Total Syntheses of Stereoisomeric Congeners of Rufomycin Natural Products Having Anti-Mycobacterial Activity

Abstract

Syntheses of bioactive rufomycin natural product congeners that differ in stereochemistry at a conserved methyl-substituted amino-hydroxy-piperidone (Ahp) moiety have been achieved for the first time. In addition to gaining new insights into the formation and conformation of the Ahp stereoisomers, we were intrigued to find that theirs reduction improves rufomycin bioactivity significantly.

The rufomycins (also known as the ilamycins) are ∼30 heptapeptide natural antibiotics produced by Streptomyces species. They have potent activity against Mycobacterium tuberculosis (Mtb). Long-standing ambiguities about their structures were resolved recently by multi-dimensional NMR spectroscopy and/or X-ray crystallography. Most rufomycins contain alanine, 3-nitro tyrosine [Tyr­(3-NO₂)], N-Me leucine, 5,6-dehydro norleucine (dhNle), and leucine residues (Figure ). Some congeners structurally diverge by virtue of the oxidation state of a conserved N-alkyl tryptophan residue and the stereochemistry of methyl-substituted amino-hydroxy-piperidones (Ahps) derived from oxidation and intramolecular cyclization of 5-hydroxyleucines having either R or S configuration at C4. Indeed, Zhou and co-workers executed structural elucidation and biological characterization of four natural rufomycins (4–7) (Figure ), having all four possible stereoisomers of the Ahp. The stereochemical diversity of the Ahp moieties reflects their origin via a non-enzymatic, spontaneous cyclization of chiral aldehyde intermediates.

1.

Structures of rufomycins 4–7 and ruf 4′–7′.

We have been intrigued by the rufomycins’ Ahp moieties because of their importance for bioactivity and stereochemical complexity. In related bioactive depsipeptide natural products having this moiety, there is a strong bias towards one of two possible epimers that is enforced by intramolecular hydrogen bonding with donors outside of the ring system. The rufomycins stand apart from those molecules because they are methyl-substituted at C4 and all four possible stereoisomers are represented. In contrast to other Ahp-containing natural products, there is no intramolecular hydrogen bonding in the structure of rufomycin 5, which suggests that the stereochemistry of the Ahp in all of the other family members is not influenced by non-covalent bonding. The possibility that the stereochemistry of C4 could influence conformational preferences and stereoisomer distributions in Ahp formations and antimycobacterial activity motivated us to synthesize rufomycin 4–7 and their des-epoxy congeners ruf 4′–7′.³ To the best of our knowledge, the complete set of rufomycins 4–7 has not been collectively synthesized and characterized. Thus, we developed a solid-phase synthetic approach through which rufomycins 4–7 (and all other known congeners) can be prepared.

Our synthesis began with preparation of the four non-commercially available amino acids (HyLeu, dhNle, N-alkyl Trp, and protected [Tyr­(3-NO₂)-OTBS], Scheme ). For the sake of compatibility with solid-phase peptide synthesis (SPPS), we chose to synthesize Fmoc-N-Me-HyLeu-OH 11 and 12 from bis-lactim ether 8 using commercially available alkyl bromides 9 and 10, respectively.

1. Syntheses of (a) Fmoc-N-Me-HyLeu-OH (11 and 12), (b) Fmoc-Tyr­(3-NO₂-OTBS)-OH (14), and (c) Fmoc-dhNle-OH (18).

To avoid any potential O-modification by SPPS coupling reagents, the reactive phenol group of 3-nitro tyrosine was protected by O-silylation to yield Fmoc-Tyr­(3-NO₂-OTBS)-OH 14.

Though dhNle has been synthesized previously, , we envisioned a higher-efficiency approach using a Neigishi coupling to access Fmoc-dhNle-OH. Commercially available iodo methyl ester 15 was treated with zinc to generate the organozinc Negishi reactant 16. It was reacted with trans-1-bromo-1-propene in the presence of Pd₂(dba)₃ and Sphos, affording Boc methyl ester 17. The desired Fmoc-dhNle-OH compound 18 could be easily obtained by tandem hydrolysis and protecting group interconversion in a three-step route.

