Synthesis, Structural Reassignment, and Antibacterial Evaluation of 2,18-Seco-Lankacidinol B

Abstract

Lankacidins are a group of polyketide natural products with activity against several strains of Gram-positive bacteria. We developed a route to stereochemically diverse variants of 2,18-seco-lankacidinol B and found that the stereochemical assignment at C4 requires revision. This has interesting implications for the biosynthesis of natural products of the lankacidin class, all of which possessed uniform stereochemistry prior to this finding. We have evaluated 2,18-seco-lankacidinol B and three stereochemical derivatives against a panel of pathogenic Gram-positive and Gram-negative bacteria.

The lankacidin group of natural products was first discovered in the extracts of Streptomyces rochei over fifty years ago and contain an unusual 17-membered macrocyclic ring and a highly substituted β-keto-δ-lactone function (Figure 1A).^[1] Lankacidin C (1) has inhibitory activity equal to or better than erythromycin against many species of Gram-positive bacteria, but is not susceptible to common forms of erythromycin resistance, making it a promising candidate for further development.^[2] In an effort to improve the therapeutic potential of lankacidins, McFarland and colleagues at Pfizer pursued structural modifications by semisynthesis, but their efforts were limited by the chemical instability of the β-keto-δ-lactone core, which is required for potent antibiotic activity (Figure 1B).^[2b] Fully synthetic methods to generate lankacidin derivatives could circumvent many of the limitations of semisynthetic approaches and enable systematic investigation of structure–activity effects in the class.^[3] Three fully synthetic routes to lankacidin natural products have been reported,^[4] including an elegant recent biomimetic approach by Hong and colleagues.^[4d] To the best of our knowledge, the effects of modifications of the β-keto-δ-lactone on antibiotic activity have not yet been systematically investigated.^[5]

Figure 1.

A) Selected members of the lankacidin class of natural products, with structural differences compared to lankacidin C (1) highlighted in blue. B) Chemical instability of the β-keto-δ-lactone core structure.

Earlier this year, Lu et al. reported the isolation and structural elucidation of 2,18-seco-lankacidinols A (3) and B (4), two new members of the lankacidin class that lack the C2–C18 bond that is common to other family members.^[6] 2,18-Seco-lankacidinol B (4) is the first acyclic member of the lankacidin class to be isolated and likely arises from a modified biosynthetic pathway.^[7] During the course of our efforts to develop a modular platform for the synthesis of lankacidin antibiotics, we synthesized acyclic intermediates that closely resembled 4, including several stereochemical variants. Herein we report the initial results from those efforts, which have provided material for microbiological evaluation and have resulted in a structural reassignment of 2,18-seco-lankacidinol B (4).

In the design of our route to lankacidins, we prioritized convergency and the use of reactions that we anticipated would have high functional group tolerance so as to streamline the synthesis of derivatives. Thus, our first efforts were aimed at the synthesis of two halves of approximately equal complexity, right half 11 and left halves 17–20 (Scheme 1). The synthesis of right half 11 commences with exposure of oxirane 5^[8] to lithium acetylide (generated from 1-bromopropene and butyllithium)^[9] followed by addition of tert-butyldimethylsilyl chloride to generate protected diol 7 in 78% yield (Scheme 1A). Oxidative deprotection of the PMB group, Mitsunobu displacement of the resulting primary alcohol with benzo[d]thiazole-2-thiol, and in situ oxidation of the resulting thioether delivers sulfone 8 in 77% yield over 2 steps. Julia-Kociensky olefination with commercially available aldehyde 9 provides dienoate 10 in 80% yield as a single stereoisomer.^[10] Reduction of the ethyl ester to the primary alcohol (DIBAL-H) followed by palladium-mediated hydro-stannylation^[11] delivers right half 11 in 84% yield. The route to 11 requires 6 steps from 5 and 6, proceeds in 40% overall yield, and has enabled the production of multi-gram quantities of this key intermediate.

