with solvent or Lewis bases.29

Figure 2. Typical exp−calcd 13C chemical shift correlations in the CBr and CBr2 training sets (both axes are in ppm).

This hybrid DFT/parametric approach often avoids more expensive PCM computations at both the structure optimization step and the chemical shift calculations, saving time without degrading the accuracy. For solvents other than chloroform, the calculated chemical shifts could be linearly scaled to best match the experimental data. This additional scaling is precedented31 and worked exceptionally well for Rychnovsky in his high profile revision of hexacylinol.7a The majority of the cases that we examined followed the suite, with only very few outliers where strong specif ic interactions with solvent required special consideration. We will address these exceptions below. First we overview the performance of our approach on a test set of halogenated natural products, many of which were unambiguously characterized by X-ray crystallographic analysis and therefore suitable for testing this method. Table 2 gives a representative subset of test structures that were examined in this study to evaluate the scope and the accuracy of the method (i.e., overall rmsd(δC), and the experimental and calculated chemical shifts for each carbon bearing a heavy atom). It is easy to see that our abbreviated and fast hybrid DFT/parametric approach compares favorably with the methods found in Tantillo’s excellent compilation on CheshireNMR.info,32 despite the fact that the compounds presented in Table 2 are multiply halogenated natural products. Several correct structures in Table 2 were accompanied by incorrect assignment of peaks in the 1H or 13C NMR spectra, which we corrected where possible (marked by asterisks in Table 2 and the SI). Oxachamigrene represents another variety of errors: its structure was incorrectly presented on the first page of Darias’ paper,57 although the NOE analysis pointed to an alternative trans-2-bromo-3-chloro configuration (confirmed by our computational analysis). Regrettably, the incorrect structure is propagated through the literature as it was copied into both 3370

DOI: 10.1021/acs.joc.7b00188 J. Org. Chem. 2017, 82, 3368−3381

The Journal of Organic Chemistry

Featured Article

Table 2. Structures of Halogenated Natural Products Validated in This Worka

3371

DOI: 10.1021/acs.joc.7b00188 J. Org. Chem. 2017, 82, 3368−3381

The Journal of Organic Chemistry

Featured Article

Table 2. continued

3372

DOI: 10.1021/acs.joc.7b00188 J. Org. Chem. 2017, 82, 3368−3381

The Journal of Organic Chemistry

Featured Article

Table 2. continued

a

Data presented in this table for each structure: (i) overall rmsd(δC) with the number of reported chemical shifts, N, used for statistical analysis; (ii) experimental (calculated) chemical shifts for each carbon bearing a heavy atom(s). bAsterisks designate cases where the reported chemical shifts of carbons bearing halogens had to be swapped with other carbons, most commonly the adjacent C−OH.

editions of the Dictionary of Marine Natural Products.33 Table 2 gives the corrected structure of oxachamigrene.



to completely botched assignments. Again, the ability to expediently compute both 13C chemical shifts and nuclear spin−spin coupling constants in a synergistic fashion imparted confidence that the structures we reviewed below are indeed misassigned. Tristichone C. Sheu and co-workers recently reported eight new halogenated chamigrane-type sesquiterpenoids, tristichones A−D and tristichols A−D.56 Most of them were assigned correctly (see Table 2 above). However, calculated 13C chemical shifts for the originally proposed structure of tristichone C matched well for all but one carbon: C(2)Cl was off by 6.6 ppm.

MISASSIGNED OR INCOMPLETELY ASSIGNED STRUCTURES

With this new, fast, computational tool, we revisited a large number of natural products containing heavy elements, mostly structures of halogenated sesquiterpenes. As a result, we discovered a rather unsettling rate of misassignment, varying from misassignment of stereochemistry of one stereogenic center 3373

DOI: 10.1021/acs.joc.7b00188 J. Org. Chem. 2017, 82, 3368−3381

The Journal of Organic Chemistry

Featured Article

The alternative regioisomer, C(2)Br and C(3)Cl, shown in Figure 3, gave a much better fit with rmsd(δ13C) of 0.99 ppm.

Figure 3. Original and revised structure of tristichone C.

Most importantly, C(2) and C(3) matched the experimental values well: C(2)Br 59.4 (58.4) ppm, C(3)Cl 72.6 (72.0) ppm, leaving little doubt that the correct structure for tristichone C is the shown 3-bromo-2-chloro-substituted chamigrene. The case of tristichone C illustrates a typical situation in which accurate calculations of C−Hal chemical shifts are critical for structure elucidation, given that (i) the proton chemical shifts are less reliable (and more solvent-dependent), and (ii) the computed SSCCs are often negligibly affected by such Cl−Br swaps. In the past, such revisions could not be realized with techniques other than X-ray, for example, Clardy’s revision of violacene,75 or Erickson’s revision of allo-isoobtusol (renamed cartilagineol).52 Misinterpretation of HRMS data, overlooked peaks, observation of a molecular ion for a minor component, or dehydrohalogenated ion is another general trend in misassignments of halogenated natural products. In the past, we10d and Koshino40 reported structure corrections for aldingenins A−D believed to be misassigned due to low quality HRMS data. We seem to have uncovered additional cases for this. Gomerone B. Halogenated sesquiterpenoids, gomerones A− C, were isolated and characterized by Cueto and co-workers in 2008.67 In 2012 gomerone C was synthesized by Carreira,69 who reported that the spectral data of their synthetic sample were identical to those of gomerone B. Based on this synthesis, Carreira proposed structural revision of gomerone C and, by logical extension, reassigned the structure of gomerone B, i.e., switched the two structures, Figure 4C. The structure of synthetic gomerone C was unambiguously determined by Xray, and unlike the original misassigned structure, it was consistent with Carreira’s mechanistic hypothesis of its formation from penultimate intermediate “19”. As Table 2 shows, it also gave an excellent match with our calculated data, rmsd(δ13C) = 1.09 ppm. However, neither original nor the newly reassigned structure of gomerone B matched well the δ/J data computed with DU8+. The biogenetic pathway for the formation of gomerones proposed by Cueto involves an electrophile-induced transannular carbocationic cyclization of 2-chloro-β-chamigrene shown in Figure 4B, which should lead to the axial orientation of the electrophilic halogen. Computed (DFT) energy of the gomerone isomer with axial chlorine at C(2) is 3.6 kcal/mol lower than that of the equatorial epimer. Both of these observations make the formation of the equatorial epimer problematic, either as a kinetic or thermodynamic product. As these sesquiterpenoids were isolated from Laurencia majuscula, we hypothesized that a similar pathway, initiated by ubiquitous electrophilic bromine could, have produced the 2bromo-3-chloro counterpart of gomerone C. This hypothesis was supported by the DU8+ computations, resulting in rmsd(δC)