The reported structures of rufomycins 4–7 and rufs 4′–7′ have Trp­(N-S-epo) and Trp­(N-prenyl) residues, respectively.³ The Fmoc-protected counterparts of those amino acids, Fmoc-Trp­(N-Prenyl)-OH 22 and Fmoc-Trp­(N-epo)-OMe 27, were synthesized from the same starting material (Scheme ). Fmoc-Trp-OMe 20 was prepared on a large scale from commercially available H-Trp-OMe HCl salt 19 by Fmoc protection. Ester 20 was tert-prenylated following Baran’s protocol with subtle modification, giving rise to Fmoc-Trp­(N-prenyl)-OMe 21. LiOH/CaCl₂-enabled hydrolysis smoothly afforded Fmoc-Trp­(N-prenyl)-OH 22 in high yield. To synthesize optically pure diol 26, we first attempted to utilize Sharpless asymmetric dihydroxylation with ester 21. The reaction proceeded smoothly with high diastereoselectivity (dr ∼ 9:1, data not shown). Unfortunately, separation of the diastereomers by column chromatography was unsuccessful. Resolution of the corresponding tosyl and chiral camphorsulfonyl alcohols also failed (data not shown). Given those challenges, we envisioned an early stage epoxide installation. First, indole ester 20 was reduced to indoline ester 23 by TFA/TES reduction. Inspired by Yamamoto’s work, we treated nucleophilic indoline 23 with known chiral epoxide 24 in the presence of W­(OEt)₆. To our delight, diol 25 was formed exclusively without any C2 ring-opening product. Stepwise DDQ oxidation, regioselective tosylation, and epoxidation afforded the desired (S)-Fmoc amino ester 27.

2. Syntheses of Fmoc-Trp­(N-prenyl)-OH (22) and Fmoc-Trp­(N-epo)-OMe (27).

With the protected amino acid building blocks, we set out to assemble rufomycin precursors via Fmoc-SPPS. We synthesized rufomcyins in a fashion similar to that used to prepare the cyclomarins, given the high degree of similarity in the structures of those natural products. , CTC resin was selected for the following reasons: (i) its mild release condition suppresses unwanted epoxide decomposition under acidic conditions and (ii) it could prevent diketopiperazine formation. Our SPPS commenced with commercial H-Leu-CTC resin 28. HATU was used in the first two couplings to overcome the attenuated reactivity of N-Me HyLeu. After each coupling reaction, Fmoc was removed by 20% 4-Me piperidine (4-MP) in DMF to yield resin-bound tripeptide 29. The coupling reagent was then switched to PyAOP to suppress any potential N-guanylation by HATU. After two iterative coupling/deprotection cycles, resin-bound pentapeptide 30 was synthesized. Couplings of tryptophans 22 and 27 with intermediate 30 were left overnight for completion. Final heptapeptide 32 was obtained after PyAOP-mediated coupling with Fmoc-dhNle-OH 18 followed by Fmoc removal. Subsequently, TBDPS and TBS protecting groups were removed simultaneously by overnight TBAF treatment, after which cyclization substrates 33 were released from CTC resin with 25% HFIP/DCM.

Given the problematic side reactions of PyBOP and HATU coupling reagents with 3-nitro tyrosine, , we envisioned minimizing side reactions using the carbodiimide coupling reagent, EDC. The EDC-mediated macrocyclization suppressed O-modification of the reactive nitro tyrosine residue without compromising the macrolactamization efficiency. We were gratified that 2 equivalents of EDC could effectively mediate the cyclization of linear peptides 33 without noticeable modification at 3-nitro tyrosine. Indeed, four rufomycin congeners (34, 34′, 35, and 35′) bearing N-Me HyLeu were isolated in acceptable yields (20–32%) over 15 steps.

