Synthesis and Structure of Mycolactone E isolated from Frog Mycobacterium

Abstract

The structure of mycolactone E, isolated from the frog pathogen Mycobacterium liflandii, was established via organic synthesis. Within the mycolactone family of metabolites, a structural variation has been seen only at the unsaturated fatty acid moiety thus far, and mycolactone E follows this observation. Interestingly, the absolute configuration of its unsaturated fatty acid matches that of the mycolactones from human mycobacteria, rather than the structurally more closely related mycolactone F from fish mycobacteria.

Buruli ulcer is a severe necrotizing skin disease caused by Mycobacterium ulcerans, but it is one of the most neglected diseases.¹ Infection with M. ulcerans, probably carried by aquatic insects,² results in progressive necrotic lesions that, if untreated, can extend to 15% of a patient’s skin surface. Currently, surgical intervention is the only realistic therapy.³ In 1999, Small and co-workers isolated the toxic metabolites, named mycolactones A and B, from M. ulcerans. Evidence from animal studies suggested that mycolactones A and B are directly responsible for the observed pathology.⁴ The gross structure of mycolactones A and B was elucidated primarily through 2-D NMR experiments.⁵ Their stereochemistry was predicted via the NMR database approach and then confirmed by total synthesis.^6,7 Through these studies, mycolactones A and B are now described as a mixture of Z-Δ^4',5'- and E-Δ^4',5'-geometric isomers (Scheme 1). Mycolactones A and B constitute the major metabolites produced by West African strains of M. ulcerans. However, several mycolactone congeners,¹ including mycolactone C, ⁸ were recently isolated from clinical isolates of M. ulcerans from Africa, Malaysia, Asia, Australia, and Mexico. With use of organic synthesis, the structure of mycolactone C was also elucidated (Scheme 1).⁹

Scheme 1.

Structures of the mycolactones isolated from human and fish mycobacteria.

Interestingly, a mycolactone-like metabolite, named mycolactone F, was isolated from the fish pathogen M. marinum as well as M. pseudoshottsii,¹⁰ and its complete structure was established with use of organic synthesis as the major tool.¹¹ Intriguingly, the absolute configuration of the 1,3-diol present in the unsaturated fatty acid of mycolactone F corresponds to the antipode of that present in the mycolactones isolated from the human mycobacteria.

A frog pathogenic mycobacterium was discovered and named M. liflandii.¹² Following importation of the West African clawed frog Xenopus tropicalis, a lethal frog disease, caused via infection of M. liflandii, appeared in the United States.¹² From lipid extracts, Small and co-workers succeeded in isolation of a partially purified mycolactone, called mycolactone E, and proposed the structure II (Scheme 2).¹³ Based on MS analysis in combination with deuterium exchange and chemical transformation, Leadlay and co-workers suggested the alternative structure I (Scheme 2). ¹⁴ In this paper, we report a structure elucidation of mycolactone E with use of organic synthesis as the major tool.

Scheme 2.

Proposed structures for mycolactone E isolated from frog pathogen mycobacterium.

Between the two proposed structures, the MS/MS fragmentation pattern, in particular a fragment ion corresponding to [(M+Na)-102 Da]⁺ reported by Leadlay and coworkers,¹⁴ supports I over II. For this reason, we chose first to focus on the proposed structure I. It is worthwhile to note that the gross structure difference between I and mycolactone F is only Me- vs. Et-group at the terminal position of unsaturated fatty acid. Therefore, we recognized the possibility that the complete structure of mycolactone E could be established via a direct extension of the strategy used for the case of mycolactone F. More specifically, we anticipated that two remote diastereomers 9a and 9b (Scheme 3) possible for I should play a central role for this work, and began the synthesis of 9a and 9b via the synthetic route used for mycolactone F.¹¹ Specifically, we first synthesized the unsaturated fatty acids 4a and 4b (Scheme 4)¹⁵ and then coupled 4a and 4b with the core alcohol 5^{7, 16} (Scheme 5), to furnish the two remote diastereomers 9a and 9b, respectively.

Scheme 3.

Two remote diastereomers possible for I.

Scheme 4.