Figure 4. Gomerones B and C. (A, B) Reference 67. (C) Reference 69.

= 1.65 ppm over 15 carbons and a good match for C(2)Cl 79.9 (78.5) and C(3)Br 70.8 (71.4) ppm based on the experimental data presented in the Cueto paper. This formal substitution of Cl for Br in gomerone C also led to significant improvement in the rff-calculated proton spin−spin coupling constants: from rmsd(JHH) of 0.75 Hz to 0.35 Hz over experimental SSCCs and improved the 1H chemical shifts match for 12 reported protons from rmsd(δH) 0.27 ppm to 0.09 ppm. We therefore suggest that the recently reassigned structure of gomerone B should be f urther revised to the shown 3-bromo-2-chloro sesquiterpenoid, Figure 4D, possessing the same stereochemistry as gomerone C (see also Table 4). As shown in Figure 4C, the last step in Carreira’s synthesis is hydrochlorination of the penultimate “19” with HCl/SnCl4. We suggest that gomerone B should be accessible via hydrobromination of the same key intermediate 19. Maneonenes, Isomaneonenes, And Lembynes. In 1978, Erickson, Clardy, and co-workers described isolation of C15halogenated nonterpenoid natural products, maneonenes, and isomaneonenes from the Hawaiian Marine Alga Laurencia nidif ica.97 Isomaneonene B was fully characterized by X-ray analysis, and the structure of isomeric isomaneonene A was proposed on the basis of its NMR. DU8+ calculations matched well the experimental spectra for both isomaneonenes A and B; see Figure 5 and Table 2. However, the C(5)Cl stereochemistry for cis-maneonenes A−C was not defined. Based on the results of DU8+ calculations we now assign their relative configuration as shown in Figure 5. It appears that they all have the 5(R) stereochemistry. cis-Maneonene A was shown to E−Z equilibrate with cismaneonene B under acidic conditions. Similar equilibration was observed for cis-maneonene C and compound 18. Although only 1 H NMR was reported for compound 18 in the Clardy paper, we 3374

DOI: 10.1021/acs.joc.7b00188 J. Org. Chem. 2017, 82, 3368−3381

The Journal of Organic Chemistry

Featured Article

The differences in 1H NMR chemical shifts are due to different solvents, chloroform-d for (12E)-lembyne A and benzene-d6 for cis-maneonene C. The 13C chemical shifts differ much less: rmsd between the two experimental sets is 0.44 ppm (for 12 carbons reported for cis-maneonene C). The DU8+ calculations for lembyne A matched the structure which we proposed for Clardy’s compound 18, Figures 5 and 6. While the 13C NMR data was not reported for compound 18, the experimental 1H NMR spectra match very well, with a systematic 0.1−0.15 ppm offset (both were recorded in benzene-d6). Lembyne B presented a challenge for us, as the originally proposed structure, i.e., the C(5) epimer of isomaneonene A, gave a poor match between the experimental and computed 13C chemical shifts. Instructively, the computed spin−spin coupling constants were acceptable, rmsd(JHH) < 0.5 Hz, pointing to a possible stereochemical misassignment of quaternary atoms and/or to potential misassignment of substituents which could be other than Br atoms. Indeed, the epimer at carbon C(13) gave a better but still marginally acceptable fit: the rmsd(δ13C) improved from 2.92 to 2.47 ppm. Additional uncertainty was introduced by the fact that the chemical shift for carbon C(13) was not reported, while acetylenic C(2) was reported at 84.08 ppm, which is inconsistent with several experimental and computational observations; see Table S3. To err on a safe side, we excluded this value from our initial comparisons. We then hypothesized that isomaneonenes A and B could be products of carbocationic transannular cyclization shown in Scheme 1. cis-Maneonene C could ionize, potentially with

Figure 5. Structures of maneonenes and isomaneonenes. Stereochemistry at C(5)Cl of cis-maneonenes A−C is assigned in this work.

Scheme 1. Plausible Carbocationic Pathways to Isomaneonenes A and B and Lembyne B

assigned its C(5)Cl stereochemistry based on computations of SSCCs, rmsd(JHH) = 0.45 Hz, and based on the hypothesis that acid-induced equilibrium between cis-maneonene C and compound 18 does not affect the stereochemistry at C(5). In 2001, Suzuki and co-workers isolated similar compounds from the Malaysian34 and Okinawan35 Laurencia species, which they named lembyne A, lembyne B,99 and (12E)-lembyne A100 (Figure 6). Our calculations show that (12E)-lembyne A, except for the erroneous swap of the reported 13C chemical shifts for acetylenic carbons C(1) and C(2), is in fact cis-maneonene C.

anchimeric assistance of the neighboring vinyl ether group, to generate a carbocation possessing an isomaneonene A framework. Nucleophilic attack at C(12) by the abundant bromine anion leads to isomaneonene A. The Z-isomer of cis-maneonene C (i.e., compound 18, Scheme 1B) could undergo a similar process to yield isomaneonene B (its structure was solved by Xray analysis).