Having the complete set of rufomycin alcohols, we first attempted to execute the oxidative Ahp formation (final synthetic transformation) via the protocol developed by Ye and used by Kazmaier and Payne in their syntheses of rufomycin congeners. In pilot experiments, diastereomeric alcohols (4S)-34′ and (4R)-35′ were oxidized by the Dess-Martin periodinane (DMP) to yield aldehydic Ahp precursors 36′ and 37′, respectively. The crude aldehydes were subjected to K₂CO₃-mediated Ahp ring closure conditions identical to those previously reported. The (4R)-aldehyde 37′ led to the formation of (4R,5R)-ruf 6′ in 52% yield (Figure c), whereas (4S)-aldehyde 36′ surprisingly also yielded (4R,5R)-ruf 6′ as the major product in a slightly lower yield (46%), as evidenced by both UPLC (Figure a) and HSQC (see Supporting Information) analyses. Since the two diastereomeric aldehydes (36′ and 37′) both yielded ruf 6′ as the major product under the same conditions, we concluded that: (i) both aldehydes interconvert via epimerization at C4 under the mildly basic reaction conditions for Ahp formation and (ii) (4R,5R)-ruf 6′ is the thermodynamically favored product via the (4R)-aldehyde 37′ intermediate. From these observations, we were concerned about the inclusion of a base in Ahp formation in stereospecific syntheses of rufomycin congeners and thus the structural identities of the synthetic rufomycins. Comparisons of NMR spectra of our products with those of the natural products and the previously reported molecules affirmed our concerns (see Supporting Information).

2.

Ahp cyclization studies of aldehydes 36, 36′, 37, and 37′ under different conditions.

The problematic epimerization during Ahp formation motivated us to pursue base-free conditions that would not effect epimerization in syntheses of ruf 4′ and 5′. Inspired by a report that an analogous hemiaminal ether formation is promoted by molecular sieves, we used that reagent in Ahp formation and were gratified to find that (4S)-aldehyde 36′ was converted into (4S,5R)-ruf 4′ and (4S,5S)-ruf 5′ in a roughly 1:1 ratio of HPLC-separable diastereomers having distinct UPLC traces (Figure b) and HSQC NMR spectra (see Supporting Information). These observations indicated that the configuration of the chiral center at C4 in the substrate was retained in the product. We applied the epimerization-free conditions to (4R)-aldehyde 37′ and isolated (4R,5R)-ruf 6′ and (4R,5S)-ruf 7′. Interestingly, the former was the major product, whereas the latter was recovered in trace amounts (Figure d). Following our base-free Ahp formation protocol, we also stereospecifically prepared epoxides containing rufomycins (4–7). As predicted, (4S)-aldehyde 36 was cyclized to afford equal amounts of rufomycins 4 and 5 (Figure e). As was the case with 6′ and 7′, the cyclization of (4R)-aldehyde 37 strongly favored (4R,5R)-rufomycin 6 over (4R,5S)-rufomycin 7 (Figure f). Across all of these reactions with C4-methyl-substituted aldehyde intermediates, we note that the Ahps having the S configuration at C5 could be formed. These observations differ from those in the chemical syntheses of natural products lacking C4 substituents wherein products with the R configuration at C5 are formed almost exclusively. Apparently, the configuration of C4 influences the stereochemical outcome of Ahp formation.