Synthesis of Side-chain. Reagents and Conditions: (a) (1) CH₂=CHMgBr, CuI, THF, followed by addition of TMSCl, i- Pr₂NEt. (2) OsO₄, NMO, followed by Pb(OAc)₄ treatment; 67% overall yield. (b) Follow the Cr-mediated catalytic asymmetric allylation reported in Reference ¹⁵. (c) Stepwise chain elongations consisting of Horner-Emmons olefination, DIBAL reduction, and MnO₂ oxidation. For the details, see Supporting Information and Reference ¹¹.

Scheme 5.

Completion of total synthesis. Reagents and Conditions: Yamaguchi esterification: (a) Cl₃C₆H₂COCl, i-Pr₂NEt, DMAP, PhH, rt, 24 h, 74%. (b) TBAF, THF, rt, 18 h, 76%.

Determination of Gross Structure

Overall, we adopted the strategy precedent in the mycolactone F series. In the mycolactone E series, however, we faced a new challenge; mycolactone E was available only in a very minute amount,¹⁷ which presented a severe limitation in choosing an analytical method for comparison of the natural and synthetic materials. With this limitation, we first recorded the MS/MS spectra for the synthetic and natural materials under the identical conditions in a side-by-side manner; although the relative peak-intensity of [(M+Na)-102 Da]⁺/[M+Na]⁺ was found delicately to depend on each measurement, the synthetic material gave the MS/MS profile very similar to that of the natural mycolactone E.¹⁸

The MS/MS profile discussed supported the proposed gross structure I, but we wished to have more direct, unambiguous evidence to establish the gross structure. An obvious choice was to rely on the NMR spectroscopy, but the amount of natural mycolactone E available was unfortunately insufficient to measure a proton NMR spectrum.¹⁹ Fortuitously, however, a ¹H NMR spectrum was recorded in one of our laboratories in 2005 (Figure 1).^13,20

Figure 1.

¹H NMR spectrum (acetone-d₆) of natural and synthetic mycolactone E. Panel A: natural mycolactone E (500 MHz). Panels D and C: before and after photochemical isomerization of synthetic mycolactone E 9a (600 MHz). Panel B: after photochemical isomerization of 9a (500 MHz). For the entire spectra shown in Panels A–D and the corresponding spectra in the 9b series, see Supporting Information.

At first glance, we felt that the ¹H-NMR spectrum of natural mycolactone E was not consistent with the proposed gross structure I; more specifically, the H5’ and H7’ resonances stood out as a singlet at 6.43 and 6.05 ppm in the ¹H-NMR spectrum of 9a, but no corresponding resonances were easily recognized in the ¹H-NMR spectrum of natural mycolactone E. However, on the basis of the knowledge gained in the mycolactone F series, we soon realized the possibility that this ¹H-NMR spectrum might not totally deny the proposed structure I.

As mentioned, the structural difference between mycolactone F and mycolactone E is minimal. Therefore, we anticipate the chemical behaviors of mycolactone E to be parallel with those observed on mycolactone F or its remote diastereomer. Although slow, we observed the geometric isomerism of mycolactone F under the standard laboratory conditions, to yield a mixture of three predominant geometric isomers. Thus, an aged sample of mycolactone E should be composed of a mixture of the corresponding geometric isomers. Then, there is a possibility that the H5’ and H7’ protons might give six independent singlets and therefore it could not be straightforward to detect these signals in the ¹H-NMR spectrum of natural mycolactone E.

In order to experimentally test this possibility, we subjected synthetic 9a to a photochemically-induced isomerization and obtained an approximately 1:1:1 mixture of the three dominant geometric isomers. With the use of the diagnostic chemical shifts identified in the mycolactone F series,¹¹ we assigned these products as all-trans-, Z-Δ^4',5'- and Z-Δ^6',7'-geometric isomers (Scheme 6). The ¹H-NMR spectrum of the sample thus obtained is shown in Figure 1. Amazingly, all the signals present in this spectrum were detected in the ¹H-NMR spectrum of natural mycolactone E, thereby confirming the proposed gross structure I.

Scheme 6.