Figure 6. Structure revisions of lembyne A, (12E)-lembyne A, and lembyne B. 3375

DOI: 10.1021/acs.joc.7b00188 J. Org. Chem. 2017, 82, 3368−3381

The Journal of Organic Chemistry

Featured Article

Lembyne B, having the butynene moiety in the exo position, would have to be a product of the yet unidentified 5-epimer of compound 18, which we will refer to as prelembyne B (Scheme 1C). Because the X = Br tentative structure for the revised lembyne B gave a marginally acceptable rmsd(δ13C) of 2.47 ppm, we entertained the notion that the reacting nucleophile could be chloride. Indeed, replacing Br with Cl at C(12) significantly improved the 13C match, while not degrading the other metrics: rmsd(JHH) = 0.43 Hz, rmsd(δ1H) = 0.23 ppm, rmsd(δ13C) = 1.29 ppm. As quaternary carbons bearing chlorine or bromine substitution are often difficult to differentiate (and also for the completeness of our candidate set), we considered the 12bromo-13-chloro- isomer (b) as an alternative (Figure 6). Surprisingly, it gave a comparable if not slightly better match: rmsd(JHH) = 0.44 Hz, rmsd(δ1H) = 0.24 ppm, rmsd(δ13C) = 1.23 ppm. A summary of the candidate structures for the revised lembyne B is presented in Table 3.

Scheme 2. Plausible Mechanism To Access the 13-Chloro Candidate Structure for Lembyne B

Table 3. Comparison of the Original and Three Candidate Structures of Lembyne B

Figure 7. ZPE-corrected B3LYP/6-311+G(d,p)//B3LYP/6-311+G(d,p) energy profile for the cation 1 → cation 2 rearrangement.

a

PCM calculations to ascertain the effect of aqueous solution on the potential cation 1−cation 2 equilibrium gave an even smaller energy gap of 2.8 kcal/mol. Summarizing the lembyne B case, we point to the misassignment of the original structure and suggest that both epi-13 candidate structures, 12-chloro and 13-chloro, are plausible (Figure 8). The accuracy of the method does not allow for

Experimental value 108.66 ppm.

The last two entries in Table 3 also show the quality of the fit in the case where the peak observed at 84.08 ppm is reassigned from C(2) (erroneous acetylenic) to C(13)Br. Again, the last two candidate structures, containing one chlorine and one bromine atom, produce better matches. Mechanistically, the 12-chloro product could readily be formed by the chloride departing from carbon C(5) of prelembyne B and attacking the oxonium cation at C(12) as shown in Scheme 1C. In contrast, formation of the potential 13-chloro product requires an additional rearrangement step, e.g., the Br migration shown in Scheme 2. To assess the plausibility of this rearrangement DFT structures of the two cations and the transition state were computed at the B3LYP/6-311+G(d,p)//B3LYP/6-311+G(d,p) level of theory. As shown in Figure 7, cation 2 is only 3.2 kcal/mol higher in energy than cation 1 (all values are zpe-corrected). The transition state, TS, was found only 11.8 kcal/mol above cation 1, implying very rapid interconversion at ambient temperature. One imaginary vibrational frequency was identified in the TS, Figure 7. Black arrows illustrate this normal mode, which is fully consistent with the expected reaction coordinate. Note that this mechanism accounts for least motion of the chloride anion to complete the reaction with the endo attack at C(13).

Figure 8. Two most likely candidates for lembyne B.

differentiation between the two candidates. To pinpoint the location of the chlorine atom one could potentially use the 35,37Cl isotope effects on 13C chemical shifts. The authors also reported one NOESY cross peak, which is in disagreement with our revision: between ethyl’s C(14)H2 (2.12 ppm) and C(5)H. Since the endo C(8)H proton, calculated within 2.04 Å from C(5)H, has an experimental chemical shift of 2.11 ppm, we hypothesize that the C(8)H−C(5)H cross peak was mistaken for the cross peak between C(5)H and ethyl’s methylene. Also, we do not have a good explanation for massspectrometry results in the original report. Boltzmann Populations and J/δ Fitting. Augmentation of the 13C chemical shift calculations with calculations of spin−spin 3376