Scheme 1.

Synthesis of coupling partners en route to 2,18-seco-lankacidinol B. A) Synthesis of right half 11. B) Synthesis of left halves 17 and 18 by means of an anti-selective aldol-lactonization cascade. C) Synthesis of left halves 19 and 20 by means of a syn-selective aldol followed by lactonization.

We designed our route to the β-keto-δ-lactone functionality to enable stereochemical control within the ring. By leveraging the Evans aldol reaction with common intermediate 15 as an electrophile, it would be possible to generate all four possible diastereomeric combinations at C4 and C5 in a predictable fashion. As shown in Scheme 1B, common intermediate 15 is prepared by an aldol reaction between β-iodoacrolein (12)^[12] and acetyl thiazolidinethione 13 (70% yield) followed by silyl protection of the resulting secondary alcohol and reductive cleavage of the auxiliary (91%, 2 steps). To access the 4R,5R and 4S,5S diastereomers, an anti-selective aldol reaction is required, and our efforts were informed by elegant work by Hong and co-workers in their synthesis of lankacidinol (2).^[4d] Thus, exposure of the boron enolate of bketo imide 16^[13] to aldehyde 15 in the presence of excess chlorodicyclohexylborane provides two anti-aldol products, which lactonize upon exposure to sodium methoxide to provide 4R,5R left half 17 (61% isolated yield) and 4S,5S left half 18 (21% isolated yield). Stereochemical assignments of the two diastereomers were made by comparison to similar intermediates in Hong˛s lankacidin synthesis^[4d] and by ¹H-¹H NOESY NMR analysis of the final compounds 4 and 23 (see Scheme 2 and Supporting Information).

Scheme 2.

Convergent coupling of right half 11 to left halves 17–20 and structural reassignment of 2,18-seco-lankacidinol B.

To access the 4R,5S and 4S,5R stereochemical variants, syn-aldol reactions are required. Under conditions reported by Evans and co-workers,^[13] syn-aldol reaction of β-keto imide 21 with aldehyde 15 followed by lactonization provides 4R,5S left half 19 in 98% yield over 2 steps. Similarly, aldolization of β-keto imide 16 followed by lactonization provides 4S,5R left half 20 in 90% yield over 2 steps. Left half 20 was transformed into its ferrocene carboxylate 22, which readily crystallized from ethyl ether. X-ray crystallographic analysis confirmed its relative and absolute stereochemistry.

To complete the synthesis of 2,18-seco-lankacidinols B bearing all four C4,C5-diastereomeric variations, we attached left half 11 to each right half 17–20 by means of Stille cross coupling reactions (Scheme 2). The use of a tin scavenger (Ph₂PO₂NBu₄)^[14] and a highly polar solvent (DMSO) greatly accelerates the reaction and results in increased isolated yields (see supporting information for optimization details).^[15] These convergent coupling reactions proceed efficiently (73–78% yield) without the need for protection of the free enol ether at C2/C3 or the primary alcohol at C18. Desilylation of each of the Stille products with hydrogen fluoride provides 2,18-seco-lankacidinols B (4, 23–25). Each of the final products 4 and 23–25 were oils, and we were unable to obtain X-ray-quality crystals of derivatives thereof. However, the structural assignments are strongly supported by the Evans’ aldol stereochemical models, nuclear Over-hauser effect data (not included in the isolation report, see supporting information), and the X-ray crystallographic data of ferrocene carboxylate 22.

Much to our surprise, the ¹H- and ¹³C-NMR spectra of 4 (Figure 2, teal traces) did not match the reported spectra^[6] of 2,18-seco-lankacidinol B (Figure 2, black traces). Most notably, the methine proton off of C5 and the C20 methyl protons were significantly shifted, as were the ¹³C signals corresponding to C3, C5, C6, and C20. Taken together, these discrepancies suggested to us that the structural differences resided on the β-keto-δ-lactone ring. We then compared the spectra of remaining three diastereomers to the reported spectra, we found that the spectra of the 4R,5S diastereomer 25 (Figure 2, red traces), which has the opposite C4 stereochemistry as the originally reported structure, matched very well, necessitating a structural reassignment of 2,18-seco-lankacidinol B from structure 4 to structure 25.