Photochemical isomerism of 9a,b. Photochemical isomerization was done in acetone with a Reonet photoreactor at 300 nm for 16 minutes, to furnish a 1:1:1 mixture of the three geometric isomers as the major products. The diagnostic chemical shifts (¹H NMR in acetone-d₆) are: 6.43 ppm (H-5’) and 7.38 (H-3’) for all trans isomer; 6.66 (H-5’) and 7.39 (H-3’) for Z-Δ^6’,7’ isomer ; 5c: 6.32 (H-5’) and 7.93 (H-3’) for Z-Δ^4’,5’ isomer.¹¹ For the time-course of photochemistry, see Supporting Information.

A parallel experiment was conducted on 9b, a remote diastereomer of 9a. As expected from our previous work,^{11, 21} 9b exhibited the ¹H-NMR spectrum virtually identical to that of 9a.¹⁸ These experiments have now established that the structure of mycolactone E is represented by either 9a or 9b.

Determination of Stereochemistry

As before,¹¹ we relied on the chiral HPLC profile, to differentiate 9a and its remote diastereomer 9b. For the HPLC experiments, we purposely used the photochemically-equilibrated mixture of 9a and 9b, with the hope that each of their geometric isomers should give a distinct retention-time and therefore the HPLC comparison could be performed on the basis of six, instead of two, distinct retention times. Experimentally, we found that a Chiralpak IA chiral column in a mixture of benzene and ethanol meets our need well; with a fine-tuning on the ratio of two solvents, all of the six remote diastereomers, i.e., all trans-, Z-Δ^4',5'-, and Z-Δ^6',7'-isomers of both 9a and 9b, gave a distinct retention time (Figure 2). ²² We then applied this HPLC condition to natural mycolactone E (Figure 2, Panel a), thereby showing that natural mycolactone E and the 1:1:1 mixture of three geometric isomers of 9a exhibit the identical HPLC profile. This conclusion was further confirmed via a co-injection of the natural and synthetic materials (Figure 2, Panels b and c). These experiments have established that the natural product used for the comparison is a mixture of the three major geometric isomers of 9a, but not 9b. At present, we have not established whether the intact natural product is a single geometric isomer or a mixture of the three geometric isomers.

Figure 2.

HPLC analysis of synthetic photochemically-isomerized mycolactone E (9a), its remote diastereomer 9b, and natural mycolactone E.²² Column: Chiral Tech, Chiralpak IA (5 µm), 250 × 4.6 mm; Solvent (isocratic): EtOH/PhH = 1.5/98.5; flow rate 1 mL/min; detection: absorption at 323 nm. Panel A: 9a. Panel B: 9b. Panel C: a ca. 1:1 mixture of 9a and 9b. Panel a: the lipid extract containing natural mycolactone E.¹⁷ Peak indicated by * corresponds to the C13’-ketone. Panel b: a ca. 1:1 mixture of mycolactone E and synthetic 9a. Panel c: a ca. 1:1 mixture of natural mycolactone E and synthetic 9b.

In conclusion, we have established the complete structure of mycolactone E isolated from the frog pathogen M. liflandii in a stepwise manner. First, the proposed gross structure I was confirmed via (1) the synthesis of 9a and its remote diastereomer 9b, (2) the photochemical isomerization of 9a and 9b into a 1:1:1 mixture of the three major geometric isomers, and (3) the ¹H NMR comparison of these mixtures with that of natural mycolactone E, recorded in 2005. Second, the stereochemistry was deduced via (1) the establishment of a HPLC analytical method to differentiate 9a from 9b and (2) the established analytical method was applied to the natural product, thereby establishing the complete structure of mycolactone E as 9a. Within mycolactones isolated from the human and fish mycobacteria, a structural variation has been seen only at the unsaturated fatty acid moiety thus far, and mycolactone E isolated from the frog mycobacterium follows this observation. Interestingly, the absolute configuration of its unsaturated fatty acid matches that of the mycolactones from the human mycobacteria, rather than the structurally more closely related mycolactone from the fish mycobacteria. Lastly, we should note that the supply of structurally well-defined mycolactone E and its remote diastereomer for biological tests is now secured via organic synthesis.

Supplementary Material

1_si_001. Supporting Information Available.

Experimental details and ¹H and ¹³C NMR spectra of key compounds (50 pages). This material is available free of charge via the Internet at http://pubs.acs.org.