DOI: 10.1021/acs.joc.7b00188 J. Org. Chem. 2017, 82, 3368−3381

The Journal of Organic Chemistry

Featured Article

dramatically from 1.21 to 0.17 Hz, leaving no doubt that tristichol D is correctly assigned. In cases that we studied, the J/δ-fitting procedure was required mostly for NMR spectra recorded in solvents other than chloroform, especially for natural products capable of forming intramolecular hydrogen bonds (alcohols, etc.). However, for nearly degenerate conformers, small corrections of the conformer ratios are often required for nonpolar solvents. Thus, it could be a useful tool in structure elucidation if applied with appropriate care. Table 4 summarizes all 16 structures which we revised computationally in this work (and also pinnatifenol for which there was no stereochemical assignment in the original paper). For example, Sims expected 11-iodolaurinterol as a result of iodination of laurinterol with iodine/silver trifluoroacetate in chloroform.61 According to our calculations, the product has a different regiochemistry, i.e., ortho to the hydroxy group, not meta substitution. Additionally, it is chlorine which is installed as an electrophile instead of iodine (for this we do not have a good rationale). Another unexpected discovery was that the sesquiterpene rigidol isolated in 199743 (a bromochamigrane diol, not to be confused with an unusual diterpene of the same name from Sapium rigidifolium101) is not the reported C(3)-alcohol. There is not enough information in the original report to decide on the precise nature of the C(3)-substitution. The calculated values for the most likely candidate, 3-hydroperoxide, are presented in Table 4, with the calculated value of 78.8 ppm for C(3) matching well the experimental value of 78.5 ppm. In addition to this computational evidence, this chemical shift favorably compares with that of C(3)−OOH in a model compound, 1methylcyclohex-2-enyl hydroperoxide, 80.1 ppm, previously described in the literature.102 Another potential structure for rigidol, i.e., the 3-acetate (see the Supporting Information), also has a good fit for the calculated value of the C(3) chemical shift (77.6 ppm). However, it is unlikely that the authors of the original report overlooked the acetate group or any other carboxylate in both the proton and 13C NMR spectra of the natural product. In contrast, the computed C(3) chemical shift for the free alcohol is off by more than 12 ppm. Isorigidol,54 which has a similar allylic alcohol moiety, C(3)−OH, has a reported chemical shift of 66.8 ppm (calculated 66.82 ppm); see Table 2. This cumulative evidence corroborates our hypothesis that the reported natural rigidol is misassigned and that it is most likely the 3-hydroperoxide.

coupling constants provides an additional independent criterion and compensates for the imperfect DFT energy calculations used to estimate the Boltzmann-weighted conformer content. This is especially important in cases of (nearly) energy-degenerate conformers, where a small error in computed relative energies translates into large errors in conformer populations. Figure 9

Figure 9. Uncertainty in the minor conformer content α as a function of ΔEDFT for a set error of 0.5 kcal/mol in energy calculations.

illustrates this point for an arbitrary two conformer system, where the absolute error of energy calculations is set, as an example, to 0.5 kcal/mol. The grayed area between the red and the blue boundaries represents the allowed values for α, i.e., the content of the minor conformer, as a function of the energy difference in kcal/mol. As the energy gap between the two conformers shrinks, accurate prediction of their relative populations becomes increasingly more challenging for the given energy calculation error. We previously suggested that a J-fitting procedure for calculated SSCCs and the experimental reported values could safely be applied to identify the correct structure as long as such fitting does not require too large an energy correction, which exceeds the energy accuracy reasonably expected for a given theory level.9b We now introduce a more rigorous quality of f it function Q, eq 1, which takes into account both 13C chemical shifts and SSCCs and which is used to double check the quality of NMR prediction based on DFT-computed Boltzmann populations against the conformer mixing via the J/δ-fitting 2

Q=

∑ (Jexp − Jcalc ) NJ

+

C C ⎛ δexp ⎞2 − δcalc ∑⎜ 8 ⎟ ⎝ ⎠

NC

(1)



where NJ is the number of reported experimental JHH and NC is the number of reported experimental 13C chemical shifts. The factor 1/8 is used to empirically balance the contributions of errors in δC and J. We did not include the computed proton chemical shifts in the quality function Q because of their excessive solvent dependence. However, we do use the rmsd(δH) values as an additional secondary criterion. The case of tristichol D, shown in Table 2, is an instructive example.56 It has two near-degenerate low energy conformers with ΔEDFT ≈ 0.2 kcal/mol. Mixing the two conformers according to the populations derived from their DFT energies (0.411:0.589) gives poor accuracy for SSCCs, rmsd(JHH) = 1.21 Hz and acceptable rmsd(δC) of 1.52 ppm. However, the J/δfitting procedure using the quality function Q inverts the conformer ratio to 0.604:0.396, with improved accuracy in rmsd(δC) = 1.43 ppm. The energy correction needed to reconcile this change is less than 0.45 kcal/mol, well within the accuracy of this DFT method. As a result, the SSCC rmsd(JHH) is improved

CONCLUSIONS A fast and accurate method for computing 13C NMR chemical shifts of carbon atoms, bearing heavy elements, via bionomial prescaling of DFT isotropic shielding values was validated on a large set of halogenated natural products. In conjunction with the rff-computed nuclear spin−spin coupling constants,9 it offers a synergistic and reliable double-criterion tool for expedient structure validation or revision based on ubiquitous 1D NMR data. Application of this approach on a test set of more than 100 halogenated sesquiterpenes and other natural products revealed a rather high rate of misassignment (∼13%), further underscoring the need for a computationally inexpensive yet accurate method for fast evaluation of NMR parameters. Sixteen structures of halogenated natural products were revised in this work. 3377

DOI: 10.1021/acs.joc.7b00188 J. Org. Chem. 2017, 82, 3368−3381

The Journal of Organic Chemistry

Featured Article

Table 4. Structures of Halogenated Natural Products Revised in This Work

a

Original or most recent assignment.

For the correctly assigned structures, we also encountered another challenge of structure validation: high rate of typos in the

tabulated NMR data. This underscores the importance of dissemination of the original NMR data (i.e., FID data deposited 3378