Figure 2.

¹H- and ¹³C-NMR spectral comparison of 4 (teal) and 25 (red) to the spectra of the natural product (black).^[6] Horizontal axes are expressed in parts per million (ppm).

We measured the inhibitory activity of 4 and 23–25 against a panel of Gram-positive and Gram-negative bacteria using levofloxacin as a positive control (Table 1). All four compounds exhibited no activity up to a concentration of 64 mgmL⁻¹ against most strains, although 4, 23, and 25 exhibited weak activity against H. influenzae ATCC 49247 (32 μgmL⁻¹). These data are consistent with the previously reported weak inhibitory activity of 2,18-seco-lankacidinol B against B. subtilis and S. aureus (obtained with disk diffusion tests).^[6] It is interesting to note that all four stereochemical derivatives of the β-keto-δ-lactone ring exhibit poor inhibitory activity, suggesting that the C18 pyruvamide function or the macrocyclic ring (or both) are required for the antimicrobial properties of the class.

Table 1:

Inhibitory activities of 2,18-seco-lankacidinols against selected strains of Gram-positive and Gram-negative bacteria.^[a,b]

	A. baumannii ATCC 19606	E. coli ATCC 25922	K. pneumoniae ATCC 43816	P. aeruginosa ATCC 27853	H. influenzae ATCC 49247	E. faecalis ATCC 29212	E. faecium ATCC 35667	S. aureus ATCC 29213	S. aureus (MRSA) ATCC 33591	S. pneumoniae ATCC 49619
Compound
4	>64	>64	>64	>64	32	>64	>64	>64	>64	>64
23	>64	>64	>64	>64	32	>64	>64	>64	>64	>64
24	>64	>64	>64	>64	>64	>64	>64	>64	>64	>64
25	>64	>64	>64	>64	32	>64	>64	>64	>64	>64
levofloxacin	0.25	0.015	0.03	2	0.03	1	4	0.25	0.12	0.5

^[a]Values given in μgmL⁻¹,

^[b]ATCC= American Type Culture Collection.

The reassignment of 4 to 25 makes it stereochemically unique among reported lankacidin-class natural products, all of which have been assigned the opposite stereochemistry at C4. This raises interesting questions about the biosynthesis of 25. The C20 methyl groups of lankacidin C (1) and lankacidinol (2) are proposed to be installed by a methyltransferase in the LkcF module of the lankacidin polyketide synthase, prior to macrocyclization.^[7] It is possible that the analogous methyl transfer in the biosynthesis of 25 proceeds with the opposite stereochemical preference, or that epimerization occurs during the biosynthesis. Alternatively, it is possible that epimerization of the C20 methyl group occurs independent of the biosynthetic machinery, and may be driven by macrocyclic conformational stability. The biosynthetic origin of the stereochemical inconsistencies at C4 remains an interesting unsolved question.

In conclusion, we have developed the first fully synthetic route to acyclic members of the lankacidin class in a linear sequence of ≤8 steps from simple precursors (≤10 steps from commercially available compounds). This route enabled investigation of stereochemical variability on the β-keto-δ-lactone core, resulting in the first chemical synthesis of 2,18-seco-lankacidinol B and evidence that the original stereochemical assignment of C4 requires revision from R to S. Additionally, we conducted the first evaluation of stereochemically diverse, acyclic lankacidin derivatives against a panel of Gram-positive and Gram-negative bacteria, the results of which suggest that the macrocyclic ring or the C18 pyruvamide (or both) are required for substantial antibacterial activity. Efforts to generate cyclic lankacidin derivatives that grant insight into origin of C4 stereochemical diversity and have improved chemical stability are underway.