Combining our Ahp cyclization results and previous studies, ,, we propose a pathway for interconversion of the four diastereomers during Ahp cyclization (Figure ). Under base-free conditions, C4 of the aldehyde precursor of the Ahp does not epimerize, yet there is reversibility in the formation of the Ahp as reflected in a mixture of (5S)- and (5R)-diastereomers. Accordingly, pairs of diastereomeric Ahps (4′ and 5′, 6′ and 7′, 4 and 5, and 6 and 7) can be formed from (4R)- and (4S)-aldehyde substrates (panels b and d–f, respectively, of Figure ). On the contrary, K₂CO₃-promoted Ahp condensation causes C4 epimerization of aldehydes 36, 36′, 37, and 37′, yielding mixtures of Ahp diastereomers (Figure a and c). Under the basic conditions, the actual ratios of isomers are governed by intrinsic differences in the products’ thermodynamic stabilities. Apparently, rufomycin 6 and ruf 6′ are the most stable epimers (4R,5R); indeed, we found that they undergo extremely slow stereochemical interconversion at C5 over a period of months. In contrast, rufomycin 7 and ruf 7′ (4R,5S) are quite unstable, yielding an ∼1:1 mixture of C5 epimers 6 and 7 or 6′ and 7′ after evaporation of HPLC fractions. Interestingly, epimeric rufomycins and/or rufs having the S-configuration at C4 (4 and 4′ or 5 and 5′) are less stable than rufomycin 6 and ruf 6′ (4R,5R). Specifically, C5 in rufomycin 4 and ruf 4′ (4S,5R) partially epimerizes within a day, whereas rufomycin 6 and ruf 6′ (4R,5R) epimerize over months. Rufomycin 5 and ruf 5′ (4S,5S) have configurational stability similar to that of rufomycin 4 and ruf 4′ (4S,5R), but both are much more stable than rufomycin 7 and ruf 7′ (4R,5S). To summarize, with C4 R-methyl Ahps, the S-configuration at C5 is thermodynamically disfavored, as evidenced by the negligible formation of rufomycin 7 and ruf 7′ and the gradual conversion of rufomycin 7 and ruf 7′ into rufomycin 6 and ruf 6′, respectively. In contrast, with C4 S-methyl Ahps, there does not appear to be a thermodynamic bias with respect to the configuration at C5, as rufomycins 4 and 5 and rufs 4′ and 5′ are formed in equal amounts and slowly interconvert into one another.

3.

Hypothetical interconversion of rufomycin congeners via chiral aldehyde intermediates.

Rufomycin natural products 4–7 are reported to have exceptional activities against Mtb, ,, making them promising drug candidates for tuberculosis drug development. As a further validation of our syntheses, we determined that our synthetic rufomycins had bioactivities mirroring those reported for the natural products (Table ). As is the case for other Ahp-containing natural products, rufomycin alcohols 34, 35, 34′, and 35′ are much less potent compared to congeners with the Ahp moiety (4–7 and 4′–7′). In contrast to bioactive Ahp-containing natural products that exist as a single epimer of the hemiaminal due to stabilization by intramolecular hydrogen bonding, we found that the C5 epimers of the rufomycins not influenced by hydrogen bonding have similar anti-mycobacterial activities. A modest determinant of potency among the congeners is the epoxy tryptophan common to rufomycins 4–7. Nevertheless, the apparent irrelevance of C5 chirality to bioactivity prompted us to execute a reduction of the Ahp moiety in ruf 4′ and assess the product’s activity. We were intrigued that desoxyruf 4′ (R₁ = (S)-Me and R₂ = H, Scheme ; see Supporting Information for the synthetic details) was 4-fold more potent than its counterparts, rufomycin 4′ and 5′. Apparently, the rigidity imparted by the cyclic Ahp moiety is very important for the bioactivities of the rufomycins, yet the hydroxyaminal is a liability for potency.

1. In Vitro Activities of Synthetic Rufomycins against M. tuberculosis H37Rv.

compound	MIC90 (nM)	compound	MIC90 (nM)
rufomycin 4	30.0	ruf 4′	122
rufomycin 5	60.0	ruf 5′	122
rufomycin 6	60.0	ruf 6′	60.9–122
rufomycin 7	60.0	ruf 7′	244
34	15300	34′	31100
35	9580	35′	31100
isoniazid	291	desoxyruf 4′	31.0

^aFreshly purified compounds were used for testing, but stereochemical instability likely influences the observed bioactivities.

3. Syntheses of Rufomycin Congeners via SPPS.

In summary, we have developed a general and efficient synthetic approach to the rufomycin family of natural products. An epimerization-free Ahp formation protocol was established, which enabled the stereospecific synthesis of all possible Ahp-containing rufomcyin congeners 4–7 and ruf 4′–7′ for the first time. Our study illustrated that the Ahp scaffold and its reduced counterpart significantly impact rufomycin potency, whereas the epoxide from Trp renders a comparatively minor influence. Guided by our present work, further structure–activity relationship studies of rufomycins are underway in our laboratories.

The data underlying this study are available in the published article and its Supporting Information.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.orglett.5c01525.

• Experimental procedures and characterization data (PDF)

The authors declare no competing financial interest.