DOI: 10.1021/acs.joc.7b00188 J. Org. Chem. 2017, 82, 3368−3381

The Journal of Organic Chemistry

Featured Article

(13) Sims, J. J.; Rose, A. F.; Izac, R. R. In Marine Natural Products, Chemical and Biological Perspectives; Scheuer, P. J., Ed.; Academic Press: New York, 1978, Vol. 1, pp 297−378. (14) For example, see: ADF2016, SCM, Theoretical Chemistry; Vrije Universiteit: Amsterdam, http://www.scm.com. General references for ADF: (a) te Velde, G.; Bickelhaupt, F. M.; Baerends, E. J.; Fonseca Guerra, C.; van Gisbergen, S. J. A.; Snijders, J. G.; Ziegler, T. J. Comput. Chem. 2001, 22, 931. (b) Fonseca Guerra, C.; Snijders, J. G.; te Velde, G.; Baerends, E. J. Theor. Chem. Acc. 1998, 99, 391. (15) Malkina, O. L.; Schimmelpfennig, B.; Kaupp, M.; Hess, B. A.; Chandra, P.; Wahlgren, U.; Malkin, V. G. Chem. Phys. Lett. 1998, 296, 93. (16) (a) Kutateladze, A. G. J. Am. Chem. Soc. 2001, 123, 9279. (b) Kutateladze, A. G.; McHale, W. A. ARKIVOC 2005, IV, 88. (17) Kaupp, M.; Malkina, O. L.; Malkin, V. G.; Pyykkö, P. Chem. - Eur. J. 1998, 4, 118. (18) (a) Elyashberg, M. E.; Williams, A. J. Nat. Prod. Rep. 2010, 27, 1296. (b) For a comparison of DFT and empirical methods for computing chemical shifts, including the HOSE algorithm, see: Elyashberg, M.; Blinov, K.; Smurnyy, Y.; Churanova, T.; Williams, A. Magn. Reson. Chem. 2010, 48, 219. (c) Elyashberg, M. E.; Williams, A. J. Computer-Based Structure Elucidation from Spectral Data. The Art of Solving Problems; Springer: Heidelberg, 2015. (19) Buevich, A. V.; Elyashberg, M. E. J. Nat. Prod. 2016, 79, 3105. (20) Smith, S. G.; Goodman, J. M. J. Am. Chem. Soc. 2010, 132, 12946. (21) (a) Grimblat, N.; Zanardi, M. M.; Sarotti, A. M. J. Org. Chem. 2015, 80, 12526. (b) Zanardi, M. M.; Suarez, A. G.; Sarotti, A. M. J. Org. Chem. 2017, 82, 1873. (22) Compiled by the National Institute of Advanced Industrial Science and Technology (AIST) Japan: http://sdbs.db.aist.go.jp. (23) (a) http://nmrshiftdb.nmr.uni-koeln.de. (b) Kuhn, S.; Schlörer, N. E. Magn. Reson. Chem. 2015, 53, 582. (24) Hawkes, G. E.; Smith, R. A.; Roberts, J. D. J. Org. Chem. 1974, 39, 1276. (25) For ωB97xD, see: (a) Grimme, S. J. Comput. Chem. 2006, 27, 1787. (b) Chai, J. D.; Head-Gordon, M. Phys. Chem. Chem. Phys. 2008, 10, 6615. (c) Wu, Q.; Yang, W. T. J. Chem. Phys. 2002, 116, 515. (26) For more computational details, including parametrization of multiply substituted carbons, see the Supporting Information. (27) Daylight Chemical Information Systems, Inc.: http://www. daylight.com/dayhtml/doc/theory/theory.smarts.html. (28) O’Boyle, N. M.; Banck, M.; James, C. A.; Morley, C.; Vandermeersch, T.; Hutchison, G. R. J. Cheminf. 2011, 3, 33. (29) Rege, P. D.; Malkina, O. L.; Goroff, N. S. J. Am. Chem. Soc. 2002, 124, 370. (30) To summarize: DU8+ implies the following components of a hybrid DFT-parametric approach: (a) structure optimization, B3LYP/ 6-31G(d); (b) magnetic shielding, ωB97xD/6-31G(d); (c) Fermi contacts, B3LYP/DU8; (d) scaling of the computed Fermi contacts according to ref 9b,c to obtain spin−spin coupling constants. (e) Scaling of isotropic magnetic shielding values according to this work were used to obtain chemical shifts. For details, see the Supporting Information. (31) See also: Costa, F. L. P.; de Albuquerque, A. C. F.; dos Santos, F. M.; de Amorim, M. B. J. Phys. Org. Chem. 2010, 23, 972. (32) See http://cheshirenmr.info. (33) Findlay, J. A.; Li, G. Can. J. Chem. 2002, 80, 1697. (34) Iliopoulou, D.; Roussis, V.; Pannecouque, C.; De Clercq, E.; Vagias, C. Tetrahedron 2002, 58, 6749. (35) Coll, J. C.; Skelton, B. W.; White, A. H.; Wright, A. D. Aust. J. Chem. 1989, 42, 1695. (36) Kaiser, C. R.; Pitombo, L. F.; Pinto, A. C. Magn. Reson. Chem. 2001, 39, 147. (37) Aknin, M.; Ahond, A.; Chiaroni, A.; Poupat, C.; Riche, C.; Kornprobst, J.-M. Tetrahedron Lett. 1989, 30, 559. (38) de Nys, R.; Coll, J. C.; Bowden, B. F. Aust. J. Chem. 1993, 46, 933. (39) McPhail, K. L.; Davies-Coleman, M. T.; Copley, R. C. B.; Eggleston, D. S. J. Nat. Prod. 1999, 62, 1618. (40) Takahashi, S.; Yasuda, M.; Nakamura, T.; Hatano, K.; Matsuoka, K.; Koshino, H. J. Org. Chem. 2014, 79, 9373.

in a digital format suitable for subsequent analysis). There have been several calls for action, the latest being coordinated by Pauli under the community project on structural correctness.106



ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.joc.7b00188. Computational details (PDF)



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. ORCID

Andrei G. Kutateladze: 0000-0003-3066-517X Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS This research is supported in part by the NSF, CHE-1362959, and the University of Denver (High Performance Computing cluster).



REFERENCES

(1) (a) Paul, C.; Pohnert, G. Nat. Prod. Rep. 2011, 28, 186. (b) Wang, B.-G.; Gloer, J. B.; Ji, N.-Y.; Zhao, J.-C. Chem. Rev. 2013, 113, 3632. (2) Buchner, W.; Scheutzow, D. Org. Magn. Reson. 1975, 7, 615. (3) (a) Sergeyev, N. M.; Sergeyeva, N. D.; Raynes, W. T. Magn. Reson. Chem. 1994, 32, 381. (b) Sergeyev, N. M.; Sandor, P.; Sergeyeva, N. D.; Raynes, W. T. J. Magn. Reson., Ser. A 1995, 115, 174. (c) Morales-Rios, M. S.; Garcia-Martinez, C.; Joseph-Nathan, P.; Zepeda, L. G. Magn. Reson. Chem. 1995, 33, 149. (d) Foris, A. Magn. Reson. Chem. 2000, 38, 813. (4) Wang, X.; Duggan, B. M.; Molinski, T. F. J. Am. Chem. Soc. 2015, 137, 12343 and references cited therein. (5) Saurí, J.; Reibarkh, M.; Zhang, T.; Cohen, R. D.; Wang, X.; Molinski, T. F.; Martin, G. E.; Williamson, R. T. Org. Lett. 2016, 18, 4786. (6) For a review, see Lodewyk, M. W.; Siebert, M. R.; Tantillo, D. J. Chem. Rev. 2012, 112, 1839. (7) (a) Rychnovsky, S. D. Org. Lett. 2006, 8, 2895. (b) Braddock, D. C.; Rzepa, H. S. J. Nat. Prod. 2008, 71, 728. (c) Saielli, G.; Nicolaou, K. C.; Ortiz, A.; Zhang, H.; Bagno, A. J. Am. Chem. Soc. 2011, 133, 6072. (d) Lodewyk, M. W.; Soldi, C.; Jones, P. B.; Olmstead, M. M.; Rita, J.; Shaw, J. T.; Tantillo, D. J. J. Am. Chem. Soc. 2012, 134, 18550. (8) Bally, T.; Rablen, P. R. J. Org. Chem. 2011, 76, 4818. (9) (a) Kutateladze, A. G.; Mukhina, O. A. J. Org. Chem. 2014, 79, 8397. (b) Kutateladze, A. G.; Mukhina, O. A. J. Org. Chem. 2015, 80, 5218. (c) Kutateladze, A. G.; Mukhina, O. A. J. Org. Chem. 2015, 80, 10838. (10) For structure revisions utilizing rf f, see: (a) Reddy, D. S.; Kutateladze, A. G. Org. Lett. 2016, 18, 4860. (b) Kutateladze, A. G. J. Org. Chem. 2016, 81, 8659. (c) Reddy, D. S.; Kutateladze, A. G. Tetrahedron Lett. 2016, 57, 4727. (d) Mukhina, O. A.; Koshino, H.; Crimmins, M. T.; Kutateladze, A. G. Tetrahedron Lett. 2015, 56, 4900. (11) (a) Morishima, I.; Endo, K.; Yonezawa, T. J. Chem. Phys. 1973, 59, 3356. (b) Pyykkö, P. Chem. Phys. 1977, 22, 289. (c) Pyykkö, P.; Wiesenfeld, L. Mol. Phys. 1981, 43, 557. (d) Pyykkö, P.; Görling, A.; Rösch, N. Mol. Phys. 1987, 61, 195. (12) (a) Carbon spectra with chlorines could be accurately calculated by Gaussian 09 with WC04 functionals. (b) Spartan calculations at the EDFT/6-31G(d) level of theory offer expedited access to accurate chemical shifts of carbons-bearing halogens, with the exception of iodine. 3379

DOI: 10.1021/acs.joc.7b00188 J. Org. Chem. 2017, 82, 3368−3381

The Journal of Organic Chemistry

Featured Article

(41) Brennan, M. R.; Kim, I. K.; Erickson, K. L. J. Nat. Prod. 1993, 56, 76. (42) Ji, N.-Y.; Li, X.-M.; Li, K.; Ding, L.-P.; Gloer, J. B.; Wang, B.-G. J. Nat. Prod. 2007, 70, 1901. (43) König, G. M.; Wright, A. D. J. Nat. Prod. 1997, 60, 967. (44) Bunyapaiboonsri, T.; Yoiprommarat, S.; Lapanun, S.; Balram, U.; Chanthaket, R.; Klaysuban, A.; Suetrong, S. Phytochem. Lett. 2016, 18, 39. (45) (a) Kobayashi, J.; Cheng, J.; Ishibashi, M.; Nakamura, H.; Ohizumi, Y. Tetrahedron Lett. 1987, 28, 4939. (b) Perry, N. B.; Blunt, J. W.; Munro, M. H. G. Tetrahedron 1988, 44, 1727. (46) Gutierrez-Cepeda, A.; Fernandez, J. J.; Norte, M.; LopezRodriguez, M.; Brito, I.; Muller, C. D.; Souto, M. L. J. Nat. Prod. 2016, 79, 1184. (47) Gutierrez-Cepeda, A.; Fernandez, J. J.; Norte, M.; Souto, M. L. Org. Lett. 2011, 13, 2690. (48) Lhullier, C.; Falkenberg, M.; Ioannou, E.; Quesada, A.; Papazafiri, P.; Horta, P. A.; Schenkel, E. P.; Vagias, C.; Roussis, V. J. Nat. Prod. 2010, 73, 27. (49) Copley, R. C. B.; Davies-Coleman, M. T.; Edmonds, D. R.; Faulkner, D. J.; McPhail, K. L. J. Nat. Prod. 2002, 65, 580. (50) Ji, N.-Y.; Li, X.-M.; Li, K.; Gloer, J. B.; Wang, B.-G. Biochem. Syst. Ecol. 2009, 36, 938. (51) Juagdan, E. G.; Kalidindi, R.; Scheuer, P. Tetrahedron 1997, 53, 521. (52) (a) Revision of allo-isoobtusol (renamed to cartilagineol): Francisco, E. Y.; Turnbull, M. M.; Erickson, K. L. Tetrahedron Lett. 1998, 39, 5289. (b) For the original misassigned structure of allo-isoobtusol, see ref 51. (53) Gonzalez, A. G.; Martin, J. D.; Martin, V. S.; Norte, M. Tetrahedron Lett. 1979, 20, 2717. (54) (a) Davyt, D.; Fernandez, R.; Suescun, l.; Mombru, A. W.; Saldana, J.; Dominguez, L.; Coll, J.; Fujii, M. T.; Manta, E. J. Nat. Prod. 2001, 64, 1552. (b) X-ray structure: Suescun, L.; Mombru, A. W.; Mariezcurrena, R. A.; Davyt, D.; Fernandez, R.; Manta, E. Acta Crystallogr., Sect. C: Cryst. Struct. Commun. 2001, 57, 286. (55) Suzuki, T.; Furusaki, A.; Hashiba, N.; Kurosawa, E. Tetrahedron Lett. 1977, 18, 3731. (56) Chen, J.-Y.; Huang, C.-Y.; Lin, Y.-S.; Hwang, T.-L.; Wang, W.-L.; Chiou, S.-F.; Sheu, H.-H. J. Nat. Prod. 2016, 79, 2315. (57) Brito, I.; Cueto, M.; Diaz-Marrero, A. R.; Darias, J.; San Martin, A. J. Nat. Prod. 2002, 65, 946. (58) Suzuki, M.; Daitoh, M.; Vairappan, C. S.; Abe, T.; Masuda, M. J. Nat. Prod. 2001, 64, 597. (59) Diaz-Marrero, A.-R.; de la Rosa, J. M.; Brito, I.; Darias, J.; Cueto, M. J. Nat. Prod. 2012, 75, 115. (60) Tsukamoto, S.; Yamashita, Y.; Ohta, T. Mar. Drugs 2005, 3, 22. (61) Izac, R. R.; Sims, J. J. J. Am. Chem. Soc. 1979, 101, 6136. (62) Kladi, M.; Vagias, C.; Papazafiri, P.; Furnari, G.; Serio, D.; Roussis, V. Tetrahedron 2007, 63, 7606. (63) Kimura, J.; Kamada, N.; Tsujimoto, Y. Bull. Chem. Soc. Jpn. 1999, 72, 289. (64) Li, X.-D.; Miao, F.-P.; Yin, X.-Li; Liu, J.-L.; Ji, N.-Y. Fitoterapia 2012, 83, 1191. (65) Ji, N.-Y.; Li, X.-M.; Ding, L.-P.; Wang, B.-G. Biochem. Syst. Ecol. 2016, 64, 1. (66) Iliopoulou, D.; Vagias, C.; Galanakis, D.; Argyropoulos, D.; Roussis, V. Org. Lett. 2002, 4, 3263. (67) Diaz-Marrero, A. R.; Brito, I.; de la Rosa, J. M.; Darias, J.; Cueto, M. Tetrahedron 2008, 64, 10821. (68) This is another instructive case when ternary carbons C−OH and C−Cl were mistakenly assigned chemical shifts 85.6 and 72.9, respectively, which should be switched according to our calculations. (69) The original proposed structures of gomerones B and C (ref 67) were revised by Carreira: Huwyler, N.; Carreira, E. M. Angew. Chem., Int. Ed. 2012, 51, 13066. (70) Diaz-Marrero, A. R.; Brito, I.; de la Rosa, J. M.; D’Croz, L.; Fabelo, O.; Ruiz-Perez, C.; Darias, J.; Cueto, M. Eur. J. Org. Chem. 2009, 2009, 1407.

(71) Gonzalez, A. G.; Martin, J. D.; Martin, V. S.; Perez, R.; Tagle, B.; Clardy, J. J. Chem. Soc., Chem. Commun. 1985, 260. (72) There is an obvious typo, 140 ppm, in the reporting of a 13C chemical shift in ref 71, ostensibly for carbon C7, which should be ∼164 ppm according to calculations. Omission of this typo improves rmsd to under 1 ppm. There is no misassignment here as the X-ray structure for ketone 4 is reported. (73) (a) Fedorov, S. N.; Radchenko, O. S.; Shubina, L. K.; Kalinovsky, A. I.; Gerasimenko, A. V.; Popov, D. Y.; Stonik, V. A. J. Am. Chem. Soc. 2001, 123, 504. (b) Aplydactone was recently synthesized: Meier, R.; Trauner, D. Angew. Chem., Int. Ed. 2016, 55, 11251. (74) Matsuura, B. S.; Kölle, P.; Trauner, D.; de Vivie-Riedle, R.; Meier, R. ACS Cent. Sci. 2017, 3, 39. (75) (a) X-ray reassignment of the original structure: Van Engen, D.; Clardy, J.; Kho-Wiseman, E.; Crews, P.; Faulkner, D. J. Tetrahedron Lett. 1978, 19, 29. (b) 13C NMR data: Crews, P.; Kho-Wiseman, E.; Montana, P. J. Org. Chem. 1978, 43, 116. (76) (a) Coll, J. C.; Skelton, B. W.; White, A. H.; Wright, A. D. Aust. J. Chem. 1988, 41, 1743. (b) For an overview of several misassignments of mertensene, coccinene, and other compounds from the Southern Australian Marine Alga, Plocamium mertensii, see: Dias, D.; Urban, S. Phytochem. Anal. 2008, 19, 453. (77) König, G. M.; Wright, A. D.; Sticher, O. J. Nat. Prod. 1990, 53, 1615. (78) Wessels, M.; König, G. M.; Wright, A. D. J. Nat. Prod. 2000, 63, 920. (79) C(8) and C(9) chemical shift values are swappeda typo in the original data (ref 78). (80) Bucher, C.; Deans, R. M.; Burns, N. Z. J. Am. Chem. Soc. 2015, 137, 12784. (81) Skepper, C. K.; Molinski, T. F. J. Org. Chem. 2008, 73, 2592. (82) Manny, A. J.; Kjelleberg, S.; Kumar, N.; de Nys, R.; Read, R. W.; Steinberg, P. Tetrahedron 1997, 53, 15813. (83) Zhao, W.; Lorenz, N.; Jung, K.; Sieber, S. A. Chem. Commun. 2016, 52, 11971. (84) Tran, T. D.; Pham, N. B.; Quinn, R. J. J. Nat. Prod. 2016, 79, 570. (85) Imaoka, T.; Akimoto, T.; Iwamoto, O.; Nagasawa, K. Chem. Asian J. 2010, 5, 1810. (86) Kong, F.-D.; Ma, Q.-Y.; Huang, S.-Z.; Wang, P.; Wang, J.-F.; Zhou, L.-M.; Yuan, J.-Z.; Dai, H.-F.; Zhao, Y.-X. J. Nat. Prod. 2017, DOI: 10.1021/acs.jnatprod.6b01061. (87) Kuniyoshi, M.; Marma, M. S.; Higa, T.; Bernardinelli, G.; Jefford, C. W. Chem. Commun. 2000, 1155. (88) Caccamese, S.; Compagnini, A.; Toscano, R. M.; Nicolo, F.; Chapuis, G. Tetrahedron 1987, 43, 5393. (89) Djuardi, E.; Bovonsombat, P.; McNelis, E. Tetrahedron 1994, 50, 11793. (90) Paul, V. J.; McConnell, O. J.; Fenical, W. J. Org. Chem. 1980, 45, 3401. (91) Kusumi, T.; Uchida, H.; Inouye, Y.; Ishitsuka, M.; Yamamoto, H.; Kakisawa, H. J. Org. Chem. 1987, 52, 4597. (92) Felder, S.; Dreisigacker, S.; Kehraus, S.; Neu, E.; Bierbaum, G.; Wright, P. R.; Menche, D.; Schäberle, T. F.; König, G. M. Chem. - Eur. J. 2013, 19, 9319. (93) Yu, X.-Q.; Jiang, C.-S.; Zhang, Y.; Sun, P.; Kurtan, T.; Mandi, A.; Li, X.-L.; Yao, L.-G.; Liu, A.-H.; Wang, B.; Guo, Y.-W.; Mao, S.-C. Phytochemistry 2017, 136, 81. (94) Atta-Ur-Rahman; Ahmad, V. U.; Bano, S.; Abbas, S. A.; Alvi, K. A.; Ali, M. S.; Lu, H. S. M.; Clardy, J. Phytochemistry 1988, 27, 3879. (95) Li, X.-D.; Miao, F.-P.; Liang, X.-R.; Wang, B.-G.; Ji, N.-Y. RSC Adv. 2013, 3, 1953. (96) Wellington, K. D.; Cambie, R. C.; Rutledge, P. S.; Bergquist, P. R. J. Nat. Prod. 2000, 63, 79. (97) Waraszkiewicz, S. M.; Sun, H. H.; Erickson, K. L.; Finer, J.; Clardy, J. J. Org. Chem. 1978, 43, 3194. (98) Dictionary of Marine Natural Products; Blunt, J. W., Munro, M. H. G., Eds.; Chapman Hall/CRC Press, 2017. (99) Vairappan, C. S.; Daitoh, M.; Suzuki, M.; Abe, T.; Masuda, M. Phytochemistry 2001, 58, 291. 3380

DOI: 10.1021/acs.joc.7b00188 J. Org. Chem. 2017, 82, 3368−3381

The Journal of Organic Chemistry

Featured Article

(100) Vairappan, C. S.; Suzuki, M.; Abe, T.; Masuda, M. Phytochemistry 2001, 58, 517. (101) Siems, K.; Jakupovic, J.; Castro, V.; Poveda, L. Phytochemistry 1993, 33, 1465. (102) Courtneidge, J. L.; Bush, M.; Loh, L.-S. J. Chem. Soc., Perkin Trans. 1 1992, 1539. (103) Ji, N.-Y.; Li, X.-M.; Wang, B.-G. Helv. Chim. Acta 2010, 93, 2281. (104) Barnekow, D. E.; Cardellina, J. H.; Zektzer, A. S.; Martin, G. E. J. Am. Chem. Soc. 1989, 111, 3511. (105) Ahmad, V. U.; Ali, M. S. Phytochemistry 1991, 30, 4172. (106) (a) Pauli, G. F.; Niemitz, M.; Bisson, J.; Lodewyk, M. W.; Soldi, C.; Shaw, J. T.; Tantillo, D. J.; Saya, J. M.; Vos, K.; Kleinnijenhuis, R. A.; Hiemstra, H.; Chen, S.-N.; McAlpine, J. B.; Lankin, D. C.; Friesen, J. B. J. Org. Chem. 2016, 81, 878. (b) Bisson, J.; Simmler, C.; Chen, S.-N.; Friesen, J. B.; Lankin, D. C.; McAlpine, J. B.; Pauli, G. F. Nat. Prod. Rep. 2016, 33, 1028.

3381

DOI: 10.1021/acs.joc.7b00188 J. Org. Chem. 2017, 82, 3368−3381